...

HIJNEN 2009 Elimination of MIcro-Organisms in Water Treatment

by user

on
Category: Documents
6

views

Report

Comments

Transcript

HIJNEN 2009 Elimination of MIcro-Organisms in Water Treatment
Elimination of micro-organisms
in water treatment
Published by
KWR Watercycle Research Institute
PO Box 1072
3430 BB Nieuwegein
The Netherlands
Tel. +31(0)306069511
Elimination of micro-organisms in water treatment
Author:
ISBN:
KWR:
Thesis:
Subject headings:
W.A.M. Hijnen
978-90-74741-92-7
BTO 2008.048
University Utrecht
Elimination of micro-organisms, safe drinking water,
E. coli, Clostridium spores, process indicator
Cover:
Beach of French coast between Wisant and Cap Gris
Nez, 2008.
Wim Hijnen
Koen and Jannie Hijnen, Wim Demmenie, Gerard
Meester
Gildeprint Drukkerijen BV
Photograph:
Advices:
Printed by:
Elimination of micro-organisms
in water treatment
Eliminatie van micro-organismen in
de waterzuivering
(met een samenvatting in het Nederlands)
Proefschrift
ter verkrijging van de graad van doctor aan de Universiteit
Utrecht op gezag van de rector magnificus, prof.dr. J.C. Stoof,
ingevolge het besluit van het college voor promoties in het
openbaar te verdedigen op donderdag 29 januari 2009 des
ochtends te 10.30 uur
door
Wilhelmus Antonius Maria Hijnen
geboren op 6 juni 1956
te Bussum
Promotoren: Prof.dr. A.H. Havelaar
Prof.dr. F. van Knapen
Co-promotor: Dr. G.J. Medema
Het onderzoek in dit proefschrift is uitgevoerd bij Kiwa Water
Research en bij een aantal waterleidingbedrijven en is gefinancierd
door het gezamenlijk onderzoeksprogramma BTO van de
Nederlandse drinkwaterbedrijven en aanvullende financiering
door het waterleidingbedrijf Waternet.
Aan Jannie, Niek, Koen en Floor
VOORWOORD
Belangrijke drijfveer voor het schrijven van dit proefschrift is geweest om de
belangrijkste mijlpalen van mijn onderzoek van de afgelopen jaren dat is
uitgevoerd in het kader van de productie van microbiologisch veilig drinkwater in
een totaal overzicht te plaatsen. Veilig drinkwater is een belangrijke
levensvoorwaarde die in onze samenleving een vanzelfsprekendheid is en waar op
wordt vertrouwd. Ik vond het een uitdaging om een onderzoek te kunnen doen
dat bijdraagt aan het wetenschappelijke inzicht in de microbiologische veiligheid
van het Nederlandse drinkwater wat laat zien dat dit vertrouwen terecht is, maar
ook welke inspanningen de Waterleidingbedrijven daarvoor moeten verrichten. De
kennis gepresenteerd in dit proefschrift is weliswaar toegesneden op de
ontwikkelde landen maar levert daarnaast hopelijk ook een bijdrage aan de
realisatie van veilig drinkwater voor mensen in de minder ontwikkelde landen.
Ik ben bij het realiseren van dit proefschrift gesteund en geïnspireerd door veel
mensen om me heen. Dit proefschrift zou niet totstand zijn gekomen zonder het
vertrouwen van en de samenwerking met de Nederlandse Drinkwaterbedrijven
uitgesproken en ondervonden in het gemeenschappelijke Bedrijfstakonderzoeksprogramma (BTO) dat Kiwa Water Research (KWR) voor hen uitvoert.
Gertjan Medema, je was een belangrijke inspriratiebron en steun in het
onderzoek. Je uitgesproken vertrouwen en je positief kritische begeleiding waren
onmisbaar om dit promotietraject te starten en het proces af te ronden. Je grote
microbiologische kennis en je enthousiasme voor het onderwerp zijn en blijven
voor mij een belangrijk voorbeeld. Ik vind het inspirerend om samen met jou dit
onderwerp verder uit te diepen de komende jaren.
Arie Havelaar, ik heb je bijdrage in het laatste deel van het wordingsproces
van het proefschrift ondervonden als zeer bijzonder en verrijkend. Je
wetenschappelijke bijdragen als pionier en kenner van de microbiologie van
drinkwater en je wijze adviezen bij het afronden van het proefschrift, zijn voor mij
zeer waardevol geweest. Ik zie ernaar uit om een gezamenlijke publicatie te maken
op grond van de kennis gepresenteerd in dit proefschrift.
Frans van Knapen, jouw steun en wijze raad hebben in de afrondingsfase
een belangrijke bijgedrage geleverd aan de kwaliteit van dit proefschrift. Dank
daarvoor.
Dick van der Kooij, dankzij jouw grote enthousiasme, inspiratie en grote
kennis heb ik me kunnen ontwikkelen tot een zelfstandige microbiologisch
onderzoeker. Je originele ideeën op het gebied van de toegepaste microbiologie
van de drinkwatervoorziening en soms bovenmenselijke inzet waren en blijven
hierbij onmisbaar. Ik vind het dan ook waardevol om samen met je dit vak verder
te mogen verkennen.
Een groot deel van het onderzoek dat in dit proefschrift is beschreven, is tot stand
gekomen in nauwe samenwerking met de collega’s van vooral het microbiologisch,
maar ook het chemisch laboratorium van Kiwa Water Research (KWR). Belangrijk
onderdeel van het onderzoeksproces is de organisatie en uitvoering dat begint bij
Harm Veenendaal, die als hoofd van het microbiologische laboratorium zorgt voor
een optimale kwaliteit van het werk in een prettige werksfeer. De onderzoeken met
de MF-sampler en de ozonopstelling zijn tot stand gekomen in nauwe en goede
samenwerking met Ellen van der Speld, Didi en Ton Braat als microbiologisch
analisten. Hans van Beveren was als chemisch analist en technicus onmisbaar bij het
verzamelen van de gegevens van de ozonproeven. Anke Brouwer, je was een
belangrijke steunpilaar bij de verschillende proeven die we hebben gedaan. Je
droeg met je kritische blik bij aan het verzamelen van waardevolle gegevens en ik
beleef veel plezier aan onze samenwerking. Marijan Uytewaal-Aarts, Carola Blokker,
Gaby van Doorn-Abelman, Anita van der Veen, jullie hebben in meer of mindere mate
bijgedragen aan de divere onderzoekingen en zorgen voor een goede werksfeer
waardoor ik altijd graag bij jullie in het laboratorium kom. Meindert de Graaf, dank
voor jouw waardevolle bijdrage en inzet bij het verzamelen van de gegevens bij de
externe projecten. Je betrokkenheid en warme belangstelling voor iedereen vind ik
bijzonder.
Bij de ontwikkeling van de MF-sampler en de onderzoeken met opstellingen is het
van groot belang om te kunnen steunen op een goede fijn technische instrumentele
afdeling. Ton van Dam, Sidney Meijering en Harry van Weegen, jullie technisch
inzicht, inzet en praktische kennis waren onontbeerlijk voor het onderzoek. Ook
dank ik de collega’s van de afdeling Behandeling Jil Verduin, Ron Jong en Erwin
Beerendonk voor de prettige samenwerking. Pim Bogaerds ben ik erkentelijk voor
zijn ondersteuning bij het verkrijgen van de literatuur.
Het onderzoek vormde onderdeel van het Aandachtsveld Microbiologische
Veiligheid dat door het team Microbiologische Waterkwaliteit en Gezondheid
wordt uitgevoerd. Patrick Smeets, Pieter Nobel, Paul van der Wielen, Leo Heijnen, Hein
van Lieverloo, Jack van de Vossenberg en Luc Hornstra, ik dank jullie voor de
vruchtbare inhoudelijke gesprekken maar ook de prettige persoonlijke contacten.
Patrick, de laatste jaren waarin we beiden bezig waren met het boekje hebben we
veel gepraat over het onderzoek en de meetresultaten maar ook over meer
persoonlijke zaken. Deze gesprekken en jouw statistische kennis waren hierbij
bijzonder en inspirerend. Paul, ik heb de laatste jaren veel gehad aan je grote
onderzoekservaring, microbiologische kennis en positief kritische kijk op veel
dingen. Ik deel dan ook met veel plezier een kamer met je. Leo, ik leer veel van je
op het gebied van de moleculaire microbiologie en ervaar ons collegiaal contact als
bijzonder prettig. Pieter, jij ook bedankt voor onze prettige samenwerking. Het
originele idee om met centrische diatomeeën te gaan werken kwam van jou. Al ben
je zo goed als weg bij ons, je voelt nog als één van ons. Hein, ook jij was bijzonder
voor mij en voelt nog als een van ons. Je grote enthousiasme en inzichten op
allerlei gebied hebben me geraakt.
Naast mijn collega’s in de “Veiligheid” dank ik ook mijn collega’s in de
“Activiteit”, Bart Wulling, Rinske Valster en Evelien Sack. Bart ik beleef altijd veel
plezier aan je spontane en warme manier waarmee je met je collega’s omgaat.
Rinske en Evelien, jullie enthousiame als collega en onderzoekers in opleiding zijn
aanstekelijk. Last but not least, Gemma van Beusekom, je bent een steun en toeverlaat
voor mij en onze club; dank hiervoor.
Een goede samenwerking met mensen in de bedrijfstak is van groot belang
geweest voor het onderzoek. Voor de langzame zandfiltratieproeven ben ik Ton
Visser en Yolanda Dullemont en de andere leden van de werkgroep Jack Schijven,
Wim Oorthuizen, Gerhard Wubbels, Jantinus Bruins, Aleksandra Magic en Martine
Rosielle zeer erkentelijk. Bram van der Veer, dank voor onze samenwerking bij de
ozonoproeven.
I thank also a number of my international colleagues and friends. Nick and Roslyn
Ashbolt, thanks for your hospitality in Australia. Nick, I regard your invitation to
visit your University and work with Jin Chung as your PhD student as a special
and a valuable experience. Cheryl Davies, Katrina Charles, David Roser, Christine
Kauchner are my friends from UNSW. Cheryl, you were a support to me in
optimization of my English writing. Thor-Axel Stenström, I have highly appreciated
your membership in the advisory committee of this Thesis and your valuable
comments on the manuscript. My visits to congresses were extra valuable because
of the warm collegiality of my Canadian friends Benoit Barbeau, Michelle Prévost and
Francoise Bichai and all the other international colleagues.
Ik had dit onderzoek niet tot stand kunnen brengen zonder groei in mijn
persoonlijke ontwikkeling, maar ook niet zonder de nodige geestelijke en
lichamelijke ontspanning. Daarbij zijn mijn dierbaren en vrienden van groot belang
geweest. Jannie, ik was soms veel ‘afwezig’ en in de laatste fase was het boekje een
belangrijk gespreksonderwerp. Ik zie er naar uit om weer samen te wandelen en
theaters te bezoeken of zo maar samen te reizen en te lezen. We genieten samen
van onze prachtige kinderen Niek, Koen en Floor die ons beiden gelukkig maken. Pa
en Ma, bedankt voor de basis die door jullie is gelegd en de mogelijkheden die
door jullie zijn geboden. Mijn broer en zussen, Hens, Ria en Tineke, en ook Pa en Ma
Hompus, schoonzussen en zwagers bedankt voor jullie altijd warme belangstelling.
Hannie Nagelkerke wil ik bedanken voor haar waardevolle professionele en
persoonlijke bijdragen in mijn persoonlijke ontwikkeling. Mijn vrienden waren
onontbeerlijk voor de nodige diepgaande gesprekken, hardloop- en
wandelactiviteiten, muzikale ontmoetingen en culturele bezoekjes. Koos SpreenBrouwer, André Jansen, Arie Noordsij, Theo Noij, Guus Heiming† en al mijn andere
vrienden: bedankt daarvoor!
CONTENT
Page
Chapter 1
Microbiologically safe drinking water
1
Abbreviations and calculations
53
Chapter 2
Indicator bacteria concentrations in water
treatment and assessment of elimination
capacity
57
Chapter 3
Enumeration of faecal indicator bacteria in
large water volumes using on site membrane
filtration to assess water treatment efficiency
73
Chapter 4
Quantitative assessment of the removal of
indicator bacteria in full-scale treatment plants
87
Chapter 5
Spores of sulphite-reducing clostridia (SSRC)
as surrogate for verification inactivation
capacity of full-scale ozonation for
Cryptosporidium
109
Chapter 6
Inactivation credit of UV-radiation for viruses,
bacteria and protozoan (oo)cysts: a review
121
Chapter 7
Elimination of viruses, bacteria and protozoan
oocysts by slow sand filtration
159
Chapter 8
Removal and fate of Cryptosporidium parvum,
Clostridium perfringens and Stephanodiscus
hantzschii in slow sand filters
173
Chapter 9
Transport of phage MS2, Escherichia coli,
Clostridium perfringens, Cryptosporidium parvum
and Giardia intestinalis in a gravel and a sandy
soil
201
Chapter 10
General discussion
229
Summary
265
Samenvatting
273
List of publications
283
Curriculum vitae
289
Chapter 1
Microbiologically safe
drinking water
___________________________________________________________________
-1-
Chapter 1
.
ABSTRACT
The management of the microbiological drinking water quality in the
Netherlands has been changed in the past ten to fifteen years. One of the
major changes was the shift from curative quality management to a more
preventive strategy. In the first part of this first Chapter the historical
perspectives of the microbiological quality of drinking water from the late
19th Century untill the beginning of the beginning of the 21th is described.
Motives for the change from curative to preventive managements were
epidemiological studies on waterborne diseases, scientific knowledge on
the dose-response data of some waterborne pathogenic micro-organisms
and shortcomings in water quality control. In the Netherlands this resulted
in the introduction of a health based microbial target and the requirement
to assess this target with Quantitative Microbial Risk Assessment (QMRA)
in the revised Drinking Water Decree in 2001. Important requirement for
QMRA is quantitative information on the capacity of drinking water to
eliminate (remove or inactivate) micro-organisms.
Overall objective of the present study was to develop a generic
methodology to collect this information. The line of approach in this study
was to develop a method as closely related to the natural conditions in
water treatment and to the daily practice of microbiological water quality
monitoring. This is described and motivated in the second part of this
Chapter where the hypothesis was introduced that beside their role as
indicator of faecal pollution, faecal indicator bacteria can be used as process
indicators for the elimination of different waterborne pathogens. In the last
part of this Chapter the overall objective is specified in a number of subgoals on the basis of research needs. In the outline paragraph the separate
Chapters are mentioned focused on the sub-goals of the study.
WATERBORNE DISEASES AND DEVELOPMENTS IN
CENTRAL WATER SUPPLY
Throughout the ages, the use of fresh water by mankind for drinking was
dictated by the way that they lived. As nomads, people used water available
from the rivers and streams in their immediate vicinity. When they settled,
and civilizations developed, more advanced centralised water supplies with
mains were installed. Well known are the aqueducts from the Roman
civilization. Already in older civilizations like that of the Egyptians, Persians
and Pakistanis evidence of centralized water supply have been found
(Wijmer, 1992). After the Roman civilization vanished from Northern
___________________________________________________________________
-2-
Chapter1
Europe, their achievements in water supply disappeared for the common
people. Only for the rich people, churches and monasteries a kind of central
water supply from wells remain in practice. It was only since the 19th
century, the century of the industrial revolution, that central drinking water
supply for the more settled communities at that time was developed. Crucial
for this development was the growing knowledge about waterborne
diseases. Even after more than hundred years this development is valued
highly as shown by a recent held internet poll by the British Medical Journal.
Sanitation and clean water was seen as the major milestone in medical
advances since 1840 (http://www.bmj.com/cgi/content/full/334/suppl_1/DC3).
Centralised public drinking water supplies were introduced in the 19th
century and one of the major drivers for this development was public health.
John Snow was the first (1849) to relate the use of contaminated water with
the incidence of cholera in the population, one of the major infectious
diseases at that time (Sheppard, 1995). After several outbreaks of cholera in
1832/33, 1848/49, 1853, 1854, 1855, 1859 and 1866, the role of drinking water
in this severe public health problem was also recognised in the Netherlands.
A national Dutch committee ‘Tot onderzoek van drinkwater’ (Weelden and
Mingelen, 1868) came to the conclusion that a ‘cholera-germ’ originating
from the faeces of infected people was the cause of the distribution of this
disease. Pasteur (1822-1895) and Robert Koch (1843-1910) were the first
scientists who showed that micro-organisms can cause diseases. Koch
developed a method to cultivate these micro-organisms on a solid medium
(Koch’s “Plattengussverfahren”). With this microbiological method Koch
isolated the bacteria which caused cholera later known as Vibrio cholerae
(Koch, 1884). He introduced the following postulates to identify a microorganism as a pathogen:
- The micro-organism must be found in all organisms suffering from
the disease, but not in healthy organisms.
- The micro-organism must be isolated from a diseased organism and
grown in pure culture.
- The cultured micro-organism should cause disease when introduced
into a healthy organism.
- The micro-organism must be re-isolated from the inoculated,
diseased experimental host and identified as being identical to the
original specific causative agent.
To demonstrate the significant role of centralised drinking water supply in
improvement of public health, the decrease in frequency of typhus in the
Dutch population, another waterborne disease, was correlated with the
___________________________________________________________________
-3-
Chapter 1
.
increase in number of inhabitants which were connected to a central
drinking water supply (Figure 1).
Nowadays the quality of drinking water in the Netherlands is usually valued
highly by consumers with regard to both perception and safety. The
consumer trust in Dutch drinking water was not always obvious in recent
history, as observed for instance around in the end of 19th century in
Rotterdam (Wijmer, 1992). Confidence of the consumers in the city of
Rotterdam in the distributed drinking water was undermined by the
distribution of turbid water caused by growth of organisms in the
distribution system. The public suspected a relationship with the drinking
water and the occurrence of diarrhoea although this was not supported by
evidence. This situation got worse due to a local waterborne outbreak of
typhus. A point source sewer leakage into the deteriorated drinking water
distribution network was found to be the cause and due to this incident the
two Water Company managers responsible voluntarily resigned (Wijmer,
1992).
70
60
Cases of typhoid
% connection to
per 100000
mains supply
100
90
50
80
40
70
30
60
20
50
10
40
0
30
1900 1910 1920 1930 1940 1950 1960 1970
Figure 1. Frequency of typhus and the percentage of inhabitants connected to
central drinking water supplies in the period 1900-1970 (from Van der Kooij,
2002; sources: CBS and VEWIN).
In the recent period of approximately 1960 until the present day only three
waterborne outbreaks have been reported in the Dutch drinking water
practice related to the consumption of faecal contaminated drinking water.
The first one related to contamination with sewage occurred in 1962 when 5
___________________________________________________________________
-4-
Chapter1
cases of typhoid fever were reported in Amsterdam (Anonymous, 1962). The
second one was a larger outbreak with 609 reported cases of mainly
gastroenteritis caused by wastewater from a Marine vessel pumped into the
drinking water mains by a mal-connection (Huisman and Nobel, 1981). The
last reported outbreak dates from 2001, with an excess of gastroenteritis
cases in a new residential area with the application of a dual water system
with grey water. A cross connection in this system caused the consumption
of partially treated river Rhine water (Fernandes et al., 2007).
During the late 20th century a number of events, observations, scientific
achievements and developments in society influenced the common practice
associated with the production of microbiologically safe drinking water in
the developed countries. Waterborne outbreaks and systematic
epidemiological studies in US and other developed countries showed that
consumers in these countries may be exposed to pathogens in drinking
water. Due to increased urbanization surface water pollution increased and
not only the amount of known waterborne pathogens increased but also the
variety including more persistent species such as some viruses and protozoa
(Cryptosporidium and Giardia). Moreover, it was demonstrated that current
water treatment is not an absolute barrier against pathogens, and compliance
with the coliform standard is not always sufficient to safeguard public health
against these waterborne pathogens. In the following parts of this
introduction these developments will be highlighted.
EPIDEMIOLOGICAL STUDIES ON WATERBORNE DISEASES
Major pathogenic micro-organisms causing waterborne diseases in the 19th
century were bacteria (V. cholerae, Salmonallae and Shigella spp.).
Epidemiological studies during the last period of the 20th century in the
United States of America (Craun, 1988; Craun, 1990; Craun et al., 1998;
Herwaldt et al., 1992; Moore et al., 1994) and United Kingdom (Galbraith et
al., 1992) showed that major causative pathogens of outbreaks of waterborne
diarrhoea in the current time, however, are no longer bacteria but viruses
and pathogenic protozoa (Cryptosporidium and Giardia). The outbreak of
cryptosporidiosis in Milwaukee (US) was one of the largest reported
outbreaks (MacKenzie et al., 1994). Moreover, when bacteria are involved
other species than V. cholerae, Salmonallae or Shigella spp. have become of
importance such as Campylobacter spp. and enteropathogenic E. coli O157. A
recent local waterborne outbreak in Canada was related to the consumption
of contaminated drinking water and both C. jejuni and E. coli O157 were
___________________________________________________________________
-5-
Chapter 1
.
identified as the main pathogens (Hrudey et al., 2003). Epidemiological
studies in the Europe also demonstrated a prominent role for Cryptosporidium
in waterborne outbreaks (Medema et al., 2006) and a shift towards
Campylobacter as the causative agent in the reported bacterial outbreaks. The
prevalence of these new waterborne pathogenic bacteria in Dutch surface
water was demonstrated recently by de Roda Husman et al., (2001) and
Heijnen and Medema (2006).
Epidemiological surveys on waterborne outbreaks are valuable in assessing
the emerging pathogens for drinking water safety. Another type of
epidemiological study is the randomised controlled trial (RCT). With such a
study Payment et al. (1991, 1997) demonstrated that a relatively high
percentage of annual gastrointestinal illnesses among the consumers (1440%) in a water supply were water-related. In these studies, however,
consumers were not blinded to the type of water they received. Follow up
RCT studies which were double- and triple-blinded (Hellard et al., 2001;
Colford et al., 2005) showed no significantly increased numbers of
gastrointestinal illnesses as a result of the consumption of less treated
drinking water. Nevertheless, the level of endemic disease due to public
drinking water systems remains difficult to quantify and well documented
waterborne outbreaks are still the main source of epidemiological
information used by microbiologists to develop new microbial regulations
on drinking water.
CONTROL OF WATERBORNE DISAESES: WATER
TREATMENT
The rationale behind treating water prior to distribution for potable use was
the idea that it would protect the public from waterborne diseases like
cholera. After further exploration with his “colony count” method during the
Hamburg Outbreak in 1892, Koch (1893) demonstrated that slow sand filters
were effective barriers against cholera and were able to reduce the number of
bacteria expressed in colony counts to less than 100 per ml. Frankland and
Frankland (1894) summarized the scientific knowledge on water
microbiology that was available at that time. They concluded that the
number of bacteria in water is related to the pollution rate of the water and
that slow sand filtration is capable to reduce the number of bacteria in water
with more than 90%.
Slow sand filtration was a first step in water treatment introduced by John
Gibb in 1804. But in the twentieth century it became clear that the process has
its limitations (rapid clogging, breakthrough at low temperatures and
___________________________________________________________________
-6-
Chapter1
incomplete removal of undesirable organic compounds). Physical/chemical
processes like coagulation and sedimentation as well as chemical disinfection
with chlorine were added to treatment practices of drinking water. John
Snow was one of the first to use chlorine in order to disinfect drinking water
(White, 1999). It was not until the beginning of the 20th century that the use of
chlorine as a disinfectant became common practice in drinking water supply.
In 1906, the first ozonation plant for disinfection was built in France. The
driving force for this development was not only public health, but also the
growing public demands for clear and aesthetically sound water. Besides
these disinfection practices, from 1920 facilities started to use sedimentation,
filtration and chlorination. Under the influence of ongoing environmental
pollution granular activated carbon (GAC) filtration was introduced to
remove organic micro-pollutants. The discovery of the production of trihalomethanes (Rook, 1974) and other disinfection by-products (DBP) by
chlorine caused a decline in the use of disinfection with chlorine in the
Netherlands. Ozonation as an alternative for chlorination was introduced,
because this disinfectant was effective against the chlorine resistant
Cryptosporidium (Finch et al., 1993). Due to the discovery of the production of
the mutagenic disinfection by-product bromate by ozone in bromide
containing waters (Haag and Hoigné, 1983), this disinfectant is still not used
worldwide in water industry. Though Havelaar et al. (2000) demonstrated in
a case study that the microbial benefits of ozone disinfection can outweigh
the health risks of bromate by a factor of 10 on the basis of a single measure
of disease burden, the Disability adjusted life-year (DALY). During the last
decade advances in technology and scientific knowledge resulted in
application of membrane filtration and UV disinfection in drinking water
production. The reason for the renewed interest for UV is that despite earlier
communications, recent studies showed high efficacy against
Cryptosporidium (Clancy et al., 1998).
CONTROL OF WATERBORNE DISEASES: WATER QUALITY
Monitoring for pathogenic micro-organisms. As described
previously, microbiologically safe drinking water was one of the major
driving forces in the development of the drinking water treatment. During
the last period of the 19th century, after the methodology breakthrough
primarily caused by the work of Koch, considerable effort was put into the
assessment of the microbiological quality of drinking water. The work was
primarily focussed on the direct isolation and identification of pathogenic
bacteria because it was assumed that bacteria in water might cause diseases.
___________________________________________________________________
-7-
Chapter 1
.
Scientific knowledge evolved to the general understanding at the end of the
19th century that not all bacteria present in water are pathogenic (Frankland
and Frankland, 1894). Moreover, direct monitoring of pathogenic microorganisms in drinking water proved to be difficult and not practical for
monitoring the safety of drinking water. Though knowledge increased over
the last hundred years this is still the current situation. There is a great
diversity of pathogenic micro-organisms of faecal origin in surface water and
usually the concentrations are low and variable. Some of these microorganisms can not be cultured, such as noroviruses. For others, methods are
complicated with low recovery efficiencies and not selective for the species
significant to human health (Cryptosporidium). The introduction of molecular
methods to detect micro-organisms in water offers additional possibilities to
detect non-culturable pathogenic species with relatively quick methods.
Currently this new analytical technology is still in the stage of development.
Faecal indicator bacteria: coliforms and E. coli. Although the
intensive microbiological studies during the last part of the 19th century did
not result in a practical strategy to assess safety of water with direct
pathogen monitoring, an alternative strategy was developed. Bacterium colicommune (corresponding approximately to Escherichia coli) was isolated by
Escherich (1885) from the faeces of a patient suffering from cholera. It was
subsequently found that this bacterium, capable of fermenting lactose with
the production of acid and gas, was a normal inhabitant of the intestinal tract
of man and many other animals and is excreted with the faeces in the
environment. Schardinger (1892) and Smith (1895) independently introduced
the examination of potable water for Escherichia coli to demonstrate pollution
with faecal matter, a serious health threat as illustrated by the presence of
cholera bacteria in faeces of infected persons. E. coli turned out to be a
member of a larger group of bacteria defined as coliforms, gram-negative
non-spore forming and rod-shaped bacteria capable to ferment lactose with
gas formation within 48 hours at 35oC in selective media with briljant green
and bile-salts. This group of coliform bacteria is used in common
microbiologically drinking water quality monitoring for faecal
contamination all over the world for most of the last century. The
widespread presence of coliforms in the environment made it clear that
probably not all coliforms were of faecal origin. Eijkman (1904) was the first
to apply higher incubation temperature (46oC) to select for coliforms strictly
related to faecal contamination and hence restricting it largely to the
detection of B. coli-commune (E. coli). Later the determination of coliform
species producing gas from lactose at 44.5oC within 24 hours was introduced
as faecal coliforms or thermotolerant coliforms in water quality monitoring.
___________________________________________________________________
-8-
Chapter1
Only recently in European regulation the use of the coliform group in water
sanitation was further restricted E. coli, because regrowth of thermotolerant
coliforms has been demonstrated in practice (Caplenas and Kanarek, 1984).
Additional faecal indicators: faecal streptococci and C. perfringens.
Around the turn of the century (19th -20th) there were other bacteria identified
as bacteria related to sewage pollution, such as faecal streptococci (Houston,
1899) and Bacterium enteritidis sporogenes later known as Clostridium
perfringens (Klein, 1897-1898). The genus Enterococcus was first described by
Thiercelin (1899). Sherman (1937) was the first one to classify the genus
Streptococcus in the Pyogenes, Viridans, Lactic and Enterococcus groups
(Godfree et al., 1997). The first descriptions of Clostridium perfringens were
given by Achalme (1891-1897) and Welch and Nuttall (1892) based on
observations in tissue autopsies from patients. They described the observed
organism as a lactose-fermenting spore-forming anaerobic bacterium. In 1899
Klein and Houston (1898-1899) proposed the use of Clostridium welchii or
Clostridium perfringens as indicator for faecal pollution. According to Bonde
(1962), they were the first to introduce the term “bacteria of indication” for
faecal pollution, for the coli-aerogenes group, faecal streptococci and
Bacterium enteritidis sporogenes (C. perfringens).
Wilson and Blair (1925) showed a relationship between the presence of
anaerobic sulphite-reducing spore-forming bacteria (observed in glucosesulphite-iron agars) and the presence of E. coli in water. In addition, Wilson
and Blair (1931) suggested that since C. welchii or C. perfringens is essentially a
faecal organism and since its spores may persist long after other indicators of
contamination such as coliform bacteria have disappeared, this anaerobe
may well serve as an indicator of intermittent pollution.
Prescott et al. (1945) reviewed the developments of coliforms and “other
intestinal bacteria which have been used as indices of pollution”, the sewage
streptococci and spore-forming, lactose-fermenting anaerobes. This review of
the potential use of both alternative organisms as faecal indicators was
mainly focussed on data comparison with the coliform test. Based on the
reviewed literature they came to the conclusion that streptococci “adds little
to the information furnished by the test of coliform organisms” and that the
presence of C. perfringens indicate a faecal pollution “so remote as to be of
little or no sanitary significance”.
Later, Bonde (1962) reviewed the historical developments on both alternative
indicators in the early days of developments in water microbiology. He
stated as follows: ‘The Royal Commission on sewage disposal in 1898 (UK;
Klein, Houston, Adeney, Frankland, MacConkey) initiated research which
revealed that concentrations of both indicators, faecal streptococci and B.
___________________________________________________________________
-9-
Chapter 1
.
enteritidis sporogenes (C. perfringens), were lower than the concentrations of E.
coli and the results indicated insufficient reliability of both microbiological
parameters.” Moreover, further developments with respect to coliform
monitoring (Eijkman, 1904; MacConkey, 1905) and determination of E. coli,
forced the enteritidis test of Klein (1895) as well as the faecal streptococci as
indicators into the background of microbiological water quality monitoring.
In his studies, Bonde (1962; 1977) found similar results: no relationship
between the presence of C. perfringens spores and coliforms/E. coli. Counts of
C. perfringens in environmental water samples were sometimes up to a factor
of 100 lower than E. coli numbers. He observed a better quantitative
relationship between Streptococcus faecalis and E. coli, but still the E. coli
numbers were usually higher. Despite these findings he reasoned in his
extended thesis that the relationship of C. perfringens with E. coli was not to
be expected. Furthermore, in his opinion C. perfringens meets the
specifications for a proper indicator to a higher degree than does E. coli. The
organism must be considered to be a faecal organism, excreted always
together with a possible faecal pathogenic organism and present in
environmental water samples in numbers often as numerical as the E. coli
numbers. The indicator is more persistent towards environmental stress and
disinfectants than the pathogens and can be identified with rapid and
unambiguous methods of determination. Cabelli (1977) supported this, by
mentioning the ubiquity of C. perfringens in nature, primarily because they
are spore-forming bacteria, as an advantage. In the more recent literature the
use of spores of sulphite-reducing clostridia (SSRC) including C. perfringens
has been used by several authors to assess the environmental pollution with
sewage of remote or intermittent nature and in situations where resistance to
disinfectants and environmental stress is at a premium (i.e. Fujioka and
Shizumura, 1985; Sartory, 1988; Sorensen et al., 1989; Hill et al., 1993; Coffey et
al., 1999; Robles et al., 2000; Buchholtz ten Brink et al., 2000).
Quantitative approach of microbiological parameters. Klein and
Houston (1898, 1899) showed the superiority of microbiological methods to
chemical methods in assessing water safety as described by Bonde (1962).
Pollution of water with sewage could be detected by B. coli at a ratio of 1 :
20.000 and with the enteritis test for Bacterium enteritidis sporogenes (C.
perfringens) at a ratio of 1 : 500.000, despite the lower prevalence of this
indicator in natural waters. With chemical methods it was hardly possible to
detect one part of sewage diluted in 1000 parts of pure water. The plate
method using solid media limited the amount of water for examination to 1
ml. Phelps (1907 ref. in Prescott et al. (1945) suggested that the reciprocal of
the highest dilution in the liquefied coliform test in which a positive result is
___________________________________________________________________
- 10 -
Chapter1
obtained should be used as the expression of the number of coliforms
present in the examined water sample. This approach was further explored
and resulted in a quantitative assessment of indicator bacteria in water
samples by means of the Most Probable Number (MPN).
The use of the membrane filtration to filter larger volumes than 1 ml
followed by incubation on solid growth media, was introduced in 1953
(Bush, 1953). Windle Taylor et al. (1955) attempted the use of membrane
filtration for coliform counting and demonstrated high number of false
positives at 37oC. By increasing the incubation temperature to 44oC and using
resuscitation, he found close agreement between the results of the dilution in
liquid media (MPN) and this membrane filter method. False positives and
overgrowth was mentioned by Bonde (1977) as major drawbacks associated
with the use of membrane filtration. With the development of more selective
culture media and improvement of the general water quality, these
drawbacks were reduced to a large extent. The use of membrane filtration as
a quantitative microbial method in order to examine larger volumes than 1
ml has become more and more common practice and is currently the most
commonly used method in microbial water quality assessment.
Dutch legislation on microbiological quality of water. Nowadays,
standards for coliforms and also for the other faecal indicators in drinking
water are implemented in drinking water legislation worldwide. Due to
methodological optimization and increased knowledge on coliforms in the
environment, standards in the Dutch Drinking Water Decree changed from
coliforms to thermotolerant coliforms in the Drinking Water Decree of 1984
(Anonymous, 1984; Table 1).
Introduction of the other indicator bacteria, faecal streptococci and Bacillus
enteritidis sporogenes (C. perfringens), to examine drinking water was
suggested previously by de Graaf (1922). It was more than 50 years later
until both indicators were implemented in Dutch legislation. These
standards for indicator bacteria in drinking water implemented in the Dutch
Decree of 1984 are described in detail by Havelaar (1983) and Van der Kooij
(2002) and were based on the harmonisation of drinking water regulations in
the European Community (80/778/EEG, Pb. L 229/11). The background of
implementation of streptococci and SSRC was explained by Havelaar (1983)
as follows:
- the higher persistence of faecal streptococci and spores of sulphitereducing clostridia (SSRC) against environmental stress compared to
coliforms;
- the higher specificity of faecal streptococci as faecal indicator;
___________________________________________________________________
- 11 -
Chapter 1
-
.
the high resistance of SSRC to common used chemical disinfectants
comparable to persistent pathogenic protozoa.
Table 1. The Maximum Tolerable Concentrations (MTC; colony forming units or
cfu) of indicator bacteria in the Dutch drinking water as prescribed in
“normbladen; N” before 1960 (Anonymous, 1942, 1956), in the first Dutch
Drinking Water Decree of 1960 and the adapted Decree in 1984
Coliformsa
N1028/N3034
1942/1956
≤1 per 50 ml
Decree 1960
Decree 1984
0 per 200 ml
0 per 300 ml
0 per 50 ml
0 per 50 ml
-
-
-
0 per 300 ml
0 per 20 ml
0 per 10 ml
0 per 100 ml
-
-
0 per 100 ml
Therm. lactose-ferm.
bacteriaa
Thermotolerant
coliformsa
Faecal enterococcib
Spores of sulphitereducing clostridiac
Methods of assessment:
Anonymous, 1985a
a
Anonymous, 1981;
b
Anonymous, 1982;
c
The choice of the sample volume of ≤ 100 ml was based on practical
considerations rather than health related considerations. Havelaar (1983)
argued that implementation of a sample volume of 100 ml in the regulations
was based on the absence of coliforms (total and thermotolerant) in volumes
of 500 – 1000 ml of “good” drinking water. Moreover, concentrations of these
indicator bacteria in contaminated water are high enough to be detected in
samples of 100 ml and higher volumes than 100 ml would introduce
“transport problems”. Also for the newly implemented SSRC standard a
sample volume of 100 ml was considered as appropriate on the basis of a
national survey in 1000 ml samples of drinking water where no spores were
detected.
Additional to the these new drinking water standards in the Dutch Drinking
Water Decree of 1984 was the requirement for source water monitoring in
order to indicate the treatment required to obtain safe drinking water from
different sources of surface water with different state of pollution (Table 2).
Again this regulation was based on a European directive related to the
___________________________________________________________________
- 12 -
Chapter1
required quality of surface water used for production of drinking water
(Anonymous, 1975).
Table 2. Microbial standards for source water quality (n/100 ml) and water
treatment
Class Description and requirements
I
II
III
Low level of contamination: simple
treatment (Rapid granular filtration
and disinfection)
Moderate domestic/industrial
contamination: Coagulation/floc
removal, rapid granular filtration,
oxidation/disinfection
Heavily domestic/industrial
contamination: as II plus Cl2,
Granular Activated Carbon
Faecal
streptococci
100
Therm.
coliforms
Coli44
20
10.000
2.000
1.000
100.000
20.000
10.000
Coliforms
10
Nomenclature of indicator organisms. As described before Klein
and Houston (1899) were the first the use the title “bacteria of indication of
faecal pollution” for three different bacteria implemented in Dutch
Drinking Water Decree. Mossel (1982) introduced the term “marker”
organisms for indicating potentially unsafe food samples and described an
ecological based nomenclature for two groups, the first group named index
organisms to provide information on the risk of occurrence of pathogens.
The second group named indicator organisms used for the purpose of
assessing the risk of inadequate bacteriological quality of a more general
nature. This nomenclature was adopted by Ashbolt et al., 2001 and
extended with the term “process indicator” (Table 3). This classification
adds to the specification of the different groups of micro-organisms that are
significant for microbiological sanitation of water.
One unambiguous and universal group/or species of organism for one
specified group of waterborne pathogen (viruses, bacteria, parasites),
however, is still not available. It is doubtful if this ever will be found.
Differences in persistence, nutritional requirements and environmental
metabolic conditions as well as in their morphological and physical
___________________________________________________________________
- 13 -
Chapter 1
.
characteristics between the potential groups and their target pathogens, but
also within both groups of organisms can be large.
Table 3. The three identified groups of faecal micro-organisms used in water
quality control (from Ashbolt et al., 2001)
Group
Index and
model organism
Faecal indicator
Process indicator
Definition
A group/or species indicative of pathogenic
presence and behaviour respectively, such as E.
coli as an index for Salmonella and F-RNA
coliphages models of human enteric viruses
A group of organisms that indicates the
presence of faecal contamination; pathogens
may be present
A group of organisms that demonstrates the
efficacy of a water treatment process
Furthermore, the rate of environmental distribution between both groups is
different. Pathogens are only excreted by infected individuals and the extent
of excretion depends on seasonal variation and presence of epidemics, but
the non-pathogenic candidates for the three classes are excreted at a higher
and stable level as part of the normal faecal microbiota. Therefore, the choice
for the use of a variety of potential candidates which meets the following
requirements
1. being present in the presence of pathogens in higher numbers;
2. being equally or more persistent in the environment than the
pathogens;
3. showing no multiplication in the environment;
4. and being determined with rapid, simple, reliable and unambiguous
methods,
is, to a large extent still the best choice for microbial water quality
monitoring with respect to safety.
FURTHER
DEVELOPMENTS
MANAGEMENT
IN
MICROBIAL
RISK
Shortcomings of faecal indicators and water treatment. Beside the
identification of new emerging pathogens in waterborne diseases as a major
finding of the epidemiological studies at the end of the 20th century, there
were two other major findings which initiated increased attention to the
___________________________________________________________________
- 14 -
Chapter1
microbiologically safety of drinking water in the developed countries. First
of all it was noticed that waterborne outbreaks of enteric diseases associated
with pathogenic bacteria, viruses and protozoa in drinking water were
reported while the coliform regulation was not violated (Dutka, 1973; Craun
et al., 1997). Craun (1997) concluded that coliform bacteria are adequate
indicators for the potential risk from pathogenic bacteria and viruses, but not
for persistent pathogens like Giardia and Cryptosporidium. However, the
failure of the coliform standard to protect against outbreaks of pathogenic
bacteria like Campylobacter spp., has also been demonstrated twice in
literature by Rosef and Mork (1985) and Rosef et al. (2001). Secondly,
waterborne outbreaks of diarrhoea caused by protozoan parasites have
been related to peak concentrations in the source water and inadequacies in
water treatment (Dykes et al., 1980; Badenoch, 1990; Craun, 1990;
Richardson et al., 1991).
Legislation: microbial risk management. Health related
developments in the drinking water industry in the eighties of the 20th
century were mainly focussed on chemical issues like disinfection byproducts (DBP’s) and other trace chemical contaminants of toxicological
significance. The increased knowledge on waterborne diseases and the
awareness of the severe consequences for vulnerable parts of the population
(immune-deficient persons) initiated a greater interest in microbiologically
aspects of drinking water during the last decade of the 20th century. A large
part of these studies and observations were carried out in the US and
resulted in legislations in that country to enhance public health protection.
The Surface Water Treatment Rule (SWTR) was introduced prescribing
treatment requirements to guarantee a certain level of virus and protozoan
(oo)cyst removal (von Huben, 1991). To balance the risks of DBP’s and
pathogens an Information Collection Rule (ICR) was published in the US to
collect actual data on both health significant contaminants (pathogens and
DBP’s) in drinking water. This resulted in the Long Term 2 Enhanced Surface
Water Treatment Rule which regulates monitoring Cryptosporidium levels in
the source water and the required removal or inactivation capacity for this
specific pathogen in addition to Giardia and viruses of new and existing
treatment facilities (USEPA, 2006).
Simultaneously, a major scientific breakthrough was the introduction of the
estimation of the probability of infection given a certain dose of microorganisms (Haas, 1983; Haas et al., 1999; Medema et al., 1996; Teunis et al.,
1996). Comparable to the regulation for mutagenic contaminants
prescribing Maximum Tolerable Concentrations (MTC) for the risk of cancer
___________________________________________________________________
- 15 -
Chapter 1
.
during a life time exposure, the US Environmental Protection Agency (EPA)
used these dose-response relationships to calculate the MTC values for
viruses, Giardia and Entamoeba histolytica in drinking water for an annually
accepted infection risk of 1 infection per 10.000 consumers, drinking 2 litres
per day (Regli et al., 1991). For the viruses this MTC-value was based on the
most infectious virus type known at that time, the rotavirus. This
development can be regarded as a major breakthrough in microbial risk
management in drinking water. In the US where this was first described,
however, these MTC-values are not implemented in the current legislation.
US authorities do not communicate an accepted risk level with Public.
Developments in the Netherlands: Quantitative Microbial Risk
assessment. In the Drinking Water Decree (Anonymous, 1984, 2001) as well
as in the revised EU-regulations (Anonymous, 1980) it is prescribed that
drinking water may not lead to an unacceptable microbial health risk. This
formulation was chosen with the awareness that absolute absence of
pathogenic micro-organisms in drinking water can not be guaranteed. Based
on the available knowledge is was assumed that the standards for indicator
bacteria, and coliforms specifically, were enough to assure that the drinking
water did not lead to unacceptable health risks.
In the Netherlands the number of registered waterborne outbreaks is very
low and only related to cross contamination of the water mains due to
human failures as presented before. Dutch drinking water was regarded as
intrinsically safer than drinking water in the US and other Western countries
with reported waterborne outbreaks because of the following reasons:
- The ground water used for Dutch drinking water is abstracted from fine
sandy soil aquifers protected by clay layers;
- The surface water sources used for Dutch drinking waters are relative
highly polluted and therefore stabilized (reservoir storage or dune
passage) prior to a treatment plant with multiple barriers;
- Microbial standards for drinking water in the Netherlands with multiple
indicators implemented in legislations since 1984 (thermotolerant
coliforms, enterococci and SSRC; Table 1) are more stringent than in
other European countries (lower MTC-values) and even more stringent
compared to non European countries using solely coliforms or E. coli.
Though these assumptions are regarded as plausible, the Dutch Drinking
Water Companies in their aspiration to distribute a high quality drinking
water started a research funded by the joint research program to verify these
assumptions.
___________________________________________________________________
- 16 -
Chapter1
Assessment of the microbiological water quality with the current microbial
standards (Table 1) is a curative control strategy with a delay time of at least
one day; consumers can be exposed to pathogens during this delay time
while the consumed water was potentially contaminated with pathogens.
Moreover, the increasing knowledge on waterborne outbreaks and the
inadequacy of the coliform or E. coli standard demonstrate the need for more
preventive strategies. Hazard Analysis and Critical Control Points (HACCP)
is such a tool, introduced in the food industry and also proposed (Havelaar,
1994; Anonymous, 2003) and applied in the drinking water industry (Hijnen
et al., 2001). Later this approach is implemented in the more generic approach
of Water Safety Plans (WSP; WHO, 2004). A related approach is the use of
Quantitative Microbial Risk Assessment (QMRA). The introduction of the
MTC values for pathogenic micro-organisms in drinking water based on an
annual risk of infection first presented by Regli et al. (1991), is a basic
requirement for quantitative risk management. Implementation of QMRA in
the revised Drinking Water Decree (Table 4) was initiated by the Dutch
Authorities (Medema and Havelaar, 1994a) in close cooperation with the
Dutch Drinking Water Companies (Van der Kooij et al., 1995; Medema, 1999).
Besides the preventive character of this new approach, it offers a framework
for new and emerging pathogens such as Cryptosporidium and noroviruses.
Moreover, it enables balancing the microbial risks with toxicological risks of
disinfection processes (by-products; Havelaar et al., 2000) and with risks
associated with other routes of exposure.
Quantitative Risk Assessment for pathogenic micro-organisms in drinking
water was described by Haas et al. (1999), Teunis et al. (1997), Haas and
Eisenberg (2001) and Medema et al. (2003) and is part of a total system
approach as presented in Figure 2. The major steps in the process of QMRA
are (i) quantitative information of the relevant pathogenic micro-organisms
in the source water and (ii) quantitative knowledge about the elimination
(removal and inactivation) capacity of water treatment processes for
pathogenic micro-organisms and the factors influencing elimination and (iii)
calculation of the exposure on the basis of these data, drinking water
consumption and the dose-response curves.
The scope of the present study is on the second step of QMRA, assessment of
the elimination capacity of a process or a treatment as a chain of processes.
The three basic processes in water treatment are coagulation/floc removal,
filtration (underground or in filters) and disinfection.
___________________________________________________________________
- 17 -
Chapter 1
.
Table 4. The Maximum Tolerable Concentrations (MTC) of indicator bacteria and
pathogenic micro-organisms in the drinking water after treatment as prescribed in
the Dutch Drinking Water Decree using surface water as the source
Parameter
Cryptosporidium
Escherichia coli
Maximum
value
0
Unit
Remarks
CFU/100ml
Note 1
CFU = colony Forming
Units
Enterococci
0
CFU/100ml
a
(entero)viruses
Note 1
Giardia
Note 1
a The purpose of the brackets in (entero)viruses was to indicate that possible other
virus groups that may be of concern for drinking water provision are included in
the Dutch Drinking Water Decree
Note 1:
According to article 4, paragraph 1, micro-organisms should not be present in tap
water in such concentrations that public health is jeopardized. Specific microorganisms such as viruses and protozoa (e.g. Cryptosporidium and Giardia) can
not be detected at the very low concentrations at which exposure is relevant to the
consumer’s health. Instead, the owner who employs surface water as source for
drinking water production should in consultation with the Inspectorate carry out a
quantitative risk assessment based on data regarding the source water quality and
treatment efficiency. The theoretical infectious risk resulting from the risk
assessment should comply with a provisional standard of one infection per 10 000
individuals per year. Verification of the (provisional) infectious risk standard
should be carried out for enteroviruses, Cryptosporidium and Giardia but also
concerns other pathogenic micro-organisms. If the assessed infection risk is greater
than aforementioned standard, the owner should consult with the VROMInspector about necessary measures. The Inspectorate can decide if a risk
assessment should be carried out for vulnerable groundwater supplies. The
expression << provisional standard >> is used to indicate the value that is verified
in practice. Adjustment of this value can therefore not be precluded
___________________________________________________________________
- 18 -
Chapter1
Know your
catchment
QMRA
Assessment of
elimination capacity
Know your source
water quality
Target your
Treatment
Protect your
distribution
Coagulation
(In)filtration
Disinfection
Microbiologically
safe drinking water
Figure 2. The total system approach for the (microbial) safety of drinking water
Dutch drinking water production systems with a QMRA
obligation. QMRA is mandatory for Drinking Water Companies in the
Netherlands with treatment facilities using surface water as the source and
for Drinking Water Companies using ground water subtracted from areas
vulnerable to contamination as appointed by the inspectorate. The major
surface water sources used for drinking water production at eight major
Dutch production facilities are the River Rhine and Meuse (Table 5).
Table 5. The source water and treatment before 2000 of eight major Dutch
drinking water production facilities with a QMRA requirement
Treatment (all covered systems)
Pre-treatment
a
Open systems
1. River Meuse
IR
CFR - O3 – RGF – GAC – PD
2. River Meuse
IR
CL2 – CFR – RGF- PD
3. River Meuse
IR
CFR - O3 – RGF – GAC – PD
4. Lake IJssel
SR
CL2 – CFR - RGF – GAC – PD
5. Drentsche Aa
SR
CFR – RGF – GAC – RGF – SSF
6. River Rhine
IR
RGF – SR - O3 - SF – GAC – SSF
7. Polderb
DP – SR
RGF – SF – O3 – GAC – SSF
8. River Meuse
DP – SR
RGF - SF – SSF
a Re-contaminated by wild life; b small fraction river Rhine water
Number and
Source water
___________________________________________________________________
- 19 -
Chapter 1
.
The treatments of these facilities in operation before 2000 are presented in
Table 5. The surface waters are pre-treated with either open pre-treatment
systems like impoundment reservoirs (IR) or small reservoirs (SR) or by
dune passage (DP) with open reclamation. These pre-treatment steps are
followed by covered unit processes in different process schemes:
coagulation/floc removal (CFR), rapid granular filtration (RGF), ozonation
(O3), chlorination (Cl2), granular activated carbon filtration (GAC), slow
sand filtration (SSF), powdered activated carbon (PAC), lime stone
softening (SF) and post-disinfection (PD) with chlorine or chlorine dioxide.
Source water monitoring: required elimination capacity.
Monitoring of the microbiological quality of the source water was made
mandatory in the former Dutch Drinking Water Decree in 1984 to prescribe
the level of water treatment (Table 3). In the further developments
implemented in the revised Water Decree of 2001 source water monitoring
was intensified resulting in monitoring for index pathogens and for E. coli
and C. perfringens as process indicators (Table 6).
In the period 2001 – 2006, Water Companies have been occupied to comply
with these new regulations. Their activities were based on an extended
interpretation of this new regulation developed by a collaborative
committee with representatives of Water Companies, Kiwa Water
Research, Water Authorities and Inspectorate and the National Institute of
Public Health and the Environment RIVM (Wetsteyn, 2005).
Table 6. Monitoring requirements (annual number of samples) of the source water
and the treated water in the revised Drinking Water Decree (2001) at a production
of 180.000 m3.d-1
Micro-organisms
E.coli
Coliforms
Enterococcci
C. perfringens
Cryptosporidium
Giardia
Enteroviruses
Bacteriophages
a - = no requirement
Groundwater
(GW)
13
13
-a
-
Surface water
(SW)
13
13
13
18
18
18
18
Treated water
GW:52 SW:365
GW:18 SW: 544
-
___________________________________________________________________
- 20 -
Chapter1
During the period of these new developments surveys have been
performed to assess the prevalence of pathogenic micro-organisms in
Dutch surface waters used as the source for drinking water production. In
an extensive study the distribution of the protozoan parasites,
Cryptosporidium and Giardia was monitored by Hoogenboezem et al. (2001)
and additional samples were analyzed for enteric viruses. This revealed
that (oo)cysts were ubiquitous in the river Rhine and Meuse, two of the
most important rivers for the Dutch drinking water production (Table 7).
Table 7. The range of required elimination capacities (log) to meet the infection
risk for the index pathogens of 10-4
Index pathogens
Dutch river waters
After reservoir storage
(24 weeks)
Cryptosporidium
4.8 – 5.9b
5.8 – 7.0a
a
Giardia
4.3 – 5.9b
6.0 – 7.8
c
Enteroviruses
6–7
2.8 – 3.6d
Campylobacter
6.5 – 7.3d,e
4.5 – 8.0d,e
a Hoogenboezem et al. (2001); b Ketelaars et al. (1995); van Breemen and
Waals (1998); c Theunissen et al. (1998); d de Roda Husman et al. (2001); e
Wubbels (1996)
From the observed concentrations the required elimination of a treatment
for Cryptosporidium ranged from 5.8 up to 7.0 log and for Giardia from 6.0
up to 7.8 log to comply with the annual infection risk of 10-4 per person.
These estimations were calculated from the average surface water
concentrations corrected for recovery using dose-response data (Regli et al.,
1991; Gerba and Rose, 1993; Rose et al., 1993), the set point of the maximum
tolerated annual infection risk of 10-4 per person and the probability of
infection assessed with the exponential model (Haas, 1983; 1993).
Theunissen et al. (1998) summarized the enterovirus concentrations at the
intake points of surface waters of the Dutch drinking water companies and
estimated a required elimination of 6 – 7 log (MTC-value for rotavirus of
1.83x10-7/l for infection risk level of 10-4). The additional enterovirus
concentrations collected by Hoogenboezem et al. (2001) and de Roda
Husman et al. (2001) indicated similar required elimination capacities.
Because the Dutch surface waters are not pristine but heavily to moderate
polluted there is a pre-treatment prior to the final and covered production
stage of the drinking water. These pre-treatment stages consisted of some
___________________________________________________________________
- 21 -
Chapter 1
.
kind of open storage reservoirs with or without preceded artificially
recharge in the sand dunes. In these pre-treatment systems the level of
pathogens in the surface water is reduced to some extent but for some
micro-organisms introduction by waterfowl (i.e. Campylobacter) or other
wild life may occur. Medema (1999) showed that waterfowl contributes to
the introduction of pathogenic protozoan (oo)cysts, and Fallacara et al.
(2001) demonstrated that in approximately 50% of the faecal shedding of
free-living waterfowl sampled at six different parks with water reservoirs
in Ohio C. jejuni was isolated. Campylobacter concentrations in surface
waters showed a large variation and consequently the required elimination
varied considerably. Introduction of this pathogen was observed in open
reservoirs and collection ponds for recollection of infiltrated river water in
concentrations of 0.2 – 40 per 100 ml (Medema and Schets 1994b). In the
same study, 25% of gull and duck faeces sampled in the neighbourhood of
these reservoirs and ponds were positive for Campylobacter. Additionally,
84% of the water samples after the large storage reservoirs of the Biesbosch
contained Campylobacter (de Roda Husman et al., 2001) and based on the
measured concentrations the required DEC value ranged from 6.5 to 7.3 log
(Table 7).
For Giardia cysts and Cryptosporidium (oo)cysts, a reduction in numbers of
2.3 and 1.4 - 1.9 log, respectively was observed in the storage reservoirs
Biesbosch with an average residence time of 24 weeks (Ketelaars et al., 1995;
van Breemen and Waals 1998) . The reduction of enteroviruses in the
Biesbosch reservoirs was in the same order of magnitude (2.1 log; de Roda
Husman et al., 2001). Thus, the required DEC values for protozoan (oo)cysts
and viruses for the treatment facilities after reservoir storage are
approximately 2 log lower than based on the concentrations observed in
the untreated river water (Table 7). The required elimination for both
protozoan parasites is in the same order of magnitude as calculated by
LeChevallier and Norton (1992) for US surface waters.
Need for information on removal of pathogens. In development and
implementation of new water treatment technologies throughout the 20th
century, improvement of water quality was a major driving force with the
protection of public health as the main rationale. The performance of these
processes was commonly assessed by monitoring the removal of
physical/chemical parameters such as suspended solids, turbidity and
dissolved organic carbon and by assessing the microbiological quality of
the water with colony counts and faecal indicator bacteria. For the
microbial parameters the performance of the processes was usually judged
___________________________________________________________________
- 22 -
Chapter1
from the quality of the produced water rather than the capacity of the
process to reduce micro-organisms.
As prescribed in the revised Dutch Drinking Water Decree (2001)
quantifying the elimination capacity of water treatment processes for
pathogenic micro-organisms present in the source water has become a
major requirement to assess microbiological safety of the drinking water.
Under the influence of the growing interest in microbiological safety of
drinking water in the last decade of the 20th century studies were initiated
to quantify the removal of the pathogenic micro-organisms by full-scale
water treatment plants (Payment et al., 1985; 1993; LeChevallier et al., 1991;
Stetler et al., 1992; States et al., 1997).
Direct monitoring of enteroviruses, Cryptosporidium and Giardia elimination
in Dutch water treatment facilities is restricted to water pre-treatment
water systems using river water directly to produce semi finished water
(Havelaar et al., 1995; Ketelaars et al., 1995; Hoogenboezem et al., 2001;
Oesterholt et al., 2007). In the nineties of the last century it was recognized
that some Water Companies with open reservoirs directly before treatment
were confronted with steady and sometimes high concentrations of
Campylobacter bacteria in the source water (Medema and Schets, 1994). This
urged these Water Companies to assess the elimination of these bacteria by
their treatment (Visser et al., 2004; Dullemont, 2004, 2006). These
investigations yielded quantitative information on the efficacy of rapid
granular filtration, ozonation and slow sand filtration to eliminate these
pathogenic bacteria in comparison with the elimination of Coli44 (Hijnen et
al., 1995, 1998; Smeets et al., 2005). In recent years, open reservoirs with
water recontamination by waterfowl immediately before the drinking
water production facility have been eliminated as much as possible.
Nonetheless, these observations clearly demonstrate the possibility of
collecting elimination data by direct pathogen monitoring in treatment.
This strategy of pathogen monitoring in the full-scale treatment is not
applicable for most Dutch locations certainly not for the total treatment.
The concentrations during water treatment (multiple barriers) and required
for an infection risk of 10-4 are too low and furthermore the pathogen
enumeration methods are too complex (Wetsteyn et al., 2005). Hence, it was
necessary to develop and validate alternative methods or strategies.
___________________________________________________________________
- 23 -
Chapter 1
.
PROCESS INDICATORS FOR PATHOGENS REMOVAL AND A
GENERAL HYPOTHESIS
Generally speaking, the pathogenic micro-organisms can be divided in
resistant and less resistant organisms. The parasitic protozoa
Cryptosporidium, Giardia and Entamoeba histolytica are resistant against the
commonly used concentrations of chemical disinfectants (chlorine, ozone).
Viruses and bacteria are more susceptible to disinfectants. Thus, the major
barriers in treatment for the resistant pathogens are physical barriers like
coagulation and filtration. For viruses and bacteria chemical disinfection is
the barrier of primary importance.
Physical process indicators. The large Cryptosporidium outbreak in
Milwaukee prompted research into the validity of commonly used
parameters in water treatment such as turbidity and more advanced
techniques like particle counting to determine removal of this pathogen.
Several authors demonstrated that setting a treatment goal for turbidity of
<0.1 NTU will help to reduce breakthrough of parasitic protozoa in water
treatment. To a certain degree, a quantitative correlation between removal
of turbidity and particles with the removal of protozoan (oo)cysts both
monitored in situ, has been observed (LeChevallier et al., 1992; Nieminsky
et al., 1995). However, a large national survey on the use of particle
counting showed no correlation between the elimination capacity of
treatment for this parameter and the elimination capacity for protozoan
(oo)cysts (McTigue et al., 1998; LeChevallier and Au, 2004). Furthermore,
the same authors and also others who performed dosing experiments
(Patania et al., 1995; Coffey et al., 1999; Swertfeger et al., 1999; Dugan et al.,
2001; Emelko, 2001) demonstrated that generally the elimination capacity
for both physical parameters underestimates the elimination capacity of the
treatment for both parasites. The poor quantitative relationship between
the removal of turbidity and particles and the removal of micro-organisms
is most likely due to the fact that there is no quantitative relationship
between the concentrations of both parameters (physical and
microbiological) in the source water as shown by McTigue et al. (1996;
1998). With both physical methods unspecific colloids in the water are
monitored which are present in much higher concentrations than the
pathogens. Most of these studies, however, demonstrated that both
parameters are valuable to control removal of colloidal particles by the
processes which certainly help in minimizing the risk of breakthrough of
these persistent pathogens in water treatment.
___________________________________________________________________
- 24 -
Chapter1
Microbial process indicators. Information on multiple indicator
bacteria in treated water, as applied in the Netherlands since 1984, offers
the opportunity to quantify the removal capacity of treatment processes for
susceptible and more resistant micro-organisms.
General hypothesis
The general hypothesis of the work described in this thesis is that Escherichia
coli and spores of sulphite-reducing clostridia can be used as process indicators
to assess the elimination capacity of water treatment processes for, respectively,
pathogenic micro-organisms that are susceptible and pathogenic microorganisms that are resistant to disinfection processes under the commonly
applied conditions in water treatment.
Monitoring of these indicators both in the source waters and the finished
waters are prescribed by legislation (the Dutch Drinking Water Decree
1984, Tables 1 and 2; revised Decree 2001, Tables 3 and 5). Beside this
required data collection, Water Companies monitor indicator bacteria
throughout their treatment facilities regularly to detect contaminations and
to check process performances. Thus, the use of faecal indicators as
microbiological process indicators is closely linked to the daily practice of
water quality monitoring in the drinking industry and a multi year data base
of full-scale systems is available. Furthermore, these indicators have been
selected on the basis of requirements related to the presence of pathogenic
micro-organisms. They are commonly present in the source waters and
related to nominal faecal contamination and indicative for some peak events
(Atherholt et al., 1998; Signor et al., 2005).
For a process indicator the requirement of being presence in the presence of
pathogens is not stricktly required. In addition to the three other previously
mentioned requirements for proper indicator micro-organisms of faecal
contamination, further requirements for an ideal process indicator are as
follows:
4. being continuously present in the source water at measurable
concentrations to enable site specific assessment of elimination;
5. being determined with a method that is sensitive enough to
determine concentrations at all stages of water treatment;
6. being eliminated at an equal or slightly lesser extent than the
pathogens by individual water treatment processes or a chain
thereof.
___________________________________________________________________
- 25 -
Chapter 1
.
Due to major interest in the use of coliforms and E. coli as faecal indicators
during the last century, scientific knowledge on the validity of the first four
requirements for this group of indicators is extensive. This revealed a
serious limitation for the total group of coliforms which multiply at
moderate temperatures in treatment steps. The failure to detect the
potential presence of parasitic protozoa and pathogenic viruses in drinking
water by coliform monitoring (E. coli included) as described before,
however, emphasize that also E. coli does not meet the requirements for a
proper microbial process indicator for all pathogens. The difference in
nature of indicator bacteria, parasitic protozoa, and viruses is evident and
the major reason for their different behaviour during treatment. It is
hypothesized that for the pathogenic bacteria such as E. coli O157 and
Campylobacter, E. coli is a proper process indicator. Scientific knowledge on
behaviour of faecal streptococci in water treatment is limited. Moreover,
the environmental persistence of these bacteria (die off and susceptibility to
chemical disinfection) is in the same order of magnitude as the persistence
of coliforms (i.e. Grabow et al., 1983). Therefore, faecal streptococci is not of
additional value to E. coli as a process indicator.
The low incidence of waterborne outbreaks of parasitic protozoan in the
Netherlands has been attributed to the benefit of multiple barriers in the
treatment. Additionally, also the required compliance with a standard for a
persistent indicator bacterium like spores of sulphite-reducing clostridia
SSRC in drinking water (Table 1 and 2) has been hypothesized as a possible
explanation. Both micro-organisms, Cryptosporidium and Clostridium,
exhibit a life-cycle with a dormant life-stage meant for environmental
transmission and survival.
In the NPR 6568 (Anonymous, 1985b; note of the Dutch normalized method
NEN6567) SSRC is described as a “biological process indicator at the
drinking water production, especially for the performance of treatment
steps like sedimentation and filtration”. The use of the anaerobic spores as
technical parameter to monitor deficiencies in filtration processes was
proposed by Taylor (1949) and Willis (1957). The use of Clostridium
perfringens as process indicator for the removal of viruses and protozoan
oo(cysts) of Cryptosporidium and Giardia in drinking water treatment was
studied later by Payment and Franco (1993). For the removal of enteric
viruses, however, the use of bacteriophages as process indicators is more
appropriate due the similar nature of those organisms (Havelaar et al., 1993;
1995; Jofre et al., 1995). The first study to propagate the use spores of
sulphite-reducing clostridia (SSRC) as a surrogate for oocyst removal as
suggested by the NPR6568 was done on pilot plant scale by Hijnen et al.
___________________________________________________________________
- 26 -
Chapter1
(1997). The use of SSRC as potential process indicator was chosen, because
of the daily routine of monitoring, the availability of historical data in
source water and the subsequent stages of drinking water production of
this faecal indicator bacterium and the improbability of growth in the
aerobic water treatment processes.
In countries like US and UK with no monitoring routine of these faecal
indicators in water treatment, the aerobic spore forming bacteria (Bacillus)
was introduced as a potential process indicator for the assessment of
protozoan (oo)cyst removal in water treatment (Rice, 1996) and applied in
research (Patania et al., 1995; Nieminski et al., 2000; Emelko, 2001; Dugan et
al., 2001). Indications for regrowth of aerobic spores in filters, however,
have been documented in literature (Galofre et al., 2004; Mazoua and
Chauveheid, 2005).
SPORES OF SULPHITE-REDUCING CLOSTRIDIA AS PROCESS
INDICATOR FOR CRYPROSPORIDIUM AND GIARDIA
The genus Clostridum. Bacteria of the genus Clostridium can be
described as gram-positive, anaerobic spore-forming, non-motile, and rodshaped. C. perfringens is the species most commonly associated with faecal
contamination (Cabelli, 1977) which ferments lactose, sucrose and inositol
with the production of gas. The bacterium causes “stormy fermentation of
milk”, reduces sulphite to sulphide, reduces nitrate, hydrolyzes gelatine,
and produces lecithinase and phosphatase.
The genus Cryptosporidium and Giardia. The genus
Cryptosporidium is a small coccidian protozoan parasite that infects a host
through the oocyst stage of its life-cycle. This is the environmentally
resistant transmission stage of the parasite which may remain in the
environment for long periods without loosing infectivity. After ingestion
by a suitable host the wall of the oocysts is opened (excystation) and the
four motile sporozoites are released which transform in either an asexual or
a sexual reproduction cycle. The oocysts are the product of the sexual
reproduction cycle in which spherical thin- and thick-walled oocysts are
produced. The thick-walled oocysts are the oocysts which are excreted with
the faeces and thus environmentally transmitted.
Giardia is a genus of protozoan parasites potentially found in water and
other media. The recent taxonomy of the genus Giardia includes the
following species and their potential hosts: G. lamblia (also called G.
intestinalis or G. duodenalis; humans and other mammals); G. muris
(rodents); G. agilis (amphibians); G. psittaci and G. ardeae (birds). As with
___________________________________________________________________
- 27 -
Chapter 1
.
Cryptosporidium, the parasite is shed with the faeces as an environmentally
robust cyst that is transmitted to a new host. In the duodenum of a new
host, the trophozoïte emerges from the cysts and completes a mitotic
division to produce two trophozoïtes that attach to the epithelial cells by
their adhesive disc and feed on the epithelial cells. During the passage of
the trophozoïtes through the intestine, part of these trophozoïtes begin to
encyst and leave the host as cysts.
Characterization of spores and (oo)cysts. The bacterial spore
(Figure 3) consists of an inner part, the core, which contains the cellular
substances necessary for the outgrowth of vegetative cells.
Figure
3.
A
schematic
image
of
a
bacterial
spore
left
(www.llnl.gov/str/March02/gifs/Raber1.jpg) and right from Driks (2007; with
permission of Copyright holder) cartoons illustrating the diversity of spore
appendages. The coat is drawn as a light gray layer. The exosporium (in the top
two spores) is indicated by a thin black line surrounding the coat. Appendages
extend from the coat or exosporium. A thick grey layer indicates the spore cortex
(peptidoglycan). Top: pili-like structures found on spores of some B. cereus strains
(ref. in Driks, 2007). Middle: pin-like structures on Clostridium bifermentans
spores (Yolton et al., 1968). Bottom: ribbon-like structures on C. taeniosporum
spores.
This core is surrounded by the inner membrane, a lipid bilayer with no
detectable fluidity, encaged by the cortex and primordial cell wall, a thick
layer of peptidoglycan. This cortex is enclosed by the spore coats, a
___________________________________________________________________
- 28 -
Chapter1
proteinaceous layer (Foster and Johnstone, 1990). Finally, several Bacillus
and Clostridium species are surrounded by the exosporium, a highly
structured, paracrystalline basal layer with an external hair—like nap,
consisting of protein, glycoprotein and most likely, lipid (ref. in Walker et
al., 2007). The composition of the exosporium of B. cereus was determined
(Matz et al., 1970) and proved to be chemically complex, consisting mainly
of protein (52%), amino and neutral polysaccharides (20%), lipids (18%),
and ash (4%). Little information was found on the composition of the
exosporium of Clostridium spores but it may be assumed that they consist
of similar constituents in different ratios. The large morphological
divergences of Clostridium spore was presented by Yolton et al. (1968). They
showed the presence of appendages on spore surface and showed five
distinct morphological spore types in 12 strains of C. bifermentans. External
appendages in the form of ribbons, pili, feathers, brushes, tubules or
swords (ref. in Walker et al., 2007; see Figure 3) on the spore surfaces are
observed frequently. Yolton et al. 1972 demonstrated that these appendages
of C. taeniosporum consist for 80% of protein, 15% of glucose, rhamnose,
glucosamine and 5% phosporus.
The spore coat is approximately 50% of the spore volume and 40-60% of the
dry weight (Murell, 1969) and consists of hexosaminepeptides with large
number of amino acids and high cystine concentrations. The density of
Bacillus spp. (B. cereus, B. subtilis, B stearothermophilus) and Clostridium
perfringens spores was assessed with several techniques by Tisa et al. (1982)
and ranged from 1.120 to 1.380 kg/m3 (average of 1.253±0.074). The circular
size of the spores of C. perfringens in a special prepared suspension
(removal of vegetative cells by washing with lysozyme, trypsine and
sodium dodecyl sulphate) was determined at approximately 1 μm (0.79 –
0.98 μm) with transmission electron microscopy (TEM) (Novak et al., 2002).
Individual ovoid spores of three strains of B. subtilis were sized by TEM;
length ranged from 1.39±0.14 - 1.42±0.18 μm and the width from 0.68±0. 05
– 0.8±0. 11 μm (Leuschner et al., 1999).
Comparable to the bacterial spores, the characteristic wall of the oocysts is
responsible for their environmental persistence. This wall consists of an
outer wall of acidic glycoprotein, a central rigid layer composed of complex
glycolipid/lipoprotein layer followed by a thick layer of filamentous
glycoprotein. Both the central and thick wall provides rigidity and elasticity
of the oocyst (Harris and Petry, 1999). Generally, similar components have
been described for the spore wall. No quantitative data of these
components were found to enable a relative comparison with the spores.
___________________________________________________________________
- 29 -
Chapter 1
.
Additionally, no information of the presence of appendages on the oocyst
wall comparable to the appendages described for bacterial spores have
been found in literature. Observations of “electrosteric” repulsion in a
liquid/solid interface study by Kuznar and Emilech (2005) may point to the
existence of protein containing appendages. These oocysts are spherical
(Figure 4) and have a mean size of 4.9 (3.9-5.9) μm and a geometric mean
density of 1,045.4 kg/m3 (Medema et al., 1998) and thus larger and less
dense than bacterial spores.
Adam (2001) describes the wall of the Giardia cysts as a 0.3 to 0.5 μm thick
and composed of an outer filamentous layer and an inner membranous
layer with two membranes. The outer portion of the cyst wall is covered by
a web of 7- to 20-nm filaments and the outer cyst wall consists of four major
proteins. The predominantly sugar component is galactosamine in the form
of Nacetylgalactosamine (refs in Adam, 2001). Earlier claims that the cyst
wall is composed of chitin (N-acetylglucosamine) have been refuted. The
cysts of Giardia are elliptical in shape with the mean largest (length) and
mean smallest diameter (width) of 12.2 and 9.3 μm, respectively. The mean
ratio between length and width is 1.3 and the geometric mean density is
1,036.2 kg/m3 (Medema et al., 1998).
Surface properties of spores and (oo)cysts. Electrophoretic mobility
studies (references in Murell, 1969) indicated that the effective surface layer
of spores is non-ionogenic. Any charge of the surface is due to adsorption
of ions. There are other studies who demonstrated that the spores of
Bacillus are negatively charged (Douglas, 1955; Watanabe and Takesue,
1976). The latter authors defined the two negatively charged groups on the
spore surface of B. megaterium to be strong acidic groups such as
phosphates and the weaker acidic groups of the carboxylates.
Another important surface property which is of influence of attachment of
micro-organisms to surfaces is hydrophobicity. Spores of Bacillus and
Clostridium were found to be more hydrophobic than the vegetative cells of
these species (Wiencek et al., 1990). This increased hydrophobicity was
attributed by the same authors to the relative abundance of protein in the
outer coat and exosporium compared to the peptidoglycan on grampositive vegetative cell-surfaces. The hydrophobic nature of SSRC was also
deduced from the sticky behaviour of these spores observed in plastic Petri
dishes used for enumeration (Hijnen, unpublished results).
___________________________________________________________________
- 30 -
Chapter1
Giardia lamblia
Cryptosporidium parvum
Clostridium bifermentans
Figure 4. A microscopic image of spores of Clostridium bifermentans labelled with
Acridine-orange and a cyst of Giardia lamblia and oocysts of Cryptosporidium
parvum labelled with fluorescein isothiocyanate (FITC)
Electrophoretic mobility of oocysts studied by Brush et al. (1998) showed
differences in surface charge of oocysts, depending on the pre-treatment of
the used oocysts. Oocysts pretreated (purified) with deionized water
exhibit no surface charge whereas oocysts pre-treated with surfactants
(defatting agents) were negatively charged. Drozd and Schwartzbrod
(1996) found that oocysts were low in hydrophobicity and additionally,
Hsu and Huang (2002) found an increasing hydrophobicity at decreasing
pH. Brush and colleagues demonstrated that the hydrophobicity of oocysts
can change as they age.
In conclusion, there are similarities and differences between the oocysts
and clostridia spores which influence their similarity in resistance and
transport behaviour during water treatment processes. Furthermore,
comparative studies are required to elucidate the validity of the use of
SSRC as a process indicator.
Environmental fate: resistance, activation, germination and
outgrowth. The outer structure in combination with the dehydrated core
components in a lattice of divalent metal ions and dipicolinic acid is the
basis of the metabolic dormancy and heat resistance of the spores (Gould
and Hurst, 1969). UV resistance is established by complex formation of
spore DNA with spore-specific, acid-soluble, low molecular-weight
proteins (Setlow, 1988). The general view on bacterial spore survival in
nature is that they are extremely persistent (Gould, 2006). Longevities of
thousands of years have been demonstrated in literature. A recent
comparative survival study showed that spores of C. perfringens were more
___________________________________________________________________
- 31 -
Chapter 1
.
persistent than oocyts of C. parvum assessed by in vitro excystation and
exclusion of propidium iodide (Medema et al., 1997).
Besides this protective structure to maintain environmental dormancy,
spores have also a sensory mechanism for favourable environmental
conditions which can initiate germination, the pre-stage of vegetative
outgrowth, within seconds. Both the sporulation and germination stage of
the genera Bacillus and Clostridium is induced on exposure to stressful
environmental conditions, and these differ between genus and species.
Bacillus spp. have their active life-cycle under aerobic conditions, whereas
Clostridium spp. germinate and grow under anaerobic conditions. The
redox conditions of the environment seem to be not essential for the
germination (references in Lewis, 1969). Clostridia spores do geminate
under aerobic conditions and Bacillus spores can germinate under
anaerobic conditions. Germination of the spores is preceded by activation,
defined by Keynan and Evenchik (1969) as breaking dormancy and
overcome the inability to germinate. Heat treatment of the spores is the
most described activator, but also low pH, reducing agents, calcium,
dipicolinic acid, ionizing radiation, chemicals and ageing. The latter
activator is presented as a similar activator as heat treatment by Keynan
and Evenchik.
The germination of spores is defined as a series of events triggered by
specific germinants leading to irreversible loss of spore resistance
properties (Johnstone et al., 1994). Possible germinants are amino acids,
sugars, alcohols, dodecylamine, enzymes, hydrostatic pressure, mechanical
actions and heat. Bacillus and Clostridium spores differ in their need for
germinants as described by Gould (1969) and the latter spores need
generally more complex germinant mixtures than Bacillus spores.
After germination a spore can be changed into a mature vegetative cell
with all the characteristics and needs for multiplication. When the
environmental conditions are not appropriate i.e. the required nutrients are
not available, re-sporulation may occur without the evolution to a mature
vegetative cell (Strange and Hunter, 1969).
Being a parasitic micro-organism Cryptosporidium needs a host for
multiplication. This is not necessary for bacterial spores and the potential
propagation of these spores in water treatment is a critical aspect and
observed for the aerobic spores as previously described. Before proposing
anaerobic spores as technical parameter, Willis (1956) was negative about
the use of these anaerobic spores because of their prevalence in sand from
drinking water filters. Later Willis (1957) proposed that “filters may be a
reservoir of anaerobes which cause intermittent appearance of sulphite___________________________________________________________________
- 32 -
Chapter1
reducers in the treated water throughout the year”. He found anaerobic
bacilli, defined as bacteria forming black colonies in anaerobic incubated
Wilson-Blair medium, frequently in the sand of filters with no relationship
with the depth. Vegetative cells of this bacterium were predominantly
present, indicating the occurrence of growth in the filter bed. He emphasized
the value of the sulphite-reduction test for the indication of deficiencies in
the filtration process as previously reported (Taylor 1949), but again doubted
the use of the parameter for the classification of the safety of the water. This
observation of vegetative anaerobic bacilli in the filter beds is controversial,
however, with respect to the assumption of growth. The way to determine
vegetative cells of sulphite-reducing clostridia in the samples was to omit the
pasteurization prior to the cultivation. Because of the following arguments
the assumption that this procedure gives valid evidence for the occurrence of
outgrowth of these anaerobic bacterial spores is disputable:
- the assumption that all black colonies in the non pasteurized samples can
be classified as vegetative bacilli (or more precise clostridia because of the
sulphite-reduction) is not correct without excluding of the presence of
other non-spore forming bacteria that can reduce sulphite in the used
medium and produce black colonies such as Salmonellae, Proteus,
Escherichia freundii, Citrobacter spp. (formerly Paracolobactrum) and certain
species of Erwinia, Flavobacterium and Achromobacter (Angelotti et al.
1962);
- secondly, the susceptibility of spores to heat is variable and depends on
the state of the sporulation and species present in the samples (Bonde
1962). In general the optimum growth range of Clostridium spp.
influences the thermoresistance of the produced spores. Spores formed
by psychrotrophic spp. at low temperature are less resistant to heat than
spores formed by thermophylic spp. at higher temperatures (Roberts and
Hitchins, 1969).
Thus, subtracting the number of SSCR assessed with pasteurization from the
number of SSRC assessed without pasteurization not necessarily reveals the
concentration of vegetative Clostridium cells but can also be regarded as the
number of heat susceptible Clostridium spores.
The outgrowth of Clostridium spores is preceded by the activation and
germination. Germination of anaerobic spores occurs only under certain
conditions. As previously described germination of the anaerobic spores
may occur under aerobic conditions but outgrowth is restricted to the
anaerobic environment. Gibbs (1964) described that activation of dormant
clostridial spores occurs only in the presence of mixtures of amino acids.
Evidence for outgrowth of Clostridium spores in the sediment-water
___________________________________________________________________
- 33 -
Chapter 1
.
interface of a lake was presented by Molongoski and Klug (1976). The
fraction of anaerobic bacteria in the total bacterial population of the
sediment was 0.001% and with Clostridium as the dominant genus. This
lake was a hyper-eutrophic lake. Thus, the large numbers of vegetative
“anaerobic bacilli” in filters reported by Willis (1957) can be regarded as
debatable results with respect to the conclusion on Clostridium growth in
the aerobic and mostly oligotrophic environment of water treatment. No
information of that kind was found in his work, but if outgrowth of
Clostridium in the filter bed had occurred, these filters were possibly
supplied with water rich in organic matter and operated discontinuously
with anaerobic periods as a result. Both at the time of introduction of the
SSRC parameter in the Dutch Drinking Water Decree in 1984 and also
during the subsequent period until now, indications of growth of SSRC in
water treatment have not been documented.
Analytical methods for Clostridium spore. To assess the
concentration of spores of sulphite-reducing clostridia in water the
environmental spores must complete activation and germination before
they can be determined by outgrowth in an appropriate medium incubated
under the optimal conditions. The method to assess concentrations of SSRC
in water and sediments used in the current study, is described in the Dutch
NEN6567 (Anonymous, 1985a) and based on the work of Havelaar et al.,
(1983). The anaerobic or micro-aerophilic incubation method of the sample
in solidified Perfringens Agar Base medium in between the bottom and the
cover of a petri-dish was a method developed during laboratory practices
in Italy (personal communications, Havelaar 2007). A comparison with the
methods of strict anaerobic incubation revealed small differences (Havelaar
et al., 1983). It was concluded that this simple and cheap incubation method
could be included in the prescribed enumeration method (Anonymous,
1985a).
RESEARCH NEEDS
Assessment of the elimination capacity of water treatment facilities for
micro-organisms has not been common practice in drinking water
production. To enable QMRA in practice of a water company, methods or
strategies have to be developed to assess this elimination capacity of the
local specific operated treatment plant.
Information on variation is of interest because waterborne outbreaks of
diarrhoea caused by protozoan parasites have been related to peak
___________________________________________________________________
- 34 -
Chapter1
concentrations in the source water or inadequacies in the treatment (Dykes
et al., 1980; Badenoch, 1990; Craun 1990; Richardson et al., 1991). Beside
obvious inadequacies in treatment, normal operated processes in a
treatment exhibit a variation in the elimination of micro-organisms.
LeChevallier and Norton (1992) for instance showed that the removal of
particles of > 5 μm (size of Cryptosporidium oocysts) by individual unit
filters in the filtration stage of a treatment may vary by as much as 1.000fold. These variations will influence the overall efficiency of a treatment
stage. Variation in the removal of (oo)cysts of Cryptosporidium and Giardia
by a conventional treatment (coagulation/ floc removal plus filtration) was
observed by Hashimoto et al. (2001). Elimination capacity for
Cryptosporidium and Giardia ranged from 2.0 – 3.2 and 1.7 – 3.1 log,
respectively.
The variability of elimination of processes for Cryptosporidium has also been
demonstrated by spiking tests on pilot plant scale or in laboratory
experiments. Emelko (2001) intensively studied the removal of C. parvum
oocysts by filtration processes under different filtration conditions.
Elimination assessed under stable operation, ripening, sub-optimal
coagulant dosing, early and late breakthrough ranged between 5.5 and <0.5
log. Also with spiking experiments on laboratory scale (dose/response data
with a continuous flow system) a high variability of the efficiency of ozone
disinfection for C. parvum in natural waters was demonstrated
(Oppenheimer et al., 2000). The average Chick/Watson inactivation
constant at 100C was 0.21 and a range of 0.08 – 0.46 l/mg.min.
This variation in the elimination of micro-organisms is caused by the
multiple variables involved in the mechanisms responsible for the
elimination in the different treatment processes (Table 10).
Inactivation which is passive die off or active destruction of the organisms,
is an elimination mechanism significant for water treatment processes such
as infiltration and soil passage, storage in reservoirs and disinfection.
Especially the efficiency of the disinfection process is influenced by a lot of
abiotic conditions. Removal and interception of micro-organisms as
mechanism, caused by coagulation and sedimentation, straining and
attachment/detachment is also under the influence of partly the same
abiotic conditions.
Due to these observations and considerations it is of importance for risk
management to integrate the level of variability in both source water
concentrations and in elimination capacity in the variability of the infection
risk assessed with QMRA. Additional statistical tools to integrate
___________________________________________________________________
- 35 -
Chapter 1
.
uncertainty calculations in QMRA have been described (Teunis et al., 1997;
Evers and Groennou, 1999; Haas et al., 1999; Medema et al., 2003).
Moreover, in the framework of WSP it is of importance to determine the
causes of these fluctuations since that can lead to measures in process
design or operation to minimize infection risks (Smeets, 2008).
Table 10. Elimination mechanisms significant for the different treatment processes
related to the applied processes and the abiotic conditions influencing (+) the
elimination
Inactivation
Removal/interception
Processes:
a
Die off Disinfect. Coag./Sed. Strain. Att./det.
Mechanisms:
Soil passage
+
+
+
Storage
+
+
Coag./flocc.
+
+
Disinfection
+
Filtration
+
+
Abiotic conditions:
+
+
+
+
- Temperature
+
- DOC
+
- UV
+
+
+
- Fte
+
+
+
- pH
+
- EGV
+
+
+
- Hydraulics
+
+
+
+
- Contact time
+
- Granular mat.b
c
+
- Grains charact.
+
+
- Chemical type
+
+
- Dosed
a Disinfection, coagulation and sedimentation, straining, Attachment and
Detachment; b Type of granular material; c size and uniformity; d
assessment and control
As a consequence of these considerations on variability of elimination capacity,
data on treatment efficiency for QMRA must be as site specific and actual as
possible.
___________________________________________________________________
- 36 -
Chapter1
Development of a method to determine the elimination of faecal
indicators. Assessing the elimination capacity of treatment processes and a
train of processes at a treatment facility is one of the current challenges in
drinking water industry (Medema, 1999). The general hypothesis is that E.
coli and SSRC can be used as process indicators for susceptible and the
more resistant pathogens, respectively. The standard microbial methods
faecal indicators can be used to assess elimination capacity of the first steps
in the treatment train. The detection limit of the current methods, however,
is too low to detect E. coli and SSRC after subsequent processes. Thus, a first
need is a method with a lower detection limit to determine the
concentrations of E. coli and SSRC in the water in the last stages of
treatment.
Comparative studies on process indicators and pathogens. The
major missing information to decide whether both E. coli and SSRC are
usefull process indicators for viruses, bacteria and protozoan (oo)cysts are
comparative data on the elimination in water treatment assessed under the
same conditions. There are a number of morphological, physical and
physiological differences between both indicator bacteria and the target
pathogens which will affect the similarity of organism removal during
treatment. To elucidate the influence of these differences on the elimination
of indicator and pathogen, comparative studies are necessary, ideally
under full-scale conditions or otherwise with spiking studies using
multiple micro-organisms under well defined experimental conditions.
These comparative studies are needed to answer the following questions:
- Is the elimination of the proposed process indicators E. coli and
SSRC in water treatment predictive for the elimination (removal
and inactivation) of pathogens and, if so, for which pathogens?
- To what extent are comparative (in)filtration and inactivation
experiments predictive for the observed elimination under field
conditions and what is the influence of the experimental scale on
the predicted elimination?
- Can these comparative experiments be used to elucidate the process
conditions influencing the variation of elimination of microorganisms?
Validation and extension of the data base with literature data.
Collecting site specific information on elimination of process indicator (E.
coli and SSRC) supplemented with comparative spiking studies, is a
relative large effort and is most likely not feasible for each specific
situation. The scientific knowledge on elimination of micro-organisms by
___________________________________________________________________
- 37 -
Chapter 1
.
water treatment processes has been extended during the last decade and
potentially includes valuable and applicable information. There is a need
for exploration of the literature and systematically assemble relevant and
reliable quantitative information in a central data base which enables to
obtain a default value of the elimination capacity of a treatment process
and the conditions of influence on elimination. These data can be used in
the tiered approach of QMRA (Medema et al., 2003) for assessing a point
estimate of the annual risk of infection. When the calculated infection risk is
close to the required safety level of 10-4 and the available full-scale data on
process indicator removal is limited, a distribution of the risk of infection
can be calculated by more comprehensive statistical methods using the
range of elimination capacities collected in this data base (Smeets, 2008).
CONCLUSIONS AND OBJECTIVES OF THE CURRENT STUDY
The developments in the supply of microbiologically safe drinking water
over the last century (1900-2000) have been described showing that most
developments occurred at the turn of both centuries. The period with the
highest impact on the production and control of microbiologically safe
drinking water was around 1900. During the 20th century the fraction of the
population in the Netherlands connected to a central drinking water
steadily increased to almost 100% with a decreasing incidence of outbreaks
of waterborne diseases.
At the end of the 20th century, however, epidemiological studies showed
that outbreaks of waterborne diseases in the developed countries were not
totally banned. Outbreak investigations showed shortcomings in both
microbiological water quality monitoring and water treatment with respect
to more persistent pathogens such as Cryptosporidium, a pathogen
considered as a new challenge to the drinking water industry (Medema,
1999). Developments in the Netherlands resulted in a shift from reactively
end product quality monitoring to proactively quantitative microbial risk
assessment QMRA. QMRA is a new method to assess the safety of drinking
water by means of calculation of the risk of infection to the consumers. This
risk of infection is estimated from the dose/response curves for the
individual pathogen on the one hand and the exposure assessment on the
other. The exposure assessment is derived from the concentration of
pathogens in the source water, the elimination capacity of the water
treatment and the drinking water consumption. A major lack in the process
of QMRA is quantitative information on the elimination capacity of water
treatment processes for pathogenic micro-organisms and the variation in
___________________________________________________________________
- 38 -
Chapter1
elimination. Teunis et al (1997) showed that the uncertainty in the estimated
removal efficiency of treatment processes dominates over uncertainties in
the estimates of the other factors used in QMRA.
The objective of the current study is to develop and evaluate methods for
the assessment of the elimination capacity of full-scale water treatment
processes particularly for bacteria and protozoan (oo)cysts. Elimination of
both environmental and spiked micro-organisms under full-scale, pilot
plant and laboratory conditions is determined and compared in order to
evaluate the suitability of these methods for the assessment of the
elimination capacity of processes and the variability of this elimination
capacity.
The objective of the study is divided in a number of sub-goals
1. Evaluation of Coli44 (incl. E. coli) and SSRC (incl. C. perfringens) as
process indicators to quantify the elimination capacity of full-scale
water treatment for the index pathogens Campylobacter and
Cryptosporidium/Giarda, respectively.
2. Develop and apply large volume sampling to assess the concentrations
of indicator bacteria in the final stages of full-scale water treatment.
3. Compare the elimination of E. coli and C. perfringens with the
elimination of pathogens in a selection of processes by comparative
spiking experiments and literature data.
4. Determine the value of challenge tests under pilot plant or laboratory
scale for the assessment of the elimination capacity of full-scale
processes.
5. Explore and analyse literature data on micro-organism removal in
water treatment as tool for assessment of the elimination capacity and
the process conditions affecting removal.
6. Develop a general strategy to quantify the elimination of pathogenic
micro-organisms in water treatment processes.
OUTLINE OF THE THESIS
In the first three Chapters of this thesis the sub-objectives one and two were
studied. Historical data on microbiological water quality control were
evaluated to explore the applicability of faecal indicator bacteria
monitoring for quantifying removal of micro-organisms in full-scale
treatments. These data are presented in Chapter 2 together with results
from a pilot plant study applying a method with decreased detection limit
___________________________________________________________________
- 39 -
Chapter 1
.
to monitor SSRC removal. In Chapters 3 and 4, a large volume method was
developed and applied. The method is an up-scaled version of the
routinely applied membrane filtration (mf) method and can be used for insitu aseptically sampling of the water (Chapter 3). The developed MFsampler was used to determine E. coli and SSRC removal under full-scale
conditions (Chapter 4).
Sub-objectives three, four and five were the basis for the following five
studies. In Chapters 5 and 6 the use of both process indicators, SSRC and E.
coli, for the elimination of persistent and susceptible organisms by ozone
and UV-disinfection was investigated. Comparative spiking experiments in
continuous-flow bench-scale ozone systems and literature reviewing and
meta-analysis of UV-disinfection studies were used. The determination of
the elimination capacity of slow sand filtration and surface infiltration
systems for viruses, bacteria and protozoan parasites was studied in
Chapters 7, 8 and 9. Columns and a pilot plant were used for this purpose
and results were compared with full-scale data of the removal of both
process indicators under the local full-scale conditions.
In the general discussion (Chapter 10) the results of all these studies
supplemented with additional experimental data are discussed with the
objective to describe a general strategy to determine the elimination
capacity of a local drinking water treatment for pathogenic microorganisms. The use of E. coli and SSRC as process indicators under fullscale conditions and the translation of these data to the index pathogens
was subject of the discussion. Furthermore, the value of experimental
methods such as chalenge tests at laboratory or pilot plant scale and
literature reviewing was evaluated.
REFERENCES
Achalme, P. 1891-1897. Compt. rend. Soc. de biol., cit. from Achalme 1897. Ann.
Inst. Past. 11:845.
Adam, R. D. 2001. Biology of Giardia lamblia. Clin. Microbiol. Rev. 14:447-475.
Angelotti, R., H. E. Hall, M. J. Foter, and K. H. Lewis. 1962. Quantitation of
Clostridium perfringens in foods. Appl. Microbiol. 10:193-9.
Anonymous. 1942. Voorschriften voor de bacteriologische keuring van drinkwater:
Hoofdcommissie voor de normalisatie in Nederland.
Anonymous. 1956. Bacteriologisch onderzoek van drinkwater: Hoofdcommissie
voor de normalisatie in Nederland. N.V. Keuringsinstituut voor
Waterleidingartikelen (KIWA).
Anonymous. 1962. Jaarverslag 1962. Gemeentewaterleidingen Amsterdam.
___________________________________________________________________
- 40 -
Chapter1
Anonymous. 1975. Richtlijn 75/440/EEG van de Raad van 16 juni 1975 betreffende
de vereiste kwaliteit van het oppervlaktewater dat is bestemd voor produktie
van drinkwater in de Lid-Staten.
Anonymous. 1980. Kwaliteit van voor menselijke consumptie bestemd water:
Publicatieblad Nr. L 229/11 van 30.08/1980.
Anonymous. 1981. Bacteriologisch onderzoek van drinkwater. Onderzoek met
behulp van membraanfiltratie naar de aanwezigheid van bacteriën van de
coligroep, Nederlands Normalisatie Instituut NEN6553.
Anonymous. 1982. Bacteriologisch onderzoek van drinkwater. Onderzoek met
behulp van membraanfiltratie naar de aanwezigheid van faecale streptococcen,
Nederlands Normalisatie Instituut NEN6564.
Anonymous. 1984. Waterleidingbesluit Staatsblad van het Koninkrijk der
Nederlanden, 220: 1-35.
Anonymous. 1985a. Bacteriologisch onderzoek van drinkwater. Onderzoek met
behulp van membraanfiltratie naar de aanwezigheid van sporen van sulfietreduucerende clostridia, Nederlands Normalisatie Instituut NEN6567.
Anonymous. 1985b. Nederlandse praktijkrichtlijn; toelichting bij Nederlands
Normalisatie Instituut NEN6567, vol. NPR6568.
Anonymous. 2001. Besluit van 9 januari 2001 tot wijziging van het
waterleidingbesluit in verband met de richtlijn betreffende de kwaliteit van voor
menselijke consumptie bestemd water, p. 1-53, vol. 31. Staatsblad van het
Koninkrijk der Nederlanden.
Anonymous. 2003. 3rd Edition of the Guidelines for Drinking Water Quality:
World Health Organisation, Geneve, Switzerland.
Atherholt, T. B., M. W. Lechavallier, W. D. Norton, and J. S. Rosen. 1998. Effect of
rainfall on Giardia and Cryptosporidium. J. Am. Water Works Assoc. 90:66-80.
Ashbolt, N. J., W. O. K. Grabow, and M. Snozzi. 2001. Indicators of microbial
water quality. In Water Quality: Guidelines, Standards and Health, p. 289-316,
World Health Organization WHO, London, UK.
Badenoch, J. 1990. Cryptosporidium in water supplies, London.
Bonde, G. J. 1962. Bacterial indication of water pollution. Teknisk Forlag,
Copenhagen.
Bonde, G. J. 1977. Bacterial indication of water pollution. Adv. Aquat. Microbiol.
1:273-364.
Buchholtz ten Brink, M. R., E. L. Mecray, and E. L. Galvin. 2000. Clostridium
perfringens in Long Island sound sediments: an urban sedimentary record. J. of
Coast. Research 16:591-612.
Bush, J. H. 1953. New bacteriological technique for testing water and sewage.
Water and Sewerage Works. 100:151.
Brush, C. F., M. F. Walter, L. J. Anguish, and W. C. Ghiorse. 1998. Influence of
pretreatment and experimental conditions on electrophoretic mobility and
hydrophobicity of Cryptosporidium parvum oocysts. Appl. Environ. Microbiol.
64:4439-4445.
___________________________________________________________________
- 41 -
Chapter 1
.
Cabelli, V. J. 1977. Clostridium perfringens as a water quality indicator. In A. W.
Hoadley and B. J. Dutka (ed.), Bacterial indicators: health hazards related with
water, vol. 635, p. 65-79ASTM STP.
Caplenas, N. R., and M. S. Kanarek. 1984. Thermotolerant Non-fecal Source
Klebsiella pneumoniae: Validity of the Fecal Coliform Test in Recreational Waters.
Am. J. Public Health 74:1273-1275.
Clancy, J. L., T. M. Hargy, M. M. Marshall, and J. E. Dyksen. 1998. UV light
inactivation of Cryptosporidium oocysts. J. Am. Water Works Assoc. 90(9):92–102.
Coffey, B. M., P. M. Huck, D. D. Maurizo, M. B. Emelko, I. P. Douglas, and J.
Van den Oever. 1999. Presented at the Water Quality Technol. Conf., Tampa,
Florida, US.
Colford, J. M., T. J. Wade, S. K. Sandhu, C. C. Wright, S. Lee, S. Shaw, K. Fox, S.
Burns, A. Benker, M. A. Brookhart, M. van der Laan, and D. A. Levy. 2005. A
Randomized, Controlled Trial of In-Home Drinking Water Intervention to
Reduce Gastrointestinal Illness. Am. J. Epidemiol. 161:472-482.
Craun, G. F. 1988. Surface water supplies and health. J. Am. Water Works Assoc.
80(2):40-52.
Craun, G. F. 1990. Waterborne Giardiasis. In E. A. Meyer (ed.), Human Parasitic
Diseases, vol. 3, p. 267-293. Elsevier Science Publication, Amsterdam, The
Netherlands.
Craun, G. F., P. S. Berger, and R. L. Calderon. 1997. Coliform bacteria and
waterborne disease outbreaks. J. Am. Water Work Assoc. 89(3):96-104.
Craun, G. F., S. A. Hubbs, F. Frost, C. R. L. Calderon, and S. H. Via. 1998.
Waterborne outbreaks of cryptosporidiosis. J.Am. Water Works Assoc. 90:81-91
de Roda Husman, A. M., H. A. M. Ketelaars, W. J. Lodder, G. J. Medema, and F.
M. Schets. 2001. De microbiologische kwaliteit van het ingenomen en
afgeleverde water van Waterwinningbedrijf Brabantse Biesbosch in 2001.
RIVM330250002 (in Dutch).
Douglas, H. W. 1955. Trans. Faraday Soc. 51:146.
Driks, A. 2007. Surface appendages of bacterial spores. Mol. Microbiol. 63:623–625.
Drozd, C., and J. Schwartzbrod. 1996. Hydrophobic and electrostatic cell surface
properties of Cryptosporidium parvum. Appl. Environ. Microbiol. 62:1227-1232.
Dugan, N. R., K. R. Fox, J. H. Owens, and R. J. Miltner. 2001. Controlling
Cryptosporidium Oocysts Using Conventional Treatment. J. Am. Water Works
Assoc. 93(12):64-76.
Dutka, B. J. 1973. Coliform are an inadequate index of water quality. J. Env. Health
36:39-46.
Dullemont, Y. J., A. Visser, J. F. Schijven, and W. A. M. Hijnen. 2004.
Eliminatiecapaciteit van langzame zandfiltratie voor micro-organismen bepaald
met doseerproeven. H2O 37(13):25-27.
Dullemont, Y. J., J. F. Schijven, W. A. M. Hijnen, M. Collin, A. Magic-Knezev,
and W. A. Oorthuizen. 2006. Presented at the Recent progress in slow sand and
alternative biofiltration, Mullheim, Germany.
___________________________________________________________________
- 42 -
Chapter1
Dykes, A. C., D. D. Juranek, R. A. Lorenz, S. Sinclair, W. Jakubowski, and R. B.
Davies. 1980. Municipal waterborne Giardiasis: an epidemiological
investigation. Ann. Int. Med. 92:165-170.
Eijkman, C. 1904. Die Gärungsprobe bei 46oC als Hilfsmittel bei der
Trinkwasseruntersuchung. Centralblat für Bakteriologie Abth. 1 Orig. 37:742.
Emelko, M. B. 2001. Removal of Cryptosporidium parvum by granular media
filtration. University of Waterloo, Ontario, Canada.
Escherich, T. 1885. Die Darmbakterien des Neugeborenen und Säuglings.
Fortschritte des Medicin 3:515-547.
Evers, E. G., and J. T. Groennou. 1999. Berekening van de verwijdering van microorganismen bij de bereiding van drinkwater. RIVM 734301016.
Fallacara, D. M., C. M. Monahan, T. Y. Morishita, and R. F. Wack. 2001. Fecal
shedding and antimicrobial susceptibility of selected pathogens and a survey of
intestinal parasites in free-living waterfowl. Avian. Dis. 45.
Fernandes, T. M. A., C. Schout, A. M. De Roda Husman, A. Eilander, H.
Vennema, and Y. T. H. P. Van Duynhoven. 2007. Gastroenteritis associated with
accidental contamination of drinking water with partially treated water.
Epidemiol. Infect. 135:818-826.
Finch, G. R., E. K. Black, L. Gyurék, and M. Belosevic. 1993. Ozone inactivation of
Cryptosporidium parvum in demand-free phosphate buffer determined by in vitro
excystation and animal infectivity. Appl. Environ. Microbiol. 59:4203-4210.
Foster, S. J., and K. Johnstone. 1990. Pulling the triggers: the mechanism of
bacterial spore germination. Mol. Microbiol. 4:137–141.
Frankland, P., and P. Frankland. 1894. Micro-organisms in water. Their
significance, identification and removal. Longmans, Green and co.,
London/New York.
Fujioka, R. S., and L. K. Shizumura. 1985. Clostridium perfringens, a reliable
indicator of stream water quality. Jour. Wat. Pol. Control Fed., 57:986-992.
Galbraith, N.S., N. J. Barrett, R. and Stanwell-Smith. 1992. Water and disease
after Croydon: a review of water-borne and water-associated disease in the UK
1937-86. J NEWWA 169-185.
Galofre, B., S. Israel, J. Dellunde, and F. Ribas. 2004. Aerobic bacterial spores as
process indicators for protozoa cysts in water treatment plants. Wat. Sci Techn.
50(1):165-72.
Gerba, C. P., and J. B. Rose. 1993. Estimating viral disease risk from drinking
water. In C. R. Cothern (ed.), Comparative Environmental Risk Assessment, p.
117-135, Lewis Publishers.
Gibbs, P. A. 1964. Factors Affecting the Germination of Spores of Clostridium
Bifermentans. J. Gen. Microbiol. 37:41-8.
Godfree, A. F., D. Kay, and M. D. Wyer. 1997. Faecal streptococci as indicators of
faecal contamination in water. J. of Appl. Microbiol. Symp. Suppl. 83:110S-119S.
Gould, G. W. 2006. History of science – spores. Lewis B Perry Memorial Lecture
2005. J. Appl. Microbiol. 101:507–513.
___________________________________________________________________
- 43 -
Chapter 1
.
Gould, G. W., and A. Hurst. 1969. The Bacterial Spore. Acacemic Press, London
and New York.
Graaf de, W. C. 1922. Een en ander over het bacteriologisch wateronderzoek.
Tijdschrift voor Microbiologie en Gezondheidsleer, Vergelijkende en Tropische
Geneeskunde, Parasitaire en Infectieziekten 7:1-22.
Grabow, W.O.K., V. Gauss-Muller, O. W. Prozesky, and F. Deinhardt. 1983.
Inactivation of Hepatitis A virus and indicator organisms in water by free
chlorine residuals. Appl. Environ. Microbiol. 46: 619-624.
Haag, W. R., and J. Hoigné. 1983. Ozonation of bromide-containing waters:
kinetics of formation of hypobromous acid and bromate. Environ. Sci. Technol.
17:261-267.
Haas, C. N. 1983. Estimation of risk due to low doses of micro-organisms: a
comparison of alternative methodologies. Am. J. Epidemiol 118:573-82.
Haas, C. N., and J. N. S. Eisenberger. 2001. Risk Assessment. In L. Fewtrell and J.
Bartram (ed.), Water Quality: Guidelines, Standards and Health, p. 161-183. IWA
Publishing, London, UK.
Haas, C. N., J. B. Rose, C. P. Gerba, and S. Regli. 1993. Risk assessment of virus in
drinking water. Risk analysis 13:545-522.
Haas, C. N., J. B. Rose, and C. P. Gerba. 1999. Quantitative microbial risk
assessment. John Wiley & Sons, New York, USA.
Harris, J. R., and F. Petry. 1999. Cryptosporidium parvum: Structural Components of
the Oocyst Wall. J. Parasitol. 85:839-849.
Hashimoto, A., T. Hirata, and S. Kunikane. 2001. Occurrence of Cryptosporidium
oocysts and Giardia cysts in a conventional water purification plant. Wat. Sci.
Techn. 43(12):89-92.
Havelaar, A. H. 1983. Microbiologisch onderzoek van drinkwater. H2O 16:105-112.
Havelaar, A. H., M. van Olphen, and Y. C. Drost. 1993. F-specific RNA
bacteriophages are adequate model organisms for enteric viruses in fresh water.
Appl. environ. Microbiol. 59:2956-2962.
Havelaar, A. H. 1994. Application of HACCP to drinking water supply. Food
Control 5:145-152.
Havelaar, A. H., M. van Olphen, and J. F. Schijven. 1995. Removal and
inactivation of viruses by drinking water treatment processes under full scale
conditions. Wat. Sci. Tech. 31(5-6):55-62.
Havelaar, A. H., A. E. M. De Hollander, P. F. M. Teunis, E. G. Evers, H. J. Van
Kranen, Johanna, F. M. Versteegh, J. E. M. Van Koten, and W. Slob. 2000.
Balancing the Risks and Benefits of Drinking Water Disinfection: Disability
Adjusted Life-Years on the Scale. Environ. Health Perpect. 108:315-321.
Heijnen, L., and G. J. Medema. 2006. Quantitative detection of E. coli, E.coli O157
and other shiga toxin producing E. coli in water samples using a culture method
combined with real-time PCR. J. Water and Health 04(4):487-498.
Hellard, M. E., M. I. Sinclair, A. B. Forbes, and C. K. Fairley. 2001. A
Randomized, Blinded, Controlled Trial Investigating the Gastrointestinal Health
Effects of Drinking Water Quality. Environ. Health Perpect. 109:773-778.
___________________________________________________________________
- 44 -
Chapter1
Herwaldt, B. L., G. F. Craun, S. L. Stokes, and D. D. Juranek. 1992. Outbreaks of
waterborne disease in the United States: 1989-90. J. Am. Water Works Assoc.
84(4):129-135.
Hijnen, W. A. M., A. Visser, and D. Van der Kooij. 1995. Verwijdering van
Campylobacter bacteriën bij de drinkwaterbereiding op productielocatie
Scheveningen, DZH. SWO 95.272, Kiwa Water Research, Nieuwegein NL.
Hijnen, W. A. M., W. M. H. van der Speld, F. A. P. Houtepen, and D. van der
Kooij. 1997. Presented at the International Symposium on Waterborne
Cryptosporidium, Newport Beach, California, US.
Hijnen, W. A. M., A. Visser, and G. J. Medema. 1998. Verwijdering van
indicatorbacterien bij de drinkwaterbereiding op productielocatie Scheveningen,
NV Duinwaterbedrijf Zuid-Holland. SWI 98.216 Kiwa Water Research
Nieuwegein NL.
Hijnen, W. A. M., T. G. J. Bosklopper, J. A. M. H. Hofmann, A. D. Bosch, and G.
J. Medema. 2001. Presented at the 10th Int. Ozone Assoc. Congres, IOA, London.
Hill, R. T., I. T. Knight, M. S. Anikis, and R. R. Colwell. 1993. Benthic distribution
of sludge indicated by Clostridium perfringens at a deep-ocean dump site. Appl.
Environ. Microbiol. 59:47-51.
Hoogenboezen, W., G. J. Medema, H. A. M. Ketelaars, G. B. J. Rijs, and J. F.
Schijven. 2001. Cryptosporidium and Giardia: occurrence in sewage, manure and
surface water. RIWA report.
Houston, A. C. 1899. Bacterioscopic Examination of Drinking-water with particular
reference to the relations of Streptocooci and Staphylococci with Water of this
Class. Supplement to the Twenty-eighth Annual Report of the Local Government
Board Containing the Report of the Medical Officer for 1898-99, 467.
Hrudey, S. E., P. Payment, P. M. Huck, R. W. Gillham, and E. J. Hrudey. 2003. A
fatal waterborne disease epidemic in Walkerton, Ontario: comparison with other
waterborne outbreaks in the developed world. Wat. Sci. Tech. 47(3):7-14.
Huisman, J., and P. J. Nobel. 1981. Enkele epidemiologische gegevens over de
gevolgen van de faecale drinkwaterverontreiniging in het Scheepsvaartkwartier
te Rotterdam in maart 1981. H2O, 6(8):199-200.
Hsu, B.-M., and C. Huang. 2002. Influence of ionic strength and pH on
hydrophobicity and zeta potential of Giardia and Cryptosporidium. Colloids Surf.
A 201:201-206.
Jofre, J., E. Ollé, F. Ribas, A. Vidal, and F. Lucena. 1995. Potential usefulness of
bacteriophages that infect Bacteroides fragilis as model organisms for monitoring
virus removal in drinking water treatment plants. Appl. environ. Microbiol.
61:3227-3231.
Johnstone, K. 1994. The trigger mechanism of spore germination: current concepts.
Soc. Appl. Bacteriol. Symp. Ser. 23:17S-24S.
Ketelaars, H. A. M., G. J. Medema, L. C. W. A. Van Breemen, D. van der Kooij, P.
J. Nobel, and P. Nuhn. 1995. Occurrence of Cryptosporidium oocysts and Giardia
cysts in the river Meuse and removal in the Biesbosch reservoirs. Aqua 44(1):108111.
___________________________________________________________________
- 45 -
Chapter 1
.
Keynan, A., and Z. Evenchik. 1969. Activation. In G. W. Gould and A. Hurst (ed.),
The Bacterial Spore. Academic Press, London and New York.
Klein, E. 1895. Centralbl. f. Bakt. 18, 737.
Klein, E. 1897-1898. Report on the Morphology and Biology of Bacillus enteritidis
sporogens; on the association of this microbe with infantile diarrhoea and with
Choler Nostras; and its relation with milk, with sewage, and with manure.
Supplement to the Twenty-seventh Annual Report of the Local Government
Board Containing the Report of the Medical Officer for 1897-98, 210.
Klein, E., and A. C. Houston. 1898-1899. Further reports on bacteriological
evidence of recent and therefore dangerous sewage pollution of elsewise potable
waters. Supplement to the Twenty-eigth Annual Report of the Local Government
Board Containing the Report of the Medical Officer for 1898-99, 498.
Koch, R. 1884. Die Aetiologie der Tuberkulose. Mitt Kaiser Gesundh. 2: 1-88.
Koch, R. 1893. Wasserfiltration und Cholera. Gesammelte Werke, vol. Bd 2 Teil 1,
1912. Georg Thieme, Leipzig.
Kuznar, Z. A., and M. Elimelech. 2005. Role of Surface Proteins in the Deposition
Kinetics of Cryptosporidium parvum Oocysts. Langmuir 21:710-716.
LeChevallier, M. W., and W. D. Norton. 1992. Examining relationships between
particle counts, and Giardia and Cryptosporidium and turbidity. J. Am. Water
Works Assoc. 84(12):54-60.
LeChevallier, M. W., W. D. Norton, and R. G. Lee. 1991. Occurrence of Giardia and
Cryptosporidium spp. in surface water supplies. Appl. Environ. Microbiol.
57:2610-6.
LeChevallier, M.W. and K. Au. 2004. Water treatment and pathogen control:
process efficiency in achieving safe drinking water: World Health Organisation,
IWA Publishing, London, UK.
Leuschner, R. G. K., A. C. Weaver, and P. J. Lillford. 1999. Rapid particle size
distribution analysis of Bacillus spore suspensions. Colloids and Surfaces B:
Biointerfaces 13:47–57.
Lewis, J. C. 1969. Dormancy. In G. W. Gould and A. Hurst (ed.), The Bacterial
Spore. Academic Press, London and New York.
MacConkey, A. 1905. Lactose-fermenting bacteria in faeces. Journal of Hygiene
5:333-379.
MacKenzie, W. R. H., N.J. , M. E. Proctor, S. Gradus, K. A. Blair, D. E. Peterson, J.
J. Kazmierczak, D. G. Addiss, K. R. Fox, J. B. Rose, and J. P. Davis. 1994. A
massive outbreak in Milwaukee of Cryptosporidium infection transmitted through
the public water supply. New. Engl. J. Med. 331:161-167.
Matz, L. L., T. Cabrera Beaman, and P. Gerhardt. 1970. Chemical Composition of
Exosporium from Spores of Bacillus cereus. J. Bacteriol. 101:196-201.
Mazoua, S., and E. Chauveheid. 2005. Aerobic spore-forming bacteria for
assessing quality of drinking water produced from surface water. Water Res.
39:5186-98.
McTigue, N.E., M. LeChevallier, and J. Clancy. 1996. Findings of the national
particle project. In Annual Conference AWWA. Toronto, Canada.
___________________________________________________________________
- 46 -
Chapter1
McTigue, N.E.e.a. 1998. National assessment of particle removal by filtration:
AWWARF, Denver CO US.
Medema, G. J., and A. H. Havelaar. 1994a. Micro-organismen in water: een
gezondheidsrisico. RIVM 289202002.
Medema, G. J., and F. M. Schets. 1994b. Campylobacter en Salmonella in open
reservoirs voor de drinkwaterbereiding. RIVM 149103002.
Medema, G. J., M. Bahar, and F. M. Schets. 1997. Survival of Cryptosporidium
parvum, Escherichia coli, faecal streptococci and Clostridium perfringens in river
water. Wat. Sci. Tech. 35(11-12):249-252.
Medema, G. J., P. F. Teunis, A. H. Havelaar, and C. N. Haas. 1996. Assessment of
the dose-response relationship of Campylobacter jejuni. Int. J. Food Microbiol.
30:101-11.
Medema, G. J., F. M. Schets, P. F. Teunis, and A. H. Havelaar. 1998.
Sedimentation of free and attached Cryptosporidium oocysts and Giardia cysts in
water. Appl. Environ. Microbiol. 64:4460-6.
Medema, G. J. 1999. Cryptosporidium and Giardia: new challenges to the water
industry. University of Utrecht, Utrecht, NL.
Medema, G. J. 2002. Effecten nieuwe wetgeving BTO 2002.123 (c). Kiwa Water
Research.
Medema, G. J., W. Hoogenboezem, A. J. van der Veer, H. A. M. Ketelaars, and W.
A. M. Hijnen, and P. J. Nobel. 2003. Quantitative risk assessment of
Cryptosporidium in surface water treatment. Wat. Sci. Tech. 47(3):241-247.
Medema, G. J., P. F. M. Teunis, M. Blokker, D. Deere, A. Davison, P. Charles,
and J. F. Loret. 2006. WHO Guidelines for Drinking Water Quality:
Cryptosporidium. World Health Organization, London UK.
Molongoski, J. J., and M. J. Klug. 1976. Characterization of anaerobic
heterotrophic bacteria isolated from freshwater lake sediments. Appl. Environ.
Microbiol. 31:83-90.
Moore, A. C., B. L. Herwaldt, G. F. Craun, R. L. Calderon, A. K. Highsmith, and
D. D. Juranek. 1994. Waterborne disease in the United States, 1991 and 1992. J.
Am. Water Works Assoc. 86(2):87-99.
Mossel, D. A. A. 1982. Marker (index and indicator) organisms in food and
drinking water. Semantics, ecology, taxonomy and enumeration. Ant.
Leeuwenhoek 48:609-611.
Murrell, W. G. 1969. Chemical composition of spores and spore structures. In G.
W. Gould and A. Hurst (ed.), The bacterial spore. Academic Press, London and
New York.
Nieminski, E. C., and J. E. Ongerth. 1995. Removing Giardia and Cryptosporidium
by conventional treatment and direct filtration. J. Am. Water Works Assoc.
87(9):96-106.
Novak, J. S., V. K. Juneja, and B. A. McClane. 2003. An ultrastructural comparison
of spores from various strains of Clostridium perfringens and correlations with
heat resistance parameters. Int. J. Food Microbiol. 86:239– 247.
___________________________________________________________________
- 47 -
Chapter 1
.
Oesterholt, F., G. Martijnse, G. J. Medema, and D. Van der Kooij, D. 2007. Health
risk assessment of non-potable domestic water supplies in the Netherlands. J
Wat Sup: Aqua 56, 171-179.
Oppenheimer, J. A., E. M. Aieta, R. R. Trussell, J. G. Jacangelo, and I. N. Najm.
2000. Evaluation of Cryptosporidium inactivation in natural waters. American
Water Works Association Research Foundation, Denver CO US.
Patania, N. L., J. G. Jacangelo, L. Cummings, A. Wilczak, K. Riley, and J.
Oppenheimer. 1995. Optimization of Filtration for Cyst Removal. American
Water Works Association Research Foundation, Denver CO US.
Payment, P., L. Richardson, J. Semiatycki, R. Dewar, M. Edwardes, and E.
Franco. 1991. A randomized trial to evaluate the risk of gastrointestinal disease
due to consumption of drinking water meeting current microbiological
standards. Am. J. Public Health, 81:703-708.
Payment, P., J. Siemiatycki, L. Richardson, G. Renaud, E. Franco, and M. Prevost.
1997. A prospective epidemiological study of gastrointestinal health effects due
to the consumption of drinking water. Int. J. Environ. Health Res. 7:5-31.
Payment, P., M. Trudel, and R. Plante. 1985. Elimination of viruses and indicator
bacteria at each step of treatment during preparation of drinking water at seven
water treatment plants. Appl. Environ. Microbiol. 49:1418-28.
Payment, P., and E. Franco. 1993. Clostridium perfringens and somatic coliphages as
indicators of the efficiency of drinking water treatment for viruses and protozoan
cysts. Appl. Environ. Microbiol. 59:2418-24.
Prescott, S. C., C. A. Winslow, and M. H. McCrady. 1945. Water Bacteriology with
special reference to sanitary water analysis, 6th ed. John Wiley & Sons, New
York.
Regli, S., J. B. Rose, C. N. Haas, and C. P. Gerba. 1991. Modeling the risk from
Giardia and viruses in drinking water. J. Am. Water Works Assoc. 83:76-84.
Rice, E.W., K. R. Fox, R. J. Miltner, D. A. Lytle, and C. H. Johnson. 1996.
Evaluating plant performance with endospores. J. Am. Water Works Assoc. 88:
122-130.
Richardson, A. J., R. A. Frankenberg, A. C. Buck, J. B. Selkon, J. S. Colbourne, J.
W. Parsons, and R. T. Mayon-White. 1991. An outbreak of waterborne
cryptosporidiosis in Swindon and Oxfordshire. Epidemiol. Infect. 107:485-95.
Roberts, T. A., and A. D. Hitchins. 1969. Resistance of spores. In G. W. Gould and
A. Hurst (ed.), The Bacterial Spore. Academic Press, London and New York.
Robles, S., J. M. Rodríguez, I. Granados, and M. C. Guerrero. 2000. Sulfitereducing clostridia in the sediment of a high mountain lake (Laguna Grande,
Gredos, Spain) as indicators of fecal pollution. Internat. Micorbiol. 3:187-191.
Rook, J. J. 1974. Formation of haloforms during chlorination of natural water.
Water Treatment Exam. 23:234-245.
Rose, J. B., C. N. Haas, and C. P. Gerba. 1993. Waterborne pathogens: assessing
health risks. Health and Environment Digest. 7:1-8.
___________________________________________________________________
- 48 -
Chapter1
Rosef, O., and A. V. Mork. 1985. Kontaminert brǿnnvann som årsak til
Campylobacter-diaré (Contaminated well water as cause of Campylobacterdiarrhoea). Nor. Vet. Todsskr. 97:891-832.
Rosef, O., G. Rettedal, and L. Lageide. 2001. Thermophilic Campylobacters in
surface water: a potential risk of campylobacteriosis. Int. J. Environ. Health Res.
11:321-7.
Sartory, D. P. 1988. Faecal clostridia and indicator bacteria levels in an eutrophic
impoundment. Water SA 14:115-117.
Schardinger, F. 1892. Ueber das Vorkommen Gährung erregender Spaltpilze im
Trinkwasser und ihre Bedeutung fiir die hygienische Beurteilung desselben.
Wien. Klin. Wochenschr. 5:421-423.
Setlow, P. 1988. Small, acid-soluble spore proteins of Bacillus species: structure,
synthesis, genetics, function, and their degradation. Annu. Rev. Microbiol.
42:319-338.
Sheppard, D. A. E. 1995. John Snow: Anesthetist to a Queen and Epidemiologist to
a Nation: A Bibliography York Point: Cornwell Edward Island, Canada.
Sherman, J. M. 1937. The Streptococci. Bacteriological Reviews 1:3.
Signor, R. S., D. J. Rosen, and N. J. B. Ashbolt, J.E. 2005. Quantifying the impact
of runoff events on microbiological contaminant concentrations entering surface
drinking source waters. J. Wat. Health 03:453-468.
Smeets, P. W. M. H., Y. J. Dullemont, and G. J. Medema. 2005. Presented at the
17th IOA conference, Strasbourg, France.
Smeets, P. M. 2008. Stochastic modelling of drinking water treatment in
quantitative microbial risk assessment. Technical University Delft, Delft NL.
Smith, T. 1895. Ueber den Nachweis des Bacillus coli communis im Wasser.
Zentralblatt für Bakteriologie 18.
Sorensen, D. L., S. G. Eberl, and R. A. Dicksa. 1989. Clostridium perfringens as a
point source indicator in non-point polluted streams. Water Res. 23:191-197.
Strange, R. E., and J. R. Hunter. 1969. Outgrowth and synthesis of
macromolecules. In G. W. Gould and A. Hurst (ed.), The Bacterial Spore.
Academic Press, London and New York.
States, S. 1997. Protozoa in River Water: Sources, Occurrence and Treatment. J.
Am. Water Works Assoc. 89(9):74-83.
Stetler, R. E., S. C. Waltrip, and C. J. Hurst. 1992. Virus removal and recovery in
the drinking water treatment train. Water Res. 26:727-731.
Swertfeger, J., D. H. Metz, J. DeMarco, A. Braghetta, and J. G. Jacangelo. 1997.
Effect of filter media on cyst and oocyst removal. J. Am. Water Works Assoc.
91(9):90-100.
Taylor, E. W. 1949. Thresh, Beale & Suckling’s The examination of waters and
water supplies, 6th ed. Churchill, London, UK.
Teunis, P. F. M., G. J. Medema, L. Kruidenier, and A. H. Havelaar. 1997.
Assessment of the risk of infection by Cryptosporidium and Giarda in drinking
water from a surface water source. Water Res. 31:1333-1346.
___________________________________________________________________
- 49 -
Chapter 1
.
Teunis, P. F. M., E. G. Evers, A. H. Havelaar, and G. J. Medema. 1996. Presented
at the American Water Quality Technology Conference, Boston, November 1996.
Theunissen, J. J. H., P. J. Nobel, R. van de Heide, H. A. M. de Bruin, D. van
Veenendaal, W. J. Lodder, J. F. Schijven, G. J. Medema, and D. van der Kooij.
1998. Enterovius concentraties bij de innamepunten van oppervlaktewater voor
de bereiding van drinkwater. RIVM 289202013.
Thiercelin, M. E. 1899. Sur un diploque saprophyte de líntestin susceptible de
devenir pathogene. Comptes rendus Société de Biologie 5:269-271.
Tisa, L. S., T. Koshikawa, and P. Gerhardt. 1982. Wet and dry bacterial spore
densities determined by buoyant sedimentation. Appl. Environ. Microbiol.
43:1307-1310.
USEPA. 2006. LT2ESWTR Long Term Second Enhanced Surface Water Treatment
Rule; Final Rule. USEPA.
van Breemen, L. C. W. A., and J. M. J. Waals. 1998. Storage of surface water in the
Netherlands: challenges of the future. Water Supply 16:375-381.
Van der Kooij, D. 2002. Presented at the 54e vakantiecursus in
drinkwatervoorziening
&
21e
Vakantiecursus
in
Riolering
en
Afvalwaterbehandeling, Delft NL.
Van der Kooij, D., Y. C. Drost, W. A. M. Hijnen, J. Willemsen-Zwaagstra, P. J.
Nobel, and J. A. Schellart. 1995. Multiple barriers against micro-organisms in
water treatment and distribution in the Netherlands. Wat. Supply 13:13-23.
Visser, A., W. A. M. Hijnen, Y. Dullemont, and G. J. Medema. 2004. Langzame
zandfilters als effectieve barriére voor micro-organismen. H2O 37(13):28-30.
von Huben, H. 1991. Surface Water: the new rules. American Water Works
Association, Denver CO US.
Walker, J. R., A. J. Gnanam, A. L. Blinkova, M. J. Hermandson, M. A. Karymov,
Y. L. Lyubchenko, P. R. Graves, T. A. Haystead, and K. D. Linse. 2007.
Clostridium taeniosporum spore ribbon-like appendage structure, composition and
genes. Mol. Microbiol. 63:629-643.
Watanabe, K., and S. Takesue. 1976. Colloid titration for determining the surface
charge of bacterial spores. J. Gen. Microbiol. 96:221-223.
Weelden, V., and Mingelen. 1868. Commissie tot onderzoek van drinkwater, 1868.
Rapport aan de Koning.
Welch, W. H., and G. H. F. Nuttall. 1892. A gas-producing Bacillus (Bacillus
aerogenes capsulatus, N.S.) capable of rapid development in the blood vessels after
death. Bulletin of the Johns Hopkins Hospital, July, 1892.
Wetsteyn, F. J. 2005. Analyse microbiologische veiligheid drinkwater. Artikelcode:
5318. VROM-Inspectorate, Haarlem NL.
White, G. C. 1999. Handbook of chlorination and alternative disinfectants. John
Wiley & Sons, inc., New York.
WHO. 2004. Guidelines for Drinking Water Quality, third edition. World Health
Organization, Geneva, SW.
Wiencek, K. M., N. A. Klapes, and P. M. Foegeding. 1990. Hydrophobicity of
Bacillus and Clostridium Spores. Appl. Environ. Microbiol. 56:2600-2605.
___________________________________________________________________
- 50 -
Chapter1
Wijmer, S. 1992. Water om te drinken. Vereniging van Exploitanten van
Waterbedrijven in Nederland VEWIN, Rijswijk NL.
Willis, A. T. 1956. Anaerobes as an index of faecal pollution in water. J. appl. Bact.
19:105-107.
Willis, A. T. 1957. Anaerobic bacilli in a treated water supply. J. appl. Bact. 20:6164.
Wilson, W. J., and E. M. M. Blair. 1925. Correlation of the sulphite reduction test
with other tests in the bacteriological examination of water. Jour. of Hyg.
XXIV:111-119.
Wilson, W. J., and E. M. M. Blair. 1931. Official Circular 96. British Waterworks
Association
Windle Taylor, E., N. P. Burman, and C. W. Oliver. 1955. Membrane filtration
technique applied to the routine bacteriological examination of water. Journal of
the institution of Water Engineers 9:248-263.
Wubbels, G.H. 1996. Inventarisatie van Thermofiele Campylobacter bacterien No.
Code 945. Haren: Gemeentelijk Waterbedrijf Groningen.
Yolton, D. P., R. N. Huettel, D. K. Simpson, and L. J. Rode. 1972. Isolation and
partial chemical characterization of the spore appendages of Clostridium
taeniosporum. J. Bacteriol. 109:881-885.
Yolton, D. P., L. Pope, M. G. Williams, and L. J. Rode. 1968. Further Electron
Microscope Characterization of Spore Appendages of Clostridium bifermentans. J.
Bacteriol. 95:231-238.
___________________________________________________________________
- 51 -
ABBREVIATIONS AND
CALCULATIONS
Microbiology:
ARS = aerobic spore formers
cfu = colony forming units
Coli37 = coliforms
Coli44= thermotolerant coliforms
FC = faecal coliforms
FRNA = F-specific-RNA phages
FS = faecal streptococci
IP = index pathogen
PI = process indicator
SSRC = spores of sulphite-reducing clostridia
Microbial risk management and statistics:
%PS = percentage of positive samples
AQL = accepted quality level
AVG = arithmetic average
QMRA = quantitative microbial risk assessment
DE = decimal elimination
DEC = decimal elimination capacity
HACCP = Hazard Analysis Critical Control Point
LVS = large volume sampling
MEC = microbial elimination capacity
MTC = maximum tolerable concentration
N = number of samples
Nc = critical mean concentration for an accepted quality level (AQL)
nd = not determined
ni = no information
mf/MF = membrane filtration
SD = standard deviation
USEPA = United States Environmental Protection Agency
v = sample volume (l)
WHO = World Health Organization
WSP = Water Safety Plans
Water treatment:
Ct = disinfectant concentration (mg/l) and contact time t (min)
CB = collimated beam
CFD = computational fluid dynamics
CFR = coagulation and floc-removal
Cl2 = chlorination
___________________________________________________________________
- 53 -
DF = dune filtration
DPB = disinfection byproduct
GAC = granular activated carbon
IR = impoundment reservoir
OCB = open collection basin
O3 = ozonation
PAC = powdered activated carbon
PD = post-disinfection
REF = reduction equivalent fluence
RGF = rapid granular filtration
SF = softening
SR = small reservoir
SSF = slow sand filtration
unit processes = the individual steps in a total treatment
CALCULATIONS OF DECIMAL ELIMINATION CAPACITY
Elimination of faecal indicator bacteria under full-scale conditions:
The decimal elimination of micro-organisms in a water treatment process is the
difference between the log-tranferred concentrations in the water before and after
a process paired by date:
DE = log10 Cin − log10 Cout = log10
Cin
Cout
(1)
The Decimal Elimination Capcity (DEC) calculated from the routinely collected
microbial data collected over a period of three years (Chapter 2), is the average
elimination described with the median or 50-percentile value of DE when >20% of
the Cin and Cout values paired by date allowed DE calculation. The 10-percentile
DE value was used to demonstrate the lower part of the distribution of these
values. When in majority (>80%) of the samples after the process (Cout) no microorganisms were detected (zero count), the ratio estimation method was used where
the zero counts were included by using the weigthed average concentrations
calculated from the total detected micro-organisms in the total examined sample
volume, including the samples with no detection
DEC = log10 C in
C out
with C =
∑ cfu
Nv
(2)
The arithmetic mean DE with the standard deviation SD was used in the large
volume sampling study in Chapter 4 to calculate DEC from a total of ten
observations of Cin and Cout paired by date, when ≥50% of the Cin and Cout values
paired by date allowed DE calculation. When in <50% of the observations DE
could not be calculated due zero counts in the water after the process, the ratio
estimation method was used.
___________________________________________________________________
- 54 -
Elimination of dosed micro-organisms in disinfection and filtration processes:
DEC of disinfection processes assessed with challenge tests was calculated from
the inactivation rate constant k of the inactivation kinetics and the applied
disinfection dose described by the first order disinfection model of Chick-Watson:
DEC = log10
Co
Ct
= kCOzone t or log10
Co
Ct
= kFUV t (3)
where Cozone is the ozone concentration (mg/l; Chapter 5) and FUV is the UV
fluence (mW/cm2) and t is the contact time (minutes; Chapter 6). Result of this
study showed that the DEC assessed in batch tests with pre-cultured organisms
over-estimate the DEC of full-scale disinfection processes.
DEC of filtration processes assessed with challenge tests with short periods of
dosing and a distinct breakthrough of micro-organisms in the filtrate during
dosing DEC is calculated by
DEC = log10 C in
Cout ,max
(4)
where Cout,max is the maximum concentration observed in the filtrate (Chapter 7).
In Chapter 8 with a challenge test during a prolonged period of time (98 days)
and in Chapter 9 from a challenge infiltration test with soil columns the DEC was
calculated from the mass balance of total number of dosed organism Md and the
total number of organisms observed in the effluent Me during the sampling
DEC = log10
Md
Me
(5)
Results of the current study show that DEC calculated from column filtration
studies on slow sand filtration and surface water infiltration must be regarded as
relative DEC values and can not be translated to the DEC of full-scale systems.
The DEC derived from a filtration process operated in a pilot plant as a dummy
of the full-scale system, showed more similarity with the DEC calculated from
elimination of natural occurring micro-organisms in the full-scale filtration
process.
___________________________________________________________________
- 55 -
Chapter 2
Indicator bacteria concentrations in
water treatment and assessment of
elimination capacity•
W.A.M. Hijnen1, Y.C. Drost1, G.J. Medema1, D. van der Kooij1 and A.H. Havelaar2
1
2
•
KWR Watercycle Research Institute, PO box 1072 , 3430BB Nieuwegein, NL
University Utrecht, PO box 80175, 3508TD Utrecht, NL
Parts of this chapter are based on:
Hijnen, W.A.M. et al., 1997/1998. The removal of indicator bacteria at the eight
surface water facilities presented in Chapter 1 (in Dutch). Reports of the joint
Research Programme of the Dutch Water Companies.
Hijnen, W.A.M., Van der Speld, W.M.H., Houtepen, F.A.P. and Van der Kooij, D.
(1997). Spores of sulphite-reducing clostridia: a surrogate parameter for assessing
the effects of water treatment on protozoan oocysts? In: Proc. 1997 International
Symposium on waterborne Cryptosporidium, ed. C.R. Fricker, J.L. Clancy and P.A.
Rochelle, AWWA Denver US, 1997.
Van der Kooij, D., Drost, Y.C., Hijnen, W.A.M., Willemsen-Zwaagstra, J., Nobel,
P.J. and Schellart, J.A. (1995). Multiple barriers against micro-organisms in water
treatment and distribution in the Netherlands. Wat. Supply, 13(2): 13-23
___________________________________________________________________
- 57 -
Chapter 2
.
ABSTRACT
The use of thermotolerant coliforms (Coli44) and spores of sulphitereducing clostridia (SSRC) as process indicators to assess the elimination
capacity of water treatment and unit processes for pathogenic microorganisms susceptible and resistant to disinfection was explored. Historical
data sets of routine monitoring programs revealed that both indicators
were detected in the finished water after treatment in variable frequencies.
Coli44 was detected incidentally, whereas SSRC was observed after every
treatment in percentages of samples ranging from 0.8 up to 23%. With data
collected after unit processes efficacy of these unit processes was calculated
to eliminate both process indicators. The percentage of positive samples,
however, decreased rapidly during treatment. This affects the accuracy of
the calculated decimal elimination capacity (DEC). Decreasing the detection
limit of the standard membrane filtration method by filtering larger sample
volumes enabled more accurate assessment of DEC. The required detection
limit of the analytical method, however, to evaluate the required DEC of a
total water treatment for pathogenic micro-organisms is ≤1 organism per
100 l which requires further research for new methods.
INTRODUCTION
Chapter one describes developments in microbiologically safe drinking
water over the years. An important development in management of
microbiological quality of drinking water in the last decade is the
introduction of health-based targets. In the revised Dutch drinking water
decree (Anonymous, 2001) a mandatory quantitative microbial risk
assessment (QMRA) is implemented for drinking water produced from
surface water to demonstrate compliance with an annual accepted infection
risk of 10-4 for pathogenic micro-organisms. An important step in this
QMRA is the determination of the removal and inactivation of pathogenic
micro-organisms. The objective of the current study is to develop and apply
methods for the assessment of the elimination capacity (removal and
inactivation) of water treatment processes. Direct measurement of
elimination of pathogens is not feasible. The general hypothesis is that the
elimination of the faecal indicator bacterium E. coli and spores of sulphitereducing clostridia (SSRC) can be used for assessment of elimination of
susceptible and resistant pathogens, respectively. First step in the study
was to explore the use of the standard microbiological methods in routine
water quality monitoring for the assessment of the elimination capacity of
___________________________________________________________________
- 58 -
Chapter 2
treatment processes. Routinely collected data of concentrations of indicator
bacteria in full-scale water treatment were evaluated. Additionally, the use
of an analytical method with lower detection limit by examining larger
volumes with the standard membrane filtration technique was investigated
in a pilot plant study.
MATERIALS AND METHODS
Routinely collected microbiological data. Analysis data on
indicator bacteria in water monitored over periods of three years by the
Water Companies with standard methods were collected and evaluated.
The data were monitored in the eight full-scale systems mentioned in
Chapter 1 (Table 5). Total coliforms (Coli37) and thermotolerant coliforms
(Coli44), faecal streptococci (FS) and spores of sulphite-reducing clostridia
(SSRC) were monitored mandatory in the source waters and the drinking
water with prescribed methods (Anonymous, 1985a,b,c). Each Water
company also measured all or selected indicators at some selected
sampling points in treatment.
Process conditions. The general operational and design conditions
of the evaluated processes presented in Table 5 of Chapter 1 were gathered
during this survey: coagulant and disinfectant dosages, pH, turbidity,
temperature, contact time, flow rate. For disinfection processes the Ctvalues (mg.l-1).min calculated from the residual disinfectant concentration
(mg.l-1) and the average contact time (min).
Pilot plant study Zevenbergen. Water treatment plant Zevenbergen
of Water Supply Company Brabant Water Ltd. uses river water after
storage in open reservoirs for the production of drinking water. Treatment
includes coagulation/flotation, chlorination, dual media filtration and GAC
filtration. The use of ozone as an alternative for chlorine was studied on
pilot plant scale (Figure 1). The specifications of the treatment stages are
described elsewhere (Hijnen et al., 1997). In this study the detection limit of
the standard SSRC analysis was decreased by examining larger sample
volumes to assess concentrations of SSRC in the last stages of treatment.
Determination of the decimal elimination capacity (DEC). The
collected historical data on indicator bacteria in the water from the eight
surface water treatment locations showed that concentrations in the source
waters were high. Percentage of detection in the examined samples was
100%. After the first applied processes, however, percentages of positive
samples (%PS) decreased and after the subsequent processes in an
increasing number of the samples indicator bacteria were absent (zero
___________________________________________________________________
- 59 -
Chapter 2
.
count). Therefore a general strategy was introduced in order to calculate
the elimination capacity of processes and a total treatment for these microorganisms.
For the first processes with high percentages of positive samples the actual
decimal elimination or DE value is calculated from the concentration of
indicator bacteria in the influent (Cin) and effluent (Cout) of the process
measured at the same time (paired data) with the following equation
DE = Log 10 C in − Log 10 C out = Log 10
C in
C out
1
The Decimal Elimination Capacity (DEC) is the 50-percentile or median
values from the calculated DE values of a process.
When in the outlet or in both the inlet and the outlet of a process, the
majority of the samples >80% were negative for the indicator bacteria, DEC
is determined with the average concentration. In order to include the
information of the “negative” samples (zero counts) in these calculations all
individual samples are considered as part of a large volume examined during
a specified period of time. The average concentration is calculated with the
following equation
C=
∑ cfu
N *v
2
_
where C is the average concentration in the inlet or the outlet of the process,
∑cfu the total number of colony forming units in N samples with volume v
___________________________________________________________________
- 60 -
Chapter 2
(l). These average concentrations are used to calculate the Decimal
Elimination Capacity with
DEC = Log 10
C in
C out
3
where Cin and Cout are the arithmetic average concentrations in the water.
This method, where the observations are used as unpaired data, is called
the ratio-estimation method (Drost et al., 1997). For statistical analysis of the
data Excel (Microsoft) was used.
Acceptable quality level (AQL) of drinking water. On the basis of
the standards for indicator bacteria in drinking water of the Dutch
Drinking Water Decree (Anonymous, 1984; MTC-values and sampling
frequencies) critical mean concentrations Nc.l-1 (95% confidence level) can
be calculated that will generate annually 0, 1 and 5% positive samples of
indicator bacteria in the drinking water, assuming Poisson distribution
(Van der Kooij et al., 1995). These Nc values for an acceptable quality level
of 1% of samples non complying with the standard for thermotolerant
coliforms (Coli44 incl. E. coli) daily monitored in 300 ml is 0.018. l-1. For
SSRC and FS weekly monitored in 100 ml the Nc is 0.035.l-1.
RESULTS
Faecal indicators in surface water treatments. Concentrations of
indicator bacteria in the source waters of the eight treatment facilities
showed a high variation, both between the locations but also at the
individual locations (Figure 2; arranged from high to low concentrations).
Coliform concentrations were usually higher than the concentrations of FS
and SSRC. At facilities 1 and 2, however, SSRC concentrations in the source
water were higher than coliform concentrations.
At four of the five locations using main disinfection with either chlorine or
ozonation (Table 1) indicator bacteria were monitored after this process.
These results demonstrate that main disinfection processes are no absolute
barrier against indicator bacteria. In percentages of the total samples
ranging from 0 – 33.6% coliforms and faecal streptococci were detected
after disinfection.
Despite higher Ct values of chlorination and a significantly lower number
of samples at location 4, the percentage of positive samples for coliforms
(Coli37 and Coli44) after this chlorination was higher than the percentage
of positive samples with these indicator bacteria after chlorination at
location 2. The difference between the performance of ozonation at location
___________________________________________________________________
- 61 -
Chapter 2
.
3 and 7 with approximately the same Ct values was even larger which
indicates that the efficacy of these full-scale disinfection processes may vary
considerably. A second important but expected result was the higher
resistance of SSRC. In 30.4 up to 65.2% of the samples these anaerobic
spores were detected after disinfection.
100000
10000
Concentration (n/l)
100000
Coli37
Coli44
FS
SSRC
10000
1000
1000
100
100
10
10
1
1
0.1
0.1
5
6
4
3
8
2
Treatment facility
7
1
5
4
3
2
6
1
Treatment facility
8
7
Figure 2 Concentrations (arithmetic mean) of Coli37, Coli44, FS and SSRC in the
source waters of eight treatment facilities calculated from the routine monitoring
programs (n = 150-10,000) over a period of 3 year and arranged in decreasing
order of Coli44 and SSRC concentrations (error bar 90-percentile-values
interpolated from the collected data)
Table 1. Percentage (%) of positive samples (number of samples) for faecal
indicator bacteria in the water after main disinfection at two locations with
chlorine and two locations with ozonation with the corresponding Ct-values
(mg/l.min.)
Location 4
Cl2 (15-100)a
2.4 (82)b
2.4 (83)
Location 2
Cl2 (9-12)
0.4 (2013)
0.1 (2013)
location 3
O3 (2)
1.8 (1520)
0.6 (1519)
Location 7
O3 (1.7-2.2)
33.6 (411)
12.2 (582)
FS
0.0 (82)
1.2 (415)
1.3 (310)
8.3 (72)
SSRC
32.9 (82)
-
30.4 (312)
65.2 (135)
Organisms
Coli37
Coli44
a
range of applied Ct values; b number of samples
The data of the finished waters revealed that 100% compliance with the
microbiological standard (Water Decree, 1984) was not achieved during the
evaluated period of three years (Table 2). The first four plants used post___________________________________________________________________
- 62 -
Chapter 2
disinfection and at the other plants post-disinfection was not used. A high
level of non compliance for Coli37 (data not presented) in the finished
water at plants with main disinfection (1, 2, 3, 4 and 7) was observed. This
was caused by regrowth of these bacteria in the filtration processes GAC or
SSF applied at the end of the treatment facilities. During summer periods
Cout of this indicator was regularly higher than Cin (negative DE value).
Consequently, this parameter is unsuitable as a process indicator for
elimination assessment.
Table 2. The average concentrations of Coli44 and SSRC in the finished waters
with or without post-disinfection of the eight treatment facilities, the percentage of
positive samples (%PS) and DEC values
Facility
(DIS)a
1 (O3)
2 (Cl2)
3 (O3)
4 (Cl2)
5
6
7 (O3)
8
Coli44 (Nc = 0.018 cfu l-1)
%PS
DEC
Cout (n/l)
0.0009b
0.09
4.1
b
0.05
0.02
2.7
0.160c;<0.0008b
1.6; 0.0
3.5; >5.3
<0.013c; <0.03b
0.0; 0.0 >4.3; >3.9
0.090
1.2
5.0
0.029
2.0
4.3
0.002
0.1
4.2
0.013
1.0
3.5
SSRC (Nc = 0.035 cfu l-1)
Cout (n/l)
%PS DEC
0.051b
2.6
3.0
b
0.32
1.9
2.8
1.375b
5.6
2.1
0.117b
1.0
3.2
0.037
1.9
4.9
0.320
23
3.0
0.008
0.8
3.4
0.014
1.4
3.4
DIS = main disinfection; Cl2/O3 = chlorine/ozone; b data after postdisinfection with chlorine dioxide; c data before post-disinfection
a
The percentage of non compliance with the standards for Coli44 (Table 2)
and faecal streptococci (data not presented) was higher after treatment
plants without a main disinfection, except for location 3. Concentrations of
SSRC in the finished waters (only measured after post disinfection when
applied) were usually higher than Coli44 concentrations again showing the
higher resistance of these indicator bacteria with a higher probability of
breakthrough. The finished water concentrations of Coli44 after five
facilities were below the Nc for Coli44 of 0.018 CFU.ml-1 for an AQL of 1%.
For SSRC, however, finished water concentrations were usually above the
Nc of this parameter of 0.035 CFU.l-1 and percentages of positive samples
were higher than 1% up to 23% at location 6.
Decimal Elimination Capacity (DEC). Using these data DEC of the
facilities was calculated using the arithmetic mean concentrations in source
___________________________________________________________________
- 63 -
Chapter 2
.
and finished water (zero counts are include as indicated before). DEC of the
facilities for these indicator bacteria ranged from 2 to 5 log (Table 2).
Application of main disinfection not necessarily revealed a higher DEC
value. The highest DEC value was determined for location 5 without either
main or post-disinfection. In agreement with the former conclusion on
higher breakthrough of SSRC, DEC of the facilities for SSRC was lower
than DEC for the more susceptible Coli44.
Decimal Elimination Capacity of unit processes. From the three
year routine data of indicator bacteria samples monitored during treatment
of some facilities, indicator bacteria concentrations in the water after the
first processes was high enough to assess the actual decimal elimination
(DE) with variation presented by the 50- (median) and 10-percentiles
assessed by interpolation (Table 3).
Table 3. DEC and median and the 10-percentile value of DE of some processes at a
number facilities for thermotolerant coliforms (Coli44) and SSRC based on three
year routinely collected data
Process
DEC
CFR
-a
CFR
0.7
RGF
1.1
RGF
RGF
0.9
O3
O3
O3
1.6
a no data collected
Coli44
SSRC
DE
Median
0.8
1.2
0.7
1.0
DE
Median
0.9
0.6
1.3
0.7
0.8
0.5
P10
0.5
0.8
0.2
0.5
DEC
1.2
0.6
1.4
1.0
0.4
P10
0.1
0.1
0.7
0.0
0.3
0.0
These data demonstrate that in 10% of the time the actual elimination in the
processes for faecal indicator bacteria is 0.3 – 0.8 log lower than the average
elimination (DE). The impact of not including the zero counts in the DE
calculations on the average elimination capacity was illustrated from these
data. DEC values with zero counts included in the calculation were in 5 out
of 9 processes higher than the median DE value (Table 3). For one process
DEC was similar to the median and at 3 processes DEC was lower than the
median.
___________________________________________________________________
- 64 -
Chapter 2
120
1 Liter (n=46)
10 Liter (n=45)
100
80
60
40
20
0
Source
water
CFR
RGF
Concentration (n/l)
% positive samples
Lower detection limit of the standard analytical method. A simple
option to decrease the detection limit of the standard analytical method is
by increasing the sample volume as demonstrated for SSRC in a pilot plant
study (Hijnen et al., 1997). The removal of these spores by
coagulation/flotation followed by dual media filtration and granular
activated carbon filtration (GAC) was monitored with 1 and 10 liter
samples filtered through 47 mm (0.45 μm) membranes. By increasing the
sample volume the percentage of positive samples after dual media
filtration increased from 22 to 84% and after the subsequent GAC-filtration
(no 1 liter samples examined) the %PS was 58% (Figure 3).
1000
1 Liter
10 Liter
100
10
1
0.1
Source
water
GAC
CFR
RGF
GAC
Figure 3. Percentage of positive samples (left) and (right) average concentration of
SSRC at the different stages of the pilot plant Zevenbergen (GAC = granular
activated carbon filtration) (error bar = SD) (data from Hijnen et al., 1997)
The cumulative distribution of the results of SSRC in 1 and 10 liter samples
is depicted in Figure 4.
100
1 Liter
10 Liter
10
1
0
25
50
% samples
75
100
Figure 4. Cumulative distribution of SSRC concentration in GAC filtrate
monitored in 1 and 10 liter samples (from Hijnen et al., 1997)
___________________________________________________________________
- 65 -
Chapter 2
.
DISCUSSION
DEC assessment of a treatment or process for indicator bacteria.
Quantification of the elimination capacity of water treatment facilities for
faecal indicator requires presence of these organisms in both the source and
finished waters. At all facilities these micro-organisms were present in the
source water. On the basis of the source water concentrations and the
detection limits of the standard analytical methods of 1 CFU per 100 or 300
ml (10 or 3.3 CFU per l), the maximum assessable DEC of a treatment for
coliforms (total Coli37 and thermotolerant coliforms Coli44) is
approximately <1 – 3 log depending on source water quality. With the data
of faecal streptococci and SSRC (incl. C. perfringens) this maximum
assessable DEC is <1 – 2.4 log. The required DEC values for viruses,
Cryptosporidium and Campylobacter for an annual infection risk level of 10-4
estimated from source water concentrations was 6 – 8 log (Chapter 1).
Consequently, with the routine data of indicator bacteria monitoring with
the standard membrane filtration method the required DEC for index
pathogens is not assessable.
The evaluation of historical routinely collected data on indicator bacteria in
the finished water of drinking water facilities revealed that in a small
number of samples these indicators were observed. In a first attempt to
describe elimination capacity of water treatment processes with the routinely
collected data on faecal indicators Drost et al. (1994) compared a number of
possible options focussed on the question how to include the zero counts in
data sets as valuable observations in the calculations of elimination. Zero
counts were replaced by the value of the detection limit (A), by the average
concentration (B) or by the actual concentration plus 1 (C). Using real data
sets and simulated data sets, DEC derived from these data manipulations
were compared with DEC assessed with the ratio-estimation method (E)
described before and with DE values using only positive samples (D;
excluding zero counts). Method assessment was done by calculation of the
residual values (RES) of the actual concentration in the outlet and the
concentration calculated from the linear regression coefficient a of Cout = a.Cin:
∧
RES = C out − C out
Conclusion and recommendation of these tests was to use paired data of Cin
and Cout if >20% of these pairs were both positive. Otherwise the ratioestimation method E was recommended as second best followed by the
method where zero counts are replaced by the average concentration of the
actual data set.
___________________________________________________________________
- 66 -
Chapter 2
Calculation of the arithmetic average concentrations of faecal indicator
bacteria in source and finished water (zero counts included), enables
assessment of higher DEC values of 2 – 5 log as demonstrated in Table 2.
These DEC values are lower than the required DEC-values of 6 – 8 log for
viruses, Campylobacter, Cryptosporidium and Giardia for an annual 10-4 infection
risk in the produced drinking water.
A second study to explore calculations of micro-organism removal in water
treatment was conducted with the (statistical) methods B and D described
above and additional mathematical methods (Evers and Groennou, 1999).
Using SSRC data from full-scale plants the study concluded that a binominal
mathematical method with fixed p value was best followed by statistical
method B (zero = average). Nonetheless, the authors concluded that the
accuracy of the determined elimination capacity depends on the quality of the
collected data. From datasets with high level of zero counts no high accuracy
can be expected and it was recommended to reduce the number of these zero
counts by analyzing larger sample volumes.
Possibility to lower the detection limit. The data of the pilot plant
study at Zevenbergen demonstrate that by simply increasing the sample
volume with a factor of 10 a higher percentage of samples with SSRC can
be obtained. This results in more accurate quantitative data on the DEC of
processes. In the presented pilot plant study without a main disinfection
step, the standard sample volume of 100 ml was increased with a factor of
100 (detection limit of 0.1 cfu l-1). For an accurate assessment of
concentrations of Coli44 and SSRC in the finished water of most treatment
facilities a detection limit of 0.01 cfu l-1 is required. Volumes of 100 liter or
more, however, can not be analyzed with the standard membrane filtration
technique and require an adapted method. The use of large volume
sampling under full-scale conditions have been explored by others (Goyal
et al. 1980; Payment et al. 1989) but these techniques are based on
membrane adsorption/elution techniques with relative complex
procedures compared to the standard microbial methods and have low
recoveries.
Variability of DEC of unit processes. Another omission in the
current data sets is the lack of information on variation in elimination.
Information about the variability in removal is of interest because
waterborne outbreaks of diarrhoea caused by protozoan parasites have
been related to peak concentrations in the source water but also by
inadequacies in water treatment (Dykes et al., 1980; Badenoch, 1990; Craun,
1990; Richardson et al., 1991). Evaluation of historical data sets can help to
identify such inadequacies as demonstrated for the ozonation process at
___________________________________________________________________
- 67 -
Chapter 2
.
facility 7 (Table 1). DEC of this ozonation operated at Ct values of 1.7-2.2
mg/l.min. was 1 log for the ozone susceptible Coli44. On the basis of
dose/response data for E. coli known in literature (Finch et al., 1988), a
much higher DEC was expected.
Beside obvious inadequacies in treatment, normal operated processes in a
treatment exhibit a variation in the elimination of micro-organisms.
LeChevallier and Norton (1992) for instance showed that the removal of
particles of > 5 μm (size of Cryptosporidium oocysts) by individual unit
filters in the filtration stage of a treatment may vary by as much as 1.000fold. These variations will influence the overall efficiency of a treatment
stage. Variation in the removal of (oo)cysts of Cryptosporidium and Giardia
by a conventional treatment (coagulation/ floc removal plus filtration) was
observed by Hashimoto et al. (2001). DEC for Cryptosporidium and Giardia
ranged from 2.0 – 3.2 and 1.7 – 3.1 log, respectively.
The variability of DEC of unit processes for Cryptosporidium has also been
demonstrated by spiking tests on pilot plant scale or in laboratory
experiments. Emelko (2001) intensively studied the removal of C. parvum
oocysts by filtration processes under different filtration conditions. DEC
assessed under stable operation, ripening, sub-optimal coagulant dosing,
early and late breakthrough ranged between 5.5 and <0.5 log. Also with
spiking experiments on laboratory scale (dose/response data with a
continuous flow system) a high variability of the efficiency of ozone
disinfection for C. parvum in natural waters was demonstrated
(Oppenheimer et al., 2000). The average Chick/Watson inactivation
constant at 10oC was 0.21 and a range of 0.08 – 0.46 l/mg.min.
Variation in elimination of micro-organisms is caused by the multiple
variables involved in the mechanisms responsible for the elimination
(inactivation and removal) in the different treatment processes (see Chapter
1). Due to these observations and considerations it is of importance for risk
management to integrate variability in both source water concentrations
and elimination in the uncertainty calculations of the infection risk assessed
with QMRA. Additional statistical tools to integrate uncertainty
calculations in QMRA have been described (Teunis et al., 1997; Evers and
Groennou, 1999; Haas et al., 1999; Medema et al., 2003). Moreover, in the
framework of Water Safety Plans (WSP, WHO 2004) it is of importance to
determine the causes of these fluctuations since that can lead to measures
in process design or operation to minimize infection risks.
___________________________________________________________________
- 68 -
Chapter 2
CONCLUSIONS
From the results of the studies presented in this chapter it can be concluded
that monitoring for faecal indicator bacteria in full-scale water treatment
facilities for the production of drinking water can be used to quantify the
removal and inactivation (elimination) capacity (DEC) of unit processes
and a complete train of processes. Evaluation of the historical collected
monitoring data with standard analytical methods yielded a first
impression of DEC of processes and treatment facilities. It also showed the
limitations because of the high percentage of zero counts (no indicators
detected) in the last stages of treatment. Methods have been used and
evaluated to determine DEC from these datasets with a high level of zero
counts, but were judged as being inaccurate. Beside this inaccuracy
assessment of variability of DEC values required to calculate the
uncertainty of infection risk levels with QMRA was also not possible. The
first impression on elimination of indicator bacteria in water treatment
showed lower DEC values than the DEC values required for index
pathogens. This emphasizes the need for improvement of accuracy of DEC
assessment of a treatment for both process indicators which requires an
analytical method with lower detection limit.
The challenge is to develop a rapid, simple and reliable isolation method
for the assessment of concentrations of indicator bacteria in volumes of
≥100 litre treated water. First results presented in this study indicated that
decreasing the detection limit of the standard membrane filtration method
by increasing sample volumes is potentially an easy and successful option.
The data collected on the elimination of Coli44 and SSRC by unit processes
showed not only variation in DEC of processes over time but also a
significant variation in DEC of similar processes operated at different
locations. This emphasizes the need for quantitative information on
elimination of micro-organisms by water treatment processes for QMRA as
site specific and actual as possible.
References
Anonymous. 2001. Besluit van 9 januari 2001 tot wijziging van het
waterleidingbesluit in verband met de richtlijn betreffende de kwaliteit van voor
menselijke consumptie bestemd water, p. 1-53, vol. 31. Staatsblad van het
Koninkrijk der Nederlanden.
Anonymous. 1984. Waterleidingbesluit Staatsblad van het Koninkrijk der
Nederlanden, 220: 1-35.
___________________________________________________________________
- 69 -
Chapter 2
.
Anonymous. 1985a,b,c. NEN6553, NEN6564; NEN6567
Badenoch, J. 1990. Cryptosporidium in water supplies, London.
Craun, G. F. 1990. Waterborne Giardiasis. In E. A. Meyer (ed.), Human Parasitic
Diseases, vol. 3, p. 267-293. Elsevier Science Publ. Amsterdam NL..
Drost, Y. C., J. T. Groennou, W. A. M. Hijnen, and D. Van der Kooij. 1997.
Statistische methoden ter bepaling van de belasting en eliminatiecapaciteit van
zuiveringsprocessesn m.b.t. micro-organismen. Kiwa Water Research SWI
97.177, Nieuwegein NL.
Dykes, A. C., D. D. Juranek, R. A. Lorenz, S. Sinclair, W. Jakubowski, and R. B.
Davies. 1980. Municipal waterborne Giardiasis: an epidemiological
investigation. Ann. Int. Med. 92:165-170.
Emelko, M. B. 2001. Removal of Cryptosporidium parvum by granular media
filtration. University of Waterloo, Ontario, Canada.
Evers, E. G., and J. T. Groennou. 1999. Berekening van de verwijdering van microorganismen bij de bereiding van drinkwater. RIVM 734301016, Bilthoven NL.
Finch, G. R., D. W. Smith, and M. E. Stiles. 1988. Dose-response of Escherichia coli
in ozone demand-free phosphate buffer. Water Res. 22:1563-1570.
Goyal, S. M., H. Hanssen and C.P. Gerba. 1980. Simple method for the
concentration of influenza virus from allantoic fluid on microporous filters.
Appl. Environ. Microbiol. 39:500-4.
Haas, C. N., J. B. Rose, and C. P. Gerba. 1999. Quantitative microbial risk
assessment. John Wiley & Sons, New York, USA.
Hashimoto, A., T. Hirata, and S. Kunikane. 2001. Occurrence of Cryptosporidium
oocysts and Giardia cysts in a conventional water purification plant Wat. Sci.
Techn. 43(12):89-92.
Hijnen, W. A. M., W. M. H. van der Speld, F. A. P. Houtepen, and D. van der
Kooij. 1997. Presented at the International Symposium on Waterborne
Cryptosporidium, Newport Beach, California US.
LeChevallier, M. W., and W. D. Norton. 1992. Examining relationships between
particle counts, and Giardia and Cryptosporidium and turbidity. J. Am. Water
Works Assoc. 84(12):54-60.
Medema, G. J., W. Hoogenboezem, A. J. van der Veer, H. A. M. Ketelaars, W. A.
M. Hijnen, and P. J. Nobel. 2003. Quantitative risk assessment of
Cryptosporidium in surface water treatment. Wat. Sci. Tech. 47(3):241-247.
Oppenheimer, J. A., E. M. Aieta, R. R. Trussell, J. G. Jacangelo, and I. N. Najm.
2000. Evaluation of Cryptosporidium inactivation in natural waters. American
Water Works Assoc. Research Found. Denver CO US.
Payment, P., A. Bérubé, D. Perrefault, R. Armon, and M. Trudel. 1989.
Concentration of Giardia lamblia cysts, Legionella pneumophila, Clostridium
perfringens, human enteric viruses, and coliphages from large volumes of
drinking water, using a single filtration. Can. J. Microbiol. 35:932-935.
Richardson, A. J., R. A. Frankenberg, A. C. Buck, J. B. Selkon, J. S. Colbourne, J.
W. Parsons, and R. T. Mayon-White. 1991. An outbreak of waterborne
cryptosporidiosis in Swindon and Oxfordshire. Epidemiol Infect 107:485-95.
___________________________________________________________________
- 70 -
Chapter 2
Teunis, P. F. M., G. J. Medema, L. Kruidenier, and A. H. Havelaar. 1997.
Assessment of the risk of infection by Cryptosporidium and Giarda in drinking
water from a surface water source. Water Res. 31:1333-1346.
Van der Kooij, D., Y. C. Drost, W. A. M. Hijnen, J. Willemsen-Zwaagstra, P. J.
Nobel, and J. A. Schellart. 1995. Multiple barriers against micro-organisms in
water treatment and distribution in the Netherlands. . Wat. Supply 13:13-23.
WHO. 2004. Guidelines for Drinking Water Quality, third edition. World Health
Organization, Geneva, SW.
___________________________________________________________________
- 71 -
Chapter 3
Enumeration of faecal indicator bacteria
in large water volumes using on site
membrane filtration to assess water
treatment efficiency•
Large volume sampling
Standard method
.
.
.
SSRC determination
W.A.M. Hijnen1, D. Veenendaal1, W.M.H. van der Speld1, A. Visser2, W. Hoogenboezem3 and D. van der Kooij1
KWR Watercycle Research Institute, PO Box 1072, 3430 BB Nieuwegein NL
Dune Water Company of South Holland Ltd., PO Box 34, 2270 AA Voorburg NL
3 Hetwaterlaboratorium, PO Box 734, 2003 RS Haarlem, NL
1
2
•
Reprinted from Water Research, 34:1659-1665, Copyright 2000, with permission from
the copyright holder, Elsevier limited.
___________________________________________________________________
- 73 -
Chapter 3
.
ABSTRACT
Sample volumes as tested in routine microbiological methods for determining
the presence and absence of faecal indicator bacteria in water are too small to
assess the actual concentration in the last stages of a water treatment.
Consequently no accurate information can be obtained about the removal
efficiency of a water treatment for micro-organisms. Therefore a method for
on site isolation of faecal indicator bacteria from large volumes (100 litre or
more) of treated water using membrane filtration (MF-sampling) was
developed and tested. The procedures for culturing the isolated microorganisms were similar to those applied in the routine methods for small
volumes using membrane filtration (mf-method). The recovery efficiency of
MF-sampling for E. coli, S. faecalis and spores of sulphite-reducing clostridia
ranged from 74.6 to 100% and only for E. coli a slight decrease with increasing
sample volume was found. Field studies revealed that MF-sampling can
easily be implemented in (routine) laboratory practice for an accurate
determination of the concentration of faecal indicator bacteria in treated water
after various treatment stages. From these data the treatment efficiency of the
involved processes and the overall treatment for those micro-organisms and
the fluctuation in micro-organism removal were determined. Such data can
be used to improve water treatment regarding the removal of micro-organisms and for quantitative microbial risk assessment. Validation of the use of
faecal indicator bacteria as a surrogate parameter for the assessment of the
effects of treatment processes on pathogenic micro-organisms needs further
investigation.
INTRODUCTION
Outbreaks of waterborne diarrhoea in the USA and the UK caused by
persistent pathogenic protozoa Cryptosporidium and Giardia (Richardson et al.,
1991; MacKenzie et al., 1994; Kramer et al., 1996) have increased interest in the
effect of water treatment on pathogenic micro-organisms. Direct monitoring
of these micro-organisms is difficult because (i) their concentrations in source
and treated water are usually low and strongly fluctuating and (ii) analytical
procedures are time-consuming and have a low and variable recovery
efficiency. Traditionally the microbiological quality of drinking water is
assessed by monitoring for non-pathogenic bacteria of faecal origin (faecal
indicator bacteria). These indicator bacteria usually include total and faecal
coliforms (FC) and in the European legislation also water quality criteria for
faecal streptoccocci and spores of sulphite-reducing clostridia (SSRC) have
___________________________________________________________________
- 74 -
Chapter 3
been defined (Anonymous, 1998). SSRC have been included in legislation in
the Netherlands since 1984 (Anonymous, 1984). Most SSRC present in raw
and treated water are of faecal origin (Clostridium perfringens) and their
presence may be indicative for the presence of persistent pathogens. More
recently it has been suggested that SSRC may be used as a process indicator
for the assessment of the capacity of water treatment processes to remove
(oo)cysts of pathogenic protozoa and viruses (Payment and Franco, 1993;
Hijnen et al., 1997).
In the Netherlands data are collected on the concentrations of faecal indicator
bacteria in water in various treatment stages to obtain quantitative
information on the removal capacity of water treatment for micro-organisms.
The concentration of indicator bacteria can easily be determined in the source
water, but after one or more treatment processes the concentration decreases
and a sample volume of 100 ml as tested in the routine membrane filtration
methods (mf-method) is too small to assess the actual concentration. SSRC
removal efficiency in treatment processes at pilot plant scale was assessed by
the enumeration of these bacteria in 10 litre samples with the mf-method
(Hijnen et al., 1997), but even larger sample volumes (100 litre or more) are
needed to determine the efficiency of the final treatment steps and the overall
treatment. Therefore the mf-method was scaled up and adapted for on site
isolation of (indicator) bacteria from large volumes of water (MF-sampling)
and enumeration in the laboratory with similar cultivation techniques used in
the routine methods. This new isolation method was tested under laboratory
and field conditions.
MATERIALS AND METHODS
MF-sampling. An MF-sampling procedure was designed to filter a
water volume of 100 litre within 1 h. With a tubing pump (Masterflex 0754950 with 07019-00 pump head), using silicone tubing, stainless steel (SS) piping
and pressure regulation, the supplied water was forced through a sterile ∅
142 mm 0.45 ìm pore size membrane filter (Millipore HAWP 142-50) in a filter
house (Schleicher & Schuell AJ6020-2). The entire system of water supply and
filter house is autoclaved for 15 minutes at 121oC before sampling. The
required flow (maximum of 250 l/hr) and an overpressure of 1 ATO is
adjusted with a flow control system and the tubing pump. When the intended
volume has passed through the membrane filter, the valve of the water
supply is closed. Thereafter the membrane filter is aseptically transferred in a
sterile glass petri dish (∅ 165 mm). When no further pre-treatment is required
___________________________________________________________________
- 75 -
Chapter 3
.
before incubation, resuscitation can be started on site by placing the
membrane filter directly onto the solid medium.
Recovery tests. Stock cultures (one batch) of Escherichia coli (WR1),
Streptococcus faecalis (WR63) and spores of Clostridium bifermentans (CP1) in
peptone/glycerol solution stored at -70oC were used for the recovery tests.
The concentration of micro-organisms in these stock cultures had been
determined periodically (Table 1) in routine quality checks of the mf-method
(0.45 μm, ∅47 mm membrane filters; Sartorius 13906-50-ACN) for
enumeration of coliforms, faecal streptococci and SSRC in environmental
samples.
The effect of the sampling volume on the recovery was tested using 550 litre
volumes of non-sterile tap water (turbidity 0.1 NTU) without residual
disinfectant stored at room temperature in an SS tank with a mixing device.
These volumes were separately inoculated with the stock cultures to achieve a
final concentration of approximately 1 CFU per litre. Faecal indicator bacteria
were isolated from 10, 20, 25, 40 and 50 litre samples with the MF-sampler.
E.coli and C. bifermentans were tested in duplicate samples; S. faecalis was
tested in single samples.
The recovery efficiencies obtained with MF-sampling and with the routine
mf-method for E. coli and C. bifermentans and for indigenous SSRC from river
water were determined in separately prepared suspensions in sterile tap
water. For this purpose 1 litre volumes were inoculated with stock cultures
for a concentration of approximately 900 CFU of E.coli and 500 CFU of C.
bifermentans per litre, respectively. From these suspensions 100 ml samples
were tested directly with the mf-method using an SS filtration device (Schleicher&Schuell AS002/3). Prior to the testing with the MF-sampling procedure
volumes of 100 ml of these suspensions were diluted in 5 litres of autoclaved
(121oC) tap water at room temperature. For the test with indigenous SSRC a
volume of 50 litre sterile tap water was inoculated with 50 ml of river water
with an SSRC concentration of 103 to 104 CFU/l. Samples of 5 and 7.5 litre
were tested in duplicate with the mf-method and single samples of 10 and 15
litre samples were tested with MF-sampling.
Microbiological analysis. Membrane filters were incubated on Lauryl
Sulphate Agar (LSA; Oxoid MM615) for 5±1 h. at 25±1oC and 14±2 h. at
44±0.5oC to cultivate E. coli and the faecal coliforms (FC). FC concentrations in
the samples of the field study were derived from the number (N) of yellow
colonies on LSA at 37±1oC for 14±2 h. Typical colonies (√N) were tested in
Brilliant Green Bile broth (Oxoid CM31) during 24 h. at 44±0.5oC and the
percentage of positive tests was used to calculate the FC concentration in the
samples. Faecal streptococci were determined by incubation on KF
___________________________________________________________________
- 76 -
Chapter 3
streptococcus Agar (Difco 0496-17-2) during 48±4 h. at 37±1oC. In the
laboratory experiments concentrations of SSRC and spores of C. bifermentans
were determined on Perfringens-Agar-Base medium (PAB; Oxoid CM587)
incubated for 48±4 h. at 37±1oC. For the field studies Shahidi Ferguson
Perfringens agar base (Difco 0811-17-0) was used for SSRC determinations.
Sample volumes of 100 ml were pasteurized in a water bath at 70±1oC for 30
minutes prior to filtration. With sample volumes of 1 litre or more the
membrane filters were pasteurized in the liquefied medium placed in an oven
at 70±1oC for 30 minutes. The cultivation procedure of the membrane filters
for SSRC analysis was carried out as previously described (Hijnen et al., 1997).
Field tests. The suitability of the MF-sampling procedure under field
conditions was tested in two treatment facilities (Figure 1).
Scheveningen
River Meuse water
Dune
passage
Lagoon
coagulation
Andijk, Lake IJssel
River Rhine water
Micro
Straining
Impoundment
Reservoir IR
Powdered Activated
Carbon PAC
Rapid Sand
Filtration
Softening
with Lime SL
Rapid Sand
Filtration RSF
Open Collecting
Basin OCB
Chemical
Disinfection Cl2
Coagulation
Sedimentation
Slow Sand
Filtration SSF
Post disinfection
Rapid Sand
Filtration
Granular Activated
Carbon Filtration GAC
Post disinfection
Figure 1 Water treatment of facility 1 (the Dune Water Company South-Holland Ltd.
and of facility 2 (PWN Water Supply Company North Holland Ltd.) where MFsampling was tested under field conditions
The concentration of faecal coliforms (FC) and SSRC was daily monitored by
the involved water supply companies in the water after several treatment
stages in a two-week period in January (water temperature of 10oC) and in
February (water temperature 3-5oC) at facility 1 and 2, respectively. Samples
of 1 up to 10 litre collected after the open collecting bassin (OCB) and after
rapid granular filtration (RGF) at facility 1 and after chlorination (CL2) at
facility 2 were tested with the routine mf-method. Samples after CL2 were
neutralized with sodium thiosulphate (30 mg/l). The MF-sampling procedure
___________________________________________________________________
- 77 -
Chapter 3
.
was used to isolate both indicator bacteria from 57-100 litre samples of the
SSF-filtrate at facility 1 and of the GAC-filtrate at facility 2, respectively. The
results were compared with the results of the routine microbiological
monitoring program conducted by the water supply companies in this
period.
Statistical analysis. The similarity of the numbers of typical colonies
obtained with the MF-sampling and the mf-method was tested by the
student-t test and the F-test (Excel software). Normal distribution of the data
was tested with the Kolmogorov-Smirnov-test (Statistical Package for Social
Sciences SPSS).
RESULTS AND DISCUSSION
Recovery tests. A method with a high recovery efficiency was
developed for an accurate determination of the concentration of indicator
bacteria in water. The average recovery efficiencies of the experiment with
increasing sample volumes obtained from large volumes (550 litre) were
estimated from the ratio between the measured and the calculated concentrations of the indicator bacteria in each separate sample (Table 1).
Table 1 The calculated and measured average concentrations (± standard deviation,
SD) as well as the recovery efficiencies determined from the colony counts in different
sample volumes from separate suspensions of E. coli, S. faecalis and C. bifermentans
in tap water tested with MF-sampling; linear regression analysis for the relation
between the sampled volumes and the colony counts (Figure 2)
Data of recovery tests
E. coli
S. feacalis
C. bifermentans
Stock cultures
Conc. (n/l)
CFUa/ml
27 (±3.3; 5c)
46 (±11.1; 30c) 102 (±36; 30c)
Calculated 1.09 (±0.22)
0.94 (±0.23)
0.93 (±0.33)
Measured
0.89 (±0.12)
0.703 (±0.23)
1.06 (±0.11)
Samples
Nb
8
4
8
Recovery
%
81.9 (±11.1)
74.6 (±24.9)
114 (±11.4)
Slope
0.787 (±0.06)
1.061 (±0.07)
1.000 (±0.09)
Lin. reg. anal.
r2
(Vol. vs CFU)d
0.967
0.991
0.950
p
<0.0001
<0.0100
<0.0001
a Colony Forming Units; b N = number of observations; c number of tested samples
from the stock cultures with mf-method; d linear regression analysis: volume versus
CFU.
The calculated concentration was derived from the average concentration in
the stock culture as determined with the routine mf-method. The average
___________________________________________________________________
- 78 -
Chapter 3
recovery efficiencies for the three indicator bacteria were ≥74.6% of the
calculated concentrations. The colony counts increased linearly with the
sample volume (Figure 2) with slopes of 1.06 and 1.00 for S. faecalis and C.
bifermentans, respectively (Table 1), indicating that the recovery was
independent of the sample volume. For E.coli a slope value of 0.787 revealed
that the recovery efficiency slightly decreased with increasing sample
volume. It is possible that this decrease was caused by a decline of the
concentration in the suspension during the experiment. These recovery
efficiencies are similar or higher than those reported for the membrane
adsorption-elution techniques using cartridge filters and elution-procedures
(≥30 up to 86%; Goyal et al., 1980; Payment et al., 1989). The recovery efficiencies of the MF-sampling and the mf-method were similar for E. coli, C.
bifermentans and for the indigenous SSRC from river water (Table 2; P ≥ 0.99,
0.37 and 0.81, respectively for 95%-interval).
Colony Forming Units (CFU)
70
E. coli
S. faecalis
C. bifermentans
60
50
40
30
20
10
0
0
10
20
30
40
50
60
Volume (l)
Figure 2 The linear regression fit of the recovery tests with faecal indicator bacteria in
tap water without residual disinfectant
Field tests. Relatively high concentrations of FC (205±94 CFU/l) were
observed in the water after dune passage sampled from the open collecting
basin (OCB) at facility 1 (Table 3). Apparently the water in this open basin
was faecally recontaminated by birds and other wild life. The SSRC
concentration in this water was lower and more fluctuating (8.3±16.3 CFU/l).
MF-sampling of the slow sand filtrate before post-disinfection showed that
___________________________________________________________________
- 79 -
Chapter 3
.
treatment (PAC, SL, and RGF followed by SSF; Figure 1) clearly reduced these
concentrations.
Table 2 Concentrations of E. coli, C. bifermentans and indigenous SSRC from river
water (arithmetic mean±SD) in separate suspensions in sterile tap water determined
with the mf-method and with the MF-sampling procedure
mf-method
MFsampling
Sample volume
(ml)
E. coli
(CFU/100 ml)
C. bifermentans
(CFU/100 ml)
SSRC
(CFU/l)
100
93.3 (±12.2)
38.3 (±18.6)
8.9 (±2.1)
(n=3)
(n=3)
(n=4)
93 (±9)
49.3 (±3.7)
9.2 (±1.2)
(n=3)
(n=3)
(n=2)
5000a
prior to MF-sampling: 100 ml of a prepared suspension mixed in 5 litre
sterile tap water
a
Average FC and SSRC concentrations in the SSF-filtrate were 0.19±0.03 CFU/l
and 0.10±0.07 CFU/l, respectively. The routine microbiological monitoring
program conducted in the same period in the SSF-filtrate yielded an average
concentration for FC of 0.09±0.54 CFU/l (145 samples of 300 ml) and for SSRC
of 0.36±1.88 CFU/l (58 samples of 100 ml), respectively. These values are in
the same order of magnitude as those determined with MF-sampling, but
relative standard deviations clearly exceeded 100% as the result of a low
percentage of positive samples (3%).
The FC concentrations in the water from the impoundment reservoir (IR) of
treatment facility 2 were 4 times lower than the concentrations of these microorganisms in the OCB of facility 1 (Table 4), but SSRC concentrations were 10
times higher. Chlorination (Cl2) strongly reduced the FC concentration to
0.01±0.3 CFU/l and these indicator bacteria were not found (<0.001 CFU/l) in
a total volume of 1.000 litre GAC-filtrate tested with MF-sampling. Average
SSRC concentrations after Cl2 and GAC were 29 and 0.43 CFU/l, respectively
and could easily be determined by testing 1 and 100 litre samples with the mfmethod and MF-sampling, respectively. No SSRC were detected in the GACfiltrate (8 samples of 1 l) during routine microbiological monitoring in this
period and also all samples after post-disinfection tested as prescribed by
legislation were negative for FC and SSRC.
___________________________________________________________________
- 80 -
Chapter 3
Table 3 Concentrations (n/l) of faecal coliforms (FC) and spores of sulphite-reducing
clostridia (SSRC) in the water of the open collection basin (OCB) after dune passage
and after powderer activated carbon, softening with lime and rapid granular
filtration (PAC/SL/RGF) and slow sand filtration (SSF) of facility 1 in January 1997
(water temperature 10.4±0.4oC)
No.
Faecal coliforms
Spores of sulphite-reducing
clostridia
OCBa PAC/SL/RGFa SSFb OCBa PAC/LS/RGFa
SSFb
1
418
47.6
0.19
<1
<0.1
0.03
2
192
39
0.19
NDc
ND
0.01
3
208
64.4
0.19
12
0.9
ND
4
299
48.5
0.21
ND
ND
0.13
5
226
54
0.24
<1
<0.1
ND
6
111
36.3
0.21
1
0.2
0.13
7
164
34.4
0.21
1
0.1
0.16
8
103
24.6
0.14
44
<0.1
0.03
9
140
26.5
0.14
ND
ND
0.2
10
193
33.4
0.17
<1
<0.1
0.12
205
40.9
0.19
8.3
0.17
0.10
avgd
94
12.6
0.03
16.3
0.33
0.07
SDd
aRoutine mf-method: 4 times 2.5 litre using ∅ 47 mm membranes (10 litre
samples); b MF-sampler: sample volume of 100 l; c ND = not determined; d
avg = arithmetic mean and SD = standard deviation
The removal efficiency of the treatment processes. Due to the low
percentage of positive samples the routine microbiological monitoring
program of both water supply companies was unfit to determine removal
efficiencies of all processes or the overall treatment for FC and SSRC.
Increasing the sample volume resulted in higher percentages of positive
samples and revealed a more accurate determination of the actual
concentrations as indicated above. For FC at facility 1 and for SSRC at facility
2, respectively this enabled the assessment of the actual Decimal Elimination
(DE), which is defined as log Cin - log Cout. From these DE-values and from the
average concentrations in the water, the decimal elimination capacities (DEC)
of unit processes and of the overall treatment without post-disinfection for
both micro-organisms in this period were calculated (Table 5).
___________________________________________________________________
- 81 -
Chapter 3
.
Table 4 Concentrations (n/l) of faecal coliforms (FC) and spores of sulphite-reducing
clostridia (SSRC) in the water of the impoundment reservoir (IR) and after
chlorination (Cl2) and granular activated carbon filtration (GAC) at facility 2 in
February 1997 (water temperature 4±1oC)
No.
Faecal coliforms
Spores of sulphite-reducing
clostridia
IRa
Cl2a
GACb
IRa
Cl2a
GACb
1
25
<0.33
<0.014
270
6.2
1.60
2
210
<0.2
<0.013
635
17.8
0.11
3
10
<0.2
<0.01
1010
43.6
0.07
4
60
<0.2
<0.01
2085
40.4
0.06
5
150
0.2
<0.01
1020
44.6
0.09
6
150
<0.2
<0.01
250
27.2
0.37
7
610
<0.1
<0.01
900
28
0.59
8
320
<0.1
<0.01
1550
28
0.66
9
70
<0.1
<0.01
1350
28
1.80
10
20
<0.1
<0.01
1050
18
0.81
49.5
1079
29
0.43
avgd
0.01
<0.001
180.5
NDc
562
14.3
0.53
SDd
0.3
a Routine mf-method: 4 times 2.5 litre using ∅ 47 mm membranes (10 litre
samples); b MF-sampler: sample volume of 100 l; c ND = not determined; d
avg = arithmetic mean and SD = standard deviation
Clearly, at treatment facility 1 SSRC concentrations were removed more
efficiently by the combination of PAC, SL and RGF than FC concentrations.
For slow sand filtration however it was the other way around. The DEC value
of this process for FC was 2.3±0.09 log, but the filters showed hardly any
elimination of the clostridia spores. This phenomenon has also been observed
in GAC filter beds in a pilot plant which were infrequently backwashed
(Hijnen et al., 1997). At facility 1 concentrations of SSRC were observed in the
filter bed of two slow sand filters and the highest numbers were detected in
the top layer (Figure 3). Most likely SSRC accumulated and subsequently
survived in the filter bed, but under anaerobic conditions multiplication can
not be excluded. These observations demonstrate the need for further
research on the use of SSRC as a surrogate parameter for the assessment of
the effects of filtration processes on protozoan (oo)cysts.
___________________________________________________________________
- 82 -
Chapter 3
Table 5 The Decimal Elimination Capacity (DEC) of unit processes and the overall
treatment of drinking water facilities Scheveningen and Andijk
Facility
Treatment processes
DEC value (log)
FC
SSRC
PAC/SL/RGF
0.7±0.13a
1.7b
SSF
2.3±0.09
0.2
Treatment
3±0.17
1.9
2
Cl2
3.6
1.5±0.25
Treatment
>4.7
3.5±0.6
aDEC±SD = arithmetic mean (avg) of DE-values; bDEC = log (avg C ) - log
in
(avg Cuit)
1
Chlorination in treatment facility 2 (average product of residual concentration
C and the contact time t (Ct) of 40 mg.min/l) reduced the average FC
concentration in the water with 3.6 log (Table 5).
This DEC for these Ct-values is low compared to values reported in literature
(Sobsey, 1989). A DEC-value of 1.5±0.25 log was calculated for the
inactivation of spores of clostridia by chlorination, indicating that these spores
are more resistant to chemical disinfection. From the concentrations of FC and
SSRC in the GAC-filtrate, assessed with MF-sampling, a DEC-value of ≥ 4.7
and of 3.5±0.6, respectively was calculated for the overall treatment.
SSRC (CFU/g of sand)
60
50
40
30
20
10
0
0
10
20
30
40
50
60
Bed depth (cm)
Figure 3 The average SSRC concentration (±SD; n=6) in the sand of two
individual slow sand filters of drinking water production plant Scheveningen
___________________________________________________________________
- 83 -
Chapter 3
.
The DE-values of unit processes and the overall treatment also gave
information about the variation in removal of the indicator bacteria during
these two-week periods in winter (Figure 4). The DE-values of unit treatment
processes and the overall treatment of both facilities for FC and SSRC showed
a normal distribution (P≥ 0.827 - 0.999, 95%-interval). Table 5 show that the
variation in DE-values of the processes at facility 1 for FC was limited (SD <
0.2 log). Also the inactivation of SSRC by chlorination at facility 2 showed
little variation (SD = 0.25 log), but for the DEC-value of the overall treatment
for SSRC the SD-value was 0.6 log (Table 5). As yet it is not clear which
processes or process conditions are responsible for variations in microorganism removal.
5.0
a
PAC/SL/RSF
SSF
4.0
DE-values
Overall treatment after dune passage
3.0
2.0
1.0
0.0
0
20
40
60
80
100
Percentage observations (%)
5.0
Cl2
Overall treatment
DE-values
4.0
3.0
2.0
1.0
b
0.0
0
20
40
60
80
100
Percentage observations (%)
Figure 4 The cumulative frequency distribution of the DE-values of unit processes
and of the overall treatment for FC and SSRC at facility 1 (a) and 2 (b), respectively,
during a two week period in the winter period of 1997
___________________________________________________________________
- 84 -
Chapter 3
The presented quantitative data about the removal efficiency of full-scale
treatment processes for indicator bacteria obtained with MF-sampling
contributes to a more accurate assessment of the risk of infection with
pathogens via drinking water (Teunis et al., 1997) necessary to determine the
need for process optimization. Furthermore, with MF-sampling the process
conditions responsible for the low DE-values may be elucidated subsequently
leading to process optimization for the removal of pathogens.
CONCLUSIONS
The routine method for determining the presence and absence of faecal
indicator bacteria (e.g. faecal coliforms (FC), faecal streptococci and spores of
sulphite-reducing clostridia (SSRC) in relatively small volumes of water using
membrane filters (mf-method) is inadequate for the assessment of micro-organism removal efficiency of all processes in water treatment. Sampling of
larger volumes with specially designed equipment (MF-sampling) in two full
scale treatment plants yielded accurate determination of the concentration of
faecal indicator bacteria (FC and SSRC) in the finished water (filtrate of SSF
and of GAC) with a detection limit of 0.01 CFU/l or less. The same cultivation
procedures are used as those applied in the routine mf-methods and the
recovery efficiencies of both methods were similar.
By determining simultaneously the concentration in the source water and in
the water after selected treatment stages the decimal elimination capacity
(DEC) and the variation in DEC of unit processes and the overall water
treatment without post-disinfection for these micro-organisms was assessed.
Such data can be used to improve the microbiological safety of the drinking
water by improving water treatment for the removal of micro-organisms. The
significance of the observed accumulation of SSRC in slow sand filters and the
relative low DEC of chlorination for FC and SSRC needs further research. This
is necessary to verify the use of these indicator bacteria as surrogate
parameters for the assessment of the effects of water treatment processes on
pathogenic micro-organisms.
REFERENCES
Anonymous. 1998. Council Directive 98/83/EC of 3 November 1998 on the quality
of water intended for human consumption. Official Journal of the European
Communities L330:32-54.
Anonymous. 1984. Waterleidingbesluit Staatsblad van het Koninkrijk der
Nederlanden, 220: 1-35.
___________________________________________________________________
- 85 -
Chapter 3
.
Goyal, S. M., H. Hanssen and C.P. Gerba. 1980. Simple method for the
concentration of influenza virus from allantoic fluid on microporous filters.
Appl. Environ. Microbiol. 39:500-4.
Hijnen, W. A. M., W. M. H. van der Speld, F. A. P. Houtepen, and D. van der
Kooij. 1997. Presented at the International Symposium on Waterborne
Cryptosporidium, Newport Beach, California US.
Kramer, M. H., B. L. Herwaldt, G. F. Craun, R. L. Calderon, and D. D. Juranek.
1996. Waterborne Disease: 1993 and 1994. J. Am. Water Works Assoc. 88:66-80.
MacKenzie, W. R. H., N.J. , M. E. Proctor, S. Gradus, K. A. Blair, D. E. Peterson, J.
J. Kazmierczak, D. G. Addiss, K. R. Fox, J. B. Rose, and J. P. Davis. 1994. A
massive outbreak in Milwaukee of Cryptosporidium infection transmitted through
the public water supply. New Engl. J. Med. 331:161-167.
Payment, P., A. Bérubé, D. Perrefault, R. Armon, and M. Trudel. 1989.
Concentration of Giardia lamblia cysts, Legionella pneumophila, Clostridium
perfringens, human enteric viruses, and coliphages from large volumes of
drinking water, using a single filtration. Can. J. Microbiol. 35:932-935.
Payment, P., and E. Franco. 1993. Clostridium perfringens and somatic coliphages as
indicators of the efficiency of drinking water treatment for viruses and protozoan
cysts. Appl. Environ. Microbiol. 59:2418-24.
Sobsey, M. D. 1989. Inactivation of health-related microorganisms in water by
disinfection processes. Wat. Sci. Tech. 21:179-195.
Richardson, A. J., R. A. Frankenberg, A. C. Buck, J. B. Selkon, J. S. Colbourne, J.
W. Parsons, and R. T. Mayon-White. 1991. An outbreak of waterborne
cryptosporidiosis in Swindon and Oxfordshire. Epidemiol. Infect. 107:485-95.
Teunis, P. F. M., G. J. Medema, L. Kruidenier, and A. H. Havelaar. 1997.
Assessment of the risk of infection by Cryptosporidium and Giarda in drinking
water from a surface water source. Water Res. 31:1333-1346.
___________________________________________________________________
- 86 -
Chapter 4
Quantitative assessment of the removal
of indicator bacteria in full-scale
treatments plants•
W.A.M. Hijnen1, G.J. Medema1, D. van der Kooij1 and A.H. Havelaar2
1
2
KWR Watercycle Research Institute, PO box 1072 , 3430 BB Nieuwegein NL
University Utrecht, PO box 80175, 3508 TD Utrecht, NL
•
Reprinted with adaptations from Wat. Sci. Tech.: Water Supply, 4(2): 47-54 with
permission from copyright holder, IWA publishing.
___________________________________________________________________
- 87 -
Chapter 4
.
ABSTRACT
The elimination of thermotolerant coliforms (Coli44) and spores of
sulphite-reducing clostridia (SSRC) in full-scale water treatment was
determined by large volume sampling. Objective was to determine the
elimination capacity of full-scale treatment processes for micro-organisms,
both vegetative bacteria and bacterial spores. In two short-periods in
winter and summer, information was collected about the elimination of
Coli44 and SSRC by the overall treatment, the contribution of the unit
processes and the variability in elimination. Coli44 concentrations in the
source waters were reduced by 3.2 to 6.3 log to an average concentration
sufficiently low to achieve more than 99% compliance with the Dutch
drinking water standard (0 in 300 ml; daily sampling). The elimination of
SSRC was lower (1.4 to 4.2 log units) and SSRC were observed occasionally
(>1%) in finished water by the routine weekly sampling of 100 ml samples.
The study also yielded much information about the elimination efficacy of
unit processes at the different locations, which enables process
optimization and improved process control. Moreover, it is demonstrated
that this quantitative information on removal of indicator bacteria by fullscale treatment systems can be used as input for quantitative microbial risk
assessment.
INTRODUCTION
In the beginning of the 20th century, bacteria that indicate faecal
contamination were introduced to monitor the microbiological safety of
drinking water (Eijkman, 1904). E. coli (or thermotolerant coliforms) and
enterococci are currently used throughout the world to monitor drinking
water quality. Outbreaks of cryptosporidiosis and other enteric diseases
through drinking water meeting the coliform standard have been reported
in developed countries (Craun et al., 1997). Thus, this indicator is not
adequate to predict the safety with regard to persistent pathogens like
Cryptosporidium, Giardia, and enteric viruses. The microbiological standards
for drinking water in the Netherlands include a standard (0/100 ml) for
sulphite-reducing clostridia spores (SSRC) since 1984. The rationale was
that compliance with this standard would require sufficient treatment to
eliminate chlorine-resistant pathogens as well. This multiple indicator
concept, in combination with other water quality issues such as disinfection
by-products, has resulted in the application of ozonation and physical
___________________________________________________________________
- 88 -
Chapter 4
treatment processes rather than chlorination as principal microbial barriers
in treatment.
The development of the Quantitative Microbiological Risk Assessment
(QMRA) for defining the microbiological safety of drinking water (Haas,
1983; Regli et al., 1991, Teunis et al., 1997; Medema et al., 2003) has stimulated
quantitative research into the effect of treatment processes on microorganisms. Because the concentration of pathogens in water normally is low,
much research is conducted with challenge tests on pilot plant scale with labcultured micro-organisms, rather than removal of indigenous microorganisms by full-scale treatment systems. However, such studies may
overestimate the removal capacities of naturally occurring micro-organisms.
A relatively simple large volume sampling (LVS) technique was developed
for assessing the concentration of faecal indicator bacteria after water
treatment (Hijnen et al., 2000). In the present study, this method was applied
to determine the elimination capacity of full-scale water treatment processes
for thermotolerant coliforms (Coli44) and SSRC as process indicators for
susceptible and resistant pathogenic micro-organisms. By monitoring the
concentration of these indicator bacteria throughout the treatment,
information was collected about the elimination capacity of overall treatment,
the contribution of unit processes and the variability of elimination. The use
of this information as input for process optimisation and Quantitative
Microbiological Risk Assessment will be discussed.
MATERIALS AND METHODS
The treatment plants and process information. Eight full-scale water
treatment plants were involved using several surface waters as source water
and a number of different unit processes (Table 1). The following unit
processes were included: coagulation/floc-removal (CFR), rapid granular
filtration (RGF), ozonation (O3), chlorination (Cl2), granular activated
carbon filtration (GAC), slow sand filtration (SSF), lime softening (LS) and
post-disinfection (PD) with chlorine or chlorine dioxide. Process conditions
(design and operational) of the unit processes were collected as well as
some physical water quality parameters (temperature and turbidity) in
order to relate these conditions with the observed removal of microorganisms.
Large volume sampling. The data of the routine monitoring
program showed that the concentration of faecal indicators in the water
was generally below the limit of detection after the first one or two unit
processes (Chapter 2). Therefore, an additional study with large volume
___________________________________________________________________
- 89 -
Chapter 4
.
sampling (LVS) was conducted. In a two-week period in winter and
summer thermotolerant coliform (Coli44) and spores of sulphite-reducing
clostridia (SSRC) were determined daily before and after unit processes,
except for the weekends. Volumes of 1 up to 10 litre were examined with
the routine membrane filtration method (∅47 mm; mf method). Faecal
indicators in 50 up to 1,100 litres samples of finished water before postdisinfection (if applied) were determined with an in-situ filtration device,
the MF-sampler (∅142 mm; Hijnen et al., 2000).
Table 1. The source water and the successive processes of the eight full-scale
treatment plants
Loc. Source water
Processes
1
River a
CFR – O3 – RGF – GAC – PD
2
River a
CFR – RGF – O3 –GAC – PD
CFR – O3 – RGF – GAC – PD
3
River a
4
Lake
CL2 – CFR – RGF – GAC – PD
5
Regional stream
CFR – RGF – GAC – RGF – SSF – (PD)e
RGF – O3 – LS – GAC – SSF – (PD)e
6
River b
c
O3 – LS – GAC – SSF – (PD)e
7
Local surface water
8
River d
RGF – SSF – PD
a After impoundment reservoirs; b After CFR, RGF, soil passage, open
recollection reservoir; c After CFR, reservoir and RGF; d After RGF, soil
passage and LS; e Stand-by post-disinfection
Microbiological analysis. Thermotolerant coliforms (Coli44) were
determined using Lauryl Sulphate Agar (LSA; Oxoid MM615).
Concentrations of Coli44 were derived from the number (N) of yellow
colonies on LSA at 37±1oC for 14±2 h. Typical colonies (√N) were tested in
Brilliant Green Bile broth (Oxoid CM31) during 24 h at 44±0.5oC and the
percentage of positive results was used to calculate the Coli44
concentration. Spores of sulphite-reducing clostridia (SSRC) were
determined on Perfringens-agar-base (PAB, Oxoid CM387) or on Shahidi
Ferguson Perfringens agar base (Difco 0811-17-0). Sample volumes of 100
ml were pasteurised in a water bath at 70±1oC for 30 minutes prior to
filtration. With sample volumes of 1 litre or more the membrane filters
were pasteurised in the liquefied medium placed in an oven at 70±1oC for
30 minutes (cultivation procedure previously described; Hijnen et al., 1997).
Calculation of the elimination capacity. The Decimal Elimination
Capacity (DEC) or log-removal of the treatments or unit processes was
___________________________________________________________________
- 90 -
Chapter 4
calculated as described in Chapter 2. Actual decimal elimination (DE) was
calculated from the Cin and Cout concentrations paired by date
DE = Log 10 C in − Log 10 C out = Log 10
C in
C out
Additionally, DEC of the treatments was calculated from the arithmetic
mean concentrations in the source and the finished waters with the same
equation (ratio estimation method; Chapter 2). Statistical analysis of the
collected data was performed with Microsoft Excel.
RESULTS AND DISCUSSION
Indicator bacteria in source water and finished water. The
concentrations of Coli44 and SSRC in the source and finished water (before
post-disinfection when applied) of the eight facilities during the monitoring
periods are presented in Table 2. The source water concentrations of both
indicators varied between <1 - >1000 CFU/l and were not related to the
season. There was a significant positive correlation between the average
concentrations of Coli44 and SSRC in the source waters (P<0.05); the
correlation, however, was not proportional (slope of 0.48) and the
regression coefficient was low (r2 = 0.43).
The concentrations of faecal indicators after treatment determined with
Large Volume Sampling (LVS) showed a distinct difference between the
concentration of Coli44 and SSRC (Table 2). Coli44 were detected in the
finished water at four locations. At two of these locations (5 and 8; Table 1)
no main disinfection is applied and there was some indication that
breakthrough of Coli44 was related to the concentration in the source water
(Figure 1). Except for location 8, the percentage of positive sample (%PS)
for Coli44 in most finished waters was ≤50% whereas %PS for SSRC in the
finished waters was significantly higher (≥67%). These indicator bacteria
were observed in the finished water at all locations. The weighted average
concentration of Coli44 varied between <0.0004 and 0.186(±0.035) CFU/l
(Table 2). SSRC concentrations were considerably higher and varied
between 0.003(±0.003) and 1.045(±0.076) CFU/l. These results clearly
demonstrate that breakthrough of persistent micro-organisms in water
treatment occurs more easily compared to breakthrough of the more
susceptible micro-organisms.
The elimination capacity of the overall treatment. Coli44 were
observed after the treatments of locations with no disinfection or higher
source water concentrations. The data showed that the concentration in the
___________________________________________________________________
- 91 -
.
___________________________________________________________________
- 92 -
(l)a
Finished water in winter
Finished water in summer
a
a
Loc.
Vol.
%PS
Avg (CFU/l)
Vol. (l)
%PS
Avg (CFU/l)
COLI44
1
1.0
0.6
1285
0
<0.0008
1833
0
<0.0005
2
7
-b
1000
0
<0.0010
3
36
970
10
0.0010
4
50
439
912
0
<0.0010
497
50
0.0160
5
970
438
3907
40
0.0010
5170
10
0.0017
6
37
140
1761
0
<0.0006
7
2.2
59
1725
0
<0.0006
2780
0
<0.0004
8
143
62
1000
100
0.186 (0.035)b
1000
100
0.145 (0.063)
SSRC
1
10
77
3305
70
0.0030 (0.06)b
5113
80
0.041 (0.058)
2
33
900
67
0.033 (0.054)
3
67
515
100
1.045 (0.076)
1079
1331
831
100
0.425 (0.543)
920
78
0.062 (0.072)
4
5
1100
540
2130
80
0.062 (0.073)
3230
90
0.028 (0.019)
6
88
555
1000
100
0.181 (0.103)
984
100
0.216 (0.074)
26.0
15
1050
100
0.202 (0.111)
1815
100
0.398 (0.208)
7
8
8
7
996
90
0.202 (0.358)
1000
100
0.706 (0.500)
a Vol. = total sampled volume; %PS = percentage of positive samples; Avg = weighted average calculated from the
total of detected colonies in the total sampled volume; b standard deviation; - = not monitored
Source water (n/l)
Winter
Summer
Table 2. The average concentrations of Coli44 and SSRC (Colony Forming Units; CFU) in the source and finished water at the
eight facilities in winter (N=10) and summer (N=10)
Chapter 4
Chapter 4
finished water Cout was not correlated with the source water concentrations
Cin for both indicators (Figure 1). This demonstrates that the concentration
in the finished water depends on the DEC of the overall treatment. The
DEC values for summer and winter periods were calculated with the ratioestimation method using the average concentrations in source water and
finished water (Drost et al., 1997; Evers and Groennou, 1999) or using the
individual DE values (n>5). DEC of the treatments for Coli44 ranged from
3.0 – 6.6 log (Table 3). Due to a low %PS in the finished water for most
facilities DEC values for Coli44 were minimal values and based on the
detection limit of the analysis. This detection limit depended on the
examined water volume during the selected period (Table 2).
For SSRC DEC values of the treatments varied between 0.9 and 4.7 log. For
elimination of SSRC no influence of the season (winter or summer) was
observed but DEC was significantly correlated with source water SSRC
concentrations Cin (Figure 2a; p<0.001). Assuming that higher source water
concentrations of faecal indicators are correlated with the general quality of
the source water, this correlation suggests that DEC of a treatment is higher
at locations with poor source water quality.
10
Coli44 DIS <Det. limit
Coli44 no DIS
Cout (CFU/l)
1
Coli44 DIS
SSRC +/- DIS
0.1
0.01
0.001
Detection limit
0.0001
0.1
1
10
100
1000
10000
Cin (CFU/l)
Figure 1. Relationship between concentrations of Coli44 and SSRC in the finished
water and the source water (DIS = main disinfection in treatment)
In general treatment is in design and operational conditions (contact times
and chemical dosages) tailored to meet the drinking water standards. A
more general water quality parameter for the source water quality is
___________________________________________________________________
- 93 -
Chapter 4
.
turbidity and for most of the locations turbidity data were collected during
the course of the study. These data showed a positive correlation of source
water turbidity and the source water concentrations of both indicators;
correlation coefficient of 0.99 (p<0.001) and 0.67 (p=0.066) for Coli44 and
SSRC, respectively. Removal of turbidity to levels of around 0.1 NTU is one
of the general quality goals used in treatment design. Hence, the higher the
source water turbidity, the higher the physical treatment effort to meet this
goal. The number of physical steps in the individual treatments is added up
for the different locations (lime softening not included) and presented with
the removal of turbidity and both indicators in Table 3. The use of a
chemical disinfection on the locations is also indicated. DEC values of the
overall treatment for Coli44 and SSRC was low at locations with two
physical processes (locations 7 and 8) and high at location 5 with five
physical processes (Table 3). The removal efficiency of turbidity ranged
from 0.5 – 2.1 log and was positively correlated with the DEC values for
SSRC (r2=0.60; p<0.05; Figure 2b).
6.0
6.0
a
DEC SSRC
Winter
Summer
b
4.0
4.0
2.0
2.0
r=0.55
p<0.05
r=0.79
p<0.001
0.0
0.0
0.0
1.0
2.0
3.0
Cin (log CFU/l)
4.0
0.0
0.5
1.0
1.5
2.0
2.5
Turbidity (log removal)
Figure 2. Relationship between the DEC (error bars = SD) of treatments for SSRC
and the concentration of SSRC in the source waters (Cin;a) and (b) the average
turbidity removal (DEC) in these treatments
For Coli44 this relationship was not significant (p>0.1). These results show
that the level of micro-organism removal assessed with Coli44 and SSRC as
process
indicators
for
susceptible
and
persistent
___________________________________________________________________
- 94 -
a
Source water (avg.)
Turb.
Coli44
SSRC
(NTU)
(CFU/l)
CFU/l)
1.7
102
7
0.2
30
20
0.6
36
67
1.2
7
33
1.5
0.8
44
1.9
88
321
3.9
244
1205
11
704
820
1.4
0.5
0.9
1.5
0.8
1.3
1.9
2.1
Turb.
(log)
3.0 (0.2)
>3.6
4.6
>3.8
>3.1
>4.8
>4.7
6.6
3.2 (0.4)
>5.2
>3.1
4.4
5.3
DEC Coli44
Winter
Summer
1.6 (1.3)
2.1 (0.3)
1.9 (0.3)
3.1 (0.4)
3.4 (0.5)
2.6 (0.2)
3.5 (0.6)
4.3 (0.5)
0.9 (0.9)
1.5 (0.2)
3.7 (0.8)
3.2 (0.5)
4.7 (0.5)
4.2 (0.3)
DEC SSRC
Winter
Summer
___________________________________________________________________
- 95 -
number of physical processes plus disinfection (D)
8 (2)
7 (2+D)
3 (3+D)
2 (3+D)
1 (3+D)
6 (3+D)
4 (3+D)
5 (5)
Location
(process)a
Table 3. The average source water quality, turbidity removal (log) and DEC(±SD) in winter and summer for Coli44 and
SSRC, arranged in increasing order of successively, number of physical processes, turbidity and Coli44 in the source waters
at the different locations
Chapter 4
Chapter 4
.
pathogenic micro-organisms is to some degree related to the removal of
suspended solids. This demonstrates the importance of physical processes
(straining, attachment/detachment) in the removal of micro-organisms in
treatment. Moreover, this explains the correlation of the source water
concentration and the DEC of the overall treatment as observed for SSRC
(Figure 2a).
Required treatment efficacy to meet the drinking water standards.
The Dutch drinking water standards require absence of Coli44 and SSRC
spores in 300 ml (daily samples) and 100 ml (weekly samples) of treated
water, respectively (Anonymous, 1984). It is obvious that a high level of
compliance with these standards (i.e. >99% of all samples taken in one
year) requires a much lower average concentration in the drinking water
than the detection limit of the method. Assuming a random (Poisson)
distribution of bacteria in the water, the average concentration of Coli44
and SSRC in drinking water should be as low as 0.018 and 0.035 per litre,
respectively, to obtain an acceptable quality level (AQL) of 1% positive
samples per year (95% CI; Van der Kooij et al., 1995). To verify these
theoretical calculations, Coli44 and SSRC data of the LVS study at the
locations were compared with microbiological data from three years of
legislative water quality monitoring in standard volumes prior to the
period of the LVS study. The average Coli44 concentration in the finished
water of locations 5 and 7 assessed with LVS was 0.0013, and <0.0004
CFU/l, respectively. This is below the required concentration of 0.018
CFU/l for an AQL1%. The percentage of positive samples non complying
with the standard during the three years of monitoring at both locations
was 0.6% at location 5 and 0.09% at location 7, thus showing that the
elimination of Coli44 of these treatments is indeed enough to obtain an
AQL1% for these indicator bacteria. In the finished water of locations 2, 3, 4
and 8 the percentage of positive samples for SSRC from the 3 year
legislative monitoring program was ≥1% (1.9, 5.6, 1.0) and <1% (0.6),
respectively); at the other locations the monitoring program was not
according to legislation (higher volumes than 100 ml were analysed). The
SSRC concentrations assessed with the MF-sampler in a short period in
winter at locations 2, 3, 4 and 8 were 0.033, 1.045, 0.457 and 0.202 CFU/l,
respectively, (Table 2) and at locations 4 and 8 in summer 0.062 and 0.706
CFU/l, respectively. These concentrations are in the same order of
magnitude or higher than the concentration of 0.035/l for the AQL1%, thus
supporting the theoretical calculations presented by Van der Kooij et al.
(1995). Further support for these theoretical calculations came from a
comparison of the results of the routine monitoring program at location 8
___________________________________________________________________
- 96 -
Chapter 4
with the MF sampling in the same periods. The percentage positive
samples for Coli44 and SSRC in the finished water during the sampling
period of this study was 2.5 and 3.4%, respectively. The concentrations of
both indicators Coli44 and SSRC assessed with the mf method in the
routine program were 0.083±0.53 (n=80) and 0.34±0.61 (n=58), respectively
and with the MF sampler 0.186±0.035 and 0.201±0.358, respectively. In
conclusion both monitoring methods showed similar concentrations larger
than required to comply with an AQL of 1%.
Elimination of Coli44 and SSRC by unit processes. The LVS-study
yielded much information about the elimination efficiency of individual
water treatment processes. To illustrate this, two locations are presented in
detail, one with (5) and one without main disinfection (7). Indicator
concentrations and performance of the unit processes are presented and
compared with DEC of the overall treatment (Figures 3 and 4). At location 5
no disinfection is applied and the local river water is treated with coagulation
and floc removal (CFR), double rapid granular filtration (RGF; sand),
granular activated carbon filtration (GAC) and slow sand filtration (SSF).
After treatment, a post-disinfection is stand-by. The elimination of Coli44 and
SSRC by the unit processes and the average concentrations after these
processes are presented in Figure 3. The results show that without chemical
disinfection the relatively high Coli44 and SSRC concentrations in the source
water were eliminated with 5.4 up to 6.6 and 4.1 up to 4.3 log, respectively.
Slow sand filtration (SSF) was responsible for the high elimination of Coli44
in the treatment. These indicator bacteria were removed with 2 – 4 log by this
process.
The elimination of both indicators in treatment of location 7 during winter
and summer is presented in Figure 4. Local surface water was pre-treated
with CFR, impoundment reservoir storage and RGF before the post-treatment
that is investigated in this study (Table 1). First process of the post-treatment
was disinfection with ozone (O3) followed by lime softening (LS), GAC and
SSF. SSF was the most important barrier for both Coli44 and SSRC in this
treatment (Figure 4). A relatively low inactivation of both SSRC and Coli44
was observed in the disinfection process with ozone.
Based on the collected data an overview of the assessed DEC values for
SSRC and Coli44 of all applied unit processes is presented in Figure 5. The
collected data revealed more DEC values for SSRC than for Coli44. DEC of
coagulation/floc-removal (CFR) for SSRC varied 0.8 log (0.8 - 1.6) with no
systematic difference between winter and summer. The observed relative
standard deviation of DEC ranged from 20 – 40%. The few data collected
___________________________________________________________________
- 97 -
Chapter 4
.
for Coli44 (0.5-1.3 log) indicated lower removal of these indicators by CFR
than SSRC.
SSRC
Elimination (log)
7
Coli44
WINTER
SUMMER
6
5
862
4
Concentrations cfu/l
2661 80.7
5.9
0.0013
Concentrations cfu/l
26.9
3.3
0.045
3
2
1
0
CFR
RGFGAC-RGF
SSF
Total
CFR
RGFGAC-RGF
SSF
Total
Processes
Figure 3. Elimination (log) of SSRC and Coli44 by the unit processes and the
overall treatment at location 5 determined in winter and summer; RGF-GAC-RGF
effect of combined processes; average concentrations in source water and after the
unit processes above the columns; error bars is SD
SSRC
6
Elimination (log)
5
4
Coli44
WINTER
SUMMER
Concentrations cfu/l
20.2 18.6
5.5
0.30
>
Concentrations cfu/l
56.9 9.9
1.6 <0.0004
>
3
2
1
0
O3
O3-LSGAC
LS-GAC
SSF
Total
O3
LS-GAC
SSF
Total
Processes
Figure 4. Elimination (log) of SSRC and Coli44 in the unit processes and the
overall treatment at location 7 determined in winter and summer; O3-LS-GAC
effect of combined processes; average concentration in source water and after the
unit processes above the columns; error bars is SD
The DEC values of rapid granular filtration (RGF) for SSRC showed a
similar variation as observed for CFR (0.9 log; 0.8 - 1.7). These filters were
more effective when they were directly applied to the source waters
(locations 6 and 8; Table 1) than when applied after CFR as a secondary floc
___________________________________________________________________
- 98 -
Chapter 4
removal. This difference is most likely caused by differences in the
operational conditions: filters at location 6 and 8 were operated at 3 -5 m/h
with contact times of approximately 20 minutes; the other filters were
operated at 9 – 12 m/h (contact times of 7 – 15 min.). Only for the RGF at
location 8 DEC for Coli44 could be calculated and this value was 1 log
lower than DEC of the same filters for SSRC.
For most of the applied disinfection processes DEC for Coli44 could not be
calculated due to 100% inactivation during the period of monitoring. For
SSRC DEC of disinfection processes varied between 0 and 3 log with the
highest value for chlorine at location 4 most likely due to the high Ct of this
process (62 - 84 mg/l.min) compared to the relative low Ct values of the
ozonation processes (1 - 2 mg/l.min). In summer this chlorine process was
more efficient in inactivation of SSRC than in winter.
GAC filtration is applied as a polishing step for the removal of micropollutants but the results showed that the filtration process also removed
both indicators. For Coli44 little information was collected but for SSRC
DEC varied from -0.6 to 1.1 log. For slow sand filters with no backwash and
long operational periods of more than one year before filter bed scraping
highly variable DEC values for SSRC were observed. Slow sand filters at
locations 5 and 7 removed SSRC with 1.0 – 2.3 log, but at locations 6 and 8 a
negative DEC was observed.
For locations 6 and 7 Coli44 was not detected in the filtrate and DEC of SSF
for these susceptible bacteria at locations 5 and 8 ranged from 2.3 - 3.8 log
for, higher than DEC for SSRC. SSF on location 5 was more effective in
elimination of both indicators, most likely because of the low filtration rate
(0.06 – 0.18 m/h) and consequently high contact time of 5 – 12 hrs (rates of
the other locations 0.25 – 0.4 m/h and contact times 3 – 4 hrs).
Variation in elimination capacity. Waterborne outbreaks described
in literature have been attributed to breakthrough of pathogens due to peak
concentrations in source water coinciding with sub-optimal or inadequate
treatment performance (Badenoch, 1990; Richardson et al., 1991; Craun,
1990). In Table 3 and Figures 2, 3 and 4 variation of indicator elimination in
the overall treatment and the unit processes is presented by the standard
deviation. The variation in elimination is considerable. The relative
standard deviation of DE of the unit processes at all locations for coli44 and
SSRC was 25% (n=8) and 47% (n=17), respectively. The variation in the
elimination calculated for the overall treatment was smaller than expected
from the sum of the variations of the unit processes; relative SD of 10%
(n=2) and 13% (n=10) for coli44 and SSRC. Consequently, the low
___________________________________________________________________
- 99 -
3
1
Coli44
CFR
SSRC
5
2
3
2
1
8
RGF
>
6
7
6
1
2
Locations
3
>>
4
Cl2
Disinfection (O3 and Cl2)
3
2
.
1
4
GAC
7
6
8
6
7
>
SSF
Winter
Summer
5
___________________________________________________________________
- 100 -
Figure 5. DEC values (error bars SD) of the unit-processes for SSRC (above) and Coli44 (below) arranged in
increasing order for SSRC in winter and summer; locations on x-axis presented for both SSRC and Coli44; no bar
means no calculation possible due to lack of data (error bars = SD)
0
1
2
3
4
5
-1
0
1
2
3
4
5
DEC (log)
Chapter 4
Chapter 4
elimination in one process appears to be compensated by a higher
elimination in the subsequent process(es). This phenomenon has been
demonstrated before for the elimination of SSRC by CFR and RGF (Hijnen
et al., 1997) and for CFR/RGF and O3 (Medema, 1999). It illustrates the
importance of the multiple barrier strategy applied in water treatment.
Further analysis of the variation in DE values for SSRC and Coli44 revealed
that the actual elimination of a number of unit processes and also of the
overall treatment was positively correlated with the log transformed
concentration in the influent of the process (Table 4), in particular for SSRC.
The meaning of this observation was unclear.
The study showed that at a low-level, breakthrough of faecal microorganisms can occur. In winter (7.5oC) Coli44 (1 CFU/100 litre) was
detected in the finished water at location 3 (arrow in Figure 6).
Table 4. Correlation of DE value calculated from paired data of Cin and Cout with
the log transferred influent concentration for a number of unit processes and total
treatments
Processes
na
CFR
0/2
RGF
1/3
Disinfection
2/4
GAC
1/3
SSF
2/3
Total Treatm.
2/5
a number significant
processes
SSRC
Coli44
2
a
Slope (r )
n
Slope (r2)
0
1/1
0.51(0.42)
1.68(0.85)
1/2
0.67(0.55)
1.00(0.75); 0.88(0.69)
0
0.62(0.72)
0
1.39(0.82); 1.28(0.96)
1/1
0.63(0.59)
0.87(0.88); 1.05(0.51)
1/1
1.12(0.75)
linear relations (p<0.01)/total number of unit
During this event low Coli44 and SSRC removal by CFR coincided with
low Coli44 inactivation by ozonation. Hence, sub-optimal performance in
one unit process can sometimes compromise the overall treatment
performance. Moreover, this breakthrough-event would not have been
detected by the routine (300 ml) Coli44-monitoring of finished water.
Pathogen removal and quantitative risk assessment. A major
revision in the revised Dutch Drinking Water Decree (Anonymous, 2001
was the new risk-based regulation for pathogens in drinking water for
surface water treatment. A quantitative risk analysis is required to
demonstrate compliance with an annual risk of infection of 10-4. This risk___________________________________________________________________
- 101 -
Chapter 4
.
level has been provisionally translated into drinking water standards for
pathogens, using dose-response data for pathogens and introducing a
safety factor of 10. The difference between the concentration of pathogens
in source water and these provisional drinking water standards is the
required treatment efficacy for pathogens.
1000
Concentration (n/l)
SSRC
Source
CFR
O3
GAC
Coli44
100
10
1
0.1
0.01
1
2
3
4
5
6
7
8
9 10 1
2
3
4
5
6
7
8
9 10
Observations
Figure 6. The concentration (CFU/l) of SSRC and Coli44 in the source water and
in the water after the successive unit processes at location 3 monitored in the same
period (arrow indicates the break-through event)
Cryptosporidium oocyst concentrations (corrected for the recovery efficiency
of the detection method) in source water of the utilities in this study were
used to calculate the required elimination for this pathogen (Table 5). SSRC
or C. perfringens have been proposed as process indicator for the removal of
resistant pathogens such as Cryptosporidium and Giardia (Payment and
Franco, 1993; Hijnen et al. 1997). The data on required removal for
Cryptosporidium were compared to the data on SSRC removal achieved by
the treatment systems (Table 5). The DEC of location 1 assessed for SSRC in
this study was higher than the required DEC for Cryptosporidium. At the
other locations, the obtained SSRC elimination was equal (location 2) or
lower than the required DEC for Cryptosporidium. The difference was
especially large at location 7 and 8.
DISCUSSION
Assessment of elimination capacity. The results of the presented
study showed that the elimination capacity of water treatment and the unit
___________________________________________________________________
- 102 -
Chapter 4
processes for susceptible (Coli44) and persistent (SSRC) indicator bacteria
can be assessed by using standard microbial methods with decreased
detection limit. The decreased detection limit was realized by examining
larger volumes of the water during treatment with the standard mf method
and in the finished water by using the in situ sampling device, the MF
sampler (Hijnen et al., 2000). The data of these limited monitoring periods
showed a higher prevalence of the persistent indicator (SSRC) in the last
stages of treatment than the more susceptible thermotolerant coliforms
(Coli44), while in general concentrations of both organisms in the source
waters were in the same order of magnitude.
Table 5. The required DEC of the water treatment to meet an annual risk of
infection of 10-4 for Cryptosporidium (concentration 2.6x10-5/l, Van der Kooij et al.
1995) and the DEC achieved by the treatment systems as determined with SSRC
Location
1
2
3
4
5
6
7
8
DEC required (log)
Cryptosporidium
2.7
3.1
3.1
4.8
4.7
4.7
3.7
4.2
DEC achieved (log)
SSRC
3.5
3.1
1.9
4.0
4.3
2.9
1.8
1.2
Variation in elimination. The data also showed considerable
variation in the elimination capacity, both within and between treatments.
In the monitoring periods of two weeks in winter and two weeks in
summer DEC of the eight treatments for SSRC ranged from 0.9 – 4.7 log.
For Coli44 the DEC of four out of the eight treatments could be quantified
and ranged from 3.0 – 6.6 log. For SSRC both in the source water
concentration and DEC of the treatments were positively correlated with
turbidity in the source waters and the turbidity reduction in treatment. This
demonstrates the relationship between elimination of microbes and
suspended solids. Furthermore it explains the positive correlation between
source water concentrations of SSRC and DEC for these anaerobic spores.
Positive correlations of the source water turbidity and turbidity reduction
in treatment with levels of Cryptosporidium and Giardia in the source water
and elimination of these pathogens in treatment of four drinking water
___________________________________________________________________
- 103 -
Chapter 4
.
facilities was observed previously (LeChevallier and Norton, 1992). At 1 log
turbidity removal in conventional treatment (coagulation/sedimentation
and filtration) parasite removal was approximately 0.9 log. In a larger
study including 66 sites, however, such a correlation was not observed
(LeChevallier et al., 1991). Thus, they argued that this relationship probably
does not hold for every treatment train and regulating drinking water
safety for parasites with requirements for turbidity removal is not justified.
Additional argument against such a regulation on turbidity removal for
microbiological safety in general is that it under-estimates the overall
elimination of microbes. Other processes like inactivation and predation (in
slow sand filters) also contribute to micro-organisms removal.
The current study also presents variation in DEC between the different unit
processes which to some degree could be related to specific process
conditions. Further research is required to demonstrate the causal
relationship between the hypothesized process conditions and the
elimination of these process indicators. As presented before, the
inactivation capacity of the full-scale ozonation and chlorination processes
for susceptible micro-organisms such as Coli44 is much lower than
expected on the basis of laboratory data. For the relative low DEC of
ozonation at location 7 a further study indicated that ozone-dosage
controlled by flow instead of the residual ozone concentration and a
relatively high DOC-content (high ozone demand) in the feed water were
important causes for the low performance (Hijnen et al., 2001). Remarkable
were the results observed for SSRC removal by GAC and slow sand
filtration. DEC was highly variable and sometimes negative (higher
concentration in filtrate than in influent). Multiplication of SSRC in these
aerobically operated filter beds as an explanation for this observation is not
plausible but can not be excluded. In previous studies (Hijnen et al., 1997;
2000) negative DEC-values for GAC and slow sand filtration have been
reported. On the basis of observations in the filter bed and the backwash
water this phenomenon was attributed to accumulation and breakthrough
in these filters with a low backwash frequency and intensity.
Finally, the collected data clearly demonstrate the benefits of the multiple
barrier concept since the observed variation in elimination capacity of the
overall treatment is lower than expected from the variation of the unit
processes.
Increased elimination at higher concentrations. The positive
correlation of the average elimination of SSRC in the eight treatments
(DEC) with the source water SSRC concentration (Figure 1) was explained
by more intensive treatment conditions due to a poor source water quality
with higher turbidity and SSRC concentration. More intensive treatment
___________________________________________________________________
- 104 -
Chapter 4
conditions are partly related to process design and partly related to
operation conditions. A similar correlation between the actual elimination
(DE) and Cin was observed, however, at individual processes operated
under nominal and more or less steady process conditions. Higher
efficiencies of physical removal processes such as coagulation/flocculation
and filtration at increased particulate concentrations have been described
before (Smoluchowski, 1917; O’Melia and Ali, 1978; Vigneswaran and
Chang, 1989). Thus, the observed correlation can be due to a physical
phenomenon. The observed correlation might also be related to dispersion
and retardation in the process; actual DE is calculated with the actual Cin
and Cout assuming that these concentrations can be paired. These
observations question the validity of describing variability in elimination
with DE as the parameter suggested in the present study. This statistical
aspect of the collected data needs further attention.
Coli44 and SSRC as process indicators in QMRA. In a first attempt
to use these data in quantitative microbial risk assessment site specific
removal data of SSRC was compared with site specific information on
required elimination of Cryptosporidium oocysts. A lack of elimination
capacity was observed at a number of locations but this analysis must be
considered as a first impression. The results of the current study show the
potential use of both faecal indicators as process indicators to collect
quantitative data on the elimination of susceptible and persistent microorganisms. More information from comparative studies is needed for
definitive conclusions on the translation of this quantitative information to
the elimination of waterborne pathogens in water treatment. There are
reasons to consider SSRC as a conservative process indicator for (oo)cysts
removal. The anaerobic spores are smaller than Cryptosporidium oocysts.
They are also more persistent than Cryptosporidium. The die-off rate of
spores C. perfringens in river water is low (0.005 log/day; Medema et al.,
1997). This is probably an important cause of the low SSRC removal in
filtering processes with a low filter bed cleaning frequency like SSF and
GAC. Preliminary results of a slow sand filter column study indeed
showed that Cryptosporidium oocysts are removed to a higher decree than
spores of C. perfringens (Hijnen et al., 2004). Thus, the LVS study showed
that SSRC is a useful and potential process indicator for describing the
removal of Cryptosporidium oocysts by coagulation/filtration and
disinfection processes, but most likely not for GAC and slow sandfiltration. The observed phenomenon of accumulation and delayed
breakthrough in these filter beds, however, might be also a factor of
concern for elimination of these persistent pathogens. In addition the use of
___________________________________________________________________
- 105 -
Chapter 4
.
Coli44 as a process indicator for the susceptible waterborne pathogens
needs further attention.
Compliance with the standards. Finally, the findings of this study
demonstrate that by additional large volume sampling for faecal indicators
in the produced drinking water the level of compliance with the regulated
standards for these indicators achieved with the applied treatment can be
assessed. At a number of locations average indicator bacteria
concentrations (Coli44 and SSRC) assessed with either small samples or
large volumes were related to the number of observed positive samples (=
non compliance with the standard of 0 in 100 or 300 ml samples) in the
same monitoring period. These findings justified the proposed maximum
average concentrations for a more than 99% compliance with the standards
of the annual mandatory microbiological drinking water control program
for Coli44 and SSRC at a 95% certainty level (Van der Kooij et al., 1995).
CONCLUSION
With Large Volume Sampling (LVS) the concentrations of faecal indicators
after treatment of the eight locations in a winter and a summer period was
determined (n = 10 to 20 per location). This study shows that a high level of
compliance with the drinking water standards for Coli44 and SSRC requires
treatment systems with a large elimination capacity (1.6 up to 4.6 log) for
these indicators. The multiple indicator concept and large volume sampling
provide quantitative information about the elimination of both vegetative
bacteria and resistant bacterial spores by treatment processes. It allows
assessment of the contribution of unit processes in the overall treatment
efficacy. In general, chemical disinfection and slow sand filtration are
important barriers against Coli44. SSRC are less effectively removed by
disinfection and slow sand filtration. Their ability to survive for long periods
may cause accumulation and eventually breakthrough through the filters
with low backwash frequency.
The efficacy of unit processes differed in the different treatment systems. In
some cases, this initiated research into the cause of low efficacy of unit
processes (such as the ozonation at location 7) and optimisation of the
process. The study also provided information about the short-term variability
of the efficacy of treatment processes. The variability in the overall treatment
efficacy was found smaller than the variation in the unit processes, illustrating
the robustness of multiple barrier treatment systems. Large Volume Sampling
allowed the detection of low-level breakthrough of micro-organisms. Relating
___________________________________________________________________
- 106 -
Chapter 4
these breakthrough events to raw water quality or process conditions may
lead to measures to prevent these risk events.
Ultimately, the data on efficacy of full-scale treatment systems against
micro-organisms can be used in quantitative microbiological risk
assessment of drinking water. SSRC data appear to be applicable in
quantitative assessment of the risk of resistant pathogens, such as
Cryptosporidium, for describing the efficacy of coagulation/filtration and
disinfection processes, but not for slow sand of GAC filtration.
Further research includes comparative studies on the removal of faecal
indicators and pathogens by unit processes and the improvement of the
enumeration methods of pathogens in the source water (recovery efficiencies,
specificity).
REFERENCES
Anonymous. 2001. Besluit van 9 januari 2001 tot wijziging van het
waterleidingbesluit in verband met de richtlijn betreffende de kwaliteit van voor
menselijke consumptie bestemd water, p. 1-53, vol. 31. Staatsblad van het
Koninkrijk der Nederlanden.
Anonymous. 1984. Waterleidingbesluit Staatsblad van het Koninkrijk der
Nederlanden, 220: 1-35.
Badenoch, J. 1990. Cryptosporidium in water supplies, London.
Craun, G. F. 1990. Waterborne Giardiasis, p. 267-293. In E. A. Meyer (ed.), Human
Parasitic Diseases, vol. 3. Elsevier Science Publ. Amsterdam, The Netherlands.
Craun, G. F., P. S. Berger, and R. L. Calderon. 1997. Coliform bacteria and
waterborne disease outbreaks. J. Am. Water Work Assoc. 89:96-104.
Drost, Y. C., J. T. Groennou, W. A. M. Hijnen, and D. Van der Kooij. 1997.
Statistische methoden ter bepaling van de belasting en eliminatiecapaciteit van
zuiveringsprocessesn m.b.t. micro-organismen. SWI 97.177, Kiwa Water
Research, Nieuwegein NL.
Eijkman, C. 1904. Die Gärungsprobe bei 46oC als Hilfsmittel bei der
Trinkwasseruntersuchung. Centralblaat für Bakteriologie Abth. 1 Orig. 37:742.
Evers, E. G., and J. T. Groennou. 1999. Berekening van de verwijdering van microorganismen bij de bereiding van drinkwater. RIVM 734301016.
Haas, C. N. 1983. Estimation of risk due to low doses of micro-organisms: a
comparison of alternative methodologies. Am. J. Epidemiol. 118:573-82.
Hijnen, W. A. M., T. G. J. Bosklopper, J. A. M. H. B. A. D. Hofman, and G. J.
Medema. 2001. Presented at the 10th Int. Ozone Assoc. Congres, London.
Hijnen, W. A. M., J. F. Schijven, P. Bonné, A. Visser, and G. J. Medema. 2004.
Elimination of viruses, bacteria and protozoan oocysts by slow sand filtration.
Wat. Sci. Technol. 50:147-154.
___________________________________________________________________
- 107 -
Chapter 4
.
Hijnen, W. A. M., W. M. H. van der Speld, F. A. P. Houtepen, and D. van der
Kooij. 1997. Presented at the International Symposium on Waterborne
Cryptosporidium, Newport Beach, California.
Hijnen, W. A. M., D. Veenendaal., W. M. H. Van der Speld, A. Visser, W.
Hoogenboezem, and D. Van der Kooij. 2000. Enumeration of faecal indicator
bacteria in large water volumes using on site membrane filtration to assess water
treatment efficiency. Water Res. 34:1659-1665.
LeChevallier, M. W., W. D. Norton, and R. G. Lee. 1991. Occurrence of Giardia and
Cryptosporidium spp. in surface water supplies. Appl. Environ. Microbiol.
57:2610-6.
LeChevallier, M. W. and W. D. Norton. 1992. Examining relationships between
particle counts, and Giardia and Cryptosporidium and turbidity. J. Am. Water
Works Assoc. 84:54-60.
Medema, G. J. 1999. Cryptosporidium and Giardia: new challenges to the water
industry. University of Utrecht, Utrecht, NL.
Medema, G. J., M. Bahar, and F. M. Schets. 1997. Survival of Cryptosporidium
parvum, Escherichia coli, faecal streptococci and Clostridium perfringens in river
water. Wat. Sci. Tech. 35:249-252.
Medema, G. J., W. Hoogenboezem, A. J. van der Veer, H. A. M. Ketelaars, and W.
A. M. Hijnen, and P. J. Nobel. 2003. Quantitative risk assessment of
Cryptosporidium in surface water treatment. Wat. Sci. Tech. 47:241-247.
O’Melia, C. R., and W. Ali. 1978. The role of retained particles in deep bed
filtration. Prog. Water Technol. 10:167-182.
Payment, P., and E. Franco. 1993. Clostridium perfringens and somatic coliphages as
indicators of the efficiency of drinking water treatment for viruses and protozoan
cysts. Appl. Environ. Microbiol. 59:2418-24.
Regli, S., J. B. Rose, C. N. Haas, and C. P. Gerba. 1991. Modeling the risk from
Giardia and viruses in drinking water. J. Am. Water Works Assoc. 83:76-84.
Richardson, A. J., R. A. Frankenberg, A. C. Buck, J. B. Selkon, J. S. Colbourne, J.
W. Parsons, and R. T. Mayon-White. 1991. An outbreak of waterborne
cryptosporidiosis in Swindon and Oxfordshire. Epidemiol Infect 107:485-95.
Smoluchowski, M. 1917. Versuch einer mathematischen Theorie der Koagulations
kinetic kolloider Losunger. Zeitschrift Physicalische Chemie 92:129.
Teunis, P. F. M., G. J. Medema, L. Kruidenier, and A. H. Havelaar. 1997.
Assessment of the risk of infection by Cryptosporidium and Giarda in drinking
water from a surface water source. Water Res. 31:1333-1346.
Van der Kooij, D., Y. C. Drost, W. A. M. Hijnen, J. Willemsen-Zwaagstra, P. J.
Nobel, and J. A. Schellart. 1995. Multiple barriers against micro-organisms in
water treatment and distribution in the Netherlands. Wat. Supply 13:13-23.
Vigneswaran, S., and J. S. Chang. 1989. Experimental testing of mathematical
models describing the entire cycle of filtration. Water Res. 23:1413–1421.
___________________________________________________________________
- 108 -
Chapter 5
Spores of sulphite-reducing clostridia
(SSRC) as surrogate for verification of
the inactivation capacity of full-scale
ozonation for Cryptosporidium•
Water
Ozone
W.A.M. Hijnen1, A.J. van der Veer2, J. van Beveren1 and G.J. Medema1
1 Kiwa
2
Ltd., PO Box 1072, 3430 BB Nieuwegein, NL
Evides, PO Box 59999, 3008 RA Rotterdam, NL
•
Reprinted from Water Science and Technology: Water Supply, 2(1): 163-170, with
permission from the copyright holder, IWA publishing.
___________________________________________________________________
- 109 -
Chapter 5
.
ABSTRACT
The inactivation of C. parvum and spores of C. perfringens by ozone
treatment in natural water was compared in a lab-scale continuous-flow
system. In addition the inactivation of the natural occurring spores of
sulphite-reducing clostridia (SSRC) in this water was monitored in one of
the lab-scale systems as well as in a full-scale ozonation process. The
survival ratio of C. parvum oocysts was determined using the CD-1 neonatal
mouse infectivity test and for C. perfringens and SSRC the survival ratio was
assessed with the standard anaerobic colony count on the iron-sulphite
medium.
The results of the lab-scale experiments revealed an inactivation rate
constant k (Chick-Watson modelling) at 10oC for C. parvum of 0.14 (SD =
0.014; P<0.001) and for C. perfringens of 0.25 (SD = 0.01; P<0.001). Moreover,
first results of monitoring the SSRC inactivation in full-scale ozonation
processes indicated that the inactivation rate constant for these wild strains
was in the same order of magnitude as determined for C. perfringens.
Further research is needed to compare inactivation ozone kinetics for C.
perfringens D10 and SSRC at different temperatures and in other natural
waters. Results of additional lab-scale experiments with C. perfringens strain
D10 indicated that the Ct of the gas-feed chamber should be incorporated
in the design of a full-scale ozonation. Moreover, setting the Ct with the
contact time was not as effective for the inactivation capacity as setting the
Ct with the ozone concentration.
INTRODUCTION
Outbreaks of waterborne diarrhoea caused by pathogenic protozoa Cryptosporidium and Giardia in the USA and the UK have increased interest in
the effect of water treatment processes on these micro-organisms. These
pathogens and especially Cryptosporidium can be present in drinking water
because they are persistent and resistant to chemical disinfection processes
in water treatment. Ozone is the only disinfectant, which may achieve a
significant inactivation capacity during treatment against Cryptosporidium
(Finch et al., 1993). The design of ozone processes for full-scale treatment
plants is based upon the results of bench-scale experiments using
artificially dosed oocysts under well-defined laboratory conditions. The
obtained Ct-inactivation relationship is used to design and operate fullscale ozone systems. Controlling ozone dose and contact time sets the
required Ct-values. However, the hydrodynamics of the contact chambers,
___________________________________________________________________
- 110 -
Chaper 5
ozone transfer and decay may not be scaled up readily from lab to fullscale. In this light, a tool to verify the validity of the use of the lab-scale
data for full-scale design would be valuable. Direct verification with
Cryptosporidium is not appropriate. The concentration of oocysts in the
incoming water is too low and dosage experiments on this scale are not
suitable. There is a need for a biological surrogate to determine the
achieved inactivation capacity and the variation of this capacity under fullscale conditions. A natural surrogate is preferable to the dosage of
fluorescent-dyed polystyrene microspheres as proposed by Mariňas et al.
(1999). Spores of sulphite-reducing clostridia (SSRC) have been proposed as
a surrogate for the assessment of the effect of water treatment processes on
persistent protozoa (Payment et al., 1993; Hijnen et al., 1997). One of the
arguments was their relatively high resistance against ozone observed
under full-scale conditions (Hijnen et al., 2000a) which can easily be
determined using a large volume sampling apparatus (Hijnen et al., 2000b).
The objective of this study was to verify the use of SSRC as a surrogate for
the estimation of the inactivation capacity of a full-scale ozonation for
Cryptosporidium.
METHODS
The experimental set up. The inactivation of Cryptosporidium parvum,
Clostridium perfringens and the natural SSRC by ozonation was determined
under lab-scale conditions and compared with the inactivation of the
natural SSRC in full-scale ozonation. To eliminate the effect of the water
quality on the inactivation kinetics (Haas et al., 1996 and Oppenheimer et
al., 2000) all experiments were carried in the influent of the full-scale ozone
process operated in a demonstration plant (100 m3/h) of Water Company
Europoort (WBE-water). This is river Meuse water, pre-treated with
impoundment reservoirs, microstraining, coagulation and sedimentation
and rapid granular filtration. The quality of this WBE-water shows no large
fluctuations for most of the relevant parameters (Table 1) and the
concentration of SSRC in this water range from 1 up 10 CFU per litre. For
the lab-scale experiments the water was transported in large stainless steel
vessels (30, 200 and 700 litre).
The ozone systems. Laboratory experiments with C. parvum and C.
perfringens were carried in a three-stage continuous-flow bench-scale
system of Montgomery Watson (MW) in the US (Figure 1 from
Oppenheimer et al., 2000). The experiments with C. perfringens and natural
SSRC were performed in the Kiwa continuous-flow system (Figure 2), a lab___________________________________________________________________
- 111 -
Chapter 5
.
scale duplicate of the full-scale WBE ozonation process. All technical
information of these three systems is summarised in Table 2.
Micro-organisms and microbiological analysis. For the experiments
with Cryptosporidium parvum the Harley Moon-isolate was produced in
neonatal Holstein calves and used not more than 2 month after isolation, as
described in Oppenheimer et al. (2000). For C. perfringens an isolate from a
patient suffering from diarrhoea caused by food infection was used (strain
D10).
Table 1 The range of values of the most relevant water quality parameters of the
WBE-water
Parameter
Average (10- and 90-percentile)
o
Temperature ( C)
13.2 (3.5 - 21.9)a
Turbidity (Ftu)
0.08 (0.012 - 0.15)
Fe (mg/l)
0.018 (0.05 - 0.045)
0.02 (<0.01 - 0.04)
NH4+ (mg/l)
PH
7.4 (7.27 – 7.54)
DOC (mg/l)
2.05 (1.75 – 2.36)
Uve (m-1)
5.3 (4.6 – 6.3)
1.57 (1.53 – 1.62)
Alkalinity (mmol CaCO3)
a Minimum and maximum temperature, respectively
Figure 1 Schematic of the MW-system (from Oppenheimer et al., 2000)
From a frozen stock culture (-70oC) in peptone/glycerol solution D10 was
inoculated anaerobically on Perfringens-agar-base plates (PAB Oxoid
CM587). After 5 days of incubation at 37oC the produced biomass was
suspended with 2 ml of sterile water in a 200 ml glass container with 150 ml
___________________________________________________________________
- 112 -
Chaper 5
sterile drinking water. The concentration of D10 in the two stock solutions
used in the experiments was 2x106 and 108 per ml, respectively. Pasteurized
(30 minutes at 70oC) and non-pasteurized samples of this solution yielded
similar concentrations, indicating a high percentage of sporulation. After
incubation for 20 - 25 days at 37oC these containers were stored at 3-5oC for 70
– 250 days. Before the inoculation of the WBE-water this D10 stock solution
was filtered through an 8 μm sterilized membrane filter (Schleicher&Schüll
AE99) to remove suspended solids and large aggregates of spores.
Table 2 Technical characteristics of the applied ozone systems
Characteristics
MW-system
Kiwa-system
WBE-system
a
a
b
Vol. chambers (l)
0.43
1.4 ; 2.6
4.100a; 4.100b
Water flow (l/h)
6.6
42
100.000
Gas flow (l/h)
3.6
12
26.000
T10/HRT (min./min.)
0.35
0.8
0.72
a Gas-feed Counter current Chambers (GCC); b Reactive Flow Chambers
(RFC)
The water samples before and after ozonation were collected in sterile bottles
with sterilized thiosulphate solution as a quenching agent. 10 ml/l of 1 %
solution and 2 ml/l of a 30 g/l thiosulphate solution were added to the
samples of the MW-system and of the Kiwa-system, respectively.
Figure 2 Kiwa-system used for the ozone experiments with spores of C. perfringens
and SSRC
___________________________________________________________________
- 113 -
Chapter 5
.
The inactivation of the oocysts of Cryptosporidium parvum was determined
with the CD-1 neonatal mouse infectivity test (Finch et al., 1993;
Oppenheimer et al., 2000). C. perfringens D10 and SSRC concentrations in
water were determined on sulphite-iron agar by the method previously
described (Hijnen et al., 1997). 1 or 0.1 ml samples with high concentrations of
D10 were inoculated directly in the liquefied medium, with or without
further dilution in 9 ml of sterile drinking water. The standard membrane
filtration method was used for sample volumes from 10 ml up to 10 litres and
for SSRC in large volumes up to 56 litre the MF-sampling technique (Hijnen et
al., 2000) was used.
Inactivation kinetics. Disinfection is the inactivation of microorganisms or the reduction of the concentration viable micro-organisms N
due to the exposure to a concentration disinfectant C during a specific contact
time t. The inactivation kinetic is most commonly described by the first order
disinfection model of Chick-Watson (1908):
dN
= −kC n N
dt
where k is the inactivation rate constant and n is the dilution constant. Based
on this model the linear relationship between the log inactivation of N and
the Ct-value ((mg/l)*min) is described by:
10
log(
Nt
) = −k * C n t
N
Nt is the microbial concentration after contact time t.
Ozone concentrations and CT-calculation. Ozone in water was
determined by the Indigo method (Bader and Hoigne, 1982). In the Kiwasystem ozone was measured by the potentiometric method with an
Orbisphere Indication instrument (model 26505) and sensor (model 2301)
calibrated by the Indigo method. The Ct-value (Ctcalc.) was calculated from
the average ozone concentration C (mg/l) and the contact time t (minutes)
in each of the successive chambers (including the gas-feed chamber). The
three in-series chambers of the MW-system can be considered as
continuously stirred tank reactors (CSTR) for which the average ozone
concentration in the chambers is the concentration monitored in the outlet
of each chamber (CO3,out). Ctcalc.-values of this system were derived from the
monitored inactivation values and the ozone concentrations by a model
specially developed by Montgomery-Watson for this system where the
hydraulic characteristics were taken into account (Oppenheimer et al.,
2000). For the plug-flow systems of Kiwa and WBE the average
___________________________________________________________________
- 114 -
Chaper 5
concentrations in the chambers were calculated from the in- and outcoming
concentrations (CO3,in + CO3,out)/2. To consider the influence of the contact
time distribution on the disinfection kinetics in the full-scale ozonation t10
was used which is defined as the contact time in which 90% of the
incoming water volume remains in the chamber (Von Huben, 1991). t10values of the different systems were verified with tracer tests (I- or Cl-) and
presented as ratio of the average hydraulic retention time (HRT) in Table 2.
RESULTS AND DISCUSSION
Inactivation of C. parvum and C.
perfringens. The inactivation of
C. parvum and C. perfringens by ozone in the MW-system was determined at
10oC in separately transported batches of the WBE-water. The values of the
water quality parameters were in the range presented in Table 1. Adjusting
the ozone dosage (constant contact time) set the Ct of the ozone process.
From the linear fit of the relation between the log inactivation and Ct-value
presented in Figure 3 the inactivation rate constant k for both microorganisms was calculated.
1
C. parvum MW system
Inactivation (log)
0.5
Clostridium perfringens MW system
C. perfringens Kiwa system
0
-0.5
-1
-1.5
-2
-2.5
-3
0
2
4
6
8
10
12
14
16
Ctcalculated ((mg/l)*min)
Figure 3 Linear relationship between the log inactivation and the Ctcalc.-value for C.
parvum and C. perfringens separately determined in WBE-water at 10oC in the MWsystem and for C. perfringens as well as the Kiwa-system, respectively (dashed lines
and bold line: the inactivation data for C. parvum in natural waters from
Oppenheimer et al., 2000 based on the minimum, maximum and median inactivation
rate constant at 10oC)
The k-value for C. parvum was 0.14 (SD = 0.014; r2 = 0.81; P<0.001). The kvalue determined for C. parvum in 16 different types of natural water
___________________________________________________________________
- 115 -
Chapter 5
.
(including the WBE-water) for a temperature of 10oC determined by
Oppenheimer et al. (2000) ranged from 0.08 up to 0.46 and the median value
was 0.21. Based on the water quality data no explanation was found for the
relatively low k-value determined for the WBE-water. From the
comparison of the disinfection data of both micro-organisms (Figure 3) it
was concluded that the inactivation kinetics of C. perfringens and C. parvum
were in the same order of magnitude. The k-value for C. perfringens was
0.25 (SD = 0.01; r2 = 0.96; P<0.001).
To verify the reproducibility of the inactivation kinetics for C. perfringens in
WBE-water the experiments carried out in the US were repeated in the Kiwasystem. The results revealed that for the involved range of Ctcalc.-values the
inactivation kinetics for both systems was similar. The k-value for the linear
inactivation kinetics determined in the Kiwa-system was 0.27 (SD = 0.016; r2 =
0.93; P<0.001).
Inactivation in the gas-feed chamber. In one of the experiments
with the Kiwa-system the log inactivation of C. perfringens in the Gas-feed
Counter current Chamber (GCC) was determined separately from the
overall log inactivation in the total system. The results revealed that a
proportional and significant part of the overall log inactivation was
obtained in the GCC (Figure 4). This indicates that also this chamber should
be incorporated in the design of an ozone process.
0
Log inactivation (log)
Gas Feed Chamber (GFC)
GFC+ Reactive Flow Chambers (RFC)
-0.5
-1
-1.5
-2
-2.5
-3
0
1
2
3
4
5
6
7
Ct10-value ((mg/l)*min)
Figure 4 The log inactivation of C. perfringens strain D10 in the gas-feed chamber
(GCC) and in the total Kiwa-system (GCC + RFC) as function of the Ct10
Variation of the contact time. In the Kiwa-system the inactivation
of C. perfringens strain D10 in WBE-water was measured at 10oC by varying
the Ct10-values of the ozone process with the contact time at a constant
___________________________________________________________________
- 116 -
Chaper 5
ozone dosage of 1.2 mg/l. Unlike the former results where Ct was set with
the ozone dosage, this experiment shows that the log inactivation did not
increase with the increasing Ct10-values (Figure 5).
From these results it was concluded that setting Ct-value with the contact
time under the described conditions with low ozone concentrations in the
reactive flow chambers (measured by the indigo-method and presented in
Figure 5) did not affect the inactivation capacity of the ozone process.
Consequently, setting the Ct-value and inactivation capacity with the ozone
dosage seems to be more effective.
0
Inactivation (log)
-0.2
-0.4
-0.6
Constant O3 concentration:
GFC = 0.270 (SD 0.03)
RFC1 = 0.095 (SD 0.04)
RFC2 = 0.050 (SD 0.02)
-0.8
-1
0.0
0.5
1.0
1.5
2.0
Ct10 ((mg/l)*min)
Figure 5 The relationship between Ct10 and the log inactivation of C. perfringens
strain D10 under the condition of a constant ozone dosage of 1.2 mg/l and varying
contact times of 4 up to 20 minutes in the Kiwa-system
Inactivation of C. perfringens and emvironmental SSRC. A
previously described study revealed that under full-scale conditions at 5
drinking water production plants in the Netherlands the log inactivation of
ozonation for SSRC easily could be assessed by large volume sampling
(Hijnen et al., 2000a). The processes were operated at a Ct-range of 1 up to 4
((mg/l)*min) and the average observed SSRC log inactivation in winter (310oC) and summer (10-20oC) was 0.7 (minimum and maximum log
inactivation ranged from 0 up to 1.2 log). In the period March and April
1998 at water temperatures of 8 – 11oC the SSRC inactivation in the fullscale WBE system was monitored intensively by sampling large volumes
(35-50 litre; 13 observations) with the MF-sampling technique. The average
log inactivation was 0.8 log (SD = 0.2; n = 13) and is plotted against the
average Ct10-value (n = 20) in this period of 1.5 (SD = 0.34) (mg/l)*min
___________________________________________________________________
- 117 -
Chapter 5
.
(Figure 6;•). In the same figure the average SSRC log inactivation
mentioned before for a number of full-scale processes (n = 5) was plotted
against the applied average Ct-values (o). For both data sets a wide range
of CT10- and log inactivation values was observed presented by an error
bar in the Figure (both parameters were not monitored simultaneously). A
comparison of these results with the inactivation kinetics of C. perfringens in
the WBE-water at 10oC show that the susceptibility to ozone of
envionmental SSRC under full-scale conditions and C. perfringens under
lab-scale conditions was in the same order of magnitude (Figure 6).
0.5
full-scale (Hijnen et al 2000)
full-scale WBE 8-11oC
0.0
Inactivation (log)
lab-scale 21oC
full-scale 21oC
-0.5
-1.0
Lab-scale results for
Cl. perfringens at
o
10 C (Figure 4)
-1.5
-2.0
-2.5
-3.0
0.0
1.0
2.0
3.0
4.0
5.0
6.0
7.0
Ct10-value ((mg/l)*min)
Figure 6 The relationship between the log inactivation of environmental SSRC and
C. perfringens in the WBE-water and the Ct10 of the full-scale and lab-scale ozone
processes (error bars indicate the range of log inactivation and Ct10)
Inactivation of environmental SSRC under full-scale and lab-scale
conditions. In the full-scale WBE-ozone process a SSRC log inactivation of
1.2 log (SD = 0.13) was observed at a water temperature of 21oC and a Ct10value of 1.0 (Figure 6;■). The inactivation of the environmental SSRC in
WBE-water was determined simultaneously in the lab-scale Kiwa-system, a
small copy of the full-scale WBE-system (□). The results indicate that the
inactivation kinetic under full- and lab-scale conditions was similar. The
relationship between the log inactivation and the Ct10-value in both systems
was described by the same linear fit (Figure 6). Moreover, comparison of
these results with the results obtained at 10oC indicate that a temperature
shift of 10oC will double the log inactivation for SSRC by ozone.
Oppenheimer et al. (2000) determined a larger effect of the temperature on
___________________________________________________________________
- 118 -
Chaper 5
the log inactivation rate of Cryptosporidium by ozone. This value increases
by a factor of 4.5 for every 10oC increase in temperature.
CONCLUSION
This study revealed that SSRC is an appropriate tool to study, control and
optimise disinfection processes with ozone under full-scale conditions for the
inactivation of Cryptosporidium. The inactivation rate constant k for
Cryptosporidium and SSRC determined at 10oC under lab-scale and full-scale
conditions in the WBE-water, respectively were in the same order of
magnitude. First results show a significant difference in the effect of
temperature on the inactivation kinetics for both micro-organisms. Further
research is needed before SSRC can be used as a surrogate for quantitative
assessment of the inactivation capacity of full-scale ozonation processes for
Cryptosporidium.
REFERENCES
Bader, H., and L. Hoigne. 1982. Determination of ozone in water by the Indigo
Method, A submitted Standard Method. Ozone Sci. and Eng. 4:169-176.
Finch, G. R., E. K. Black, L. Gyurék, and M. Belosevic. 1993. Ozone inactivation of
Cryptosporidium parvum in demand-fiee phosphate buffer determined by in vitro
excystation and animal infectivity. Appl. Environ. Microbiol. 59:4203-4210.
Haas, C. N., J. Joffe, U. Anmangandla, J. G. Jacangelo, and M. Heath. 1996. Water
quality and disinfection kinetics. J. Am. Water Works Assoc. 88:95-103.
Hijnen, W. A. M., W. M. H. van der Speld, F. A. P. Houtepen, and D. van der
Kooij. 1997. Presented at the International Symposium on Waterborne
Cryptosporidium, Newport Beach, California, US.
Hijnen, W. A. M., D. Veenendaal., W. M. H. Van der Speld, A. Visser, W.
Hoogenboezem, and D. Van der Kooij. 2000a. Enumeration of faecal indicator
bacteria in large water volumes using on site membrane filtration to assess water
treatment efficiency. Water Res. 34:1659-1665.
Hijnen, W. A. M., J. Willemsen-Zwaagstra, P. Hiemstra, G. J. Medema, and D.
van der Kooij. 2000b. Removal efficiency of full scale water íreatment processes
for spores of sulphite-reducing clostridia as a surrogate parameter for protozoan
(oo)cysts. Wat. Sci. Tech. 41:165-171.
Mariňas, B. J., J. L. Rennecker, S. Teefy, and E. W. Rice. 1999. Assessing ozone
disinfection with nonbiological surrogates. J. Am. Water Works Assoc. 91:79-89.
Oppenheimer, J. A., E. M. Aieta, R. R. Trussell, J. G. Jacangelo, and I. N. Najm.
2000. Evaluation of Cryptosporidium inactivation in natural waters. Am. Water
Works Assoc. Res. Foundation, Denver, US.
___________________________________________________________________
- 119 -
Chapter 5
.
Payment, P., and E. Franco. 1993. Clostridium perfringens and somatic coliphages as
indicators of the efficiency of drinking water treatment for viruses and protozoan
cysts. Appl. Environ. Microbiol. 59:2418-24.
von Huben, H. 1991. Surface Water: the new rules. Am. Water Works Assoc.
Denver, US.
Watson, H. E. 1908. A note on the variation of the rate of disinfection with change
in the concentration of the disinfectant. J. Hyg. 8:536-592.
___________________________________________________________________
- 120 -
Chapter 6
Inactivation credit of UV radiation for
viruses, bacteria and protozoan
(oo)cysts in water: a review•
W.A.M. Hijnen1, E.F. Beerendonk1 and G.J. Medema1
1 Kiwa
Water Research Ltd., P.O. Box 1072, 3430 BB Nieuwegein, NL
•
Reprinted from Water Research, 40: 3-22, Copyright 2006, with permission from
the copyright holder, Elsevier limited.
___________________________________________________________________
- 121 -
Chapter 6
.
ABSTRACT
UV disinfection technology is of growing interest in the water industry
since it was demonstrated that UV radiation is very effective against
(oo)cysts of Cryptosporidium and Giardia, two pathogenic micro-organisms
of major importance for the safety of drinking water. Quantitative
Microbial Risk Assessment, the new concept for microbial safety of
drinking water and waste water, requires quantitative data of the
inactivation or removal of pathogenic micro-organisms by water treatment
processes. The objective of this study was to review the literature on UV
disinfection and extract quantitative information about the relation
between the inactivation of micro-organisms and the applied UV fluence.
The quality of the available studies was evaluated and only high-quality
studies were incorporated in the analysis of the inactivation kinetics.
The results show that UV is effective against all waterborne pathogens. The
inactivation of micro-organisms by UV could be described with first-order
kinetics using fluence-inactivation data from laboratory studies in
collimated beam tests. No inactivation at low fluences (offset) and/or no
further increase of inactivation at higher fluences (tailing) was observed for
some micro-organisms. Where observed, these were included in the
description of the inactivation kinetics, even though the cause of tailing is
still a matter of debate. The parameters that were used to describe
inactivation are the inactivation rate constant k (cm2/mJ), the maximum
inactivation demonstrated and (only for bacterial spores and Acanthamoeba)
the offset value. These parameters were the basis for the calculation of the
Microbial Inactivation Credit (MIC = “log-credits”) that can be assigned to
a certain UV fluence. The most UV resistant organisms are viruses,
specifically Adenoviruses, and bacterial spores. The protozoon
Acanthamoeba is also highly UV resistant. Bacteria and (oo)cysts of
Cryptosporidium and Giardia are more susceptible with a fluence
requirement of <20 mJ/cm2 for a MIC of 3 log.
Several studies have reported an increased UV resistance of environmental
bacteria and bacterial spores, compared to lab-grown strains. This means
that higher UV fluences are required to obtain the same level of
inactivation. Hence, for bacteria and spores, a correction factor of 2 and 4
was included in the MIC calculation, respectively, whereas some
wastewater studies suggest that a correction of a factor of 7 is needed under
these conditions. For phages and viruses this phenomenon appears to be of
little significance and for protozoan (oo)cysts this aspect needs further
investigation. Correction of the required fluence for DNA-repair is
___________________________________________________________________
- 122 -
Chapter 6
considered unnecessary under the conditions of drinking water practice (no
fotorepair, dark repair insignificant, esp. at higher (60 mJ/cm2) fluences)
and probably also wastewater practice (fotorepair limited by light
absorption). To enable accurate assessment of the effective fluence in
continuous flow UV systems in water treatment practice, biodosimetry is
still essential, although the use of Computational Fluid Dynamics (CFD)
improves the description of reactor hydraulics and fluence distribution. For
UV systems that are primarily dedicated to inactivate the more sensitive
pathogens (Cryptosporidium, Giardia and pathogenic bacteria), additional
model organisms are needed to serve as biodosimeter.
INTRODUCTION
The first application of UV irradiation in drinking water as disinfection
process was in 1910 in Marseille (Henry et al., 1910), after the development
of the mercury vapour lamp and the quartz tube and establishing the
germicidal effect of UV irradiation. According to Wolfe (1990) and Hoyer
(2004) general application was hampered because of high costs, poor
equipment reliability, maintenance problems and the advent of
chlorination (cheaper, more reliable and potential to measure disinfectant
residual). Due to the increased information on the production of hazardous
oxidation by-products during chlorination and ozonation, UV irradiation
gained more attention; low pressure UV produces almost no by-products.
Also, unlike chemical disinfectants, the biological stability of the water is
not affected by low pressure lamps. In Europe, UV has been widely applied
for drinking water disinfection since the 1980’s, for the control of incidental
contamination of vulnerable groundwater and for reduction of
Heterotrophic Plate Counts (Kruithof et al., 1992). The breakthrough of UV
applicability as a primary disinfection process in the US and Europe came
after the discovery of the high efficacy of UV irradiation against
Cryptosporidium (Clancy et al., 1998) and Giardia. Chemical disinfection with
chlorine is not effective against these pathogens and ozone applied at low
CT-values to limit formation of bromate has relatively little effect on the
infectivity of the protozoan (oo)cysts. In contrast, infectivity of these
pathogens is significantly reduced by UV fluences that can readily be
applied in drinking water treatment. UV is now regarded as being broadly
effective against all pathogens, bacteria, protozoa and viruses, that can be
transmitted through drinking water.
The introduction of the Quantitative Microbiological Risk Assessment
(QMRA) to define the microbiological safety of drinking water (Haas, 1983;
___________________________________________________________________
- 123 -
Chapter 6
.
Regli et al., 1991; Teunis et al., 1997; Medema et al., 2003) is a development
of growing interest. Besides knowledge about the presence of pathogenic
micro-organisms in the source water, QMRA requires quantitative
knowledge about the capacity of water treatment processes, including UV
disinfection, to eliminate (remove or inactivate) pathogenic microorganisms.
The aim of this study was to evaluate available literature data and create a
well-defined database which enables calculation of the Microbial
Inactivation Credit (MIC) of UV disinfection for viruses, bacteria and
protozoan (oo)cysts in water. Most studies are lab-scale based. For fullscale chemical disinfection processes with chlorine and ozone it was
demonstrated that the MIC for E. coli was lower than expected from the
applied CT-values and the known dose-response curves determined under
laboratory conditions (Hijnen et al., 2000; 2004a). Therefore literature was
evaluated to verify the influence of process conditions on the MIC of fullscale UV-disinfection processes. The results of this review were also used
to identify further research needs.
MATERIALS AND METHODS
Selection of the reviewed literature. Literature on the inactivation of
viruses, bacteriophages, bacteria, bacterial spores and protozoan (oo)cysts
by UV irradiation in water was collected and evaluated on technical and
microbiological aspects. In the review, only those studies were used where
inactivation was assessed using generally accepted microbial culturing
(solid media or tissues) or animal infectivity methods. Similarly, only
studies were evaluated in which assessment of the UV fluence was clearly
described and based on either UV sensor measurements, fluence
calculations and/or Reduction Equivalent Fluence (REF) assessment with
(bio)dosimetry. The description and quality of the microbiological data,
quality assurance and reproducibility of the experimental data, as well as
the availability and quality of the technical and experimental conditions
were reviewed. Only those studies where process and experimental
conditions were well documented were used.
Experimental conditions. The inactivation of micro-organisms by
UV irradiation has been studied under different experimental conditions.
Many studies used a collimated beam apparatus (CB-tests) under benchscale and well-defined laboratory conditions. A volume of inoculated water
is irradiated during varying periods of time under a lamp emitting UV
light. Other studies used continuous flow systems (CF-systems) in a
laboratory, in a pilot- or demonstration plant or under full-scale conditions
___________________________________________________________________
- 124 -
Chapter 6
where the water passes a reactor with one or more UV lamps and UV
irradiation and contact time vary over the reactor.
Drinking water and wastewater studies. UV disinfection is applied
in the drinking water industry and for disinfection of treated wastewater.
Studies from both applications were reviewed and identified as such.
Papers on wastewater studies (WWS) usually describe disinfection of
secondary effluent with or without an additional pre-treatment. In the
drinking water studies (DWS), water is generally of much lower turbidity
and higher UV transmission than in waste water. When relevant, the
influence of these parameters is discussed.
UV fluence data. UV fluence cannot be measured directly, so it has
to be inferred from monitoring the UV irradiance with a UV sensor and the
time that the micro-organisms are exposed to UV. For the Collimated Beam
experiments, the average UV irradiance and contact time are wellcharacterised and have small confidence intervals. For continuous flow
systems the average UV fluence can be calculated from the same
parameters. However, the confidence intervals are much larger, due to the
much larger variation in contact time and in UV irradiation at different
points in the reactor, compared to the Collimated Beam experiments. By
modelling the hydraulic retention time in the UV reactor using
Computational Fluid Dynamics (CFD) accuracy of these calculations has
increased the last few years. Alternatively, the Reduction Equivalent
Fluence of CF-systems can be determined with biodosimetry (Qualls and
Johnson, 1983a; Sommer et al., 1999; Österreichisches normungsinstitut,
1999; Hoyer, 2004; USEPA, 2003). Biodosimetry is performed by
challenging the UV reactor with a micro-organism with calibrated UV
inactivation kinetics (biodosimeter) assessed with CB-tests in the test water.
With the measured inactivation of the biodosimeter and the calibration
curve, the Reduction Equivalent Fluence or REF (mJ/cm2) can be
calculated.
Inactivation kinetics of the micro-organisms. The inactivation
kinetics of a large number of pathogenic micro-organisms and indicator
micro-organisms that are significant to the microbial safety of water have
been calculated from studies where UV fluence has been determined under
optimal conditions: CB-tests with “drinking water” (low turbidity and high
UV transmission).
Inactivation by UV is based on the damage caused to the nucleic acids
(DNA/RNA) of the cell or virus. Primarily the formation of pyrimidine
dimers, but also of other photoproducts of nucleic acids and nucleic acid
lesions (von Sonntag et al., 2004), inhibit replication and transcription and
hence, prevent the cell or virus from multiplying. The UV absorbance of
___________________________________________________________________
- 125 -
Chapter 6
.
DNA peaks around 260 nm; at lower and higher wavelengths the
absorbance decreases. Below 230 nm the absorbance increases again. Most
studies used low-pressure mercury lamps with a major wavelength output
(85%) at 254 nm (monochromatic (MC) UV radiation) but for some microorganisms the UV inactivation was (also) determined with polychromatic
(PC) UV radiation from medium pressure lamps.
The UV sensitivity of the selected micro-organisms is described by the
parameters of the inactivation kinetics. Inactivation is defined as the
reduction of the concentration of culturable micro-organisms N due to the
exposure to a concentration disinfectant C during a specific contact time t.
The inactivation kinetic for chemical disinfectants is most commonly
described by the first-order disinfection model of Chick-Watson (1908) and
the same model can be applied for UV disinfection. The inactivation of
micro-organisms is usually described by the log inactivation of N. Based on
the first-order model the linear relationship between log inactivation and
the UV dose or fluence is described by:
10
log(
Nt
) = − k * Fluence
N
(1)
Nt is the microbial concentration after contact time t. Fluence is the product
of the UV fluence rate (mW/cm2) and the exposure t (mWs/cm2 = mJ/cm2).
In the literature, two main deviations from first-order UV disinfection
kinetics have been observed. Some authors (Knudson, 1985; Hoyer, 1998;
Sommer et al., 1998; Mamane-Gravetz and Linden, 2005a) observed no
inactivation of bacteria or bacterial spores at low UV fluences followed by a
normal log-linear relationship at higher UV fluences. This can be described
by a shoulder model and is presented by the following equation
DI = − k * Fluence − b
(2)
where DI is the decimal inactivation 10log(Nt/N), b is the y-intercept, a
negative value since the curve is crossing the fluence-axis at the UV fluence
where log-linear relationship starts (offset). The second deviation from the
linear kinetics is no further increase in inactivation at high fluences, called
tailing. Tailing is excluded in the k-value calculation in this review, by
excluding the inactivation data at higher fluences in studies where tailing
was observed (from the plots of inactivation versus fluence).
Influence of process conditions. The inactivation kinetics can be used to
determine the disinfection efficacy or Micro-organism Inactivation Credit
(MIC; log) of full-scale UV systems and to assess the fluence requirement to
obtain a certain MIC. However, for translation of CB-results to full-scale
___________________________________________________________________
- 126 -
Chapter 6
UV systems it is essential to know the effect of process conditions on the
efficiency of the radiation process. In contrast to oxidative disinfection
processes with chemicals like chlorine and ozone, the efficacy of UV
disinfection is not affected by conditions like temperature, pH (Severin et
al., 1983; USEPA, 2003) and reactive organic matter. UV absorbance by
organic and inorganic matter is included in the UV fluence calculation. But
the following factors may affect the efficiency of UV disinfection at fullscale:
- Factors related to the micro-organisms: physiological state (preculturing, growth phase), strain diversity, repair mechanisms and
particle association;
- Factors related to the fluence assessment: fluence-distribution due to
the distribution of the hydraulic retention time, adsorption, reflection
and refraction of UV light through the water and lamp intensity (aging
and fouling).
Several studies have addressed these aspects and are discussed to
determine whether adaptation of the required fluence for a certain MIC,
calculated by applying the UV sensitivity data to full-scale UV systems, is
required and, if possible, to quantify to what extent.
RESULTS AND DISCUSSION
UV sensitivity of micro-organisms. The UV sensitivity of the microorganisms is described with the inactivation rate constant k (cm2/mJ). A
UV sensitive micro-organism has a high k-value and requires a low fluence
for inactivation according to equation 1.
Viruses and bacteriophages. The number of studies where UV
sensitivity of specific pathogenic viruses and bacteriophages is determined
under well-defined laboratory conditions with collimated beam apparatus
(CB-tests) ranged from 1 to 6. The total number of data per virus or
bacteriophage ranged from 3 up to 109 (Table 1 and 2). The calculated kvalue (no shoulder; intercept = 0) showed a narrow 95% confidenceinterval (CI) and a high goodness-of-fit (13 out of 18 r2>0.85). The six
authors describing inactivation of seeded Poliovirus type 1 yielded a total
of 61 data points presented in Fig. 1. The inactivation rate constant k
calculated for a UV fluence range of 5 to 50 mJ/cm2 was 0.135 (95%CI=0.007; r2=0.79). Due to the observed tailing by Sommer et al. (1989) and
Maier et al. (1995), MICmax (the maximum observed MIC) is set at 5.4 log
(≥50 mJ/cm2).
___________________________________________________________________
- 127 -
.
___________________________________________________________________
- 128 -
Studies (n)
UV
ka (±95%CI; r2)
MICmax
UV fluence
2
(Log)
(mJ/cm )
b,c,d,e,f,g
Poliovirus type 1
6 (61)
5 – 50
MC
0.135 (0.007; 0.79)
5.4
Adenovirus ST2,15,40,41
5 (98)g,h,i,j,k
8 – 306
MC
0.024 (0.001; 0.87)
6.4
i
8 – 184
MC
0.018 (0.004; 0.88)
3.0
Adenovirus ST40
1 (29)
Adenovirus ST2,41
1 (18)k
30 – 90
PC
0.040 (0.003; 0.77)
4.3
5 – 50
MC
0.102 (0.006; 0.78)
4.1
Rotavirus SA-11
5 (55)b,d,e,k,l
5 – 30
PC
0.154 (0.011; 0.92)
4.6
Rotavirus SA-11
1 (11)k
4 – 49
MC
0.106 (0.010; 0.67)
5.5
Calicivirus feline, canine
3 (29)i,m,n
4 – 33
MC
0.190 (0.008; 0.96)
5.7
Calicivirus bovine
1 (20)k
Calicivirus bovine
1 (20)k
2 – 15
PC
0.293 (0.010; 0.97)
5.9
e,l,o
5 – 28
MC
0.181 (0.028; 0.70)
5.4
Hepatitis A
3 (13)
5 – 40
MC
0.119 (0.006; 0.97)
4.8
Coxsackie virus B5
2 (12)h,l
a linear regression, intercept = 0; b Chang et al., 1985; c Harris et al., 1987; d Sommer et al., 1989; e Wilson et al., 1992; f
Maier et al., 1995; g Meng and Gerba, 1996; h Gerba et al., 2002; i Thurston-Enriquez et al., 2003; j Thompson et al., 2003;
k Malley et al., 2004; l Battigelli et al., 1993; m De Roda Husman et al., 2003; n Duizer et al., 2004; o Wiedenmann et al.,
1993
Table 1 UV sensitivity of viruses for monochromatic (MC) and polychromatic (PC) UV radiation determined with collimated
beam tests
Chapter 6
5 (109)b,c,d,e,f
1 (11)f
4 (30)e,h,i,j
1 (4)d
1 (14)k
1 (3)k
1 (5)k
5 – 139
12 – 46
2 – 12
9 – 35
1 – 39
5 – 20
10 – 50
UV fluence (mJ/cm2)
MC
PC
MC
MC
MC
MC
MC
UV
0.055 (0.002; 0.93)
0.122 (0.009; 0.92)
0.396 (0.025; 0.85)
0.128 (0.014; 0.98)
0.140 (0.010; 0.96)
0.232 (0.080; 0.90)
0.084 (0.003; 0.99)
ka (±95%CI; r2)
MICmax
(Log)
4.9
5.3
4.0
3.8
5.6
4.6
4.2
a
___________________________________________________________________
- 129 -
linear regression, intercept = 0; b Havelaar et al., 1990; c Wilson et al., 1992; d Meng and Gerba, 1996; e Sommer et al.,
1998; f Malley et al., 2004; g Mamane-Gravetz et al., 2005b; h Battigelli et al., 1993; i Oppenheimer et al., 1993; j Sommer
et al., 2001; k Clancy et al., 2004
MS2-phages
MS2-phages
φX174
PRD1
B40-8
T7
Qβ
Studies (n)
Table 2 UV sensitivity of bacteriophages for monochromatic (MC) and polychromatic (PC) UV radiation determined with
collimated beam tests
Chapter 6
.
Fluence
(mJ/cm2)
2 – 10
0.5 – 6
0.6 – 5
1–5
3–8
0.6 – 4
1 – 12
0.5 – 3
1–7
1 – 15
1.5-9
2.5 - 16
5 – 78
48 – 64
MC
MC
MC
MC
MC
MC
MC
MC
MC
MC
PC
MC
MC
PC
UV
0.515 (0.047; 0.83)
0.880 (0.124; 0.65)
0.889 (0.060; 0.87)
1.308 (0.087; 0.95)
0.468 (0.053; 0.89)
1.341 (0.113; 0.94)
0.400 (0.040; 0.92)
1.079 (0.077; 0.99)
0.642 (0.082; 0.85)
0.506 (0.049; 0.71)
0.539 (0.070; 0.64)
0.312 (0.032; 0.85)
0.059 (0.007 0.91)
0.060 (0.027; 0.81)
k (±95%CI; r2)
Offseta
(mJ/cm2)
0
0
0
0
0
0
0
0
0
0
0
0
12.3
18
MICmax
(Log)
5.6
5.3
5.0
5.9
4.7
5.8
4.4
3.0
5.5
6.0
5.2
4.6
4.0
3.0
a
___________________________________________________________________
- 130 -
Offset is threshold-value >0: linear regression with intercept ≠ 0; b Chang et al., 1985; c Wilson et al., 1992; d Butler et
al., 1987; e Antopol et al., 1979; f Sommer et al., 2000a; g Harris et al., 1987; h,i Sommer et al., 1989, 1996a; j Zimmer et al.,
2002; k Oguma et al., 2002 ; l Sommer et al, 1998; m Hijnen et al., 2004b (continuous flow system)
Salmonella typhi
Campylobacter jejuni
Yersinia enterocolitica
Shigella dysenteriae
Shigella sonnei
Vibrio cholerae
Legionella pneumophila
Legionella pneumophila
Escherichia coli O157
Escherichia coli
Escherichia coli
Streptococcus faecalis
Bacillus subtilis
C. perfringens
Studies
(data)
2 (26)b,c
2 (27)c,d
2 (34)c,d
1 (9)c
1 (9)b
1 (10)c
1(15)c
1 (4)e
2 (16)c,f
6 (41)b,d,g,h,i,j
1 (23)k
2 (19)b,g
4 (30)b,h,l,m
1 (9)m
Table 3 UV sensitivity of bacteria and bacterial spores for monochromatic (MC) and polychromatic (PC) UV
radiation determined with collimated beam tests
Chapter 6
k (±95% CI; r2)
Range (mJ/cm2)
Intercept (95%)
MICmax
n
3
3
3
3
3
3
9
37
Ratio UVcalculated/REF (±SD)
1.33 (0.07)
0.81 (0.02)
0.79 (0.10)
1.00 (0.28)
0.73 (0.19)
0.62 (0.14)
0.59 (0.03)
0.83 (0.25)
___________________________________________________________________
- 131 -
System
Study
Model organism
B. subtilis
Aa
DWS
a
B. subtilis
B
DWS
B. subtilis
DWS
Ca
B. subtilis
Da
DWS
a
B. subtilis
DWS
E
B. subtilis
DWS
Fa
b
B. subtilis
DWS
G
MS2 phage
WWS
Hc
a Sommer et al., 2000b; b Hijnen et al., 2004b; c Watercare, 2002
Table 5 Calculated UV fluence versus fluence assessed with biodosimetry (REF)
C. parvum
0.243 (0.08; 0.49)
0.5 – 6.1; PC
1.502 (0.538)
3.0
6 (38)a,b,c,d,e,f
a,c,f,g
C. parvum
4 (65)
0.225 (0.07; 0.37)
0.9 – 13.1; MC
1.087 (0.403)
3.0
Giardia muris
0.122 (0.178; 0.81)
1.5 – 11; MC
1.303 (1.280)
2.4
1 (4)h
Giardia lamblia
nd
0.05 – 1.5; MC
nd
2.5
1 (2)i
j
0.021 (0.004; 0.94)
43 – 172; MC
0.499 (0.449)
4.5
Acanthamoeba spp.
1 (16)
a,bClancy et al., 2001, 2002; c Craik et al., 2001; d Shin et al., 2001; e Morita et al., 2002; f Rochelle et al., 2004; g Bolton et
al., 1998; h Craik et al., 2000; i Linden et al., 2002 ; j Maya et al., 2003
Studies (data)
Table 4 UV sensitivity of protozoa and Acanthamoeba spp. for monochromatic (MC) and polychromatic (PC) UV radiation
determined with collimated beam tests
Chapter 6
Chapter 6
.
In Fig. 1 the fluence-response curves for Adenoviruses serotypes (ST)2, 15,
40 and 41 are also presented. These data demonstrate that the UV
sensitivity of these serotypes for monochromatic UV radiation shows small
differences. Adenovirus is the most persistent virus type presented in Table
1. This conclusion is supported by the recently published study of
Nwachuku et al. (2005) who found k-values for serotypes 1 and 6 in the
same order of magnitude as for the types presented in Table 1. ThurstonEnriquez et al. (2003) found the lowest k-value of 0.018 cm2/mJ with
Adenovirus ST40. Malley et al. (2004) determined the UV sensitivity of
Adenovirus ST2 and ST41 for polychromatic (PC) UV radiation (medium
pressure (MP) lamps). Up to a UV fluence of 90 mJ/cm2, the UV sensitivity
was a factor of 1.7 higher than that observed for monochromatic (MC) UV
radiation (low pressure lamps) (Table 1), but above this fluence they
observed tailing (Fig. 1). By using bandpass filters they distinguished the
germicidal effect of different wavelengths in the polychromatic UV light at
fluence-ranges up to 90 mJ/cm2 and showed that at wavelengths of 220 and
228 nm UV was significantly (a factor of 5 – 7) more effective in
inactivating Adenovirus ST2 than UV light with a wavelength of 254 nm.
Poliovirus type 1
Inactivation (log)
8
7
Adenovirus type 2,15,40,41
8
y = 0.135x;
R2 = 0.79
y = 0.024x;
R2 = 0.87
7
6
6
5
Chang 1985
4
Watercare, 2002
y = 0.085x;
R2 = 0.86
3
2
Oppenheimer 1995
Oppenheimer 1995
Maier 1995
Watercare, 2002
0
50
100
150
UV fluence (mJ/cm2)
Meng 1996 ST40
Thompson 2003 ST2,15
Thurston 2003 ST40
Gerba 2002 ST2
Meng 1996 ST41
Malley 2004 ST41
Malley 2004 ST2,41 MP
3
2
1
Thompson 2003
0
y = 0.040x;
R2 = 0.77
4
Sommer 1989
Harris 1987
Wilson 1992
1
y = 0.018x;
R2 = 0.88
5
Meng 1996
0
200
0
100
200
300
UV fluence (mJ/cm2)
400
Figure 1 UV fluence-response curves for Poliovirus and Adenovirus (regular font
DWS, italic: WWS)
The fluence-response data and lines for Rotavirus type SA11 and for three
types of Caliciviruses are presented in Fig. 2. Again Sommer et al. (1989)
showed no further increase in inactivation at fluences above 50 mJ/cm2 and
Malley et al. (2004) showed that monochromatic UV radiation was less
efficient than polychromatic UV radiation for inactivation of rotavirus SA11. Using MP-lamps, the k-values were 1.7 times higher (Table 1).
Caliciviruses from different non-human hosts (feline, canine and bovine)
___________________________________________________________________
- 132 -
Chapter 6
showed highest UV sensitivity for the bovine type (Malley et al., 2004) and
the same study showed a 1.5 times higher inactivation with polychromatic
UV radiation compared to monochromatic UV. The k-value for inactivation
of feline Calicivirus (0.106 cm2/mJ) was in the same order of magnitude as
that observed for Rotavirus, Poliovirus and Coxsackie virus B5 (Table 1).
Hepatitis A virus was more sensitive to UV radiation.
Rotavirus SA-11
6
Calicivirus
6
2
Inactivation (log)
y = 0.102x; R = 0.78
5
5
4
4
3
3
Chang 1985
2
y = 0.190x
R2 = 0.96
y = 0.293x
2
R = 0.96
y = 0.106x
R2 = 0.67
Thurston, 2003 feline 9
Duizer 2004 canine 48
Duizer 2004 feline 9
de Roda 2003 feline 9
Malley 2004 bovine MP
Malley 2004 bovine LP
de Roda 2003 canine 48
2
Sommer 1989
Battigelli 1993
Wilson 1992
1
1
Malley 2004 LP
Malley 2004 MP
0
0
0
50
100
2
UV fluence (mJ/cm )
150
0.0
20.0
40.0
2
UV fluence (mJ/cm )
60.0
Figure 2 UV fluence-response curves for rotavirus and calicivirus
Noroviruses are part of the human caliciviruses and are not culturable.
Using RT-PCR, Watercare (2002) determined that environmental Norovirus
was less effectively inactivated by UV than the other viruses as determined
with culture assays. In their wastewater treatment plant, ten samples before
and 10 samples after UV were analysed; seven samples before UV and one
sample after UV were positive for Norovirus. The mean fluence was 23
mJ/cm2 . From these (presence/absence) data, a 0.8 log inactivation was
estimated for a UV fluence of 20 mJ/cm2. At higher fluences (40 and 70
mJ/cm2) all samples were negative. However, it is uncertain to which
degree inactivation assessed with RT-PCR is representative for inactivation
assessed with infectivity assays.
UV sensitivity of bacteriophages used or proposed as model organisms for
the assessment of the REF of a UV system on full-scale is also presented in
Table 2. MS2 phage is the most persistent of the tested phages with a kvalue of 0.055 cm2/mJ. k-Values of the other bacteriophages ranged from
0.128 cm2/mJ for PRD1 up to 0.396 cm2/mJ for PhiX174. The fact that the kvalues of these micro-organisms are in the same order of magnitude as
observed for most pathogenic viruses (compare Table 1 and 2) supports the
use of these micro-organisms as surrogates for virus inactivation by UV.
Only Adenoviruses are more resistant to UV.
___________________________________________________________________
- 133 -
Chapter 6
.
Recently, Mamane-Gravetz et al. (2005b) demonstrated that MS2 is more
(three times) sensitive to low wavelengths near 214 nm emitted by MPlamps compared to 254 nm output of LP-lamps, an observation in line with
those of Malley et al. (2004) for Adenoviruses as described before. The kvalues determined by Mamane-Gravetz et al. (2005b) at wavelengths of 254
and 214 nm were 0.055 and 0.161 cm2/mJ, respectively. These values are in
the same order of magnitude as the k-values calculated in this study for LPand MP-lamps, respectively (Table 2). The UV fluence of the MP-lamp in
the Malley study was calculated based on the average irradiance measured
by a UV-sensor and weighted by a germicidal factor at each wavelength
(based on the DNA absorbance, relative to 254 nm). Thus, the fluence of
LP- and MP-lamps was compensated for the wavelengths emitted by these
lamps. Malley et al. (2004) argued that this weighting may have been biased
for MP-lamps. On the other hand their results may indicate a higher
inactivation efficiency of MP-lamps compared to LP-lamps, a conclusion
supported by the observations of Mamane-Gravetz et al. (2005b).
Bacteria and bacterial spores. Bacteria (vegetative cells) are
significantly more susceptible to UV radiation than viruses and therefore
less extensively studied. In Fig. 3, fluence-response curves for some
selected pathogenic bacteria are presented. With the exception of E. coli in
five studies, only one or two studies were found for individual pathogenic
bacteria. Wilson et al. (1992) tested the UV sensitivity of seven of the ten
bacterial species presented in Table 3.
Inactivation (log)
7
7
y = 0.805x
R2 = 0.93
6
6
y = 1.784x
R2 = 0.85
5
5
4
y = 0.738x
R2 = 0.99
y = 0.889x
R2 = 0.87
4
y = 0.515x
R2 = 0.83
3
2
y = 0.737x - 1.127
R2 = 0.94
2
Salmonella Wilson 1992
E.coli O157 Sommer 2000
Salmonella Chang 1985
1
3
E.coli O157 Wilson 1992
1
Campylobacter Butler 1987
Yersinia Butler 1987
Campylobacter Wilson 1992
0
Yersinia Wilson 1992
0
0
5
2
UV fluence (mJ/cm )
10
0
5
UV fluence (mJ/cm2)
10
Figure 3 UV fluence-response curves for pathogenic bacteria
The number of data points ranged from 4 up to 41 for E. coli. The k-values
varied from 0.312 cm2/mJ for Streptococcus faecalis to 1.341 cm2/mJ for
Vibrio cholerae (both Wilson et al., 1992). Linear regression analysis showed
___________________________________________________________________
- 134 -
Chapter 6
low variation (95% confidence interval) and high goodness-of-fit (r2).
Sensitivity of Legionella pneumophila published in literature was highly
variable. From the data of Antopol et al. (1979) and Wilson et al. (1992) a kvalue of 1.079 and 0.400, respectively, was determined (Table 3). Knudson
(1985) published a higher sensitivity (k = 1.916) and the k-value presented
by Oguma et al. (2004) was 0.62. The latter author also demonstrated that
the sensitivity of both L. pneumophila and E. coli to monochromatic and
polychromatic was similar (Table 3).
Aerobic spores of Bacillus subtilis and anaerobic spores of Clostridium
perfringens are clearly less sensitive to UV than the vegetative bacterial
cells (Table 3) and also most of the viruses and phages (Table 1 and 2). The
data on the UV sensitivity of C. perfringens were derived from a continuous
flow-system with medium pressure lamps, in which the UV fluences were
determined with biodosimetry (REF) with UV calibrated spores of B.
subtilis.
Pathogenic protozoa. Interest in UV as a disinfection process for
water has increased after Clancy et al. (1998) showed that Cryptosporidium
parvum oocysts were highly susceptible to UV when the effect on the
infectivity was assessed with the neonatal mouse model. Since then, Clancy
and several other authors have studied inactivation of Cryptosporidium
parvum and Giardia muris by UV radiation (Table 4). Figs. 4 and 5 show
substantial inactivation of (oo)cysts of both protozoa at low UV fluences
(<20 mJ/cm2) by LP- and MP-lamps. Recently, Johnson et al. (2005)
demonstrated a similar UV sensitivity for C. hominis oocysts which
predominates in human cryptosporidiosis infections. Based on the
regression analysis of these fluence-response data, efficacy of LP- and MPlamps for oocyst inactivation is in the same order of magnitude (Table 4).
Comparison of these k-values with the k-values from Table 1 and 2 show
that these protozoa are more sensitive to UV than viruses, but less sensitive
compared to most bacteria. The regression analysis of the accumulated data
shows a low goodness-of-fit (r2 = 0.37; 0.49 and 0.81) and positive intercept
values. Furthermore, Craik et al. (2000, 2001) observed considerable tailing
for a number of inactivation data at high UV fluences (Fig. 4c). Qian et al.
(2004) described the protozoan data with a statistical method (Bayesian
meta-analysis) which resulted in the UV fluence requirement curves
presented in Fig. 4 and 5 (USEPA, 2003). These curves were calculated for
an inactivation requirement of up to 3 log and could be described by a loglog relationship (Log inactivation of Giardia = 1.2085 LN UV fluence +
0.0715; r2 = 0.99; Log inactivation of Cryptosporidium = 1.2344 LN UV
fluence – 0.1283; r2 = 0.99).
___________________________________________________________________
- 135 -
Chapter 6
0
1
2
3
0
10
UV fluence (mJ/cm )
2
20
Clancy 2000
Clancy 2002
Craik 2001
Shin 2001
Morita 2002
Rochelle 2004
0
1
2
3
4
4
6
5
USEPA (2003)
Low-pressure
5
6
50
100
150
Bolton 1998
Clancy 2000
Craik 2001
Rochelle 2004
UV fluence (mJ/cm2)
0
USEPA (2003)
Medium-pressure
.
0
1
2
3
4
5
6
50
100
150
UV fluence (mJ/cm2)
0
Craik 2001 MP
Craik 2001 LP
Fluence > 20
___________________________________________________________________
- 136 -
Figure 4 UV fluence-response curves for Cryptosporidium parvum (multiple strains) and LP- and MP-lamps (a, b) and (c)
tailing in the inactivation data observed by Craik et al., 2001 for fluences above 20 mJ/cm2
Inactivation (log)
Chapter 6
Inactivation (log)
5
4
USEPA 2003
3
2
Craik 2000 Giardia muris
Maya 2003 Acanthamoeba ssp.
Karanis 1992 A. Rhysodes
Karanis 1992 A.quina/lugdunensis
Wolfe 1990 A.castellantii
Chang 1985 A.castellantii
1
0
0
50
100
150
2
UV fluence mJ/cm
200
Figure 5 UV fluence-response curves for Giardia muris and Acanthamoeba
One study described the inactivation of Acanthamoeba spp. by UV in CBtests using CD1 neonatal mouse model test to measure infectivity (Maya et
al., 2003). Just as observed for B. subtilis, an offset UV fluence is required for
this organism to see an effect on infectivity of these pathogens. This offset
value of 30 mJ/cm2 as well as the low k-value of 0.021 calculated from the
successive log-linear relationship, show that the sensitivity of this microorganism and of the most resistant virus type Adenovirus to UV are in the
same order of magnitude.
Process conditions. The k-values summarized in Tables 1, 2 and 3
can be used to determine the scale of new full-scale UV treatment processes
or to calculate the inactivation efficiency of operational UV systems.
Translation of UV sensitivity assessed with CB-tests and seeded microorganisms to the efficiency of UV disinfection under full-scale conditions,
however, is influenced by factors related to the micro-organisms and by
factors related to the fluence assessment. This is similar to the translation of
lab-scale tests for chemical disinfection to full-scale, as illustrated for ozone
by Smeets et al. (2005; 2006). The literature on the influence of these factors
has been reviewed. For pathogenic micro-organisms two studies were
___________________________________________________________________
- 137 -
Chapter 6
.
found (Fig. 6) that investigated the inactivation of environmental
pathogenic micro-organisms under different conditions. For indicators,
more studies were available. Most of the evaluated data came from
wastewater studies and to a lesser extent from drinking water studies.
Inactivation of seeded or environmental indicator micro-organisms
(coliforms, enterococci, clostridia spores, FRNA-phages, Bacillus spores) has
been determined in either CB-apparatus or CF-systems. Inactivation data
are presented in Fig. 7-11, where CB-test results are separated from results
from CF-systems. The findings are reviewed in the two following
paragraphs.
Micro-organism related factors. The UV sensitivity of seeded and
environmental micro-organisms is compared in some studies under
identical conditions or by comparing results from a study with
environmental organisms with the overall data for seeded organisms tested
in CB-apparatus or CF-systems (DWS, Fig. 6-9).
Salmonella
Inactivation (log)
7
Adenovirus
5
6
CF-system environ. WWS ST1,2,3,4,7 REF MS2
CF-system environ. WWS Non detect REF MS2
CB tests, DWS studies (Fig. 1)
4
5
Watercare 2002
y = 0.040x
2
R = 0.11
3
4
y = 0.038x + 2.067
2
R = 0.98
3
2
2
y = 0.172x
2
R = 0.87
1
y = 0.024x
2
R = 0.87
1
CB-test environ. WWS, Maya 2003
CB-test Seeded DWS, Ref. Table 3
0
0
0
20
40
60
2
UV fluence (mJ/cm )
80
0
20
40
60
80
100
2
UV fluence (mJ/cm )
Figure 6 Comparison of UV fluence-response curves for seeded and environmental
Salmonella and Adenoviruses
Environmental Salmonella, faecal coliforms and enterococci in CB-tests in
wastewater (Maya et al., 2003) were more resistant to UV light than the
seeded micro-organisms of the same species (DWS, Fig. 6, 8 and 9). A
higher UV resistance of environmental spores compared to seeded spores
(which were surviving isolates from the environmental spores) was also
observed for Bacillus spp. (Mamane-Gravetz et al., 2005a; CB-tests) and
sulphite-reducing clostridia SSRC (Hijnen et al., 2004b; CF-system), both
DWS (Fig. 10). A higher resistance to UV of environmental bacteria is also
demonstrated by the inactivation data of thermotolerant coliforms in the
WWS studies of Watercare (2002) and Gehr and Nicell (1996) as shown in
___________________________________________________________________
- 138 -
Chapter 6
Fig. 8. Based on the difference in k-value between the Watercare data and
the DWS data assessed in CB tests, the k-value decreased a factor of seven
(from 0.506 to 0.066; Fig. 8).
CB-tests
Inactivation (log)
6
Watercare 2002; MS2
y = 0.049x;
R2 = 0.79
Ref. Table 2
y = 0.055x;
R2 = 0.93
5
4
Nieuwstad 1994
y = 0.025x + 0.87
R2 = 0.94
Watercare 2002
y = 0.041x
R2 = 0.40
5
4
Gehr 2003
y = 0.051x
R2 = 0.99
3
Continuous flow
6
Watercare 2002; FRNA
y = 0.047x;
R2 = 0.81
Havelaar 1987
y = 0.092x
R2 = 0.98
3
2
2
MS2 DWS
MS2 WWS UVT 39-46%
FRNA WWS UVT 39-46%
MS2 WWS UVT ni
1
0
0
50
100
150
MS2 WWS calc. Fluence
FRNA WWS, calc. Fluence
FRNA , WWS calc. Fluence
1
0
0
200
2
50
100
150
2
200
UV fluence (mJ/cm )
UV fluence (mJ/cm )
Figure 7 UV fluence-response curves for seeded MS2 FRNA phages and
environmental FRNA phages determined under different conditions (ni = no
information)
CB-tests
Continuous flow
8
8
E. coli seeded; DWS, CF-systems
Inactivation (log)
y = 0.506x;
2
R = 0.73
E. coli environ.; WWS, calc. Fluence, Havelaar 1987
Therm. coliforms;WWS, REF MS2, Watercare 2002
Therm. coliforms;WWS, calc.fluence, Gehr 1996
6
6
y = 0.187x - 1.154;
2
R = 0.98
4
y = 0.199x
2
R = 0.50
4
y = 0.218x;
2
R = 0.95
2
y = 0.168x;
2
R = 0.53
y = 0.066x;
R2 = 0.53
2
E. coli seeded; DWS, ref. Table 3
E. coli sew. sludge isolate; DWS, Sommer 1998
Therm. coliforms; WWS, Maya 2003
Total coliforms; WWS, Chang 1985
0
0
25
50
75
0
2
UV fluence (mJ/cm )
100
0
25
50
75
2
100
UV fluence (mJ/cm )
Figure 8 UV fluence-response curves for seeded and environmental coliforms (E.
coli, thermotolerant coliforms and total coliforms) determined under different
conditions
In a study with a CF-system (LP-lamps) operated at a fluence of 25 mJ/cm2
complete inactivation of environmental Poliovirus (type 1) in a chalk well
___________________________________________________________________
- 139 -
Chapter 6
.
water was observed with an estimated inactivation of more than 2.3 log
(Slade et al. 1986; 21 samples of 0.15-1 m3 tested over a period of one year).
This might indicate little or no difference in sensitivity between
environmental and lab-cultured Polioviruses; the latter are inactivated with
3.4 log at this fluence calculated from the k-value of Table 1. In the WWSstudy of Watercare (Watercare,2002; Simpson et al., 2003; Jacangelo et al.,
2004) the UV sensitivity of environmental F-specific RNA phages (FRNA)
was comparable to the UV sensitivity of seeded MS2 phages tested under
similar conditions (CB-tests; Fig. 7). A k-value of 0.049 cm2/mJ was
calculated for the seeded MS2-phages while for environmental FRNA
tested under similar conditions in CB-tests resulted in a k-value of 0.047
cm2/mJ (Fig. 7). In contrast, a lower UV resistance of environmental
Adenoviruses was observed in a CF-system (Watercare study) compared to
seeded Adenoviruses tested in DWS-studies (CB-tests, Fig. 6). Because 35%
of the observations in this study yielded a higher inactivation than could be
detected (>2.7 up to >3.3 log at a fluence range of 23 up to 50 mJ/cm2; Fig.
6), the difference is even larger; these data have not been used in the kvalue calculation presented in Fig. 6. Predominant Adenovirus types in the
Watercare study were serotypes 1, 2, 3, 4 and 7, with less commonly
serotypes 5, 8, 11, 13, 15, 19, 25 and 29. The higher susceptibility of the
environmental Adenovirus in this study could be the result of the absence
of Adenovirus, type 40, the most persistent serotype.
CB-tests
6
Continuous flow
6
Inactivation (log)
Enterococci CF-system WWS, REF MS2, Watercare, 2002
2
y = 0.312x; R = 0.85
5
Enterococci CF-system WWS, calc. Fluence, Havelaar 1987
5
4
4
3
3
y = 0.146x - 0.316
2
R = 0.99
2
y = 0.070x
2
R = 0.55
y = 0.131x
2
R = 0.83
2
1
1
S. faecalis CB-test DWS, ref. Table 3
Enterococci CB-test WWS, Maya, 2003
0
0
0
20
40
60
80
2
UV fluence (mJ/cm )
0
20
40
60
80
2
UV fluence (mJ/cm )
Figure 9 UV fluence-response curves for seeded and environmental enterococci
determined under different conditions
___________________________________________________________________
- 140 -
Chapter 6
Overall, increased UV resistance of environmental micro-organisms was
more explicit for the bacterial spores and for vegetative bacteria, and was of
less significance for FRNA-phages and viruses. To the authors knowledge
no data have been published on the UV sensitivity of environmental
(oo)cysts of Cryptosporidium and Giardia. The observed difference in UV
sensitivity for bacteria and spores may be attributable to the physiological
state of the micro-organisms, strain diversity, DNA-repair mechanisms and
particle association. These factors are discussed in more detail below.
a) Physiological state. The physiological state of micro-organisms
affects the sensitivity to environmental stress factors such as UV
radiation. Martiny et al. (1990), Mofidi et al. (2002) and Malley et al.
(2004) showed that UV sensitivity was related to the growth-phase
of the bacteria with the highest sensitivity in the active growth
phase and lower sensitivity in the stationary phase.
CB-tests: Bacillus spp.
Inactivation (log)
6
5
CB and CF: Clostridia spp.
6
Qualls 1983 ATCC6633
Chang 1985 ATCC6633
Sommer 89, 98, 96a ATCC6633
y = 0.059x - 0.738
Wild strain
Environ.
R2 = 0.91
C. perfringens seeded MP-CF, DWS REF B. subtilis
SSRC environ. MP-CF, DWS REF B. subtilis
5
C. perfringens environ. LP-CF, WWS REF MS2
C. perfringens environ. LP-CB, WWS
4
4
Hijnen 2004b
y = 0.060x - 1.077
R2 = 0.81
3
3
2
Gehr 2003
y = 0.027x
R2 = 0.82
2
y = 0.018x
R2 = 0.91
1
1
Watercare 2002
2
y = 0.021x;R = 0.55
0
0
0
50
100
2
UV fluence (mJ/cm )
150
Hijnen 2004b
y = 0.022x
2
R = 0.66
0
25
50
75
100
UV fluence (mJ/cm2)
125
Figure 10 UV fluence-response curves for seeded and environmental bacterial
spores determined under different conditions
b) Strain variation. Different strains of one species may have different
UV sensitivity, as demonstrated for E. coli by Sommer et al.,1998,
Sommer et al., 2000a (Fig. 8) and Malley et al. (2004). UV sensitivity
of different E. coli strains in these studies varied by a factor of 5.8
and 3.7, respectively. The latter study demonstrated a higher
sensitivity of E. coli O157:H7 compared to non-pathogenic/toxic
strains. In contrast, Clancy et al. (2002) and Rochelle et al. (2004)
showed that the high inactivation efficiency of UV radiation for
Cryptosporidium was observed in multiple strains of C. parvum. The
similar UV sensitivity observed for C. hominis (Johnson et al., 2005)
___________________________________________________________________
- 141 -
Chapter 6
.
suggests that this high sensitivity of Cryptosporidium oocysts is
common for all sub-species.
c) Repair. Exposure to UV results in damage to the nucleic acids of the
cell. Although also other components of the cell may be damaged
by UV, micro-organisms may still retain metabolic functions such as
enzyme activity. Over time, organisms have developed mechanisms
to repair DNA damage as a result of exposure to UV from the sun.
The mechanisms of repair are comprehensively described in von
Sonntag et al. (2004). Two types of repair have been described: darkrepair and photo-reactivation.
Dark-repair does not require light and has been demonstrated in
almost all bacteria. Spores have no active metabolism, but repair
starts upon germination. Viruses have no metabolism so cannot
repair damage to their genome themselves. However, several
viruses have been shown to use the repair enzymes of the host cell.
This is suggested as the cause of the high resistance of Adenovirus,
a double-stranded DNA virus, which can use the host cell’s repair
mechanism, while RNA-viruses may not. Some viruses even carry
the genes for repair-enzymes (Lytle, 1971; refs. von Sonntag), but
this is not the case for viruses that are transmitted via water.
Photo-reactivation occurs in conditions of prolonged exposure to
(visible) light and is specifically targeting pyrimidine-dimers. For
bacteria several CB studies demonstrated repair after light
exposure. The significance of this phenomenon to the required
fluence to achieve a certain inactivation can be deduced from the
influence of repair on the inactivation kinetics (dose-response
curves), but also from the occurrence of these repair mechanisms
under conditions of disinfection practice.
Most photo-reactivation studies with CB tests used low fluences
and optimal conditions for light exposure for repair (thin layer of
fluid). The results show that under these conditions fluence
requirement increases with increasing fluence (lower k-values).
Quantitative data showed a 2.8 – 4.6 higher UV fluence requirement
for 1 - 3 log inactivation of Legionella pneumophila (Knudson, 1985);
based on these data k-value decreased a factor of 3.2. Oguma et al.,
(2004) observed a comparable log-repair at an initial inactivation of
3 log after UV disinfection with LP- and MP-lamps and complete
photo-reactivation. For E. coli Bernhardt (1994) showed an increased
offset value and decreased k-value. For several bacteria spp. (E. coli,
Yersinia enterocolitica, Salmonella typhi and Vibrio cholera) he
calculated an increased fluence requirement for a 4 log inactivation
___________________________________________________________________
- 142 -
Chapter 6
ranging from a factor of 1.8 up to 4.2 to account for complete photoreactivation. For E. coli similar enhancement of fluence requirement
was observed by Hoyer (1998; 3.5 times for 4 log inactivation).
Sommer et al. (2000a) showed a decrease in k-value after photoreactivation and also demonstrated that dark-repair is of less
importance for E. coli. The latter observation was confirmed by
Zimmer et al. (2002), who demonstrated that photo-reactivation of
E. coli did not occur after MP-lamps, an observation also supported
by Oguma et al. (2002; 2004).
Morita et al. (2002) demonstrated photo-reactivation and dark-repair
of DNA in Cryptosporidium parvum with the endonucleasesensitivity site assay. The animal infectivity, however, was not
restored. Furthermore, they concluded that UV radiated oocysts are
able to excyst but have lost their ability to infect host cells. Similar
observations were reported by Shin et al. (2001) and Zimmer et al.
(2003). Craik et al. (2000) and Linden et al. (2002) came to the same
conclusion for Giardia muris and Giardia lamblia cysts, respectively.
Belosevic et al. (2001), however, showed the ability of DNA-repair
by some Giardia spp. after UV radiation with MP-lamps. This was
also presented by Kruithof et al. (2005); in vivo reactivation (darkrepair) of G. muris cysts occurred at fluence values as low as 25
mJ/cm2, but not at 60 mJ/cm2, after prolonged time of incubation
(3, 14 and 20% reactivation after 10, 20 and 30 days, respectively).
Also DNA-repair of G. lamblia cysts after exposure to
monochromatic UV irradiation is recently reported (Shin et al.,
2005). An extensive study was presented on repair in C. parvum
oocysts by Rochelle et al. in 2004. Identification of possible DNA
repair genes in C. parvum showed that the oocysts contain all of the
major genetic components of the nucleotide excision repair
complex. Nevertheless, inactivation displayed by oocysts
immediately after UV exposure or displayed by oocysts after UV
exposure followed by various repair conditions were generally in
the same order of magnitude. This suggests that repair of UV
induced damage in C. parvum after UV exposure in drinking water
is not likely to occur.
d) Particle association. Higher resistance of particle associated faecal
bacteria has been observed in secondary effluents. This
phenomenon was demonstrated by Qualls et al. (1983b) and
Havelaar et al. (1987) for thermotolerant coliforms and the
enterococci, respectively. A lower inactivation rate was observed in
non-filtered effluent of sewage water plants compared to filtered
___________________________________________________________________
- 143 -
Chapter 6
.
samples (pore size 8 μm). Recently, Örmeci and Linden (2002)
applied different techniques (extraction with EGTA, filtration of 5
μm filters with or without homogenization by blending) to separate
particle- and non-particle associated coliforms and showed an
increased resistance of environmental coliforms associated with
particles to UV (Fig. 11). Aggregates of B. subtilis spores were
artificially made with clay in a Jar Test apparatus by MamaneGravetz and Linden (2004) and caused a reduction in inactivation
efficiency. The k-value decreased from 0.0617 for the suspended
spore-clay solution to 0.0579 cm2/mJ for the aggregated spore-clay
suspension. The same authors published a new study in 2005, in
which they found evidence for a correlation of hydrophobicity of
spores with aggregation. Aggregation may be a cause of tailing (no
further increase of inactivation at higher fluence) observed in the
kinetics. The k-values of isolated environmental Bacillus strains in
the tailing phase of the kinetics were similar to the k-values of the
original and natural Bacillus spore population. This indicates that a
shielding effect of aggregation or particle association is a significant
factor in the low susceptibility of environmental Bacillus spores, and
also observed for the environmental clostridia spores as presented
in Fig. 10.
8
Seeded viruses and phages
Inactivation (log)
7
y = 0.049x
2
R = 0.77
6
y = 0.083x
2
R = 0.79
5
Environmental coliforms
5
Detection-limit
4
3
4
2
3
Watercare 2002:
örmeci 2002:
Polio 2.5 NTU; UV254 44%
Polio 0.2 NTU; UV254 46%
MS2 5.3 NTU; UVT254 39%
MS2 2.5 NTU; UVT254 44%
MS2 0.2 NTU; UVT254 46%
2
1
0
0
100
2 200
UV fluence (mJ/cm )
Untreated
1
EGTA extraction
5 um filtration
5 um filtration and blending
0
0
50
UV fluence (mJ/cm2)
100
Figure 11 Effect of water quality and association with particles on UV fluenceresponse curves for viruses and phages and for environmental coliforms,
respectively
___________________________________________________________________
- 144 -
Chapter 6
Fluence related factors. Variability in fluence may be caused by
water quality (adsorption, reflection and refraction) and the distribution of
the hydraulic retention time in continuous flow systems.
a) Water quality. The presence of UV absorbing organic and inorganic
compounds in water will reduce the UV fluence but results showed
that fluence values can easily be corrected for the UV transmission
of the water. Havelaar et al. (1990) placed 0.22 μm membrane
filtered secondary effluent (UV transmission of 40-60%) in the UV
pathway of the collimated beam apparatus and showed no decrease
of the inactivation rate constant for MS2 phages after correction of
the fluence for the transmittance. In studies with CF-systems both
Schoenen et al. (1995) and Sommer et al. (1997) showed that the
inactivation efficiency assessed at similar sensor readings was more
sensitive for change in lamp intensity than for change in the water
transmittance.
The influence of water quality on the efficiency of UV disinfection
can be demonstrated by comparing results of drinking water
studies (DWS) with wastewater studies (WWS) conducted under
similar conditions. CB-tests with seeded Polioviruses in secondary
effluent with high turbidity and low UV transmission were
published by Oppenheimer et al. (1995), Watercare (2002) and
Thompson et al. (2003) (Fig. 1). In the first two studies a lower
inactivation was observed in wastewater than in drinking water
studies. In the Watercare study, the inactivation rate constant k of
seeded Poliovirus in filtered secondary effluent was a factor 1.6
lower than the k-value in calculated from the drinking water studies
(Fig. 1). The same study (Watercare, 2002) observed a slightly lower
inactivation of MS2 in secondary effluent compared to the
inactivation of MS2 in drinking water at the same fluence (Fig. 7).
CB-tests were conducted with secondary effluent with high and low
turbidity (0.2 - 2.5 NTU and UVT of 40 - 68%). They demonstrated
that there was no impact of turbidity on the inactivation of seeded
Polioviruses and MS2 phages (Fig. 11). In addition, data from the
WWS-study of Thompson et al. (2003) showed no decreased
inactivation of seeded Poliovirus and Adenovirus compared to
inactivation of these organisms tested in DWS-studies (Fig. 1).
b) Fluence determination in CF-systems. In the literature, only few
drinking water studies have been published where fluence-response
curves were determined with continuous flow systems (CFsystems). Results from the studies of Martini et al. (1990), Schoenen
et al. (1991) and Bernhardt et al. (1992, 1994) showed lower
___________________________________________________________________
- 145 -
Chapter 6
.
inactivation rate constants for E. coli (Fig. 8) when compared to the
k-value determined from CB-tests. The fluence in the CF-systems in
these studies were based on information of the supplier of the UV
equipment or on actinometry.
Information about the precision of fluence calculations can be
obtained with biodosimetry. Spores of Bacillus subtilis and MS2
phages are used as model organisms in biodosimetry assays to
assess the Reduction Equivalent Fluence (REF) of CF-systems.
Sommer et al. (2000b) determined the REF with UV254 calibrated
spores of B. subtilis of more than 30 commercial available CFsystems and presented the results of six systems. In one system, the
REF was equal to the UV fluence calculated according to the
supplier’s instructions, four systems showed that that REF was 19 –
38% lower than the calculated fluence and in one system REF was
33% higher than the calculated fluence (Table 5). The average ratio
of the REF to the calculated fluence was 0.83 with a relatively high
standard deviation of 0.25 (Table 5). The overestimation of the
effective fluence (REF) by fluence calculation was supported by data
presented by Hijnen et al. (2004b). In conclusion, calculated fluence
data in CF-systems frequently do not match those obtained by
biodosimetry. Biodosimetry is essential to determine the efficacy of
CF-systems (DVGW, 1997; Sommer et al., 2000b; USEPA, 2003). The
introduction of Computational Fluid Dynamics for fluence
calculations (no data presented) is improving the quality and
precision of fluence calculations.
c) Reflection. Reflection caused by the construction materials of the
UV reactor will have an influence on the inactivation efficiency
determined by biodosimetry (Sommer et al., 1996b). This factor is of
greater influence in single lamp systems than in multiple lamp
systems because of the higher surface-volume ratio.
GENERAL DISCUSSION
Kinetics of UV inactivation. Most of the inactivation data can be
adequately described with the first-order disinfection model, at least for a
certain fluence range. An offset UV fluence before inactivation starts, i.e., a
shoulder model, is observed for Bacillus spores and Acanthamoeba spp. The
simple inactivation model, where the shoulder is given as an offset of the
first-order model, is used in this study.
Another deviation from first-order kinetics is the reduction of inactivation
rate at higher UV fluences (tailing). This is observed in several drinking
___________________________________________________________________
- 146 -
Chapter 6
water studies with CB-tests (Polioviruses, rotaviruses, E. coli, C. parvum and
G. muris) and also for environmental bacteriophages and bacteria in
wastewater studies in CF-systems. Tailing normally starts after at least 99%
of the initial available micro-organisms are inactivated and is observed to a
larger extent in the more UV susceptible micro-organisms. For the most
resistant organisms (Adenoviruses, MS2 phages, bacterial spores and
Acanthamoeba spp.) tailing was not observed. The cause of tailing is still
under debate. Several causes have been hypothesised, such as experimental
bias, hydraulics, aggregation of micro-organisms or a resistant
subpopulation, but no conclusive evidence is available for any of these. For
micro-organisms where tailing is observed, we have used the first-order
model only for the fluence range that yielded a linear relation with the
inactivation in the experiments. Because of the observed tailing,
extrapolation of this inactivation rate to higher fluences is yielding
uncertain results. For use in QMRA, the higher fluences can be assumed to
yield (at least) the same inactivation credits as the highest fluence in the
linear relation.
Significance for water disinfection. This study provides an
extensive overview of the efficacy of UV disinfection for viruses, bacteria
and bacterial spores and protozoan (oo)cysts, obtained from the reviewed
literature. The k-values that were calculated from the reviewed studies can
be used in QMRA and treatment design to determine the efficacy of a UV
fluence in the inactivation of the range of reported bacterial and viral
pathogens and indicator organisms. For Cryptosporidium and Giardia, the
logarithmic functions given in USEPA (2003) were used for calculating the
inactivation efficacy.
In the group of pathogenic micro-organisms, viruses are generally more
resistant than Cryptosporidium, Giardia and the bacterial pathogens.
Adenovirus 40 is the most UV resistant waterborne pathogen known.
Acanthamoeba is also very resistant. Bacterial spores, esp. environmental
spores of Clostridium are also resistant to UV, with k-values that are
comparable to the Adenoviruses (Fig. 10).
Correction of the required fluence: micro-organism related factors.
Based on the increased UV resistance observed for environmental
Salmonella, enterococci, thermotolerant coliforms, FRNA phages and spores
of sulphite-reducing clostridia, correction of the fluence requirement for
inactivation of bacteria and bacterial spores from the environment seems
appropriate. The evaluated studies suggest a two times increased fluence
requirement for bacteria and four times for bacterial spores in drinking
water. For wastewater this is most likely not enough and based on Fig. 8 a
factor of seven seems more appropriate. The results of environmental
___________________________________________________________________
- 147 -
Chapter 6
.
polioviruses (Slade et al., 1986) and FRNA phages and Adenoviruses
(Watercare, 2002) indicate that such a correction is not needed for phages
and viruses. However, further research is needed to support these findings.
Similarly, studies on the increased resistance of environmental protozoan
(oo)cysts to UV are appropriate.
Data from the evaluated studies indicated that photo-reactivation can result
in a significant increase of the required fluence for bacteria to achieve the
same level of inactivation as without photo-reactivation. Dark-repair does
not seem to be very significant for the UV disinfection practice for most
pathogens. Though for Giardia dark-repair was observed in two studies at
lower (5-25 mJ/cm2) fluences, but not at higher fluences (60 mJ/cm2).
Consequently, correction of the required fluence of full-scale UV
disinfection because of photo-reactivation of bacteria and in the case of
Giardia, also because of dark-repair at low fluences, could be necessary. For
viruses, it is assumed that repair is included in the available fluenceresponse curves, as suggested for the double-stranded DNA Adenoviruses.
After UV disinfection of drinking water photo-reactivation is not likely to
occur but in the case of wastewater disinfection light exposure is likely.
Translation of the presented photo-reactivation data to full-scale
conditions, however, is not straight forward. These data have been
observed under conditions favouring the induction of photo-repair (low
fluence values, thin layer with optimal conditions for reactivation). The
conditions in wastewater practice will be less favourable for exposure to
light and hence for photo-reactivation to occur. Furthermore, the applied
UV fluences in practice are usually higher than applied in the reviewed
studies. Lindenauer and Darby (1994) and Gehr and Nicell (1996) showed a
decrease in repair at higher fluences due to tailing in the inactivation
kinetics. The former author also hypothesized that extended DNA damage
at higher fluence values will reduce the potential for photo-repair. From
their wastewater study and that of Whitby and Palmateer (1993), Gehr and
Nicell (1996) suggested that in practice the overall impact of photo-repair
might be negligible, because of the limited exposure to light and therefore
limited induction of photo-repair. In conclusion this needs further
verification, but we assume that the necessity for a fluence correction as a
result of photo-reactivation in UV disinfection practice is less than
suggested by the experimental data in the photo-reactivation studies.
Correction of the required fluence: fluence related factors. Most of
the studies that have been reviewed have been executed under wellcontrolled laboratory conditions in which UV fluence was assessed with
sensors and seeded micro-organisms. Information about the efficacy of UV
systems under full-scale conditions was limited and those which have been
___________________________________________________________________
- 148 -
Chapter 6
evaluated, generally showed lower inactivation efficiency than in the
laboratory. This reduced efficiency may be caused by factors related to the
micro-organisms as described previously, but also by imperfections in the
calculation of the fluence to which the micro-organisms are exposed in fullscale UV systems. The latter can be largely overcome by applying
biodosimetry to full-scale UV systems to determine the Reduction
Equivalent Fluence (REF). This is already enforced for the application of
UV systems in drinking water practice in Austria (Österreichisches
normungsinstitut, 1999). In Germany, a similar protocol is used as
guideline (DVGW, 1997), and in the USA, the draft EPA Ultraviolet
Disinfection Guidance Manual also appoints credits for inactivation of
Cryptosporidium on the basis of biodosimetry (USEPA, 2003). Commonly
used biodosimeters are spores of B. subtilis or MS2 phages. Cabaj et al.
(1996), however, demonstrated that the REF decreases with increased
broadening of the fluence distribution and increased inactivation rate
constant of the used model organism. Consequently, susceptible model
organisms (high k-value) are more sensitive to a broad fluence distribution,
which will enlarge the gap between the REF and the arithmetic mean
fluence. MS2 phages and spores of B. subtilis are less sensitive to UV than
most other pathogenic micro-organisms (Table 1, 2, 3 and 4). The EPA
manual (USEPA, 2003) introduced a REF bias based on effects of fluence
distribution and inactivation rate constants to account for the difference in
sensitivity between model organism and target pathogens. Another
approach is the use of alternative model organisms. E. coli is suggested and
also this review indicates that it can be used as model for the more
susceptible bacteria and also Cryptosporidium and Giardia. More recently,
Clancy et al. (2004) suggested two potential bacteriophages Qβ and T7 as
model organisms. The use of T7 as UV dosimeter was previously proposed
by Rontó et al. (1992). The k-values of these organisms (Table 2) are more in
the range of the k-values calculated for the more sensitive pathogens. For
MP-systems, the germicidal fluence is usually obtained with the DNA
absorbance spectrum to weigh the effectiveness of the different
wavelengths. The action spectra (the relative sensitivity to different UV
wavelengths) of adenoviruses and the model organism MS2 phage differed
from the action spectrum of DNA as demonstrated by Malley et al. (2004)
and Mamane-Gravetz et al. (2005b). Thus, with different action spectra of
model organisms used as biodosimeter different REF values will be
calculated for polychromatic UV radiation. Therefore, more information
about differences in action spectra between pathogens and potential
___________________________________________________________________
- 149 -
Chapter 6
.
biodosimeter organisms is required to increase the precision of the fluence
determination of MP-systems.
Required fluence table. The accumulated knowledge in this review
was used to create the required fluence of LP-lamps for a MIC of 1, 2, 3 or 4
log inactivation for most micro-organisms relevant to microbiological
safety of water (Table 6). The required fluence is calculated from the kvalues presented in Tables 1, 3 and 4 with correction for the increased UVsensitivity of environmental bacteria and bacterial spores. For five bacteria
species this was based on specific literature data and for the other bacteria,
this was set at a factor of three, whereas for wastewater disinfection higher
correction seems appropriate. Correction for environmental organisms was
not necessary for the viruses (see data on FRNA-phages in Figure 7) and for
the protozoa no data are available. Increased fluence requirement because
of DNA-repair did not seem necessary for viruses and protozoa, although
for Giardia at low fluences of 5-25 mJ/cm2 results are still conflicting. For
bacteria fluence correction for dark-repair is not necessary and further
study has to elucidate whether correction for photo-reactivation is
required.
Research items. Based on this review a number of knowledge gaps
are identified. More quantitative information is needed to estimate the
effect of micro-organism related factors like environmental species, DNArepair (esp. of Giardia) and differences in spectral sensitivity influencing the
fluence requirement of UV disinfection under full-scale conditions.
Biodosimetry is a powerful tool to determine germicidal fluence values of
CF-systems, but to determine REF for the whole range of relevant microorganisms with different UV sensitivities, additional model organisms are
needed. In the application of medium-pressure lamps with polychromatic
UV light, further development of fluence assessment is of importance using
biodosimetry with proper weighting for spectral sensitivity in connection
with fluence calculation models. Daily UV process control needs further
research into accurate description of the distribution of water flows and UV
intensity over UV reactors, using CFD, to obtain simple, reliable and cheap
in situ process control systems. Independent verification with biodosimetry
is still essential. Systems to measure the germicidal fluence on-site in water
treatment practice, using micro-organisms or compounds “naturally”
present in the water would allow on-site verification of the efficacy of UV
systems in practice.
___________________________________________________________________
- 150 -
Chapter 6
Table 6 The UV fluence (mJ/cm2) requirements for a MIC of 1 up to 4 log by
monochromatic UV radiation for viruses, bacteria, bacterial spores and protozoan (oo)cysts
based on the k-values with or without correction for environmental species; for bacteria in
wastewater a higher correction for environmental species is needed and further research has
to clarify the need for a higher fluence to account for photoreactivation; for Giardia
increased fluence requirement because of dark repair is a factor for further research
Bacillus subtilisa
Required fluence (mJ/cm2)
1
2
3
4
56
111
167
222
Adenovirus type 40
56
111
167
-b
Clostridium perfringensa
45
95
145
-b
Adenovirus type 2,15,40,41
42
83
125
167
Acanthamoebac
40
71
119
167
50
-d
-b
MIC required (log):
Adenovirusa
(no type 40)
25
Calicivirus canine
10
21
31
41
Rotavirus SA-11
10
20
29
39
Calicivirus feline
9
19
28
38
Coxsackie virus B5
8
17
25
34
Streptococcus faecalisa
9
16
23
30
Legionella pneumophilad
8
15
23
30
Poliovirus type 1
7
15
22
30
6
13
19
26
6
12
17
51
Hepatitis A
6
11
17
22
Calicivirus bovine
5
11
16
21
O157d
5
9
14
19
5
9
14
18
3
6
12
-e
Giardia USEPAc
2
5
11
-e
Campylobacter jejunid
3
7
10
14
Yersinia enterocoliticad
3
7
10
13
Legionella pneumophilad
3
6
8
11
Shigella dysenteriaed
3
5
8
11
Vibrio choleraed
2
4
7
9
Shigella
sonneid
Salmonella
E. coli
E.
typhia
colia
Cryptosporidium
USEPAc
a environmental
spp.; b MICmax < 4 log; c no correction for environmental spp.
(research needed); d corrected for environmental spp.; e no value due to tailing
___________________________________________________________________
- 151 -
Chapter 6
.
CONCLUSIONS
The accumulated literature data on the inactivation kinetics of disinfection
with UV irradiation demonstrate that the process is effective against all
pathogenic micro-organisms relevant for the current drinking water
practices. The inactivation of micro-organisms by UV could be described
with first-order kinetics using fluence-inactivation data from laboratory
studies in collimated beam tests. No inactivation at low fluences (shoulder)
and no further increase of inactivation at higher fluences (tailing) was
observed for some micro-organisms. The former deviation from the loglinear kinetics is included in MIC calculations and the latter was used to
determine the maximum (observed) MIC-values. The parameters that were
used to describe the inactivation are the inactivation rate constant k
(cm2/mJ), the maximum inactivation demonstrated and (only for bacterial
spores and Acanthamoeba) the offset parameter. The most persistent
organisms known are viruses, specifically Adenoviruses, and bacterial
spores. From the protozoa Acanthamoeba was highly UV resistant. Bacteria
and (oo)cysts of Cryptosporidium and Giardia are more susceptible with a
fluence requirement of <20 mJ/cm2 for a MIC of 3 log.
Several studies have reported an increased UV resistance of environmental
bacteria and bacterial spores, compared to lab-cultured organisms. This
means that higher UV fluences are required to obtain inactivation. Hence,
for bacteria and spores, a correction factor of two and four was included in
the MIC calculation, respectively, and data from the wastewater studies
show that a higher correction is required under these conditions. For
phages and viruses this phenomenon appears to be of little significance and
for protozoan (oo)cysts this aspect needs further attention. For application
in drinking water, no correction for repair seems necessary for most
pathogens. The results on repair for Giardia are conflicting, but no repair
occurred at higher fluences (60 mJ/cm2). For application in wastewater, the
occurrence of photo-reactivation of bacteria is a subject for further research.
To enable accurate assessment of the effective fluence in continuous flow
UV systems in water treatment practice, biodosimetry is of great
importance. In the case of MP-lamps more information about differences in
spectral sensitivity
between pathogens and potential biodosimeter
organisms is needed to increase the precision of the fluence determination.
For UV systems that are primarily dedicated to inactivate the more
sensitive pathogens (Cryptosporidium, Giardia, pathogenic bacteria),
additional model organisms are needed to serve as biodosimeter.
___________________________________________________________________
- 152 -
Chapter 6
REFERENCES
Anonymous. 1999. Anlagen zur Desinfektion von Wasser mittels Ultravioletstrahlen, Anforderungen und Prüfung, Önorm M 5873-1. Österreichisches
Normungsinstitut, Vienna Au.
Antopol, S. C., and P. D. Ellner. 1979. Susceptibility of Legionella pneumophila to
Ultraviolet radiation. Appl. Environ. Microbiol. 38:347-348.
Battigelli, D. A., M. D. Sobsey, and D. C. Lobe. 1983. The inactivation of Hepatitis
A virus and other model viruses by UV irradiation. Wat. Sci. Tech. 27:339-342.
Belosevic, M., S. A. Craik, J. L. Stafford, N. F. Norman, J. C. Kruithof, and D.
Smith. 2001. Studies on the resistance/reactivation of Giardia muris cysts and
Cryptosporidium parvum oocysts exposed to medium-pressure ultraviolet
radiation. FEMS Microbiol. Letter 204:197-203.
Bernhardt, H. 1994. Desinfektion aufbereiteter Oberflächenwässers mit UV
Strahlen. GWF Wasser-Abwasser 135:677-689.
Bernhardt, H., O. Hoyer, B. S. Hengesbach, D., P. Karanis, H. Moriske, and C.
von Sonntag. 1992. Desinfektion aufbereiteter Oberflächenwässers mit UV
Strahlen- erste Ergebnisse des Forschungsvorhabens. GWF Wasser- Abwasser
133:632-643.
Bolton, J. R., B. Dussert, Z. Bukhari, T. Hargy, and J. L. Clancy. 1998. Presented at
the Proceedings of the American Water Works Association Water Quality
Technology Conference, San Diego, US, Nov. 1998.
Butler, R. C., V. Lund, and D. A. Carlson. 1987. Susceptibility of Campylobacter
jejuni and Yersinia enterocolitica to UV radiation. Appl. Environ. Microbiol. 53:375378.
Cabaj, A., R. Sommer, and D. Schoenen. 1996. Biodosimetry: model calculations
for UV water disinfection devices with regard to dose distribution. Water Res.
30:1003-1009.
Chang, J. C. H., S. F. Ossoff, D. C. Lobe, M. H. Dorfman, C. M. Dumais, R. G.
Qualls, and J. D. Johnson. 1985. UV inactivation of pathogenic and indicator
micro-organisms. Appl. Environ. Microbiol. 49:1361-1365.
Chick, H. 1908. An investigation of the laws of disinfection. J. Hyg. 8:92.
Clancy, J. L., Z. Bukhari, T. M. Hargy, J. R. Bolton, B. W. Dussert, and M. M.
Marshall. 2001. Using UV to inactivate Cryptosporidium. J. Am. Water Works
Assoc. 92:97-104.
Clancy, J. L., T. M. Hargy, D. A. Battigelli, M. M. Marshall, D. Korich, and W. L.
Nicholson. 2002. Susceptibility of multiple strains of C. parvum to UV light. Am.
Water Works Assoc. Res. Found. Denver CO, US.
Clancy, J. L., T. M. Hargy, M. M. Marshall, and J. E. Dyksen. 1998. UV light
inactivation of Cryptosporidium oocysts. J. Am. Water Works Assoc. 90:92–102.
Clancy, J. L., K. Fallon, T. M. Hargy, E. Mackey, and H. Wright. 2004. Presented at
the American Water Works Association Water Quality Technology Conference,
Nov. 2004, San Antonio, US.
___________________________________________________________________
- 153 -
Chapter 6
.
Craik, S. A., G. R. Finch, J. R. Bolton, and M. Belosevic. 2000. Inactivation of
Giardia muris cysts using medium-pressure ultraviolet radiation in filtered
drinking water. Water Res. 34:4325-4332.
Craik, S. A., D. Weldon, G. R. Finch, J. R. Bolton, and M. Belosevic. 2001.
Inactivation of Cryptosporidium parvum oocysts using medium- and low-pressure
ultraviolet radiation. Water Res. 35:1387-1398.
de Roda Husman, A. M., E. Duizer, W. Lodder, W. Pribil, A. Cabaj, P. Gehringer,
and R. Sommer. 2003. Presented at the 2nd Int. Congress on ultraviolet
technologies, Vienna, Au.
Duizer, E., P. Bijkerk, B. Rockx, A. de Groot, F. Twisk, and M. Koopmans. 2004.
Inactivation of caliciviruses. Appl. Environ. Microbiol. 70:4538-4543.
DVGW. 1997. UV-Desinfektionsanlagen für die Trinkwasserversorgung –
Anforderungen und Prüfung, Technical Standard W294. In D. V. d. G.-u.
Wasserfaches (ed.), Bonn, GR.
Gehr, R., and J. Nicell. 1996. Pilot studies and assessment of down stream effects
on UV and ozone disinfection of a physicochemical wastewater. Water Qual.
Res. J. Can. 31:263-281.
Gehr, R., M. Wagner, P. Veerasubramanian, and P. Payment. 2003. Disinfection
efficiency of peracetic acid, UV and ozone after enhanced primary treatment of
municipal wastewater. Water Res. 37:4573-4586.
Gerba, C. P., D. M. Gramos, and N. Nwachuku. 2002. Comparative inactivation of
enteroviruses and adenovirus 2 by UV light. Appl. Environ. Microbiol. 68:51675169.
Haas, C. N. 1983. Estimation of risk due to low doses of micro-organisms: a
comparison of alternative methodologies. Am. J. Epidemiol. 118:573-82.
Harris, G. D., V. D. Dean Adams, D. L. Sorensen, and M. S. Curtis. 1987.
Ultraviolet inactivation of selected bacteria and viruses with photoreactivation of
the bacteria. Water Res. 21:687-692.
Havelaar, A. H., C. C. E. Meulemans, W. M. Pot-Hogeboom, and J. Koster. 1990.
Inactivation of bacteriophage MS2 in wastewater effluent with monochromatic
and polychromatic ultraviolet light. Water Res. 24:1387-1393.
Havelaar, A. H., W. M. Pot-Hogeboom, W. Koot, and R. Pot. 1987. F-specific
bacteriophages as indicators of the disinfection efficiency of secondary effluent
with ultraviolet radiation. Ozone Sci. and Eng. 9:353-368.
Henry, V., A. Helbronner, and M. Recklinghausen. 1910. Nouvelles recherches
sur la sterilization de grandes quantited d’eau par les rayons ultraviolets. Comp.
Rend. Acad. Sci. 151:677–680.
Hijnen, W. A. M., G. J. Medema, and D. van der Kooij. 2004a. Quantitative
assessment of the removal of indicator bacteria in full-scale treatment plants.
Wat. Sci. Technol.: Water Supply 4 47-54.
Hijnen, W. A. M., A. J. van der Veer, E. F. Beerendonk, and G. J. Medema. 2004b.
Increased resistance of environmental anaerobic spores to inactivation by UV.
Wat. Sci. Technol.: Water Supply 4:54-61.
Hijnen, W. A. M., D. Veenendaal., W. M. H. Van der Speld, A. Visser, W.
Hoogenboezem, and D. Van der Kooij. 2000. Enumeration of faecal indicator
___________________________________________________________________
- 154 -
Chapter 6
bacteria in large water volumes using on site membrane filtration to assess water
treatment efficiency. Water Res. 34:1659-1665.
Hoyer, O. 1998. Testing performance and monitoring of UV systems for drinking
water disinfection. Wat. Supply 16:424-429.
Hoyer, O. 2004. Presented at the European Conference UV Karlsruhe, UV
radiation, Effects and Technologies, Karlruhe, GR, 2003.
Jacangelo, J. G., P. Loughran, B. Petrik, D. Simpson, and C. Mclloy. 2004.
Removal of enteric viruses and selected microbial indicators by UV irradiation of
secondary effluent. Wat. Sci. Tech. 47:193-198.
Johnson, A. M., K. Linden, K. M. Ciociola, R. De Leon, G. Widmer, and P. A.
Rochelle. 2005. UV inactivation of Cryptosporidium hominis as measured in cell
culture. Appl. Environ. Microbiol. 71:2800-2802.
Knudson, G. B. 1985. Photoreactivation of UV irradiated Legionella pneumophila
and other Legionella species. Appl. Environ. Microbiol. 49:975-980.
Kruithof, J. C., P. C. Kamp, B. J. Martijn, M. Belosevic, and G. Williams. 2005.
Presented at the Third Int. Congress on Ultraviolet Technologies, Whistler, BC
Canada, May 24-27.
Kruithof, J. C., R. C. Van der Leer, and W. A. M. Hijnen. 1992. Practical
experiences with UV disinfection in The Netherlands. J. Water SRT-Aqua 41 8894.
Linden, K. G., G. Shin, G. Faubert, W. Cairns, and M. D. Sobsey. 2002. UV
Disinfection of Giardia lamblia cysts in water. Environ. Sci. Technol. 36:2519-2522.
Lindenauer, K. G., and J. L. Darby. 1994. Ultraviolet disinfection of wastewater:
effect of dose on subsequent photoreactivation. Water Res. 28:805-817.
Lytle, C. D. 1971. Host-cell reactivation in mammalian cells I. survival of ultraviolet-irradiated Herpes virus in different cell-lines. Int. Jour. of Rad. Biol. 19:329337.
Maier, A., D. Tougianidou, A. Wiedenmann, and K. Botzenhart. 1995. Detection
of poliovirus by cell culture and by PCR after UV disinfection. Wat. Sci. Tech.
31:141-145.
Malley, J. P., N. A. Ballester, A. B. Margolin, K. G. Linden, A. Mofidi, J. R.
Bolton, G. Crozes, J. M. Laine, and M. L. Janex. 2004. Inactivation of pathogens
with innovative UV technologies. AWWARF, Denver CO, US.
Mamane-Gravetz, H., and K. G. Linden. 2004. Presented at the American Water
Works Association Water Quality Technology Conference, San Antonio, US.
Mamane-Gravetz, H., and K. G. Linden. 2005a. Relationship between
physiochemical properties, aggregation and UV inactivation of isolated
environmental spores in water. J. Appl. Microbiol. 98:351-363.
Mamane-Gravetz, H., K. G. Linden, A. Cabaj, and R. Sommer. 2005b. Spectral
sensitivity of Bacillus subtilis spores and MS2 coliphage for validation testing of
ultraviolet reactors for water disinfection. Environ. Sci. Technol. 39:7845-7852.
Martiny, H., T. Schubert, and H. Rüden. 1990. Use of UV radiation for the
disinfection of water V. Communication: Microbiological investigations with
bacteria of the exponential and the stationary phase in cold and in warm
drinking water. Zbl. Hyg. 190:380-394.
___________________________________________________________________
- 155 -
Chapter 6
.
Maya, C., N. Beltrán, B. Jiménez, and P. Bonilla. 2003. Evaluation of the UV
disinfection process in bacteria and amphizoic amoeba inactivation. Wat. Sci.
Technol.: Wat. suppl. 3:285-291.
Medema, G. J., W. Hoogenboezem, A. J. van der Veer, H. A. M. Ketelaars, and W.
A. M. Hijnen, and P. J. Nobel. 2003. Quantitative risk assessment of
Cryptosporidium in surface water treatment. Wat. Sci. Tech. 47:241-247.
Meng, Q. S., and C. P. Gerba. 1996. Comparative inactivation of enteric
adenoviruses, poliovirus and coliphages by ultraviolet irradiation. Water Res.
30:2665-2668.
Mofidi, A. A., P. A. Rochell, C. I. Chou, H. M. Mehta, K. G. Linden, and J. P.
Malley. 2002. Presented at the American Water Works Association Water
Quality Technology Conference, Seattle US, November 10-14.
Morita, S., A. Namikoshi, T. Hirata, K. Oguma, H. Katayama, S. Ohgaki, N.
Motoyama, and M. Fujiwara. 2002. Efficacy of UV irradiation in inactivating
Cryptosporidium parvum oocysts. Appl. Environ. Microbiol. 68:5387-5393.
Nwachuku, N., C. P. Gerba, A. Oswald, and F. D. Mashadi. 2005. Comparative
inactivation of adenovirus serotypes by UV light disinfection. Appl. Environ.
Microbiol. 71:5633-5636.
Oguma, K., H. Katayama, and S. Ohgaki. 2002. Photoreactivation of Escherichia coli
after low- or medium-pressure UV disinfection determined by an endonuclease
sensitivity site assay. Appl. Environ. Microbiol. 68:6029-6035.
Oguma, K., H. Katayama, and S. Ohgaki. 2004. Photoreactivation of Legionella
pneumophila after inactivation by low- or medium pressure ultraviolet lamp.
Water Res. 38(11), 2757-2763.
Oppenheimer, J. A., J. E. Hoagland, J.-M. Laine, J. G. Jacangelo, and A. Bhamrah.
1993. Presented at the Water Environmental Federation Specialty Conference
Series: Planning, design & operation of effluent disinfection systems, Alexandria,
VA.
Oppenheimer, J. A., J. G. Jacangelo, J. M. Laîné, and J. E. Hoagland. 1995.
Presented at the American Water Works Association Water Quality Technology
Conference, New Orleans US, November 12-16.
Örmeci, B., and K. G. Linden. 2002. Comparison of UV and chlorine inactivation
of particle and non-particle associated coliforms. Wat. Sci. Technol.: Wat. suppl.
2:403-410.
Qian, S. S., D. M., D. C. Schmelling, M. M., K. G. Linden, and C. Cotton. 2004.
Ultraviolet light inactivation of protozoa in drinking water: a Bayesian metaanalysis. Water Res. 38:317 – 326.
Qualls, R. G., M. P. Flynn, and J. D. Johnson. 1983b. The role of suspended
particles in ultraviolet disinfection. J. Water Pollut. Control Fed. 55:1280–1285.
Qualls, R. G., and J. D. Johnson. 1983a. Bioassay and dose measurement in UV
disinfection. Appl. Environ. Microbiol. 45:872–877.
Regli, S., J. B. Rose, C. N. Haas, and C. P. Gerba. 1991. Modeling the risk from
Giardia and viruses in drinking water. J. Am. Water Works Assoc. 83:76-84.
___________________________________________________________________
- 156 -
Chapter 6
Rochelle, P. A., A. A. Mofidi, M. M. Marshall, S. J. Upton, B. Montelone, K.
Woods, and G. Di Giovanni. 2004. An investigation of UV disinfection and
repair in Cryptosporidium parvum. AWWARF, Denver CO, US.
Rontó, G., S. Gáspár, and A. Bérces. 1997. Phage T7 in biological UV dose
measurement. J. Photochem. Photobiol. B Biol. 12:285-294.
Schoenen, D., V. Zemke, and A. Kolch. 1991. The influence of reflection of the UVrays on the disinfection of drinking water. Zbl. Hyg. 191:396-405.
Schoenen, D. K., A., Gebel, J. and Hoyer, O. 1995. UV-Desinfektion von
Trinkwasser, Grundlagen und Anvorderungen für die praktische Anwendung.
DVGW Schriftenreihe Wasser 86.
Severin, B. F., M. T. Suidan, and R. S. Engelbrecht. 1983. Effects of temperature
on ultraviolet light disinfection. Environ. Sci. Technol. 17:717-721.
Shin, G., Z. Bohrerova, K. G. Linden, and G. Faubert. 2005. Presented at the Third
Int. Congress on Ultraviolet Technologies, Whistler, BC Canada, May 24-27.
Shin, G., K. G. Linden, M. J. Arrowood, and M. D. Sobsey. 2001. Low-pressure
UV inactivation and DNA repair potential of Cryptosporidium parvum oocysts.
Appl. Environ. Microbiol. 67:3029-3032.
Simpson, D., J. G. Jacangelo, P. Loughran, and C. Mcllroy. 2003. Investigation of
potential surrogate organisms and public health risk in UV irradiated secondary
effluent. Wat. Sci. Tech. 47:37-43.
Slade, J. S., N. R. Harris, and R. G. Chisholm. 1986. Disinfection of chlorine
resistant enteroviruses in groundwater by ultraviolet irradiation. Wat. Sci. Tech.
18:115-123.
Smeets, P. W. M. H., Y. J. Dullemont, and G. J. Medema. 2005. Presented at the
17th IOA conference, Strasbourg, France.
Smeets, P. W. M. H., A. W. C. Van der Helm, Y. J. Dullemont, L. C. Rietveld, J. C.
Van Dijk, and G. J. Medema. 2006. Inactivation of Escherichia coli by ozone
under bench-scale plug flow and full-scale hydraulic conditions. Water Res.
40:3239-3248.
Sommer, R., A. Cabaj, and T. Haider. 1996b. Microbiocidal effect of reflected UV
radiation in devices for water disinfection. Wat. Sci. Tech. 34:173-177.
Sommer, R., A. Cabaj, W. Pribil, and T. Haider. 1997. Influence of lamp intensity
and water transmittance on the UV disinfection of water. Wat. Sci. Tech. 35:113118.
Sommer, R., A. Cabaj, W. Pribil, T. Haider, and G. Hirschmann. 2000b. Presented
at the Wasser Berlin, IOA, European-African-Asian-Australian Group, Berlin GR,
October 23-26.
Sommer, R., A. Cabaj, T. Sandu, and M. Lhotsky. 1999. Measurement of UV
radiation using suspensions of micro-organisms. J. Photochem. Photobiol. B:Biol.
53:1-6.
Sommer, R., W. Pribil, S. Appelt, P. Gehringer, H. Eschweiler, H. Leth, A. Cabaj,
and T. Haider. 2001. Inactivation of bacteriophages in water by means of nonionizing (UV-253.7 nm) and ionizing (gamma) radiation: a comparative
approach. Water Res. 35:3109-3116.
___________________________________________________________________
- 157 -
Chapter 6
.
Sommer, R., T. Haider, A. Cabaj, W. Pribil, and M. Lhotsky. 1998. Time fluence
reciprocity in UV disinfection of water. Wat. Sci. Tech. 38:145-150.
Sommer, R., M. Lhotsky, T. Haider, and A. Cabaj. 2000a. UV inactivation, liquidholding recovery, and photoreaction of Escherichia coli OH157 and other
pathogenic Escherichia coli strains in water. Jour. Food Prot. 63:1015-1020.
Sommer, R., G. Weber, A. Cabaj, J. Wekerle, G. Keck, and G. Schauberger. 1989.
UV inactivation of micro-organisms in water. Zbl. Hyg. 189:214-224.
Teunis, P. F. M., G. J. Medema, L. Kruidenier, and A. H. Havelaar. 1997.
Assessment of the risk of infection by Cryptosporidium and Giarda in drinking
water from a surface water source. Water Res. 31:1333-1346.
Thompson, S. S., J. L. Jackson, M. Suva-Castillo, W. A. Yanko, Z. El jack, J. Kuo,
C. Chen, F. P. Williams, and D. P. Schnurr. 2003. Detection of infectious human
Adenoviruses in tertiary-treated and ultraviolet-disinfected wastewater. Wat.
Environ. Res. 75:163-170.
Thurston-Enriquez, J. A., C. N. Haas, J. G. Jacangelo, K. Riley, and C. P. Gerba.
2003. Inactivation of Feline Calicivirus and Adenovirus Type 40 by UV radiation.
Appl. Environ. Microbiol. 69:577-582.
USEPA. 2003. Ultraviolet disinfection guidance manual. EPA 815-D-03-007. EPA,
Washington DC, US.
von Sonntag, C., A. Kolch, J. Gebel, K. Oguma, and R. Sommer. 2004. Presented
at the European Conference UV Karlsruhe, UV radiation. Effects and
Technologies, Karlsruhe GR, September 22-24.
Watercare. 2002. Pilot plant investigations, surrogate study, results and
recommendations. Disinfection review group report Water Care Services Ltd.,
New Sealand.
Whitby, G. E., and G. Palmateer. 1993. The effect of UV transmission, suspended
solids, wastewater mixtures and photoreactivation on micro-organisms in
wastewater treated with UV light. Wat. Sci. Tech. 27:379-386.
Wiedenmann, A., B. Fischer, U. Straub, C. H. Wang, B. Flehmig, and D.
Schoenen. 1993. Disinfection of Hepatitis A virus and MS2 coliphage in water by
ultraviolet irradiation: comparison of UV susceptibility. Wat. Sci. Tech. 27:335338.
Wilson, B. R., P. F. Roessler, E. van Dellen, M. Abbaszadegan, and C. P. Gerba.
1992. Presented at the American Water Works Association Water Quality
Technology Conference, November 15-19, Toronto CN.
Wolfe, R. L. 1990. Ultraviolet disinfection of potable water. Environ. Sci. Technol.
24:768-773.
Zimmer, J. L., and R. M. Slawson. 2002. Potential repair of Escherichia coli DNA
following exposure to UV radiation from both medium- and low-pressure UV
sources used in drinking water treatment. Appl. Environ. Microbiol. 68:32933299.
Zimmer, J. L., R. M. Slawson, and P. M. Huck. 2003. Inactivation and potential
repair of Cryptosporidium parvum following low- and medium-pressure
ultraviolet irradiation. Water Res. 37:3517-3523.
___________________________________________________________________
- 158 -
Chapter 7
Elimination of viruses, bacteria and
protozoan oocysts by slow sand
filtration●
W.A.M. Hijnen1, J.F. Schijven2, P. Bonné3, A. Visser4 and G.J. Medema1
1 Kiwa
Water Research Ltd., PO Box 1072, 3430 BB Nieuwegein, NL
National Institute of Public Health and the Environment, PO Box 1, 3720 BA
Bilthoven, NL
3 Waternet, PO Box 8169, 1005 AD, Amsterdam, NL
4 Dune Water Company South Holland, PO Box 34, 2270 AA Voorburg, NL
2
•
Reprinted for Water Science and Technology, 50(1): 147-154, with permission from
copyright holder, IWA publishing.
___________________________________________________________________
- 159 -
Chapter 7
.
ABSTRACT
The decimal elimination capacity (DEC) of slow sand filters (SSF) for
viruses, bacteria and oocysts of Cryptosporidium has been assessed from
full-scale data and pilot plant and laboratory experiments. DEC for viruses
calculated from experimental data with MS2-bacteriophages in the pilot
plant filters was 1.5 – 2.0 log. E. coli and thermotolerant coliforms (Coli44)
were removed by full-scale filters and in a pilot plant filter with 2 - 3 log. At
full-scale, Campylobacter bacteria removal was 1 log more than removal of
Coli44, which indicates that Coli44 is a conservative surrogate for these
pathogenic bacteria. Laboratory experiments with sand columns showed 2
- 3 and >5 - 6 log removal of spiked spores of sulphite-reducing clostridia
(SSRC; C. perfringens) and oocysts of Cryptosporidium, respectively.
Consequently, SSRC is not a good surrogate to quantify oocyst removal by
SSF. Removal of indigenous SSRC by full-scale filters is less efficient than
observed in the laboratory columns probably due to continuous load of
these filter beds with spores, accumulation and retarded transport. It
remains to be investigated if this also applies to oocyst removal by SSF. The
results additionally showed that the schmutzdecke and accumulation of
(in)organic charged compounds in the sand increased the elimination of
micro-organisms. Removal of the schmutzdecke reduced DEC for bacteria
with ±2 log, but did not affect removal of phages. This clearly indicates that
besides biological activity, both straining and adsorption are important
removal mechanisms in the filter bed for larger micro-organisms than
viruses.
INTRODUCTION
Recently the Dutch Drinking Water Decree has been revised (Anonymous,
2001). Drinking water companies should demonstrate sufficient elimination
of pathogenic viruses, bacteria and (oo)cysts of Cryptosporidium and Giardia
by treatment to comply with an annual infection risk lower than 10-4 per
person. This requires quantitative knowledge about the elimination
capacity of processes. A first quantitative microbial risk assessment at the
plants of Amsterdam Water Supply (AWS) and the Dune Water Company
South-Holland (DWS) revealed a lack of knowledge about the efficacy of
their slow sand filters for the removal of micro-organisms. In collaboration
with the National Institute of Public Health and the Environment (RIVM)
and Kiwa Water Research they started a project to determine the decimal
elimination capacity DEC of slow sand filtration for viruses, bacteria and
___________________________________________________________________
- 160 -
Chapter 7
protozoan oocysts. Additional goals were to investigate the effect of the
schmutzdecke and the sand on elimination, and to verify the validity of
surrogates used to assess pathogen elimination capacities.
The project was started by evaluation literature for removal of pathogenic
micro-organisms. Simultaneously, full-scale data of the elimination of
indigenous micro-organisms such as thermolerant coliforms (Coli44),
spores of sulphite-reducing clostridia (SSRC) and Campylobacter, were
collected and reviewed. Both activities did not answer all the questions.
Moreover, the low and variable removal of SSRC by full-scale filters
seriously doubted the use of these indicators as surrogate for protozoan
(oo)cyst removal. Therefore further experimental research on pilot and
laboratory scale has been carried out. The results of all these investigations
are presented and discussed in this paper.
MATERIALS AND METHODS
Full-scale data analysis. The elimination of thermotolerant
coliforms Coli44 and spores of sulphite-reducing clostridia SSRC by the SSF
at two plants of AWS and one plant of DWS was determined from data of
three years of routine sampling (n= 32 up to 5184) and of a two-week
period in winter and summer with daily large volume sampling (n≤20).
Periodically the influent of the filters at the location of DWS contained
Campylobacter bacteria and simultaneously concentrations of these
pathogenic bacteria before and after SSF were measured by large volume
sampling. The Decimal Elimination Capacity DEC was calculated from the
log transformed average concentration in influent and effluent with DEC =
log C in - log C out . The average concentration is the total number of colony
forming units divided by the total sampled volume (number of samples x
sample volume) in the selected period.
Pilot plant experiments. Under full-scale situation elimination of
viruses could not be determined. Therefore elimination of bacteriophage
MS2 by two slow sand filters of the AWS pilot plant Leiduin has been
determined with a challenge test. MS2 is an icosahedral phage with a
diameter of 27 nm is a conservative surrogate for viruses (Schijven et al.,
2003). E. coli WR1 was co-injected as a reference to the elimination of
indigenous Coli44. The influent of SSF1 was surface water pre-treated by
coagulation floc removal, rapid sand filtration 1, dune infiltration followed
by an open collection reservoir, rapid sand filtration 2, ozonation, softening
and granular activated carbon filtration. SSF2 was supplied with the filtrate
of rapid sand filter 2 without additional treatment. DOC, turbidity and pH
___________________________________________________________________
- 161 -
Chapter 7
.
of the influent of SSF1 was 1.5 mg C/l, 0.1 FTU and 8.0, respectively and of
the influent of SSF2 2.1 mg C/l, 0.7 FTU and 8.0, respectively.
The filters (surface of 2.56 m2, bed depth of 1.5 m and 0.3 mm diameter
sand) were operated at a filtration rate of 0.3 m/h. The filter bed porosity
determined with a tracer (NaCl) test was 0.27. The schmutzdecke of SSF1
was scraped 12 days before the experiment. SSF2 was tested with a
smutzdecke of 81 days. Then the filter was scraped and tested again after 4
days. Temperature ranged from 9.4 up to 11.7oC. Before the start of the
spiking experiments MS2 bacteriophages and E. coli WR1 were added and
well mixed in the water above the filter bed to avoid dilution effects (this
was done before each test with different spiking concentrations). Spiking of
a low and a high concentration of MS2 and WR1 to the influent of the filters
(13 ml of suspension per min) lasted two periods of 24 hours, respectively.
The effluent concentrations of both micro-organisms had been monitored
during a ten days period.
Column experiments. Due to the relative large scale and possible
health risk in the pilot plant, removal of oocysts of Cryptosporidium has
been determined in columns and compared with elimination of spores of
Clostridium perfringens D10, a possible surrogate for oocyst removal in
treatment (Hijnen et al., 2000). MS2 and WR1 were co-injected as references
with the pilot plant and full-scale results. A full-scale filter of DWS was
sampled at four layers (20-40, 40-60, 60-80 and 80-100 cm) and the AWS
pilot plant filter at five layers (20-40, 50-70, 80-100, 110-130 and 130-150 cm)
for ripened sand. In the DWS and AWS columns (diameter of 9 cm) the 40
cm sand bed was packed with 10 and 8 cm sand of each separate layer of
sampled sand, respectively. Each separate layer was introduced in the
column with water and mechanically packed by ticking against the Perspex
column. Underneath this sand bed a layer of gravel (diameter of 1 - 2 mm)
placed on a rough iron mesh. The influent of the columns was the influent
of SSF1 sampled in a large RVS-tank (700 litre) and transported to the
experimental location. Temperature of the column influent varied between
8 and 13.5oC. Before the spiking experiment the columns were operated at a
filtration rate of 0.08 m/h during 2-3 weeks. 48 hours before the spiking test
filtration rate was increased to 0.3 m/h. The water was inoculated with
micro-organisms in a separate RVS-vessel located directly above the sand
columns (no dilution effect). During two hours this inoculated influent was
supplied to the columns and during the following 30 hours the
concentration of micro-organisms in the filtrate was monitored. To verify
the effect of co-injection two separate columns with AWS sand were tested,
one spiked with MS2 and E. coli WR1 (AWS1) and the other two (AWS2
___________________________________________________________________
- 162 -
Chapter 7
and DWS) spiked with a cocktail of MS2, WR1, Clostridium perfringens (D10)
and Cryptosporidium parvum.
Microbiological methods. The analytical methods used for Coli44
and SSRC in the full-scale study were described in detail before (Hijnen et
al., 2004). The most probable numbers of thermophilic Campylobacter
bacteria in the water before and after SSF of DWS was enumerated directly
in 1 ml or in 10 ml - 100 litres by membrane filtration on 0.22 μm pore size
filters. The samples are pre-cultured in Preston-bouillon (Ribeiro and Price,
1984) in 3 x 3 portions of a decimal dilution for 48 h at 42±0.5oC under
micro-aerophilic conditions. Each of these pre-cultures are incubated on
solid Karmali-medium (Karmali et al., 1986) for 48±2hrs at 42±0.5oC under
micro-aerophilic conditions. Typical colonies are grey and glancing and cell
material was microscopically confirmed as mobile and spiral formed cells.
The methods of production and storage of the stock-solutions and
enumeration methods of MS2 bacteriophages and E. coli WR1 used in the
pilot plant study have previously been described (Schijven, 2001). WR1
used for the column tests were pre-cultured in autoclaved tap water with 1
mg/l glucose-C. The production of stock-solution and enumeration method
of spores of C. perfringens D10 was described previously (Hijnen et al.,
2002). Cryptosporidium parvum oocysts (Moredun; harvested by
sedimentation and differential centrifugation) were spiked from a 1 ml
suspension (108 oocysts). For oocyst counting, samples of 1-200 ml were
analysed directly without concentration using Direct Fluorescence Assaytechnique with the chemscan (Chemunex). Samples were filtered and
prepared for scanning with the chemscan-kit (Chemunex 200 k0009-01 with
IMS) including a mounting medium (85 μl), 25 mm 2.0 μm membrane
filters and support Pad. The membrane filters were labelled with 100 µl 1:1
in deionized water diluted monoclonal antibody reagents (Oxoid; diluted
1:1 in deionised water) for 30 min. at 37oC. The filters were scanned and
counted spots were microscopically confirmed based on colour, form and
size.
RESULTS AND DISCUSSION
Elimination of bacteria and bacterial spores by full-scale filters.
The process conditions of the slow sand filters operated at the two locations
of AWS and at DWS showed no large differences (Table 1). From the
concentrations of Coli44 in influent and effluent observed in three separate
periods DEC-values of >1.5 up to 3.2 log were calculated (Table 1). The
average DEC for these indicator bacteria was 2.6±0.5 log. This is higher
___________________________________________________________________
- 163 -
Chapter 7
.
than the average DEC of 1.9±0.5 log calculated from a number of studies in
literature (Poynter and Slade, 1977; Slade, 1978; Cleasby et al., 1984;
Bellamy et al., 1985; Ellis, 1985). Removal of Campylobacter bacteria by SSF at
the DWS location Scheveningen was 3.4±0.6 log, 1 log more than Coli44
removal by these filters (Table 1). This indicates that E. coli is a
conservative surrogate parameter for Campylobacter removal by SSF.
Elimination of SSRC by SSF varied significantly. The maximum removal
was 1.8 log, but at locations Leiduin (AWS) and Scheveningen (DWS) data
also showed little to no elimination and even a small increase in the SSRC
numbers (negative DEC). Multiplication in anaerobic zones cannot be
excluded. However, accumulation and retarded transport seem the most
plausible explanations for this phenomenon. Spores not retained
irreversibly by straining or attachment to the sand grains, may persist long
enough to pass the sand filter eventually in concentrations that may even
be higher than in the influent at the time of monitoring. This phenomenon
for these spores has also been described by Schijven et al. (2003) and
indicates that DEC depends on the duration of the contamination. Hence, it
explains the positive correlation between elimination and the influent
concentration observed at Leiduin. DEC of 1.8 and -0.2 log10 was observed
at an average SSRC influent concentration of 45 and 0.16 spores/l,
respectively. These results raise serious doubts about the use of these
spores as a quantitative surrogate for Cryptosporidium oocysts removal by
SSF. On the other hand, provided that no irreversible attachment or
physical straining occurs, the same phenomenon of breakthrough in the
long run may also be the case for protozoan (oo)cysts removal by SSF.
Pilot plant study. Figure 1 shows the breaktrough curves of MS2
and WR1 in the challenge test. The difference between the average influent
and maximum effluent concentration was used to estimate DEC. The
tailing after the spiking period of 48 hrs is caused by slow detachment of
the attached micro-organisms (Schijven, 2001). The filters SSF1 and SSF2
reduced the concentration of both organisms with 1.7 up to 4.2 log10 (Table
2). A concentration increase of about 3 log (Fig. 1) did not significantly
affect DEC. The schmutzdecke of 81 days at SSF2 showed a significant
effect on the elimination of WR1. Before cleaning of the filter bed DEC was
3.9 - 4.2 and after cleaning 2 - 2.8 log. MS2 elimination however was hardly
influenced by the filter bed cleaning and was 2.1±0.6 log. This value is
similar to the average removal of indigenous enteroviruses by SSF of 1.9
log observed by others (Slade, 1978; Ellis, 1985).
___________________________________________________________________
- 164 -
1,8
1,7
2,2
1,9
Bacteriophage MS2
Low Cina High Cin
1,7
1,8
3,9
2,0
4,2
2,8
E.coli WR1
Low Cin High Cin
2,1
2,3
___________________________________________________________________
- 165 -
SSF2
81
SSF2
4
a C = influent concentration
in
Filter Time (days) after
schmutzdecke removal
SSF1
12
Table 2 The Decimal Elimination Capacity (log10(Cin/Cmax, out)) of the MS2 bacteriophages and E. coli (WR1) of SSF1 and SSF2
of the pilot plant of AWS at 10oC
Location:
Weesperkarspel AWS
Leiduin AWS
Scheveningen DWS
Filtration rate (m/h)
0.4
0.25
0.25
Bed length (m)
1.5
1.5
1.0
Surface cleaning freq. (year)
1
2
4-5
Grain size (mm)
0.15-0.6
0.13-0.37
0.3-1.8
3.2a; ndb; ndc
2.9a; 2.3b; 2.2c
DEC (log10) Coli44
2.2a; >1.5b; >3.7c
ndd
nd
4.1a; 3.0b; 3.2c
DEC (log10) Campylobacter
a
b
c
a,e
b,e
c,e
DEC (log10) SSRC
1.6 ; 1.5 ; 1.0
(1.8) ; -0.2 ; -0.1
1.8a; 0.0b; -0.2c
a three years routine monitoring in small volumes; b daily large volume sampling in two weeks in winter; c
daily large volume sampling in two weeks in summer; d nd = not determined; e 1st DEC high (45/l) and 2nd
and 3rd DEC low (0.16/l) SSRC level in influent Cin
Table 1 The decimal elimination capacity (DEC) of full-scale slow sand filters
Chapter 7
Chapter 7
.
Concentration org./ml
A 3 log10 removal of polioviruses by SSF was found by Windle-Taylor
(1969). Poynter and Slade (1978) reported a DEC of 3.5 and 2.8 log for MS2
phages and polioviruses. Without a schmutzdecke WR1 removal was
2.3±0.4 log. This value is close to MS2 removal by the same filters and E. coli
removal observed in full-scale filters and in literature.
1000000
100000
10000
Filter bed
cleaning
1000
100
MS2 in
MS2 out
WR1 in
WR1 out
10
1
0.1
0.01
0.001
0
2
4
6
8 10 12 14 16 18 20 22 24
Operational time (days)
Figure 1 The concentrations of bacteriophage MS2 and E. coli WR1 in the
influent and effluent of SSF2 during two separate spiking experiments with 24
hours of low and 24 hours of high concentrations; between these experiments, on
day 11, the filter bed surface was cleaned (Schmutzdecke was scraped off)
Column experiments. Due to the relative large scale and possible
health risk in the pilot plant, removal of oocysts of Cryptosporidium has
been determined in columns packed with ripened sand sampled 20 cm
from the top of the filter bed (without schmutzdecke; worst case). Main
objective of this part of the study was to compare the elimination of
bacteria, viruses and protozoan oocysts under the same conditions; thus
MS2 bacteriophages, E. coli WR1 and spores of Clostridium perfringens D10
were co-injected with C. parvum. The influent concentrations of the microorganisms ranged from 1.0x103 to 1.7x105 per ml.
The oocyst concentration in the influent of 103 – 104 per ml were removed
completely (<1 per 200 ml) by both columns with AWS and DWS sand.
This resulted in a DEC of at least 5 - 6 log (Table 3). In literature DEC values
of SSF for protozoan (oo)cysts assessed with spiking experiments were in
the same order of magnitude. For Giardia cysts 4 - 6 log and for
Cryptosporidium oocysts >4,5 - >6.5 log were described (Bellamy et al., 1985;
Schuler et al., 1991; Timms et al., 1995). Fogel et al. (1993) determined
concentrations of indigenous Giardia and Cryptosporidium (oo)cysts in the
___________________________________________________________________
- 166 -
Chapter 7
influent and effluent of a full-scale plant over a period of 1.5 year. From
these data a much lower DEC value of 1.2 and 0.3 log for Giardia and
Cryptosporidium, respectively, was calculated.
Figure 1 shows that MS2 phages and E. coli were hardly removed by the
AWS2 column (Figure 2), a result also observed in column AWS1 and DWS
(Table 3). MS2 and WR1 results of the AWS columns showed that coinjection with the other organisms did not affect the elimination of MS2 and
WR1. The columns removed spores of C. perfringens 2 - 3 log and more
efficient than MS2 and WR1 but less efficient than C. parvum. This indicates
that SSRC is not a suitable quantitative surrogate for Cryptosporidium oocyst
removal by SSF. Based on comparative spiking studies a similar conclusion
was reported by Emelko (2001) for aerobic spores as a surrogate for oocyst
removal by rapid granular filtration.
1
MS2-phage
Cout/Cin
0.1
E. coli
Cl. perfringens
0.01
0.001
0.0001
0.00001
0
500
1000
1500
2000
Operational time (min.)
Figure 2 Breakthrough curves from column AWS2 (no oocysts of C. parvum
detected in filtrate)
The column with DWS sand (grain size of 0.3-1.8 mm) showed more
elimination of the bacteria and bacterial spores than the column AWS2 with
finer sand (0.13-0.37 mm; Table 3). This difference was possibly caused by a
higher Carbon and iron oxyhydroxide content in the sand. In the DWS
sand concentrations of these components were 0.055% (dry weight) and 2.4
mg/g, respectively, and in the AWS sand <0,002% and 0.027 mg/g,
respectively. Thus DWS sand had a larger adsorption capacity, which
indicated that attachment plays a significant role in the elimination of
bacteria.
___________________________________________________________________
- 167 -
Chapter 7
Table 3 The Decimal Elimination Capacity (log10(Cin/Cmax
compared to DEC assessed for full-scale filters
.
out))
of the columns
C. perfringens
C. parvum
E.coli
a
a
SSRC
Coli44
MS2
Column AWS1; 0.4 m
nsb
0.2
0.4
nsb
Column AWS2; 0.4 m
0.3
0.2
2.3
>5.3
Full-scale AWS; 1.5 m
2.0c
1.7
-0.2 - 1.8d
nda
Column DWS; 0.4 m
0.4
0.1
3.2
>6.5
e
d
Full-scale DWS; 1.0 m
2.5
nd
-0.2 – 1.8
nd
a indigenous Coli44 and SSRC; b ns = no spiking; nd = not determined; c
average DEC for Coli44 from pilot plant filters in Table 2; d range of values
for SSRC from Table 1; e average DEC for Coli44 from full-scale filters in
Table 1
In Table 3 DEC of the column experiments (0.4 m of ripened sand) were
compared with DEC assessed in the pilot and full-scale filters of AWS (1.5
m of filter bed) and DWS (1.0 m of filter bed). This revealed that the
removal of MS2-phages and WR1 per unit length in the columns was much
less than observed in the slow sand filters. This may be attributed to the
fact that ripened sand from the schmutzdecke of the sampled filters have
not been included in the columns to investigate elimination under worst
case scenario. Moreover, also the higher porosity of the fresh packed
columns of about 0.40 compared to the porosity of the ripened filter bed
SSF1 of the pilot plant of 0.27 may have contributed to the observed
difference in removal capacity.
Although scaling up of the column results to full-scale conditions was not
possible, the resemblance of the breakthrough curves of MS2 and WR1
under both conditions substantiates extrapolation of the relative results to
full-scale filters. It was concluded that SSF must be very effective
eliminating peak concentrations of persistent oocysts of C. parvum and
spores of C. perfringens. However, the low and variable DEC of full-scale
filters for SSRC and also for oocysts found by Fogel et al. (1993) suggested
that overall elimination of both biological particles by SSF assessed over a
long period of time is influenced by accumulation and retarded transport.
Destructive sampling of the columns for analysis of retained microorganisms (Schijven et al., 2003) demonstrated that the most significant
removal mechanism for the small MS2 bacteriophages was adsorption. The
larger C. perfringens spores and E. coli were removed both by adsorption
___________________________________________________________________
- 168 -
Chapter 7
and physical straining and for the oocysts of Cryptosporidium physical
straining was the main removal mechanism. From the latter observation
one can conclude that elimination in the filter bed is irreversible and
retarded transport as proposed for SSRC is not possible. However,
unpublished data showed that oocysts might be transported deeper into
sandy soil due to changes in pH and conductivity. Also changes in
hydraulic conditions may attribute to transport of oocysts through sand
beds (Emelko, 2001). The same author emphasised the need for further
research to elucidate the significance of seeding experiments with high
concentrations as a tool to assess DEC of full-scale filters for
Cryptosporidium.
CONCLUSIONS
Based on the results of this study with E. coli (Coli44) and MS2bacteriophages DEC of slow sand filters for bacteria and viruses is
quantified at 2 – 3 and 1.5 - 2 log, respectively. One log higher was the
elimination of indigenous Campylobacter bacteria by full-scale filters, which
indicated that Coli44 is a conservative surrogate for these pathogenic
bacteria. MS2-bacteriophages can be regarded as a conservative surrogate
for virus removal (low attachment and survival rate). Moreover, viruses are
the most critical micro-organisms for the performance of slow sand filters
because they are retained the least. The column experiments showed that
SSF will have a large efficacy in eliminating peak concentrations of
persistent micro-organisms like spores of sulphite-reducing clostridia (>2 3 log) and (oo)cysts of Cryptosporidium and Giardia (> 5 log). This may be
more in full-scale filters with a schmutzdecke. On the basis of these results
SSRC is not a good quantitative surrogate for protozoan oocyst removal by
SSF. Further research is needed to elucidate if and to what extent
accumulation and retarded transport, the suggested phenomenon for the
low and variable removal of indigenous SSRC by full-scale filters, will
affect elimination of oocysts of Cryptosporidium in the long run.
Additionally, this study showed that filters with a schmutzdecke have 1 - 2
log greater capacity to eliminate bacteria, whereas elimination of viruses is
not affected by the schmutzdecke. This positive effect of the schmutzdecke
on elimination will also count for protozoan oocysts, because straining is an
important removal mechanism for these organisms. Furthermore, results
also suggested that charged (in)organic components accumulated in the
sand might have a significant effect on the removal efficiency of organisms
removed by adsorption.
___________________________________________________________________
- 169 -
Chapter 7
.
REFERENCES
Anonymous. 2001. Besluit van 9 januari 2001 tot wijziging van het
waterleidingbesluit in verband met de richtlijn betreffende de kwaliteit van voor
menselijke consumptie bestemd water, p. 1-53, vol. 31. Staatsblad van het
Koninkrijk der Nederlanden.
Bellamy, W. D., G. P. Silverman, D. W. Hendricks, and G. S. Logsdon. 1985.
Removing Giardia cysts with slow sand filtration. J. Am. Water Works Assoc.
77:52-60.
Cleasby, J. L., D. J. Hilmoe, and C. J. Dimitracopoulos. 1984. Slow sand and direct
in-line filtration of a surface water. J. Am. Water Works Assoc. 76:44-55.
Ellis, K. V. 1985. Slow sand filtration, p. 315-354, Critical Reviews in
Environmental Control, vol. 15. CRC Press.
Emelko, M. B. 2001. Removal of Cryptosporidium parvum by granular media
filtration. University of Waterloo, Ontario, Canada.
Fogel, D., J. Isaac-Renton, R. Guasparini, W. Moorehead, and J. Ongerth. 1993.
Removing Giardia and Cryptosporidium by Slow Sand Filtration. J. Am. Water
Works Assoc. 85:77-84.
Hijnen, W. A. M., G. J. Medema, and D. van der Kooij. 2004. Quantitative
assessment of the removal of indicator bacteria in full-scale treatment plants.
Wat. Sci. Technol.: Water Supply 4:47-54.
Hijnen, W. A. M., A. J. Van der Veer, J. Van Beveren, and G. J. Medema. 2002.
Spores of sulphite-reducing clostridia (SSRC) as surrogate for verification of the
inactivation capacity of full-scale ozonation for Cryptosporidium. Wat. Sci.
Technol.: Wat. suppl. 2:163-170.
Hijnen, W. A. M., J. Willemsen-Zwaagstra, P. Hiemstra, G. J. Medema, and D.
van der Kooij. 2000. Removal of sulphite-reducing clostridia spores by full scale
water treatment processes as a surrogate for protozoan (oo)cysts removal. Wat.
Sci. Tech. 41:165-171.
Karmali, M. A., A. E. Simor, M. Roscoe, P. C. Flemming, C. S. Smith, and J. Lane.
1986. Evaluation of a blood-free, Charcoal-based, selective medium for the
isolation of Campylobacter organisms from feces. J. Clin. Microbiol. 23:456-459.
Poynter, S. F. B., and J. S. Slade. 1977. The removal of viruses by slow sand
filtration. Prog. Water Tech. 9:75-88.
Ribeiro, C. D., and T. H. Price. 1984. The use of Preston Enrichment broth for the
isolation of thermophilic Campylobacters from water. J. Hyg. Camb. 92:45-51.
Schijven, J. F. 2001. Virus removal from groundwater by soil passage. Technische
Universiteit Delft, Delft, the Netherlands.
Schijven, J. F., H. A. M. de Bruin, S. M. Hassanizadeh and A. M. de Roda
Husman. 2003. Bacteriophages and clostridium spores as indicator organisms for
removal of pathogens by passage through saturated dune sand. Water Res.
37:2186–2194.
Schuler, P. F., M. M. Ghosh, and P. Gopalan. 1991. Slow Sand and Diatomaceous
Earth Fi1tration of Cysts and Other Particulates. Water Res. 25:995-1005.
___________________________________________________________________
- 170 -
Chapter 7
Slade, J. S. 1978. Enteroviruses in slow sand filtered water. J. Inst. Water Eng. Sci.
32:530-536.
Timms, S., J. S. Slade, C. R. Fricker, R. Morris, W. O. K. Grabow, K. Botzenhart,
and A. P. Wyn Jones. 1995. Remova1 of Cryptosporidium by Slow Sand Filtration.
Wat. Sci. Tech. 31:81-84.
Windle Taylor, E. 1969. The removal of viruses by slow sand filtration. Rep.
Results bac. chem. Biol. Exam. London Waters 1969-1970.
___________________________________________________________________
- 171 -
Chapter 8
Removal and fate of Cryptosporidium
parvum, Clostridium perfringens
and small-sized centric diatoms
(Stephanodiscus hantzschii) in
slow sand filters●
W.A.M. Hijnen1, Y.J. Dullemont2, J.F. Schijven3, A.J. Hanzens-Brouwer1, M.
Rosielle4 and G.J. Medema1
1Kiwa
Water Research, PO Box 1072, 3430 BB NieuwegeinNL
Waternet, Provincialeweg 21, 1108 AA Amsterdam, NL
3 National Institute of Public Health and the Environment, RIVM, PO Box 1, 3720
BA Bilthoven, NL
4 Het Waterlaboratorium, PO Box 734, 2003 RS Haarlem, NL
2
•
Reprinted from Water Research, 41: 2151-2162, Copyright 2007, with permission
from the copyright holder, Elsevier limited.
___________________________________________________________________
- 173 -
Chapter 8
.
ABSTRACT
The decimal elimination capacity (DEC) of slow sand filtration (SSF) for
Cryptosporidium parvum was assessed to enable quantitative microbial risk
analysis of a drinking water production plant. A mature pilot plant filter of
2.56 m2 was loaded with C. parvum oocysts and two other persistent
organisms as potential surrogates; spores of Clostridium perfringens (SCP)
and the small-sized (4-7 µm) centric diatom (SSCD) Stephanodiscus
hantzschii. Highly persistent micro-organisms that are retained in slow sand
filters are expected to accumulate and eventually break through the filter
bed. To investigate this phenomenon, a dosing period of 100 days was
applied with an extended filtrate monitoring period of 150 days using large
volume sampling. Based on the breakthrough curves the DEC of the filter
bed for oocysts was high and calculated to be 4.7 log. During the extended
filtrate monitoring period the spatial distribution of the retained in the filter
bed was determined. These data showed little risk of accumulation of
oocysts in mature filters most likely due to predation by zooplankton. The
DEC for the two surrogates, SCP and SSCD, was 3.6 and 1.8 log,
respectively.
On basis of differences in transport behaviour, but mainly because of the
high persistence compared to the persistence of oocysts, it was concluded
that both spores of sulphite-reducing clostridia (incl. SCP) and SSCD are
unsuited for use as surrogates for oocyst removal by slow sand filters.
Further research is necessary to elucidate the role of predation in
Cryptosporidium removal and the fate of consumed oocysts.
INTRODUCTION
One of the pathogens of major concern for the drinking water industry is
Cryptosporidium, a persistent pathogenic protozoan and cause of a number of
outbreaks of waterborne diarrhoea documented in the USA and the UK
(Richardson et al., 1991; MacKenzie et al., 1994). Studies have demonstrated
the failure of regular water quality monitoring with Escherichia coli to indicate
the absence of this pathogen in drinking water (Harwood et al., 2005). This
shortcoming of current quality control is overcome by the use of
Quantitative Microbial Risk Assessment (QMRA) to define the
microbiological safety of drinking water (Haas, 1983; Medema et al., 2003).
In the new Dutch Drinking Water Decree “Waterleidingbesluit”
(Anonymous, 2001) a maximum acceptable annual infection risk of 10-4 has
been introduced for pathogens of fecal origin including Cryptosporidium.
___________________________________________________________________
- 174 -
Chapter 8
The required risk level has to be demonstrated with QMRA which is based
on dose-response relation in human volunteers and exposure assessment.
For exposure assessment quantitative knowledge about the presence of
pathogenic micro-organisms in the source water and the capacity of water
treatment processes to eliminate these micro-organisms is required, along
with data on drinking water consumption.
Slow sand filtration is one of the oldest water treatment processes used to
produce microbiologically safe drinking water. Quantitative information of
how effective these filters are in removing pathogens, however, turned out
to be limited (LeChevallier and Au, 2004). It is only since the last part of the
20th century that some studies were published on the elimination capacity
for protozoan (oo)cysts (Bellamy et al., 1985; Schuler et al., 1991; Fogel et al.,
1993; Timms et al., 1995). In most of these studies the actual Decimal
Elimination Capacity (DEC; log removal) of slow sand filters for Giardia
and Cryptosporidium (oo)cysts was assessed by dosing peak concentrations,
resulting in high DEC-values of 4 to more than 6 log in matured and
relatively small sized filter beds. In contrast with these findings, one study
(Fogel et al., 1993) determined removal of environmental Giardia and
Cryptosporidium (oo)cysts by full-scale filters by prolonged sampling and
showed breakthrough during periods with low temperature (<5oC)
resulting in low average DEC-values of 1.2 and 0.3 log, respectively. In the
Netherlands, slow sand filters are operated as the last stage of a multiple
barrier treatment. Hence, assessment of the DEC for environmental
protozoan (oo)cysts is not feasible. The removal of spores of sulphitereducing clostridia (SSRC; including Clostridium perfringens) as a surrogate
for protozoan (oo)cysts (Payment and Franco, 1993; Hijnen et al., 1997) by
full-scale filters was monitored (Hijnen et al., 2000; 2004). Results showed
low DEC for SSRC, ranging from -0.2 to 1.8 log. Direct comparison of the
removal of dosed spores of C. perfringens and oocysts of C. parvum in a
short sand column (0.5 m) showed 2.3 - 3.2 log removal of the spores and
more than 5 log removal of the oocysts, indicating that environmental
SSRC is a conservative surrogate for protozoan (oo)cysts (Hijnen et al.,
2004). However, the clear discrepancy between the high removal of dosed
spores and the low removal of SSRC by mature full-scale filters sand casts
doubts on the validity of translating column study results to the larger fullscale filters. Because spores are very persistent (Medema et al., 1997), it was
hypothesized that spores accumulate in the filter bed and they might
detach which causes remobilization and delayed breakthrough. This
phenomenon was first suggested as cause for the relative low removal of
SSRC by granular activated carbon filters (GAC) with high SSRC
concentrations in the filter bed and the back wash water (Hijnen et al.,
___________________________________________________________________
- 175 -
Chapter 8
.
1997). Additional observations supporting this hypothesis have been
described in literature for protozoan-sized microspheres in soil (Harvey et
al., 1995), 1 µm fluorescent microspheres in GAC and expanded clay filters
(Persson et al., 2005), aerobic spores in GAC filters (Mazoua and
Chauveheid, 2005) and C. parvum oocyst transport in sand columns (Harter
et al., 2001; Bradford and Bettahar, 2005; Hijnen et al., 2005) and E. coli and
SSRC in the latter study. Furthermore in a column study it was
demonstrated that the removal of spores depended on the duration of the
seeding (Schijven et al., 2003). As with SSRC, oocysts of C. parvum are also
persistent, but oocysts are larger by a factor of 3 - 4 and have potentially
different surface properties. As a consequence, it remains unknown
whether the phenomenon of reduced elimination by sand filters over time
due to accumulation could apply to these pathogens as well.
To investigate this remaining question a study was conducted with the
objective of describing and quantifying the removal, accumulation and
delayed breakthrough of C. parvum oocysts in a slow sand filter of some
extent (2.56 m2) operated under full-scale conditions mimicking as much as
possible long term loading conditions. Spores of C. perfringens as well as a
small-sized algae, the centric diatom Stephanodiscus hantzschii, were dosed
simultaneously to compare removal of different kind of persistent microorganisms and to investigate the potential use of both organisms as
surrogate for the assessment of protozoan oocysts removal by slow sand
filtration (Hijnen et al., 1997; Akiba et al., 2002).
MATERIALS AND METHODS
Slow sand filter and the experimental set up. Slow sand filtration in
the water treatment Leiduin of Water Company Waternet is the last stage in
a treatment train with rapid sand filtration, ozonation and granular activated
carbon (GAC) filtration. The source water is pre-treated water from the River
Rhine after dune passage and regained by an open collecting system. The
filter used for this study was part of a pilot plant, a dummy system of the
full-scale treatment. The filter has been in operation for 3 years without
surface scraping at a flow rate of 1 m3 h-1 and a filtration rate of 0.3 m h-1
prior to carrying out the present study. The influent, GAC filtrate, had a low
DOC (1.5 mg C l-1) and turbidity (0.1 FTU), a pH of 8.0 and temperature
ranging from 8.2 to 18.8oC. The filter bed (2.56 m2 and depth of 1.5 m) was
filled with sand (d50 = 0.28 mm; porosity of 0.41) on top of a support of
Nicolon cloth (0.7 mm) and 0.4 m gravel pack. The experimental set up is
presented in Fig. 1 and will be described in detail hereafter.
___________________________________________________________________
- 176 -
Direct anaerobic colony count
in pasteurized samples after
dosing
Influent; filtrate of granular
activated carbon filters
98 days dosing
Spores of C. perfringens in a
washed suspension in sterile
tap water
Direct microscopic count in
settled samples before and
after dosing
Cells of S. hantzschii in a
cultivated Diatom Medium
Hemoflow cross-flow filtration
Col. count as influent
Membrane filtration
Microscopic count as influent
- 177 -
Figure 1. Diagram of the experimental set up
Isolation/analytical methods:
Hemoflow cross-flow filtration
USEPA Method 1623
250 days effluent sampling
Filter bed sampling at day 184 and 253
Filter bed (2.56 m2): 1.5 m of sand (Ø 0.28 mm) and rate of 0.3 m h-1;
Analytical methods:
Direct microscopic count, and
no IMS purification in samples
after dosing
Dosed micro-organisms:
UV irradiated Oocysts of
C. parvum in washed faecal
suspension (Cerivine origin)
Chapter 8
Chapter 8
.
Cryptosporidium oocysts. C. parvum oocysts suspensions in watery
faeces (Cervine origin; Moredun scientific Ltd., Midlothian, UK; 2x109) were
used. Each tube of suspension was purified by filtering through plankton
netting with a mesh width of successively 500, 300 and 100 µm, the netting
was flushed with sterile water which was mixed with the filtrate. This
suspension was mixed during 30 minutes (CAT S50) and subsequently
settled during 30 minutes. The water was decanted and the sediment was
flushed twice with sterile water. The accumulated water suspension was
centrifuged 10 minutes (4000 RPM) and the pellet was concentrated to a
volume of approximately 100 ml with an estimated number of 4.5x108. This
pre-treated suspension was divided into three portions each dosed during
one week to the influent. For safety reasons the oocyst suspensions were
inactivated with UV in a collimated beam apparatus (calculated UV fluence
of ≥10 mJ.cm-2) prior to the dosage.
Clostridium perfringens. A suspension of spores of C. perfringens (SCP
strain D10; an isolate from a patient suffering from diarrhoea caused by food
infection) was prepared with a total number of spores of 2.2x109 following a
previously described method (Hijnen et al., 2002). The level of sporulation of
the suspension was 100% and tested by enumeration on Perfringens-agarbase plates (PAB Oxoid CM587) with and without pasteurization for 15
minutes at 60oC.
Stephanodiscus hantzschii. S. hantzschii is a small-sized centric
diatom (SSCD) with an average size of 5 µm. SSCD-cells were obtained from
a pre-cultured strain of the Culture Collection of Algae and Protozoa (CCAP
1079/4; Dunstaffnage Marine Laboratory, Argyll, UK, Th. Pröschold). The
strain was inoculated into a gently stirred 150 ml sterilized Diatom Medium
(DM) for fresh water algae (Beakes et al., 1988) and incubated for 2-3 weeks
at 20±2oC with a light-dark cycle of 12:12 hours (white light intensity of 50
µE m-2 s-1). The growth and conditions of the algae were monitored
regularly by light microscope. The concentration of cells in the final
suspension of ±1.5x108 ml-1 was determined using a haemocytometer.
Dosage procedure. Appropriate volumes of the pre-treated oocyst, the
C. perfringens and S. hantzschii suspensions were added to 150 l of drinking
water in a 200-l stainless steel (SS) vessel with cover. This diluted and stirred
suspension, which was freshly prepared every week, was dosed
continuously into the mains of the influent of the filter at a flow rate of 13 ml
min-1 during the first two weeks. Because in this vessel biofouling occurred,
from week three the oocysts suspension was dosed (0.7 ml min-1) separately
from a separate 20 l suspension in drinking water prepared fresh weekly in a
30-l SS vessel with cover and stirrer, cooled in a container with melting ice.
___________________________________________________________________
- 178 -
Chapter 8
Dosing of the SCP suspension was also from this cooled 20 l vessel after
seven weeks. The total dosage period was 98 days (14 weeks) in which
dosage failed for SCP and SSCD in week 4 and for oocysts in week 5 (tubing
not correctly connected; actual dosage period was 91 days).
Large volume sampling. Large volume sampling was considered
necessary to determine organism concentration in the filtrate. Water volumes
of 1,000 l or more were concentrated on-site using a Hemoflow cross-flow
ultrafilter (HF80S; Fresenius, Germany; Simons et al., 2001). The water
volumes sampled with the Hemoflow were concentrated to a total volume of
800 ml HF-concentrate. Approximately 5% of the HF-concentrate was used
to determine the number of SSRC (including C. perfringens); the residual HFconcentrate was used to enumerate the number of oocysts.
The concentration of small-sized centric diatoms (SSCD) in large volumes of
effluent was determined using the Membrane Filtration or MF-sampler
(Hijnen et al., 2000). Algal cells were isolated on poly-carbonate membrane
filters (2 µm; Sartorius) and re-suspended in tap water by 2 minutes of ultrasonic treatment (low energetic; Branson 5200, Danbury, USA). The recovery
of this method was 53 - 100% (n=6) determined with chemically purified
centric diatoms in the size range of 6-7 µm (M. Rosielle, personal
communications).
Sampling program. The filter bed load of C. parvum oocysts and
spores of C. perfringens as SSRC, was determined weekly or more by
sampling the influent of the filter as well as the water volume above the filter
bed. The concentration of S. hantzschii dosed to the influent was determined
as SSCD in the water above the filter bed. Because of the relative high
background concentration of SSCD in the influent of the pilot plant, the
SSCD concentration was also determined before the dosage point of S.
hantzschii. The effluent of the slow sand filter was sampled weekly during
the dosing period and a period of 22 weeks after the dosage stopped to
determine delayed breakthrough.
Analysis. Oocysts of Cryptosporidium in the influent and the HFconcentrate were enumerated with the EPA method 1623 (USEPA, 2001).
Recovery of the HF-concentrate was tested by addition of 5 ml Colorseed™
(BTF Decisive Microbiology, North Ryde, NSW, Australia) suspension (99
oocysts ±1.3). The HF-concentrate was purified by immuno-magnetic
separation (IMS) prior to labeling and microscopic counting. The adsorbed
oocysts were eluted from the beads with HCl (0.1 N) and this suspension
was neutralized with KOH (1 N). The concentrate was collected on a
membrane filter (Millipore, 1,2 µm RTTP) and labeled with FITC (2 hours at
37oC). The membrane was microscopically counted with epi-fluorescence
microscopy (Leica DM RXA, magnification 312.5).
___________________________________________________________________
- 179 -
Chapter 8
.
The concentration of spores of sulphite-reducing clostridia (SSRC, including
C. perfringens) was determined either directly or after concentration by
membrane filtration or HF-filtration. All samples were pasteurized before
enumeration. In the first weeks SSRC in the effluent was measured by
membrane filtration (1 or 10 l; (∅47 mm, 0.45 µm pores; Sartorius 13906-50ACN) and pasteurization of the membranes (30 min. at 70±1oC; Dutch
NEN6567). Given that no SSRC were detected in these samples, SSRC in the
effluent was subsequently monitored in the HF-concentrate. Fifty ml of this
concentrate was pasteurized (15 min. at 60±1oC; ISO-method 6461/2-1986 for
C. perfringens spores) prior to cultivation in PAB medium (Oxoid CM 587)
under anoxic conditions during 48±4 hour at 37±1oC. A real-time PCR
method was used to confirm that the SSRC colonies observed in the PAB
medium were spores of C. perfringens. Cell material was isolated from the
black colonies, suspended in 1 ml sterile water and subsequently tested with
the real-time PCR for DNA of C. perfringens. The primers which codes with
A-toxine gen of C. perfringens (Yoo et al., 1997) were used to identify the C.
perfringens.
For enumeration of the breakthrough of cells of S. hantzschii the
concentration of SSCD with specific cell morphology and size (4 – 7 µm) is
determined microscopically in concentrates of the influent and the effluent,
prepared by sedimentation. Water samples were pretreated with lugolsolution until the sample was light orange to stop biological activity. These
samples were pored into Hydro-bios sedimentation chambers (Hydro-bios
Apparatenbau, Kiel-Holtenau, GMB) with appropriate size (10 – 50 – 100 ml)
and settled during 24 - 72 hours. Water was decanted from these tubes and
the sediment was examined microscopically (Olympus IX70, Olympus,
Zoeterwoude, NL; magnification – 400x).
Enumeration of retained micro-organisms. After the dosage of microorganisms was stopped, delayed breakthrough of the dosed microorganisms was monitored. After 184 days, when oocyst concentration in the
filtrate was below the detection limit (DL) of the analysis, filtration was
stopped and the water volume above the filter bed was lowered to the
surface. At two locations the saturated filter bed was scraped with a scoop to
sample the Schmutzdecke (approximately 2-3 mm) and the following 50 cm
of the saturated filter bed was sampled with a soil core sampler (veenboor;
Eijkelkamp, Giesbeek, The Netherlands). This was repeated after 253 days
where sand from the first 5 cm and from layers deeper than 50 cm was
collected. At first the presence of C. parvum oocysts in the upper part (0 – 1
cm) was monitored microscopically by fixing sand grains on a Dynal slide.
The grains were labeled with FITC (2 hours at 37oC) and completely scanned
___________________________________________________________________
- 180 -
Chapter 8
for oocysts. To determine the concentration of retained micro-organisms in
the sand, 3 - 4 gram of the sampled sand from the two locations in the filter
bed was suspended in 100 ml sterile tap water and treated for 2 minutes in a
low energetic ultra-sonic cleaner (Bransonic 5510, Branson ultrasonic,
Danbury, USA). The number of micro-organisms eluted from the sand was
enumerated in this sonicated sand suspension (UTSusp.) with the analytical
methods described above. Oocysts and spores in the Schmutzdecke material
and the sand of the first cm in the filter bed (0 – 1) were examined in two
separate sub-samples of 3 - 4 gram. A third sub-sample was treated
ultrasonically in 100 ml sterile Laureth-buffer (USEPA, 2001) to verify the
elution efficiency of the ultra-sonic treatment for oocyst enumeration. Fifty
ml of the UTSusp. was used to enumerate oocysts and spores and the other
50 ml was used to enumerate the SSCD with the methods previously
described. Microscopic counting was used to enumerate SSCD in the
UTSusp. and because these suspensions had a high content of suspended
solids, the collected concentrations must be seen as an indication of the order
of magnitude of centric diatoms in the filter bed. Statistical differences
between concentrations of retained micro-organisms in the filter bed were
analyzed with the Student t-test or the Wilcoxon-test using SPSS (14.0).
ATP and zooplankton analysis in the sand. The concentration of
adenosine tri-phosphate (ATP) in the sand was measured (Magic-Knezev
and van der Kooij, 2004). The presence of zooplankton in the sand was
measured in separate sand samples of approximately 500 gram taken from
the filter bed at the end of the operational time. 30 gram of sand was mixed
intensively in tap water. The invertebrates were separated from the
sediment by separation in a MgSO4 solution in tap water (49 g l-1). The sand
slurry was tested for the presence of larger invertebrates by filtration over
500 µm sieve and subsequently pored in an Anderson glass tube
(Anderson, 1981) filled with the MgSO4 solution. After 15 – 20 minutes of
sedimentation the zooplankton sample (±800 ml) was taken at the upper
sampling port and subsequently at the lower sampling port. The
suspensions were sieved subsequently through a 30 µm sieve and loaded in
a counting chamber for microscopic examination (Olympus IX70, Olympus,
Zoeterwoude, NL; magnification 100x) to count the number and identify
the zooplankton in the sand.
Elimination of the micro-organisms and mass balance. The Decimal
Elimination Capacity (DEC) of the filter bed for the tested micro-organisms
_
was calculated from the average concentration in the influent C in and the
_
effluent C out , the latter calculated by
___________________________________________________________________
- 181 -
Chapter 8
.
_
C out =
∑N
∑V
_
out
R out
(1)
out
where Nout is the number of oocysts counted, Vout the sampled volume (l)
and Rout the recovery in the tested samples.
_
DEC = log10
C in
(2)
_
C out
DEC was also calculated from the mass balance by the following
equation
⎡M ⎤
DEC m = log10 ⎢ d ⎥
⎣Me ⎦
(3)
where DECm is the elimination capacity on the basis of the mass balance,
−
n
C out ,i + C out ,i +1
i =1
2
M d = t d C in and M e = ∑ t i
the total number of micro-
organisms dosed to the influent and found in the effluent, respectively, td is
the time period of dosing (hours) and dti is the i-th time interval of n
intervals between two successive samples with concentrations Cout,i and
Cout, i+1.
The mass of micro-organisms (numbers) accumulated in the filter bed Mb
after an operational time of 184 and 253 days is derived from
−
M b = ∑ M l ,n = ∑ (C l ,n 1000)(d l ,n A1000)
(4)
_
where Ml,n is the total number of micro-organisms and C n ,l the average
concentration of micro-organisms (N ml-1) in n layers (l) with a thickness of
dl,n (m) of the filter bed, respectively with a surface area A of 2.56 m2. Ml,n in
the un-sampled layers was calculated from the average concentration
extrapolated from the concentrations of the next upper and lower layers.
Because of the inaccuracy of the concentrations in the filter bed, mass
balance calculations was not done for SSCD.
Calculation of Collector and Sticking Efficiencies. Assuming
elimination in the filter bed was only due to attachment and detachment,
the sticking efficiency of the spores and oocysts was calculated from the
colloid filtration model (Yao et al., 1971) described with the equation
LN
C
3 (1 − θ )
=−
αηL
C0
2 dc
(5)
___________________________________________________________________
- 182 -
Chapter 8
where dc is the diameter of the collector, α the sticking efficiency, η the
single collector collision efficiency, and L the length of the column. η was
calculated with the optimized equation presented by Tufenkji and
Elimelech (2004a). For the calculations the following parameters values
were used: bulk water density 999.703 kg m-3; Hamaker constant for
bacterium glass water interface 6.2x10-21 J (Rijnaarts et al., 1995); sizes (m) of
spores of C. perfringens 1.5x10-6 and oocysts of C. parvum 4.9x10-6 (Medema
et al., 1998); and ρp C. perfringens 1270 (Tisa et al., 1982) and oocysts 1045
kg.m-3, respectively (Medema et al., 1998).
RESULTS
Cryptosporidium removal. Oocysts of C. parvum were dosed to the
GAC filtrate, the influent of the SSF. Based on the low concentration of
environmental oocysts observed in the source water during the experiment
(routine monitoring data, not presented) and the preceding removal in the
rapid sand filter and GAC-filtration, the background concentration of
oocysts in the influent was negligible. The oocyst concentration measured in
the influent after dosage and in the water above the filter bed was constant
with an average concentration of 314.6 (±161) N l-1 (Table 1; average recovery
of 70.6 (±13) %; n = 25).
Table 1. The concentration of dosed Cryptosporidium oocysts, spores of C.
perfringens (SCP) and environmental and small-sized centric diatoms SSCD (4 –
7 µm; incl. S. hantzschii) in the influent and effluent of the slow sand filter
Samples
na
n+a
Sampled
Detected
Conc.
STDc
Vol. (l)
organisms
n/l
Oocysts inb
33
33
3.3
733
314.6
161
Oocysts outb
22
7
26.613
18
0.0016
NDc
SCP in
20
20
2.0
2270
1135.0
1014.8
SCP out
21
16
2.363
336
0.142
0.509
d
SSCD in BG
21
17
1.51
140
92.9
889.4
SSCD in
17
17
0.98
274
279.6
92.8
SSCD out
29
29
25.94
187
7.22
8.46
a n and n+, number of samples and number of positive samples; b corrected for
recovery; c STD = standard deviation; ND = not determined; d Background
concentration of environmental SSCD (oocysts and SCP below detection limit).
During the dosing period of 98 days and an additional period of 163 days
after dosing had stopped, 26 m3 of the filtrate was examined for oocysts.
___________________________________________________________________
- 183 -
Chapter 8
.
After the first 29 days and during a period of approximately 50 days, 14
oocysts were detected in approximately 50% of the samples of the filtrate
(Fig. 2).
C. parvum oocysts (N/l)
10000
Influent
1000
Effluent
100
Effluent <DL
10
1
Sampling of the
filter bed
0.1
0.01
0.001
0
50
100
150
200
250
10000
Influent
Spores (N/l)
1000
Effluent 70°C MF<DL
100
Effluent 60°C HF
Effluent 60°C HF <DL
10
1
0.1
0.01
0.001
0
50
100
150
200
250
Centric diatoms (N/l)
10000
Increased dosage
24 hours
1000
100
10
1
Influent before dosage
Influent after dosage
Influent before <DL
Effluent
0.1
0.01
0
50
100
150
200
250
Operational time (days)
Figure 2. Concentration of C. parvum oocysts, spores of C. perfringens and small
sized centric diatoms (SSCD) in the influent and the effluent of the slow sand filter
(DL is detection limit)
___________________________________________________________________
- 184 -
Chapter 8
A second shorter period (25 days) of oocyst breakthrough was observed
shortly after dosage was stopped. From the total number of 18 oocysts
detected in the volume sampled during the total monitoring period of 247
days an average concentration of 0.0016 oocysts l-1 was calculated (Table 1;
average recovery of 43.6 (±15.8) %; n = 22) resulting in a DEC-value of 5.3
log.
Removal of spores of sulphite-reducing clostridia. Also for SSRC the
background concentration of environmental SSRC in the influent was
neglected (data derived from routine sampling and not presented). The
average concentration of spores of C. perfringens (SCP; confirmed with
molecular typing on a few samples) was 1135 (±1015) N l-1 (Table 1). The
temporary decline in influent concentration after 40 days (Fig. 2) was solved
by switching dosage from the large 150-l vessel used for S. hantzschii dosage
to the smaller and permanently cooled 20-l vessel used for C. parvum oocysts.
During the first weeks no SCP were observed in effluent sample volumes of
1 and 10 l and breakthrough of these spores was detected starting from day
36 and further (Fig. 2). The highest breakthrough of SCP was at day 44, 15
days after the first highest oocyst breakthrough (Fig. 2). Based on the total
number of SCP observed in sampled effluent volume of 2.363 l the average
concentration was 0.142 (±0.509) SCP l-1 (Table 1) resulting in a DEC-value of
3.9 log for these organisms
Removal of small-sized centric diatoms. Compared to the
Cryptosporidium and SCP, SSCD (4 – 7 µm) were present in higher and more
variable concentrations in the source water of this treatment plant.
Concentrations ranged from <1 up to 5.4x105 l-1. In early spring or late
autumn diatom bloom in the source water reservoir caused peak
concentrations. The average SSCD concentration in the source water was a
factor of approximately 200 higher than the concentration of SCP and was
reduced only slightly (0.4 log) by the rapid sand filters (data not presented).
Because of these high levels of SSCD in the source water, SSCD were also
determined in the influent before dosage of S. hantzschii (Fig. 2). The
concentrations of SSCD in the filter influent were highly variable and only
slightly below the level of dosed S. hantzschii. Thus, environmental SSCD
concentrations contributed 40 – 50% on average to the total load of the test
filter with these micro-organisms (Table 1) and could not be separated from
cells of S. hantzschii in the microscopically enumeration method. The results
of SSCD monitoring in the effluent showed a clear breakthrough (Fig. 2),
much larger than observed for C. parvum oocyst and C. perfringens spores.
After an initial high peak in the effluent, breakthrough stabilized. Because of
the mixed loading of S. hantzschii with environmental SSCD, and the
variability herein, the DEC for these micro-organisms was not calculated
___________________________________________________________________
- 185 -
Chapter 8
.
from the average concentration in the influent and effluent but from the
actual concentrations determined at the same day (n = 12). The average DEC
was 1.8 ± 0.6 log (range of 0.9 – 2.6 log).
Concentration of retained micro-organisms. An important objective
of the experiment was to monitor delayed breakthrough behaviour of
oocysts and its potential surrogates, SCP and SSCD. The filter bed was not
disturbed until breakthrough of oocysts was below the detection limit (DL;
0.002 n l-1; Fig. 2). After 184 days sand samples were taken from the filter bed
at two places. This interruption of the filtration process and locally
disruption of the filter bed caused no additional oocyst breakthrough (Fig. 2).
The maximum concentration of oocysts per ml sand was 160 (±151, n = 6 and
range of 57 – 426) observed in the top of the filter bed (Fig. 3a) using the
method of elution in sterile water (ultrasonic treated sand suspension;
UTSusp.) and declined with increasing depths to a detection limit of 1
oocysts ml-1 after 43 cm. Ultrasonic treatment in 100 ml laureth buffer (LB)
applied to the samples of the Schmutzdecke (SM) and the first centimetres
did not increase the oocyst recovery from the sand (Fig. 3b).
C. parvum
1000
1000
100
Concentration (N/ml)
C. perfringens
- LB (n=4)
+LB (n=2)
1000000
b
100000
10000
100000
10
100
d
1000000
1000
1
SM
0.5 cm 3 cm
184 days
SM
0.5 cm 3 cm
100
10000
SM 0.5 cm 3 cm
253 days
SM
184 days
10
0.5 cm 3 cm
253 days
1000
100
1
184 days
253 days
253 days <DL
a
0.1
SM
0
10
184 days
253 days
253 days <DL
c
1
25
50
75
100
Depth (cm)
125
150
0
SM
25
50
75
100
Depth (cm)
125
150
Figure 3. The concentration (error bars = range of values; DL is detection limit) of
C. parvum oocysts and of C. perfringens spores over the total filter bed (a and c)
and the first 3 cm (b and d) of the filter bed after 184 and 253 days of operation (86
and 157 days after finishing with dosing, respectively) with the additional
concentrations of oocysts obtained after laureth buffer elution (+LB; n = 2) during
the sampling at day 184 (b)
___________________________________________________________________
- 186 -
Chapter 8
When the experiment was stopped after 253 days of operation the filter bed
was sampled again on two different locations. The concentration of oocysts
in the first 5 cm of the filter bed was lower than the concentrations
determined at day 184 (Fig. 3a).
The difference in oocyst concentration observed after 184 and 253 days in the
layer 1-5 cm was significant (P=0.03). Due to the high standard deviations of
average concentrations at the top of both samplings the difference was not
significant. All oocyst concentrations were corrected individually for the
recovery of Colorseed™ seeded in the UTSusp. (24.8 ±8.1%). The high
performance of the elution method in sterile water was also demonstrated by
direct microscopic counting of oocysts in the samples. Part of the
schmutzdecke samples and sand samples of the first centimetre were
examined microscopically for the presence of oocysts. One layer of sand
grains was fixed in the well of a Dynal slide with a diameter of 9.26 mm and
all fields of 0.39 mm2 (average grain number of three) were scanned (total of
518 grains on the top and partly the side). No oocysts were detected and on
the basis of the average grain size of 0.3 mm which results in a grain number
of 70,771 N ml-1 of sand and a scan factor of 0.6 (under and part of the sides
of the grains could not be scanned) a detection limit of 228 oocysts ml-1 of
sand was calculated, which was slightly higher than the maximum observed
numbers determined by the elution method (Fig. 3a).
Appropriate fractions of the UTSusp. were used to determine SCP (1 – 10 ml)
and SSCD (50 ml) concentrations. The SCP concentration in the sand at day
184 was a factor of 100 higher than the oocyst concentration (Fig. 3a and 3c)
and declined to 1 – 10 ml-1 in the first 50 cm of the filter bed. The large
difference in accumulation in the top of the filter bed between oocysts and
spores could not be explained by the difference in influent concentration and
observed breakthrough (Fig. 2). This indicates that there is a difference in fate
of the retained oocysts and spores. For SCP the concentrations in the sand
sampled at day 253 were in the same order of magnitude as monitored after
184 days (Fig. 3b). 100% of a total of 90 SCP colonies which were molecularly
tested, were identified as C. perfringens.
Microscopic counting of SSCD in the UTSusp. was hampered by the high
content of suspended solids and thus, the calculated concentrations show the
order of magnitude of the presence of these organisms in the filter bed. At
day 184, low concentrations of SSCD (<0.3 – 8.9 N ml-1) were observed in the
sand. These concentrations were a factor of 10 – 1000 below the numbers of
oocysts and spores (Fig. 3) and indicate a relative low adsorption capacity of
the sand for SSCD, which agreed well with the high level of breakthrough in
this filter (Fig. 2). At day 253 the estimated concentrations of SSCD in the
schmutzdecke material (430 – 2029 N ml-1) and the sand from the layer of 1 ___________________________________________________________________
- 187 -
Chapter 8
.
5 centimetre (<0.4 – 492 N ml-1) were clearly higher than the SSCD
concentrations observed at day 184. A relation with the autumn bloom of
SSCD in the source water is suspected; day 253 was in the beginning of
November. Two of the three peak SSCD concentrations observed in the
influent of the filter bed during this period were monitored in between the
filter bed sampling at day 184 and day 253 (Fig. 2).
ATP and zooplankton in the sand. The spatial distribution of ATP in
the filter bed showed a higher (micro)biological activity in the schmutzdecke
and the first 5 cm of the bed than deeper in the bed (Fig. 4). ATP
concentrations after 184 and 253 days of operation were similar. The
determination of zooplankton in the sand of the filter bed revealed high
concentrations of Nematoda, Testacea (observed species Centropyxis, Cyphoderia
and Euglypha), Oligochaeta, Rotatoria (observed species Colurella, Lecane,
Lepadella and other unidentified species), Nauplii and Harpacticoida and some
Cyclopoida in the top 5 centimeters of the filter bed (Fig. 4). Next to these
organisms Ciliata were detected in lower concentrations at filter depths of 50
and 100 cm.
250
1000
Nematoda
Testacea
Polychaeta
Rotifera
Nauplius
Harpacticoida
Cyclopoida
Ciliophora
184 days ATP
253 days ATP
Organisms per 30 g
ATP ng mL-1
200
150
100
100
10
50
1
0
SM
0.5
3
12.5
Depth (cm)
27.5
41.5
2.5
50
100
Depth (cm)
Figure 4. The ATP concentration (±SD), after 184 and 253 days of operation (at
depth of <50 cm, ATP < 10 ng mL-1 of sand) and the zooplankton concentration per
30 gram of sand after 260 days at three depths
Mass balance and CFT parameters. The mass balance for oocysts and
spores (in numbers) was calculated to determine the elimination and
accumulation in the filter bed (Table 2). From the calculated number of
___________________________________________________________________
- 188 -
Chapter 8
organisms dosed to the influent and passed though the filter bed in the
filtrate the DECm was calculated. The DECm for oocysts and SCP was 4.7 and
3.6 log, respectively, approximately 0.5 log lower than the DEC calculated
from the average concentrations. The DECm-values are most likely more
accurate than DEC calculated from the average filtrate concentration where
zero counts are included as zero, thus underestimating the actual average
concentration. The accumulated number of C. parvum oocysts Mb determined
in the filter bed after 184 days was only 1.8% of the total dosed oocyst
number and decreased to 0.2% at day 253. For spores of C. perfringens, Mb
after 184 and 253 days was 32.5 and 45% of the dosed spores, respectively.
The difference in Mb between both sampling dates is most likely caused by
the variability in SCP concentrations determined on four different locations
in the filter bed (fig. 3d).
Table 2. The mass balance data (dosed organisms (Md) and organisms in filtrate
and filter bed (Me; Mb) and the colloid filtration parameters (collector efficiency η
and collision efficiency α)
Md
C. parvum
5.4x108
C. perfringens
1.9x109
Me
9.8x103
5.2x105
Mb 184 daysa
Mr 184 days (%)c
9.7x106 ± 3.3x106
1.8 ± 0.6
7.0x108 ± 4.0x108
32.5 ± 17.2
Mb 253 daysb
1.1x106 ± 2.6x105
8.9x108 ± 3.5x107
Mr 253 days (%)c
0.21 ± 0.05
45 ± 3.3
η
α
0.0112
0.207
0.006
0.310
Number of oocysts and spores in 0.3 m of the filterbed sampled at two
locations; b Number of oocysts and spores in 0.05 m of the filter bed
sampled at two locations; c Recovered mass (% of dosed numbers) in filter
bed after 184 en 253 days of operation
a
The collector efficiency η for oocysts in the filter bed of the present study was
a factor of two higher than η for spores of C. perfringens (Table 2). Despite the
lower η a higher collision efficiency α was calculated for the removal of the
spores, indicating a higher attachment in the filter bed compared to the
oocysts.
___________________________________________________________________
- 189 -
Chapter 8
.
DISCUSSION
Removal of Cryptosporidium and surrogates. The results of the study
clearly demonstrate the high capacity of mature slow sand filters with a well
developed schmutzdecke for removal of Cryptosporidium from water. During
the prolonged loading of the pilot filter very few oocysts passed the filter bed
and on the basis of the collected data a Decimal Elimination Capacity DECm
of 4.7 log was calculated. This removal rate was similar to the removal rates
assessed with short term dosing experiments in slow sand filters operated on
laboratory and pilot plant scale (Bellamy et al., 1985; Schuler et al., 1991;
Timms et al., 1995; Hijnen et al., 2004). The present study also showed that
slow sand filters have a high capacity to remove spores of C. perfringens (SCP;
DECm of 3.6 log). Consequently, with the used experimental set up low DECvalues for both persistent micro-organisms due to delayed breakthrough
could not be demonstrated. The height and duration of the loading with
oocysts and spores in the current experiment was probably not enough to
achieve sufficient accumulation and consequently increase of concentration
in the effluent (Schijven et al., 2003). Moreover, the tested pilot filter was not
pre-loaded with environmental oocysts and SSRC prior to the experiment,
but a pre-loading with environmental small-sized centric diatoms (SSCD 4-7
µm) is plausible since the influent contained significant amounts of
environmental SSCD (Fig. 2). This might explain the relatively low capacity
of the filter bed to remove SSCD (DEC of 1.8 log).
Removal mechanisms and delayed breakthrough. Accumulation and
delayed breakthrough of micro-organisms in filters affecting DEC over time
is determined by the ruling removal mechanisms. It is not likely that in a
fixed filter bed like the slow sand filter, micro-organisms that were removed
by straining are remobilized. Delayed breakthrough is restricted to persistent
micro-organisms that are removed by reversible attachment and show to
some degree detachment. On the basis of the results of this study and
observations by others, each of the mentioned processes, straining,
attachment and detachment and survival, will be discussed further for the
removal of the three tested micro-organisms.
Straining as removal mechanism for biocolloids in filters is expected to
depend on the ratio between the colloid and collector size (dp/dc) and based
on geometric modelling, straining is expected to be of less importance at
ratios of <0.05 (Sakthivadivel et al., 1966; 1969) and <0.154 (Herzig et al.,
1970). The ratios for oocysts, SCP and SSCD in the pilot filter of the current
study are 0.018, 0.005 and 0.014 – 0.0025, respectively, indicating a minor role
of straining. In current literature, however, the significance of straining to
colloid removal of the size of bacteria and protozoa is still in debate. One
___________________________________________________________________
- 190 -
Chapter 8
study showed the variability of the role of straining and simple extrapolation
of size of grain and microbe is inappropriate (Hijnen et al., 2005). The large
difference in removal of SSCD and oocysts, organisms with the same size, in
the current study with a slow sand filter containing a well developed
schmutzdecke, supports this conclusion. There are studies where straining is
presented as a significant removal mechanism in freshly packed columns
(Bradford et al., 2003, Bradford and Bettahar, 2005; Tufenkji et al., 2004b;
Schijven et al., 2007). The negligible effect of the schmutzdecke on the oocyst
removal in slow sand filters observed in other studies (Bellamy et al., 1985;
Timms et al., 1995; Hijnen et al., 2004) question mark the role of straining in
slow sand filtration. In conclusion, straining will have contributed to the
overall removal of oocysts, SCP and SSCD in the pilot plant filter with a
prolonged pre-filtering period of three years, but is most likely not the
dominating removal mechanism.
Attachment and detachment of biocolloids like micro-organisms to sand is
described by the colloid filtration theory. The parameter α is the attachment
efficiency, calculated from the removal efficiency and the single-collector
removal efficiency η, a theoretical parameter based on the physical processes
involved in colloid transport and colloid to collector interactions. Assuming
complete removal by sorption, the collision efficiency α calculated for the
removal of oocysts in the 1.5 m filter bed was 0.207 (Table 2). This value is
slightly lower than the α values of 0.26 – 0.37 observed for C. parvum oocysts
in a column with glass beads of 0.328 mm at approximately the same
hydraulic conditions as the present study (Tufenkji and Elimelech, 2005). On
the basis of a recent review on Cryptosporidium-sand interactions it was
concluded that physicochemical filtration plays an important role in oocyst
removal in saturated porous media (Tufenkji et al., 2006). The collision
efficiency for the removal of spores of C. perfringens in the filter bed was 0.310
(Table 2) and SCP concentrations at the top were higher compared to the
oocyst concentrations (Fig. 3). The difference in α-value calculated for the
removal of spores and oocysts, indicate that attachment of the spores of C.
perfringens to the sand was more efficient than the attachment of the oocysts
of C. parvum, an observation described previously (Hijnen et al., 2005). This
indicates the different surface properties of spores and oocysts. The zetapotential of the C. perfringens spores in the influent was -32.6 (±2.5) mV
(Schijven et al., 2007), whereas for oocysts higher zeta-potentials of -6 mV
have been reported by Tufenkji and Elimelech (2005) at pH 8 (similar as pH
of the influent of the present study) and also others (Brush et al., 1998; Dai
and Hozalski, 2003). In the top of the filter bed a relatively high Fe-hydroxide
(0.55 mg/kg) concentration was determined which will enhance attachment
___________________________________________________________________
- 191 -
Chapter 8
.
of the mostly negative charged micro-organisms. The low estimated level of
SSCD in the filter bed indicates unfavourable surface properties of these
micro-organisms for attachment, which is supported by the observed low
removal of SSCD in the pilot filter and also in the full-scale rapid sand filters
(0.4 log), the first stage of the treatment. Similar low reduction of fluorescent
micro algae (0.4 – 15 µm) of 0.4 – 1 log in GAC and expanded clay filters was
reported in literature (Persson et al., 2005).
Continuation of the breakthrough of oocysts, SCP and SSCD during
approximately 50 days after the dosage was stopped (Fig. 2) and a secondary
peak of oocyst concentration in this period, an observation also described by
others in column studies (Harter et al., 2001; Hijnen et al., 2005), demonstrate
that the attachment of these organisms in the filter bed is reversible. Davies et
al. (2005a) demonstrated remobilization of micro-organisms including C.
parvum oocysts retained in soil in response to changes in ionic strength (from
tap water to rain water). The virtually flat tail of the breakthrough curves of
all three micro-organisms after dosing was stopped (Fig. 2) indicates a high
detachment rate as was described for SCP (Schijven et al., 2003) and for C.
parvum oocysts (Bradford et al., 2005) retained in sand column studies.
Finally, for delayed breakthrough the removed micro-organisms must
survive (negligible inactivation rate and no mass reduction) in the filter bed.
The oocysts, spores and SSCD were primarily retained in the Schmutzdecke
and the first 5 – 10 cm of the filter bed of the present study (Fig. 3), the part
with the increased microbiological activity (Fig. 4) and the highest
accumulation of (in)organic components (data not presented). Timms et al.,
(1995) showed similar spatial distribution of oocysts in a slow sand filter bed
monitored immediately after a dosing experiment and 80% of the oocysts
being present in the top layer of the filter bed. Our study revealed, however,
a low recovery of oocysts in the filter bed determined 86 days after dosing
was stopped (Table 2). In the next 69 days of filtration a further decline in
mass of retained oocysts was observed. In contrast with these results, the
recovery of C. perfringens spores in the filter bed was significantly higher,
despite the lower mass load and removal rate and no inactivation of the
retained numbers of spores was observed (Table 2). The mass accumulation
and the log-linear inactivation or mass reduction rate of spores and oocysts,
respectively in the filter bed (Fig. 5) was estimated from the mass load,
observed removal rate and numbers retained in the filter bed. The estimated
inactivation rate of SCP was 0.005 d-1 (Fig. 5), similar to the inactivation rate
assessed for environmental SCP in river water (Medema et al., 1997). The
estimated mass reduction rate of retained oocysts was 0.014 – 0.02 d-1 (Fig. 5),
values lower than estimated from a study where the mass reduction of
retained oocysts in soil exposed to faecal wastes inoculated with C. parvum
___________________________________________________________________
- 192 -
Chapter 8
oocysts was determined (Hutchison et al., 2005). From the published oocyst
mass reduction data in the soil of the latter study log-linear mass reduction
rates of 0.136 and 0.153 log d-1 were estimated. King et al., (2005) showed
lower mass reduction rates of C. parvum oocysts in untreated surface water
from a reservoir; estimated rates were 0.0027, 0.0069 and 0.0076 d-1 at 15, 20
and 25oC, respectively. The same authors noticed no reduction occurred in
autoclaved samples of the same water and from their results King et al.,
(2005) hypothesized an influential role of predation in oocysts mass
reduction.
Numbers of organisms (log)
10.0
9.0
8.0
Numbers observed in bed
7.0
Oocysts no decay
Oocysts decay 0.019 - 0.014
6.0
Spores no decay
Spores decay 0.005
5.0
0
50
100
150
200
250
300
Operational time (days)
Figure 5. Simulated accumulation of mass of spores and oocysts (numbers) in the
filter bed with and without decay rates estimated from the mass determined after 184
and 253 days of operation
Disintegration, predation and inactivation of oocysts. The observed
reduction of oocysts in the sand over time can be caused by disintegration or
predation. Microscopic evidence for disintegration of the retained oocysts in
this study, though, was not obtained. The observed oocysts in the filter bed
were intact and only to some degree, differential staining of the oocysts in
the schmutzdecke with the monoclonal antibody was noticed. From the
results of the zooplankton analysis (Fig. 4) it was deduced that predation of
oocysts was a possible cause of the large oocyst reduction in the filter bed.
Eight different species were observed in the filter bed in relative high
numbers. Two of these species, Testacae (Testate amoebae) and Rotatoria, are
family of zooplankton species for which predation of oocysts have been
documented (Fayer et al., 2000; Stott et al., 2001, 2003). The results of the mass
recovery of spores in the filter bed show that spores of sulphite-reducing
clostridia are less susceptible to predation.
___________________________________________________________________
- 193 -
Chapter 8
.
The presence of zooplankton like protozoa, Rotifera, Nematoda and
Oligochaeta in slow sand filters and the significant role of predation in the
removal of bacteria was shown by several studies (i.e. Lloyd, 1996; WeberShirk and Dick, 1997). Predation as an oocyst reduction mechanism depends
on the presence of suitable predators which is not always the case as
demonstrated by Davies et al. (2005b). In air dried and sieved surface soil
samples from drinking water supply catchment areas inoculated with C.
parvum oocysts they observed hardly any reduction in total numbers during
180 days of incubation at 4, 20 and 35oC. In the absence of predation,
inactivation of the retained oocysts is the only process which will diminish
the risk of delayed breakthrough of reversible attached infectious oocysts.
Inactivation rates of oocysts in the soil samples of the study of Davies et al.,
(2005b) estimated by loss of viability (fluorescence in situ hybridisation,
FISH) ranged from 0.015 – 0.022 d-1. Using cell culture-Taqman PCR assay,
King et al. (2005) demonstrated higher rates in loss of infectivity of C. parvum
oocysts inoculated in surface water of 0.01, 0.045 and 0.049 d-1 at 15, 20 and
25oC, respectively. In addition this inactivation rate adds to the observed
mass reduction rate in reducing the risk of remobilization and delayed
breakthrough of infectious oocysts.
SSRC and SSCD as surrogates. The results of this study demonstrate
clearly differences between the behaviour of Cryptosporidium oocysts, SSRC
and SSCD in slow sand filters. Based on geometric considerations removal of
oocysts and SSCD by straining is similar and higher than removal of SSRC
by straining, but the importance of this removal mechanism is uncertain. The
present study indicates that attachment is of more importance for the
removal of SSRC and oocysts than for the removal of SSCD, but for all three
micro-organisms a high degree of detachment is observed. Finally, oocysts
are less persistent than SSRC and SSCD because of a higher inactivation rate
and the susceptibility to predation. Therefore the risk of accumulation of
infectious oocysts in biological active slow sand filters is expected to be low.
This implicates that a lowering of DEC of slow sand filters for infectious
oocysts as observed for SSRC and SSCD, is less likely to occur. Furthermore,
it is assumed that the environmental oocysts observed by Fogel et al., 1993 in
the filtrate of full-scale slow filters after delayed breakthrough and causing
low DEC were most likely not infectious. From these considerations it is
concluded that SSRC and to a larger extent SSCD, are too conservative
surrogates for the assessment of capacity of slow sand filters to eliminate
Cryptosporidium oocysts due to differences in size, surface properties and
persistence. The results of this study suggest that the higher persistency of
both surrogates due to low inactivation rates and susceptibility to predation
is the major basis for this conclusion.
___________________________________________________________________
- 194 -
Chapter 8
CONCLUSIONS
The results of this study demonstrate the high capacity of mature slow
sand filters to remove Cryptosporidium oocysts and spores of C. perfringens.
The observed decimal elimination capacity DECm was 4.7 and 3.6 log,
respectively. The relative low DEC of 1.8 log observed for the small sized
centric diatoms (SSCD) with a similar size as the oocysts was attributed to
low attachment and accumulation followed by delayed breakthrough, a
phenomenon also observed for SSRC. On the basis of this study, the risk of
delayed breakthrough of infectious oocysts is low because of a rapid
decline in oocyst concentration in the filter bed most likely caused by
predation. The results of this study indicate that environmental SSCD and
SSRC persist longer in the filter bed. From this finding in combination with
the difference in observed DEC of the filter bed for the three organisms, it
was concluded that SSCD and SSRC are too conservative parameters to be
useful as surrogates for the assessment of the elimination capacity of slow
sand filters for Cryptosporidium oocysts. Further investigations are necessary
to elucidate the role of predation and the ultimate fate of the ingested oocysts
for Cryptosporidium removal and the effect of temperature and filter bed
scraping on this removal mechanism.
REFERENCES
Akiba, M., S. Kunikane, H.-S. Kim, and H. Kitazawa. 2002. Algae as surrogate
indices for the removal of Cryptosporidium oocysts by direct filtration. Wat. Sci.
Technol.: Wat. suppl. 2:73-80.
Anderson, M. T. 1981. Improved method for separating zooplankton from
detritus. Prog. Fishculturist 43:42-4.
Anonymous. 2001. Besluit van 9 januari 2001 tot wijziging van het
waterleidingbesluit in verband met de richtlijn betreffende de kwaliteit van voor
menselijke consumptie bestemd water, p. 1-53, vol. 31. Staatsblad van het
Koninkrijk der Nederlanden.
Beakes, G., H. M. Canter, and G. H. M. Jaworski. 1988. Zoospores ultrastructure
of Zygorhizidium affluens Canter and Z. planktonicum Canter, two chytrids
parasitizing the diatom Asterionella formossa. Hassall. Can. J. Bot. 66:1054-1067.
Bellamy, W. D., G. P. Silverman, D. W. Hendricks, and G. S. Logsdon. 1985.
Removing Giardia cysts with slow sand filtration. J. Am. Water Works Assoc.
77:52-60.
Bradford, S. A., and M. Bettahar. 2005. Straining, attachment and detachment of
Cryptosporidium oocysts in saturated porous media. J. Environ. Qual. 34:469-478.
___________________________________________________________________
- 195 -
Chapter 8
.
Bradford, S. A., J. Šimůnek, M. Bettahar, and M. T. van Genuchten. 2003.
Modeling colloid attachment, straining and exclusion in saturated porous media.
Environ. Sci. Technol. 37:2242-2250.
Brush, C. F., M. F. Walter, L. J. Anguish, and W. C. Ghiorse. 1998. Influence of
pretreatment and experimental conditions on electrophoretic mobility and
hydrophobicity of Cryptosporidium parvum oocysts. Appl. Environ. Microbiol.
64:4439-4445.
Dai, X., and R. M. Hozalski. 2003. Evaluation of Microspheres as Surrogates for
Cryptosporidium parvum Oocysts in Filtration Experiments. Environ. Sci. Technol.
37:1037-1042.
Davies, C. M., N. Altavilla, M. Krogh, C. M. Ferguson, D. A. Deere, and N. J.
Ashbolt. 2005b. Environmental inactivation of Cryptosporidium oocysts in
catchment soils. J. Appl. Microbiol. 98:308-317.
Davies, C. M., C. Kaucner, N. Altavilla, N. J. Ashbolt, C. M. Ferguson, M. Krogh,
W. A. M. Hijnen, G. J. Medema, and D. A. Deere. 2005a. Fate and transport of
surface water pathogens in watersheds. AWWARF, Denver CO, US.
Fayer, R., J. M. Trout, E. Walsh, and R. Cole. 2000. Rotifers ingest oocysts of
Cryptosporidium parvum. J. Eukaryot. Microbiol. 47:161-163.
Fogel, D., J. Isaac-Renton, R. Guasparini, W. Moorehead, and J. Ongerth. 1993.
Removing Giardia and Cryptosporidium by Slow Sand Filtration. J. Am. Water
Works Assoc. 85:77-84.
Haas, C. N. 1983. Estimation of risk due to low doses of micro-organisms: a
comparison of alternative methodologies. Am. J. Epidemiol. 118:573-82.
Harter, T., S. Wagner, and E. R. Atwill. 2001. Colloid transport and filtration of
Cryptosporidium in sandy soils and aquifer sediments. Environ. Sci. Technol.
34:62-70.
Harvey, R. W., N. E. Kenner, A. Bunn, D. MacDonald, and D. Metge. 1995.
Transport behavior of groundwater protozoa and protozoan-sized microspheres
in sandy aquifer systems. Appl. Environ. Microbiol. 61:209-217.
Harwood, V. J., A. D. Levine, T. M. Scott, V. Chivukula, J. Lukasik, S. R. Farrah,
and J. B. Rose. 2005. Validity of the indicator organisms paradigm for pathogen
reduction in reclaimed water and public health protection. Appl. Environ.
Microbiol. 71:3163-3170.
Herzig, J. P., D. M. Leclerc, and P. LeGoff. 1970. Flow of suspensions through
porous media: application to deep filtration. Ind. Eng. Chem. 62:8-35.
Hijnen, W. A. M., A. J. Brouwer-Hanzens, K. Charles, and G. J. Medema. 2005.
Transport of MS2 phage, Escherichia coli, Clostridium perfringens, Cryptosporidium
parvum and Giardia intestinalis in a gravel and a sandy soil. Environ. Sci. Technol.
39:7860-7868.
Hijnen, W. A. M., J. F. Schijven, P. Bonné, A. Visser, and G. J. Medema. 2004.
Elimination of viruses, bacteria and protozoan oocysts by slow sand filtration.
Wat. Sci. Technol. 50:147-154.
Hijnen, W. A. M., W. M. H. van der Speld, F. A. P. Houtepen, and D. van der
Kooij. 1997. Presented at the International Symposium on Waterborne
Cryptosporidium, Newport Beach, California.
___________________________________________________________________
- 196 -
Chapter 8
Hijnen, W. A. M., A. J. Van der Veer, J. Van Beveren, and G. J. Medema. 2002.
Spores of sulphite-reducing clostridia (SSRC) as surrogate for verification of the
inactivation capacity of full-scale ozonation for Cryptosporidium. Wat. Sci.
Technol.: Wat. suppl. 2:163-170.
Hijnen, W. A. M., D. Veenendaal., W. M. H. Van der Speld, A. Visser, W.
Hoogenboezem, and D. Van der Kooij. 2000. Enumeration of faecal indicator
bacteria in large water volumes using on site membrane filtration to assess water
treatment efficiency. Water Res. 34:1659-1665.
Hutchison, M. L., L. D. Walters, T. Moore, D. J. I. Thomas, and S. M. Avery. 2005.
Fate of pathogens present in livestock wastes spread onto fescue plots. Appl.
Environ. Microbiol. 71:691-696.
King, B. J., A. R. Keegan, P. T. Monis, and C. P. Saint. 2005. Environmental
temperature controls Cryptosporidium oocyst metabolic rate and associated
retention of infectivity. Appl. Environ. Microbiol. 71:3848-3857.
LeChevallier, M. W., and K. Au. 2004. Water treatment and pathogen control:
process efficiency in achieving safe drinking water. World Health Organisation,
IWA Publishing, London, UK.
Lloyd, B. J. 1996. The significance of protozoal predation and adsorption for the
removal of bacteria by slow sand filtration. In N. Graham and R. Collins (ed.),
Advances in slow sand filtration and alternative biological filtration. John Wiley
& Sons, West Sussex, UK.
MacKenzie, W. R. H., N.J. , M. E. Proctor, S. Gradus, K. A. Blair, D. E. Peterson, J.
J. Kazmierczak, D. G. Addiss, K. R. Fox, J. B. Rose, and J. P. Davis. 1994. A
massive outbreak in Milwaukee of Cryptosporidium infection transmitted through
the public water supply. New. Engl. J. Med. 331:161-167.
Magic-Knezev, A., and D. van der Kooij. 2004. Optimisation and significance of
ATP analysis for measuring active biomass in granular activated carbon filters
used in water treatment. Water Res. 38:3971-3979.
Mazoua, S., and E. Chauveheid. 2005. Aerobic spore-forming bacteria for
assessing quality of drinking water produced from surface water. Water Res.
39:5186-98.
Medema, G. J., M. Bahar, and F. M. Schets. 1997. Survival of Cryptosporidium
parvum, Escherichia coli, faecal streptococci and Clostridium perfringens in river
water. Wat. Sci. Tech. 35:249-252.
Medema, G. J., W. Hoogenboezem, A. J. van der Veer, H. A. M. Ketelaars, and W.
A. M. Hijnen, and P. J. Nobel. 2003. Quantitative risk assessment of
Cryptosporidium in surface water treatment. Wat. Sci. Tech. 47:241-247.
Medema, G. J., F. M. Schets, P. F. Teunis, and A. H. Havelaar. 1998.
Sedimentation of free and attached Cryptosporidium oocysts and Giardia cysts in
water. Appl. Environ. Microbiol. 64:4460-6.
Payment, P., and E. Franco. 1993. Clostridium perfringens and somatic coliphages as
indicators of the efficiency of drinking water treatment for viruses and protozoan
cysts. Appl. Environ. Microbiol. 59:2418-24.
___________________________________________________________________
- 197 -
Chapter 8
.
Persson, F., J. Långmark, G. Heinicke, T. Hedberg, J. Tobiason, T. Stenström, and
M. Hermansson. 2005. Characterisation of the behaviour of particles in biofilters
for pre-treatment of drinking water. Water Res. 39:3791-3800.
Richardson, A. J., R. A. Frankenberg, A. C. Buck, J. B. Selkon, J. S. Colbourne, J.
W. Parsons, and R. T. Mayon-White. 1991. An outbreak of waterborne
cryptosporidiosis in Swindon and Oxfordshire. Epidemiol. Infect. 107:485-95.
Rijnaarts, H. H. M., W. Norde, E. J. Bouwer, J. l. Lyklema, and A. J. B. Zehnder.
1995. Reversibility and mechanism of bacterial adhesion. Colloids Surf. B. 5:5-22.
Sakthivadivel, R. 1969. Clogging of a Granular Porous Medium by Sediment.
University of California.
Sakthivadivel, R. 1966. Theory and Mechanism of Filtration of Non-colloidal Fines
Through a Porous Medium. University of California.
Schijven, J. F., M. Colin, Y. J. Dullemont, W. A. M. Hijnen, A. Magic-Knezev and
W. Oorthuizen. 2007. Removal of micro-organisms by slow sand filtration. RIVM
703719019/2007 (in Dutch).
Schijven, J. F., H. A. M. de Bruin, S. M. Hassanizadehb, and A. M. de Roda
Husman. 2003. Bacteriophages and clostridium spores as indicator organisms for
removal of pathogens by passage through saturated dune sand. Water Res.
37:2186–2194.
Schuler, P. F., M. M. Ghosh, and P. Gopalan. 1991. Slow Sand and Diatomaceous
Earth Fi1tration of Cysts and Other Particulates. Water Res. 25:995-1005.
Simmons, O. D., M. D. Sobsey, C. D. Heaney, F. W. Schaefer, and D. S. Francy.
2001. Concentration and detection of Cryptosporidium oocysts in surface water
samples by method 1622 using Ultrafiltration and capsule filtration. Appl.
Environ. Microbiol. 67:1123-1127.
Stott, R., E. Matsushita, and A. Warren. 2001. Protozoan predation as a
mechanism for the removal of Cryptosporidium oocysts from wastewaters in
constructed wetlands. Wat. Sci. Tech. 44:194-198.
Stott, R., E. May, E. Ramirez, and A. Warren. 2003. Predation of Cryptosporidium
oocysts by protozoa and rotifers: implications for water quality and public
health. Wat. Sci. Tech. 47:77-83.
Timms, S., J. S. Slade, C. R. Fricker, R. Morris, W. O. K. Grabow, K. Botzenhart,
and A. P. Wyn Jones. 1995. Remova1 of Cryptosporidium by Slow Sand Filtration.
Wat. Sci. Tech. 31:81-84.
Tisa, L. S., T. Koshikawa, and P. Gerhardt. 1982. Wet and dry bacterial spore
densities determined by buoyant sedimentation. Appl. Environ. Microbiol.
43:1307-1310.
Tufenkji, N., D. R. Dixon, R. Considine, and C. J. Drummond. 2006. Multi-scale
Cryptosporidium/sand interactions in water treatment. Water Res. 40:3315 – 3331.
Tufenkji, N., and M. Elimelech. 2004a. Correlation equation for predicting singlecollector efficieny in physicochemical filtration in saturated porous media.
Environ. Sci. Technol. 38:529-536.
Tufenkji, N., and M. Elimelech. 2005. Spatial distribution of Cryptosporidium
oocysts in porous media: evidence for dual mode deposition. Environ. Sci.
Technol. 39:3620-3629.
___________________________________________________________________
- 198 -
Chapter 8
Tufenkji, N., G. F. Miller, J. F. Ryan, R. W. Harvey, and M. Elimenech. 2004b.
Transport of Cryptosporidium oocysts in porous media: role of straining and
physicochemical filtration. Environ. Sci. Technol. 38:3620-3629.
USEPA. 2001. Method 1623: Cryptosporidium and Giardia in Water by
Filtration/IMS/FA. USEPA, Washington DC, US.
Weber-Shirk, M. L., and R. I. Dick. 1997. Biological mechanisms in slow sand
filters. J. Am. Water Works Assoc. 89:72–83.
Yao, K. M., M. Habibia, and C. R. O'Melia. 1971. Water and wastewater filtration:
concepts and applications. Environ. Sci. Technol. 5:1105-1112.
Yoo, H. S., S. U. Lee, K. Y. Park, and Y. H. Park. 1997. Molecular typing and
epidemiological survey of prevalence of C. perfringens types by multiplex PCR. J.
Clin. Microbiol. 35:228-232.
___________________________________________________________________
- 199 -
Chapter 9
Transport of MS2 Phage, Escherichia coli,
Clostridium perfringens, Cryptosporidium
parvum and Giardia intestinalis
in a Gravel and a Sandy Soil •
Wim A.M. Hijnen1, Anke J. Brouwer-Hanzens1, Katrina J. Charles2 and Gertjan
Medema1
1
2
Kiwa Water Research, PO BOX 1072, 3433 BB Nieuwegein, NL
Cooperative Research Centre for Water Quality and Treatment, Centre for Water
& Waste Technologies, University of New South Wales, UNSW-Sydney, 2052
NSW Australia
•
Reproduced with permission from Hijnen, W.A.M., Brouwer-Hanzens, A.J.,
Charles, K. and Medema, G.J. (2005) Transport of MS2 phage, Escherichia coli,
Clostridium perfringens, Cryptosporidium parvum and Giardia intestinalis in a
gravel and a sandy soil. Environmental Science and Technology 39, 7860-7868.
Copyright 2005, American Chemical Society.
___________________________________________________________________
- 201 -
Chapter 9
.
ABSTRACT
To define protection zones around groundwater abstraction wells and safe
setback distances for artificial recharge systems in water treatment,
quantitative information is needed about the removal of micro-organisms
during soil passage. Column experiments were conducted using natural
soil and water from an infiltration site with fine sandy soil and a river bank
infiltration site with gravel soil. The removal of phages, bacteria, bacterial
spores, and protozoan (oo)cysts was determined at two velocities and
compared with field data from the same sites. The microbial elimination
rate (MER) in both soils was generally >2 log, but MER in the gravel soil
was higher than that in the fine sandy soil. This was attributed to enhanced
attachment, related to higher metal-hydroxides content. From the high
sticking efficiencies and the low influence of flow rate on MER it was
deduced that straining played a significant role in the removal of
Escherichia coli and Cryptosporidium parvum oocysts in the gravel soil. Lower
removal of oocysts than the 4-5 times smaller E. coli and spores in the fine
sand indicates that the contribution of straining is variable and needs
further attention in transport models. Thus, simple extrapolation of grain
size and particle size to the extent of microbial transport underground is
inappropriate. Finally, the low MER of indigenous E. coli and Clostridium
perfringens observed in the soil columns as well as under field conditions
and the second breakthrough peak found for Cryptosporidium and spores in
the fine sandy soil upon a change in the feedwater pH indicate a significant
role of detachment and retardation to microbial transport and the difficulty
of extrapolation of quantitative column test results to field conditions.
INTRODUCTION
Soil passage is frequently used as pretreatment in production of drinking
water in The Netherlands in river bank filtration (RBF; 5%) and artificial
recharge (AR) in open basins (13%) or deep wells (1%) by several water
suppliers. It is an intensive filtration process with long contact times and an
effective barrier for pathogenic micro-organisms such as viruses, bacteria,
and protozoa. How effective it is, however, is not known and is a question
of growing interest since the introduction of quantitative microbial risk
assessment for drinking water safety (Haas et al., 1999). In 1980 a minimum
water travel time of 60 days as a protection zone around groundwater
abstraction wells was formalized in The Netherlands (Anonymous, 1980).
This travel time was assumed to cause sufficient die off of pathogenic
___________________________________________________________________
- 202 -
Chapter 9
bacteria from contamination sources (Knorr, 1937). In the past decades,
however, viruses, and more recently protozoa like Cryptosporidium and
Giardia, have been recognized as pathogens of major concern in the water
industry (Craun et al., 1997; MacKenzie et al., 1994; Gerba et al., 1990). These
organisms have been related to waterborne diseases because of their
persistence in the environment, resistance to water treatment, and high
infectivity. These organisms are different from bacteria in survival, surface
properties, and size. Moreover, it has become clear that die off in
groundwater is not the only process that governs the transport of microorganisms. For viruses it was demonstrated that attachment to soil particles
was more important than survival in the groundwater (Schijven, 2001).
Therefore, viruses and maybe protozoa could be transported over longer
distances in soil and thus be more significant to the microbial safety of
groundwater. To verify the safety of the 60-days set back distance
guideline, but also to assess the microbial safety of vulnerable groundwater
systems and of RBF and AR systems with shorter water travel times, more
quantitative information is needed on the elimination capacity of soil
passage systems.
A number of field studies have been carried out that established either
removal of indigenous micro-organisms or lab-cultured seeded microorganisms (Schijven et al. 1999, 2000, 2001; Van Olphen et al., 1993; Medema
and Stuyfzand, 2002). These studies showed that soil passage poses a very
effective barrier to micro-organisms, but critical situations may arise
(Medema and Stuyfzand, 2002). Such situations are intrusion of
contaminations to unconfined aquifers above groundwater wells, water
abstraction during RBF from a gravel aquifer, with increased risk during
high flow events, or short circuiting during recollection in AR systems.
Field studies are valuable but hampered by some drawbacks. The
concentration of pathogens in the field is generally too low to assess
removal, and only non hazardous model micro-organisms (Escherichia coli,
bacteriophages, and spores of clostridia) can be used in spiking studies
(Schijven et al., 2000). Moreover, the removal process is complex and
influenced by a range of factors which vary considerably between sites.
Hence, the effect of specific conditions such as soil characteristics, water
velocity, and water quality variations are difficult to assess under field
conditions. Some of these
disadvantages can be overcome by column-based studies with spiked
micro-organisms. Information can be collected on microbial transport in
different soil types under well defined and standardized conditions as
described in the literature. The importance of attachment and the surface
properties of bacteriophages, bacteria, and soil and of water quality
___________________________________________________________________
- 203 -
Chapter 9
.
parameters has been elucidated by column experiments (Burge and Enkiri,
1978; Sobsey et al., 1980; Bales et al., 1991; Jin et al., 1997; Goldschmid et al.,
1972; Fletcher and Marshall, 1982; Scholl et al., 1990; McCaulou et al., 1994).
More recently, transport of the oocysts of Cryptosporidium in soil columns
was studied (Harter et al., 2001; Logan et al., 2001; Bradford and Bettahar,
2005; Tufenkji et al., 2004a), and results indicate the importance of straining
on the removal of these larger organisms. The significance of column
studies increases when results are related to field conditions of the selected
soils and validated by field studies, as described for phage MS2 in dune
sand by Schijven (2001). The number of such studies is still limited.
Furthermore, there is a dearth of comparative studies describing
elimination of the spiked model organisms relative to pathogens and of labcultivated versus indigenous strains of these spiked model organisms. The
objective of the present study was to compare the elimination rate of FRNA
phage MS2 as a model for viruses (Havelaar, 1993), indicator bacteria E. coli
and spores of Clostridium perfringens as models for bacterial and protozoan
pathogens (Payment and Franco, 1997; Hijnen et al., 2000), and
Cryptosporidium and Giardia (oo)cysts in two different soil systems: an AR
site with fine dune sand and a RBF site with a gravel aquifer. For the AR
site elimination of MS2 phages in soil columns was compared with
elimination under field conditions as published by Schijven et al. (1999). For
the RBF site elimination of the lab-cultivated phages and indicator bacteria
was compared with the elimination of indigenous FRNA phages and
indicator bacteria under column as well as field conditions, the latter
published by Medema and Stuyfzand (2002).
MATERIALS AND METHODS
Soil and Water from the Infiltration Sites. TheAR infiltration site is
located in Castricum, The Netherlands, which is situated in the coastal
dunes, where pretreated river Lek water is recharged in open basins
without unsaturated zones. Saturated soil was collected at a depth of 3 m
from a flow path 3 m from the edge of the open infiltration basin. The other
infiltration site studied is the RBF site located in Roosteren, The
Netherlands, where river Meuse water is abstracted through a fluvial
gravel aquifer. During low river flow soil from the river bank was collected
from a drill at a depth of 0.05-0.65 m. The soil samples were stored for 1.5
months in the dark at 5°C before the experiment. The water used for the
experiment was the natural water of both infiltration sites: for Castricum
this is pretreated river Lek water (coagulation, sedimentation, and
___________________________________________________________________
- 204 -
Chapter 9
filtration) and Roosteren untreated river Meuse water. A 600 L water
sample was taken in two separate stainless steel (SS) vessels and stored for
1 week at approximately 10°C. River Meuse water was stirred continuously
to avoid sedimentation.
Soil and Water Analysis. After treatment of the soil with HCl and
H2O2 to remove cementing materials, the grain size distribution of the <2
mm fraction was measured using a FRITSCH Laser Particle Sizer A22
(Laval Lab inc., Canada). The clay and silt fractions were converted to the
traditional grain size analysis with pipet and sieve analysis (Konert and
Vandenberghe, 1997). Total organic material and carbonate were
determined using the methods of Stuyfzand and van der Jagt (1997). Soil
pH and the electrical conductivity (EC) were determined after shaking 20 g
for 2 h with 50 mL of ultrapure water in the decanted fluid at 20°C. Free
(amorphous and exchangeable) Fe, Mn, and Al were determined in 2.5 g of
soil after shaking for 2 h in the dark with 100 mL of an ammoniumoxalate/oxalic acid mixture (16.1 and 10.9 g.L-1, respectively) at pH 3. Mn
and Al were quantified by ICP-MS and Fe by AAS-flame
photospectroscopy. The effective cation exchange capacity (CEC) of the soil
was assessed by the AgTU method (Stuyfzand and van der Jagt, 1997).
Water quality of the infiltration water in the storage tanks was determined
using standard methods.
Column Setup. The soil column setup consisted of four Perspex
columns (Ø9 cm; height 50 cm), located in a temperature-controlled dark
room (16±1°C). Two columns were filled with Castricum soil and two with
Roosteren soil. The columns were packed with humid soil and placed on a
shaker for 0.5 and 1 min. at amplitudes of 6 and 10, respectively. The bed
settled 3-5 cm; consequently, the surface was replenished with soil to
eliminate the headspace. Subsequently, the columns were closed and
supplied with water from 30 L SS vessels connected to the columns with 5
mm PVC piping. These vessels were regularly refilled with Lek water and
Meuse water supplied to the Castricum and Roosteren columns,
respectively. For the experiment with spiked river water 30 L SS vessels
were filled with river water, inoculated with the micro-organisms, and
continuously mixed with a stirrer. The columns were operated at a constant
flow rate or fluid approach velocity U (i.e., volume of water per unit time
per unit cross sectional area of media) with a peristaltic pump (Gilson,
minipuls 2; Gilson inc., Middleton) located at the outlet of the filters. One
column of each site was operated at U = 0.9 m.day-1. U under field
conditions at the Castricum site can be lower, and at the Roosteren site
higher rates have been observed during high-flow events. Therefore, the
___________________________________________________________________
- 205 -
Chapter 9
.
second column of the Castricum and Roosteren sites was operated at 0.5
and 2.5 m.day-1, respectively.
Tracer Test and Breakthrough Modeling. The medium dispersivity
L, porosity θ, and pore velocity v were determined with a tracer test using
sodium chloride (275 and 310 mg L-1 NaCl for Castricum and Roosteren
columns, respectively). The conductivity was monitored continuously with
two sensors connected to data log units (Campbell 21X, Campbell Scientific
Inc., Utah). A one-dimensional convection-dispersion model (CDE) was
used to describe the breakthrough curves with the following equation
(CXTFIT; Toride et al., 1997)
δC
δ 2C
δC
= D. 2 - v
δx
δx
δt
(1)
where C is the electrical conductivity, D the coefficient of hydrodynamic
dispersion (m2.h-1), x the distance in the columns (m), v the average pore
velocity (m.h-1), and t the time (hours).
Removal of Micro-organisms. Water flow was started 7 days prior
to the experiment to equilibrate the columns. In the days before the
challenge experiment, concentrations of indigenous F-specific RNA
bacteriophages (FRNA phages), thermotolerant coliforms (Coli44), and
spores of C. perfringens were determined in the inlet water and outlet water
of the columns (Cout,indigenous). For the experiment 10 L of Lek water and 26 L
of Meuse water in the 30 L SS vessels were inoculated with lab-cultured
micro-organisms and the average inlet concentration during the spike Cin
was measured in a 1 L sample of inoculated water stored next to the setup.
The following organisms were spiked: FRNA phage MS2 (Cin = 1.6x109
±6.3x108 plaque-forming particles PFP.L-1), E.coli (WR1) (Cin = 1.2x106
±6.6x105 colony-forming particles CFP.L-1), C. perfringens (D10) (Cin =
7.9x105 ±1.1x105 CFP.L-1), ocysts of C. parvum (Cin = 1.4x106 ±7.5x105 L-1) and
cysts of G. intestinalis (Cin = 1.6x106 ±5.0x105 L-1). The inoculated water in
the vessel was stirred continuously with a magnetic stirrer. For 24 h the
columns were challenged with these suspensions, and thereafter the supply
of unspiked water was restored. Breakthrough of the spiked organisms was
determined by sampling complete column filtrate in separate containers of
different volumes over 2 weeks, which were analyzed for each organism.
Breakthrough curves were constructed by plotting the ratio between Cout
and Cin (n.L-1) equal to C/Co against the number of pore volumes, corrected
for the time interval of the sampling. The micro-organisms elimination rate
(MER; log) was calculated based on the mass balance calculated by
___________________________________________________________________
- 206 -
Chapter 9
−
t ch * Q * C in
)
MER = log10 (
t
Q
C
−
C
(
*
*
(
))
∑ s ,i
out
out ,indigenous
(2)
where tch and ts,i are periods (h) of challenge test and sampling for sample i
and Q is the volume load (L.h-1).
Micro-organisms Suspensions and Enumeration Methods. Methods
for production and storage of stock solutions MS2 phages and E. coli WR1
and for enumeration of FRNA phages and Coli44 have been described
previously (Schijven, 2001). Stock solution procedures for spores of C.
perfringens D10 and the enumeration method were described by Hijnen et
al. (2002). C. parvum oocysts (Moredun; harvested by sedimentation and
differential centrifugation) were spiked from a 2 mL suspension (108
oocysts; viability 70%, PI staining) and G. intestinalis cysts (H3 isolate,
Waterborne, harvested by sedimentation and differential centrifugation)
from a 50 mL suspension (108 H3 isolate; viability 58%, PI staining).
(Oo)cyst were counted in samples of 1-3.000 mL using a direct fluorescence
assay technique with the Chemscan-RDI (Chemunex SA, Ivry-sur-Seine
Cedex, France). Samples were filtered and prepared for scanning with the
Chemscan kit (Chemunex 200 k0009-01 with IMS) including a mounting
medium (85 μL), membrane filters (25 mm 2.0 μm), and support Pad. The
membrane filters were labeled with 100 μL diluted monoclonal antibody
reagents (Oxoid Ltd., Hampshire, U.K.) 1:1 in deionized water for 30 min.
at 37°C. The filters were scanned, and counted spots were microscopically
confirmed based on fluorescence, morphology, and size.
Statistical Methods. The data were analyzed with the
nonparametric Sign test using SPSS (SPSS Inc., Chicago, IL) to test for
median differences, assuming a non-symmetric distribution.
Modeling Removal Data. The conceptual model used in modeling
micro-organisms removal has been described by several authors (Schijven
et al., 1999; Bales et al., 1991; McCaulou et al., 1994; Toride et al., 1997;
Šimůnek et al., 1998). On the basis of the results, this was only possible for
MS2 phage breakthrough. The Hydrus-1D model (Šimůnek et al., 1998) was
used to model the breakthrough curves. It consists of a first-order
dispersion model for water transport supplemented with equations for
attachment and detachment of MS2 to soil and a first-order inactivation
kinetic with different rate coefficients for the free and attached phages. The
concentration in the water (C; PFP.L-1) was described by
δ 2C
δC
δC
ρB
= D
-v
- k att C - μ1C + k det
s (3)
2
θ
δx
δx
δt
___________________________________________________________________
- 207 -
Chapter 9
.
The concentration of attached MS2 (S; PFP.g-1) is described by
ρ β
θ
δS
ρB
= k att C - k det
s
θ
δt
- μ
s
ρB
s
θ
(4)
Subject to boundary conditions C = Co at x = 0 and δC/δx = 0 at x = ∞. Here
D is the hydrodynamic dispersion coefficient (m2.h-1), x is the distance (m);
katt and kdet, are the attachment and detachment rate coefficients (h-1),
respectively, for a one kinetic site, μ1 and μs (h-1) are the inactivation rate of
the free and attached phages, respectively, ρB is the dry bulk density (kg.m3), and θ is the porosity.
Calculation of Collector and Sticking Efficiencies. Yao et al. (1971)
presented the first model for the transport of colloidal particles from the
pore fluid to the vicinity of the porous medium described with the
following equation
LN
C
3 (1 − θ )
=−
αηL
C0
2 dc
(5)
where dc is the diameter of the collector, α the sticking efficiency, η the
single collector collision efficiency, and L the length of the column.
Rajagopalan and Tien (1976) developed a semiemperical equation to solve
η, an equation recommended by Logan et al. (1995) who compared different
filtration models. Recently, Tufenkji and Elimelech (2004b) optimized the
RT model by including the influence of hydrodynamic and van der Waals
interactions, resulting in the following equation 6
−0.715
0.052
0.053
η 0 = 2.4 As 1 / 3 N Pe
N R−0.081 N vdW
+ 0.55 AS N R1.675 N A0.125 + 0.22 N R−0.24 N G1.11 N vdW
where the Happel porosity dependent parameter As = 2(1-γ5)/(2-3γ+3γ5-2γ6)
and γ=(1-θ)1/3; the peclet number NPe = Udc/D with the Diffusion
coefficient D = k B (T + 273) /(3πd p μ ) , Boltzmann’s constant kB = 1.38x10-23
J.K-1, temperature T (oC), the colloid diameter dp (m), and the dynamic
viscosity μ (M.L-1.T-1); interception number NR = dp/dc; van der Waals
number NvdW = A/kB(T+273) with A being the Hamaker constant (J);
attraction number NA = A/(12πdp2U); gravity number Ng = 2/9dp2(ρp ρB)g/µU, where ρp is the particle density in kg.m-3 and gravitational
constant g = 9.8806 m.s-2. For the calculations the following parameters
values were used: bulk water density 999.703 kg.m-3; Hamaker constant for
bacterium glass water interface (Rijnaarts et al., 1995) 6.2x10-21 J; sizes (m) of
the micro-organisms MS2 2.1x10-8 (Havelaar, 1993), E. coli 1.5x10-6, C.
perfringens 1.5x10-6, protozoan (oo)cysts C. parvum 4.9x10-6 and G. intestinalis
10.8x10-6 (Medema et al., 1998); and ρp of MS2 and E. coli 1085 kg.m-3
___________________________________________________________________
- 208 -
Chapter 9
(Bouwer and Rittman, 1992), C. perfringens 1270 (Tisa et al., 1982) and
(oo)cysts 1045 and 1036 kg.m-3, respectively (Medema et al., 1998).
RESULTS AND DISCUSSION
Soil Characteristics and Water Quality. Geochemical analysis of
the soils revealed that Castricum soil was a calcareous sand with high
uniformity (Uc = 1.6) and a fine grain size. It had a low content of organic
matter and metal hydroxides (Table 1). In contrast, Roosteren soil was
coarser with a high estimated coarse fraction (>2 mm) of 25% and thus a
lower uniformity. This soil was richer in organic matter and metal
hydroxides (Fe- and Al-ox) which resulted in a higher CEC. The Ca content
and pH of both soils were high. Levels of most water quality parameters in
the Meuse water were higher than those in the pretreated Lek water (Table
1), which corresponds to the higher content of organic and inorganic matter
in Roosteren soil.
The hydrodynamic parameters of the four columns were calculated from
the tracer tests (Table 2). The porosity was 36% and 32% for Castricum and
Roosteren soil, respectively. The low uniformity of the Roosteren soil
caused a 10 times higher dispersivity compared to the Castricum soil
column at a velocity of 0.9 m.day-1.
Transport of FRNA Phages. Prior to the spiking experiment, the
concentration of FRNA-phages in the feedwater and the filtrate of the
columns was determined. These phages were observed in the Meuse water
(700 – 1000 PFP.L-1), but not in the filtrate of the Roosteren columns
supplied with this water. The MS2 spike concentration was 1000 times
higher than that of the other micro-organisms, which resulted in higher
concentrations in the filtrate and clear breakthrough curves (Figure 1). The
micro-organisms elimination rate, MER, observed in the columns is
presented in Table 3. Modeling the breakthrough curves of MS2 with the
Hydrus-1D model resulted in a good fit of the observed data for most
columns (Figure 1; Table 2). The evident scatter in Roosteren water at 0.9 m
d-1 resulted in a poor fit (Figure 1; R2 = 0.63). The rate of inactivation in the
feed water μ1 was assumed to be equal to the rate of decrease of the
maximum breakthrough concentration during the challenge test and
calculated using linear regression of the natural log-transformed data.
For Castricum water μ1 was 0.120 h-1 (Table 2; column 0.9 m.day-1) and for
Roosteren water μ1 was 0.106 h-1 (column 2.5 m.day-1). In addition, the
estimated inactivation rate of the attached phages μs (Table 2) was lower
than μ1.
___________________________________________________________________
- 209 -
.
0.04 – 0.57
0.1 – 0.37
11.4 – 11.7
76.6 – 76.7
1.8 – 1.9
<0.1 - <0.1
HCO3- (mg.L-1)
NH4 (mg N.L-1)
Turbidity (Ftu)
(mg.L-1)
Ca (mg.L-1)
DOC (mg.L-1)
Fe (mg.L-1)
< 100
FRNA (PFP.L-1)
700 – 1,000
3,000 – 6,200
2,400 – 3,200
0.14 – 0.13
2.5 – 2.5
72 – 72.6
8.2 – 8.3
1.18 – 1.0
0.29 – 0.37
163 – 180
569 – 574
8.0 – 8.4
Meuse
watera,b
Mn-ox. (g/kg dw)
Al-ox. (g/kg dw)
Fe-ox. (g/kg dw)
Org. matter (%dw)
Ca (% dw)
Uc (d60/d10)
Clay (%; ≤2 μm)
Grain size (d50; mm)
Gravel (%; >2 mm)
CEC (meq/kg dw)c
EC (μS.cm-1)
pH
Parameter
0.02 (0.02)d
0.22 (0.32)d
0.58
(1.2)d
0.6 (0.1)d
2.5
1.6 (0.12/0.19)
2.43
(0.85)d
0.18 (0.21)d
0
19.3 (14)d
95
8.73
Soil characteristics
Castricuma
0.19
1.11
5.11
2.7
2.1
5.3 (0.11/0.56)
4.2
0.5
25
67
98.3
8
Roosterena
a
___________________________________________________________________
- 210 -
Lek water = Castricum; Meuse water = Roosteren;b two samples, before and after challenge period;c dw = dry
weight; d between brackets values of Castricum soil sample studied by Schijven (1999, 2001)
20 – 48
C. perfr. (CFP.L-1)
Coli44
2 – 10
162 – 170
EC (μS.cm-1)
pH
(CFP.L-1)
759 – 768
7.7 – 8.4
Parameter
Mg
Water Quality
Lek watera,b
Table 1. Water quality of the infiltration waters and geochemical characteristics of the soils
Chapter 9
0.238 (± 0.002)
0.89
Porosity θ (%)
Dispersivity αL (cm)
R2 tracer
0.97
0.167 (± 0.399)
36
Castricum 0.9
0.92
2.54
0.009 (± 0.006)
0.95
0.004 (± 0.003)
0.91
kdett (h-1)
R2
0.004 (± 0.003)
2.385 (± 0.048)
0.63
0.99
18.44 (± 0.005)
32
Roosteren 2.5
2.49
7.77
0.92
0.004 (± 0.001)
4.509 (± 0.424)
0.076 (± 0.010)
-0.106 (± 0.020; R2 = 0.92)
-0.069 (± 0.021)
0.99
2.287 (± 0.013)
32
Roosteren 0.9
0.90
2.81
___________________________________________________________________
- 211 -
0.622 (± 0.018)
0.724 (± 0.018)
katt
(h-1)
0.098 (± 0.016)
-0.056 (± 0.017)
μs (h-1)
-0.120 (± 0.030; R2 = 0.93)
36
Velocity U (m.d-1)
Velocity v (m.d-1)
μ1 (h-1)
Castricum 0.5
0.46
1.27
Table 2. Hydrodynamic parameters of the columns and Hydrus-1D parameters (±Standard Deviation) for MS2 transport in
the Castricum and Roosteren soil
Chapter 9
Chapter 9
.
Blanc and Nasser (1996) reported lower μs in loamy soil and sandy soil
saturated with groundwater at 10 and 23oC of 0.007 – 0.013 and 0.018 –
0.019 (h-1), indicating that the values in Table 2 are relatively high.
Calculated from the empty pore volume contact time and μ1, it was
estimated that inactivation accounted for 13.6% and 11.4% of the overall
removal of MS2 in the Castricum columns 0.5 and 0.9 m.day-1, respectively.
Castricum
0.1
0.5 m/d
0.9 m/d
C/Co
0.01
0.001
0.0001
0.00001
0.000001
0
5
10
15
20
25
Pore volumes
Roosteren
0.1
0.9 m/d
C/Co
0.01
2.5 m/d
0.001
0.0001
0.00001
0.000001
0
10
20
30
40
50
60
Pore volumes
Figure 1. The breakthrough curves of MS2-phages in the Castricum and the
Roosteren columns (open and closed circles are the observed values and lines
indicate the Hydrus-1D model fit)
In the Roosteren columns 0.9 and 2.5 m.day-1 contribution of inactivation to
the overall elimination was less, 5.7% and 4.6%, respectively. From this
consideration it was deduced that attachment was the most important
process in overall MS2 removal in both columns, while detachment rates
___________________________________________________________________
- 212 -
Chapter 9
were a factor of 100 – 1000 lower (Table 2). In Roosteren soil operated at 0.9
m.day-1, the attachment rate coefficient katt was calculated at 2.385 ± 0.048 h1, 4 times higher than the k
att calculated for the Castricum column operated
at the same velocity. Despite the lower collision efficiency η of Roosteren
soil, the sticking efficiency  in this soil was a factor of 10 higher than that
calculated for Castricum soil (Table 3).
Table 3a. Micro-organisms Elimination Rate (MER; log) of the Castricum Soil
Columns for Spiked Micro-organisms, and the Calculated Collector Efficiency η
and Sticking Efficiency α (Tufenkji and Elimelech, 2004b)
column 0.5 m.d-1
column 0.9 m.d-1
MER
η;α
MER
η;α
MS2a
3.3
1.035;0.003
2.2
0.680;0.003
E. coli
4.7
0.048;0.085
4.2
0.032;0.113
C. perfringens
≥5.0
0.067;0.064
≥4.5
0.042;0.092
C. parvum
3.9
0.084;0.040
3.3
0.059;0.048
G. intestinalis
6.2
0.250;0.021
>6.2
0.172;0.031
a
corrected for elimination caused by inactivation (10%)
Table 3b. Micro-organisms Elimination Rate (MER; log) of the Roosteren Soil
Columns
column 0.9 m.d-1
column 2.5 m.d-1
MER
η;α
MER
η;α
MS2a
3.4
0.390;0.019
1.8
0.188;0.021
E. coli
4.8
0.019;0.562
4.1
0.009;1.023
C. perfringens
≥2.4
0.032;0.169
≥3.0
0.013;0.516
C. parvum
>6.7
0.038;0.402
>7.2
0.017;0.959
G. intestinalis
>7.4
0.120;0.140
>6.8
0.051;0.299
a
corrected for elimination caused by inactivation (5%)
___________________________________________________________________
- 213 -
Chapter 9
.
Adsorption of micro-organisms to solid surfaces is governed by
electrostatic forces when surfaces are charged, and by intermolecular forces
governed by the hydrophobic or hydrophilic composition of both surfaces
(Ryan and Elimelech, 1996). For micro-organisms the isoelectric point (IEP),
defined as the pH where net surface charge of suspended colloids is zero, is
generally below 7.0 (Fletcher and Marshall, 1982; Harden and Harris, 1953).
Thus, at neutral pH, micro-organisms are negatively charged and they
preferably attach to the positively charged sites on the negatively charged
mineral grains of the soil. pH values of the Lek and Meuse water were
comparable. However, the relatively high pH of the Castricum soil
compared to the pH of Roosteren soil (Table 1) is one possible explanation
for the differences in katt values between both soils.
Hydroxides of iron, aluminum, and manganese are the most common
sources for positively charged sites at neutral pH. While the IEP of quartz is
2.0, the IEP of Fe-hydroxide-coated quartz is 8.5 (Stumm and Morgan,
1981). The positive effect of metal hydroxide coating on bacterial and viral
attachment has been demonstrated experimentally (Scholl et al., 1990; Ryan
et al., 1999). Castricum and Roosteren markedly differed in metal hydroxide
content (Table 1). The higher metal content in Roosteren soil yields more
attachment sites and thus a higher katt-value. Roosteren soil was also higher
in organic matter, but the role of organic matter in attachment of microorganisms to soil is undecided and strongly depends on the nature of the
accumulated organic matter, which influences the surface charge as well as
the hydrophobic/hydrophilic character. Consequently, the higher MER
observed in the Roosteren soil for MS2 can probably be attributed to the
presence of a more positively charged surface (metal oxides) and possibly
to the presence of more hydrophobic organic matter.
The results clearly demonstrate that the MER of MS2 was lower at a higher
water velocity (Table 3). Mass transport to the soil surface and microorganism-surface interactions depends on water flow velocity, diffusion
characteristics and surface properties of the micro-organisms and soil. It
has been proposed that based on the colloid filtration theory, colloid
removal in filters is proportional to the water velocity v-2/3 (Yao et al., 1971),
neglecting dispersion, inactivation and detachment. This ratio has been
observed in experiments for poliovirus and bacteriophage ФX174 (Schijven,
2001; Jin et al., 1997; Wang et al., 1981). On the basis of this ratio and the
MER observed at 0.9 m.day-1, the MER for the lower velocity in the
Castricum is estimated at 3.3 log and for the Roosteren soil columns with
higher velocity 1.7 log. These values were almost similar to the observed
MER-values for MS2 in these columns (Table 3).
___________________________________________________________________
- 214 -
Chapter 9
Schijven et al. (2003) also studied MS2 removal in Castricum soil columns
and determined less than 1 log removal in a bed of 1.4 m operated at a
higher velocity of 1.5 m.day-1 and lower temperature of 5 ± 3oC than in the
current study. Model parameters μ1, μs, and katt calculated from the
experiments of Schijven et al. (2003) were 0.003, 0.004, and 0.071 h-1,
respectively, clearly lower than the values presented in Table 3. Because
decay explains only 10% of the overall removal, we attribute the higher
elimination of MS2 in the current study with Castricum soil to increased
attachment. Interception and sedimentation are most likely processes of
minor importance to contact of small phages with soil particles. More
relevant is diffusion. The difference in diffusion constant D for MS2
calculated for the temperatures of both studies (16 ± 1 and 5 ± 3oC) was a
factor of 1.4, explaining only a small part of the large difference in katt
observed between both studies. Geochemical characteristics of the soil
samples tested in both studies are presented in Table 1. This showed that,
besides more inactivation and attachment at higher temperature in the
current study, the Castricum soil studied by Schijven was coarser with a
lower organic content and CEC-value, which may have contributed to the
higher attachment of MS2 in our columns.
Transport of Escherichia coli. Indigenous thermotolerant coliforms
(Coli44) in the Lek water (Table 1) were removed completely by the
Castricum soil columns, however, after the Roosteren soil columns low
numbers of Coli44 (3 – 97 CFP.L-1) were observed. On the basis of the
concentration of Coli44 in Meuse water (Table 1), the MER of unseeded
Coli44 was 2.8 and 1.5 log in the columns operated at 0.9 and 2.5 m.day-1,
respectively (C/C0 of 0.0012 - 0.0021 and 0.03 – 0.04, respectively; Figure 2).
Spiked E. coli WR1 was eliminated to a larger extent by both soils than
MS2. The relatively low spike concentrations of these bacteria (106 CFP.L-1)
combined with high removal caused little breakthrough; the outlet
concentrations of the Roosteren columns increased only slightly (p = 0.007
and 0.018). As observed for MS2, Roosteren soil columns removed WR1 to
a higher extent than Castricum soil columns (p = 0.015), despite the lower η
values (Table 3) due the higher grain size. The calculated sticking
efficiencies  of 1.023 for Roosteren soil at a flow rate of 2.5 m.day-1 was
slightly above unity. Assuming all parameters used to calculate η were
valid, the high MER must be caused by an additional removal mechanism
not accounted for in the colloid filtration model, i.e., straining. A parameter
related to straining is the interception number NR, reflecting the ratio
between the collector and colloid sizes. It is described in the literature that
straining becomes important when this ratio is >5% (Herzig et al., 1970),
___________________________________________________________________
- 215 -
Chapter 9
.
which was true only for the largest of the tested micro-organisms, G.
intestinalis, in the finer Castricum soil (6%).
Velocity 0.9 m d-1
0.1
0.01
0.001
0.5
0.0001
0.00001
C/Co salt
C/Co
1.0
Castricum
Roosteren
a
0.000001
0.0
0
10
20
30
Pore volumes
Low and high velocity
0.1
1.0
Castricum
Roosteren
0.001
0.5
0.0001
C/Co salt
C/Co
0.01
0.00001
b
0.000001
0
20
40
60
80
0.0
100
Pore volumes
Figure 2. Breakthrough curves of E. coli and the tracer (C/Co; EC, solid line
Castricum and dashed line Roosteren); (a) columns with similar velocity and (b)
columns with low (Castricum 0.5 m day-1) and high velocity (Roosteren 2.5 m day1).
However, more recently Bradford et al. (2002, 2003) and also Tufenkji et al.
(2004a) showed the importance of straining when this ratio is <5%, even as
low as 0.2% (Bradford et al. 2003). NR is calculated from the average
diameter of the grains and there is no correction for the grain size
distribution. Straining occurs in the small sized pores. Pore size distribution
is influenced by the uniformity and shape of the grains as well as the
___________________________________________________________________
- 216 -
Chapter 9
content of sediments. Roosteren soil has a lower uniformity and porosity
with higher sediment content of organic and inorganic nature compared to
Castricum soil (Table 1), resulting in a higher presence of small sized pores.
This will result in lower permeability or hydraulic conductivity of the soil
and consequently inhibition of microbial transport.
Horizontal and vertical hydraulic conductivity of a representative
Roosteren soil sample with Uc of 2.4 was 5 and 1 m.day-1, respectively, and
lower than the horizontal and vertical hydraulic conductivity of about 12
and 6 m.day-1 observed for the Dutch dune sand areas (Stuyfzand; personal
communication).
The relatively low influence of flow rate on the removal of E. coli in both
soils also point to an additional removal by straining. Though significantly
higher removal was observed in Castricum and Roosteren soil with
decreased flow rate (p = 0.018 and <0.001, respectively), the observed
increase in MER was low compared to the effect observed for MS2 and not
proportional to the water velocity v-2/3. Straining is expected to be
independent of the flow rate.
Transport of Spores of C. perfringens. Concentrations of indigenous
C. perfringens spores in both river waters were of the same order of
magnitude as those observed for Coli44 (Table 1). However, these
indigenous micro-organisms were observed after both soils in higher
concentrations than Coli44: 9 – 18 CFP and 76 – 293 CFP.L-1 after Castricum
and Roosteren columns, respectively. Castricum columns at 0.9 and 0.5
m.day-1 showed 0.4 and 0.6 log removal of the indigenous spores,
respectively. The MER calculated for the indigenous spores in the
Roosteren columns at 0.9 and 2.5 m.day-1 were 1.2 and 1.6 log, respectively.
During the challenge experiment C. perfringens D10 was spiked in a
concentration of about 100 - 10,000 times the concentration of indigenous
spores in the river water, but this had no significant effect on the C.
perfringens concentrations in the outlet, showing a high removal of seeded
spores. The minimum MER-values are presented for spiked C. perfringens
spores in Table 3.
Transport of Protozoan (Oo)cysts. The high MER values obtained
for the protozoan (oo)cysts (Table 3) indicate low mobility of these
organisms in soil. Neither oocysts of C. parvum nor cysts of G. intestinalis
were observed in the column filtrate of Roosteren soil, demonstrating
removal of >6.7 – >7.4 log. In the filtrate of the Castricum soil columns,
however, C. parvum oocysts were observed frequently (Figure 3) and a G.
intestinalis cyst was detected once (1 per 20 mL). The difference in (oo)cyst
breakthrough between both soils is significant. In 34 and 25 samples with a
___________________________________________________________________
- 217 -
Chapter 9
.
total sampled volume of 9.75 and 10.4 L for Castricum and Roosteren soil in
both columns 410 and 0 oocysts were counted, respectively. η values
calculated for the columns with a similar fluid approach velocity of 0.9
m.day-1 showed a slightly higher η value in the Castricum soil than in the
Roosteren soil. For the  values this was the other way around: a 10 times
higher  value for the Roosteren soil than for the Castricum soil (Table 3).
Castricum 0.5 m d-1
1
1.0
C. perfringens
Cryptospordium
0.01
0.001
0.5
C/Co salt
C/Co
0.1
0.0001
0.00001
0.000001
0.0
0
10
20
30
Pore volumes
Castricum 0.9 m d-1
1
1.0
C. perfringens
0.1
Cryptospordium
C/Co
0.001
0.5
0.0001
C/Co salt
0.01
0.00001
0.000001
0.0
0
20
40
60
80
100
Pore volumes
Figure 3. Breakthrough of tracer (C/C0; EC solid lines) and of C. parvum oocysts
and C. perfringens observed in the Castricum sandy soil
As discussed for E. coli, the higher MER for C. parvum in Roosteren soil
compared to Castricum soil again indicates that the average grain size is
not a proper generic predictor for microbial transport in soil. Previous
___________________________________________________________________
- 218 -
Chapter 9
studies (Harter et al., 2001; Logan et al., 2001) demonstrated a positive
correlation of the level oocysts breakthrough with the grain size in a higher
grain size range of 0.3 mm up to 1.4-2.4 mm (Uc of 1.7 to 2.1). Oocysts
breakthrough was approximately 0.2% (MER of 2.7 log) in a 0.1 m column
with 0.42 to 0.5 mm sieved and acid-treated sand (Harter et al., 2001).
Compared to this breakthrough, 0.01% oocyst breakthrough observed in a
0.5 m column with fine and highly uniform Castricum soil (Table 1), is
relatively high. The contribution of attachment to the high removal of
oocysts in Roosteren soil is illustrated by the high  values (Table 3), but, as
discussed for E. coli, an  value close to unity for G. intestinatlis indicates a
supplementary effect of straining to the removal. Breakthrough of the
oocysts (C/Co) in the Castricum soil column was higher with elevated
water velocity (Figure 3), but the effect on MER was small (Table 3).
Enhancement of C. parvum transport by increased flow rates has been
described before (Harter et al. 2001), but this study used a medium coarse
and coarse sands (0.42 to 2.4 mm). Since NR for this Castricum soil is 0.033
(3.3%) the contribution of straining to the removal process in this soil
(Bradford et al., 2003) is the most likely explanantion for the low influence
of flow rate on the MER.
A second breakthrough with a considerably higher maximum number of
oocysts was observed in the period after the challenge test in both columns.
The difference in the observed number of oocysts in the sub-samples of the
Castricum columns was a factor of 100, the concentrations showed a
subsequent increasing and decreasing trend and this trend was observed in
both columns. Simultaneously, an increase in C. perfringens concentration in
the filtrate was observed (Figure 3). This second breakthrough of oocysts
coincided with an increase of 0.7 pH unit in the Lek water used to feed the
columns (Table 1). As described before, water quality influences the
geochemical characteristics of the soil (Table 1). Besides this indirect effect
of water quality on microbial transport, pH and ionic-strength influences
the attachment/detachment of colloids in granular media, as demonstrated
by several authors (Burge and Enkiri, 1978; Sobsey et al. 1980; Goldschmid
et al. 1972; Ryan et al. 1999). An increase in pH increases the negative charge
on the surfaces of oocysts (Drozd And Schwartzbrod, 1996) and soil, and
enhances the electrostatic repulsion between attached oocysts and soil. This
may have caused an increased detachment rate of reversibly attached
oocysts and most likely the same accounts for the increase in spores in the
column effluent. A secondary breakthrough was also observed by Harter et
al. (2001) and they concluded that long-term, low-level elution is a potential
source of oocyst transport. Similar observations of high retardation and a
___________________________________________________________________
- 219 -
Chapter 9
.
multi-peaked breakthrough of labeled flagellates and protozoan-sized
microspheres were described by Harvey et al. (1995). They attributed this to
grain-surface interactions in the subsurface transport behavior of these
kinds of particles. More recently, the role of electrosteric repulsion in
oocysts adhesion to quartz surfaces has been proposed. This repulsive force
cannot be captured by the classic DLVO theory and is attributed to the
presence of proteins on the surface of the oocysts extending in the solution
(Kuznar and Elimelech, 2004).
Organism Comparison. Elimination of bacteriophages, bacteria,
spores and protozoan oocysts in the soil of the two infiltration sites was
determined simultaneously in this comparative study. The Microbial
Elimination Rate (MER) in both soils was generally > 2 log but the sequence
of the MER for the different sized organisms differed. In the gravel soil of
the RBF site of Roosteren the sequence was Giardia ≥ Cryptosporidium ≥ C.
perfringens ≥ E. coli > MS2 (Table 3). In the sandy soil of the AR site,
however, the sequence was Giardia > C. perfringens > E. coli >
Cryptosporidium > MS2 phages (Table 3), indicating that particle and grain
sizes do not universally govern removal of micro-organisms through soils.
From two observations it was deduced that this difference in sequence of
organism removal between both soils was caused by increased straining in
the Roosteren soil. First, the sequence in Roosteren soil corresponds to the
sequence in size of these organisms, and second, the difference in MER
between Castricum and Roosteren soil increased with size of the particles
(Table 3). The ratio between values of the different sized organisms was as
previously described (Logan et al., 1995). The sticking efficiencies indicated
lower attachment of MS2 compared to the other organisms, although an
additional effect of straining in this soil too cannot be excluded. This
emphasizes the complexity of the involved processes and the relative
contribution of straining, inactivation and attachment/detachment to the
overall removal in different soils. Furthermore, it demonstrates that the
semiempirical filtration models must be extended with parameters that are
related to the permeability of the soil such as pore size distribution and
hydraulic conductivity.
Indigenous Micro-organisms and Retarded Transport. Remarkably,
the MER observed for the indigenous Coli44 in the Roosteren soil columns
(2.8 and 1.5 log) determined prior to the challenge test, was a factor of 1.7 to
2.7 lower than the MER determined for spiked E. coli WR1 in these
columns. For C. perfringens the difference in removal of indigenous and
spiked spores was even larger (>1.9 up to 11-fold) and these MER values
were lower than those observed for (seeded) MS2. These observations
___________________________________________________________________
- 220 -
Chapter 9
indicate that the elimination rate in soil assessed with peak loading of
spiked micro-organisms most likely over-estimates the elimination of
indigenous micro-organisms under field conditions. It is hypothesized that
this observation is caused by the presence of high concentrations of
indigenous Coli44 and C. perfringens accumulated in the collected soils
from both infiltration sites used for this experiment. Detachment of a
fraction of these accumulated micro-organisms probably caused the low
MER for unseeded Coli44 and C. perfringens. Similar observations of
accumulation of C. perfringens spores, delayed breakthrough (retardation),
and as a consequence a low elimination rate, have been described for slow
sand filters (Hijnen et al. 2004). Clostridial spores are persistent and survive
longer than E. coli (Medema et al., 1997), which might explain why
indigenous C. perfringens showed the lowest MER values in three of the
four columns. The results of a column study (1.4 m of length; 1.5 m.day-1)
by Schijven et al. (2003), using the same soil and C. perfringens spores (D10),
support the observations in the current study. They determined a MER (log
C/Co) of approximately 4 log during the peak dose (24 h), which decreased
and leveled off to ±2.3 log in the following 23 days of filtration. Modeling of
these data revealed a high detachment rate (Schijven et al. 2003). A high
detachment rate will cause a high level of remobilization and further
transport of the spores. The second breakthrough peak for Cryptosporidium
oocysts in this study and the study of Harter et al. (2001) also indicates that
detachment may play an important role in retarded transport for these
microbes in the underground.
Translation of Column Data to Field Conditions. The quantitative
data of micro-organisms removal during soil passage obtained in this study
came from column experiments. It is important to evaluate how these
results relate to micro-organisms removal under field conditions.
Therefore, column data for MS2, E. coli, and C. perfringens (indigenous and
spiked) were compared to field observations at the same sites, as described
by others (Schijven et al. 1999, 2001, 2003; Medema and Stuyfzand, 2002),
plotting elimination rates against the travel time (Figure 4).
A field study in Castricum AR site at low temperature (5 ± 3oC; Schijven et
al., 1999), demonstrated that removal of MS2 declined as a logarithmic
function of travel time (R2 = 0.96; n = 7) (Figure 4a). MS2 removal in
Castricum soil columns determined under the same temperature (Schijven
et al., 2003) fitted well in this relationship. The high MER of MS2 observed
in the current column study with Castricum soil at 16oC did not fit in the
presented relationship (Figure 4a). As discussed before, a higher
___________________________________________________________________
- 221 -
Chapter 9
.
FRNA-phages, Castricum and Roosteren
0
2
R = 0.96; p<0.001
Log(C/Co)
-2
-4
-6
MS2 Castr. column (Schijven, 2003)
MS2 Castr. field (Schijven, 1999)
-8
MS2 Castr. column
MS2 Roost. column
ind. FRNA-phages Roost. field (Medema, 2002)
-10
0.01
0.1
1
10
a
100
E. coli, Roosteren
0
Log(C/Co)
-2
R2 = 0.93; p<0.01
-4
-6
ind. E. coli column
-8
spiked E. coli column
b
ind. E. coli field (Medema, 2002)
-10
0.01
0.1
1
10
100
C. perfringens, Roosteren
0
Log(C/Co)
-2
R2 = 0.93; p<0.01
-4
-6
-8
ind. C. perfringens column
spiked C. perfringens column
c
ind. C. perfringens field (Medema, 2002)
-10
0.01
0.1
1
Travel time (days)
10
100
Figure 4. Breakthrough of MS2 and FRNA-phages, E. coli and spores of C.
perfringens as function of the contact time (days; notice log-scale) in columns
(closed symbols) and field studies (open symbols) combined in a logarithmic
relation
___________________________________________________________________
- 222 -
Chapter 9
inactivation rate, μ1, as well as a higher attachment rate, katt, in our columns
probably caused this difference.
The MER of the indigenous E. coli and C. perfringens observed in the
Roosteren soil columns and observed under field conditions (Medema and
Stuyfzand, 2002) also fitted in a logarithmic function with travel time
(Figure 4b,c; R2 = 0.93 and 0.94, respectively). This was not the case for
spiked MS2 (Figure 4a), E. coli and C. perfringens (Figure 4b,c) in the
Roosteren soil columns, however. The following differences in the
experimental conditions of the column and field study may have
contributed to this difference. Soil for the column study was collected from
the first meter of the travel distance in the fluvial gravel aquifer. This part
of the river bank is heavily loaded with untreated river Meuse water and
most likely contains larger amounts of organic and inorganic matter (Table
1), compared to the deeper layers in the gravel aquifer. This result in lower
permeability of this part of the infiltrated soil compared to the deeper parts.
Moreover, field observations were done during extreme flood peaks in
winter (Medema and Stuyfzand, 2002). Under these circumstances high
flow rates decrease the travel time to the production well from 45-65 days
to 10-14 days.
Despite these differences, the relatively low elimination rates of indigenous
E. coli and C. perfringens in both columns and in the Roosteren aquifer can
be seen as additional support for the occurrence of accumulation, survival
and remobilization during microbial transport in the underground.
Furthermore, the sequence of MER observed under field conditions was
FRNA-phages > E. coli > C. perfringens, indicating the role of the persistence
of these micro-organisms as discussed before. Additionally, it shows that
quantitative data about the removal of micro-organims over distance or
travel time obtained with columns experiments using spiked microorganisms cannot simply be extrapolated to the field.
By studying transport of multiple micro-organisms in two different soils
with column experiments and comparing theses data with field behavior
from the same infiltration sites, the variability of microbial transport in the
underground is demonstrated. The largest travel distances can be predicted
for viruses from the MS2 results. Despite a relatively high inactivation rate
of MS2 in the water, modeling the breakthrough curves showed a high
contribution of attachment to the overall removal of these phages. The
variability of the contribution of attachment and straining and the
importance of the hydraulic conductivity to the transport of microorganisms in soil was demonstrated by the larger MER-values observed in
the coarse gravel soil compared to the fine sandy soil. Furthermore, higher
___________________________________________________________________
- 223 -
Chapter 9
.
breakthrough of Cryptosporidium oocysts compared to E. coli and spores of
C. perfringens in the sandy soil, demonstrate the importance of surface
properties of the micro-organisms to the transport behavior. Finally, the
significance of detachment and retardation to microbial transport in the
underground was deduced from the low MER of indigenous microorganisms, observed in the columns as well as under field conditions and
of the second breakthrough peak of Cryptosporidium found after the sandy
soil. These results show that the sequence in transport of viruses, bacteria
and protozoan (oo)cysts in the underground is governed by the size of the
organisms (viruses > bacteria > protozoa) and the hydraulic conductivity of
the soil (straining), as well as by the surface properties of both microorganism and soil, related to their attachment/detachment behavior, and
the inactivation rate. Thus, modeling microbial transport in the
underground to assess travel times and to set protection zones requires
knowledge of the hydrological and geochemical field conditions and
quantification of process parameters related to straining, attachment and
detachment and inactivation.
REFERENCES
Anonymous. 1980. Guidelines and recommendations for the protection of
Groundwater intake areas, Commissie bescherming Waterwingebieden (CBW).
VEWIN-RID, Rijswijk, NL.
Bales, R. C., S. R. Hinkle, T. W. Kroeger, and K. Stocking. 1991. Bacteriophage
adsorption during transport through porous media: chemical perturbations and
reversibility. Environ. Sci. Technol. 25:2088-2095.
Blanc, R., and A. Nasser. 1996. Effect of effluent quality and temperature on the
persistence of viruses in soil. Wat. Sci. Tech. 33:237-242.
Bouwer, E. J., and B. E. Rittman. 1992. Comment on use of colloid filtration theory
in modeling movement of bacteria through a contaminated aquifer. Environ. Sci.
Technol. 26:400-401.
Bradford, S. A., and M. Bettahar. 2005. Straining, attachment and detachment of
Cryptosporidium oocysts in saturated porous media. J. Environ. Qual. 34:469-478.
Bradford, S. A., J. Šimůnek, M. Bettahar, and M. T. van Genuchten. 2003.
Modeling colloid attachment, straining and exclusion in saturated porous media.
Environ. Sci. Technol. 37:2242-2250.
Bradford, S. A., S. R. Yates, M. Bettahar, and J. Šimůnek. 2002. Physical factors
affecting the transport and fate of colloids in saturated porous media. Water
Resour. Res. 88:1327-1340.
Burge, W. D., and N. K. Enkiri. 1978. Virus adsorption in five soils. J. Environ.
Qual. 7:73-76.
___________________________________________________________________
- 224 -
Chapter 9
Craun, G. F., P. S. Berger, and R. L. Calderon. 1997. Coliform bacteria and
waterborne disease outbreaks. J. Am. Water Work Assoc. 89:96-104.
Drozd, C., and J. Schwartzbrod. 1996. Hydrophobic and electrostatic cell surface
properties of Cryptosporidium parvum. Appl. Environ. Microbiol. 62:1227-1232.
Fletcher, M., and K. C. Marshall. 1982. Are solid surfaces of ecological significance
to aquatic bacteria. Adv. Microb. Ecol. 6:199-236.
Gerba, C. P., and J. B. Rose. 1990. Viruses in source and drinking water. In G.A.
McFeters (ed.) 'Drinking water Microbiology: progress and recent
developments', Springer-Verlag New-York Inc.
Goldschmid, J., D. Zohar, J. Argaman, and Y. Kott. 1972. Effect of dissolved salts
on the filtration of coliform bacteria in sand dunes. In P. Jenkins (ed.), Advances
in water pollution research. Pergamon Press, Oxford, UK.
Haas, C. N., J. B. Rose, and C. P. Gerba. 1999. Quantitative microbial risk
assessment. John Wiley & Sons, New York, USA.
Harden, V. P., and J. O. Harris. 1953. The isoelectric point of bacterial cells. J.
Bacteriol. 65:198-202.
Harter, T., S. Wagner, and E. R. Atwill. 2001. Colloid transport and filtration of
Cryptosporidium in sandy soils and aquifer sediments. Environ. Sci. Technol.
34:62-70.
Harvey, R. W., N. E. Kenner, A. Bunn, D. MacDonald, and D. Metge. 1995.
Transport behavior of groundwater protozoa and protozoan-sized microspheres
in sandy aquifer systems. Appl. Environ. Microbiol. 61:209-217.
Havelaar, A. H. 1993. Bacteriophages as models of enteric viruses in the
environment. ASM News 59:614-619.
Herzig, J. P., D. M. Leclerc, and P. LeGoff. 1970. Flow of suspensions through
porous media: application to deep filtration. Ind. Eng. Chem. 62:8-35.
Hijnen, W. A. M., J. F. Schijven, P. Bonné, A. Visser, and G. J. Medema. 2004b.
Elimination of viruses, bacteria and protozoan oocysts by slow sand filtration.
Wat. Sci. Technol. 50:147-154.
Hijnen, W. A. M., A. J. Van der Veer, J. Van Beveren, and G. J. Medema. 2002.
Spores of sulphite-reducing clostridia (SSRC) as surrogate for verification of the
inactivation capacity of full-scale ozonation for Cryptosporidium. Wat. Sci.
Technol.: Wat. suppl. 2:163-170.
Hijnen, W. A. M., D. Veenendaal., W. M. H. Van der Speld, A. Visser, W.
Hoogenboezem, and D. Van der Kooij. 2000. Enumeration of faecal indicator
bacteria in large water volumes using on site membrane filtration to assess water
treatment efficiency. Water Res. 34:1659-1665.
Jin, Y., M. V. Yates, S. S. Thompson, and W. A. Jury. 1997. Sorption of viruses
during flow through saturated sand columns. Environ. Sci. Technol. 31:548-555.
Knorr, N. 1937. Die Schutzzonenfrage in der Trinkwater-hygiene. Das Gas- und
Wasserfach 80:330-355.
Konert, M., and J. Vandenberghe. 1997. Comparison of laser grain size analysis
with pipette and sieve analysis: a solution for the underestimation of the clay
fraction. Sedimentology 44:523-535.
___________________________________________________________________
- 225 -
Chapter 9
.
Kuznar, Z. A., and M. Elimelech. 2005. Role of Surface Proteins in the Deposition
Kinetics of Cryptosporidium parvum Oocysts. Langmuir 21:710-716.
Logan, B. E., D. G. Jewett, R. G. Arnold, E. J. Bouwer, and C. R. O’Melia. 1995.
Clarification of clean-bed filtration models. J. Environ. Engineer. 121:869-873.
Logan, J. L., T. K. Stevik, R. L. Siegrist, and R. M. Ronn. 2001. Transport and fate
of Cryptosporidium parvum oocysts in intermittent sand filters. Water Res.
35:4359-4369.
MacKenzie, W. R. H., N.J. , M. E. Proctor, S. Gradus, K. A. Blair, D. E. Peterson, J.
J. Kazmierczak, D. G. Addiss, K. R. Fox, J. B. Rose, and J. P. Davis. 1994. A
massive outbreak in Milwaukee of Cryptosporidium infection transmitted through
the public water supply. New. Engl. J. Med. 331:161-167.
McCaulou, D. R., R. C. Bales, and J. F. McCarthy. 1994. Use of short-pulse
experiments to study bacteria transport through porous media. J. Contam.
Hydrol. 15:1-14.
Medema, G. J., M. Bahar, and F. M. Schets. 1997. Survival of Cryptosporidium
parvum, Escherichia coli, faecal streptococci and Clostridium perfringens in river
water. Wat. Sci. Tech. 35:249-252.
Medema, G. J., F. M. Schets, P. F. Teunis, and A. H. Havelaar. 1998.
Sedimentation of free and attached Cryptosporidium oocysts and Giardia cysts in
water. Appl. Environ. Microbiol. 64:4460-6.
Medema, G. J., and P. J. Stuyfzand. 2002. Presented at the 4th international
symposium on artificial recharge, Adelaide, Australia, September 22-26.
Payment, P., and E. Franco. 1993. Clostridium perfringens and somatic coliphages as
indicators of the efficiency of drinking water treatment for viruses and protozoan
cysts. Appl. Environ. Microbiol. 59:2418-24.
Rajagopalan, R., and C. Tien. 1976. Trajectory analysis of deep-bed filtration with
the sphere-in-a-cell porous media model. AIChE J. 22:869-873.
Rijnaarts, H. H. M., W. Norde, E. J. Bouwer, J. l. Lyklema, and A. J. B. Zehnder.
1995. Reversibility and mechanism of bacterial adhesion. Colloids Surf. B. 5:5-22.
Ryan, J. N., M. Elimelech, R. A. Ard, R. W. Harvey, and P. R. Johnson. 1999.
Bacteriophage PRD1 and silica colloid transport and recovery in an iron oxidecoated sand aquifer. Environ. Sci. Technol. 33:63-73.
Ryan, J. N., and M. Elimelech. 1996. Colloid mobilization and transport in
groundwater. Colloid. surf. A 107:1-56
Schijven, J. F., W. Hoogenboezem, S. M. Hassanizadeh, and J. H. Peters. 1999.
Modelling removal of bacteriophages MS2 and PRD1 by dune recharge at
Castricum, Netherlands. Water Resour. Res. 35:1101-1111.
Schijven, J. F., and S. M. Hassanizadeh. 2000. Removal of viruses by soil passage:
overview of modeling, processes and parameters. Crit. Rev. Envrion. Sci. Tech.
31:49-125.
Schijven, J. F. 2001. Virus removal from groundwater by soil passage. Technische
Universiteit Delft, Delft, the Netherlands.
Schijven, J. F., H. A. M. de Bruin, S. M. Hassanizadehb, and A. M. de Roda
Husman. 2003. Bacteriophages and clostridium spores as indicator organisms for
___________________________________________________________________
- 226 -
Chapter 9
removal of pathogens by passage through saturated dune sand. Water Res.
37:2186–2194.
Scholl, M. A., A. L. Mills, J. S. Herman, and G. M. Hornberger. 1990. The
influence of mineralogy and solution chemistry on the attachment of bacteria to
representative aquifer materials. J. Contam. Hydrol. 6:321-336.
Šimůnek, J., M. Šejna, and M. T. van Genuchten. 1998. The HYDRUS-1D software
package for simulating one-dimensional movement of water, heat, and multiple
solutes in variably-saturated media. Version 2.0, IGWMC-TPS-70. Groundwater
Modeling Center, Colorado School of Mines, Golden, Colorado, 251 pp.
Sobsey, M. D., C. H. Dean, M. E. Knuckles, and R. A. Wagner. 1980. Interaction
and survival of enteric viruses in soil materials. Appl. Environ. Microbiol. 40:92101.
Stumm, W., and J. J. Morgan. 1981. Aquatic chemistry, An Introduction
Emphasizing Chemical Equilibria in Natural Waters. Second edition. Wiley, New
York US.
Stuyfzand, P. J., and H. van der Jagt. 1997. Toelichting op de anorganischechemische analysemethoden voor grond in het LAM van Kiwa. Kiwa,
KOA96.116, Nieuwegein, NL.
Tisa, L. S., T. Koshikawa, and P. Gerhardt. 1982. Wet and dry bacterial spore
densities determined by buoyant sedimentation. Appl. Environ. Microbiol.
43:1307-1310.
Toride, N., F. J. Leij, and M. T. van Genuchten. 1997. The CXTFIT code for
estimating transport parameters from laboratory or field tracer experiments.
Version 2.0 US Salinity Laboratory, Agricultural Research Service US
Department of Agriculture, Report 137, Riverside, California US.
Tufenkji, N., and M. Elimelech. 2004b. Correlation equation for predicting singlecollector efficieny in physicochemical filtration in saturated porous media.
Environ. Sci. Technol. 38:529-536.
Tufenkji, N., and M. Elimelech. 2005. Spatial distribution of Cryptosporidium
oocysts in porous media: evidence for dual mode deposition. Environ. Sci.
Technol. 39:3620-3629.
Tufenkji, N., G. F. Miller, J. F. Ryan, R. W. Harvey, and M. Elimenec. 2004a.
Transport of Cryptosporidium oocysts in porous media: role of straining and
physicochemical filtration. Environ. Sci. Technol. 38:3620-3629.
Van Olphen, M., E. van der Baan, and A. H. Havelaar. 1993. Virus removal at
river bank filtration. H2O (in Dutch) 26:63-66.
Wang, D. S., C. P. Gerba, and J. C. Lance. 1981. Effect of soil permeability on virus
removal through soil. Appl. Environ. Microbiol. 42:83-88.
Yao, K. M., M. Habibia, and C. R. O'Melia. 1971. Water and wastewater filtration:
concepts and applications. Environ. Sci. Technol. 5:1105-1112.
___________________________________________________________________
- 227 -
Chapter 10
General discussion
___________________________________________________________________
- 229 -
Chapter 10
.
INTRODUCTION
Major scientific breakthroughs in drinking water safety occurred in the 19th
century. The recognition of the role of water in transmission of pathogenic
micro-organisms to the public and the protection of the public by means of
water treatment and monitoring for faecal pollution using faecal indicator
bacteria such as E. coli were the most prominent developments in that
period. During the major part of the 20th century both water treatment and
microbiological water quality monitoring to protect public health evolved
gradually (Chapter 1). In the last decade of 20th century a number of
observations and scientific achievements caused a more rapid development
in the field of microbiological safety of drinking water. A growing amount
of epidemiological studies showed the significance of persistent and highly
virulent waterborne pathogens for drinking water safety and the
shortcomings of both water treatment and end product monitoring for E.
coli to guarantee microbial safety. These three aspects were exhibited in the
large waterborne cryptosporidiosis outbreak in Milwaukee in 1993. At the
same time, dose-response studies of relevant waterborne pathogens were
published in literature which enabled a quantitative calculation of health
risks of exposure to pathogens, i.e. through the consumption of drinking
water.
On the basis of this information and the awareness that the current end
product monitoring of faecal indicators is a curative strategy with a delay
time during which consumers can be exposed to pathogens, Dutch
authorities implemented a health-based target for microbial safety in the
revised Drinking Water Decree (Anonymous, 2001). Water Companies
using surface water as the source water should demonstrate compliance
with an annual infection risk level of 10-4 per person for a number of index
pathogens: enteroviruses, Campylobacter bacteria and two parasitic
protozoa, Cryptosporidium and Giardia (Wetsteyn et al., 2005). Compliance
with this infection risk should be demonstrated by means of a Quantitative
Microbial Risk Assessment (QMRA) which is part of the total system
approach of Water Safety Plans, covering the whole water system from
source to tap (WHO, 2004).
The most direct approach would be to monitor tap water for the index
pathogens. However, the concentrations at which these pathogens pose an
infection risk above 10-4 per person per year (pppy) are extremely low,
much lower than the detection limit of pathogen detection methods. Hence,
the concentration of pathogens in tap water has to be estimated from their
___________________________________________________________________
- 230 -
Chapter 10
concentration in source water and their removal by water treatment. The
general objective of the studies presented in this thesis was to develop and
evaluate methods for the assessment of the elimination capacity of fullscale water treatment processes for the index pathogens.
The line of approach in this study was to develop a method as closely
related to the natural conditions in water treatment and to the daily
practice of microbiological water quality monitoring. Microbiological water
quality monitoring for multiple faecal indicator bacteria including the
persistent spores of sulphite-reducing clostridia SSRC (incl. C. perfringens)
in surface waters and finished drinking waters is mandatory in the
Netherlands since 1984. Besides this mandatory program most Water
Companies monitor these indicator bacteria also after selected processes in
treatment. The gerneral hypothesis of this study was therefore that
Escherichia coli and spores of sulphite-reducing clostridia can be used as
process indicators to assess the elimination capacity of water treatment
processes for, respectively, pathogenic micro-organisms that are susceptible
and pathogenic micro-organisms that are resistant to disinfection processes
under the commonly applied conditions in water treatment.
Important rationale for the increased interest in microbiologically safe
drinking was the growing knowledge on health risks coming from the
pathogenic protozoa, Cryptosporidium and Giardia (Medema, 1999). For that
reason the current study focused on SSRC to describe protozoa removal
and subsequently turned the attention to E. coli to describe removal of
Campylobacter. In the growing understanding that viruses are also
pathogens of interest for water safety (de Roda Husman, 2001; Fernandes et
al., 2007) and these pathogens are the most critical micro-organisms in
water (in)filtration (Schijven, 2001), filtration experiments were conducted
in this thesis to determine the elimination of bacteriophages as a process
indicator for viruses (Havelaar, 1986).
E. COLI AND SSRC AS PROCESS INDICATORS UNDER FULLSCALE CONDITIONS
Routine water quality monitoring. The water companies have large
amounts of historical monitoring data on the prevalence of faecal indicator
bacteria in their source waters and the water further down the treatment,
assessed with standard microbiological methods. These data sets
demonstrate the continuous presence of both process indicators in Dutch
surface waters used for drinking water production and showed the
___________________________________________________________________
- 231 -
Chapter 10
.
potential of both indicators to quantify efficiency of processes to eliminate
micro-organisms (Chapter 2). These data yield valuable quantitative
information of the first processes in water treatment on the elimination of
these indicators expressed as the Decimal Elimination Capacity (DEC).
These routinely collected data with a very small amount of positive
samples, however, are not suited to determine accurately DEC of a total
treatment and its variability (Chapter 2; Drost et al., 1997; Evers and
Groennou, 1999).
Large volume sampling. The current study showed that
modification of the standard membrane filtration method to examine a
larger water volume yielded a simple and reliable method with a low
detection limit to monitor faecal indicator bacteria in a total water
treatment chain (Chapters 2, 3 and 4). Larger volumes can be examined by
either the standard membrane filtration technique (mf method; up to 10
litres, Chapter 2) or by an up-scaled membrane filtration method (larger
membrane) with a special device developed for in situ sampling (Figure 1),
the MF sampler (Chapter 3; volumes of 10 – 1000 litres).
Figure 1. The MF-sampler used under full-scale conditions
In microbiological studies in water treatment large volume sampling has
been used by others (Goyal et al., 1980; Van Olphen et al., 1993; Payment et
al., 1989, 1991, 1993; LeChevallier and Norton, 1991). These studies used
filtration and/or adsorption followed by elution. However, these methods
are not rapid, simple, reliable and unambiguous and relatively expensive
for routine monitoring. Because the MF sampler method was attuned to the
common laboratory practices to detect indicator bacteria in water,
implementation of the method in water quality monitoring by the Water
___________________________________________________________________
- 232 -
Chapter 10
Companies was easy. This was demonstrated by the studies described in
Chapter 3 and 4 where data on elimination of Coli44 and SSRC by full-scale
treatment plants were collected in collaborative studies with Dutch Water
Companies. And also by the use of the method in drinking water practices
by the Water Companies (personal communications) and in other
microbiological research projects (Hoogenboezem et al., 2001b; Medema
and Stuyfzand, 2002; Smeets et al., 2006).
Elimination of process indicators by full-scale treatment. The study
in eight Dutch treatment facilities presented in Chapter 4 clearly showed
that traces of faecal contamination (Coli44 and SSRC) were occasionally
found in finished waters. SSRC where detected more frequently and at
higher concentrations than the susceptible Coli44. This was not only in
treatments with primary chemical disinfection but also in a treatment
without this process. This clearly points to differences in transport and fate
of susceptible and persistent faecal micro-organisms in water treatment.
A considerable variation in the decimal elimination capacity (DEC)
between the eight full-scale treatment facilities for Coli44 and SSRC was
observed. For Coli44 the DEC ranged from 3.0 – 6.6 log and for SSRC from
1.2 – 4.5 log. The observed differences in the DEC between different
treatment facilities for SSRC were attributed to design and operational
conditions which have been tailored to the quality of the source water,
conclusions also drawn from the work of others for protozoan oocysts
(LeChevallier and Norton, 1991, 1992; Smeets et al., 2007).
Monitoring of Coli44 and SSRC in large volumes demonstrated also a
considerable variation in DEC values of the same type of processes at the
eight different facilities (Chapter 4). The largest differences were observed
for chemical disinfection processes, slow sand filtration and granular
activated carbon filtration. Important observation was the relatively low
and variable inactivation capacity of full-scale chlorination and ozonation
for Coli44, lower than expected from disinfection kinetics established in
laboratory studies. This is a very significant observation for water
treatment practice, as it shows that simple extrapolation from laboratory
disinfection studies to full scale water treatment, as is common in the water
sector, may seriously overestimate the treatment efficacy. Consequently,
inactivation kinetics for indigenous micro-organisms under full-scale
conditions deviates considerably from the kinetics of pure cultures assessed
under well defined laboratory conditions, mostly batch tests. This was also
observed in a literature study on UV (Chapter 6). Basis for this discrepancy
are factors related to conditions of the studied micro-organisms such as
physical protection (particle association), physiological or metabolic
___________________________________________________________________
- 233 -
Chapter 10
.
conditions and variation in susceptibility within one species. Different
susceptibility of lab-cultured and environmental micro-organisms is
demonstrated in other studies (Hijnen et al., 2004a; Smeets et al., 2005). Also
process related factors are responsible for the different inactivation kinetics
in lab-scale and full-scale disinfection studies. Low inactivation of Coli44
and SSRC observed in a local full-scale ozonation was evaluated by a
Hazard Analysis Critical Control Point (HACCP) study which is an integral
part of a Water Safety Plan (WHO, 2004). A team of process technicians and
experts designated the high DOC content of the water, the applied ozone
dosing strategy in combination with poor hydraulics of the full-scale
system as major causes (Hijnen et al., 2001). Evidence for the role of poor
hydraulic conditions on inactivation efficacy was obtained in the studies
presented by Van der Veer et al. (2005) and Smeets et al. (2006).
The study in Chapter 4 demonstrated the high efficacy of slow sand filters
to remove Coli44 (2 – 3 log). The DEC of these filters for SSRC, however,
was low and variable (0 – 2 log). This phenomenon was attributed to
accumulation, survival and delayed breakthrough deduced from high
SSRC concentrations in the sand bed (Chapter 3). Similar observations were
reported for granular activated carbon GAC filters (Hijnen et al., 1997)
where high SSRC concentrations were detected in both the filter bed and
the backwash water. More recently, low removal of aerobic and anaerobic
spores by GAC filtration were observed in literature (Galofré et al., 2004;
Mazoua and Chauveheid, 2005) and these authors suggested a similar
explanation for their results. The validity of the proposed explanation for
the low spore removal in slow sand filters and soil infiltration with no filter
cleaning was supported by results presented in Chapters 7, 8 and 9 of this
thesis.
In conclusion the DEC of full-scale processes can be assessed with faecal
indicator bacteria as process indicators. By increasing the sample volume of
the standard analytical methods the elimination capacity of the total
treatment can be assessed more accurately with additional information on
the variability of the DEC. The DEC of full-scale treatment for
environmental micro-organisms is potentially lower than elimination
capacities assessed for the same processes under well defined conditions in
challenge studies under pilot plant or laboratory conditions commonly
used to assign log credits to water treatment processes (LeChevallier and
Au, 2004; von Huben, 1991; USEPA, 2006). This underlines the importance
of obtaining quantitative information on elimination of micro-organisms
under full-scale and site specific conditions as proposed and studied in this
thesis.
___________________________________________________________________
- 234 -
Chapter 10
Results, limitations and remaining questions. But the results also
revealed some limitations and remaining questions concerning the
methodology:
- The proposed MF method to assess DEC of water treatment processes
with indicator bacteria Coli44 and SSRC is primarily focussed on the
index
pathogens
Campylobacter
and
Cryptosporidium/Giardia.
Enteroviruses, however, are also important pathogens for water safety
(Regli et al., 1991) and implemented in the revised Dutch Drinking
Water Decree (Anonymous, 2001). Bacteriophages, especially F-specific
RNA phages (FRNA), are described as appropriate process indicators
for the removal of enteroviruses in water treatment (Havelaar, 1986;
1993; Schijven, 2001). A standard for bacteriophages was not included
in the drinking water regulations. Additional research on
bacteriophages as process indicators for (entero)viruses is
recommended.
- The MF method for indicator bacteria is limited to one organism at a
time. Successive membrane sampling at one sampling point enables to
monitor multiple bacterial species, however. Recently, a new method
using cross-flow membrane filtration has been introduced for detection
of multiple organisms, including bacteriophages. The observed
recoveries for bacteriophages, bacteria and protozoa were as high as the
recoveries of the MF-sampler method for bacteria and spores
(Veenendaal and Brouwer, 2007). This method was first described by
Leclerc et al. (1977) as a technique to continuously and automatically
determine E. coli in 100 ml of water. The technique was also applied to
overcome the problems with low recoveries of the dead end filtration
technique used to assess concentrations of Cryptosporidium and Giardia
(oo)cysts (Simmons et al, 2001) and explored for multiple microorganism detection (Morales-Morales et al., 2003; Hill et al., 2005, 2007).
This cross-flow filtration method was successfully applied in the
experiment of Chapter 7 and further investigations are recommended
to explore the use of this new LVS method in drinking water practices.
- In the course of this study and the performance of the provisional risk
assessments it became apparent that for some processes site specific
full-scale information on the elimination of Coli44 and SSRC was not
assessable because concentrations were too low to detect (infiltration
and soil passage, slow sand filters, post-disinfection) or the processes
were not in operation under full-scale conditions (UV disinfection).
Consequently, alternative methods are needed to assess DEC of these
processes.
___________________________________________________________________
- 235 -
Chapter 10
.
-
The methodology with the two process indicators is valuable to
establish elimination capacity and robustness of treatment for water
safety, but is not suited to control the treatment processes for an
optimal DEC and for momentary loss of DEC. The method has a delay
time of at least 24 hours. These temporarily “bad day” conditions with
potentially high concentrations of micro-organisms in the finished
water may contribute to a large extent to the annual risk of infection
(Smeets, 2008). The results of Chapter 2 and 4 have shown that Coli44
and SSRC can detect “bad day” conditions and help to elucidate how
operational process conditions such as dosing, flow rates and contact
times affect the DEC. This enables the design of an on line control
strategy for optimal elimination conditions and may in the long run
lead to process models for predicting the elimination of microorganisms on the basis of on line process parameters. Analyzing actual
full-scale process and elimination data by statistical techniques such as
multiple-regression analysis (Wiersema, 1999; Haas et al., 2001) is an
example of such studies. Both studies identified pH, coagulation doses
and polymers, temperature, turbidity as conditions affecting the
elimination rate in coagulation processes. The outcome of such studies,
however, largely depends on the quality of the collected data base on
process parameters and microbial elimination data. Another example of
such studies is presented by Smeets (2008) who combined process
models for disinfection with full-scale elimination data and used
stochastic modelling to describe elimination capacity and estimate
micro-organisms concentrations in treated water.
- In the process of QMRA site specific quantitative information is
required on the elimination of index pathogens which in most cases is
not available as indicated before. It is unsure how well both process
indicators can be translated to the elimination of these index pathogens.
Some data on Coli44 as indicator for Campylobacter already indicated
the value of this process indicator. But for the Clostridium spores there
were observations at GAC filtration and slow sand filtration which cast
doubt on the use of this parameter as process indicator for the parasitic
protozoa for these processes.
Consequently, additional research was necessary to find alternative
strategies for the assessment of the DEC of selected processes, to elucidate
the most significant conditions that affect the variability in DEC and to
compare the elimination of process indicators (PI), E. coli and spores of
sulphite-reducing clostridia, with the elimination of the index pathogens
(IP) under similar conditions.
___________________________________________________________________
- 236 -
Chapter 10
ADDITIONAL METHODS TO ASSESS DEC, TO STUDY
CONDITIONS AND TO VALIDATE PROCESS INDICATORS
To overcome the limitations of the use of faecal indicators as process
indicators for DEC assessment and to address part of the remaining
questions the use of challenge tests or dosing experiments was explored
(Chapters 5, 7, 8 and 9). Dosing micro-organisms to full-scale drinking
water treatment processes is generally not allowed and feasible. Therefore
dosing experiments were done at pilot or laboratory scale under controlled
conditions, mimicking full-scale conditions. In this type of study
elimination can be quantified precisely for selected micro-organisms, water
type and process conditions. Dosing experiments with multiple organisms
under the same conditions enabled comparison of the elimination of the
different pathogens and their process indicators under similar conditions in
order to assess their elimination ratio (IP/PI ratio) and to investigate the
influence of specific process variables. The F-specific RNA phage MS2 was
dosed in these studies as model for virus removal.
Challenge tests in bench-scale experiments intrinsically deviate from the
full-scale conditions where multiple conditions affect the overall
elimination. Consequently, the result of such tests potentially over- or
under estimate the real elimination under normal practices. In the course of
this study verification of the validity of these challenge tests for assessment
of DEC of full-scale processes was part of the objectives. For this
verification three strategies have been applied:
- when elimination data of environmental Coli44 and SSRC under fullscale conditions are available:
o comparison of elimination of both environmental process
indicators under the challenge test and full-scale conditions;
o comparison of elimination of dosed E. coli and spores of C.
perfringens under the challenge test conditions with Coli44
and SSRC elimination under full-scale conditions.
- when no information of elimination of the process indicators are
available: comparison with elimination data from scientific literature of
either full-scale systems or other challenge tests. Quantitative analysis
of literature data is presented in Chapter 6 and a separate review report
(Hijnen and Medema, 2007).
___________________________________________________________________
- 237 -
Chapter 10
.
DISINFECTION CHALLENGE TESTS
With respect to disinfection processes the study focussed on the two
disinfection processes of major interest in the Dutch drinking water
industry, ozonation and disinfection with ultraviolet radiation (UV).
Batch experiments. Disinfection studies under well defined
conditions on laboratory scale are usually batch tests where pre-cultured
micro-organisms are exposed to different doses of disinfectant (chlorine,
ozone, UV). These inactivation kinetic studies are relatively simple and
inexpensive. From the obtained dose-response curves the first order
inactivation rate constant k specific for a combination of micro-organism
and disinfectant is derived. These studies yield basic information on the
inactivation kinetics and susceptibility of the different micro-organisms to a
disinfectant (k value). Moreover, the effects of some water quality
conditions such as temperature and pH can be quantified which is of
interest for process modelling. With the k values and the quantified effect
of temperature and water quality on these k values disinfection processes
are designed to achieve a desired inactivation capacity (USEPA, 2006).
Translation of these batch test results to ‘real world’ disinfection processes,
however, will over-estimate the inactivation efficacy of these processes
(Chapter 2, 3, 4 and 6). The conditions in a full-scale continuous flow
system are not as optimal as the conditions in a batch laboratory system
(variability in hydraulics and dosages). Furthermore, environmental microorganisms potentially have different susceptibility to disinfectants than
pre-cultured organisms.
Continuous flow systems. In continuous flow bench-scale systems
process conditions can be created that are close to full-scale conditions. In
these systems inactivation was less than expected from simple translation
of batch study results and closer to the inactivation of micro-organisms
observed under full-scale conditions for ozone and UV (Chapter 5; Hijnen
et al., 2004b; Chapter 6). With continuous flow systems the lower
susceptibility of environmental micro-organisms for UV disinfection
(Hijnen et al., 2004a) and ozone (Smeets et al., 2005) compared to labcultured species of the same group was demonstrated which accounts, in
part, for the discrepancy between disinfection efficacy observed in lab and
pilot tests. This emphasizes the need for further research on the difference
between the susceptibility of environmental and lab-cultured microorganisms.
___________________________________________________________________
- 238 -
Chapter 10
FILTRATION CHALLENGE TESTS
In this thesis the lack of quantitative information on removal of index
pathogens by slow sand filtration and surface water infiltration and the
IP/PI ratios for these processes was addressed by challenge testing with
pre-cultured organisms on pilot plant and laboratory scale (Chapters 7, 8
and 9).
Laboratory column tests. Small scale column tests operated under
conditions simulating full-scale filtration conditions are relatively simply
and cheap to perform. The experiences in the course of this study showed
that the quantitative elimination data collected which such studies,
however, can not be translated directly to full-scale conditions (Chapter 7
and 9). The filtrations conditions in the columns deviate too much from the
full-scale filtration conditions. This conclusion was derived from
comparison of column test data with elimination data of environmental
process indicators under full-scale filtration conditions. The columns were
challenged with multiple micro-organisms under variable filtration
conditions. This yielded comparative quantitative information on the
relative elimination of viruses, bacteria and bacterial spores and
Cryptosporidium and Giardia by slow sand filtration and surface infiltration.
Furthermore, with these studies the relative influence of the geo-chemical
and morphological properties of the sand and the infiltration rate on the
overall elimination in sand beds was demonstrated. Under both conditions,
slow sand filtration and surface water infiltration, the column studies
showed that the average grain size of the sand is less important for
microbial elimination than the geo-chemical properties of the sand.
Microbial transport in filter material or soil is governed to a large extent by
the size of micro-organisms. The infiltration column study (Chapter 9),
however, revealed that in a soil column with fine and uniform sand
Cyptosporidium oocysts (4 μm) were removed to a lower extent than E. coli
and spores of C. perfringens (both approx. 1 μm), an observation also
described by Nobel et al. (1999). This emphasizes the role of surface
interactions in the transport of micro-organisms through sand or soil. The
higher effect of filtration rate on elimination of bacteriophages compared to
bacteria, bacterial spores and (oo)cysts demonstrate the additional effect of
straining in the elimination of the larger micro-organisms. These results
show that column studies are useful to compare removal of different microorganisms in order to assess the PI/IP ratio and to understand the impact
of process conditions on the elimination of micro-organisms. This
___________________________________________________________________
- 239 -
Chapter 10
.
information is necessary for further understanding and modelling of
microbial transport in (in)filtration processes.
Challenge tests on pilot plant scale. Slow sand filters (SSF) of a
pilot plant operated as dummy of a full-scale SSF process (Chapters 7, 8;
Dullemont et al., 2006; Schijven et al., 2007) were used to determine the
elimination of viruses, bacteria and bacterial spores and protozoan
(oo)cysts. Comparison of the results with results of elimination of
environmental micro-organisms under full-scale conditions revealed that
challenge tests on pilot plant scale were valid to assess the DEC of full-scale
filters for bacteria (E. coli and Campylobacter) but not for SSRC (Chapter 8).
The elimination of dosed spores of C. perfingens was much higher than the
elimination of “natural” SSRC by full-scale filters. Evidence for the
suggested mechanism of delayed transport as explanation for the variable
and sometimes low DEC of persistent micro-organisms by filters with no
filter back wash was given in Chapter 8. The rapid decline of oocyst
concentrations in the challenged filter bed suggested a lower and most
likely negligible risk of delayed breakthrough of oocysts in these filters. The
role of predation in this oocyst decline was hypothesised, since
zooplanktonic organisms which have been described to ingest oocysts are
present in these biological filters. The potential role of these organisms as
transport vector to the outlet of the filter bed is an issue of further research.
In the pilot pant study the large positive effect of the presence of a
Schmutzdecke on a slow sand filter on the elimination of E. coli was
demonstrated (Chapter 7). The results on MS2 bacteriophage removal
showed that the Schmutzdecke has a minor effect on virus removal.
Another condition that is of importance for the DEC of slow sand filtration
is temperature. The DEC of the process for E. coli and MS2 phages assessed
in challenge tests varied with 3 log in a temperature range of 7 - 16oC
(Hijnen et al., 2006).
Challenge tests on pilot plant scale to assess the efficiency of conventional
processes (coagulation/filtration) as a microbial barrier in water treatment
have been used increasingly over the last decade. Especially studies on
Cryptosporidium and Giardia removal by conventional treatment have been
presented (Lodgsdon et al, 1981; West et al, 1994; Patania et al, 1995;
Nieminski et al, 1995; Dugan et al, 2001; Emelko, 2001; Huck et al., 2001). In
these studies effects of sub-optimal process operation on micro-organism
breakthrough and water safety have been addressed since the recognition
of the significant effect of such treatment inadequacies on water safety
(Badenoch, 1990; Richardson et al., 1991; Craun, 1990). In challenge tests the
loss of elimination capacity of such processes was clearly demonstrated and
___________________________________________________________________
- 240 -
Chapter 10
quantified at sub-optimal coagulant doses and conditions (pH, mixing) and
risk moments in filtration (increasing pressure drops, backwashing
moments and start up) as summarized in a review (Hijnen and Medema,
2007).
In conclusion, pilot plant studies operated as a dummy of the full-scale
processes are valuable studies to assess the DEC and to elucidate the
influence of process conditions on the DEC.
QUANTITATIVE ANALYSIS OF LITERATURE DATA
The international literature on microbial inactivation and elimination is a
significant source of quantitative data and was extended in the last decade
(LeChevallier and Au, 2004; Hijnen and Medema, 2007). The review of the
first authors described the elimination capacity of a broad range of water
treatment processes without further quantitative analysis. A more
quantitative method of literature reviewing was explored in this study for
UV (Chapter 6) and by Hijnen and Medema (2007; third edition) for the
major treatment steps currently used in treatment facilities
(coagulation/floc removal, rapid granular filtration, slow sand filtration).
In this method quantitative literature data were used to calculate a default
value, the Microbial Elimination Capacity (MEC) of the process for viruses,
bacteria and bacterial spores and prototzoan (oo)cysts. The MEC was
calculated from the DEC values in those literature studies qualified for
inclusion in the calculation. This qualification was based on evaluating the
study design/report: experimental conditions, microbial assays and
process information. The MEC-value was the weighted average value of
the DEC values derived from the different studies. The weighting was
based on a Full-scale index (FSI), with an increasing value of successively
lab-scale, pilot plant and full-scale with either dosed or environmental
organisms. With the scale of the studied processes on the y-axis and the
conditions of the used micro-organisms on the x-axis, the data were ranked
in an x-y matrix with values of 1 – 5. Drawback of this weighting system is
that the same rank number between 1 and 5 represents either low
conformity with the full-scale conditions or low conformity with the
environmental pathogen of concern while the effect of both conditions on
the DEC is probable not equal. Because data sets from full-scale studies
with environmental micro-organisms were limited and in most cases
evenly distributed over these data sets, the influence of this weighting on
the calculated MEC was also limited (Hijnen and Medema, 2007). This
diminishes the significance of the noticed drawback of this weighting on
___________________________________________________________________
- 241 -
Chapter 10
.
the assessed MEC values. The FSI was further used as a descriptive
parameter to qualify the studies with respect to their conformity with the
full-scale conditions. The range of the MEC was given by the minimum and
maximum observed the DEC values. The effect of process conditions on the
DEC values was presented by describing the major process conditions in
the review and summarizing specific experimental data addressing the
effect of selected process conditions.
The literature review yielded MEC values of conventional treatment
processes (coagulation and filtration) and slow sand filtration for viruses
(incl. enteroviruses and bacteriophages), bacteria (incl. indicator bacteria
and Campylobacter) and bacterial spores (incl. Clostridium and Bacillus),
Cryptosporidium and Giardia. For conventional treatment processes the
highest MEC-values were observed for protozoan (oo)cysts, but in general
the MEC values of the different organisms were in the same order of
magnitude. The variability of the MEC values was high and only for
obvious conditions low or high elimination could be attributed to process
conditions (free sedimentation versus lamellae separation; pre-oxidation).
The UV review (Chapter 6) yielded a required fluence table for a range of
micro-organisms relevant for drinking water safety in which fluence for
bacterial pathogens was corrected for the discrepancy between dose-effect
observed in batch tests for pre-cultured organisms and in continuous flow
systems for environmental organisms. Viruses and Acanthamoeba are the
most persistent pathogens. Susceptibility of bacteria and Cryptosporidium
and Giardia to UV were in the same order of magnitude.
These default elimination capacities derived from literature were used in
QMRA calculation for specific locations with no or limited quantitative
data on process indicator elimination (Medema et al., 2006). The MEC was
used as an input value for a point estimate of the annual infection risk from
index pathogens in drinking water of the location. In a more
comprehensive stochastic calculation the uncertainty level of this annual
infection risk was derived from the range in the MEC value, assuming a
simple triangular distribution.
PROCESS INDICATOR VALIDATION
The IP/PI ratio. To validate the use of process indicators to establish
the elimination of the index pathogens, comparative studies on elimination
are needed. The ratio between the index pathogen and the process indicator
elimination (IP/PI ratio) is a parameter to evaluate this aspect of the
process indicator use. The ideal IP/PI ratio is 1.0, but on the basis of the
___________________________________________________________________
- 242 -
Chapter 10
results presented in this thesis these ratios will be variable. The impact of
these ratios on the prediction of the elimination of the index pathogen is
demonstrated in Figure 2, assuming that this ratio is the slope of a linear
relationship between the elimination of both organisms, independent from
the extent of removal. This shows that the over- or under-estimation of the
elimination of the index pathogen will increase with the increasing
elimination capacity of the process.
DEC index pathogen (log)
6
ratio 1.5:
extra safe
5
4
ratio 1.0:
DECPI = DECIP
3
ratio 0.5:
unsafe
2
1
0
0
1
2
3
DEC process indicator (log)
4
Figure 2. The effect of the use of the IP/PI ratios on the predicted DEC values at a
ratio of respectively 0.5, 1.0 and 1.5
The IP/PI ratios assessed from the comparative studies with multiple
micro-organisms presented in this thesis supplemented with additional
studies from literature are summarized in Figure 3 and 4. In these Figures
the ideal IP/PI ratio of 1.0 is depicted by a straight line.
Process indicator validation for Cryptosporidium. Figure 3 shows
for Clostridium spores (SSRC and C. perfringens) as potential process
indicator for Cryptosporidium removal by slow sand filtration and
infiltration a wide range of ratios of <1.0 to >>1.0. Consequently, these
spores are not suitable as process indicator for Cryptosporidium for these
processes. The ratio assessed for SSRC/Cryptosporidium inactivation by
ozone of 0.8 was close to the safe ratio of 1.0. For the use of Clostridium
spores in conventional water treatment two IP/PI ratios of 0.9 and 1.0 were
derived from a direct comparative study (Payment and Franco, 1993) and
an indirect comparative study (Hijnen et al., 2003; data of IP and PI
elimination collected in two separate periods). The use of Bacillus spores as
process indicator was intensively studied in the US. For a more
pronounced conclusion on the IP/PI ratio of bacterial spores/
___________________________________________________________________
- 243 -
Chapter 10
.
Cryptosporidium elimination by conventional treatment results from two
external comparative studies were used (Huck et al., 2001; Prevost et al.,
2007, large review of data from 15 studies). These studies demonstrated
that IP/PI ratio for these aerobic spores as process indicator for
Cryptosporidium was close to the ideal value of 1.0 with the majority >1.0
(safe area).
The MEC values of conventional treatment processes (coagulation and
rapid granular filtration) presented in the literature review (Hijnen and
Medema, 2007) can also be used to calculate IP/PI ratios. These ratios were
1.3 - 1.6 and in agreement with the data presented in Figure 3.
DEC IP (Cryptosporidium )
8
SSF (C. perfringens)
SSF (SSRC)
Infiltration (C. perf.)
Infiltration (SSRC)
Conv. treatm. (C. perf.)
Conv. treatm. (SSRC)
RGF (Bacillus)
Ozone (C.perfringens)
Conv. treatm. (Bacillus)
IP/PI = 1.5
7
6
IP/PI = 1.0
5
4
3
2
1
0
0
1
2
3
4
5
6
7
8
DEC PI ( spores)
Figure 3. The IP/PI ratio of Cryptosporidium and bacterial spores C.
perfringens (C. perf.), sulphite-reducing clostridia (SSRC) and aerobic spores
(Bacillus) for a number of water treatment processes: Slow sand filtration (SSF;
Chapters 7,8), Infiltration (Chapter 9, Nobel et al., 1999), Conventional treatment
(Payment and Franco, 1993; Hijnen et al., 2003; Prevost et al., 2007) and rapid
granular filtration (Huck et al., 2001)
Process indicator validation for Campylobacter. The IP/PI ratios
for Coli44 (incl. E. coli) as process indicator for Campylobacter elimination by
rapid granular filtration, slow sand filtration and ozone (Figure 4) showed
that Coli44 complies with the requirement of a proper process indicator.
Most ratios were in the safe area of >1.0, but not too high (>1.5).
Predictive value of PI for “bad day” elimination conditions. From
the results presented above it can be concluded that the IP/PI ratio exhibit
___________________________________________________________________
- 244 -
Chapter 10
a level of variation caused by the differences in characteristics of pathogen
and process indicator related to the elimination process. This variation in
IP/PI ratio can be included in a stochastic QMRA modelling tool used to
calculate the level of uncertainty of the infection risk. Precondition is that
the variation is random and not related to specific process conditions where
the fate of process indicator and pathogen significantly differs.
Consequently, the reaction of IP and PI in treatment process on specific
conditions affecting elimination must be as similar as possible. Differences
in persistence and surface properties between micro-organisms are such
characteristics of interest for the fate of PI and IP in disinfection processes
and filtration processes.
6
DEC IP (Campylobacter )
IP/PI = 1.5
5
RGF (Coli44)
IP/PI = 1.0
SSF (Coli44/E.coli)
Ozone (E.coli)
4
3
2
1
0
0
1
2
3
4
5
6
DEC PI (Coli44 incl. E. coli )
Figure 4. The IP/PI ratio of Campylobacter and Coli44 (incl. E. coli) for a
number of water treatment processes: rapid granular filtration (RGF; Hijnen et al.,
1998), slow sand filtration (Hijnen et al., 1995; Chapter 7, Dullemont et al., 2006)
and ozone (Smeets et al., 2005)
Difference in persistence of micro-organisms and the distribution in
disinfectant dose of full-scale disinfection processes are two major
characteristics which affect the inactivation of these different microorganisms and consequently the IP/PI ratios. Generally spoken the doseeffect relation of disinfection processes can be assessed more accurately
with persistent micro-organisms than with the more susceptible microorganisms. This was demonstrated by ozone studies with C. parvum oocysts
and bacterial spores (Oppenheimer et al., 2000; Chapter 5, Hijnen et al.,
___________________________________________________________________
- 245 -
Chapter 10
.
2004b) and UV studies with bacterial spores and bacteriophages (Qualls
and Johnson, 1983; Sommer et al., 1999; Hijnen et al., 2004a; USEPA, 2006).
Dose assessment of UV systems is done with UV calibrated pre-cultured
aerobic spores of B. subtilis or MS2 bacteriophages with relatively low and
known susceptibility to UV. These process indicators are being used as
biodosimeters for certification of full-scale UV systems. More recently B.
subtilis is also suggested as biodosimeter for full-scale ozone systems
(Broséus et al., 2008). The more susceptible micro-organisms are better
process indicators to study the variability in efficacy of disinfection
processes related to the hydraulics of the continuous flow system. In a
model study on UV biodosimetry Cabaj et al. (1996) demonstrated the
integrated effect of the susceptibility of micro-organisms and the hydraulics
of a UV reactor on the inactivation capacity. At a constant dose of 40
mJ/cm2 their numerical modelling showed a clear decrease in inactivation
capacity (or Reduction Equivalent Dose; RED) with increased susceptibility
of the micro-organisms (k-value) and dose distribution. Consequently, the
broadness of the dose distribution has more effect on the inactivation of
susceptible micro-organisms than on the resistant ones. This was also
demonstrated for ozone disinfection by Smeets et al. (2006). An increasing
broadness of dose distribution expressed as decreasing number of
continuously stirred tank reactors (CSTR), showed the largest decrease in
inactivation of micro-organisms with the highest k-value. The higher
impact of dose distribution of full-scale ozonation and chlorination
processes on the inactivation efficacy of susceptible micro-organisms such
as E. coli implicate that this indicator bacterium can be used as process
indicator to determine if the dose distribution in the contact chambers is
adequate for the required log-inactivation. The log-inactivation for
pathogenic organisms that are equally or more resistant (viruses and
protozoan (oo)cysts) will never exceed the log-inactivation of E. coli. More
recently, Computational fluid dynamics (CFD) has become an engineer
based tool of growing interest in design and control of full-scale water
treatment disinfection and has been used to describe the dose distribution
in UV systems (Ducoste et al., 2005) and more recently also in full-scale
ozonation contactors (Li et al., 2006).
The difference in persistence of micro-organisms has also consequences for
elimination and the IP/PI ratio in filtration processes with low or no filter
bed cleaning as demonstrated for the elimination of SSRC and
Cryptosporidium in slow sand filtration and surface water infiltration
(Chapter 7, 8 and 9). While attached anaerobic spores will persist long
under these conditions, attached oocysts are susceptible to predation
___________________________________________________________________
- 246 -
Chapter 10
(Chapter 8) and potentially disappear under these conditions, although not
yet everything is clear on the fate of these ingested oocysts.
On the other hand, the similar reaction of attached spores of C. perfringens
and oocysts of Cryptosporidium in soil columns on a pH increase in the
water presented in Chapter 9 indicates similarity in surface attachment of
both micro-organisms and fate in filter beds with short contact times and
frequent filter bed back washing. Similarity of reactions of spores and
oocysts on coagulation and filtration conditions were also notified by
others (Huck et al., 2001; Emelko, 2001). This confirms the predictive value
of bacterial spore elimination for elimination of oocyst in these water
treatment processes.
APPLICATION OF THE METHODS IN QMRA
An example of QMRA. During the last few years Quantitative Microbial
Risk Assessment (QMRA) was applied by the Dutch Drinking Water
Companies as a tool to assess drinking water safety. The provisional
QMRA’s used the methodologies described in this thesis to determine the
elimination capacity of water treatment. An example of such calculations is
presented in Figure 5.
Sample volumes: 0.1 - 1.0 -10 -100 liter
12
DEC (log)
10
8
SSRC/Cryptosporidium
E. coli/Campylobacter
Challenge
test PI
6
4
2
0
Step 1
Step 2
Step 3
Step 4
DECr
Figure 5. The cumulative DEC of a train of full-scale water treatment processes
with four treatment steps for Coli44 and SSRC and the required DEC (DECr) for
the index pathogens to meet the statutory infection risk of 10-4 pppy (error bars
presents the range of DECr based on minimum and maximum IP concentrations in
the source water; steps 1-4, RGF, Ozone, GAC filtration, SSF)
___________________________________________________________________
- 247 -
Chapter 10
.
Indicator bacteria were monitored in variable sample volumes depending
on the location in treatment. For the final step in treatment (SSF), the DEC
was determined by challenge tests on pilot pant scale (PI; Chapters 7, 8;
Dullemont et al., 2006).
On basis of these treatment data and data on the concentration of the index
pathogens in the source water, the annual risk of infection for
Cryptosporidium and Campylobacter in the drinking water of this locations
was 1.1x10-6 and 6.6x10-5 per person per year (pppy), respectively. Using the
more comprehensive method with stochastic modelling of water quality
and treatment efficiency data (Medema et al., 2003) the average annual risk
of infection for both index pathogens was 1.9x10-7 and 6.0x10-5 with 97.5percentile values of 1.5x10-6 and 1.7x10-4, respectively.
This example shows that the average and range of the annual risk of
infection for both index pathogens could be assessed with full-scale
monitoring of both process indicators in variable sample volumes and
additional pilot plant research. The average risk for Cryptosporidium and
Campylobacter is below 10-4 pppy. The IP/PI ratios for both
pathogen/process indicator combinations are usually >1.0 (Figure 3 and 4),
suggesting that the risk of infection is probably lower.
Significance of variability of the DEC in QMRA. The variability in
the exposure assessment depends on the variability in the concentrations of
index pathogens in the source water but also on the variability in treatment
efficiency (Teunis et al., 1997; Medema et al., 2003). Assessment of
variability in the DEC by determination of actual Decimal Elimination (DE)
with paired Cin and Cout concentrations by date as applied in the current
study is common practice in microbiological risk assessment studies
(LeChevallier and Norton, 1991; Nieminski et al., 1995; Teunis et al., 1997).
The observed variability of DEC assessed with the described microbial
methods is affected by the quality of the assessed data. This quality is
influenced by the accuracy of the applied methods. By decreasing the
detection limit of the standard microbial methods for indicator bacteria by
a factor of 1000 or more the accuracy of assessment of DEC was increased
as demonstrated in Chapters 3 and 4.
Basic assumptions in this approach are that the assessed DEC is
independent of the source water concentration and is predictive of the
probability of breakthrough of the micro-organisms and the concentrations
in the finished water. Under normal conditions treatment is continuously
loaded with a low, but variable level of micro-organisms of faecal origin
(faecal indicators and pathogens) in the raw water. Peak contaminations
may occur infrequently, for instance after heavy rainfall. A large part of the
___________________________________________________________________
- 248 -
Chapter 10
micro-organisms that enter the treatment are initially inactivated or
retained (largely in the filter beds) but the retained organisms may be
remobilized over time. Therefore, transport of microbes will demonstrate a
distribution of retention time or retardation depending on the mechanisms
influencing their transport behaviour such as co-aggregation, and breaking
of these aggregates and adsorption/desorption. These phenomena will
occur in processes such as coagulation/floc-removal and the filtration
processes, but also in pipes and (contact) reservoirs along the treatment.
This means that the retention time of micro-organisms in water treatment
may differ very significantly from the retention time of water in the system.
The higher this retardation, the higher the probability of spatial and
temporal clustering at the end of treatment and the less there is a direct
correlation between Cin and Cout at any point in time. Moreover, in daily
practices influent and effluent sampling at the processes will not be
synchronized on the basis of the average water retention time.
The observed positive correlation between the DEC and Cin as observed in
this thesis for SSRC and sometimes for E. coli (Chapter 4 and 9) and by
others for oocysts (Smeets et al., 2007; Assavasilavasukul et al., 2008) is a
clear indication for this retardation. One possible explanation for a higher
DEC at higher Cin values is a positive correlation between Cin and the load
of suspended solids causing a higher DEC due to co-aggregation or
increased filter bed ripening. The observed positive correlation of Cin with
turbidity and of turbidity removal with DEC of treatment plants for oocysts
(LeChevallier and Norton, 1991; Dugan et al., 2001) and for SSRC (Chapter
4) are indications for the validity of this hypothesis. The other explanation
for the correlation of DEC and Cin is the aspect of retardation (accumulation
and delayed transport) as hypothesized for environmental SSRC in slow
sand filters and GAC filters (Chapter 4), for environmental centric diatoms
in slow sand filters (Chapter 8) and environmental FRNA phages, Coli44
and SSRC in soil passage (Chapter 9). Additional indications for retarded
microbial transport during water treatment are i) the lower variation in
elimination of overall treatment plants for environmental Coli44 and SSRC
(relative standard deviation 10 – 13%) compared to the variation in
elimination of unit processes (relative standard deviation 25 – 47%) in
Chapter 4 and ii) the increased spatial heterogeneity and over-dispersion of
dosed aerobic bacterial spores after a treatment train (Gale et al., 1997, 2002)
although propagation of these microbes during treatment can not be
excluded.
This over-dispersion or clustering of pathogens in drinking water by
treatment has two major implications on the QMRA process. It will cause
___________________________________________________________________
- 249 -
Chapter 10
.
an over-dispersion in the distribution of exposure to pathogens among the
public. Due to this over-dispersion the average risk at a constant mean dose
concentration is lower than the risk related to a Poisson distributed dose
(Haas, 2002; Havelaar et al., 2004). Secondly, the actual decimal elimination
assessed by paired concentrations on date as an input of stochastic
modelling (Medema et al. 2003) may over-estimate the variability of the
DEC calculated from such data and as a result both the infection risk itself
and its variability. This was confirmed by the work done by Smeets (2008)
who showed that the current method using paired Cin and Cout to calculate
elimination over-estimates the risk of infection by one order of magnitude.
Important in this respect are the observations described in Chapter 4 and
mentioned above and also described by Medema (1999), that variability of
one step in treatment is dependent on the performance of the previous
process. Thus, the variability in the DEC of a treatment train with multiple
barriers is lower than expected from the sum of the variability of the
individual processes. These observations emphasize that for a proper and
realistic description of the uncertainty of the risk of infection in drinking
water the variability in an overall treatment should be assessed rather than
summation of the variability of the individual processes.
In conclusion the proposed methodology to quantify the DEC of specific
full-scale systems by microbiological monitoring is an important input for
QMRA and microbial risk management (WSP). Interpretation of these data
with respect to the uncertainty of the assessed infection risk from the
drinking water needs comprehensive statistical tools to account for the
variability in the collected microbial data. Smeets (2008) developed and
applied statistical methods with stochastic modelling by fitting the
calculated distribution of concentrations with the observed distribution.
This reduced the over-estimation of variability in DEC caused by the use of
paired data. Besides the use of statistical methods for data interpretations,
these tools can be used in the development of a microbiological monitoring
strategy which accounts for the previous considerations on variability.
GENERIC METHODOLOGY TO ASSESS ELIMINATION
CAPACITY OF WATER TREATMENT PROCESSES FOR
PATHOGENS
On the basis of the studies presented in this thesis supplemented with data
presented in this discussion a generic concept of site specific assessment of
elimination capacity of water treatment processes is proposed and
presented in Figure 6.
___________________________________________________________________
- 250 -
Chapter 10
Full-scale IP/PI monitoring 1
Feasible
QMRA
IP not feasible
Full-scale PI monitoring 2
IP/PI ratio data base
Feasible
Small scale mimick FS
Challenge tests IP/PI 2
PI not feasible
Pilot full-scale (dummy)
Challenge tests IP/PI 3
Small scale well defined
Challenge tests IP/PI 4
WSP
PI and on line process data
5
Multi-regression process model
Literature data reviewing and analysis
Microbial Elimination Capacity (logcredit)
Process conditions 3,4
Figure 6. Generic methodology of assessment of elimination capacity of water
treatment processes (IP and PI = index pathogen and process indicator, FS = fullscale; solid line =DEC values and dotted line = indirect information for DEC
assessment)
Basic principle in the proposed methodology is that the elimination should
be determined under local specific conditions as much as possible with
simple and reliable methods which fit in the daily drinking water practices.
The solid lines in this diagram represent information on elimination of
micro-organisms, information which has to be collected regularly in
accordance with the regulations (Anonymous, 2001). The dotted lines mean
information on the IP/PI ratio which has been collected on the basis of
comparative quantitative data (treatment monitoring data or challenge test
results).
Description of the proposed generic methodology:
1. The methodology starts with direct monitoring for index pathogens (IP)
and usage of these data in QMRA and Water Safety Plans (WSP). It was
demonstrated that for some systems this was a realistic possibility
(Chapter 1), but for most it is not. By measuring index pathogens (IP)
and process indicators (PI) simultaneously, the elimination of process
indicators IP/PI ratios can be determined. On the basis of further
considerations it has to be decided to what extent pathogen monitoring
in treatment has to be continued or monitoring of faecal indicator
bacteria as process indicators (PI) can be used as an alternative (next
best method 2; cheaper and hence more intensive).
___________________________________________________________________
- 251 -
Chapter 10
.
2. From the present study it was concluded that the next best method is
monitoring elimination of E. coli and SSRC in the full-scale water
treatment facility. The historical database on faecal indicators
concentrations in water treatment shows that large volume samples are
needed for a reliable estimate of the DEC of downstream treatment
processes and its variability (Chapters 2, 3 and 4). This method allows
site specific assessment of the DEC of the PI, but requires additional
information on IP/PI ratios (Chapters 5, 6, 7, 8 and 9) to translate these
data to the index pathogens.
3. When the elimination capacity is not assessable with process indicator
monitoring, challenge tests with index pathogens and/or process
indicators in a pilot of the process operated under full-scale conditions
(dummy) is an alternative the method (Chapters 7 and 8). The
quantitative data of the index pathogen elimination can directly be used
in QMRA. When for safety reasons the dummy is only challenged with
the process indicator IP/PI ratios are required to translate the observed
elimination to the index pathogen removal. These ratios can be assessed
with small scale experiments (Chapters 5, 7 and 9) and/or obtained
from literature reviews (Chapter 6, Hijnen and Medema, 2007).
4. When both methods (2 and 3) are not applicable the use of small scale
tests under well defined conditions (collimated beam, disinfection, jartests, column tests) is an alternative method which yields only relative
quantitative data. Additional literature data assessed under more
realistic process conditions (Chapter 6, Hijnen and Medema, 2007) are
required to translate the results of these small tests to full-scale
conditions.
5. Additional methods of interest are the use of statistical modelling of
full-scale data of process conditions and elimination capacity by multiregression analysis (Wiersema, 1999; Haas et al. 2001) or the use of
process models (Schijven, 2001; Van der Wielen et al., 2006; Schijven et
al., 2008) in combination with stochastic modelling of elimination data
as demonstrated more recently by Smeets (2008). This enables a more
direct and continuous control of water safety by on line process data
and can be used as supportive information for QMRA. Example of such
on line process monitoring control of the DEC of processes based on
microbiological elimination data is the dose (fluence) control of a fullscale UV disinfection at the drinking water production plant Berenplaat
of Evides (Medema and Beerendonk, 2005). These methods are
additional because application depends on the availability of
elimination data assessed with the above described methods. The
___________________________________________________________________
- 252 -
Chapter 10
success of these methods depends largely on the quality of the available
data bases.
Choices of interest in the proposed methodology which have not been
regulated in the current legislation (Anonymous, 2001) and guideline of the
Inspectorate (Wetsteyn, 2005) must be subject of further consult with the
authorities.
IS DUTCH DRINKING WATER SAFER THAN US DRINKING
WATER?
The revised Dutch Drinking Water Decree with a health-based target for
pathogens in drinking water and the use of QMRA (Anonymous, 2001) is
different from the approach in the US where most of the scientific basis for
this approach was developed (Haas et al., 1983, 1999). US authorities do not
communicate an acceptable risk level for drinking water with the public in
their regulations. Moreover, they also have no extended water quality
assessment with faecal indicators for their water safety. In stead they
adopted an approach to regulate treatment design and operation and
assign generic log credits for removal and inactivation of Cryptosporidium,
Giardia and viruses to adequately designed and operated treatment
processes (USEPA, 2006).
The results acquired in this thesis are all on state-of-the-art treatment
systems in the Netherlands. The performance of these systems is monitored
and controlled on a routinely basis using process parameters such as
turbidity, disinfectant concentration, retention time etc. Nonetheless, the
results of the microbial monitoring with Coli44 and SSRC in large volumes
as two different process indicators in full-scale treatments show that the
variability of the DEC of the applied processes is considerable with
moments of low DEC (Chapter 4). The significance of this variability is
demonstrated in the literature on waterborne disease outbreaks that were
associated with moments of low DEC of the treatment (Hrudey et al., 2003).
Hence, large volume monitoring for faecal indicator bacteria in water
treatment yields more perception of these conditions which is relevant as
input for the mandatory QMRA in Dutch regulations and further
developments of Water Safety Plans. In the current US regulations generic
log-credits are assigned to the applied state-of-the-art processes in water
treatment which are operated according to prescribed conditions. No
further site specific microbiological monitoring is mandatory to
demonstrate the actual log-credits. On the basis of these differences in the
___________________________________________________________________
- 253 -
Chapter 10
.
regulations, the microbiological safety of the Dutch drinking waters is
locally more specified and quantified than the microbiological safety of the
US drinking waters. Therefore, the higher safety of the Dutch drinking
waters compared to the US drinking waters as posed in Chapter 1, can not
be verified on the basis of scientific evidence at the present and still
remains an assumption based on circumstantial evidence.
FURTHER RESEARCH NEEDS
Process indicator for viruses. In the work of this thesis the main
focus was on elimination of the index pathogens Cryptosporidium/Giardia
and Campylobacter. Data on virus elimination was assessed in Chapter 6 for
UV and Chapters 7 and 9 on removal of the F-specific RNA phage MS2, a
conservative model organism for virus transport in filtration processes
(Schijven, 2001). In the microbial risk assessment of the Inspectorate
(Wetsteyn et al., 2005) both somatic coliphages and FRNA phages are
mentioned as potential process indicators for virus removal in treatment.
Microbiological water quality monitoring in drinking water practices,
however, is still largely based on indicator bacteria and data on elimination
of bacteriophages in full-scale treatment plants are very limited.
Two full-scale studies indicated the value of these bacteriophages as
process indicators for virus removal by conventional treatment; the study
of Payment and Franco (1993) and the study of Havelaar et al. (1995). From
the elimination data of human enteric viruses and coliphages by three fullscale drinking water systems with large volume sampling an IP/PI ratio of
1.3±0.5 was calculated (Payment and Franco, 1993). The study of Havelaar
et al. (1995) yielded an IP/PI of 0.9±0.3 for FRNA phages. In both studies
the faceal indicator bacteria were also monitored and Payment proposed C.
perfringens as a surrogate for virus and Cryptosporidium/Giardia removal in
drinking water treatment. He observed an IP/PI ratio for viruses and C.
perfringens of 1.0±0.3 (n=3). The IP/PI ratios for enteroviruses and faecal
indicators (Coli44 and SSRC) calculated from the data presented by
Havelaar was 0.7±0.3. Similar ratios of 0.6 – 0.7 were also observed in
comparative (in)filtration studies with challenge tests using MS2 phages as
a model virus (Chapters 7 and 9, Dullemont et al., 2006; Nobel et al. 1999).
Thus, these studies indicate that faecal indicator bacteria potentially are
removed to a larger extent in filtration processes than viruses and do not
satisfy the requirement for a proper process indicator for virus removal by
filtration processes. Because MS2 is regarded as a conservative process
indicator for virus removal (Schijven, 2001) it is questionable whether this
___________________________________________________________________
- 254 -
Chapter 10
conclusion is correct, however. Using the MEC values of conventional
treatment processes for viruses, bacteria and bacterial spores (Hijnen and
Medema, 2007) virus/indicator ratios of >1.0 were calculated for most
processes (range of 0.7 – 1.6). These ratios are based on indirect comparison
and therefore less reliable, but they suggest that when lacking data on
bacteriophage elimination, data on elimination of faecal indicators can be
used as crude estimate of elimination of viruses. This hypothesis must be
verified by further comparative research on the elimination of viruses,
environmental bacteriophages and faecal indicator bacteria in full-scale
water treatment.
Further omissions in knowledge on process indicators. In the
process of QMRA it became apparent that for some processes quantitative
information on elimination of micro-organisms was still missing because of
lack of data on process indicator removal as well as a lack of data on IP/PI
ratios. Examples of such processes are Granular Activated Carbon filtration
(GAC), post-disinfection with chlorine dioxide and ultrafiltration.
- GAC filtration is applied at every surface water treatment plant and to
some decree faecal indicators Coli44 and SSRC can be used to assess
DEC, but quantitative information on IP/PI ratio is still missing.
- Post-disinfection is used at the end of a treatment train in a number of
Dutch locations and is potentially an effective process in inactivation of
susceptible pathogens (bacteria and viruses). The Coli44 concentrations
at the end of those treatments, however, are too low to assess DEC.
- Ultrafiltration is applied both as pre-treatment step and a polishing step
in water treatment. Because of the high potential of ultrafiltration as a
barrier for viruses, bacteria and parasitic protozoa (DEC of 5 log or
more) concentration of faecal indicators or other particulates in the
source water must be high to demonstrate this DEC. Applied as a pretreatment process such high DEC values can be demonstrated by
particle counting as presented for the drinking water production plant
Heemskerk PWN (Kruithof et al., 2001), but used as a polishing step in
treatment this is not possible.
GENERAL CONCLUSIONS AND RECOMMENDATIONS
On the basis of the studies described in this thesis the general conclusion is
that the elimination of E. coli and SSRC by water treatment processes under
full-scale conditions can be assessed with a reliable method with a low level
of detection which fits in the daily routine of microbial water quality
monitoring. The DEC of water treatment for both process indicators is
___________________________________________________________________
- 255 -
Chapter 10
.
determined site specific and related to the full-scale conditions as much as
possible. Application of the method showed a large variability of DEC for
these environmental micro-organisms which is potentially lower than DEC
assessed from challenge tests with pre-cultured organisms.
Challenge tests with pre-cultured organisms can be used to assess
elimination of micro-organisms in water treatment processes. Challenge
tests on pilot plant scale operated as a dummy of the full-scale process will
generate quantitative data with a high predictive value for DEC of a fullscale system. DEC determined in small scale challenge tests (batch or
columns) must be regarded as indicative for the DEC of the full-scale
process. With these tests the relative DEC of different micro-organisms
(pathogens and process indicators, IP/PI ratio) can be assessed and the
influence of process conditions on these DEC values.
Quantitative information on elimination of micro-organisms by water
treatment processes presented in literature can be used to calculate an
average DEC or default value of the Microbial Elimination Capacity of
these processes. These MEC values can be used when local specific
conditions are missing and as reference data for the DEC assessed with
challenge tests.
Spores of sulphite-reducing clostridia (SSRC; incl. C. perfringens) can be
used as a process indicator for Cryptosporidium and Giardia (oo)cyst
elimination
by
conventional
water
treatment
processes
(coagulation/filtration) and ozone but not for slow sand filtration and
surface water infiltration. The IP/PI ratios assessed for these processes
were too variable and sometimes too conservative (»1.0). From the
accumulated IP/PI ratios for Coli44 (incl. E. coli) and Campylobacter it was
concluded that these indicator bacteria are proper process indicators for
slow sand filtration, rapid granular filtration and ozonation. The IP/PI
ratios for MS2 as a model virus and both faecal indicators showed that
these were usually <1.0 indicating that viruses are removed to a lower
extent than both indicators.
The methods described in this thesis has been used in practice by Water
Companies to optimize their water treatment processes and to perform the
mandatory Quantitative Microbial Risk Assessment to estimate the annual
risk of infection with the level of uncertainty for Campylobacter,
Cryptosporidium and Giardia in drinking water.
Finally a number of issues can be identified as subjects which need to be
investigated further.
___________________________________________________________________
- 256 -
Chapter 10
-
-
-
-
-
-
The Hemoflow method with cross-flow filtration is a new method to
determine multiple micro-organisms including bacteriophages in large
water volumes. The use of this method as an alternative for the MFsampling for both faecal indicators must be explored further in
drinking water practice.
Bacteriophages are the preferred process indicators for enterovirus
elimination in water treatment. In this discussion it has been
hypothesised that E. coli and SSRC removal are indicative for virus
elimination and can be used in QMRA when data on bacteriophages are
not available. This simplification of elimination assessment, however,
needs further supportive evidence by collecting elimination data of
indicator bacteria and bacteriophages in comparative studies under
full-scale as well as under challenge test conditions.
In the revised Dutch Drinking Water Decree (Anonymous, 2001) the
statutory parameters to determine microbial safety of the drinking
water are E. coli and spores of C. perfringens. The change from
thermotolerant coliforms (Coli44) to E. coli is small since the methods
are almost identical. Increasing the specificity of the analytical methods
for the anaerobic spores SSRC to determine spores of C. perfringens,
however, may have a larger impact on the use of this process indicator
in QMRA and needs to be studied further.
Granular Activated Carbon (GAC) filtration is not primarily intended
as a microbial barrier in water treatment. The results of the current
study on faecal indicators indicated the potential of the process to
eliminate micro-organisms. Since these processes are operated at the
end of water treatment trains with potentially low process indicator
concentrations, additional challenge tests are required to assess this
potential.
A process which is not addressed in this study is post-disinfection with
low disinfectant doses. On the basis of the observations described in
this thesis on chemical disinfection it seems inappropriate to use the
same dose-effect data to verify inactivation efficiency as used for main
disinfection processes. Especially because these processes are operated
under conditions of low doses and long contact times. Challenge tests
are required to evaluate the efficacy of these processes for susceptible
micro-organisms (Campylobacter and viruses).
Predation is an elimination mechanism which is of benefit for
elimination in biological filtration processes like slow sand filtration
and GAC filtration. The ingested micro-organisms which include
waterborne pathogens, may survive in these zooplankton micro-
___________________________________________________________________
- 257 -
Chapter 10
.
organisms. Since these grazing organisms are also observed in the
filtrate of these filters, these organisms can become a vector of
waterborne pathogens. Further studies on this are proposed.
REFERENCES
Anonymous. 2001. Besluit van 9 januari 2001 tot wijziging van het
waterleidingbesluit in verband met de richtlijn betreffende de kwaliteit van voor
menselijke consumptie bestemd water, p. 1-53, vol. 31. Staatsblad van het
Koninkrijk der Nederlanden.
Assavasilavasukul, P., B. L. T. Lau, G. W. Harrington, R. F. Hoffman, and M. A.
Borchardt. 2008. Effect of pathogen concentrations on removal of Cryptosporidium
and Giardia by conventional drinking water treatment. Water Res. 42:2678 – 2690.
Badenoch, J. 1990. Cryptosporidium in water supplies, London UK.
Badenoch, J. 1995. Cryptosporidium in water supplies: second report of the group of
experts. Department of the Envrionment, Department of Health, London UK.
Broséus, R., B. Barbeau, and C. Bouchard. 2008. Verification of full-scale ozone
contactor inactivation performance using biodosimetry. J. Envrion. Engin. 4:314.
Cabaj, A., R. Sommer, and D. Schoenen. 1996. Biodosimetry: model calculations
for UV water disinfection devices with regard to dose distribution. Water Res.
30:1003-1009.
Clancy, J. L., T. M. Hargy, M. M. Marshall, and J. E. Dyksen. 1998. UV light
inactivation of Cryptosporidium oocysts. J. Am. Water Works Assoc. 90(9):92–102.
Craun, G. F. 1990. Waterborne giardiasis. In E. A. Meyer (ed.), Human Parasitic
Diseases, vol. 3, p. 267-293. Elsevier Science Publ. Amsterdam NL.
de Roda Husman, A. M. 2001. Human viruses in H2O. H2O 8:18-20.
Drost, Y. C., J. T. Groennou, W. A. M. Hijnen, and D. Van der Kooij. 1997.
Statistische methoden ter bepaling van de belasting en eliminatiecapaciteit van
zuiveringsprocessen m.b.t. micro-organismen. Kiwa Water Research SWI 97.177,
Nieuwegein NL.
Ducoste, J., K. Linden, D. Rokjer, and D. Liu. 2005. Assessment of Reduction
Equivalent Fluence Bias Using Computational Fluid Dynamics. Environ Eng Sci
22:615-628.
Dugan, N. R., K. R. Fox, J. H. Owens, and R. J. Miltner. 2001. Controlling
Cryptosporidium Oocysts Using Conventional Treatment. J. Am. Water Works
Assoc. 93(12):64-76.
Dullemont, Y. J., J. F. Schijven, W. A. M. Hijnen, M. Collin, A. Magic-Knezev,
and W. A. Oorthuizen. 2006. Presented at the Recent progress in slow sand and
alternative biofiltration, Mullheim, Germany.
Emelko, M. B. 2001. Removal of Cryptosporidium parvum by granular media
filtration. University of Waterloo, Ontario, Canada.
Evers, E. G., and J. T. Groennou. 1999. Berekening van de verwijdering van microorganismen bij de bereiding van drinkwater. RIVM 734301016, Bilthoven, NL.
___________________________________________________________________
- 258 -
Chapter 10
Fernandes, T. M. A., C. Schout, A. M. De Roda Husman, A. Eilander, H.
Vennema, and Y. T. H. P. Van Duynhoven. 2007. Gastroenteritis associated with
accidental contamination of drinking water with partially treated water.
Epidemiol. Infect. 135:818-826.
Gale, P., P. A. H. van Dijk, and G. Stanfield. 1997. Drinking water treatment
increases micro-organism clustering; the implications for microbiological risk
assessment. J Water SRT-Aqua 46:117-126.
Gale, P., R. Pitchers, and P. Gray. 2002. The effect of drinking water treatment on
the spatial heterogeneity of micro-organisms: implications for assessment of
treatment efficiency and health risk. Water Res. 36:1640-1648.
Galofre, B., S. Israel, J. Dellunde, and F. Ribas. 2004. Aerobic bacterial spores as
process indicators for protozoa cysts in water treatment plants. Wat. Sci Techn.
50:165-72.
Goyal, S. M., H. Hanssen and C.P. Gerba. 1980. Simple method for the
concentration of influenza virus from allantoic fluid on microporous filters.
Appl. Environ. Microbiol. 39:500-4.
Haas, C. N. 2002. Conditional dose–response relationships for microorganisms:
development and application. Risk analysis 22:455– 463.
Haas, C. N., K. French, G. R. Finch, and R. K. Guest. 2001. Data review on the
physical/chemical removal of Cryptosporidium. Am. Water Works Assoc.
Research Foundation, Denver CO, US.
Havelaar, A. H. 1986. F-specific RNA bacteriophages as model viruses in water
treatment processes. University of Utrecht, RIVM Bilthoven, the Netherlands.
Havelaar, A. H., M. van Olphen, and Y. C. Drost. 1993. F-specific RNA
bacteriophages are adequate model organisms for enteric viruses in fresh water.
Appl. environ. Microbiol. 59:2956-2962.
Havelaar, A. H., M. van Olphen, and J. F. Schijven. 1995. Removal and
inactivation of viruses by drinking water treatment processes under full scale
conditions. Wat. Sci. Tech. 31(5-6):55-62.
Havelaar, A. H., M. J. Nauta, and J. T. Jansen. 2004. Fine-tuning Food Safety
Objectives and risk assessment. Int. J. Food Microbiol. 93:11-29.
Hijnen, W. A. M., A. Visser, and D. Van der Kooij. 1995. Verwijdering van
Campylobacter bacteriën bij de drinkwaterbereiding op productielocatie
Scheveningen, DZH. SWO 95.272, Kiwa Water Research, Nieuwegein NL.
Hijnen, W. A. M., W. M. H. van der Speld, F. A. P. Houtepen, and D. van der
Kooij. 1997. Presented at the International Symposium on Waterborne
Cryptosporidium, Newport Beach, California US.
Hijnen, W. A. M., A. Visser, and G. J. Medema. 1998. Verwijdering van
indicatorbacterien bij de drinkwaterbereiding op productielocatie Scheveningen,
NV Duinwaterbedrijf Zuid-Holland. SWI 98.216 Kiwa Water Research
Nieuwegein NL.
Hijnen, W. A. M., T. G. J. Bosklopper, J. A. M. H. Hofmann, A. D. Bosch, and G.
J. Medema. 2001. Presented at the 10th Int. Ozone Assoc. Congres, London, UK.
___________________________________________________________________
- 259 -
Chapter 10
.
Hijnen, W. A. M., J. S. Vrouwenvelder, and S. A. Manolarakis. 2003.
Microbiologische
aspecten
van
de
levering
van
huishoudwater:
beleidsonderbouwende monitoring. KWR 02.095B. Kiwa Water Research,
Nieuwegen, the Netherlands.
Hijnen, W. A. M., A. J. Van der Veer, E. F. Beerendonk, and G. J. Medema. 2004a.
Increased resistance of environmental anaerobic spores to inactivation by UV.
Wat. Sci. Technol.: Wat. suppl. 4:55-61.
Hijnen, W. A. M., G. J. Medema, E. Baars, T. G. J. Bosklopper, A. J. van der Veer,
and R. T. Meijers. 2004b. Influence of DOC on the Inactivation Efficiency of
Ozonation Assessed with Clostridium perfringens and a Lab-Scale Continuous
Flow System. Ozone Sci. Eng. 26:465-473.
Hijnen, W. A. M. 2006. Verslag van de BTO workshop “Langzame zandfiltratie als
veiligheidsbarrière” 25 oktober 2006. BTO2007.002, Kiwa Water Research,
Nieuwegein NL.
Hijnen, W. A. M., and G. J. Medema. 2007. Elimination of micro-organisms by
drinking water treatment processes: a review BTO 2003.13-3 Third edition. Kiwa
Water Research, Nieuwegein, the Netherlands.
Hill, V. R., A. M. Kahler, N. Jothikumar, T. B. Johnson, D. Hahn, and T. L.
Cromeans. 2007. Multistate Evaluation of an Ultrafiltration-Based Procedure for
Simultaneous Recovery of Enteric Microbes in 100-Liter Tap Water Samples.
Appl. Environ. Microbiol. 73:4218-4225.
Hill, V. R., A. L. Polaczyk, D. Hahn, J. Narayanan, T. L. Cromeans, J. M. Roberts,
and J. E. Amburgey. 2005. Development of a Rapid Method for Simultaneous
Recovery of Diverse Microbes in Drinking Water by Ultrafiltration with Sodium
Polyphosphate and Surfactants. Appl. Environ. Microbiol. 71:6878 - 6884.
Hoogenboezen, W., G. J. Medema, H. A. M. Ketelaars, G. B. J. Rijs, and J. F.
Schijven. 2001a. Cryptosporidium and Giardia: occurrence in sewage, manure and
surface water. Amsterdam: RIWA report.
Hoogenboezem, W., F. Donker, E. Gijsbers, and S. Strating. 2001b. Beëindiging
veiligheidsdesinfectie op pompstation Mensink? H2O 34:21-24.
Huck, P. M., M. B. Emelko, B. M. Coffey, D. D. Maurizo, and C. R. O'Melia. 2001.
Filter operation effects on pathogen passage. American Water Works Assoc.
Reserach Foundation, Denver CO US.
Ketelaars, H. A. M., G. J. Medema, L. C. W. A. Van Breemen, D. van der Kooij, P.
J. Nobel, and P. Nuhn. 1995. Occurrence of Cryptosporidium oocysts and Giardia
cysts in the river Meuse and removal in the Biesbosch reservoirs. J Water SRTAqua 44(1):108-111.
Kruithof, J. C., R. C. Van der Leer, and W. A. M. Hijnen. 1992. Practical
experiences with UV disinfection in The Netherlands. J. Water SRT-Aqua 41:8894.
Kruithof, J.C., P. C. Kamp, H. C. Folmer, M. M. Nederlhof, and S. C. J. M. van
Hoof. 2001. Development of a membrane integrity monitoring strategy for the
UF/RO Heemskerk drinking water treatment plant. Wat Supply 1(5-6):261-271.
___________________________________________________________________
- 260 -
Chapter 10
Lechavallier, M. W., W. D. Norton, and R. G. Lee. 1991. Giardia and
Cryptosporidium spp. in filtered drinking water supplies. Appl. Environ.
Microbiol. 57:2617-2621.
LeChevallier, M. W., and K. Au. 2004. Water treatment and pathogen control:
process efficiency in achieving safe drinking water. World Health Organisation,
IWA Publishing, London, UK.
LeChevallier, M. W., and W. D. Norton. 1992. Examining relationships between
particle counts, and Giardia and Cryptosporidium and turbidity. J. Am. Water
Works Assoc. 84(12):54-60.
Leclerc, F., D. A. Mossel, P. A. Trinel, and F. Gavini. 1977. Microbiological
monitoring: a new test of fecal contamination. In A. W. Hoadley and B. J. Dutka
(ed.), Bacterial indicators: health hazards associated with water. American
society for testing and materials, Philadelphia, US.
Li, J., Zhang, J., J. Miao, J. Ma, and W. Dong. 2006. Application of computational
fluid dynamics (CFD) to ozone contactor optimization. Wat Sci Technol: Wat
Supply 6(4):9-16.
Logsdon, G. S., J. M. Symons, R. L. Hoye, and M. M. Arozarena. 1981. Alternative
filtration methods for removal of Giardia cysts and cyst models. J. Am. Water
Works Assoc. 2:111.
Mazoua, S., and E. Chauveheid. 2005. Aerobic spore-forming bacteria for
assessing quality of drinking water produced from surface water. Water Res.
39:5186-98.
Medema, G. J., and F. M. Schets. 1994. Campylobacter en Salmonella in open
reservoirs voor de drinkwaterbereiding. RIVM 149103002, Bilthoven NL.
Medema, G. J. 1999. Cryptosporidium and Giardia: new challenges to the water
industry. University of Utrecht, Utrecht, NL.
Medema, G. J., and P. J. Stuyfzand. 2002. Presented at the 4th international
symposium on artificial recharge, Adelaide, Australia, September 22-26.
Medema, G. J., W. Hoogenboezem, A. J. van der Veer, H. A. M. Ketelaars, W. A.
M. Hijnen, and P. J. Nobel. 2003. Quantitative risk assessment of
Cryptosporidium in surface water treatment. Wat. Sci. Tech. 47(3):241-247.
Medema, G.J. and E. F. Beerendonk. (2005) Testing UV-fluence monitoring of fullscale UV at Berenplaat: Kiwa Water Reserach, KWR 05.028, Nieuwegein, NL.
Medema, G.J., J.F. Loret, T. Stenström, and N.J. Ashbolt. 2006. Quantitative
Microbial Risk Assessemnt in Water Safety Plans. Report for the European
Community, EC. Kiwa Water Research, Nieuwegein NL; Suez EnvironmentalCIRSEE, LePecq, FR; Swedish Institute for Infectious Disease Control, Stockholm,
SW; University of New South Wales, Sydney, AU.
Morales-Morales, H. A., G. O. Vidal, J., C. M. Rock, D. Dasgupta, K. H. Oshima,
and G. B. Smith. 2003. Optimization of a Reusable Hollow-Fiber Ultrafilter for
Simultaneous Concentration of Enteric Bacteria, Protozoa, and Viruses from
Water. Appl. Environ. Microbiol. 69:4098-4102.
___________________________________________________________________
- 261 -
Chapter 10
.
Nieminski, E. C., and J. E. Ongerth. 1995. Removing Giardia and Cryptosporidium
by conventional treatment and direct filtration. J. Am. Water Works Assoc.
87(9):96-106.
Nobel, P. J. e. a. 1999. Effecten van begrazing op waterkwaliteit. KOA 98.204, Kiwa
Water Research, Nieuwegein NL.
Oppenheimer, J. A., E. M. Aieta, R. R. Trussell, J. G. Jacangelo, and I. N. Najm.
2000. Evaluation of Cryptosporidium inactivation in natural waters. American
Water Works Assoc. Research Found. Denver CO US.
Patania, N. L., J. G. Jacangelo, L. Cummings, A. Wilczak, K. Riley, and J.
Oppenheimer. 1995. Optimization of Filtration for Cyst Removal. American
Water Works Assoc. Research Foundation, Denver CO US.
Payment, P. 1991. Fate of human enteric viruses. coliphages, and Clostridium
perfringens during drinking-water treatment. Can. J. Microbiol. 37:154-157.
Payment, P., A. Bérubé, D. Perrefault, R. Armon, and M. Trudel. 1989.
Concentration of Giardia lamblia cysts, Legionella pneumophila, Clostridium
perfringens, human enteric viruses, and coliphages from large volumes of
drinking water, using a single filtration. Can. J. Microbiol. 35:932-935.
Payment, P., and E. Franco. 1993. Clostridium perfringens and somatic coliphages as
indicators of the efficiency of drinking water treatment for viruses and protozoan
cysts. Appl. Environ. Microbiol. 59:2418-24.
Prévost, M., K. Jaidi, P. Payment, and B. Barbeau. 2007. Presented at the Water
Quality and Technology Conference, Charlotte, US.
Qualls, R. G., and J. D. Johnson. 1983. Bioassay and dose measurement in UV
disinfection. Appl. Environ. Microbiol. 45:872–877.
Regli, S., J. B. Rose, C. N. Haas, and C. P. Gerba. 1991. Modeling the risk from
Giardia and viruses in drinking water. J. Am. Water Works Assoc. 83:76-84.
Richardson, A. J., R. A. Frankenberg, A. C. Buck, J. B. Selkon, J. S. Colbourne, J.
W. Parsons, and R. T. Mayon-White. 1991. An outbreak of waterborne
cryptosporidiosis in Swindon and Oxfordshire. Epidemiol. Infect. 107:485-95.
Schijven, J. F. 2001. Virus removal from groundwater by soil passage. Technische
Universiteit Delft, Delft, the Netherlands.
Schijven, J. F., M. Colin, Y.J. Dullemont, W.A.M. Hijnen, A. Magic-Knezev, W.
Oorthuizen. 2007. Removal of micro-organisms by slow sand filtration: RIVM
703719019/2007 (in Dutch).
Schijven, J. F., M. Colin, Y. J. Dullemont, W. A. M. Hijnen, A. Magic-Knezev, W.
Oorthuizen, G. Wubbels, B. J., S. A. Rutjes, and A. M. de Roda Husman. 2008.
Effect filtratiesnelheid, temperatuur en korrelgrootte op de verwijdering van
micro-organismen door langzame zandfiltratie. RIVM 330204002/2008,
Bilthoven (in Dutch).
Simmons, O. D., M. D. Sobsey, C. D. Heaney, F. W. Schaefer, and D. S. Francy.
2001. Concentration and detection of Cryptosporidium oocysts in surface water
samples by method 1622 using Ultrafiltration and capsule filtration. Appl.
Environ. Microbiol. 67:1123-1127.
___________________________________________________________________
- 262 -
Chapter 10
Smeets, P. W. M. H., Y. J. Dullemont, and G. J. Medema. 2005. Presented at the
17th IOA conference, Strasbourg, France.
Smeets, P. M. 2008. Stochastic modelling of drinking water treatment in
quantitative microbial risk assessment. Technical University Delft, Delft NL.
Smeets, P. W. M. H., A. W. C. Van der Helm, Y. J. Dullemont, L. C. Rietveld, J. C.
Van Dijk, and G. J. Medema. 2006. Inactivation of Escherichia coli by ozone
under bench-scale plug flow and full-scale hydraulic conditions. Water Res.
40:3239-3248.
Smeets, P. W. M. H., J. C. Van Dijk, G. Stanfield, L. C. Rietveld, and G. J.
Medema. 2007. How can the UK statutory Cryptosporidium monitoring be used
for Quantitative Risk Assessment of Cryptosporidium in drinking water? J. Wat. &
Health 05:107-118.
Sommer, R., A. Cabaj, T. Sandu, and M. Lhotsky. 1999. Measurement of UV
radiation using suspensions of micro-organisms. J. Photochem. Photobiol. B:Biol.
53:1-6.
Teunis, P. F. M., G. J. Medema, L. Kruidenier, and A. H. Havelaar. 1997.
Assessment of the risk of infection by Cryptosporidium and Giarda in drinking
water from a surface water source. Water Res. 31:1333-1346.
USEPA. 2006. LT2ESWTR Long Term Second Enhanced Surface Water Treatment
Rule; Final Rule USEPA.
Van der Veer, A. J., R. M. Schotsman, B. Martijn, R. Schurer, and J. C. van Dijk.
2005. Presented at the 17th IOA Conference, Strasbourg Fr.
Van der Wielen, P. W. J. J., M. Blokker, and G. J. Medema. 2006. Modelling the
length of microbiological protection zones around phreatic sandy aquifers in The
Netherlands. Wat. Sci. Technol. 54:63-69.
Van Olphen, M., E. van der Baan, and A. H. Havelaar. 1993. Virus removal at
river bank filtration. H2O (in Dutch) 26:63-66.
Veenendaal, H. R., and A. J. Brouwer-Hanzens. 2007. A method for the
concentration of microbes in large volumes of water. Kiwa Water Research,
Nieuwegein NL.
von Huben, H. 1991. Surface Water: the new rules. American Water Works
Association, Denver CO US.
West, T. P., D. P. Meyerhofer, A. DeGraca, S. Leonard, and C. P. Gerba. 1994.
Presented at the Annual Conf. American Water Works Assoc., Denver CO US.
Wetsteyn, F. J. 2005. Analyse microbiologische veiligheid drinkwater., Artikelcode:
5318. VROM-Inspectorate, Haarlem NL.
WHO. 2004. Guidelines for Drinking Water Quality, third edition. World Health
Organization, Geneva, SW.
Wiersema, I. 1999. Influence of process factors on the removal of indicator bacteria
by coagulation processes during drinking water production. Doctoral report 9735. Wageningen university, WUR, NL.
___________________________________________________________________
- 263 -
Summary
Summary
RATIONALE
Ever since the recognition of the important role of water in the transmission
of pathogenic micro-organisms in the 19th century, microbiological safety of
drinking water has been a major research issue for microbiologists in
drinking water industry. In central water supply the safety of drinking
water is controlled by means of treatment and water quality assessment.
Chapter 1 describes the major historical developments in microbiological
water quality assessment and regulations related to the transmission of
waterborne pathogens of faecal origin. Standards for bacteria of faecal
origin such as Escherichia coli and coliforms in drinking water as indicators
for the presence of these pathogens have been implemented in legislations
world wide. Additional drinking water standards for other indicators of
faecal pollution such as enterococci and spores of sulphite-reducing
clostridia (SSRC) have been implemented in regulations of selected parts of
a number of countries in the world among which also the Netherlands and
other European countries.
Recent epidemiological studies and outbreak reports in the United States
(US) and the United Kingdom (UK) have demonstrated the importance of
persistent pathogens such as Cryptosporidium and Giardia and some
enteroviruses and of the pathogenic bacteria Campylobacter and E. coli O157
for drinking water safety. Inadequacies in water treatment in combination
with the high virulence of these pathogens assessed in dose-response
studies and the failure of faecal indicators to detect them in drinking water
were important causes of the growing concern for these waterborne
pathogens.
A major scientific breakthrough was the introduction of the estimation of
the probability of infection given a certain dose of a selected pathogen. This
enabled to introduce quantitative health-related targets in the microbial
risk management of drinking water which can be verified by means of
Quantitative Microbial Risk Assessment (QMRA).
Intrinsically the safety of Dutch drinking water was assumed to be higher
than the drinking water in the US and the UK because of comprehensive
water treatment with multiple barriers and standards with persistent
(Clostridium spores). The few reported waterborne outbreaks in the
Netherlands were related to cross connections with sewage and not to
___________________________________________________________________
- 265 -
Summary
.
inadequate drinking water quality. Though this assumption is regarded as
plausible, in their aspirations to distribute a high quality drinking water
Dutch Water Companies started an investigation in the joint research
program to verify this. Simultaneously Dutch Authorities anticipated on
these developments by implementing an accepted infection risk level of 1
person per 10.000 consumers per year (10-4) for drinking water in the
revised Drinking Water Decree. This health-related target should be
verified by means of QMRA. Important step in this QMRA is the exposure
assessment by monitoring pathogens in the source water and
determination of the elimination (inactivation and removal) of these
pathogens in water treatment.
OBJECTIVES
Main objective of the current study was to develop a general strategy to
assess elimination capacity of water treatment processes for pathogens. For
this purpose the following aspects were investigated:
- the potential use of faecal indicators Coli44 (incl. E. coli) and SSRC (incl.
C. perfringens) as process indicators to assess pathogen elimination in
full-scale water treatment plants.
- the value of comparative challenge tests with pre-cultured organisms
for the assessment of elimination capacity of full-scale processes, to
study the effect of process conditions and to validate the use of process
indicators.
- the use of literature data to assess elimination capacity of water
treatment processes for pathogens and the effect of process conditions
on this.
INDICATOR
BACTERIA
TO
ASSESS
CAPACITY OF FULL-SCALE TREATMENT
ELIMINATION
Research after the use of faecal indicators as process indicators to
determine removal of pathogens in full-scale water treatment is described
in the Chapters 2, 3 and 4. Drinking water Companies have generated a
large amount of data on microbiological quality of the source water and the
drinking water and the water after selected steps in treatment. These data
proofed to be useful in assessment of the decimal elimination capacity
(DEC; logremoval) of the first processes of water treatment inclusive the
variation of this capacity, but the data were unsuited to do the same for the
overall treatment (Chapter 2). To achieve the latter a method had to be
___________________________________________________________________
- 266 -
Summary
developed with which these indicator bacteria could be detected in the
drinking water. For this purpose larger water volumes were examined
either with the standard membrane filtration method used to determine the
concentration of indicator bacteria in water or with a special developed
device, the MF-sampler. With the MF sampler the limit of detection of the
standard method was lowered by a factor of 1000 or more (Chapter 3). The
large volume sampling method (LVS) method yielded reproducible results
with a high recovery of 75-100% assessed with dosage experiments and
could easily be implemented in the daily routine of the drinking water
practice. With this LVS method the concentration of Coli44 and SSRC in the
water during water treatment of ten facilities was measured in a
collaborative investigation with the drinking water Companies (Chapter 4).
Both indicators were detected in the finished waters with SSRC in higher
concentrations than Coli44. The study yielded information on DEC of
individual processes and overall treatment facilities for both indicators and
its variability. A variation in elimination was observed for individual
processes over time and for similar processes at different locations.
Moreover, the results clearly indicated that elimination of environmental
micro-organisms under full-scale conditions is lower than expected from
studies presented in literature applying challenge tests with pre-cultured
micro-organisms. These results emphasize the need to assess DEC of fullscale processes as local specific as possible preferably with environmental
micro-organisms.
LIMITATIONS OF THE INDICATOR BACTERIA METHOD
The indicator bacteria method with LVS appeared to have a number of
limitations. First of all the attention is mainly focused on the elimination of
bacteria and protozoa and no information is collected on the elimination
capacity of treatment for viruses. At a number of processes (slow sand
filtration, surface water infiltration and post disinfection) the
concentrations of indicator bacteria were too low to enable DEC assessment
or some processes (UV disinfection) were not applied during the course of
the study. Furthermore, the method is not suited to control DEC of
processes in the daily drinking water routine. For this purpose knowledge
is required on the process conditions directly affecting the DEC. Finally, the
predictive value of Coli44 and SSRC for the elimination of Campylobacter
bacteria and both protozoa, Cryptosporidium and Giardia had to be
elucidated.
___________________________________________________________________
- 267 -
Summary
.
Because of these limitations and additional questions alternative strategies
were applied to determine the DEC of processes. Challenge tests were
performed with set ups at different scales (pilot plant and laboratory) and
different pre-cultured micro-organisms, among which bacteriophages as a
process indicator for virus removal.
ELIMINATION BY DISINFECTION
In the framework of the study two comparative studies have been
conducted on the efficacy of disinfection processes. In Chapter 5 the
susceptibility of spores of Clostridium perfringens and oocysts of
Cryptosporidium parvum has been compared for ozonation in a continuous
flow set up as model for a full-scale ozonation installation. The study
showed that SSRC can be applied as process indicator for inactivation of
Cryptosporidium by full-scale ozonation. The susceptibility of both
organisms was in the same order of magnitude with an index
pathogen/process indicator (IP/PI) ratio of 0.8. The inactivation data
assessed with the applied continuous flow laboratory set up were
predictive for inactivation of SSRC under full-scale ozone conditions. The
inactivation in the gas feed chamber was significant and proportional to the
calculated Ct-values (mg/l.min). When Ct of the process is controlled with
the ozone concentration the dose-effect relation is in agreement with the
Chick-Watson inactivation kinetics. Setting Ct in the range of 0.5 – 1.5
(normal for the Dutch situation) with flow and contact time at a constant
low ozone dose showed to be ineffective to control the inactivation capacity
of the process.
The susceptibility to Ultraviolet radiation (UV) of micro-organisms relevant
for drinking water safety has been assessed and compared in Chapter 6 on
the basis of literature data. Challenge test studies with a laboratory set up
(collimated beam) and continuous flow systems as well as studies with
environmental micro-organisms organisms in full-scale UV systems were
used. This evaluation showed a higher inactivation of pre-cultured
organisms in the collimated beam set up compared to the inactivation of
environmental organisms of the same species in continuous flow systems.
This discrepancy is caused by factors related to the micro-organisms and to
the process conditions. These factors were evaluated on the basis of the
available literature. A required UV fluence table for relevant microorganisms was described in which the required fluence for bacteria and
bacterial spores was corrected for this discrepancy between laboratory data
___________________________________________________________________
- 268 -
Summary
and field observations. For viruses such a correction seemed to be not
necessary and due to a lack of data such correction was not possible for the
protozoa.
ELIMINATION BY FILTRATION
Filtration processes applied in drinking water production focused on the
elimination of micro-organisms are slow sand filtration and soil passage. In
Chapter 7 an investigation is presented on the elimination capacity of slow
sand filtration assessed under different conditions. The results
demonstrated that DEC for Coli44 determined under full-scale conditions
was 2-3 log and in the same order of magnitude as DEC determined for E.
coli in pilot plant filters operated in a pilot plant as dummy of the same
treatment facility. DEC assessed in a column test was clearly lower,
possibly as a result of the lack of a schmutzdecke in the column and the
difference in hydraulic conditions of the filter beds. In a filter with a
schmutzdecke the DEC for E. coli was 1-2 log higher than in the same filter
without this layer. The effect of the schmutzdecke on the removal of the
bacteriophage MS2 as a conservative model for viruses was marginal.
These dosed organisms were removed with 1.5-2.0 log by these filters.
Coli44 appeared to be a safe process indicator for the elimination of
Campylobacter bacteria.
Removal of environmental SSRC by full-scale slow sand filtration ranged
from -0.2 – 1.8. In a column test with short columns (0.4 m) removal of
dosed spores of C. perfringens was higher (2.3 – 3.2 log). Oocysts of
Cryptosporidium were removed in these columns even more (>5.3 - >6.5
log). These data implicate that SSRC is not a proper process indicator for
Cryptosporidium elimination by slow sand filtration. Hypothesis for the
variable and sometimes low elimination of Clostridium spores by these
filters was accumulation, survival and delayed transport. Further research
after this phenomenon was done in a pilot plant filter as dummy of the fullscale filters (Chapter 8). Elimination of spores of C. perfringens, UV
inactivated oocysts of Cryptosporidium parvum and centric diatoms
(Stephanodiscus hantzschii) was determined with a prolonged challenge test
with a filter operated three years in advance. Centric diatoms are persistent
organisms with a size similar to the size of oocysts which were present in
the influent in considerable higher concentrations than environmental
SSRC. The DEC calculated from the experiment for oocysts, spores and
diatoms was 4.7, 3.6 and 1.8 log, respectively. The duration of the test of
three months was too short to demonstrate the occurrence of accumulation
___________________________________________________________________
- 269 -
Summary
.
and delayed transport for spores and oocysts. On the basis of the low DEC
for centric diatoms naturally present in the influent in sometimes high
concentrations and the spatial distribution of spores of C. perfringens in the
filter bed determined at the end of the dosage experiment it was concluded
that this process is valid for both persistent organisms. However, most
likely not for persistent organisms which are sensitive to predation such as
oocysts. Mass balance calculations revealed a large loss of oocysts numbers
in the filter bed. In the zooplankton population observed in the sand
species were identified capable of ingesting oocysts. Because the notified
predators were observed in the filter bed and in the filtrate, delayed
transport and breakthrough of viable oocysts in predators as a vector was
suggested as an issue for further research.
The elimination capacity of soil passage under field conditions is usually
not simple to assess. In Chapter 9 a challenge test is described with
columns filled with soil material from two surface water infiltration sites in
the Netherlands loaded with the local surface water spiked with MS2
bacteriophages, E. coli and spores of C. perfringens and oocysts of
Cryptosporidium and Giardia. From both sites quantitative information was
available on the elimination of dosed MS2 phages and indigenous
bacteriophages and indicator bacteria during soil passage. This experiment
with different organisms and two soil types demonstrated that besides size
of micro-organisms and soil grains, surface properties of organism and soil
and soil characteristics such as uniformity and conductivity affect the
transport of micro-organisms. The delayed and second breakthrough peaks
of oocysts and spores in the same soil columns and the low DEC of
environmental Coli44 and SSRC, emphasize the importance of desorption
in transport of micro-organisms in soil. From the results it was concluded
that both straining and adsorption/desorption are mechanisms which,
dependent on the conditions in different ratio contribute to the elimination
of micro-organisms in soil passage. The IP/PI ratio for Clostridium spores
and protozoan (oo)cysts assessed from this experiment was variable and
higher or lower than 1.0 meaning that SSRC is not a proper process
indicator for the removal of these index pathogens by soil passage.
Moreover, it was concluded that these column tests are not applicable for
the assessment of the elimination capacity of soil passage under field
conditions.
___________________________________________________________________
- 270 -
Summary
GENERIC METHODOLOGY FOR ELIMINATION CAPACITY
ASSESSMENT
On the basis of the discussion in Chapter 10 a generic methodology was
described to assess the elimination capacity of local water treatment
processes for index pathogenic micro-organisms. This information is the
input for the mandatory quantitative microbial risk assessment which has
to be performed for drinking water production locations using surface
water or vulnerable groundwater sites as the source water.
The most optimal methods are the assessment of elimination of indigenous
index pathogens and indicator bacteria (Coli44 and SSRC) in treatment by
large volume sampling. Because for index pathogens this is not feasible for
most locations and certainly not for the total treatment, Coli44 is considered
as an environmental process indicator for the elimination of Campylobacter
by most processes and SSRC as a process indicator for Cryptosporidium and
Giardia for coagulation/floc-removal, rapid granular filtration and
ozonization. Further research is required to determine the necessity and
possibility to add monitoring of bacteriophages as a specific process
indicator for virus elimination. Because for some processes monitoring for
elimination of indicator bacteria is not possible due to low influent
concentrations challenge tests with pre-cultured micro-organisms are
necessary to collect additional quantitative data. Quantitative data on
elimination of these pre-cultured in challenge tests can be applied for DEC
of full-scale processes when the tests were performed in a pilot plant
operated as a dummy of the full-scale plant. Data from challenge tests on
laboratory scale with multiple organisms (batch inactivation tests or
filtration columns tests) must be regarded as indicative for DEC under fullscale. The value of these tests lies in the assessment of the index
pathogen/process indicator ratio (IP/PI) and the elucidation of specific
conditions on the elimination process. When quantitative data on
elimination of indicator bacteria in the full-scale treatment and from
additional challenge tests are not available, literature is an important
potential source of quantitative information of elimination of microorganisms in water treatment. Additionally these data can be used by the
translation of challenge tests results to full-scale conditions.
___________________________________________________________________
- 271 -
Summary
.
GENERAL CONCLUSION AND RECOMMENDATIONS FOR
FURTHER RESEARCH
A methodology has been developed to assess the Decimal Elimination
Capacity (DEC) of full-scale water treatment processes for waterborne
pathogenic micro-organisms. The methodology is tailored to the natural
conditions of the water treatment and to the daily monitoring practice of
microbiological water quality as much as possible.
An issue of further research is the determination of the DEC of full-scale
treatment for viruses using environmental bacteriophages as potential
process indicators. In relation to this the application of ultrafiltration in the
recently developed hemoflow method has to be evaluated. With this
method multiple micro-organisms (i.e. bacteriophages and faecal indicator
bacteria) are determined in large volumes of water. For a number of water
treatment processes additional knowledge on the elimination capacity and
the IP/PI ratios has to be collected. These processes are granular activated
carbon filtration, postdisinfection with chlorine dioxide and ultrafiltration.
The effect of the intended change in legislative water quality monitoring
from spores of sulphite-reducing clostridia to specific C. perfringens
inclusive vegetative cells, an analysis without pasteurization, on the use of
this parameter as a process indicator is an issue of future attention.
Predation in biological filters is potentially a beneficial elimination
mechanism in water treatment for water safety provided that the predators
are no vectors of transport for the ingested pathogens. This is an issue of
importance for further research.
___________________________________________________________________
- 272 -
Samenvatting
Samenvatting
AANLEIDING
Sinds de erkenning van de belangrijke rol die water speelt in de overdracht
van ziekteverwekkende micro-organismen in de negentiende eeuw is de
microbiologische veiligheid van drinkwater een belangrijk onderwerp voor
microbiologische onderzoekers en de drinkwaterbedrijven. In de centrale
drinkwatervoorziening wordt de microbiologische veiligheid gecontroleerd
door waterzuiveringsprocessen en de bepaling van de microbiologische
waterkwaliteit. In Hoofdstuk 1 worden de belangrijkste historische
ontwikkelingen op het gebied van de microbiologische waterkwaliteit en
de regelgeving met betrekking tot wateroverdraagbare ziekteverwekkers
van fecale herkomst beschreven. Normen voor bacteriën van fecale
herkomst zoals Escherichia coli en coliformen in drinkwater als indicator
voor de aanwezigheid van deze ziekteverwekkers zijn wereldwijd
opgenomen in de wetgeving. Aanvullende normen voor andere
indicatoren voor fecale verontreiniging zoals enterococcen en sporen van
sulfiet-reducerende clostridia (SSRC) zijn opgenomen in de regelgeving
van een aantal Europese landen waaronder ook Nederland.
Recente epidemiologische studies en rapporten over uitbraken in de
Verenigde Staten (US) en het Verenigd Koningrijk (UK) hebben de
belangrijke betekenis van persistente ziekteverwekkers (pathogenen) zoals
Cryptosporidium and Giardia en sommige virussen als ook de bacteriën
Campylobacter en E. coli O157 voor de veiligheid van drinkwater
aangetoond. Onvolkomenheden in de waterzuivering in combinatie met de
hoge virulentie van deze pathogenen bepaald in dosis-respons studies en
het falen van de fecale indicatoren om de aanwezigheid van deze
pathogenen in drinkwater aan te tonen, waren belangrijke oorzaken voor
de groeiende zorg over deze wateroverdraagbare pathogenen.
Een belangrijke wetenschappelijke doorbraak was de introductie van de
schatting van de kans op infectie, gegeven een bepaalde dosis van een
pathogeen micro-organisme. Dit maakte het mogelijk in het management
van de microbiologische kwaliteit van drinkwater, een kwantitatief aan de
gezondheid gerelateerd doel te introduceren dat met Quantitative
Microbial Risk Assessment (QMRA) kan worden geverifieerd.
De veiligheid van het Nederlandse drinkwater werd als veiliger ingeschat
dan het drinkwater in de US en UK vanwege de uitgebreidere
___________________________________________________________________
- 273 -
Samenvatting
.
waterzuivering met meerdere barrières en normen voor persistente
indicator bacteriën (Clostridium sporen). Bovendien waren het lage aantal
gerapporteerde drinkwateruitbraken in Nederland gerelateerd aan
kruisverbindingen met afvalwater en niet aan slecht drinkwater. Hoewel
deze aanname als plausibel werd beschouwd, startten de Nederlandse
drinkwater
bedrijven
een
onderzoek
in
het
gezamenlijke
onderzoeksprogramma (BTO) om dit te verifiëren. Dit past in hun streven
om een hoge kwaliteit drinkwater te distribueren. Gelijktijdig anticipeerde
de Nederlandse Overheid op deze ontwikkelingen door in de aanpassing
van
het
Nederlandse
Waterleidingbesluit
een
geaccepteerd
infectierisiconiveau van 1 persoon per 10.000 consumenten per jaar (10-4) op
te nemen voor drinkwater dat wordt geproduceerd uit oppervlaktewater of
uit grondwater afkomstig uit een kwetsbare waterwinning. Een belangrijke
stap in de kwantitatieve microbiologische risicoanalyse (QMRA)waarmee
dit doel moet worden geverifieerd, is de bepaling van de blootstelling aan
pathogenen via drinkwater. Dit wordt bepaald door in het ruwe water van
een drinkwaterpompstation de concentratie van deze pathogenen te meten
en vast te stellen in welke mate deze worden geëlimineerd (inactivatie en
verwijdering) in de waterzuivering.
DOELEN
Het belangrijkste doel van de gepresenteerde studie was om een algemene
strategie te ontwikkelingen waarmee de eliminatiecapaciteit van
waterzuiveringsprocessen voor pathogenen kan worden bepaald. Voor dit
doel werden de volgende aspecten onderzocht:
- evaluatie van de mogelijkheid om de fecale indicator, thermotolerante
bacteriën van de coligroep (Coli44 inclusief E. coli) en sporen van sulfiet
reducerende clostridia (SSRC inclusief C. perfringens) te gebruiken als
procesindicator
om
de
eliminatie
van
pathogenen
door
waterzuiveringsprocessen in de praktijk te bepalen.
- de waarde van vergelijkende doseerproeven met voorgekweekte microorganismen om de eliminatiecapaciteit van praktijkprocessen te
bepalen, de invloed van de procescondities te bestuderen en het
gebruik van procesindicatoren te valideren.
- het gebruik van literatuurgegevens om de eliminatiecapaciteit van
zuiveringsprocessen en de invloed van de procescondities hierop te
bepalen.
___________________________________________________________________
- 274 -
Samenvatting
INDICATORBACTERIËN OM DE ELIMINATIECAPACITEIT
ONDER PRAKTIJKCONDITIES TE BEPALEN
Het onderzoek naar het gebruik van fecale indicatoren als
procesindicatoren om de verwijdering van pathogenen door de
waterzuivering in de praktijk te bepalen, wordt beschreven in de
Hoofdstukken 2, 3 en 4. Drinkwaterbedrijven hebben een groot aantal
gegevens verzameld over de microbiologische kwaliteit van het ruwe
water, het drinkwater en het water op een aantal geselecteerde locaties in
de zuivering. Deze gegevens waren geschikt om de decimale
eliminatiecapaciteit (DEC; logremoval) van de eerste processen in de
zuivering te bepalen inclusief de variatie van deze capaciteit, maar niet om
dit te doen voor de gehele zuivering (Hoofdstuk 2). Om dit laatste te
bereiken, moest een methode worden ontwikkeld om deze indicator
bacteriën tot in het drinkwater te kunnen aantonen. Hiervoor werden grote
volume monsters onderzocht met de standaard toegepaste membraan
filtratie methode of met een speciaal ontwikkelde opstelling, de MFsampler. Met de MF-sampler werd de detectielimiet van de standaard
methode verlaagd met een factor 1000 of meer (Hoofdstuk 3). Deze LVS
methode leverde reproduceerbare resultaten op met een hoge
opbrengstfactor van 75-100% die was bepaald met doseerproeven. De
methode kon eenvoudig in de dagelijkse drinkwater routine worden
geïmplementeerd. Met deze LVS methode werd de concentratie Coli44 en
SSRC in het water tijdens de drinkwaterzuivering van tien locaties gemeten
in een gezamenlijk onderzoek met de drinkwaterbedrijven (Hoofdstuk 4).
Beide indicatoren werden gedetecteerd in de eindproducten van deze tien
zuiveringen waarbij SSRC in hogere concentraties werden gevonden dan
Coli44. De studie leverde informatie op over de DEC van individuele
processen en van de totale zuivering voor beide indicatoren en de variatie
hierin. Een variatie in eliminatie door individuele processen in de tijd en
door dezelfde processen op verschillende locaties werd waargenomen.
Bovendien toonden de resultaten duidelijke indicaties dat de eliminatie van
natuurlijke micro-organismen onder praktijkcondities lager is dan
verwacht op grond van studies gepresenteerd in de literatuur waarbij
doseerproeven met voorgekweekte micro-organisms werden gebruikt.
Deze resultaten benadrukken de behoefte om DEC van praktijkprocessen
zoveel als locatiespecifiek te bepalen en bij voorkeur met natuurlijke microorganismen.
___________________________________________________________________
- 275 -
Samenvatting
.
BEPERKINGEN VAN DE METHODE MET DE INDICATORBACTERIËN
De methode bleek echter ook een aantal beperkingen te hebben. Zo is de
aandacht voornamelijk gericht op de eliminatie van bacteriën en protozoën
en wordt geen informatie verzameld over de eliminatiecapaciteit van de
zuiveringen voor virussen. Voor een aantal processen bleken de
concentraties indicatorbacteriën te laag om de DEC te kunnen bepalen of
sommige processen (UV desinfectie) werden nog niet op praktijkschaal
toegepast. De methode is niet geschikt om in de dagelijkse drinkwater
routine de DEC van processen te controleren. Hiervoor is kennis nodig
over de procescondities waarmee de DEC van een proces kan worden
gecontroleerd. Tot slot was onduidelijk in welke mate beide
indicatorbacteriën voorspellend waren voor de verwijdering van
Campylobacter bacteriën en beide protozoën, Cryptosporidium en Giardia.
Vanwege deze beperkingen en aanvullende vragen werden alternatieve
methoden toegepast om de DEC van processen te kunnen bepalen.
Hiervoor werden doseerproeven uitgevoerd met opstellingen op
verschillende schaal (proefinstallatie en laboratorium) en met verschillende
voorgekweekte micro-organismen, waaronder ook bacteriofagen als
procesindicator voor de verwijdering van virussen.
ELIMINATIE DOOR DESINFECTIE
In het kader van dit onderzoek zijn twee vergelijkende studies uitgevoerd
naar de effectiviteit van desinfectieprocessen. In Hoofdstuk 5 werden de
gevoeligheid van sporen van Clostridium perfringens en oöcysten van
Cryptosporidium parvum voor ozonisatie vergeleken in een continu
doorstroomde opstelling als model voor een praktijk ozoninstallatie. De
studie toonde aan dat SSRC gebruikt kan worden als procesindicator voor
de inactivatie van Cryptosporidium door ozonisatie in de praktijk. De
gevoeligheid van beide organismen was in dezelfde orde van grootte met
een index pathogeen/procesindicator (IP/PI) verhouding van 0,8. De
gegevens over de inactivatie bepaald met de continu doorstroomde
laboratoriumopstelling waren voorspellend voor de inactivatie van SSRC
waargenomen in een praktijk ozonproces. De inactivatie bij de ozoninbreng
(begassingsruimte) was aanzienlijk en proportioneel met de berekende Ctwaarde in deze ruimte (ozon concentratie C maal de contacttijd t;
mg/l.min). Wanneer de Ct van het proces wordt ingesteld met de
ozonconcentratie dan is de dosis-effect relatie in overeenstemming met de
___________________________________________________________________
- 276 -
Samenvatting
inactivatiekinetiek beschreven door Chick-Watson. Het instellen van Ct
waarden in de range van 0.5 – 1.5 m/l.min (gebruikelijk voor de
Nederlandse processen) met het debiet en de contacttijd als
stuurparameters bij een constante lage ozondosis bleek niet effectief om de
inactivatiecapaciteit van het desinfectieproces te beïnvloeden.
De gevoeligheid van micro-organismen voor Ultraviolet stralen (UV) is
bepaald en vergeleken op basis van literatuurgegevens in Hoofdstuk 6.
Doseerproeven met een laboratoriumopstelling (collimated beam) en
continu doorstroomde systemen evenals studies met natuurlijke microorganismen in praktijk UV systemen werden hiervoor gebruikt. Deze
evaluatie toonde een hogere inactivatie van voorgekweekte organismen in
de collimated beam opstelling vergeleken met de inactivatie van
natuurlijke organismen van dezelfde soort in continu doorstroomde
systemen. Deze discrepantie wordt veroorzaakt door factoren gerelateerd
aan de organismen en aan de procescondities. Deze factoren werden
geëvalueerd op basis van de beschikbare literatuur. Een vereiste UV
fluence tabel voor de relevante micro-organisms is beschreven waarin de
vereiste fluence voor bacteriën en bacteriesporen is gecorrigeerd voor deze
discrepantie tussen laboratoriumgegevens en praktijkobservaties. Op
grond van een vergelijking van literatuurgegevens was een dergelijke
correctie voor virussen niet noodzakelijk. Wegens een gebrek aan gegevens
was een dergelijke correctie voor de protozoën niet mogelijk.
ELIMINATIE DOOR FILTRATIE
Filtratieprocessen die in de productie van drinkwater worden toegepast
met als primair doel micro-organismen te elimineren zijn langzame
zandfiltratie en bodempassage. In Hoofdstuk 7 wordt een onderzoek
gepresenteerd naar de eliminatiecapaciteit van langzame zandfiltratie
bepaald onder verschillende condities. De resultaten toonden aan dat de
DEC voor Coli44 bepaald onder praktijkcondities 2-3 log was. Deze DEC
was in dezelfde orde van grootte als de DEC bepaald voor E. coli in een
filter van een proefinstallatie functionerend als een dummy van de
zuiveringslocatie. DEC bepaald in een doseerproef met kolommen op
laboratoriumschaal was duidelijk lager, mogelijk als gevolg van de
afwezigheid van een schmutzdecke in de kolommen en verschillen in
hydraulische condities van de filterbedden. In een proefinstallatie filter met
een schmutzdecke was de DEC voor E. coli 1-2 log hoger dan in hetzelfde
filter zonder deze laag. Het effect van de schmutzdecke op de verwijdering
van de bacteriofaag MS2 als een conservatief model organisme voor
___________________________________________________________________
- 277 -
Samenvatting
.
virussen was marginaal. Deze gedoseerde micro-organismen werden
verwijderd met 1,5-2,0 log door deze filters. Coli44 bleek een veilige
procesindicator voor de eliminatie van Campylobacter bacteriën door dit
proces.
Verwijdering van natuurlijke SSRC door praktijk langzame zandfilters
varieerde van -0,2 tot 1,8 log. In een doseerproef met korte kolommen op
laboratoriumschaal (0,4 m) was de verwijdering van sporen van C.
perfringens hoger (2,3 – 3,2 log). Oöcysten van Cryptosporidium werden in
deze kolommen zelfs met meer dan 5,3 en 6,5 log verwijderd. Deze
gegevens impliceren dat SSRC geen geschikte procesindicator voor
eliminatie van Cryptosporidium door langzame zandfiltratie is. De
hypothese voor de variabele en soms lage eliminatie van Clostridium sporen
door deze filters is accumulatie, overleving en vertraagd transport. Nader
onderzoek naar dit fenomeen werd uitgevoerd in een proeffilter bedreven
als een dummy van de praktijk filters (Hoofdstuk 8). Eliminatie van sporen
van C. perfringens, UV geïnactiveerde oöcysten van Cryptosporidium parvum
en centrische diatomeeën (Stephanodiscus hantzschii) werd bepaald in een
langdurige doorseerproef met een filter met een voorgeschiedenis van drie
jaar normale filtratie. Centrische diatomeeën zijn persistente organismen
met een grootte gelijk aan de grootte van oöcysten en deze organismen
waren van nature aanwezig in het influent in aanzienlijke hogere
concentraties dan natuurlijke SSRC. De DEC berekend uit de gegevens van
dit experiment voor oöcysten, sporen en diatomeeën was respectievelijk
4,7, 3,6 and 1,8 log. De duur van de test van drie maanden was te kort om
het process van het optreden van accumulatie en vertraagd transport voor
sporen en oöcysten te laten zien. Op basis van de lage DEC voor de
natuurlijke centrische diatomeeën die soms in hoge aantallen in het influent
werden waargenomen én de waarnemingen van de ruimtelijke verdeling
van sporen van C. perfringens in het filterbed bepaald aan einde van het
doseerexperiment werd geconcludeerd dat dit proces optreedt voor beide
persistente organismen. Echter waarschijnlijk niet voor persistente
organismen die gevoelig zijn voor predatie zoals oöcysten. Massabalans
berekeningen toonden een groot verlies aan van de oöcyst in het filterbed.
In de zoöplankton populatie die werd waargenomen in het zand, werden
soorten geïdentificeerd die oöcysten kunnen opnemen. Omdat deze
predatoren werden gevonden in het filterbed én in het filtraat, werd
vertraagd transport en doorbraak van levensvatbare oöcysten aanwezig in
een predator als een vector gesuggereerd als een onderwerp voor nader
onderzoek.
___________________________________________________________________
- 278 -
Samenvatting
De eliminatiecapaciteit van bodempassage onder veldcondities is normaal
gesproken niet eenvoudig te bepalen. In Hoofdstuk 9 wordt een
doseerexperiment
beschreven
waarbij
kolommen
gevuld
met
bodemmateriaal van twee oppervlakte infiltratie locaties in Nederland
werden belast met het plaatselijke oppervlaktewater waaraan MS2
bacteriofagen, E. coli en sporen van C. perfringens en (oö)cysten van
Cryptosporidium en Giardia waren gedoseerd. Van beide locaties waren
gegevens bekend over de eliminatie van gedoseerde MS2 fagen en
natuurlijke bacteriofagen en indicatorbacteriën tijdens bodempassage. Dit
experiment met verschillende organismen en twee bodemtypen maakte
duidelijk dat naast de grootte van micro-organisms en korrelgrootte van
het bodemmateriaal, oppervlakte eigenschappen van organismen en de
bodem en andere karakteristieken van de bodem als uniformiteit en
doorlatendheid, het transport van micro-organismen beïnvloeden. De
vertraagde en tweede doorbraak piek van oöcysten en sporen in dezelfde
bodemkolommen en de lage DEC van natuurlijke Coli44 en SSRC,
benadrukken de betekenis van desorptie voor transport van microorganismen in de bodem. Op basis van de resultaten werd geconcludeerd
dat zowel zeefwerking als adsorptie/desorptie de verwijderingsmechanismen zijn die, afhankelijk van de condities, in verschillende
verhouding bijdragen aan de eliminatie van micro-organismen bij
bodempassage. De IP/PI-verhouding voor Clostridium sporen en protozoën
(oö)cysten bepaald uit dit experiment was variabel en groter of kleiner dan
1,0 waardoor SSRC geen geschikte procesindicator is voor de verwijdering
van deze index pathogenen door bodempassage. Bovendien bleek de
waargenomen verwijdering van de gedoseerde micro-organismen in de
kolommen beduidend hoger te zijn dan de verwijdering van de natuurlijke
micro-organismen in dezelfde kolommen en bepaald onder veldcondities
van beide locaties.
ALGEMENE METHODIEK OM DE ELIMINATIECAPACITEIT
TE BEPALEN
Op basis van een discussie in Hoofdstuk 10 over de resultaten van het
onderzoek is een algemene methodiek beschreven om de decimale
eliminatiecapaciteit van lokale waterzuiveringsprocessen (DEC) te bepalen
voor index pathogenen. Deze DEC is de input voor de verplichte
kwantitative microbiologische risicoanalyse (Quantitative Microbial Risk
Assessment) die moet worden uitgevoerd voor de drinkwaterproductie
locaties die oppervlaktewater of kwetsbaar grondwater gebruiken als de
___________________________________________________________________
- 279 -
Samenvatting
.
bron. De meest optimale methoden zijn de bepaling van de eliminatie van
natuurlijk voorkomende index pathogenen en van natuurlijk voorkomende
indicatorbacteriën (Coli44 en SSRC) in de zuivering door groot volume
bemonsteringen. Omdat dit voor de index pathogenen maar beperkt
mogelijk is en niet voor de gehele zuivering, wordt Coli44 beschouwd als
natuurlijke procesindicator voor de verwijdering van Campylobacter door de
meeste processen en SSRC als procesindicator voor de verwijdering van
Cryptosprodium en Giardia door coagulatie, snelfiltratie en ozonisatie.
Verder onderzoek is nodig om vast te stellen of aanvulling met het meten
van bacteriofagen als een specifieke procesindicator voor virussen mogelijk
en noodzakelijk is. Omdat voor sommige processen ook het meten van de
verwijdering van natuurlijke indicatorbacteriën niet mogelijk is door een te
laag aanbod in het water en de index pathogeen/-procesindicator (IP/PI)
verhouding bepaald moet worden, zijn doseerproeven met voorgekweekte
organismen noodzakelijk om aanvullende kwantitatieve gegevens te
verzamelen. Deze gegevens kunnen worden toegepast om de DEC van
praktijk processen te bepalen wanneer de testen zijn uitgevoerd in een
proefinstallatie bedreven als een dummy van de praktijk zuivering.
Gegevens van doseerproeven op laboratoriumschaal met meerdere
organismen (batch desinfectieproeven of filtratieproeven met kolommen)
moeten worden beschouwd als indicatief voor de DEC onder
praktijkcondities. De waarde van deze testen ligt in het vaststellen van de
index pathogeen/procesindicator (IP/PI) verhouding en het ophelderen
van de invloed van specifieke condities op het eliminatieproces. Bij het
ontbreken
van
kwantitatieve
gegevens
over
concentraties
indicatorbacteriën in de praktijkinstallatie en van aanvullende
doseerproeven, is de literatuur een belangrijke bron van kwantitatieve
informatie
over
de
eliminatie
van
micro-organismen
door
waterzuiveringsprocessen. Aanvullend kunnen deze gegevens worden
gebruikt om gegevens van doseerproeven te vertalen naar de
praktijkcondities.
ALGEMENE CONCLUSIE
VERDER ONDERZOEK
EN
AANBEVELINGEN
VOOR
Een algemene methodiek is ontwikkeld en toegepast om de Decimale
Eliminatie Capaciteit (DEC) van waterzuiveringsprocessen op
praktijkschaal voor ziekteverwekkende micro-organismen te bepalen. De
methodiek is zoveel als mogelijk toegesneden op de natuurlijke condities in
___________________________________________________________________
- 280 -
Samenvatting
de waterzuivering en de dagelijkse praktijk van microbiologische kwaliteits
controle.
Een onderwerp voor verder onderzoek is de bepaling van de DEC van
zuiveringen voor virussen waarbij gebruik wordt gemaakt van natuurlijke
bacteriofagen als mogelijke procesindicatoren. In verband hiermee is de
toepassing van ultrafiltratie in de recent ontwikkelde hemoflow methode
van belang om nader te bekijken. Met deze methode worden meerdere
micro-organismen waaronder bacteriofagen en fecale indicatorbacteriën in
grote volumes water bepaald. Voor een aantal waterzuiveringsprocessen
ontbreekt kennis over de eliminatiecapaciteit en de IP/PI verhouding. Deze
processen zijn aktieve koolfiltratie, nadesinfectie met chloordioxide en
ultrafiltratie. Het effect van de voorgenomen verandering in de wettelijke
waterkwaliteitscontrole van sporen van sulfiet-reducerende clostridia naar
specifiek C. perfringens inclusief vegetatieve cellen waarbij een
pasteurisatiestap achterwege wordt gelaten, kan het gebruik van deze
organismen als procesindicator beïnvloeden en is daarom een punt van
nader onderzoek. Predatie in biologische filters is potentieel een voordelig
eliminatie mechanisme in de waterzuivering voor de microbiologische
veiligheid van water mits de predatoren niet functioneren als vector voor
transport voor de opgenomen pathogenen. Dit is een belangrijk onderwerp
voor vervolgonderzoek.
___________________________________________________________________
- 281 -
List of publications
List of publications
PEER REVIEWED JOURNALS
Hijnen, W.A.M., Y.F. Dullemont, J.F. Schijven, A. Brouwer-Hanzens, M. Rosielle
and G.J. Medema. 2007. Removal and fate of Cryptosporidium parvum, Clostridium
perfringens and small-sized centric diatoms (Stephanodiscus hantzschii) in slow
sand filters. Water Res. 41: 2151-2162.
Hijnen, W.A.M., E.F. Beerendonk and G.J. Medema. 2006. Inactivation credit of
UV radiation for viruses, bacteria and protozoan (oo)cysts in water: a review.
Water Res. 40: 3-22.
Hijnen, W.A.M., A.J. Brouwer-Hanzens, K. Charles and G.J. Medema. 2005.
Transport of MS2 phage, Escherichia coli, Clostridium perfringens, Cryptosporidium
parvum and Giardia intestinalis in a gravel and a sandy soil. Environ. Sci. &
Technol., 39: 7860-7868.
Hijnen, W.A.M., E. Baars, A.J. van der Veer, Th.G.J. Bosklopper, R.T. Meijers
and G.J. Medema. 2004. Influence of DOC on the inactivation efficicency of
ozonation assessed with C. perfringens and a lab-scale continuous flow system.
Ozone: Sci. Eng. 26: 465-473.
Hijnen, W.A.M, D. Veenendaal, W.M.H. van der Speld, A. Visser, W.
Hoogenboezem and D. van der Kooij. 2000. Enumeration of faecal indicator
bacteria in large water volumes using on site membrane filtration to assess water
treatment efficiency. Water Res. 34(5):1659-1665.
Hijnen, W.A.M., R. Jong and D. van der Kooij. 1999. Bromate reduction in a
denitrifying bioreactor. Water Res. 33:1049-1053.
Hijnen, W.A.M., R. Voogt, H.R. Veenendaal, H. van der Jagt and D. van der
Kooij. 1995. Bromate reduction by denitrifying bacteria. Appl. Environ.
Microbiol. 61:239-244.
Hijnen, W.A.M. and D. van der Kooij. 1992. The effect of low concentrations of
assimilable organic carbon (AOC) in water on the biological clogging of sand
beds. Water Res. 26(7):963-972.
Hijnen, W.A.M. and D. van der Kooij. 1992. AOC removal and accumulation of
bacteria in experimental sand filters. Revue des sciences de l'eau, 5 (special
issue):17-32.
PEER REVIEWED CONGRES JOURNALS
Hijnen, W.A.M. and G.J. Medema. 2005. Inactivation of viruses, bacteria, spores
and protozoa by ultraviolet irradiation in drinking water practice: a review. Wat.
Sci. Technol.: Wat. Supply, 5(50):93-99.
Hijnen, W.A.M., J.F. Schijven, P. Bonné, A. Visser and G.J. Medema. 2004.
Elimination of viruses, bacteria and protozoan oocysts by slow sand filtration.
Wat. Sci. and Technol. 50(1):147-154.
___________________________________________________________________
- 283 -
List of publications
.
Hijnen, W.A.M., A.J. van der Veer, E.F. Beerendonk and G.J. Medema. 2004.
Increased resistance of environmental anaerobic spores to inactivation by UV. Wat.
Sci. Technol.: Wat. Supply, 4(2):54-61.
Hijnen, W.A.M., G.J. Medema and D. van der Kooij. 2004. Quantitative assessment
of the removal of indicator bacteria by full-scale treatment. Wat. Sci. Technol.: Wat.
Supply, 4(2):47-54.
Hijnen, W.A.M., A.J. van der Veer, J. van Beveren and G.J. Medema. 2002. Spores of
sulphite-reducing clostridia (SSRC) as surrogate for verification inactivation
capacity of full-scale ozonation for Cryptosporidium. Wat. Sci. and Technol.: Wat.
Supply, 2(1):163-171.
Hijnen, W.A.M., J. Willemsen-Zwaagstra, P. Hiemstra , G.J. Medema and D. van
der Kooij. 2000. Removal of sulphite-reducing clostridia spores by full scale water
treatment processes as a surrogate protozoan (oo)cysts removal. Water Sci. and
Technol. 41:165-171.
Hijnen, W.A.M., D. Koning, J.C. Kruithof and D. van der Kooij. 1988. The effect
of bacteriological nitrate removal on the concentration of bacteriological biomass
and easily assimilable organic carbon compounds in water. Wat. Supply 6:265273.
INTERNATIONAL PROCEEDINGS
Hijnen, W.A.M. and G.J. Medema. 2007. Microbial Elimination Capacity of
conventional treatment for viruses, bacteria and protozoan (oo)cysts.
Proceedings Water Quality Technology Conference WQTC, Charlotte, US.
Hijnen, W.A.M., A.T. Lugtenberg, H. Ruiter, R.R.J. Vink and G.J. Medema. 2007.
Decay Rate Index for E. coli and enterococci in fresh and salt bathing waters.
Proceedings Water Quality Technology Conference WQTC, Charlotte, US.
Hijnen, W.A.M., Y.J. Dullemont, J.F. Schijven, K. Bosklopper and G.J. Medema.
2006. Assessment of the capacity of slow sand filtration to eliminate
Cryptosporidium oocysts. Proceedings Water Quality Technology Conference
WQTC, Denver, US.
Hijnen, W.A.M., Th.G.J. Bosklopper, J.A.M.H. Hofman, A.D. Bosch and G.J.
Medema. 2001. Improvement of the disinfection efficiency of the full-scale
ozonation of the River-lake waterworks of Amsterdam Water Supply. In proc.
Int. Ozone Assoc. Congres, september 2001, London.
Hijnen, W.A.M., F.A.P. Houtepen, W.M.H. van den Speld en D. van der Kooij. 1997.
Spores of sulphite-reducing clostridia: a surrogate parameter for assessing
the effects of water treatment processes on protozoan (oo)cysts? In proc. AWWA
Symposium on Cryptosporidium 2-3 march, 1997, USA.
Hijnen, W.A.M., J.W.N.M. Kappelhof, J.P. van der Hoek, D. van der Kooij,
A.J.H.F. Creusen and L.A.C. Feij. 1993. Biological filtration for the removal of
AOC and biomass from ground water after denitrification. In proc. European
Water Filtration Congress, Oostende, 15-17 March 1993: pp. 2.73-2.85.
___________________________________________________________________
- 284 -
List of publications
OTHER WATER JOURNALS AND BOOKS
Hijnen, W.A.M., P.J. Stuyfzand and G.J. Medema. 2004. Initial pathogen sorption
and transport in soil. In American Water Works Assoc. Research Foundation
report “Fate and transport of surface water pathogens in watersheds”. Davies,
Kauchner, Altavilla, Ashbolt, Ferguson and Deere, UNSW Australia.
Hijnen, W.A.M. en G.J. Medema. 2003. Kwantificering van de “multiple barriers”.
H2O 36(19):24-27.
Hijnen, W.A.M. and P.W.M.H. Smeets. 2004. Ozone as disinfectant in drinking
water production. PAO course “UV en andere desinfectietechnieken in de
drinkwaterbereiding en de afvalzuivering”, 8-9 November 2004, Delft University
of Technology, the Netherlands.
Hijnen, W.A.M., P.N.A.M. Nuhn D. van der Kooij en G.J. Medema. 1999.
Verwijdering
van
ziekteverwekkende
micro-organismen
bij
de
drinkwaterbereiding: en kwantitatieve aanpak. H2O 32(7):27-31.
Hijnen, W.A.M. G.J. Medema en E. Koreman. 1999. Veilig hergebruik van
spoelwater. H2O 32(7):25-27.
Hijnen, W.A.M., J. Bunnik, J.C. Schippers, R. Straatman and H.C. Folmer. 1998.
Determining the clogging potential of water used for artificial recharge in deep
sandy aquifers. Third International Symposium on artificial Recharge, TISAR
1998.
Hijnen, W.A.M., R. Jong, W.C. van Paassen, L.A.C. Feij en D. van der Kooij. 1997.
Bereiding van biologisch stabiel drinkwater na biologische nitraatverwijdering
uit grondwater. H2O 30:149-153.
Hijnen, W.A.M., W.A. Oorthuizen, A. de Ruyter en D. van der Kooij. 1996.
Microbiologische beoordeling en juiste behandeling van nieuw filtermateriaal
verkorten de inlooptijd van langzame zandfilters. H2O 29:163-164
Hijnen, W.A.M. 1996. Nabehandeling van grondwater na nitraatverwijdering met
het autotrofe denitrificatie met zwavel en denitrificatie met het ethanol vastbed
proces. In Kiwa mededeling 124, Nitraatverwijdering, KWR Nieuwegein NL.
Hijnen, W.A.M., R. Voogt en D. van der Kooij. 1995. Nitraatreducerende
bacteriën zetten bromaat om in bromide. H2O 28:390-391.
Hijnen, W.A.M. 1994. Effecten van methaan en ammonium op de
microbiologische kwaliteit van het water. Kiwa mededeling 123, Behandeling
van methaanhoudend grondwater, KWR Nieuwegein, NL.
Hijnen, W.A.M., J. Verdouw, J.C. Schippers en D. van der Kooij. 1993.
Bestimmung des Verstopfungspotentials von Schluckbrunnen. In DVGWSchriftenreihe Wasser 85:131-149.
Hijnen, W.A.M. und Dirk van der Kooij. 1992. Biologische Kolmation von
Schluckbrunnen unter dem Einfluß des AOC-Gehalts des Wassers. WasserAbwasser 133(3):148-153.
Hijnen, W.A.M., G.K. Reijnen, R.H.M. Bos, G. Veenendaal en D. van der Kooij.
1992. Lagere Aeromonas-aantallen in het drinkwater van pompstation
Zuidwolde door verbeterde ontgassing en vernieuwen van het filtergrind. H2O
___________________________________________________________________
- 285 -
List of publications
.
25:370-375.
Hijnen, W.A.M., G.K. Reijnen en D. van der Kooij. 1992. Groei van Aeromonas in
filters gevoed met methaan- en ammoniumhoudend grondwater. Kiwa-VWN
bundel Colloquium “Aeromonas, vóórkomen, bestrijden en betekenis”, KWR
Nieuwegein, NL.
Hijnen, W.A.M. en D. van der Kooij. 1990. Verstopping van infiltratieputten door
bacteriegroei onder invloed van het AOC-gehalte van water. H2O 23:142-148.
Hijnen, W.A.M., J.C. Kruithof en D. van der Kooij. 1990. Bacteriologische
kwaliteit en
het
AOC-gehalte
van
grondwater
na
biologische
nitraatverwijdering. H2O, 23:720-726.
Hijnen, W.A.M. en D. van der Kooij. 1989. Verstopping van infiltratieputten
onder invloed van het gehalte assimileerbare organische koolstof (AOC) van het
water. Kiwa mededeling 106, KWR Nieuwegein NL.
Hijnen, W.A.M en D. van der Kooij. 1984. Rol van microbiologische processen bij
het verstoppen van persputten. Kiwa mededeling 79, KWR Nieuwegein NL.
AS CO-AUTHOR
Van der Kooij, D., W.A.M. Hijnen, E. Cornelissen, S. van Agtmaal, Koos Baas
and G. Galjaard. 2007. Elucidation of membrane biofouling processes using
bioassays for assessing the microbial growth potential of feed water. Proceedings
AWWA membrane technol. Conf. Tampa Bay, US.
Cornelissen, E., W.A.M. Hijnen, P. Wessels, D. van der Kooij and D. Biraud.
2007. Assessment of the efficiency of air-water flushing for the removal of
biomass from surfaces in a laboratory test. Proceedings AWWA membrane
technol. Conf. Tampa Bay, US.
Dullemont, Y.J., J.F. Schijven, W.A.M. Hijnen, M. Colin, A. Magic-Knezev and
W.A. Oorthuizen. 2006. Removal of microorganisms by slow sandfiltration. In
Recent progress in slow sand and alternative biofiltration processes ed. R.
Gimbel, N.J.D. Graham and M.R. Collins, IWA puplishing, London, UK.
Smeets, P.W.M.H., W.A.M. Hijnen and T.A. Stenström. 2006. Efficacy of water
treatment processes, Chapter 4 in ‘Quatitative Microbial Risk Assessment in
Water Safety Plan, Medema, G.J., J. Loret, T.A. Stenström and N. Ashbolt (ed.),
report for the European Commission under the fifth Framework Programm,
Theme 4: Energy, environment and sustainable development (contract
EVK1CT200200123), Kiwa Water Research, Nieuwegein, NL.
Van der Veer, W.A.M. Hijnen and D. van der Kooij. 2005. Pilot plant studies on
AOC and Biofilm Formation Rates of ozonated water after filtration. In proc. of
17th world congress of IOA, August 22-25, Strasbourg, France.
Chung, J., W.A.M. Hijnen, G. Vesey, N.J. Ashbolt. 2004. Potential Cryptosporidium
oocyst surrogates for sand filtration and the importance of their surface
properties. Int. Conf. on Cryptosporidium and Giardia, Amsterdam, NL.
Visser, A., W.A.M. Hijnen, Y.J. Dullemont, G.J. Medema. 2004. Langzame
zandfilters als effectieve barrières voor micro-organismen. H2O 37(12):26-28.
___________________________________________________________________
- 286 -
List of publications
Dullemont, Y.J., A. Visser, J.F. Schijven en W.A.M. Hijnen. 2004.
Eliminatiecapaciteit van langzame zanfiltratie voor micro-organismen bepaald
met doseerproeven. H2O 37(13):22-24.
Medema, G.J., W. Hoogenboezem, A.J. van der Veer, H.A.M. Ketelaars, W.A.M.
Hijnen and P.J. Nobel. 2003. Quantitative risk assessment of Cryptosporidium
in surface water treatment. Wat. Sci. Tech. 47(3):241-247.
Bosklopper, K., W.A.M. Hijnen, Y.J. Dullemont en G.J. Medema. 2003. HACCP:
risicoanalyse en procesverbetering in de praktijk. H2O 36(19):34-36.
Medema, G.J., J. Groennou, W.A.M. Hijnen, P. Teunis, L. Kruideniers, J.
Willemsen-Zwaagstra, A. Havelaar and D. van der Kooij. 1999. Frequency
distributions of Cryptoporidium and Giardia in raw water and elimination
capacity of water treatment. In “Cryptosporidium and Giardia: new challenges to
the water industry. Thesis, University of Utrecht.
Medema, G.J., Hijnen, W.A.M., P.J. Nobel, and D. van der Kooij. 1997.
Cryptosporidium and Giardia- the Dutch perspective. Chartered Institution of Water
and Environmental Management, Techn. Papers CIWEM symposium, 4 Dec. 1997,
United Kingdom.
Jong, R.C.M., J.W.N.M. Kappelhof, W.A.M. Hijnen en A.J.H.F. Creusen. 1997.
Beheersing van een denitrificerende ethanol vast bed bioreactor. H2O 30(5):140141.
Van der Kooij, D., W.A.M. Hijnen, L.W. van Breemen, F.A.P. Houtepen, J.
Willemsen-Zwaagstra. 1995. Removal of micro-organisms in surface water
treatment in the Netherlands. Proc. Water Quality Technol. Conf. Nevember 1995,
New Orleans:2277-2286.
Van Puffelen, J., P.J. Buijs, P.N.A.M. Nuhn and W.A.M. Hijnen. 1995. Dissolved Air
Flotation in potable water treatment: the Dutch experience. Wat. Sci. Technol.
31:146-157.
Van der Kooij, D., Y.C. Drost, W.A.M. Hijnen, J. Willemsen-Zwaagstra, P.J.
Nobel and J.A. Schellart. 1994. Multiple barriers against micro-organisms in
water treatment and distribution in the Netherlands. In. Conf. IWSA, Kruger
National Park South Africa, 13-18 March, 1994.
Hoek, J.P. van der, R.C. Jong, J.W.N.M. Kappelhof, W.A.M. Hijnen, A.J.H.F.
Creusen, A.J. Bekkers, L.A.C. Feij. 1993. Nitrate removal from ground water by
biological filtration using the fixed bed/ethanol process. In Proceedings:
European Water Filtration Congress, Oostende, 15-17 March 1993, pp. 2.55-2.66.
Kruithof, J.C., R.Chr. Van der Leer en W.A.M. Hijnen. 1992. Practical experiences
with UV disinfection in the Netherlands. J Water SRT-Aqua 41(2):88-94.
Hoek, J.P. van der, W.A.M. Hijnen, C.A. van Bennekom and B.J. Mijnarends.
1992. Optimization of the sulpher-limestone filtration process for nitrate removal
from groundwater. J Water SRT-Aqua 41(4):209-218.
Kappelhof J.W.N.M., J.P. van der Hoek and W.A.M. Hijnen. 1991. Experiences
with fixed bed denitrification using ethanol as substrate for nitrate removal from
ground water. In IWSA international Workshop Inorganic nitrogen compounds
and water supply, November 27 - 29, Hamburg.
___________________________________________________________________
- 287 -
List of publications
.
Kooij, D. van der and Hijnen, W.A.M. 1990. Criteria for defining the biological
Stability of drinking water as determined with AOC-measurements. Proceedings
of the AWWA Water Quality Technology Conference in San Diego, November
1990.
Van der Kooij, D., W.A.M. Hijnen and J.C. Kruithof. 1989. The effects of
ozonation, biological filtration and distribution on the concentration of easily
assimilable organic carbon in drinking water. Ozon. Sci. & Eng. 11:297-311.
Kruithof, J.C., J.A.M. van Paassen, W.A.M. Hijnen, H.A.L. Dierx and C.A. van
Bennekom. 1985. Experiences with nitrate removal in the eatern Netherlands.
Proceedings of the congress "Nitrates in water" Paris, october 22-24, 1985.
Van der Kooij, D. and W.A.M. Hijnen. 1984. Substrate utilization by an oxalateconsuming Spirillum species in relation to its growth in ozonated water. Appl.
Environ. Microbiol. 47:551-559.
Van der Kooij, D. en W.A.M., Hijnen. 1984. Mogelijkheden van AOC-bepalingen
bij het vaststellen van de concentratie van gemakkelijk afbreekbare organische
verbindingen in water. H2O 17(12):249-252.
Van der Kooij, D. en Hijnen W.A.M. 1984. Aanwezigheid en bestrijding van
Legionella pneumophila, de veroorzaker van de veteranenziekte, in
warmtapwatersystemen. H2O 17(18):387-391.
Van der Kooij, D. and W.A.M. Hijnen. 1983. Nutritional versatility of a starchutilizing Flavobacterium at low substrate concentrations. Appl. Environ.
Microbiol. 45:804-810.
Van der Kooij, D. en W.A.M. Hijnen. 1983. Verwijdering van organische stoffen
door micro-organismen bij filtratie-processen. H2O 16(13):306-311.
Van der Kooij, D., J.P. Oranje and W.A.M. Hijnen. 1982. Growth of Pseudomonas
aeruginosa in tap water in relation to utilization f substrates at concentrations of
a few micrograms per liter. Appl. Environ. Microbiol. 44:1086-1095.
Van der Kooij, D., A. Visser & W.A.M. Hijnen. 1982. Determining the
concentration of easily assimilable organic carbon in drinking water. J.Am. Wat.
Wks Ass. 74:540-545.
Van der Kooij, D. and W.A.M. Hijnen. 1981. Utilization of low concentrations of
starch by a Flavobacterium species isolated from tap water. Appl. Environ.
Microbiol. 41:216-221.
Van der Kooij, D., A. Visser and W.A.M. Hijnen. 1980. Growth of Aeromonas
hydrophila at low concentrations of substrates added to tap water. Appl. Environ.
Microbiol. 39:1198-1204.
Nijssen, J en W.A.M. Hijnen. 1978. Afvalwaterzuivering op de hogere school voor
levensmiddelentechnologie. H2O 11(21):489-490.
___________________________________________________________________
- 288 -
Curriculum vitae
CURRICULUM VITAE
Wim Hijnen is geboren op 6 juni 1956 te Bussum. Hij heeft in 1972 het
HAVO diploma behaald aan het Leeuwenhorst College te
Noordwijkerhout. Vervolgens is hij gaan studeren aan de Hogere
Agrarische School te ’s-Hertogenbosch. De studierichting die hij na een
algemeen eerste jaar koos was Levensmiddelentechnologie. In 1977
studeerde
hij
daar
in
af
met
als
specialismen
Afvalwaterzuiveringstechnologie en Technische Microbiologie. Ter
afronding van zijn studieperiode in ’s-Hertogenbosch heeft hij in 1978 zijn
onderwijsbevoegdheid gehaald.
Zijn loopbaan startte bij het Keuringsinstituut voor Waterleidingartikelen
te Rijswijk op 16 augustus 1978 als microbiologisch onderzoeksanalist en
laboratoriumcoördinator. Vanaf circa 1980 is hij onderzoeks- en
adviesprojecten gaan doen. Allereerst op het gebied van biologische
verstopping van infiltratieputten en biologische nitraatverwijdering uit
grondwater. Vanaf 1987 heeft hij het laboratorium verlaten om als
projectleider te gaan werken in de toenmalige Sectie Microbiologie. De
onderwerpen waaraan hij vanaf die periode microbiologisch
zuiveringsonderzoek is gaan doen, zijn biologisch stabiel water na
ozonisatie, Aeromonas nagroei, biologische afbraak van bromaat en
microbiologische veiligheid van drinkwater, in het bijzonder verwijdering
van ziekteverwekkende micro-organismen in de waterzuivering. Op grond
van
deze
activiteiten
heeft
hij
zitting
gehad
in
Onderzoeksbegeleidingscommissies en de werkgroep Microbiologie van
het Speurwerkonderzoeksprogramma. Naast deze projecten op het gebied
van de openbare drinkwatervoorziening, heeft hij ook microbiologische
projecten uitgevoerd in het kader van hergebruik van afvalwater en
zwemwater.
In 1982 heeft hij het diploma Milieukunde van de PBNA gehaald. In 1989
werd het PHBO diploma Bioprocestechnologie behaald en in 1998 heeft hij
deelgenomen aan de 10e Eijkmancursus voor levensmiddelenmicrobiologie.
In 2003 is hij voor 3,5 maand Fellow visitor geweest aan de University of
New South Wales (Sidney, Autralia), waar hij samen met een PhD student
laboratoriumonderzoek heeft gedaan naar de verwijdering van microorganismen bij zandfiltratie. Vanaf 2005 is hij een eigen PhD traject gestart
om zijn doctorstitel te halen aan de Universiteit van Utrecht. In zijn huidige
functie bij KWR Watercycle Research Institute is hij zelfstandig
onderzoeker en adviseur op het gebied van de microbiologische veiligheid
en activiteit in de watercyclus.
___________________________________________________________________
- 289 -
Fly UP