Removal of 'Giardia lamblia' Cysts by
Drinking Water Treatment Plants
Washington Univ., Seattle
Prepared for
Municipal Environmental Research Lab.
Cincinnati, OH
Mar 84
PB84-162874
el teaetars
»>».» • '•
-------
EPA-600/2-84-069
March 1984
REMOVAL OF GIARDIA LAMBLIA CYSTS BY DKITIKING WATER TREATMENT PLANTIE
by
Foppe B. DeKalle
Jogeir Engeset
William Lawrence
University of Washington
Seattle, Washington 98195
Granc No. R806127
Project Officer
Gary L gsclon
Drinking Wat»»r Research Division
Municioal Environmental Researcr. Laboratory
Cincinnati, Ohio 45268
MUNICIPAL ENVIRONMENTAL rESEARCH LABORATORY
OFFICE OF RESEARCH AND DEVELO»MEOT
U.S. ENVIRONMENTAL PROTECTION ARENCY
CINCINNATI, r«IO 45268
-------
TECHNICAL REPORT DATA
(Please read /ntirucnons un me reverie oe/ore comaltnntl
RtPORT .NO.
EPA-600/2-84-069
12.
3 RECIPIENT
r s AC=ESSJOAI i
16287^
TITLE AND SUBTITLE
Removal of Ciardi'a Iambi la Cysts by Drinking Water
Treatment Plants
5. REPORT DATE
March 1984
6. PERFORMING ORGANIZATION CCCtE
AUTHOHISI
Foppe B. DeWalle, Jogeir Engeset, William Lawrence
8. PERFORMING ORGANIZATION REPORT NC
PERFORMING ORGANIZATION NAME AND ADDRESS
Dept. of Environmental Health
University of Washington
Seattle, Washington 98195
10. PROGRAM ELEMENT NO.
BNC1A
11. CCN7RAC7
-------
DISCLAIMER
•Hie information in this document has been funded wholly or in part by the
United States Environmental Protection Agency under assistance agreement
number R806127 to the University of Washington. It has been subject to the
Agency's administrative review, and it has been approved for publication as
an EPA document. Mention of trade names or commercial products does not
constitute endorsement or recommendation for use.
ii
-------
FOREWORD
The U.S. Environmental Protection Agency was created because of
increasing public and government concern about the dangers of pollution to the
health and welfare of the American people. Noxious air, foul water, and
spoiled land are tragic testimonies to the deterioration of our natural envi-
ronment. The complexity of that environment and the interplay of its com-
ponents require a concentrated and integrated attack on the problem.
Research and development is that necessary first step in problem
solution, and it involves defining the problem, measuring its impact, and
searching for solutions. The Municipal Environmental Research Laboratory
develops new and improved technology and systems to prevent, treat, and
manage wastewater and solid and hazardous waste pollutant discharges from
municipal and community sources, to preserve and treat public drinking water
supplies, and to minimize the adverse economic, social, health, and aesthetic
effects of pollution. This publication is one of the products of that
research and is a most vital communications link between the researcher and
the user community.
This report presents the results and conclusions from pilot plant fil-
tration research on the removal of Giardia lamblia cysts and cyst-sized
particles from drinking water. Granular media filters and a diatomaceous
earth filter were evaluated in this study.
Francis T. Mayo
Director
Municipal Environmental Research
Laboratory
iii
-------
ABSTRACT
A study was conducted to evaluate the removal of Giardia lamblia cysts
and cyst-sized particles from Cascade Mountain waters. Methods included
coagulation/sedimentation and filtration, or direct filtration using three
2.3 L/min (0.6 gpm) pilot treatment units and diatonaceous earth (DE)
filtration using a 3.8 L/min (1 apnv/ft2J DE pilot filter. The units were
located at the University of Washington. The results were verified through
field testing using a 75 L/min (20 gpm) pilot unit (Waterboy, Neptune Micro-
floe) in field trials at Hoquiam and Leavenworth, Washington.
Ihe study noted greater than 99.9% removal of spiked cysts under
optimum conditions, although removal percentages decreased greatly at lower
spiking levels. Both the University of Washington pilot unit and the field
unit established the importance of a minimum alum dosage (10 mg/L) / an
optimum pH range, and intermediate flow rates of 4.9 m/hr (2 gpm/ft ) to 9.8
m/nr (4 gpm/ft ). Effluent turbidity and cyst-sized particles passing the
filter increased rapidly when the above conditions were not attained or when
sudden changes occurred in plant operation. When no coagulants were used
during filtration, only 48% of the spiked cysts were removed, and 47% of the
turbidity. A cyst spike in the pilot unit in Hoquiam using alum as
coagulant resulted in an 81% cyst removal, and the spike at Leavenworth
using a polymeric flocculant gave a 72.1% removal. Producing a low
turbidity filter effluent with alum or polymeric flocculant was difficult
when the water temperature was 3 C. Further research in low temperature
direct filtration is necessary to improve the removal efficiency under these
conditions. DE filtration proved effective both for turbidity, particle and
cyst removal. The addition of 0.0075 mg/L nonionic polymer showed some
improvement in efficiency. Cyst removals ranged from about 99% to 99.y9%.
This report was submitted in fulfillment of Grant NO. R-8061/27 by the
University of Washington under the sponsorship of the US Environmental
Protection Agency. Ohis report covers the period from September 1978 to
March 1982, and work was completed at that date.
IV
-------
Foreword iii
Abstract iv
Figures vi
Tables ix
1. introduction 1
Characteristics of Organism 1
Characteristics of Disease 2
Prevalence of Organism 4
Giardiasis Outbreaks 5
2. Conclusions 11
3. Experimental Procedures 13
Collection and Enumeration of Giardia Cysts 13
Design and Testing of the 2.3 L/min. (0.6 gpn) Water Treatment
Pilot Plants 22
Testing of Coagulation/Filtration and Direct Filtration at the
University of Washington 24
Testing of Diatomaceous Earth Filter at the University of
Washington 25
Testing of Direct Filtration in Hoguiam and Leavenworth ... 28
Hoquiam Water Treatment Plant 29
Leavenworth Water Treatment Plant 31
4. Results 32
Method Evaluation: Collection, Enumeration of Giardia Cysts
and QVQC 32
Testing of University of Washington Pilot Plant 41
Testing of Coagulation/Filtration and Direct Filtration at
University of Washington 47
Testing of Diatomaceous Earth Filter at the University of
Washington 73
Testing of Direct Filtration in Hoquiam and Leavenworth ... 78
References 100
102
-------
FIGURES
Numbers Page
1 Sucrose gradient technique to recover cysts from stool specimens 14
2 Modified sucrose gradient technique to recover cysts 15
3 Schematic of 293mm Millipore Filter Unit used to recover C_.
lamblia cysts f ran water 18
4 Procedure for recovery of G. lamblia cysts with the 293mm
Millipore filter 19
5 Procedure for recovery of cysts from dilute water suspensions . . 20
6 Water treatment pilot plant at University of Washington 23
7 Schematic of the DE filter system 26
8 Cross section of coagulation, flocculation and mixed media
filtration compartments of the Waterboy-27 30
9 Size distribution of serially diluted Giardia suspension in
distilled water at (1) 5%, (2) 2.5%, (3) 1.25% and (4) 0.625%
of the stock solution 33
10 Linearity of two counting methods for enumerating Giardia cysts . 34
11 Coefficient of variation for two methods used for enumerating
Giardia cysts 35
12 Results of 47mm diameter membrane filter recovery test using
Lake Union water spiked with Giardia cysts. (1) before
recovery, (2) recovered cysts and (3) background counts .... 37
13 Percent recovery of Giardia cyst by different 47mm diameter
membrane filters from two types of water 38
14 Percent recovery of cysts by 293mm diameter, 5ym pore size
membrane filters, (A) Millipore and (B) Nuclepore 39
15 Effects of pH on the zeta potential of fixed G. lamblia cysts,
(A) different suspensions and (B) same suspension . 43
VI
-------
Hunters Page
16 Tracer evaluation of the rapid mix tanks 44
17 Tracer evaluation of the flocculation tanks 45
18 Tracer evaluation of the sedimentation tanks 46
19 Turbidity in filter influent and effluent of Run no. 4 52
20 Turbidity in filter influent and effluent of Run no. 5 53
21 Turbidity in filter influent and effluent of Run no. 6 54
22 Turbidity in filter influent and effluent of Run no. 7 56
23 Effect of alum dosage on direct filtration process, Filter B . . 58
24 Effect of alum dosage on direct filtration process, Filter C . . 59
25 Effect of pH on direct filtration performance, Filter B 60
26 Effect of pH on direct filtration performance, Filter C 61
27 Effect of pH increase on filter performance 63
28 Effect of flowrate on direct filtration efficiency. Filter B . . 64
29 Effect of flowrate on direct filtration efficiency, Filter C . . 65
30 Particle removal at different filter depths 66
31 Sampling schedule for 20L filter effluent sample at different
filtration rates 68
32 Percentage of total number of filter effluent cysts present in a
20L sample collected according to Figure 31 69
33 Effect of alum dosage on cyst removal 72
34 Effect of pH on cyst removal 74
35 Characteristics of DE filter run with Celite 503 filter aid at
20 mg/L body feed 75
36 Typical data from a DE filter run using Hyflo Super-Gel as
filter aid. Body feed rate, 20 mg/L 76
37 Typical data from a DE filter run using Celite 512 as filter aid.
Body feed rate, 20 mg/L 77
vii
-------
Number Page
38 Effect of alum dosage on particle and turbidity removal during
field work at Hcquiam 80
39 Effect of pH on particle and turbidity removal during field work
at Hoquiam 82
40 Effect of pH changes during Run no. 9 at Hoquism 83
41 Effect of filtration rate on particle and turbidity at Hoquiam . 85
42 Effect of high filtration rate on filter performance at Hcquiam.
Alum dosage 15 mg/L, pH 6.7 and filter loading 15 m/hr
(6.1 gpro/ft2) 86
43 Relationship between effluent turbidity and particle removal at
Hoquiam 87
44 Relationship between effluent turbidity and median particle
removal 88
45 Turbidity removal at Hoquiam Water Treatment Plant 90
46 Effect of alum dosage and pH on turbidity removal at Hoquiam
Water Treatment Plant 91
47 Effect of alum dosage on particle and turbidity removal at
different tenperatures during field work ?t Leavenworth .... 93
48 Effect of pH on particle and turbidity removal at different
temperatures during field work at I/eavenworth 94
49 Effect of Cat Floe T polymer dosage on particle and turbidity
removal and rate of headless buildup at Leavenworth 96
50 Frequency distribution of particle removal at different effluent
turbidities during alun and polymer treatment at Leavenworth . 97
51 Effect of polymer dosage on turbidity removal at Leavenworth
Water Treatment Plant 98
viii
-------
TAELES
Number Page
1 Summary of Laboratory Recovery Bates of G lamblia cysts with
Millipore Pellicon Cassette Unit 40
2 Zeta Potential (Electrophoretic Mobility) of Buffered Formalin
Fixed Giardia lamblia cysts at Varying pH Values and Cyst
Concentrations 42
3 Zeta Potential of a Fixed Giardia lamblia Cys Suspension at
Different pH Values 42
4 Results of a Single Dose Spike of Giardia Cysts into Flocculation
Compartment of Pilot Plant - Run #1 48
5 Results of Continuous Spike of Giardia Cysts into Pilot Plant -
Run #2 50
6 Results of Continuous Spike of Giardia Cysts Directly Introduced
into Dual Media Filters - Run #3 51
7 Performance cf Each Filter Run with Cysts Added Directly to the
Filter 57
8 Cyst Removal During Direct Filtration at UW Pilot Plant 71
9 Filter Runs with Cysts Using DE Filter 79
-------
SECTION 1
INTRODUCTION
A stud/ was undertaken to evaluate the removal of Giardia iambiia cysts
by drinking water plants. Hie first phase of the study was devoted to a
laboratory-scale evaluation of Giardia removal efficiency by coagulation,
flocculation, and filtration. In addition, a diatonaceous earth filter was
tested. The second phase consisted of a pilot- scale .evaluation of Giardia
cyst and cyst-size particle removal from drinking water at locations in the
State of Washington that were suspected of harboring cysts in the raw water.
All laboratory water treatment plant experiments were conducted with
unflitered Seattle tap water to which cysts were added. The cysts that were
used to spike the water were isolated front the feces of human giardiasis
patients. The cysts were recovered from the spiked water using membrane
filtration techniques. Giardia cysts present in the membrane retentate were
enumerated with a henracyiometer and a Coulter Counter.
Currently Giardia lamblia is the most commonly identified pathogen in
waterborne outbreaks in the U.S. and the protozoan is especially predominant
in the Pacific Northwest, Rocky Mountain states and New England.
CHARACTERISTICS OF ORGANISM
Giardia lamblia is a pathogenic intestinal parasite found in humans and
certain animals. The multiflagellated protozoa belong to phylum
Sarmastigojijora, subphylum Mastigophora, class Zcomastigophorasida, order
Diplomonadorida, family Hexamitidae, and subfamily Octomitinae. The
organism was first observed by Antony van Leeuwenhoek in 1681 while studying
his own feces (Dobell, 1932). During the mid and latter part of the 19th
century, the organism was observed and studied by many workers. The genus
was named by Joseph Kunstler in 1882, but until Charles Wardell Stiles
established the name Gj.ardia lamblia in a lettei to Kofoid and Christiansen
(Kofoid and Christiansen, 1915), the organism had been synonymously known as
Giardia intestinal is. qj.ardia duodenalisf or Giardia enterica.
The organism has two stages in its life cycle: the reproductive
trophozoite stage and the dormant cyst stage. The trophozoite is
pear-shaped with a broad anterior end that comes to a blunt point
-------
posteriorly. The dorsal surface is convex, whereas the ventral surface,
•which contains a large sucking disk, is somewhat concave. Another name for
the sucking disk is the striated disk, because of the striated appearance of
the pellicle which is caused by its alternating light and dense lines. The
trophozoite is 9 to 21 urn long, 5 to 15 urn wide, and 2 to 4 urn thick. The
organism is bilaterally symmetrical with eight flagella. Its basal bodies
arise near the midline at the level of the two anterior vesicular nuclei.
TWo of the flagella emerge anterolatcrally, two posterolaterally, two
ventrally, and two caudally. The parasite has no true axostyle, as has been
previously reported. Rather, what has been observed is the intracytoplasmic
axonentes of the ventral flagella and the associated groups of microtubules.
TWo media bodies are composed of bundles of microtubules arranged either
irregularly or sometimes united in ribbons, Their function is obscure,
though it has been suggested that they may help support the posterior end of
the • organism, be involved in its energy metabolism, or have something to do
with formation of the new sucking disk. The trophozoites reproduce by
binary fission (Levine, 197<»). The ovoid- to ellipsoidal-shaped cyst of Q.
Iambiia is surrounded by a hyaline cyst wall approximately 0.3 urn thick and
composed of thin fibrous elements interspersed with fine particles
(Sheffield and Bjorvatn, 1977). The cyst is smaller than the trophozoite (8
to 12 x 7 to 10 urn). A peripherally situated lacunar system is separated
f ran the plasma membrane and cyst wall by a thin layer of cytoplasm. The
flagellae of the trophozoite are believed lost or reabsorbed upon
encystment. But the intracytoplasmic portions (axonemes) of at least six
flagellae are retained. Newly formed cysts have two nuclei, whereas mature
cysts have four. Although nuclei have been observed in close apposition,
none have been seen dividing. Exactly when division or doubling of the
other organelles takes place is uncertain. But during excystation, two
trophozoites emerge from each cyst.
CHARACTERISTICS OF DISEASE
JLu Iambi ia has been the most common pathogenic intestinal parasite in
the United States ever since the Centers for Disease Control (ly/9)
initiated the Intestinal Parasite Surveillance Report in January 1976. An
estimated 7 percent of the adult population harbor the parasite (Schultz,
1975). The intestinal disease caused by G*. lamblia is called giardiasis.
Symptoms of the disease appear from 2 to 3b days after exposure to the cyst.
in most cases, however, the incubation period is about 1 to 2 weeks. The
cyst is the only form cf the organism's two life stages infectious to man.
If ingested, the vegetative trophozoite will be destroyed during passage
through the early stages of the digestive system, whereas the cyst will
survive until it reaches the small intestine. The environmental conditions
there support the emergence of the trophozoites, which divide rapidly and
can build up to enormous numbers. A single diarrheic stool can contain 14
billion parasites, and a stool from a moderate infection may contain 300
million cysts (Chandler and Read, 1961).
-------
In nrost Giardia infections, the diagnosis can be nade by stool
examination. In some cases, it nay take nxsre than one speciren to confirm
the disease when using direct smear and concentration techniques, tfaen it
becomes necessary to e:anine norc than one stool specimen, the probability
for a positive identification will increase by e:anining stools on alternate
rather than consecutive days. Diagnosis also appears to be easier in early
acute infections rather than established ones. In the acute stage, stools
are frequently watery or loose and nay contain mostly trophozoites and few
cysts because of rapid bowel transit.
In a series of controlled e>:periments with prison volunteers, Rendtorff
(1954) studied different epidemic-logical problems of various human
intestinal protozoans, among them G. lanblia. One of the objectives of the
study was to establish the minimum number of cysts capable of producing an
infection, by feeding known numbers of cysts to the volunteers. Of the five
men who received only 1 cyst, no one became infected. Vhen the dosage was
increased to 10 cysts per person, both volunteers became positive, thus
indicating that the critical number for infection is somewhere between 1 and
10 cysts per person.
The acute stage of infection is manifested by a sudden onset of
explosive, watery, often foul smelling diarrhea, marked abdominal flatulence
and distention, foul gas, nausea, anorexia, and cramps, which are usually
upper or midepigastric. Less frequently there is vomiting, chills,
lew-grade fever, headache, and belchinq. The acute stage usually lasts only
3 to 4 days and is often not recognized at the time as being due to
giardiasis. In some cases, the acute stage may last for months, leading to
malabsorption, debility, and significant weight loss. This latter situation
appears to be more common in children than adults which perhaps explains why
giardiasis was formerly considered a disease of childhood.
Acute infections can develop into long-standing subacute or chronic
infections. The most coixion symptoms include intermittent mushy and foul
smelling stools, abdominal flatulence and distention, primary upper
intestinal cramps, nausea, anorexia, foul belching, heartburn, headache,
constipation, weight loss, and fatigue. The symptoms may either bo
persistent or recurrent and are usually milder than during the acute stage
of the infection. Although most individuals with giardiasis are
symptomatic, many are a5yr.ptomatic and may never become symptomatic. But
the potential exists in some for development of intermittent chronic
symptoms.
protozoan does not lyse or rupture host cells, but appears to feed
on mucous secretions. A dense coating of trophozoites on the intestinal
epithelium interferes with the absoriJtion of fats and other nutrients, which
can trigger the onset of disease. The gallbladder may become intected,
which can cause jaundice and colic. A few cases of urticaria have been
reported (irebster, 1958; Wolfe, 1979) , and erythema multiforme (Koncnenko,
1976) and arthritic symptoms (Goodbar, 1977) have been found associated with
giardiasis.
-------
Usually the parasite disappears spontaneously fror. tne infected
individual but, that may take from a few days to several months. Once a
person has recovered from giardiasis, there are indications tnat some
resistance to re-infection has developed. The degree of resistance may vary
among individuals, and there is some uncertainty as to whether it is of
permanent or temporary nature.
(tost of those infected with JL. Iambiia today aiTfc treated with drugs.
The most effective and commonly prescribed are quinacrinc (Atabrine) ana
metronidazol (Flagyl), but both have potential problems. Quinacrine may
cause serious toxic effects in a small percentage of those taking it,
including toxic psy-dissis, vomiting, fever, and exfoliative dermatitis.
Metronidazol is a suspected carcinogen and mutagen. Neither of the drugs has
been proven safe for use by pregnant women. If used at all during
pregnancy, they should be administered only to those wonen with severe
symptoms definitely attributable to giaruiasis where benefit is judged to
outweigh potential risk.
PREVALENCE OP ORGANISM
Although the incidence of giardiasis does vary from one area to
another, G. .lamblia is a cosmopolitan parasite. According to the public
health laboratories in the United States, the sta-.es with the largest
percentage of G. '-amblia positive stool specimens in 1978 (Centers for
Disease Control, 1979) were Arizona, Arkansas, California, and Washington.
Of the total number of stool specimens examined in these four states, more
than 8 % were positive for fi*. iambi ia. These figures do not necessarily
mean, however, that the same states top the list of baterborne outbreaks or
total number of reported cases.
Waterborre outbreaks of giardiasis have occurred primarily in the
mountainous areas of this country particularly in New England, the Pacific
Northwest, and the Rocky Mountains. Co] or ado has experienced more outbreaks
than any other state and this probably reflects increased surveillance ana
investigation. Another possible explanation for the higher incidence of
giardiasis in the mountainous areas is the general concept about water
quality. High mountain lakes and streams are assumed to be free from
pollution and, therefore, when used as domestic water supplies, chlorination
is usually the only treatment. Often the chlorine dosage is low and
adequace contact time is not always provided. The potential for JL. Imblia
to be present in the mountain regions, is increased by tne heavy
recreational usage in many of these areas. When considering the high
percentage of asymptomatic carriers in the adult population, there is a
possibility of direct human contamination ci the water or inoirect
contamination through cross-transmission to animals. In the lowland areas
the water source is known to be cont inated and appropriate treatment
facilities are built, establishing the bender necessary to protect the
public.
-------
Many isolated cases of giardiasis involve people who use the outdoors
for recreation or work and drink untreated water. Many conceive crystal
clear lakes and sparkling streams in the mountains with no permanent human
habitation as the ultimate in water purity. Little thought is given to the
often great potential for contamination of these waters by fellow users of
the area or by wildlife.
A study aimed at identifying animal reservoirs of G. Iambi ia in
Colorado and New Mexico (Davis and Hibler, 1979) found a significant number
of beaver, coyotes, cattle, cats and dogs infected with ciarflia. when
exposed to G, iambi ij> cysts of human origin, the majority of the beaver,
bighorn sheep, dogs, pronghorn deer, mule deer, and raccoons became
infected. Human volunteers and dogs ingesting cysts from a naturally
infected beaver and mule deer were shedding cysts within one to two weeks
after exposure, thus emphasizing the potential for cross-species
transmission of Ciardia.
Another study, to assess the prevalence of Ciardia infection in aquatic
mammals in Washington State (Frost et al., 1980), found a significant number
of positive beaver and muskrat. During the three year investigation, the
percentage of Ciardift-positive animals increased each year, reaching 19.0
percent for the beaver and 42.6 percent for the muskrat. The juvenile
beaver and muskrat showed a higher positivity than the adults and jL-jging by
the number of cysts excreted, the beaver had a higher level of infection
than the muskrat. Positive animals were found both in protected ana
nonprotected watersheds, suggesting ti.at pathogen-free surface waters may be
difficult to find.
ttie information on cyst survivability in water is limited. Working
with human volunteers, Rendtorff and Holt (1954) found the cysts to retain
their infectivity after 16 days of storage at 8 C. Davis and Hibler (1979)
successfully infected dogs with cysts that had been stored in tne
refrigerator for 21 days. Some of the earliest work on infectivity and
storage was dona by Fantham and Porter (1916). A female kitten was fed food
contaminated with £L. Iambi ia cysts from a stool specimen that had been kept
for 74 days. No information was given on how the stool specimen was stored.
After nine days, cysts were recovered from the cat feces and the animal.
si.owed signs of diarrhea. Boeck (1921) found fiiardia to be viable after 32
days when stored in distilled water at 12 to 20 C, and at least 66 days when
sealed under a cover slip on slides. The eosin-stain technique used by
early researchers including Boeck to determine viability, however, is of
questionable value.
GIARDIASIS OUTBREAKS
During the period 1971 to 1978 a total of 24 outbreaks of waterborne
giardiasis were reported, affecting more than 7,000 persons. Although
reporting has generally improved in recent years, more waterborne outbreaks
occur than are reported. The majority of these outbreaks were caused by the
drinking of untreated surface water or surface water in which chlorination
w«is the only treatment. Only a few involved filtered water.
-------
One of the first outbreaks o- c-cstrcor.teritis in wliich G. lanblia was
irplicated as the probable ociolcxjic:.! agent, cxxurred in Portland, Oregon,
frcn October 1954 to Parch 1555 (Veazie ct al, 197?). The Oregon State
Board of Health estimated that at least 50,000 cases occurred during thct
period. I!uch controversy concer- ...g the pathogenicity of G. Iambi la existed
at that tine. In an effort to pirpont the cau^e of the illness a survey was
made of a group of people, most of whom were symptomatic. The
bacteriological studies revealed no enteric pathogens and the incidence of
intestinal protozoa other than G. Iambi ia ciid not differ from what had been
found in similar groups in the past. However, there was an abnormally high
prevalence of Giardia infection. 1*je flagellate was found in 44% of those
studied during the outbreak, in contrast to 73 of those e::amined during
nonepidenic periods. The source and mode of spread were never
satisfactorily detemined, but the water supply could well nave been
involved. Heavy rains with a subsequent increase in water turbidity was
reported during the period of the outbreak.
Ttie first waterborne outbreak of giardiacis documented in the U.S.
occurred at Aspen, Colorado, during December 1965 through January 1966
(lloore et al, 156?). A survey of 1,094 skiers who had vacationed in Aspen
during the two months showed that at least 123 hed developed syrptoms
characteristic of giardiasis. Ohe city received appror.iinately half of its
water fron a distant mountain creek c.nd half fron three wells, roth sources
were chlorinated, but colifom contamination had been noticed intermittently
during the x/intcr. A survey of the sparsely populated creek area revealed
no obvious possibility of sewage contani nation. However, tracers placed in
the Aspen scv/erage system were detected in two of the three wells. ATI
engineering evaluation discovered leaking sewer mains near the wells and £.
Iambi ia cysts were isolated from the sex/age in these lines. A parasitologic
survey of Aspen residents detected only a mociest level of Gicrdia infection.
H:C largest outbreak of giardiasis and the first where a Gj. lacblia
cyst v/as recovered from the municipal water supply, occurred in none, I'ew
York, during November 1974 to June 1975 (Shaw et al, 1977) . It was also the
first time that water from an outbreak had been sham to infect laboratory
animals. A total of 350 residents had laboratory-confimed giorciasis and
an epidomiologic study estimated that more than 5,300 persons may have been
symptomatic.
first sign of an epidemic surfaced in early January 1975 when £.
Iambi ia v/as identified in stool spccir.cns from eight of 23 persons in Home
with gastroenteritis. Since early Itovcnber, however, local health
department personnel had been investigating an increased incidence of
diarrhea. IXiring this investigation, C. Irr.iblia was the only pathogen
commonly identified. A random household survey in the city indicated an
overall attack rate of 10.6%. !to correlation was found between illness and
daily activity, animal contact, or consumption of food, but a significant
association was discovered between having giardiasis and using water from
the rity system as opposed to using water fron private wells.
-------
Rone used a surface water s TCC located several miles to the north of
the city. From the intake at Fi- . Creek the water was piped to a reservoir
where chlorine and airnonia were added, first at the inlet and then at the
outlet as the water entered the distribution system. Ho other treatnent was
provided. Ihe water ct most sampling points in the distribution system was
negative for coliforms but the total bacteria count was high, indicating an
inadequate disinfection. In an attempt to isolate G. laablia fron the
municipal water supply* raw water was filtered through a small pressure
filter. At the end of each filter run the sand filter wan backwashed. The
backwash water was collected, coagulated and flocculated and the flee
allowed to settle. The sediment was used fcr microscopic examination or
aliquots were fed to pathogen-free dogs. In one of the sediment sanples
examined microscopically one JL. Iambiia cyst was found. Further evidence
of contamination of the raw water supply by the parasite was obtained when
G. Iambiia was found in some of th«2 dogs.
The source of infection was never established. However, the watershed
was found to be more hecvily populated than city officials expected and
there were some questions about the sanitary disposal procedures at sor.ie of
the settlements. Ito animal survey was conducted to assess the potential
for contamination by wildlife.
Until 197G, all the reported outbreaks of giardiasis in which municipal
\rater supplies were implicated, had involved surface water with chlonnation
as the only treatnent. In late April and early ttey of 1976, local
physicians in Canas, Washington, reported the occurrence of approximately 25
cases of giardiasis. This became the first reported outbreak involving a
filtered water supply (Kirner et al., 1978). The epidemiological
investigation that followed, showed that approximately 600 people had
clLaical signs of the infection.
The city of Camas used both surface water and deep well water as
sources of supply. The surface water sources were Boulder Creek and Jones
Creek which came from adjoining watersheds. Both sources were generally of
excellent quality including la? turbidity based on existing standards. Fran
the intake, the water flowed by gravity to a direct filtration systcn.
Unlike most direct filtration pJants, the injection of pretreataent
cherticals occurred immediately prior to the fa-/o multimedia pressure filters.
Chlorine was added in a transmission main about 1.5 hours upstrean of the
water treatment plant. Chlorine was not added to the filtered water except
during three separate failures of the upstream chlorination equipment. The
seven wells were primarily used to augment the surface water supply during
periods of high doraand or when the flow in Boulder and Jones Creeks was low.
As a safety measure the well water v:as chlorinated, but no additional
treatment was necessary.
Fost of the confirmed: giardiasis cases initially reported were located
in areas of the comunity most likely to receive surface water. Itence, the
surface water system was cuspected of being the source of the G. laiablia
cysts. A survey of the watersheds indicated no human habitation and most of
-------
the roads were found to be in very poor condition making access difficult.
No obvious source of contamination was observed but signs of beaver activity
were in evidence. With the help of professional trappers a total of seven
beavers were trapped in the watersheds, three of which were found to be
infected with £ Iambi ia.
Of the treated water samples collected for bacteriological examination
during the outbreak, only one was unsatisfactory. However, G. Iambi ia cysts
were recovered from both the raw and treated water at different locations
and times. An inspection of the water treatment plant in search of clues
that might explain how cysts could escape into the distribution system
revealed a cross connection between raw and filtered water in the coagulant
feed line. Further, there had been loss of media in both filters and thc-
coarse garnet had regions of mounding which could cause short circuiting.
The effectiveness of the coagulation process was questioned because of
insufficient control of the coagulant feed rate and the short detention time
prior to filtration. A subsequent analysis of the filtration process using
a particle counter indicated a 75% removal of particles in a 7 to ?.S micron
size range which incorporates the size of a JL. Iambi ia cyst.
On three different occasions during the month of April, the
chlorination equipment on the raw water main had been out of service due to
mechanical difficulties. During that time the chlorination was performed
manually, but after review of the emergency chlorination, procedures, it was
concluded that large amounts of water arrived at the treatment plant without
adequate chlorination. The time differential between the chlorination
equipment failures and the majority of detected giaruiasis cases correlated
closely to the incubation period for the disease. Even so, the chlorination
equipment failures cannot explain all the cases since the earliest signs of
the outbreak were evidenced prior to the first breakdown at the chlorination
plant.
Die second outbreak of giardiasis to involve a filtered surface water
supply occurred in Berlin, New Hampshire, in the spring of 1977. In a two
week period in early April, 100 cases of G. Iambi la infection were
diagnosed. By the time th^ outbreak subsided in the middle of May,
estimates based on subsamples of persons in community-wide surveys indicated
that 3,450 people had experienced gastronintestinal illness, 1,656 of which
were symptomatic for giardiasis (Lopez et al.r 1980). Among the remaining
segment of the population exhibiting no signs of gastrointeritis, an
estimated 5,197 people had asymptomatic G. Iambiia infection. The first 100
confirmed cases of giardiasis were randomly distributed in the city. Since
a preliminary analysis revealed no events or meals common to these cases, a
waterborne epidemic was suspected.
Berlin had two independent sources of water, the Upper Ammonoosuc River
and the Androscoggin river. The watersheds of the two rivers had no known
large point sources or discharges. Hunting, fishing and other forms of
recreation were permitted but no public sanitary facilities were available
on the upper Ammonoosuc watershed. Water from the two sources was treated
-------
separately and supplied to identifiable areas within the distribution
system, although seme areas received a mixture. Die older of the treatment
plants, receiving water from the Upper Ammonoosuc River, consisted of eight
pressure filters^ No provision was made for chemical pretreatment, and
turbidity monitoring equipment was not available. The water was chlorinated
prior to distribution. The Androscoggin plant was put in service just
before the outbreak, replacing an older filtration plant. The new plant
provided conventional treatment including chemical coagulation,
clarification, rapid sand filtration, and chlorination.
{L. Iambiia cysts were first identified in the Berlin water system by
the Androscoggin Valley Hospital Laboratory (Lippy, 1978). Water drawn from
a laboratory tap was passed through an improvized gauze filter overnight and
the filter material was found positive for cysts when examined
microscopically. Samples of raw and finished water at both treatment plants
and of water collected from the distribution system were also G. lamblia
positive. A survey conducted in the Ammanoosuc watershed to determine the
source of contamination disclosed a beaver lodge upstream of the treatment
plant intake. Four beavers were eventually trapped, but only one had £.
lamblia infection. Since there was ample opportunity for human fecal
contamination of the raw water, it could not be determined whether this
animal was an unlucky victim of water contaminated with &. lamblia of human
origin or whether the beaver served as a major contributing source of the
organism in the water. A similar survey of the Androscoggin watershed was
not seriously considered because of its large size and thus the source of
the G. lamblia cysts at the Androscoggin treatment plant was never
determined. However, because of the recreational activities in the area the
human aspect could not he completely ruled out. Furthermore, residential
sewage disposal violations were known to occur along the upstream portion of
the river.
The operation at both plants was studied to develop remedial action
that would prevent £L lamblia cysts from passing through the treatment
process. At the Ammonoosuc plant, no chemicals were used to condition the
water prior to filtration which made cyst passage through the filters very
likely. Mudballs and mounding of the filter medium in some of the filters
further impaired the efficiency of the filtration process. The chlorine
dosage and contact time were inadequate to inactivate the G. lamblia cysts.
The Androscoggin plant had experienced some floe carry-over to the
filters, but this was not considered a serious problem, It was discovered,
however, that air bubbles were escaping from the joints in the slab of the
backwash channels during air scour of the filters. The escape of air
through the joints indicated the possibility for raw water to seep through
the joints during filtration and to contaminate the filtered water. The
possibility was confirmed by a static hydraulic test of the backwash
channel. It showed that over 3% of the plant output was not filtered.
Another outbreak of giardiasis involving filtered surface water
occurred at Leavenworth, Washington, from January through Nay 1980. A
-------
survey conducted in early May indicated that as many as 600 people might
have been affected. Among the city's water customers 27% in the group
surveyed had experienced diarrhea with symptoms characteristic of giardiasis
(Austin and Barter, 1980). For people on private wells the incidence was
only 3%f and each of the persons with Giardia infection had been exposed to
Leavenworth water either through restaurants or work.
A source of supply for Leavenworth is the Icicle River. Raw water
turbidity is normally less than 0.5 NIU. There was no permanent human
habitation above the water intake, but the watershed was open for recreation
witfi several Forest Service campgrounds located on the river. However,
during the time period of the outbreak these camps were not likely to have
been inhabited and the sewage disposal for the camps was contained and not
likely to contaminate the river. From the intake structure the water flowed
by gravity to a direct filtration plant. The plant was designed for
chemical addition, coagulation, and filtration, but because of the low raw
water turbidity no chemical pretreatment had ever been practiced.
Surveillance and filtering activities were conducted at tne water
treatment plant. G. Iambiia cysts were recovered from the filtered water.
This implicated the water supply as responsible for the outbreak. The
actual source of the contamination was never determined. According to
personnel at the Forest Service ranger station, there were many good beaver
habitats at higher elevations, but signs of beaver activity in the area had
not been reported. Ihe inspection at the treatment plant also revealed a
significant loss of filter media which required that all four filters be
rebuilt.
10
-------
SECTION 2
CONCLUSIONS
The study demonstrated that to achieve cyst removal efficiencies
greater than 95%, unit process operation must be optimized. Process
variables of primary importance were coagulant type, coagulant dosage, and
the pH of the raw water when alum was used as the primary coagulant. Raw
water turbidity, filtration rate, and sudden changes in plant throughput
were shown to be of secondary importance. Raw water temperature was a key
variable when temperatures were <5° c. Cold water temperatures slowed the
rate of -alum floe formation which resulted in a significant amount of floe
forming throughout the depth of the filter and effluent piping.
During field operation with the LSEPA mobile pilot plant, the effect of
low water temperatures on the coagulation and flocculation process was
particularly noticeable when using alum, but also with polymers. It was
felt that if the detention time in the flocculator had been longer, higher
removal efficiencies may have been possible.
The polymers tested as primary coagulants did not perform as well as
alum. The removal efficiency was generally 10% less for turbidity and 11%
less for cyst-sized particles. On the other hand, the filter runs were
longer and the necessity for close pH monitoring experienced during alum
treatment was not required.
The role of the operator is critical in order to optimize unit process
operation. Sudden changes in the raw water may require immediate
adjustments of chemical feed and pH as was demonstrated during field
operation. Under such conditions, good plant records become important.
Jar tests were initially used to obtain information on optimum plant
operating conditions. However, because the pilot plants were operated as
direct filtration plants, the data from the jar tests were found to be of
limited value since the tests provided no information about tfte
filterability of the floe. Rather than relying on the jar test, optimum
operating conditions could be determined quickly and reliably by stepwise
changing the major process variables one at a time while monitoring the
filtered water quality. This approach was also used successfully to provide
important information to the full scale plant during concisions of rapidly
changing raw water quality.
11
-------
Diatonaceous earth (DE) filtration is an effective method for removing
G. Iambiia cysts. With a precoat of 1.0 kg/m , more than 99.35% of the
cysts added were removed at the beginning of the filter run. As the
thickness of the filter cake increased the removj3. ranged from 99.61 to
99.96%. The efficiency of the DE filter can be unproved by the addition of
a nonionic polymer, added with the body fead. A 0.0075 mg/L dosage of
Magnifloc 985N increased the cyst removal from 99.94 to 99.99%. Larger
dosages would reduce the length of the filter run.
Information obtained from pilot plant work such as this can be a
valuable aid in improving a full-scale plant operation. Equally important,
it can be a tool to gather information useful in the design of water
treatment facilities.
12
-------
SECTION 3
EXPERIMENTAL PROCEDURES
COLLECTION AMD ENUMERATION OF CIARDIA CYSTS
Experimental procedures only are described in this section. The
results of these experimental procedures may be found in Section 3, Results.
Collecting Giardia Qysts from Human Fftces frcm Q'ja.rflJcIf?''5; Patients
The fecal samples were received from the State of Washington
Parasitology Laboratory, Department of Social and Health Services. The
cysts were separated from the human feces using a sucrose gradient technique
as modified from the one used by Sheffield and Bjorvath (1977). These feces
had been examined and confirmed for presence of Giardia cysts by Ms. Yvonne
Fichteneau. The stools had been preserved by the addition of 5% formalin to
inactivate pathogenic bacteria.
To isolate the cysts, the feces were emulsified in approximately 20 mL
distilled water and were passed through 3 layers of gauze (60 to 100 urn mesh
equivalent). The procedure is outlined in Figure 1. The filtrate was
subsequently centrifuged at 1400 rpn for 3 min. The supernatant was poured
off and the sediment was resuspended in 5 mL distilled water. This
suspension was pipetted onto a discontinuous density gradient of 5 mL each
of 1.5 M, 1.0 M, 0.75 M and 0.50 M sucrose in a conical centrifuge tube
followed by centrifugation at 2200 rpm for 30 min. The cysts weie then
collected from the H_0 - 0.5 M sucrose and 0.5 M-0.75 M sucrose interface,
by means of a capillary pipette. In our procedure, microscopic examination
of the cysts showed the absence of any extraneous debris ana further
filtration through 20 urn and 5 urn filters was not necessary. The cysts in
this sucrose solution were then diluted with distilled water to 1 L and were
kept at 4 C. The technique was gradually modified as listed in Figure 2.
Cyst Recovery Using the 47 mm Filter Technique
Aliquots of 10 mL and 20 mL of distilled, tap end Lake Union water
samples were spiked with known concentrations of Giardia cysts as determined
by hemacytoneter. The aliquots were subsequently filtered through 47 mm
diameter 5 urn pore size membrane filters to isolate the cysts.
13
-------
Preserve stools In 5 ml in SZ
buffered formalin
Examine a feces smear to
ascertain presence of cysts
Nix 1.0 g of feces with 40 mL
distilled Mater
1
Pour suspension through
sieve with 60-100 un mesh
or three layer of gsuze
Centrifuge filtrate at
1400 rpm for 3 min.
1
Discard supernatant, add 5 mL
distilled water, and mix to
form suspensions
I
Prepare discontinuous
sucrose density gradient
of 5 n: layers of 1.5, 1.0.
0.75. 0.5 H sucrose In 40 mL
conical centrifuge tube, add
5 ml suspension on top
i
Centrifuge at 2500 rpm for
30 mm.; cysts collected
at Mater/0.5K interface and
0.5/0.75 H Interface are
removed with capillary
pi pet; suspension is diluted
to 1 I. for stock solution
Figure 1. Sucrose gradient technique to recover cysts from stool specimens,
14
-------
Examine a feces smear to ascertain
presence Oi" cysts.
Transfer feces to 2000 nL beater. Add
distilled water to 500 nLand homogenize.
Pour suspension through 3 or 4 layers of
cheesecloth. Transfer liquid to 250 mL round
centrifuge bottles and centrifuge at 650 xg
(2100 rpm) for 2 mm.
Aspirate off supernatant and resuspend sediments
with distilled water. Pour into four 50 mL
centrifuge tubes, luke volume to 45 ml in each
tube.
Mash by ccntrifuging at 650 xg (2100 rpm) for
2 nun. Discard supernatant and resuspend
sediments. Repeat using 2 drops Dawn
dishwashing liquid per 15 mLof suspension.
Wash with distilled water until s.-pernatant
is reasonable clear.
Resuspend sediments in 25 mL distilled water
and layer onto 25 mL. 1.0 M sucrose, in two
50 mL centrifuge tubes. Centrifuge at 800 xg
(2400 rpm) for 1C •nin.
Aspirate off 3/4 of supernatant apsearing above
the oa-.d of cysts. Pour remainder of suspension
into bO mL centrifuge tube and centrifuge at
BOO xg (2430 rpm) for 2 mm.
Aspirate off supernatant and resuspend cysts et
the bottom and repeat washing procedure t*ict.
Final sediments are resuspended and stored at 4°C.
Figure 2. Modified sucrose gradient technique to recover cysts.
15
-------
Elution of the cysl:s from each neribrane filter was accomplished by
placing the filter with cysts in a small flask, adding equivalent amounts of
distilled water and electrolyte totaling initial volume of water that was
filtered. The flask was then gently agitated so that a flow of water passed
over the membrane's upper surface.
The effluent was examined for presence of ciardia by direct examination
of effluent after centrifugation and by use of the Coulter Counter.
ethodoloo for Csr
Two counting techniques were used for the enumeration of the Giardia
jamblia cysts. The first technique was microscopic counting using different
counting chambers/ and the second was an electric current displacement
techrique using a ZBI Coulter Counter and Channelyzer (Coulter Electronics,
Hialeah, FL) calibrated to measure particle densities in the Giardia size
range (8 to 12 urn) . The instrument measured the reduction in current
between two sides of a small orifiue before and during the time that a small
particle passes through the opening. The air rent reduction was proportional
to particle volume.
A day Adams Model 4011 Spencer Bright-Line Counting Chamber was used
for the microscopic counting. Three alternative means for counting the
cysts wert: used, depending upon their density in the solution. The cysts
were counted _n a volume of 0.02 mm for suspected high densities or when £.
^amblia cyst counts exceeded 20,000 cys^s/mL. The multiplication factor is
50,000 times the number of cysts counted to give cysts per raL. The cysts
were counted in a volume of 0.1 rmr for moderate densities in the range of
5,000 to 20,000 cysts/mL. The multiplication factor is 10,000 times the
nixnber of cysts-courted to give cysts per mL. The cysts were counted in a
volume of 0.9 nnr primarily for low densities. The multiplication factor is
1111 times the number of cysts counted to give cysts per mL; i.e., less than
5,000 cysts per mL. The specimens were stained with 5ft Lugol's Iodine prior
to pipetting into the hemacytometf.'r. A low detection limit in a large
amount cf water was accomplished by concentrating the cysts and particles by
passing the water through a membrane filter and resuspending cysts and
particles in & small volume.
The electric current displacement method used a Coulter Electronics
Coulter Counter particle counter, irodel ZBI with a 100 urn aperture tube. A
0.9% I sot on solution was used roth as diluting medium and electrolyte
solution to allow flow of the electric current. In this method, the counts
were based on the current interruption when a particle passed through the
aperture, and were made per 0.5 nt, of cyst-containing solution. A size
frequency distribution Channelyzer coupled to the Counter was used to verify
the counts of particles in the same size range as ciardia. This counting
technique was not specific for ciardta cyots.
16
-------
The cyst-sized particles in the water samples were enumerated with the
Coulter Counter. Replicate counts were made, and Deans were calculated and
compared to the mean values of the initial concentrations.
Recovery Usin the 293 it Membrane Filter
Stock suspensions of 10 to 10 cysts/mL of Ciardia lamfalia cysts were
prepared and stored at 4 C.
Appropriate volumes of these stock suspensions were then added to 10 L
of distilled water to give different final concentrations. Nitrogen gas was
used to pressurize the stainless steel vessel to pass the water through the
filter unit (5 urn pore size) at 10 psi (Figure 3) . The time to filter 10 L
through the membrane averaged 2.5 minutes.
The filter was then removed and carefully placed in a shallow plastic
container of slightly larger diameter than the filter. Distilled water (0.5
L) was then added and the entire assembly agitated by means of a small
shaker for 3 min (Figure 4) .
A two-step concentrating method was selected to achieve a final volume
of about 5 to 15 irL. This involved spinning of the 0.5 L wash in a
centrifuge for 20 min at 1,000 rpa, about 20 mL of the centrate in each of
four tubes. Respinning of the 80 mL in two conical tubes under the same
conditions and collecting and combining the two 5 to 7 mL sediments
permitted the sample to be enumerated by the Coulter technique. The
technique was gradually modified to that listed in Figure 5.
The Milllpore Pell icon Cassette Unit was a multiple surface area
cassette filter. The unit contained nine 465 CRT (0.5 ft ) Miilipore
membranes with a pore size of 1.2 urn stacked on top of each other, separated
by plates of acrylic plastic. The filtration rate was 0.07 m/hr (0.01
gpo/ft ). The design of the unit was such that a retentate recirculated by
a peristaltic pump with a retentate/filtrate ratio of 1:2 enabled a
continuous flow of water over the filter area. Particles greater than 1.2
urn in diameter remained in the retentate, the volume of which was readily
controlled.
The Hillipore Pell icon Cassette Unit was tested at the Ryderwooc,
Washington, reservoir in which beaver -3 had been sighted ana C. lanbiia
cysts had been detected in the upstream sediment. The unit was hooked UL- to
the raw water intake inside the treatment plant with chlorination as the
only preceding treatment. Over a three hour period, 93 L were passed
through the unit with a retentate/filtrate ratio of 1:2. 100 mL of the
retentate and wash were kept and examined for G. l«nb\|p cysts by the
following procedure. The oop"*ntratlon step represents a 98% volume
reduction.
17
-------
Figure 3. Schematic of 293mm Mllllpore Filter Unit used to recover G. lanblla
cysts from water. ~
18
-------
10 liter* dictllled water
-•uipendon of ey§t» added
Pea* through 293 ta (3.0 fts pore elie) Kllllport
Filter et 10 pet with Nitrogen (•*.
Filter reewved aod cysti wathed off
t>r 4|ltatloa la 0.5 1. HO.
Centrifuf* retent«:e 9 1.009 RTf for
20 alo. lo four 125 «i-lub«».
Ucalo "•ediocat" (appron- 20 oL « 4).
Tranifer to two 30 oL conical tubes and
re-cmtrifuge.
Retain "acdiaent" (approi: 3 aL > 2).
Figjre 4. Procedure for recovery of G. lamb Ha cysts wUh the 293nm
MIHIpore filter. ~
19
-------
Pass suspension through 293 «at (5.0 ua pore
*iie) Nuclepore Filter at 10 psi with
nitrogen gas. 0.2 ua filter on nitrogen tank.
Reaove filter and place in shallow dish.
Cysts washed off b> agitation of filter in
250 ml HjO for 3 min. (platform shaker.
Tooth-aster Cocpany. Racine, Wisconsin).
Reaove filter and rinse thoroughly.
*
Centrifuge retentate at 350 xg (1500 RPM)
for 10 min. in eight 50 mi conical bottom
tubes.
Aspirate off supernatant from each tul-e to
8 B! final volume. Transfer renaming volune
to two SO o£ centrifuge tubes and recentrifuge
Aspirate off supernatant from each tube to 5 ai
final volume. Transfer remaining volume to one
15 ml centrifuge tube and centrifuge at 350 xg
(1500 RPM) for 10 min.
4
Aspirate off supernatant to 1 nt final volume.
Figure 5. Procedure for recovery of cysts from dilute water suspensions,
20
-------
1. 50 mL of retentate centrifuged at 2000 rpm for 10 min.
2. Approximately 5 mL of sediment was passed through a discontinuous
sucrose gradient (Sheffield and Bjorvatn, 1977) .
3. The volumes of water between the HJ3 - 0.5 M layer and 0.5 - 0.75 M
layer were pipetted and examined under a microscope at 280 X for Giardia
cysts.
4. Ihe above procedure was also used for 50 mL of the wash from the
filter unit.
rohoretic Mobil it
Tests for electrophoretic mobility (EM) and zeta potential (ZP) were
carried out to determine how it varied for formalin fixed Giardia cysts at
different pH values using a Zeta Meter. The experiments were conducted
using a plexiglass Riddick type II electrophoresi s cell (Zeta Meter, Inc.,
New York, NY) with a 4.4 mm diameter cell tube and cell constant of 62. The
cell had a platinum-irridium anode and cathode. Measurements were made at a
distance of 0.147 diameters from the tube wall, which is the distance of no
electro-osmotic fluid-flow. The voltage used for the experiments varied from
200 volts for a 0 to 300 micromho/cm to 50 volts for 700 to 1500 micromho/cm
suspensions. Solutions of higher conductivity experienced more rapid
thermal overturn due to heating of the solutions and the tube contents had
to be replaced more often. A total of 10 or more individual cysts was
measured in each batch with regard to their travel distance in the cell
tube. As the distance between the electrodes was 10 cm, the voltage decline
ranged f ran 20 to 5 volts/an. The EM is calculated as
EM = Cyst Travel Distance/Tiroe Interval divided by Volt/Electrode
Distance
with the units of um/sec/volt/sec. The present study used a 98 urn tracking
distance for each cyst. The EM was converted to ZP (M volts) using the
Helmholtz-Smoluchowski multiplication factor expressed as:
ZP = EM-4TT
(Vfc = vixcosity of liquid; Dfc = dielectric constant of liquid;
f (K ) = Henry relaxation correction)
a
The pH was measured with a Model 5 Corning pH meter (Corning, Glass,
Corning, NY), standardized daily with pH buffer, while the turbidity was
measured with a continuously rt cor ding low range Hach 1720A turbidimeter
(Hach Chemical Co., Loveland, CO .ind standardized daily as sugge"-.ed in the
manual. To verify the readings of the flow-through turbidimeters, grab
21
-------
samples of the influent and effluent were analyzed daily on a DRT-100 (H.F.
Instruments, Ft. Meyers, PL) bench top turbidimeter.
DESIGN AND TESTING OF THE 2.3 I/MIN (0.6 GIN) WATER TREATMENT PILOT PLANTS
The University of Washington pilot plant used chlorinated unfiltered
Seattle tap water as "raw" water. In addition, it could be supplied by Lake
Union water pumped frcm the lake adjacent to the campus. It was constructed
from 3/4 in plywood, coated with fiberglass. The unit consisted of three
individual but identical treatment plants, each designed for 2.3 L/min (0.6
gpjn). This design allowed the plant operator to vary the capacity of the
pilot plant and thereby the surface loading on the filters whils keeping
flow conditions and all the design factors identical at all times (Figure
6).
The pilot plant was 137 cm (54 in) wide, 229 on (90 in) long, and 122
cm (48 in) high. Physical dimensions of the individual units and design
factors are given below.
Rapid Mix
The dimensions were 15.2 cm by 15.2 on (6 in) with a maximum water
depth of 22.2 en (8.75 in). At maximum depth the theoretical detention time
was 2.3 min at 2.3 L/min flowrate A variable speed,impeller produced
G-values ranging from approximately 300 sec to 1000 sec .
Flocculators
The flocculator consisted of three compartments, each 21.6 cm (8.5 in)
by 22.9 cm (9 in) and 35.6 cm (14 in) deep, measured from the overflow weir,
producing a 23.2 min theoretical detention time. Calculated G-values
ranging from 30 sec to 150 sec could be attained by changing the speed
and surface area of the paddle blades making tapered flocculation possible.
Sedimentation Basins
The sedimentation basins were 45.7 cm (18 in) wide, 182.9 cm (72 in)
long, and 91.4 en (36 in) deep, measured from the overflow weir. With the
2.3 L/min (0.6 gpn) design flow rate, the theoretical detention time was 5.6
hr and the surface loading 3.9 m/d (96 gpd/ft ). The tanks were designed
with a baffled inlet zone and could be shortened to decrease detention time
and increase surface loading by means of a divider wall.
Filter Columns
The study used two 10.8 cm (4.25 in) diameter plexiglass filter columns
fitted with Turbitrol PC Media consisting of 50.8 on of 0.92 mm effective
size anthracite (UO1.28) and 25.4 cm of 0.40 mm effective size sand
(UO1.30). The columns had headless taps at 10.2 cm (4 in) intervals.
22
-------
CHEMICAL
FEEDPORTS
ii ii n ii
INFLUENT
RAPID
MIX
•si
X
X
X
_
X
{
o<
f
Jx
TT
X
X
X
— 1
SLOW
MIX
SEDIMENTATION
3AS IN
OVERFLOW
HEIR
MIXING
CHANNEL
DUAL MEDIA
FILTERS
TURBIDITY
METERS
II
EFFLUENT
Figure 6. Water treatment pilot plant at University of Washington.
23
-------
Conductivity tests were done to determine the retention times in the
rapid mix and flocculation tanks. Sufficient amouncs of Nad were added to
the tanks and conductivity was measured at fixed time intervals.
The retention time for the sedimentation tank was determined
spectrophotometrically using an inorganic blue dye to prevent salt
stratification. Samples were taken every 10 to 15 min and transmittance was
measured by a Bausch and Lonb Spectronic 20 spectrophotometer.
It seemed reasonable to assume that the rapid mix tank and to some
degree the flocculation tanks would behave as completely stirred reactors.
The sedimentation tank on the other hand, would most likely show a
combination of characteristics, some typical of a plug flow and others of a
completely mixed reactor.
The tracer was added in the same manner for all retention time tests.
A concentrated solution of tracer was metered into the line feeding water to
the tank being tested. This was continued until the tracer concentration in
the reactor had reached a constant j.evel, at which time the feed was
discontinued and the sampling of the reactor effluent initiated. Eftluent
samples were collected until practically all the tracer had been displaced
from the reactor.
To determine the actual retention times of the various processes the
concentration of tracer in the effluent was plotted vs. time on rectangular
coordinates. The resulting tracer decay curve was divided into segments of
equal time increments. The moment (time x concentration) of each segment
about the origin was then computed and the sum of moments was divided by the
sum of the concentrations to give the actual retention time.
Additional analyses were performed to obtain information on the flow
regime and general performance of the tanks. If F(t) is the fraction of
tracer retained in the tank for a duration less than tinrc t, then the
fraction remaining in the tank longer than tine t must be l-F(t). For a
single compartment, completely mixed reactor,
l-F(t)=e-(t/T)
where T is the theoretical retention time. A semilog plot of l-F(t) vs. t/T
for a completely mixed reactor would yield a straight line. Deviations from
a straight line, if any, could be used to provide information on the tank's
dead space ratio and plug-flow and mixed-flow fractions by employing the
relationship derived by Rebhun and Argaman (1965).
TESTING OF (EKSUIATIOJv'FILTBATION ANT DIRECT FILTRATION AT THE UNIVERSITY OF
WASHINGTON
24
-------
JAR TESTS
Jar tests using 3 L beakers and unfiltered Seattle tap water
(chlorinated Tolt Reservoir water) were done to determine the optimum alum
dosage during the water treatment pilot plan.*: run. The test was done at 100
rpai rapid mix for 2 min, followed by slow mix ac ?0 rpn for 20 min. pH was
adjusted with 0.1 N NaOH to 6.5. The jar test was done at two different
sets of alum dosages with settling tintr of 30 min and 60 min. The
turbidity of each sample was tested after settling,.
Continuous testing with Giardia spiking
Continuous runs with the coagulation/filtration pilot plant used alum
dosages of 10 mg/L with pH kept at 6.7 by addition of lime. The rapid mix
was run at 500 rpn while the slow mix v&s kept at 22 rpn (G=48 sec ) .
Figure 6 shows the cyst introduction and watrr sampling points.
raw water flow rate was 2.3 L/min (0.6 gpm) with a corresponding
loading rate of 4.9 m/hr (2 gpVfO . Using unfiltered chlorinated
The
filter
Seattle tap water, initial particulate concert rations of the "raw" tap water
and initial turbidity were also recorded pric.: to the start of the run.
Prior to sp.king of the influent Seattle tap water with a Giardia cyst
suspension, perticulates/mL in the cyst-size range were established.
Spiking with Giardia was carried out at different points in order to
establish removal efficiencies for each sequence of unit processes.
a) A single spike of cysts introduced .''.to the first flocculation
compartment of unit A.
b) A continuous dose of cysts introduced at the chemical feed port A.,
just ahead of entry into the rapid mix compartment.
c) A continuous dose of cysts bypassing flocculation-sedimentacion and
introduced into Filters B and C to determine losses in the filtration step
alone.
TESTING OF DIATOMACEOUS EARTH FILTER AT THE UNIVERSITY OF WASHINGTON
Diatomaceous Earth fPEl Filter Performance
The DE test filter was a 0.1 nr (1 ft ) pressure filter, operated at
3.8 I/min (1.0 gpm). A schematic of the filter system is -shown in Figure 7.
The operation of the filter consisted of three steps: precoating,
filtration, and filter cleaning.
25
-------
PUMP
FLOW METER
VALVE
PRECOAT /-, PRECOAT RECYCLE
PRECOAT TANK
-R-
II
FILTERED HATER
-B-
DE
FILTER
DRAIN
RAW WATER CYST BODY FEED
TANK SUSPENSION TANK
Figure 7. Schematic of the DE filter system.
26
-------
In the preooat tank a slurry was prepared by adding the desired amount
of diatonite to tap water. The slurry was recirculated through the filter
at high rate while keeping the contents of the tank well mixed. A gradual
buildup of diatonite on the filter septum could be observed and the water in
the tank finally became clear and free of diatomite.
While positive pressure was maintained in the filtration chamber, the
appropriate valves were oroned and closed to change from preooat mode to
filtration mode. During filtration, a small amount of diatomite body feed
was continuously added to th° raw water. This addition of fresh diatomite
to the preooat filter cake meant that layers of clean diatomite were
constantly rejuvenating the filter and thereby slowing down the headless
buildup due to particles plugging the filter cake pores. The thickness of
the filter cake steadily increased during the run.
Filter runs were terminated when headless exceeded 30 psi. The filter
cake was removed from the septum and the spent diatomite discharged to
waste. Septum and filtration chamber were carefully sluiced to make the
filter ready for a new precoating.
The initial work with the DC filter was aimed at determining the amount
of preooat required for adequate initial reduction of turbidity and
particles in the 8 to 12 urn range. Several different grades of diatcmite,
obtained from the Manville Products Corp., Denver, GO, were used.. The
amounts of precoat material applied to the septum ranged from 0.5 kg/m (0.1
lb/ftn to_1.2 kg/nT (0.24 Ib/fO. The results indicated that 1.0 kg/nT
(0.2 Ib/ft ) vvould be adequate for all grades of diatcmite, giving a 56 to
96% and 70 to 91% initial reduction in turbidity and cyst-sized particles,
respectively. For the very fine grades such as Standard Super-Gel and
Filter-Gel smaller quantities of precoat did not result in a significant
increase in turbidity or cyst-sized particles in the filtered water.
However, this finding was more of academic interest. Because of the
relatively high initial headloss, the finest grades wero not judged to be
good candidates for full scale water treatment applications.
The initial runs with the DE filter were made without the addition of
G. Iambiia cysts to the raw water. The runs were designed to gain knowledge
about the filter's performance with respect to particle and turbidity
removal Lor different grades ot diatomite. In addition, the amount of body
feed was varied from 10 to 40 mg/L to investigate its effect on the rate of
headloss buildup across the filter cake. Influent and effluent turbidities
«re monitored continuously with a Hach Model 1720A turbid iineter (Hach
Company, Loveland. CO), while particle analyses were performed on influent
and effluent grab samples
-------
Diator.iaceous Earth (DEI Filtration with (Jiardia lariblia cysts
The cysts used for the DE rtTis were oitrectcd from stool specinens as
described earlier. The concentre.ticn of the stock solution ranged from 1.0
x 10 to 4.6 x 10J cysts/mL, and was stored at 4 C until needed.
Cysts were added to the raw water at the samo location as the body
feed, cither as a slug or., as a constant continuous dosage. Ihe slug
contained a total of 3.0 x 10 cysts, added in 10 sec using a Fin Lab Pump
(Fluid Ifetering, Inc., Oyster Bay, NY). For the continuous cyet addition,
the parasite was metered into the raw water line with a Buchler Polystaitic
Pump, Itodcl 2-6100 (Budiler Instruments, Inc., Fort Lee, MJ). Different rav/
water cyst concentrations were used during these runs, ranging from 1.5 x
10* to 9.0 x 105 cysts/L.
The filter effluent sampling schedule was determined from a series of
tests in which a salt solution was added to the raw water in place of cysts.
Ihe conductivity of the filter effluent was nonicored continuously to
determine: 1) how fast a 10 sec slug would pass through the filter, and 2)
the time required to reach a constant effluent concentration when a
continuous dosage was added to the filter influent. It was found that the
entire slug would have reached the filter effluent in 10 nin. That neant,
in order to trap all the cysts escaping the filter, a 38 L sanple would need
to be collected. This was not an unreasonably large volume to process by
the technique developed for this study. When a constant dosage was added to
the raw water, the effluent concentration had attained its raximum and
constant level after 10 nin. By adding cysts for 15 min r.nd sampling the
filter effluent during *:he lart 5 min, a 19 L sanple containing an average
effluent concentration of cysts was collected.
Ml cyat runs used Hyflo Super-Gel as filter aid. Based on results
frcra preceding runs, without cyst addition, a 1.0 kg/p. (0.2 Ib/ft ) precoat
and 20 ng/L body feed was judged most suitable for the 3.8 L/min filtration
rate and raw water quality. During two of the four runs a 0.0075 mg/1
dosage of the nonionic polymer Ilagnifloc 985N was added to the raw water for
the duration of the run.
TESTING OF DISECT FILTRATION IN KOQUIAIi AID LEAVENKOKTH
The last part of the study was used to validate the laboratory results
in the field by using a mobile pilot plant. In addition, the pilot plant
was to be compared with the full-scale plant to determine any diccrepanciea.
All plants were intended to operate at conditions giving maximum cyst
reipovalr. The mobile pilot plant was tested at Kcquiam and Leaven»'orth,
Washington, by treating a portion of the raw water. A comparison was also
made between wrter Duality generated by the pilot plant and the drinking
water generated by tne city water plant at each location.
28
-------
Die tests were conducted with a USEPA pilot drinking water treatment
unit, the Waterboy-27 (Neptune Microfloc, Corvallis, Oregon) which was
modified by extending the depth of the sand filter compartment Dy 83.8 on
(33 in) to provide for more headless buildup and prevent negative pressures
within the filter as shown in Figure 8. The upper boundary of the filter
bed was dropped front 76.2 cm (30 in) to 124.5 cm (49 in) below the top of
the unit. At Hoquiam the water was tapped from the water main through an
unused chlorine injection port. Transportation of the raw water to the
pilot plant by a 7.6 cm (3 in) line was provided by the pressure of the
main. Water was pumped into the plant by two centrifugal pumps in series
able to deliver a maximum of 75.7 L/min (20 gpm). After injection of
chemicals the water was passed through three static in-line mixers, Model
2-50-541-5 (Kenics, Danver, MA), whereafter it entered the flocculator,
which provided for an 8 min detentio/i time at the cannon operating condition
of 62 L/min plant flow (4.1 gpra/ft in the filters). The water overflowed
ito the filter conpartment with a 78.7 cm (31 in) average water head above
the filter. The filters consisted of 45.7 cm (18 in) MS-4 anthracite (e.s.
1.0 to 1.1 mm, u.c. < 1.7), 22.9 cm (9 in) of MS-6 sand (e.s. 0.42 to 0.55
mm, u.c. < 1.8), 7.6 on (3 in) of MS-21 fine garnet (e.s. 0.18 to 0.28 rtrn,
u.c. < 2.3), 7.6 on (3 in) of MS-22 course garnet (e.s. 1 to 2 mm), 10.2 cm
(4 in) of 0.95 cm (3/8 in) gravel, and 12.7 can (5 in) of 1.9 cm (3/4 in)
gravel (Neptune Microfloc, Oorvallis, Oregon), the top support plate was
perforated with 0.63 cm (0.25 in) openings 5.1 cm (2 in) apart from center
to center to provide a total perforated area of 45.5 cm (0.049 ft ).
Support gravel was located below the top plate followed by the bottom
support plate perforated by 0.36 cm (0.14 in) openings 3.8 cm (1.5 in) apart
fron center to center, to provide a total perforated area of 47.4 cm (0.051
ft ). The filter effluent was then pimped to the 4731 L (1250 gal) backwash
water tank which had an overflow at the top. At the end of the filter run
the 246 L/min (65 gpm) backwash water pump delivered the 3690 L (975 gal)
effective liquid volume to the bottom of the sand filters at a rate of 67.2
cm/tain (16.5 gpm ft ) for 15 min. The backwashing resulted in a 23% bed
expansion which was less than the 50% expansion commonly used.
HOQUIAM WATER TREATMENT PLANT
Hoquiam is located in Grays Harbor County, Washington, approximately 21
kilometers (13 miles) from the Pacific Ocean and 80 kilometers -50 miles)
north of the Columbia River and the Oregon border. The city has two main
sources of water: Davis Creek and the west fork of the Hoquiam River. A
third source, the Little Hoquiam River, is used only in case of emergency
and bypasses the treatment plane.
The present water treatment plant at Hcquianuwas completed in 1975. It
was designed for a maximum flow of 11,400 m /day (3 mgd) anc serves a
population of approximately 10,500. It was a conventional plant providing
coagulation, flocculation, sedimentation and filtration. At maximum flow
the detention time in the flocculator was 6.5 min. From the flocculation
basin the water overflowed into the rectangular sedimentation basin which
provided 49 min retention at maximum flow. The clarified water was numped
to the three mixed media filters which were operated at a maximum filtration
29
-------
IN-L!Kt
(t
n
H
t
WPID
HJJtB
« f ~* *"*
rr'
i i
1 I
1 l
I 1
1 I
I i
1 1
ill
• i
s
I •
— ft
y
\
^fc *"** •
1
I
1
1
1
I
1 1
• «•
r-j---v^-r^:rra
tin ttfsiA rjLTia
1
•
T^W.I
$
AXTNRMlTt
JB'Of «-«
SCAD 9'0f ns-6
fltC C Gtn«i-1 i
COABLSC GtY.cT 3
SlLtCA QfQf J/fe
6-Of IS
^^ ^j
«•?;
Figure 8. Cross section of coagulation, flocculatlon and mixed media
filtration compartments of the Wattrboy-27.
30
-------
rate cf 12.2 nv/hr (5 gpo/ft ) . Backwash was initiated by loss of head
through the filters, when operating in automatic node. Cnce a backwash
cycle was initiated, it automatically backwagied each filter in sequence.
Half cf each filter was backvashed at a time, The backwacft rate was 36.6
n/hr (IS gpa/fO and the water was supplied iy clearwell pimps. The
filtered water was chlorinated in the clearwell and pooped to the city
reservoir.
The plant noraally used aim as the prinary coagulant, sooetines in
coBtoination with a nonionic polyser as a coagulant did. During periods of
low turbidity, however, only polymer was used as a coagulant. The sane
polyoer was also used as filter aid. Soda ash was used for pi! control
during coagulation and flocculation and for final pH adjustment in the
clearwell.
LBWIWOKZH WATS* TREA2&CT KANT
Lesvenwcrth is located at the eastern foothills of the Cascade Mountain
range near the Wenatchee River in Chelan County, Washington. The two
sources of water used by the city are Icicle Creek and shallow wells, with
Icicle Creek the laain source for the city's 2,400 residents. Because of the
dry climate, the per capita water usage is very high compared to the west
sice of the raoun tains.
The flow to the 13,300 niVday (3.5 ngd) direct filtration plant was
controlled by an electric butterfly valve operated by signals froa the
storage roeervcir just outside the city. Tho polymer used as coagulant was
added directly to the 30.5 CD (12 in) raw water line before it entered the
baffled flccculator. The retention tine at nwudmn flow was about 9 min.
From the flocculate* the water flowed via txto inlet tlune to the four mixed
nsdia filters operated at a rate of 12.5 n/hr (5.1
Unlike many other plants, the filter operation was controlled by
siphons. The siphoning waa initiated by applying vacuun and the siphon was
broken by allowing air to be sucked in, all of which was controlled by a
scries of oolenoid valves. The filters were backvaetted one at a time. As
the headless increased, tho water level above the filter media would rise
until it raekde contact with a sensor. At that time the inlet siphon would
be broken ana the bada»ash siphon initiated. The three filters remaining in
the filter node would supply the backwash water.
The filtered water was chlorinated in the clearvell ana flowed by
gravity to the storage reservoir . A booster puap woo available for use if
necesnary, when the plant was operated at high flow rates-
31
-------
SBCTIOCI <
RESULTS
CWLUWTIOI!: OJLLDCnC*!, EJUlHSATIOn OT niAPDIA CYSTS A!£> QVOC
Statistical tyflflSt'.ion of Cy~t ntunerr.tior^ *3ptApiqut?3
The linearity or proportionality of both counting methods \.-ao evaluated
by using a 1:10 diluted stock suspension containing appropriately 4x10
cysts/L as measured with a hcrucytorc'ter end Coulter Counter follo.:oc! by a
sequential 1:2 dilution to obtain ID*:, 5site direct' en. For c:xu-plo, bcncd on the 5% suspension v/hich liad a
count of 25,000 cycts/rL, the 8 tiir.es nore dilute ca*:penRion of 0.625S
should have had a calculated count of 3125 cyjts/rL. Ilic actucl count was
20-10 cycts/rL or only Ct>1 of the calculated anount, indicating that at low
cysts concentration the counting of the cysts in the acjuarec of *d»c snail
volurne of the counting char.ibcr ray nir.c actual cysts.
coefficient of correlation Ixttwecn the particle counts and the
Dilution percentage v/as 0.96 for the Coulter Counter end 0.90 fcr the
htJTVTcyborcotor. The ntcndard deviation using the Coulter Courtt-r, ha/evcr,
van rnjch lo^cr than that of the henncytcnetcr. ihile the average
coefficients o£ variation of the Coulter Counter was oj la; as 1.9fAt it v:ao
as high as 74.2% for the hcrccy nctcr indicating that the Coulter Counter
is r.»rc precise. In Figure 11, the coefficient of variation as related to
32
-------
DIWiETER (M1CPCXS)
78 9
90
eo
70
«0
20
10
, i, i . i ,-;-»,144. T^-f-i
!Jljlflr4J|wH:b]
i:nmp;nrr ;-pr-
4 * *-*-«-** 4 •—• *4-4-* *^i -» t *--*•* *
( » » « -*-t-T f ^"» » » f? • » *~*"i t t-t^-
. 4 ^r-* fc-. I *-*-»-( 4^* -» »-»-^4 *— -*-^.
f'i^f- l.-,....,*-,i, /T\+ , -* »-*
. * * ( - 4- * i I .--A. *-- Jf-* * *--.
?00 300 400
CtLL V(XUH£ (CIJ8IC »«
Igure 9. Size distribution of serially diluted GUrdU suspension In
distilled water at (1) 5i, (2) 2.Si. (3) 1.25X and (4) 0.625*
of the stock solution.
33
-------
I
c/»
bJ
d
a.
O
ui
>•
o
I
§
i
100.000
10.000
1000
100.000
10.000
1000
0.1
COULTER COUNTER
HEHACYTOMETER
0.5 1.0
' 0 10.0
PERCENTAGE DILUTION OF STOCK SUSPENSION
Figure 10. Linearity of two counting methods for enumerating Giardia cysts.
-------
§
S
i
g
i
I
3.0
2.0
1.0
1000
5000 10.000
PARTICLE CONCENTRATION (NO/ML)
50.000
Figure 11. Coefficient of variation for two methods used for enumerating
Giardia cysts.
35
-------
the concentration of the cysts, shows a ru.nj.rum around 20,000 cysts/rL for
both methods, indicating that the rest precise results are obtained at this
concentration.
TV.e lower detection lirit of the Coulter Counter was la/er than that of
the hcmacytcneter. The minimum amount of particles that could bo ttetected
v/ith the fomer ncthod in 0.5 irL of solution was 250. Tnc lower limit for
tlie hcnacytcneter in 0.0001 mL of solution was 1 cyst. This indicates that
the forrier method was 5000 tines more sensitive. The above results
therefore clearly indicate that the reproducibility and ranee of the Coulter
Counter were greater than for the hcnacytcneter. Ikx/cver, the irethod was
nonspecific for cysts and included other particles with the sane size: r^nge
as cysts.
Evaluation of 47rm Ilembrane Recovery *?>3chnique
An erianple of a recovory test using La):e Union \;ater, one of the water
sources available for the University of I-'ashington pilct plant, spiked v/ith
cysts is shown in Figure 12. She counts In the size rence of
Giardia were 19,564 pErticles/rt, boforn and 13,11?. particles/rrL after tho
recovery accompanied by a small apparent decrease in size of the particles.
These counts were substantially above the 220 particles/mL background count
of Lake Union water in the 8 to 1?. ur. size range.
The average recoveries of the 5.0 urn ."dllipore and Huclepore ncnbrr.nes are
sham in Figure 13. The flillipore membrane recovery using Ciarc'ia cysts in
distilled water ranged from SS'i to 90^ v/ith an average of 75.2". The
average recovery using Lake Union \»?ter was 77.0% using Ilillinore and 72.3?.
using iTuclepore. These results indicate no major differences bob/cen the
membranes even thougn the former hcd a sponge-like structure and the latter
had a pinpoint-hole structure.
Evaluation of 293 mm IVynhrane Recovery Ttochniquo
Ihe average 5rccovcrir of cycts at initial concentrations ranging frcm
cyst/rL to 10 cj-sts/nL measured with the hcmacytonetcr \.-as 20" (Figure
14) using ttie Ilillipore ra3nbrane and C51 \iitn the !!uclcpore mer.ibranc
indicating that the nembrene structure may have had an effect when using the
293 nm membrane. At concentrations below 1 cj-st/irL the recoveries became
highly variable due to the la: nur.-Jber of cysts that could bo enumerated.
For example, the recovery of duplicate runs at 0.1 cyct/rcL v;as 753 and 23",
rcsr«ctively. The la-;er reco\?ery using the 293 mn lillipore filter, r-s
compared to the 47 mm unit, may be due to the greater difficulty of removing
the entrapped rycts from the surf ceo of the Jarre r.icnbrene by agitation
using the chair., bath.
rvalueticn o rtanbreno Cassette Unit
She cassette unit was evaluated and showed a recovery of 0.-14 to 2.34"
(Table 1) which was la/er than observed for the 2?3 mm na.ibrExe. flic la/
3b
-------
DIAMETER (MiCSON METER)
78 9
60
50
40
30
20
10
200 300 400 600
CELL VOLUME (COBIC HICP.ON METERS)
•igure 12. Results of 47rmi diameter membrane filter recovery test using
Lake Union water spiked with Giardja cysts. (1) Before recovery
(2) recovered cysts and (3) ba'ckgroJnd counts.
37
-------
17
g
Ul
s
3
kJ
3
g
QC
W
_
—
et
UJ
S
Ul
i
Ul
"
k*
flg
£
S
s
0
-------
100
60
2 60
S
Ul
£ 40
20
10J
B
~ -O"
103
INITIAL CYST CONCENTRATION (NO/ML)
Figure 14. Percent recovery of cysts by 293mm diai.ieter, 5 ym pore size
membrane filters, (A) Millipore and (B) Nuclepore.
39
-------
TABLE 1. SUMMARY OF LABORATORY RECOVERY RATES OF G. LAMBLIA CYSTS WITH MILLIPORE PELLICON CASSETTE UNIT
Initial Volume of Final concentration Final volume Dilution Equivalent retentate Recovery
concentration of stock spike added of membrane retentate of retentate factor concentration percentage
(cysts/ml) to 20 L (mL) and Hash (cysts/ml) and wash (cysts/ml)
10.460
2,960
2,960
2,040
5.480
16
20
20
20
16
320
330
305
350
440
120
270
270
325
135
7.5
13.9
13.5
16.25
8.4
4,600
6,920
5,880
3.850
4.820
0.44
2.34
2.0
1.9
0.88
Mean Recovery Rate (n * 5) Is 1.5
S.D. = 0.72
-------
recovery nay be due to entrapment of cysts in the mesh separating the
stacked filters.
No cysts of fi. Iambiia were microscopically observed during the
recovery tests at Ryderwood, although numerous diatoms and other protozoa
were visible in the retentate sediment. Both retentate and wash were then
examined for its particle size distribution with the Coulter Counter,
resulting in 23 out of 560 particles/mL counted in the Giacd^T ~ize range.
2eta Potential of tysts
The zeta potential values for the fixed Giardia lamblia cysts clearly
show a decreasing potential at decreasing pH values (Tables 2 ana 3).
However, even at low pH values the cysts retain their negative charge
(Figure 15). The Zeta potential was always more negative than -20 mv in the
range of pH 5 to pH 10.
TESTING OF UNIVERSITY OF WASHINGTON PILOT PLANT
Unit process detention times for the 2.3 L/min (0.6 gpm) pilot plants
were determined by addition of Nad. or dye. The tracer concentration was
measured and plotted as a function of time to determine the retention time.
A second plot on semilogarithmic coordinates gave information about the
overall performance of the unit process reactor. The fraction of tracer
remaining at a given time (l-F[t]) was plotted as a function of the ratio
between the time of tracer measurement and theoretical retention time (tA) •
The actual retention time for the rapid mix was 2.1 min compared to 2.3
man as was estimated theoretically. The tank had a completely mixed flow
regime as evidenced by the semilogarithmic plot (Figure 16). At the
theoretical retention time T, 67% of the tracer had been displaced and only
19% remained at 1.5T.
The three flocculation tanks were studied individually and in series.
By itself, each of the compartments behaved as a completely mixed reactor.
However, as expected, with the tanks in series, the flow regime was
approaching plug flow (Figure 17). Although not intended for application to
stirred reactors, the relationship between retention time, dead space and
flew regime developed by Rebhun and Argaman (1965) can provide useful
information on flocculator performance. Applying this relationship to the
tracer data, the three tanks in series were approximately 53% plug flow. At
the theoretical retention time, 23.2 nun, 64% of the tracer had been
displaced and only 12% remained after one and one half times the theoretical
retention time. The actual retention time was 17.3 min.
The dye testing of the sedimentation tank revealed that a fair amount
of mixing was occurring throughout the tank (Figure 18). Only about 15% of
the flow was plug flow. It was believed that the flow regime could be
improved by constructing a better baffled inlet zone, although some of the
41
-------
TABLE 2. ZETA POTENTIAL (ELECTROPHORETIC MOBILIGY) OF BUFFERED FORMALIN FIXED
GIARDIA LAMBLIA CYSTS AT VARYING PH VALUES AND CYST CONCENTRATIONS
. f
1.
2.
3.
4.
5.
6.
Cyst Cone.
(0/ml)
4 x 104
5.25 x 104
4 x 104
4 x 104
4 x 104
2.14 x 104
Spec. Conductance
(microhmos)
1,800
340
3,000
3,500
4,000
12,000
PH
3.5
4.3
5.6
6.0
8.0
10.0
ZP(x) in mv.
corrected
-17.4
-14.4
-24.7
-31.3
-33.8
-39.2
N
6
8
6
14
6
4
SD
5.3
7.0
3.6
8.4
10.9
9.0
*Giardia suspensions used in #1, #3, #4, and #5 were from same
stock suspension. #2 and #6 were from different stock suspension.
TABLE 3. ZETA POTENTIALS OF FIXED GIARDIA LAMBLIA CYST SUSPENSION AT
DIFFERENT PH VALUES
Specific
Conductance
(microhmos)
340
1,200
1,600
3.200
PH
3.8
5.5
7.5
10.0
ZP(x)
-21.0
-25.5
-27.1
-37.3
N
10
10
10
10
SD
7.2
5.6
3.3
4.3
42
-------
-40
£ -30
S
o.
«S
5 -20
-10
6 7
pH
10
Figure 15. Effects of pH on the zeta potential of fixed (3. lamblia cysts,
(A) different suspensions and (B) same suspension.
43
-------
5
g
O
_J
Si
Mgure 16. Tracer evaluation of the rapid mix tanks.
44
-------
2
1
1
0.9
0.8
0.7
0.6
0.5
0.4
0.3
0.2
0.1
0.0)
0.0)
O.C7
O.C6
O.OS
0.04
0.03
0.02
0.01
I I
0 0.5 1.0 1.5
t/T
2.0
2.5
Figure 17. Tracer evaluation of the flocculation tanks.
-------
1
0.9
0.8
0.7
0.6
0.5
0.4
0.3
0.2
0.1
0.09
O.OB
0.07
0.06
0.05
0.04
0.03
O.OZ
0.01
0 0.5 1.0 1.5 2.0
t/T
Figure IS. Tracer evaluation of the sedimentation tanks.
46
-------
mixing was likely caused by density currents. The incoming water was around
10 C compared to an ambient temperature of 22 C. The theoretical
retention time was estimated to 5.6 hrs, whereas the actual, as determined
by the dye test, was 3.9 hrs. At the theoretical retention time, 38% of the
dye remained in the tank and 12% was still left in the tank after two times
the theoretical retention time. The tracer data indicated no ^ad space.
With the objective to determine the optimum speed of the flocculator
paddles, a series of tests was performed with a 10 mg/L alum coagulant
dosage and pH 6.7, adjusted by the addition of lime. The rapid mix speed
was kept constant at 500 rpn throughout the tests. Following the selection
of the mixing speed to be evaluated, the pilot plant was operated for 40
rain. Samples were collected from each of the three flocculation
compartments and allowed to settle in a jar for 30 min, at which time the
turbidity of the settled water was determined. Based on these tests, a 22
rpn mixing speed was selected. Ihe corresponding Gt and G values were
calculated to be 49800 and 48 sec" , respectively. The sample collected
from the first compartment with less than 6 min flocculation time did not
settle well compared to the sample from the third compartment with 17.3 min
flocculation.
TESTING OF OOKUIATIO^FILTRATION AND OIRBCT FILTRATION AT UNIVERSITY OF
WASHINGTON
jar Tests
Batch tests with different alum dosages showed that the lowest turbidity
after settling was obtained at a dosage of 8 to 12 mg/L using 1 L beakers.
(font?nUflltfi Tfesfcing with Giflfflj^ ^piking - Conve.ntT Pflpl Trt»a,tanent
The first seven runs were made using the coagulation/sedimentation unit
followed by filtration. Runs made thereafter were direct filtration runs
bypassing the sedimentation unit.
The results of the single spike of Giardia cysts (Run 1), added to the
first flocculaticn compartment of unit A are summarized in Table 4. This
run was performed primarily to determine removal of cyst-sized participates,
cysts and turbidity by both flocculation, sedimentation and filtration. No
cysts were detected in the filter effluent while high removals of particles
in the cyst-sized range were observed together with high turbidity removals.
The turbidity removal by the filter was more tnan 96% ana the run was
terminated after 80 hrs due to high headless in Filters B and C (Figure 6).
Tn this figure and in later figures showing University of vasrungton pilot plant
filter run data, the data points identified asr*verflow are for water
applied to the filters.
47
-------
TABLE 4 RESULTS OF A SINGLE DOSE SPIKE OF G1ARDIA CYSTS INTO FLUCCULATION
COMPARTMENT OF r>ILOT PLANT - RUN HI
Sampling point Sairpllng tine Participate Participate No. cysts
(hrs. after concentration removal7 In found in 201.
spike) (no/ml) preceedlng concentrate
process
Turbidity Turbidity Keadloss
(NTU) removal* In (ft)
preceedlng
process
(I)
Tap water --
plant Influent
1. overflow
from sedimenta-
tion basin
2. filter C
effluent
3. filter C
effluent
before run
1.5
1.5
4.5
2017*
20.01
.348
3.7
• m •
99.0
98.3
81.5
0
0
0
0.44
0.58
.02
.022
-31
96.6 6
96.2 7.4
•Approximately 170.0UO cysts added as single dose to first flocculation compartment at Aj.
^Concentration of Glardla size partlculates In Influent tap water Is calculated from Coulter enumeration of
7.5 L. water passed through S.Oum 293ran Nuclepore filter and processed as in section 2.4.
TPart1culate removal Is calculated as 100 - (concentration out/concentration In). Initial concentration of
partlculates for filter efficiency (s Glardla size paniculate concentration In overflow wetr. Initial con-
centration of partlculates for sedimentation efflccncy Is Glardla size participate concentration of Influent
tap water.
^Turbidity removal Is calculated similarly to that of participate removal.
-------
A continuous addition of Ciardia cysts (Run 2) added to the chemical
intake port is sunmarized in Table 5. Cyst-sized particle removal is
defined as removal of 8 to 12 urn participates as observed on the Coulter
Counter. Particulate removal was 99.0% for coagulation/sedimentation and
90% for filtration. Mean turbidity removal by filtration as monitored
periodically from Filters B and C effluents was 87% and 92%, respectively.
Total cyst-sized particle removal for the entire treatment train was 99.9%.
Three cysts were observed by microscopic examination in two different
efflurr.c samples; i.e., at the beginning and at the end of the run.
Estimated removal of cysts by coagulation/sedimentation and filtration was
99.8% or above at an influent concentration of 225 cysts/L and 0.05 cysts/L
in the effluent.
After a ripening period a high quality effluent was produced while the
headless showed an approximately linear increase with time.
In the third run Giardia cysts were added directly to the dual media
Filters B and C. Cyst-sized participate removal by filtration throughout
the run averaged 74% and 62% for B and C, respectively. Poor floe formation
and subsequent low turbidity removal were probably a result of inadequate
lime feeding. Cyst-sized particulate removal by sedimentation was greater
than 99% in spite of a low turbidity removal of 44% CIab..e 6).
When adding 984 cysts/L at the influent of Filter B in Run 3, 8 cysu/20
L were recovered in the effluent, corresponding to the 99.96% removal. When
622 cysts/L were added at the influent of Filter C, the effluent
concentration was 5 cysts/20 L, which is also a 99.96% removal.
The influent and effluent quality data of Run 4 (low pH) sho.,-ed an
average cyst-sized particle removal of 99.9%, while the turbidity removal
was 85%. The 1093 cysts/L in the influent of Filter C correspond with a
worst effluent concentration of 0.6 cyst/L in the effluent, which was a
99.95% removal.
The effluent turbidity of the fourth filter run is shown in Figure 19,
together with the cyst effluent concentration. The increase of hcadloss
with time was primarily accounted for in the 'jop 5 cm (2 in) of the filters.
Run 5, conducted at high pH (7.2), showed an average particle removal
of 80% and turbidity removal of 76%. Cysts were added to Filter C only, at
a concentration of 23 cysts/L. The filter effluent concentration was 0.75
cysts/L corresponding to a 96.74% removal. The higher pH resulted in a slow
rate of headless buildup, but uniform throughout the depth of the filter.
The effluent quality during Run 6 with no pH adjustment (pH 6.4) showed
a significant improvement ever Run 5 with respect to turbidity and
cyst-sized particle removals (Figures 20 and 21). The cyst removal,
however, was essentially the same. Of the 30 cysts/L added to the influent
of Filter C, 1 cyst/L was recovered from the effluent or a 96.67% removal.
The gradual headless buildup was primarily restricted to the top one-third
of the filter.
49
-------
TABLE 5. RESULTS OF CONTINUOUS SPIKE OF GIARDIA CYSTS INTO PILOT
PLANT - RUN #2
Sampling
point
Tap water
1. overflow
from sedimen-
tation basin
2. filter C
effluent
3. filter Z
effluent
4. filter C
effluent
5. filter B
effluent
6. filter C
effluent
7. filter B
effluent
B. filter C
effluent
9. filter B
effiuent
10. filter C
effluent
11. backwash
sample from
filter C at
end of run
Sampling
t'TO
(hrs. after
spike)
.before run
2
2
7
7
23
23
47
47
98
98
100
Particulate
concentration
(no/mL)
2017
10.4
0.54
1.33
1.22
1.34
0.47
0.78
1.04
2.02
.099
46SO
Paniculate
removal In
preceedlng
process
(X)
—
99.5
94.8
87.3
88.3
87.1
95.5
92.5
90.0
80.6
90.5
...
No. cysts
found in
20 L. con-
centrate*
—
0
1
0
0
0
0
0
0
1
1
720T
Turbidity
(NTU)
.50
0.2
.037
.038
.026
.03
.018
.028
.019
.03
.018
™
Turbidlty
removal In
preceedlng
process
(t)
...
60
81.5
87
87
85
91
86
90.5
B5
91
...
Headless
(ft)
• •
--
1.7
2.2
2.0
3.1
3.3
4.2
4.7
5.4
6.0
•-
100
4.7 x 103
2.0 x 10°
12. sedimen-
tation basin
after run
•An estimate of Glardla cysts per 20 L. sample was calculated by microscopic examination
of 0.16 ml of sediment from 20 nil concentrate as recovered in section 2-4. Final number
Is extrapolation from estimate of sediment volume (approximately 0.20 ml).
the dual media filters was calculated by
liter of backwash water recovered as in
ml of sediment were counted (of a total
of backwash results in figure given.
be crudely estimated by dividing the
the sum of the number of cysts in the
YAn estimate of the number of cysts trapped by
microscopically examining the sediment from 1
section 2-4. Six cysts in approximately 0.04
if 0.25 ml sediment). Extrapolation to 20 L.
Removal of Giardia .sts by sedimentation can
number of cysts in the sedimentation basin by
backwash and the sedimentation basin.
BAn estimate of total partlculates/ml removal by flocculation-sedimentation during
the entire run was calculated by siphoning 4 L. (representsting 81 of the total surface
area) of the settled floe at t»-c bottom of the sedimentation basin and enumerating by
the Coulter method.
AAn estimate of the total of cysts removed by flocculation-sedimentation was determined
by hemacytometer counts of approximately n.25 ">!• of sediment from 10 ml of floe
centrlfuged. ' JJQ
-------
TABLE 6. RESULTS OF CONTINUOUS SPIKE OF GIAkDIA CYSTS INTO PILOT PLANT - RUN P3
Sailing point
Tap »at«r
1. orerflcm
fron sedimen-
tation basin
2. filter B
effluent
3. filter C
effluent
4. filter B
effluent
5. filter C
effluent
6. overflow
7. filter 8
effluent
8. filter C
effluent
9. bactw*(>-
sa-3>le froa
filter c
Sampling tine Partlculate
(hrs. after concentration
spUe) (no/si.)
before run 2017
2 3.78
2 1.3
2 2.5
7.5 0.81
7.5 0.88
16.5 16.1
16.5 3.5
16.5 3.92
22 1840
Partlculate No. cysts
renewal in found In 201.
proceeding concentrate
process
It]
...
99.8 cysts fed
directly
to filters
65.6 8
34.0 2
78.6 1
76.7 5
91.2
78.3 8
75.7 3
1.2 x 10b
Turbidity Turbidity
(MTU) removal in
proceeding
process
(X)
0.5
.28 44
.12 57.1
.12 S7.1
.165 41
.17 39
.35 30
.2 42.8
.2 42.8
...
Headless
(ft)
..
"
.7
.83
.92
l.t
--
1.2
1.25
-•
•Cstlnatc of cysts calculated
AAn eitloite of the totil nuxfaer of cysts In bactoMSh socple icas calculated by extrapolating nemacytoneler
coutts of cysts In a total 0.25 nL of baclvash sedincnt fron 1 liter filtered through S.Oun Kuciepore filter
and procotuxl as In section 2-4.
-------
fc
•••
o
TURBIDITY
O OVERFLOW
O FILTER B
FILTER C
in
S
5
0 5 10 15 20 25 30 35 40 45 50 55 60 65
TIME (MRS)
Figure 19. Turbidity in filter influent, and effluent of Run No. 4.
52
-------
0.8
0.6
0.4
0.2
• CYSTS/L
TURBIDITY
O OVERFLOW
O FILTER B
A FILTER C
l i i
0.8
0.6
52
tst
0.4
0.2
02 4 6 8 10 12 14 16 18 20 22 24 26
TIME (MRS)
Figure 20. Turbidity in filter influent and effluent of Run No. 5.
53
-------
U3
n
PO
c
3-
*<
•d.
3
CO
1
n
Ol
CL
n
n
70
3
0.3
0.2
0.1
• CYSTS/L
TURBIDITY
O OVERFLOW
0 FILTER B
A FILTER C
1.8
1.6
1.4
1.2
1.0
0.8
0.6
0.4
0.2
10 20 30 40 SO 60 70 80 90 100 110 120
TIME (HRS)
-------
Because of high pH (7.2), the seventh filter run showed a lower
turbidity removal (Figure 22) than earlier runs with the same coagulant
dosage (10 mg/L) and the pH ranging from 6.3 to 6.7. The largest increase
in headloss developed at the anthracite/sand interface.
The results summarized in Table 7 generally show higher cyst removals
at high spiking levels than at low spiking levels. This can be explained by
the inherent limitations in the cyst enumeration technique. Low levels are
difficult to determine accurately. Therefore, the investigator may
sometimes have to settle for an upper boundary value which often will
overestimate the number of cysts in the sample and underestimate the
removal.
Direct Filtration at University of frfcsfoington Pilot Plant
This subsequent part of the study was devoted to the evaluation of the
8 to 12 urn particle and turbidity removal efficiency of the University of
Washington pilot unit. The unit was operated in the direct filtration node
treating chlorinated unflitered Tolt Reservoir water.
The filter runs were conducted at different alum dosages, pH values and
flow rates. The main parameters measured during the testings were: removal
of particles in the ciardia size range, turbidity removal, length of filter
run, headloss buildup at different depths in the filter and particle
distribution at different filter depths.
The effect of alum dosage on the filtration efficiency is shown in
Figures 23 and 24 for Filters B and C operated under identical conditions.
Data were collected both at low (5.5 m/hr, [2.3 gpm/ft ] and 6.0 m/hr [2.5
gpnH) and high (9.6 m/hr [3.9 gpm/ft*] and 13.5 m/hr [5.6 gprn/ft*])
filtration rates for identical Filters B and C. The filter runs without any
alum addition showed a 59% cyst-sized particle removal efficiency ana 10%
turbidity removal. The turbidity removal reached a maximum plateau at a 10
mg/L alum dosage while the particle removal did not increase further above 7
mg/L. The data of the University of Washington pilot plant show that
particle removal exceeds turbidity removal below a dosage of 10 mg/L alum,
possibly due to the inability of the filters to trap the small particles
causing the turbidity, while still retaining the cyst-sized particles.
Dosages above 15 mg/L greatly shorten the filter run and decrease the
removal efficiency. Increasing the flow rate in the alum dosing range of 15
to 30 mg/L resulted in a slightly lower particle removal. The direct
filtration runs at the high flow rate were less consistent than at the low
flow rate and the run at 17 mg/L alum and 13.5 m/hr (5.6 gpm/ft ) showed an
unexpected lower removal efficiency and greater rate of headloss buildup.
The pH was also of major importance in the participate removal and
highest removals were observed at pH 6.5 and within a range of 5.6 to 7.0
55
-------
0.6
0.5
g 0.3
0.2
0.1
• CYSTS/L
TURBIDITY
O OVERFLOW
0 FILTER B
- A FILTER C
10 20
30
40
50
TIME (HRS)
60
70
80
90
2.0
1.8
1.6
1.4
1.2
1.0
O.C
0.6
0.4
0.2
0
100
Figure 22. Turbidity in filter influent and effluent of Run No. 7.
56
-------
Table 7. Performance of each filter run with
cysts added directly to filter
Run No.
3 6.7
4 6.3
(no lime)
5 7.2
(high
lime)
6 6.4
(no lime)
7 7.2
Cyst
Dosage
984/1 (B)
622/1 (C)
1093/1 (C)
23/1 (C)
30/1 (C)
2.3/1 (C)
% Cyst-
sized
Particle
Removal
93.1
91.5
30.6
95.6
95.2
% Cyst
Removal
99.96
99.95
96.74
96.67
30.4
(high
lime, low
spiking
level)
57
-------
10 20
All* DOSAGE («G/l)
30
Flqure 23. Effect of alum dosage on direct filtration process. Filter B.
-------
10 20
ALUH DOSAGE (HG/1)
Figure 24. Effect of alum dosage on direct filtration process. Filter C.
59
-------
5.5
2.26 3W/FT
ALUM DOSAGE: 15 NG/1
FILTEP: B
pH
Figure 25. Effect of pH on direct filtration performance,
Filter B.
60
-------
u. ae
a
100
90
80
70
60
50
90
80
70
60
50
4
3
2
1
90
80
70
60
50
40
30
20
10
0
5T5
FIOUMTE: 2.35 GW/FT'
ALW DOSAGE: 15 HS/l
FUTER: C
-y-
-O-
-o-
6.0
6.5
7.0
PH
Figure 26. Effect of pH on direct filtration performance,
Filter C.
61
-------
(Figures 25, 26) using a dosage of 15 mg/L. Dounling of the flow rate
resulted in a slightly lower particle removal but did not affect the
turbidity removal. A rfl increase frcn 6.5 to 7.0 resulted in a greater rate
of headless buildup at the higher flow rate than at the lower flow rate.
Addition of soda ash was required to counteract the pH decrease resulting
from the alum addition. A typical dosage of 7 ng/L soda ash was required to
maintain the pH at 6.5 when using 15 mg/L alua. High rate runs generally
produced poorer quality water* and the quality deteriorated even further
when the pH was changed at the high filtration rates. The results in Figure
27 shew that when the pa was temporarily changed frcn 6.5 to 6.8 the
effluent turbidity increased from 0.12 to 0.65 NIU in Filter B and frcn 0.11
to 0.33 NIU in Filter C operated in an identical fashion. The influent
turbidity was 0.90 MID. The lower efficiency was also reflected in the
decreased rate of headless buildup.
Increasing the flow rate at a dosage of 15 mg/L alvm and pH 6.5
resulted in a gradual decrease of turbidity and particle-removal. Turbidity
renewal greatly decreased above 17.1 n/hr (7.0 gprV-t ) in Filter B._ and
particle removal decreased substantially above 12.9 ro/hr (5.3 gpn/ftr) in
Filter C. The longest filter runs were observed below a flow rate of 6.1
n/hr (2.5 gpm/f t ). Selecting flow rates above 17.6 n/hr (7.2 gpm/ft)
resulted in a sharp increase in the rate of headloss buildup. The length of
the theoretical water column that could be passed through the filter before
backwashing showed a pattern parallel to that of the length of the filter
run indicating that substantially longer runs were obtained below a flow
rate of 6.1 n/hr (2.5 gpm/ftT) (Figures 28 and 29).
Samples were carefully withdrawn front the three way valves at each of
the headloss ports at different tines during the the high flow rate filter
runs. The results in Figure 30 show that the cyst-sized particles were
gradually filling the voids in the upper parts of the filter. Thus a
relatively sharp downward moving front existed between the filled voids and
the empty pore spaces. The data indicate that at the 15 mg/L alum dosage
the particle concentration in the upper portion of the filter was about
two-fold higher than in the influent due to its accumulation and
flocculation/sedimentation. This increase was about ten-fold at 20 mg/L
alum and about fifteen-fold at the 30 mg/L dosage. Some differences were
noted between the last two sampling ports after the filter; i.e., the port
below the screen containing the filter media and the last port after passage
of the effluent through the turbidity meters. The last port gave lower
particle counts at the 30 mg/L alun run, possibly indicating some particle
settling or attachment to the effluent tubing.
9
It was concluded that optimum operation of direct filtration was
achieved at an alum dosage of 10 mg/L alum, and a pH of 6.5. The longest
filter runs were obtained at a rate below 6.1 n/hr (2.5 gpn/ft^). Sudden
operational changes such as a pH increase caused a rapid effluent
deterioration with respect to turbidity and cyst-sized particles. Quality
deterioration was more pronounced when the conditions were changed rapidly
as opposed to gradually.
62
-------
0.5
0
0.6
o.s
di ;•;
w»- 0.2
0.1
s 0.2
0.1
0
tr* A __
Sg 10
g= 8
S- 6
ffe! *
4*J CK &
n 100
SI 90
80
CHANGE OF pH 6.S to 6.8
FILTER B
FILTER C
\
^ ALUM DOSAGE IS K/L
' aOWRATE 5.36W/FT2
Vl
I I I I
2 3 4 S 6
DURATION OF FILTER RUN (HOURS)
Figure 27. Effect of pH increase on filter performance.
63
-------
100
90
80
70
60
SO
90
70
60
50
9
B
7
6
5
4
3
2
1
40 —
30 —
20 —
10 —
pH: 6.5
H.W DOSAK:
FILTER: B
ISKS/l
400 s
300 °
m4
zoo 3
n
100 I
FLOWRATE (GPH/R2)
Figure 28. Effect of flowrate on direct filtration
efficiency, Filter B.
64
-------
PH : 6.5
All* DOSAGE IS «A
FILTER: C
flOWRATE (GPM/nZ)
Figure 29. Effect of flowrate on direct filtration efficiency, Filter C.
65
-------
1000 —
100 —
3!
i
a.
^
c*.
FRTE* B 4.2
O 0:30 RR
3-30 HR
FILTER C 5.8 GWFT2
« 0.30 HR
A 3.30 HR
— ALUM DOSAGE 20 KG/t
10 20
rarrn IK FTIT:K (INCH)
Figure 30. Particle removal at different filter depths.
G6
-------
Direct Filtration at University of Washington with Giardia Iambiia tysts
During the final runs. Filter C was the only filter used for cyst
spiking. Furthermore, it was decided for practical reasons to add the cysts
as a slug rather than have continuous feeding throughout the run.
Continuous feeding would have required extremely large quantities of cysts,
because of the high cyst removal efficiency of the filter and the relatively
large number of cysts necessary for reliable detection and enumeration in
the filter effluent. These large quantities were just not available.
Tto determine when the peak concentration of cysts would appe«. in the
filter effluent, a series of conductivity tests were made. A salt "-ilution
was added to the plant infl'ient at the cyst addition port. TV* salt
solution was added for 30 sec, the same time period that had been selected
for the cyst addition. Conductivity was monitored continuously at the final
flocculation tank effluent, the overflow from the distribution trough which
also functioned as a constant head tank for the filters and the etfluent
from Filter C. This test was repeated for many different filter loading
rates to cover the entire range of normal operation. Knowing the filter
loading rate, Figure 31 could be used to determine when sampling started ana
stopped if the objective was to collect a 20 L filter effluent sample, half
of which preceded the peak of the effluent cyst concentration ana half of
which followed the peak. Figure 32 shows what percentage of all the cysts
passing through the filter was captured in the 20 L sample.
£. Iambiia cyst stock solutions were prepared by ertracting the cysts
from stool specimens of giardiasis patients, provided by hospitals ana
pathology laboratories throughout the State of Washington. The procedure for
extraction has been described earlier. The5resulting stock solutions ranged
in concentration from 1.2 x 10 to 5.0 x 10 cysts/mL, and were stored at 4
C until needed.
Cyst suspensions for the pilot plant runs were prepared immediately
before being added to the plant influent. The total cyst concentration
selected for the run determined the amount of stock solution used.
Distilled water was used as diluent to give a final volume of 180 mL, wV. A
was pumped into the plant influent line.
The cyst addition port was located on the plant influent line, ahead of
the coagulant feed manifold and opposite the pH adjustment port. A static
mixer, Kenics Model 1/2-10-321-5, separated the cyst/pH adjustment feed
manitold and the coagulant feed manifold. The static mixer provided gooa
dispersion of the cysts and uniform pH of the raw water before any chemical
coagulant was added. The cyst suspension was pumped into the feed line,
using a FMI lab pump, calibrated to deliver the 180 mL volume in exactly 30
sec.
Earlier tests aimed at determining cyst losses during the coagulation
and flocculation process, had shown some variability when parameters like
pH, and coagulant dosage, as well as type of coagulant used were changed.
Therefore, during the actual cyst runs both filter influent and effluent
67
-------
60
SO
40
30
T»EAK> CONCENTRATION
1.0 1.5
FLOW RATE (L/HIN)
2.0
Figure 31. Sampling schedule for 20L filter effluent sample at different
filtration rates.
68
-------
60i
sol
X>t
-L I i
1.0 1.5
FLOW RATE (L/HIN)
2.0
rlgure 32. Percentage of total number of filter effluent cysts present 1n 20L
sample collected according to Figure 31.
69
-------
were sampled to determine the cyst removal efficiency of the filtration
process. All samples collected were processed as described earlier, which
included membrane filter filtration, centrifugation ana microscopic
examination.
Die cyst dosage during the runs ranged from 2.0 x 10 to 21.5 x 10
cysts. For the majority of runs, however, approximately 20 x 10 cysts were
added to the raw water. The total number of organisms actually reaching the
filter depended upon two main factors: first, the loss of cysts in the rapid
mix and flocculation tanks, due to disintegration and attachment, and
second, the filter loading rate. In order to provide a constant reaction
time for the coagulation and flocculation processes, the flowrate through
the plant was kept constant at 2.3 L/min. For a low filter loading rate,
this meant that a proportionately large amount of the cysts reaching the
distribution trough would be wasted through the overflew. As the flow rate
to the filter increased, so did the cyst load, assuming the dosage to the
plant remained relatively constant.
These direct filtration runs were designed to investigate factors such
as coagulant dosage and pH, especially its affect on alum coagulation (Table
8). The filter loading rate was kept relatively constant at 9.8 m/hr (4
As expected, with no coagulant being added to the water, the filter
performed poorly with regard to both cyst removal and turbidity reduction.
More than halt the cysts, 52%, passed through the filter and the effluent
turbidity remained relatively high. At optimum conditions, however, cyst
removal was consistently high. An alum-dosage of 12 mg/L, pH 6.2, and a
filter loading rate of 4.9 m/hr (2 gpro/ft*) .- would give a 99.73% removal of
cysts at the end of the one hour filter ripening period. Later in the run,
cyst reduction was 99.94% and the effluent turbidity was constant at 0.02
NIU. The influent turbidity was 1.2 MTU. An increase in the filter loading
rate to 9.8 m/hr (4 gpVft ) did not have any adverse effect on the filter's
ability to remove cysts. In fact, at the end of the filter ripening period,
the cyst reduction was 99.94%, slightly higher than at the lower flowrate.
Seven hours into the run it had improved to 99. 98%, even though the effluent
turbidity was 0.2 NIU, compared to 0.02 NIU ot the lower loading rate
(Figure 33) .
A reduction in the coagulant dosage led, as expected, to an increase in
the number of cysts passing thiough the filter. At a 7 mg/L alum dosage,
99.75% of the cysts were removed one hour into the run, and 09.98% after 16'
hrs. The lower alum dosage also resulted in an increase in the filter
ripening period to 1.5 hrs and a higher effluent turbidity, 0.03 NiU. A
further reduction in the alum dosage to 4 mg/L had a more dramatic effect.
The filter ripening period was increased to approximately 2 hrs and only
64.2% of the cysts added to the plant after 2.5 hrs of operation were
removed in the filter. The effluent turbidity was 0.5 NIXJ, but slowly
decreasing. The effluent turbidity at 72.5 hrs ranained relatively high at
0.4 NIU, whereas the cyst removal had increased to 91.8%.
7.0
-------
Table 8. Cyst Removal During Direct Filtration at UW Pilot Plant
Alum
Coagul. Filter rllter
and Loading Infl. Cvst fiapsed Infl. tffl. Turbidity
Bun Dotage Rate cVst Removal lime Turbidity Turbidity Removal
Ho. Hg/L pH ">/hr Dosage t Hrs-Htn MTU HID J_
72
73
74
76
77
78
79
no
Rl
B?
None
12.0
12. 0
12.0
7.0
4.0
12.0
12.0
12.0
Ot-
rinc 11
5.0
6.5
6.2
6.2
6.2
6.2
6.2
6.8
5.6
S.6
6.4
6.0
6.0
4.3
9.6
9.2
9.6
B.5
10.0
8.5
9.6
8.2
9.6
9.6
9.6
9.6
6.6-10-
3.R-I06
4.2-I06
7.3 106
B.7-106
9.8-106
9. 4 -1C0
I0.7-106
8. 8-106
10.3-IC6
8.4-106
IO.O-106
9.8-106
9.8-10*
9.R-IO&
48.01
99.733
99.943
99.936
99.979
99.750
99.870
64.23
91.81
95.44
99.41
99.83
99.84
9S.90
99.911
4:30
1:15
26.00
1:00
7:00
1:00
16.00
2:30
72.30
1:00
10-00
1-00
7:00
1:110
2i.no
0.73
1.24
1.19
1:37
1.14
1.94
0.81
1.31
1.35
0.95
1.02
1.73
1.78
0.3:.
O.RO
0.39
0.03
0 19
0.04
0.02
0.24
0.03
0.52
0.37
U.2B
0.04
0.03
0.02
T 23
0.27
46.58
97.74
98.40
97.23
98.07
87.63
S6.30
60.31
72.59
70.53
96.27
98.09
98.93
75.00
66.25
-------
1M-
i a
Of
Cylt
-------
During earlier runs with alus as the primary coagulant, the overall
performance of the filtration plant was fond to be very sensitive to
changes in pH. Therefore, scree of the cyst runs were designed specifically
to investigate the inportance of proper pH control on cyst reraov.xl. It had
been shewn that 99.98% ot the cysts could be renewed during the filtration
process with proper pretreatnent, using 12 ng/L alum at pH 6.2. Keeping the
alun dosage unchanged but lowering the pH to 5.6 did not dramatically affect
the cyst renewal. After one hour of operation, at the end of the filter
ripening period, 99.83% of the cysts were ra&oved by the filter* and 6 hrs
later 99.84% were reccved. The effluent turbidities were 0.03 MR} ana 0.2
NIU, respectively. An increase in pH to 6.8 dropped the cyst reaoval one
hour into the run to 95.44% and the effluent turbidity was 0.3 NXU, showing
scce fluctuation. After 10 hrs the cyst reduction had inprcved to 99.41%
and the turbidity was at 0.04 MIU (Figure 34) .
The only polyaer used during the cyst runs was Cat-Floe Tl (Calgon
Cbrp.). A 5 ng/L dotsage was determined the optisun and tho pH was kept at
6.4, the natural pH of the raw water, The cysts were added to the plant
influent after one And 21 hrs of operation with removal efficiency of 95.9*
and 99.91%, respectively. Even though good cyst renoval was achieved ciuring
ttiis run, the effluent turbidity was 0.2 NTU after one hour and increased
slightly to 0.3 VTU 21 nrs into the run. These value* were considered
relatively high in ccnparieon to the excellent filter performance when alum
was uitod as coagulant.
TESTING OF DIATUtACIDUS EAKTH KILTER AT THE UNIVERSITY OP WASHINGTON
As expected, the rgcult* from the Initial runs nhowod that the
cyst-else particle removal by the DC filter wa» generally better than the
reduction in turbidity, oe tho -wvetal type« of filter Aid tested. the Sent
parfomorn were the finer grates, ecpac tally tn the very bwjtnmrxj ct the
run. Later in the run, hcwver. none Bpeelttcally outperfonwd the oU«r».
Snot typical data Are ahoxi in Figure* 35, 36 ant 37.
The noflt notlcoaDle difference between these run» WAB t>w rate of
buildup which M*J» »low«at for the criareevt grade*. Thin was alno
nan tf O6 ted ty longer filter r>jn«. The length ot the run depenrted not only
on the type ot diatmlte UVK), trjt to a nigmt leant o>gn*e on ttw tvrxnt of
body food arVta«1 to tne filter. The Iftzy (rod rate r.tnrjed ?rm 10 to 40
nq/U Dxxtjh the raw water u4ed f-^c tr«r»e run* «•*• of niyn Quality,
turbidity mrrully rarsjif*^ frcra 0.5 to 0.9 *ft\j during tni« tun* o! t/w
txnty f««4 rat«» l«r»» Uvm 20 *q/L re^jlfe^ tn r«Utlvely »hnrt filter run«,
usually fron lf> to 46 hr«. Similarly, tM duration ot tr>e filter run
:f
one cf the run* ucin^ KyHo Sup»r-Ctl mi filter aid, a
ntration of tt* noniomc pniyrwr, HrTiifloc MSN (African
Co., Mr/ne, KJ) wac *ddbd to U>e tew wat«f. The n&et nctic«»atl« « feet oj
the 0.0075 «g/L polynK addition W&B a aignificant trpjovarwrt in t.»-e
rt fluent qutlity in the wry tvainninq ef th« run. AC the run
progr««einq, the efficiency o? Cw DC f'lter ie«r«d to be eirJtltr to
rum; where no polyivr had been *dded. Tho irprovanent in effluem
73
-------
100
.. w
\
s
»7
t.k
ft.S J.5
of
-------
100
90
5 80
3
W
of
70
to
M
LOSS
-- TUS31DITY
~ CYST SIZLD TAPTICLtS
30
25
20
10
1W
2iO
'»<;ur« 3i>. Characteristics of DL ftlter run with Cellte 503 filter aid at
20 mQ/L body fetd.
-------
100
90
80
70
60
5C
*HEADLOSS
---4r-- TURBIDITY
_ m.._ CYST SIZE PAPTICUS
30
25
20 §
15
10
50 100 150
T!HE (HOOPS)
Z DC
250
lure 36. Typical data from a DE filter run using Hyflo Super-Cel as filter
aid. Body feed rate, 20 rng/L.
-------
IOC
90
80
70
60
50
•HEADLOSS
•TURBIDITY
CYST SIZED PARTICLES
30
25
2C
15
10
50
100 150
W1t( HOURS)
200
250
Figure 37. Typical d;ta from a DE filter run using Celite 512 as filter aid.
Body feed rate, 20 mg/L.
77
-------
was paralleled by a more rapid increase in headless across the filter. When
the run was terminated, it was approximately 25% shorter than similar runs
where no polymer was added. It was assumed that the single most important
factor for the decrease in the duration of the run was the polymer addition.
All effluent samples collected during this series of tests were very
low in cysts. In fact, after concentrating the 38 L or 19 L samples to 1 mL
with an average recovery of 88.5%, no cysts were detected in 5 of the 12
concentrates, for practical reasons, only aboug 25% of the 1 mL volune was
examined. Because of the low counts, the actual cyst removal efficiency of
the DE filter could not be determined. Only the boundary values could be
determined. However, the data showed that diatomite filtration was
effective in removing G. Iambiia cysts, even in the very beginning of the
filter run when the precoat acted as the only barrier. The only decrease in
the filter's ability to trap the cyst particles was recorded when the dosage
at the end of the run was increased six times. This decrease in performance
was less evident when a polymer was added to the raw water. Generally the
polymer addition improved the removal efficiency, but tended to shorten the
filter run because of a more rapid rate of headloss buildup, especially
towards the end of the run. A better method might be to add polymer only in
the beginning. The results are shown in Taole 9.
TESTING OF DIRECT FILTRATION IN HOQUIAH AND LEAVQMOKffl
Results of EPA Pilot Unit at Hoquiam
The filter runs were conducted at different coagulant dosages, pH
values and filtration rates. The main parameters determined were turbidity
removal, removal of particles in the 8 to 12 urn range, length of filter run
and headloss buildup at different depths in the filter.
The major factors influencing treatment efficiency were coagulant
dosage, pH and filtration rate. Figure 38 shows the effect of alum dosage
at pH 6.7 and 10 m/hr (4.1 gpm/ftz). The data indicate that the particle
removal reached a maximum above a dosage of 10 mg/Lr whil" the turbidity
removal was already at its maximum at 8 mg/L alum. Adding 0.04 mg/L of a
nonionic polymer, Magnifloc 985N (American Cyanamid Co., Wayne, New Jersey),
led to no improvement in the particle or turbidity removal. The rate of
headloss buildup, however, increased fron 5.3 cxi/hr (2.1 in/hr) to 18.3
cm/hr (7.2 in/hr), greatly reducing the length of the filter run. The
duration of a filter run could be improved by lowering the coagulant dosage,
but the treatment efficiency would suffer as a result.
In general, the rate of headloss buildup was linear with time and the
i yarity of runs were terminated due to turbidity breakthrough before the
205 on (80 in) to 230 cm (90 in) of available head had been exhausted. The
lowest rates of headloss buildup were observed at high pH values and low
filtration rates, whereas high rates of headloss buildup were the rule for
low pH values and high filter loading rates. The headloss profile at the
end of the different filter runs unowed a rather uniform distribution
throughout the filter, with only a slightly smaller buildup at the top.
This indicated that the floes penetrated the bed sufficiently.
78
-------
TABLE 9. FILTER RUNS WITH CYSTS USING DE FILTER
Cyst Addition
Run
No.
63
64
65
66
Number
Polymer
Added Sluq
3.0-106
3.0-106
No
3.0-106
3.0-106
Yes 3-°'106
3.0-106
No
res
Added
Contin.
Cysts/1
1.5- 10s
9.0-105
1.5-105
9.0-105
4.5-105
4.5-105
E loosed Time
Hrs-.Hin
0:05
0:20
2:00
2:30
3:00
0:05
0:20
2:00
2:30
3:00
3:00
3:00
Removal (R)
S
99. 35 'R< 99. 78
99.65'R
99.b3< R
99.61< R<99.96
99.03
-------
QALUM
AALUM + 0.04 KG/I 985N
riALUM * 2.8 KG/L HAGNIFLOC
WITH INFL. TURB B.3 NTU
AFTER RAIN
ALL TEST AT pH « 6.7; FLOW 4.1 GPM/FT
5 10
ALUM DOSAGE (HG/L)
IS
Figure 38. Effect of alum dosage on particle and turbidity removal during
field work at Koquiam.
80
-------
A series of runs were made with polymer as primary coagulant or
coagulant aid in combination with alum. The data indicated that 3.4 mg/L
Catfloc T (Calgon Corp., Pittsburgh, PA) or 2.1 mg/L Hagnifloc 573C
(American Cyanaroid Co., Wayne, NJ) were required to obtain a larger than 90%
particle and turbidity removal. A 5 mg/L alum dosage with polymer coagulant
aid showed no major improvement in particle and turbidity removal as
compared to the same polymer dosage by itself. When the alum dosage was
increased to 7.1 mg/L the filter's effectiveness improved and better than
90% removal of turbidity £jxJ cyst-sized particles was observed at the 7.1
mg/L alun dosage in combination with either 1.7 mg/L Catfloc T or 2.0 mg/L
Hagnifloc 573C. Without changing the polymer dosage, no significant
improvement was evident when the alum dosage was increased beyond 8.1 mg/L.
The addition of polymer generally tended to decrease the rate of headless
buildup, possibly by forming floes that penetrated deeper into the filter
bed.
Another important parameter for controlling particle and turbidity
removal, particularly when alum was used as coagulant, was pH. The data in
Figure 39 show that at Hoquiam the highest removals were obtained at a pH
value of 6.7, which remained optimum throughout the study period, with only
one exception. High removals were still observed in the 6.4 to 7.0 pH
range, but lower removals were noted outside this range, lowering the pH to
6.0 resulted in particle and turbidity removals below 90% and caused a high
rate of. headloss buildup. Similarly, an increase in the pH to 7.4 resulted
in a major decrease in turbidity removal, but a lesser decrease in the
removal of particles. In the present study, the pH was manipulated by the
addition of hydrochloric acid (HC1) or soda ash (Na CO..). The latter could
be dosed more accurately than the lime previously used. To maintain optimum
pH during alun addition, soda ash was always required to counteract the pH
decrease caused by the alun. The addition of alum decreased the pH as it
decreased the alkalinity as given by:
A12(S04)3-14.J H20 + 3 Na2C03 + 3H2O •* jAHOH^ * + ^B2S04 * 14'3 "2° * 3 °°2
During the pilot plant study, the addition of 10 mg/L alum typically
resulted in a pii decrease of 0.27 pH units. Addition of 10 mg/L soda ash
generally increased the pH 0.46 units.
The effect of sudden changes in pH is demonstrated in Figure 40. The
data were obtained from two runs with an alum dosage of 10 mg/L and a 10
ro/hr (4.1 gpra/ft) filtration rate. During the first run the pH changed
fron 6.8 to 7.3, resulting in a decrease in particle removal and the rate of
headloss buildup. An increase in effluent turbidity was also noticed,
approaching the influent turbidity of 1.7 NIU. The increase to pH 6.9 only
changed the rate of headloss buildup. However, the change from pH 6.9 to
7.3 had the largest impact as evidenced by an increase in effluent turbidity
of 0.5 NHJ. In addition, particle removal and rate of headloss buildup
decreased. When the pH was brought back to 6.8 the effluent turbidity
responded immediately and returned to its initial value.
81
-------
so
O 8 MG/L ALUM
10 HG/L ALUM
D 23.4 MC/L ALUM WITH INFL.
TURB. 3.3 NTU AFTER RAIN
PH
Figure 39. Effect of pH on particle and turbidity removal during field work
at Hoquiam.
82
-------
00
u»
? lOfl
w "*
80
£ 1.5
S
| 1.0
•- s
| " 0.5
^ 0
3.0
§ S 2.0
UJ CZ
»8 -
0
7.5
7.0
6.S
~-q
---- 0 ---
FLOW 4.1 GPK/rr2. 10 HG/L ALUM
1 1 1
-o—
— cx
FLOW 4.1 GPH/FT. 10 MG/L ALUM
4 0 1 2 3
LENGTH OF FILTER RUN (HOURS)
1 1 1 1
Figure 40. Effect of pH changes during Run No. 9 at Hoquiam.
-------
Increasing the filtration rate from 10 m/hr (4.1 gpm/ft ) to 15 m/hr
(6.1 gpn/fc) was generally detrimental to the overall performance of the
filter (Figure 41). At an alum dosage of 15 mg/L and a pH of 6.7, the
higher filtration rate resulted in a gradual decrease ir turbidity and
particle removals. Further, the higher filtration rate resulted in a more
rapid buildup of headloss and a significant shortening of the filter run due
to an early turbidity breakthrough (Figure 42). Because of the short filter
run, less than half as much high quality water was produced at this
filtration rate as would normally be expected at 10 n/hr (4,1 gpm/ft ) witn
the same raw water quality.
A very low turbidity and particle removal (49 and 48%) was experienced
on September 2, following a rainstorm which increased the influent turbidity
fron 1.2 to 8.3 NTU. A more than doubling of the alun dosage to 23.4 mg/L,
in combination with a lowering of the pH to 6.0 was required to bring the
effluent quality to within normal operating values. In addition to an
increase in turbidity during the heavy rain, the pH of the raw water
decreased from 7.3 to 6.8. These results indicate that optimum process
conditions can change rapidly, within a few hours, as changes occur in the
quality of the raw water.
A relationship was established between e-ffluenc turbidity end particle
removal (Figure 43). An effluent turbidity below 0.05 NIU corresponded with
a median (50% of the values) particle removal of 95.1%, while an effluent
turbidity between 0.05 and 0.1 NTU was associated with a 94.3% particle
removal. A surprisingly large number of samples (33%) with an effluent
turbidity below 0.05 NIU had particle removals below 90%. This was observed
especially in the beginning of the run directly after the filter ripening or
during the running phase when the influent particle concentration declined
temporarily. The polymer plus alum and polymer runs did not produce
effluent turbidities below 0.1 NIU, but high median particle removals of
95.3 and 92.6% respectively, were noted for effluent turbidities between
0.10 and 0.20 NIU. These results are further summarized in Figure 44 which
shows a relationship between median particle removal and effluent turbidity
range. Greater than 90% median particle removal was observed for effluent
qualities below 0.2 NIU but not for values above it.
The study also evaluated the removal of actual Giardia lamblia cysts by
the pilot unit, cysts recovered fran human stool specimens were added to
the raw water, ahead of any chemical addition, during an 8 min spike. The
cysts were recovered fran the influent and effluent using a membrane
filtration technique. Of the 1.67 x 10 cysts added to the raw water, 1.06
x 10 remained in the water just before entering the filter according to the
membrane filtration technique. The filler effluent contained a total of 2.6
x 10 cysts, representing at least an 81% cyst removal. This corresponded
with a 99% removal of particles in the 8 to 12 urn range as determined by the
particle counter and a 94% turbidity removal.
Full Scale Plant at Hoauiam
84
-------
I
<=>-(
55
8
o as
Iff v> *~*
S 3— i —
II!
* •— uj
5
FLOW
Figure 41. Effect of filtration rate on particle and turbidity at Hoouiam.
85
-------
M
3!
cc
UJ
_J
o
100 -
90 -
80 -
1.0
0.5
Ul
1 2 3
LENGTH OF FILTER RUN (MRS)
Figure 42. Effect of high filtration rate on filter performance at Hoquiam.
Alum dosage 15 mg/L, pH 6.7 and filter loading 15 tn/hr (6.1
gpm/ftZ).
86
-------
'.NT
1D1TY
O -s 0.05 MTU
0.05 - < 0.10
0,10 - < 0.20
V 0.20 - < 0.30
0.30 - < 0.50
PERCENTAGE OF SAMPLES WITH LESS THAN
CORRESPONDING PARTICLE REMOVAL
Figure 43. Relationship between effluent turbidity and particle removal at
Hoquiam.
87
-------
100
O ALUM
^ ALUM AND POLYMER
POLYMER
V AL»« AND POLYMER AT
LEAVENWORTH
40
0 0.1
0.5
EFFLUENT TURBIDITY (NTU)
Figure 44. Relationship between effluent turbidity
and median particle removal
88
-------
The City of Hoquaim water treatment plant used both alum and polymer
for pretreatment. Normal practice was to add the polymer, at 1 mg/L or
less, as primary coagulant and filter aid during periods of low turbidity.
When the raw water turbidity exceeded 1.5 NTU, alum was used as primary
coagulant and the polymer as coagulant aid and filter aid. The alun dosage
could be as high as 30 mg/L depending upon the raw water quality. Alun was
also used to precoat the filters following backwash. The pH was controlled
by the addition of soda ash.
The chemicals, including chlorine gas for prechlorination, were added
to the raw water line about 30 meters (100 ft) upstream of the flocculator.
Powdered activated carbon was added just ahead of the flocculator for
removal of color. The flew ranged from 8400 m (1.8 mgd) to 12,100 m /day
(3.2 mgd). At average flow, the retention time in the flocculator was 9
min, and 67 min in the sedimentation basin, corresponding to an overflow
race of 61.6 m/day (1467 gpd/ft ). The filter loading rate was 9.0 m/hr
(3.7 gpo/ft2).
The plant results in Figure 45 shew that the percent turbidity removal
increased with increasing influent turbidity, whereas the effluent turbidity
was not greatly affected by the higher influent values. A comparison of
turbidity removals with and without alum addition indicates that no major
benefit resulted fran its use. This is further illustrated in Figure 46
where an alum dosage of 20 mg/L resulted in turbidity removals ranging front
50 to 98%.
The high removal variability was primarily due to fluctuations in pH
during coagulation and flocculation. The pH ranged from 6.6 to 7.4 and the
lower removals were observed at the higher pH values. The apparent
inability of alum to affect the overall performance of the plant at other
dosages was again related to operating at high pH values.
Data obtained by the EPA pilot plant treating the exact same water, had
indicated the pH optimum to be at 6.7 pH units. At this pH, an alum dosage
of 10 mg/L and a 10 m/hr (4.1 gpm/ft ) filtration rate, the pilot plant
reduced the turbidity to 0.03 NTU, a 98.3% reduction. In fact, the lowest
daily average effluent turbidity at the full scale plant during the study
period, 0.25 MTU, occurred at a process pH of 6.7. The alum dosage was 22
mg/L ano the filter loading rate 7.2 m/hr (3.0 gpm/ft ). The raw water
turbidity that day was 2.0 MTJ, thus yielding an 87.5% reduction. It was
felt that the alum dosage could have been reduced without adversely
affecting plant performance. Possible benefits from an efficiency
standpoint, would be lower effluent turbidity and longer filter runs.
Some operational changes at the plant were considered as a result of
the pilot plant work. One of them, a closer monitoring of the raw water pH
as it entered the flocculator, was well underway toward the end of the
study. This included reducing the amount of soda ash added to the raw
water. Instead, additional soda ash was added to the clearwell to increase
the pH before distribution as a corrosion control measure.
89
-------
I
£
100
so
100
50
ALUM ADDITION
£3
ADD1T10N
1 2 3 4 5 6
RAW WATER TURBIDITY (NTU)
Figure 45. Turbidity removal at Hoquiam Water Treatment Plant.
90
-------
100
6.0 6.5
50
s
7.0
7.5
O
O
°So
'
— 0
§
!|§
I I
0 80 o
o o00, °
Q-O—O
1
20
ALUM DOSAGE (MG/L)
30
Figure 46. Effect of alum dosage and pH on turbidity removal at Hoquiam
Water Treatment Plant.
91
-------
EPA Pilot Plant at Leavenworth
Wlater to the pilot plant was supplied from a fire hydrant located on
the iaw water line directly adjacent to the City of Leavenworth Water
Treatment Plant. The raw water turbidities were very low during the fall,
generally around 0.3 NHJ, with a range of 0.22 to 0.85 NIU. During
September and October the water temperature averaged 8.5 C, alkalinity 24.0
mg/L and pH 6.8.
Host of the testing was conducted at a filtration rate of 10.7 m/hr
(4.4 gpm/ft ) which corresponded to the maximum filter loading rate at the
full scale plant. The optimum alum dosage during this time period was 15
mg/L, resulting in a 90% reduction in turbidity. The corresponding
cyst-sized particle removal was 96%, with the maximum 98% occurring at an
alum dosage of 13 mg/L (Figure 47). The pH optimum was 6.7. At pH 6.4 and
7.1 both turbidity and particle removals were reduced (Figure 48).
The influent turbidity, as indicated in Figure 48, showed sane
variability. However, the highest recorded value, 0.85 MTU, was an isolated
peak associated with a heavy rainstorm. The more moderate fluctuations did
not seem to have much impact on the effluent quality. For the most part,
the effluent turbidity would vary from 0.02 to 0.03 NTU when the plant was
operated at or near optimum conditions. Since this threshold value was
below the limit of sensitivity claimed by the manufacturer of the
flow-through turbidimeters used, it was verified by grab samples on a bench
model.
During the month of November the average water temperature dropped to
3 C, the alkalinity decreased to 12.5 mg/L and the pH to 6.4. These
changes had a noticeable impact on the effluent quality. The turbidity
removal decreased from 90 to 50%, and only 48% of the particles were
retained by the filter. To improve upon the plant's performance, a new
evaluation of the optimum operating conditions was made. It indicated that
the optimum alum dosage had been reduced from 15 to 7 mg/L and the pH
optimum increased from 6.7 to 7.0. However, the performance was poor
compared to earlier runs, and the effluent turbidity did not reach a stable
value following filter ripening, but decreased rather slowly throughout the
run. As a result the turbidity removal was at times lower than the 50%
experienced when no coagulant was used, and was never better than 61%. It
was suspected that because of the low water temperature, the 8 min retention
time in the flocculator was inadequate for proper floe formation. Hence,
the necessary pretreatment was not achieved before filtration. To further
investigate this theory, raw and finished water was analyzed for aluminum.
It was not surprising to find that at times as much as 70% of the alum
coagulant added was passing through the filter. Wnen the Magnifloc 985N was
used as filter aid, an increase in particle removal up to a dosage of 0.026
mg/L was noted. A lesser improvement was observed in turbidity removal.
The addition of Calgon L-650E as filter aid lowered turbidity and particle
removal.
92
-------
O TESTS AT 8.5°C
TESTS AT 3°C
201
ALUM DOSAGE (HG/L)
Figure 47. Effect of alum dosage and turbidity removal at different
temperatures during field work at Leavenworth.
93
-------
3.0
'-2.0
;i.o
100
50
0
100
50
A
O TESTS AT 8.5°C /^
J \
A TESTS AT 3°C A
Table 48. Effect of pH on particle and turbidity removal at different
temperatures during field work at Leavenworth.
94
-------
Polymers as primary coagulant at 2 C showed removals comparable to that
experienced with alum at this low temperature (Figure 49) . Maximum
turbidity removal of 59% was realized "at a dosage of 0.2 to 0.4 mg/L,
compared to 43% renoval with no coagulant. The particle removal decreased
at low temperatures from 45 to 12% when 0.2 to 0.4 mg/L of polymer was
added. To optimize particle removal uncter these conditions, a polymer
dosage of 3.5 mg/L was required. However, the high polymer dosage decreased
the turbidity removal to 35%, apparently a result of the colloidal
restabilization by the polymer.
A greater cyst-sized particle renoval was noticed when lower effluent
turbidities were reached. A frequency distribution plot (Figure 50) of the
alum runs at 8 C and 3 C show that a median particle removal of more than
90% was realized at an effluent turbidity of 0.1 NTU. Median particle
removals of 64 and 68% were realized at an effluent turbidity of 0.1 to 0.2
NTU and 0.2 to 0.3 NTU, respectively, while lower particle rentals were
associated with higher effluent turbidity values. The results from the runs
with polymer (Cat Floe T) as primary coagulant showed a trend opposite that
of the alum data, as the lowest particle removals were observed at the
lowest effluent turbidity values. An effluent turbidity of 0.1 co 0.2 NTU
gave a 53% median removal. At effluent turbidities between 0.1 and 0.3 NTU,
better particle removals were obtained with alum than polymer, but the
opposite was true above 0.3 NTU.
Tb determine the ability of the pilot plant to remove cysts, 1.25 x 10
cysts were added to the raw water over a 320 min period, and the filter
influent and effluent sampled and analyzed for cysts. Prior to the cyst
addition, a salt solution had been added to the influent water and traced
through the plant to determine .suitable sampling times. The plant was
operated at 10.7 m/hr (4.4 gpn/ft ) filtration rate with 1.2 mg/L Cat Floe T
as the coagulant. The raw water turbidity was 0.33 NTU and the temperature
1 C. During the cyst addition the effluent turbidity was 0.19 NTU, a
42.4% reduction. The three filter influent samples recovered a calculated
867 cysts while 242 cysts were recovered from the effluent corresponding to
a 72.1% removal. The particle removal was 53.3%.
at Leavenworth
The full scale plant was operated at polymer dosages ranging from 0.2
to 0.6 mg/L of Cat Floe T. No improvement in turbidity removal was noted at
higher dosages (Figure 51). In addition, 0.06 mg/L L-650E was added to the
inlet flume, ahead of the filters, as a filter aid. At an influent
turbidity of 0.30 NTU which was quite common during dry weather, the
effluent turbidity was 0.13 NTU or a 57% removal. Higher removals were only
associated with higher raw water turbidities. A 0.21 mg/L dosage of Cat
Floe T at an influent turbidity of 1.0 NIU resulted in an 84% removal.
During most of the study period, some of the solenoid valves
controlling filter operation did not function properly. The result was a
loss of vacuum. This sometimes occurred during the night when the plant was
95
-------
2345
POLYHEh FLOCCULANT (MG/L)
Figure 49. Effect of Cat Floe T polymer dosage on particle and turbidity
removal and rate of headless buildup at Leavenworth.
96
-------
100
80
—a 60
V«UJ
>• o
£5
UJ
ce 3:
So
t— UJ
Si
0
IOC
80
60
|i 40
UJ
oc :r
t_>
20
o_ «—
o
o
S o
O 0.01 - < 0.025 NTU EFFLUENT TURBIDITY
A 0.025 - < 0.05
V 0.05 - < 0.10
D 0.10 - < 0.20
EH 0.20 - < 0.30
9 0.30 - < 0.10
< 0.40
10
PERCENT OF MEASUREMENTS
WITH LESS THAN INDICATED PARTICLE REMOVAL
Figure 50. Frequency distribution of particle removal at different effluent
turbidities during alum and polymer treatment at Leavenworth.
97
-------
INFLUENT 2.6 NTU
FILTERED THROUGH
WHATMAN 40 at 15"
MERCURY PRESSURE
0.1
0.2 0.3 0.4
POLYMEK FLOCCULANT (MG/L)
0.5
0.6
Fiqure 51. Effect o," polymer dosage en turbidity removal
at Leavenwnrth Wat^r T refitment Plant
98
-------
unattended. Without vacuum none of the siphons could be initiated. As a
result, if a filter reached terminal headless, the inlet siphon would be
broken. With no vacuum it could not be backwashed. Bus could, in fact,
shut down the entire plant until the vacuum was reestablished.
Summary of Field Activities •
During the nearly seven months of field operation, a total of 49 runs
were made. Of these, 30 were performed at Hoquiam between Nay 7 and
September 5, 1980.
Although the City of Hoquiam Water Treatment Plant was a conventional
plant whereas the pilot plant was operated as a direct filtration plant,
much of the information obtained by _ this study could be and was used to
evaluate the full scale plant operation. Having identified the relative
importance of the key unit process variables, a few operational changes
relating to chemical addition were made. The benefit from these changes was
improved pH control during alum coagulation and flocculation and better
utilization of the nonionic polymer when used as a filter aid.
From September 17 through November 29, 1980, the pilot plant was
operated at Leavenworth. A total of 19 runs were made. Cold weather during
the last two weeks of operation caused the water temperature to decrease
significantly, fluctuating between 2 and 3 C. The result was poor floe
formation, particularly when alum was used as the coagulant.
The city's treatment plant did at times experience operational proolems
due to equipment malfunctioning, however, these were later corrected ana new
equipment was installed to improve the rapid mix process.
99
-------
Boeck WC. On the longevity of human intestinal protozoan cysts. Amer. Jour
Hygiene 1:527-540, 1921.
Center for Disease Control. Giardiasis, Vail Colorado. Morbidity, Mortality
Weekly Report 27:155, 1979.
Center for Disease Control. Intestinal parasite surveillance. Annual Sunmary,
Atlanta, 1979.
Chandler AC and Read CP. Introduction to parasitology. John Wiley and Sons,
New York, 1961, p. 10.
Davis KB and Hibler CP. Animal reservoir and cross species transmission of
Giardia. In: Waterborne transmission of giardiasis. USoPA, Cincinnati,
EPA 600/19-79-001, 1979.
Dobell CA. The discovery of intestinal protozoa in man. Proc Royal See Med
^3:1-15, 1920.
Fantham HB and Porter A. The pathogenicity of Giardia (lamblia) intestinalis
to men and experimental animals, Brit Med Jour £: 139-141, 1916.
Frost F, Plan B, Liechty B. Giardia prevalence in cosmercially trapped
rranmals. Jour Environm Health 42_: 245-249, 1980.
Gcodbar JP. Join synptcms in giardiasis. Lancet 1:1010-1011, 1977.
Kimer J, Littler JD, Angelo LA. A waterborne outbreak of giardiasis in
Camas. Jour Aner Water Works Assoc 70:35-40, 1978.
Kafoid CA and Christiansen EH. On the life history of Giardia. Proc Nat Acad
Sci 1:547, 1915.
Konenenko VM. Erythema multiform exudatinum in a child with lamblia
cnolycystitis. Peuiatr Akush Genekol 2_:30-31, 1976.
Levine ND. Giardia lamblia; classification, structure, identification. In:
Waterborne transmission of Giardiasis, USE i 600/19-79-001, Cincinnati, 1979.
Lippy EC. Tracing a giardiasis outbreak at Berlin, New Hampshire. Aner
Water Works Assoc 512-520, 1978.
IOC
-------
Lopez CE, Dykes AC, Juranek DD, Sinclair SP, Conn JM, Christie RW, Lippy EC,
Schultz MG, Mires MH. Waterborne giardiasis: a connunity wide outbreak of
disease and a high rate of asymptomatic infection. Amer Jour Epid 112:495-506,
1980.
Moore GT, Cross VM, McGuire D, Mollahan CS, Gleason NN, Healy GR and Newton
IB. Epidemic giardiasis at a ski resort. New England J Med 281;402-407, 1969.
Rebhun M, and Argaman Y. Evaluation and Hydraulic Efficiency of Sedimentation
Basins. Jour SED, ASCE 9.1:37, 1965.
Rsndtorff RC and Holt CJ. The experimental transmission of human intestinal
protozoan parasites IV: Attempts to transmit Endomoeba coli and Giardia
lamblia cysts by water. Amer Jour of Hygiene 60_: 327-328, 1954.
Shaw PK, Brodsky RE, Lyman DO, Wood BT, Hibler CP, Healy GR, McCleod KI, Stahl
W and Schultz MG. A comnunity-wide outbreak of giardiasis with evidence of
transmission by a municipal water supply. Ann Intern Med £7:426-432, 1975.
Sheffield HG and Bjorvatn B. Ultrastructure of the cyst of Giardia lamblia.
Amer Jour Trop Med Hygiene 26_: 23-30, 1977.
Schultz MG. Giardiasis. Jour Amer Med Assoc 222:1383-1384, 1975.
Veazie L, Brownlee I and Sear HJ. *n outbreak of gastrointeritis associated
with Giardia lamblia. In: Waterborne transmission of giardiasis. USEPA,
Cincinnati, EPA 600/19-79-001, 1979.
Webster BH. Human infection with Giardia lamblia; analysis of 32 cases.
Amer Jour Digest Disease 3_:64-71, 1958.
Wolfe MS. Managing the patient with giardiasis: clinical diagnostic and
therapeutic aspects. In: Waterborne transmission of giardiasis. USEPA,
Cincinnati, EPA 600/19-79-001, 1979.
101
-------
APPENDIX
Electron Microscopy of Giardia lamblia Cysts
DANIEL L. LUCHTEL,* WILLIAM P LAWRENCE. AND FOPPE B DtWALLE
Department of Environmei. . Health. School of Public Health and Community Medicine. Vnn ersily at
Washington. Seattle. Washington 98195
The flagellated protozoan Giardia lamblia is a recognized public health prob-
lem. Intestinal infection can result in acute or chronic diarrhea with associcled
symptoms in humans As part of a study to evaluate removal of G lamblia cysts
from drinking water by the processes of coagulation and dual-media filtration, we
developed a methodology by using 5 0-/im-porosity membrane filters to evaluate
the filtration efficiency. We found that recovery rates of G lamblia cysts by
membrane filtration varied depending upon the type and diameter of the mem-
brane filter Examination of membrane-filtered samples by scanning electron
microscopy revealed flexible and flattened C lamblia cysts on the filter surface.
This feature may be responsible for the low recovery rates with certain filters
and, moreover, may have implications in water treatment technology Formation
of the cyst wall is discussed Electron micrographs of cysts' apparently undergoing
binary fission and cysts exhibiting a possible bacterial association arc shown
Exposure to the waterborne pathogen Giardia
lamblia is a current public health problem (10)
as exemplified by recent outbreaks of giardiasis
reported from Vail, Colo. (7), Berlin, N.H (15),
and Camas, Wash. (11) These outb.eaks oc-
curred in municipalities that use surface water
for drinking purposes. Each of their seemingly
adequate water treatment facilities failed to fol-
low proper treatment procedures of the raw
water. G. lamblia cysts were detected in the
finished water at both Berlin and Camas The
percentages of stool specimens positive for G
lamblia cysts reported by U S state laboratories
in 1976 were 0 2 in California, 9 6 in Colorado.
10.6 in Minnesota, 9.5 in Maine, and 6.3 in Wash-
ington (2)
The work reported here is part of a study that
determined the efficiency of a water treatment
plant for removing G. lamblia cysts. Experi-
ments showed that >99% of the cysts introduced
into a water treatment pilot plant can be re-
moved by the processes of cnagulation-floccula-
tion, sedimentation, and dual-media filtration
(W. P. Lawrence, Masters thesis. University of
Washington, Seattle, 1979). The efficiency of
cyst removal was evaluated by filtering the fin-
ished water from the pilot plant. In also evalu-
ating the reproducibility of our filtration proce-
dure with known concentrations of cysts, we
found that the recovery rates of cysts that were
passed through two different types (MiUipore
and Nuclepore) and diameters (47 and 293 mm)
of membrane filters varied considerably.
Electron microscopy was used to determine
the possible causes of these various rates We
found that the cyst Mall of G lamblia is remark-
ably flexible and concluded that the interaction
of the flexible cyst wall in the filter pore ma\
explain the different recovery rates on different
types and sizes of fillers.
MATERIALS AND METHODS
Fecal material was collected from human giardiasis
patients in cooperation with the Washington Slate
Paroaitology Laboratory. Seattle The material was
fixed in either 5% buffered Formalin or 2*1 glutaral-
dehyde in 0 1 M cacodjlate. which was dune immedi-
ately after positive identification of G lamblia cists
in the feces A given quantity of the fecal material was
diluted 1.2 in distilled water, stirred into a liquid
suspension, and Tillered through three Ia\ers of gauze
that approximated a 50- to 80-|im-mesh sieve The
filtrate was centnfuged at 400 x g After the super-
nntant was decanted, the sediment was emulsified with
an equal amount of distUled water
We used the method of Sheffield and Bjonatn (30)
to further separate the cysts from other fecal material
A 5-ml amount of the fecal suspension was added to a
discontinuous densitv sucrose gradient consisting of 5
ml each of I 5. 1.0. 075. and 05 M sucrose solutions
added successively to a 40-ml conical centrifuge lube
After centnfugalion for 30 mm at 1.000 X /». approxi-
mately 4 ml was collected b\ capillar\ pipette from
both the water-05 M sucrose and 05 M-075 M su-
crose interfaces This, suspension, consisting of cxsts
and small noncvst paniculate debn<- *a« diluted 10-
fold wnh distilled water and centnfuged for 3 in 5 mm
at 400 x g The sediment, rnnsisting of a high number
of cysts relatively free of debris, was again diluted 10-
fold with distilled water and kept at 4°C until use We
eliminated the final nitration, a* recommended b\
Sheffield and Bjorvatn (20), through a 20-/im Tiller to
remove any remaining debris
Known quantities ofcvsus were added to an exper-
imental water supply and tented in a pilot water trcai-
*Reprinted with permission from Applied and Environmental Microbi-
ology, Oct. 1980, Vol. 40, wo. 4, pp. 821-832.
102
-------
822 LUCHTEL, LAWRKNCE. AND DeWALLE
APPL. ENVIKOK MICROBIOL.
ment plant for the efficiency of cyst removal (W P.
Lawrence. Masters thesis. University of Washington,
Seattle. 1979) It was necessary to develop a quanti-
tative method with a known recovery efficiency that
would retain any cysts still remaining in the finished
water after passing through the water treatment plant.
We developed a recovery method that used mem-
brane filters of 5fim pore sue to retain G lamblta
cysts. We first tested two filters of a small diameter
(47 mm) We soon found that it was necessary to test
more expensive. larger-diameter filters (293 mm) to
maintain filtering efficiency for the relatively large
volumes of water from the treatment plant. The re-
covery efficiency of the filters was tested in the follow-
ing way.
Aqueous suspensions of fixed G lamblia cysts were
passed by vacuum through 5 0-um-porosily Millipore
(Millipore Corp.. Bedford, Mass ) or 5.3-fim-porosily
Nuclepore (Nuclepore Corp. Pleasaoton. Calif) mem-
brane filters Concentrations of cysts before and after
filtration were determined uy enumeration on a Clay-
Adams model 4011 Spencer Bnght Une hemacytom-
eter and collaborated with counts on a Coulter
Counter (Coulter Electronics. Hiajeah, Fla) Cysts
were removed from the 47-mm filters b> immersing
each filter in 10 ml of distilled water in a small flask
and agitating gently by hand The filter was then
discarded, and the liquid was examined for presence
and quantity of G lamblia cysts The larger 293-mm
membrane filters were processed by using a two-step
centnfugation process summarized in Fig I Kecovery
rates of cysts from the different types and diameters
of filters were then calculated
For the scanning electron microscopic studies.
aqueous suspensions of fixed G lamblia cysts were
20 liters of prefiltered (5 0 ,aa) tap water
1
Pass through 293-mm (5 (tyro-pore sue)
Nuclepore filter at 10 lb/in* with nitrogen
gas. 0.2-fim filter on nitrogen tank
1
Filter remowd and placed in shallow dish
Cysts washed off by agitation of filter in 0 3
bl«r of water (platform shaker.
Toothmaster Co, Racine, Wis)
I
Centrifuge retentate at 1.500 rpm for 10
mm in eight 50-ml conical bottom tubes
1
Retain "sediment" (approximately
10 ml x 8)
1
Transfer to two 50-ml conical tubes and
recentnfuge
J
Retain "sediment" (approximately
5 mix 2)
1
Enumerate on Coulter Counter and
compare with initial cumenlration
FlC 1. Summary of method uted for the recoiviy
of G. lamblia cysts from 293 mm diameter 5 O-/JJTI-
porosity Suclepore filters
filtered bv gravity through 47-mm-diameler 50-pm-
porositv Millipore or Nuclepare membrane fillers The
filters were air dried, and small pieres of the fillers
were cut out and stuck 0:1(0 stubs covered w ith double-
stick tape. Oihyr cyst suspensions were critical point
dried to avoid membrane filtration and sir drying The
aqueous suspensions were post fixed in 1% OsO. in 0 15
M cacodylate, dehydrated in ethanol, and critical-
point dried with COi After each step of the postfua-
tion and dehydration procedure, the suspen. .ons were
bnefiy ceninfuged. and the fluid was decanted For
the critical-point drying step, the suspensions were
enclosed in BEEM capsules (Better Equipment for
Electron Microscopy. Inc. Bronx. N Y.) capped on the
two ends with 5 0-pm-porosity Nuclepore filters (a
modification of the procedure of Hayunga [8]). After
cniical-poml drying, the BEEM capsules were ope:ied.
and the dried cysts were t>pnnkled onto stubs covered
with double-stick tape The stubs were coated with
gold-palladium in a Demon Vacuum Desk-1 sputter
coaler and viewed in a JEOL JS.M-35 scanning elec-
tron microscope (JEOL. Tokyo. Japan)
For the transmission electron microscopy studies,
aqueous suspensions of fixed G lamblia cysts were
post filed in osmium, dehydrated in ethanol. and
embedded in Epon Thin sections were stained with
uranyl acetate and lead citrate and viewed with a
JEOL JEM 100S electron microscope.
RESULTS
Since a subsequent part of the overall study is
concerned with the efficiency of a water treat-
ment pilot plant for the removal of G lamblia
cysts (Lawrence and DeWaile. manuscript in
preparation), we needed to develop and evaluate
a quantitative method with a known recovery
efficiency that could be used to determine the
number of cysts in a given volume of water
Known quantities of cysts were filtered, and the
recovery efficiency was determined Four differ-
ent methods were checked against each other.
Recovery rates of G. lamblia cysts with the
47-mm-diameter 5 0-um-porosity Miliipore and
Nuclepore filters were comparable (Fig. 2) The
same recovery rate, approximately 75?, was
found when the 293-mm-diameter Nuclepore fil-
ter was used (Fig 3) A significantly lower recov-
ery rate, approximately 255r, was found after
filtering cysts with the 293-mm-diamel»r Milli-
pore filter. Coulter Counter and hemac>tometer
counts of the filtrates showed that no cysts
passed through the filters The reasons for the
less than 100% recovery from the filters and the
strikingly lower recovery on the large Millipore
filter were unclear Therefore, it was decided to
study the filter surface with scanning electron
microscopy.
Cysts collected on either atr-dned Millipore
or Nuclepore membrane filters exhibited dis-
torted or flattened cyst walls (Fig. 4 to 11) The
pattern of such flattening of the cyst wall was
103
-------
VOL. 40.1980
ULTRASTRUCTURE OF C1ARDIA CYSTS 823
100
o
o
«
OC 0
» 100
u
^
v
to"
10'
510-
Cyst Cone. Ino/mll
FlC. 2. Recovery rates (D of G lamblia cysts from
4A) 47 mm Nuclepore 5 0 iim-porosily filters and (B)
47-mm Milhpore S j-ian-porosuy filters.
100
o
u
0)
OC
_ SO
Ol
u
10"
10=
Cyst Conc.lno/mll
FIG 3 Recovery ratet> (9)ofG lambha cysts from
(A) 293-mm Nuclepore S 0-fim porosity filter!, and (B)
293 mm Millipore S 0-p/n porosity fillers.
different for cysts collected on Millipore filters
compared with cysts on Nuclepore filters The
surface of the Millipore filter consists of inter-
ir.eshed strands (Fig. 4 and 5). and the diameter
of the individual strands is much smaller than
the 5.0-um pore size The distortion of the cyst
on the Millipore surface seemed to be deter-
mined to some extent by how it rested on the
small individual strands (Fig. 5). For the cysts
retained on the surface of a Nuclepore Tiller, the
pattern of distort ion Mas distinctly different (Fig
6 and 7), apparently because of the smoothness
of the Nucltpwrti surface. A fairly uniform, run-
like structure was apparent around those cysts
that rested on the flat surface of the Tiller (Fig.
7 and 8). Cysts that overlapped the filter pore
were sharply bent into the pores (Fig 6, 10, and
11). Overall, more cysts per unit of area were
readily seen on the Nuclepore than on the Mil-
lipore filters. Although there seemed to be fewer
cysts on the Millipore Tillers, it was more difficult
to delect the cysts on the rough Millipore sur-
face.
Wj observed sectioned material by transmis-
sion electron microscopy (Fig 12) to confirm the
presence of Giardia cysts. CysU prepared via
cntical-pcjit drying were not flattened (Fig. 13,
see also Fig. 14 to 17) Rather, such specimens
appeared ovoid or spherical and agreed with the
transmission electron microscopic observations.
The possible forces that may act on cysts to
distort them during the processes, of nitration
and air drying are considered below.
Some additional observations were made on
the material that had been prepared for electron
microscopy. Some of the cysts appeared to shew
a process of division (Fig 8 and 9) One
"stretched" cyst was found, apparently an arti-
fact caused by the preparative procedures (Fig.
10).
With the scanning electron microscope, a va-
riety of material was observed on Ihe cyst wall.
This was particularly evident on cntical-poml-
dneo specimens (Fig 13 to 17) Air-dned cysts
were usually free of such material (Fig. 6). Oc-
casionally, bacterium-like structures were asso-
ciated with the cysts (at the upper nght and
lower left of the double cyst shown in Fig. 10
and at the nght of the cyst shown in Fig. 15).
One cyst in Ihe sections prepared for transmis-
sion electron microscopy showed a bactenum-
like structure associated with Ihe cyst wall (Fig.
18)
Our transmission electron microscopic prepa-
rations usually showed a rather wide space be-
tween the o/ganism and the cyst wall (Fig 12
and 19) A peripheral array of vesicles was char-
acteristic for most organisms. A dense-staining
material coated the inside surface of these pe-
ripheral vesicles. A few larger peripheral lacunae
were seen (asterisk in Fig 19) The inner surfaces
of the lacunae were lined with a dense-staining
material A dense material also coated the ii.ner
surface of the cyst wall and the surface of the
encysted organism
DISCUSSION
Information about ihe biology of Giardia or-
ganisms, Ihe incidence of giardiasis, and the
ullraMructure of these parasitic protozoans is
reviewed in three recent publications (1, 10,13).
Several scanning electron microscopy studies on
the trophozoite (4,17,23) complement transmis-
sion electron microscopy studies (3, 6, 18, 19.
additional references in 13) Previous ultrastruc-
tural studies of the cvst are those of Sheffield
and Bjorvatn (20), Sheffield (19). and Tombes
etal. (21).
104
-------
824 LUCHTEL, LAWRENCE. AND DeWALLE
APPL. ENVIRON. MICROBIOL
FlC. 4-7. Scanning electron micrographs showing cysts collected on filters by gravity filtration and then
air dried.
Pic. 4. A luu magnification nttr that shows thret G. lamblia cyttt (arrows) on a S.O-tan-porosity Millipore
filter. Bar. 20 urn.
Fie. 5. A higher-magnification o>u> of the middle cyst shown in Fig. 4. The cyst is flattened and distorted.
The distortion* setm to depend on hou- the cyst rests on the contours of the filter surface. Bar. S jun.
Fie. 6. A lou'-magnification rieu- of a 5.0-pm-porosity Nuclepore filter that shows several cystf (arrows)
and some unidentified debris, presumably consisting of fecal material and ruptured cysts (arrowhead). Bar,
20um.
Fie. 7. A higher-magnification tieu of a cyst, comparable to those shown in Fig. 6. The cyst i* flattened on
the filter surface and typically shows a thin outer rim or flange. The central conitx portion of the cyst is
caused by the encysted organism. Bar, 2 pm.
Flexible cyst wall. Although the above ul-
trastructural studies (and this study) provide
detailed information about the structure of the
trophczoite and the cyst, it seems worthwhile to
begin this discussion by referring to the earlier
work of Filice (5), who observed fresh, unfixed
preparations of cysts. He noted that the cyst is
a flexible structure since he saw that the organ-
ism could move about inside and deform the
cyst wall. He also observed that the cyst wall
had enough strength to keep its shapp when the
protoplasm within disintegrates and, also, that
the cysts do not explode when immersed in
distilled water.
The most striking feature of the cyst wall
shown in our initial observations on air-dried
preparations is its flexible nature, even after the
cyst is fixed in glutaraldehyde or Formalin. Such
105
-------
VOL. 40.1980
ULTRASTRUCTURE OF GIAKD1A CYSTS 825
Pic. &-1 J. Air-dried cysts collected on Nuclepore filter*. Apparently, encysted organisms are able to divide,
and the cyst wall U then restructured to enclose separately each of the two neu-ly formed organisms.
FIG. 8. A *ingte cyst in which the organism inside appears to be in the process of dividing. Bar, S pm.
Fic. 9. A double cyst, apparently formed after an organism within a single cyst had divided. The arrows
indicate a line of demarcation that separates the two cysts. Presumably, this double cyst breaks apart to form
two separate cysts. Bar, 5 tun,
Fie. 10. A double cyst that hat been stretched artifartually during the preparation and filtration proce-
dure*. Bar, 5 pn.
FlG. 11. A cyst that has become distorted, apparently because of settling into a pore of the filter. Bar, 2 tan.
flattened shapes for Giardia cysts (Fig. 5 and 7)
are not consistent with the ovoid outlines of
cysts shown by various light microscopic studies
(13) and the transmission electron microscopy
observations of Sheffield and Bjorvatn (20). We
then confirmed that our material was Giardia
cysts by transmission electron microscopy (Fig.
12) and subsequently showed that ovoid cysts
could be prepared for scanning electron micro-
scopic observation if the cystr are critical-point
d:ied (Fig. 13). Although we did not check epch
set of variables independently (filtering versus
not filtering; air drying versus critical-point
drying), most of the 'flattening is probably due
to the surface tension of water as the specimen
is being air dried. Some of the critical-point-
dried cysts were somewhat distorted (iasert. Fig.
13), possibly due to some transient air drying
during the several fluid exchanges before the
critical-point-drying step. But overall, although
the critical-point-dried cysts underwent several
filtering and centrifugetion steps, they retained
their ovoid shape. On the other hand, filtration
had some effect on the cyst morphology as the
106
-------
826
107
-------
VOL. 40.1980
ULTRASTHUCTURE OF GIAKDIA CYSTS 827
cysts appeared distinctly different on the Milli-
pore surface (Fig. 5) compared with those on the
Nuclepore surface (Fig 7)
Tombeset al (21) studied the cysts of Giardia
collected from a variety of mammals, including
humans. The morphology of the cysts they col-
lected from humans is different from that ob-
served by us. The cysts they studied by phase
microscopy had the typical elliptical shape, by
scanning electron microscopy, the cysts seemed
to be distorted, having a cuboidal shape Possible
reasons for our different results are difficult to
decide upon since Tombes et al. used a variety
of fixation and pieparative techniques, and for
any particular micrograph, the data are not
given as to how the cysts were fixed, whether
the material was fixed immediately or after some
initial filtrations (sucrose flotation techniques
were not used), how long the material was stored
in aldehyde before drying, and whether the cysts
were air dried or critical-point dried Overall,
Tombes et al. noted no consistent differences in
cysts after air or critical-point drying We found
substantial differences in cyst morphology when
cysts were air dried or critical-point dried We
suggest that a possible pn>. e Jura I =rror that
Tombes et al. mention in their discussion may
be a significant factor in our different results
Sucrose flotation technique. We used the
sucrose flotation method of Sheffield and Bjor-
vatn (20) to prepare suspensions of cysts They
apparently fixed the cysts after the sucrose pro-
cedure. If so, they obtained remarkably good
fixation after a lengthy concentration process
We fixed the fecal material before the sucrose
flotation Fur kboratorv diagnosis of giardiasis
in unfixed stools, the basic method is a zinc
sulfate flotation method i!3) With this tech-
nique, the cytoplasm of the cells is plasmoiyzed
by the hypertonic zinc sulfate solution, and the
cytoplasm is characlenst-cally concentrated at
one side of the cyst (see Fig 23 in reference 13)
Although the cyst wall is apparently stable
throughout the zinc sulfate flotation process, it
seems much more delicate when sucrose flota-
tion is used Levine (14) observed that Giardia
cysts concentrated by sugar flotation shmel and
become unrecognizable in a matter of minutes
Stevens, in a discussion after Lex ine's paper (14;,
noted that there was no morphological effect on
the cysts with the sucrose flotation technique if
the cysts were removed immediately from the
interface and placed in physiological saline With
the methodology of SKifield and Bjorvatn (20),
the suspensions ar' diluted 10-fold with water
after collecting them from the interfaces
Another possible effect of the sucrose flotation
method is that it may change the width of the
space between the cyst wall and the organism.
Sheffield (19) believes that these spaces are not
caused by the different isolomc pressures of the
flotation solutions. We found a much wider
space between the cyst wall and the organism
than that shown by Sheffield and co-workers
(19, 20) or the cyst shown by transmission elec-
tron microscopy in the study of Nemanic et al
(18). The material studied by Nemanic et al.
(18) was not exposed to a sucrose flotation tech-
nique as the organisms were prepared for elec-
tron microscopy by washing pieces of gut and
centnfuging the wash. Perhaps species differ-
ences may be a factor in comparing our results
with those ff Nemanic et al. (18) but the reasons
for our results being different from those of
Sheffield and Bjorvatn (20) are not apparent
unless they fixed the cysts af*cr sucrose flotation.
Perhaps selection of micrographs may be a con-
tributing factor as Sheffield, in a discussion after
his paper (19), states that a variety of cyst types
were seen; that is, cysts in which the cytoplasm
was closely applied to the cyst wall, whereas
others showed large, open areas between cyto-
plasm and wall. We also saw sections of cysts in
which the cytoplasm was closely applied to the
cyst wall, but since most of the sectioned cysts
showed an open space (Fig 12), our interpreta-
tion is that the organism does not occupy the
entire space of the cyst The nmlike structure
on air-dned cysts (Fig 7) would also indicate
that the cyst wall collapsed into a space not
occupied by the encysted organism We ob-
served that the cyst walls are usually 0 15 to 0 25
fim thick, which 15 less than the 0 3-fim thickness
observed by Sheffield and Bjorvatn (20).
Composition of cyst wall. The composition
of the cyst wall is unknown Filice (5) was not
able to obtain any positive histochemical infor-
mation, although he did show that it was Feul-
FIG 12 A fnuiuniuioii electron micrograph of encysted Giardia organisms The rys/ uaf/.s usually form
smooth timid outlines, although a couple ufexample1- of acutely folded cyst u>alls (arrous) can be seen (alia
tff insert of Fig IS) Bar.lOiun
Vic 13 Smooth, ovoid cyst-, after critical point drying These cysts are embedded in a mat w dump of
debru. bacteria, and fecal material A lou'magnification micrograph of the entire clump is »houn in the
lou'tr right insert The upper left inter! khou-K example* of -angle, isolated rysl« after critical point drying
Such rys/» may »Aou) same moderate degree of distortion Bar. 10 |im Loirer right insert bar. 100 jun Upper
left insert bar. S |im
108
-------
Pic. 14-17. X variety of cyst morphologies as seen a/for critical point drying. Atmott alt cysts had tome
•orl of material or tiebri* stuck on the cyst wall. In some cases, structures thai could be identified as bacteria
were attached to th* cyst mall (Fig. IS). In other cases, unidentified fibrous forms were teen on the cyst walls
(Fig. 16 and 17). Bars, 2 ion
PlC. 18. A transmission electron micrograph showing a structure, presumably bacterial in nature, attached
to the cyst wall. Tht fibrous coat of the attached structure seems to interact with the fibrous cyst uvll. The
insert shows a low magnification i-ieu of the entire cyst and attached structure. Bar. J /tin. Insert bar. 2 iim.
109
-------
VOL. 40.1980
ULTRASTRUCTURE OF GIARDIA CYSTS 829
Fie. 19. An encysted organism with its typical array of peripheral vesicles fatso see Fig. 12). K, nuclei: A,
ajconemes of the flagetlae: S, microtubule-ribbon complexes of the fragmented sucking disk. The arrow points
to a portion of the cyst trail that has apparently retained a staining density simi/or to the slr.ining density of
the inner surface of the cyst wall. The asterisk is in a peripheral lacuna. Bar, 1 fan.
gen stain negative; it did not stain with a iipid
stain, Sudan IV, and it did not seem to be
affected by various enzyme digestions (pepsin,
trypsin. and papain). Iii any case, the cyst wall
is not fixed adequately with aldehydes to with-
stand the surface tension of water during air
drying, and its flexible nature, even after fixa-
tion, may lower the filtering efficiency of various
water filtration plants. The nature of the cyst
wall needs to be taken into consideration when
110
-------
830 LUCHTEL. LAWRENCE, AND DEWALLE
APPL ENVIRON MICROBIOL
various cyst model systems are being tested For
example. Logsdon et al. (16) used 9-fim-diameter
radioactive microspheres as a model for Giardia
cysts because the cysts are difficult to obtain,
delect, and count, whereas the radioactive mi-
crosphere<; are similar in size to Giardia cysts
and are eas> to trace Our observations suggest
that such microsphcres woi'lt- be filtered more
efficiently than Giardia cysts in pilot water fil-
tration plant!)
Loss of cysts during membrane filtrati jn.
The maximum rate of recovery obtained from
the Millipore and Nuclepore membrane filters
was /5% A number of factors may account for
the 2o* loss. Cysts may remain attached or
embedded in the filter after the recovery proce-
dure, adhere to nonfilter surfaces of the nitration
assembly, pass through the filter, or be de-
stroyed during the filtration or centrifugation
process or both.
The filters were agitated by hand as vigorously
as possible without destroying the filters It was
later suggested that perhaps a better method
would be to vigorously and systematically wash
the filter surfaces with strong streams of distilled
water with 001% Tween 20 from a capillary'
pipette We did not test such a washing proce-
dure Compared with the unidimensional surface
of the Nuclepore filter, the convoluted fibrous
structure of the Mdlipore filter may permit cysts
and other material to become embedded within
the depth of the filter, and, by our recovery
procedure, tnt cysts would not be readily warned
out Such differences in the filter chiiractenstics
may explain the difference in recovery rates
between the 293-mm-diameter MUlipore filler
and the 293-mm-diameter Nuclepore filter.
What is still puzzling are the comparable recov-
ery' rates of the 47-mm-diameter Millipore and
the 47-mm-diameter Nuclepore filters However,
a 293-mm-diameter filter has approximately 39
times more surface area than a 47-mm-diameter
filter. Thus, although there may be some differ-
ence in recovery' rates for the 47-mm-diameter
Millipore and Nuclepore filters, perhaps we were
not *
-------
832 LUCHTEL. LAWRENCE. AND DeWALLE
J C Hoff M). Waterbome triiunnition of gurdiaua
Ujt Environmental Protection Agency. Cincinnati.
OhK.
15. Uppy. E- C. 1978 Tracing a giardiasis outbreak at Berlin,
New Hampshire J Am. Water Works AMOC 70:412-
520
16 Logadon, G. &, I. M. Symoiu, and R. L. Hoye. 1979
Water fiKralion terhniqua for removal of cjnu and
cyit nuxLla. p 240-256 In W Jakutxmtki and J C
Hoff
diatnbution in Guidia mum and Giardia lamblia J
Infect Dm 140:2! 2-228.
APPL. ENVIRON MICROBIOL
19 Sheffield. H. G. 1979 The uluutruciural aipecu of
Cun/ia.p 9-21 InW Jakubowalu and J C HoFfled).
Waterbome irinsmiuiuon of nardia.ui \J-S Environ-
mental Protection Ae.enr> Cincmnau. Ohiu
20 Sheffield. H. G . and B. Bjonatn. 1977 Ulirastructure
of the CJT" of Giardia lamblia Am J Trop Med H)g
18:23-30
21 Tombe*. A- S, S. S. I^ndfnrd. and U D. Williuna.
1979 Surface morphology of Ciardia c>*sta reco\ered
from a variety of host*, p. 22-37. In W Jakubomlu and
J C Hoff led I. Waterbome tranunisaon of giardiaiu
US Environmentel Praucuon Agency. Cincinnati.
Ohio
22 TYater, W. 1964 The cytoplasm of proune*. p 81-137
la J Brwhet and A E Muiky led ). The cell. vol. 6
Academic Pteaa. Inc. New York.
23 Wataon. J. H, L.. J. Goodwin, ud K. 5. Rjtjan. 1979
Cionfia lamblia in human duodenum and bile Micron
10.61-64.
112
------- |