DTIC ADA241527: Evaluation of Military Field-Water Quality. Volume 8. Performance of Mobile Water-Purification Unit (MWPU) and Pretreatment Components of the 600-GPH Reverse Osmosis Water Purification Unit (ROWPU), and Consideration of Reverse Osmosis (RO) Bypass, Potable-Water Disinfection, and Water-Quality Analysis Techniques

Survival, Water, Medical Field Manuals

Military Manuals

Defense Technical Information Center

Document text

m 


•  *» 


AD-A241  527 


\ 


T  | 


;il  % 


f 


AD 

UCRL-21008  Vol.  8 
Approved  for  public  release 
Distribution  unlimited 


Evaluation  of  Military  Field-Water  Quality 


Volume  8.  Performance  of  Mobile  Water-Purification  Unit  (MWPU) 
and  Pretreatment  Components  of  the  600-GPH  Reverse  Osmosis  Water 
Purification  Unit  (ROWPU),  and  Consideration  of  Reverse  Osmosis  (RO) 
Bypass,  Potable-Water  Disinfection,  and  Water-Quality  Analysis  Techniques 


R,  E.  Selleck 
Z.  Ungun 
G.  Chester 
V.  Diyamandoglu 
J.  I.  Daniels 
B.  Mariflas 


May  1990 


m 

m . 


immM 


Supported  by 

U.S.  Army  Medical  Research  and  Development  Command 
Fort  Detrick,  Frederick,  MD  21701 
Project  Order  82PP2817 
Project  Officer:  Dr.  Stephen  A.  Schaub 


astir 


X  JL  U 


x  Vi  O 


ii  a'jjuagmaMaifeiaai 


REPORT  DOCUMENTATION  PAGE 


Form  Approved 
OM»Ho.on*4il» 


ia.  report  security  classification 
UNCLASSIFIED 


2*  SECURITY  CLASSIFICATION  AUTHORITY 


2b.  OKIASSIFICATION/ DOWNGRADING  SCHEDULE 


4  FIRFORMING  ORGANIZATION  RCPORT  NUMEER(S) 

UCRL-21008  Vol.  8 


a.  NAMI  OF  FIRFORMING  ORGANIZATION 
Lawrence  Uvarmora 

National  Laboratory 


<c  AOORIJS  (Gty,  Statt,  and  ilf  Coda) 

Environmantal  Sciancaa  Division 
P.0.  Box  5507,  L-453 
Livaraora,  CA  94550—0617 


a.  NAMI  OF  FUNOING/  SPONSORING 

ORGANIZATION  U.S.  Army  Medical 
lasearch  and  Davelopmant  Command 


It  AO  DM  Is  (City,  SUf,  and  WCodii 
Fort  Datrick 

Fradarick,  MD  21701-5012 


lb  RISTRICTIVE  MARKINGS 


3  OlSTRlfUTION  i  AVAILABILITY  OF  REPORT 

Approved  for  public  release; 
distribution  unlimited. 


5.  MONITORING  ORGANIZATION  RIPORT  NUMBER(S) 


Ta.  NAME  OF  MONITORING  ORGANIZATION 

U.S.  Amy  Biomedical  Resaarch  and 
Davalopmant  Laboratory 

AOORESS  (C/ty,  srata,  TnSTFcaSSJ 
ATTNi  SGRD-UBZ-C 
Fort  Datrick  (Building  568) 


5HMfM3E3in  G  nnarTfniTTT  inaRrmiTJVTZM.'i’iRrr.i  tti 


Army  Project  Ordar  82PP2817 


I  ■EMI-JLH  g-UfllN-TOTlim  H.- 


ifljT«^^:>?r™.'iJnrTT.?7T7rai: 


. .  wnouwv  wumf  U«*unsf  non,  evaluation  Of 

85^SS!lS«!“fufaiysi5S°Dn!t1SoM!' 


■  4  v.vr  a?n  is  in  m 

IFWTOT* 


NAL  AUTHORS) 

R.E.  Sallack.  Z.  Ungun.  G.  Chastar.  V.  Diyamandoglu,  J.I.  Daniala,  and  B.  Marina* 


tl.  SUPPLEMENTARY  NOTATION 


I  11*.  TYPE  OP  RIPORT  |1JD.  TIME  COVIRIO  114.  DATE  OP  REPORT  ( Y$tr,  Month,  My) 

FINAL  _ I  FROM  11-86  TO  5-90  I  May  1990 


ATI  COOES 
GROUP  I  5UI*GROUP 


24 

07 

06 

11 

II.  SUIJICT  TCRMS  (Contmua  on  ravarj*  i)  n«c****ry  and  NFamiFy 

Mobil*  Water  Purification  Unit;  Revnroa  Osmosis  Water 
Purification  Unit;  Pretraatmant;  Filtration;  Diainfection; 
Water-Quality  Analysis:  Reverse  Osmoaia  Bvpaaa _ 


19,  ABSTRACT  (Cohdnua  on  rtvam  ,p  nataiMry  and  Fdatwiy  by  Mock  number)  ""  "  r  "  truwi  . 

A  comparison  is  made  between  the  performances  of  the  600-gph  Ravarse  Osmosis  Water 
Purification  Unit  (ROWPU)  operated  in  the  bypass  mode  and  the  Mobile  Water  Purification  Unit 
(MWPU,  also  referred  to  as  an  ERDLAT0R).  Generally,  the  performance  of  the  MWPU  is 
significantly  better  than  the  pretreatment  units  of  the  ROWPU  in  terms  of  removing  both 
turbidity  and  pathogenic  organisms.  It  is  recommended  that  the  practice  of  bypassing  ths 
reverse  osmosis  (RO)  components  of  the  ROWPU  be  avoided  unless  it  can  J>a  dsmonstratad  claarly 
that  the  cartridge  filters  will  remove  the  cysts  of  infectious  organisms  effectively  and 
reliably.  If  the  ROWPU  must  be  operated  in  the  bypass  mode,  it  is  recommended  that  the 
dose  of  disinfectant  used  be  made  equal  to  that  currently  employed  in  the  field  for  untreated 
raw  water. 

The  analytical  methods  used  to  determine  total  dissolved  solids  (TDS)  and  residual 
free  chlorine  wilfh  the  new  Water-Quality  Monitor  (WQM)  are  also  reviewed  briefly.  The 
limitations  of  ths  methods  used  to  calibrate  the  TDS  and  f rae— chlorine  probes  of  ths  new 


20.  OIJ TRIIUTION  /  AVAIIAIILIT Y  OP  AISTRACT 
□  UNCLASSIRHOAJNUMITEO  gj  SAMI  AS  RPT 


ZZa.  NAMI  OP  RESPONSIBLE  INDIVIDUAL 
Mary  Frances  Bostian 


DO  Form  1473.  JUN  8« 


21  ABSTRACT  SECURITY  CLASSIFICATION 
□  otic  USERS  UNCLASSIFIED  _ 


22b.  TELEPHONE  Mud*  Area  Coda)  22c.  OFFICE  SYMIOL 

301-663-7325  SGRD-RMI-S 


Prav/oui  fdWOfll *r* OblO/aT#,  SECURITY  CLASSIFICATION  OR  THIS 

UNCLASSIFIED 


UNCLASSIFIED 


AD 

UCRL-21008Vol.8 


Evaluation  of  Military  Field-Water  Quality 

Volume  8.  Performance  of  Mobile  Water-Purification  Unit  (MWPU) 
and  Pretreatment  Components  of  the  600-GPH  Reverse  Osmosis  Water 
Purification  Unit  (ROWPU),  and  Consideration  of  Reverse  Osmosis  (RO) 
Bypass,  Potable-Water  Disinfection,  and  Water-Quality  Analysis  Techniques 


R.  E.  Selleck,*  Z.  Ungun,41  G.  Chester,*  V.  Diyamandoglu* 
J.  I.  Daniels,  and  B.  Marinas* 


Environmental  Sciences  Division 
Lawrence  Livermore  National  Laboratory 
University  of  California 
P.  O.  Box  5507 
Livermore,  CA  94550 


Ac<M*l*sil»a  For 

"NT  If?  UVUfcl 

me  ui< 

UuMnooojio  ert 

Just  if  I.0&1.  loa 


n 

!'l 


May  1990 


*f . . 

ms'.,  rtbul  luai/ 


Supported  by 

U.S.  Army  Medical  Research  and  Development  Command 
Fort  Detrick,  Frederick,  MD  21701 

Project  Order  82PP2817 


Av*LL'U>U  1 1,» 


th 


Dlat 


A»hM.  uno/nr 
Kp#o  1;\ I 


Project  Officer:  S.  A.  Schaub 


Principal  Investigators  at  Lawrence  Livermore  National  Laboratory: 
L.  R.  Anspaugh,  J.  I.  Daniels,  and  D.  W.  Layton 


Approved  for  public  release;  distribution  unlimited 

The  findings  of  this  report  are  not  to  be  construed  as  an  official  Department 
of  the  Army  position  unless  so  designated  by  other  authorized  documents. 


*  Sanitary  Engineering  and  Environmental  Health  Research  Laboratory  [Building  112], 
University  of  California,  Richmond  Field  Station,  Richmond,  CA  94804. 


FOREWORD 


This  report  is  the  eighth  volume  of  a  nine-volume  study  entitled  Evaluation  of 
Military  Field-Water  Quality.  Titles  of  the  other  volumes  are  as  follows:  Vol.  1,  Executive 
Summary:  Vol.  2,  Constituents  of  Military  Concern  from  Natural  and  Anthropogenic 
Sources:  Vol.  3,  Opportunity  Poisons:  Vol.  4,  Health  Criteria  and  Recommendations  for 
Standards:  Vol.  5,  Infectious  Organisms  of  Military  Concern  Associated  with  Consumption: 
Assessment  of  Health  Risks,  and  Recommendations  for  -Establishing_Relat£cL  Standards; 
Vol. 6,  Infectious  Organisms  of  Military..  Concern  Associated  with  Nonconsumptive 
Exposure;  Assessment  of  Health  Risks,  and  Recommendations  _fQr  _Establisliing.  Related 

Standards;  Vol.  7,  Performance.  Evaluation  of  the  600-GPH . Reverse  -Osmosis  Water 

Purification  Unit  (ROWPU);  Reverse  Osmosis  (RO)  Components;  and  Vol.  9,  Rata  ..for 
Assessing  Health  Risks  in  Potential  Theaters  of  Operation  for  U.S,  Military. Forces. 

The  nine  volumes  of  this  study  contain  a  comprehensive  assessment  of  the 
chemical/  radiological,  and  biological  constituents  of  field-water  supplies  that  could  pose 
health  risks  to  military  personnel  as  well  as  a  detailed  evaluation  of  the  field-water- 
treatment  capability  of  the  U.S.  Armed  Forces.  The  scientific  expertise  for  performing  the 
analyses  in  this  study  came  from  the  University  of  California  Lawrence  Livermore 
National  Laboratory  (LLNL)  in  Livermore,  CA;  the  University  of  California  campuses 
located  in  Berkeley  (UCB)  and  Davis  (UCD),  CA;  the  University  of  Illinois  campus  in 
Champaign-Urbana,  IL;  and  the  consulting  firms  of  IWG  Corporation  in  San  Diego,  CA, 
and  V.J,  Ciccone  &  Associates  (VJCA),  Inc.,  in  Woodbridge,  VA.  Additionally  a  Department 
of  Defense  (DoD)  Multiservice  Steering  Group  (MSG),  consisting  of  both  military  and 
civilian  representatives  from  the  Armed  Forces  of  the  United  States  (Army,  Navy,  Air 
Force,  and  Marines),  as  well  as  representatives  from  the  U.S.  Department  of  Defense,  and 
the  U.S.  Environmental  Protection  Agency  provided  guidance,  and  critical  reviews  to  the 
researchers.  The  reports  addressing  chemical,  radiological,  and  biological  constituents  of 
field-water  supplies  were  also  reviewed  by  scientists  at  Oak  Ridge  National  Laboratory  in 
Oak  Ridge,  TN,  at  the  request  of  the  U.S.  Army.  Furthermore,  personnel  at  several  research 
laboratories,  military  installations,  and  agencies  of  the  U.S.  Army  and  the  other  Armed 
Forces  provided  technical  assistance  and  information  to  the  research  on  topics  related  to 
field  water  and  the  U.S.  military  community. 


iii 


ACKNOWLEDGMENTS 


The  authors  extend  their  appreciation  to  the  scientists  and  staff  of  the 
Environmental  Sciences  Division  at  the  Lawrence  Livermore  National  Laboratory  (LLNL), 
as  well  as  to  Dr.  Stephen  A.  Schaub  and  his  colleagues  at  the  U.S.  Army  Biomedical 
Research  and  Development  Laboratory  (USABRDL),  for  their  cooperation  and  assistance.  A 
special  thank  you  is  extended  to  the  scientists  and  staff  of  the  Sanitary  Engineering  and 
Environmental  Health  Research  Laboratory  at  the  University  of  California,  Richmond 
Field  Station,  and  at  the  University  of  California,  Berkeley,  for  their  efforts  in  preparing  this 
document.  The  authors  also  express  their  gratitude  to  those  individuals  and  organizations 
in  the  reverse  osmosis  industry  who  provided  information  about  reverse  osmosis 
technology  that  was  intrinsic  to  our  research,  and  to  Dr.  G.C.  White  for  his  review  of  the 
material  related  to  disinfection  procedures  and  analytical  methods  for  the  assessment  of 
field-water  quality. 


i  v 


TABLE  OF  CONTENTS 


Foreword .  iii 

Acknowledgments. .  iv 

List  of  Tables .  v  i 

List  of  Figures . ix 

Preface .  x 

Abstract .  1 

Introduction . .. .  1 

Concepts  of  Deep-Bed  Filtration . 2 

Theory .  5 

Particle  Transport . 5 

Particle  Attachment . 12 

Coagulation  and  Coagulents .  14 

Jar  Tests .  15 

Initial  Mixing  and  Mean  Velocity  Gradients..... .  16 

Cartridge  Filters .  18 

i  Diatomaceous-Earth  (Precoat)  Filters .  19 

Description  of  the  600-gph  ROWPU  Pretreatment  System .  21 

Coagulation . 21 

Multimedia  Filter .  23 

ROWPU  Cartridge  Filters .  23 

Description  of  the  Mobile  Water  Purification  Unit .  24 

Background .  25 

Treatment  System .  27 

TheERDLATOR . 29 

Conditioning  Chemicals .  32 

Precoating-Filter  Operation .  33 

Performance  of  Deep-Bed  Filters  Operated  in  the  Direct-Filtration  Mode .  34 

Determination  of  Best  Type  and  Optimum  Dosage  of  Coagulent .  36 

General  Background  Papers .  42 

600-gph  ROWPU  Pretreatment  Units .  52 

Performance  of  Cartridge  Filters .  58 

Performance  of  Precoat  Filters  Operated  in  the  Direct-Filtrations  Mode .  59 

Performance  of  the  MWPU  (ERDLATOR) .  68 

Nuclear  Warfare  Agents .  73 


v 


Chemical  Warfare  Agent  GB .  74 

Disinfection . 74 

Heating .  76 

Iodine .  76 

Chlorination .  78 

Analytical  Methods . 81 

Comparison  of  WQAS  and  WQM  Equipment .  82 

Physical  Characteristics .  82 

Analytical  Capabilities . . .  84 

Summary,  Conclusions,  and  Recommendations . 89 

Direct  Filtration .  89 

Theoretical  Removal  of  Particles  in  the  600-gph  ROWPU  Multimedia  Filter .  89 

Observed  Removal  of  Microorganisms .  90 

Limits  of  Process  Applicability .  93 

Filter-Bed  Composition .  93 

Cartridge  Filters .  94 

Diatomaceous-Earth  (Precoat)  Filtration .  95 

The  ROWPU  Bypass .  96 

Coagulation  Control .  98 

Determination  of  Optimal  Coagulent  Dose .  99 

Initial  Mixing .  100 

Disinfection .  101 

Analytical  Methods . 101 

Suggested  Improvements .  102 

References . 103 


v  i 


Volume  8 


LIST  OF  TABLES 


1.  Effect  of  coagulant  dosage  and  type  on  the  removal  of  Giardia 

muris  cysts  and  coliform  bacteria  by  direct  filtration  with 
preliminary  flocculation . 

2.  Types  of  filter  media  used  in  the  direct-filtration  studies  of 

Virginia  surface  waters . 

3.  Details  of  the  successful  filter  runs  with  Cat-Floe  T1  coagulant  for 

raw-water  turbidities  exceeding  10  NTU . 

4.  Average  removal  of  total  coliform  bacteria  and  particles  in  the  7- 

to  12- urn  size  range  for  direct  dual-media  filtration  without 
chlorination . 

5.  Raw-water  characteristics  of  Monocacy  River . 

6.  Average  removal  of  turbidity  observed  for  a  filtration  system 

similar  to  that  in  the  600-gph  ROWPU  pretreatment  section.. . 

7.  Average  removal  of  microorganisms  observed  for  a  filtration 
system  similar  to  that  in  the  600-gph  ROWPU  pretreatment  section.  .  .  . 

8.  Average  removal  of  total  aerobic  bacteria  observed  for  a  filtration 

system  similar  to  that  used  in  the  600-gph  ROWPU  pretreatment 
section  for  various  feed-water  flows  .  . . 

9.  Average  removal  of  microorganisms  from  a  turbid  water  at  three 

pH  values  by  a  multimedia  filter  similar  to  that  used  in  the 
600-gph  ROWPU . 

1 0.  Observed  removal  of  total  bacteria  and  fecal  coliform  bacteria  by 

the  MWPU  without  chlorination  in  1952 . 

1 1.  Observed  concentration  of  the  spores  of  B.  subtilus  var.  niger  in 
the  ERDLATOR  floe  and  effluent,  and  the” filtered-water  effluent 

of  the'MWPU . . .  , 

i  2.  Average  daily  removal  of  turbidity  and  total  coliform  bacteria 
from  Potomac  River  water  in  1967  with  the  standard  3000-gph 
MWPU . 

13.  U.S.  Army  field-water-quality  standards  and  capabilities  of  Army 

water-quality-analysis  sets  . . 

14.  Specifications  for  WQM  parameters . 


35 

49 

51 

53 

55 

56 

56 


57 


59 

70 


71 


72 

83 

86 


VM 


Volume  8 


15.  Summary  of  recommended  maximum  limits  on  raw-water-quality 

characteristics  for  direct  filtration  (gravity  filters) .  94 


600-gph  ROWPU  with  that  of  the  3000-gph  MWPU  when  operated 

in  accord  with  Army  operation  manuals— no  prechlorination  . .  97 


viii 


Volume  8 


LIST  OF  FIGURES 

1 .  Classification  of  modes  of  filter  operation .  4 

2.  The  single-collector  theory  according  to  Yao  e£ai. .  6 

3.  Comparison  of  numerical  and  analytical  solutions  of  Eq.  7 .  9 

4.  Differential  volume  element  in  the  filter  bed .  9 

5.  Comparison  of  observed  and  theoretical  removal  in  a  deep-bed  filter.  ...  11 

6.  The  effect  of  the  mean  velocity  gradient,  G,  on  the  aggregation 

rate  of  suspended  particles  for  rapid-mix  devices  of  various 
geometries .  17 

7.  Water-processing  block  diagram  for  the  600-gph  ROWPU .  22 

8.  ROWPU  multimedia  filter . 24 

9.  The  backwash  water  system  for  the  600-gph  ROWPU  multimedia 

filter .  25 

10.  Diagram  of  the  600-gph  ROWPU  cartridge  filter .  26 

11.  The  1500-gph  MWPU  and  supporting  equipment  in  operating  position.  ...  28 

1  12.  Cutaway  view  of  Mobile  Water  Purification  Unit .  28 

13.  Cross-sectional  flow  diagram  of  ERDLATOR  (original  testing  version).  .  .  30 

14.  Cross  section  of  the  1500-gph  MWPU  (ERDLATOR)  assembly .  31 

15.  Cross  section  of  diatomite  filter,  final  testing  device  for  use  in  the 

Mobile  Water  Purification  Unit .  33 

16.  Comparison  of  jar-test  results  with  the  passage  of  kaolinlte 

turbidity  through  rapid  sand  filters .  37 

17.  Effect  of  polyelectrolyte  dose  on  the  passage  of  0.1- ym  latex 

beads  through  a  deep-bed  filter .  37 

18.  Removal  of  MS2  bacteriophage  by  uncoated  and  polyelectrolyte 

(PE)  coated  diatomaceous-earth  (DE)  filter  aid .  65 

19.  Theoretical  removal  of  destabilized  particles  in  the  600-gph 

ROWPU  multimedia  filter .  . .  91 


i  x 


PREFACE 


This  work  is  Volume  8  of  the  series,  Evaluation  of  Military  Field-Water  Quality,  and 
it  is  concerned  primarily  with  the  performance  of  the  mobile  water-purification  unit 
(MWPU)  and  the  pretreatment  components  of  the  600-GPH  reverse  osmosis  water 
purification  unit  (ROWPU)  available  in  1986.  A  performance  evaluation  of  the  reverse 
osmosis  (RO)  components  of  the  ROWPU  appears  in  Volume  7,  the  companion  volume  to 
Volume  8.  Also  considered  in  this  volume  are  (1)  the  efficacy  of  the  operation  of  the 
ROWPU  in  a  mode  that  bypasses  the  RO  components,  (2)  the  potable-water  disinfection 
processes  available  to  the  U.S.  Armed  Forces,  and  (3)  the  water-quality  analysis  techniques 
applicable  to  field  water.  Together,  Volumes  7  and  8  represent  a  complete  assessment  of  the 
treatment,  disinfection,  and  water  quality  analysis  procedures  now  used  by  U.S.  military 
forces. 


Volume  8 


ABSTRACT 


A  comparison  is  made  between  the  performances  of  the  600-gph  Reverse  Osmosis 
Water  Purification  Unit  'ROWPU)  operated  i  i  the  bypass  mode  and  the  Mobile  Water 
Purification  Unit  (MWPU,  frequently  referred  to  as  an  ERDLATOR  because  the  equipment 
was  developed  at  the  Engineer  Research  and  Development  Laboratory  at  Fort  Belvoir, 
VA).  Generally,  the  performance  of  the  MWPU  is  significantly  better  than  t 
pretreatment  units  of  the  ROWPU  in  terms  of  removing  both  turbidity  and  pathogenic 
organisms.  It  is  recommended  that  the  practice  of  bypassing  the  reverse  osmosis  (RO) 
components  of  the  ROWPU  be  avoided  unless  it  can  be  demonstrated  clearly  that  the 
cartridge  filters  will  remove  the  cysts  of  Infectious  organisms  effectively  and  reliably.  If 
the  ROWPU  must  be  operated  in  the  bypass  mode,  it  is  recommended  that  the  dose  of 
disinfectant  used  be  made  equal  to  that  currently  employed  in  the  field  for  untreated  raw 
water. 

The  analytical  methods  used  to  determine  total  dissolved  solids  (TDS)  and  residual 
free  chlorine  with  the  new  Water-Quality  Monitor  (WQM)  are  also  reviewed  briefly.  The 
limitations  of  the  methods  used  to  calibrate  the  TDS  and  free-chlorine  probes  of  the  new 
WQM  are  discussed. 


INTRODUCTION 

The  U.S.  Army  is  considering  the  use  of  the  600-gph  Reverse  Osmosis  Water 

Purification  Unit  (ROWPU)  in  an  alternate  configuration,  here  called  the  "bypass  mode". 

Only  the  multimedia  and  cartridge  filters  would  be  utilized,  followed  by  chemical 

disinfection.  The  advantage  of  this  is  that  the  loss  of  between  1/2  and  2/3  of  the  filtered 

water  as  RO  reject  water  would  be  avoided.  A  second  benefit  of  this  arrangement  would 

be  cutting  the  power  requirements  in  half,  because  roughly  50%  of  the  power  supplied  to 

the  ROWPU  is  used  in  the  reverse  osmosis  (RO)  process.^ 

The  treatment  system  in  the  bypass  mode  would  include  the  following  components: 

(1)  a  garnet-sand-anthracite  deep-bed  multimedia  pressure  filter,  (2)  a  5-pm  (nominal 

rating)  cartridge  filtration  unit,  and  (3)  chlorination  with  a  chlorine  dose  equivalent  to  5 

to  1C  mg/L  of  free-chlorine  residual,  depending  on  the  pH,  temperature,  and  other 

feed-water-quality  parameters.  Three  criteria  were  suggested  by  the  U.S.  Army  planners 

2 

to  be  considered  in  the  treatment  of  raw  waters  in  the  bypass  mode.  These  criteria  were 
as  follows: 


l 


Volume  8 


•  The  water  must  be  free  of  "acute  toxicity  from  Industrial,  agricultural, 
domestic,  or  natural  resource  contributions." 

•  The  water  must  satisfy  potability  requirements  regarding  NBC  (nuclear, 
biological,  and  chemical)  warfare  agents. 

•  The  water  must  meet  palatability  requirements. 

Modifications  to  the  treatment  train  have  been  considered  by  the  U.S.  Army 
planners.  Two  of  the  alternatives  under  consideration  are  (1)  reducing  the  pore  size  of 
the  cartridge  filters  from  a  nominal  rating  (defined  by  a  military  test)  of  5  pm  to  an 
absolute  rating  (the  diameter  of  the  largest  spherical  particle  passing  through  a  filter)  of 
3  pm,  and  (2)  using  alternative  disinfectants  that  might  have  viricidal  and  cysticidal 
properties  superior  to  those  of  chlorine.  A  major  concern  expressed  by  the  U.S.  Army 
planners  was  the  health  risk  inherent  in  the  failure  to  remove  pathogenic  viruses  and  cysts 
when  bypassing  the  RO  section  of  the  600-gph  ROWPU. 

The  objectives  of  this  report  are  to  review  and  summarize  the  state  of  knowledge 
concerning  primarily  the  removal  of  microorganisms  achieved  with  multimedia  deep-bed 
filters  operated  in  the  direct-filtration  mode  equivalent  to  that  proposed  for  the  600-gph 
ROWPU  bypass,  and  to  compare  the  results  with  those  that  can  be  expected  for  the 
current  3000-gph  Mobile  Water  Purification  Unit  (MWPU).  In  this  way  the  U.S.  Army  can 
judge  better  the  risks  associated  with  bypassing  the  RO  section  of  the  ROWPU  and  can 
compare  them  with  the  risks  associated  with  the  use  of  the  MWPU. 

CONCEPTS  OF  DEEP-BED  FILTRATION 

The  pretreatment  portion  of  the  current  600-gph  ROWPU  contains  a  deep-bed 
multimedia,  filter.  Deep-bed  filtration  or  depth  filtration. may  be  defined  as  a  process  in 
which  a  fluid  suspension  is  passed  through  a  filter  composed  of  granular  or  fibrous 
materials.  The  suspended  solids  are  deposited  on  the  surfaces  of  the  grains  or  fibers  along 
the  entire  filter  depth  as  the  water  flows  through  the  filter  media. 

A  second  general  type  of  filtration  is  surface  fiitration.  In  this  case  the  solids  are 
strained  from  the  water  at  the  surface  of  the  filter  medium.  Straining  occurs  if  the  sizes 

4 

of  the  particles  in  suspension  are  greater  than  about  0.2  times  the  filter  grain  size. 

Precoat  filters  of  the  type  used  in  the  MWPU  are  commonly  considered  to  be  surface 

filters.  A  filter  may  act  both  as  a  surface  and  deep-bed  filter,  depending  primarily  on  the 

physical  and  chemical  characteristics  of  the  suspension  and  the  filter  media,  as  well  as 

•j 

the  rate  of  filtration. 


2 


Volume  8 


In  water  treatment,  the  process  of  destroying  the  stabilizing  forces  that  keep  the 

3 

colloidal  particles  apart  is  referred  to  as  "chemical  coagulation."  According  to  Weber, 
coagulation  is  the  overall  process  of  colloidal  particle  aggregation,  Including  both 
destabilization  and  transport,  whereas  flocculation  is  just  the  particle  transport  that 
occurs  after  destabilization.  Following  the  mixing  of  the  coagulating  chemicals  with  the 
colloidal  suspension,  several  complex  reactions  take  place.  These  reactions  are  fast, 
taking  place  in  less  than  one  second.  At  this  point  in  the  process,  the  colloids  are 
destabilized,  and  any  particle  formed  is  very  small. 

The  rate  of  agglomeration  at  which  the  destabilized  particles  form  visible  floes  is 
dependent  on  the  number  of  opportunities  for  particle-particle  contact.  When  the  water 
containing  the  destabilized  particles  is  passed  through  a  granular  deep-bed  filter  column, 
flocculation  takes  place  at  a  greatly  accelerated  rate  because  of  the  large  number  of 
opportunities  for  contact  between  the  destabilized  particles  alforded  by  the  passage  of 
the  water  in  the  tortuous  flow  pathways.  The  floe  particles  then  become  attached  to  or 
adsorbed  by  the  surface  of  the  filter  grains.  As  the  filtration  run  progresses,  the  pores  in 
the  upper  section  of  the  filter  medium  gradually  fill  with  floe,  and  the  particles  move 
down  into  the  filter  to  find  an  available  surface  for  attachment.  Finally,  the  pressure 
drop  across  the  filter  bed  increases  to  an  undesirable  level  or  breakthrough  occurs. 
'Consequently,  filter  backwashing  is  required. 

The  terms  direct  and  conventional  water  filtration  apply  to  the  mode  of  filter 
operation.  In  direct  filtration,  all  solids  (those  occurring  naturally  in  the  raw  water  and 
those  added  during  the  treatment)  must  be  removed  and  stored  in  the  filter  bed.5  On  the 
other  hand,  in  conventional  water  filtration  some  of  the  solids  are  removed  in  a 
sedimentation  basin  located  upstream  from  the  filter.  According  to  the 
Coagulation-Filtration  Committee  of  the  American  Water  Works  Association  (AWWA), 
direct  filtration  refers  to  a  treatment  system  In  which  filtration  is  not  preceded  by 
sedimentation,  and  consequently  those  systems  that  do  not  use  any  chemicals  for 
pretreatment  are  not  considered  direct-filtration  systems.6  The  main  steps  in  the 
direct-filtration  process  are  the  following;  (1)  the  addition  of  the  destabilizing  chemicals, 
(2)  complete  or  partial  flocculation  with  no  settling,  and  (3)  filtration.  The  flocculation 
step  may  be  eliminated  depending  on  the  raw-water  quality.5  The  flow  schemes  shown  in 
Fig.  1  illustrate  the  differences  between  conventional  and  direct  filtration,  as  well  as  the 
differences  between  the  various  versions  of  direct  filtration. 


3 


Volume  8 


Typical  Conventional  Filter  Planti 


A. 

Older 

planti 


B. 

Recent 

dailgn 

trendi 


Direct  Filtration 


C. 

In-line 

tiltrationi 


D. 

Direct 

filtration 


E. 

In-line 

with 

contact 

hniln 


Nonlonlc  polymer 
0.08  to  0.5  mg/L 
or  activated  silica 


Nonlonlc  polymer 
0.05  to  0.5  mg/L 
or  activated  silica 


Figure  I.  Classification  of  modes  of  filter  operation.  Adapted  from  Culp. 


U 


Volume  8 


THEORY 

7 

According  to  O'Melia  and  Stumm,  filtration  is  similar  to  coagulation  in  the  sense 
that  removal  within  a  deep-bed  filter  column  involves  at  least  two  separate  steps: 
(l)  particle  transport  and  (2)  particle  attachment.  Particle  transport,  a  physical-hydraulic 
process,  is  principally  affected  by  parameters  governing  the  mass  transfer.  On  the  other 
hand,  particle  attachment,  basically  a  chemical  process,  is  influenced  by  both  chemical 
and  physical  parameters. 

Particle  Transport 

7  8 

O'Melia  and  Stumm  and  Yao  et  al.  studied  deep-bed  filtration  from  a  microscopic 
point  of  view.  According  to  those  authors,  the  suspended  solids  are  removed  by  the 
following  two-step  process:  (1)  the  mass  transport  of  the  suspended  solids  to  the 
immediate  vicinity  of  the  solid-liquid  interface  of  the  filter  media,  and  (2)  the 
attachment  of  particles  to  the  media.  A  simplified  mass-transport  model  was  postulated 
in  which  a  single  grain  of  the  filter  media  (henceforth  called  the  collector)  is  separated 
from  the  surrounding  grains  and  fixed  in  space.  The  water  flows  in  the  direction  of  the 
gravitational  force  in  a  laminar  regime.  Stokes'  equations  were  used  to  compute 
laminar-flow  velocities  around  the  spherical  collector.  Coagulation  theory  was  then  used 
to  compute  the  number  of  contacts  that  would  occur  per  unit  time  between  a  collector  of 
diameter  d^  and  a  uniform  suspension  (i.e.,  ail  particles  having  the  same  size)  of  spherical 
particles  of  diameter  d^,  as  shown  in  Fig.  2. 

The  transport  mechanisms  by  which  suspended  particles  *  are  collected  are 
interception,  sedimentation,  and  diffusion.  According  to  Fig.  2,  a  suspended  particle  will 
touch  or  intercept  the  collector  when  the  particle  is  contained  in  a  flowing-fluid 
streamline  that  passes  within  a  distance  of  dp  /2  of  the  collector  (point  A  in  Fig.  2). 
Inertial  and  gravitational  forces  may  also  cause  the  particle  to  deviate  from  a  streamline 
(i.e.,  sedimentation)  and  collide  with  the  collector  (path  B  in  Fig.  2).  The  third 
mechanism,  Brownian  motion  or  diffusion,  may  also  cause  the  particles  to  cut  across  the 
streamlines  and  collide  with  the  collector  (path  C  in  Fig.  2).  In  water  suspension  the 

inertial  effect  on  particle  collection  is  very  small  and  has  been  ignored  by  O'Melia  and 

7  8 

Stumm  and  Yaoet  al.  in  their  works.  Other  possible  forces  such  as  electrostatic  or  van 
der  Waals  forces  have  also  been  ignored. 

A  materials  balance  may  be  written  about  a  differential  volume  to  obtain  the 

temporal  and  spatial  variation  in  n,  the  number  of  suspended  particles  per  unit  volume  of 

8 

water: 


5 


Volume  8 


Figure  2.  The  single-collector  theory  according  to  Yao  et  aL  A  a  Interception, 
B  a  sedimentation,  C  »  diffusion,  U  a  velocity  (well  removed  from  collector)  at  which 
particles  approach  the  collector.  Adapted  from  Yao  et  al.8 


4  n  _  -  _  „  2 

r-+  v  •  Vn  =  D  V  n  + 
o  t  p 


J2H  E _ 

3tr  •  y  •  d 


3  n 

lz' 


(1) 


in  which 

2  - 1 

D  s  the  particle  Brownian-motion  diffusion  coefficient  (I  T  ) 

—  ^  _  I 

v  -  streamline-flow  velocity  (a  vector)  (LT  ) 

-2 

g  =  gravitational  acceleration  (LT  ) 

m  =  mass  of  a  suspended  particle  (M) 

-3 

p  and  p  =  densities  of  the  water  and  the  suspended  particles  respectively  (ML  ) 
M  =  water  dynamic  viscosity  (ML  T  ) 

z  =  coordinate  in  the  direction  of  gravity  (L) 

t  =  time  (T) 

dp  =  diameter  of  spherical  particle  (L) 

Vn  =  gradient  in  the  number  of  particles,  general  (vector  notation) 

L,M,T  =  units  of  length,  mass,  time 


Volume  8 


The  first  term  on  the  left-hand  side  of  the  equation  gives  the  temporal  variation 

of  n  at  any  point  (x,y,z).  The  second  term,  (v*  Vn)  describes  the  effect  of  advection  on 
the  concentration  at  that  point.  The  first  term  (DpV  n)  on  the  right-hand  side  of  the 
equation  describes  the  effect  of  diffusion,  and  the  second  term  the  effect  of  gravitational 
settling  if  the  particles  settle  in  accord  with  Stokes'  law.  The  effect  of  interception  on 
particle  collection  is  included  in  the  boundary  conditions  used  to  integrate  Eq.  1 . 

Equation  1  has  been  solved  analytically  by  considering  the  following  assumptions: 
<1 )  It  =  0  <steady  state);  (2)  all  the  suspended  particles  that  touch  the  collector 
disappear  instantaneously  from  the  system  without  increasing  the  size  of  the  collector 
(the  collector  is  always  clean);  and  (3)  only  one  removal  mechanism  is  predominant  at  a 
specific  suspended  particle  size  d  .  The  results  obtained  were  expressed  in  terms  of  a 

r 

single-collector  efficiency  n,  which  is  defined  as 


n  =  rg/UnAg  , 


(2) 


in  which 
r 

g 


n 


=  rate  at  which  the  suspended  particles  touch  or  strike  the  collector  (number  of 
particles*  T"^); 

=  velocity  at  which  the  particles  approach  the  collector,  well  removed  from  the 
collector  (LT" 1 ); 

=  particle  concentration  upstream  from  the  collector  (number  of 


particles  •  L"^);  and 

2  2 

=  cross-sectional  area  of  the  collector  =  trdg/4  (L  ). 


The  analytical  solutions  based  on  the  assumptions  described  previously  are: 


Interception  only: 
Sedimentation  only: 


n,  =  3  «p/dg)2/2 

(Pp  -  P)  g  < 
nC=  18  yU 


Diffusion  only:  =  4(Dp/Udg)^  , 


(3) 

(4) 

(5) 


7 


Volume  8 


in  which 

D  =  kT  /3  IT  yd  (6) 

P  a'  H  p 

and 

k  =  Boltzmann  constant  =  1.39  x  10"*^erg/K 

Tfl  =  absolute  temperature  (K). 

Equations  3,  4,  and  5  are  plotted  In  Fig.  3  for  the  special  case  of 

2  3 

U  =  2  gal/(mln  •  ft  ),  d  =  0.5  mm,  p  a  1,05  g/cm  ,  at  25#C.  A  numerical  solution  of 

O  r 

Eq,  l  (curved  line  shown  in  Fig.  3)  demonstrates  that  the  overall  collector  efficiency  can 
be  obtained  without  significant  error  simply  by  adding  Eqs.  3,  4,  and  5; 

n=VnG+nD  *  '  (7) 

The  results  shown  in  Fig.  3  demonstrate  that 

•  There  exists  a  size  of  the  suspended  particles  for  which  the  removal  efficiency 
is  a  minimum.  For  conditions  typical  of  conventional  water-filtration  practice, 
this  size  is  about  l  pm  and  particles  of  greater  or  lesser  size  than  1  pm  are 
removed  more  effectively. 

•  For  example,  for  particles  smaller  than  l  pm,  the  collector  efficiency 
increases  with  decreasing  particle  size,  and  particle  removal  is  accomplished 
by  diffusion. 

•  For  particles  greater  than  l  pm,  the  collector  efficiency  Increases  rapidly  with 
Increasing  particle  size,  and  removal  is  accomplished  by  interception  and/or 
sedimentation. 

Conceptually,  a  single  isolated  collector  is  very  different  from  a  packed  bed  of 
collectors.  With  the  collectors  in  contact  with  each  other,  the  flow  streamlines  are  not  as 
pictured  in  Fig.  2.  Even  so,  the  single-collector  theory  has  been  applied  directly  to 
deep-bed  filtration  with  reasonably  good  results,  as  will  be  demonstrated  subsequently. 

The  number  of  single  collectors  contained  in  a  differential  volume  element  (i.e., 
having  dimensions  so  tiny  that  particle  behavior  can  be  described  using  differential 
equations)  of  bed  of  thickness  6z  (see  Fig,  4)  is: 


8 


Volume  8 


Figure  3.  Comparison  of  numerical  and  analytical  solutions  of  Eq.  7. 

(U  ■  2  gal/(min  •  ft2),  d„  *  0.3  mm,  p_  ■  1.03  g/cm3  and  T  ■  23‘C).  Adapted  from  Yao 

,8  8  p 

et  al. 


Figure  4.  Differential  volume  element  in  the  filter  bed.  (Q  =  flow  rate, 
A  =  cross-sectional  surface  area). 


9 


Volume  8 


Number  of  collectors  =  (1  -  e)  A6z/V^,  in  which  e  is  the  bed  porosity,  A  is  the  total 
bed  area,  and  is  the  volume  of  a  single  collector.  The  rate  at  which  all  of  the 
collectors  contained  In  the  filter  bed  are  removing  particles  from  the  water  is  rR  (Eq.  2) 
times  the  number  of  collectors.  The  rate  of  removal  per  unit  volume  of  all  of  the 
collectors,  r  ,  is 

S 

rg  s  rR  x  number  of  collectors/volume  of  all  collectors 
=  3nUn/2d  . 

O 


Also,  the  rate  at  which  the  particles  are  removed  from  the  water  per  unit  volume  of  the 
water,  rL,  Is 


(8) 


The  change  In  the  particle-number  concentration  with  depth  at  any  time  may  be 
obtained  from  a  material  balance  written  about  the  differential  element  of  bed  volume  of 
thickness  6z  shown  in  Fig.  4,  or 


(cA6z)‘  rL  *  Q6n 
or 


Urn  rL  =  (U/e)(dn/dz) ,  (9) 

6z+  dz 


where  Q  =  U  *A.  Substituting  Eq.  8  into  Eq.  9  gives 


(10) 


in  which  X  is  the  filter  coefficient. 

Q 

Yao  et  al.  checked  the  validity  of  Eq.  7  with  Eq.  10.  The  results  of  this  comparison 
are  shown  In  Fig.  5.  The  observed  and  theoretical  minimum  efficiencies  were  both 
located  at  a  particle  size  of  approximately  I  ym,  but  the  observed  collector  efficiencies 
were  In  general  much  greater  than  the  theoretical.  This  means  that  the  theory  can  be 
used  to  obtain  a  quantitative  estimate  of  the  particle  size  most  likely  ;«.»  pass  through  a 
deep-bed  filter,  but  only  a  qualitative  estimate  of  the  proportion  removed. 


10 


Volume  8 


Figure  3.  Comparison  of  observed  and  theoretical  removals  In  a  deep-bed  filter, 
(dg  a  0.397  mm,  e  a  0.36,  U>  2  gal/(mln  •  ft2 ),  T  *  23*C,  z  a  3.3  in.,  pp  a  1.03  g/cm3, 
clean  filter).  Adapted  from  Yao  et  ah8 


It  was  assumed  In  the  previous  discussions  that  all  particles  that  strike  or  touch  a 
collector  will  stick  to  the  collector.  This  would  be  the  case  if  all  the  suspended  particles 
were  completely  destabilized.  In  general  this  is  not  the  case  and  a  parameter  called  the 
collision  efficiency  factor,  o,  is  sometimes  used  to  correct  for  a  system  of  partially 
destabilized  particles,  or 


(11) 


The  collision  efficiency  factor  represents  the  ratio  of  the  particles  that  stick  to  the 
collector  to  the  total  number  of  particles  that  strike  the  collector. 

As  stated  in  the  assumptions,  the  single-collector  theory  was  developed  for  the  case 
of  absolutely  clean  collectors.  Filtration,  however,  Is  an  Intrinsically  transient  process 
because  the  particles  deposited  on  the  collectors  change  the  geometry  of  the  Interstitial 


11 


Volume  8 


4 

spaces  in  the  filter  bed,  as  well  as  the  nature  of  the  collector  surfaces.  The  deposited 
particles  first  act  as  additional  collectors,  but  they  eventually  restrict  the  flow  through 
the  filter  bed.  Typically,  an  initial  increase  in  the  filter  efficiency  is  observed,  followed 
by  a  monotonlc  decrease. 

The  change  in  filter  efficiency  with  the  deposition  of  particles  in  the  filter  bed  was 
investigated  by  Ives,  among  others.  The  following  semiempirical  expression  was  proposed 
for  the  filter  coefficient,  X  (Eq.  1 0): 

X  =  XQ[  1  +  (  8po/e)]y  •  [1  -  (o/e)]2  •  [1  -  (o/ou)]x  ,  (1 2) 

in  which 

X^  =  filter  coefficient  for  a  ciean  bed 

0  e 
3p  =  packing  constant  =  ■pj 

e  a  bed  porosity  for  a  clean  filter 

a  a  volume  of  particles  deposited  per  unit  bed  volume 

ou  a  ultimate  value  of  a  when  the  filter  bed  becomes  ineffective  at  a  specified 
depth  (X+  0  as  a+cu) 
x,y,z  a  empirical  constants. 

In  Eq.  12  the  first  term,  [1  +  (0pa/e)]y,  gives  the  effect  of  the  Increase  in  specific  filter 
surface  (i.e.,  collector  surface  area  per  unit  filter  volume)  available  for  deposition  on  the 
filter  coefficient,  which  results  in  an  initial  improvement  period.  The  second  term, 
[1  -  (o/e)]2,  accounts  for  the  decrease  in  the  specific  filter  surface  when  the  void  spaces 
start  to  be  filled.  The  third  term,  [1  -  ( o/ou)]x,  gives  the  increase  in  interstitial-water 
flow  velocity  caused  by  the  restriction  in  the  flow  passageways.  Thus,  the  performance  of 
a  deep-bed  filter  will  at  first  increase  (the  first  term  in  Eq.  12  is  predominant)  and  then 
deteriorate  (second  and  third  terms  are  predominant)  with  time.  The  durations  of  the 
initial  Improvement  and  the  subsequent  deterioration  periods  depend  upon  the  relative 
magnitudes  of  the  exponents  x,  y,  and  z.  The  pressure  drop  across  the  filter  bed  will 
always  Increase  with  time. 

Particle  Attachment 

The  transport  model  described  in  the  previous  section,  albeit  elegant,  applies  to  the 
case  of  completely  or  nearly  completely  destabilized  particles.  It  has  little  significance  if 
the  particles  are  not  destabilized. 

i  2 


Volume  8 


Whether  or  not  a  particle  will  stick  to  a  collector  is  controlled  by  the  surface 

7 

properties  of  both  the  particle  and  the  collector.  According  to  O'Melia  and  Stumm, 
there  are  two  significant  models  that  have  been  used  to  describe  the  interactions  between 
suspended  particles  and  the  filter  media.  The  first  model  is  based  on  the  electrical 
double-layer  theory.  According  to  this  theory,  the  solid  side  of  a  solid-liquid  interface 
assumes  an  electrical  charge,  called  the  primary  charge,  which  depends  on  the  surface 
chemistry  of  the  solid.  An  equivalent  number  of  charges  of  opposite  sign  (secondary 
charges)  form  in  the  aqueous  phase  to  counterbalance  the  primary  charge.  The  counter 
charges  are  provided  by  the  ions  dissolved  in  the  water,  and  such  ions  are  called 
cn  filter  ions.  Because  most  clayey  colloids,  bacteria,  cysts  of  protozoa,  viruses,  etc., 
carry  negative  primary  charges,  the  counterions  are  usually  cations.  Some  of  the 
counterions  are  closely  associated  with  the  charged  particles,  and  move  with  the  particles 
when  an  electrical  potential  is  applied  to  the  colloidal  solution.  The  net  charge  on  the 
particle,  Including  that  of  the  closely  associated  counterions,  determines  a  parameter 
called  the  zeta  potential. 

Theory  demonstrates  that  if  a  particle  with  its  associated  zeta  potential  approaches 
another  particle  (or  filter  collector)  with  a  zeta  potential  of  the  same  sign,  then  an 
electrostatic  repulsive  force  occurs  between  the  particles  that  prevents  a  closer 
approach.  If  a  strong  electrolyte  Is  added  to  the  system,  and  especially  if  the  electrolyte 
contains  cations  of  high  charge  density,  then  the  zeta  potential  will  be  depressed  because 
the  cations  will  be  adsorbed  on  or  near  the  solid  surfaces.  The  particles  may  then  collide 
when  carrying  sufficient  kinetic  energy  to  overcome  the  residual  electrostatic  repulsive 
forces.  They  will  also  stick  together  if  the  short-range  van  der  Waals'  attractive  forces 
are  greater  than  the  repulsive  forces  at  this  proximity. 

O'Melia  and  Stumm  state  that  particle  attachment  brought  about  by  charged 
synthetic  or  natural  polymers  (or  polyelectrolytes),  which  have  been  successfully  used  as 
filter  conditioners  or  coagulating  agents  in  water  treatment,  cannot  be  characterized  by 
the  double-layer  model.  The  model  neglects  the  importance  of  chemical  forces,  such  as 
coulombic  attraction,  when  the  suspended  particle  and  the  filter  collector  are  of  opposite 
charge,  and  it  Is  only  valid  for  lyophobic  surfaces  and  simple  electrolytes/  The  second, 
multifaceted  theory,  called  the  bridging  theory,  has  been  developed  In  recent  years  to 
explain  Interactions  such  as  ion  exchange,  hydrogen  bonding,  and  the  formation  of 
coordinative  bonds  and  linkages  that  can  outweigh  electrostatic  forces  when  polymer 
coagulation  is  used. 


13 


Volume  8 


COAGULATION  AND  COAGULANTS 


As  shown  in  Fig.  1,  a  coagulant  (or  coagulants)  is  always  added  ahead  of  a  deep-bed 
filter,  regardless  of  the  mode  of  operation.  The  coagulants  most  commonly  used  in  the 
United  States  are  alum  [Al^SO^)^  •  UH^O],  hydrated  ferric  chloride  (FeCI^  •  61^0), 
and  various  nonionic  or  cationic  synthetic  polymers.  A  polymer  is  called  a  polyelectrolyte 
if  the  monomers  from  which  it  is  synthesized  contain  ionizable  groups.  Cationic  polymers 
gain  a  positive  charge  upon  ionization,  and  anionic  polymers  gain  a  negative  charge. 
Polymers  that  do  not  contain  any  ionizable  groups  are  called  nonionic.  Research  has 
revealed  that  both  cationic  and  anionic  polymers  are  capable  of  destabilizing  negatively 
charged  colloidal  particles  such  as  those  found  in  natural  waters,  but  anionic  polymers 
usually  have  to  be  used  in  conjunction  with  another  electrolyte  such  as  NaCl  or  CaClj,  or 
with  a  coagulant  such  as  alum. 

Early  Investigations  demonstrated  that  the  double-layer  theory  alone  did  not 
adequately  explain  the  observed  mechanisms  of  coagulation  with  polymers.  The 

3 

chemical-bridging  theory  was  then  developed  to  explain  the  observed  behavior. 
According  to  this  theory,  a  molecule  of  an  organic  polymer  or  an  inorganic  polymer 
resulting  from  the  hydrolysis  of  alum  and  ferric  chloride  will  become  attached  to  a 
colloidal  particle  at  one  or  more  sites.  The  number  of  sites  occupied  by  a  polymer 
molecule  depends  on  the  nature  of  the  various  electrostatic  or  chemical  forces,  and  the 

3 

charge  and  structure  of  the  polymer  and  the  colloid. 

A  nonionic  polymer,  or  even  a  polyelectrolyte  having  the  same  polarity  as  that  of 
the  colloid,  may  be  adsorbed  by  the  surface  of  the  colloid.  Some  segments  of  the  polymer 
molecule  will  be  adsorbed  by  the  surface,  and  some  will  extend  into  the  solution  phase. 
According  to  the  bridging  theory,  those  segments  that  extend  into  the  solution,  called 
loops  and  tails,  form  the  bridges  between  the  colloidal  particles.  There  must  be  a 
minimum  polymer  size  (i.e.,  minimum  molecular  weight)  required  to  bridge  the 
potential-energy  barrier  existing  between  two  negatively  charged  colloidal  particles.7 

Cationic  polymers  can  destabilize  the  negatively  charged  colloids  commonly 
encountered  in  natural  waters  and  wastewaters  by  both  bridge  formation  and  charge 
neutralization.  As  a  result,  cationic  polymers  do  not  require  a  minimum  molecular  weight 
to  be  effective  since  they  can  be  adsorbed  specifically  by  negative  colloids  and  have  their 
primary  charge  neutralized.  Ghosh  et  £l.  state  that  studies  of  polymer  configuration  on 
colloidal  surfaces  by  electron  paramagnetic  resonance  have  shown  that  cationic 
polyelectrolyte  segments  are  often  adsorbed  almost  completely  by  the  particle  surface 
without  many  loops  or  tails  extending  into  the  solution.  This  requires  that  a  mechanism, 


14 


Volume  8 


in  addition  to  bridging,  be  operative,  because  the  bridging  mechanism  depends  on  the 
presence  of  the  extending  segments.  Those  investigators  concluded  that  charge 
neutralization  plays  an  important  role  in  cationic-polymer  coagulation. 

3AR  TESTS 

It  appears  from  the  foregoing  that  the  destabilization  of  colloids  with  the  cationic 
polyelectrolytes  is  achieved  primarily  by  charge  neutralization  and  Interparticle  bridging. 
Both  of  these  mechanisms  imply  a  stoichiometric  reaction  between  the  colloid  and  the 
coagulant,  and  both  mechanisms  may  cause  particle  restablllzation  with  overdosing.^  The 
theory  of  coagulation  has  not  yet  been  sufficiently  developed  to  determine  the  optimum 
coagulant  dose  or  coagulant  type  that  should  be  used  in  direct  filtration;  therefore,  these 
determinations  must  be  made  experimentally. 

The  optimum  coagulant  dosage  is  most  commonly  determined  by  jar  tests.  The  jar 
test  is  a  laboratory  method  used  to  determine  the  effectiveness  of  various  coagulants, 
optimum  coagulant  and  coagulant-aid  dosages,  optimum  pH  for  coagulation,  and  the  most 
effective  order  in  which  the  various  chemical  coagulants  should  be  added  to  the 
suspension.  The  typical  jar-test  apparatus  consists  of  six  agitator  paddles  mechanically 
1  coupled  to  operate  at  the  same  speed,  which  can  be  varied  from  10  to  100  rpm.  The 
coagulation  containers  commonly  are  l-  or  2-L  beakers.  The  general  jar-test  procedure 
consists  of  flash-mixing  the  chemicals  by  agitating  at  100  rpm  for  1  min,  followed  by 
flocculation  at  a  stirring  rate  of  20  to  70  rpm  for  10  to  30  min.  The  residual  turbidity  of 
the  sample  is  measured  after  30  min  of  quiescent  settling.10 

According  to  Benefield  et  al.1 1  the  optimum  dose,  as  determined  by  jar  tests,  often 
does  not  match  that  observed  in  the  treatment  plant.  One  reason  for  this  is  that  the  jar 
test  is  a  batch  test,  whereas  the  static  flash-mixing  tanks  used  in  many  water-treatment 
plants  are  continuou&-fiow  devices  that  provide  significant  backmlxing.  A  batch  reactor 
is  analogous  to  a  plug-flow  reactor  and  is  Inherently  more  efficient  than  a  backmlxing 
reactor  for  most  reactions  (Vrale  and  lorden;1^  Benefield  e£  al.1  ^  consequently,  the  dose 
of  the  coagulant  required  for  the  flash  mixers  in  the  treatment  plant  may  be  greater  than 
that  indicated  by  the  jar  tests. 

v  This  situation  applies  to  the  type  of  coagulant  mixing  presently  used  in  the 
ERDLATOR  of  the  MWPU.  Agreement  between  jar  tests  and  plant  performance  may  be 
better  for  plants  utilizing  non-backmixing  devices,  such  as  that  used  in  4fce  600-gph 
ROWPU,  where  rapid  mixing  is  initiated  in  a  pipe  elbow  and  continued  at  a  relatively  high 
Reynolds  number  in  the  pipe  leading  to  the  multimedia  filter. 


15 


Volume  8 


Other  tests  used  to  ascertain  the  optimum  coagulant  type  and  dose  include 
electrophoretic-mobility  measurements  (zeta  potential)  and  more  recently,  particle-size- 
distribution  analyses.  Zeta  potential  is  closely  aligned  with  the  simple  double-layer 
theory  of  coagulation  and  hence  is  disparaged  by  those  who  believe  that  bridging  plays  the 

7 

dominant  role  in  cationic  polyelectrolyte  coagulation. 

INITIAL  MIXING  AND  MEAN  VELOCITY  GRADIENTS 

The  aluminum  and  ferric  iron  contained  in  alum  and  ferric  chloride,  respectively, 

will  generate  positively  charged  poiymer-ilke  chains  when  given  sufficient  time  and  the 

appropriate  water-quality  characteristics.  According  to  Weber,  the  hydrolysis  and 

polymerization  of  these  salts  of  Al(III)  and  Pe(III)  are  very  rapid  and  require  a  uniform  pH 

and  coagulant  concentration.  The  instantaneous  blending  of  the  alum  and  ferric-chloride 

feed  solutions  with  the  water  is  thus  essential.  The  instantaneous  blending  of  the 

high-molecular-weight  organic  polymers  with  the  water  is  less  essential  because  the 

polymers  do  not  have  to  be  formed  within  the  system,  and  their  rates  of  adsorption  by  the 

colloids  are  slower  due  to  their  larger  size.  Regardless  of  this  distinction,  it  is  apparent 

that  the  rate  at  which  a  coagulant  is  mixed  with  the  raw  water  is  nearly  as  Important  as 

the  optimum  coagulant  dose.  For  example,  the  poor  mixing  of  a  polyelectrolyte  with  the 

water  could  result  in  an  overdose  on  some  of  the  suspended  particles,  and  an  underdose  on 

others.  The  result  would  be  poor  coagulation,  even  though  the  average  dose  is  correct. 

The  mean  velocity  gradient  G  (defined  mathematically  as  the  square  root  of  P/Cty), 

where  P  is  the  mixing-power  input,  V  is  the  volume  of  water  being  mixed  such  that  P/V  is 

the  power  dissipation  per  unit  volume,  and  u  is  the  dynamic  viscosity  of  water)^®  is 

commonly  used  to  describe  the  intensity  of  mixing  required  in  both  chemical-mixing  and 

flocculation  devices.  The  greater  the  value  of  G,  the  more  rapid  the  mixing  rate  for  a 

device  of  a  given  geometry.  This  rule  can  not  be  extrapolated  to  devices  of  different 

12 

geometries,  as  demonstrated  by  the  results  of  Vrale  and  3orden. 

Those  investigators  conducted  some  rapid-mixing  experiments  with  different  types 

of  reactors,  using  alum  as  the  coagulant.  They  employed  one  completely  backmixed, 

stirred-tank  reactor  (CSTR)  and  four  turbulent  pipe-flow  reactors  of  various 
12 

geometries.  The  efficiency  of  the  coagulation  process  was  determined  by  measuring 
the  residual  turbidity  subsequent  to  a  period  of  slow  flocculation  and  sedimentation.  An 
apparent  turbidity-removal  rate  was  then  estimated  from  the  observed  turbidity  values. 
Some  of  the  results  of  their  study  are  presented  in  Fig.  6. 


Volume  8 


The  performance  of  the  CSTR  mixer  (Unit  2)  was  the  poorest,  with  the  rate  of 
aggregation  actually  decreasing  when  G  was  increased.  It  was  also  observed  that  the 
minimum  mixing  speed  had  to  be  greater  than  150  rpm  (G  s  350  s“b  to  achieve  a  uniform 
mixture  in  the  CSTR.  For  mixing  speeds  less  than  150  rpm,  the  water  was  not  mixed  at 
the  bottom  of  the  reactor.  Unit  1,  one  of  the  pipe-flow  reactors,  provided  a  slow  rate  of 
blending  and  was  used  to  determine  the  effect  of  the  background  mixing  in  the  piping 
following  the  in-line  turbulence  inducers.  The  performance  of  Unit  1  is  not  depicted  in 
F  ig.  6.  Unit  3,  which  had  a  complicated  design  and  no  practical  application,  was  intended 
to  serve  as  the  standard  of  best  performance,  although  this  did  not  prove  to  be  the  case. 
Unit  4,  a  combination  of  Units  1  and  3,  had  a  similar  performance  to  that  of  Unit  3.  Unit 
5,  which  performed  the  best  of  all  the  mixing  devices  tested,  consisted  of  an  annular  ring 
inserted  in  a  round  pipe.  The  ring  had  six  holes  for  alum  injection  into  the  flow  stream, 
with  the  holes  facing  downstream.  Significant  turbulence  was  created  by  expansion  of  the 
flow  downstream  from  the  annular  ring. 


[Mmh  velocity  gradient )  G(rU 

Figure  6.  The  effect  of  the  mean  velocity  gradient,  G,  on  the  aggregation  rate  of 
suspended  particles  for  rapid-mix  devices  of  various  geometries.  Adapted  from  Vrale  and 
Jorden.*  ^ 

Vrale  and  3orden  ^  concluded  that  the  completely  backmixed  stirred-tank  reactor 
(CSTR)  of  the  type  often  used  in  water-treatment  plants  provide  the  least -satisfactory 
way  of  mixing  two  miscible  fluids  rapidly,  and  that  G  by  itself  Is  not  a  sufficient  criterion 
for  judging  the  rate  of  initial  aggregation  for  devices  with  different  geometries.  The 
location  of  coagulant-application  points  and  the  turbulence  intensity  at  and  immediately 


17 


Volume  8 


downstream  from  the  chemical-addition  points  also  should  be  considered.  These 
researchers  also  indicated  that  an  optimal  design  for  colloid  destabilization  with  alum 
would  be  optimal  for  destabilization  with  cationic  poiyelectroiytes  because  the  particle- 
destabilization  mechanisms  are  similar.  This,  in  turn,  is  dependent  upon  the  local  and 
temporal  concentrations  of  the  coagulant  and  the  colloids  because  the  adsorption 
reactions  are  rapid  and  irreversible. 

CARTRIDGE  FILTERS 
13 

Cole  et  al.  define  a  cartridge  filter  as  a  cylindrical  object  consisting  of  a 

disposable  or  cleanable  porous  medium,  associated  with  an  appropriate  plastic  or  metal 

structure.  Individual  cartridges  or  multiples  of  cartridges  may  be  installed  in  a  filter 

housing.  In  most  Instances,  cartridge  filters  are  required  to  remove  particles  of  size 

14 

ranging  from  fractions  of  a  micrometer  to  40  ym  (Warring  gives  this  size  range  as  1  to 
500  ym),  and  they  are  used  in  many  industrial  applications.  Materials  such  as  cotton, 
wool,  rayon,  cellulose,  fiberglass,  polypropylene,  acrylics,  nylon,  asbestos,  etc.,  or  various 
combinations  of  these  materials,  are  commonly  used  as  filter  media  in  the  disposable 
cartridge  elements.  Materials  such  as  stainless  steel,  Monel  alloys,  and  ceramics  are  used 
in  the  reusable  filter  elements. 

Cartridge  filters  are  classified  Into  two  categories,  "depth"  and  "surface"  filters. 

Depth-type  filters  capture  particles  throughout  the  total  thickness  of  the  medium  within 

the  interstices  of  the  internal  structure.*  These  filters  normall^have  a  structure  that 

gradually  increases  in  density  toward  the  center  of  the  element.1  Thin  media,  such  as 

paper  or  woven-wire  cloths  are  described  as  surface-type  filters  because  most  of  the 

13 

particle  capture  occurs  at  or  near  the  filter  surface.  According  to  Cole  et  al.,  the  terms 

"surface"  and  "depth"  are  relative  rather  than  absolute  because  removals  vary 

considerably  with  the  size  and  chemical  properties  of  the  suspended  particles,  the  rigidity 

of  the  medium,  the  uniformity  of  the  element  pore  sizes,  and  the  thickness  of  the  medium. 

The  absolute  size  rating  of  a  cartridge  filter  Is  defined  as  the  largest  hard  spherical 

particle  that  will  pass  through  the  filter  under  specified  conditions,  or  sometimes  the 

largest  opening  in  the  filter.  The  "nominal"  size  rating  is  a  term  used  by  the  military 

and  is  defined  by  a  test  apparently  used  only  by  the  military.*^  Increments  of  a  fine  test 

dust  are  introduced  upstream  from  the  filter.  Small  portions  of  the  filter  effluent  are 

then  examined  microscopically  for  particle  counts.  The  counts  are-, converted 

mathematically  to  weight  values,  and  the  size  at  which  98%  of  the  particles  by  weight  is 

retained  on  the  filter  is  called  the  "nominal  size".  The  test  does  not  give  reproducible 

1 3 

results  and  has  never  been  widely  accepted. 


18 


Volume  8 


DIATOMACEOUS-EARTH  (PRECOAT)  FILTERS 

A  precoat  filter  consists  of  a  rigid,  semiflexible,  or  flexible  septum  (or  screen)  on 

14 

which  filter  aid  (or  medium)  is  deposited.  Usually  a  filter  system  consists  of  several 
septa  which  are  housed  in  a  pressure  vessel.  Each  of  the  septa  supports  a  thin  layer  of 
filter  aid  that  has  been  deposited  hydraulically  on  one  side  of  the  septa  at  the  beginning  of 
the  filtration  cycle}  this  thin  layer  is  called  filter  cake. 

The  screen  or  the  septum  on  which  the  aid  is  formed  is  basically  a  strainer,  and  the 
principal  mechanism  Involved  in  the  removal  of  the  solids  is  believed  to  be  mechanical 
straining  or  direct  interception.1**  The  septum,  the  filter  cake,  and  the  particles  which 
are  filtered  from  the  raw  water  all  contribute  to  the  straining  of  additional  suspended 
particles  from  the  water  being  filtered.  In  straining,  particles  larger  than  the  pore  size  of 
the  filter  cake  are  collected  on  the  upstream  surface  of  the  filter.  Adsorptive  forces  also 
provide  a  small  positive  contribution  to  the  removal  of  suspended  solids.  The  adsorptive 
forces  arise  from  the  nonuniformity  of  the  upstream  surface  of  the  filter,  which  results 
from  the  fact  that  the  pores  are  not  uniform  in  shape  or  direction  and  the  surface  is  not 

I L 

perfectly  smooth. 

The  desired  properties  of  the  precoating  filter  aid  are  that  it  be  a  finely  divided 
1  powder,  light  weight  and  insoluble  in  the  liquid  to  be  filtered.  The  aid  should  have  no 
effect  upon  the  chemical  properties  of  the  filtrate,  and  should  be  fine  enough  to  remove 
essentially  all  of  the  suspended  solids. 

The  filter  aid  used  In  most  precoat  filters  is  manufactured  from 
dlatomaceous  earth,  which  consists  of  the  siliceous  fossil  remains  of  dead  aquatic  algae 
called  diatoms.  The  diatoms  are  processed  by  crushing,  calcining  (heating  to  a  high 
temperature  without  fusing),  and  classification  to  yield  a  fine,  porous,  multishaped, 
angular  media,  ranging  in  size  from  about  5  to  50  pm.  Processed  dlatomite  filter  aid  has 
the  required  characteristics  for  water  filtration.  The  structure  permits  the  formation  of 
a  rigid,  porous  (>90%  porosity)  filter  cake  that  will  retain  solids  well  as  the  water  or  other 
liquid  containing  those  solids  passes  through  the  filter.  Perlite,  which  comes  from  a 
siliceous  rock,  is  another  material  which  can  be  used  in  precoat  filters.15 

The  clarifying  ability  is  the  parameter  used  to  characterize  the  performance  of 
dlatomite  filter  aid.  The  number  that  quantifies  this  ability  is  called  the  "clarity  index"} 
it  is  inversely  proportional  to  the  permeability  of  the  aid.  Thus,  as  the  permeability  of 
the  diatomite  increases  to  allow  more  throughput,  its  clarifying  ability  decreases  and 
larger  suspended  solids  pass  into  the  filtrate. 


19 


Volume  8 


The  precoat  filtration  run  or  cycle  consists  of  four  steps1  (l)a  thin  layer  of  filter 
aid  is  applied  to  the  septum  to  form  the  filter  precoat}  (2)  the  water  is  filtered  until  a 
predetermined  pressure  drop  across  the  filter  is  reached?  (3)  the  fouled  filter  aid  is 
expelled  from  the  filter  elements  by  backwashing;  and  (4)  the  water  In  the  filter  shell 
containing  the  backwash  sludge  is  released  to  waste. 

Precoat  filter  aid  is  applied  in  one  of  the  two  following  ways:  (1)  the  aid  is  applied 
directly  to  the  filter  septum  by  filtering  water  containing  the  aid  at  the  start  of  the  run, 
or  (2)  a  previously  prepared  filter-aid  slurry  is  recirculated  through  the  filter  until  the 
filter  effluent  returning  to  the  slurry  tank  is  clear.  Generally  this  precoat  has  a  thickness 
of  1/16  to  1/8  In.  The  procedure  to  be  used  should  be  selected  according  to  the  porosity 
of  the  septum  and  the  fineness  of  the  filter  aid.  The  recirculating  procedure  is 
advantageous  when  the  septum  openings  are  large  and  the  filter  aid  is  fine.16 

As  water  containing  suspended  particles  is  filtered  during  a  filtration  run,  the 
precoat-cake  surface  gradually  becomes  fouled  with  a  layer  of  particles.  This  Increase  In 
the  thickness  of  the  filter  cake  will  result  in  clogging  of  the  filter  and  an  Increased 
pressure  drop  across  the  filter.  This  effect  can  be  mitigated  to  some  extent  by 
continuously  Injecting  small  amounts  of  filter  aid  into  the  raw  water  as  it  enters  the 
filter,  which  helps  to  maintain  a  highly  porous  filter  cake.  This  additional  filter  aid  is 
called  body  feed. 

The  total  filter-aid  requirement  Includes  the  amount  of  material  used  to  precoat  the 
septa,  as  well  as  that  added  as  body  feed.  The  optimal  amount  of  precoat  material  to  be 
used  in  any  filter  will  be  the  minimum  amount  needed  to  protect  the  filter  element  from 
clogging  while  producing  an  effluent  of  the  desired  quality.  In  general,  body-feed 
requirements  vary  in  proportion  to  the  raw-water  turbidity,  although  the  type  of 
suspended  solids  also  affects  the  amount  of  body  feed  required.16 

If  the  primary  mechanism  of  suspended  solids  removal  with  diatomaceous  earth 
filtration  is  straining,  then  it  follows  that  the  removal  of  suspended  solids  Is  a  function  of 
the  particle  size  of  the  filter  aid.  Tien  and  Payatakes**  state  that  straining  Is  effective 

when  the  size  of  the  particles  Is  greater  than  0.2  times  the  filter-grain  size. 

13 

Cheremislnoff  and  Azbel  state  that  dlatomaceous-earth  filter  aids  have  good  efficiency 

14 

in  retaining  particles  of  sizes  less  than  1  ym.  Warring  does  not  give  a  quantitative 
measure  for  particle  size  but  states  that  dlatomaceous-earth  filter  aids  can  provide 
extremely  fine  filtering  and  are  used  frequently  when  very  high  purity  is  required,  such  as 
in  the  filtering  of  sugars,  edible  oils,  or  water.  The  particle  size  of  commercially 
available  dlatomaceous-earth  filter  aids  ranges  from  5  to  50  ym. 


20 


Volume  8 


Prior  to  filtration,  the  raw  water  to  bo  filtered  may  be  coagulated,  flocculated,  and 
allowed  to  settle.  However,  all,  some,  or  none  of  these  three  processes  may  be  used)  the 
choice  will  be  made  based  on  the  raw-water  quality. 

The  modes  of  filter  operation  shown  in  Fig.  1  for  deep-bed  filters  apply  equally  well 
to  precoat  filters.  The  term  direct  filtration  is  also  used  for  precoat  filters  when  the 
coagulated  water  is  not  settled  prior  to  filtration. 

DESCRIPTION  OF  THE  600-GPH  ROWPU  PRETREATMENT  SYSTEM 

According  to  Carnahan  et  al.,17  a  conventional  clarification  system  with  separate 

coagulation-flocculation  and  sedimentation  basins  could  not  be  used  in  the  600-gph 

ROWPU  pretreatment  system  because  of  restrictions  Imposed  by  the  U.S.  Army  on  the 

overall  dimensions  of  the  ROWPU.  Instead,  the  pretreatment  used  consists  of  direct 

filtration  by  a  multimedia  pressure  filter  with  in-line  polyelectrolyte  addition,  followed  by 

processing  through  a  cartridge-filtration  unit.  This  pretreatment,  aside  from  the 

cartridge  unit,  is  somewhat  in  accord  with  scheme  C  In  Fig.  1,  except  that  a  cationic 

polyelectrolyte  is  used  in  lieu  of  the  alum,  activated  silica,  or  nonionic  polymer,  and  the 

filtration  rate  is  7  gal/(mln  *  ft  ),  Instead  of  the  5  gai/(min  •  ft  )  shown  in  the  scheme. 

As  shown  in  Fig.  7,  the  multimedia  filter  removes  solids  that  pass  through  the  Input- 

water  strainer,  as  well  as  the  smaller  solids  that  are  acted  upon  by  the  coagulant.1  A 

cationic  polyelectrolyte  and  sodium  hexametaphosphate  are  added  ahead  of  the 

multimedia  filter.  Citric  acid  Is  added  after  the  multimedia  filter  and  before  the 

18  19 

cartridge-filter  unit  according  to  Fig.  7.  *  Sodium  hexametaphosphate  and  citric  acid 

are  added  primarily  to  control  the  rate  of  RO  membrane  fouling.  The  citric  acid  is  added 

18 

in  quantities  sufficient  to  adjust  the  water  pH  to  a  value  between  5  and  8.  Chlorine  is 

added  after. the  RO  unit  to  prevent  chlorine  from  damaging  the  membrane  elements. 

Carnahan  et  al.,17  state  that  the  multimedia  filter  was  designed  to  operate  at  rates 

ranging  from  5  to  10  gal/(mln  •  ft2)  because  this  range  provides  the  optimum  trade-off  in 

1 8 

size  versus  filtration  efficiency.  The  nominal  filtration  rate  Is  stated  to  be 
2 

6.5  gal/(mln  •  ft  ),  which  is  equivalent  to  a  total  ROWPU  feed-water  flow  of  32  gal/mln. 
COAGULATION 

The  cationic  polyelectrolyte,  Cat-Floe  Tl  (manufactured  by  Caigon,  Pittsburgh,  PA) 
is  used  In  the  ROWPU  pretreatment  system.17  It  is  pumped  to  the  filter-feed  line  with  an 
American  Lewa  positive-displacement  pump  with  a  200-mL/min  capacity.  According  to 


21 


Volume  8 


Product 

water 


* 

Figure  7.  Water-processing  block  diagram  for  the  600-gph  ROWPU.  From 
'  TM  5-4610-213-10.18 

the  600-gph  ROWPU  operator's  manual,*8  the  chemical  feed  pump  is  calibrated  to  a  flow 
rate  of  60  mL/min,  which  corresponds  to  a  dosage  of  5.2  mg/L  for  the  nominal  ROWPU 
flow  rate. 

The  chemical-feed  flow  rate  is  adjusted  during  operation  to  Improve  the 
performance  of  the  multimedia  filter  if  necessary.  This  Is  achieved  by  using  a  turbidity 
tube,  the  bottom  of  which  contains  a  white  bull's-eye  set  in  a  black  background  disk.  A 
600-mL  flltered-water  sample  is  drawn  into  the  tube  from  the  cartridge-filter  drain.  The 
criterion  for  water  clarity  is  that  both  the  white  bull's-eye  and  the  black  disk  at  the 
bottom  of  the  tube  should  be  seen  clearly.  If  both  are  not  visible,  the  polymer  dosage 
should  be  altered.  This  is  accomplished  by  changing  the  chemical-feed  flow  rate.  The 
U.S.  Marines  adjust  the  dosage  in  accord  with  the  600-gph  ROWPU  operator's  manual.*8 
It  Is  uncertain  what  operating  procedure  is  followed  by  the  U.S.  Army. 

Carnahan  et  al.,*^  state  that  the  ROWPU  chemical-feed  system  provides  a  mean 
velocity  gradient  of  300  s“*  for  approximately  0.1  s  and  800  s"  for  2  s.  This  was 
checked  for  a  total-flow  rate  of  34.5  gal/min  and  a  2-in.  i.d.  multimedia  filter  feed-water 
pipe  6  ft  long  with  two  right-angle  elbows.  The  mean  velocity  gradient  G  was  estimated 


22 


Volume  8 


to  be  approximately  3000  s  .or  the  elbow  where  the  coagulant  is  injected,  and 
approximately  800  s’*  for  the  pipe.  The  residence  time  for  the  pipe  was  about  2  s.  These 
numbers  check  weil  with  those  reported  by  Carnahan  jrt  al.,* '  except  for  the  discrepancy 
in  the  G  value  at  the  elbow. 

MULTIMEDIA  FILTER 

The  multimedia  filter  in  the  600-gph  ROWPU  is  a  pressure  filter  (Cuiligan  mixed- 
media  filter)  operated  under  a  nominal  pressure  of  30  pslg.*2  The  dimensions  and 
characteristics  of  the  media  installed  in  the  filter  are  given  in  Fig.  8.  The  minimum 

j  8 

operating  pressure  is  reported  to  be  23  psig. 

The  backwash-water  system  for  the  multimedia  filter  is  shown  schematically  in 
Fig.  9.* 8  Brine  from  the  RO  elements  is  used  as  the  backwash  water.  The  backwash 
pump  is  a  centrifugal  pump  with  a  maximum  capacity  of  120-gai/mln  water  flow  and 
70-psig  pressure  and  is  controlled  by  a  gate  valve  as  shown  in  Fig.  9.  The  backwash 
system,  which  is  provided  with  a  timer,  operates  automatically.  The  backwash  operation 
lasts  approximately  20  min  after  the  backwash  pump  Is  started.  During  backwash,  the 
water-flow  rate  varies  automatically  from  0  to  between  70  and  120  gal/mln  (0  to  between 
1  14  and  24  gal/[mln  •  ft2])  while  washing  and  rinsing,  which  fully  fluldlaes  the  multimedia 
bed. 1,1 7,1 8  According  to  the  operator's  manual,  the  filter  should  be  backwashed  after 
20  h  of  operation  or  when  the  filter  pressure  drop  exceeds  the  Initial  drop  by  3  psi.  The 
Initial  pressure  drop  is  usually  about  2  psi.*2 

Carnahan  et  al.,  7  give  the  results  of  some  ROWPU  multimedia  filtration  tests, 
from  which  it  was  estimated  that  the  filter  retained  approximately  3300  g  of  solids  just 
before  backwashing  was  required.  It  was  also  observed  that,  by  using  two  backwash  cycles 
of  15  and  ?4.gal/(mln  •  ft2),  the  filter  was  cleaned  sufficiently  to  regain  the  Initial 
pressure  drop  of  2  psi. 

ROWPU  CARTRIDGE  FILTERS 

The  effluent  of  the  multimedia  filter  is  polished  with  5-ym  nominal  sire  rating 
cartridge  filters,*  (The  definition  of  nominal  Is  given  In  the  section  entitled  Cartridge 
Filters).  The  purpose  of  this  unit  is  to  prevent  the  carry-over  of  solids  or  organics  to  the 
reverse  osmosis  (RO)  components.*2 

Eight  replaceable  filter  elements  are  housed  in  a  single  pressure  vessel  as  Illustrated 
in  Fig,  10.  The  filter  elements  consist  of  polypropylene  cord  woven  around  stainless  steel 
cores.  The  water  flows  in  a  spiral  motion  around  and  Into  the  stainless  steel  core,  and 


23 


volume  8 


3  In. 


14  In. 


7  In. 
3  In. 


30  In.  (i.d.) 


Riw-w»ttr  Inlet 


=535^ 


1/8  X  1/8-In.  Plastic  pellets 


Anthrioits  ooal 


Calolnad  aluminum 
illloats 


Garnet  tend 


Small  iliad  garnet  gravaT 


Madlum-tiiad  garnet  gravel 


Effective 

diameter 

(mm) 


0.80 


0.42 


0.30 


L>—  Fllterad-watar  outlet 


Speeiflo 

gravity 

1.2 


1.5 


2.5 


3.95 


Density 

Uniformity  lb/ft* 

coefficient  (kg/m3) 

—  45 

(720) 


2 


52 

(830) 


1.74  — 

1.2  135 

(2100) 


Figure  8.  ROWPU  multimedia  filter.  From  Small  etal..1 


then  out  of  the  bottom  of  the  pressure-containment  vessel.  The  filter  elements  are 
classified  in  the  group  called  "depth"  filters  with  high  contaminant-holding  capacity.20 
The  tubular  elements  are  40  in.  long  and  2-3/4  in.  in  diameter,  which  gives  a  surface  area 
of  2.4  ft2  per  element,  or  a  total  of  19.2  ft2  per  installation.21  The  filtration  rate  is  thus 
1.8  gal/(min  •  ft2)  (14.4  m/h)  for  a  ROWPU  flow  rate  of  34.5  gal/min. 

According  to  the  manufacturer  of  the  cartridge-filter  elements  (Fllterite 
Corporation,  Timonium,  MD),  the  polypropylene  filter  media  is  compatible  for  use  wltn 
strong  acids  and  concentrated  bases,  and  at  temperatures  up  to  275°F  (135°C). 
Polypropylene,  however,  is  affected  by  oxidizing  agents  such  as  free  chlorine.  This  may 
inhibit  or  exclude  prechlorination  of  the  raw  water  if  the  600-gph  ROWPU  is  to  be 
operated  in  the  bypass  mode  (i.e,,  by-passing  the  ROWPU  RO  section). 

DESCRIPTION  OF  THE  MOBILE  WATER  PURIFICATION  UNIT 

A  Mobile  Water  Purification  Unit  (MWPU)  was  developed  and  tested  by  the 
Engineering  Research  and  Development  Laboratory  (ERDL)  at  Fort  Beivoir,  Virginia.  The 
development  of  this  equipment  was  the  U.S.  Army's  response  to  a  need  for  a  lightweight, 


24 


V  WIUIIIV  u 


Baokwaih 

watar 

connection 


Canvaa 

brlna 

hoaa 


2-In.  Swivel 
adaptar 


2-In.  Doubla  I 
hoia  nlppla 


Figure  9.  The  backwash  water  system  for  the  600-gph  ROWPU  multimedia  filter.  From 
TM  5-4610-213-10.18 


mobile  system  that  was  reliable  and  capable  of  producing  potable  water  for  field  troops  in 
a  variety  of  hostile  environments.  Concern  was  focused  on  operation  with  cold  water,  and 
the  equipment  was  designed  for  freshwater  point  sources. 

BACKGROUND 

Water  fit  for  drinking  is  now  used  by  the  field  Army  for  drinking,  bathing,  kitchen 
tasks  and  laundry.  If  local  water-treatment  facilities  are  not  available,  then  raw  fresh 
waters  must  be  treated  by  field  equipment  to  provide  for  these  needs.  The  equipment 
must  be  reliable  and  capable  of  treating  water  from  almost  any  surface  source.  While  the 


2.5 


volume  b 


Figure  10.  Diagram  of  the  600-gph  ROWPU  cartridge  filter.  From  TM  3-4610-213-10. 


18 


water  points  are  checked  by  the  Medical  Corps  for  suitability,  easily  treatable  water  Is 

not  always  found,  and  the  equipment  may  be  called  upon  to  purify  waters  that  are  turbid, 

colored,  polluted  with  natural  biological  pathogens,  or  have  extreme  pH  values.  Nuclear, 

22 

biological,  and  chemical  warfare  agent  removal  is  also  a  concern. 

The  terrain  and  weather  conditions  In  which  the  water  treatment  Is  to  take  place 
are  unlikely  to  be  kind  to  the  equipment.  For  example,  the  equipment  will  be  used  in 
Isolated  locations  where  the  only  protection  for  the  unit  Is  that  which  Is  actually  part  of 
the  unit  Itself.  It  must  be  capable  of  operating  at  high  and  low  ambient  temperatures,  in 
rough  terrain,  in  rain,  sleet,  high  winds,  and  at  night.  The  equipment  may  be  moved  on 
short  notice,  and  should  be  self-contained  so  that  it  can  be  started  up  soon  after  arrival  at 
the  next  water  point.  Because  the  troops-often  must  operate  the  equipment  under  adverse 
conditions,  it  should  be  simple  to  operate  and  rugged.  Moreover,  a  minimum  number  of 
troops  should  be  required  to  operate  it. 


26 


Volume  8 


The  development  of  the  Mobile  Water  Purification  Unh  (MWPU)  began  in  1949,  and 

testing  was  completed  in  1955.  This  work  was  conducted  by  the  U.S.  Army  in  conjunction 

with  special  studies  performed  by  institutions  funded  by  the  Army.  All  three  processes 

used  for  water  treatment  in  the  MWPU  were  studied:  (l)  coagulation  and  clarification,  (2) 

dlatomaceous-earth  filtration,  and  (3)  disinfection.  Research  projects  were  funded  at 

Harvard  University  on  water-disinfecting  agents,  at  New  York  University  on 

water-coagulation  procedures,  at  the  University  of  Illinois  on  filter  septa  and  water 

filterabllity,  and  at  Johns  Hopkins  University  on  filter  aids.  The  Sanitary  Engineering 

Branch  of  ERDL  conducted  basic  research  and  development  efforts  on  related  studies. 

Two  prototype  water-purification  units  were  tested  using  the  solids-contact 

clarifier,  filters,  and  disinfection  processes.  Field  operations  were  carried  out  at  sites  at 

Lake  Michigan  in  Illinois,  at  several  lake  and  river  sites  in  southern  Virginia,  and  at  the 

Potomac  River  site  of  ERDL  at  Fort  Belvolr,  VA.  Fifty  hours  of  operation  were  logged  at 

each  site.  The  two  test  models  used  were  similar,  except  for  the  solids-contact  clarifier. 

One  had  an  Army-designed  clarifier  and  the  other  an  off-the-shelf  commercial  clarifier, 

23 

which  could  serve  the  same  function  as  the  Army  modei. 

The  results  of  these  tests  gave  U.S.  Army  designers  the  criteria  to  be  used  for 

equipment  having  the  required  military  characteristics.  They  decided  what  design 

modifications  should  be  made  to  accommodate  the  high  and  low  ambient  temperatures  in 

which  the  equipment  would  be  operated,  and  they  recommended  that  the  "Water 

Purification  Unit,  Mobile,  Electrified,  1500-GPH  Capacity"  should  be  designed  and 

fielded.  They  also  recommended  the  expansion  of  the  project  to  include  development  of  a 

23 

3000-gph  capacity  model.  Most  data  on  MWPU  performance  have  been  published  for 
the  3000-gph  model. 

TREATMENT  SYSTEM 

The  MWPU  with  supporting  equipment  in  place  Is  illustrated  in  Fig.  11.  A  cutaway 
view  of  the  processes  contained  in  the  truck-mounted  van  is  shown  in  Fig.  12.  The 
treatment  processes  consist  of  the  following:  solids-contact  clarification}  disinfection 
with  calcium  hypochlorite}  external  solids  concentration}  and  diatomaceous-earth 
filtration.  The  solids  contact  clarifier,  designed  at  ERDL,  is  called  the  ERDLATOR.  The 
rate  at  which  the  water  is  treated  by  the  MWPU  is  controlled  by  the  hydraulic  loading  on 
the  ERDLATOR.  The  MWPU  is  fielded  in  three  versions,  600,  1500,  and  3000  gph. 


27 


T  VIVIIM&  g 


Volume  8 


The  raw  water  is  pumped  to  the  clarifier  with  a  self-priming,  2-hp  centrifugal 
pump.  Large  debris  is  excluded  with  a  strainer  installed  on  the  suction  hose  as  shown  in 
Fig.  11.  The  effluent  from  the  ERDLATOR  is  pumped  through  two  diatomaceous-earth 
filters.  The  filter  effluent  then  flows  by  gravity  to  two  3000-gal  capacity  storage  tanks  in 
series.  The  treated  water  is  pumped  by  a  second  self-priming,  2-hp  pump  from  the  second 
storage  tank  to  the  distribution  system.  Wastewaters  from  the  system,  which  include  the 
concentrated  sludge  derived  from  the  ERDLATOR,  as  weli  as  the  backwash  of  the  filters, 
are  released  by  gravity  to  a  discharge  point  located  downstream  from  the  raw  water 
intake.  Normally,  the  equipment  can  be  Installed  in  less  than  one  hour  by  three  soldiers 

2ti 

and  routinely  operated  by  one. 

The  van-type  body  used  to  hold  the  treatment  units  for  the  3000-gph  model  is 

25 

258  in.  long,  98  in.  wide,  88  in.  high,  and  Is  carried  on  a  2.5-ton  truck.  The  van  interior 

is  insulated  and  heated.  When  the  unit  Is  operational  at  temperatures  between  0  and 

26 

-40°F,  a  gas-burning  heater  is  used  to  warm  the  treatment  system;  other  heating 

devices  may  be  used  to  prevent  freezing  of  the  effluent  in  the  distribution  system.  The 

shipping  weight  of  the  van  is  9400  lb.  The  total  weight  is  much  greater  when  operating, 

due  to  the  weight  of  water  being  orocessed;  the  van  is  leveled  and  supported  by  four  jacks 

23 

mounted  on  the  van  frame. 

All  the  required  power  is  supplied  by  a  military  standard  10-kW  generator  that 
weighs  1500  lb.  Fifty  feet  of  insulated  power  cable  is  used  so  that  the  generator  may  be 
operated  remotely  from  the  water-purification  unit,  thereby  minimizing  the  noise  to 
which  the  operators  are  subjected.  The  generator  is  skid-mounted  and  transported  on  a 
1.5-ton  cargo  trailer, 

The  ERDLATOR 


Figures  13  and  14  present  cross-sectional  and  three-dimensional  views,  respectively, 
of  the  ERDLATOR  used  in  the  MWPU.  The  raw  water  is  first  metered  and  then 
introduced  into  two  small  mixing  launders  located  above  the  floccuiator.  Dissolved  gases 
that  may  be  supersaturated  in  the  feed  water  are  removed  to  a  considerable  extent  in 
these  chambers  by  water  forced  through  four  aspirator  nozzles,  as  shown  in  Fig.  14. 

Solutions  of  calcium  hypochlorite,  ferric  chloride,  and  pulverized  limestone  slurry 
are  all  added  to  the  water  in  the  influent  launders  in  the  production  model  of  the 
ERDLATOR.  (Figure  13  indicates  the  addition  of  chemicals  at  different  points  in  the 
apparatus;  this  was  the  mixing  arrangement  used  in  the  prototype  model.)  The 
arrangement  used  to  add  the  ccagul&nt  (ferric  chloride)  to  the  water  makes  calculation  of 
the  mean  velocity  gradient  G  impossible. 


29 


Volume  8 


Figure  13.  Cross-sectional  flow  diagram  of  ERDLATOR  (original  testing  version). 
Adapted  from  Ruiz  and  Schmitt.^ 


From  the  launders  the  water  then  passes  to  the  cylindrical  flocculation  chamber, 
which  is  separated  into  three  or  more  sections  with  horizontal,  flat-metal  disks.  The  disks 
are  attached  to  a  vertical  hollow  shaft  and  are  spaced  equally  along  the  shaft.  The  disks 
serve  to  divide  the  chamber  into  compartments,  thereby  shortening  the  required  mixing 
time  and  avoiding  major  opportunity  for  short-circulating.  The  shaft  is  rotated  at  about 
100  rpm  to  provide  velocity  gradients  for  the  flocculation  process.  This  agitation  also 
serves  to  detach  gas  bubbles  from  the  solids.  The  gas  can  escape  either  through  holes 
drilled  in  the  disks  or  in  the  hollow  shaft.  The  mixing  chamber  contains  approximately 
40%  of  the  total  volume  of  930  gal  held  in  the  ERDLATOR  tank  of  the  3000-gph  MWPU. 
This  gives  a  mean  flocculation  time  of  about  7-1/2  min.  The  flat-metal  disks  have  an 
edge  velocity  of  24  ft/sat  100  rpm.26 

The  water  leaving  the  flocculation  chamber  is  directed  vertically  upward  in  the 
solids  contact  clarifier  by  shallow  baffles  that  reverse  the  rotation  of  the  water  as  it 
leaves  this  chamber.  A  well-defined  ferric  chloride-limestone  slurry  sludge  blanket  is 
formed  in  the  clarifier,  as  indicated  in  Fig.  13.  The  excess  sludge  is  drawn  off  at  a  point 
located  about  10  in.  below  the  circular  dual-edge  coiiecxor  weirs.  The  waste  sludge  is 
allowed  to  concentrate  by  gravity  in  a  separate  sludge  concentrator,  as  Jllustrated  in 
Figs.  13  and  14. 


30 


Volume  8 


Influent  launder 


Figure  14.  Cross  section  of  the  1500-gph  MWPU  (ERDLATOR)  assembly.  Adapted  from 
TM  5-4610-218-1  2. 25 


The  vertical  upflow  water  velocity  in  the  clarification  zone  is  about  3  ft/mln 
(1.22  gal/fmin  •  ft  ])  at  the  section  of  maximum  horizontal  area  for  the  3000-gph  model. 
It  was  recommended  that  this  "rise  velocity"  should  be  reduced  for  water  temperatures 
below  40°F  to  allow  for  floe  formation.  The  total  mean  residence  time  in  the  ERDLATOR 
including  the  flocculation  zone  is  about  20  min.  The  turbidity  of  the  effluent  typically 
ranges  in  value  from  0.5  to  2.0  turbidity  units.22 


3  i 


Volume  8 


The  effluent  from  the  ERDLATOR,  together  with  the  clear  supernatant  from  the 
sludge  concentrator,  flows  to  a  wet  well.  The  water  is  then  pumped  to  the 
dlatomaceous-earth  filters.  A  float  operated  valve  triggers  a  switch  to  notify  the 
operator  if  the  wet-well  level  drops  too  low. 

Conditioning  Chemicals 

Ferric  chloride,  the  primary  coagulant  used  in  the  ERDLATOR,  is  dissolved  in  water 
and  is  pumped  to  the  ERDLATOR  with  a  dual-sided  electrical  diaphragm  pump.  The  dose 
applied  is  commonly  in  the  range  of  50  to  75  mg  coagulant/L  feedwater*  ^ 

A  slurry  of  the  pulverized  limestone  is  prepared  in  a  small  dual-compartment  slurry 
feeder.  The  tank  contains  agitators  that  revolve  vertically.  Measuring  cups  attached  to 
the  agitators  convey  the  chemical  to  a  funnel  leading  into  the  slurry  tank.  The  feed  rate 
is  controlled  by  hand,  and  the  amount  of  chemical  to  be  added  is  determined  by  the  soldier 
on  duty.  Commonly,  the  limestone  dose  rate  varies  from  50  to  150  mg  of  limestone  per 
liter  of  influent  feed  water.  The  same  slurry  feeder  is  used  to  prepare  the 

diatomaceous-earth  body  coat  for  the  filters.  The  slurries  are  pumped  to  the  ERDLATOR 
or  the  filters  with  a  pump  located  Inside  the  slurry  tank. 

The  pulverized-limestone  particles  tend  to  become  enmeshed  with  and  add  weight  to 
the  flee  particles  formed  by  the  ferric  chioride  coagulant.  In  their  research  work,  the 
U.S.  Army  found  in  many  different  test  situations  that  the  use  of  pulverized  limestone 
brought  stability  to  the  coagulation  process  and  provided  water  of  relatively  uniform 
quality  in  the  effluent.  Limestone  also  provides  some  protection  against  low-pH  waters. 
The  enhanced  sludge-blanket  stability  also  makes  it  possible  to  bring  the  MWPU  on-line 
quickly.^ 

A  solution  of  calcium  hypochlorite  is  prepared  and  pumped  to  the  ERDLATOR  wi  !h 

one  side  of  the  dual-sided  diaphragm  pump,  which  is  also  used  to  pump  the  ferric  chloride 

solution.  This  disinfectant  also  Increases  the  rate  of  coagulation.  The  dosage  is  adjusted 

by  the  operator  to  give  the  same  total-residual  chlorine  in  the  fully  treated  potable  water 

as  that  specified  by  medical  personnel.  Equipment  and  chemicals  are  available  for 

determination  of  either  free-residual  or  total-residual  chlorine.  Typical  doses  of  calcium 

77 

hypochlorite  range  from  3  to  7  mg/L. 

Powdered  activated  carbon  may  also  be  added  to  control  strong  tastes  and  odors. 

The  carbon  is  applied  in  the  mixing  zone  of  ■•he  ERDLATOR  at  the  samejpoint  as  the 

ferric  chloride,  calcium  hypochlorite,  and  pulverized  limestone.  The  activated-carbon 

77 

dose  recommended  for  this  treatment  is  2  to  10  mg/L. 


32 


Volume  8 


Air  pump 


Air  blMd 


External  prtnt'r*  gaga 


Filter  alamantt 
( wirt  ■wound  with 
diatomite  filter- 
aid  caka) 


Diatomite  praooat  funnal 


Viawlng  window 

Coagulated  water  from 
ERDLATGR  wat  wall 


Internal 
prMiure  gaga 


EXT'.""."!  Baokwath  diioharga 


Filtered  affluent 


Figure  15.  Cross  section  of  diatomite-f .liter,  final  testing  device  for  use  in  the  Mobile 

50 

Water  Purification  Unit.  From  Donahew. 


Precoat-F liter  Operation 


The  water  stored  in  the  ERDLATOR  wet  well  is  pumped  by  a  centrifugal  pump  to 

50 

the  diatomaceous-earth  filters,  as  illustrated  in  Fig.  15.  The  water  enters  the  vertical 

cylindrical  shell  of  the  filters  under  pressure  and  is  forced  through  six  diatomaceous-earth 

filter  elements.  The  filtered  water  then  flows  by  gravity  to  two  3000-gal  storage  tanks 

2 

arranged  in  series.  The  total  effective  surface  area  of  the  filter  elements  Is  10  ft  for 
each  of  the  two  filters.  ^ 

The  wire-wound  filter  elements  are  precoated  with  a  slurry  of  diatomaceous  earth. 

Diatomaceous  earth  is  also  added  continuously  to  the  influent  water  to  give  longer  filter 
25 

runs.  A  controller  installed  on  the  filtered-water  line  is  used  to  control  the  flow  rate 

through  the  filter  element.  The  maximum  filtration  rate  permitted  by  the. controller  is 

2  2 
3.0  gal/(min  •  ft  )  and  the  nominal  rate  is  2.5  gal/(mln  •  ft  )  for  a  3000-gph  production 

rate. ^ 


33 


Volume  8 


Eventually  the  pressure  drop  across  the  filter  elements  builds  up  to  such  an  extent 
that  the  filter  cake  and  captured  solids  must  be  removed  and  the  filter  elements 
recoated.  This  is  accomplished  by  the  "air-bump"  method.  Air  is  trapped  In  two  places  in 
the  filter  housing  during  a  filter  run,  and  the  pressure  on  the  filter  elements  is  relieved 
almost  instantaneously  when  the  air-relief  valve  located  on  top  of  the  filter  is  opened  by 
the  operator  to  initiate  backwash.  The  trapped  air  dislodges  the  filter  cake  and  captured 
solids  in  its  rush  to  the  air-relief  valve.  The  expended  filter  aid  is  then  washed  to  waste. 
The  loss  of  production  of  filtered  water  due  to  the  time  that  it  takes  to  backwash  is 
minimal,  amounting  to  only  2%  according  to  estimates  for  operation  under  combat 
conditions.26 

The  section  of  the  van  that  contains  the  two  dlatomaceous-earth  filters  also 
Includes  the  filter  pumps.  The  pumps  are  used  to  provide  water  to  the  filters,  as  well  as 
to  flush  the  filters  after  backwashing.  Also  included  are  pressure  gauges,  flow 
controllers,  air-relief  valves,  precoat  tanks,  and  the  necessary  pipes  and  valves. 

PERFORMANCE  OF  DEEP-BED  FILTERS  OPERATED  IN  THE  DIRECT-FILTRATION 
MODE 

Filtration  is  a  process  intended  to  remove  particulates  such  as  bacteria,  viruses, 
algae,  protozoan  cysts,  and  clay  from  water  and  other  liquids.  Sand  and  its  derivatives 
are  commonly  used  as  deep-bed  filter  media,  and  garnet  and  coal  have  been  used  since  the 
1960s.  The  multimedia  deep-bed  filter  used  in  the  600-gph  ROWPU  Is  shown  in  detail  in 
Fig.  8. 

The  conventional  steps  for  granular-media  filtration  Include  coagulation, 
flocculation,  sedimentation,  and  filtration,  as  shown  in  schemes  A  and  B  in  Fig.  1.  No 
provision  Is  included  in  the  treatment  train  of  the  600-gph  ROWPU  for  flocculation  or 
sedimentation.  This  approach  represents  one  form  of  the  process  called  "direct 
filtration".  Many  papers  have  been  published  about  the  proper  operation  of  the 
flocculation  and  sedimentation  steps  for  filters  operated  in  the  conventional  mode,  but 
because  the  focus  of  this  report  is  the  ROWPU,  those  papers  have  been  omitted  from 
further  consideration  herein, 

For  good  performance  of  a  conventional  filtration  plant,  the  consensus  is  that  proper 

coagulation  is  the  most  Important  of  the  three  steps  (coagulation,  flocculation,  and 

29-31 

sedimentation)  that  may  precede  filtration.  Incomplete  particle  destabilization 

occurs  with  coagulant  doses  that  are  either  too  high  or  too  iowj  this  permits  particles  to 
pass  through  the  filter.  An  Environmental  Protection  Agency  (EPA)  study  on  the 
removal  of  Giardia  muris  cysts  and  coliform  bacteria  gives  evidence  that  although  some 


34 


Volume  8 


Table  1.  Effect  of  coagulant  dosage  and  type  on  the  removal  of  Giardia  muris  cysts  and 
coliform  bacteria  by  direct  filtration  with  preliminary  flocculation.*  (20-  to  30-mln 
flocculation  time,  20#C  water  temperature,  and  low-turbldity  feed  water.) 


Coagulant 

Coagulant  dose 
(mg/L  feedwater) 

Filtered- 
water  turbidity 
(NTU)b 

Cyst  removal 
(%) 

None 

— 

0.35 

59  to  94 

Alum 

1.8 

0.60  to  0.65 

94 

Alum 

1.9 

0.77  to  0.79 

23 

Alum 

2.1  to  2.2 

0.46  to  0.63 

63  to  88 

Alum 

7 

0.32 

>95 

Alum 

9.8 

0.26 

99.7 

Alum  nonionic  polymer 

11  4  0.01 

0.08 

99.75 

Coliform  removal 

<%) 


None 

— 

0.54  to  0.72 

44  to  62 

None 

— 

0.26  to  0.47 

96.4  to  98.1 

Alum 

30 

0.24  to  0.42 

98.7  to  98.9 

Alum 

30 

0.22 

99.2 

Cationic  polymer 

0.5 

0.16  to  0.21 

97.7  to  98.7 

a - rs — 

Adapted  from  Logsdon  and  Fox. 

13  NTU  s  nephelometric  turbidity  unit. 


removal  is  obtained  with  even  Inadequate  coagulation,  the  performance  of  the  filter  Is 
much  enhanced  with  properly  coagulated  water.  Evidence  of  this  enhancement  Is  given  by 
the  data  presented  In  Table  l. 

Because  the  coagulation  step  is  so  important,  It  behooves  the  equipment  operator  to 
use  the  best  type  and  optimum  dosage  of  coagulant  in  order  to  achieve  maximum  filter 
performance.  A  number  of  tests  have  been  proposed  that  might  be  used  to  ludge  the 
appropriate  coagulant  dosage  and  type.  These  Include  the  far  test,  zeta-potentlal 
measurements,  and  particle-size-distribution  analysis.  The  advantages  and  limitations  of 
these  tests  will  be  discussed  in  detail  subsequently. 


35 


Volume  8 


One  technique  for  monitoring  filter  performance  involves  the  use  of  turbidity 
measurements,  and  another  is  by  particle-size-distribution  analysis  of  the  filter  feed  and 
effluent  waters.  Early  filter  breakthrough  can  be  detected  with  continuous  or  frequent 
turbidity  measurements.  Logsdon  and  FoxJ*  state  that  this  is  particularly  Important  for 
filters  required  to  remove  viruses  or  Giardia  cysts.  To  date  particle-size-distribution 
analysis  has  not  been  used  much  for  the  control  of  the  filtration  process,  but  turbidity  has 
been  almost  always  measured  since  the  early  times  of  rapid  sand  filtration. 

The  impact  of  backwashing  is  also  noteworthy.  Backwashing  is  necessary  when 
32 

turbidity  rises.  Backwashing  should  be  initiated  as  soon  as  the  operator  notices  rising 
turbidity,  even  though  the  turbidity  of  the  effluent  is  still  <  1  NTU.  In  an  EPA  study  to  be 
discussed  in  a  later  section,  it  was  found  that  "a  small  increase  in  turbidity  can  be 
associated  with  a  dramatic  increase  in  cyst  concentration." 

When  the  filter  is  returned  to  service  after  backwashing,  poor  effluent-water  quality 
may  be  experienced  for  a  short  time.  Filtering  to  waste  until  the  quality  of  the  filtered 
water  improves  is  a  commonly  used  strategy.  Harris  found  that  when  a  nonionic 
polymer  was  added  to  the  backwash  water,  low-turbidity  filtered  water  was  produced 
immediately  after  the  filter  run  began.  The  nonionic  polymer  was  thought  to  Improve  the 
adsorption  of  particles. 

The  seemingly  close  association  of  turbidity  removal  with  the  removal  of 
microorganisms  in  deep-bed  filtration  made  it  necessary  to  review  direct-filtration  papers 
dealing  solely  with  turbidity.  Some  of  those  studies  will  be  covered  in  detail  in  the 
following  sections  of  this  report. 

DETERMINATION  OF  BEST  TYPE  AND  OPTIMUM  DOSAGE  OF  COAGULANT 

35 

Adin  qnd  Rebhun  injected  a  cationic  poiyelectrolyte  (Cat-Floe)  into  a  1/2-in.  pipe 
leading  to  two  2-1/4-in.  l.d.  sand-filter  columns.  The  blending  between  the  coagulant  and 
the  water  containing  20  mg/L  of  kaolinite  clay  was  achieved  by  the  turbulence  created 
between  the  jet  derived  from  the  coagulant-injection  needle  and  the  flow  of  the  water 
through  the  pipe  and  two  90-deg  elbows.  The  two  filters  consisted  of  sand  columns,  5  and 
15  cm  deep,  with  an  effective  grain  size  of  0.62  mm.  The  hydraulic  loading  was  held 
constant  at  5  m/h  (2.0  gal/[min  •  ft  ]).  3ar  tests  were  performed  simultaneously  with  the 
filter  studies  us  shown  in  Fig.  16.  The  optimal  jar-test  coagulant  dosage  closely  matched 
the  dosage  giving  the  best  filter  performance,  but  the  filter  was  much  more  tolerant  to 
off  dosages. 


36 


volume  s 


Dot*  (mg  polymer/L  fttd  w*t*r)  X10* 

Figure  16.  Comparison  of  jar-test  results  with  the  passage  of  kaolinlte  turbidity  through 
rapid  sand  filters,  where  C/CQ  is  the  ratio  of  the  turbidity  of  the  treated  water  to  that  of 
the  untreated  water  (kaolinlte  *  20  mg/L,  cationic  polymer,  filtration  rate  ■  5  m/h). 

i  j 

Adapted  from  Adin  and  Rebhun. 


Filter  run  time  (min) 


Figure  17.  Effect  of  polyelectrolyte  dose  on  the  passage  of  0.1-  pm  latex  beads  through  a 
deep-bed  filter.  (Bead  concentration  =  9.7  mg/L,  filtration  rate  a  2  gal/[min  •  ft4'].) 
Adapted  from  Habibian  and  O'Melia.^ 


37 


Volume  8 


36 

Habibian  and  O'Melia  conducted  a  study  to  investigate  the  role  of  chemical 
parameters  in  direct  filtration.  Suspensions  of  different  sizes  of  latex  beads  were 
coagulated  with  various  cationic  polymers  and  filtered  through  six  test-filter  columns 
operating  under  the  same  conditions  but  receiving  different  polymer  dosages.  No 
information  was  given  on  the  type  of  filter  media  used  or  the  kind  of  mixing  employed. 
One  filter  was  operated  as  a  control  with  no  polymer}  two  were  underdosed  relative  to  the 
results  of  jar  tests;  two  were  overdosed;  and  one  was  operated  at  the  optimum  jar-test 
dosage.  The  filtration  rate  was  kept  constant  at  2  gai/(mln  •  ft  ),  and  all  filter  media 
were  precoated  with  polymer  prior  to  the  start  of  a  filtration  run. 

The  results  of  the  experiments  for  the  0.1-pm  latex  particles  and  the  Cat-Floe 
polymer  are  shown  in  Fig.  17.  The  optimum  jar-test  coagulant  dosage  was  determined  to 
be  0.07  mg/L.  The  control  and  the  two  test  columns  receiving  gross  underdoses  of  the 
coagulant  (runs  1,  2,  and  3  in  Fig.  17)  gave  constant  concentrations  of  the  latex  particles 
In  the  filtered  water,  as  determined  by  turbidity  soon  after  the  start  of  the  filter  runs. 
The  passage  of  the  particles  through  the  filter  remained  constant  for  a  while  and  then 
increased  rapidly  to  nearly  100%.  The  two  filters  receiving  gross  overdoses  of  the 
coagulant  (runs  5  and  6  in  Fig.  17),  performed  poorly  almost  immediately  after  the  start 
of  the  filter  runs.  At  the  optimum  dose  as  determined  by  the  jar  tests  (run  4  In  Fig,  17), 
the  passage  of  the  particles  through  the  filter  attained  a  maximum  value  of  approximately 
40%  soon  after  the  Initiation  of  the  filter  run,  and  then  decreased  to  a  constant  minimal 
value  of  less  than  10%.  Filtration  in  this  run  was  so  effective  that  the  7.3-ft  (of  water 
head)  allowable  pressure  drop  across  the  filter  column  was  reached  in  less  than  3  h  of 
continuous  operation. 

Jar-test  results  for  the  1.099-  ym  latex  particles  using  polyethyleneimlne  coagulants 
with  molecular  weights  ranging  from  600  to  100,000,  but  with  almost  equal  charge 
densities  (as  determined  by  acid-base  titrations),  indicated  that  the  optimum  coagulant 
dosage  was  nearly  the  same  for  ail  the  polymers  regardless  of  their  molecular  weights, 
except  for  the  lowest-moleculer-weight  polymer  that  also  had  a  higher  charge  density. 
This  indicated  that  charge  neutralization  plays  a  significant  role  In  the  coagulation  of 
latex  particles  by  polyethyleneimlne.  Subsequent  electrophoretic-mobility  measurements 
Indicated  that  the  optimum  dosage  occurred  at  negative  zeta  potentials.  This  finding 
confirms  the  effect  of  charge  neutralization,  but  since  optimum  dost^e  did  not  occ.ut  at 
zero  zeta  potential  (zero  mobility),  it  also  Indicates  that  complete  charge  neutralization 

is  not  requited  for  effective  coagulation.. From  these  results  the  investigators  concluded 
that  charge  neutralization  plays  a  role  In  coagulation,  although  it  Is  not  the  sole 
mechanism.® 


38 


Volume  8 


As  stated  previously,  Yao's  theoretical  and  experimental  studies  showed  that  the 

size  of  the  suspended  particles  affects  the  removal  efficiency  of  the  filter  markedly.  In 

36 

order  to  evaluate  the  validity  of  this  prediction,  Habibian  and  O'Melia  filtered 
suspensions  of  three  sizes  of  latex  particles  (0.109,  1.099,  and  7.6  ym)  through  three 
shallow  filter  beds  0.8  in.  deep.  The  coagulant,  PEI- 1 8  (poiyethyleneimine),  was  added 
continuously  to  the  filter  Influents.  The  polymer  dosages  were  0.176,  0.032  and 
0.024  mg/L,  respectively.  The  Initial  latex-particle  concentrations  in  the  the  three  filter 
influents  were  approximately  the  same  (30,  48  and  32  mg/L,  respectively).  The 
particle-removal  efficiency  was  greatest  for  the  7.6-ym  particles.  It  was  also  quite  high 
for  the  0.109  ym  particles  for  which  the  maximum  allowed  pressure  drop  across  the  filter 
bed  (85  in.)  was  reached  before  particle  breakthrough.  Filtration  of  the  1.099- ym 
particles  was  the  least  efficient,  and  breakthrough  was  observed  at  a  filter  pressure  drop 
of  5  In.  of  water  head.  The  Investigators  concluded  that  the  trend  In  particle-removal 
efficiency  as  a  function  of  particle  size  was  In  qualitative  agreement  with  Yao’s  transport 
model. 

37 

Stump  and  Novak  Investigated  the  performance  of  various  polymers  In  a  process 
consisting  of  rapid  mechanical  mixing,  followed  by  20  min  of  flocculation,  and  then 
filtration  through  a  multimedia  filter  (anthracite,  sand,  and  garnet).  The  filtration  rate 
1  was  5  gal/(mln  •  ft2).  The  feed  water,  a  100-mg/L  kaollnlte  clay  suspension,  had  a 
turbidity  of  80  formazln  turbidity  units  (FTU).  The  optimum  coagulant  dosage  obtained  by 
the  jar  tests  provided  the  longest  filter  runs.  However,  the  jar  tests  failed  to  predict 
which  polymer  would  perform  the  best  in  the  filter  runs.  It  was  concluded  that  the  jar  test 
could  not  be  used  to  select  the  polymer,  but  It  could  be  used  to  determine  the  optimal 

coagulant  dosage  for  a  given  coagulant. 

37 

Stump  and  Novak  also  concluded  that  the  low-molecular-weight  cationic  polymers 
(molecular  weight  <10,000)  gave  poor  removals,  whereas  those  with  molecular  weights  in 
excess  of  1  x  10^  created  excessive  filter  pressure  drops.  They  recommended  a  polymer 
molecular  weight  within  the  range  of  10,000  to  200,000  for  direct  filtration.  They  also 
recommended  that  the  mean  velocity  gradient,  G,  In  the  flash  mixer  should  be  Increased 
with  the  molecular  weight  of  the  cationic  polyelectrolyte.  A  range  of  about  200  to  950  s'* 
was  recommended  for  the  polyelectrolytes  deemed  useful  for  water  coagulation. 

Electrophoretic-mobility  measurements  (zeta  potential)  and  more  recently 
partlcle-size-dlstrlbutlon  analyses  are  also  used  to  help  ascertain  the  optimum  coagulant 
dose.  According  to  O’Melia  and  Stumm/  zeta  potential  cannot  provide  an  accurate 
estimate  of  the  forces  affecting  the  particle  attachment  in  water  filtration,  and 
consequently  its  use  as  a  monitoring  tool  for  correct  coagulant  dose  Is  limited. 


Volume  8 


38 

Letterman  et  al.  performed  a  study  to  ascertain  whether  zeta  potential  could  be 
used  to  prejudge  the  performance  of  the  cationic  polyelectrolyte  Cat-Floe  T  In  the  direct 
filtration  of  water  having  a  turbidity  of  32  FTU  derived  from  a  mix  of  bentonlte-kaollnlte 
clays.  The  coagulant  was  blended  with  the  feed  water  through  a  pipe  tee  leading  Into  a 
mechanically  stirred  flocculator  containing  four  turbine  Impellers.  The  mean  velocity 
gradient  G  was  not  estimated  for  the  pipe  mixer.  The  mean  flocculation  period  was  2  to 
10  min,  depending  on  the  feed  flow,  and  the  mean  velocity  gradient  could  be  varied 
between  0  and  700  s"1  in  the  flocculator.  The  water  was  filtered  through  3-in.  i.d. 
dual-media  filter  columns  (anthracite  and  sand).  The  filter  rate  was  varied  from  2.5  to 
7.5  gal/(min  •  ft  )  during  the  course  of  the  study. 

The  polyelectroiyte  dosage  was  adjusted  in  a  filter  run  performed  at  a  filtration  rate 
of  7.5  gal/(min  •  ft  )  to  obtain  a  minimum  filter  effluent  turbidity.  This  corresponded  to 
particle  zeta  potentials  in  the  range  of  -5  to  +12  mV  in  the  suspension  leaving  the 
flocculator.  The  investigators  stated  that  these  results  were  in  good  agreement  with  the 
principal  author's  previous  studies  on  direct  filtration  of  natural  suspensions  from  Lake 

IQ 

Michigan,  using  cationic  polyelectrolytes.  In  these  former  tests  the  particle  zeta 
potentials  ranging  from  -4  to  +13  mV  corresponded  to  an  Interval  of  minimum  turbidity  in 
the  filtered  water.  It  was  concluded  that  zeta-potential  measurements  at  or  near  zero 
Indicated  the  best  coagulant  dosage. 

Yeh  and  Ghosh5  conducted  studies  with  the  objective  of  developing  methods  to 
select  polymers  for  direct  filtration.  They  Investigated  the  influence  of  polymer 
molecular  weight  and  charge  density  on  filtration  efficiency,  as  well  as  the  effects  of 
mixing  energy  on  polymer-particle  Interactions,  by  using  zeta-potentlal  measurements,  jar 
tests,  colloid  titration,  and  particle-size-distribution  analysis.  Cationic  polymers 
(molecular  weights  ranging  from  1200  to  5x10^)  were  Investigated,  Including  Cat-Floe  T 
(molecular  weight  =  1x10  ).  Studies  were  performed  both  in  batch-  and  continuous-flow 
modes.  The  batch  tests  consisted  of  the  jar  tests  and  zeta-potentlal  measurements.  In 
the  continuous-flow  filtration  studies  a  3.7-L  mixing  chamber  with  a  variable-speed  mixer 
was  used  ahead  of  a  2.34-crn  (1-in.)  i.d.  filter.  The  filter  contained  silica  sand  with  an 
effective  size  of  1.0  mm  and  a  uniformity  coefficient  of  1.2,  packed  to  a  depth  of 
15.2  cm. 

It  was  found  that  the  coagulant  dosage  that  gave  the  minimum  residual  turbidity  In 

the  jar  tests  also  corresponded  to  a  zero  zeta  potential.  The  partlcie-size-distrlbutlon 
analysis  conducted  in  the  continuous-flow  studies,  on  the  other  hand,  -  indicated  a 
substantially  higher  optimum  polymer  dose  when  measured  after  3  min  of  rapid  mixing. 
The  filtration  runs  indicated  that  the  coagulant  dose  obtained  from  the 


40 


Volume  8 


particle-size-distribution  analysis,  rather  than  that  given  by  the  jar  tests,  gave  the  best 
filter  performance.  The  investigators  concluded  that  particle-size-distributlon-analysis 
was  the  preferable  method  to  determine  the  optimum  polymer  dosage  for  direct  filtration. 

Yeh  and  Ghosh^  concluded  that  low-  to  medium-molecular-weight  cationic  polymers 

(molecular  weights  from  10,000  to  100,000)  perform  best  in  direct  filtration,  which  is  in 

38 

agreement  with  the  results  of  Letterman  et  ah  discussed  previously.  The  investigators 
recommended  that  rapid  mixing  should  not  be  continued  for  prolonged  periods  of  time 
because  it  might  cause  fioc  breakup,  especially  when  hlgh-molecular-welght  polymers  are 
used.  They  also  concluded  that  for  most  direct  filtration  operations,  slow  flocculation 
following  rapid  mixing  is  not  necessary,  especially  if  the  suspended-solids  concentration  in 
the  raw  water  is  30  mg/L  or  higher. 

9 

Ghosh  et  al.  Investigated  the  relationships  between  polymer  molecular  weight, 
charge  density  and  dosage,  mixing  conditions  and  the  flocculated  partlcle-size- 
dlstributions  in  batch  reactors.  Three  groups  of  commercial  cationic  polyelectrolytes 
were  tested,  including  the  Cat-Floe  group.  The  study  results  indicated  the  following) 
(1)  both  charge  neutralization  and  bridging  play  a  role  in  coagulation  with 
polyelectrolytes,  (2)  the  optimum  polymer  dosage  Is  independent  of  polymer  molecular 
weight  In  most  cases,  and  (3)  a  strong  correlation  exists  between  the  optimum  polymer 
dosage  and  its  charge  density.  It  was  observed  that  the  optimum  coagulant  dosage 
decreased  as  a  power  function  of  the  number  of  cationic  charges  per  molecule,  and  the 
power  was  close  to  unity  for  the  type  of  suspensions  coagulated  In  their  study  (silica  and 
bentonite  clay). 

o 

Ghosh  et  al.  also  compared  particle-size  distribution  and  zeta  potentials  subsequent 
to  rapid  mixing  and  flocculation  in  a  mechanically  stirred  batch  reactor.  They  found  that 
the  polymer  dosage  yielding  the  largest  mean  size  of  the  particles  corresponded 
approximately  to  a  zero  zeta  potential,  regardless  of  how  the  mean  size  was  computed. 
These  findings  tend  to  support  the  concept  that  charge  neutralization  plays  an  important 
role  in  polyelectrolyte  coagulation,  as  does  the  applicability  of  zeta-potential 
measurements  for  the  determination  of  the  optimum  polymer  dose. 
Particle-size-distribution  analyses  also  Indicated  that  a  significant  amount  of  coagulation 
occurred  during  the  first  phase  of  rapid  or  flash  mixing.  The  optimum  mean  velocity 
gradient  G  was  800  s"*  for  most  of  the  cationic  poiyelectrolytes  tested,  including  the 
Cat-Floe  group.  Increasing  G  beyond  800  s"1  failed  to  improve  the  coagulation  rate. 

The  experimental  data  of  Ghosh  et  al.  also  Indicated  that  the  charge  density  of  the 

9 

Cat-Floe  group  of  polymers  was  independent  of  the  pH  of  the  suspension. 


41 


Volume  8 


GENERAL  BACKGROUND  PAPERS 
39 

Robeck  etai.  observed  removals  of  attenuated  poliovirus  Type  1  (Mahoney  strain) 
ranging  from  1  to  50%  when  uncoagulated  water  was  filtered  through  a  dual-media  filter 
with  filtration  rates  of  2  to  6  gal/(min  •  ft  ).  Properly  coagulated  water  with  no  in-line 
flocculation  chambers  and  no  settling  tank  gave  rise  to  90-99%  removals  of  the  virus,  with 
most  removals  being  >98%.  Two  kinds  of  raw  water  were  used:  (1)  a  blend  of  hard  ground 
water  with  demineralized  water  and  (2)  water  from  the  Little  Miami  River,  OH.  The 
raw-water  temperatures  varied  from  17  to  19°C;  the  pH  varied  from  8.1  to  7.7,  and  the 
alkalinity  ranged  from  100  to  200  ppm.  The  feed-water  virus  concentration  was 
maintained  in  the  region  of  10,000  pfu/mL.* 

The  dual-media  filter  contained  16  in.  of  anthracite  coal  and  8  in.  of  Muscatine 

sand,  and  the  raw  water  was  coagulated  with  10  ppm  of  alum  when  the  raw-water 

+ 

turbidity  was  10  3TU  or  less.  The  alum  dose  was  increased  for  higher  raw-water 
turbidities  (about  40  JTU).  These  higher  turbidities  and  alum  doses  tended  to  shorten  the 
filter  runs.  Combinations  of  alum  and  polyelectroj  ;tes  were  tested  In  a  few  of  the  runs. 
One  run  lasted  52  h  with  a  pressure  drop  across  the  filter  reaching  7  ft  of  water  head. 

From  their  experiments,  Robeck  et  al.  concluded  that  (1)  an  increase  In  filtered 
water  turbidity  of  less  than  0.5  JTU,  indicating  a  floe  breakthrough,  was  usually 
accompanied  by  a  breakthrough  of  virus,  (2)  a  virus  penetration  of  the  filter  could  occur 
without  a  turbidity  increase,  even  though  the  turbidity  might  be  as  low  as  0.5  3TU, 
(3)  poiyelectrolyte  doses  as  low  as  0.05  mg/L  helped  to  increase  the  floe  strength  and 
prevent  virus  breakthrough,  and  (4)  more  than  98%  of  the  viruses  could  be  removed  in  the 
dual-filter  media  at  filtration  rates  of  2  to  6  gal/(rnin  •  ft  )  if  a  low  but  well-mixed  dose 
of  alum  was  fed  just  ahead  of  the  filters. 

A  task  group  of  the  American  Water  Works  Association  (AWWA)^®  collected  and 
analyzed  data  available  at  the  time  (1980)  relative  to  the  performance  of  direct-filtration 
water-treatment  plants.  The  objectives  of  the  study  were  the  following:  (1)  to  obtain 
data  on  direct  filtration,  (2)  to  analyze  the  results  and  identify  problems,  (3)  to  suggest 
feasible  solutions  to  the  problems  identified,  and  (4)  to  recommend  areas  for  further 
research  and  development.  Multimedia  filters  were  installed  in  a  few  of  the  plants,  but 
most  of  the  plants  studied  included  just  dual-media  filters.  Turbidity,  color,  algae  types 


*  pfu  =  plaque-forming  unit. 

^  JTU  =  Jackson  turbidity  unit. 


4  2 


Volume  8 


and  concentration,  and  water  temperature  were  judged  to  be  the  most  important 
feed-water  quality  parameters.  Filter  media,  pressure  drop,  and  filtration  rate,  as  well  as 
the  mode  of  operation  and  type  and  dosage  of  coagulants,  were  considered  to  be  the  most 
Important  operating  parameters.  From  the  results  obtained  in  this  study,  it  was  difficult 
to  distinguish  trends  in  the  operating  parameters  that  would  point  to  an  improvement  in 
the  process  performance. 

Problems  identified  with  the  raw-water  quality  included  the  following:  (l)a  color 
greater  than  about  30  to  40  Hazen  units  tended  to  move  through  the  filter  beds; 
(2)  turbidities  greater  than  about  15  FTU  wore  poorly  removed;  (3)  certain  types  of 
diatoms  tended  to  clog  the  filters;  (4)  coagulation  became  slow  at  low  temperatures;  and 
(5)  coagulant  performance  could  be  poor  at  the  elevated  pH  values  associated  with  algae 
blooms.  Other  problems  were  related  to  the  presence  of  iron  and  manganese.  The  AWWA 
committee  concluded  that  problems  with  direct  filtration  may  be  encountered  with  raw 
waters  containing  more  than  40  Hazen  units  of  color,  5  FTU  of  turbidity,  2000  asu/mL 
[(one  areal  standard  unit  (asu)  equals  400  ym^)]  of  algae,  0.3  mg/L  of  iron,  and  0.05  mg/L 
of  manganese.  They  also  recommended  that  water-flow  rates  should  be  decreased  at  low 
temperatures  in  order  to  permit  the  completion  of  the  coagulation  process;  otherwise, 
post-flocculation  may  occur  in  the  filtered  water. 

The  daia  related  to  the  types  and  doses  of  coagulants  used  in  the  direct  filtration 
plants  surveyed  varied  widely,  with  no  operational  problems  readily  identified  with  the 
single  exception  of  backwashing.  It  was  observed  that  the  use  of  polyelectrolytes  as 
coagulants  caused  stickiness  in  the  filter  media,  which  made  the  cleaning  of  the  filter  bed 
with  backwashing  difficult.  The  formation  of  mudballs  on  the  surface  of  both  the  dual- 
and  multimedia  filters  was  also  noted. 

Trends  were  not  readily  apparent  in  the  information  collected  on  flash-mixing  and 
in-line  flocculation.  Despite  these  inconclusive  results,  the  Committee  recommended  that 
in-line  mixing  of  the  coagulants  with  the  water  should  be  favored  over  mechanical  mixing. 

The  size  of  the  filter  media  used  proved  to  be  a  significant  parameter.  Anthracite 
media  smaller  than  0.8  mm  decreased  the  length  of  the  filter  runs  considerably  and  were 
subject  to  algae  binding.  Based  on  this  information,  the  Committee  recommended  an 
effective  size  uf  1.1  mm  with  a  low  uniformity  coefficient  for  the  anthracite  layer.  The 
depth  of  this  layer  should  be  in  the  range  of  37.5  to  90  cm  (15  to  36  in.). 

It  was  found  that  the  sand  layer  in  sand-anthracite  dual-media  beds  controlled  the 
filtered  water  quality.  The  depth  of  the  sand  bed  should  be  about  20  to  .30  cm  (8  to 
12  in.),  and  the  effective  size  of  the  media  of  0.45  mm,  commonly  used 


43 


Volume  S 


in  practice,  appeared  adequate.  A  third  layer  of  garnet  sand  (effective  size  of  0.2  to 
0.3  mm)  was  also  used  in  some  of  the  plants  surveyed.  It  appeared  that  those  plants  could 
successfully  treat  waters  containing  turbidities  up  to  500  FTU,  and  color  up  to  1000 
Hazen  units,  at  the  expense  of  shorter  filter  runs. 

The  filtration  rates  ranged  f*-om  about  1  to  6  gal/(min  •  ft  )  in  the  plants  surveyed. 
The  AWWA  committee  concluded  that  the  higher  rates  could  be  applied  with  careful  use 
of  mixing  energy  and  coagulant  aids,  and  recommended  a  filtration  rate  in  the  region  of  4 
to  6  gal/(min  •  ft  ).  The  only  microbiological  data  considered  by  the  Committee  were 
those  reported  by  Robeck  et  ai.^  in  1962?  they  concluded  that  waters  derived  from  the 
direct-filtration  process  must  be  disinfected.  They  also  pointed  out  the  increased  need 
for  a  good  water-quaiity-monitoring  program  because  the  direct-filtration  process  utilizes 
little  detention  time  for  coagulation  ahead  of  the  filters. 

A  review  of  the  advantages  and  disadvantages  of  direct  filtration  was  also  prepared 
by  Trussell  et  al.^1  They  decided  that  the  quantity  and  physicochemical  properties  of  the 
particulate  matter  determined  the  success  or  failure  of  direct-filtration  systems.  Other 
important  factors  considered  included  bacterial  quality,  tastes,  odors,  and  color  of  the 
raw  water.  Direct  filtration  should  be  viewed  with  extreme  caution  when  the  raw  water 
contains  high  levels  of  coliform  bacteria  because  it  does  not  provide  as  many  barriers  of 
protection  as  conventional  filtration  systems.  Moreover,  heavy  dosing  of  powdered 
activated  carbon  to  remove  tastes  and  odors  causes  significant  decreases  In  filtered-water 
output.  Other  observations  made  by  Trussell  and  co-workers  included  the  increase  in 
performance  resulting  from  the  installation  of  in-line  flocculation,  and  the  significant 
increase  in  filter  pressure  drop  caused  by  a  bottom  layer  of  garnet  without  a 
compensating  increase  in  filtered-water  quality.  A  very  coarse  top  layer  composed  of 
pumice  or  plastic  media  was  also  recommended. 

The  East  Bay  Municipal  Utility  District^ L  (EBMUD)  headquartered  in  Oakland,  CA, 
investigated  the  replacement  of  alum  as  a  coagulant  with  cationic  polymers  in  three 
direct-filtration  water-treatment  plants.  Six  cationic  polymers,  four  of  which  were 
polymers  of  diallyldimethyl  ammonium  chloride  and  two  of  epichlorohydrindimethylamine, 
were  tested.  The  coagulated  water  was  filtered  through  dual-media  filters  (3  in.  of 
0.9-mm  effective  size  anthracite  and  27  in.  of  0.5-mm  effective-size  sand,  both  layers 
with  a  uniformity  coefficient  of  1.4).  The  filtration  rates  utilized  ranged  from 
2c,0  to  2.5  gal/(min  •  ft  )  with  alum  coagulation  and  from  3.5  to  4.0  gal/(min  •  ft  )  with 
clay-polymer  coagulation  in  two  -  of  the  treatment  plants,  -and  from 


44 


Volume  S 


2  2 
2.8  to  3.3  gal/(min  •  ft  )  with  alum  coagulation  and  from  4.0  to  4.5  gal/(min  •  ft  )  with 

clay-polymer  coagulation  in  the  third  plant. 

It  was  found  that  the  removals  of  turbidity  achieved  with  alum  and  the 

polyelectrolytes  were  virtually  identical  when  the  raw-water  turbidity  exceeded  about  3.5 

nephelometric  turbidity  units  (NTU).  The  alum  performed  considerably  better  than  the 

polyelectrolytes  for  raw-water  turbidities  in  the  range  of  2  to  3  NTU.  The 

polyelectrolytes  gave  acceptable  turbidity  removals  when  artificial  turbidity  (clay)  was 

added  to  the  low-turbidity  feed  waters.  In  this  case  the  typical  filter-run  length  increased 

from  23  h  with  alum  coagulation  to  47  h  with  clay-polymer  coagulation. 

The  EBMUD  Investigators  emphasized  the  increase  in  the  potential  for  the 

unreacted  polymer  to  pass  through  the  filter  if  the  clay  dosage  was  low  relative  to  the 

polymer  dosage.  Quantitative  data  could  not  be  obtained  because  the  polyelectrolyte- 

concentration  detectability  limits  of  the  analytical  methods  used  were  too  high. 

Nevertheless,  those  investigators  recommended  the  monitoring  of  polymer  residuals  in  the 

treated  water  for  future  applications.  It  was  also  noted  that  the  coagulative  properties  of 

the  polyelectrolytes  tested  tended  to  change  with  age  (storage).  Changes  in  chlorine 

demand  and  the  growth  of  microorganisms  were  also  noted. 

High  alum  residuals  in  filtered  water,  inadequate  plankton  removals,  and  powdered 

activated  carbon  breakthrough  were  the  main  operational  problems  encountered  in  1978  in 

Q 

the  2  x  10  -gai/d-capacity  Alfred  Merrit  Smith  Water  Treatment  Facility  of  the  Southern 

43 

Nevada  Water  System.  Raw  water  derived  from  Lake  Mead  was  filtered  at  a  constant 

2 

rate  of  5  gal/(min  •  ft  )  through  dual-media  filters  (20-in.  depth  of  0.6-  to  0.7-mm 

effective  size  anthracite  and  10  in.  of  0.45-mm  effective  size  sand).  Alum  and,  when 

necessary,  powdered  activated  carbon  were  mixed  with  the  raw  water  in  mechanical  flash 

mixers  without  in-line  flocculation  or  sedimentation. 

Pilot-plant  tests  demonstrated  that  the  operational  problems  cited  previously  could 

be  ascribed  to  insufficient  alum  flocculation  times.  The  carryover  of  alum  was  reduced 

significantly,  and  activated  carbon  doses  as  high  as  20  mg/L  could  be  used  for  taste  and 

43 

odor  control  when  the  water  was  preflocculated  for  a  period  of  15  to  20  min. 

Additional  pilot-plant  tests  were  then  performed  by  a  consulting  engineer.  Both 
direct  filtration  with  and  without  flocculation  and  conventional  filtration  were 
investigated,  using  the  filter  media  installed  in  the  existing  plant.  A  few  runs  were  also 
made  utilizing  a  larger-grain-size  anthracite  layer  with  an  effective  grain  diameter  of 
0.95  mm  and  a  uniformity  coefficient  of  i.5.  It  was  demonstrated  that  the  conventional 
treatment  gave  the  best  filter  performance,  but  the  improvement  over  the  other  schemes 
did  not  warrant  the  additional  expense  of  installing  sedimentation  basins  for  this 


45 


Volume  S 


particular  lake  water.  Direct  filtration  with  30  min  of  preliminary  flocculation  yielded 
longer  filter  runs}  more  consistent  performance;  better  turbidity  and  plankton  removals; 
more  even  distribution  of  pressure  drop  through  the  filter  bed;  elimination  of 
activated-carbon-particle  breakthrough  at  high  filtration  rates;  and  less  coagulant  usage 

than  direct  filtration  without  flocculation.  These  results  confirmed  the  findings  of  the 

.  43 

preliminary  pilot-plant  studies. 

The  plankton  problems  merit  further  discussion.  It  was  found  that  the  species  or 
type  of  plankton  was  much  more  important  than  the  total  plankton  count.  Specifically, 
relatively  large  numbers  of  anabaena  and  anacystis  species  of  bluegreen  algae  decreased 
the  filter  performance  significantly.  Finally,  the  results  of  these  studies  demonstrated 
that  a  0.95-mm-effective-size  anthracite  layer  was  preferable  to  the  0.6-  to  0.7-mm 
anthracite  used  in  the  plant  filters.  The  filter  with  the  larger-size  anthracite  gave  longer 
filter  runs  with  turbidity  and  plankton  removals  comparable  to  those  of  the 
smaller-anthracite-slze  filter. 

32 

Research  reported  by  the  University  of  Washington  and  the  U.S.  EPA  shows  a 
relationship  between  the  turbidity  of  the  filtrate  and  the  removal  of  cyst-sized  particles 
(5  to  13  ym).  Filtrate-turbidity  levels  must  be  consistently  lower  as  desk  od  removals  of 
cyst-sized  particles  increase.  It  was  found  that  for  removals  >99.0%,  an  overwhelming 
majority  of  the  filtered  water  samples  (88%)  had  turbidities  less  than  0.10  NTU.  For 
removals  of  90.1  to  95.0%  only  about  a  third  of  the  samples  had  turbidities  of  less  than 
0.10  NTU. 

33 

Logsdon  et  a).  Investigated  the  removal  of  cyst  models  and  the  cysts  of  G.  muris 
in  direct  filtration  with  a  dual-media  filter  (46  cm  of  anthracite  with  an  effective  size  of 
1.27  mm  and  15  cm  of  sand  with  an  effective  size  of  0.36  mm).  Three  sets  of 
experiments  were  run,  each  with  a  different  type  of  coagulant.  Alum,  or  alum  plus  a 
cationic  or  .nonionic  polymer,  was  blended  with  the  feed  water  as  It  was  fed  through  three 
in-line  static  mixers  in  series.  (No  coagulant  was  added  to  the  water  in  some  of  the  initial 
test  runs.)  The  solution  was  then  flocculated  for  about  20  to  30  min  in  a  cascade  of  CSTR 
reactors.  The  filtration  rate  on  the  dual-media  filter  was  varied  from  about  2.7  to 
7.4  mm/s  [4  to  11  gal/(min  •  ft  )]  during  the  course  of  the  experiments.  The  most 
commonly  used  filtration  rate  was  2.7  mm/s  [4  gal/(min  •  ft*)]. 

The  water  used  in  the  experiments  was  obtained  from  a  gravel  pit.  This  water  was 
characterized  by  low  turbidity,  and  experiments  were  conducted  at  a  water  temperature 
of  about  20#C  and  a  pH  of  8.2.  The  cysts  of  G.  muris  or  cyst  models  (radioactive  beads) 


Volume  8 


were  added  continuously  to  the  in-line  flash  mixers  to  yield  initial  concentrations  ranging 
from  about  470  to  190,000  cysts  per  liter.  The  coagulants,  when  used,  were  added  in 
amounts  sufficient  to  maintain  a  filtered-water  turbidity  of  less  than  1  NTU. 

The  removal  of  the  cysts  ranged  from  59  to  94%  when  no  coagulant  was  added  to  the 
raw  water.  The  removals  increased  to  99.5%  or  more  during  periods  of  stable  operation, 
when  alum  was  added  in  quantities  sufficient  to  yield  turbidities  of  <*  0.3  NTU  in  the 
filtered  water,  and  the  initial  feed-water  cyst  count  was  >10,000/L.  Three  out  of  20 
experimental  runs  yielded  filtered-water  cyst  concentrations  >  100/L  at  a  filtration  rate 
of  2.7  mm/s  [4  gal/(min  •  ft2)]. 

It  was  concluded  that  "a  filter  treating  water  dosed  with  an  adequate  amount  of 
coagulant  and  operated  in  a  manner  that  prevents  filtered-water  turbidity  Increases 
should  remove  a  very  substantial  portion  of  the  Glardia  cysts."  It  was  also  pointed  out 
by  the  investigators  that  the  continuous  passage  of  a  small  number  of  cysts  through  a 
dual-media  filter  operated  in  the  direct  filtration  mode  cannot  be  ruled  out. 

Some  other  observations  of  interest  were  also  made  by  Logsdon  et  aL  Increasing 
the  filtration  rate  by  30,  100,  or  150%  in  less  than  10  s  increased  the  passage  of  turbidity 
and  cysts  through  the  filter.  The  simultaneous  application  of  alum  and  a  nonionic  polymer 
helped  to  mitigate  this  effect  at  20#C,  but  not  at  10#C.  Sudden  surges  in  flow  rate  also 
dislodged  the  turbidity  and  cysts  stored  »n  the  filter.  For  example,  increasing  the 
filtration  rate  abruptly  from  1 1  to  27  m/h  then  back  to  11  m/h  [4.5  to  11  gal/(min  •  ft2), 
then  to  4.5  gal/(min  •  ft2)]  for  a  period  of  2  min  increased  the  filtered-water  turbidity 
about  4-fold  and  the  cyst  concentration  about  25-foid.  Finally,  initial  tilter-rlpening 
periods  (allowing  equipment  to  operate  for  a  while  to  Improve  performance)  lasting  about 
0.5  h  were  evidenced  in  the  experiments.  The  filtered-water  cyst  concentrations  were 
about  10  to  25  times  greater  than  those  during  stable  operation  (immediate  performance 
monitoring)  during  these  initial  periods. 

Rebhun  et  al.^  determined  the  removal  of  organic  colloids  by  direct  filtration.  The 
study  apparatus  used  (for  feeding,  dosing,  flowmeter,  and  flow-control  arrangements)  was 
similar  to  that  used  in  an  earlier  work  discussed  previously  in  the  section  on  determination 
of  the  best  type  and  optimum  dosage  of  coagulant.^  Two  filter  columns,  one  with  a  5-cm 
i.d.  and  15-cm  media  depth,  and  the  other  with  a  4-cm  i.d.  and  100-cm  media  depth,  were 
used.  Both  columns  had  the  same  dual  media»  A  coarse  quartz  sand  with  a  1.21-mm 
effective  size  and  1.17  uniformity  coefficient  on  top  of  a  medium-grade  quartz  sand  with 
a  0.62-mm  effective  size  and  1.21  uniformity  coefficient.  The  media  porosities  were  0.4 
and  0.37,  respectively. 


47 


Volume  8 


The  feed  water  contained  10  mg/L  of  a  humic  acid  dispersion,  and  the  coagulants 
tested  were  alum,  polyelectrolytes,  and  alum  plus  polyelectrolytes.  The  mean  residence 
time  given  for  coagulation  was  only  2  to  3  min.  It  was  found  that  effective  filtration 
could  be  achieved  with  a  combination  of  alum  and  a  cationic  polymer  if  the  doses  of  the 
coagulants  were  carefully  selected.  Alum  or  cationic  polymers  (Magnafloc  R-140  and 
R- 139,  Cat-Floe  C),  added  singly,  failed  to  give  acceptable  results. 

McCormick  and  Klng^  determined  the  removal  of  turbidity,  color,  algae,  and  total 
coliform  bacteria  from  various  surface  waters  located  in  the  Commonwealth  of  Virginia 
with  a  direct-filtration  pilot  plant.  The  results  of  this  study  are  of  special  interest  here 
because  the  modes  of  operation,  filtration  rates,  and  type  of  coagulants  used  were  often 
similar  to  those  used  in  the  600-gph  ROWPU. 

Alum  and/or  cationic  polyelectrolytes  were  blended  with  the  raw  water  in  a 
mechanical  flash  mixer.  The  blended  water  could  then  be  Introduced  Into  a  mechanically 
stirred  flocculation  basin  or  bypassed  to  the  filter-pump  well.  (The  flocculation  basin  was 
not  utilized  in  most  of  the  experiments.)  The  water  was  then  pumped  to  one  of  the  three 
9.2-cm-i.d.  gravity-filter  columns  that  contained  the  types  of  filter  media  listed  in 
Table  2. 

The  mean  residence  times  were  3,  28,  and  <  10  min  in  the  flash  mixer,  flocculation 

basin,  and  filter-pump  well,  respectively.  The  mean  velocity  gradients  utilized  in  the 

flash  mixer  were  1290  s"1  for  alum  coagulation  and  1650  s”1  for  polyelectrolyte 

coagulation.  The  mean  velocity  gradient  was  varied  between  approximately  20  and  63  s“* 

in  the  flocculation  basin  when  the  floe  basin  was  used.  The  water  was  usually  filtered  at  a 

2 

rate  of  12.6  m/h  [5  gal/(min  •  ft  )]  and  backwashed  at  a  rate  of  50.4  m/h 
2 

[20  gal/(min  •  ft  )]  for  7  min.  No  mudball  formation  was  observed.  A  successful  filter 
run  was  defined  as  that  in  which  the  filtered-water  turbidity  was  always  <  1  NTU  for  a 
period  of  at  least  8  h. 

The  quality  of  the  surface  waters  tested  varied  considerably.  For  example,  the 
raw-water  turbidity  ranged  from  a  minimum  of  1.5  to  a  maximum  of  24  NTU.  The 
observed  ranges  in  the  other  raw-water  quality  constituents  measured  were  the  following! 
0  to  42  APHA*  color  units,  185  to  20,000  algae  clumps/ml,  (not  always  determined),  1.5  to 
12.5°C  water  temperature,  and  0  to  2300  coliform  bacteria/ 100  mL.  The  optimal  doses  of 
the  coagulants  ranged  from  5  to  40  mg/L  for  alum,  and  from  l  to  5  mg/L  for  Cat-Floe  Tl. 
Other  polyeiectrolytes  investigated  were  Cat-Floe  21,  Cat-Floe  T,  and  Magnafloc  572  C. 

The  polyeiectrolytes  were  at  times  mixed  together  with  the  alum  in  the  flash  mixer. 


*  APHA  =  American  Public  Health  Association. 


48 


Volume  8 


Table  2.  Types  of  filter  media  used  in  the  direct-filtration  studies  of  Virginia  surface 

waters.® 


Filter 

number 

Media 

type 

Media 
depth  (cm) 

Effective 
size  (mm) 

Uniformity 

coefficient 

l 

coal 

25 

1.0  to  1.1 

<1.7 

sand 

30 

0.42  to  0.49 

1.3  to  1.5 

garnet 

23 

0.21  to  0.25 

1.6  to  2.1 

2 

coal 

‘  51 

1.3 

1.35 

sand 

25 

45 

1.4 

3 

coal 

51 

1.7 

l.l 

sand 

25 

0.45 

1.4 

— 
From  McCormick  and  King. 


Firm  conclusions  could  not  be  drawn  for  all  of  the  process  variables  investigated  by 
45 

McCormick  and  King,  but  the  following  proved  to  be  readily  apparenti 

•  The  raw-water  turbidity  was  10  NTU  In  39  experimental  runs.  These  waters 
were  filtered  successfully  80%  of  the  time.  The  failures  were  ascribed  to 
premature  breakthrough  of  turbidity  (13%)  and  failure  to  maintain  a  constant 
coagulant  dosage  (3%). 

•  The  raw-water  turbidity  exceeded  10  NTU  In  1 4  experimental  runs.  The  water 
was  filtered  successfully  In  only  three  of  those  runs,  and  the  results  for  the 
three  successful  runs  are  summarized  In  Table  3.  Failures  were  ascribed  to 
premature  breakthrough  of  turbidity  or  a  poor-quality  effluent  at  all  times. 

•  The  polyelectrolytes  tested  commonly  performed  better  than  the  alum, 
although  it  was  noted  that  the  alum  was  superior  for  color  removal.  A  mix  of 

alum  and  a  cationic  polymer  worked  the  best  of  all.  Premature  turbidity 

#  .  .. 

breakthrough  occurred  in  43%  of  all  the  alum  runs,  30%  of  the  polymer  runs, 


49 


Volume  8 


and  28%  of  the  runs  in  which  both  were  applied  to  the  raw  water.  It  also 
appeared  that  the  polyelectrolytes  would  have  been  superior  to  alum  in  terms  of 
the  projected  length  of  a  filter  run. 

Large  algae  concentrations  (range  of  7000  to  20,000  algae  clumps/mL)  tended 
to  clog  all  of  the  filters  tested.  The  high  pH  associated  with  algae  blooms  also 
increased  the  amount  of  alum  that  had  to  be  added  to  the  raw  water.  This,  in 
turn,  resulted  in  a  pressure  drop  across  the  filters. 

A  filtered-water  turbidity  of  less  than  0.10  NTU  also  gave  total  coliform 
bacteria  counts  below  the  detectable  limits  of  the  analytical  method  used  in  the 
Investigation. 

It  was  determined  that  the  effective  size  of  the  anthracite  layer  used  in  Filter 
No.  3,  1.7  mm  (see  Table  2),  was  too  large.  Early  breakthrough  occurred  in 
most  of  the  runs  performed  with  that  filter.  The  rate  of  Increase  of 
filter-pressure  drop  was  significantly  greater  In  the  multimedia  bed  (Filter 
No.  1)  than  the  dual-media  bed  (Filter  No.  2)  In  which  the  effective  size  of  the 
anthracite  was  1.3  mm.  It  was  concluded  that  the  media  contained  In  Filter 
No.  2  represented  the  most  effective  combination  of  media  sizes  and  bed 
depths. 

The  effects  of  other  parameters  Investigated  could  not  be  judged  from  the 
relatively  limited  amount  of  data  taken  In  the  study.  These  Included  water 
temperature  (which  was  always  lathei  low,  1.5  to  12.5°C),  an  increase  in 
filtration  rate  from  12.6  m/h  to  20.16  m/h  (5  to  8  gal/[mln  •  ft^]),  and  direct 
filtration  with  and  without  in-line  flocculation.  It  did  appear  that  preliminary 
flocculation  might  have  Improved  the  performance  of  the  multimedia  filter 
because  the  25  cm  (10  in.)  of  anthracite  coal  placed  in  that  filter  was  not 
sufficient  to  provide  the  additional  detention  time  required  for  good  bed 
flocculation. 

The  investigators  concluded  that  the  raw  water  should  have  a  turbidity  of  less 
than  10  NTU,  a  color  of  less  than  15  APHA  units,  and  an  algae  concentration  of 
less  than  1000  algae  clumps/mL  in  order  to  be  treated  reliably  with  direct 
filtration  at  a  filtration  rate  of  12.6  m/h  [5  gal/(mln  •  ft^)]. 


Volume  8 


Table  3.  Details  of  the  successful  filter  runs  with  Cat-Floe  T!  coagulant  for  raw-water 
turbidities  exceeding  10  NTU.a 


Average  turbidity  (NTU) 

Raw  Filtered  Filter  Temp  Coagulant 

water  water  no.15  Modec  (#C)  dose  (mg/L) 


10.4 

0.40 

3 

R 

5.7 

3.7 

1 6.5 

0.45 

2 

R 

7 

3.5 

12.0 

0.9 

2 

RF 

4 

5.0 

a - ZT5 — - 

Adapted  from  McCormick  and  King. 
b  See  Table  2. 

c  R  stands  for  rapid  mixing  followed  by  direct  filtration  and  RF  stands  for  rapid  mixing 
and  flocculation  followed  by  direct  filtration. 


Hand  et  al.  conducted  a  bench-scale  study  to  compare  the  suitability  of  direct 
filtration  versus  conventional  filtration  for  the  upgrading  and  extension  of  the 
pretreatment  facilities  of  the  Sweetwater  Authority  Water  Treatment  Plant  in  San  Diego, 
CA.  The  existing  plant  operating  In  the  "in-line"  filtration  mode  (see  Fig.  1)  had 
experienced  short  filter  runs,  and  product-water  turbidities  exceeding  the  0,3-NTU 
turbidity  standard  established  by  the  State  of  California  Department  of  Health  Services. 

The  raw-water  quality  appeared  to  be  good  enough  to  be  filtered  successfully  with 
the  direct-filtration  mode.  The  raw-water  turbidity  was  less  than  5.3  NTU  50%  of  the 
time,  and  less  than  14  NTU  90%  of  the  time.  However,  It  was  decided  from  the 
bench-scale  batch  tests  that  changing  from  direct  filtration  to  conventional  treatment, 
Including  flocculation  and  sedimentation,  would  serve  to  improve  the  performance  of  the 
filtration  plant  significantly. 

Cleasby  et  al/*7  investigated  the  ability  of  a  10-cm-i.d.  dual-media  filter  column 
(40  cm  of  l.34-mm  effective-size  anthracite  with  a  uniformity  coefficient  of  1.18,  and 
30  cm  of  0.43-mm  effective-size  sand  with  a  uniformity  coefficient  of  1.53)  to  remove 
particles,  using  particle  counts,  turbidity,  total  collform  bacteria,  and  standard  plate 
count  as  Indicators,  The  Influent  water  was  taken  from  a  gravel  pit,  which  normally 
would  have  supplied  high-quality  water,  but  during  the  testing  period  was  subject  to 
several  unanticipated  algal  blooms.  Water  pH  was  adjusted  to  6.8  to  7.8,. during  the 


51 


Volume  8 


high-pH  periods  of  algae  blooms.  Alum  or  Cat-Floe  T  was  blended  with  the  raw  water  in 
static  mixers,  and  no  flocculation  time  was  provided  other  than  that  occurring  in  the  feed 
lines.  The  coagulated  water  was  filtered  through  the  dual-media  filter  at  filtration  rates 
ranging  from  6.8  to  13.5  m/h  [2.8  to  5.5  gal/(min  •  ft2)]. 

The  raw-water  temperature  varied  considerably  during  the  period  of  testing 
(October  1981  to  November  1982)  as  shown  in  Table  4.  The  optimum  alum  dosage  ranged 
within  the  relatively  narrow  limits  of  6.1  to  12.3  mg/L  for  raw-water  turbidities  varying 
from  about  0.4  to  8  NTU.  The  optimum  Cat-Floe  T  dosage  ranged  widely  from  0.09  to 
0.84  mg/L.  The  results  listed  in  Table  4  indicate  that  in  general,  alum  coagulation  is 
superior  to  that  achieved  with  cationic  polyelectrolyte  as  far  as  effluent  quality  is 
concerned. 

Other  conclusions  of  this  study  were  as  foliowss  (1)  poorer  filtrate  is  found  at  the 
beginning  of  the  run,  affirming  the  Importance  of  filtering  to  waste  during  ripening  at  the 
start  of  a  run  for  a  period  of  time,  especially  where  Glardia  cysts  or  7-  to  12>um  particles 
are  of  concern;  (2)  removal  of  total  coliform  bacteria  was  >,86%;  (3)  water  as  cold  as  2°C 
did  not  seem  to  affect  the  removal  efficiencies;  (4)  longer  filter  cycles,  but 
poorer-quality  filtrate,  were  obtained  using  the  cationic  polymer;  and  (5)  prechlorination 
with  alum  was  essential  during  periods  of  severe  algae  blooms  (prechlorination  with  the 
polyelectrolyte  proved  Ineffective)*  Finally,  Cleasby  et  al»  recommended  that  the 
raw-water  turbidity  should  not  exceed  about  16  NTU  with  "small"  algae  populations  or 
1 1  NTU  with  "moderate"  algae  populations  for  the  cationic  polymer  to  be  effective. 

600-GPH  ROWPU  PRETREATMENT  UNITS 

Small  et  al.1  conducted  a  study  on  the  removal  of  microorganisms  by  the  600-gph 
ROWPU  pretreatment  system  without  chlorination  in  1979-80.  River  water  (Monocacy 
River,  MD)  was  first  coagulated  and  then  filtered  through  a  multimedia  pressure  filter 
(Culligan  Model  MD30  Mixed-Media  Filter)  similar  to  the  one  Installed  in  the  current 
600-gph  ROWPU.  The  filtered  water  was  then  pumped  under  pressure  through  cartridge 
filters.  The  raw  water  was  coagulated  with  a  cationic  polymer  (Cat-Floe  T)  at  an  average 
dosage  of  5.0  mg/L  (4.3  to  5.6  mg/L).  Three  filtration  rates  of  6.5,  7.6,  and 
8.7  gal/(min  •  ft  )  were  investigated  in  nine  experimental  runs  lasting  9  h  each.  These 
filtration  rates  are  equivalent  to  feed-water  flows  to  the  ROWPU  of  30,  35,  and 
40  gal/min.  The  operation  of  these  pieces’ of  equipment  was  as  described  in  the  technical 
manual  for  the  600-gph  ROWPU. ^ 


52 


Volume  8 


Table  4.  Average  removal  of  total  coliform  bacteria  and  particles  in  the  7-  to  12-ptn  size 
range  for  direct  dual-media  filtration  without  chlorination.®  (Filtration  rates  were  varied 
from  6.8  to  13.5  m/h.) 


Coagulant 

Temp 

(#c) 

Coliform  bacteria 

Particles  (7  to  12  um  In  size) 

Influent 
(number/ 
100  mL) 

Removal  after 
filter  ripening 
(*) 

Influent 
(number/ 
100  mL) 

Removal  after 
filter  ripening 
(%) 

Alum 

7  to  14 

1300 

91 

2320 

98.8 

Cat-Floe  T 

4  to  5 

8200 

96.5 

1170 

96.7 

Cat-Floe  T 

3  to  4 

1500 

89.7 

370 

87.0 

Alum 

2 

1600 

96 

2190 

99.0 

Cat-Fioc  T 

3 

640 

89 

1620 

98.0 

Alum 

6  to  7 

350 

91.3 

2860 

94.0 

Alum 

17 

90 

86 

13040 

99.0 

Cat-Floe  T 

21  to  23 

50 

86 

1350 

85.5 

Alum 

20  to  24 

550 

89 

1640 

96.5 

Cat-Fioc  T 

23 

170 

86.5 

340 

87.5 

a  From  Cleasby  et  al.^ 


The  cartridge-filter  elements  had  the  same  nominal  size  rating  as  those  used  in  the 
current  600-gph  ROWPU  (5  um)v  but  they  had  less  surface  area.  Six  instead  of  eight 
polypropylene  Fllterite  elements  were  housed  in  one  pressure  vessel. 

The  ranges  in  the  temperature,  pH,  turbidity,  TDS  concentration,  and  alkalinity  of 
the  Monocacy  River  water  are  shown  in  Table  5.  The  water  had  a  total  organic  carbon 
(TOC)  content  of  approximately  5.1  mg/L,  and  an  immediate  chlorine  demand  of  about 
1 .4  mg/L.  Cultures  of  Bacillus  Rlobigll  spores  and  poliovirus  I,  LSc  strain  were  well  mixed 
with  the  river  water  prior  to  adding  the  coagulant.  The  feed-water  concentrations  ranged 
from  220  to  15,000  colony- forming  units  (cfu)/mL  for  B.  globlgil,  14  to  8500  cfu/mL  for  jn 
situ  Escherichia  coll,  440  to  1720  plaque-forming  units  (pfu)/mL  for  poliovirus,  .354  to 
24,290  cfu/mL  for  total  ]n  sjtu  bacterial  counts,  and  204  to  11,190  cfu/mL  for  total  ]n 
situ  enterics. 


53 


Volume  8 


The  average  removals  of  turbidity  are  shown  in  Table  6  for  various  run  times  up  to 
9  h.  As  might  be  anticipated,  most  of  the  turbidity  removal  took  place  in  the  multimedia 
deep-bed  filter,  and  the  removals  increased  with  filter  ripening.  Table  7  lists  the 
observed  removal  of  the  microorganisms  averaged  over  all  run  times,  and  Table  8  shows 
the  average  removal  of  the  total  aerobic  bacteria  (determined  by  standard  plate  count) 
for  the  three  water  flows  investigated.  Again  the  results  support  the  contention  that 
most  bacteria  and  viruses  are  removed  In  the  multimedia  filter  and  almost  none  in  the 
cartridge  filters.  The  data  shown  in  Table  8  are  erratic,  making  it  difficult  to  judge  the 
effect  of  feed-water  flow  on  the  removal  of  total  aerobic  bacteria,  except  possibly  to  say 
that  removal  may  have  decreased  at  the  highest  feed-water  flow  tested  (40  gal/mln).  The 
Investigators  concluded  that  a  40-gal/min  feed-water  flow  probably  should  be  avoided,  and 
that  operations  at  35  gal/mln  may  be  acceptable. 

Small  et  al,1  also  concluded  that  the  removals  of  the  microorganisms  shown  In 
Tables  7  and  8  are  not  meaningful  in  terms  of  providing  an  effective  barrier  against  the 
transmission  of  infectious  diseases.  The  major  line  of  defense  against  pathogens  would 
have  to  be  disinfection  if  the  RO  section  of  the  ROWPU  were  to  be  bypassed.  They  also 
pointed  out  that  the  results  of  this  study  cannot  be  used  to  predict  the  removal  of 
amoebic  cysts.  They  did  suspect  that  the  cysts  would  be  better  removed  than  the  bacteria 

due  to  their  larger  size. 

48 

3.  B.  Duncan  conducted  tests  on  the  removal  of  B.  globlgil  spores,  E.  coll, 
Saccharomyces  cervislae  (a  yeast)  as  a  cyst  simulant  for  Entamoeba  histolytica.  He  made 
total  plate  counts  and  total  enteric  bacteria  (as  defined  by  Small  et  al.  )  with  two  5.25-in. 
i.d.  filter  columns  packed  with  the  multimedia  used  in  the  600-gph  ROWPU  (see  Fig.  8). 
Soil  from  Frederick  County,  MD,  was  collected,  ground,  and  sifted  through  175-mesh 
screens.  The  fines  were  then  mixed  with  Fort  Detrick,  MD,  tap  water  to  yield  feed-water 
turbidity  levels  of  50  and  120  NTU.  A  particle-slze-dlstrlbution  analysis  Indicated  that 
most  of  the  suspended  particles  had  sizes  of  <  50  pm.  The  suspension  pH  was  then  adjusted 
to  5.3,  7.3,  and  9.3  with  carbon  dioxide  or  sodium  carbonate,  and  then  the  water  was 
pumped  through  the  filters  at  a  filtration  rate  of  6.3  gal/(mln  •  ft^).  The  two  filter 
columns  were  arranged  in  parallel  and  the  run  lasted  for  4  h.  A  filtration  rate  of 
6.3  gal/(mln  •  ft  )  is  equivalent  to  a  total  ROWPU  feed-water  flow  of  31  gal/min. 

The  turbid  feed  water  was  seeded  first  with  the  microorganisms  and  then  a  Cat-Floe 
polyelectrolyte  in  a  pipeline  leading  to  the  filters.  The  polyelectrolyte  dosage  was  not 
stated  In  the  preliminary  report.  The  flow  was  then  divided  and  delivered  equally  to  the 


54 


Volume  8 


Table  5.  Raw-water  characteristics  of  Monocacy  River.3 


Run 

Date  time  (h) 

Turbidity 

(NTU) 

pH 

TDS 

(mg/L) 

Alkalinity 
(mg/L  as  CaCOj) 

Temperature 

(°C) 

Plant  flow 
(gal/mln) 

9/17 

0 

3.5 

7.9 

278 

120.4 

19.4  to  22.3 

30 

3 

3.5 

8.0 

275 

116.8 

6 

3.7 

8.4 

270 

116.8 

9 

4.0 

8.4 

258 

116.8 

9/18 

0 

6.0 

7.9 

262 

113.1 

20.6  to  22.8 

35 

3 

5.9 

8.2 

275 

116.8 

6 

5.5 

8.4 

258 

113.1 

9 

5.6 

8.4 

322 

113.1 

9/23 

0 

4.7 

7.8 

288 

170.4 

22.2  to  24.0 

40 

3 

3.9 

7.8 

267 

120.4 

6 

3.8 

8.4 

262 

113.1 

9 

4.0 

8.4 

260 

116.8 

10/9 

0 

3.5 

7.9 

274 

153.9 

14.4  to  17.7 

40 

3 

4.2 

8.1 

286 

173.8 

6 

3.8 

8.4 

267 

158.9 

9 

3.8 

8.5 

260 

158.9 

10/10 

0 

2.4 

8.1 

282 

158.9 

14.2  to  14.8 

30 

3 

2.3 

7.9 

267 

148.9 

6 

2.2 

7.8 

270 

151.4 

9 

2.4 

7.8 

294 

153.9 

10/15 

0 

4.3 

7.8 

289 

148.9 

11.2  to  13.8 

35 

3 

4.3 

7.8 

300 

158.9 

6 

3.1 

8.0 

278 

158.9 

9 

3.3 

8.1 

272 

156.4 

10/22 

0 

3.0 

7.7 

282 

171.3 

12.7  to  14.8 

35 

3  ' 

3.1 

7.7 

305 

166.3 

3.0 

7.7 

285 

161.4 

9 

3.1 

8.0 

286 

158.9 

10/23 

0 

3.2 

7.7 

309 

156.4 

10.6  to  12.5 

40 

3 

2.9 

7.5 

293 

158.9 

6 

2.9 

8.0 

304 

163.8 

9 

3.2 

7.8 

285 

158.9 

10/29 

0 

7,1 

7.6 

398 

213.5 

10.0  to  12.4 

30 

3 

6.7 

7.9 

424 

213.5 

6 

5.1 

8.0 

445 

228.4 

9 

5.0 

8.0 

452' 

248.2 

““ 

a  Adapted  from  Small  etal.^ 


55 


Volume  8 


Table  6.  Average  removal  of  turbidity  observed  for  a  filtration  system  similar  to  that  in 
the  600-gph  ROWPU  pretreatment  section.® 


Elapsed  time  (h) 

Turbidity  removal  (*) 

Across  multimedia  filter 

Across  system 

0 

67.3 

74.4 

3 

81.4 

84.0 

6 

83.5 

86.4 

9 

85.9 

88.1 

a  Adapted  from  Small  et  al.* 


Table  7.  Average  removal  of  microorganisms  observed  for  a  filtration  system  similar  to 
that  in  the  600-gph  ROWPU  pretreatment  section.® 


Organism 

Microorganism  removal  (%) 

Across 

multimedia 

filter 

Across 

cartridge 

filters 

Overall 

B.  Klobiftii  spores 

98 

1 

98 

E.  coll _ 

91 

3 

94 

Poliovirus  I 

72 

8 

80 

Total  aerobic  bacteria15 

74 

9 

83 

Total  enteric  bacteria*5 

84 

3 

87 

a  Adapted  from  Small  et  al? 
13  As  defined  by  Small  et  al.1 


56 


Volume  8 


Table  8.  Average  removal  of  total  aerobic  bacteria  observed  for  a  filtration  system 
similar  to  that  used  in  the  600-gph  ROWPU  pretreatment  section  for  various  feed-water 
flows.3 


Total  aerobic  bacteria  removal 


Feed-water  flow  (gal/min) 

Across 

multimedia 

filter 

Across 

cartridge 

filter 

Across 

system 

30 

86 

3 

90 

35 

45 

48 

93 

40 

78 

-45 

33 

a  Adapted  from  Small  etal.^ 

b  Average  of  three  runs  per  feed-water  flow;  across-system  value  would  be  sum  of 
across-multi media  and  across-cartridge  filter  values,  only  If  perfect  sampling  and 
recovery  occurred. 


two  test-filter  columns.  Chlorine  was  added  to  the  feed  water  entering  one  of  the 
columns  while  the  other  column  was  used  as  a  control.  The  concentrations  of  the 
microorganisms  seeded  to  the  feed  water  were  determined  after  adding  the 
polyelectrolyte.  Only  the  results  of  the  control  studies  are  discussed  herein.  (Much  more 
information  is  required  to  evaluate  the  disinfection  studies  than  that  which  was  presented 
in  the  preliminary  report.! 

The  initial  concentrations  in  the  filter  influent  of  the  organisms  tested  were 
commonly  in  the  range  o*  thousands  of  cfu/mL  except  for  E.  coil,  which  had  arithmetic 
mean  initial  counts  of  68,000  cfu/100  mL  at  a  pH  of  5.5,  none  detectable  at  a  pH  of  7.5, 
and  1200  cfu/100  mL  at  a  pH  of  9.5.  The  observed  removals  of  the  microorganisms  are 
summarized  in  Table  9.  (No  information  on  turbidity  removal  was  given  in  the 
preliminary  report.;  In  all  cases,  removal  was  inconsequential,  as  far  as  providing  an 
effective  secondary  barrier  against  the  transmission  of  waterborne  diseases.  Disinfection 
(chlorination  in  this  case)  is  the  primary  barrier.  However,  the  coagulant  dosage  may 
have  been  less  than  optimal  for  the  large  concentrations  of  suspended  solids  contained  in 
the  feed  water. 


57 


Volume  8 


Follow-up  tests  were  run  by  Duncan  using  river  water  obtained  from  the 

Monocacy  River,  MD.  The  raw  water  had  a  turbidity  of  19  NTU  and  a  pH  of  8.3. 

6 

Approximately  10  cfu/ml  of  B.  globigli  spores  were  added  to  the  raw  water.  The 
multimedia  filter  removed  99.86%  of  the  spores  on  the  average,  and  the  cartridge  filter 
removed  70.79%  of  the  remaining  spores.  The  river-water  pH  was  then  decreased  to  7.0 
with  carbon  dioxide  and  the  test  was  repeated.  The  multimedia  filter  removed  nearly  all 
of  the  B.  globigli  spores  (initial  concentration  =  10^  cfu/mL).  Unfortunately,  the 
information  made  available  to  our  investigators  does  not  state  whether  the  water  was 
prechlorinated  or  not.  In  view  of  the  results  shown  in  Table  9  for  unchlorinated  water, 
one  has  to  suspect  that  chlorine  may  have  been  used  in  these  follow-up  tests. 

PERFORMANCE  OF  CARTRIDGE  FILTERS 

The  current  600-gph  ROWPU  pretreatment  system  Includes  eight  Filterite 
polypropylene  cartridge  filters,  as  shown  in  Fig.  10.  These  filters  have  a  nominal  size 
rating  of  5  urn.  Most  bacteria  and  all  viruses  are  smaller  than  this  size  rating  and  are  also 
smaller  than  the  3*pm  absolute  rating  now  being  considered  as  a  replacement  for  the  5-pm 
cartridge  filters.  An  important  contribution  to  pathogen  removal  to  be  made  by  the 
cartridge  filter  could  be  in  the  area  of  cyst  removal.  Glardia  lamblla  cysts  have  a 
minimum  dimension  of  7  pm,  and  E.  histolytica  cysts  are  commonly  larger  (6  to  18  pm). 

It  appears  that  very  little  research  relevant  to  the  purposes  of  our  study  has  been 
performed  with  cartridge  filters.  Long*7  evaluated  17  cartridge  filters  having  various 
size  ratings  for  removal  of  a  microsphere  cyst  model  (5.7  pm  in  diameter).  The 
microspheres  were  filtered  from  a  solution  containing  40,000  to  65,000  microspheres/mL, 
diluted  with  tap  water.  No  further  information  was  included  in  the  article  regarding  pH, 
temperature,  etc.  More  than  99.99%  of  the  microspheres  was  removed  in  10  of  the 
17  filters  tested.  It  is  interesting  to  note,  however,  that  filters  with  size  ratings  of  1.0, 
2.0,  and  5.0  pm  did  not  satisfy  the  99.99%  removal  criteria,  whereas  some  filters  with 
ratings  of  2.0,  3.0  and  10.0  pm  did.  In  addition,  two  filters  rated  at  the  same  size,  5  pm, 
had  widely  varying  particle  removals  (99.96  vs  21.4%). 

It  was  not  possible  to  determine  from  the  article  whether  the  filter  used  in  the 
600-gph  ROWPU  was  tested.  In  view  of  the  large  discrepancies  in  the  results,  however,  it 
is  clear  that  the  performance  of  cartridge  filters  may  be  highly  variable. 


58 


Volume  8 


Table  9.  Average  removal  of  microorganisms  from  a  turbid  water  for  three  pH  values  by 
a  multimedia  filter  similar  to  that  used  in  the  600-gph  ROWPU.a  (Filtration  rate  =  6.3 
gal/(min  •  ft^)  and  feed-water  turbidities  of  50  and  120  NTU.) 


Microorganism  removal  b  (%) 

Organism 

pH  *  5.5 

pH  =  7.5 

pH  =  9.5 

B.  Rlobigii  spores 

81 

83 

90 

E.  coll 

63 

NDC 

75 

S<  cerevisiae  (6-  to  18-ym  elliptic  yeast  ceils) 

84 

NDC 

94 

Total  aerobic  bacteria*1 

80 

70 

87 

Total  enteric  bacteria*1 

82 

42 

63 

a  Adapted  from  Duncan^ 

15  Arithmetic  averages. 

c  None  detected  In  the  water  supplied  to  the  multimedia  filters. 
d  As  defined  by  Small  et  al.1 


Carbon-impregnated  cartridge-filter  elements  were  tested  during  the  early  stages  of 
the  development  and  design  of  the  600-gph  ROWPU.^  Cartridge-filter  elements  ranging 
from  3  to  20  yn,  in  size  rating  were  tested.  It  was  found  that  the  filter  elements  with  the 
smaller  size  ratings  gave  better -quality  effluents,  but  a  trade-off  had  to  be  made  between 
filter  life  and  effluent  quality.  The  data  showed  that  a  filter  element  with  a  size  rating 
of  5ym  produced  a  water  with  a  turbidity  of  <0.5  FTU,  and  had  an  average  useful  life  of 
72  h.  The  useful  life  is  apparently  designated  as  the  time  required  to  yield  a  20  psid1*  or 
25  psid  pressure  drop  across  a  filter  element. 

Radoski50  investigated  the  removal  of  diesel  fuel  from  tap  water  with  carbon- 
containing  cartridge  filters.  It  was  claimed  that  the  filtered  water  was  free  of  any 
objectionable  taste  or  odor  at  a  water  temperature  of  14  to  19°C  when  the  feed  water 
contained  1  mg/L  of  the  diesel  fuel. 

PERFORMANCE  OF  PRECOAT  FILTERS  OPERATED  IN  THE  DIRECT-FILTRATION 
MODE 

In  i  very  early  literature  review,  Cummins^ 1  states  that  size  distribution  and 
particle  shape  of  the  filter-aid  media  are  very  important  factors  in  dlatomaceous-earth 
filtration. 


59 


Volume  8 


The  performance  of  a  diatomaceous-earth  filter  aid  can  be  improved  by  coating  it 

with  an  electropositive  coating  medium.  The  filter-aid  coating  is  attained  either  by 

mixing  the  diatomite  and  coating  material  prior  to  application  of  the  filter  aid  or  by 

filtering  water  with  the  coagulant  in  it  through  the  filter  aid  after  it  has  been  applied.  A 

1936  patent  by  Cummins  states  that  a  filter  aid  coated  with  alum  is  more  effective  than 

an  uncoated  aid  in  removing  suspended  solids  from  water.  Diatomaceous  earth  coated 

with  alum  is  less  permeable  to  water  than  uncoated  diatomaceous  earth  of  the  same 

grade,  but  the  coarser  grades  coated  with  alum  are  more  permeable  than  the  uncoated 

finer  grades,  although  each  may  possess  similar  capacities  for  the  removal  of  turbidity, 

coliform  organisms,  and  color.  It  was  concluded  that  the  use  of  alum  coated  filter  aid  was 

52 

advantageous  due  to  the  reduction  in  pressure  drop. 

The  Department  of  Water  and  Power  of  Los  Angeles,  CA,  carried  out  a  study  on 
diatomaceous-earth  filtration  for  the  city's  water  supply  in  1931. 33  It  was  found  that 
slimy  materials,  large  concentrations  of  algae,  and  close-textured  diatoms  shortened  the 
filter  runs.  Turbidity  resulting  from  more  than  20  ppm  of  suspended  silt  particles,  as  veil 
as  any  fine  floe  reaching  the  filter,  were  "troublesome."  It  was  also  found  that 
dissolved  and  entrained  gases  had  no  appreciable  effect  on  the  filter  efficiency,  provided 
that  sufficient  backpressure  was  maintained  to  prevent  the  release  or  entrainment  of 
gases  in  the  filter. 

The  openings  in  the  septa  used  in  the  Los  Angeles  study  varied  from  25  to  250  ym. 
The  septum  with  the  largest  openings  required  a  longer  time  to  precoat,  but  the 
precoating  time  was  fairly  constant  for  septa  with  openings  finer  than  160  ym.  The 
septum  with  the  smallest  openings  had  the  largest  initial  pressure  drop,  tended  to  clog 
rapidly,  and  was  difficult  to  clean.  The  clogging  in  the  precoated  filter  elements  was 
caused  by  the  finer  filter-aid  particles  contained  in  the  precoat,  and  was  more  noticeable 
with  the  thicker  septa.  It  was  concluded  that  the  largest  opening  capable  of  retaining  the 
filter  aid  should  be  used  to  minimize  pressure  drop  and  clogging. 

Microorganism  removal  in  the  tests  agreed  with  the  previously  observed  0  to  96% 
removal  for  bacteria  derived  from  treatment  of  swimming-pool  water,  depending  on  how 
long  Into  the  filter  run  the  sample  was  taken.  The  passage  of  bacteria  through  the  filter 
increased  rapidly  as  the  size  of  the  filter  aid  increased.  For  the  most  economical 
operating  conditions  employing  the  coarse  filter  aids  and  thin  precoats,  there  was  a 
tendency  for  bacterial  breakthrough,  and  this  was  especially  evident  at  pressure  drops 
across  the  filter  elements  >  10  psi. 


Volume  8 


52 

Oulman  et  al.  conducted  a  study  to  determine  and  evaluate  the  effects  of  adding 
cationic  polyelectrolytes  to  the  feed  water  continuously  or  during  the  precoating 
operation  of  the  diatomaceous-earth  filters.  The  parameters  investigated  included 
polymer  dose,  filter-cake  permeability,  pH,  and  removal  of  turbidity,  bacteria,  and  color. 
The  effects  of  coating  the  filter  aid  with  the  polyelectrolytes  were  explained  by  the  fact 
that  the  coating  increases  the  contact  area  between  the  suspended  particles  and  filter  aid, 
and  decreases  the  probability  of  desorption  of  an  adsorbed  particle.  It  was  demonstrated 
that  filter  aid  coated  with  polyelectrolytes  provides  a  more  complete  removal  of 
turbidity,  bacteria,  and  color  colloids  with  a  lower  pressure  drop  than  the  same  filter  aid 
without  a  polyelectrolyte  coat. 

A  1965  American  Water  Works  Association  (AWWA)  Task  Group  Report*^  stated 
that  many  of  the  municipal  diatomaceous-earth  filtration  systems  functioning  at  the  time 
were  inadequate  for  the  following  reasons:  (1)  the  equipment  for  effective  precoating, 
filtering,  and  backwashing  was  inadequate,  (2)  the  hydraulic  loading  rate  of  6  to 

8  gal/(mln  •  ft  )  commonly  used  in  practice  was  too  high  for  continuous  operation  at 

fixed  installations,  and  (3)  the  equipment  was  unable  to  supply  filter  aid  adequately  as 
body  feed  during  the  run. 

The  AWWA  report  also  stated  that  most  of  the  municipal  experience  with  turbidity 
removal  had  been  restricted  to  waters  low  in  turbidity  and  of  good  bacteriological  quality 
that  required  little  pretreatment  except  prechlorination.  There  was  no  agreement  as  to 
the  maximum  level  of  turbidity  that  could  be  removed  without  difficulty,  but  it  was 
agreed  that  direct  diatomaceous-earth  filtration  could  not  be  recommended  for  a  grossly 
or  even  moderately  polluted  supply.  Appropriate  pretreatment  (conventional  or  chemical 
pretreatment  to  increase  filterability)  could  widen  the  application  areas  for 

diatomaceous-earth  filters.  Finally,  the  AWWA  Task  Group  recommended  that  the 

filtration  rate  be  held  to  1  gai/(min  •  ft  )  for  public  water  supplies  unless  available  data 
justify  an  increase  in  the  rate. 

Syrotynski'*^  reviewed  the  performance  of  diatomaceous-earth  filtration  plants  in 
the  State  of  New  York.  No  coagulants  were  reported  to  be  added  to  the  filter-feed 
waters.  The  results  of  their  survey  indicated  that  the  total  counts  of  bacteria  in  the 
filtered  water  did  not  correlate  with  either  initial  count  or  the  filtered-water  turbidity. 
In  a  subsequent  report,  Syrotynski  and  Stone^  reported  that  the  total  microscopic  count 
is  decreased  about  60  to  90%  with  direct  diatomaceous-earth  filtration  with  no 
coagulation.  Body  feed  was  utilized  and  the  filtration  rate  was  held  constant  at 
1.0  gal/(min  •  ft^). 


61 


Volume  8 


56 

Hunter  et  al.  investigated  the  removal  of  total  coliform  bacteria  with  direct 
dlatomaceous-earth  filtration  and  no  in-line  coagulation,  with  and  without  coated  filter 
aids.  The  removal  of  the  coliform  bacteria  increased  from  90  to  99.86%  as  finer  grades  of 
diatomaceous  earth  were  tested.  The  initial  feed-water  count  of  the  coliform  bacteria 
exceeded  19,000  organisms/ 100  mL  for  these  runs.  Various  levels  of  body  feed  were 
utilized  with  a  precoat  of  0.1  lb/ft2. 

Those  investigators  concluded  thatt  (1)  the  greater  the  permeability  of  the  filter 
aid,  the  greater  the  number  of  coliform  organisms  that  are  able  to  penetrate  it} 
(2)  "complete"  coliform  removal  is  possible  using  very  fine  filter  aid  for  influent  coliform 
concentrations  of  several  thousand  organisms  per  100  mL}  (3)  beyond  that  level,  higher 
influent  coliform  concentrations  give  higher  effluent  concentrations}  (4)  an  increase  from 
1  to  2  gal/(min  •  ft2)  in  the  filtration  rate  did  not  change  the  removal  of  the  coliform 
bacteria  significantly}  (5)  increasing  the  body-feed  filter-aid  dosages  from  7.3  to  60  ppm 
did  not  change  the  removal  of  the  bacteria  significantly}  and  (6)  coated  filter  aids 
improved  the  removal  of  the  bacteria,  but  the  improvement  was  not  sufficient  to  merit 
the  use  of  such  aids. 

Burns  et  al.  coated  three  different  grades  of  diatomlte  filter  aid  (Hyflo  Supercel, 
Celite  543,  and  Ceiite  560  manufactured  by  3ohns-Manvllle  Products  Corp.,  Manville,  N3) 
with  a  cationic  polyelectrolyte  (Purlfloc  601,  manufactured  by  Dow  Chemical  Co., 

Midland,  MI).  The  coating  operation  consisted  of  mixing  the  diatomlte  for  about  3  min  in 

57 

a  "dilute  solution"  of  the  polyelectrolyte  and  pumping  the  mixture  through  the  filter. 
The  filter  cake  was  then  washed  with  distilled  water  to  remove  excess  polyelectrolyte.  A 
1.5-in.  diameter  flat,  horizontal  septum  (100-mesh  stainless-steel  wire  screen)  was  used  in 
those  experiments.  The  following  three  suspensions  were  fllteredi  (1)  a  1-g/L  calcium 
bentonite  clay  with  particle  sizes  of  0.8  to  1.0  pm,  (2)  a  4-g/L  Black  Hills  clay  with 
particle  sizes  of  0.45  to  0.8  pm,  and  (3)  30-mg/L  Folger's  Instant  Coffee  filtered  through  a 
0.45-pm  Millipore  filter.  The  coffee  solution  was  intended  to  serve  as  a  source  of  color 
colloids. 

The  uncoated  diatomaceous-earth  filter  aids  did  not  remove  any  of  the  clayey 
suspended  solids.  The  removal  with  coated  filter  aids  Increased  with  the  weight  of 
polyelectrolyte  added  per  unit  weight  of  the  filter  aid.  Two  different  mechanisms  were 
postulated  to  explain  the  results  of  the  study.  First,  the  polyelectrolyte  coating  on  the 
dlatomaceous-earth  filter  aid  ensures  nearly  100%  removal  of  the  suspended  clay  particles 
until  its  charge  capacity  is  exhausted;  and  second,  any  subsequent  removal  is^chieved  by 
straining  only.  The  effectiveness  of  the  polyelectrolyte  coating  deteriorated  for  pH 
values  in  excess  of  about  6  to  7. 


62 


Volume  8 


The  coated  filter  aid  did  not  fully  eliminate  the  color  produced  by  the  coffee 
solution  under  any  circumstances.  It  was  concluded  that  the  color  was  associated  at  least 
in  part  with  either  nonreactive  colloids  or  dissolved  substances.  The  suspended-solids 

removal  was  around  80%  with  the  fully  coated  filter  aids. 

58 

A  study  carried  out  at  the  University  of  Texas  in  1974  reports  on  the  removal  of 
an  E.  coll  T2  bacteriophage  by  direct  diatomaceous-earth  filtration.  The  system  used  did 
not  include  in-line  flocculation,  and  the  filter  aids  were  coated  with  a  cationic 
polyelectrolyte.  Filter  aids  were  Hyflo,  Hyflo  B,  Hyflo  D,  and  Cellte  560  (manufactured 
by  Johns-Manville  Products  Corp.).  Coatings  were  Purlfloc  C-31  (manufactured  by  Dow 
Chemical  Co.),  aluminum  and  ferric  salt  hydrates.  The  experiments  were  performed  with 
a  0.15-lb/ft2  precoat,  50-mg/L  body  feed,  and  a  1  gal/(min  •  ft2)  filtration  rate.  The 
bacteriophage  was  added  to  dechlorinated  tap  water  in  amounts  sufficient  to  give  a  feed 
water  concentration  of  1700  pfu/L. 

From  a  series  of  2-h  runs  two  filter-aid  configurations  were  selected  for  additional 
study:  (1)  Hyflo  filter  aid  with  0.07  mg/L  of  polyelectrolyte,  and  (2)  Hyflo  D  alone  (no 
polyelectrolyte).  Two  12-h  runs  were  made.  No  observable  breakthrough  occurred  after 
12  h  with  the  pretreated  Hyflo,  though  breakthrough  occurred  after  2  h  with  Hyflo  D 
alone.  It  was  concluded  that  a  good  removal  of  viruses  (>99%)  could  be  achieved  only 
with  a  polyelectrolyte  coagulant  plus  an  adequate  precoat  with  certain  of  the  tested  aids. 
Testing  was  also  done  to  affirm  that  the  polyelectrolyte  had  no  viricidal  effect  on  T2 
phage  after  4  h  at  a  polyelectrolyte  concentration  of  0.081  mg/L. 

A  second  stage  of  testing59  at  the  University  of  Texas  was  conducted  with 
procedures  identical  to  those  described  previously  for  the  T2  phage  study.  In  this  case 
poliovirus,  Mahoney  Type  I  was  added  to  tap  water;  some  additional  test  runs  with  T2 
phage  are  reported  as  well.  A  series  of  2-h  runs  was  made.  With  coated  filter  aids  and 
polyelectrolyte  pretreatment  >98  %  removal  of  virus  was  obtained.  The  uncoated  filter 
aid,  however,  gave  only  62%  removal  of  poliovirus  after  2  h,  as  opposed  to  90%  removal 
of  phage  T2  after  the  same  period  of  time. 

Two  12-h  runs  were  made  with  Hyflo  filter  aid  and  polyelectrolyte.  After  12  h  of 
continuous  operation,  there  was  no  significant  breakthrough  of  poliovirus. 

Two  4-h  runs  were  made  with  two  filter  aids  to  Investigate  the  breakthrough 
phenomenon.  One  of  the  filter  aids  was  coated,  the  other  one  uncoated.  After  4  h, 
removal  of  poliovirus  had  dropped  to  90%  for  the  uncoated  filter  aid  and  to  98%  for  the 
coated  filter  aid. 


63 


Volume  8 


Finally,  a  series  of  2-h  runs  using  T2  phage  with  three  kinds  of  filter  aid  at  flow 
rates  of  1  and  2  gal/(min  *  ft  )  and  pH  values  of  6.7  and  9.5  were  performed.  From  the 
results  obtained,  It  was  concluded  that  fliter-ald  grade  and  flow  rate  did  not  affect  T2 
phage  removal.  However,  pH  did  affect  removal,  with  >99.0%  at  pH  9.5  and  92.3%  at  pH 
6.7  for  one  filter  aid. 

It  was  concluded  that  >98  %  removal  of  phage  T2  and  poliovirus  can  be  obtained 
when  a  coagulant  is  added  to  the  filter  feed  water.  Filtration  rate,  body-feed  rate,  and 
fiiter-aid  grade  did  not  appear  to  affect  the  removal  of  the  viruses  significantly. 
Surprisingly,  it  was  found  that  "the  same  virus  removals  were  obtained  either  with  or 
without  body  feed}"  hence  the  body-feed  rate  was  set  to  maintain  cake  porosity. 

Amlrhor  and  Engelbrecht60  investigated  the  removal  of  bacteriophage  MS 2  using 
direct  dlatomaceous-earth  filtration  with  Celite  560  (manufactured  by  Oohns-Manville 
Products  Corp.)  without  body  feed.  A  0.5-in.  thick  precoat  of  diatomaceous  earth  was 
applied  to  the  filter,  and  filtration  took  place  at  a  filtration  rate  of  1  gal/(min  •  ft2). 
The  filter  aid  was  coated  with  a  cationic  polymer,  Purlfloc  C-31  (manufactured  by  Dow 
Chemical  Co.),  at  quantities  equivalent  to  0.2  and  0.4  mg  of  polyelectrolyte  per  gram  of 
diatomaceous-earth  filter  aid.  The  virus  was  seeded  to  a  dlstllled-deionized  water 
buffered  to  pH  levels  of  6,  7,  and  8. 

Virtually  no  virus  was  removed  by  the  uncoated  diatomaceous-earth  filter  aid  as 
shown  in  Fig.  18.  The  coated  filter  aid  gave  about  90%  removal  or  better  for  a  run  time 
of  7  h  and  a  pH  of  6.0.  The  results  of  other  tests  not  given  herein  demonstrated  that  the 
removal  of  the  virus  decreased  with  increasing  pH,  and  breakthrough  occurred  when  the 
absorptive  capacity  of  the  polyelectrolyte  was  exhausted.  A  number  of  experiments  were 
also  conducted  with  an  uncoated  filter  aid  and  the  poiyelectrolyte  added  to  the  filter  feed 
water.  The  results  demonstrated  that  the  dose  of  a  coagulant  must  be  tailored  to  the 
character  of  the  feed  water  in  direct  filtration.  The  lowest  dose  of  poiyelectrolyte  used 
in  this  study  (0.25  mg/L)  gave  the  best  removal  of  the  virus  for  the  operating  conditions 
used. 

Recent  studies  by  Lange  et  al.^  demonstrated  a  decrease  in  the  removal  of 
total-collform  bacteria  with  an  increase  in  the  filtration  rate  The  authors  performed  the 
studies  on  low-turbidity  water  (4.5  to  5.4  NTU)  at  filtration  rates  of  1,  2,  and 
4  gal/(min  •  ft  ).  Unlike  the  results  reported  by  Hunter  et  al.,  the  removal  of  both 
total  coliform  bacteria  and  standard  plate  count  bacteria  increased  significantly  when  the 
diatomaceous  earth  was  coated  with  alpm.  For  example,,  the  range  in  coliform  bacteria 
removal  increased  from  about  30  to  70 %  without  alum  coating  to  about  96  to  99.9%  with 
coating.  The  authors  also  concluded  that  the  most  Important  factor  in  removal  of 


64 


Volume  8 


0  1  2  3  4  6  6  7  8  9  10  11  12  13  14  16 

Filtration  time  (h) 


Figure  18.  Removal  of  MS 2  bacteriophage  by  uncoated  and  polyelectrolyte  (PE)  coated 
diatomaceous-earth  (DE)  filter  aid.  [pH  a  6.0{  virus  concentration  a  4  x  108  pfu/mLj 
filtration  rate  =  1  (gal/min  •  ft2)].  Adapted  from  Amlrhor  and  Engelbrecht.60 


bacteria  is  the  grade  of  the  diatomaceous  earth.  A  coarse  grade  of  diatomite  gave  a  28% 
removal  of  collform  bacteria  and  a  38%  removal  of  standard  plate  count  bacteria, 
whereas  a  fine  grade  of  diatomite  gave  a  99.9%  removal  of  the  conforms  and  a  99.8% 
removal  of  the  standard  plate-count  bacteria. 

The  cysts  of  E.  histolytica  and  G.  lamblia  are  similar  in  size  and  surface  electrical 
charge.  They  are  dissimilar  in  that  the  pathogenicity  of  E.  histolytica  has  long  been 
known,  whereas  only  recently  has  the  pathogenicity  of  G.  lamblia  been  recognized.  Since 
the  mld-70's,  research  has  been  conducted  in  removal  and  inactivation  techniques  for  G. 
lamblia.  Both  G.  lamblia  and  l:.  histolytica  are  protozoans  that  are  pathogenic  to 
humans.  They  are  similar  in  size,  the  cysts  of  G_.  lamblia  being  ovoid  in  shape,  14  to 
16  pm  long  and  6  to  12  pm  wide.  E.  histolytica  has  a  spherical  cyst  form  10  to  15  pm  in 
diameter.  They  are  also  similar  in  symptoms  of  disease,  i.e.,  both  cause  diarrhea.  Of 


65 


Volume  8 


concern  to  the  mobile  field  Army,  the  debilitating  effects  of  this  condition  could  render 
military  personnel  incapable  of  action.  Because  water  supply  has  been  incriminated  for 
carriage  of  both  protozoan  cysts,  the  Army  is  concerned  about  removal  of  cysts  from  Its 
water  supply. 

62 

According  to  an  early  paper  by  Kominek,  diatom aceous-earth  filters  are  effective 
in  the  removal  of  chlorine-resistant  organisms,  such  as  cysts,  as  well  as  suspended  solids. 
Kominek  hypothesized  that  diatomaceous  earth  forms  a  porous  cake  with  exceedingly  fine 
openings,  thereby  providing  excellent  filtration  without  the  refinement  in  pretreatment 
technique  required  for  rapid  sand  filtration. 

Leading  investigators  of  diatomaceous-earth  filtration  for  drinking-water  treatment 

were  with  the  U.S.  Army  during  World  War  I!.63  The  data  contained  in  an  Army  report6^ 

of  Duly  1944  resulted  from  the  combined  efforts  of  the  Office  of  the  Surgeon  General 

(OTSG)  and  the  National  Institutes  of  Health.  The  experiments  are  described  In  the  Army 

33 

publication  and  were  summarized  recently  by  Logsdon  et  al.  The  Intent  of  those  studies 
was  to  determine  the  removal  of  the  cysts  of  E.  histolytica  with  diatomaceous-earth 
filtration.  Several  diatomaceous-earth  filtering  systems  were  tested,  from  small  batch 
systems  (4-L  volume)  to  large  continuous-flow  systems  (73  L/mln).  The  precoating  water 
was  contaminated  in  the  only  run  out  of  a  total  of  13  where  more  than  one  cyst  was 
recovered  from  the  filtered  water.  It  was  concluded  by  both  the  Army  and  Logsdon  that 
virtually  all  cysts  of  E.  histolytica  can  be  removed  with  direct  diatomaceous-earth 
filtration. 

£3 

Dones  and  Brady  studied  the  removal  of  Schistosoma  mansonl  cercariae  by 

"  2 

filtration  through  diatomaceous  earth  In  a  small  model  filter.  A  precoat  of  0.15  lb/ft 
was  used  without  body  feed,  and  either  tap  or  raw  river  water  was  used  In  the 
experimental  runs.  The  removal  of  the  cercariae  was  investigated  in  nine  runs.  Varying 
filtration  rates  [0.3  to  11.1  gal/(min  •  ft2)]  and  applied  pressures  (0  to  17  psi)  were  used, 
as  well  as  three  types  of  diatomaceous-earth  filter  aids  (Sorbo-Cel  503,  Speedplus,  and 
4200+kaolln).  Referring  to  the  filtered  water  quality,  the  authors  state  "...there  were  no 
cercariae  recovered  in  any  of  the  samples  examined."^ ^ 

Baumann  and  Babbitt^  also  examined  the  removal  of  the  cysts  of  E.  histolytica 
with  porous  filter  septa  both  with  and  without  filter  aid.  They  were  looking  for  a  septum 
that  would  remove  cysts  without  filter  aid,  on  the  hypothesis  that  application  of  the  filter 
aid  is  the  most  likely  procedure  for  error.  One  of  the  more  interesting  conclusions  made 
in  that  report  is  that  "a  septum  that  will  effect  a  removal  of  roughly  75  to-£0%  of  the 
applied  bacteria  will  remove  also  100%  of  the  applied  cysts. They  also  concluded  that 
septa  with  a  mean  pore  size  of  18.6  pm  or  less  will  remove  all  of  the  cysts. 


66 


Volume  8 


r  y  /o 

Logsdon  and  several  associates0'  and  DeWalle  et  al.  found  diatomaceous-earth 
filtration  to  be  effective  in  removing  G.  latnblia  cysts,  G.  muris  cysts,  and  9-ym 
radioactive  beads  that  served  as  models  of  G.  lamblla. 

The  Drinking  Water  Research  Division  of  the  Environmental  Protection  Agency 
(EPA)  conducted  a  study  in  1981  on  the  removal  of  Giardia  cysts  and  cyst  models  from 
drinking-water  supplies  by  alternative  filtration  methods.  3  Radioactive  microspheres 
9  ym  in  diameter  were  used  as  Giardia  cyst  models  in  the  first  phase  of  the  diatomaceous- 
earth  filtration  study.  Those  spheres  are  very  similar  to  Giardia  cysts  in  size  and  are 

/  n 

easily  traced.  Logsdon  et  al.  pointed  out  that  data  procured  from  the  microsphere 
studies  gave  conservative  results  because  in  some  of  their  corroborative  runs  better 
removals  were  obtained  with  G.  lamblla  than  with  the  microspheres.  G.  muris  cysts, 
which  are  similar  in  size,  shape,  and  surface  electrical  charge  to  G.  lamblla  cysts,  were 
also  used  later  on  In  the  diatomaceous-earth  filtration  studies  because  it  proved  difficult 
to  obtain  viable  G.  lamblla  cysts  on  a  continuous  basis. 

Low-turbidity  water  was  obtained  from  a  local  gravel  pit  on  the  premise  that  it 
closely  resembled  the  mountain  streams  through  which  giardiasis  is  commonly  spread. 
The  diatomaceous-earth  test  filter  had  an  area  of  0.1  m*  (1  ft^)  and  was  operated  as  a 
pressure  filter  with  a  slurry  feeder  for  body  feed.  The  hydraulic  loading  rates 
investigated  were  2.4  and  3.5  m/h  [1.0  and  1.4  gal/(min  •  ft2)],  and  the 
body-feed-to-turbidity  (mg/L  to  NTU)  ratios  ranged  from  2»1  to  35j1.  The  entire  effluent 
from  the  test  filter  was  filtered  through  5-um  membrane  sampling  filters  to  determine  the 
removals  of  the  cysts  and  cyst  simulants. 

The  microspheres  were  dosed  on  a  continuous  basis,  and  the  filter  was  operated  for 
periods  ranging  from  2  to  27  h,  with  a  typical  run  lasting  from  6  to  8  h.  The  G.  muris 
cysts  were  dosed  in  one  slug  of  20  x  106  to  40  x  106  cysts,  just  ahead  of  the 
filter-pressure-containment  vessel.  Filtration  was  continued  until  at  least  99%  of  the 
cysts  could  be  collected  on  or  pass  through  the  filter.33  After  the  run  the  membrane 
filter  was  washed  and  the  retained  cysts  were  concentrated  and  counted. 

The  results  of  the  radioactive  microsphere  tests  indicated  that  in  most  cases  more 
than  99.9%  of  the  microspheres  were  removed  by  the  diatomaceous-earth  filter.  The 
removal  of  the  cysts  appeared  to  be  unrelated  to  the  reduction  in  turbidity.33  The 
passage  of  turbidity  at  a  time  in  the  run  when  cysts  do  not  pass  is  attributed  to  the 
difference  in  the  sizes  of  the  several  kinds  of  particles.  Cysts  have  a  smallest  dimension 
of  about  7  ymj  clays  and  bacteria  can  be  as  small  as  1  ym,  and  thus  would  be  capable  of 
passing  through  the  filter-cake  pores,  while  the  cysts  would  be  strained  out. 


Volume  8 


The  microsphere-removal  efficiencies  were  found  to  be  related  to  certain  operating 
conditions.  The  removals  increased  with  the  application  of  the  precoat  up  to  1.0  kg/m2. 
Additional  precoat  did  not  improve  performance.  The  removal  consistently  equaled  or 
exceeded  99.9%  with  a  precoat  of  1.0  kg/m2  and  body  feed.  Finally,  the  test  results  also 
indicated  that  the  removal  capability  of  the  diatomaceous-earth  filter  usually  improved 
during  a  filter  run.  This  was  caused  by  the  increase  in  cake  depth  resulting  from  the 
collection  of  the  body  feed  and  suspended  solids.^ 

The  G.  murls  cyst  studies  gave  good  removal  at  both  the  2.4  and  3.5  m/h  [1.0  and 
1.4  gal/(min  •  ft2)]  filtration  rates.  Removal  was  greater  than  99.8%  with  the  exception 
of  one  run  at  99.36%.  That  particular  run  was  also  distinguished  by  an  above-normal 
amount  of  dlatomite  in  the  filter  effluent.  The  results  of  this  test  series  also  confirmed 
the  finding  given  previously  that  cyst  removal  does  not  correlate  with  turbidity  removal. 
It  was  concluded  that  diatomaceous  earth  filtration  with  body  feed  and  a  precoat  of 
1.0  kg/m2  removes  Glardia  cysts  effectively  and  economically  as  long  as  the  filters  are 
operated  and  maintained  properly,  and  as  long  as  the  filtration  rate  does  not  exceed  that 
used  in  the  study  [3.5  m/h  or  1.4  gai/(mln  •  ft2)]. 

DcWallc  ct  al.  added  Glardia  cysts  to  unfiltered  Seattle  tap  water  with  a  turbidity 
in  the  range  of  0.6  to  0.9  NTU.  The  seeded  water  was  not  coagulated  or  disinfected.  The 
ranges  in  precoat  and  body  feed  investigated  were  0.5  to  1.2  kg/m2  and  10  to  40  mg/L, 
respectively.  The  Glardia  cysts  were  added  as  a  slug  to  the  filter  Influent  (3.0  x  10^  cysts 
in  10  s).  Removal  ranged  from  99.03  to  >99.87%  In  12  separate  determinations.  The  best 
removal  was  obtained  with  the  following  operating  conditions:  a  precoat  of  1.0  kg/m2,  a 
body  feed  of  20  mg/L,  an  hydraulic  loading  rate  of  3.8  L/min  [0.09  gal/(mln  *  ft2)]  and  a 
water  pH  of  6.7.  The  removal  of  the  cysts  increased  as  the  bed  matured  during  the  course 
of  a  filter  run. 

Lange  et  al.^  reported  Glardia  cyst  removals  of  99.9%  for  seven  grades  of 

diatomaceous-earth  filter  aids  subjected  to  filtration  rates  of  2.44  to  9.76  m/h  [1  to 
2 

4  gal/(mln  •  ft  )]  in  the  water-temperature  range  of  5  to  19°C.  Low-turbidity  waters, 
approximately  5  NTU,  were  used  in  the  testing. 

PERFORMANCE  OF  THE  MWPU  (ERDLATOR) 

69 

Black  and  Spaulding  conducted  research  on  the  then  new  Army  diatomaceous- 
earth  filtration  technique.  Half  of  their  experimental  runs  were  made  with  just  filtration, 
and  the  other  half  included  coagulation  with  alum  and  sedimentation  prior  to  filtration. 
The  concentrations  of  E.  coli  were  determined  with  the  multiple-tube  dilution  technique. 


68 


Volume  8 


The  untreated  water  when  filtered  "seldom  produced  negative  tests  for  E_.  coli  in  all  five 

tubes  even  when  filtered  turbidity  was  at  a  minimum."  Better  results  were  obtained  with 

the  coagulated  and  settled  water.  Measurable  turbidity  (although  fairly  low)  was  found  in 

most  filtrates  obtained  from  the  untreated  water. 

Special  field  studies  were  conducted  in  the  1950's  to  assess  the  performance  of  the 

28 

MWPU  under  closely  controlled  conditions.  Table  10  shows  results  of  one  series  of 

biological  testing  on  the  prototype  MWPU.  The  coagulation  process  alone  provided  good 

reductions  of  bacteria.  Of  the  six  runs  made  without  the  addition  of  chlorine,  fecal 

coliform  bacteria  were  detected  in  the  filtered  water  of  only  one  of  the  runs. 

Three  nonpathogenic  organisms  (Serratia  marcescens,  Bacillus  subtills  var.  niger, 

and  T3  coliphage)  were  used  in  another  series  of  MWPU  tests.  "Very  substantial" 

quantities  of  S.  marcescens  and  T3  coliphage  were  removed  by  the  MWPU  without 

disinfection.  For  example,  an  average  of  99 .6  &  T3  coliphage  was  removed  by  the 

22 

ERDLATOR,  followed  by  dlatomaceous-earth  filtration. 

Bacterial  counts  showed  removals  in  the  ERDLATOR  (the  clarifier)  of  81  to  92%  in 

yet  another  series  of  tests  when  contaminated  water  with  490  to  33,000  bacteria/ 100  mL 
26 

was  treated.  The  test  water  was  Potomac  River  water  and  standard  additions  of  FeCl^ 

and  pulverized  limestone  were  used  without  prechlorination.  The  turbidities  ranged  from 

25  to  150  ppm  in  the  raw  water,  and  from  less  than  l  to  3  ppm  in  the  filtered  water. 

26 

Testing  was  performed  by  the  National  Institutes  of  Health  on  cysts  of 

E.  histolytica  at  a  concentration  of  5500  cysts/gal.  Cysts  were  recovered  in  the 

ERDLATOR  (the  clarifier)  effluent,  ranging  from  0  to  8  cysts/gal. 

B.  subtllus  was  used  as  the  test  organism  to  assess  the  MWPU's  ability  to  remove 

bacterial  spores  in  cold  water  (35  to  36°F).  Table  1 1  shows  the  results  obtained  without 

chlorination.  As  the  floe  bed  matured  from  the  beginning  of  the  run  through  the  sixth 

hour,  the  removal  of  spores  by  coagulation  alone  increased  from  95.45  to  97.36%.  The 

performance  of  the  filter  improved  substantially  when  20  ppm  of  dlatomite  body  feed  was 

added  to  the  filter  influent  after  3  h  of  operation.  Spore  removal  increased  from  about 

86%  without  body  feed  to  an  average  of  98.86%  with  body  feed.  The  overall  decrease  in 

3  4 

the  bacterial  spores  was  from  about  10  -  to  10  -fold  for  the  entire  unit  during  the  time 
when  body  feed  was  added  to  the  filter. 

Lindsten  and  Schmitt^  conducted  a  series  of  tests  In  1967  on  the  performance  of  a 
"standard  3000-gal/h,  ERDLATOR-type... water-purification  unit."  The  tests  were 
performed  to  allow  comparison  of  expanded  perlite  with  dlatomaceous  earth  for  use  in 
Army  filters. 


69 


Volume  8 


Table  10.  Observed  removal  of  total  bacteria  and  fecal  coliform  bacteria  by  the  MWPU 
without  chlorination  in  1952, a  (Raw-water  source:  Potomac  River  Water,  temperature  = 
56  to  88°F,  turbidity  a  28  to  1 10  ppm  during  the  testing  period.") 


Date 

Sample  identity 

Total  bacteria  count 
(number/mL) 

Fecal  coliform  bacteria 
(MPN/100  mL) 

23  May 

feed  water 

1050 

4900 

ERDLATOR  effluent 

6 

NDC 

29  May 

feed  water 

18 

610 

ERDLATOR  effluent 

NDC 

NDC 

20  June 

feed  water 

29 

5400 

ERDLATOR  effluent 

NDC 

NDC 

filtered  water 

NDC 

NDC 

1  July 

feed  water 

1300 

7900 

ERDLATOR  effluent 

2 

13 

filtered  water 

NDC 

NDC 

10  July 

feed  water 

1030 

4900 

ERDLATOR  effluent 

5 

49 

filtered  water 

1 

1.8 

7  July 

feed  water 

160 

1100 

ERDLATOR  effluent 

3 

46 

a  Adapted  from  Ruiz  and  Schmitt.23 


b  No  further  Information  Is  available  "...the  Potomac  River  Is  subject  to  wide  variations 
in  quality,  typical  of  surface  sources  draining  densely  populated  areas."23  Also,  the 
reporting  of  turbidity  in  ppm  is  as  shown  in  the  original  report. 

c  NDb  below  the  detectability  limit  of  the  analytical  method  used. 


The  removal  of  turbidity  and  total-coliform  bacteria  was  determined  for  several 
grades  of  dlatomite  and  one  grade  of  perlite.  The  tests  were  performed  daily  for  a  total 
of  50  d,  and  the  individual  test  runs  lasted  6  to  8  h  on  each  day.  The  filtration  rate  was 
held  constant  at  25  gal/min  per  filter  or  2.5  gal/(min  •  ft  ^).  The  Potomac  River  water 
used  as  the  water  source  in  the  teuts  did  not  have  to  be  fortified  with  either  turbidity  or 
coliform  bacteria.  The  general  tangos  in  two  of  the  quality  characteristics  (turbidity  and 
total-coliform  organisms)  of  the  raw  water  are  given  in  Table  i  2. 

The  raw  water  was  coagulated  but  not  chlorinated  in  the  ERDLATOR  in.most  of  the 
test  runs,  but  a  few  tests  were  made  without  the  addition  of  any  conditioning  chemical. 
The  average  daily  results  are  shown  in  Table  12  for  the  particular  test  conditions  of 


70 


Volume  8 


Tible  11.  Observed  concentration  of  the  spores  of  ft.  subtilus  var.  niger  in  the 
ERDLATOR  floe  and  effluent,  and  the  fiitered-water  effluent  of  the  MWPU.a 
(Raw-water  source:  pond  water,  temperature  =  35  to  36°F,  pH  7.2  to  7.6,  filtration 
rate  -  2.5  gai/(min  •  ft^),  FeClj  dose  =  49  ppm,  limestone  dose  =  227  ppm,  body 
feed  =  20  ppm,  feed  concentration  of  spores  -  1.8  x  lQ^/mL  on  a  continuous  basis,  no 
chlorination.) 


Time  (hours 
of  operation) 

ERDLATOR  effluent  ERDLATOR  floe  Filtered  water 

(number  of  organisms/mL)  (10^  organisms/nr.L)  (number  of  organisms/mL) 

1.5 

— 

2.75 

2.0 

— 

4.6 

__b 

2.5 

— 

5.6 

__b 

3.0 

6000 

6.7 

17 

4.0 

4600 

7 

170 

5.0 

3200 

7 

2 

6.0 

3700 

8 

11 

Si  70 

Adapted  from  Pruett  and  Lindsten. 

k  Body  feed  net  added. 


2 

diatomite  precoat  (0.10  Ib/ft  )  and  body  feed  (40  to  43  ppm)  only  and  coagulation  without 
chlorination.  The  results  were  obtained  with  three  different  grades  of  diatomite,  but  this 
grade  variation  appeared  to  be  insignificant.  The  dosages  of  FeCi^  and  limestone  listed  in 
Table  12  were  obtained  by  dividing  the  mass  rate  of  addition  of  the  chemical  to  the 
ERDLATOR  by  the  mass  flow  of  the  water  through  the  F.RDLATOR. 

Troubles  with  the  MWPU  were  encountered  on  test  days  46  and  47.  The  residual 
turbidity  in  the  filtered  water  was  considerable  and  the  observed  removal  of  the  coliform 
bacteria  was  relatively  poor.  The  presence  of  algae  was  noted  on  one  of  those  days. 
Excluding  days  46  and  47,  the  arithmetic  mean  removal  of  turbidity  was  about  72%  in  the 
ERDLATOR,  and  greater  than  99%  in  the  filters.  The  overall  reduction  in  turbidity  was 
about  500-fold  when  taken  across  the  entire  treatment  train.  The  corresponding  values 
ior  the  total  coliform  bacteria  were  about  86%  for  the  ERDLATOR,  more  than  99%  for 
the  filters,  and  300-fold  for  the  entire  system. 

No  conditioning  chemicals  were  added  to  the  ERDLATOR  on  days  51  through  53,  and 
the  raw  water  was  seeded  with  the  cysts  of  Naegleria  gruberi,  a  nonpatbogenic  cyst 
simulant  for  E.  histolytica.  The  filtered  water  was  not  analyzed  for  N.  gruberi  cysts  in 
one  of  the  three  runs,  but  no  cysts  were  recovered  in  the  remaining  two  runs  for  which  the 


Volume  8 


Table  12.  Average  daily  removal  of  turbidity  and  total  coliform  bacteria  from  Fotomac 
River  water  in  1967  with  the  standard  3000-gph  MWPU.a  (pH  =  7.7  to  8.2,  alkalinity  =  43 
to  84  mg/L  as  CaCC>3,  temperature  =  59  to  79°F,  filtration  rate  =  2.5  gal/min  •  ft2,  no 
chlorination,  diatomite  precoat  =  0.10  lb/ft2,  and  body  feed  =  40  to  43  ppm.) 


Day 

Doses  of  conditioning 
chemicals 

Turbidity  (NTU) 

Total  coliforms 
(number  /  mL ) 

FeCl  3 
(ppm) 

cacoy 

(ppm) 

Raw  ERDLATOR 
water  effluent 

Filter 

effluent 

Raw 

water 

""kRDLATOR 

effluent 

Filter 

effluent 

l 

25 

80 

47 

1  4 

0.2 

575 

62 

1 

2 

28 

80 

50 

1 1 

0.1 

1150 

69 

1 

3 

28 

80 

46 

12 

0 

1000 

104 

2 

6 

43 

80 

54 

34 

0.2 

1 175 

164 

1 

7 

41 

83 

59 

\7 

0 

1925 

165 

5 

8 

43 

86 

52 

16 

0 

10700 

1070 

67 

46 

22 

60 

51 

NRC 

4 

600 

NRC 

43 

47  b 

73 

167 

46 

1  1 

2 

325 

44 

58 

48 

40 

178 

47 

7 

0 

390 

33 

10 

11 

44 

102 

59 

19 

0.1 

6525 

370 

l 

12 

32 

102 

47 

12 

0 

5500 

350 

1 

13 

30 

107 

49 

8 

0 

5625 

227 

2 

16 

23 

112 

44 

13 

0.2 

2275 

181 

l 

17 

21 

1  10 

48 

10 

0.1 

2  425 

184 

3 

18 

21 

84 

52 

10 

0.2 

1900 

177 

4 

21 

34 

80 

46 

12 

0 

1000 

83 

0 

a  7  1 

Adopted  from  Lindsten  and  Schmitt. 
b  Algae  present. 
c  NR  =  not  reported. 


input  concentrations  were  3200  and  20  cysts  per  gallon.  The  Potomac  River  water  had  a 

turbidity  of  20  and  6  NTU  on  those  two  days.  The  temperature  was  not  reported.  The 

2 

diatomite  precoat  was  0.10  Ib/ft  and  the  body  feed  was  39  to  40  ppm.  It  was  concluded 

that  the  cysts  of  N_.  gruberi  could  be  effectively  removed  from  the  water  by 

sedimentation,  and  subsequent  filtration  with  body  feed  through  the  precoated  filters  at  a 

2 

filtration  rate  of  2.5  gal/(min  •  ft  ). 


72 


Volume  8 


The  MWPU  was  also  tested  under  closely  controlled  conditions  in  isolated,  restricted 
military  locations  where  full-scale  "live-agent"  tests  were  possible  —that  is,  where  the 
actual  chemical  or  nuclear  warfare  agent  was  used  rather  than  a  simulant.  The  results 
are  summarized  briefly  as  follows  for  the  nuclear  and  chemical  warfare  agents. 

NUCLEAR  WARFARE  AGENTS 

72 

The  results  of  tests  using  nuclear  warfare  agents  were  reported  by  Lindsten  et  ah  . 

♦  The  ERDLATOR  in  combination  with  the  coagulant  ferric  chloride  and  the 
limestone  water  conditioner  removed  80  to  86%  of  2-month-oid  fission  products  from 
water. 


•  The  small  amount  of  radioactive  floe  particles,  which  was  carried  over  from  the 
coagulation  step,  was  removed  by  the  diatomaceous-earth  filters. 

•  Conventional  treatment  in  the  MWPU  followed  by  post-treatment  with  mixed-bed 
ion-exchange  resin  removed  an  a
…[truncated]