You are on page 1of 332

AWWA Microbial/Disinfection

Research By-Products Research


Foundation council

Filter Operation
Effects on Pathogen
Passage __

Subject Area:
Water Treatment
Filter Operation
Effects on Pathogen
The mission of the Awwa Research Foundation is to advance the science of water to improve
the quality of life. Funded primarily through annual subscription payments from over 1,000 utili
ties, consulting firms, and manufacturers in North America and abroad, AwwaRF sponsors
research on all aspects of drinking water, including supply and resources, treatment, monitoring
and analysis, distribution, management, and health effects.
From its headquarters in Denver, Colorado, the AwwaRF staff directs and supports the efforts
of over 500 volunteers, who are the heart of the research program. These volunteers, serving on
various boards and committees, use their expertise to select and monitor research studies to ben
efit the entire drinking water community.
Research findings are disseminated through a number of technology transfer activities, includ
ing research reports, conferences, videotape summaries, and periodicals.
Filter Operation
Effects on Pathogen
Prepared by:
Peter M. Huck and Monica B. Emelko
NSERC Chair in Water Treatment, Department of Civil
Engineering, University of Waterloo, Waterloo, ON N2L 3G1

Bradley M. Coffee and Danielle D. Maurizio


Water Quality Division, Metropolitan Water District of Southern
California, 700 Moreno Avenue, La Verne, CA 91750

Charles R. O'Melia
Department of Geography and Environmental Engineering,
The Johns Hopkins University, Baltimore, MD 21218

Sponsored by:
Microbial/Disinfection By-Products Research Council

Jointly funded by:


Awwa Research Foundation
6666 West Quincy Ave.
Denver, CO 80235-3098
and
U.S. Environmental Protection Agency
Washington, DC 20460

Published by the
Awwa Research Foundation
and American Water Works Association
Disclaimer

This study was funded by the Awwa Research Foundation (AwwaRF). AwwaRF assumes no responsibility for the
content of the research study reported in this publication or for the opinions or statements of fact expressed in the report.
The mention of trade names for commercial products does not represent or imply the approval or endorsement of AwwaRF.
This report is presented solely for informational purposes.

Library of Congress Cataloging-in-Publication Data has been applied for.

Copyright 2001
by
Awwa Research Foundation
and
American Water Works Association
Printed in the U.S.A.

ISBN 1-58321-170-5

Printed on recycled paper


CONTENTS

LIST OF TABLES........................................................................................................................ vii

LIST OF FIGURES.......................................................................................................................^
FOREWORD............................................................................................................................... xxi
PREFACE..................................................................................................................................xxiii
ACKNOWLEDGMENTS.......................................................................................................... xxv

EXECUTIVE SUMMARY...................................................................................................... xxvii


CHAPTER 1: INTRODUCTION AND OBJECTIVES................................................................. 1
CHAPTER 2: BACKGROUND..................................................................................................... 5
Regulatory Background.............................................................................................................. 5
Treatment Options for Cryptosporidium .................................................................................... 6
Removal Mechanisms of Granular Media Filtration.................................................................. 7
Operational Factors Affecting Particle Removal...................................................................... 10
Surrogates for Cryptosporidium Removal During Water Treatment....................................... 17
Cryptosporidium Removal During Drinking Water Treatment................................................ 22
CHAPTER 3: METHODS AND RESEARCH PLATFORMS.................................................... 29
Experimental Design................................................................................................................. 29
Facilities Description................................................................................................................ 33
Seeding and Sampling Methodologies ..................................................................................... 41
Analytical Methods................................................................................................................... 48
CHAPTER 4: INFLUENCE OF OPERATIONAL EFFECTS ON BENCHMARK
SYSTEMS......................................................................................................................... 57
Controls.....................................................................................................................................59
Stable Filter Operation.............................................................................................................. 62
Suboptimal Coagulation............................................................................................................ 67
Ripening.................................................................................................................................... 76
Breakthrough............................................................................................................................ 88
Hydraulic Step.......................................................................................................................... 98
CHAPTER 5: MITIGATING OPERATIONAL EFFECTS....................................................... 111
Experimental Plan................................................................................................................... Ill
Mwd Pilot Plant Results ......................................................................................................... 115
MWD Demonstration Plant Results ....................................................................................... 122
Summary................................................................................................................................. 127
CHAPTER 6: INCREASING OUR UNDERSTANDING (BENCH-SCALE
INVESTIGATIONS)...................................................................................................... 131
Direct Filtration Studies.......................................................................................................... 131
Conventional Treatment Studies............................................................................................. 158
Overall Summary.................................................................................................................... 172
CHAPTER 7: ROBUSTNESS.................................................................................................... 175
Classifying Filter Performance................................................................................................ 175
Robustness Development........................................................................................................177
MWD Pilot Plant Results........................................................................................................ 184
MWD Demonstration Plant Results........................................................................................ 189
MWD Full-Scale Results ........................................................................................................ 192
Summary .................................................................................................................................195
CHAPTER 8: CONCLUSIONS..................................................................................................197
Optimized Pathogen Removal................................................................................................. 197
Breakthrough Phase................................................................................................................. 197
Suboptimal Coagulation.......................................................................................................... 198
Hydraulic Step......................................................................................................................... 198
Ripening.................................................................................................................................. 199
Preoxidation............................................................................................................................ 199
Robustness............................................................................................................................... 199
Bench-Scale Studies................................................................................................................200
CHAPTER 9: REGULATORY IMPLICATIONS .....................................................................201
Optimized Filtration................................................................................................................ 201
Credit for Removals Above a Certain Minimum Level..........................................................201
Individual Filters.....................................................................................................................201
Pre-oxidation........................................................................................................................... 202
CHAPTER 10: GUIDANCE TO UTILITIES ............................................................................203
Suboptimal Coagulation..........................................................................................................203
Breakthrough Phase................................................................................................................. 203
Ripening.................................................................................................................................. 204
Hydraulic Step.........................................................................................................................204
APPENDIX A: RECOVERY DATA - CRYPTOSPORIDIUM PARVUM AND GIARDIA
LAMBLIA ........................................................................................................................207

APPENDIX B: DETAILED TASK 2 DATA AND TABLES ...................................................213

APPENDIX C: CLEAN BED FILTER PERFORMANCE - THE JOHNS HOPKINS


EXPERIMENTS .............................................................................................................245

APPENDIX D: DETAILED TEST RESULTS - THE JOHNS HOPKINS EXPERIMENTS... 255

REFERENCES............................................................................................................................271

LIST OF ABBREVIATIONS..................................................................................................... 281

vi
LIST OF TABLES

Table 2.1 Operational factors and related particle removal mechanisms.................................. 10


Table 2.2 Surrogate parameters for physical removal of Cryptosporidium............................... 18
Table 2.3 Removal of Cryptosporidium by drinking water filters during stable
operation........................................................................................................................ 23
Table 3.1 Operating conditions examined during Examining Benchmark Systems
experiments.................................................................................................................... 30
Table 3.2 Rationale for selection of design of Mitigating Operational Effects
experiments.................................................................................................................... 32
Table 3.3 Design of "Mitigating Operational Effects" experiments.......................................... 33
Table 3.4 Process configuration of pilot scale plants ................................................................ 34
Table 3.5 Nominal raw water quality of pilot scale plants........................................................ 35
Table 3.6 Characteristics of Influent Model Waters Used at Johns Hopkins............................ 38
Table 3.7 Characteristics of Dismal Swamp Water Used at Johns Hopkins............................. 38
Table 3.8 Jar coagulation protocol............................................................................................. 44
Table 3.9 Seeding and sampling specifics................................................................................. 46
Table 3.10 Quality assurance and control data comparing Cryptosporidium slides read
at the University of Waterloo and the commercial laboratory used by MWD.............. 52
Table 4.1 Statistical comparisons for Task 2............................................................................. 60
Table 4.2 Summary of pilot plant seeding data during control experiments............................. 61
Table 4.3 Summary of pilot plant seeding data during stable operation................................... 64
Table 4.4 Filter performance (removal of turbidity, particles, C.parvum, B. subtilis,
E. coli, MS-2, and Giardid) during Task 2 ripening experiments................................. 84
Table 4.5 Filter performance (removal of turbidity, particles, Cryptosporidium, and
Bacillus) during Task 2 breakthrough, onset of breakthrough, and end of run
experiments.................................................................................................................... 90
Table 4.6 Filter performance (removal of turbidity, particles, Cryptosporidium, and
Bacillus) during Task 2 hydraulic step experiments...................................................... 101
Table 5.1 Summarized filter specifications at MWD pilot plant during Task 3........................ 112
Table 5.2 Summarized chemical dosages and raw water quality at MWD pilot plant
during Task 3 ................................................................................................................. 113
Table 5.3 Experimental matrix for Task 3 pilot-plant experiments at MWD............................ 114
vii
Table 5.4 Experimental matrix for Task 3 demonstration-scale experiments at MWD............ 116
Table 5.5 Summarized raw water quality for demonstration-scale experiments at MWD........ 123
Table 6.1 Effects of alum dosage on the ripening of filter performance................................... 136
Table 6.2 Effects of alum dosage on the breakthrough of filter performance........................... 137
Table 6.3 Comparison of turbidity (NTU) removal by jar test and conventional mode
filtration......................................................................................................................... 159
Table 6.4 Comparison of UVas4 absorbance (710 cm) removal by jar test and
conventional mode filtration.......................................................................................... 159
Table 7.1 Filter performance classifications.............................................................................. 176
Table 7.2 Statistical significance of robustness index for different types of
experiments at MWD pilot plant.................................................................................... 187
Table 7.3 Statistical significance of robustness index for effects of preoxidation
at MWD pilot plant........................................................................................................ 187
Table 7.4 Statistical significance of robustness index for effects of filter aid at
MWD pilot plant............................................................................................................ 188
Table 7.5 Statistical significance of robustness index comparing dual- and tri-media
filters at MWD pilot plant.............................................................................................. 188
Table 10.1 Qualitative Importance of Filter Operational Effects.............................................. 205

Table A. 1 Cryptosporidium recovery data from MWD............................................................ 208


Table A. 2 Giardia recovery data from MWD.......................................................................... 209
Table A. 3 Cryptosporidium recovery data from the University of Waterloo - Ottawa water.. 210
Table A. 4 Cryptosporidium recovery data from the University of Waterloo - MWD water... 211

Table B. 1 Experimental schedule at Ottawa pilot plant........................................................... 214


Table B. 2 Ottawa pilot plant water temperature and filter headloss, turbidity, and particle
counts............................................................................................................................. 215
Table B. 3 Ottawa pilot plant filter removal of turbidity and particles...................................... 216
Table B. 4 Ottawa pilot plant filter removal ofC.parvum and B. subtilis. ............................... 220
Table B. 5 Summary of Ottawa pilot plant seeding data........................................................... 225
Table B. 6 Experimental schedule at MWD pilot plant............................................................. 227
Table B. 7 MWD pilot plant influent water quality................................................................... 228
Table B. 8 MWD pilot plant filter headloss, turbidity, and particle counts............................... 229
viii
Table B. 9 MWD pilot plant filter removal of turbidity and particles....................................... 231
Table B. 10 MWD pilot plant filter removal of C. parvum and B. subtilis ............................... 235
Table B. 11 MWD pilot plant filter removal of E. coli, MS-2 and G. lamblia.......................... 239
Table B. 12 Summary of MWD pilot plant seeding data.......................................................... 243

IX
LIST OF FIGURES

Figure 1.1 Conceptual Representation of the Relationship between Pathogen Passage


and Operational Parameters Such as Turbidity and Particle Counts............................. 2
Figure 3.1 Microorganism losses to seeding apparatus and equipment during no
coagulant and no media control experiments (mean standard deviation.................... 48
Figure 4.1 Results of control experiments without coagulant (with and without
filter media).................................................................................................................... 62
Figure 4.2 Effect of coagulation on removal of C. parvum by filters........................................ 71
Figure 4.3 Effect of coagulation on removal of B. subtilis by filters......................................... 71
Figure 4.4 Effect of coagulation on removal of particles by filters........................................... 72
Figure 4.5 Cryptosporidium removal vs. filter effluent turbidity (optimal
coagulation conditions).................................................................................................. 74
Figure 4.6 Cryptosporidium removal vs. filter effluent turbidity (suboptimal
coagulation conditions).................................................................................................. 74
Figure 4.7 Effect of various 'no coagulant'conditions on removal of C. parvum
by filters in Ottawa......................................................................................................... 75
Figure 4.8 Effect of various 'no coagulant'conditions on removal of B. subtilis by
filters in Ottawa.............................................................................................................75
Figure 4.9 Turbidity and particle response of filter during ripening experiment
on November 10, 1998 at Ottawa pilot plant................................................................. 78
Figure 4.10 Particle and C. parvum response of filter during ripening experiment
on November 10, 1998 at Ottawa pilot plant................................................................. 79
Figure 4.11 Particle and B. subtilis response of filter during ripening experiment
on November 10, 1998 at Ottawa pilot plant................................................................. 79
Figure 4.12 Turbidity response of filter during ripening experiment on
October 20, 1998 at MWD pilot plant........................................................................... 81
Figure 4.13 Particle response of filter during ripening experiment on
October 20,1998 at MWD pilot plant........................................................................... 81
Figure 4.14 Effect of ripening on removal of C. parvum by filters........................................... 83
Figure 4.15 Effect of ripening on removal of B. subtilis by filters............................................ 83
Figure 4.16 Effect of ripening on removal of particles by filters.............................................. 85
xi
Figure 4.17 Turbidity and particle response of filter during onset of
breakthrough experiment on January 21,1999 at Ottawa pilot plant........................... 92
Figure 4.18 Particle and microorganism response of filter during onset of
breakthrough experiment on January 21,1999 at Ottawa pilot plant........................... 92
Figure 4.19 Turbidity and particle response of filter during breakthrough experiment
on December 9,1998 at Ottawa pilot plant................................................................... 94
Figure 4.20 Particle and microorganism response of filter during
breakthrough experiment on December 9,1998 at Ottawa pilot plant.......................... 94
Figure 4.21 Effect of breakthrough, onset of breakthrough, and end of run
conditions on removal of Cryptosporidium by filters.................................................... 96
Figure 4.22 Effect of breakthrough, onset of breakthrough, and end of run
conditionsonremovalof Bacillus by filters.................................................................. 96
Figure 4.23 Effect of breakthrough, onset of breakthrough, and end of run
conditions on removal of particles by filters.................................................................. 97
Figure 4.24 Turbidity and particle response of filter during hydraulic step
experiment on June 7,1999 at Ottawa pilot plant......................................................... 102
Figure 4.25 Particle and microorganism response of filter during hydraulic
step experiment on June 7,1999 at Ottawa pilot plant.................................................. 102
Figure 4.26 Turbidity and particle response of filter during hydraulic step
experiment on June 15, 1999 at Ottawa pilot plant....................................................... 103
Figure 4.27 Particle and microorganism response of filter during hydraulic
step experiment on June 15,1999 at Ottawa pilot plant................................................ 103
Figure 4.28 Turbidity response of filter during hydraulic step experiment on
April 6,1999 at MWD pilot plant.................................................................................. 105
Figure 4.29 Particle response of filter during hydraulic step experiment on
April 6,1999 at MWD pilot plant.................................................................................. 105
Figure 4.30 Effect of hydraulic step on removal of Cryptosporidium by filters....................... 108
Figure 4.31 Effect of hydraulic step on removal of Bacillus by filters...................................... 108
Figure 4.32 Effect of hydraulic step on removal of particles by filters..................................... 109
Figure 5.1 Schematic of MWD pilot plant used during Task 3 experiments ............................ 112
Figure 5.2 Effect of preoxidation on median filter effluent turbidity at MWD's pilot plant
under optimized and suboptimal coagulation conditions for dual-media filters
(permanganate turbidity/TOC test used 6.5 mg/L humic acid)..................................... 117
xii
Figure 5.3 Effect of preoxidation on median filter effluent particle counts from the
dual-media filters at MWD's pilot plant under optimized and suboptimal
coagulation (permanganate turbidity/TOC test used 6.5 mg/L humic acid).................. 118
Figure 5.4 Effect of preoxidation on median filter effluent particle counts from the tri-
media filters at MWD's pilot plant under optimized and suboptimal coagulation
(permanganate turbidity/TOC test used 6.5 mg/L humic acid)..................................... 118
Figure 5.5 Effect of filter aid on median filter effluent turbidity at MWD's pilot plant
under optimized and suboptimal coagulation conditions............................................... 120
Figure 5.6 Effect of filter aid on median filter effluent particles at MWD's pilot plant
under optimized and suboptimal coagulation conditions............................................... 120
Figure 5.7 Effect of media configuration on median filter effluent turbidity at MWD's
pilot plant under optimized and suboptimal coagulation conditions............................. 121
Figure 5.8 Effect of media configuration on median filter effluent particles at MWD's
pilot plant under optimized and suboptimal coagulation conditions............................. 121
Figure 5.9 Effect of coagulation conditions on median filter effluent turbidity for the
dual-media filter at MWD's demonstration plant.......................................................... 124
Figure 5.10 Effect of coagulation conditions on median filter effluent particles for the
dual-media filter at MWD's demonstration plant ......................................................... 124
Figure 5.11 Effect of higher coagulant dosages on median filter effluent particles for the
dual-media filter at MWD's demonstration plant (cationic polymer
dosage = 1.0 mg/L)........................................................................................................ 126
Figure 5.12 Effect of oxidant on median filter effluent particles for the dual-media filter
at MWD's demonstration plant (cationic polymer dosage = 1 mg/L)............................ 126
Figure 5.13 Effect of filter media configuration on the 90th percentile effluent turbidity at
MWD's demonstration plant.......................................................................................... 127
Figure 5.14 Effect of treatment options on median filter effluent particles for baseline
(stable filter operation) experiments at the MWD pilot plant........................................ 128
Figure 5.15 Effect of treatment options on median filter effluent particles for suboptimal
coagulation experiments at the MWD pilot plant.......................................................... 129
Figure 6.1 Effect of alum dosage on headloss development by the anthracite layer during
a filter run (direct filtration mode, filtration rate = 4.9 m/h [2 gpm/ft2])...................... 133
Figure 6.2 Effect of alum dosage on headloss development by the entire dual media bed
during a filter run (direct filtration mode, filtration rate = 4.9 m/h [2 gpm/ft2])........... 133
xiii
Figure 6.3 Turbidity of the anthracite and dual media effluents during a filter
run at an alum dosage of 4.5 mg/L (direct filtration mode, filtration rate =
4.9.m/h[2gpm/ft2])................_
Figure 6.4 Turbidity of the anthracite and dual media effluents during a filter
run at an alum dosage of 6.8 mg/L (direct filtration mode, filtration rate =
4.9 m/h [2 gpm/ft2]).....»................................................................................................ 134
Figure 6.5 Turbidity of the anthracite and dual media effluents during a filter
run at an alum dosage of 9 mg/L (direct filtration mode, filtration rate =
4.9 m/h [2 gpm/ft2])....................................................................................................... 135
Figure 6.6 Turbidity of the anthracite and dual media effluents during a filter
run at an alum dosage of 13.5 mg/L (direct filtration mode, filtration rate =
4.9 m/h [2 gpm/ft2]).................................,..................................................................... 135
Figure 6.7 Comparison of particle concentrations and turbidity of the filter
effluent during a filter run (direct filtration mode, filtration rate =
4.9 m/h [2 gpm/ft2], alum = 4.5 mg/L)..................................,............,.......................... 138
Figure 6.8 Comparison of particle concentrations and turbidity in the filter
effluent during a filter run (direct filtration mode, filtration rate =
4.9 m/h [2 gpm/ft2], alum = 6.8 mg/L).......................................................................... 138
Figure 6.9 Comparison of particle concentrations and turbidity in the filter
effluent during a filter run (direct filtration mode, filtration rate =
4.9 m/h [2 gpm/ft2], alum = 9 mg/L)............................................................................. 139
Figure 6.10 Comparison of particle concentrations and turbidity in the filter
effluent during a filter run (direct filtration mode, filtration rate =
4.9 m/h [2 gpm/ft2], alum = 13.5 mg/L)................,....................................................... 139
Figure 6.11 Headless development across the anthracite layer and the entire
dual media filter during a filter run at an alum dosage of 13.5 mg/L
(direct filtration mode, filtration rate = 4.9 m/h [2 gpm/ft2])........................................ 141
Figure 6.12 Comparison of turbidity and particle concentration of dual media
effluents during the ripening period at an alum dosage of 13.5 mg/L
(direct filtration mode, filtration rate = 4.9 m/h [2 gpm/ft2])........................................ 141
Figure 6.13 Comparison of turbidity and particle concentration of dual media
effluents during the breakthrough at an alum dosage of 13.5 mg/L
(direct filtration mode, filtration rate = 4.9 m/h [2 gpm/ft2])........................................ 142
xiv
Figure 6.14 Log removal of particles during a filter run at an alum dosage of
13.5 mg/L (direct filtration mode, filtration rate = 4.9 m/h [2 gpm/ft2])...................... 142
Figure 6.15 Effect of alum dosage on headless development by the anthracite layer
during a filter run (direct filtration mode, filtration rate = 14.7 m/h [6 gpm/ft2])......... 144
Figure 6.16 Effect of alum dosage on headless development by the entire dual media bed
during a filter run (direct filtration mode, filtration rate = 14.7 m/h [6 gpm/ft2])......... 146
Figure 6.17 Effect of alum dosage on turbidity removal by the anthracite layer during a
filter run (direct filtration mode, filtration rate = 14.7 m/h [6 gpm/ft2])....................... 146
Figure 6.18 Effect of alum dosage on turbidity removal by the entire dual media bed
during a filter run (direct filtration mode, filtration rate = 14.7 m/h [6 gpm/ft2])......... 147
Figure 6.19 Effect of alum dosage on the removal of particles by the entire
dual media bed during a filter run (direct filtration mode, filtration rate =
14.7 m/h [6 gpm/ft2])..................................................................................................... 149
Figure 6.20 Effect of alum dosage on the log removal of particles by the entire
dual media bed during a filter run (direct filtration mode, filtration rate =
14.7 m/h [6 gpm/ft2])..................................................................................................... 149
Figure 6.21 Effect of alum dosage on the ripening of turbidity by the entire
dual media (direct filtration mode, filtration rate = 14.7 m/h [6 gpm/ft2])................... 151
Figure 6.22 Effect of alum dosage on the ripening of particles by the entire dual
media bed (direct filtration mode, filtration rate = 14.7 m/h [6 gpm/ft2]).................... 151
Figure 6.23 Effects of pH on headloss development by the anthracite layer during a
filter run (direct filtration mode, filtration rate = 4.9 m/h [2 gpm/ft2])......................... 153
Figure 6.24 Effects of pH on headloss development by the entire dual media bed
during a filter run (direct filtration mode, filtration rate = 4.9 m/h [2 gpm/ft2])........... 153
Figure 6.25 Effects of pH on turbidity removal by the anthracite layer during a filter
run (direct filtration mode, filtration rate = 4.9 m/h [2 gpm/ft2]).................................. 154
Figure 6.26 Effects of pH on turbidity removal by the entire dual media bed
during a filter run (direct filtration mode, filtration rate = 4.9 m/h [2 gpm/ft2])........... 154
Figure 6.27 Effects of pH on the removal of particles by the entire dual media bed
during a filter run (direct filtration mode, filtration rate = 4.9 m/h [2 gpm/ft2])........... 155
Figure 6.28 Effects of pH on the log removal of particles by the entire dual media
bed during a filter run (direct filtration mode, filtration rate = 4.9 m/h [2 gpm/ft2]).... 155

xv
Figure 6.29 Effects of pH on the ripening of turbidity by the entire dual media
(direct filtration mode, filtration rate = 4.9 m/h [2 gpm/ft2])........................................ 156
Figure 6.30 Effects of pH on the ripening of particles by the entire dual media bed
(direct filtration mode, filtration rate = 4.9 m/h [2 gpm/ft2])........................................ 157
Figure 6.31 Effects of alum dosage on headloss development by the anthracite
layer during a filter run (conventional treatment mode, filtration rate =
4.9 m/h [2 gpm/ft2])....................................................................................................... 161
Figure 6.32 Effects of alum dosage on headloss development by the entire
dual media bed during a filter run (conventional treatment mode,
filtration rate = 4.9 m/h [2 gpm/ft2])............................................................................. 161
Figure 6.33 Effects of alum dosage on turbidity removal by the anthracite layer
during a filter run (conventional treatment mode, filtration rate =.
4.9 m/h [2 gpm/ft2])....................................................................................................... 162
Figure 6.34 Effects of alum dosage on turbidity removal by the entire dual media
bed during a filter run (conventional treatment mode, filtration rate =
4.9 m/h [2 gpm/ft2])....................................................................................................... 162
Figure 6.35 Effects of alum dosage on the removal of particles by the entire dual
media bed during a filter run (conventional treatment mode, filtration rate =
4.9 m/h [2 gpm/ft2])....................................................................................................... 164
Figure 6.36 Effects of alum dosage on the log removal of particles by coagulation
and the entire dual media bed during a filter run (conventional treatment mode,
filtration rate = 4.9 m/h [2 gpm/ft2])............................................................................. 164
Figure 6.37 Effects of alum dosage on the ripening of turbidity by the entire dual media
(conventional treatment mode, filtration rate = 4.9 m/h [2 gpm/ft2])........................... 166
Figure 6.38 Effects of alum dosage on the ripening of particles by the entire dual media
bed (conventional treatment mode, filtration rate = 4.9 m/h [2 gpm/ft2])..................... 166
Figure 6.39 Effects of alum dosage on headloss development by the anthracite
layer during a filter run (conventional treatment mode, filtration rate =
14.7 m/h [6 gpm/ft2])..................................................................................................... 167
Figure 6.40 Effects of alum dosage on headloss development by the entire dual
media bed during a filter run (conventional treatment mode, filtration rate =
14.7 m/h [6 gpm/ft2])..................................................................................................... 167

xvi
Figure 6.41 Effects of alum dosage on turbidity removal by the anthracite layer
during a filter run (conventional treatment mode, filtration rate =
14.7 m/h [6 gpir^ft2])..............................................................:...................................... 168
Figure 6.42 Effects of alum dosage on turbidity removal by the entire dual media
bed during a filter run (conventional treatment mode, filtration rate =
14.7 m/h [6 gpm/ft2])..................................................................................................... 168
Figure 6.43 Effects of alum dosage on the removal of particles by the entire dual
media bed during a filter run (conventional treatment mode, filtration rate =
14.7 m/h [6 gpm/ft2])..................................................................................................... 169
Figure 6.44 Effects of alum dosage on the log removal of particles by coagulation
and the entire dual media bed during a filter run (conventional treatment mode
filtration rate = 14.7 m/h [6 gpm/ft2])........................................................................... 169
Figure 6.45 Effects of alum dosage on the ripening of turbidity by the entire dual
media (conventional treatment mode, filtration rate = 14.7 m/h [6 gpm/ft2])............... 170
Figure 6.46 Effects of alum dosage on the ripening of particles by the entire dual media
bed (conventional treatment mode, filtration rate = 14.7 m/h [6 gpm/ft2])................... 171
Figure 7.1 Representation of the filtration system robustness concept..................................... 179
Figure 7.2 Time-series data from pilot plant comparing use of ozone as a
pre-oxidant..................................................................................................................... 179
Figure 7.3 Percent occurrence for particle data from pilot plant............................................... 180
Figure 7.4 Normalized percentage of particles passing through MWD pilot plant
filters (data normalized such that 100 percent equals the total particles from the
filter without ozone)....................................................................................................... 181
Figure 7.5 Time-series data for turbidity and particles from stable filter operation
test at MWD pilot plant................................................................................................. 185
Figure 7.6 Time-series data for turbidity and particles from suboptimal coagulation
test at MWD pilot plant................................................................................................. 185
Figure 7.7 Box-and-whisker plot of turbidity and particle robustness indices comparing
the different types of experiments at MWD pilot plant................................................. 186
Figure 7.8 The effect of coagulation on turbidity robustness index for at MWD's
demonstration-scale treatment plant (dual-media filters operated at
3.5 gpm/ft2; chlorine added as a preoxidant)................................................................. 190

xvn
Figure 7.9 The effect of coagulation on particle robustness index at MWD's
demonstration-scale treatment plant (dual-media filters operated at 3.5 gpm/ft2;
chlorine added as a preoxidant).....................................................;............................... 190
Figure 7.10 The effect of preoxidation, coagulant dose, and media configuration on the
turbidity robustness index at MWD's demonstration-scale treatment plant
(1 mg/L polymer present).............................................................................................. 191
Figure 7.11 The effect of preoxidation, coagulant dose, and media configuration on the
particle robustness index at MWD's demonstration-scale treatment plant
(1 mg/L polymer present).............................................................................................. 191
Figure 7.12 Influent turbidity and coagulant dosages for full-scale testing at MWD's
F.E. Weymouth filtration plant...................................................................................... 193
Figure 7.13 Combined filter effluent turbidity and filter effluent particle counts from a
single filter at MWD's F. E. Weymouth filtration plant................................................ 193
Figure 7.14 Particle robustness index (PRI90) for two filters at MWD's F. E. Weymouth
filtration plant................................................................................................................ 194
Figure 7.15 Modified particle robustness index (PRIgo/goai) for two filters at MWD's F. E.
Weymouth filtration plant.............................................................................................. 195

Figure C.I Clean bed performance of the dual media filter under standard conditions............ 251
Figure C.2 Effects of pretreatment chemistry and filtration rate on clean bed filter
performance...................................................................................................................252
Figure C.3 Effects of filtration rate on clean bed filter efficiency............................................. 253

Figure D. 1 Jar test results of settled turbidity at pH 7 (raw water: turbidity = 7 NTU,
DOC = 1.5 mg/L)........................................................................................................... 256
Figure D.2 Jar test results of settled UV254 Absorbance at pH 7 (raw water: turbidity =
7 NTU, DOC = 1.5 mg/L).............................................................................................. 256
Figure D.3 UV254 absorbance of the dual media effluents during a filter run
(direct filtration mode, filtration rate = 4.9m/h [2 gpm/ft2])......................................... 257
Figure D.4 UV254 absorbance of the anthracite and dual media effluents during a
filter run at an alum dosage of 13.5 mg/L (direct filtration mode, filtration rate =
4.9 m/h [2 gpm/ft2])....................................................................................................... 257

XVlll
Figure D.5 Effect of alum dosage on the removal of UV254 absorbance by
anthracite layer during a filter run (direct filtration mode, filtration rate =
14.7 m/h [6 gpm/ft2])..................................................................................................... 258
Figure D.6 Effect of alum dosage on the removal of UVa54 absorbance by the entire dual
media bed during a filter run (direct filtration mode, filtration rate =
14.7 m/h [6 gpm/ft2])..................................................................................................... 258
Figure D.7 Effect of alum dosage on the removal of DOC by anthracite layer during a
filter run (direct filtration mode, filtration rate = 14.7 m/h [6 gpm/ft2])....................... 259
Figure D.8 Effect of alum dosage on the removal of DOC by the entire dual media bed
during a filter run (direct filtration mode, filtration rate - 14.7 m/h [6 gpm/ft2])......... 259
Figure D.9 Jar test results for the effects of pH on settled turbidity (raw water: turbidity =
7 NTU, DOC = 1.5 mg/L).............................................................................................. 260
Figure D.10 Jar test results for the effects of pH on filtered turbidity (raw water: turbidity
= 7 NTU, DOC = 1.5 mg/L).......................................................................................... 260
Figure D.I 1 Jar test results for the effects of pH on settled UV254 (raw water: turbidity
= 7 NTU, DOC = 1.5 mg/L).......................................................................................... 261
Figure D.I 2 Jar test results for the effects of pH on filtered UV254 (raw water: turbidity
= 7 NTU, DOC = 1.5 mg/L).......................................................................................... 261
Figure D.I 3 Effects of pH on the removal of UV254 absorbance by anthracite layer during
a filter run (direct filtration mode, filtration rate = 4.9 m/h [2 gpm/ft2])...................... 262
Figure D.I 4 Effects of pH on the removal of UV254 absorbance by the entire dual media
bed during a filter run (direct filtration mode, filtration rate = 4.9 m/h [2 gpm/ft2]).... 262
Figure D. 15 Effects of pH on the removal of DOC by anthracite layer during a filter run
(direct filtration mode, filtration rate = 4.9 m/h [2 gpm/ft2])........................................ 263
Figure D.I 6 Effects of pH on the removal of DOC by the entire dual media bed during a
filter run (direct filtration mode, filtration rate = 4.9 m/h [2 gpm/ft2])......................... 263
Figure D.I 7 Jar test results for the effects of alum dosage on settled turbidity (raw water:
turbidity - 7 NTU, DOC = 5 mg/L)............................................................................... 264
Figure D.I 8 Jar test results for the effects of alum dosage on settled UV254 (raw water:
turbidity - 7 NTU, DOC = 5 mg/L)............................................................................... 264
Figure D.I 9 Jar test results for the effects of alum dosage on filtered turbidity (raw water:
turbidity = 7 NTU, DOC = 5 mg/L)............................................................................... 265

xix
Figure D.20 Jar test results for the effects of alum dosage on filtered UV254 (raw water:
turbidity = 7 NTU, DOC = 5 mg/L)............................................................................... 265
Figure D.21 Jar test results for the effects of alum dosage on dissolved organic carbon
(raw water: turbidity = 7 NTU, DOC = 5 mg/L)........................................................... 266
Figure D.22 Effects of alum dosage on the removal of UV254 absorbance by anthracite
layer during a filter run (conventional treatment mode, filtration rate
= 4.9 m/h [2 gpm/ft2])................................................................................................... 266
Figure D.23 Effects of alum dosage on the removal of UV254 absorbance by the entire
dual media bed during a filter run (conventional treatment mode, filtration rate
= 4.9 m/h [2 gpm/ft2])................................................................................................... 267
Figure D.24 Effects of alum dosage on the removal of DOC by anthracite layer during a
filter run (conventional treatment mode, filtration rate = 4.9 m/h [2 gpm/ft2])............ 267
Figure D.25 Effects of alum dosage on the removal of DOC by the entire dual media bed
during a filter run (conventional treatment mode, filtration rate
= 4.9 m/h [2 gpm/ft2])................................................................................................... 268
Figure D.26 Effects of alum dosage on the removal of UV254 absorbance by anthracite
layer during a filter run (conventional treatment mode, filtration rate
= 14.7 m/h [6 gpm/ft2])................................................................................................. 268
Figure D.27 Effects of alum dosage on the removal of UV254 absorbance by the entire
dual media bed during a filter run (conventional treatment mode, filtration rate
= 14.7 m/h [6 gpm/ft2])................................................................................................. 269
Figure D.28 Effects of alum dosage on the removal of DOC by anthracite layer during a
filter run (conventional treatment mode, filtration rate = 14.7 m/h [6 gpm/ft2]............ 269
Figure D.29 Effects of alum dosage on the removal of DOC by the entire dual media bed
during a filter run (conventional treatment mode, filtration rate
= 14.7 m/h [6 gpm/ft2])................................................................................................. 270

xx
FOREWORD

The Awwa Research Foundation is a nonprofit corporation that is dedicated to the


implementation of a research effort to help utilities respond to regulatory requirements and
traditional high-priority concerns of the industry. The research agenda is developed through a
process of consultation with subscribers and drinking water professionals. Under the umbrella of
a Strategic Research Plan, the Research Advisory Council prioritizes the suggested projects
based upon current and future needs, applicability, and past work; the recommendations are
forwarded to the Board of Trustees for final selection. The foundation also sponsors research
projects through the unsolicited proposal process; the Collaborative Research, Research
Applications, and Tailored Collaboration programs; and various joint research efforts with
organizations such as the U.S. Environmental Protection Agency, the U.S. Bureau of
Reclamation, and the Association of California Water Agencies.

This publication is a result of one of these sponsored studies, and it is hoped that its
findings will be applied in communities throughout the world. The following report serves not
only as a means of communicating the results of the water industry's centralized research
program but also as a tool to enlist the further support of the nonmember utilities and individuals.

Projects are managed closely from their inception to the final report by the foundation's
staff and large cadre of volunteers who willingly contribute their time and expertise. The
foundation serves a planning and management function and awards contracts to other institutions
such as water utilities, universities, and engineering firms. The funding for this research effort
comes primarily from the Subscription Program, through which water utilities subscribe to the
research program and make an annual payment proportionate to the volume of water they deliver
and consultants and manufacturers subscribe based on their annual billings. The program offers a
cost-effective and fair method for funding research in the public interest.

A broad spectrum of water supply issues is addressed by the foundation's research


agenda: resources, treatment and operations, distribution and storage, water quality and analysis,
toxicology, economics, and management. The ultimate purpose of the coordinated effort is to
assist water suppliers to provide the highest possible quality of water economically and reliably.
xxi
The true benefits are realized when the results are implemented at the utility level. The
foundation's trustees are pleased to offer this publication as a contribution toward that end.

Water utilities are becoming increasingly concerned about the ability of their water
treatment processes to remove microbial contaminants. With this concern comes the question of
what affects the passage of pathogens through filters. This report describes research to determine
the impacts of filter design and operational parameters on pathogen removal. The researchers
performed bench- and pilot-scale experiments to evaluate the phases of the filter cycle (e.g.,
ripening, normal operation, and breakthrough) as well as the impact of suboptimal events (e.g.,
coagulation upsets, hydraulic surges, and sudden water quality changes) on the passage of
pathogens through filters.

Edmund G. Archuleta, P.E. James F. Manwaring, P.E.


Chair, Board of Trustees Executive Director
Awwa Research Foundation Awwa Research Foundation

xxn
PREFACE

The Microbial/Disinfection By-Products Research Council was established in 1995 as a


vehicle for the selection and funding of research to provide scientific information in the areas of
health effects, exposure assessment, risk assessment, and prevention and control of
contamination by microbes and disinfection by-products in drinking water. The council is
composed of representatives designated by the U.S. Environmental Protection Agency (USEPA),
the Awwa Research Foundation (AwwaRF) Board of Trustees, the Association of State Drinking
Water Administrators, the National Resources Defense Council, the National Environmental
Health Association, or their designees. Sources of funding for this research include the USEPA
and AwwaRF, along with other interested parties. The council disburses these funds for research
deemed to be of the highest urgency and importance in resolving critical research issues in
drinking water.

xxm
ACKNOWLEDGMENTS

The authors of this report are indebted to the following water utilities and individuals for
their cooperation and participation in this project:

University of Waterloo, Waterloo, Ont.:


W.B. Anderson, L.M. Collins, S. Giuseppin, A.M. Semple, R.M. Slawson,
J.L. Zimmer
Metropolitan Water District, La Verne, Calif:
H. Baribeau, R.A. Becerra, A.M. Chitre, R. DeLeon, C.C.P. Huang, S.
Liang, H.M. Mehta, K.N. Scott, M.C. Simpson, M. Tan, F.A. Vergel de Dios, H.
Wen
The Johns Hopkins University, Baltimore, Md.:
J.-Y. Shin
Regional Municipality of Ottawa-Carleton, Ottawa, Ont.:
I.P. Douglas, A. Proulx, J. Van Den Oever

The project benefited greatly from the input of three technical advisors. They were Dr.
William D. Bellamy (CH2MHill, Englewood, Colo.), Dr. Gary S. Logsdon (Black & Veatch,
Cincinnati, Ohio), and Dr. Eva C. Nieminski (State of Utah, Department of Environmental
Quality, Salt Lake City, Utah). The project team gratefully acknowledges the valuable input
provided by these advisors. The team also acknowledges contributions from K.L. Edwards, W.
Miller, S.Y. Jasim and D. Guerrieri of the Windsor Utilities Commission, Windsor, Ont.

In-kind support for this project was provided by the Natural Sciences and Engineering
Research Council of Canada (NSERC) in the form of a Collaborative Research and Development
(CRD) grant and an Industrial Research Chair at the University of Waterloo. The current
Chair/CRD partners include: NSERC, the Regional Municipality of Waterloo, the Region of
Ottawa-Carleton, the Windsor Utilities Commission, the Ontario Clean Water Agency (OCWA),
the City of Brantford, the City of Toronto, Conestoga-Rovers & Associates Limited, RAL
Engineering Ltd., Stantec Consulting Ltd., EPCOR Water Services, PICA USA Inc., Azurix
North America and the University of Waterloo.

xxv
The authors would also like to acknowledge the support and valuable input of the
AwwaRF Project Advisory Committee: T. Bavan (City of Atlanta, Atlanta, Ga.), W. Becker
(Becker & O'Melia, Mt. Airy, Md.), K. R. Fox (USEPA, Cincinnati, Ohio), M. Hotaling
(Newport News Waterworks, Yorktown, Va.) and N. Pizzi (EE&T, Inc., Twinsburg, Ohio) and
of AwwaRF Project Manager Kathryn Martin.

xxvi
EXECUTIVE SUMMARY

INTRODUCTION AND OBJECTIVES

The primary goal of drinking water supply systems is to protect public health by
providing water that is free of microbial and chemical contaminants. The emergence of parasitic
protozoa such as Giardia lamblia and Cryptosporidium parvum as etiological agents of
waterborne disease has prompted renewed interest in evaluating the efficacy of water treatment
processes. Increasingly stringent regulations for drinking water quality, including the Interim
Enhanced Surface Water Treatment Rule (IESWTR), will require effective removal of these
organisms. Although disinfection or inactivation plays a crucial role in this regard, physical
removal is also important. Stage 1 of the Disinfectants/Disinfection By-Products (D/DBP) Rule
may place constraints on the levels of inactivation that can be achieved with disinfection.
Although alternative treatment processes such as membrane filtration offer excellent removal
capabilities of parasitic pathogens, they are only beginning to become cost-effective at large
scale. Recent attention has also been focused on UV, but the use of this technology for protozoan
inactivation in drinking water is very new. A multi-barrier approach to pathogen removal implies
an increased importance of granular media filtration.

Our conceptual representation of the relationship between pathogen passage and


operational parameters such as turbidity and particle counts is presented in Figure ES.l. The left
vertical axis shows turbidity or particle counts while the right vertical axis shows pathogen
concentration in filter effluents. The horizontal axis represents the time of approximately one
filter cycle, at a variable scale to clearly depict the stages of filter ripening.

As shown in Figure ES.l, effluent turbidity must always remain below a specified goal or
standard, in this case 0.1 NTU. After backwash, both turbidity and particle counts increase
during filter ripening. The two peaks of particle passage that may occur during, ripening result
from backwash water remnants and particle non-attachment associated with the period that
particles are being retained and act as subsequent collectors. Pathogen passage through the filters
may increase during this time for the same reasons. Events such as coagulation upsets and
hydraulic surges result in suboptimal pretreatment and may consequently cause filter effluent
xxvii
——— Turbidity T
»•"•"• Pathnopnc

WD
O

Filter Cycle

Time

Figure ES. 1 Conceptual representation of the relationship between pathogen passage and
operational parameters such as turbidity and particle counts

concentrations of turbidity, particles, and pathogens to increase. As the filter becomes loaded,
particles and pathogens begin to break through the filter, initiating a backwash cycle.

Evaluating the various phases of the filter cycle (i.e. ripening, normal operation, and
breakthrough) as well as events (i.e. coagulation upsets, hydraulic surges, and sudden water
quality changes) allow for a description of the robustness of the removal process. The
development of practical design and operating guidelines for minimizing pathogen passage
during filtration is facilitated by a quantitative understanding and description of the pathogen
removal robustness of the filtration process.

The overall goal of this project was to evaluate the impact of filter design and operational
parameters on pathogen removal during drinking water treatment and to develop practical
guidelines and strategies for estimating and minimizing pathogen passage through filtration.

xxvm
Specific objectives in pursuit of this goal were:

1. To develop a succinct, comprehensive summary of the appropriate literature on


particle and pathogen removal. Special emphasis was placed on including
investigations that evaluate the effect of filter operational factors and that addressed
pathogen or surrogate removal under non-optimal conditions.

2. To assess the degree of pathogen removal that can be reasonably expected from
"benchmark" filtration systems (i.e. relatively standard design) under optimized
operation and the following conditions: sub-optimal coagulation, filter ripening,
breakthrough (end-of-run) and hydraulic surges.

3. To investigate strategies for minimizing pathogen passage through drinking water


filters. These strategies addressed design and operating considerations. Chemical
parameters such as pretreatment (coagulation) strategy and backwash water chemistry
were evaluated. Physical factors that were investigated included: media configuration,
flow rate, and direct vs. conventional filtration. The impacts of pre-oxidation and
water quality changes were also examined.

4. To develop a practical measure of a filter's robustness, i.e. its ability to consistently


maintain superior levels of pathogen removal during periods of elevated turbidity and
particle counts. This robustness measure is intended to provide utilities a common
basis for quantifying the effect on pathogen passage of different treatment
configurations and operational conditions.

METHODS AND RESEARCH PLATFORMS

Experimental Design

The research approach consisted of five specific tasks designed to provide concrete
outcomes of practical value to the water industry. These tasks were:

1. a literature review,

xxix
2. experiments investigating the influence of operational effects on pathogen passage in
benchmark systems (referred to as Examining Benchmark Systems),

3. experiments investigating strategies for mitigating filter operational effects on


pathogen passage (referred to as Mitigating Operational Effects),

4. development of a concept of robustness, and

5. synthesis of experimental outcomes into useful strategies for the water industry.

A review of the relevant filtration literature was critical at the outset of this research.
Although the filtration and particle removal literature is voluminous, limited information on
pathogen removal under conditions of process stress is available. The main objective of this
review (which is presented in Chapter 2) was to consolidate and update reviews and studies
already completed and to focus on non-steady-state and sub-optimal filtration performance and
its impact on particle and pathogen removal.

The first of the two experimental tasks defined removals of pathogens and performance
indicators (turbidity, particle counts, and aerobic spores) in benchmark systems (i.e., typical
design and operating conditions); these experiments are referred to as "Examining Benchmark
Systems." They were conducted at the Ottawa pilot plant in Canada and the Metropolitan Water
District of Southern California's (MWD) La Verne pilot plant in California. These locations
represent two basic types of coagulation: a relatively low dose optimized for particle removal
(MWD treating Colorado River water) and a relatively high dose for combined TOC and particle
removal (Ottawa River water). Inactivated Cryptosporidium parvum oocysts and pure-cultured
Bacillus subtilis spores were seeded at both locations. Inactivated Giardia lamblia, as well as
viruses (MS-2 bacteriophage) and bacteria (Escherichia coli) were also seeded at MWD in a
limited number of experiments. The experiments were designed to assess pathogen removal that
could be reasonably expected from benchmark systems and to evaluate the sensitivity of
pathogen removal during typical filter cycles and conditions of process stress. In this task no
attempt was made to improve pathogen removal or mitigate adverse conditions. Six basic
conditions were investigated during these experiments. They were:

xxx
1. controls,

2. stable filter operation,

3. sub-optimal coagulation,

4. ripening,

5. breakthrough, and

6. hydraulic step.

Several sub-conditions within suboptimal coagulation and breakthrough were


investigated. "End-of-run" experiments were performed at MWD because it was not possible to
actually achieve breakthrough. Experiments for each of the six principal conditions were
conducted at least in triplicate at each location.

The second major experimental task ("Mitigating Operational Effects") examined


strategies for reducing the influence of operational effects on the removal of pathogens. Various
physical and chemical factors affecting pre-treatment and filtration were investigated at bench-,
pilot- and demonstration-scale. These experiments were conducted with surrogates (turbidity and
particle counts) for cost reasons. A one-to-one correspondence between pathogens and surrogates
was not expected, however, practically useful relationships within a specific context (e.g.,
coagulation effects) were anticipated from the Examining Benchmark Systems experiments.

Experimental data were used to develop a quantitative framework describing the


robustness of pathogen removal by filtration; this was done to provide utilities with a tool or
framework with which to assess their own treatment processes. The study results^, were also
synthesized into recommendations of practical value to the water industry in minimizing
pathogen passage through filters.

Research Platforms and Experimental Approach

As mentioned previously, the Task 2 (Examining Benchmark Systems') experiments were


performed using two pilot-scale facilities. The locations were chosen to encompass two different

xxxi
coagulation regimes. The Ottawa location also experienced low temperatures (1 C). Each pilot
plant was operated to mimic as closely as possible the full-scale treatment plant at the same
location. In addition to the seeded organisms, turbidity, particle counts and filter hydraulic
performance were monitored in each experiment.

The filters at both pilot plants contained media depths and sizes typical of the utilities'
full-scale plants (and typical of many existing treatment plants). The operational mode chosen
was conventional treatment with dual-media filtration. Further operating details regarding the
research platforms are specified in Chapter 3. The Ottawa pilot plant used a high coagulant dose
(~40 mg/L alum and 2 mg/L activated silica) to achieve both TOC and particle removal. The
MWD pilot plant used a low coagulant dose (5 mg/L alum and 1.5 mg/L cationic polymer) for
particulate removal only. Chlorine (~2 mg/L) was added at rapid mix as a pre-oxidant at both
pilot plants. The optimized coagulation conditions were selected to meet the 0.1 NTU turbidity
goal of the Partnership for Safe Water, a voluntary treatment optimization program sponsored by
the U.S. Environmental Protection Agency and the American Water Works Association.

The pilot-scale filters in both locations were seeded with jar-coagulated suspensions of
~108 formalin-inactivated Cryptosporidium parvum and ~107 to 109 Bacillus subtilis spores.
Except for a small number of experiments, seeding of microorganisms was done into the filter
influent, using a procedure established by members of the project team in previous
investigations. The data collected for the seeding experiments consisted of replicate samples
(either four or five) taken from the filter influent and filter effluent at each location. The filter
influent and filter effluent data were collected over a one-hour period when the seed suspension
was added at the filter influent. Each of these tests was then repeated at least three times. A
single-factor, analysis of variance (ANOVA) statistical test was selected to interpret the data.

Methods

Formalin-inactivated Cryptosporidium parvum oocysts were used for seeding in both


pilot plant locations. Samples of filter influent and effluent were membrane-filtered and
enumeration was using the immunofluorescence assay (IFA). Particle counting and turbidity
measurements followed standard techniques. Further details regarding the methods and the
quality assurance/quality control program may be found in Chapter 3.
xxxii
RESULTS

The key results from the study are presented in this section. The terms
"Cryptosporidium" and "C. parvum" are used throughout this report. Both refer specifically to
Cryptosporidium parvum oocysts, but C. parvum is used to improve readability and save space.

Influence of Operational Effects on Benchmark Systems

Optimized Pathogen Removal

Under optimal operating conditions, there was a 2-log difference in Cryptosporidium


removal at two pilot plants operated to produce similar effluent turbidity (<0.1 NTU) and
particles (<20/mL). The pilot plant in Ottawa removed approximately 5 logs (even at
temperatures as low as 1 C) while that at MWD removed approximately 3 logs. The reasons for
this difference are not understood, although there were major differences in coagulation regimes
and some differences in pilot plant characteristics. Ottawa used a relatively high alum dose for
both particle and color removal, while MWD used a lower dose. Further investigations should
attempt to determine the reasons for the substantial and reproducible differences in observed
Cryptosporidium removals.

Breakthrough

The breakthrough phase at the end of the filter cycle represents a vulnerable period for
pathogen passage. A substantial deterioration of oocyst removal capability was observed even in
the early stages of breakthrough, when filter effluent particle counts had just begun to rise. At
this time turbidity had not always increased. As mentioned previously, MWD breakthrough
could not be achieved in the pilot plant for technical reasons, and the conditions tested were
referred to as "end-of-run".

Figure ES.2 shows results for breakthrough and end-of-run conditions in the form of a
box-and-whisker plot. The number of observations for each condition is indicated on the figure.
The designation "Yes" indicates that the mean log removal for that condition was statistically
different from the mean log removal at optimal conditions (a = 5%).

xxxin
Metropolitan Ottawa
/• •/
n = 32
6-

cD
6
n = 35 •

5- •5

Yes;
4- Yes; n=12 •4
n=12 , -- "
Ye s; '.
'- • n = 12 ; •3

"
• •
•2

(=:

1 • • 1

0- ————
Optimal End of Run Optimal Onset of Breakthrough
Breakthrough

Figure ES.2 Effect of breakthrough, onset of breakthrough, and end of run conditions on
removal of Cryptosporidium by filters

Because changes in turbidity may occur only after substantial deterioration of


Cryptosporidium removal, turbidity values may not be an adequate predictor of increasing oocyst
passage. Particle counts may provide a more sensitive indication of pathogen passage, but still
fall short of predicting the full extent of the deterioration of Cryptosporidium removal at the
beginning of breakthrough.

These results indicate that breakthrough, as measured by turbidity or perhaps even


particles, should be avoided. This may mean that filter rans should be terminated even earlier
than has been the practice.

In terms of explaining the observed deterioration in Cryptosporidium removal, oocyst


passage through the filter (not detachment of previously captured oocysts) appears to be an
important mechanism, however the experimental design tested passage, not detachment. It is
recommended that further investigations be conducted to assess the importance of detachment
for pathogen passage during breakthrough.
xxxiv
Suboptimal Coagulation

Suboptimal coagulation substantially deteriorated pathogen removal (Figure ES.3) - even


at turbidities which were below 0.3 NTU. This finding is of major significance for utilities.

Although turbidity may be an adequate indicator of poor coagulation performance,


particle counts provide a more sensitive indication.

As expected, the complete absence of coagulant resulted in almost no Cryptosporidium


removals (Figure ES.3). The lack of coagulant over several hours gave marked deterioration,
although some removal capability remained.

Hydraulic Step

The testing showed that changes in hydraulic loading did not deteriorate Cryptosporidium
removal, with the exception of one experiment where turbidity, particles (to some extent), and
Cryptosporidium levels all increased substantially.

Metropolitan Ottawa

n = 32 Yes; :
n=12
n = 35
DD ,

s
o
I4
£ • •
.
•5 3
•3 ••
Yes;
n=12 ;
Yes;

i, ••
n-= 12 Yes;

cb :
n=8
-

<= =
Optimal Suboptimal No Optimal Suboptimal No
Coagulants Coagulants

Figure ES.3 Effect of coagulation on removal of C. parvum by filters


xxxv
Although it is possible that the timing of the hydraulic step within the filter cycle may influence
the release of previously collected particles, the experiments were not designed to assess this.

Possibly, turbidity may be an adequate monitoring parameter for measuring significant


hydraulic events. Hydraulic step events should be investigated further to attempt to define
conditions under which they may or may not be significant in terms of pathogen passage.

Ripening

Minimal or moderate deterioration of pathogen removal - compared to suboptimal


coagulation or breakthrough - may result during filter ripening. However, if breakthrough is
avoided, ripening becomes a more significant part of the filter cycle for pathogen passage. In
addition, in filters with longer ripening periods and/or for mono-media filters, oocyst passage
during ripening may be more important than observed in the dual-media filters evaluated in this
research. In full-scale plants, oocysts present in backwash remnant water might also contribute to
increased oocyst passage through a filter during ripening. This was not evaluated in the present
study.

Mitigating Operational Effects

Pilot- and Demonstration-Scale Investigations

A number of available treatment options exist to potentially improve the benchmark


treatment systems for which results have been reported above. A number of these were
investigated in pilot plant experiments at MWD. For stable filter operation conditions, changing
or eliminating the oxidant, adding filter aid, or using tri-media filters all changed the filter
effluent turbidity by less than 0.02 NTU (as a median value). Particles (Figure ES.4) were more
sensitive to treatment alternatives and showed a significant deterioration when chlorine was
removed from the treatment scheme. Alternative oxidants, filter aid, or tri-media all exhibited a
minor influence on the median effluent particles. A similar effect was observed when treatment
changes were applied to the suboptimal coagulation tests. Thus, although pre-oxidation improved
particle control in virtually all cases, pre-oxidation cannot compensate for poor coagulation.

xxxvi
Eliminate CI2 UseO3 Use MnCV Add filter aid Use tri-media

Figure ES.4 Effect of treatment options on median filter effluent particles for baseline (stable
filter operation) experiments at the MWD pilot plant

It should be noted that these results are specific to the source water tested. The sensitivity
of coagulation, preoxidation or other design conditions may depend on influent water quality.

Demonstration-scale results confirmed the pilot results above, but also added insight into
the sensitivity of particle removal to the coagulation conditions. At demonstration-scale, the
alum and the polymer dosages were independently varied. When alum was used as a single
coagulant, particle control was more sensitive to the dosage than when alum was used in
conjunction with a polymer. These data indicate that multiple coagulants may improve treatment
robustness.

Bench-Scale Investigations

The bench scale investigations also demonstrated sensitivity to coagulant dose, with
direct filtration systems being much more sensitive than conventional systems.

In all bench scale experiments conducted, ripening of particles occurred earlier than that
of turbidity. The results suggested that turbidity might be a better or more conservative monitor
of filter performance during the ripening period.

xxxvii
The bench-scale investigations also demonstrated that in dual-media filters, the lower
sand layer serves as a second barrier for particles when the performance of the anthracite layer is
not effective.

Robustness

A simple robustness index, incorporating both average quality and some measure of
deviation from stable operation, was developed in this project. Three formulas were evaluated
for use as the robustness index. Using turbidity as the primary measurement, and choosing the
95th percentile value, the robustness index was written as follows:

T T
_2L + _5°_
T
J 50
T
•* goal

where:

TRI95 = turbidity robustness index using the 95th percentile, dimensionless

r50 , T95 = 50th and 95th percentile turbidity, NTU

Tgoal = filter turbidity goal, NTU

The "goal" term, Tgoal , represents a utility- or plant-specific performance goal for
turbidity. A lower value of the robustness index indicates a treatment process that is meeting the
water quality goal with relatively low variation.

This index was found useful for comparing treatment conditions, but its utility is quite
dependent on the strictness of the treatment goal chosen. The presence of pre-oxidation made a
greater contribution to robustness than the use of a filter aid.

REGULATORY IMPLICATIONS

This section provides comments that are intended to be helpful to those reviewing this
report to gain information that might be helpful in regulation development.

xxxv
Optimized filtration

This study clearly showed that optimized filtration is plant- and source water- specific in
terms of the performance achievable for Cryptosporidium removal. Maximizing the removal of
turbidity through filtration will maximize Cryptosporidium reduction. Avoiding high turbidity as
a result of operating events will reduce Cryptosporidium passage.

Credit for removals above a certain minimum level

This is likely to be a more important issue for Cryptosporidium than for Giardia, because
of the difficulty in Cryptosporidium inactivation. Credit for removals above a certain level must
of course be demonstrated. This investigation has demonstrated that removals can be highly
variable within a filter cycle. Filter robustness therefore becomes important. Plants employing
coagulation to remove organics may be more sensitive to coagulant dosage than those practicing
only paniculate removal. The different results obtained at the two pilot plants in this study have
shown that simply optimizing a filter for turbidity removal (<0.1 NTU) may not necessarily
optimize or maximize Cryptosporidium removal. It may therefore be inappropriate to set single
log removal values for all conventional or direct filtration processes.

Individual filters

Data from this project support individual filter monitoring (turbidity and particles).
Whether or not a 0.3 NTU turbidity standard should be applied to individual filters could not be
answered by this study. The results from this study do however indicate that a given turbidity
value in the breakthrough phase may indicate a greater potential for pathogen passage than the
same value in the ripening stage.

Pre-oxidation

Data from this project indicate that the removal of pre-oxidation should be carefully
considered for its effect on filter particulate removal performance.

xxxix
GUIDANCE TO UTILITIES

This section presents practical suggestions to aid utilities in mitigating the effects of filter
operation effects on pathogen passage.

Suboptimal Coagulation

1. Source water quality changes may have the same effect on filter performance as
suboptimal coagulation.

2. Utilities should carefully consider the effects of reduced coagulant dosages. Jar tests
and ideally pilot filter investigations should be performed to assess impacts on filter
performance and robustness prior to any full-scale implementation.

3. Multiple coagulants may improve robustness; redundant (on-line) chemical feed


systems add reliability.

4. The benefits of a pre-oxidant for mitigating potential suboptimal coagulation


incidents should be considered.

5. If a pre-oxidant is in use, a utility should proceed cautiously with removing it.

Breakthrough Phase

Given the demonstrated importance of the breakthrough phase for pathogen passage in
this investigation, it is recommended that utilities:

1. Consider using particle counters on selected filters to monitor early breakthrough.

2. Specify maximum headless and maximum filter run time to wash before
breakthrough.

3. Weigh power and labor savings of regularly scheduled backwash against water
quality.

4. Stagger backwashes on time.

xl
5. Consider monitoring within the filter bed to provide early warning of breakthrough.
(Additional work would be required to demonstrate the feasibility and usefulness of
this approach at full-scale.)

6. Consider operator training and manager education (within the context of global
improvement).

Breakthrough mitigation is likely to be highly plant-specific.

Ripening

To mitigate effects of ripening it is recommended that utilities:

1. Carefully adjust coagulants to minimize spiking (i.e., watch out for suboptimal
operation).

2. Use a turbidimeter as a conservative measure of performance during ripening.

3. Consider alternate media designs (perhaps dual media) that may minimize ripening.

4. Consider adding coagulants to backwash (or filter influent).

It is noted that ripening is essentially unavoidable, absent filter-to-waste. Utilities may


therefore wish to install filter-to-waste capability, if this is not already present.

Although not investigated in this study, discussion within the project team identified two
possible additional strategies for mitigating ripening that utilities might consider. These were to
consider resting filters following backwashing and to ensure that backwashing is optimized.
Although cleaning must be sufficient to avoid operational problems such as mudballs, it has been
hypothesized that cleaning that is too thorough may impair performance during the ripening
period.

Hydraulic Step

To mitigate the possible effects of hydraulic steps (sudden increases in flow) it is


recommended that utilities:

xli
1. Evaluate the effect of backwashing on the occurrence of hydraulic spikes.

2. Consider the trade-off between starting off-line filters and imposing a hydraulic
increase on on-line filters.

3. Control plant influent flows as much as possible to minimize hydraulic changes to the
filters.

xlii
CHAPTER 1: INTRODUCTION AND OBJECTIVES

The primary goal of drinking water supply systems is to protect public health by
providing water that is free of microbial and chemical contaminants. The emergence of parasitic
protozoa such as Giardia lamblia and Cryptosporidium parvum as etiological agents of
waterborne disease has prompted renewed interest in evaluating the efficacy of water treatment
processes. Increasingly stringent regulations for drinking water quality, including the Interim
Enhanced Surface Water Treatment Rule (D3SWTR), will require effective removal of these
organisms. Although disinfection or inactivation plays a crucial role in this regard, physical
removal is also important. Stage 1 of the Disinfectants/Disinfection By-Products (D/DBP) Rule
may place constraints on the levels of inactivation that can be achieved with disinfection.
Although alternative treatment processes such as membrane filtration offer excellent removal
capabilities of parasitic pathogens, they are only beginning to become cost-effective at large
scale. Recent attention has also been focused on UV, but the use of this technology for protozoan
inactivation in drinking water is very new. A multi-barrier approach to pathogen removal implies
an increased importance of granular media filtration.

Our conceptual representation of the relationship between pathogen passage and


operational parameters such as turbidity and particle counts is presented in Figure 1.1. The left
vertical axis shows turbidity or particle counts while the right vertical axis shows pathogen
concentration in filter effluents. The horizontal axis represents the time of approximately one
filter cycle, at a variable scale to clearly depict the stages of filter ripening.

As shown in Figure 1.1, effluent turbidity must always remain below a specified goal or
standard, in this case 0.1 NTU. After backwash, both turbidity and particle counts increase
during filter ripening. The two peaks of particle passage that may occur during ripening result
from backwash water remnants and particle non-attachment associated with the period that
particles are being retained and act as subsequent collectors. Pathogen passage through the filters
may increase during this time for the same reasons. Events such as coagulation upsets and
hydraulic surges result in suboptimal pretreatment and may consequently cause filter effluent
Filter Cycle
Time

Figure 1.1 Conceptual representation of the relationship between pathogen passage and
operational parameters such as turbidity and particle counts

concentrations of turbidity, particles, and pathogens to increase. As the filter becomes loaded,
particles and pathogens begin to break through the filter, initiating a backwash cycle.

Evaluating the various phases of the filter cycle (i.e. ripening, normal operation, and
breakthrough) as well as events (i.e. coagulation upsets, hydraulic surges, and sudden water
quality changes) allow for a description of the robustness of the removal process. The
development of practical design and operating guidelines for minimizing pathogen passage
during filtration is facilitated by a quantitative understanding and description of the pathogen
removal robustness of the filtration process.

The overall goal of this project was to evaluate the impact of filter design and operational
parameters on pathogen removal during drinking water treatment and to develop practical
guidelines and strategies for estimating and minimizing pathogen passage through filtration.
Specific objectives in pursuit of this goal were:

1. To develop a succinct, comprehensive summary of the appropriate literature,


including worldwide references, discussing particle and pathogen removal. Special
emphasis was placed on integrating the various strategies for optimizing filtration
processes with the most current information regarding pathogen removal through the
water treatment process. Although the primary focus of this project pertained to
removal of pathogens during filtration, the literature review was approached from a
multi-barrier perspective when applicable (e.g. chemical pretreatment considerations).
2. To assess the degree of pathogen passage that can be reasonably expected from
"benchmark" filtration systems (i.e. relatively standard design and optimized
operation) and evaluate the sensitivity and consistency of pathogen removal by these
systems during periods of elevated turbidity and particle counts. Key scenarios that
were investigated included: stable operation, suboptimal coagulation, complete
coagulation failure, filter ripening, hydraulic surges and particle breakthrough (end-
of-run). The benchmark systems represented two basic types of chemical
pretreatment: a relatively low coagulant dose optimized for particle removal and a
relatively high coagulant dose for combined TOC and particle removal.
3. To investigate strategies for minimizing pathogen passage through drinking water
filters. These strategies address design operating considerations. To present strategies
relevant to a wide range of utilities, several chemical and physical factors and
configurations were investigated. Chemical parameters such as pretreatment
(coagulation) strategy and backwash water chemistry were evaluated. Physical factors
that were investigated include: media configuration, backwash physics, flow rate, and
filtration strategy (i.e. direct vs. conventional). The impact of pre-oxidation and water
quality changes were also examined.
4. To develop a practical measure of a filter's robustness, i.e. its ability to consistently
maintain superior levels of pathogen removal during periods of elevated turbidity and
particle counts. Specifically, this robustness measure will provide utilities a common
basis for quantifying the effect on pathogen passage of different treatment
configurations and operational conditions, during both typical and atypical operation.
CHAPTER 2: BACKGROUND

A number of investigators have completed extensive studies and reviews in the areas of
optimizing chemical pretreatment, optimizing granular media filtration, applying detection
methods for pathogens, and evaluating removal of pathogens during drinking water treatment.
The major focus of this brief literature review was to consolidate and update reviews and studies
already completed and to focus on non-steady-state and suboptimal performance of physical
processes for removal of particles and pathogens. Emphasis was placed on integrating the
various strategies for optimizing filtration processes with the most current information regarding
pathogen removal through the water treatment process. A similar, more detailed critical review
of the filtration literature and its applications to pathogen passage is available in Emelko (2001).

REGULATORY BACKGROUND

Under the authority of the Safe Drinking Water Act Amendments of 1996, the U.S.
Environmental Protection Agency (USEPA) promulgated Stage 1 of the Disinfectants/
Disinfection By-Products (D/DBP) Rule (USEPA, 1998a) and the Interim Enhanced Surface
Water Treatment Rule (IESWTR) (USEPA, 1998b) in December 1998. These rules become
effective in January 2002.

The USEPA proposed several new regulatory activities under the IESWTR. The most
substantial elements pertain to turbidity performance and turbidity monitoring. Only surface-
water systems that serve more than 10,000 people and have rapid granular filters will be required
to comply with the new turbidity provisions described below.

The combined filter effluent turbidity must be less than or equal to 0.3 NTU in at least
95 percent of the measurements taken each month. The combined filtered effluent turbidity must
never exceed 1 NTU. Compliance shall be determined based on measurements of the combined
filter effluent at 4-hour intervals.
In addition, continuous turbidity monitoring of each individual filter is required and
exceptions must be appropriately reported to the State. Depending on the extent and frequency of
the exceptions, a system must conduct a self-assessment of the filter in accordance with
guidelines issued by the USEPA (USEPA, 1991). If an individual filter has turbidity levels
greater than 2.0 NTU based on two consecutive measurements, 15 minutes apart, at any time in
each of two consecutive months, the system will arrange for an assessment by either the State or
an approved third party.

In summary, major modifications are expected for U.S. utilities as a result of the
IESWTR. Three areas of substantial uncertainty are (1) how to compare the relative effectiveness
of treatment alternatives to implement the most cost-effective improvements; (2) how to use the
continuous turbidity monitoring data that will be required to improve treatment performance;
and, (3) how to determine the most significant filter operational effects during an individual filter
assessment.

The Canadian Drinking Water Guidelines address the issue of pathogens such as
Cryptosporidium parvum and Giardia lamblia, but do not provide detailed requirements for filter
performance such as those in the IESWTR.

TREATMENT OPTIONS FOR CRYPTOSPORIDIUM

Studies of Cryptosporidium parvum inactivation by several chemical disinfectants


(ozone, chlorine dioxide, chlorine, and monochloramine) have indicated that oocysts were 30
times more resistant to ozone than Giardia cysts exposed to the same disinfection conditions
(Korich et al., 1990). More recently, Oppenheimer et al. (2000) have reported that the Ct
requirement for Cryptosporidium was 7 to 10 times greater than for Giardia at 25°C, and 36
times greater at 5°C. Chlorine and monochloramine have been found to be considerably less
effective at inactivating oocysts (Korich et al., 1990). Other studies have concluded that less
ozone was necessary for inactivating Cryptosporidium, however, oocysts were still
approximately ten times more resistant to inactivation than were Giardia cysts (Owens et al.,
1994). Although of the chemical disinfectants ozone has generally provided the strongest
disinfection, at practical operating conditions it often cannot provide the complete
inactivation/removal of Cryptosporidium oocysts that would be required at a treatment facility.

Finch et al. (1994) and Gyiirek et al. (1996) indicated disinfectant synergy and concluded
that previous research based on the disinfection capabilities of chemicals used singly
substantially underestimated the disinfection efficacy of chemicals used sequentially, which is
often done in water treatment practice. UV disinfection technologies have also demonstrated
effective inactivation of Cryptosporidium (e.g., Campbell et al., 1995; Clancy et al., 1996;
Bukhari et al., 1998). Their operation is more effective in cleaner waters such as filter effluents
(e.g., Bukhari et al., 1999). Despite progress in the development of disinfection technologies that
can achieve reasonable Cryptosporidium inactivation, the traditional physico-chemical barriers
used in drinking water treatment remain critical to achieving desirable levels of Cryptosporidium
removal.

While optimized coagulation and clarification contribute a substantial barrier against


Cryptosporidium (e.g., Kelly et al., 1995; Fox et al., 1998; Dugan et al., 1999; Edzwald et al.,
1999), filtration is one of the most critical and successful physico-chemical barriers against
Cryptosporidium passage through water treatment. Although alternative treatment processes such
as pressure-driven membranes offer excellent removal capabilities of parasitic pathogens and
they are becoming increasingly cost-competitive, their use can require significant capital
investment. A component of conventional water treatment operations, rapid granular media
filtration has demonstrated efficacy as a barrier against Cryptosporidium. Full-scale
Cryptosporidium removals from 2 to >4 log have been reported in the literature (e.g., Baudin and
Laine, 1998; Nieminski and Ongerth, 1995).

REMOVAL MECHANISMS OF GRANULAR MEDIA FILTRATION

The size of particulate matter removed by granular media filters typically ranges from 0.1
to 1000 urn. Particle removal during filtration occurs when particles deviate from the fluid
streamlines due to gravitational forces, diffusion gradients, and inertial effects of momentum
(e.g., O'Melia and Stumm, 1967). The relative impact of these effects depends on water quality
and physical characteristics of the particles and filter media. Particle removal can occur either by
straining or attachment (Montgomery, 1985).

The particle removal mechanisms of granular media filtration involve three distinct steps:
transport, attachment, and detachment. The mechanistic basis for removal by particle deposition
requires particle transport to distances close to collectors (i.e., media grains, with or without
previously-deposited particles) where attachment can occur (e.g., O'Melia and Stumm, 1967;
Elimelech, 1991). Particles deposited in filters can also detach and be transported to the bulk
fluid. Detached particles may re-attach at further depths or pass through the filter and appear in
the filter effluent (e.g., Amirtharajah, 1988). O'Melia (1985) and Amirtharajah (1988) provide
thorough summaries of filtration theory.

Diffusion and sedimentation are the dominant transport mechanisms during water
filtration (Ives, 1982). The combination of these two transport mechanisms yields a minimum net
transport efficiency for particles that are approximately 1 jam in size (Amirtharajah, 1988). Given
that Cryptosporidiwn oocysts are typically 3-6 pm in size, they are near the minimum net
transport efficiency.

Particle attachment to either a media grain or previously retained particles depends on the
surface properties of these materials. O'Melia and Stumm (1967), Tobiason and O'Melia (1988),
and Raveendran and Amirtharajah (1995) specified several mechanisms that affect attachment
during filtration. They include: London-van der Waals (LVDW) forces, electrical double-layer
(EDL) forces, hydrodynamic forces, steric forces, Born forces, structural forces, and chemical or
bridging forces.

When the magnitude of the hydrodynamic forces (consisting of lift and drag) on particles
exceeds that of adhesive forces, particles are detached and reach lower depths within the filter;
an avalanche effect of arriving particles may be an alternative mechanism of detachment.
Attachment and detachment occur simultaneously in filter layers that have reached a saturated
but metastable configuration of deposited particles (Amirtharajah, 1988). Many studies have
demonstrated that optimized particle destabilization minimizes "premature" detachment (e.g.,
Roebeck et al., 1964; O'Melia, 1985).

8
There are several possible sources of particles in filter effluents during breakthrough.
Particles may pass through the filter directly from the influent, they may attach and detach from
the filter, and they may enter the filter at a smaller size, form floes on the surface of the media,
and then detach from the media (Lawler et al., 1995). The particle size distribution data of Ginn
et al. (1992) demonstrated a decrease in removal efficiency of larger particles as particle deposits
in the bed increased, however, this trend was not evidenced in turbidity measurements. This
finding was consistent with Moran et al. (1993a) who concluded that ripening and breakthrough
were strongly dependent on particle size. That investigation demonstrated that while removal of
smaller particles increased (or ripened) for the longest duration, removal of intermediate sizes
(such as those of Cryptosporidium) ripened early but decreased substantially as deposits in the
bed increased (Moran et al., 1993a).

Ginn et al. (1992) proposed a conceptual model filtration model consisting of four
phases. From a mechanistic standpoint, this model was generally consistent with the findings of
Moran et al. (1993a) that indicated that ripening and breakthrough were not distinct stages of
filter operation, but occurred simultaneously for different sized particles. The first phase of the
Ginn et al. (1992) model is filter ripening during which attachment increases as particles are
deposited and subsequently act as collectors. Effective filtration is the second phase of the
model; increasing interstitial flow velocities result in increased detachment and decreased
attachment, however, attachment is still high enough to remove many of the influent particles.
Effluent turbidity begins to breakthrough during the third phase as a result of increasing
interstitial flow velocities caused by clogging of pores; attachment continues to decrease while
detachment increases. Preferential tubular passages (wormholes) of flow begin to form and result
in the significant decrease of particle attachment. Detachment peaks and begins to decrease
because of the continued concentration of flows in the wormhole passages. After breakthrough,
wormhole flow begins to dominate and the filtering capacity of the bed is almost exhausted; this
is the fourth phase of the filtration model. At this time, attachment and detachment do not occur
and filter effluent turbidity and particles rapidly approach their influent values.

Several important water treatment ramifications can be drawn from the reported
experimental and theoretical work regarding particle detachment during filtration. The
experimental findings of Ginn et al. (1992) and Moran et al. (1993a,b) underscored the
inadequacy of particle count data as a sole tool for assessing filter removals of pathogens such as
Cryptosporidium. The demonstrated deterioration in removals (or early breakthrough) of oocyst-
sized particles in the later portions of filter cycles suggested that end-of-run operating conditions
may be particularly vulnerable in terms of Cryptosporidium removal, or more specifically,
Cryptosporidium detachment (Ginn et al., 1992; Moran et al., 1993a,b).

The filtration model proposed by Ginn et al. (1992) provides a conceptual framework for
attachment and detachment mechanisms during filtration; it can be generally applied to
understanding and predicting operational effects on particle and more specifically
Cryptosporidium removal by filters.

OPERATIONAL FACTORS AFFECTING PARTICLE REMOVAL

The mode and conditions of operating water treatment processes can greatly impact
related particle removal mechanisms. Both pretreatment conditions and filtration conditions
directly impact the particle removal efficiency of filters. Table 2.1 summarizes operational
conditions and their related particle removal mechanisms.

Table 2.1
Operational factors and related particle removal mechanisms

Operational Particle Removal Mechanism Some Key


Factors Transport Attachment Detachment References
Coagulation Roebeck et al., 1964
O'Melia and Stumm, 1967
O'Melia, 1985

Filter Aid Black etal., 1965


Yao et al., 1971
Habibian and O'Melia, 1975,
Zhuetal., 1996

Hydraulic Changes Cleasby et al., 1963


Tuepker and Bauescher, 1968
Logsdon et al., 1981
Sharmaetal., 1992

(continued)
10
Table 2.1 (continued)
Operational Particle Removal Mechanism Some Key
Factors Transport Attachment Detachment References
Backwash Strategy Amirtharajah, 1988
Raveendran and Amirtharajah, 1995
Coltonetal., 1996

Temperature Ives and Sholji, 1965


Ives, 1982

Ripening O'Melia and Ali, 1978


Amirtharajah, 1985
Moranetal., 1993a

Breakthrough Ginnetal., 1992


Moranetal., 1993a,b
Lawleretal., 1995

Media Ives and Sholji, 1965


Trussell et al., 1980____ ___

Coagulation

Although considerable reductions in both particles and turbidity can occur during
filtration without any chemical pretreatment, it has been repeatedly demonstrated that proper
coagulation is a critical step in maintaining good particle removal during filtration (e.g., O'Melia
and Crapps, 1964; Roebeck et al., 1964; Trussell et al., 1980). As discussed previously, several
particle attachment mechanisms occurring during filtration have been specified (e.g., O'Melia
and Stumm, 1967; O'Melia, 1985); they include particle destabilization forces which are directly
affected by coagulation. The relative strength of attachment forces affects the degree of non-
attachment that occurs in filters; the degree of attachment to forces like hydrodynamic shear also
dictates detachment. Chemical parameters such as pH and ionic strength, which are impacted by
chemical pretreatment, have been experimentally identified as significant factors affecting
particle detachment in packed bed filters when a constant hydrodynamic force is applied (e.g.,
McDowell-Boyer, 1992; Ryan and Gschwend, 1994).

It stands to reason that coagulation would similarly impact the removal of pathogenic
colloidal particles such as Cryptosporidium oocysts. The importance of coagulation processes for
11
improving filter removal efficiencies of Giardia cysts has been well documented. Several
investigations have demonstrated little (<1 log) to no removal of Giardia cysts by GAC filters
(Patania et al., 1995), sand and dual-media filters (Al-Ani et al., 1986), and tri-media filters
(Horn et al., 1988) during no coagulation conditions. Even suboptimal coagulation conditions
have been shown to affect the pathogen removal efficiency of filters. In a pilot-scale direct
filtration plant, Logsdon et al. (1981) found that mean Giardia muris cyst removals decreased by
approximately 1 to 2.5 log during periods of suboptimal and minimal coagulation compared to
optimal operating conditions. Similar decreases in cyst removal as a result of suboptimal
coagulation conditions have been demonstrated at other direct (Ongerth and Percoraro, 1995)
and conventional pilot plants (Patania et al., 1995).

Filter Aid

Polymers (polyelectrolytes) act as filter aids and can improve the quality of filter
effluents. When a polymer is used as a filter aid, both transport and attachment mechanisms can
be facilitated by the formed polymer-particle floes which are generally larger and stronger than
those achieved with conventional chemical pretreatment alone (Zhu et al., 1996). Several studies
have demonstrated that cationic polymer addition during filtration is similar to coagulation and
flocculation in that particle destabilization must be optimized (e.g., Yao et al., 1971; Habibian
and O'Melia, 1975; Zhu et al., 1996). The use of filter aids has demonstrated improvements in
filtrate quality (Conley and Hsiung, 1969), resistance to early breakthrough (Conley and Pitman,
1960), and reduction in the magnitude and duration of ripening (Tuepker and Buescher, 1968),
however, it can also result in significant increases in headloss. Appropriate addition of filter aids
should result in higher attachment forces and, therefore, should be concurrent with optimized
backwashing (Amirtharajah, 1988).

Patania et al. (1995) investigated the impact of filter aid addition on Cryptosporidium and
Giardia removal by GAC/sand filters and concluded that the use of filter aid did not improve the
removal of either microorganism under the conditions studied. The authors noted that their filter
influent Cryptosporidium concentrations were below their method detection limit, resulting in
inconclusive oocyst data. The conclusion that filter aid did not substantially enhance protozoan
or at least Giardia removal was consistent with the findings of Ongerth et al. (1989). This

12
apparent discrepancy between improved filtrate turbidity but unchanged cyst/oocyst removal
may suggest that the use of filter aid improves filtrate quality by enhancing the removal of non-
oocyst/cyst sized particles. It is possible that filter aid induced particle bridging considerably
increases the size of particles originally in the 1 um range, which are known to have a minimum
transport efficiency (Yao et al., 1971), thereby increasing their transport and removal efficiencies
while the removals of larger particles remain largely unaffected.

Hydraulic Changes

Hydraulic conditions can significantly impact the quality of filter effluents. It is generally
recognized that filter performance is adversely affected by non-steady flow (e.g., Trussell et al.,
1980). Hydraulic changes can occur suddenly or gradually; at the extreme they encompass
sudden starts and stops in operation. In general, hydraulic changes disrupt the equilibrium
between particle attachment forces and hydraulic shear (detachment) forces (e.g., Logsdon,
1999). Therefore, it is reasonable to surmise that the relative impacts of hydraulic changes are
inextricably linked to other operational factors such as coagulation conditions; however, some
general conclusions regarding the impact of hydraulic changes on filtrate quality can be drawn.

Cleasby et al. (1963) and Tuepker and Buescher (1968) showed that large flow rate
changes cause deterioration of filtered water quality by the detachment of previously retained
particles. The degree of deterioration was related to the magnitude and rapidity of the rate change
and independent of the duration of the disturbance. These relationships support observations that
declining rate filters may provide better performance than constant rate filters (e.g., Hudson,
1959; DiBernardo and Cleasby, 1980). Subsequent experiments by Hilmoe and Cleasby (1986)
found the opposite result: no significant differences between declining rate and constant rate
filters. The authors speculated that the previously reported poorer effluent quality achieved by
constant rate filtration might have been caused by the constant rate control system used by
DiBernardo and Cleasby (1980), which might have inadvertently resulted in continuous flow rate
fluctuations or surges.

Cleasby et al. (1963) also revealed additional complexity in speculating on the risk of
pathogen passage through filters when flow rates changes or hydraulic steps are applied because
they demonstrated that particle passage through filters following a disturbance was dependent on
13
the composition of the filter influent. This result underscores that the balance between
attachment and detachment forces is affected by multiple factors such as coagulation, the
ongoing ripening and breakthrough of different sized particles, and hydraulics. Fitzpatrick et al.
(1999) demonstrated that large and sudden changes in flow have a dramatically deteriorated
particle removal by filters while smaller changes that were implemented gradually did not
always increase particle counts. Similar effects of increased flow rates on Giardia removal have
been observed, however, the increases in cyst passage were considerably higher than the
increases in turbidity (Logsdon et al., 1981); these results suggested that increases in
Cryptosporidium passage through filters could potentially be expected under similar operating
conditions.

Backwashing Strategy

During backwashing, hydrodynamic shear is primarily responsible for detachment of


particles. Raveendran and Amirtharajah (1995) experimentally and theoretically described the
interactions between suspended particles and media grains coated with previously deposited
particles under fluidization conditions. A calculation of the interaction forces that control particle
detachment indicated that solution chemistry affects particle detachment during backwashing.
Optimization of operating conditions for stronger attachment during normal filter operation may
possibly make it more difficult to remove particles during subsequent filter backwashing,
however. If media cleaning is ineffective it may lead to poorer quality effluent during the initial
stages of filtration (e.g., Amirtharajah and Wetstein, 1980; Amirtharajah, 1993). Ensuring
thorough removal of pathogenic particles during backwashing may also prevent later release
during ripening or normal filter operation. The simultaneous application of air scour with
subfluidization water wash (collapse pulsing) has been identified as an optimal backwashing
strategy (e.g., Amirtharajah and Wetstein, 1980; Amirtharajah et al., 1991; Amirtharajah, 1993).
The use of collapse-pulsing backwash strategies has been shown to reduce the number of oocyst-
sized particles in filter effluents during ripening (Colton et al., 1996).

Temperature

Since particles are transported and move relative to water to reach a grain surface so that
they can attach and be removed, they will experience a viscous drag that is proportional to water
14
viscosity (e.g., Ives and Sholji, 1965). The more viscous the water (e.g., colder water
temperatures), the more slowly particles move relative to the water to reach a grain surface,
reducing the probability of removal (e.g., Ives and Sholji, 1965). Experiments conducted by Ives
and Sholji (1965) demonstrated that colder water temperatures result in considerably decreased
particle removal when all other factors such as raw water quality remained constant.

In normal water treatment practice the effects of water temperature on particle transport
and subsequent removal by filters are not easy to delineate because other operations such as
clarification are also affected by water temperature. Seasonal changes also often result in
considerably different raw water qualities making it almost impossible to label certain effects as
exclusively temperature-related. The main importance of evaluating the role of temperature as it
affects pathogen passage through filters is twofold. First, for given water treatment plants it is
important to determine the level of stable operation removals across the range of water
temperature or seasonal conditions experienced. Second, it is important to determine if other
operational conditions or events such as coagulation upsets, ripening, breakthrough, etc. have
similar implications for pathogen passage during warm and cold water conditions.

Ripening

It has been suggested that 90% of the particles that pass through a well-operated filter do
so during ripening (e.g., O'Melia and Ali, 1978; Amirtharajah, 1985). The two peaks of particle
passage that occur during ripening result from backwash water remnants (Amirtharajah, 1988)
and the period that particles are being retained and subsequently acting as collectors (O'Melia
and Ali, 1978); during this second peak particle passage is primarily due to non-attachment. It is
likely that these mechanisms also affect pathogen passage through filters during this period.

A variety of findings regarding protozoan passage through granular media filters during
ripening have been reported. Logsdon et al. (1981) reported that Giardia cyst passage through
filters was significantly higher during ripening than during stable operation, even at low effluent
turbidities. Similar findings were obtained at two pilot plants studied by Patania et al. (1995); the
differences between cyst removals during stable filter operation and ripening were less dramatic,
however. At a third pliot-plant studies by Patania et al., (1995), Giardia removals during

15
ripening were comparable to those achieved during stable filter operation. These data suggest
that multiple factors may affect protozoan removal during ripening.

Breakthrough

Possible sources of breakthrough in filters include particles that pass through directly
from the influent (non-attachment) or particles that become detached (Lawler et al., 1995).
According to the Ginn et al. (1992) model and according to other studies (e.g., Moran et al.,
1993b), non-attachment and detachment occur during breakthrough conditions. As particle
detachment and non-attachment increase, increased pathogen passage through filters would also
be expected.

Logsdon et al. (1981) demonstrated that turbidity breakthrough at the end of a filter cycle
(when filter effluent turbidity was above 0.4 NTU) could be accompanied by a substantial
passage of Giardia cysts, even if they were not present in the filter influent. A considerable
increase in cyst passage was also observed during early breakthrough conditions when filter
effluent turbidity was just above 0.2 NTU. Patania et al. (1995) investigated Giardia passage
through filters during breakthrough when effluent turbidities increased from 0.1 NTU to 0.2
NTU or higher. They found that while Giardia removal was approximately 0.5 log lower during
breakthrough during stable operation. These data suggested that increased Cryptosporidium
passage could also be expected during breakthrough, especially at filter effluent turbidities above
0.2 NTU.

Media Specifications

Important aspects of media specifications are size, shape, depth, uniformity, and the
choice of single-, dual-, or multi-media (Trussell et al., 1980). Media size affects the length of
time to turbidity breakthrough and limiting headloss (e.g., Trussell et al., 1980). Media shape
affects headloss (rate of headloss build up) and is associated with the ability of filters to remove
particles (e.g., Trussell et al., 1980). Angular media have demonstrated better turbidity and
particle removal compared to smoother media (e.g., Trussell et al., 1980). Lower uniformity
coefficients of anthracite media have also demonstrated improved removal of oocyst-sized
particles (Yohe et al., 1999).

16
Filtration theory practice has indicated that reverse-graded multi-media typically provide
better participate (e.g., Rimer, 1968; Conley, 1972) and protozoan pathogen removal (e.g.,
Logsdon et al., 1985) at the cost of faster headless buildup. Other studies, however, have
indicated comparable particulate (e.g., Tate and Trussell, 1978, Billica et al., 1999) and
protozoan pathogen (e.g., Emelko et al., 1999; Swertfeger et al., 1999) removal by dual- and
multi-media filters. These results indicate the theorized particle removal superiority of mixed-
media over dual-media is far from consistent.

SURROGATES FOR CRYPTOSPORIDWM REMOVAL DURING WATER


TREATMENT

Several surrogates for viable Cryptosporidium oocysts have been evaluated. Potential
surrogates for the cyst or oocyst removal efficiency of drinking water treatment processes have
included turbidity, particle counts, heterotrophic plate counts (HPC), aerobic spores (typically
Bacillus subtilis), UV254, dissolved organic carbon (DOC), and chemically inactivated
Cryptosporidium oocysts. Table 2.2 lists and summarizes the main observations from several
studies that evaluated surrogates for the physical removal of Cryptosporidium oocysts during
water treatment.

Particle counting allows for real-time monitoring and is more sensitive than turbidity for
treatment optimization (e.g., Hargesheimer et al., 1992; Arora et al., 1995). However, particle
counting lacks the ability to discern between particles in the same size range and to detect very
small changes in concentration that would likely be associated with Cryptosporidium
concentrations (e.g., Hargesheimer et al., 1992). Several studies have indicated that while
turbidity and particle removals are indicative of treatment efficiency, they are not indicative of
Cryptosporidium removal by treatment processes (e.g., LeChevallier et al., 1991b; Nieminski
and Ongerth, 1995; Swaim et al., 1996).

Several pilot- and full-scale studies have demonstrated that organism-sized particles and
turbidity are approximate indicators of pathogen removal by drinking water treatment processes,
but not reliable surrogates (e.g., LeChevallier and Norton, 1992; Nieminski and Ongerth, 1995).
Plummer et al. (1995) reached similar conclusions about turbidity, as well as UV254 and DOC.

17
Table 2.2
Surrogate parameters for physical removal of Cryptosporidium
Study Surrogate Parameters Major Observations

<B in 0)
en k.
c c Q «
2 •c (O .C
Q.
o> Q.
1 T3
"•e s.
5 1 s O
1 s 2 O b
0 *• Q. (- io OQ 13 Q 2
LeChevallier et al., 1991b Turbidity/particles indicative of treatment efficiency, not oocyst removal.
Giardia and Cryptosporidium removed similarly.

LeChevallier and Norton, 1992 Turbidity/particles indicative of treatment efficiency, not oocyst removal.

Nieminski, 1994 Turbidity/particles indicative of cyst/oocyst removal.


oo
Cryptosporidium more difficult to remove than Giardia.
HPCs not good surrogate for cyst/oocyst removal.

West etal., 1994 Particle removal underestimated oocyst removal.

Charles et al., 1995 Cryptosporidium removed at similar, but lower levels than Giardia.

Kelley etal., 1995 Turbidity/particles indicative of treatment efficiency, not oocyst removal.

Nieminski and Ongerth, 1995 Turbidity/particles indicative of treatment efficiency, not oocyst removal.
Particles more indicative of oocyst removal than turbidity.
HPCs not reliable surrogates for oocyst removal.
Cryptosporidium removed at similar, but lower levels than Giardia.

Ongerth and Pecoraro, 1995 Similar Cryptosporidium and Giardia removals.

Patania et at., 1995 Turbidity/particles indicative of treatment efficiency, not oocyst removal.
Similar Cryptosporidium and Giardia removals, not directly related.

Plummer et al., 1995 Turbidity, UV254, and DOC indicative of treatment efficiency, but not
oocyst removal.______________________________
(continued)
Table 2.2
Surrogate parameters for physical removal of Cryptosporidium (continued)
Study Surrogate Parameters Major Observations

1
^ * m 2 <"

I8'§
>. 4ja
1 ? .s§
S '*&1
llsS
C>3 TO
loH
0
Sog ;2 CL TO > O .^
O I- CL h- CD 3C OQ Z) Q 2
Coallieretal., 1996 y ^ Spore removals more sensitive than turbidity and similar to particles.

Lytle etal., 1996 / / /^ /i "f J Spore removals closely parallel particle removals.
Spores removals decline more rapidly than particles and turbidity.
Oocyst-sized particles easiest to remove, not good surrogate.

Rice etal., 1996 . Spore removals useful for determining treatment efficiency.
Spore removal closely and conservatively match particle removal.

Swaimetal., 1996 Cyst and oocyst removals not well correlated with particle removals.
Removal of cyst sized particles lower than cyst removal.
Comparable Cryptosporidium and.Giardia removals.

Scott etal., 1997 Turbidity/particles indicative of treatment efficiency, not oocyst removal.
Spores conservative indicator of oocyst removal.

YatesetaL, 1997b Spore removal not directly related to particle and turbidity removal.

Baudin and Laine, 1998 Turbidity/particles indicative of treatment efficiency, not oocyst removal.

Edzwald and Kelley, 1998 Turbidity/particles indicative of treatment efficiency, not oocyst removal.

Fox etal., 1998 Trends in turbidity and particle removal through clarifier generally
indicative of oocyst removal.
(continued)
Table 2.2
Surrogate parameters for physical removal of Cryptosporidium (continued)

Study Surrogate Parameters Major Observations

I
•B £
•g %> % o
o ex oi
§ (O
Q.
.•= 0) Q.
CL <n
2 !9 T5 1 S
•e '-E 15
"5 1 O Q
0. 09
o .2 2 S X OQ 1 § 1
y y" y y y Turbidity, particles and spores indicative of treatment efficiency.
Nieminski and Belliamy, 1998 S
Mostly non-detects of cysts/oocysts in finished water.

Duganetal., 1999 Positive correlation between removals of oocysts, particles, and


turbidity through coagulation and settling.
S3
Turbidity removal most conservative indicator of oocyst removal
O through filtration.

Edzwald et al., 1999 Particle and cyst removal trends generally consistent across
clarification and filtration.
Particle removal provides a conservative estimate of cyst removal.

Emelkoetal., 1999 Microsphere removals were generally comparable to oocyst removals.

Swertfeger et al., 1999 Microsphere and spore removals were more similar to cyst and
oocyst removals than were turbidity and particle removals.
Turbidity and particle removals lower than cyst/oocyst removals.
1-2 urn size range. " 5-15 um size range.
b 2-5 urn size range. ' other particle size measurements were also made.
c 2-6 um size range. 1 cumulative particles >1 um.
d 2-15 pm size range. k cumulative particles >2 um.
e 3-5 um size range. 1 cumulative particles >3 um.
' 3-6um size range. m cumulative particles >5 um.
NS
9 4-7 um size range. not stated.
Patania et al. (1995) indicated that achieving a goal of 0.1 NTU was indicative of effective
cyst/oocyst removal. Although the risk of Cryptosporidium passage appeared to increase with
increasing filtrate turbidity in several studies (e.g., Hall et al., 1995; Hall and Croll, 1996;
Nieminski and Ongerth, 1995), Fuller et al. (1995) did not observe significant oocyst passage
during the first hour of operation after backwash when filter effluent turbidity was high (filter
ripening). HPCs yield an estimate of the total number of viable bacteria that can be successfully
grown on plates and have been used to determine treatment efficiencies and distribution system
integrity in several studies (e.g., Ferguson et al., 1990). Evaluating treatment processes for
pathogen removal based on HPCs may be difficult because of the varying composition of
bacterial populations that constitute HPCs. Several pilot- and full-scale studies have
demonstrated that removal of heterotrophic bacteria was not an effective surrogate for predicting
cyst/oocyst removal (e.g., Nieminski; 1994; Nieminski and Ongerth, 1995; Nieminski and
Bellamy, 1998).

Aerobic spores, primarily Bacillus, have been used for evaluating treatment efficiency
(e.g., Coallier et al., 1996; Rice et al., 1996). They are present in most surface waters, pose no
public health threats, and are not indicative of fecal contamination (Jakubowski et al., 1996). The
spores are approximately 1 um in size, close to the size range of Cryptosporidium oocysts (Rice
et al., 1996). They are also highly resistant to disinfection and their removals closely parallel
particle removal (e.g., Coallier et al., 1996; Rice et al., 1996).

Studies performed by Scott et al. (1997) evaluated Bacillus spores, turbidity, and particle
counts as surrogates for pathogen removal. Turbidity and particle counts yielded conclusions
similar to those discussed above (e.g., Nieminski and Ongerth, 1995; Patania et al., 1995)
whereas spores demonstrated a significant correlation with Cryptosporidium removal at both
pilot- and full-scale. Lytle et al. (1996), Nieminski and Bellamy (1998), and Swertfeger et al.
(1999) also concluded that aerobic spores were indicative of treatment efficiency, however, they
did not conclude that the spores were adequate surrogates for oocyst removal. These studies
suggest that although spores may be a more conservative indicator of treatment efficiency than
particles or turbidity, they are not necessarily robust indicators of Cryptosporidium removal
during water treatment.

21
Due to the lack of adequate surrogates for removal viable Cryptosporidium oocysts,
chemically inactivated oocysts are commonly used for treatment evaluations. The use of
chemically inactivated oocysts is preferable to that of viable oocysts because of the potential
health risks associated with the use and release of viable Cryptosporidium. It has been suggested
that chemically inactivated Cryptosporidium oocysts may not be ideal surrogates for viable
oocysts due to differences in surface charge (e.g., Lytle and Fox, 1994), however, bench-scale
studies have indicated that formalin-inactivated oocysts and viable oocysts of C. parvum are
comparably removed by both dual- and tri-media filters (Emelko, 2001).

CRYPTOSPORIDIUM REMOVAL DURING DRINKING WATER TREATMENT

Filtration is one of the most critical and successful physico-chemical barriers against
Cryptosporidium passage through water treatment. Several studies have assessed
Cryptosporidium oocyst removal by granular media filters operated at or near to optimized stable
operating conditions; they are summarized in Table 2.3.

Full-scale oocyst removals have been reported anywhere from 2-3 log (e.g., Kelley et al.,
1995; Nieminski and Ongerth, 1995) to >4 log (e.g., Baudin and Laine, 1998). Pilot-scale
Cryptosporidium removal data have suggested that filters can achieve anywhere from 2-3 log
(e.g., West et al., 1994; Kelley et al., 1995; Fox et al., 1998), 3-4 log (Yates et al., 1997b), and
>5-log (e.g., Patania et al., 1995; LeChevallier et al., 1991b) removal of oocysts. Differences in
analytical reliability, processed sample volume, method detection limits, and influent
microorganism concentrations can all contribute to reported differences in the Cryptosporidium
removal capacities of filters between studies. The range of oocyst removals described in Table
2.3 underscores the need for accurate and thorough description of experimental conditions and
further investigation of Cryptosporidium removals by filters during optimized and vulnerable
operating conditions; however, some general conclusions can be drawn from these data.

Pilot- and full-scale studies performed by Nieminski (1994) and Nieminski and Ongerth
(1995) indicated comparable pilot-scale removals of Cryptosporidium by conventional and direct
filtration processes. Differences were observed at full-scale, however, where direct filtration

22
Table 2.3
Removal of Cryptosporidium by drinking water filters during stable operation
Study Type of Filter Filter C. parvum Seeded Observations of Relevance
Treatment Type Loading Log Cone. to Present Study
(gpm/ft2) Removal (#/L)'
LeChevallier et al., 1 991 b Varied - 66 full-scale plants varied varied >2.4* - Rapid sand and GAC effluents had higher oocyst
2 pilot-scale conventional plants GAC/sand 0.75-1.2 >5.3 NS concentrations than dual- and tri-media effluents - possibly
related to different raw water quality.
Rapid sand more effective than GAC for oocyst removal.
Many non-detects in filter effluent samples.

LeChevallier & Norton, 1992 3 full-scale conventional plants anth./sand NS >2.4 _ Oocyst/cyst occurrence related to raw water levels.
GAC/sand NS >2.5 - Many non-detects in filter effluent samples.
anth./sand NS >2.3 -

Nieminski, 1994 0.5-gpm conventional pilot-scale anth./sand 2.5 avg. 2.8 ~104" No difference between conventional and direct filtration for
0.5-gpm direct filtration pilot-scale anth./sand 2.5 avg. 2.9 ~104"
cyst and oocyst removal.
to 600-gpm conventional full-scale anth./sand NS avg. 2.1 ~107tt Direct filtration yielded better cyst and oocyst removal
600-gpm direct filtration full-scale anth./sand NS avg. 2.7 ~10m but influent quality was very different.

West et al., 1994 Pilot-scale direct filtration anth. 6 2.1-3.3 ~102-103t Many non-detects in filter
effluent samples.
anth. 14 1.9-3.2 ~102-103t

Charles et al., 1995 Bench-scale direct filtration mixed 4 2.3 - 3.3* ~104T Lack of coagulant decreased removal by >2 logs.
2-gpm conventional pilot plant mixed 4 2.2 - 4.5* -100f Rapid mix conditions impacted removal by <1 log.
Stable/optimal removals not specified.

Hall et al., 1 995 Pilot-scale dissolved air flotation (DAF) sand 2.5 2.9 - 3.4* -2x1 04t
followed by rapid gravity anth./sand 2.5 3.9* 2x1 04t
filtration (RGF) GAC 2.5 4.4* 2x1 04t
sand 2.5 >3.6* 300-800* No differences between single, dual, and GAC filters.
anth./sand 2.5 3.1 - >4.4* 300-8001 Lower removal during ripening (-0.4 log)
GAC 2.5 3.2-4.1* 300-800' - Half the effluent samples were non-detects.
sand 2 >3.4* 400f
sand 4.1 3.5** 400f
Pilot floe blanket clarification and RGF sand 2.5 3.9* 700f
(continued)
Table 2.3
Removal of Cryptosporidium by drinking water filters during stable operation (continued)
Study Type of Filter Filter C. parvum Seeded Observations of Relevance
Treatment Type Loading Log Cone. to Present Study
(gpm/ft2) Removal (#/L)"
Kelleyetal., 1995 1 mgd conventional plant sand NS >1.0 Many non-detects in filter effluent samples.
3 mgd conventional plant sand NS >0.9 -

-10""
Nieminski & Ongerth, 1995 0.5-gpm conventional pilot-scale anth ./sand 5.8 1.9-4.0 Fluctuating turbidity resulted in variable cyst and
-104"
0.5-gprn direct filtration pilot-scale anth./sand 5.8 1.3-4.0 oocyst concentrations.
4.8 1.9-2.8 -10W Direct filtration yielded better cyst and oocyst
600-gpm conventional full-scale anth./sand
4.8 2.6 - 2.9 -10m removal, but influent quality was very different.
600-gpm direct filtration full-scale anth./sand

Ongerth & Pecoraro, 1995 1-gpm direct filtration pilot plant tri-media 5 2.5-3.2 -400-3000 Inadequate coagulation decreases removal to -1 ..5 log.

GAC/sand 3-6 -0.2-3.1** ~103-104t Chemical pretreatment critical for oocyst removal.
Patania et al., 1995 Conventional pilot-plant
anth./sand 3-6 0.4-1.1** ~103-104t Media design, filter aid, filtration rate: less important.
8 2.3 - 3.7** -103t Cyst /oocyst removals decreased by -0.5-1 log
N) Pilot-scale in-line filtration anth./sand
Conv. pilot w/ low-rate surface filtration sand -2 2.2 - 4.7** -103-104t during ripening.
Ripening effect observed at 2 pilot plants. At a third pilot
pilot plant, no difference between ripening/stable.
Stable and breakthrough removals - comparable.
Many influent and effluent non-detects.
Large range of filter influent concentrations.

Timmsetal., 1995 Pilot-scale slow sand filtration slow sand 0.12-0.16 >4.5 -4000 No oocysts found in filter effluent.

Edzwald et al., 1996 Pilot-scale contact filtration anth./sand 3 avg. 4.7 7210 Filter influent concentrations changed due to chemical
Pilot-scale dissolved air flotation (DAF) anth./sand 3 -2-4.1 2-631 pretreatment conditions and affected log removals.
followed by RGF anth./sand 6 -2 - 2.2 2-631

deep anth. 12 3.7 - >4.3* -103t During ripening, cyst /oocyst removals -0.5-1 log lower
Swaimetal., 1996 Pilot-scale direct filtration
16 3.6 - >4.3* -103t than during stable operation.
deep anth.
12 3.5-4.1* -103t Cyst and oocyst removals >3 log during ripening.
deep dual
16 3.6 - >4.3* -103t
deep dual

1.3,6 -1.7-4.3 -108tt Did not distinguish between dual- and tri-media removals.
Scott etal., 1997 6-gpm conventional pilot plant dual/tri
(continued)
Table 2.3
Removal of Cryptosporidium by drinking water filters during stable operation (continued)
Study Type of Filter Filter C. parvum Seeded Observations of Relevance
Treatment Type Loading Log Cone. to Present Study
(gpm/ft2) Removal (#/L)"
Yatesetal., 1997b Pilot-scale direct filtration (alum) anth./sand 6.0 avg. 3.7 -10° n FeCI3 coagulation provided better removals than alum.
Pilot-scale direct filtration (FeCb) anth./sand 6.0 avg. 4.5 -108tt

Baudin and Laine, 1998 12.7 mgd conventional full-scale anth./sand 4.1 >4 _ During ripening, oocyst removals -1 log lower than during
31.7 mgd conventional full-scale GAC 2.9 2.0-3.0 - stable operation.
Stable and breakthrough removals -comparable.
Ripening and breakthrough turbidities not specified.

Edzwald and Kelley, 1998 Pilot DAF (FeCI3) and RGF anthVsand 3.0 2 2510* Coagulant selection affected DAF removal of oocysts and
K) 6.0 2 2510* subsequent filter influent concentration.
Pilot DAF (alum) and RGF anth./sand 3.0 3.9 251 0T
6.0 2.2 251 0T

Foxetal., 1998 Conventional pilot plant anth./sand 2.0 2-3 -105 Study focused on clarifier removals, filtration only generally
mentioned. Stable/optimal removals not specified.

Duganetal., 1999 Conventional pilot plant sand 2.0 -3.0-3.7 -84-480 Under-coagulation resulted in lower removals.
anth./sand 2.0 -3.0-3.7 -84-480 Comparable removals in sand and dual-media filters.
anth./sand 4.1 -1.6-3.1 -84-480 Removal decreased by > 1 log over time at higher rates.
Increased media depth did not consistently result in
increased oocyst removal.
Stable/optimal removals not specified.

Edzwald etal., 1999 DAF pilot (alum/polymer) T=2-3°C anth./sand 6 >3.7 -107§5 No oocysts/cysts were found in filter effluent.
Lamella pilot (alum/polymer) T=2-3°C anth./sand 6 >4.7 ~10658 Filter influent concentrations and subsequent log removals
Lamella pilot plant (alum) T=2-3°C anth./sand 6 >4.8 ~106§5 were different due to different removals by clarifiers.
DAF pilot (alum/ polymer) T=13-14°C anth./sand 6 >3.0 -1075S
Lamella pilot (alum/polym.) T=13-14°C anth./sand 6 >4.3 ~1065§
Lamella pilot (alum) T=13-14°C anth./sand 6 >3.9 -10655
(continued)
Table 2.3
Removal of Cryptosporidium by drinking water filters during stable operation (continued)
Study Type of Filter Filter C. parvum Seeded Observations of Relevance
Treatment Type , Loading Log Cone. to Present Study
(gpm/ft2) Removal (#/L)'
Swertfeger et al., 1999 Conventional pilot plant (summer) sand 2.5 1.8-3.3 -10!>-10° TT All media configurations statistically comparable.
anth./sand 5.0 1.6-3.4 -105-106tt Removals adjusted (empty column removals subtracted).
deep dual 5.0 3.4-4.2 -1 05-1 06 n Almost entire effluent processed.
Conventional pilot plant (winter) sand 2.5 2.5 - 3.0 -105-106tt Polystyrene bead removal somewhat consistent with
anth./sand 5.0 3.1-3.2 -105-106Tt oocyst removal in one experiment.
deep dual 5.0 2.9 - 4.0 -105-108tt Oocyst were not coagulated, only pretreated with a small
amount (0.1-0.5 mg/L) of ferric sulfate as filter aid.

Not applicable.
"s Not stated.
' Measured concentration in filter influent unless otherwise stated.
1 Seeded concentration in raw water.
K) ^Unclear as to whether measured or calculated filter influent concentrations were used.
"Theoretical influent concentration based on hemocytometer counts and dilution calculations. Filter influent concentration was not stated.
n Total number of cysts/oocysts seeded at filter influent.
§§ Total number of cysts/oocysts seeded at clarifier influent.
* Removals based on raw water concentration (not filter influent
concentration).
** Removals based on influent clarifier concentration (not filter influent concentration).
removals surpassed those of conventional filtration. The authors speculated that the differences
between the full-scale oocyst removals were attributable to differences in filter influent quality.

Investigations of media type and design have demonstrated that these parameters have
little impact on oocyst removals by filters. Hall et al. (1995) did not find performance differences
between sand and dual-media filters when the filters had similar filtrate quality (measured by
turbidity). Other pilot studies found Cryptosporidium removals by sand, anthracite/sand, and
GAC/sand filters statistically comparable (e.g., Dugan et al., 1999; Swertfeger et al., 1999).
Investigations of media design (i.e., relative media depth) have also indicated that it does not
substantially impact oocyst removal by filters (e.g., Patania et al., 1995; Swertfeger et al., 1999).

Pilot-scale studies have suggested that differences in filtration rate in rapid gravity filters
do not substantially impact Cryptosporidium removal (e.g., Patania et al., 1995). Increases in
filtration rate also showed no adverse effects on oocyst removals by slow sand filters (Timms et
al., 1995). Investigations of the effects of filtration rate over time, however, have demonstrated
that oocyst removals can decrease by >1 log after a period of operation at a higher rate.

Patania et al. (1995) concluded that filter aid does not substantially impact oocyst
removal by filters, however, several authors have demonstrated that optimal chemical
pretreatment is critical to maximizing Cryptosporidium oocyst removals during filtration. Patania
et al. (1995) examined conventional filtration, low-rate surface filtration, and in-line filtration at
pilot-scale and demonstrated that filtration was ineffective for oocyst removal without chemical
pretreatment. Other pilot-scale studies have also indicated that suboptimal coagulation conditions
decrease oocyst removal by filters by at least 1 log (e.g., Charles et al., 1995; Ongerth and
Pecoraro, 1995; Dugan et al., 1999). Charles et al. (1995) concluded that rapid mix conditions
were less critical than coagulation and impacted oocyst removals by < 1 log.

Cryptosporidium removals of >3 log have been maintained during filter ripening, despite
a decrease in removals when compared to stable operation (e.g., Swaim et al., 1996). Several
pilot-scale studies have indicated that oocyst removals decrease by approximately 0.5-1 log
during filter ripening (e.g., Hall et al., 1995; Patania et al., 1995; Swaim et al., 1996). These
findings were confirmed at full-scale by Baudin and Lame (1998) who demonstrated an ~1 log
deterioration in oocyst removals during filter ripening.
27
Two studies have concluded that oocyst removals are comparable during turbidity
breakthrough and stable filter operation (Patania et al., 1995; Baudin and Laine, 1998). Patania et
al. (1995) noted that the filter effluent turbidity had only increased by approximately 0.1 NTU
during their evaluation of filter breakthrough. The authors speculated that oocyst removal might
have deteriorated if sampling had continued beyond this point.

28
CHAPTER 3: METHODS AND RESEARCH PLATFORMS

EXPERIMENTAL DESIGN

The research approach consisted of five specific tasks designed to provide concrete
outcomes of practical value to the water industry. These tasks were:

1. a literature review,
2. experiments investigating the influence of operational effects on pathogen passage in
benchmark systems (referred to as Examining Benchmark Systems),
3. experiments investigating strategies for mitigating filter operational effects on
pathogen passage (referred to as Mitigating Operational Effects),
4. development of a concept of robustness, and
5. synthesis of experimental outcomes into useful strategies for the water industry.

A review of the relevant filtration literature was critical at the onset of this research.
Although the filtration and particle removal literature is voluminous, limited information on
pathogen removal under conditions of process stress is available. The main objective of this
review (which has been presented in Chapter 2) was to consolidate and update reviews and
studies already completed and to focus on non-steady-state and suboptimal filtration
performance and its impact on particle and pathogen removal.

The first of the two experimental tasks defined removals of pathogens and performance
indicators (turbidity, particle counts, and aerobic spores) in benchmark systems (i.e., typical
design and operating conditions); these experiments are referred to as "Examining Benchmark
Systems." They were conducted at the Ottawa pilot plant in Canada and the Metropolitan Water
District of Southern California's (MWD) La Verne pilot plant in California. These locations
represent two basic types of coagulation: a relatively low dose optimized for particle removal
(MWD treating Colorado River water) and a relatively high dose for combined TOC and particle
removal (Ottawa River water). Inactivated Cryptosporidium parvum oocysts and pure-cultured
Bacillus subtilis spores were seeded at both locations. Inactivated Giardia lamblia, as well as
viruses (MS-2 bacteriophage) and bacteria (E. coli) were also seeded at MWD. Because

29
disinfection provides substantial bacterial reduction even during non-optimized coagulation or
filtration, bacteria were seeded in a limited number of experiments. The experiments were
designed to assess pathogen removal that could be reasonably expected from benchmark systems
and to evaluate the sensitivity of pathogen removal during typical filter cycles and conditions of
process stress. Six basic conditions were investigated during these experiments. They were:

1. controls,
2. stable filter operation,
3. suboptimal coagulation,
4. ripening,
5. breakthrough, and
6. hydraulic step.

The six conditions studied during the Examining Benchmark Systems experiments are
listed in Table 3.1 along with their rationale and specific experimental conditions. It will be
noted that several sub-conditions within suboptimal coagulation and breakthrough were
investigated. The "end-of-run" experiments at MWD were performed because it was not possible
to actually achieve breakthrough. Experiments for each of the six principal conditions were
conducted at least in triplicate at each location.

Table 3.1
Operating conditions examined during Examining Benchmark Systems experiments

General Operating Condition Specific Experiments Experimental Rationale


Controls no media in filter, no serves as control experiment to measure extent of
coagulation any seeded organism losses to seeding apparatus
and pilot plant
Stable operation warm water defines highest attainable removal and acts as
baseline to which other operating conditions are
compared
cold water (T < 3°C) defines highest attainable removal under cold water
(Ottawa) conditions
spring runoff (Ottawa) gauges effect of sudden water quality change
seeding at rapid mix establishes if pre-coagulation in a jar followed by
subsequent seeding to filter influent is
representative of pilot-plant coagulation and
subsequent filtration
(continued)

30
Table 3.1 (continued)

General Operating Condition Specific Experiments Experimental Rationale


Suboptimal coagulation coagulant underfeed demonstrates the role of pre-treatment
(including no coagulation) loss of coagulant aid addresses the relative importance of coagulant aid
(Ottawa) in pathogen removal
no coagulants since determines extent of coagulant impact on filtration
backwash (i.e., worst case scenario); also serves as control
experiment
temporary coagulation addresses effect of temporary coagulation failure
failure (Ottawa) (e.g., pump failure)
no pre-coagulation of determines if coagulant in seed suspension (which
pathogen seed suspension is fed into filter influent) acts as filter aid
(Ottawa)

pre-coagulation of seed determines if coagulant in seed suspension can act


suspension only (no plant as a filter aid without plant-level coagulation
coagulation) (Ottawa)
Ripening peak filter effluent turbidity determines pathogen passage associated with the
during ripening operational period that potentially represents one of
the most significant deteriorations of filtrate quality
Hydraulic step sudden increase in flow determines any deterioration in pathogen removal
associated with rapid changes in filtration rate
(e.g., another filter out of service) that are known
to cause deterioration of filtrate quality
Breakthrough end-of-run (terminal determines when pathogen passage increases
headloss - MWD) relative to baseline at the end of a filter cycle and
relative to indicators like turbidity
onset-of-breakthrough determines degree of pathogen passage at the end of
(Ottawa) a filter cycle relative to small changes in
indicators like turbidity
breakthrough (Ottawa) determines pathogen passage in relation to larger
changes in turbidity and particle counts
All experiments were performed at both MWD and Ottawa unless otherwise noted.

The filters at both pilot plants contained media depths and sizes typical of the utilities'
full-scale plants (and typical of many existing treatment plants). The operational mode chosen
was conventional treatment with dual-media filtration. Further operating details regarding the
research platforms are specified in a subsequent section of this chapter.

31
The second major experimental task examined strategies for mitigating operational
effects for the removal of pathogens. Various physical and chemical factors affecting pre-
treatment and filtration were investigated at bench- and pilot-scale. These experiments were
conducted with surrogates (turbidity and particle counts) for cost reasons. A one-to-one
correspondence between pathogens and surrogates was not expected, however, practically useful
relationships within a specific context (e.g., coagulation effects) were anticipated from the
Examining Benchmark Systems experiments. The concepts summarized in Table 3.2 guided the
experimental design of the Mitigating Operational Effects experiments. The specific factors
considered for the Mitigating Operational Effects experiments are listed in Table 3.3.

Table 3.2
Rationale for selection of design of Mitigating Operational Effects experiments

Key Issues Rationale Experimental Considerations


Design vs. operation Several key operating factors were Media design
investigated during the "Examining
Benchmark Systems" experiments.

Several design factors have been shown to Conventional vs. direct filtration
affect turbidity and particle removal, yet
their effects on pathogen removal and Presence vs. absence of pre-oxidation
process robustness are not clearly known.
Flow rate

Pathogen detachment vs. Pathogens exit filters because of non- Chemical factors likely contribute
non-attachment attachment or detachment (the release of most to non-attachment and can
previously deposited organisms). include:
coagulation regime
PH
use of filter aid

Physical factors likely contribute


most to detachment and can
include:
hydraulics
design factors

Pre-treatment Proper coagulation is critical for successful


(coagulation) particle removal performance; thus non-
optimal coagulation has the potential to
significantly compromise pathogen
removal.

Backwashing Backwashing has both physical and Backwash protocol


chemical aspects that may be important in
reducing pathogen passage, particularly Backwash chemistry
during ripening.___________________________

32
Table 3.3
Design of Mitigating Operational Effects experiments

Grouping Factor Research Platform


Chemical factors Coagulation chemistry Johns Hopkins University
pH (bench-scale)

Physical factors Flow rate Johns Hopkins University


Media (bench-scale)

Impact of pre-oxidation Oxidant presence and type MWD


Media (pilot-scale)
Filter aid

The Examining Benchmark Systems and Mitigating Operational Effects data were used to
develop a quantitative framework describing the robustness of pathogen removal by filtration;
this was done to provide utilities with a tool or framework with which to assess their own
treatment processes. The study results were also synthesized into recommendations of practical
value to the water industry in minimizing pathogen passage through filters.

FACILITIES DESCRIPTION

Pilot-Scale

The pilot-scale facilities used to evaluate the influence of operational effects on pathogen
passage are described below. The two benchmark systems during the Examining Benchmark
Systems experiments represented two basic types of chemical pretreatment: a relatively low
coagulant dose optimized for particle removal (MWD) and a relatively high coagulant dose for
combined TOC and particle removal (Ottawa). The pilot-plant process configurations and
nominal raw water qualities are listed in Table 3.4 and Table 3.5, respectively. Both pilot plants
were operated to mimic, as closely as possible, the full-scale process at their respective locations.
For the experiments examining bench-mark systems, this included prechlorination.

33
Table 3.4
Process configuration of pilot-scale plants

Utility / Process Specifics MWD Region of


Ottawa-Carleton
City La Verne, California Ottawa, Ontario
Plant name La Verne Pilot Plant Brirtania Pilot Plant
Design capacity, gpm 12 (6/train) 16 (8/train)
45 L/min 61 L/min
Preoxidation
Ozone yes" no
Chlorine yesa yes
Potassium Permanganate yes" no
Chemicals
Alum yes yes
Coagulant aid (cationic polymer) yes no
Coagulant aid (activated silica) no yes
Filter aid (non-ionic high MW polymer) yesb no
Rapid Mix
G, sec'1 600 in-line
Hydraulic detention time, min 1.7
Flocculation
G for stages 1,2, and 3, sec'1 75, 50, 25 60,40, 20
Hydraulic detention time, min. 20 30
Sedimentation
Hydraulic detention time, min. 80 100
Post-Sedimentation Oxidation
Ozone n/a n/a
Chlorine n/a n/a
Filtration: Examining Benchmark Systems
Hydraulic loading rate, gpm/ft2 4 2.7
Surface area, ft2 0.20 0.20
Filter 1 - media dual-media (20/8) dual-media (16/11)
anthracite/sand (in.)
Filtration: Mitigating Operational Effects
Surface area, fr 0.20 n/a
Filters 3 and 4
filter media dual-media
anthracite/sand, in. 20/8
hydraulic loading rate, gpm/ft2 4
Filters 5 and 6
Filter media tri-media n/a
anthracite/sand/ilmenite, in 20/8/3
hydraulic loading rate, gpm/ft2____________________4___________________
"Only one of these oxidants was used in a given experiment.

b Used only in specific "Mitigating Operational Effects" experiments.

34
Table 3.5
Nominal raw water quality of pilot-scale plants

Utility/Water Quality Metropolitan Water District of Region of


Southern California Ottawa-Carleton
City La Verne, California Ottawa, Ontario
USA Canada
Source water Colorado River Water (CRW) & Ottawa River
CRW/SPW blend

Qualitative description Imported from Colorado River. Large Few upstream inputs, some
of source water reservoirs dampen water quality logging in past
fluctuations.

Temperature, °C
nominal value 18 11
range 11-26 1-27

TOC/DOC, mg/L
nominal value 3 6.0
range 2.5-3 5-7.

Turbidity, NTU
nominal value 1.1 3
range 0.2-3.5 1-30

PH
nominal value 8.3 7.2
range 7.9-8.6 6.7-7.6

Alkalinity, mg/L as CaC03


nominal value 130 22
range 120-140 15-40

Hardness, mg/L as CaC03


nominal value 32
range 25-40

Total Dissolved Solids,


mg/L 650
nominal value 530-720
range

*Not available

35
La Verne Pilot Plant

MWD's La Verne pilot plant was selected as a benchmark systems utilizing a relatively
low coagulant dose optimized for particle removal. The La Verne pilot plant contains two trains,
each capable of operating at 22 L/min (6 gpm), and simulates a conventional filtration plant
(rapid mix-flocculation-sedimentation-filtration) with pre-ozonation. During this project, La
Verne treated Colorado River water (CRW) or a blend of CRW and State Project water (SPW).

Raw water could be preoxidized with ozone through four contactors (15.2 cm [6 in]
diameter x 5.2 m [17 ft] height), or with chlorine or permanganate in the rapid mix basin. It was
then coagulated using alum or ferric chloride and a cationic polymer, flocculated, and settled.
Clarified water was pumped to filters; filter aid could be added to this filter influent line. For
each train, three 15.2 cm [6 in] filters were operated in a constant-rate, rising-head mode. Each
filter included a fixed nozzle surface wash system and could be backwashed with either
chlorinated or chloraminated water. Continuous on-line measurements were recorded by a
supervisory control and data acquisition (SCADA) data logging program every 5 minutes. Other
operational specifics are presented in the context of individual experiments discussed in
subsequent sections of this report.

Ottawa Pilot Plant

The Ottawa pilot plant is constructed only of stainless steel, glass and inert fluorocarbons.
Raw water is typically pumped at 50-60 L/min (13-16 gpm) to a constant head tank. From the
constant head tank, the flow is split between two identical process trains. During this research,
Side 1 was operated with pre-chlorination.

Metering pumps inject treatment chemicals into the feed line; the chemicals are mixed in
line. The water then enters a 3-cell under/over flocculation tank. Flocculated water then passes
into an inclined plate sedimentation tank. Settled water is collected in a settled water storage
tank, before being sampled and fed to a dual-media filter (anthracite/sand) and a filter adsorber
(not used in this project). The filters, which are 15.2 cm (6 in.) diameter, can operate in either the
constant rate or declining rate mode (the constant rate mode was used in this research). Filtrate is
collected in a backwash water storage tank with dedicated storage cells. Water for backwashing

36
is then pumped from the dedicated cell back to the filter or contactor on the side of the plant
from which it was collected. An air compressor and injection port allow for air-scour during
backwashing. Continuous on-line measurements are recorded by a SCADA data logging
program every minute.

The experiments at Ottawa were conducted under the overall direction of the University
of Waterloo. Microorganism sample processing and analysis were conducted at the University of
Waterloo.

Bench-scale Investigations at Johns Hopkins University

Model Waters

Filter experiments and jar tests were conducted using model raw waters with
characteristics listed in Table 3.6. Turbidity was provided by 0.8 mg/L of hematite (a-Fe2O3)
synthesized following the procedures used in a previous study (Becker, 1995). Negatively
charged spherical polystyrene latex particles (Interfacial Dynamics Corporation, Portland,
Oregon) with a diameter of 4 urn were used as a surrogate for Cryptosporidium oocysts. Water
from the Great Dismal Swamp in southeastern Virginia was used as a source of dissolved natural
organic carbon. Two different batches of this water were used for the laboratory experiments at
the Johns Hopkins University. The first batch was collected in March, 1997 and used for the low
rate direct filtration experiments. The second batch was obtained in December, 1999 and used for
all the other experiments. The major characteristics of this water are listed in Table 3.7. Since the
chemical properties of the Dismal Swamp water have been consistent over time, the new batch of
NOM from this source is not expected to significantly differ from previous batches.

Jar Tests

The coagulant dosage for the direct filtration experiments was determined using
laboratory jar tests with a 48-hour flocculation period. This long time was used to provide
extensive contact opportunities so that the optimum dose obtained would not be limited by the
low turbidity and DOC of the water being coagulated and flocculated, thereby being a useful
estimate of the optimum dose for a direct filtration system. After one hour of sedimentation

37
Table 3.6
Characteristics of influent model waters used at Johns Hopkins

Parameter Target value Average value of model raw water


Turbidity 7.3 NTU 7.1 NTU
Particles Diameter = 4 um; Diameter = 4 um;
Concentration = Concentration =
1.1 x 103 particles/mL 1.1 x 103 particles/mL
Dissolved organic carbon 1.5mg/L* 1.4mg/L
5.0 mg/Lj 4.7mg/L
absorbance 1.00/I Ocm* 0.998/I Ocm
3.00/10cmt 2.89 710 cm
pH 7.0 7.01
NaHCOa 10"4 M (buffer) n.aj
NaCl 10'3 M n.aj:__________
* for direct filtration experiments
t for conventional mode filtration experiments
J n.a. = not available

Table 3.7
Characteristics of dismal swamp water used at Johns Hopkins

Parameter First batch* Second batchf


Dissolved organic carbon 54.6 mg/L 11 0.9 mg/L
UV254 absorbance 2.88 /cm 5.94 /cm
PH 4.15 3.71
SUVAJ 5.27 L/mg-m 5.36 L/mg-m
* collected on March 27,1997
t collected on December 4,1999
% specific ultraviolet light absorbance

following flocculation, a 50 mL sample was gently withdrawn from each jar using a wide mouth
pipette and measured for pH, settled turbidity, and settled UV254 absorbance. The samples were
then filtered through a 1.2 um glass fiber filter and measured for filtered turbidity, filtered UV254
absorbance, and DOC. During alum addition, pH was monitored and immediately adjusted to pH
7 using small amounts of 0.1 N NaOH. Reagent grade aluminum sulfate was used as the sole
coagulant (A12(SO4)3'18H2O). To allow appropriate comparison with the results obtained at the
University of Waterloo and MWD, the concentration of 18 HaO alum used in laboratory

38
experiments at Johns Hopkins was converted and expressed in this report as an equivalent
concentration of alum with 14.3 H2O (A12(S04)314.3 H2O). The optimum alum dosage for these
jar tests was found to be 9 mg/L (as Al2(SO4)3'14.3 t^O) and this dosage was used as a
benchmark dosage for the direct filtration experiments.

The optimum alum dosage for the conventional filtration experiments was determined
from jar tests using a standard coagulation and sedimentation protocol (30 minutes flocculation
at 25 rpm followed by 60 minutes sedimentation). For the model raw water used in this study,
the optimum alum dosage was found to be 36 mg/L for all experimental parameters tested. For
the conventional-mode filtration experiments, the dose of NaOH needed to achieve pH 7 was
determined for each desired alum dosage using a 500 mL sample of model raw water and was
then used in the batch pretreatment mode.

Filtration Procedure

Specifications for Filter Column and Dual-Media. A laboratory filtration apparatus was
constructed at Johns Hopkins. The filter column was fabricated of acrylic plastic, 1.0 m in height
and 51 mm in internal diameter. Anthracite and sand obtained from the Metropolitan Water
District of Southern California (MWD) were used in a dual-media bed. A depth of anthracite of
450 mm with a porosity of 50 per cent overlies 250 mm of sand media with a porosity of 46
percent. The densities of the two media are 1.58 kg/L and 2.64 kg/L for the anthracite and sand,
respectively. Each of the two filter media was thoroughly washed with distilled water to remove
fines - 30 times for anthracite and 15 times for sand - after which it was dried overnight at 110°C.
The media were then equilibrated with ambient humidity and temperature after which the desired
amounts were weighed and loaded into the columns. Then, the media were backwashed to
remove fine residues. Sieve analyses were conducted for the anthracite and sand media used in
the filtration experiments at the Johns Hopkins laboratory. Samples were taken from the same
batches of media used in the installation of the laboratory dual-media filter. The experimental
procedure followed the ASTM methods (ASTM, 1993). Specifications measured for the
anthracite indicate that it had an effective size of 0.96 mm and a uniformity coefficient of 1.35.
Similar specifications for the sand indicate an effective size of 0.38 mm and a uniformity
coefficient of 1.39.

39
The influent was fed with a pulseless gear, pump (Cole-Parmer Instrument Company,
Vernon Hills, Illinois). An inlet port was equipped with a diffuser to distribute the influent
evenly across the filter area at the top of the column. A stainless steel mesh was installed at the
bottom of the dual-media to support the bed. A particle counter (Met One, Model WGS267,
Grants Pass, Oregon) was connected in-line to the filter effluent.

Three sampling ports were installed on the column. One port located 10 mm above the
top of the anthracite media was used to obtain influent samples. A second port located in the bed
10 mm above the sand-anthracite interface was used to obtain samples of water that had been
filtered by the anthracite and represented influent to the sand layer. The third port was located at
the bottom of the filter column and was used to obtain samples of the dual-media filter effluent.
Headlosses were calculated from the measurements of piezometric head at these same three
locations.

Direct Filtration Procedure. Direct filtration experiments were conducted using a model
raw water with characteristics listed in Table 3.6. The effect of filtration rate was studied at low
(4.9 m/h [2 gpm/ft2]) and high (14.7 m/h [6 gpm/ft2]) filtration rates. The effect of pH was
investigated at pHs of 6, 7, and 8. The model raw water was prepared 24 hours prior to the
filtration experiments to provide sufficient time for the natural organic matter to equilibrate with
the hematite surfaces. Aluminum sulfate (A12(SO4)3'18 H2O) was used as the sole coagulant.
Sodium hydroxide (NaOH) was used to achieve the desired pH of the filter influent. The
concentration of NaOH feeding solution was determined for each desired alum dosage using a
500 mL sample of model raw water. All chemicals were fed using a dual-channel peristaltic
pump (Masterflex pump, Cole-Parmer Instrument Company, Vernon Hills, Illinois) with the flow
rate less than 2 % of the total flow rate of the filter influent.

Conventional Treatment Procedure. Filtration experiments were done in conventional


mode at both low and high filtration rates using a model raw water high in dissolved organic
carbon. All other parameters of the raw water remained the same as in the previous experiments.
Coagulation and sedimentation were conducted in a batch mode after which the settled water was
added continuously to the dual-media filter. The desired amount of alum was added to 400 L of
model raw water contained in a cylindrical tank (Cole-Parmer Instrument Company, Vernon

40
Hills, Illinois), and mixed by an impeller mixer (Cole-Parmer Instrument Company, Vemon
Hills, Illinois), followed by an addition of NaOH with the amount determined to yield a final pH
of 7. The dose of NaOH needed to achieve pH 7 was determined for each desired alum dosage
using a 500 mL sample of model raw water and was then used in the batch pretreatment mode.
Flocculation was facilitated using a paddle mixer (20 x 50 cm) at 25 rpm for 30 minutes. The
paddle was then removed and the water was allowed to settle quiescently for 3 hours. Three
hundred liters of water were gently withdrawn from just below the water surface, measured for
turbidity, particle concentration, UV254 absorbance and DOC, and used as an influent to
filtration. This settled water was then applied to the filter using a pulseless gear pump (Cole-
Parmer Instrument Company, Vernon Hills, Illinois).

Analytical Methods

Samples (50 mL) were taken from each sampling location and analyzed for turbidity
(Hach, 2100A, Loveland, Colorado), UV254 absorbance (Shimadzu, UV160, Columbia,
Maryland), and dissolved organic carbon (Tekmar-Dohrmann, Phoenix 8000, Cincinnati, Ohio).
Samples were filtered through 0.45 |^m cellulose membrane filters (Millipore, Bedford,
Massachusetts) for the measurements of UV254 and dissolved organic carbon. Particle
concentrations in the filter effluent were measured on-line using a laser diode light-blockage
particle counter (Met One, WGS267, Grants Pass, Oregon). Results for particles > 2 urn in size
are presented in this study. Aluminum sulfate ^2(804)3'18H2O) was used as sole coagulant,
with all dosages expressed as commercial grade alum, Al2(SO4)s 14.3H2O.

SEEDING AND SAMPLING METHODOLOGIES

All of the pilot-scale seeding experiments employed continuous seeding of


microorganisms during specific points of the filter cycle. Almost all of the seeding experiments
conducted during this study involved microorganism seeding at the filter influent; a very limited
number of experiments examined microorganism seeding at rapid mix. Seeding of the filter
influent was chosen based on previous experience (Yates et al., 1997a,b) and after careful
consideration by the project team. This protocol was selected in preference to seeding at the
rapid mix to ensure reliably detectable levels of seeded organisms in the filter effluent. Seeding
at the rapid mix would have entailed large and likely poorly reproducible losses through the
41
sedimentation tank and would have produced filter influent concentrations that varied widely
over the seeding time.

Prior to seeding, Cryptosporidium, Giardia, and Bacillus concentrations were determined


by triplicate counts on a hemocytometer. A pure culture of E. coli (ATCC 11775) was
maintained on R2A media. Prior to an experiment, an aliquot of this culture was streaked onto a
TSA (Tryptic-Soy agar) plate and incubated at 37°C for 24 hours. An aliquot of this culture was
inoculated into 50 mL TSB (Tryptic-Soy broth) and incubated at 37°C for 8 hours on an orbital
shaker set at 80 rpm. This actively growing suspension was washed twice by centrifugation-
resuspension in phosphate buffer (Standard Method, Method 9050 C.I a), and resuspended in 10
mL filter-sterilized filter effluent water. E. coli stock concentration was determined by triplicate
plating of appropriate dilution on m-Endo media, and incubated at 37°C for 24 hours.

MS-2 coliphage was kept at 4°C in saline-calcium solution. E. coli Famp (ATCC 15597)
was infected with MS-2 and inoculated on tryptone agar plates using the double agar layer
procedure. After a 24-hour incubation at 37°C, the plates were washed with 5 mL aliquots of
saline-calcium solution. The bacteria-phage suspension was centrifuged, and the supernatant
filtered on 0.2 um cellulose acetate membranes to remove the bacteria. The filtrate containing
the phage was stored in glass at 4°C. The MS-2 stock solution concentration was determined
using the double agar layer procedure and triplicate plating.

The feedstock microorganism suspension was prepared by adding these microorganism


spikes to a 1.5 L seed suspension of chlorinated and quenched raw water. Each vial containing
the concentrated microorganism stock was rinsed into the feedstock ten times with the
chlorinated/quenched raw water. Additional samples were collected from the feedstock to
confirm microorganism concentration. The targeted filter influent concentrations of the seeded
microorganisms were typically: 105 oocycts or cysts per liter (Cryptosporidium and Giardia),
104-106 spores per liter (Bacillus'), 106 CFU/L (E. coli), and 106 PFU/L (MS-2).

Jar Coagulation Protocol

During pilot-scale testing, liquid aluminum sulfate (alum) was used in combination with
cationic polymer (MWD) or activated silica (Ottawa) to achieve the desired conditions for

42
turbidity and particle removal (except in suboptimal coagulation experiments). The cationic
polymer or activated silica was dosed during rapid mix simultaneously with alum. Chlorine was
added at rapid mix at a dosage of ~2 mg/L. The chlorination dosage was to achieve the benefits
of pre-chlorination or pre-oxidation, but was not necessarily sufficient to meet Surface Water
Treatment Rule disinfection requirements. During the MS-2 and E. coli seeding experiments at
MWD, no chlorine residual was present at the seeding location because these organisms are
easily inactivated by free chlorine (Wolfe et al., 1989a). This was achieved by quenching the
chlorine residual with sodium thiosulfate.

Microorganism pre-treatment was achieved via a jar-coagulation. In most cases, the jar
coagulation conditions were identical to the pilot plant coagulation conditions during the specific
experiments. Some experiments were performed to specifically examine the effects of the
coagulant added to the seed suspension (e.g., potential filter aid effects); during these, the jar
coagulation conditions were different than the pilot plant coagulation conditions. The coagulated
microorganism suspensions were added at the filter influent (except in a few experiments where
they were added at the rapid mix) immediately after jar coagulation. The microorganism jar
coagulation protocol is provided in Table 3.8.

Seeding Protocol

A peristaltic pump was used to add the feedstock to the pilot plant filter influent water
(and in some cases the raw water). The seed suspensions were introduced into the filter influent
water approximately 0.6 to 0.9 m (2 to 3 ft) above the filter media so that mixing with the filter
influent water would occur prior to filtration. Microorganisms were typically seeded into the
filter influent for one hour, at a rate of 25 mL/min. A limited number of experiments required
microorganism seeding for longer periods of time (5 or 8 hours); the seed flow rates were
adjusted accordingly (to 5 and 3.1 mL/min respectively). The microorganism seed suspension
was continuously fed to the pilot plant filter influent until the last set of filter influent and
effluent samples was collected. In some experiments (e.g., those examining detachment-related
phenomena), additional samples were collected after microorganism seeding had ceased. A
second peristaltic pump was continuously operated to re-circulate filter influent water from a few
centimeters above the surface of the filter media. To prevent carryover of microorganisms

43
Table 3.8
Jar coagulation protocol

1 . While stirring pilot plant raw water on a stir plate, add NaOCl to achieve a dosed chlorine
concentration of 2 mg/L.
*
2. Stir chlorinated sample on stir plate for 15 minutes.
*
3. Quench chlorine by adding sodium thiosulfate (^28203) to a final 2:1 molar ratio of
3'.Cl, adding 2x more Na2S2Os as a safety factor.

4. Stir chlorinated/thiosulphate-quenched sample on stir plate for 5 minutes.


V
5. Pour 0.5 L of chlorinated/quenched raw water into a 2-L jar (to be used on the jar test
apparatus).
*
6. Vigorously shake and add the microorganism spike to the 1.0 L of chlorinated/quenched raw
water.
*
1. (a) Using an additional 0.5 L of chlorinated/quenched raw water, rinse the microorganism
spike container 10 times, adding all of the rinse water to the 2-L jar.
(b) Pour 0.5 L of chlorinated/quenched raw water into a 2 L jar (for a final volume of
-1.5 L).
*
8. Jar coagulate the samples with the jar test apparatus. This step is site specific (e.g. detention
times, rpms, etc.)
MWD: Ottawa:
- 90 sec @ 100 rpm - 5 sec @ 100 rpm
- 20 min @ 50 rpm - 15 min @ 30 rpm
- 80 min settling - 15 min @ 15 rpm
- 60 min settling
*
9. After settling, gently mix jar coagulated suspension so that floes are essentially broken up.

between experiments, both this line and the microorganism feed line were flushed for at least ten
minutes after the completion of an experiment. These lines were also flushed for several minutes
while the flow rates were confirmed, prior to the start of each experiment.

Sampling Protocol

Microorganism samples were collected at the filter influent and effluent locations. The
filter influent location was approximately 5 to 8 cm (2 to 3 in) above the surface of the filter

44
media; the effluent was collected at the column exit (upstream of the turbidimeter and particle
counter). Additional control samples were collected from the microorganism feedstock
suspension. Filter influent and effluent samples were collected in 250-mL and 1-L glass Wheaton
bottles, respectively. Aliquots from the Cryptosporidium and Giardia feedstock were collected in
5-mL glass chromatography vials. Bacillus spores, E. coli, and MS-2 were collected in 15 mL
polypropylene centrifuge tubes. Sampling containers for Cryptosporidium and Giardia were
washed, autoclaved, and rinsed with a few milliliters of a buffered detergent solution prior to use.
(l x phosphate buffered saline [PBS] with final concentrations of: 0.1% sodium dodecyl sulfate,
0.1% Tween 80, and 0.01% Sigma Antifoam A and final pH of 7.4) The excess surfactant
solution was discarded and sodium thiosulphate (^28203) was added to each sample bottle for a
final concentration of 0.01%. Prior to a seeding experiment, 1-L negative controls were collected
at the filter influent and effluent locations.

Most experiments were performed during the early to mid portion of the filter cycle, after
at least four hours of filter operation; ripening and breakthrough conditions were an exception to
this criterion. With the exception of ripening, all experiments were conducted after a period of
stable operation during which filter effluent turbidities were continuously below 0.1 NTU.
Microorganisms were typically seeded for one hour, as noted previously, with samples collected
at 15, 30, 45 and 55 (or 60) minutes after the start of seeding. Seeding and sampling information
is summarized in Table 3.9. Only Cryptosporidium and Bacillus sampling times are noted;
however, when Giardia, E. coli, and MS-2 were sampled, they were sampled at the same time as
Cryptosporidium.

Calculation of Microorganism Concentration and Removal

Filter influent and effluent microorganism concentrations were determined by the same
analytical method for a given microorganism at each laboratory. Microorganism removals (logio)
were calculated by subtracting the log of the filter effluent concentration from the log of the filter
influent concentration. When no microorganisms were detected, the concentration was reported
as 0, however, removal was calculated by using a concentration of 1 microorganism/sample
volume processed and normalized to 1 L. For example, a value of 1 oocyst/L would be used in

45
Table 3.9
Seeding and sampling specifics

General Operating Seeding Sampling


Condition
Stable.operation 1-hour seeding period (0-60 minutes) samples collected at 15, 30,45, and 55
minutes (Ottawa)
seeding during first half of filter cycle
after at least 4 hours of operation samples collected at 10, 20,40, and 60
minutes (MWD)
filter effluent turbidity < 0.1 NTU

Suboptimal coagulation 1-hour seeding period (0-60 minutes) samples collected at 10, 20,40, and 60
minutes (MWD)
seeding during first half of filter cycle
after at least 4 hours of operation samples collected at 15,30,45, and 55
minutes (Ottawa)

No coagulation 1-hour seeding period (0-60 minutes) samples collected at 15, 30,45, and 55
minutes (Ottawa)
seeding during first half of filter cycle samples collected at 10, 20,40, and 60
after at least 4 hours of operation minutes (MWD)

Ripening 30-minute seeding period capturing both samples collected at 5,10,15,20, and 25
before and after peak filter effluent minutes
turbidity during ripening

Hydraulic step seeding 20 hours into filter cycle (during


stable operation)

MWD 8.4-hour (502 minute) seeding period samples collected at 485, 500, 505, and
520 minutes
hydraulic step initiated at 502 minutes
(end of seeding)

Ottawa 5-hour (300 minute) seeding period samples collected at 280, 295, 300, 305,
310, 320, and 360 minutes
hydraulic step initiated at 300 minutes

End-of-run 1-hour seeding period (0-60 minutes) samples collected at 15, 30,45, and 55
minutes

Breakthrough 1-hour seeding period (0-60 minutes) samples collected at 15,30,45, and 55
minutes

December 1999 1-hour (-60-0 minutes) B. subtilis seeding samples collected at 15,30,45, and 55
experiments 1-hour (0-60 minutes) C. parvum seeding minutes

46
the log removal calculation if no oocysts were found in a 1-L sample; a value of 2 oocysts/L
would be used in the calculation if no oocysts were found in a 500-mL sample.

Microorganism Losses to Seeding Apparatus and Pilot Plant

Control experiments were conducted at MWD and Ottawa to determine microorganism


losses to the seeding apparatus and pilot plant equipment. These experiments consisted of
removing the media from the filters and seeding the columns with microorganisms to determine
system losses.

Cryptosporidium, Bacillus, and Giardia (MWD only) were seeded during the control
experiments. All of the media were removed from the filters, except at Ottawa where 2-3 cm of
support gravel could not be removed from the column. No coagulants were used in the pilot
plants or the jar coagulation step during these experiments. Cryptosporidium and Bacillus losses
at Ottawa are shown in Table B.4, while those at MWD are given in Tables B.10 and B.12.
Giardia losses at MWD are also shown in Tables B.ll and B.12. At each location, the mean
log 10 loss for a given microorganism was based on four replicate samples.

The control experiments at MWD indicated 0.02 ± 0.11, -0.03 ± 0.03, and 0.03 ± 0.08 log
loss (mean ± standard deviation) of Cryptosporidium, Bacillus, and Giardia, respectively.
Despite the few remaining centimeters of gravel at Ottawa, the Cryptosporidium and Bacillus
losses were only 0.09 ±0.12 and 0.15 ± 0.06 log respectively. The microorganism losses are
summarized in Figure 3.1. Overall, these results suggested that the oocyst, cyst, and spore
removals observed during the subsequent pilot-scale experiments were due to filtration, with
only minimal system losses. Combined with the analytical method recovery data discussed
below, these data demonstrated considerable reliability of the microorganism removal data
presented in this report.

47
C. parvum
. subf/7/s
G.

Pilot Plant
Figure 3.1 Microorganism losses to seeding apparatus and equipment during no coagulant
and no media control experiments (mean ± standard deviation).

ANALYTICAL METHODS

Cryptosporidium and Giardia

Microorganism inactivation and preservation

Samples of C. parvum and G. lamblia were preserved in a penicillin/streptomycin


solution because recent research has indicated that this preservative may more closely represent
oocyst behavior in the natural environment (Li et al., 1997). The cysts and oocysts were obtained
from a commercial laboratory (Waterborne, Inc., New Orleans, LA). For each experiment, ~107
o

cysts and ~ 10 oocysts were obtained; they were inactivated with 5 to 10% formalin (final

48
concentration) in IX PBS with 0.01% Tween 20 to prevent cyst/oocyst clumping. All
microorganism stocks were refrigerated at 4°C in the dark until use.

Enumeration of Spike Suspension

Prior to pilot-scale cyst/oocyst seeding, the stock suspension was briefly vortexed and a
small portion of the suspension (< 100 uL in total) was removed to enumerate the cyst or oocyst
concentration. The stock concentration was determined with a hemocytometer (Petroff-Hausser
Bacterial Counting Chamber, Hausser Scientific Corporation, Horsham, PA) in triplicate. The
entire hemocytometer grid was used in the cyst/oocyst enumeration process.

Analytical Protocol

Cryptosporidium and Giardia were measured in feedstock suspensions and filter influent
and effluent samples. Filter influents were analyzed in volumes of 10, 50 and 100 mL at MWD
and 2.5, 5, 10, and 100 mL at Ottawa. Filter effluents were analyzed in volumes ranging from 5
mL to 1 L, depending on the operating condition studied. Sample volumes were chosen to yield
between 10 and 2000 oocysts/cysts per membrane. All pipettes and glassware were pre-rinsed
with the buffered detergent solution to prevent cyst/oocyst losses by adsorbtion to the materials.

Processing varied little between samples collected at the various pilot plants. At MWD,
filter housings (Swinnex; Millipore Corp., Bedford, MA) containing pre-wetted, 27 mm, 0.45
um cellulose acetate filters (Sartorius Corp., Hayward, CA) were connected to disposable
syringes. The filters were blocked by passing ~2 mL of 1% bovine serum albumen (BSA)
through the membranes. Samples were then passed through the membranes at a vacuum pressure
of 10-15 psi. The syringes were rinsed with the buffered detergent solution and the membranes
were blocked again with 2 mL of 1% BSA. The membranes were removed from the housings
and placed on support membranes on a manifold (Hoefer Scientific, San Francisco, CA); weights
held the membranes down. This was followed by the immunofluorescence assay (IFA) described
below. The membranes were kept wet with IX PBS and covered until mounted on slides.

At the University of Waterloo, samples from Ottawa were filtered through 0.40 um
polycarbonate membranes. The filter membranes were placed on top of support membranes
placed on a manifold (Hoefer Scientific, San Francisco, CA) maintained at a vacuum pressure of
49
12.7 cm (5 in) of mercury; weights held the membranes in place. The membrane filters were
blocked by passing ~2 mL of 1% BSA through the membranes. Samples were then directly
filtered on the manifold. The glassware that had contained the samples was then rinsed with the
buffered detergent solution. The detergent rinse followed by an additional 2 mL of BSA was also
filtered through the membranes; this was followed by the IFA described below. If necessary, the
membranes were kept wet with IX PBS and covered until sample mounting on slides. Further
details regarding the development of this method for Cryptosporidium concentration are
provided in Emelko (2001).

All Cryptosporidium and Giardia identification was performed with IFA techniques
(USEPA, 1996), using the Hydrofluor combination Cryptosporidium and Giardia Kit (Strategic
Diagnostics, Newark, DE). Slides from MWD were shipped to a commercial laboratory (CH
Diagnostic & Consulting Services, Inc., Loveland, CO) for enumeration; slides from Ottawa
were enumerated at the University of Waterloo.

Recovery

None of the results reported in this study were corrected for method recovery.
Cryptosporidium recovery experiments were performed on waters from the MWD and Ottawa
pilot plants; MWD also performed Giardia recovery experiments. Recoveries were based on at
least five replicate samples of oocysts/cysts spiked into filter influent and effluent waters from
each pilot plant; spike concentrations were determined with a hemocytometer. The processed
sample volumes were those typically used during the experiments (1-10 mL of influent and 0.5-1
L of effluent). Cryptosporidium recoveries for samples processed at MWD (MWD water) ranged
from 16 to 54%, with a mean recovery of 31%; Giardia recoveries ranged from 28 to 90%, with
a mean of 53%. At the University of Waterloo, method recoveries for MWD and Ottawa waters
ranged from 50 to 86%, with a mean of 73%. The MWD Cryptosporidium and Giardia recovery
data are summarized in Table A.I and Table A.2, respectively; University of Waterloo
Cryptosporidium recovery data are summarized in Table A.3 and Table A.4.

The good protozoan recoveries were likely attributable to the direct filtration method that
did not require elution steps. The use of this method was possible due to the high seeded
cyst/oocyst concentrations and associated small sample volumes. Although some differences in
50
recoveries between the MWD and University of Waterloo methods exist, they were not
considered critical because of the overall consistency in recovery at each location. Most
importantly, there were no substantial deviations in recovery between filter influent and effluent
samples at either laboratory; therefore, the resulting log removals were essentially unaffected by
recovery differences.

Quality Assurance/Quality Control

Sample handling, identification, preservation, transportation and storage were done


according to established procedures as described in detail in the methods section and following
the USEPA ICR Methods for Protozoa Analysis (USEPA, 1996). Samples containing protozoa
were transported on ice and shipped via overnight courier (if necessary) for processing. The
samples were processed immediately or refrigerated as specified in the ICR methodology.

The quality assurance/quality control (QA/QC) programs at the University of Waterloo


and MWD ensured that accurate protozoan data were produced. The programs included recovery
studies and the analysis of spike suspension samples, negative controls from the waters studied,
and method blanks and controls. Because of the use of multiple laboratories in this project, inter-
laboratory comparisons of selected samples were carried out. One of these QA/QC experiments
involved conducting recovery studies by processing spiked samples into waters from the same
location (MWD filter influent and effluent) and processing them at both laboratories (MWD and
the University of Waterloo). As is evident in Table A.3 and Table A.4, recovery from MWD
water was comparable to that from Ottawa water when the same method was used, suggesting
the analytical method was consistent for all waters studied.

To ensure comparable slide reading of processed Cryptosporidium samples, several


samples were read at both the University of Waterloo and the commercial laboratory used by
MWD. Limited comparisons included sending slides twice to each laboratory. Table 3.10
summarizes the results of the inter-laboratory slide reading comparison; these data suggest
general agreement between both laboratories over the entire range of counts observed during the
course of this study. Although some considerable relative differences were observed between
samples counted at the two laboratories, the counts at a given laboratory were generally
internally consistent. As is demonstrated in Table 3.10, inter-laboratory differences in counts
51
Table 3.10
Quality assurance and control data comparing Cryptosporidium slides read at the University of
Waterloo and the commercial laboratory used by MWD

Sample UW Count Commercial Laboratory Difference Relative Log


Count toUW Difference
Location #1 (#/slide) #2 (#/slide) #1 (#/slide) #2 (#/slide) (log UW - log CL)*
Filter 6542 6780 — -4 -0.02
influent 6219 6445 — -4 -0.02
6859 6208 — 9 0.04
6632 6003 — 9 0.04
21 13 — 38 0.21
9 13 2 12 36 0.20
297 266 185 220 28 0.14
175 150 — 14 0.07
79 72 — 9 0.04
71 101 — -42 -0.15
9 7 — 22 0.11
1176 883 — 25 0.12

Filter 761 790 _ -4 -0.02


effluent 14 9 — 36 0.19
24 23 14 23 21 0.10
20 23 — -15 -0.06
19 19 3 0 92 1.10
43 45 53 45 -11 -0.05
0 0 — 0 0
0 0 — 0 0
Average log difference is calculated for samples read twice by each laboratory.

typically had a negligible impact on logio counts which were used to calculate the logio removals
that formed the basis of the conclusions in this report.

Bacillus

Enumeration of Spike Suspension

Prior to pilot-scale cyst/oocyst seeding, the microorganism stock suspension was briefly
vortexed and a small portion of the suspension (< 100 uL in total) was removed to enumerate the
cyst or oocyst concentration. The stock concentration was determined with a hemocytometer
(Petroff-Hausser Bacterial Counting Chamber, Hausser Scientific Corporation, Horsham, PA) in
triplicate. The entire hemocytometer grid was used in the cyst/oocyst enumeration process.
52
Analytical Protocol

The analysis for B. subtilis (ATCC 6051) was performed according to a previously
described method (Rice et al., 1996). This method generally consisted of filtration of samples
onto 47 mm, 0.45 um gridded cellulose acetate membranes (Pall Gelman Corporation #66278,
Ann Arbor, MI) and growth at 37°C for 24 hours on plates of nutrient agar with trypan blue
(0.015 g/L). Spores were identified by their blue color. Typically, duplicate sample volumes of
0.1 L and 1.0 L were used to enumerate filter influent and effluents respectively.

MS-2 Bacteriophage

Viral seeding studies were performed with the bacteriophage MS-2 (strain ATCC 15597-
Bl; American Type Culture Collection, Rockville, MD). The bacterial host was E. coli (strain
ATCC 15597). MS-2 was used as a model virus because it is similar in size (0.025 um), shape
(icosahedron [20-sided]), and nucleic acid (RNA) to poliovirus and hepatitis virus (as reported in
Jacangelo et al., 1997). This bacterial virus was also used for performing microbial challenge
experiments recommended by the U.S. Environmental Protection Agency (USEPA, 1990). Virus
propagation was performed using the double-agar overlay procedure (Adams 1959) as described
by Wolfe et al. (1989a). Effluent samples were collected and analyzed by the double-agar
overlay procedure. The results of the viral experiments were based on triplicate analyses.

Escherichia coli

E. coli culture and seeding procedures were similar to those described in Ferguson et al.
(1990) and Wolfe et al. (1989a,b). An E. coli strain isolated from MWD's raw water was
maintained in lauryl-tryptose broth and periodically streaked onto R2A agar to verify its purity.
Approximately 20 hours before each seeding experiment, 1 L of tryptic soy broth was inoculated
with 3 mL of a fresh lauryl-tryptose broth culture of E. coli. The culture was placed in a shaker
incubator, set at 35°C and 80 rpm, and incubated for -15 hours. The cells were then harvested
and washed twice with deionized water in a centrifuge at 5°C and 3,000 rpm x g for 10 minutes.
The supernatant was aspirated out, and the pellets suspended in 20 mL of sterile seed-water
solution. This suspension was then diluted into 1 L of sterile plant influent water. After seeding,
E. coli were enumerated in triplicate within 4 hours by membrane filtration onto m-Endo agar.

53
Selected colonies on M-Endo were streaked for purity and identified biochemically to ensure that
the organisms were E. coli of the same biotype used for the seeding experiments. The results of
the E. coli experiments were based on triplicate analyses.

Particle Counting

A standard protocol was used at all sites to verify the calibration of the particle counters
using commercially available, calibrated, mono-disperse polymer microspheres (Duke Scientific
Corp.; Palo Alto, CA). Each particle counting instrument was calibrated (by the manufacturer)
according to ASTM F 658-87 and met the resolution requirements of USP 788. The particle
counters measured total particles from 2-150 urn, with the data reported as cumulative particles
>2 urn. PCX particle counters (Met One, Grants Pass, OR) were used at MWD. An IBR particle
counter (IBR, Grass Lake MI) was used at Ottawa. A Met One, Model WGS267 particle counter
(Grants Pass, OR) was used for the bench-scale experiments at Johns Hopkins University.

Turbidity

At pilot-scale, turbidity was monitored using on-line turbidimeters that were calibrated
using dilute formazin solutions as specified by the manufacturer. Calibration was checked by
comparison with a bench-top turbidimeter with an accuracy of ±2%, using standards of 0.80 and
6.6 NTU. Hach model 1720C turbidimeters (Hach Co., Loveland, CO) were used at plant
influent, filter influent, and filter effluent locations at MWD and Ottawa. An additional turbidity
meter was used at the filter effluent sampling location in Ottawa (ABB Model 7997/201,
Calgary, AB). Filter influent turbidity was measured by grab samples analyzed with a Hach
Model 21 OOP hand held turbidimeter (Hach Co., Loveland, CO.).

Headless

Differential pressure transducers continuously measured headloss at the MWD and


Ottawa pilot plants.

54
Chemical Parameters

TOC samples were analyzed by the ultraviolet/persulfate oxidation method (APHA,


AWWA, WEF, 1995) using an organic carbon analyzer. DOC included filtration with a pre-
washed 0.45 urn nylon filter. UVa54 was measured using a spectrophotometer set at a wavelength
of 254 nm. These samples were first filtered through a pre-rinsed 0.45 um filter to remove
turbidity, which could interfere with UV measurement. Alkalinity was determined by titration
(APHA, AWWA, WEF, 1995). Sample pH was measured with grab samples analyzed by a pH
meter that was calibrated daily, using pH 7.00 and 9.18 or 10.0 buffer solutions.

55
CHAPTER 4: INFLUENCE OF OPERATIONAL EFFECTS ON BENCHMARK
SYSTEMS

INTRODUCTION

The objective of the experiments described in this chapter was to assess the influence of
operational effects on benchmark systems. This work was conducted as Task 2 of the project. In
this task no attempt was made to optimize or improve pathogen removal. Rather the goal was to
document performance of these systems under optimal and a variety of other conditions.

Chapter 3 described the experimental facilities, key raw water quality parameters in each
location, experimental design and procedures for these experiments. Key points are summarized
here for the convenience of the reader.

The experiments were performed using two pilot-scale facilities. The locations were
chosen to encompass two different coagulation regimes. The Ottawa location also experienced
low temperatures (1°C). Each pilot plant was operated to mimic as closely as possible the full-
scale treatment plant at the same location. Cryptosporidium parvum (C. parvwri) and Bacillus
subtilis (B, subtilis) were seeded at both locations. Additional microorganisms were seeded in
some experiments at MWD (Giardia lamblia, MS-2 bacteriophage and Escherichia coli [E.
coli]). Turbidity, particle counts and filter hydraulic performance were monitored in each
experiment.

Pilot-scale, dual-media filters at the Ottawa pilot plant and at MWD were seeded with
jar-coagulated suspensions of ~108 formalin-inactivated C. parvum oocysts. A total of ~107 B.
subtilis spores were seeded in Ottawa during initial experiments and then 109 for the remainder
of the study. At MWD 109 spores were seeded for all experiments. The Ottawa pilot plant used a
high coagulant dose (-40 mg/L alum and 2 mg/L activated silica) to achieve both TOC and
particle removal. The MWD pilot plant used a low coagulant dose (5 mg/L alum and 1.5 mg/L
cationic polymer) for particulate removal only. Chlorine (~2 mg/L) was added at rapid mix as a
pre-oxidant at both pilot plants. The optimized coagulation conditions were selected to meet the
0.1 NTU turbidity goal of the Partnership for Safe Water, a voluntary treatment optimization

57
program sponsored by the U.S. Environmental Protection Agency and the American Water
Works Association.

The seeding protocol has been described in detail in Chapter 3. Except for a small
number of experiments that are specifically identified later, seeding of microorganisms was done
into the filter influent. In many experiments this occurred over a period of approximately one
hour, although in some experiments this seeding period was extended. During this seeding
period, several pairs of filter influent and effluent samples were taken for the enumeration of the
seeded microorganisms. Each experimental condition was replicated several times.

The six basic conditions investigated during these experiments were:

1. controls,

2. stable filter operation,

3. suboptimal coagulation,

4. ripening,

5. breakthrough, and

6. hydraulic step.

The control experiments, involving no coagulation addition and no media in the filters,
were performed to quantify the possible losses of seeded organisms to the system.

The reference condition that provided the benchmark for comparison was termed stable
operation. This represented the optimal pathogen removal performance that could be obtained in
each location and involved sampling in the middle of the filter cycle.

The additional operational conditions that were investigated involved two in which the
pilot-scale systems were deliberately manipulated away from optimal performance, and two that
examined portions of the normal filter cycle during which performance is expected to be less
than optimal. The deliberate manipulations examined suboptimal coagulation and a sudden

58
increase in flow to the filter (hydraulic step experiments). Both of these operational conditions
may occur to varying degrees in full-scale treatment plants.

The portions of the filter cycle examined where non-optimal removals are expected were
ripening and breakthrough. Ripening is an unavoidable part of every filter cycle, although in
some plants the water produced during this period is filtered to waste rather than sent to the
distribution system. In principle, turbidity or particle breakthrough at the end of a run can be
avoided by terminating the run sufficiently early. There is of course a cost associated with this
shortening of the filter run. A particular focus of the experiments conducted was the beginning
stage of breakthrough (termed early breakthrough or onset-of-breakthrough), when filter effluent
quality might still be considered acceptable by traditional turbidity or particle criteria.

In this chapter, each of these experimental conditions is discussed in a separate


subsection. Within each subsection, the results from the two plant locations (Ottawa and MWD)
have been integrated.

As discussed in Chapter 3, the data collected for the seeding experiments consisted of
four or five sample pairs taken from the filter influent and filter effluent at each location. The
filter influent and filter effluent samples were collected over a one-hour period when the seed
suspension was added at the filter influent. Each of these tests was then repeated at least three
times. The single-factor, analysis of variance (ANOVA) statistical test was selected to interpret
the data using a standard statistical package (Microsoft Excel 97, version SR-2). Table 4.1 shows
the statistical comparisons performed during Task 2.

Appendix B contains the detailed Task 2 data, including experimental schedules, raw
water quality, and experimental results for microorganisms, particles and turbidity.

CONTROLS

As mentioned previously, control experiments were performed to quantify the possible


losses of seeded microorganisms to the pilot plant systems. One of these experiments was
performed at each location: on August 4, 1998 at MWD and on August 24, 1999 at Ottawa. In
these experiments there was no media in the filters and no coagulant was added. As was standard
59
Table 4.1
Statistical comparisons for Task 2

Stable Filter Suboptimal Ripening Breakthrough Hydraulic Step


Operation Coagulation
Control 0 __ D

Operation
Suboptimal 6,
Coagulation
Ripening
Breakthrough
Hydraulic Step

0 Denotes tests where statistical comparisons were performed.

practice, the microorganisms were seeded in the filter influent. Thus these experiments were
designed to give an indication of possible adsorption of seeded microorganisms on surfaces
within the pilot plant, including any sample tubing. Additional control experiments in which
filter media was present but coagulants were not added were also conducted. These latter
experiments will also be discussed in the section on Suboptimal coagulation because they in fact
represent a total coagulant feed failure.

Results for the control experiments are summarized in Table 4.2. In both locations the
removals of C. parvum and B. subtilis were less than 0.10 log units. Removals of G. lamblia and
particles were monitored at MWD and they were in the same range. It should be kept in mind
that for particles the comparison is made between plant influent (rather than filter influent as in
the case of the seeded microorganisms) and filter effluent, because filter influent particle counts
were not measured. At both locations the removals for all parameters are calculated on the basis
of four influent-effluent sample pairs.

These results convincingly show that there were essentially negligible losses of seeded
microorganisms to the pilot plant apparatus. Therefore, the removals attributed to filtration under
each of the tested operating conditions can be attributed to the filters themselves.

60
Table 4.2
Summary of pilot plant seeding data during control experiments
Date Experiment Research Log Removal (Mean ± 1 Standard Deviation) Filter Effluent Value
Platform (Mean ± 1 Standard Deviation)
C. parvum B. subtilis E. coli MS-2 G. lamblia Particles Particles Turbidity
__________________________________________________________.___________(#/mL)_____(NTU)
8/24/99 No Coagulants in Ottawa 0.09 ±0.12 0.1 ± 0.06 ...
Plant/Jar, No Media
8/4/98 No Coagulants, MWD 0.02 ± 0.11 -0.03 ± 0.03 - - 0.03 ± 0.08 -0.2 ± 0.01 7400 ± 0.00 1.00 ± 0.00
No Media
1.0 i.o
C. parvum Removal B. subtilis Removal

Ottawa MWD Ottawa MWD


0.5 -• - •o.s

0.0 •• -•0.0

-0.5 - • • • -0.5

-1.0 -1.0
Nocoag.; Extended duration; Nocoag.; Extended duration; Nocoag.; Extended duration; Nocoag.; Extended duration;
no media nocoag. no media nocoag. no media nocoag. no media nocoag.

Figure 4.1 Results of control experiments without coagulant (with and without filter media)

Figure 4.1 summarizes the results of control experiments with and without filter media
(all without coagulation). The box-and-whisker plot shows the minimum, 25th percentile,
median, 75th percentile, and maximum values for removals. Removals were very similar in all
cases, and the median removal was never higher than 0.25 logio.

STABLE FILTER OPERATION

Experimental Design

A number of stable operation experiments were carried out at both pilot locations over
the course of the Task 2 investigations. The purpose of these experiments was to document the
best removals that could be obtained under optimal conditions in each location. As mentioned
previously, the seeding and sampling were conducted in the early to mid portion of the filter
cycle, after ripening was complete. Because these experiments provided a baseline against which
all other conditions were compared, they were conducted throughout Task 2 to ensure that the
baseline did not shift. In all, eight stable operation experiments were conducted at Ottawa, and

62
eleven at MWD. In all of these experiments, the microorganisms were seeded at the filter
influent, as was standard practice in Task 2.

In addition, several stable operation experiments were conducted in which seeding was
done at the rapid mix. One of these experiments was conducted in Ottawa, while two were done
at MWD. The purpose of these tests was to document filter performance when the seeded
microorganisms had been exposed to the entire pilot plant coagulation, flocculation and
sedimentation sequence. The reasons why this approach was not used as the normal seeding
protocol have already been discussed in Chapter 3.

Results and Discussion

A summary of the results for the stable operation experiments is presented in Table 4.3.
Detailed results may be found in Appendix B. The results are discussed in detail below.
However, the most striking finding is that there was more than a two logic difference in C.
parvum removals between the two locations, despite essentially the same effluent turbidity
values and very similar (and low) filter effluent particle counts. The reasons for this difference
are not definitively known.

At Ottawa, 5.5 ± 0.4 logio removal of C. parvum was obtained, while at MWD 3.0 ± 0.7
log 10 was observed. The filter influent concentrations in both locations were similar:
approximately 106 per liter in Ottawa and 105 to 106 per liter at MWD. C. parvum was always
found in the filter effluent samples at MWD, typically at concentrations of at least one hundred
oocysts per liter. (At Ottawa, filter effluent C. parvum concentrations were usually less than ten
oocysts per liter and often zero.) Thus analytical problems with enumerating oocysts in the filter
effluent samples are not considered responsible for the lower removals observed at MWD.
Because of the high observed removals, the Ottawa data were scrutinized very carefully by the
project team. The team is confident in the results obtained.

Although there was some variation in removals calculated on the basis of individual
influent-effluent sample pairs, the calculated removals from run to run were quite reproducible as
indicated by the relatively low standard deviations. The Ottawa experiments included runs at low
temperature. No deterioration in performance was observed at temperatures as low as 1°C.

63
Table 4.3
Summary of pilot plant seeding data during stable operation
Date Experiment Research Log Removal (Mean ± 1 Standard Deviation) Filter Effluent Value
Platform (Mean ± 1 Standard Deviation)
C. parvum B. subtilis E. coli MS-2 G. lamblia Particles Particles Turbidity
(#/mL) (NTU)
8/6/98 Stable Filter Operation Ottawa 4.9 ± 0.21 3.2 ± 0.29 3.7 ± 2.9 0.02 ± 0.00
9/9/98 Ottawa 5.7 ± 0.06 - 3.8 ± 0.10 0.9 ± 0.2 0.02 ± 0.00
9/23/98 Ottawa 5.8 ± 0.03 - - - 2.8 ± 0.24 8.7 ± 5.6 0.03 ± 0.00
10/6/98 Ottawa 5.8 ± 0.15 4.6 ± 0.18 0.2 ± 0.1 0.02 ± 0.00
3/9/99 Ottawa 5.2 ± 0.38 2.1 ± 0.14 - - 4.1 ± 0.10 0.4 ± 0.1 0.03 ± 0.00
5/31/99 Ottawa 5.6 ± 0.20 4.6 ± 0.05 - - 3.7 ± 0.18 1.2 ± 0.6 0.03 ± 0.00
7/27/99 Ottawa 5.6 ± 0.02 4.5 ± 0.24 - - 3.0 ± 0.22 5.1 ± 1.5 0.04 ± 0.00
1/19/00 Ottawa 5.3 ± 0.36 4.2 ± 0.01 - - *
4.8 ± 0.6 0.03 ± 0.00
7/15/98 MWD 2.6 ± 0.07 2.0 ± 0.13 - 4.6 ± 0.13 2.2 ± 0.09 6.3 ± 1.2 0.05 ± 0.00
7/28/98 MWD 3.3 ± 0.07 2.7 ± 0.26 - 4.5 ± 0.37 2.6 ± 0.04 4.5 ± 0.4 0.05 ± 0.00
8/18/98 MWD 4.1 ± 0.65 2.3 ± 0.17 .8 ± 0.30 3.4 ± 0.04 1.5 ± 0.2 0.05 ± 0.00
9/22/98 MWD 3.8 ± 0.16 1.9 ± 0.46 .8 ± 0.37 4.4 ± 0.49 2.8 ± 0.01 5.2 ± 0.2 0.05 ± 0.00
9/29/98 MWD . 3.2 ± 0.37 0.6 ±0.11 .3 ± 0.06 3.3 ± 0.14 3.1 ± 0.4 0.05 ± 0.00
10/27/98 MWD 3.2 ± 0.15 2.3 ± 0.17 0.5 ± 0.20 .0 ± 0.09 4.2 ± 0.88 2.5 ± 0.14 10 ± 3.2 0.05 ±. 0.00
1 1/24/98 MWD - 2.1 ± 0.86 .1 ± 0.11 2.0 ± 0.02 32 ± 1.4 0.05 ± 0.00
12/15/98 MWD 2.9 ± 0.11 1.8 ± 0.06 0.3 ±0.17 .1 ± 0.06 1.8 ± 0.02 37 ± 1.8 0.05 ± 0.00
2/9/99 MWD 2.1 ± 0.15 1.9 ± 0.07 0.6 ± 0.25 .0 ± 0.08 2.2 ± 0.01 21 ± 1.7 0.06 ± 0.00
3/9/99 MWD 2.4 ± 0.18 1.9 ± 0.04 .3 ± 0.04 2.9 ± 0.07 3.9 ± 0.6 0.05 ± 0.00
4/27/99 MWD 2.9 ± 0.29 1.9 ± 0.06 0.5 ±0.14 .0 ± 0.05 2.2 ± 0.02 27 ± 1.2 0.05 ± 0.00

Ottawa Average 5.5 ± 0.37 3.8 ± 7.07 _ _ 3.6 ± 0.63 3.1 ± 5.5 0.03 ± o.o;
MWD Average 3.0 ± 0.66 2.2 ± 0.52 0.5 ± 0.18 1.4 ± 0.44 4.4 ± 0.51 2.5 ± 0.50 13.8 ± 12.6 0.05 ± 0.00

7/20/99 Stable - Seeded at Rapid Ottawa 1.3 ± 0.93 1.1 ± 0.92 . . 3.0 ± 0.10 5.7 ± 1.5 0.06 ± 0.01
Mix
3/16/99 Stable - Seeded at Rapid MWD 0.2 ± 0.06 2.2 ± 0.02 0.7 ±0.12 2.2 ± 0.13 2.9 ± 0.05 3.4 ± 0.4 0.06 ± 0.00
Mix
5/1 1/99 MWD 2.4 ± 0.12 2.2 ± 0.02 - - 2.7 ± 0.03 9.5 ± 0.8 0.05 ± 0.00

MWD Average 1.3 ± 1.17 2.2 ± 0.02 . - 2.8 ± 0.12 6.5 ± 3.3 0.05 ± 0.00
Removals of B. subtilis were lower in both locations than for C. parvum: 3.8 ± 1.1 logio
in Ottawa, versus 2.2 ± 0.5 logio at MWD. Although removals were substantially higher in
Ottawa, the difference between the two locations was not as great as for C. parvum. In both
locations the seeded concentration of B. subtilis was lower than that of C. parvum. Although B.
subtilis spores were invariably detected in the filter effluent samples at Ottawa and always at
MWD, it is possible that the lower seeded concentrations contributed to lower observed
removals. However it would appear that B. subtilis gives a conservative indication of a filter's
ability to remove C. parvum under stable operating conditions. It would also appear that
differences in B. subtilis removals in different filters indicate to at least some extent differences
to be expected in C. parvum removals. The overall reproducibility of the calculated B. subtilis
removals in these experiments was almost as good as for C. parvum.

The other microorganisms (G. lamblia, E. coli and MS-2) were seeded only at MWD. An
average G. lamblia removal of 4.4 + 0.5 logio was observed. This is higher than that seen at
MWD for C. parvum, as would be expected based on other investigations. Removals of E. coli
and MS-2 under stable operation conditions were very low: 0.5 ± 0.2 logio and 1.4 + 0.4 logio
respectively. Lower removals are expected for particles of this size. As a surrogate for C.
parvum, G. lamblia will in all likelihood overestimate removals, whereas the other two
organisms would not appear to be useful surrogates for C. parvum.

Table 4.3 also summarizes particle removals (> 2(im) and particle filter effluent
concentrations for the two locations. In Ottawa the mean particle removal during the stable
operation experiments was 3.6 ± 0.6 logio, while at MWD it was 2.5 ± 0.5 logio. Mean filter
effluent particle numbers in the two locations were approximately 3 per mL in Ottawa and 14 at
MWD. The MWD average was influenced by several runs with effluent particle numbers above
20 per mL. It will be recalled that for particle counts and turbidity, removals are based on raw
water rather than filter influent values, because particles were not measured in the filter influent.
Although raw water values were roughly similar in both locations (on the order of 5,000 per mL)
it must also be kept in mind that different particle counters were used. Further, many of the filter
effluent particle counts were at or near the detection limit of the instrument, particularly in
Ottawa.

65
Given these qualifications, it is not possible to quantitatively compare the particle
removals determined for stable operation in this study to pathogen removals (C. parvum and G.
lamblid) under the same conditions. However, different observed particle removals in different
filters (with roughly similar influent particle counts) may be indicative of differences in C.
parvum and G. lamblia removals by these filters.

It is questionable whether the observed different filter effluent particle counts in Ottawa
and MWD represent a real difference. However, given the substantial difference in C. parvum
removal between these two locations, it is possible that small differences in particle counts may
be indicative of measurable differences in C. parvum removal of the two filtration systems.
Certainly in the breakthrough experiments discussed later in this chapter, small increases in
effluent particle counts later in the run in Ottawa signaled a much greater deterioration in the
filter's ability to remove C. parvum.

Filter effluent turbidity values for the two locations are also shown in Table 4.3. As with
the particle data, the values shown correspond to the times at which the microorganism samples
were taken. Log removals for turbidity are not calculated because they are limited both by the
relatively low influent values and also by the fact that many filter effluent values are very close
to the instrument detection limit. In all but one stable operation experiment, filter effluent
turbidity was at or below 0.05 NTU. The mean filter effluent turbidity in Ottawa was 0.03, while
that at MWD was 0.05 NTU. Both of these values are considered to be indicative of excellent
filtration performance and it could be argued that there is no meaningful difference between the
overall average values obtained in the two locations. If there is a real difference, it is extremely
subtle and could not be reliably used to predict the differences in C. parvum removal observed at
the two locations.

Table 4.3 also summarizes results for the two experiments at MWD and one at Ottawa
where microorganisms were seeded at the rapid mix in the pilot plant rather than being jar
coagulated offline and seeded at the filter influent. It is immediately apparent that the
microorganism removals are much lower than for seeding in the filter influent. At Ottawa, filter
effluent particle counts and turbidity are slightly above the mean values in the normal stable

66
operation experiments, although they still represent excellent performance. At MWD, the
corresponding values are within the range of normal stable operation performance.

An important factor to keep in mind when interpreting the results of seeding at the rapid
mix is the much lower filter influent concentrations of seeded organisms. This can be observed
by examining the detailed data tables in Appendix B. These much lower influent concentrations
are due to substantial losses through the sedimentation step and also because of the fact that the
seed is dispersed over a longer period of time because plug flow conditions do not exist through
the flocculation and sedimentation basins. At MWD, filter influent C. parvum concentrations
were on the order of 3,000 oocysts per mL, which is at least two orders of magnitude lower than
when seeding at the filter influent. One of the MWD experiments gave essentially no C. parvum
removal although removals for other parameters were observed. The reason for this apparently
anomalous result is not known. In the other experiment the calculated removal (2.4 logio) was
within the range of removals reported for seeding at the filter influent. At Ottawa, the filter
influent levels were on the order of several hundred oocysts/L and no oocysts were detected in
the filter effluent samples. Therefore, the calculated removals should be treated as a lower
bound. Similarly, filter influent concentrations of B. subtilis spores were quite low in Ottawa and
close to normal levels at MWD. For MWD, the calculated removals were the same as for normal
stable operation although the number of experiments involving rapid mix seeding is small.
Overall, the results of experiments involving seeding at the rapid mix are not considered to
accurately reflect the oocyst removal capabilities of the filters in either pilot plant.

SUBOPTIMAL COAGULATION

Experimental Design

As discussed previously, two coagulation regimes were evaluated as pretreatment in this


study: a low dose optimized for particle removal (MWD treating primarily Colorado River
water) and a high dose for combined TOC and particle removal (Ottawa River water).

Chapter 3 discussed major water quality characteristics and operating conditions of the
two pilot plants. It will be recalled that the coagulant dosage in Ottawa was nearly eight times

67
greater than at Metropolitan and the coagulation pH was lower. Both of the pilot plants received
raw water which was low in turbidity (averages of 1.6 to 2.5 NTU at Ottawa and 0.6 to 1.1 NTU
at Metropolitan) and particles (averages of 4,400 to 7,300 particles/mL at Ottawa and 3,000 to
4,000 particles/mL at Metropolitan). Major differences between the raw waters included
temperature (Ottawa's coldest water was 1°C, whereas Metropolitan's coldest water was 13°C)
and alkalinity (Ottawa's alkalinity was ~20 mg/L as CaCOs, whereas Metropolitan's alkalinity
was 110-130 mg/L as CaCO3).

To investigate coagulation, three basic types of experiments were performed. They were
(1) no coagulation; (2) optimized coagulation; and, (3) suboptimal coagulation.

No coagulation. These experiments served as controls to determine microorganism


losses through the pilot-plant facilities and also to estimate the worst-case condition of total
coagulant failure. Previous seeding of C. parvum oocysts at MWD's pilot plant indicated a loss
of approximately 0.3-log (50 percent) of oocysts when no chemicals were added to the water
(Yates et al., 1997a,b). In the tests reported here, each seeding experiment consisted of four
replicate filter influent samples and four replicate filter effluent samples taken at 15-min
intervals over one hour. At Ottawa, coagulants were removed from the treatment process in
several different scenarios. These scenarios were as follows:

1. Extended duration coagulant loss with no coagulant in either the pilot plant or the
jar-test apparatus. This experiment compares most closely to the MWD controls
because coagulation was discontinued and the filter then backwashed prior to
microorganism seeding.
2. Short-term coagulant loss with no coagulant in either the pilot plant or the jar test
apparatus. These tests were performed when the pilot plant coagulation was turned
off a relatively short period of time prior to seeding the microorganisms. These tests
may simulate the actual performance of a treatment plant which experiences a short-
term failure of the primary coagulant feed system. In these short-term failures,
substantial amounts of floe are likely present in the filter and may act as an effective
collector of the microorganisms.

68
3. Short-term coagulant loss with no coagulant in the pilot plant. In these experiments,
no coagulant was fed to the pilot plant but coagulants were added to the jar-test
apparatus.
4. Short-term coagulant loss with no coagulant in the jar. In these experiments,
coagulant was fed to the pilot plant, but no coagulant was added to the jar-test
apparatus. The previous condition and this condition may be used to provide
information on the primary mechanisms of microorganism control by coagulation.
5. Extended duration coagulant aid loss. In this experiment the activated silica feed was
discontinued, but otherwise coagulation remained as normal. This experiment, while
simulating a coagulant aid failure, was conducted to investigate whether the use of
activated silica was an important factor in the high observed removal of C. parvum
oocysts under optimal coagulation conditions in Ottawa.

Optimized coagulation. This experiment set (termed Stable Operation and discussed in
the last section) served as another control and was used to determine the maximum removal of
oocysts and endospores through the pilot-pant filters. It is this optimized condition which has
been the focus of much previous work (Patania et al., 1995; Yates et al., 1997a,b; Nieminski &
Ongerth, 1995). In this experiment, seeding continued for one hour and four replicate samples
were taken from the filter influent and filter effluent at 15-min intervals. Seeding began 2-4
hours into the filter run, which was after the filter ripening period. Previous seeding of
C. parvum oocysts at MWD's pilot plant showed removals of >3 logic during stable filter
operation (Yates et al., 1997b).

Suboptimal coagulation. Because the residence time for water in treatment plants is
relatively short, in locations where influent water quality can change rapidly, coagulant dosage
optimization can be difficult to attain. Even though the treatment plant may be capable of
producing low-turbidity water when high-turbidity water consistently enters the plant, the speed
of the raw water quality changes relative to the coagulation response may lead to increased
particle passage. The slow response of the coagulation process to changing raw water conditions
was implicated in the Milwaukee cryptosporidiosis outbreak (Fox and Lyttle, 1996). In the
present investigation, the suboptimal coagulation experiments determined the effects of changing
coagulation conditions (without a change in raw water quality) on pathogen passage. The

69
experiments, where the coagulant dose was changed 40 to 60 percent from optimum, were then
compared to the no-coagulant and optimized coagulation tests. The targeted suboptimal turbidity
was 0.2-0.3 NTU. In some tests, however, the target effluent turbidity was exceeded.

Results and Discussion

Figure 4.2, Figure 4.3 and Figure 4.4 summarize the removal of seeded C. parvum and B.
subtilis and naturally present particles at both locations. Results for the various partial
coagulation scenarios in Ottawa are discussed later. Complete data may be found in Appendix B.

Each removal is expressed as the logio difference between paired sets of data taken at the
filter influent and filter effluent. The number of data points used in the statistical comparisons
(single-factor analysis of variance) is shown on the figures. The 'Yes' designation means that the
mean for a given condition was statistically different from the optimal or stable operation
condition at the 5% level.

In general, similar trends were seen for all three parameters at both locations. Suboptimal
coagulation adversely affected removals in a substantial way. At both Metropolitan and Ottawa,
significantly greater logic removals were obtained during optimized coagulation (i.e., 2-4 hours
into the filter cycle when effluent turbidity was 0.10 NTU or less) than during suboptimal
coagulation or coagulant failure. At both locations, average removals were reduced by just over 2
logio under suboptimal coagulation. All differences were statistically significant at the one
percent level (a = 0.01) as well as at the five percent level. Removals of C. parvum were higher
than those of B. subtilis under both optimal and suboptimal conditions. The differences in B.
subtilis removals between suboptimal and optimal conditions were also substantial and were
statistically significant at the one percent level (a = 0.01). The decrease in particle removals in
both locations was also statistically significant compared to optimal conditions. (It will be
recalled that particle removals are calculated based on raw water values.) As discussed
previously, the essentially zero removals of both organisms for the no-coagulant condition
confirmed that seeded organisms were not being lost in the pilot plants.

70
Metropolitan Ottawa
/ -

n = 32 Yes; '•
"S 6-
I— — I n=12 - -6
M : n = 35
_O
L_!!_J ;
"5

. 1 4'
O
-4
5
•• '. -3
!§ "' "" Yes;
0 n=12
|2- Yes; -2
6
0 1-
—j— n=12 Yes;
n=8 '• - 1

0- = ^ rl r^n -0
Optimal Suboptimal No Optimal Suboptimal No
Coagulants Coagulants

Figure 4.2 Effect of coagulation on removal of C. parvum by filters

Metropolitan
7 . ——————————————— Ottawa
r———————————————————————————————————————————— ,

6- . -6
/—s
n = 20
60 ^ • •5
.2
**_-' n = 44

!'•
• •

1 4- •4
Yes; -3
n=12
^*

i
'§ 2- Yes; Yes; •2
05 n=ll Yes; „ 8
1 - n=12 • 1

0-
=^=
dn •0
Optimal Suboptimal No Optimal Suboptimal No
Coagulants Coagulants

Figure 4.3 Effect of coagulation on removal of B. subtilis by filters

71
/ -
Metropolitan Ottawa -/
6- -6
n = 28 :
CtO 5 "
-5
NW-

1 4- '. n = 4/1 -4
o i i

Yes; ;
1 n=12 -3
— 1 i Yes:
Yes:
I? 2 - n=12 n=8 : -2

__ ± =
cu
1 - 1 | 1 Yes; - 1
n=12
n- CZ3CZD .n
Optimal Suboptimal No Optimal Suboptimal No
Coagulants Coagulants

Figure 4.4 Effect of coagulation on removal of particles by filters (based on raw water and
filter effluent particle counts)

The suboptimal coagulation conditions were achieved by reducing the optimum


coagulant dosage (alum and polymer at Metropolitan or alum and activated silica at Ottawa) by
40 to 65 percent. At Metropolitan, the coagulant reduction resulted in an average effluent
turbidity of 0.16 NTU, which is well below the IESWTR specified level of 0.3 NTU. At Ottawa,
the coagulant reduction resulted in an average effluent turbidity of 0.56 NTU. The suboptimal
coagulation experiments at Ottawa varied more substantially than at Metropolitan, most
noticeably in terms of C. parvum and turbidity. This reflected the greater difficulty in 'hitting'
the target suboptimal conditions in Ottawa, and may indicate that such regimes are very
vulnerable to under dosing (close to the coagulant demand).

The relationship between seeded C. parvum or B. subtilis and particle removal was
examined under all of the coagulation conditions (results not shown). At Metropolitan, C.
parvum and B. subtilis removal were highly correlated to particle removal (r2 values of 0.87 and
0.82, respectively). At Ottawa, the strength of the correlation was not as high (0.60 and 0.25 for
C. parvum and B. subtilis, respectively).

72
Figure 4.5 and Figure 4.6 show the effect of coagulation condition on filter effluent
turbidity and C. parvum removal. The C. parvum removals are shown for individual influent-
effluent data pairs. In Figure 4.5 (optimal conditions) turbidity was always less than 0.1 NTU.
Although Ottawa C. parvum removals were almost always greater than MWD's, in each location
removals calculated from individual influent-effluent sample pairs varied considerably. This
shows the need for replication in this type of work.

Figure 4.6 (suboptimal coagulation conditions) shows that in all MWD experiments of
this type the filter effluent turbidity was in the range of 0.1 to 0.2 NTU. Some Ottawa data are
available for this range, but in other Ottawa experiments the effluent turbidity was closer to 1
NTU. When turbidity was 0.1 to 0.2 NTU, the Ottawa C. parvum removal did not appear to
decrease, whereas MWD's did. This would suggest that the sensitivity of turbidity for monitoring
coagulation impacts on C. parvum removal may be site specific and perhaps dependent on the
coagulation regime used. It is possible that particle counts may be a more sensitive indicator of
poor coagulation performance.

Figure 4.7 and Figure 4.8 show removals of C. parvum and B. subtilis under various
'coagulant removal' scenarios in Ottawa. The 'extended duration; no coagulant' data are the 'no
coagulant' results discussed just above. When coagulant was absent for only a short duration
(several hours prior to and during seeding), C. parvum removals were seriously impaired (by
several log units) but at least 2 log removal did occur. B. subtilis removals were reduced by
about the same extent under this condition, but because they were lower to begin with, they were
essentially zero. The absence of coagulant in the jar had a very limited effect as long as
coagulant was present in the pilot plant. Hypothetically, this may indicate that the presence of
coagulant in the filter is more important for C. parvum removal than coagulation of the oocysts
themselves. The absence over several hours of coagulant in the pilot plant (although it was
present in the jar) had a somewhat more noticeable but still modest effect on removals. This
lends support to the hypothesis mentioned just above.

73
• MWD
o Ottawa
4--

3--

-I-
0.5 1.0 1.5
Effluent Turbidity (NTU)

Figure 4.5 C. parvum removal vs. filter effluent turbidity (optimal coagulation conditions)
(Source: Huck et al. 2000. Reprinted figure 2 with permission from Springer-Verlag.)

.00
f 5"
•MWO
5 o Ottawa
E 4 o
0
|3 °0

' 0

1 ••

0
0.0 0.5 1.0 1.5
Effluent Turbidity (NTU)
Figure 4.6 C. parvum removal vs. filter effluent turbidity (suboptimal coagulation conditions)
(Source: Huck et al. 2000. Reprinted figure 3 with permission from Springer-Verlag.)

74
/ /

I6'
o

^^•^MM
6
S3"
CS
|-y-| ' ^5
e
£ 4- '4

3
o
S
12"
2

1
o-" —————|————— ————|———————————————|———————————————
0
no media no coag. no coagulants no silica; w/alum no coagulants no coagulants
in plant in jar

Figure 4.7 Effect of various 'no coagulant' conditions on removal of C. parvum by filters in
Ottawa

/ - • /

§ 6- •6

Jl 5- •5
"e3
o 4- •4

PH ^ - C=*= -K= . •3

1 2- LjtJ
•2
o
a
09 1 ' • 1

•— •*• • —i— .. ' T !——i——I T l——i———————i———————i——————; . A

Nocoae.; Extended Short duration; Extended Short duration: Short duration;


no media duration; no coag. duration no coagulants no coagulants
no coag. no silica; w/alum in plant in jar
Figure 4.8 Effect of various 'no coagulant' conditions on removal of B. subtilis by filters in
Ottawa

75
The absence of activated silica for the entire run had essentially no effect on the removal
of either organism. Although it is possible that some silica remained in the filter from previous
runs, this finding suggests that the use of silica (which was one difference between Ottawa and
MWD) was not responsible for the very high C. parvum removals seen in Ottawa.

In summary the results of the coagulation investigations at two locations indicate that
even at filter effluent turbidity levels less than 0.3 NTU, substantial deterioration of filtration
performance may result if coagulation is not optimized. C. parvum removals were more sensitive
to coagulation conditions than was turbidity removal. The sensitivity of turbidity for measuring
coagulation impacts on C. parvum removal may depend on the coagulation regime. Filter
effluent particle monitoring may provide a sensitive measure of coagulation performance and C.
parvum removal. Plants using a relatively high alum dose (such as Ottawa), may be able to
provide some reduced level of C. parvum removal by filtration during a short term (several hour)
coagulant feed failure. (A short term coagulation failure was not tested at MWD).

RIPENING

Experimental Design

To assess the effect of ripening on the passage C. parvum and B. subtilis (and E. coli,
MS-2, and G. lamblia at MWD) through filters, several experiments were conducted during
ripening at both the Ottawa and MWD pilot plants following several optimized filtration cycles.
Increased microorganism passage during filter ripening (relative to optimal operation) was
expected based on the literature (Chapter 2).

During ripening at Ottawa, the filter effluent turbidity and particle concentration typically
increased during the first several minutes of the experiment and then decreased toward the end of
the seeding period. The seeding period during the ripening experiments at Ottawa was only 30
minutes because the filter typically ripened to stable operating conditions of filter effluent
turbidities < 0.1 NTU and particle concentrations < 5-10 particles/mL during the first 30 minutes
of filter operation after backwashing. Samples were collected one after another at Ottawa, at 5,
10,15,20, and 25 minutes after seeding commenced.

76
During ripening at MWD, the filter effluent turbidity and particle concentration also
typically increased during the first few minutes of the experiment and then decreased toward the
end of the seeding period. The duration of the ripening period of the MWD filter was
approximately 30-40 minutes, very similar to that of the Ottawa filter. Microorganisms were
seeded for one hour at MWD. Note that for the first ripening experiment (October 6, 1998),
sampling times of 15, 30, 45, and 60 minutes were used; during the subsequent ripening
experiments, samples were collected at 10, 20, 40, and 60 minutes to capture more of the filter
spike. Also, during the experiment on October 6, seeding was started approximately ten minutes
after the start of the filter run, whereas during the other ripening experiments, seeding
commenced immediately at the start of the filter run. Therefore, the first sample for the first
ripening experiment was collected approximately fifteen minutes later into the filter run than
during the other ripening experiments.

Results

Unlike the stable filter operation experiments, both traditional performance measures
(turbidity and particle counts) and filter effluent microorganism concentrations varied throughout
the ripening period. Peak filter effluent turbidity and particle counts during ripening at Ottawa
ranged from 0.41 to 0.69 NTU and 91 to 840 particles/mL respectively. The turbidity and
particle response of the Ottawa filter during the November 10, 1998 ripening experiment is
shown in Figure 4.9 (It should be noted that particle counts are shown on a logarithmic scale and
turbidity on an arithmetic scale.) This figure depicts a sharp filter effluent turbidity and particle
peak approximately 2 to 3 minutes after the filter returned to service after backwashing. The
seeding period and sampling times are noted in Figure 4.9 and indicate that microorganism
seeding captured the entire period of elevated turbidity and particles associated with filter
ripening while sampling captured most of the spikes; these spikes lasted less than 30 minutes.
The filter effluent turbidity and particle counts peaked at approximately 0.67 NTU and 840
particles/mL during the November 10, 1998 experiment (Figure 4.9). During the October 27,
1998 experiment filter effluent turbidity and particle counts peaked at 0.69 NTU and 174
particles/mL respectively; they peaked at 0.41 NTU and 91 particles/mL during the November 3,
1998 experiment. Although the magnitude of the turbidity and particle spikes varied somewhat
between experiments, the duration of the ripening period was generally comparable among the

77
10000 1.0
Seeding Particles > 2 \im
A A A A
Turbidity
Sampling times
1000 0.8

0.6
(N
Al

o
t 0.4

0.0
0.5 1.0 1.5 2.0
Filter Run Time

Figure 4.9 Turbidity and particle response of filter during ripening experiment on
November 10, 1998 at Ottawa pilot plant

three ripening experiments conducted at Ottawa. The instantaneous turbidity and particle data
(which did not necessarily indicate the peak turbidity and particle concentrations that occurred
during ripening) are available in Appendix B (Table B.3).

C. parvum passage through the Ottawa filter during ripening appeared to have a dynamic
pattern similar to that of turbidity and particles; consistent with the elevated turbidity and particle
counts, more oocysts were detected at the filter effluent during ripening. Filter effluent C.
parvum and particle concentrations of the Ottawa filter during the November 10, 1998 ripening
experiment are shown in Figure 4.10. (Again, it should be noted that particle counts are shown
on a logarithmic scale and oocyst numbers on an arithmetic scale.) This figure indicates that the
elevated filter effluent oocyst concentration that occurred at 5 minutes was concurrent with

78
10000 150
Particles > 2pm
C. parvum (oocysts/L)

0.5 1.0 1.5 2.0


Filter Run Time (hours)

Figure 4.10 Particle and C. parvum response of filter during ripening experiment on
November 10,1998 at Ottawa pilot plant

10000 8000
Seeding Period
Particles > 2pm
B. subtilis (CFU/L) - • 7000

1000

0.5 1.0 1.5 2.0


Filter Run Time (hours)

Figure 4.11 Particle and B. subtilis response of filter during ripening experiment on November
10, 1998 at Ottawa pilot plant

79
elevated particle counts and somewhat elevated turbidity (Figure 4.9). This 110 oocyst/L value
during filter ripening at Ottawa on November 10, 1998 accompanied the highest instantaneous
particle concentration during microorganism sampling (145 particles/mL) and one of the highest
turbidity values (0.19 NTU) achieved during sampling during the ripening experiments. During
the other two ripening experiments (October 27 and November 3, 1998), maximum oocyst
concentrations were somewhat lower (82 and 62 oocysts/L respectively). The corresponding
maximum particle counts were 31 and 58 particles/mL and the maximum turbidities were 0.09
and 0.32 NTU respectively.

As can be seen from Figure 4.11, B. subtilis spore passage through the Ottawa filter was
somewhat consistent with the C. parvum data, however, filter effluent spore concentrations
during ripening did not always track the turbidity and particle trends as closely as did the oocyst
levels. The detailed instantaneous C. parvum and B. subtilis data are available in Appendix B
(Table B.4).

The turbidity response of the MWD pilot plant filter during October 20, 1998 ripening
experiment is shown in Figure 4.12. Figure 4.12 shows a clear ripening peak of turbidity at
approximately 10 minutes after the filter returned to service. Microorganism seeding captured
the entire turbidity spike at MWD, which lasted approximately 30 minutes. Instantaneous
turbidity values associated with microorganism seeding are available in Appendix B (Table B.9).

The particle response of the MWD filter for the October 20, 1998 ripening experiment is
shown in Figure 4.13. The particle response does not show the initial increase that was observed
with turbidity (Figure 4.12) because the particle count data were only being reported at five-
minute intervals. However, the data do show a marked decrease from ~1,000 particles/mL to -10
particles/mL within 30 minutes. The filter effluent particle concentration also appeared to be
influenced by the seed suspension that was fed to the filters. Shortly after the seeding was
terminated, the filter effluent particle concentration decreased from ~15 particles/mL to ~7
particles/mL (Figure 4.13), however, this effect was considered minor since these filter effluent
particle concentrations were all generally consistent with those observed during the stable
operation experiments at MWD. Such an effect was not observed at Ottawa, however. Although
it could very likely have been associated with microorganism seeding during the ripening

80
0.25
Seeding duration
• Plant Influent
Sampling times
-Filter Effluent
•0.2 ^
S

0.15 I"

•0.1

£
C
0.05

-I-
0.5 1.0 1.5 2.0 2.5 3.0
Filter Run Time (hours)

Figure 4.12 Turbidity response of filter during ripening experiment on October 20, 1998 at
MWD pilot plant

10,000

Seeding duration
1,000 T Sampling times • Plant Influent
• Filter Effluent

I
|
100 •
Al

1
10 •

0.0 0.5 1.0 1.5 2.0 2.5 3.0


Filter Run Time (hours)

Figure 4.13 Particle response of filter during ripening experiment on October 20,1998 at
MWD pilot plant

81
experiment, it was not observed during other experiments. The effect in Figure 4.13 also could
have been somewhat coincidental with the continued ripening of the MWD filter.

Consistent with the moderately elevated turbidity and particle counts, C. parvum and B.
subtilis were also detected at increased levels at the filter effluent during the ripening
experiments at MWD. During the October 20, 1998 experiment, filter effluent levels of
C. parvum and B. subtilis peaked at 264 oocysts/L and 750 CFU/L respectively. The measured
turbidity and particle counts at these sample times were 0.17 NTU and 31 particles/mL
respectively. The other ripening experiments conducted at MWD yielded similar peak filter
effluent C. parvum concentrations, turbidity, and particle counts of 243 oocyst/L, 0.07 NTU, and
44 particles/mL on October 6, 1998; 478 oocysts/L, 0.16 NTU, and 49 particles/mL on October
14, 1998; 712 oocysts/L, 0.15 NTU, and 36 particles/mL on November 3, 1998; and 198
oocysts/L, 0.14 NTU, and 68 particles/mL on November 10, 1998. The detailed turbidity and
particle data are available in Table B.9. The detailed C. parvum and B. subtilis data are available
in Appendix B (Table B.10).

E. coli, MS-2 bacteriophage and G. lamblia (during one run) removal during ripening
were also examined at MWD. While filter effluent G. lamblia spikes during ripening were
similar to those observed for C. parvum, spikes in the filter effluent concentrations of E. coli and
MS-2 bacteriophage were not reproducibly observed. These data are summarized in Appendix B
(Table B. 11).

Removals obtained during ripening are presented as box-and-whisker plots for C.


parvum, B. subtilis and particle counts in Figure 4.14, Figure 4.15 and Figure 4.16, respectively.
The removals are compared to stable (optimal) operation; the differences were tested for
statistical significance at the 5% level. The average removals obtained during each ripening run
at Ottawa and MWD are summarized in Table 4.4.

Discussion

Moderate increases in filter effluent C. parvum concentrations as a result of ripening were


observed at both Ottawa and MWD (Table B.4 and Table B.10); this result was also true for B.
subtilis and G. lamblia (based on one experiment) passage at MWD (Table B.ll). These data

82
Metropolitan Ottawa

Yes;
= 32 n=14
6- • -• 6

n = 35 No;
1 5- • n = 20 •• 5

4.. ••4
Q$
E
3 •• ••3
t
••2

•• 1

Optimal Ripening Optimal Ripening

Figure 4.14 Effect of ripening on removal of C. parvum by filters

Metropolitan Ottawa

6- Yes; ;• 6
n=15
n = 20
5- • -5

§ •
¥ 4- n = 44 • -4

j| 3 • • -3
Yes;
^ n = 20
2• • -2

| J_
1 • ' -• 1
,

'
n.
Optimal Ripening Optimal Ripening

Figure 4.15 Effect of ripening on removal of B. subtilis by filters

83
Table 4.4
Filter performance (removal of turbidity, particles, C. parvum, B. subtilis, E. coli, MS-2, and G. lamblia) during Task 2 ripening
experiments
Date Experiment Research Log Removal (Mean ± 1 Standard Deviation) Filter Effluent Value (Mean ± 1 Standard Deviation)
Platform
C. parvum B. subtilis E. coli MS-2 G. lamblia Particles Particles Turbidity
(#/mL) (NTU)
10/27/98 Filter Ripening Ottawa 5.0 ± 0.67 1.4 ± 0.84 - - - 3.1 ± 0.48 9.0 ± 12 0.07 ± 0.02
00 3.0 ± 0.46 21 ± 22 0.17 ± 0.09
11/3/98 Ottawa 5.1 ± 0.73 2.1 ± 0.69
11/10/98 Ottawa 5.0 ± 0.75 2.9 ± 1.93 - - . 2.6 ± 0.67 36 ± 61 0.16 ± 0.05
10/6/98 MWD 3.0 ± 0.52 .6 ± 0.24 0.1 ± 0.10 . - 1.9 ± 0.09 33 ± 8.5 0.06 ± 0.00
10/13/98 MWD 2.8 ± 0.11 .8 ± 0.23 0.7 ± 0.32 0.3 ± 0.67 3.9 ± 0.58 2.2 ± 0.14 29 ± 14 0.09 ± 0.05
10/20/98 MWD 3.4 ± 0.28 .4 ± 0.31 0.2 ± 0.15 1.0 ± 0.04 2.3 ± 0.20 21 ± 14 0.09 ± 0.06
1 1/3/98 MWD 2.4 ± 0.51 .3 ± 0.22 0.4 ± 0.15 1.1 ± 0.27 2.3 ± 0.23 18 ± 12 0.08 ± 0.05
MWD 3.2 ± 1.10 .5 ± 0.28 0.4 ± 0.20 1.0 ± 0.20 2.2 ± 0.30 30 ± 25 0.08 ± 0.04
11/10/98
. _ _ 2.9 ± 0.55 21.9 ± 37 0.13 ± 0.07
Ottawa Average 5.1 ± 0.66 2.1 ± 1.36
MWD Average 2.9 ± 0.64 1.5 ± 0.28 0.4 ± 0.27 0.9 ± 0.49 2.2 ± 0.23 26. 1 ± 15.3 0.08 ± 0.04
Metropolitan Ottawa

6- -6

n = 28
5 - -5

oJ Yes;
I 4' n=15 -4
[ n = 44 •

1 Yes-
JJ 3 - n = 20 • -3

2- | T | •2

1 - - 1

n. .n
Optimal Ripening Optimal Ripening

Figure 4.16 Effect of ripening on removal of particles by filters (based on raw water and filter
effluent particle counts)

suggested reduced attachment of these microorganisms during ripening as compared to stable


filter operation. These microorganism passage trends generally tracked changes in filter effluent
turbidity and particle counts. In other words, filter effluent oocyst levels decreased during the
ripening period. Although the instantaneous turbidity and particle count data (Table B.3 and
Table B.9) were generally indicative of filter performance, the actual values were not necessarily
indicative of C. parvum passage through the filters because specific particle count or turbidity
values were not necessarily correlated with specific filter effluent C. parvum concentrations.

The box and whisker plots and statistical comparisons (Figure 4.14, Figure 4.15 and
Figure 4.16) were based on the entire run and include the entire range of removals observed
during the defined sampling periods. The effect of ripening on filter effluent C. parvum and B.
subtilis concentrations was generally correlated with filter effluent particle counts and turbidity,
however, the elevated values for these parameters were short-lived and did not result in dramatic
differences in overall removals relative to stable (optimized) filtration. At Ottawa, C. parvum and
B. subtilis removals during ripening were 5.1 ± 0.7 logio and 2.1 ± 1.4 logio respectively (mean ±
standard deviation); at MWD they were 2.9 ± 0.6 and 1.5 ± 0.3 logio respectively.
85
During ripening, the median removals of C.parvum were lower than during stable
(optimal) filter operation (Figure 4.14). Oocyst removals during ripening at Ottawa were
significantly different (5% significance level) from those observed during the stable operation
experiments; the C. parvum removal experiments failed to yield a statistical difference between
stable operation and filter ripening at MWD. When only early ripening at MWD was considered
(sample times of 10, 20, and 40 minutes), however, the difference between oocyst removals
during stable filter operation and ripening became statistically significant (5% significance level,
comparison not shown). In general, these data suggested a brief, minimal to moderate increase in
C. parvum passage through filters that was concurrent with elevated filter effluent turbidity and
particle counts. As indicated previously, the detailed C. parvum concentration and removal data
are available in Appendix B (Table B.4 and Table B.10). (It should be noted that these
experiments were designed to evaluate the passage of oocysts present in the filter influent during
ripening, not the passage of oocysts that might be present in the backwash remnant water. The
significance of the latter would be site-specific but could lead to increased oocyst passage during
ripening in some instances.)

Trends in B. subtilis removal by the pilot.filters (Figure 4.15) were qualitatively


comparable to those observed for C. parvum. At Ottawa, B. subtilis removals during ripening
were lower and significantly different (5% significance level) from those achieved during the
stable filter operation experiments; the same result was found at MWD. For B. subtilis the
differences between stable operation and ripening were greater than for C. parvum. The relative
differences between removals of B. subtilis and C. parvum during ripening and stable operation
indicate that B. subtilis spores are probably not good quantitative surrogates for C. parvum
oocyst removal by filtration.

G. lamblia, E. coli, and MS-2 bacteriophage removal data during ripening for the MWD
filter are presented in Table B.ll and summarized in Table 4.4. Higher logio removal of G.
lamblia was obtained than for C. parvum. However, because of the limited availability of G.
lamblia, this experiment was only conducted once. Considerably lower removals of E. coli and
MS-2 were obtained than for the other organisms. For E. coli, 0.5 ± 0.2 logio removal was
obtained during stable filter operation. This very low removal, although not unexpected, was
complicated by the presence of some E. coli in the control sample; nonetheless, it indicated that

86
filtration is not as effective a barrier to bacterial passage as to larger cysts and oocysts. Likewise,
the MS-2 bacteriophage removal was 1.5 ± 0.5 log 10. MS-2 is significantly smaller than the other
organisms and this lower removal was also not unexpected.

When compared to stable filter operation, the removal of particles by the pilot filters also
decreased during ripening at both Ottawa and MWD (Figure 4.16); although not large, this
difference was statistically significant (5% significance level) at both locations. These data were
generally consistent with the microorganism removal data in indicating that ripening resulted in
lower removals relative to stable operation and that the Ottawa pilot filter achieved higher
removals than the MWD filter regardless of the operating condition (stable operation or
ripening). Although generally indicative of treatment efficiency and oocyst removal during stable
operation and ripening at MWD, particle removals as measured in this study were inadequate as
quantitative surrogates for oocyst removal at Ottawa (Figure 4.14 and Figure 4.16).

Filter effluent particle concentrations, turbidity, oocyst concentrations and spore


concentrations were elevated for a relatively short period of time during the ripening experiments
at Ottawa and MWD. However, the decrease in oocyst removal during ripening at Ottawa was
approximately 0.5 logio and was statistically significant (5% significance level), whereas the 0.1
logio difference at MWD failed to demonstrate a significant difference from stable operation.
The Ottawa result was consistent with the findings of Patania et al. (1995), Hall et al. (1995), and
Swaim et al. (1996) which demonstrated a 0.5 to 1.0 logio deterioration in oocyst removal during
filter ripening. The MWD result was consistent with other studies, such as LeChevallier et al.
(199la), which did not yield statistically different oocyst removals between ripening and stable
filter operation. The MWD G. lamblia results were also generally consistent with those reported
in the literature (Logsdon et al., 1981) and demonstrated higher cyst passage through filters
during ripening as compared to stable filter operation. The results reported in the present study
were less dramatic than those reported by Logsdon et al. (1981), however.

87
BREAKTHROUGH

Experimental Design

Both non-attachment and detachment occur during breakthrough conditions (Ginn et al.,
1992; Moran et al., 1993b). As particle detachment and non-attachment increase, increased
pathogen passage through filters would also be expected. To assess the passage through filters of
C. parvum and B. subtilis (and E. coli, MS-2, and G. lamblia in limited experiments at MWD)
several experiments were conducted at MWD and Ottawa. All of the experiments were
performed after periods of stable operation.

For the experiments at MWD, the headless was allowed to rise above the terminal
condition of 183 cm (6 ft.). Unfortunately, neither turbidity nor particle breakthrough occurred at
the terminal headloss condition. Therefore, the experiments were performed as "end-of-run"
experiments at approximately 72 hours into the filter cycle with approximately 183 cm (6 ft.) of
total headloss. The filters overflowed to drain at ~260 cm (8.5 ft.) of total headloss; thus, the
filter run lengths (and resultant headloss) could not be increased substantially in order to force
breakthrough. Jar-coagulated C. parvum, B. subtilis, E. coli, MS-2, and G. lamblia were seeded
into the filters for one hour; samples were collected at 15, 30, 45, and 60 minutes after the start
of seeding.

For the experiments planned for Ottawa, breakthrough was defined as the period after
which filter effluent turbidity exceeded 0.2 NTU. Since breakthrough is difficult to predict,
seeding and sampling did not necessarily commence immediately at filter effluent turbidities of
0.2 NTU. Similar to the end-of-run experiments at MWD, jar-coagulated C. parvum and B.
subtilis were seeded into the filters for one hour; samples were collected at 15, 30, 45, and 55
minutes after the start of seeding.

Early breakthrough experiments were also conducted at Ottawa; they were termed
"onset-of-breakthrough." Filter effluent turbidities of <0.3 NTU were targeted because the
Interim Enhanced Surface Water Treatment Rule (IESWTR) requires filter effluent turbidities
below 0.3 NTU in greater than 95% of measured samples. The first onset-of-breakthrough
experiment at Ottawa was conducted on January 21, 1999 when filter effluent turbidities were
increasing but still below 0.1 NTU. (In fact, this experiment was intended to be a stable
operation experiment and only captured onset-of-breakthrough fortuitously). As during the end-
of-run and breakthrough experiments, jar-coagulated C. parvum and B. subtilis were seeded into
the filter for one hour and samples were collected at 15, 30, 45, and 55 minutes after the start of
seeding.

Because of the striking results obtained in the January 21, 1999 onset-of-breakthrough
experiment, two additional such experiments were performed at Ottawa on December 20 and 22,
1999 when filter effluent turbidities were between 0.2 and 0.3 NTU. Jar-coagulated B. subtilis
spores were seeded for one hour and then C. parvum oocysts were seeded for one hour as
breakthrough commenced. Samples were only collected during the hour of C. parvum seeding at
15, 30, 45, and 55 minutes after the start of oocyst seeding. The staggered seeding was planned
to help elucidate mechanistic behavior during early breakthrough (detachment versus non-
attachment) by accumulating B. subtilis spores on the filter, just before the filter effluent quality
started substantially deteriorating. If spores were subsequently detected in the filter effluent
when influent concentrations were low (ideally near 0 CFU/L), they would be indicative of
detachment. The C. parvum sampling during seeding was intended to detect non-attachment. It
should be noted, however, that these experiments were not designed to conclusively determine
whether non-capture or release of previously deposited oocysts was the dominant mechanism of
oocyst passage into filter effluents during breakthrough. Rather, the experiments were designed
to determine the effects of filter operation during early breakthrough on pathogen passage into
filter effluents while providing some insight into mechanistic behavior.

Results

Table 4.5 summarizes the results obtained during the breakthrough, onset-of-
breakthrough and end-of-run experiments. Detailed results are included in Appendix B. Filter
effluent turbidity and particle concentrations during the end-of-run experiments at MWD were
similar to those obtained during the stable filter experiments (Table B.9). Furthermore, the filter
effluent turbidities and particles remained constant throughout the end-of-run seeding period.
Similar results were found for filter effluent concentrations of C. parvum, B. subtilis, E. coli,
MS-2, and G. lamblia. The December 1 and 8, 1999 E. coli experiments were considered suspect

89
Table 4.5
Filter performance (removal of turbidity, particles, and microorganisms) during Task 2 breakthrough, onset-of-breakthrough, and end-
of-run experiments

Date Experiment Research Log Removal (Mean ± 1 Standard Deviation) Filter Effluent Value
Platform (Mean ± 1 Standard Deviation)
C. parvum B. subtilis E. coli MS-2 G. lamblia Particles Particles Turbidity
(#/mL) (NTU)
11/17/98 End-of-run MWD 2.9 ± 0.28 3.0 ± 0.18 1.0 ± 0.32 2.1 ± 0.13 4.8 ± 0.12 2.7 ± 0.04 7.9 ± 0.1 0.05 ± 0.00
12/1/98 MWD 2.3 ± 0.17 2.7 ± 0.04 2.0 ±0.10 - 2.2 ± 0.04 18 ± 1.5 0.05 ± 0.00
12/8/98 MWD 2.1 ± 0.16 2.5 ± 0.41 1.8 ± 0.22 - 2.0 ± 0.05 28 ± 2.6 0.05 ± 0.00
VO
O MWD Average 2.4 ± 0.43 2.7 ± 0.30 2.0 ± 0.19 - 2.3 ± 0.28 75 ± 8.9 0.05 ± 0.00

1/21/99 Onset-of- Ottawa 3.0 ± 0.52 2.2 ± 0.21 3.3 ± 0.18 1.8 ± 0.7 0.06 ± 0.01
12/20/99 Breakthrough Ottawa 1.5 ± 0.20 0.5 ± 0.35 2.5 ± 0.46 32 ± 36 0.28 ± 0.03
12/22/99 Ottawa 1.4 ± 0.05 0.9 ± 0.08 - - - 0.29 ± 0.02
Ottawa Average 2.0 ± 0.83 1.2 ± 0.7P 2.9 ± 0.56 7(5.7 ± 28.2 0.21 ± 0.77

1 1/25/98 Breakthrough Ottawa 1.8 ± 0.03 1.2 ± 0.16 1.5 ± 0.14 249 ± 74 0.94 ± 0.09
12/9/98 Ottawa 1.7 ± 0.06 1.0 ± 0.09 1.7 ± 0.05 140 ± 16 0.50 ± 0.04
1/13/99 Ottawa 1.5 ± 0.68 1.3 ± 0.18 3.0 ± 0.05 3.7 ± 0.4 0.61 ± 0.15
Ottawa Average 1.6 ± 0.38 1.2 ± 0.18 2.1 ± 0.71 737 ± 112 0.68 ± 0.27
because elevated E. coli levels from an unknown source were observed in the filter effluent
negative control samples. The filter effluent water typically had a free chlorine residual of
approximately 0.6 mg/L, except during the seeding period. Both the backwash water and the
filter influent water were analyzed for E. coli, but very few or no colonies were present. The
filter effluent was tested for E coli on a non-seeding day, and no colonies were present. Another
filter that receives the same water at the same conditions was also tested and no colonies were
found. The elevated levels of E. coli in the control samples might have resulted from sample or
laboratory contamination, but no source could be confirmed. As a result, logic removals of E.
coli were not calculated for these experiments. With the exception of the suspect E. coli data, the
filter effluent microorganism concentration data during the end-of-run experiments at MWD
suggested that this period of operation was a somewhat stable operating period at the end of the
filter cycle.

In contrast to the end-of-run data, the filter performance data indicated that the onset-of-
breakthrough that occurred in Ottawa was a very dynamic period during which filter effluent
turbidity and particle counts rapidly changed; Figure 4.17 presents these data from the January
21, 1999 experiment. The filter effluent turbidity and particle counts ranged from 0.04 to 0.07
NTU and 0.3 to 4.3 particles/mL respectively during the seeding and sampling period. During
the December 20, 1999 experiment, the filter effluent turbidity and particle counts ranged from
0.25 to 0.32 NTU and 7.5 to 116 particles/mL respectively during sampling. On December 22,
1999 particle counts were not available, but filter effluent turbidity ranged from 0.25 to 0.35
NTU during sampling. The detailed performance data are available in Appendix B (Table B.3).

Increases in filter effluent turbidity and particle counts were accompanied by substantial
increases in filter effluent C. parvum concentrations (relative to those obtained during stable
filter operation) at Ottawa. As is evident from Figure 4.18, these increases occurred even when
filter effluent particle counts were relatively low (< 20 particles/mL). Figure 4.18 also indicates
that although an increase in filter effluent B. subtilis concentrations was observed during the
onset-of-breakthrough experiments, this increase was not as severe as the increase in filter
effluent oocyst concentrations. The December 20 and 22, 1999 data were generally consistent
with the January 21, 1999 data and indicated even greater passage of oocysts through the filter

91
20
-*- Particles 2:2pm Seeding Period
-«- Turbidity
o Sampling Times

CM
Al

0.0
10:00 AM 11:00 AM 12:OOPM 1:00 PM
Time

Figure 4.17 Turbidity and particle response of filter during onset-of-breakthrough


experiment on January 21,1999 at Ottawa pilot plant

20 2500
-»- Particles > 2pm Seeding Period
• C. parvum (oocysts/L)
A B. subtilis (CFU/L)

15 -

10-
I
<N
Al
8

10:00 AM 11:00 AM 12:OOPM 1:00 PM


Time

Figure 4.18 Particle and microorganism response of filter during onset-of-breakthrough


experiment on January 21,1999 at Ottawa pilot plant

92
during early breakthrough conditions. (As noted earlier the turbidity during the December 1999
onset-of-breakthrough experiments was measurable higher than in the January 1999 experiment.)
Detailed microorganism concentration and removal data are provided in Appendix B (Table
B.4).

The experiments designated as "breakthrough" were conducted at filter effluent


turbidities above 0.2 NTU when both filter effluent turbidity and particle counts continued to
rapidly change at Ottawa; Figure 4.19 presents these data from the December 9, 1998
^

experiment. The filter effluent turbidity and particle counts ranged from 0.42 to 0.62 NTU and
106 to 183 particles/mL respectively. During the November 25, 1998 experiment, the filter
effluent turbidity and particle counts ranged from 0.72 to 0.98 NTU and 13.1 to 324 particles/mL
respectively. On January 13, 1999 particle counts ranged from 2.4 to 5.8 particles/mL during
breakthrough; filter effluent turbidity ranged from 0.38 to 0.82 NTU. Although the particle
counts were very low during the January 13, 1999 experiment, subsequent examination of the
particle counter revealed that it was operating properly. Similarly, the turbidity meters were also
evaluated and found to be in proper working order. Although the particle data were suspiciously
low during the January 13, 1999 experiment, the data were not disregarded because no
equipment errors could be found. The detailed performance data are available in Appendix B
(Table B.3).

The elevated filter effluent turbidity and particle counts during the breakthrough
experiments at Ottawa were accompanied by high filter effluent C. parvum concentrations
relative to those obtained during the stable operation, end-of-run, and onset-of-breakthrough
experiments. The >104 oocysts/L filter effluent C. parvum concentrations observed during the
December 9, 1998 breakthrough experiment (Figure 4.20) were also consistently observed
during the November 25, 1998 and January 13, 1999 experiments. Figure 4.20 also indicates that
generally similar filter effluent B. subtilis concentrations were observed during the breakthrough
experiments. It should be noted that the filter effluent microorganism concentrations in Figure
4.20 are presented on a logio scale. Detailed microorganism concentration and removal data are
provided in Appendix B (Table B.4).

93
•»- Particles £ 2pm
•-Turbidity
o Sampling Times

0.0
7:30:00 PM 9:30:00 PM 11:30:OOPM 1:30:00 AM 3:30:00 AM 5:30:00 AM 7:30:00 AM
Time

Figure 4.19 Turbidity and particle response of filter during breakthrough experiment on
December 9,1998 at Ottawa pilot plant

250 1.E+05
•*- Particles £ 2pm
• C. parvum (oocyst/L)
A B. subtilis (CFU/L)

1.E+00
7:30 PM 9:30 PM 11:30PM 1:30 AM 3:30 AM 5:30 AM 7:30 AM
Time

Figure 4.20 Particle and microorganism response of filter during breakthrough experiment on
December 9,1998 at Ottawa pilot plant

94
Discussion

The end-of-run experiments at MWD yielded filter effluent turbidities and particle counts
that were similar to those observed during stable operation and microorganism concentrations
that were generally consistent with stable operation. In contrast, the onset-of-breakthrough
experiments at Ottawa demonstrated a relatively modest degradation of the traditional
performance parameters that was accompanied by tremendous increases in filter effluent C.
parvum and B. subtilis concentrations. These data suggested that small increases in particle
counts during early breakthrough operating conditions could signal substantially increased non-
capture of oocysts.

By seeding B. subtilis first and then C. parvum during the actual sampling, the December
20 and 22, 1999 onset-of-breakthrough experiments at Ottawa were designed to investigate
whether the passage of oocysts through the filter during early breakthrough conditions was
largely a function of non-attachment. The high effluent oocyst concentrations observed would
suggest this. Although the high concentration of spores in the filter effluent would suggest some
detachment, this interpretation was unclear because during sampling more spores were present in
the filter influent than in the effluent. To accurately determine the dominant mechanism of
microorganism passage through filters during breakthrough, the experiments would have to be
repeated in a slightly different manner.

The box and whisker plots and statistical comparisons (Figure 4.21, Figure 4.22 and
Figure 4.23) were based on the entire run and include the entire range of removals observed
during the defined sampling periods. As discussed above, the elevated filter effluent C. parvum
and B. subtilis concentrations during end-of-run, onset-of-breakthrough, and breakthrough
operation were loosely correlated with increasing filter effluent particle counts and turbidity. The
Ottawa data particularly emphasized that no clear relationships between particle concentrations,
turbidity, and microorganism concentrations were observed during onset-of-breakthrough or the
breakthrough experiments.

Table 4.5 has summarized the microorganism, turbidity, and particle removals at Ottawa
and MWD during the end-of-run, onset-of-breakthrough, and breakthrough experiments. At
MWD, C. parvum and B. subtilis removals during end-of-run operation were 2.4 ± 0.4 logic and
95
Metropolitan Ottawa
7 •

n = 32
6-

5-
n = 35 diH
Yes
1 4- Yes n=12
o n = 12
1
§ Yes
I 3- n=12
S. 2 •
(J
<==j
1 -

0- ————
Optimal End of Run Optimal Onset of Breakthrough
Breakthrough

Source: Reprinted from Proceedings ofAWWA 2000 Water Quality Technology Conference, by permission.
Copyright ©2000, American Water Works Association

Figure 4.21 Effect of breakthrough, onset-of-breakthrough, and end-of-run conditions on


removal of C. parvum by filters

Metropolitan Ottawa

6
n = 20
5-

I 4' n = 44 Yes
9
o n = 12
1 Yes
1—
—Ti—
—1 n=12 :
•ft

%1 1 ics
"9 1
2" •
n 12
09

1 - c=^=3 :

0 •

Optimal End of Run Optimal Onset of Breakthrough


Breakthrough

Figure 4.22 Effect of breakthrough, onset-of-breakthrough; and end-of-run conditions on


removal of B. subtilis by filters

96
Metropolitan Ottawa

6- •6

n = 28
5 • •5

Yes;
4- 11 - „ - 0 •4
44 Yes;
No; n=12 '•
n-12
3• •3

2- \ •2

1 • • I

0•
Optimal End of Run Optimal Onset of Breakthrough
Breakthrough

Figure 4.23 Effect of breakthrough, onset-of-breakthrough, and end-of-run conditions on


removal of particles by filters (based on raw water and filter effluent particle counts)

2.7 ± 0.3 logio respectively (mean ± standard deviation). At Ottawa, the onset-of-breakthrough
C. parvum and B. subtilis removals were 2.0 ± 0.8 and 1.2 ± 0.8 logio respectively; they were 1.6
± 0.4 and 1.2 ± 0.2 logio respectively during breakthrough.

Although the average filter effluent microorganism concentrations were not significantly
elevated (5% significance level) at MWD during end-of-run operation relative to stable
operation, the box and whisker plots summarizing the C. parvum and B. subtilis removal data
indicated significantly different (at the 5% level) median removals of both of these
microorganisms during the end-of-run when compared to stable operation at MWD (Figure 4.21
and Figure 4.22). Median C. parvum removals were lower during end-of-run conditions than
during stable operation, whereas B. subtilis removals were higher. Though significant differences
in particle removal were not observed (Figure 4.23), these differences in microorganism
removals were statistically significant (5% significance level). This result might have been
partially associated with the somewhat variable filter influent microorganism concentrations at
MWD during the end-of-run experiments, however, the filter influent concentrations were not

97
statistically different from those achieved during the stable operation experiments (5%
significance level).

Figure 4.21 and Figure 4.22 also indicate that the median C. parvum and B. subtilis
removals at Ottawa during the onset-of-breakthrough and breakthrough experiments were lower
and statistically different (5% significance level) than those observed during stable operation. As
discussed previously, both C parvum and B. subtilis removals were severely compromised
during breakthrough, and even onset-of-breakthrough conditions, when filter effluent particle
counts and turbidity were still relatively low. These results were consistent with the statistically
significant differences (5% significance level) observed in particle removals during onset-of-
breakthrough and breakthrough filtration relative to stable operation (Figure 4.23)

Oocyst removal during end-of-run operation at MWD was approximately 0.6 logio lower
than during stable operation. Compared to stable operation, the onset-of-breakthrough and
breakthrough experiments at Ottawa resulted in approximately 3.5 logic and 4 logio lower
removals of C. parvum, respectively. The Ottawa findings were in general agreement with the
findings of Logsdon et al. (1981) that demonstrated that turbidity breakthrough at the end of a
filter cycle could be accompanied by considerable passage of Giardia cysts. The Ottawa onset-
of-breakthrough results were very different from those obtained during the Patania et al. (1995)
study of Giardia and C. parvum passage through filters during breakthrough when effluent
turbidities increased, from 0.1 NTU to 0.2 NTU or higher. Those researchers found that while
Giardia removal was approximately 0.5 logio lower during breakthrough, no difference between
C. parvum removals during stable operation and breakthrough was observed. It is possible that
other factors such as chemical pretreatment may impact the degree of pathogen passage that
occurs during early breakthrough filtration.

HYDRAULIC STEP

Experimental Design

It is known that rapid filtration rate changes can significantly degrade the quality of filter
effluents by creating a change in shear forces that dislodges attached particles. Experiments

98
evaluating the effect of hydraulic changes were performed at both the Ottawa and MWD pilot
plants. To assess the effect of these changes on the removal of C. parvum and B. subtilis (and
MS-2 at MWD), hydraulic steps were imposed during stable (optimized) operating conditions.
Each of the hydraulic step experiments consisted of a 25% increase in filtration rate over less
than one minute and was achieved by opening the filter effluent valves. The higher rate was
maintained throughout the remainder of the filter cycle. The experiments were designed to
represent a scenario that could occur in practice, such as when one filter is put out of service
resulting in increased flow rates through the other filters that remain in operation. The
experimental design tested the hypothesis that microorganisms accumulated within the filter
during stable operation may be released immediately following a sudden increase in hydraulic
loading.

During the hydraulic step experiments, the microorganisms were seeded in the filter
influent over an extended period of time (5-hours at Ottawa, 8-hours at MWD). The seeding
occurred during stable filtration conditions to allow the microorganisms to accumulate in the
filter (as had been demonstrated in the previously described stable operation experiments). The
hydraulic step was imposed immediately after the seeding period at a time when filter influent
microorganism concentrations were expected to be low (ideally near 0 microorganisms/L). Thus
microorganisms appearing in the effluent would be as a result of detachment rather than non-
capture. In this regard the seeding strategy was different than for any of the other conditions
tested.

At Ottawa, samples were collected prior to, during, and after the hydraulic step. The flow
increase occurred at 300 minutes into the filter cycle. Samples were collected at 280 and 295
minutes to ensure that the filter was removing microorganisms at levels comparable to those
achieved during the stable filter operation experiments. A sample was collected at 300 minutes to
collect what passed through the filter as the hydraulic step occurred. Samples were also collected
at 5 minute intervals after the hydraulic step was imposed (at 305, 310 and 315 minutes) to
account for any subsequent effects on water quality. Additional samples were collected during
the second and third experiments (at 320 and 360 minutes) to determine if microorganism
removals returned to baseline levels (i.e., those achieved during stable filter operation) after the
change in flow rate.

99
At MWD samples were collected prior to and after the hydraulic step. The hydraulic step
was initiated at 502 minutes; samples were collected at 485 and 500 minutes to determine
microorganism removals during stable operating conditions and 505 and 520 minutes to assess
the effects of the hydraulic step. .

Results

Table 4.6 summarizes the results of the hydraulic step experiments. Detailed data are
included in Appendix B.

The hydraulic step experiments were conducted in Ottawa on June 7, 15, and 22, 1999.
Even though the same protocol was followed during each experiment, the resulting impact on
water quality differed among the experiments. On June 7, both filter effluent particles and
turbidity increased considerably for a period of approximately 30 minutes following the flow
increase (Figure 4.24). The filter effluent turbidity during stable filter operation portion prior to
the hydraulic step was approximately 0.06 NTU and the filter effluent particle concentration (> 2
um) was approximately 0.5 particles/mL. As a result of the hydraulic step, the filter effluent
turbidity and particle concentration temporarily increased up to 0.37 NTU and 297 particles/mL
respectively. Furthermore, during the course of that filter cycle, neither the filter effluent
turbidity nor the particle concentration returned to the baseline levels that had been achieved
prior to the hydraulic step (Figure 4.24). Generally consistent with the stable operation
experiments, filter effluent concentrations of C. parvum and B. subtilis were relatively low
(respectively <10 oocysts/L and <100 CFU/L) during the stable operation portion of the filter
cycle prior to the hydraulic step. The increase in filter effluent turbidity and particle
concentration (Figure 4.24) was commensurate with an increase in filter effluent
microorganisms, with effluent concentrations of C. parvum reaching 4,412 oocysts/L and B.
subtilis reaching 2,000 CFU/L (Figure 4.25). Detailed turbidity, particle, and microorganism
concentration and removal data are available in Appendix B (Table B.3 and Table B.4).

The hydraulic step had a very different impact on filter effluent water quality on June 15,
1999. The filter effluent turbidity only slightly increased whereas the particle concentration
peaked considerably (Figure 4.26). The filter effluent turbidity during stable filter operation prior
to the hydraulic step was approximately 0.05 NTU and the filter effluent particle concentration
100
Table 4.6
Filter performance (removal of turbidity, particles, C. parvum and B. subtilis) during Task 2 hydraulic step experiments

Date Experiment Research Log Removal (Mean ± 1 Standard Deviation) Filter Effluent Value
Platform (Mean ± 1 Standard Deviation)
C. parvum B. subtilis E. coli MS-2 G. lamblia Particles Particles Turbidity
(#/mL) (NTU)
6/7/99 Hydraulic Step Ottawa 1.7 ± 2.52 1.4 ± 1.12 - - 2.5 ± 1.20 105 ± 121 0.18 ± 0.13
6/14/99 Ottawa 4.3 ± 0.66 2.6 ± 0.48 2.7 ± 1.01 71 ± 124 0.05 ± 0.01
6/22/99 Ottawa 3.1 ± 1.02 2.1 ± 0.80 - - - - -
4/6/99 MWD 1.3 ± 0.66 2.0 ± 0.14 1.5 ± 0.14 2.3 ± 0.11 17 ± 4.9 0.05 ± 0.00
4/13/99 MWD 1.9 ± 0.68 2.1 ± 0.66 1.4 ± 0.29 2.3 ± 0.10 18 ± 3.9 0.05 ± 0.00
4/20/99 MWD 2.2 ± 0.99 2.5 ± 0.37 1.1 ± 0.46 2.5 ± 0.17 11 ± 4.6 0.05 ± 0.00
Ottawa Average 3.1 ± 1.78 2.1 ± 0.93 2.6 ± 1.06 87 ± 119 0.11 ± 0.11
MWD Average 1.8 ± 0.82 2.2 ± 0.46 1.3 ± 0.35 2.3 ± 0.15 15 ± 5.2 0.05 ± 0.00
«- Particles > 2pm
o Sampling Times
•- Turbidity

-•0.3 S

3:00 PM 4:00 PM 5:00 PM 6:00 PM


Time

Figure 4.24 Turbidity and particle response of filter during hydraulic step experiment on
June 7,1999 at Ottawa pilot plant

450 10000
Seeding Period Particles > 2pm
400 C. parvum (oocysts/L)
B. subtilis (CFU/L)
350-
:-1000

300

<o
250 - •
:- 100
200-

150 -•

100- :- 10

50- Hydraulic Ste


0 oocysts/L

1
3:00 PM 4:00 PM 5:00 PM 6:00 PM
Time

Figure 4.25 Particle and microorganism response of filter during hydraulic step experiment
on June 7, 1999 at Ottawa pilot plant

102
450 0.5

» o » o o o Sampling Times
- Turbidity____
0.4
350- :

2 300 - :
- 0.3 3
a.

2 200
a> - 0.2
o
Hydraulic Step
| 150 +

100-•
- 0.1

50

0.0
2:00 PM 3:00 PM 4:00 PM
Time

Figure 4.26 Turbidity and particle response of filter during hydraulic step experiment on
June 15,1999 at Ottawa pilot plant

450 50
Seeding Period
•*- Particles 2. 2pm
400 • C. parvum (oocysts/L)
A B. subtilis (CFU/L)
40
350 -

. 300 - -

- 30
' 250

200 - :
20
Hydraulic Step
150 - :

100 -•
- 10
50 -•

2:00 PM 3:00 PM 4:00 PM


Time

Figure 4.27 Particle and microorganism response of filter during hydraulic step experiment
on June 15, 1999 at Ottawa pilot plant

103
(> 2 urn) was approximately 1.0 particles/mL. As a result of the hydraulic step the filter effluent
turbidity and particle concentration temporarily increased to 0.09 NTU and 381 particles/mL
respectively. Unlike the June 7 experiment, both the filter effluent turbidity and particle
concentration returned to the baseline levels that were being achieved prior to the hydraulic step
(Figure 4.26). Despite the considerable increase in filter effluent particles (over a period of
approximately 30 minutes), no appreciable changes in filter effluent C. parvum and B. subtilis
concentrations relative to the stable operation period of the filter cycle were observed (Figure
4.27). While relatively low concentrations of B. subtilis spores were found in the filter effluent
during the stable and hydraulic step portions of the experiment (<50 CFU/L), almost no C.
parvum oocysts were detected.

A third hydraulic step experiment was performed at Ottawa on June 22, 1999. Filter
effluent turbidity and particle data were not available during this experiment due to difficulties
with the data acquisition system. These problems made it impossible to exactly pinpoint when
the hydraulic step (in relation to the associated sampling) occurred. Filter effluent turbidity and
particle data collected prior to the experiment indicated that the filter was likely operating at less
than optimal conditions just prior to the hydraulic step; the effluent particle counts were slightly
above what was typically observed at the Ottawa pilot plant during stable filter operation. A
slight increase in filter effluent oocyst concentrations was observed as a result of the hydraulic
step, however, the filter effluent spore concentrations did not demonstrate this trend (Table B.4).
Although the filter effluent oocyst concentrations were slightly elevated as a result of the
hydraulic step, the increase was not as dramatic as that which occurred during the June 7
experiment (Table B.4).

Similar hydraulic step experiments were performed at MWD on April 6, 13, and 20,
1999. Unlike the Ottawa experiments, the impact of the hydraulic step on water quality was
fairly comparable in each of the experiments. During the April 6 experiment, for example, the
filter effluent turbidity remained essentially unchanged and well below 0.1 NTU (Figure 4.28)
while only a slight increase in filter effluent particle concentration (>2 um) was observed (Figure
4.29). In general, during the hydraulic step experiments at MWD, the filter effluent turbidity was
approximately 0.05 NTU while the particle concentration ranged from 8-24 particles/mL). The
filter effluent microorganism concentrations (C. parvum, B. subtilis, and MS-2) generally
104
l.U

\
Seeding duration o <o o
Sampling times
^^ 0.20
P 1.5
— Plant Influent £

£, — Filter Effluent j5
•3 •fi
0.15 f-
1
nintfluent
ITt FiEfflutenrt
b

p 0
ro
£

0.5
1
0.05

Flowrate
increase
0.0 • 0.00
18 19 20 21 22 23 24 25 26 27 28 29 30

Filter Run Time

Figure 4.28 Turbidity response of filter during hydraulic step experiment on April 6, 1999 at
MWD pilot plant

1000
Seeding duration
Samp ing times

—Plant Influent
100
—Filter Effluent

E
a.
Al
$
~3 10 •

Flowrate
increase
18 19 20 21 22 23 24 25 26 27 28 29 30

Filter Run Time

Figure 4.29 Particle response of filter during hydraulic step experiment on April 6, 1999 at
MWD pilot plant (based on raw water and filter effluent particle counts)

105
decreased after the completion of seeding, regardless of the implementation of the hydraulic step
(Table B.10 and Table B.ll). Coupled with "the particle and turbidity data (Table B.9), the
microorganism data suggest that the hydraulic step had little impact on filter effluent
microorganism concentrations at MWD.

The original experimental objective of having very low influent microorganism


concentrations at the point of and following the hydraulic step was not achieved. However, this
turned out to be not important because of the generally low filter effluent microorganism
concentrations seen as a result of the hydraulic steps.

Discussion

All of the hydraulic step experiments at MWD and at least one of these experiments at
Ottawa (June 15) failed to yield substantial increases in filter effluent microorganism
concentrations (C. parvum, B. subtilis, and MS-2). These data suggest that little detachment of
microorganisms occurred as a result of 25% increase in flow hydraulic steps. A considerable
increase in filter effluent C. parvum concentrations occurred during the hydraulic step
experiment at Ottawa on June 7 and a more modest increase was observed on June 22. The
release of oocysts during these hydraulic step experiments was likely due to at least a moderate
amount of detachment from the filters; this was particularly evident in the June 7 data where the
filter effluent concentrations of oocysts were higher than the influent concentrations.

The June 7 and 15 hydraulic step experiments at Ottawa underscore the finding that
particle counts were not directly indicative of microorganism passage through filters during
hydraulic step conditions. Essentially the same peak particle concentrations occurred during the
June 15 (Figure 4.26) and June 7 experiments (Figure 4.24). In contrast, the filter effluent
microorganism concentration was dramatically higher on June 7 (Figure 4.25) than on June 15
(Figure 4.27). Moreover, the filter effluent particle spike on June 15 was not accompanied by an
increase in effluent oocyst concentration, emphasizing that increases in performance measures
such as particle counts are not necessarily directly indicative of microorganism passage through
filters.

106
The box and whisker plots and statistical comparisons were based on the entire run and
included data from both before and after the hydraulic step. The microorganism removal data
should be considered with this point in mind. The difficulty in interpreting removal data for these
experiments is associated with the filter influent microorganism concentrations during these
experiments. As noted above, the filter influent microorganism concentrations were lower than
during the other experiments because the microorganisms were seeded over a longer period of
time (5 hours at Ottawa and 8 hours at MWD as opposed to 1 hour for most of the other
experiments at both research platforms). In addition, the filter influent microorganism
concentrations were actually decreasing at both Ottawa and MWD after the implementation of
the hydraulic step because seeding had already ceased. Therefore, the logio removals calculated
for samples collected after the hydraulic step are affected by the fact that the filter influent oocyst
concentration was decreasing over time.

The hydraulic step had varied effects on water quality at Ottawa and relatively consistent
effects on water quality at MWD. At Ottawa, C. parvum and B. subtilis removals were 3.1 ± 1.8
logio and 2.1 ± 0.9 logio (mean ± standard deviation) respectively and at MWD they were 1.8 ±
0.8 and 2.2 ± 0.5 logio respectively. Microorganism removals during the hydraulic step
experiments at Ottawa and MWD are summarized in Table 4.6.

The box and whisker plots summarizing C. parvum removals at MWD and Ottawa during
stable filter operation (optimized operation) and the hydraulic step experiments indicate that
oocyst removals were generally higher at Ottawa during both operating periods (Figure 4.30).
These data also indicate that oocyst removals were significantly lower (5% significance level)
during the hydraulic step experiments at each location than during the experiments conducted at
optimized operating conditions (Figure 4.30). As discussed above, these differences in oocyst
removals were at least in part affected by the lower filter influent oocyst concentrations during
the hydraulic step experiments and the inclusion of data from both before (essentially stable
operating conditions) and after the filtration rate increase.

Trends in B. subtilis removal (Figure 4.31) by the filters were generally comparable to
those observed for C. parvum (Figure 4.30). At Ottawa, B. subtilis removals during optimal
operation were higher and significantly different (5% significance level) from those achieved

107
Metropolitan Ottawa'
/ •
•' -
n = 32
6- •6
Y«:s; • ;
n = 35 | n = 21
5' •5

4- Ye s;
n = 12 • •

" •3

•2

1 • •• 1

n•
Optimal Hydraulic Optimal Hydraulic
Step Step

Figure 4.30 Effect of hydraulic step on removal of C. parvum by filters (please see
accompanying text regarding interpretation)

Metropolitan Ottawa

6- -6
n = 20
5- •5

4- n == 44 •4
Yes;
No; n = 21
n=12
3- -3

2- • •
•2

1 - • 1

0- •0
Optimal Hydraulic Optimal Hydraulic
Step Step

Figure 4.31 Effect of hydraulic step on removal of B. subtilis by filters (please see
accompanying text regarding interpretation)

108
Metropolitan Ottawa

6- •6

n = 28
5 • •5
Yes;
n=13
I 4'
0
do
o
•4
o
n = 44
§ No;
04
« 3- n iz •3

r-f-i
2 • •2

1 • • 1

0- ————
Optimal Hydraulic Optimal Hydraulic
Step Step

Figure 4.32 Effect of hydraulic step on removal of particles by filters (please see
accompanying text regarding interpretation; based on raw water and filter effluent particle
counts)

during the hydraulic step experiments. B. subtilis removals at MWD were slightly higher during
optimal operation than during the hydraulic step experiments, however, this difference was not
statistically significant (5% significance level). A similar result was found with the particle
removal data (Figure 4.32).

Filter effluent particle concentration, turbidity, oocyst concentration and spore


concentrations increased for a short period of time during two hydraulic step experiments at
Ottawa. Even though the increase in effluent microorganism concentrations was temporary, the
number of oocysts that passed through the filter was substantial (given that a high number of
oocysts were seeded into the filter during the experiment). This result was consistent with the
findings of Cleasby et al. (1963), Tuepker and Buescher (1968), and Fitzpatrick et al. (1999) who
showed that sudden, large flow rate changes cause deterioration of filtered water quality. These
findings were also consistent with those of Logsdon et al. (1981) who similarly demonstrated

109
that increases in Giardia passage through filters could be expected as a result of hydraulic
changes.

In contrast, the hydraulic step experiments at MWD and the June 15 experiment at
Ottawa demonstrated that a 25% increase in flow through the filters could leave filter effluent C.
parvum, B. subtilis, and MS-2 concentrations and turbidity essentially unaffected while either
very small (MWD) or moderate (Ottawa) spikes in filter effluent particle concentrations (> 2 jj.ni)
occurred. The difference between these results and those from the Ottawa experiments that
resulted in considerable passage of microorganisms suggests that the balance between
attachment and detachment forces may be variable and result in performance from no risk to a
high risk of microorganism release from a filter as a result of hydraulic changes. These results
further suggest that it may be desirable to attempt to identify the factors that affect these forces
so that the potentially severe effects of hydraulic changes (such as those observed during the
June 7 experiment at Ottawa) can be minimized.

110
CHAPTER 5: MITIGATING OPERATIONAL EFFECTS

While Task 2 defined the ability of a benchmark (i.e., representative) filtration process to
respond to operational challenges, Task 3 investigated ways to potentially mitigate filter
operation effects. Essentially, the focus of Task 3 was to increase the robustness of the particle
removal processes. For this task, both pilot-scale and demonstration-scale experiments were
completed at MWD's facilities.

EXPERIMENTAL PLAN

MWD Pilot Plant Operation

The pilot plant consisted of two, 6 gpm parallel trains - one with a pre-oxidant and one
without. Each train was operated with two parallel filters: dual- and tri-media. Table 5.1 presents
filter specifications and a simplified schematic of the pilot plant is shown in Figure 5.1. Chlorine,
ozone, and permanganate were evaluated as pre-oxidants in a blend of approximately 75 percent
Colorado River water and 25 percent State Project water. Blending of the two source waters was
performed to meet the total dissolved solids secondary maximum contaminant goal of 500 mg/L,
which is a treatment objective for MWD's full-scale plants. Table 5.2 shows summarized
chemical requirements and raw water properties at MWD's pilot plant during Task 3. On-line
instruments continuously monitored and recorded turbidity, particles, headless, and filter flow.

Description of Pilot Plant Experimental Matrix

Table 5.3 lists the experimental factors for pilot plant operation at MWD. All of the tests used
either 2.0 mg/L chlorine, 1.0 mg/L ozone, or 0.38 mg/L potassium permanganate (as
permanganate, MnO4~) as a pre-oxidant. A parallel train was always operated as a control without
any oxidant, but with the same coagulant dosages. Each oxidant condition was applied during a
two-week cycle consisting of four experiments. The four experiments included baseline (i.e.,
stable filter operation), simulated runoff event (i.e. turbidity/TOC spike), hydraulic step, and
suboptimal coagulation. Each cycle was performed with and without the application of 0.01
mg/L nonionic filter aid. Each test consisted of one 24 hour filter run.
Ill
Table 5.1
Summarized filter specifications at MWD pilot plant during Task 3

Parameter Media Dual-media Tri-media

Filtration rate (gpm/ft2) 4.0 4.0

Depth (in.) Anthracite 20 v 20


Silica sand 8 8
Ilmenite sand - 3

Effective size (mm) Anthracite 1.0-1.1 1.0-1.1


Silica sand 0.43-0.50 0.43-0.50
Ilmenite sand 0.20 - 0.25

Uniformity coefficient Anthracite <1.65 <1.65


Silica sand <1.65 <1.65
Ilmenite sand <1.80

Pre-oxidant: Pre-oxidant:
2.0 mg/L C12
OR
1.0 mg/L O3 NONE
OR
0.38 mg/L MnO4"

Filters Filter 4
Dual-media Dual-media
4.0 gpm/ft2 4.0 gpm/ft2

Figure 5.1 Schematic of MWD pilot plant used during Task 3 experiments

112
Table 5.2
Summarized chemical dosages and raw water quality at MWD pilot plant during Task 3

Parameter Value
Optimized coagulant dose Alum (mg/L) 5.0
Cationic polymer (mg/L) 1.5

Suboptimal coagulant dose Alum (mg/L) 1.8


Cationic polymer (mg/L) 0.53

Filter aid, when used Nonionic polymer (mg/L) 0.01

Pre-oxidant, when used Chlorine (mg/L) 2.0


Ozone (mg/L) 1.0
Permanganate (mg/L) 0.38

Coagulation/filtration pH 7.6 - 8.3

Raw water Alkalinity (mg/L as CaCO3) 105 -120


pH 7.8 - 8.3
Temperature (°C) 13-27
TOC (mg/L) 2.4 - 2.9
Turbidity (NTU) 0.75 - 3.4

113
Table 5.3
Experimental matrix for Task 3 pilot-plant experiments at MWD

Parameters Number Conditions


Operating challenges 4 Baseline, turbidity/TOC spike,
hydraulic step, suboptimal coagulation
Pre-oxidation 4 None*, chlorine, ozone, permanganate
Filter aid 2 On/off
Filter media design 2 Dual- and tri-media

The no-oxidant experiments were performed in parallel with the other experiments as a control

The baseline - or stable filter operation - experiment was considered the benchmark
experiment to determine optimal filter performance for a particular pre-oxidant. The presence of
filter aid was not considered part of the benchmark experimental conditions, but was added as a
potential mitigation strategy.

During the simulated runoff event experiment, kaolin (product 883-4; Aldrich;
Milwaukee, Wis.) and montmorillonite (product 152-2; Aldrich) clay suspensions (1:1 wt. ratio)
were fed at the pilot-plant influent prior to the addition of any oxidant or coagulant. The clays
were added to increase the influent turbidity by approximately 10 NTU. Humic acid (product
675-2; Aldrich) was fed at the same location at a dosage calculated to yield a TOC increase of
1.0 mg/L, or 30-40 percent. The turbidity suspension and the TOC solution were fed for the
entire duration of the filter runs. These conditions were chosen to simulate increased turbidity
and organic matter resulting from a watershed runoff event.

The filtration rates were increased by 25 percent over a one-minute period during the
hydraulic step experiments. The rate increases were imposed approximately 20 hours into the
filter runs. Hydraulic step experiments were chosen because relatively minor changes in the
filtration rate may cause a decline in overall particle and organism removal, especially
considering the potential release of previously attached particles and organisms.

114
Suboptimal coagulation was the last experiment conducted during each two-week cycle.
Suboptimal coagulation resulted from a reduction in coagulant dosage by 65 percent. For these
tests, alum was reduced from 5.0 to 1.8 mg/L and cationic polymer was reduced from 1.5 to
0.53 mg/L. As a result, filter effluent turbidities increased to 0.1-0.2 NTU, which was a
substantial increase but still resulted in turbidities which would meet the USEPA's requirements
in the Interim Enhanced Surface Water Treatment Rule. To ensure that the previous coagulation
conditions did not influence the suboptimal coagulation experiments, the coagulant doses were
changed approximately 24 hours before the beginning of the filter runs.

MWD Demonstration Plant Operation

Demonstration plant data - collected as part of a separate AWWA Research Foundation


project (Stanley, Coffey, and Rector, 1998) - were also incorporated into this project. For the
demonstration plant testing, combinations of six primary alum dosages (0, 1, 3, 5, 7, and
10 mg/L) and four cationic polymer dosages (0, 1, 2.5, and 5 mg/L) were applied to one dual-
and one tri-media filter at a filtration rate of 3.5 gpm/ft2. Limited tests also evaluated the use of
chlorine and ozone as pre-oxidants and also the use of higher coagulant dosages. Table 5.4 lists
the experimental matrix for MWD's demonstration plant operation.

MWD PILOT PLANT RESULTS

Turbidity/TOC seeding conditions

Although the supplemental turbidity feed (a mixture of kaolin and montmorillonite) was
consistently fed during each turbidity/TOC feed experiment, the humic acid seeding varied from
approximately 2.0 to 6.5 mg/L. The seeding goal was to increase the plant influent turbidity by
10 NTU and TOC by 1.0 mg/L. Humic acid was initially fed at 2.0 mg/L to increase the TOC. It
was initially estimated that approximately 50 percent by weight of the humic acid was carbon;
therefore, a 2.0 mg/L feed of humic acid should have resulted in a 1.0 mg/L TOC increase.
However, analytical results (using the persulfate-UV oxidation method for TOC) showed
minimal change in TOC concentration downstream of the feed location.

115
Table 5.4
Experimental matrix for Task 3 demonstration-scale experiments at MWD

Parameters Number Conditions


Alum dose (mg/L) 8 0,1,3,5,7, 10, 20*, 30*, 40*
Cationic polymer dose (mg/L) 4 0,1,2.5,5
Pre-oxidation 3 chlorine, ozonef, nonet
Filter media design 2 Dual- and tri-media

*Only limited tests were conducted at these dosages


fOnly limited tests were conducted without an oxidant or with ozone

Controlled, bench-top experiments were then conducted using organic free water and the
humic acid to determine the carbon content of the humic acid. These tests showed that the humic
acid contained 31 percent carbon (on a weight basis), which still could not explain the missing
TOC from the mass balance at the pilot-scale. Further dilution tests were performed using pilot
plant influent water. Humic acid was spiked for some samples and humic acid and 10NTU
turbidity was spiked for the others. It was thought that the increased turbidity may adsorb or bind
to the TOC and make it unavailable to the analytical method. Results showed that samples
without the turbidity spike had TOC values near the target. The samples that had the turbidity
increase had TOC values slightly less than the target concentration. The increased turbidity
appeared to "mask" some TOC, but not substantially.

In the tests reported here, the TOC increase to the source water is presented using
calculated values and could not be analytically verified by a mass balance. For these tests, it is
not known whether the turbidity or the TOC exerted more influence on the filter performance.

Effect of Pre-Oxidation

Figures 5.2 through 5.4 compare the three oxidation conditions under three treatments
(stable filter operation, turbidity/TOC spike, and subbptimal coagulation) for dual-media filters.
Figure 5.2 shows the results of median filter effluent turbidities for the dual-media filters.

116
Minimal differences in filter effluent turbidity were observed for the stable filter operation
conditions for any oxidant. However, when the turbidity/TOC suspensions were spiked into the
raw water (simulating runoff), the presence of an oxidant improved turbidity control. Note that
2.0 mg/L humic acid was added during the chlorine and ozone (without filter aid) experiments;
6.5 mg/L humic acid was added during the ozone with filter aid and permanganate experiments.
The turbidity improvement during runoff simulation ranged from 0.04 NTU for chlorine to
greater than 0.4 NTU for permanganate. Under suboptimal coagulation conditions, the oxidant
choice did not improve turbidity control; for chlorine and ozone, the presence of an oxidant even
slightly increased the median filter effluent turbidity.

Figure 5.3 shows the median filter effluent particles sized >2 |nm for the same tests as
above. Unlike the turbidity results in Figure 5.2, the pre-oxidized filter train always
outperformed the non-preoxidized filter train - often by more than one order of magnitude.
Chlorine exhibited the largest effect (with an average improvement of particle removal of 1.3
log), followed by ozone (0.6 log improvement) and permanganate (0.5 log). Note, however, that
the oxidant dosages were not necessarily optimized for particulate control.

1.00

a with preoxidation
•without preoxidation

0.00
Baseline Runoff Suboptimal Baseline Runoff Suboptimal Baseline Runoff Suboptimal

Chlorine Ozone Permanganate

Figure 5.2 Effect of preoxidation on median filter effluent turbidity at MWD's pilot plant
under optimized and suboptimal coagulation conditions for dual-media filters (runoff test with
permanganate used 6.5 mg/L humic acid)

117
10,000

1,000-
(I)
0)
o

Ewith preoxidation
•without preoxidation

Baseline Runoff Suboptimal Baseline Runoff Suboptimal Baseline Runoff Suboptimal

Chlorine Ozone Permanganate

Figure 5.3 Effect of preoxidation on median filter effluent particle counts from the dual-
media filters at MWD's pilot plant under optimized and suboptimal coagulation (runoff test with
permanganate test used 6.5 mg/L humic acid)

10,000 T

Ewith preoxidation
•without preoxidation

Baseline Runoff Suboptimal Baseline Runoff Suboptimal Baseline Runoff Suboptimal

Chlorine Ozone Permanganate

Figure 5.4 Effect of preoxidation on median filter effluent particle counts from the tri-media
filters at MWD's pilot plant under optimized and suboptimal coagulation (permanganate
turbidity/TOC test used 6.5 mg/L humic acid)

118
The median filter effluent turbidities for the tri-media filters exhibited similar trends as
the dual-media filters, with the tri-media filters typically producing lower turbidities (data not
shown). The median filter effluent particles (>2 j^m) for the tri-media filters are shown in Figure
5.4. Similar trends were observed to that of the dual-media filters, but the tri-media filters
produced a larger difference in particle removal between the preoxidized and non-preoxidized
filters.

Effect of Filter Aid

Figures 5.5 and 5.6 show the effect of filter aid on dual-media filters in the pilot plant.
Median turbidity and particle values were compared for baseline runoff and suboptimal
coagulation experiments for filters with and without pre-oxidation. The use of filter aid did not
consistently effect either filter effluent turbidity or particles. Thus, it appears that the use of filter
aid was not an effective control strategy to mitigate against suboptimal coagulation or runoff
conditions for this water.

Effect of Filter Media Design

Median filter effluent turbidities for baseline runoff and suboptimal coagulation
experiments comparing dual- and tri-media filters are presented in Figure 5.7. Few substantial
differences were observed between dual- and tri-media filters, as measured by median turbidity.
Using chlorine preoxidation, the tri-media filter produced lower turbidities than the dual-media
filter for all three experiments. Without preoxidation, the dual-media filter performed better
during the baseline and turbidity/TOC spike experiments, but performed worse during the
suboptimal coagulation experiment.

Median filter effluent particles (>2 jam) for baseline runoff and suboptimal coagulation
experiments comparing dual- and tri-media filters are presented in Figure 5.8. For the filter
receiving chlorine-preoxidized water, the tri-media filter consistently performed better than the
dual-media filter. Without preoxidation, the dual- and tri-media filters performed almost
identically.

119
0.25

^without filter aid


•with filter aid

0.00
Baseline Turbidity Spike Suboptimal Baseline Turbidity Spike Suboptimal
Chlorine preoxidation Without preoxidation

Figure 5.5 Effect of filter aid on median filter effluent turbidity at MWD's pilot plant under
optimized and suboptimal coagulation conditions

1,0007

^without filter aid


•with filter aid

Baseline Turbidity Spike Suboptimal Baseline Turbidity Spike Suboptimal


Chlorine preoxidation Without preoxidation

Figure 5.6 Effect of filter aid on median filter effluent particles at MWD's pilot plant under
optimized and suboptimal coagulation conditions

120
0.25

^dual-media
•tri-media

0.00
Baseline Turbidity Spike Suboptimal Baseline Turbidity Spike Suboptimal
Chlorine preoxidation Without preoxidation

Figure 5.7 Effect of media configuration on median filter effluent turbidity at MWD's pilot
plant under optimized and suboptimal coagulation conditions

1,OOOT

I
S.
n dual-media
E •tri-media
UJ

u_
I

Baseline Turbidity Spike Suboptimal Baseline Turbidity Spike Suboptimal


Chlorine preoxidation Without preoxidation

Figure 5.8 Effect of media configuration on median filter effluent particles at MWD's pilot
plant under optimized and suboptimal coagulation conditions

121
Effect of Hydraulic Step

The filtration rate increase of 25 percent over a one-minute period did not substantially
deteriorate the filter performance, as measured by turbidity and particles. These data were
analyzed differently than for the other experiment types. For the hydraulic step experiments, the
turbidity and particle counts from one hour prior and one hour immediately following the one-
minute period were collected for each experimental condition. Then the median and 95th
percentiles from before and after the hydraulic increase were compared. There was no statistical
difference (at a confidence of 95 percent) for the median turbidity, 95th percentile turbidity, or
median particle counts resulting from the hydraulic step. The 95th percentile particle counts after
the hydraulic step, however, were greater than before the hydraulic step but the difference was
not substantial (less than 5 particles/mL). Thus, at the conditions tested, the hydraulic step
experiments did not deteriorate filtration performance.

MWD DEMONSTRATION PLANT RESULTS

The demonstration plant was also operated in a mode which allowed an assessment of the
sensitivity of full-scale treatment to coagulation, pre-oxidation, and media type. For each of the
demonstration-scale tests, the plant was operated at either 2.0 or 4.0 mgd, depending on whether
one or two filters were in service. At 2.0 mgd, the fiocculation time was 83 minutes, the
sedimentation time was 275 minutes, and the filtration rate was 3.5 gpm/ft2. At 4.0 mgd, the
flocculation and sedimentation times were halved, but the filtration rate remained constant
because a second filter was brought into service. Each test consisted of two, consecutive 24 hour
filter runs. Table 5.5 presents the raw water quality during the demonstration-scale testing
period.

122
Table 5.5
Summarized raw water quality for demonstration-scale experiments at MWD

Parameters Units Median,


range
Temperature °C 19,
11-25
Turbidity NTU 1.4,
0.7-3.0
Particle counts #/mL 3,000,
1,300-10,000
TOC mg/L 2.7,
2.6-3.3
UV254 cm'1 0.047,
0.038-0.075
Alkalinity mg/L as CaCOs 110,
99-120
pH 8.3,
8.2-8.4

Effect of Coagulation

Figure 5.9 shows the effect of the coagulation conditions on the dual-media filter effluent
turbidity. The presence of multiple coagulants versus single coagulants improved the ability of
the filter to produce low turbidity levels. For example, when the alum dose was 3 mg/L and no
polymer was added, the median filter effluent turbidity was 0.21 NTU. The addition of 1.0 mg/L
cationic polymer decreased the turbidity to 0.05 NTU. The use of alum as the sole coagulant
could produce filter effluent turbidities less than 0.05 NTU, although the alum dose required to
achieve that turbidity was 10 mg/L or above. In contrast, the use of cationic polymer as a single
coagulant could not produce consistently low turbidities.

123
0.7

Median
Turbidity 0:
(NTU)

0.2!

1 Cationic
2.5 Polymer
5 (mg/L)

Alum (mg/L) 10

Figure 5.9 Effect of coagulation conditions on median filter effluent turbidity for the dual-
media filter at MWD's demonstration plant (with chlorine as preoxidant)

10,001

1,001

Median
Particles 1
(#/mL)

1 Cationic
2.5 Polymer
5 (mg/L)

Alum (mg/L)

Figure 5.10 Effect of coagulation conditions on median filter effluent particles for the dual-
media filter at MWD's demonstration plant (with chlorine as preoxidant)

124
Similar performance can be seen from the median particle counts leaving the dual-media
filter (see Figure 5.10). The particle performance showed that—when alum was used alone—a
local minimum for particle densities were produced at 7 mg/L of alum (11 particles/mL). The
dual-media filter consistently produced higher median levels of particles (22-28 particles/mL)
when either more or less alum was added. When cationic polymer was added, the sensitivity of
particle reduction to alum dosage was reduced.

Higher dosages of alum than the 10 mg/L dosage used in most of these tests did not
improve particle removal at the demonstration plant. Figure 5.11 shows the median filter effluent
particle concentration from the dual-media filter after it was dosed with 10, 20, 30, and 40 mg/L
alum. In fact, more particles passed through the filter at 30 and 40 mg/L than when a lower
dosage was used.

Effect of Pre-Oxidation

The effect of pre-oxidation was also evaluated at the demonstration plant. In these tests,
the cationic polymer dosage was fixed at 1.0 mg/L, and the alum dosages selected were 1, 5, and
10 mg/L. Two, 24-hour filter runs were performed with chlorine, ozone, or no oxidant and the
on-line turbidity and particle count data were compared. Figure 5.12 shows the median filter
effluent particle performance from the dual-media filter. Similar to the pilot-scale testing, these
tests showed that the presence of an oxidant significantly improves particle removal. The particle
removal differences between oxidants were relatively minor. Note also that the effluent particle
concentrations were higher during the pre-oxidation tests than during the tests described
immediately above. One likely cause of this difference was that a higher blend of State Project
water was used (due to blend requirements at the facility where the demonstration plant is
located) and the TOC was elevated compared to the prior tests (TOC of 3.1-3.3 mg/L during
these tests). The additional TOC may have exerted an additional coagulant demand.

Effect of Media Configuration

Tri-media filters did not consistently improve the combined filter effluent turbidity or
particle counts. Figure 5.13 depicts the improvement in the 90th percentile turbidity which occurs
using tri-media filters. Each of these values was obtained by subtracting the 90th percentile

125
1,000

10 20 30
Alum dose (mg/L)

Figure 5.11 Effect of higher coagulant dosages on median filter effluent particles for the dual-
media filter at MWD's demonstration plant (cationic polymer dosage =1.0 mg/L)

i.ooor

100;
Median
Particles
(#/mL)

Alum Dose (mg/L)

Figure 5.12 Effect of oxidant on median filter effluent particles for the dual-media filter at
MWD's demonstration plant (cationic polymer dosage =1.0 mg/L)

126
Improvement
in 90th
Percentile 0.10;
Turbidity
(NTU)

1 Cationic
2.5 Polymer
5 (mg/L)

Alum (mg/L)

•\th
Figure 5.13 Effect of filter media configuration on the 90 percentile effluent turbidity at
MWD's demonstration plant. A positive value shows that tri-media improved performance. A
negative value (shown with black on top of the bars) demonstrated that tri-media did not improve
performance.

turbidity of the dual-media filter from the 90th percentile turbidity of the tri-media filter. Bars
with a black top surface indicate negative values - in these cases, the tri-media filter produced
higher effluent turbidity than the dual-media filter. The tri-media filter provided treatment
improvements typically at low alum dosages and higher cationic polymer dosages.

SUMMARY

Pilot-Plant Results

A number of available treatment options exist to potentially improve the benchmark


treatment system chosen for this work. Figure 5.14 shows the net effects of potential treatment
changes when applied to stable filter operation conditions. Based on the median filter effluent
turbidity (data not shown), changing or eliminating the oxidant, adding filter aid, or using tri-
media filters all changed the filter effluent turbidity by less than 0.02 NTU (as a median value).

127
Eliminate CI2 Use03 Use MnO4' Add filter aid Use tri-media

Figure 5.14 Effect of treatment options on median filter effluent particles for baseline (stable
filter operation) experiments at the MWD pilot plant

Particles were more sensitive to treatment alternatives and showed a significant deterioration
when chlorine was removed from the treatment scheme. Alternative oxidants, filter aid, or tri-
media all exhibited a minor influence on the median effluent particles.

A similar effect was observed when treatment changes were applied to the suboptimal
coagulation tests (see Figure 5.15). Removing the oxidant impaired treatment in all cases, but
changing the oxidant (to ozone or permanganate) or adding filter aid or tri-media filtration
produced a smaller effect.

128
Eliminate CI2 UseO3 Use MnO4" Add filter aid Use tri-media

Figure 5.15 Effect of treatment options on median filter effluent particles for suboptimal
coagulation experiments at the MWD pilot plant

Based on these results, the following conclusions are offered:

• Coagulation chemistry remained the most important barrier to particle (and potential
pathogen) passage.

• Pre-oxidation improved particle control in virtually all cases. Pre-oxidation cannot,


however, compensate for poor coagulation.

• Filter aid and tri-media did not provide significant treatment improvements under the
conditions tested.

Note that these results are specific to the source water tested. The sensitivity of
coagulation, preoxidation or other design conditions may depend on influent water quality.

Demonstration-scale results

The demonstration-scale results confirmed the pilot results above, but also added insight
into the sensitivity of particle removal to the coagulation conditions. In the pilot-scale tests, the

129
alum and the cationic polymer were adjusted equally on a percentage basis. At the
demonstration-scale, the alum and the polymer dosages were independently varied. When alum
was used as a single coagulant, particle control was more sensitive to the dosage than when alum
was used in conjunction with the polymer. These data indicate that multiple coagulants may
improve treatment robustness.

130
CHAPTER 6: INCREASING OUR UNDERSTANDING (BENCH-SCALE
INVESTIGATIONS)

This chapter presents the results of bench-scale investigations conducted at the Johns
Hopkins University. In terms of project organization, these investigations were part of Task 3,
which examined ways of mitigating operational effects on pathogen passage. The experiments at
Johns Hopkins used laboratory scale filters to increase our understanding of the effects of the
following factors: coagulant dosage, filtration rate, filtration mode (direct versus conventional)
and pH. Modeling was also conducted.

DIRECT FILTRATION STUDIES

In this section the results obtained using the laboratory dual-media filter facility in a
direct filtration mode are presented. First, the effects of coagulant dosage on filter performance at
a conventional filtration rate (4.9 m/h [2 gpm/ft2]) are discussed. Alum was the sole coagulant
used. Alum dosages of 4.5, 6.8, 9.0, and 13.5 mg/L were investigated at pH 7. The rationale for
this range of doses is presented below. The performance of the dual-media filter at a high
filtration rate (14.7 m/h [6 gpm/ft2]) was also studied at alum dosages of 6.8, 9, 11.3, and 13.5
mg/L. Comparison is made with the results from the conventional filtration rate studies. The
effects of pH on the filter performance in direct filtration at the conventional rate were studied at
pH 6 and 8. The results are compared with the filter performance at pH 7. Model simulations of
clean filter bed performance were made using the same particle and media characteristics as
applied in the experimental studies and the results are analyzed in relation to the experimental
conditions. These results are described in Appendix D.

Effects of Alum Dosage

The alum dosages for the direct filtration experiments were determined based on jar test
results. A jar test was conducted at pH 7 using the model raw water with the same characteristics
as applied in the direct filtration experiments. Flocculation was extended to 48 hours in this jar
test in order to provide sufficient contact opportunities for the coagulation and flocculation of
turbidity and natural organic matter. The detailed procedure for this jar test is described in the
Materials and Methods section. As illustrated in Figure D. 1 and Figure D.2, the optimum alum
131
dosage for this jar test was 9.0 mg/L for both settled turbidity and settled UV254 absorbance. This
dosage was then used as an estimate for the optimum dosage in the direct filtration experiments.
Filter performance at suboptimum alum dosages was tested by decreasing the alum dosages by
25% and 50%. Overdosing of alum was tested by increasing the dosage by 50%. Four filtration
experiments were conducted at alum dosages of 4.5, 6.8, 9.0 and 13.5 mg/L. Each of these
experiments lasted for four hours. Additional experiments using alum dosages of 9.0 and 13.5
mg/L were conducted for runs of 24 hour duration to study the effects of coagulant dosage on
filter performance at breakthrough.

The development of headloss by the anthracite layer and by the entire filter bed at four
alum dosages is presented in Figure 6.1 and Figure 6.2. At alum dosages of 4.5 and 6.8 mg/L, no
headloss was developed during these 4 hour filter runs, indicating that particle and turbidity
removals were negligible at these dosages. At an alum dosage of 9.0 mg/L, the headloss after 22
hours of filtration was 54 cm across the anthracite layer and 99 cm across the entire dual-media
bed. When a dosage of 13.5 mg/L of alum was used, headloss development was somewhat
greater: after 21 hours, headloss was 57 cm across the anthracite layer and 120 cm across the
entire bed. At these two higher dosages, the rate of headloss development across the dual-media
filler increased at about 11 hours into the filter run. The rate of headloss development across the
anthracite layer, on the other hand, slightly decreased at this time. These headloss results indicate
that a modest change in coagulant dosage for this water, from 6.8 to 9 mg/L of alum, transformed
a filter operated in a direct filtration mode from one that was ineffective into a unit that provided
considerable removal of particulate matter. In addition, these headloss data demonstrate that the
bottom sand layer of the dual-media filter participated in the filtration process throughout the run
when sufficient coagulant was added to allow the filter to function.

The turbidities of the anthracite and dual-media effluents are presented as functions of
filtration time at the four alum dosages in Figure 6.3, Figure 6.4, Figure 6.5, and Figure 6.6.
When alum was added at concentrations of 4.5 and 6.8 mg/L (Figure 6.3 and Figure 6.4), the
turbidity at both locations rapidly increased to the raw water level in 20 minutes and did not
improve for the rest of the filter run. Removal was negligible, as also indicated by the headloss
data in Figure 6.1 and Figure 6.2. When alum was added at 9.0 mg/L (Figure 6.5), the turbidity
of the effluent from the anthracite layer was initially measured at 1.8 NTU; it then declined to

132
80

70
* 4.5 mg/L
60 D 6.8 mg/L
O 9.0 mg/L
v xxxx
XX * X 000°°°
50
X 13.5 mg/L „ O °°
c
40 x X *0X °° 0'

30
x x o
x o °
20

10

0
9 12 15 18 21 24
Time (hour)

Figure 6.1 Effect of alum dosage on headloss development by the anthracite layer during a
filter run (direct filtration mode, filtration rate = 4.9 m/h [2 gpm/ft2]).

140
» 4.5 mg/L
120 D 6.8 mg/L
O 9.0 mg/L
100
X 13.5 mg/L
•&
5 80

X 60

40
*cx g o o
20

0
9 12 15 18 21 24
Time (hour)

Figure 6.2 Effect of alum dosage on headless development by the entire dual-media bed
during a filter run (direct filtration mode, filtration rate = 4.9 m/h [2 gpm/ft2]).

133
1U.U

9.0 .A 6J 60*6*6* 6 6 66 6 d 6 6 (
8.0 *G
*<§
7.0

o
.1? 5.0
1 4.0

3.0 • Anthracite
2.0 O Filtrate

1.0 O
00 < tp .......
0.0 0.5 1.0 1.5 2.0 2.5 3.0 3.5 4.0
Time (hour)

Figure 6.3 Turbidity of the anthracite and dual-media effluents during a filter run at an alum
dosage of 4.5 mg/L (direct filtration mode, filtration rate = 4.9 m/h [2 gpm/ft2]).

r
9.0 666666 oooooo o o o <
8.0 . -
7.0 -8
S-, o
1
6.0 t>
t1 5.0 •
4.0 - O -

3.0
• Anthracite
2.0 O Filtrate
1.0 - 0
00 < BO ,,,,,,,
0.0 0.5 1.0 1.5 2.0 2.5 3.0 3.5 4.0

Time (hour)

Figure 6.4 Turbidity of the anthracite and dual-media effluents during a filter run at an alum
dosage of 6.8 mg/L (direct filtration mode, filtration rate = 4.9 m/h [2 gpm/ft2]).

134
7.0 r

6.0 * Anthracite
O Filtrate »*

5.0

4-°

•5
£ 3.0

n f> oonnnnnn
9 12 15 18 21 24
Time (hour)

Figure 6.5 Turbidity of the anthracite and dual-media effluents during a filter run at an alum
dosage of 9.0 mg/L (direct filtration mode, filtration rate = 4.9 m/h [2 gpm/ft2]).

8.0

7.0 * Anthracite
O Filtrate
6.0

5" 5.0

£ 4.0
3
£ 3.0

2.0

1.0

0.0 'OOP

9 12 15 18 21
Time (hour)

Figure 6.6 Turbidity of the anthracite and dual-media effluents during a filter run at an
alum dosage of 13.5 mg/L (direct filtration mode, filtration rate = 4.9 m/h [2 gpm/ft2]).

135
about 0.28 NTU over 90 minutes and stabilized at that value until it started to break through at
10 hours of filter run. The turbidity of the dual-media filtrate increased initially to 0.64 NTU; it
then decreased to slightly below 0.20 NTU after 20 minutes of filtration and stabilized there until
it reached the breakthrough at 22 hours of filter run. The effects of alum dosage on the ripening
of turbidity and particle removals by the anthracite and by the dual-media filter are listed and
summarized in Table 6.1. The effects of alum dosage on the breakthrough of turbidity and
particle removals are listed in Table 6.2. At an alum dosage of 13.5 mg/L (Figure 6.6), the
turbidity of the anthracite effluent rose initially to about 1.7 NTU, declined rapidly to about 0.30
NTU after 30 minutes, and then started to break through after 6 hours of the filter run. The
turbidity of the filter effluent increased over the first few minutes of filtration to a value of 1.4
NTU, ripened in 20 minutes to a level less than 0.10 NTU, and remained at that level until the
breakthrough occurred at about 19 hours into the filter run. These results indicate extensive
removal of turbidity by the dual-media bed at alum dosages of 9.0 and 13.5 mg/L, consistent
with the headless profiles in Figure 6.1 and Figure 6.2. They confirm significant removal by the
lower sand layer throughout the run and extensive removal by the lower layer during the initial
or ripening stages of the run. The results are also in agreement with the predictions from the
model simulation studies described in Appendix D (see Figure C.I through Figure C.3). An alum
dosage of 9.0 mg/L, indicated by 48-hour jar tests, provided effective filtration. A higher dosage
(13.5 mg/L alum) improves effluent turbidity, lowering it below 0.10 NTU, and increases
headless. Lower alum dosages (4.5 and 6.8 mg/L of alum) were completely ineffective in this
system.

Table 6.1
Effects of alum dosage on the ripening of filter performance

Alum dosage* Anthracite layer Dual-media


(mg/L)
Turbidity Particle Turbidity Particle
concentration concentration
9.0 90min Not measured 20min 15 min
13.5 30min Not measured 20min lOmin
Ripening did not occur at alum dosages of 4.5 and 6.8 mg/L.

136
Table 6.2
Effects of alum dosage on the breakthrough of filter performance

Alum dosage Anthracite layer Dual-media


(mg/L)
Turbidity Particle Turbidity Particle
concentration concentration
4.5 Immediate Immediate Immediate Immediate
6.8 Immediate Immediate Immediate Immediate
9.0 10 hours Not measured 22 hours 20 hours
13.5 6 hours Not measured 19 hours 17 hours

Data for the turbidities and particle concentrations in the effluent from the dual-media
bed as functions of filtration time at the four alum dosages are presented in Figures 6.7 to 6.10.
With an alum dosage of 4.5 mg/L (Figure 6.7), particle concentrations and turbidities in the bed
effluent reached raw water levels rapidly and simultaneously, and then remained at those levels
for the rest of the 4-hour run. When an alum dosage of 6.8 mg/L was used (Figure 6.8), a small
removal of particles was observed while filtrate turbidity quickly reached and maintained the raw
water level. A small, trivial amount of ripening occurred with particle concentrations over the
remainder of the experiment. These results confirm again that these alum dosages are too low to
accomplish effective filtration. They also suggest that, for filters operated in a direct filtration
mode in a substantially under dosed chemical regime, effluent particle counts and turbidities
track each other. When alum was added at 9.0 mg/L, effluent turbidity and particle
concentrations ripened in 15 to 20 minutes, respectively. Filtrate particle concentrations
decreased from the beginning of the filter run; there was no indication of a rise to a peak particle
concentration during the ripening period. With an alum dosage of 13.5 mg/L (Figure 6.10),
filtrate turbidity initially increased, reached a peak, and then ripened in about 20 minutes. Filtrate
particle concentrations were highest at the start of the run, declining rapidly to a minimum in less
than 10 minutes, after which a decline to concentrations of 1 particle/mL or less was maintained.
Breakthrough of particles was observed at about 17 hours into the filter run. These results
indicate that, when a filter is operated in the direct filtration mode with an effective coagulant

137
1U.U 1Z-VV

9.0 *••*•••** *******

8.0 4^™J^^^*t58J^8^®*^l^8**^'^^^^i;8**^^(^^ 1000 g-


^P 5t
7.0 « 800 |
tf
X^S

g 6.0 • Al

£• 5.0 . 600 1
•3
o a
1 4.0
t-i
400 J
3.0 _u
73
2.0 • Filtrate Turbidity 200 1
1.0 - o Particle Concentration
.0.0 J 0
0. 0 0.5 1.0 1.5 2.0 2.5 3.0 3.5 4. 3
Time (hour)

Figure 6.7 Comparison of particle concentrations and turbidity of the filter effluent during a
filter run (direct fill ration mode, filtration rate = 4.9 m/h [2 gpm/ft2], alum = 4.5 mg/L).

10.0 r\nn
^w

9.0
'soo
^ ^s#«S«?fi'Ss'»5!*S^^
8.0
700 ^
7.0
600 ^
1 6-0 • * (S
500 AI
£> 5.0 c
3 400 |
"1 4.0 .*
3.0 300 |
U
0 * Filtrate Turbidity
2.0 200 •§
o Particle Concentration 1
' * 100 *
1.0
00 ^ n
0.0 0.5 1.0 1.5 2.0 2.5 3.0 3.5 4.0
Time (hour)

Figure 6.8 Comparison of particle concentrations and turbidity in the filter effluent during a
filter run (direct filtration mode, filtration rate = 4.9 m/h [2 gpm/ft2], alum = 6.8 mg/L).

138
1.2

* Filtrate Turbidity
1.0
o Particle Concentration 200

0.8
150
£ 0.6
•o
100 |

0.2
<^W*
»I 5u

» •••••• «••••••• ^*+*<


0.0
6 9 12 15 18 21 24
Time (hour)

Figure 6.9 Comparison of particle concentrations and turbidity in the filter effluent during a
filter run (direct filtration mode, filtration rate = 4.9 m/h [2 gpm/ft2], alum = 9.0 mg/L).

l.O low
Q
O
1.4 » * FiltrateTurbidity o«< , 1600
Q

1.2
o Particle Concentration o - 1400
% ,J %
1
I
o
O o2| 1200
1.0 ojffl
^ CsH 1000 Al
0.8 * pg
' 1 - 800 I
0.6 * ~a* 600
X
ParComticle

1JU ^ •...«. »i.. ...>*


0.4
400
0.2 i £• 200

0.0 f\

0 3 6 9 12 15 18 21
Time (hour)

Figure 6.10 Comparison of particle concentrations and turbidity in the filter effluent during
a filter run (direct filtration mode, filtration rate = 4.9 m/h [2 gpm/ft2], alum = 13.5 mg/L).
139
dosage, ripening as measured by filtrate turbidity is slower than ripening as indicated by total
particle concentrations for sizes > 2 urn. Stated another way, these results indicate that turbidity
is a more sensitive parameter for indicating filtrate quality than particle concentration at the start
of filtration runs under these conditions (direct filtration, appropriate chemical dosages). If the 4
um latex particles added to the model water (Chapter 3) are a good surrogate for
Cryptosporidium parvum oocysts in this process, then filtrate turbidity measurements provide a
conservative indicator of Cryptosporidium removal during filter ripening in these systems.

The effects of two alum dosages, 9.0 and 13.5 mg/L, on the removal of UV absorbance
(UV254) are presented in Figure D.3. At an alum dosage of 9.0 mg/L, the UV254 absorbance of the
filtrate increased to a peak about one hour after the start of filtration and then decreased
gradually for the remainder of the 4-hour run. When a dosage of 13.5 mg/L of alum was used,
the peak occurred at a lower value of UV254 absorbance and ripening was complete in about 30
minutes. The results in Figure D.3 indicate removals of UV254 absorbance in the order of 90%
(one log removal). Since the DOC in the Dismal Swamp water source is readily treated by
coagulation, this result is reasonable.

In Figure 6.6, and Figures 6.10 through Figure 6.14, the performance of the dual-media
filter operated for 21 hours at an alum dosage of 13.5 mg/L is presented. The purpose of these
experiments was to observe filter behavior during stable operation over a complete filter run. The
rate of headless development across the bottom or sand layer of the filter (Figure 6.11) increased
significantly at about 11 hours into the filter run. The headloss results indicate that most of the
removal accomplished by the bed during the first 11 hours of filter operation occurred in the
anthracite layer. After that time, the rate of headloss increased significantly in the sand layer and
decreased somewhat in the anthracite layer, indicating significant penetration of the anthracite by
particulate matter. This is confirmed by the data for turbidity and UVa54 absorbance presented in
Figure 6.6 and Figure D.4. The UV254 absorbance observed in the effluents from the anthracite
layer and the dual-media bed at an alum dosage of 13.5 mg/L are presented as functions of
filtration time in Figure D.4. The UV254 absorbance of the anthracite effluent remained at about
the same level as that in the bed effluent filtrate for about 11 hours, after which it started to
increase rapidly. At the end of the run it reached values near the filter influent (1 per 10 cm),
indicating little or no removal of DOC by the anthracite layer in the latter stages of the run.

140
1 *t\J

* Anthracite ' '


120 o0 o<
O Filtrate _r>O
Cr
100 0 °
O
^
^ 80 o
CO 0
3
•o o
60 o ***<
o ^ ^* •^v-w
40 0 ° ° ° 4. • *
CDO 0 ' ^ «. * *

.20, d^0 * * * * *
iP ^» *
n ^ (^
9 ' 12 - 15 18- 21
Time (hour)

Figure 6.11 Headloss development across the anthracite layer and the entire dual-media
filter during a filter run at an alum dosage of 13.5 mg/L (direct filtration mode, filtration rate =
4.9 m/h [2 gpm/ft2]).

l.OU 10

1.40 * * FiltrateTurbidity 14 2"


o Particle Concentration
1.20 .2 |
o a
£3
fce- i.oo 10 S
0 » o
1" 0.80 8 1
o c
o
1
H 0.60 * 6 c
o
^ o O
0.40 * 4 -3
a
0.20 2 fc
o V*»
4 •»»»» Oo On ** * * *O * ** * <
n nn OnOO /% O ft O rt ^rtrtrtft O
n
10 20 30 40 50 60
Time (minute)

Figure 6.12 Comparison of turbidity and particle concentration of dual-media effluents during
the ripening period at an alum dosage of 13.5 mg/L (direct filtration mode, filtration rate = 4.9
m/h [2 gpm/ft2]).

141
l.OU LOW

1.40 » FiltrateTurbidity ° 0 - l 1600


o Particle Concentration fc
o 1400
1.20
* 0 £
1200
1.00 (N

•§
Al
% ° <P^ 1000
0.80 •
°^fe 800
0.60
3J& * 600
0.40 ^^* 400
0.20 200
4 * * V^> *
0.00 < 0
14 15 16 17 18 19 20 21
Time (minute)

Figure 6.13 Comparison of turbidity and particle concentration of dual-media effluents during
the breakthrough at an alum dosage of 13.5 mg/L (direct filtration mode, filtration rate = 4.9 m/h
[2 gpm/ft2]).

4.5

4.0

3.5

3.0

2.5

2.0 f

s 1.5

1.0

0.5

0.0
0 9 12 15 18 21
Time (hour)

Figure 6.14 Log removal of particles during a filter run at an alum dosage of 13.5 mg/L
(direct filtration mode, filtration rate = 4.9 m/h [2 gpm/ft2]).

142
In contrast, the UV254 absorbance data for the bed effluent remained low and tracked the
turbidity measurements for the entire run.

Ripening of turbidity and particle removal is illustrated in Figure 6.12. Turbidity removal
by the dual-media effluent ripened at about 20 minutes from the start of the filter run, whereas it
took only 10 minutes for the ripening of particle removal. Figure 6.13 shows the breakthrough
behavior of the dual-media filter in terms of turbidity and particle concentration. The turbidity of
the filtrate from the dual-media bed increased above 0.10 NTU after about 19 hours of filtration.
Particle concentration exceeded 10 particles/mL after 17 hours of filtration and rapidly increased
for the remaining of the filter run. The turbidity of the effluent from the anthracite layer
increased or broke through at about 6 hours (Figure 6.6). These, headloss, turbidity, and UV254
absorbance data indicate that the bottom sand layer, in addition to providing significant removal
during the ripening period at the start of a run, also provided substantial removal, a type of
multiple barrier, over the entire last half of the run. The results for particle removal by the bed
are presented in another way in Figure 6.14. Particle removals by the dual-media bed are
expressed as "log removal" and plotted as a function of filtration time in Figure 6.14. After the
ripening period, log removals of 4.0 or greater are observed up to a filtration time of about 16
hours, after which they deteriorate rapidly.

High Filtration Rate

The objective of this study was to investigate the performance of the dual-media filter at
various coagulant dosages when operated at a high filtration rate (14.7 m/h [6 gpm/ft2]). Four
filtration experiments were conducted at alum dosages of 6.8, 9.0, 11.3, and 13.5 mg/L. Each of
these experiments lasted until both the turbidity and particle counts reached breakthrough.

The development of headloss across the anthracite layer with four different alum dosages
is illustrated in Figure 6.15. In general, the headloss increased with increasing alum dosage. A
slight decrease in the rate of headloss development was observed after the capacity of the
anthracite layer was exhausted. This result is confirmed by the behavior of the other parameters
such as turbidity and UV254 absorbance, which are discussed subsequently. When 6.8 mg/L of
alum was used, the headloss development was negligible for the first four hours, indicating no
significant removal by this layer, and then increased for the remainder of the filter run. The final
143
120 |-

100

00

* 6.8 mg/L
O 9.0 mg/L
A 11.3 mg/L
* * * * X 13.5 mg/L

246 10 12
Time (hour)

Figure 6.15 Effect of alum dosage on headless development by the anthracite layer during a
filter run (direct filtration mode, filtration rate = 14.7 m/h [6 gpm/ft2]).

headloss at breakthrough with this alum dosage was 64.5 cm. At an alum dosage of 9.0 mg/L, the
headloss increased for 3.5 hours, after which the rate of headloss development decreased
significantly. The final headloss across the anthracite media with.this alum dosage was 53.5 cm
and was obtained after 5 hours of filter operation. Comparing the results from the 6.8 and 9.0
mg/L alum dosages, it can be seen that a modest change in coagulant dose immediately enables
the filter to function effectively at around the optimum coagulant dosage. When a dosage of 11.3
mg/L of alum was used, the final headloss at breakthrough was 89 cm, which was also obtained
at 5 hours into the filter run. A decrease in the rate of headloss was observed at about 3.5 hours
but was not as obvious as in the lower alum dosages. At 13.5 mg/L of alum dosage, the final
headloss at the bottom of the anthracite layer was 97 cm. The rate of headloss development
significantly decreased after 4 hours of filter run. Except for the case with 6.8 mg/L of alum
addition, the final headloss at breakthrough increased with increasing alum dosage. From the
results at alum dosages of 9.0, 11.3, and 13.5 mg/L, it can be seen that, when operated at high

144
filtration rate, the anthracite layer was exhausted at 3.5 to 4 hours, after which most of the
particulate matter passed through this layer and was removed in the lower sand media. This is
confirmed by the increase in the rate of headloss development by the entire dual-media filler that
occurred at about the same time and indicates that particles that pass through the anthracite
media or aggregates dislodged from this media are effectively removed by the lower sand media
when the performance of the upper anthracite layer of the dual-media filter starts to deteriorate.

The effect of alum dosage on the headloss development across the entire dual-media bed
is presented in Figure 6.16. As with the results from the anthracite layer, the headloss
development was greater with increasing alum dosages. Slight increases in the rate of headloss
development occurred during each run and appear to be associated with the decreases in the rate
across the anthracite layer. At an alum dosage of 6.8 mg/L, total headloss development was
negligible but slightly higher than in the anthracite layer for the first 3.5 hours of filter run. This
confirms the view that most of the particles pass through the filter media during the ripening
period at this alum dose. The headloss then increased to 162.2 cm at the end of the run. At alum
additions of 9.0 and 11.3 mg/L, the final headlosses at breakthrough were 146.5 and 171.1 cm,
respectively. At these two dosages the rate of headloss increased slightly after 2.5 hours of filter
operation. When the alum dosage was 13.5 mg/L, the rate of headloss slightly increased after 2.5
hours and significantly decreased at 5 hours. The final headloss at this alum dosage was 270 cm.

The effect of alum dosage on turbidity removal by the anthracite layer is demonstrated in
Figure 6.17. At an alum dosage of 6.8 mg/L, no removal of turbidity was observed during the
first 3 hours. The turbidity then decreased until 6 hours, after which it started to break through
rapidly. This result is consistent with the headloss development discussed earlier. When higher
alum dosages (9.0, 11.3, and 13.5 mg/L) were used, the turbidity removal by the anthracite layer
was not affected significantly by the change in alum dosage. The breakthrough of turbidity at this
layer occurred at about 1.5 hours into the filter run.

145
300

250

*6.8mg/L
O9.0mg/L
A11.3mg/L
X 13.5

6 8 10 12
Time (hour)

Figure 6.16 Effect of alum dosage on headless development by the entire dual-media bed
during a filter run (direct filtration mode, filtration rate = 14.7 m/h [6 gpm/ft2]).

A 0
X40
*6.8mg/L
A° O9.0mg/L
All.3mg^
X 13.5 mg/L

10 12
Time (hour)

Figure 6.17 Effect of alum dosage on turbidity removal by the anthracite layer during a filter
run (direct filtration mode, filtration rate = 14.7 m/h [6 gpm/ft2]).

146
* 6.8 mg/L
O 9.0 mg/L
A 11.3 mg/L
X 13.5 mg/L

10 12
Time (hour)

Figure 6.18 Effect of alum dosage on turbidity removal by the entire dual-media bed during
a filter run (direct filtration mode, filtration rate =14.7 m/h [6 gpm/ft2]).

The turbidity removal by the entire dual-media bed is illustrated in Figure 6.18. At an
alum dosage of 6.8 mg/L, ripening occurred for about 6 hours, after which the turbidity
increased. At alum dosages of 9.0 and 11.3 mg/L, turbidity started to break through at 3.5 hours
and 4 hours into the filter run, respectively. When the alum dosage was 13.5 mg/L, breakthrough
of turbidity occurred after 3 hours of filter run. As demonstrated in the headloss profiles, the
lower sand layer functions effectively and serves as a secondary barrier for the removal of
turbidity for a considerable time after the effectiveness of the upper anthracite layer is
significantly decreased.

The turbidity breakthrough occurred much faster than expected at this filtration rate when
compared to the results at the conventional filtration rate. The reason is not clear from the data
obtained. The Dismal Swamp water used as a source of natural organic matter in these filtration
studies has a higher coagulant demand than the DOC in many natural surface waters, and
therefore the alum dosages applied in the study at Johns Hopkins are greater than the dosages
normally practiced at these DOC concentrations in the field. These higher alum dosages generate

147
more precipitates, which consequently increase the headloss development and shorten the
duration of filter run. However, it was not expected that, at a given alum dosage, the filter run
length would be reduced so much when operated at the higher filtration rate.

The profiles of particle concentration during the full filter operation are shown in Figure
6.19. When the alum dosage was 6.8 mg/L, particles ripened in 5 hours and then immediately
started to break through. At alum additions of 9.0 and 11.3 mg/L, breakthrough occurred at 3 and
4 hours respectively. At an alum dosage of 13.5 mg/L, no significant particle breakthrough was
observed during the entire 12-hour filter run. This is significantly different from what was
observed for turbidity, where breakthrough began after about 3 hours. Except when alum dosage
was 13.5 mg/L, particle breakthrough was observed earlier than that of turbidity. Considering
only the three lower dosages (6.8, 9.0, and 11.3 mg/L), it is plausible that particle counts can
serve as a better indicator of deteriorating filter performance than turbidity at the end of the filter
operation.

Figure 6.20 depicts the log removal of particles during the full filter run. When the alum
dosage was 6.8 mg/L, a maximum of only 1.5 log removal could be achieved. At alum dosages
of 9.0, 11.3, and 13.5 mg/L, the best log removals were 3.6, 4.2, and 4.3, respectively. The
removal dropped below 3.0 log 10 in 1.3, 2.5, and 6 hours when alum dosages were 9.0, 11.3,
and 13.5 mg/L, respectively.

Figure D.5 and Figure D.6 illustrate, respectively, the removal of UV254 absorbance by
the anthracite layer and the entire dual-media bed during a full filter run. At an alum dosage of
6.8 mg/L, UV254 absorbance removal has a very extensive ripening period. At the higher alum
dosages, about 95% removal is maintained after short ripening. The removal of UV254
absorbance by the dual-media bed was slightly higher when alum dosages were 11.3 and 13.5
mg/L as compared to 9.0 mg/L. The removal by the anthracite layer was not significantly
changed at these three dosages. At all alum dosages, the removals by the entire dual-media bed
were not dramatically better than the removals by the anthracite media. These results indicate
that the dissolved NOM is rapidly and effectively converted into particulate form at alum dosage
of 9.0 mg/L and higher. Any NOM remaining in solution after the alum reacts is not removed by
either media.

148
1800

1600
» 6.8 mg/L
I 1400 O 9.0 mg/L
A 11.3 mg/L
I 1200
<N
Al
X 13.5 mg/L
1000

I
<L>
U
800

Time (hour)

Figure 6. 19 Effect of alum dosage on the removal of particles by the entire dual-media bed
during a filter run (direct filtration mode, filtration rate = 14.7 m/h [6 gpm/ft2]).

ex >KX
» 6.8 mg/L
13.5 mg/L
O 9.0 mg/L
A 11.3 mg/L
X 13.5 mg/L

10 12
Time (hour)

Figure 6.20 Effect of alum dosage on the log removal of particles by the entire dual-media
bed during a filter run (direct filtration mode, filtration rate = 14.7 m/h [6 gpm/ft2]).

149
The removals of DOC at four alum dosages are illustrated in Figure D.7 and Figure D.8.
The general trend of the DOC removal is consistent with that of UV254 absorbance. With 6.8
mg/L of alum, about 90% removal was achieved after 5 hours of ripening. At higher dosages,
ripening was much shorter or not obvious, and 90% of removal was maintained until the end of
the filter run. A slight decrease in removal was observed with these dosages at the end of the
runs. As with UV254 absorbance, no significant increase in the removal of this dissolved material
was achieved by the sand layer as compared to the removal by anthracite layer.

The effect of alum dosage on the duration of ripening period is depicted in Figure 6.21
and Figure 6.22. As shown in Figure 6.21, the turbidity of the dual-media filter effluent ripened
slightly faster with alum dosages of 11.3 and 13.5 mg/L than at 9.0 mg/L. At dosages of 11.3 and
13.5 mg/L, ripening occurred at about 5 minutes, while at 9.0 mg/L of alum ripening occurred a
little later. For any of the alum dosages tested, the turbidity during the stable operation period did
not reach below 0.10 NTU. The turbidity after ripening was maintained below 0.10 NTU when
the lower filtration rate (2 gpm/ft2) was applied. Figure 6.22 illustrates the ripening behavior of
particles at alum dosages of 9.0, 11.3 and 13.5 mg/L. Ripening was reached at 4 minutes when
alum dosages were 11.3 and 13.5 mg/L. When 9.0 mg/L of alum was used, ripening occurred at
about 6 minutes. At all alum dosages, particle removal ripened faster than the turbidity removal.
This result is identical to that shown in conventional filtration rate (2 gpm/ft2). At the initial
stages of a filter run, turbidity can be used as a better (more conservative) indicator of possible
passage of pathogens during the ripening period than particle counts.

Effect of pH

The effect of pH on filter performance was investigated in the direct filtration mode
operated at the conventional filtration rate. Jar tests were performed at pH 6 and 8 with extended
flocculation time (48 hours). Two filtration experiments were conducted: at pH 6 with an alum
dosage of 6.8 mg/L and at pH 8 with an alum dosage of 9.0 mg/L.

Figure D.9 through Figure D.12 illustrate the results of jar tests conducted at pH 6 and 8.
At pH 6, the optimum alum dosage was 9.0 mg/L for all settled parameters (settled turbidity and
settled UV254 absorbance) and 8.1 mg/L for all filtered parameters. At pH 7 the optimum alum
dosage was found to be 9.0 mg/L both from jar test and direct filtration experiments. The
150
3.0

2.5 O 9.0 mg/L


A 11.3 mg/L
2.0 X 13.5 mg/L

1.0

0.5
'oooo.
0.0
i
10 20 30 40 50 60
Time (minute)

Figure 6.21 Effect of alum dosage on the ripening of turbidity by the entire dual-media bed
(direct filtration mode, filtration rate = 14.7 m/h [6 gpm/ft2]).

20

18 x O 9.0 mg/L
?16 A 11.3 mg/L
B 14 <> X 13.5 mg/L

S 12
§ 10

S 6
1 4
V

S, 2

0
10 20 30 40 50 60
Time (minute)

Figure 6.22 Effect of alum dosage on the ripening of particles by the entire dual-media bed
(direct filtration mode, filtration rate = 14.7 m/h [6 gpm/ft2]).
151
optimum dosage at pH 8 was 11.7 mg/L for all settled parameters, 9.9 mg/L for filtered turbidity,
and 10.8 mg/L for filtered UV254 absorbance. In summary, the optimum alum dosage generally
decreased as the pH of coagulation decreased. This might be due to the fact that the charge of
aluminum species becomes increasingly positive at lower pH and the negative charge of natural
organic matter is diminished. It can require less alum to react with NOM and neutralize the
charge on the NOM as the pH of coagulation is lowered.

One direct filtration experiment was conducted at pH 6 with the alum addition of 6.8
mg/L. When operated at pH 7, the performance of the filter with this dosage was inadequate,
with negligible removal of turbidity and particles (see Figure 6.4 and Figure 6.8). At pH 6, the
performance of the filter was not excellent but significantly improved compared to the case of
pH 7 at a given dosage. Another filter experiment was conducted at pH 8 with the alum dosage
of 9 mg/L. This concentration of alum has been considered as the optimum dosage at pH 7.
However at the high pH, negligible removal was observed whereas the filter functioned very
effectively at pH 7.

The profiles of headloss development by the anthracite layer and the dual-media filter at
pH 6 and 8 are presented in Figure 6.23 and Figure 6.24. The headloss at the end of the filter run
was 80 cm across the dual-media bed at pH 6 and 6.8 mg/L of alum. This is still lower than the
typical final headloss of the filter operated at optimum condition (100 to 120 cm), but
significantly higher when compared to the negligible headloss build-up at pH 7 with the same
alum dosage. It can be pointed out that the removal of particles was enhanced by lowering the
pH. The contribution of the anthracite layer to the headloss development across the dual-media
was relatively small, suggesting that most of the removal was achieved by the sand layer. The
headloss across both the anthracite layer and the dual-media bed did not increase at all when
operated at pH 8 with the optimum alum dosage optimum for pH 7.

The removal of turbidity is presented in Figure 6.25 and Figure 6.26. At pH 6, the
turbidity of the anthracite effluent was around 4 NTU and was markedly improved to 0.2 - 0.5
NTU across the sand media, which again confirms the idea that the underlying sand layer
functions as an effective trap for the particles that pass through the anthracite layer when the

152
25

O alum = 6.8 mg/L, pH 6


20 A alum = 9.0 mg/L, pH 8

OO
15

I 10
X O O

A A AAAAAAAAAAAAi,

6 9 12 15 18 21 24
Time (hour)

Figure 6.23 Effects of pH on headloss development by the anthracite layer during a filter run
(direct filtration mode, filtration rate = 4.9 m/h [2 gpm/ft2]).

90

80 O alum = 6.8 mg/L, pH 6


70 A alum = 9.0 mg/L, pH 8
.00

60

50

40

30
oo
20
A A A A AAAAAAAAAAAAn
10

0
0 3 6 9 12 15 18 21 24
Time (hour)

Figure 6.24 Effects of pH on headloss development by the entire dual-media bed during a filter
run (direct filtration mode, filtration rate = 4.9 m/h [2 gpm/ft2]).

153
7 , A AAAA* A A AAAAAAAAAAi,

OO O O
£ 4 O O
rs
(2 3

2
O alum = 6.8 mg/L, pH 6
1 A alum = 9.0 mg/L, pH 8

0
6 9 12 15 18 21 24
Time (hour)

Figure 6.25 Effects of pH on turbidity removal by the anthracite layer during a filter run (direct
filtration mode, filtration rate = 4.9 m/h [2 gpm/ft2]).

AAAAAAAA AAAAAAAAAA^,

O alum = 6.8 mg/L, pH 6


A alum = 9.0 mg/L, pH 8

£ 4
13

13

'21
9 12 15 18 24
Time (hour)

Figure 6.26 Effects of pH on turbidity removal by the entire dual-media bed during a filter run
(direct filtration mode, filtration rate = 4.9 m/h [2 gpm/ft2]).

154
250

alum = 6.8 mg/L, pH 6


alum = 9.0 mg/L, pH 8

9 12 15 18 21 24
Time (hour)

Figure 6.27 Effects of pH on the removal of particles by the entire dual-media bed during a
filter run (direct filtration mode, filtration rate = 4.9 m/h [2 gpm/ft2]).

alum = 6.8 mg/L, pH 6


alum = 9.0 mg/L, pH 8

9 12 15 18 21 24
Time (hour)

Figure 6.28 Effects of pH on the log removal of particles by the entire dual-media bed
during a filter run (direct filtration mode, filtration rate = 4.9 m/h [2 gpm/ft2]).

155
performance of the upper layer is inadequate. At pH 8, no removal of turbidity was observed as
was expected and confirmed by constant headless at this pH.

As shown in Figure 6.27 and Figure 6.28, at lower pH, the particle concentration
decreased modestly after ripening and then slowly increased until the end of the filter run. On the
other hand, at higher pH the particle concentration reached a maximum of 200 particles/mL after
which it slowly decreased.

The reduction in UV254 absorbance and DOC was relatively effective compared to the
removal of other parameters such as particle concentration (Figure D.I3 through Figure D.I6).
At pH 6 and with 6.8 mg/L of alum, the UV254 absorbance of the anthracite effluent stayed at
0.1-0.2/10 cm. The dual-media bed reduced the UV254 absorbance to a level below 0.1/10 cm
during the entire filter run. This corresponds to more than 90% removal from the raw water level
and is close to the efficiency of the filter operated at optimum alum dose of 9 mg/L at pH 7.

Ripening was observed at pH 6 with 6.8 mg/L of alum using only the particle
concentration measurements, which occurred at about 15 minutes after the start of the filter run;
significant amount of turbidity passed through the bed at all times (Figure 6.29 and Figure 6.30).

AA*

O alum = 6.8 mg/L, pH 6


£ A alum = 9.0 mg/L, pH 8
£ 4
''•£
T3

10 20 30 40 50 60
Time (minute)

Figure 6.29 Effects of pH on the ripening of turbidity by the entire dual-media (direct filtration
mode, filtration rate = 4.9 m/h [2 gpm/ft2]).
156
1O £.J\J

*> 16 O 00

SB J» K
0. <
"2 200 *
"" A AA A A^ 4^A AA^ AA^AAAAA ^ A A
g 0 A A^ ^ ^ AAA^^ A^AA ^AA^^A^ , 1
£- 12 °°oo ^^
g 150 I
<N 10 <N
A Al
Al O alum = 6.8 mg/L, pH 6
1 8
A alum = 9.0 mg/L, pH 8 100 |
g 6 .. u
o
CJ
C *l^ . A c
o o
U 4 v"^ {* A /vO O
J-
"o vO0v ACto^. Oo O O ^A O 50 .20
1 2
A ^^^^
o ^^s^^^
o oo^ s\s^
oo O
o ^O
°°oo^O< , c
CU 0,

oU ^ i————
A^AAA. . - ———— , . —— , —- ^ ,
—. --_ r - i
- — ———————————— . A
v

0 10 20 30 40 50 60
Time (minute)

Figure 6.30 Effects of pH on the ripening of particles by the entire dual-media bed (direct
filtration mode, filtration rate = 4.9 m/h [2 gpm/ft2]).

Summary

A dual-media direct filtration process treating a water moderate in turbidity and low in
hardness, alkalinity and NOM was studied at conventional (4.9 m/h [2 gpm/ft2]) and high (14.7
m/h [6 gpm/ft2]) filtration rates. Alum was used as the sole coagulant at dosages of 4.5, 6.8, 9.0,
and 13.5 mg/L for conventional rate experiments and 6.8, 9.0, 11.3, and 13.5 mg/L for high
filtration rate experiments. The optimum dosage for the direct filtration experiments was
estimated using jar tests with 48 hour fiocculation. The effects of pH on the performance of the
filter in direct filtration mode were studied at the conventional filtration rate and compared with
the results from jar tests. The results are summarized as follows:

1. The jar tests with extended flocculation provide a useful estimate for the optimum
coagulant dosage for direct filtration. The dual-media filter performs effectively when
operated in direct filtration mode with the optimum alum dosage estimated from these
jar tests, while it deteriorates significantly when the dosage is decreased. Relatively
modest reductions in coagulant dosage from the optimum can transform an effective
filter into an ineffective one.
157
2. Turbidity measurements provide a more sensitive monitor of filter performance than
particle concentration determinations during the ripening phase of a filter run. If the
negatively charged latex particles with a diameter of 4 urn added to the model water
in these experiments are a useful surrogate for Cryptosporidium oocysts, then
turbidity measurements are a more conservative indicator of the possible presence of
these pathogens than are particle concentration determinations during filter ripening.
3. Generally, particle concentration is considered to be a preferred parameter for
monitoring filter breakthrough at the end of a run. Except for one experiment
operated at high filtration rate with alum dosage of 13.5 mg/L, the measurements of
particle concentrations appear to be a more conservative (better) indicator of filter
breakthrough (and possibly Cryptosporidium breakthrough) than do measurements of
turbidity.
4. In a dual-media filter, the second, lower layer of media provides important removal
during the ripening phase of the process, contributes to removal throughout the main
part of the run, and can dominate removal for a significant time after the effectiveness
of the upper layer has deteriorated.
5. Higher filtration rates resulted in significantly shorter filter operation compared to the
conventional rate although the general profile of the performance of the filter was
similar to the results obtained at the conventional filtration rate.
6. The optimum alum dosage was decreased at lower pH. At suboptimum dosage, the
efficiency of the filter improved significantly at pH 6 when compared at pH 7. On the
other hand, higher pH deteriorated the filter performance when alum dosage that was
optimum at pH 7 was used.

CONVENTIONAL TREATMENT STUDIES

Effect of Alum Dosage

In order to determine the optimum alum dosage for the conventional mode filtration
studies, a conventional jar test was performed prior to filtration experiments using the model raw
water with a high concentration of natural organic matter (DOC = 5 mg/L). Results of the jar test
are listed in Table 6.3 and Table 6.4, and illustrated in Figure D.I7 through Figure D.21. The

158
Table 6.3
Comparison of turbidity (NTU) removal by jar test and conventional mode filtration

Jar test Conventional treatment


Alum dosage Settled turbidity Filtered turbidity Sedimentation Dual-media
supernatant effluent
27mg/L 4.2 2.4 3.6 2.2 - 2.9
36 mg/L 0.49 0.24 0.67 0.10-0.15
45mg/L 0.39 0.16 0.24 0.08 - 0.09

Table 6.4
Comparison of UV254 absorbance (710 cm) removal by jar test and conventional mode filtration

Jar test Conventional treatment


Alum dosage Settled UV254 Filtered UV254 Sedimentation Dual-media
absorbance absorbance supernatant effluent
27 mg/L 1.3 0.90 0.550 0.348
36 mg/L 0.28 0.23 0.172 0.146
45 mg/L 0.13 0.22 0.165 0.117

optimum alum dosage based on settled water turbidity and UV254 absorbance measurements was
found to be 36 mg/L. For the filtered parameters including turbidity, UV254 absorbance, and
DOC, optimum dosage was 32 mg/L. Based on the settled parameters, an alum concentration of
36 mg/L was selected as an optimum dosage for the conventional treatment experiments.
Experiments were also made with both a 25% decrease and increase of alum in order to study the
effect of alum dosage on the ripening and performance of the filter. The corresponding alum
dosages were 27 mg/L for under dosing and 45 mg/L for overdosing. The conventional treatment
experiments were conducted at both a low filtration rate (4.9 m/h [2 gpm/ft2]) and a high
filtration rate (14.7 [6 gpm/ft2]). For high filtration rate studies, alum dosages of 36 and 45 mg/L
were tested.

159
Figure 6.31 and Figure 6.32 illustrate the effect of alum dosage on the development of
headloss when the filter was operated at low filtration rate. When 27 mg/L of alum was added,
headloss development was higher compared to higher dosages. At alum dosages of 36 and 45
mg/L, headloss increased very slowly over the entire 27 hours of filter run and the difference
between these two results was negligible. The profile of headloss development in the
conventional treatment mode is distinguished from that in the direct filtration mode by two
aspects. First, the rate of headloss development is significantly slower in conventional mode in
comparison to the direct filtration mode. Second, higher alum dosages result in slower headloss
development when the filter is operated at conventional mode and the difference due to the
change in alum dosage is not significant. In direct filtration mode, the headloss increased more
rapidly with higher alum dosages and the difference in the rate of headloss development was
significant. When the filter is operated in direct filtration mode, higher alum dosages increase the
particle loading to the filter media and consequently result in more rapid headloss development.
On the other hand, in the conventional mode, higher alum dosage promotes the floe formation of
particles/NOM and increases the settlability of floes in the coagulation and sedimentation
processes prior to filtration, hence reducing the amount of total particle mass applied to the filter
media. In the previous studies on the effect of alum dosages at direct filtration mode, overdosing
of alum resulted in some possible disadvantages including rapid build-up of headloss and hence
shortened filter service period. In the conventional mode, however, these disadvantages were not
observed when alum was added at overdose concentration.

The effects of alum dosage on turbidity removal by the anthracite layer and entire dual-
media filter are shown in Figure 6.33 and Figure 6.34. When alum was added at 27 mg/L, the
turbidity continuously increased over the entire filter run, reaching final turbidities of 3.9 NTU
by the anthracite layer and 2.9 NTU by the dual-media. At alum dosages of 36 mg/L, the
turbidity of the anthracite effluent reached 0.09 NTU after ripening and increased to 0.26 NTU at
the end of the filter run. The turbidity of the dual-media filtrate remained below 0.10 NTU for 20
hours of filter run and increased slightly thereafter. When the alum dosage of 45 mg/L was used,
the turbidity remained below 0.10 NTU for both anthracite effluent and dual-media filtrate over
the entire filter run. Breakthrough of particles was not observed until the end of the run at these
dosages. The turbidity results suggest that a 25% decrease in alum dosage from the optimum

160
It

12 * * •

10 *****
? * * * <
^
W
8 vv"

5 A ^ A A Ak A A A A ™
I

"« ^
Q>
4
4; ^ 27 in&L

2
A45mg/L
0
() 3 6 9 12 15 18 21 24 27
Time (hour)

Figure 6.3 1 Effects of alum dosage on headloss development by the anthracite layer during a
filter run (conventic>nal treatment mode, filtration rate = 4.9 m/h [2 gpm/ft2]).

45

40 .»*
35 i .»**
-, 30

f 25 - ^.'**
t
« ^HKAAAA
^•^AAAAA A
» A A A *A* A
A A OOv" ^^ O
•v-v--'^ •*• oO^O
O •" *
32 i H^O OOO v v v ^
15 '

» 27 mg/L
10

5 A 45 mg/L
0
C 3 6 9 12 15 18 21 24 27
Time (hour)

Figure 6.32 Effects of alum dosage on headloss development by the entire dual-media bed
during a filter run (conventional treatment mode, filtration rate = 4.9 m/h [2 gpm/ft2]).
161
H.J

4.0
A A
***** *
3.5
^ *

* * 27 mg/L
5, 2.5
O36mg/L
1 2.0 A 45 mg/L

5 1.5

1.0 -

0.5 L
,., Ui«Mii . 1 1 o 1 1 j A»A A ^ 2 SA° i jon rrri ^
,

0 3 6 9 12 15 18 21 24 2'7
Time (hour)

Figure 6.33 Effe:cts of alum dosage on turbidity removal by the anthracite layer during a filter
run (conventional treatment mode, filtration rate = 4.9 m/h [2 gpm/ft2]).

3.5

3.0

2.5
_«~~*
.... — • — ••" '
1 2.0 Jj*^ • 27 mg/L
£ O 36 mg/L
f 1.5 ^ A 45 mg/L
H

1.0 t
>
0.5

0.0 4
9 12 15 18 21 24 27
Time (hour)

Figure 6.34 Effects of alum dosage on turbidity removal by the entire dual-media bed during
a filter run (conventional treatment mode, filtration rate = 4.9 m/h [2 gpm/ft2]).

162
concentration changes the filter performance from effective to ineffective, while overdosing of
alum does not make dramatic difference. The results for turbidity removal by conventional
filtration are listed in Table 6.3.

The effect of alum dosage on particle removal is shown in Figure 6.35. When the alum
addition was 27 mg/L, particle concentration continuously and rapidly increased and nearly
reached the raw water concentration at the end of the run. As with the results for turbidity, the
removal of particles was negligible at this alum dosage. At 36 mg/L of alum addition, a slight
increase in particle concentration was observed at the later part of the filtration period, but
considering the low headloss at this stage it is not obvious that this is due to the breakthrough of
the filter. As shown in Figure 6.36 the removal of particles reached approximately 3 log removal
after ripening and then slightly decreased to 2 log removal. The log removals of particles in the
conventional filtration were calculated as a decimal log of particle counts in the filter effluent
subtracted from that of raw water. Therefore, it represents the efficiency of the entire treatment
procedure including coagulation and sedimentation as well as filtration. At an alum dosage of 45
mg/L, particle concentration stayed constantly low during the filter run, which maintained at 3.5
log removal. Breakthrough was not observed at this alum dose. The high quality of the anthracite
and dual-media effluents shown in the results for turbidity and particle concentration is due to
the fact that most of the turbidity and particles are removed in the coagulation and sedimentation
processes prior to filtration when coagulant dosages are appropriate. This is related to and
confirmed by low headloss development at high alum dosages. The result in conventional mode
is in contrast to the direct filtration results in which better removal of turbidity and particles
contributed to faster headloss development and shortened the filter run.

The removals of UV254 absorbance by the anthracite layer and the dual-media bed are
presented in Figure D.22 and Figure D.23. The removals of DOC are illustrated Figure D.24 and
Figure D.25. Comparison of UV254 absorbance removal by jar test and conventional filtration
experiments is presented in Table 6.4. The UV254 absorbance of the sedimentation effluent were
0.55, 0.17, and 0.17/10 cm at alum dosages of 27, 36, and 45 mg/L, respectively. At an alum
dosage of 27 mg/L, the UV254 absorbance of the anthracite layer and dual-media effluents both
stayed at approximately 0.35/10 cm. When alum addition was 36 and 45 mg/L, the UV254

163
800
_ . o- - - 27 mg/L
36mg/L
45 mg/L

Figure 6.35 Effects of alum dosage on the removal of particles by the entire dual-media bed
during a filter run (conventional treatment mode, filtration rate = 4.9 m/h [2 gpm/ft2]).

45 mg/L
o- - - 27 mg/L
36 mg/L
45 mg/L

12 15
Time (hour)

Figure 6.36 Effects of alum dosage on the log removal of particles by coagulation and the
entire dual-media bed during a filter run (conventional treatment mode, filtration rate = 4.9 m/h
[2 gpm/ft2]).

164
absorbance of the anthracite and dual-media beds were 0.15 and 0.12/10 cm, respectively. No
difference was found between the anthracite layer and dual-media bed at these dosages.

The ripening behavior of turbidity and particle concentration at alum dosages of 36 and
45 mg/L is presented in Figure 6.37 and Figure 6.38, respectively. Ripening of the filter was not
sensitive to the alum dosages tested in this study. At both dosages, ripening of turbidity occurred
after 13 minutes from the start of filter run. Ripening of particle concentration occurred at 10
minutes for these alum dosages. As with the results from direct filtration studies, during the
initial stage of the filtration cycle, turbidity can be used as a better or conservative indicator of
filter ripening than particle concentration. Two differences were noticed in ripening behavior of
the filter operated in conventional mode when compared to the direct filtration mode: (1)
ripening was less sensitive to the alum dosage in conventional mode than in direct filtration
mode, and (2) ripening was more rapid in conventional filtration than in direct filtration. In direct
filtration, ripening was completed in 20 - 25 minutes for turbidity and 15-20 minutes for
particle concentration.

High Filtration Rate

Studies in conventional mode filtration were performed at a filtration rate that was three
times higher (14.7 m/h [6 gpm/ft2]) at alum dosages of 36 and 45 mg/L. The general behavior of
the filter is similar and in agreement to the results obtained at the lower filtration rate. As shown
in Figure 6.39 and Figure 6.40 the rate of headloss development was slower than in the direct
filtration mode and the effect of alum dosage was not significant.

The turbidity of the anthracite layer effluent remained at 0.20 NTU at 45 mg/L of alum
addition and 0.25 NTU when the alum dosage was 36 mg/L (Figure 6.41). Breakthrough was
observed at 6 hours into the filter run when alum was added at 36 mg/L. At this alum dosage, the
turbidity of the dual-media filtrate was about 0.18 - 0.20 NTU for 7 hours of the filter run after
which it started to break through (Figure 6.42). For an alum addition of 45 mg/L, turbidity
remained at 0.11 - 0.12 NTU during the entire filter run.

In Figure 6.43 and Figure 6.44 the removal of particles at two alum dosages are
illustrated. With 36 mg/L of alum, particle concentration continuously increased and reached 1.2

165
0.20

0.18
O 36 mg/L
0.16
oo A 45 mg/L
O
0.14

°-12 &AA
A
O O
O
0.10 O O
A A
A A A A A
0.08
A A

0.06
0.04

0.02

0.00
0 10 20 30 40 50 60
Time (minute)

Figure 6.37 Effects of alum dosage on the ripening of turbidity by the entire dual-media
(conventional treatment mode, filtration rate = 4.9 m/h [2 gpm/ft2]).

40

35 * 27 mg/L
f •
30 O 36 mg/L « *4
A 45 mg/L
25
§ 20

8 15
I *
o 10
O

I 5
*A'

0
10 20 30 40 50 60
Time (minute)

Figure 6.38 Effects of alum dosage on the ripening of particles by the entire dual-media bed
(conventional treatment mode, filtration rate = 4.9 m/h [2 gpm/ft2]).

166
iJ

O <>
20
t

o o ° ° A A A
• ——— Jti* 2 * A A A
0 '

1 10

5 O 36 mg/L
A 45 mg/L

0
()123456789
Time (hour)

Figure 6.39 Effects of alum dosage on headloss development by the anthracite layer during a
filter run (conventicmal treatment mode, filtration rate = 14.7 m/h [6 gpm/ft2]).

80

70
o <
60 :
»

000 0 0 ° 0 ° ° A ,I
"? 5° ' Pff^!*!! A^AA A AA

1 40

1 30

20
O 36 mg/L
10 A 45 mg/L

0
C(123456789
Time (hour)

Figure 6.40 Effects of alum dosage on headloss development by the entire dual-media bed
during a filter run (conventional treatment mode, filtration rate = 14.7 m/h [6 gpm/ft2]).

167
0.35 0 0 <

0.30

S 0.25 <X^X>00000<> 0000°


^^^AAA^AA^A A A A A
£ 0.20 <
jo ^^ AAA A A A A A A A
i

£ 0.15 > L

0.10 < ,
i
O 36 mg/L
0.05 A 45 mg/L

0.00
()123456789
Time (hour)

Figure 6.41 Effects of alum dosage on turbidity removal by the anthracite layer during a filter
run (conventional 1treatment mode, filtration rate = 14.7 m/h [6 gpm/ft2]).

0.30

0.25 o o <>

^ 0.20 ^^
4Q3> O
OOOOOOO O O O
> O
£• 0.15
."2
AJIMAAAAAAA A A A A ii
0.10
k
i L

0.05 O 36 mg/L
A 45 mg/L

0.00
0 123456789
Time (hour)

Figure 6.42 Effects of alum dosage on turbidity removal by the entire dual-media bed during a
filter run (conventional treatment mode, filtration rate = 14.7 m/h [6 gpm/ft2]).

168
5E
•^^

E
<N

1
•g 20
cd

10

6789

Figure 6.43 Effects of alum dosage on the removal of particles by the entire dual-media bed
during a filter run (conventional treatment mode, filtration rate = 14.7 m/h [6 gpm/ft2]).

Figure 6.44 Effects of alum dosage on the log removal of particles by coagulation and the
entire dual-media bed during a filter run (conventional treatment mode, filtration rate = 14.7 m/h
[6 gpm/ft2]).
169
log removal at the end of the filter run. At 45 mg/L of alum, particle concentration remained at
1.7-1.8 log removal. As in the direct filtration studies, turbidity and particle concentration show
that the performance of the filter is less effective when operated at a high filtration rate.

Figure D.26 and Figure D.27 show the removal of UV254 absorbance by the anthracite
layer and the dual-media filter. The removal of UV254 absorbance at this filtration rate was
generally less effective than at the lower filtration rate. The UV254 absorbance of the anthracite
effluent stayed at 0.18/10 cm for 36 mg/L of alum and 0.12/10 cm for 45 mg/L of alum. The
UV254 absorbance of the dual-media effluent was approximately the same as that of the
anthracite effluent at both dosages.

Ripening of turbidity and particles are presented in Figure 6.45 and Figure 6.46. Ripening
was very short or not obvious for both turbidity and particle removal.

0.25

0.20 o ooooooooooooooo

0.15

A A A A A A
AAAAAAA AAAAAAAAAA 4 4
1 0.10

0.05 O 36 mg/L
A 45 mg/L

0.00
10 20 30 40 50 60
Time (minute)

Figure 6.45 Effects of alum dosage on the ripening of turbidity by the entire dual-media
(conventional treatment mode, filtration rate = 14.7 m/h [6 gpm/ft2]).

170
50

45 O 36mg/L
A 45 mg/L

! 35
J 40

<N 30
Al
<>o
C -,c
o 25
o<>

§
U IS
l;>
.
o
I 10

10 20 30 40 50 60
Time (minute)

Figure 6.46 Effects of alum dosage on the ripening of particles by the entire dual-media bed
(conventional treatment mode, filtration rate = 14.7 m/h [6 gpm/ft2]).

Summary

A conventional treatment process treating a water high in natural organic matter and
moderate in turbidity was studied at both low (4.9 m/h [2 gpm/ft2]) and high (14.7 m/h [6
gpm/ft2]) filtration rates. Coagulation and sedimentation were provided in a batch mode followed
by continuous addition of treated water to the dual-media filter. The optimum alum dosage was
determined from laboratory jar tests and the effects of alum dosage on conventional mode
filtration were investigated with both under dosing (27 mg/L) and overdosing (45 mg/L) of alum
as well as with the optimum dosage (36 mg/L). The following conclusions are made:

1. When operated in conventional filtration mode, the quality of dual-media filter


effluent during the stable operation period (after ripening) was in good agreement
with the jar test results using standard flocculation and sedimentation protocol.

171
2. When operated in conventional mode, the performance of the dual-media filter
improved with higher alum dosage. Possible disadvantages of overdosing suggested
in the direct filtration mode (faster headloss development and shorter filter run) were
not observed in conventional filtration mode. When the alum dosage was above
optimum, the dual-media filter operated in conventional mode produced better filtrate
quality for a longer filter service period than in direct filtration mode. When the alum
dosage was below optimum, the filter performance in the conventional mode was not
effective enough to be placed into service, but not as completely inadequate as in the
direct filtration mode.

3. Ripening of turbidity and particle removals was more rapid when operated in
conventional mode than in direct filtration mode. The ripening behavior of the filter
was not sensitive to a further increase in alum dosage once the dosage was above
optimum. The selection of a conservative indicator for the ripening is not obvious due
to the rapid stabilization of effluent quality for both turbidity and particle
concentration. Significant breakthrough of filter performance was not observed when
the alum dosages were above optimum. However, particle concentration started to
increase earlier than turbidity when the dosages were at or above optimum, which is
in agreement with the direct filtration results suggesting the use of particle
concentration as an earlier indicator of filter breakthrough.

4. In conventional mode, the performance of the dual-media filter was still adequate but
less reliable at a higher filtration rate than at a conventional filtration rate. The
general filtration behavior was similar to that of conventional filtration rate.

OVERALL SUMMARY

Laboratory dual-media filtration experiments were conducted (a) in direct filtration mode
using a model raw water moderate in turbidity and low in dissolved organic carbon and (b) in
conventional filtration mode treating a water moderate in turbidity and high in dissolved organic

172
carbon. Effects of chemical pretreatment on the filter performance were investigated at
conventional and high filtration rates. The following conclusions are made:

1. All packed bed filtration systems are sensitive to the dosage of coagulant used in
pretreatment. The results of this research indicate, in addition, that direct filtration
systems are much more sensitive to coagulant dose than are conventional systems.
Considerable differences were observed in the filter performance depending on
whether it was operated in direct filtration or conventional filtration mode. The
response of the dual-media filter to the changes in alum dosage was more sensitive in
direct filtration mode. An increase in alum dosage above optimum resulted in slight
improvement in filtrate quality whereas suboptimum dosage significantly deteriorated
the filter performance. The filter performance at suboptimum alum dosages was
completely inadequate in direct filtration. The filtrate quality immediately reached
breakthrough and did not improve for the rest of the filter run. A modest decrease in
alum dosage from the optimum changes filter performance from effective to
ineffective. Increasing in alum dosage in direct filtration mode, while improving filter
performance, indicated some possible disadvantages including rapid development of
headless and consequently shorter filter service time. The increase in alum dosage
results in a higher loading of particles and aluminum precipitates to the media,
resulting in rapid headless development and hence a shorter filter run. Determining
the correct coagulant dosage is more important for direct filtration than for
conventional filtration.
In conventional mode, however, the effect of alum dosage on the filter performance,
while obvious, was not as dramatic as for direct filtration. The disadvantages of
overdosing in direct filtration were not witnessed in conventional filtration mode.
Higher alum dosage enabled better filter performance for a longer period of time. In
conventional mode, as the alum dosage is increased, floe formation and settlability of
floes are enhanced in the coagulation-sedimentation procedure and consequently
more particles are removed prior to filtration. The amount of particles loaded to the
filter decreases with the increase in alum dosage, resulting in a longer filter run.

173
2. The laboratory jar tests provided useful estimates for the optimum dosage for
filtration in either direct filtration or conventional filtration mode. The conventional
filtration results agreed well with the jar test results using standard coagulation and
sedimentation procedures. Jar tests with 48 hour extended flocculation were used in
determining the optimum alum dosage for direct filtration experiments and proved to
be a reasonable approximation. However, since this extended time cannot be easily
adapted to full-scale systems, and the filter performance responds fairly sensitively to
the coagulant dosage in direct filtration, an appropriate method should be carefully
established on site for determining the optimum dosage in treatment facilities
employing direct filtration.

3. In all laboratory filtration experiments conducted at Johns Hopkins, ripening of


particle removal (i.e. lower particle counts) occurred earlier than did ripening of
turbidity, and, in most cases, breakthrough of particles started earlier than
breakthrough of turbidity. If the latex particles of 4 um in diameter used in this study
are surrogates for Cryptosporidium parvum, these results suggest that turbidity can be
used as a better or more conservative monitor of filter performance during the
ripening period, while particle concentration measurements can be considered an
earlier indicator of deteriorating filter performance during the breakthrough period.

4. In dual-media filtration, the lower sand layer serves as a multiple barrier for particles
when the performance of the anthracite layer is not effective. When chemical
pretreatment is inadequate, particles that pass through the anthracite layer reach the
sand layer and can be removed due to the smaller size of the media particles. Model
simulations of clean bed efficiency also indicate that, when chemical pretreatment is
inadequate, the performance of the sand layer is still effective due to smaller grain
size and consequently higher single collector efficiency. With adequate pretreatment,
when the performance of the anthracite starts to breakthrough, the particles dislodged
from the anthracite media are effectively removed in the lower sand media.

174
CHAPTER 7: ROBUSTNESS

One of the objectives of this project was to develop and apply a filtration robustness
concept for particle removal. It was desired to be used as a practically oriented tool to plan pilot-
scale studies, interpret pilot-scale results, and monitor full-scale data. Robustness was defined as
the ability of a filtration system to provide excellent particle/pathogen removal under normal,
stable operating conditions and to minimally deviate from this performance during moderate to
severe process upsets. This chapter describes the development of a practical measure of a
treatment plant's or filter's ability to consistently maintain superior levels of particle removal (1)
during the entire filter cycle and (2) in response to in-plant and raw-water events. Various ways
of quantifying this concept of robustness were explored.

CLASSIFYING FILTER PERFORMANCE

Many studies in the literature compare filter performance under optimal (or near-optimal)
treatment conditions. Deviations from optimal operations must also be characterized so that the
relative risk of pathogen passage under various conditions can be readily assessed and the overall
risk minimized by water treatment plant staff.

A classification of filter performance is presented in Table 7.1. In this case, performance


is considered "normal" or "upset" depending on whether the filter effluent turbidity is below
some specified level, in this case 0.10 NTU (the Partnership for Safe Water Goal) for most of the
time. The quasi-steady state condition occurs during normal operation. It is during "normal
operation" that filters perform at peak levels.

Also occurring during normal performance are non-steady-state periods such as filter
ripening, hydraulic surges, and end-of-run breakthrough. Ripening and breakthrough fall under
the general classification of normal conditions because they are inherent to the filtration process.
In most operating plants, some level of hydraulic surge is unavoidable and may result in only
brief and/or minimal turbidity increases.

175
Table 7.1
Filter performance classifications

Normal Upset
Criteria Futer effluent turbidity < 0.10 NTU Filter effluent turbidity > 0.10 NTU
with no major operational changes

Operational Quasi-steady state • Suboptimal coagulation


Periods ("stable filter operation") • Suboptimal flocculation
Included Non-steady-state events such as • Changing influent water quality
ripening, end-of-run breakthrough,
and hydraulic surges

Upset or suboptimal filter performance is defined as occurring when the filter effluent
turbidities exceed 0.10 NTU most of the time. The events that typically lead to this condition
significantly impair particle attachment, detachment, or transport through the filters. Events such
as impaired coagulation (pump failures, chemical misfeeds, inadequate mixing, etc.), impaired
flocculation (insufficient residence times, flocculator shutdown, etc.), and changing water quality
can all result in suboptimal filter performance.

The filter performance classification described previously and in Table 7.1 is useful in
developing practically oriented tools to interpret data from pilot- or full-scale studies assessing
the variability of treatment performance. Robustness may be viewed as a measure of filtration
system reliability. Considerations of robustness are distinct from (maximum) filtration
efficiency, which is commonly measured only during stable filter operation. Indicators of process
robustness can incorporate water production, water quality, and treatment variability factors and
potentially integrate them into one or more key parameters that can be readily applied to daily
plant operations.

Several tools currently exist to assess filter performance data. In general, most of these
tools address either water quality or water production. A common measure of water quality is the
average turbidity over the entire filter cycle; a typical measure of water production is the amount
of water treated per unit surface area of the filter. Direct measures such as turbidity and particle
counts can be used to provide real-time water quality data. Unfortunately—although turbidity
176
and particle counts indicate general treatment efficiency—they are often weak predictors of
pathogen removal and therefore may offer limited value as a performance measure for
optimizing pathogen passage during filtration.

A number of filter performance indices, such as unit filter run volume (Trussell et al.,
1980) and filter performance index (Montgomery Consulting Engineers, 1985), have been used
to compare filter performance during stable filter operating conditions. These indices are
calculated from direct measurements and indicate either water production alone or the
relationship between water production and quality. Indices such as these can be used in real-time
optimization; however, none of the indices currently available has been evaluated for its
applicability in estimating pathogen passage through filters, and none directly addresses the. issue
of performance robustness.

Stochastic approaches such as Monte Carlo analyses are used to estimate the uncertainty
of model outputs (such as the risk of eryptosporidiosis infection) resulting from the uncertainty
of input parameters (raw water quality, method recovery, treatment efficiency, etc.). Teunis and
co-workers (1997) suggested that the uncertainty and variability in treatment removal efficiency
dominates uncertainties in other factors contributing to the risk of Cryptosporidium parvum or
Giardia lamblia infection. Similarly, neural networks can be used as a tool for expressing subtle
relationships between various water quality and operational parameters and their effects on
turbidity or oocyst removal. Neural networks have been successfully applied to the prediction of
alum dosages and effluent turbidities associated with enhanced coagulation.

ROBUSTNESS DEVELOPMENT

The primary objective of this task was to produce a practically oriented tool which could
be used to monitor and assess treatment robustness. After some initial investigation, it was
decided that neural networks and advanced statistical techniques such as Monte-Carlo analysis
were too sophisticated for easy implementation at most water treatment plants. Thus, a simpler
approach was desired.

177
Description of Concept

Robustness is defined here as the ability of a filtration system to provide excellent


particle/pathogen removal under normal operating conditions and to deviate minimally from this
performance during moderate to severe process upsets. This concept of filtration robustness is
analogous to a vehicle suspension, which is designed to provide a smooth and safe ride by
absorbing the impact from variable road conditions. Figure 7.1 graphically depicts the concept of
robustness (Coffey et al., 1998). In this figure, the horizontal axis represents influent water
quality or some other operating condition such as coagulant dose. The vertical axis represents a
measure of filter performance. As shown, the performance of a robust system does not vary
widely with changing operating conditions. Though the performance of the optimal filtration
system is greater in this hypothetical example, its optimal performance is limited to a narrow
band of operating conditions. Depending on both the relative performance differences between
the optimal and robust systems and the breadth of the operating conditions, the robust system
may be preferred.

Quantifying Robustness—Initial Investigations

Three preliminary methods for evaluating the robustness of particle passage were applied
to data collected during Task 3 at MWD's pilot plant. For the tests described here, both trains of
the pilot plant were operated with dual-media filters, using chemical dosages of 5.0 mg/L alum,
and 1.5 mg/L cationic polymer (not necessarily an optimized chemical dosage). One train of the
pilot plant used ozone as a pre-oxidant and the other train did not use any pre-oxidant. For these
data, graphs depicting a time-series analysis, percent occurrence, and normalized particle passage
all provided valuable insight into the treatment effectiveness.

Figure 7.2 shows time-series data from the pilot plant. Without ozone, more particles
passed through the filter than for the train with ozone. However, with ozone, an increasing
number of particles passed through the dual-media filter beginning after approximately one hour
of operation. Though Figure 7.2 is certainly useful—at least qualitatively—in evaluating the

178
Optimal system

(0

I
0)
Q. * v Robust system
s

-f-
Influent Water Quality or Coagulant Dose

Figure 7.1 Representation of the filtration system robustness concept

10,000

1,000-

I
8
£

Filter Operation Time (hours)

Figure 7.2 Time-series data from the MWD pilot plant comparing use of ozone as a pre-
oxidant

179
relative filter performance, other methods may provide greater insight. Figure 7.3 (a percent
occurrence graph) shows that the 90th-percentile particle occurrence—which may be a better
predictor of the public health risk than the average particle concentration—was -50 particles/mL
with preoxidation and ~80 particles/mL without preoxidation.

A third method for evaluating particle passage was to use a normalized particle passage
graph (Figure 7.4). This figure shows that the treatment with ozone passed less than half as many
total particles as the treatment without ozone. The relative number of particles passed may
provide an indication of a problem filter or treatment (assuming that the particle counters were
carefully calibrated and maintained). Figure 7.4 also indicates some potential for treatment
improvements. For example, using a filter-to-waste system for 1 hour would reduce the total
number of particles passed by ~25 percent for the treatment with ozone, but only ~15 percent for
the treatment without ozone. Thus, this type of analysis may help utilities determine where
limited financial resources should be allocated for maximum treatment improvement.

100%

10 100 1000
Particles (#/mL)

Figure 7.3 Percent occurrence for particle data from the MWD pilot plant

180
1.00

Without ozone
to
Q. 0.75 • '

CO
0_
'o
§ 0.50
t>
CO
LU

1
S 0.25 With ozone

0.00
0 6 12 18 24
Filter Operation Time (hours)

Figure 7.4 Normalized percentage of particles passing through MWD pilot plant filters (data
normalized such that 100 percent equals the total particles from the filter without ozone)

Development of a Robustness Index

The figures discussed above (Figures 7.2 to 7.4) indicate that some combined measure of
particle passage which incorporates both average quality and deviation from stable operation
may be useful. Three formulas were subsequently evaluated for use as the robustness index.
Using turbidity as the primary measurement, the robustness index was written as follows:

'90 '50
(7.1)
'50 'goal

T95 T50
7H/9595 = -2 (7.2)
-*T50 JTgoal

and

T
•* 9?_ , .
T50
(7.3)
T T
J 50 i goal

181
where:

IK/90 . TRI95 » TRI99 = turbidity robustness index using the 90th, 95th and 99th percentile,
dimensionless

T50 , T90 , T95 , T99 = 50th, 90th, 95th, and 99th percentile turbidity, NTU

Tgoai = filter turbidity goal, NTU

The "goal" term, Tgoal , represents a utility- or plant-specific performance goal for
turbidity. In addition to the turbidity robustness index, a similar equation may be used to
characterize the particle performance of the filter (PRI). In this case, particle data are used
instead of turbidity data and the particle robustness indices are labeled accordingly.

The physical significance of the robustness index may be seen by examining the two
terms of the index. The first term, the ratio of the 90th, 95th or 99th percentile to the 50th percentile
(median), represents the uniformity of the turbidity or particle performance over a specified
duration (e.g., a single filter run or a 24-hour period). The second term—the ratio of the median
turbidity/particle count to the goal—represents how well the filter is performing overall.

Using any equation above, a lower value of the robustness index indicates a treatment
process which is meeting the water quality goal with relatively low variation. For example, if the
particle counts always met a treatment goal of 25 particles/mL and did not deviate from that
goal, the value of the robustness index would be equal to 1.0. Thus, a higher robustness index
value indicates that the treatment is either not meeting its goal (on average) or the variability is
high. Note that depending on the selected treatment goal, the robustness index could be less than
unity. The minimum possible value for the robustness index is 0.5 (when the second term
approaches zero because the median performance greatly exceeded the goal.)

Robustness Index Examples

One additional limitation of the robustness index presented above is that the magnitude of
the index depends upon two terms. As a result, the index could exceed one (1.0) even when the

182
individual percentiles each were less than the goal chosen. An alternative robustness index which
would avoid this condition is:

TRIwlsoal = 1?- (7.4)


goal

where:

TRIw/goai = turbidity robustness index using the 90th percentile to goal ratio.

For this index, a specific value could be chosen (e.g., 2.0) and treatment plant operators
could attempt to maintain this level. A index value of two (2.0) would indicate that 10 percent of
the time, the turbidity exceeded the goal. For most of the examples shown below, the two-term
robustness indices were used. However, this alternative robustness index could also be used.

Preoxidation example. The pilot-plant data shown in Figure 7.2 was processed using an
equation for the 95th percentile particle robustness index (PRI95 ). The particle goal was 25
particles/mL (>2 urn). This particle goal was chosen because (1) it is measurable with reasonable
precision and accuracy; (2) it is attainable during normal operation; and (3) it is occasionally
exceeded. For both filters—with and without preoxidation—the PRI95 was 1.9 even though the
filter with preoxidation always produced fewer particles than the filter without preoxidation. The
PRI9S was equivalent in this case because the particle performance of the filter which received
preoxidized water was less uniform. Thus, these two filters achieved equivalent robustness, as
defined by the equations above.

Suboptimal coagulation example. A second example was used to evaluate robustness


using pilot-plant tests with optimal and suboptimal coagulation. For each experiment, the pilot-
plant filters usually operated for 24 hours. Thus, for a 24-hour filter run, the 90th, 95th, and 99th
percentiles capture the highest turbidity or particle data for 144, 72, and 14 minutes, respectively.
Note that the data during these periods do not need to be consecutive. For the pilot-plant
analysis, the turbidity and particle goals were 0.1 NTU and 25 particles/mL (>2 urn),
respectively.

183
The 90th or 95th percentile robustness equation may be more applicable when comparing
major differences in treatment performance or possibly when evaluating combined filter effluent
samples. The 99th percentile robustness equation may be more desirable when the objective is to
evaluate the effects of ripening or hydraulic surges because the duration of these events may not
be adequately captured within 72 min/day. The disadvantage of the second equation is that it is
heavily influenced by noise in the particle or turbidity data.

An example of a robust filter run is given in Figure 7.5 as a time series of turbidity and
particles. This preoxidized (with ozone), optimal coagulation, dual-media filter achieved a TRIg$
of 0.73 and a PRIgs of 1.1 which indicates that the filter was producing stable levels of turbidity
and particles which consistently met their prescribed goals. Figure 7.6 shows an example of a
much less robust filter run (using the same filter, pre-oxidant, and filter aid, but at decreased
alum and cationic polymer doses). The TRIgs for this run was 1.4 and the PRIys was 3.9. For this
example, the turbidity and particle values were of poorer quality and less stable.

MWD PILOT PLANT RESULTS

The 95th percentile index values (TRIgs and PRlgs) were used to compare the robustness
of the various treatments at MWD's pilot plant. Stable filter operation (baseline) and hydraulic
step experiments using filters which received preoxidized water generally produced the lowest
(best) robustness index values. The 25 percent filtration rate increase during the hydraulic step
experiments negligibly increased turbidity and particles. The highest (worst) robustness index
values were produced by the suboptimal coagulation and runoff (turbidity/TOC spike)
experiments on all filters.

184
0.25 1,000
fR/95 = 0.73
PR/95 = 1.1

- 0.20 -

2- 100 i-
1 0.15-
CO
0.

0>
0.10
LU i
10 "J
<5
Turbidity
0.05-

0.00
12 18 24

Length of Run (hours)

Figure 7.5 Time-series data for turbidity and particles from stable filter operation test at
MWD pilot plant

0.25 1,000

~ 0.20 •

100 S-
§ 0.15

&

0.10
LU
:• 10
I

0.05-

0.00
12 18 24

Length of Run (hours)

Figure 7.6 Time-series data for turbidity and particles from suboptimal coagulation test at
MWD pilot plant

185
15
-50

10

s.
a

Baseline Suboptimal Hydraulic Runoff Baseline Suboptimal Hydraulic Runoff


Coagulation Step Coagulation Step
Turbidity Particles

Figure 7.7 Box-and-whisker plot of turbidity and particle robustness indices comparing the
different types of experiments at MWD pilot plant. Plot indicates minimum, 25th percentile,
median, 75th percentile, and maximum values.

Tables 7.2 through 7.5 show the mean value and the statistical significance of the
robustness indices for several comparisons of paired data. Statistical significance was determined
using the paired Student's Mest. A significance level of five percent (a = 0.05) and a
hypothesized mean difference of zero were used to compare the data. Figure 7.7 shows the
corresponding box-and-whisker plot for the robustness indices for all of the experimental
challenges during Task 3 at MWD.

Table 7.2 compares the robustness index for the different types of experiments: baseline
(stable filter operation), suboptimal coagulation, hydraulic step, and runoff (turbidity/TOC
spike). Three groupings of data are presented for each experimental challenge: (1) all dual- and
tri-media filters (with and without preoxidation); (2) only filters that received preoxidized water;
and, (3) only filters that received water without preoxidation. The robustness indices for the
suboptimal coagulation and runoff (turbidity/TOC spike) experiments were significantly greater
than for the baseline experiments for all groupings. For the hydraulic step experiments, the

186
Table 7.2
Statistical significance of robustness index for different types of experiments at MWD pilot plant

Turbidity performance (TRIyi) Particle performance (PRIgs)


Baseline Suboptimal Hydraulic Runoff Baseline Suboptimal Hydraulic Runoff
coagulation step coagulation step
All filters Mean 0.85 1.4 0.91 2.2 1.8 3.7 1.7 8.5
(with and Different
without than Yes Yes Yes Yes No Yes
preoxidation) baseline?
Dual- and tri- Mean 0.79 1.4 0.85 1.6 1.3 2.5 1.1 5.1
media filters Different
with than Yes Yes Yes Yes No Yes
preoxidation baseline?
Dual- and tri- Mean 0.91 1.4 0.96 2.8 2.3 5.0 2.3 12
media filters Different
without than Yes No Yes Yes No Yes
preoxidation baseline?

robustness index was either not statistically different than for the baseline tests, or the magnitude
of the difference was quite small. Figure 7.7 shows these data graphically. Thus, decreasing
coagulation effectiveness by reducing both the alum and cationic polymer dosages or by
increasing turbidity and organic matter (and not changing the coagulant dosages) impaired
treatment performance. Though preoxidation could reduce the effect of the treatment challenge
(e.g., a PRIgs of 2.5 with preoxidation compared to a PRIgs of 5.0 without preoxidation), the
treatment still deteriorated substantially from baseline operations.

Table 7.3 quantifies the effect of preoxidation. In this table, the robustness index for the
train with preoxidation versus the train without preoxidation are grouped for all of the treatment
challenges. For both turbidity and particle robustness indices, adding a pre-oxidant significantly
improved treatment (reduced the particle and turbidity robustness index).

Table 7.3
Statistical significance of robustness index for effects of preoxidation at MWD pilot plant

Turbidity (TRI95) Particles (PRI9i)


With Without With Without
preoxidation preoxidation preoxidation preoxidation
Mean 1.2 1.5 2.5 5.3
Statistical difference? Yes Yes
187
Table 7.4
Statistical significance of robustness index for effects of filter aid at MWD pilot plant

Turbidity (TRI95) Particles (PRI9S)


With Without With Without
filter aid filter aid filter aid filter aid
Dual- and tri-media Mean 1.3 1.3 3.3 4.5
(with and without _ . . , ,.„ 0 XT
. , .. ,
preoxidation) Statistical difference? No No
Dual- and tri-media filters Mean 1.1 1.2 1.9 3.1
(with preoxidation) Statistical difference? No No
Dual- and tri-media filters Mean 1.6 1.4 4.7 6.0
(without preoxidation) Statistical difference? Yes Yes

Table 7.4 compares results for the experiments with and without filter aid. The three
groups of data were: (1) all filters; (2) filters receiving preoxidized water; and, (3) filters
receiving non-preoxidized water. For the first two groups, there was no effect of filter aid on
turbidity and particle robustness indices. However, without preoxidation, there was a statistically
significant (though small) treatment impairment when filter aid was used as measured by
turbidity. However, the particle robustness index indicated that filter aid substantially improved
treatment in the absence of a pre-oxidant.

Table 7.5
Statistical significance of robustness index comparing dual- and tri-media filters at
MWD pilot plant
Turbidity (TRI95) Particles (PRI95)
Dual-media Tri-media Dual-media Tri-media
All filters Mean 1.4 1.3 4.4 3.5
(with and without
preoxidation) Statistical difference? Yes No
All filters Mean 1.3 1.1 2.8 2.2
(with preoxidation) Statistical difference? Yes No
All filters Mean 1.6 1.5 5.9 4.7
(without preoxidation) Statistical difference? Yes No

188
A comparison of robustness based on media type (dual- versus tri-media) is shown in
Table 7.5. Based on turbidity, the tri-media filters improved treatment. However, based on
particle robustness, the improvement was not statistically significant. Though the mean
difference between the dual- and tri-media filters were greater for the particle robustness than for
the turbidity robustness, the variability of the particle data prevented the difference from being
statistically significant.

MWD DEMONSTRATION PLANT RESULTS

Pilot-scale testing during Task 3 at MWD indicated that treatment robustness was heavily
influenced by coagulation conditions. The demonstration plant data—collected as part of a
separate AWWA Research Foundation project (Stanley, Coffey and Rector 1998)—also
supported this result. In the pilot-scale tests, suboptimal coagulation was achieved by reducing
both the alum and cationic polymer dosages by equal amounts (typically by ~65 percent) to
achieve a turbidity goal of 0.2-0.3 NTU. For the demonstration plant testing, combinations of six
alum dosages (0, 1, 3, 5, 7, and lOmg/L) and four cationic polymer dosages (0, 1, 2.5, and
5 mg/L) were applied to one dual- and one tri-media filter at a filtration rate of 8.6 m/h
(3.5 gpm/ft2). Limited tests also evaluated the use of chlorine and ozone as pre-oxidants.

Figures 7.8 and 7.9 show the 90th percentile turbidity and particle robustness indices,
respectively, of a 37.2 m2 (400 ft2), dual-media filter operated at 8.6 m/h (3.5 gpm/ft2) with
chlorine used as a pre-oxidant. (For the demonstration-plant tests, the 90th and 95th percentile
robustness indices were similar. The 90th percentile indices were chosen for comparison with the
full-scale data below.) These data show that the use of multiple coagulants—in this case, alum
and cationic polymer—improved treatment robustness. For example, for all of the tests which
used 3 to 10 mg/L of alum and 1 to 5 mg/L of cationic polymer, the turbidity robustness index
(TRIgo) varied from 0.7 to 1.1. When operating below these coagulant conditions (as was the case
for the pilot plant suboptimal experiments), the robustness indices increased substantially.

Figures 7.10 and 7.11 show the effects of preoxidation on the turbidity and particle
robustness indices for both dual- and tri-media filters. Each pre-oxidant was tested at three
different alum doses (note that permanganate was not tested at the demonstration plant). In all of

189
6.0

O.Ol
Cationic
Polymer
Dose (mg/L)
Alum Dose (mg/L)

Figure 7.8 The effect of coagulation on turbidity robustness index at MWD's demonstration-
scale treatment plant (dual-media filters operated at 3.5 gpm/ft2; chlorine added as a preoxidant)

10.0|

x~
0>
•o

I
•o0>
••e
S.

0.0
Cationic
Polymer
Dose (mg/L)
Alum Dose (mg/L)

Figure 7.9 The effect of coagulation on particle robustness index at MWD's demonstration-
scale treatment plant (dual-media filters operated at 3.5 gpm/ft2; chlorine added as a preoxidant)

190
E Dual-media
'Tri-media

1 5 10 10
Alum Dose (mg/L)

Figure 7.10 The effect of preoxidation, coagulant dose, and media configuration on the
turbidity robustness index at MWD's demonstration-scale treatment plant (1 mg/L polymer
present)

n Dual-media
•iri-media

10 1 5 10
Alum Dose (mg/L)

Figure 7.11 The effect of preoxidation, coagulant dose, and media configuration on the particle
robustness index at MWD's demonstration-scale treatment plant (1 mg/L polymer present)

191
these tests, the cationic polymer dose was fixed at 1.0 mg/L. Treatment robustness in these tests
was influenced by (in order of decreasing magnitude) alum dosage, pre-oxidant presence, and
media configuration. Note that the suboptimal coagulation conditions for the pilot plant in
Tasks 2 and 3 (which treated a similar water quality as the demonstration plant) were 1.8 mg/L
alum and 0.53 mg/L of cationic polymer. Clearly, the pilot plant was being operated in a mode
which was highly sensitive to the coagulant dosage.

MWD FULL-SCALE RESULTS

The robustness concept was initially developed to compare the ability of various design
or operating conditions during the pilot phase (Task 3) to achieve excellent particle removal with
minimal variation. Necessarily, because of the large experimental matrix, only two coagulant
dosages were tested at the pilot-scale (optimal and suboptimal). Data collected during the pilot
testing lasted 24 hours (the length of one filter run). Testing at the demonstration plant allowed a
more in-depth evaluation of coagulant combinations and dosages. Demonstration-plant testing
established that the use of multiple coagulants may improve treatment robustness. Data shown
for the demonstration plant consisted of two complete 24 hour filter runs.

At the full-scale, control of experimental variables was obviously limited. To evaluate the
robustness concept at full-scale, therefore, on-line data from one of MWD's full-scale plants, the
F. E. Weymouth filtration plant, was collected for calendar year 1999. Both the pilot- and
demonstration-scale facilities were fed with the same source water as the full-scale plant. Figure
7.12 shows the influent turbidity and resultant alum and cationic polymer dosages used at full-
scale. The alum dosage ranged from 4.0 to 6.0 mg/L (median value of 4.5 mg/L) and the cationic
polymer dosage ranged from 1.5 to 3.5 mg/L (median value of 1.5 mg/L). These chemical
dosages were bracketed by the chemical dosages used at the pilot- and demonstration-scale.

Figure 7.13 shows the combined filter effluent turbidity and the filter effluent particles
from one of the filters (filter number 2). The Weymouth plant contains 48 filters, each with a
surface area of 177 m2 (1,900 ft2). The combined filter effluent turbidity remained low during the
entire year and ranged from 0.05 to 0.09 NTU (as a daily median value of grab samples collected
every two hours). The particles leaving filter number 2* however, were more variable and ranged

192
10.0

8.0 ••

01

Jan-99 Mar-99

Figure 1.12 Influent turbidity and coagulant dosages for full-scale testing at MWD's
F. E. Weymouth filtration plant

0.15 200

0.00
Jan-99 Mar-99

Figure 7.13 Combined filter effluent turbidity and filter effluent particle counts from a single
filter at MWD's F. E. Weymouth filtration plant

193
from less than 1 particle/mL to nearly 400 particles/mL (median value of 4 particles/mL). Filter
number 2 was representative of other filters based on a visual comparison of particle
concentrations leaving each filter. A substantial increase in the number of particles passing
through the filters was observed during August and September 1999. The alum and polymer
dosages were increased during September in an attempt to reduce the filter effluent particles.
Note that the combined filter effluent turbidity remained low during the entire duration of the
elevated particle event.

The particle robustness index (P/Z/po) was calculated for two filters (filters 2 and 24) and
is shown in Figure 7.14. Note that these two filters performed similarly. The robustness index
varied from 0.6 to ~6 (median value of 0.9) for the entire year. This value fell within the same
range as the pilot- and demonstration-scale tests and proved useful in identifying potential
treatment deficiencies. The source of the increased particles and robustness, however, could not
be definitively established. No significant correlation to influent water quality or operating
conditions was observed. A weak correlation was observed with water temperature because the
increased passage of particles occurred during August and September. One hypothesis is that a
diatom bloom in the source water may have resulted in increased particle passage during these
months.

8.0
•Filter 24
-Filter 2

6.0..
X
o>
13
c

£ 4.0 •'

J3
O
cr

£ 2.0
a.

0.0 M-1 -4-"-


Jan-99 Mar-99 May-99 Jul-99 Sep-99 Nov-99 Jan-00
Date

Figure 7.14 Particle robustness index (PRIgd) for two filters at MWD's F. E. Weymouth
filtration plant
194
20
Filter 24
-Filter 2

8[15

10 +

en
-
uo o«;
r
Possible robustness index goal

0-1-
Jan-99 Mar-99 Nov-99 Jan-00

Figure 7.15 Modified particle robustness index (PRIgo/goai) for two filters at MWD's F. E.
Weymouth filtration plant.

The modified robustness index (PRI90/goal) was also used to evaluate the full-scale data.
Figure 7.15 shows this modified index for filters 2 and 24 (the same filters as shown in Figure
7.14). This figure shows a similar trend as the PRI90, but the departure from normal performance
is magnified. Figure 7.15 also shows a possible goal for the modified robustness index. The
physical significance of the index goal of 2.0 is that the 90th percentile particle concentration
was double the goal. This simplified index may be more useful for plant operators who desire to
modify plant performance to achieve the goal.

SUMMARY

The robustness indices presented here represent a first attempt to develop an index which
accounts both for average treatment performance and variability. It was desired to develop an
index, which could maintain physical significance, could be easily understood, and could be
implemented at a water treatment plant with minimal sophistication.
195
The index was successfully applied to pilot-, demonstration-, and full-scale treatment.
Within the parameters tested during these experiments, the following observations were made
using the robustness index and standard statistical techniques:

• Chemical coagulation conditions controlled treatment performance, followed by the


presence of a pre-oxidant, filter media configurations, and the presence of filter aid.

• The presence of dual coagulants improved the robustness of particle removal


compared to single coagulants.

Because the robustness approach deliberately valued simplicity, it suffered from some
limitations. Some of these limitations are:

• Values of the robustness index were heavily influenced by the treatment goal term.
Selecting a treatment goal which is too high minimizes the variability term in the
index.

• The robustness index can be less than one. This effect results once again from setting
a treatment goal substantially above normal operations.

• The relative weighting of the two components (average performance and variability)
is equal.

The simplified robustness index which incorporated the ratio of the 90th percentile (or
some other chosen percentile) to the goal represents an alternative approach which may be more
easily implemented at full-scale.

196
CHAPTER 8: CONCLUSIONS

This chapter presents the key conclusions that can be drawn from this investigation of
filter operational effects on pathogen passage. A significant proportion of the work was
conducted on benchmark systems at pilot-scale and involved evaluating the removals of seeded
inactivated Cryptosporidium parvum oocysts and several surrogates. Additional pilot-scale
investigations examined ways of mitigating filter operation effects. Bench-scale investigations
contributed additional understanding and project results were evaluated in terms of filtration
robustness.

OPTIMIZED PATHOGEN REMOVAL

Under optimal operating conditions, there was a 2-log difference in C. parvum removal
between two pilot plants operated to produce similar effluent turbidity (<0.1 NTU) and particles
(< 20/mL). The pilot plant in Ottawa removed approximately 5-logs while that at MWD removed
approximately 3-logs. The reasons for this difference are not understood, although there were
major differences in coagulation regimes and some differences in pilot plant characteristics.
Ottawa used a relatively high alum dose for both particle and color removal, while MWD used a
lower dose. Further investigations should attempt to determine the reasons for the substantial and
reproducible differences in observed Cryptosporidium removals.

At Ottawa, the excellent removals were also achieved at temperatures as low as 1°C.

BREAKTHROUGH PHASE

The breakthrough phase at the end of the filter cycle represents a vulnerable period for
pathogen passage. A substantial deterioration of oocyst removal capability was observed even in
the early stages of breakthrough, when filter effluent particle counts had just begun to rise. At
this time turbidity had not always increased.

197
Because changes in turbidity may occur only after substantial deterioration of C. parvum
removal, turbidity values may not be an adequate predictor of increasing oocyst passage. Particle
counts may provide a more sensitive indication of pathogen passage, but still fall short of
predicting the full extent of the deterioration of C. parvum removal at the beginning of
breakthrough.

These results indicate that breakthrough, as measured by turbidity or perhaps even


particles, should be avoided. This may mean that filter runs should be terminated even earlier
than has been the practice.

In terms of explaining the observed deterioration in C. parvum removal, oocyst passage


through the filter (not detachment of previously captured oocysts) appears to be an important
mechanism, however, the experimental design tested passage, not detachment. It is
recommended that further investigations be conducted to assess the importance of detachment
for pathogen passage during breakthrough.

SUBOPTIMAL COAGULATION

Suboptimal coagulation substantially deteriorated pathogen removal, even at turbidities


which were below 0.3 NTU. This finding is of major significance for water utilities.

Although turbidity may be an adequate indicator of poor coagulation performance,


particle counts provide a more sensitive indication.

As expected, the complete absence of coagulant resulted in almost no C. parvum


removals. The lack of coagulant over several hours gave marked deterioration, although some
removal capability remained.

HYDRAULIC STEP

The testing showed that changes in hydraulic loading did not deteriorate C. parvum
removal, with the exception of one experiment where turbidity, particles (to some extent), and C.
parvum all increased substantially. Although it is possible that the timing of the hydraulic step
198
within the filter cycle may influence the release of previously collected particles, the experiments
were not designed to assess this.

Possibly, turbidity may be an adequate monitoring parameter for measuring significant


hydraulic events. Hydraulic step events should be investigated further to attempt to define
conditions under which they may or may not be significant in terms of pathogen passage.

RIPENING

Minimal or moderate deterioration of pathogen removal, compared to suboptimal


coagulation or breakthrough, may result during filter ripening. However, if breakthrough is
avoided, ripening becomes a more significant part of the filter cycle for pathogen passage. In
addition, in filters with longer ripening periods and/or for mono-media filters, oocyst passage
during ripening may be more important than observed in the dual-media filters evaluated in this
research. In full-scale plants, oocysts present in backwash remnant water might also contribute to
increased oocyst passage through a filter during ripening. This was not evaluated in the present
study.

PREOXIDATION

Experiments conducted as part of this investigation showed that preoxidation may


substantially improve particulate control. Preoxidation was found to partially mitigate the effects
of suboptimal coagulation.

The choice of oxidant (chlorine, ozone or permanganate) may be less important than
presence of any pre-oxidant. However, of these oxidants, only ozone is likely to provide any
disinfection of Cryptosporidium.

ROBUSTNESS

A simple robustness index, incorporating both average quality and some measure of
deviation from stable operation, was developed in this project. This index was found useful for

199
comparing treatment conditions, but its utility is quite dependent on the strictness of the
treatment goal chosen.

The presence of pre-oxidation made a greater contribution to robustness than the use of a
filter aid.

BENCH-SCALE STUDIES

The bench-scale investigations also demonstrated sensitivity to coagulant dose, with


direct filtration systems being much more sensitive than conventional systems.

In all bench-scale experiments conducted, ripening of particles occurred earlier than that
of turbidity. The bench-scale results suggested that turbidity might be a better or more
conservative monitor of filter performance during the ripening period. Note that these differences
were not always observed at pilot-scale.

The bench-scale results also demonstrated that in dual-media filters, the lower sand layer
serves as a second barrier for particles when the performance of the anthracite layer is not
effective.

200
CHAPTER 9: REGULATORY IMPLICATIONS

This chapter provides comments that are intended to be helpful to those reviewing this
report to gain information that might be useful in regulation development.

OPTIMIZED FILTRATION

This study clearly showed that optimized filtration is plant- and source water-specific in
terms of the performance achievable for Cryptosporidium parvum removal. Maximizing the
removal of turbidity through filtration will maximize C. parvum reduction. Avoiding high
turbidity as a result of operating events will reduce C. parvum passage.

CREDIT FOR REMOVALS ABOVE A CERTAIN MINIMUM LEVEL

This is likely to be a more important issue for Cryptosporidium parvum than for Giardia
lamblia, because of the difficulty in C. parvum inactivation. Credit for removals above a certain
level must of course be demonstrated. This investigation has demonstrated that removals can be
highly variable within a filter cycle. Filter robustness therefore becomes important. Plants
employing coagulation to remove organics may be more sensitive to coagulant dosage than those
practicing only paniculate removal. The different results obtained at the two pilot plants in this
study have shown that simply optimizing a filter for turbidity removal (<0.1 NTU) may not
necessarily optimize or maximize C. parvum removal. It may therefore be inappropriate to set
single log removal values for all conventional or direct filtration processes.

INDIVIDUAL FILTERS

Data from this project support individual filter monitoring (turbidity and particles).
Whether or not a 0.3 NTU turbidity standard should be applied to individual filters could not be
answered by this study. The results from this study do however indicate that a given turbidity

201
value in the breakthrough phase may indicate a greater potential for pathogen passage than the
same value in the ripening stage.

PRE-OXIDATION

Data from this project indicate that the removal of pre-oxidation should be carefully
considered for its effect on filter particulate removal performance.

202
CHAPTER 10: GUIDANCE TO UTILITIES

This chapter presents practical suggestions to aid utilities in mitigating the effects of filter
operation effects on pathogen passage.

SUBOPTIMAL COAGULATION

1. Source water quality changes may have the same effect on filter performance as
suboptimal coagulation.
2. Utilities should carefully consider the effects of reduced coagulant dosages. Jar tests
and ideally pilot filter investigations should be performed to assess impacts on filter
performance and robustness prior to any full-scale implementation.
3. Multiple coagulants may improve robustness; redundant (on-line) chemical feed
systems add reliability.
4. The benefits of a pre-oxidant for mitigating potential suboptimal coagulation
incidents should be considered.
5. If a pre-oxidant is in use, a utility should proceed cautiously with removing it.

BREAKTHROUGH PHASE

Given the demonstrated importance of the breakthrough phase for pathogen passage in
this investigation, it is recommended that utilities:

1. Consider using particle counters on selected filters to monitor early breakthrough.


2. Specify maximum headless and maximum filter run time to wash before
breakthrough.
3. Weigh power and labor savings of regularly scheduled backwash against water
quality.
4. Stagger backwashes based on time.
5. Consider monitoring within the filter bed to provide early warning of breakthrough.
(Additional work would be required to demonstrate the feasibility and usefulness of
this approach at full-scale.)

203
6. Consider operator training and manager education (within the context of global
improvement).

Breakthrough mitigation is likely to be highly plant-specific.

RIPENING

To mitigate effects of ripening it is recommended that utilities:

1. Carefully adjust coagulants to minimize spiking (i.e., watch out for suboptimal
operation).
2. Use a turbidimeter as a conservative measure of performance during ripening.
3. Consider alternate media designs (perhaps dual-media) that may minimize ripening.
4. Consider adding coagulants to backwash (or filter influent).

It is noted that ripening is essentially unavoidable, absent filter-to-waste. Utilities may


therefore wish to install filter-to-waste capability, if this is not already present.

Although not investigated in this study, discussion within the project team identified two
possible additional strategies for mitigating ripening that utilities might consider. These were to
consider resting filters following backwashing and to ensure that backwashing is optimized.
Although cleaning must be sufficient to avoid operational problems such as mudballs, it has been
hypothesized that cleaning that is too thorough may impair performance during the ripening
period.

HYDRAULIC STEP

To mitigate the possible effects of hydraulic steps (sudden increases in flow) it is


recommended that utilities:

1. Evaluate the effect of backwashing on the occurrence of hydraulic spikes.


2. Consider the trade-off between starting off-line filters and imposing a hydraulic
increase on on-line filters.

204
3. Control plant influent flows as much as possible to minimize hydraulic changes to the
filters.

Table 10.1 summarizes the project team's qualitative assessment of the relative
importance of various operational effects on pathogen passage. (The Project Advisory
Committee also contributed to this evaluation.) The range of values reflects the varied experience
of the participants. However, there was general agreement that suboptimal coagulation posed the
greatest threat or challenge. Breakthrough was next in importance, although with proper
operation it can be largely avoided. Ripening and hydraulic step were considered less important,
with the latter drawing the most variety of opinion.

Table 10.1
Qualitative importance of filter operational effects1

Operation 1 2 3 4 5
Suboptimal coagulation XX xxxx
xxxx
Ripening X xxxx XXX
XX
Hydraulic step XXX XX XXX XX
Breakthrough X xxxx xxxx
X
Scale of 1 to 5, with 5 being highest

205
APPENDIX A

Recovery Data - Cryptosporidium parvum and Giardia lamblia

207
Table A.I
C. parvum recovery data from MWD
Location Sample Volume Seeded Concentration Measured Concentration Recovery
(mL) (oocysts/L) (oocysts/L) (%)
Filter influent 10 2.0EK)5 5.5E404 28
6.1E-KM 31
4.8E+04 24
5.6E+04 28
4.2E-+04 21
4.0E+04 20
5.4E+04 27
10 l.OE+04 1.6E+03 16
2.3E403 23
2.6E-KB 26
10 2.0E405 5.7E+04 28
5.8E404 29
5.2E404 26
7.8E-KM 39
7.2E404 36
7.5E+04 38
4.6E+04 23
10 1.1E404 2.7E-KB 25
2.5E+03 23
4.4E-H)3 40

AVERAGE 27
SW. DEV. 6

Filter effluent 500 2.1E-H)2 6.4E+01 31


7.0E+01 34
6.0E-K)! 29
7.6E-K)! 37
6.4E401 31
7.2E+01 35
5.2E401 25
500 2.0E-+02 8.4E+01 41
8.8E401 43
8.8E+01 43
5.0E-K)! 25
9.0E401 44
1.1E+02 54
9.6E-H)! 47

AVERAGE 37
SW. DEV. 9

208
Table A.2
Giardia recovery data from MWD

Location Sample Volume Seeded Concentration Measured Concentration Recovery


(mL) (cysts/L) (cysts/L) (%)
Filter influent 10 2.1E+05 l.OE+05 50
1.2E+05 58
6.9E+04 33
l.OE+05 49
7.7E+04 37
5.8E+04 28
9.2E+04 44
10 l.OE+04 4.3E+03 42
4.6E+03 45
3.3E+03 32
10 2.0E+05 9.8E+04 50
l.OE+05 51
1.1E+05 58
1.2E+05 60
1.3E+05 ' 64
1.3E+05 66
9.8E+04 50
10 9.8E+03 5.3E+03 54
5.4E+03 55
7.0E+03 71

AVERAGE 50
STD. DEV. 12

Filter effluent 500 2.1E+02 9.0E+01 44


7.8E+01 38
7.4E+01 36
l.OE+02 50
8.8E+01 43
1.3E+02 63
1.7E+02 83
500 2.0E+02 1.1E+02 55
1.8E+02 90
1.3E+02 65
8.8E+01 43
1.4E+02 70
1.5E+02 73
1.2E+02 61

AVERAGE 5*
STD. DEV. 17

209
Table A.3
C. parvum recovery data from the University of Waterloo - Ottawa water

Location Sample Volume Seeded Concentration Measured Concentration Recovery


(mL) (oocysts/L) (oocysts/L) (%)
Filter influent 100 l.OE+06 7.0E+05 70
7.2E+05 72
8.5E+05 85
8.2E+05 82
7.1E+05 71
10 l.OE+06 7.0E+05 70
8.0E+05 80
5.9E+05 59
7.6E+05 76
8.0E+05 80

AVERAGE 74

Filter effluent 500 l.OE+03 7.1E+02 71


7.3E+02 73
8.6E+02 86
7.6E+02 76
6.9E+02 69
8.4E+02 84
6.0E+02 60
5.9E+02 59
7.5E+02 75
8.0E+02 80
500 l.OE+02 7.2E+01 72
8.4E+01 84
7.0E+01 70
7.6E+01 76
7.4E+01 74
6.4E+01 64
7.8E+01 78
7.6E+01 76
8.2E+01 82
5.6E+01 . 56

AVERAGE 73

210
Table A. 4
C. parvum recovery data from the University of Waterloo - MWD water

Location Sample Volume Seeded Concentration Measured Concentration Recovery


__________(mL)_______(oocysts/L)_______(oocysts/L)____
Filter influent 1 l.OE+06 7.8E+05 78
6.7E+05 67
5.7E+05 57
8.1E+05 81
7.0E+05 70
AVERAGE 71

Filter effluent 500 l.OE+03 7.0E+02 70


6.5E+02 65
8.0E+02 80
8.3E+02 83
7.2E+02 72
500 l.OE+02 5.0E+01 50
8.4E+01 84
5.6E+01 56
7.6E+01 76

AVERAGE 71

211
APPENDIX B

Detailed Task 2 Data and Tables

(Examining Benchmark Systems)

213
Table B. 1
Experimental schedule at Ottawa pilot plant

Day Date Test Condition C. parvum B. subtilis Comments


T 7/21/98 Lab Recovery Experiment y n UW Pilot plant water.
R 7/23/98 Lab Recovery Experiment y n UW Pilot plant water.
W 7/29/98 Lab Recovery Experiment y n UW Pilot plant water.
R 7/32/98 Lab Recovery Experiment y n UW Pilot plant water.

R 8/6/98 Stable Filter Operation (Shakedown #1) y n Seed: Cryptosporidium only.

W 9/9/98 Stable Filter Operation y n Bacillus method being validated - incubator problem.
W 9/23/98 Stable Filter Operation y n Bacillus method being validated - incubator problem.

T 10/6/98 Stable Filter Operation y y


W 10/21/98 Ripening y y Pump failure at start of experiment.
T 10/27/98 Ripening y y
T 11/3/98 Ripening y y
T 1 1/10/98 Ripening y y
W 11/25/98 Breakthrough y y Rapid increase in turbidity - sampling at 0.8 NTU.

W 12/9/98 Breakthrough y y Breakthrough occurred at 2 am - sampling at 0.5 NTU.

W 1/13/99 Breakthrough y y
R 1/21/99 Onset of Breakthrough y y
T 2/9/99 No Coagulants in Plant, No Coagulants in lar y y
R 2/18/99 Sub-optimal Coagulation y y Spore pump failure for a few minutes at T = 30.

T 3/9/99 Stable Filter Operation y y


T 3/23/99 Sub-optimal Coagulation y y
T 3/30/99 — — — Spring runoff until end of April.

R 4/8/99 Stable Filter Operation During Runoff y n


T 4/13/99 Stable Filter Operation During Runoff y n
T 4/20/99 . ... ... Spring runoff
T 4/27/99 — — ... Spring runoff

T 5/4/99 Sub-optimal Coagulation y y


T 5/11/99 Lab Recovery Experiment y n Ottawa pilot plant water.
W 5/12/99 Lab Recovery Experiment y n Ottawa pilot plant water.
T 5/18/99 Lab Recovery Experiment y n Ottawa pilot plant water.
R 5/20/99 Lab Recovery Experiment y n Ottawa pilot plant water.
M 5/31/99 Stable Filter Operation y y
M 6/7/99 Hydraulic step y y Seeded for 5 hours prior to hydraulic step.
M 6/14/99 Lab Recovery Experiment y n MWD pilot plant water.
T 6/15/99 Hydraulic step y y Seeded for 5 hours prior to hydraulic step.
T 6/22/99 Hydraulic step y y Seeded for 5 hours prior to hydraulic step.
T 6/29/99 No Coagulants in Plant, Coagulants in Jar y y
T 7/13/99 No Coagulants in Plant, Coagulants in Jar y y
T 7/20/99 Stable - Seeded at Rapid Mix y y
T 7/27/99 Stable Filter Operation y y
W 8/4/99 No Coagulants in Plant, No Coagulants in Jar y y
W 8/18/99 Stable Filter Operation, No Coagulans in Jar y y
T 8/24/99 No Coagulants, No Media y y

M 12/13/99 No Coagulants Since Backwash y y


F 12/17/99 No Silicate in Plant/Jar y y
M 12/20/99 Onset of Breakthrough y y
W 12/22/99 Onset of Breakthrough y V

214
Table B. 2
Ottawa pilot plant water temperature and filter headloss, turbidity, and particle counts
Date Type of Run Raw Water Settled Water Rate of Headloss Turbidity Particle Counts >= 2 mm (#/mL)
Temperature Temperature (AfVhr) Median Median Log Removals Median Median Log Removals
PI FE Median 95th Percentile PI FE Median 95th Percentile
8/6/98 Shakedown (Stable) 23.5 23.5 0.085 1.04 0.02 1.7 1.7 4896 2.3 3.3 3.5
9/9/98 Stable Operation 19.8 20.1 0.092 1.08 0.02 1.7 1.7 5043 0.9 3.8 3.9
9/23/98 " *
17.9 16.9 1.17 0.03 1.6 1.7 5324 7.8 2.8 3.0
10/6/98 "
14.5 14.5 0.059 2.14 0.02 2.0 2.0 7251 0.2 4.5 4.8
3/9/99 "
3.3 2.3 0.095 2.35 0,03 1.9 1.9 4533 0.3 4.2 4.3
5/31/99 "
21.0 18.7 0.044 2.54 0.03 1.9 1.9 5665 0.9 3.8 3.9
7/27/99 "
25.6 24.7 0.044 1.00 0.04 1.4 1.8 4427 4.8 3.0 3.3
1/19/00 " *
2.7 1.6 0.086 0.03 * * * * *
4.1
8/18/99 Stable Operation, No Coagulants in Jar 21.8 21.0 0.051 1.07 0.03 1.6 1.6 5110 0.4 4.1 4.3
7/20/99 Stable Operation - seeded at Rapid Mix 23.8 22.7 0.051 1.39 0.08 1.3 1.5 4958 5.5 3.0 3.0
10/27/98 Filter Ripening 11.6 12.0 0.067 2.42 0.07 1.6 1.7
" 6691 2.7 3.4 3.7
1 1/3/98 8.4 8.8 0.078 5.56 0.14 1.6 1.8 12695 10.6 3.1 3.6
11/10/98 "
9.0 9.4 0.044 1.93 0.14 1.1 1.2 5029 7.3 2.8 3.3
1 1/25/98 Breakthrough 6.1 6.4 0.062 2.97 0.98 0.5 0.6
" 7695 247.4 1.5 1.7
12/9/98 5.3 5.6 0.044 2.37 0.51 0.7 0.7
" 6537 136.2 1.7 1.8
1/13/99 1.3 3.7 0.083 • 1.52 0.55 0.4 0.6 3905 4.2 3.0 3.1
1/21/99 Onset-of-breakthrough 1.2 3.7 0.096 1.47 0.05 1.5 1.5 3723 1.3 3.5 3.8
12/20/99 n
2.3 3.3 0.127 4.53 0.26 1.1 1.2 6252 9 2.8 2.9
12/22/99 "
1.7 2.1 0.098 4.32 0.25 1.2 * * * *
1.6
12/17/99 No Silicate in Plant/Jar 3.5 4.1 0.057 4.58 0.03 2.2 2.2 12485 3.3 3.6 3.6
2/9/99 No Coagulants in Plant/Jar 2.2 2.8 0.111 2.52 0.74 0.5 0.6
" 7416 185.9 1.6 1.7
8/4/99 24.4 23.6 0.019 1.28 0.55 0.4 0.4 3967 554.3 0.9 0.9
6/29/99 No Coagulants in Plant, Coagulants in Jar 23.7 22.7 0.014 1.66 0.74 0.4 0.4 5725 432.1 1.1 1.2
7/13/99 "
21.8 21.1 0.021 1.77 0.68 0.4 0.5 6361 499.6 1.1 1.2
12/13/99 No Coagulants Since Backwash 3.7 6.1 0.013 4.11 2.27 0.3 0.3 10271 5117.0 0.3 0.3
8/24/99 No Coagulants in Plant/Jar, No Media * * * * * * * * * * *

2/18/99 Suboptimal Coagulation 2.6 2.1 0.150 2.05 0.98 0.3 0.3
11 3583 583.5 0.8 0.8
3/23/99 1.4 3.8 0.113 2.71 0.90 0.5 0.5 3740 70.9 1.7 1.7
5/4/99 «
11.5 11.6 0.139 2.54 0.15 1.2 1.3 5829 19.5 2.5 2.5
4/8/99 Runoff 2.2 3.1 0.069 8.53 0.04 2.4 2.4
" 14185 1.2 4.1 4.2
4/13/99 4.3 3.8 0.059 13.22 0.03 2.6 2.6 15882 2.2 3.8 4.0
677/99 Hydraulic Step 22.0 20.0 0.051 2.09 0.09 1.4 1.5
" 5853 0.2 4.4 4.7
6/15/99 21.2 20.2 0.055 1.81 0.06 1.5 1.6
" 5740 1.4 3.6 3.8
6/22/99 * * * * * * * * * * *

Data are not available.


Table B. 3
Ottawa pilot plant filter removal of turbidity and particles
Date Type Seed Length of Turbidity (NTU) Particles (#/mL)
of Duration Run PI FI FE Log >=2 |im
Experiment at Sampling at Sampling Removal PI FE Log
(min) (h) (PI to FE) Removal
8/6/98 Shakedown 15 1.04 0.02 1.72 4966 1.9 3.42
(Stable Filter Operation) 30 1.04 0.02 1.72 4949 8.0 2.79
45 .04 0.02 1.72 4868 2.7 3.26
55 .04 0.02 1.72 4915 2.1 3.37

9/9/98 Stable Filter Operation 15 .10 0.02 1.74 5031 0.9 3.75
30 .08 0.02 1.73 5024 0.9 3.75
45 .08 0.02 1.73 5026 0.6 3.92
55 .08 0.02 1.73 5008 1.0 3.70

9/23/98 Stable Filter Operation 15 .17 0.04 1.47 5317 17 2.50


30 .17 0.03 1.59 5329 7.5 2.85
45 1.17 0.03 1.59 5255 5.6 2.97
55 1.17 0.03 1.59 5225 4.8 3.04

10/6/98 Stable Filter Operation 15 2.14 0.02 2.03 7382 0.2 4.57
30 2.14 0.02 2.03 8381 0.1 4.92
45 2.19 0.02 2.04 7258 0.2 4.56
55 2.18 0.02 2.04 7021 0.2 4.55

3/9/99 Stable Filter Operation 15 2.35 0.03 1.89 4818 0.3 4.21
30 2.35 0.03 1.89 4583 0.4 4.06
45 2.35 0.03 1.89 4278 0.3 4.15
55 2.35 0.03 1.89 4291 0.4 4.03
5/31/99 Stable Filter Operation 15 2.59 0.04 1.81 5757 0.8 3.86
30 2.54 0.03 1.93 5548 0.9 3.79
45 2.54 0.03 .93 5543 1.0 3.74
55 2.53 0.03 .93 5662 2.0 3.45

7/27/99 Stable Filter Operation 15 1.00 0.05 .30 4472 7.3 2.79
30 1.00 0.04 .40 4325 4.6 2.97
45 1.35 0.04 .53 8342 4.1 3.31
55 1.24 0.04 .49 4796 4.3 3.05
1/19/00 Stable Filter Operation 15 . 0.02 . . 4.4 _
30 - 0.03 - - 5.3 -
45 . 0.03 . - 5.3 -
55 - 0.03 - - 4.3 -

8/18/99 Stable Filter Operation 15 1.07 0.03 1.55 5093 0.4 4.10
no coagulants in jar 30 .07 0.03 1.55 5110 0.4 4.11
45 .07 0.03 1.55 5174 0.4 4.11
55 .07 0.03 1.55 4999 0.4 4.10
7/20/99 Stable Filter Operation 150 .40 0.08 .24 4973 5.4 2.96
seeded at rapid mix 165 .39 0.07 .30 4801 4.4 3.04
180 .39 0.06 .36 4983 4.9 3.01
195 .39 0.06 .36 4861 5.1 2.98
210 .39 0.05 .44 4846 4.6 3.02
225 .39 0.05 .44 5162 5.0 3.01
240 .37 0.05 .44 4898 5.6 2.94
(continued)
216
Table B.3 (continued)
Ottawa pilot plant filter removal of turbidity and particles

Date Type Seed Length of Turbidity (NTU) Particles (#/mL)


of Duration Run PI FI FE Log >=2 urn
Experiment at Sampling at Sampling Removal PI FE Log
(min) (h) (PI to FE) Removal
255 1.36 0.05 1.43 4968 6.9 2.86
270 1.34 0.05 1.43 5093 9.2 2.74

8/24/99 No Coagulants, No Media 15 _ _ . _ _ _


no coagulants in plant or jar 30 - - - - - -
45 - - - - - -
55 - - - - - -

12/13/99 No Coagulants, with Media 15 4.07 2.27 0.25 1.1E+0 4815 0.35
4
no coagulants in plant or jar 30 4.07 2.27 0.25 l.OE+0 5172 0.30
4
no coagulants since 45 4.07 2.27 0.25 1.1E+0 5250 0.31
backwash 4
55 4.24 2.27 0.27 l.OE+0 5452 0.28
4

2/9/99 No Coagulants, with Media 15 2.61 0.69 0.58 7748 143 1.73
no coagulants in plant or jar 30 2.52 0.74 0.53 7336 189 1.59
45 2.45 0.79 0.49 7218 237 1.48
55 2.41 0.83 0.46 6862 254 1.43

8/4/99 No Coagulants, with Media 15 .28 0.58 0.34 3978 521 0.88
no coagulants in plant or jar 30 .28 0.55 0.37 3941 555 0.85
45 .28 0.55 0.37 3940 586 0.83
55 .24 0.55 0.35 3932 599 0.82

6/29/99 No Coagulants, with Media 15 .66 0.72 0.36 5645 403 .15
no coagulants in plant 30 .66 0.74 0.35 5643 438 .11
coagulants in jar 45 1.66 0.75 0.35 5906 447 .12
55 1.64 0.76 0.33 5736 462 .09

7/13/99 No Coagulants, with Media 15 1.78 0.65 0.44 6424 453 .15
no coagulants in plant 30 1.77 0.68 0.42 6378 500 .11
coagulants in jar 45 1.75 0.69 0.40 6265 505 .09
55 1.75 0.72 0.39 6281 536 .07

12/17/99 No Silica, with Media 15 4.53 0.03 2.18 1.2E+0 3.3 3.58
4
no silicate in plant or jar 30 4.58 0.03 2.18 1.3E+0 2.8 3.65
4
no coagulants since 45 4.92 0.03 2.21 1.3E+0 3.6 3.56
backwash 4
55 4.79 0.03 2.20 1.3E+0 4.6 3.44
4

2/18/99 Suboptimal Coagulation 15 2.05 0.98 0.32 3631 589 0.79


(52% reduction to 20 mg/L) 30 2.05 0.98 0.32 3515 557 0.80
45 2.03 0.98 0.32 3503 595 0.77
55 2.03 0.98 0.32 3479 612 0.75
(continued)

217
Table B.3 (continued)
Ottawa pilot plant filter removal of turbidity and particles

Date Type Seed Length of Turbidity (NTU) Particles (#/mL)


of Duration Run PI FI FE Log >=2|im
Experiment at Sampling at Sampling Removal PI FE Log
(min) (h) (PI to FE) Removal
3/23/99 Suboptimal Coagulation 15 2.71 0.89 0.48 3776 66 1.76
(47% reduction to 20 mg/L) 30 2.71 0.90 0.48 3876 71 1.74
45 2.71 0.95 0.46 3815 72 1.72
55 2.79 0.98 0.45 3799 75 1.71

5/4/99 Suboptimal Coagulation 15 2.54 0.13 1.29 5818 20 2.47


(49% reduction to 22 mg/L) 30 2.54 0.14 1.26 5793 19 2.48
45 2.54 0.17 1.17 5774 22 2.41
55 2.54 0.19 1.13 6007 24 2.40

10/27/98 Filter Ripening 5 2.42 0.09 1.43 6663 31 2.34


10 2.42 0^08 1.48 6566 6.4 3.01
15 2.43 0.07 1.54 6646 3.5 3.28
20 2.42 0.05 1.68 6590 2.3 3.46
25 2.42 0.06 1.61 6836 2.1 3.51
11/3/98 Filter Ripening 5 5.61 0.32 1.24 12739 58 2.34
10 5.59 0.19 1.47 12808 26 2.69
15 5.56 0.14 1.60 12702 12 3.04
20 5.56 0.09 1.79 12383 6.2 3.30
25 5.52 0.10 1.74 12757 4.1 3.49
11/10/98 Filter Ripening 5 1.93 0.19 1.01 4989 145 1.54
10 1.93 0.22 0.94 5142 18 2.46
15 1.93 0.13 1.17 5081 4.8 3.02
20 1.92 0.13 1.17 4960 2.7 3.26
25 1.92 0.11 1.24 4864 7.7 2.80
1/21/99 Onset-of-breakthrough 15 1.47 0.05 1.47 3746 1.1 3.53
30 1.47 0.05 1.47 3788 1.3 3.46
45 1.47 0.06 1.39 3641 2.3 3.20
55 1.47 0.06 1.39 3669 2.5 3.17
12/20/99 Onset-of-breakthrough 15 4.48 0.26 .23 6216 8.7 2.86
30 4.57 0.26 .24 6115 9.1 2.83
45 4.48 0.26 .23 6128 25 2.38
55 4.45 0.32 .14 6243 84 1.87
12/22/99 Onset-of-breakthrough 15 4.27 0.25 .23 _ . _
30 4.32 0.30 .16 . - .
45 4.32 0.30 1.16 . - .
55 4.31 0.30 1.16 - - -

11/25/98 Breakthrough 15 2.97 0.80 0.57 7726 147.3 1.72


30 2.97 0.98 0.48 7744 311.2 1.40
45 3.08 0.98 0.50 8097 294.6 1.44
55 3.08 0.98 0.50 8216 242.7 1.53
12/9/98 Breakthrough 15 2.38 0.50 0.68 6419 147.2 1.64
30 2.38 0.47 0.70 6387 131.2 1.69
45 2.36 0.48 0.69 6334 123.6 1.71
55 2.35 0.55 0.63 6233 158.1 1.60
(continued)

218
Table B.3 (continued)
Ottawa pilot plant filter removal of turbidity and particles

Date Type Seed Length of Turbidity (NTU) Particles (#/mL)


of Duration Run PI FI FE Log >=2 \im
Experiment at Sampling at Sampling Removal PI FE Log
(min) (h) (PI to FE) Removal
1/13/99 Breakthrough 15 1.52 0.65 0.37 3891 3.3 3.07
30 1.52 0.40 0.58 3999 3.7 3.03
45 1.52 0.62 0.39 3829 3.6 3.03
55 1.52 0.76 0.30 3860 4.3 2.95

4/8/99 Runoff 15 8.52 0 2.33 14258 1.5 3.98


30 8.61 0 2.46 14097 1.5 3.97
45 8.56 0 2.33 14591 1.1 4.12
55 8.52 0 2.33 14628 1.1 4.12

4/13/99 Runoff 15 13.2 0.03 2.64 15836 2.2 3.86


0
30 13.2 0.03 2.64 15865 2.3 3.84
3
45 13.2 0.03 2.64 15905 2.1 3.88
2
55 13.2 0.03 2.64 15867 1.6 4.00
2

6/7/99 Hydraulic Step 280 2.10 0.06 1.54 5816 0.5 4.07
(25% increase in flow) 295 2.10 0.07 1.48 5593 2.9 3.29
300 2.10 0.10 1.32 5802 1.9 3.48
305 2.10 0.37 0.75 5859 145.2 1.61
310 2.10 0.28 0.88 5957 278.9 1.33
315 2.09 0.22 0.98 5880 199.4 1.47

6/15/99 Hydraulic Step 280 1.79 0.05 .54 5835 0.9 3.81
(25% increase in flow) 295 1.80 0.05 .55 5749 1.5 3.58
300 1.80 0.06 .45 5863 15.7 2.57
305 1.80 0.08 .35 5728 328.5 1.24
310 1.80 0.04 .70 5851 140.0 1.62
315 1.80 0.04 .64 5791 10.8 2.73
320 1.80 0.05 .60 6083 1-4 3.62

6/22/99 Hydraulic Step 280 . . . . . .


(25% increase in flow) 295 - - - - - -
300 - - - - - -
305 . - - - - .
310 - - - - - .
315 - . . - - -
320 - - - . - -
360 - - - - - -

219
Table B. 4
Ottawa pilot plant filter removal of C. parvum and B. subtilis

Date Type Seed C. parvum B. subtilis


of Duration (oocysts/L) (CFU/L)
Experiment at Sampling FI FE Log FI FE Log
(min) Removal Removal
8/6/98 Shakedown *f 15 1.6E+5 0 >4.9 - - -
(Stable Filter Operation) 30 1.1E+5 0 >4.7 - - -
45 1.8E+5 4 4.7 - - .
55 2.7E+5 2 5.1 - - -

9/9/98 Stable Filter Operation *f 15 1.1E+6 0 >5.7 . . _


30 7.9E+5 0 >5.6 . . .
45 9.7E+5 2 5.7 . . »
55 9.2E+5 0 >5.7 - - -

9/23/98 Stable Filter Operation *f 15 1.2E+6 0 >5.8 l.OE+5 l.OE+3 2.0


30 1.4E+6 0 >5.8 8.8E+4 O.OE+0 >4.9
45 1.3E+6 0 >5.8 1.3E+5 O.OE+0 >5.1
55 1.4E+6 2 5.8 1.1E+5 O.OE+0 >5.0

10/6/98 Stable Filter Operation *r 15 1.4E+6 0 >5.8 _ . _


30 1.4E+6 0 >5.8 . . .
45 1.4E+6 2 5.8 . . .
55 1.4E+6 4 5.5 - - -

3/9/99 Stable Filter Operation 15 .6E+6 4 5.6 7.9E+4 7.8E+2 2.0


30 .7E+6 18 5.0 8.2E+4 6.7E+2 2.1
45 .7E+6 26 4.8 8.5E+4 4.4E+2 2.3
55 .7E+6 6 5.5 9.5E+4 l.OE+3 2.0
5/31/99 Stable Filter Operation f 15 .1E+6 2 5.7 4.1E+6 9.8E+1 4.6
30 .OE+6 4 5.4 5.7E+6 1.2E+2 4.7
45 .1E+6 4 5.4 4.7E+6 1.2E+2 4.6
55 .2E+6 0 >5.8 5.4E+6 1.5E+2 4.6
7/27/99 Stable Filter Operation f 15 7.6E+5 0 >5.6 3.1E+5 5.0E+0 4.8
30 8.0E+5 0 >5.6 2.7E+5 9.0E+0 4.5
45 7.3E+5 0 >5.6 2.7E+5 l.OE+1 4.4
55 8.2E+5 0 >5.6 2.3E+5 1.4E+1 4.2
1/19/00 Stable Filter Operation 15 4.2E+5 2 5.3 9.9E+5 6.4E+1 4.2
30 4.4E+5 2 5.3 1.1E+6 6.6E+1 4.2
45 5.0E+5 1 5.7 l.OE+6 6.4E+1 4.2
55 4.9E+5 7 4.8 8.8E+5 5.6E+1 4.2
8/18/99 Stable Filter Operation f 15 7.8E+5 0 >5.6 6.3E+4 5.5E+1 3.1
no coagulants in jar 30 7.0E+5 0 >5.5 7.0E+4 4.3E+1 3.2
45 8.2E+5 0 >5.6 6.1E+4 3.6E+1 3.2
55 8.2E+5 0 >5.6 6.6E+4 3.1E+1 3.3
(continued)

220
Table B.4 (continued)
Ottawa pilot plant filter removal of C. parvum and B. subtilis
Date Type Seed C. parvum B. subtilis
of Duration (oocysts/L) (CFU/L)
Experiment at Sampling FI FE Log FI FE Log
(min) Removal Removal
7/20/99 Stable Filter Operation T 150 30 0 >1.2 5.0E+2 4.2E+1 1.1
seeded at rapid mix 165 710 0 >2.6 3.3E+2 3.5E+1 1.0
180 300 0 >2.2 6.1E+2 2.2E+1 1.4
195 210 0 >2.0 4.7E+2 1.9E+1 1.4
210 120 0 >1.8 5.7E+2 1.5E+1 1.6
225 90 0 >1.7 4.1E+2 1.5E+1 1.4
240 10 0 >0.7 3.3E+2 9.0E+0 1.6
255 0 0 >0.0 O.OE+0 1.7E+1 -1.2
270 0 0 >0.0 4.7E+2 7.0E+0 1.8
8/24/99 No Coagulants, No 15 2.9E+5 1.70E+0 0.2 2.9E+5 1.9E+5 0.2
Media 5
no coagulants in plant or 30 3.3E+5 3.50E+0 0.0 4.4E+5 3.4E+5 0.11
jar 5
45 3.7E+5 2.80E+0 0.1 5.0E+5 4.1E+5 0.09
5
55 3.4E+5 3.30E+0 0.0 4.7E+5 2.9E+5 0.21
5

12/13/99 No Coagulants, with 15 4.2E+5 2.4E+05 0.2 1.4E+6 1.4E+6 0.00


Media *
no coagulants in plant or 30 4.8E+5 2.6E+05 0.3 1.8E+6 1.3E+6 0.14
jar
no coagulants since 45 4.2E+5 2.4E+05 0.2 4.2E+6 1.7E+6 0.39
backwash
55 4.8E+5 2.5E+05 0.3 3.9E+6 1.7E+6 0.36

2/9/99 No Coagulants, with 15 5.8E+5 1820 2.5 6.7E+4 l.OE+4 0.81


Media *
no coagulants in plant or 30 5.4E+5 2740 2.3 7.6E+4 3.6E+4 0.33
jar
45 4.3E+5 5990 1.9 l.OE+5 3.4E+4 0.47
55 4.2E+5 6180 1.8 7.1E+4 5.4E+4 0.12

8/4/99 No Coagulants, with 15 2.4E+5 240 3.0 3.1E+5 1.1E+5 0.45


Media
no coagulants in plant or 30 2.6E+5 200 3.1 3.7E+5 2.1E+5 0.25
jar
45 3.8E+5 190 3.3 3.2E+5 3.2E+5 0.00
55 2.9E+5 430 2.8 3.2E+5 3.2E+5 0.00

6/29/99 No Coagulants, with 15 6.5E+5 0 >4.8 1.3E+5 5.6E+2 2.4


Media §
no coagulants in plant 30 6.2E+5 0 >4.8 3.2E+5 7.7E+2 2.6
coagulants in jar 45 6.9E+5 0 >4.8 3.4E+5 8.2E+2 2.6
55 6.6E+5 0 >4.8 3.1E+5 9.2E+2 2.5

7/13/99 No Coagulants, with 15 1.2E+6 0 >5.1 7.8E+4 5.1E+2 2.2


Media §
no coagulants in plant 30 1.1E+6 0 >5.0 2.2E+5 7.0E+2 2.5
coagulants in jar 45 l.OE+6 0 >5.0 5.9E+4 4.4E+2 2.1
55 1.1E+6 0 >5.0 7.9E+4 6.3E+2 2.1
(continued)

221
Table B.4 (continued)
Ottawa pilot plant filter removal of C. parvum and B. subtilis

Date Type Seed C. parvum B. subtilis


of Duration (oocysts/L) (CFU/L)
Experiment at Sampling FI FE Log FI FE Log
(min) Removal Removal
12/17/99 No Silica, with Media T 15 4.8E+5 2 5.4 7.6E+5 2.8E+2 3.4
no silicate in plant or jar 30 4.9E+5 0 >5.4 1.2E+6 3.0E+2 3.6
no coagulants since 45 4.7E+5 9 4.7 1.1E+6 5.0E+2 3.3
backwash
55 4.0E+5 8 4.7 1.1E+6 3.8E+2 3.5

2/18/99 Suboptimal Coagulation 15 4.5E+5 6680 1.8 6.4E+4 5.3E+3 1.1


30"
(52% reduction to 20 5.2E+5 19093 1.4 6.7E+4 9.2E+3 0.9
mg/L)
45 5.3E+5 28747 1.3 7.1E+4 7.2E+3 1.0
55 6.2E+5 26047 1.4 6.5E+4 1.1E+4 0.8

3/23/99 Suboptimal Coagulation 15 1.4E+6 200 3.8 3.4E+4 8.9E+2 1.6


(47% reduction to 20 30 1.5E+6 710 3.3 5.7E+4 6.8E+2 1.9
mg/L)
45 1.6E+6 1570 3.0 5.2E+4 5.3E+2 2.0
55 2.1E+6 2550 2.9 4.2E+4 1.4E+3 1.5

5/4/99 Suboptimal Coagulation 15 1.3E+6 4 5.5 7.1E+3 8.8E+2 0.9


(49% reduction to 22 30 1.5E+6 12 5.1 8.3E+3 9.2E+2 1.0
mg/L)
45 1.6E+6 8 5.3 7.6E+3 9.7E+2 0.9
55 1.7E+6 8 5.3 7.0E+3 7.5E+2 1.0

10/27/98 Filter Ripening 5 1.1E+6 82 4.1 1.5E+5 5.6E+2 2.4


10 1.3E+6 22 4.8 1.5E+5 9.0E+3 1.2
15 1.4E+6 16 4.9 1.8E+5 7.6E+3 1.4
20 1.4E+6 4 5.5 2.0E+5 4.2E+3 1.7
25 1.4E+6 2 5.8 l.OE+4 8.2E+3 0.1

11/3/98 Filter Ripening 5 l.OE+6 62 4.2 6.5E+4 2.3E+3 1.5


10 1.3E+6 - . 7.5E+4 2.3E+3 1.5
15 1.3E+6 18 4.9 1.7E+5 5.6E+2 2.5
20 1.4E+6 2 5.8 2.2E+5 l.SE+3 2.2
25 1.3E+6 4 5.5 2.1E+5 1.7E+2 3.1

1 1/10/98 Filter Ripening 5 9.1E+5 110 3.9 6.3E+4 O.OE+0 >4.8


10 9.6E+5 24 4.6 1.8E+5 5.4E+3 1.5
15 l.OE+6 4 5.4 1.7E+5 4.5E+3 1.6
20 1.1E+6 2 5.7 2.1E+5 6.6E+3 1.5
25 1.1E+6 4 5.4 1.9E+5 O.OE+0 >5.3

1/21/99 Onset-of-breakthrough * 15 7.1E+5 194 3.6 7.9E+4 5.6E+2 2.2


30 6.8E+5 298 3.4 9.5E+4 3.3E+2 ?,.5
45 7.9E+5 1948 2.6 9.7E+4 9.4E+2 2.0
55 7.2E+5 2162 2.5 9.5E+4 8.9E+2 2.0

12/20/99 Onset-of-breakthrough 15 5.7E+5 l.OE+4 1.8 4.8E+5 7.8E+4 0.8


30 6.1E+5 2.2E+4 1.4 3.7E+5 7.2E+4 0.7
45 6.0E+5 1.7E+4 1.6 8.2E+4 4.2E+4 0.3
55 6.4E+5 3.4E+4 1.3 6.7E+4 5.7E+4 0.1
(continued)

222
Table B.4 (continued)
Ottawa pilot plant filter removal of C. parvum and B. subtilis

Date Type Seed C. parvum B. subtilis


of Duration (oocysts/L) (CFU/L)
Experiment at Sampling FI FE Log FI FE Log
(min) Removal Removal
12/22/99 Onset-of-breakthrough 15 6.6E+5 2.5E+4 .4 3.3E+5 3.4E+4 1.0
30 8.0E+5 3.2E+4 .4 3.6E+5 4.8E+4 0.9
45 7.8E+5 3.8E+4 .3 1.9E+5 2.9E+4 0.8
55 8.5E+5 4.1E+4 .3 1.9E+5 2.9E+4 0.8

1 1/25/98 Breakthrough 15 8.1E+5 1.4E+4 .8 8.5E+4 5.4E+3 1.2


30 9.4E+5 1.8E+4 .7 1.5E+5 5.2E+3 1.5
45 9.0E+5 1.7E+4 .7 9.5E+4 7.6E+3 1.1
55 9.5E+5 1.6E+4 .8 1.3E+5 9.1E+3 1.2

12/9/98 Breakthrough 15 9.7E+5 2.1E+4 1.7 4.2E+4 3.2E+3 1-1


tt
30 9.6E+5 1.5E+4 1.8 3.1E+4 3.6E+3 0.9
tt
45 8.5E+5 1.8E+4 1.7 6.9E+4 6.8E+3 1.0
tt
55 9.2E+5 1.9E+4 1.7 6.8E+4 7.8E+3 0.9
tt
1/13/99 Breakthrough §§ 15 6.6E+5 2460 2.4 3.1E+4 1.8E+3 1.2
30 5.3E+5 1.9E+4 1.5 3.5E+4 1.1E+3 1.5
tt
45 5.5E+5 7.4E+4 0.9 2.4E+4 2.0E+3 1.1
tt
55 5.4E+5 3.8E+4 1.2 3.2E+4 2.0E+3 1.2
tt
4/8/99 Runoff 15 1.2E+6 16 4.9 . . .
30 1.5E+6 8 5.3 . - -
45 1.5E+6 4 5.6 - - -
***
55 1.4E+6 4 5.5 - - -

4/13/99 Runoff 15 1.3E+6 4 5.5 _ _ ^


30 1.2E+6 4 5.5 - . -
45 1.3E+6 10 5.1 - - -
55 1.5E+6 6 5.4 - - -

6/7/99 Hydraulic Step f 280 2.0E+5 0 >5.0 4.6E+4 6.5E+J 2.8


(25% increase in flow) 295 2.2E+5 10 4.3 4.6E+4 1.2E+2 2.6
300 2.0E+5 1522 2.1 1.3E+4 4.7E+2 1.4
305 2.1E+3 4412 -0.3 8.2E+3 2.0E+3 0.6
310 1.3E+3 3384 -0.4 5.3E+3 2.1E+3 0.4
315 9.0E+2 3270 -0.6 2.4E+3 1.1E+3 0.3

6/15/99 Hydraulic Step T 280 1.8E+5 0 >4.9 4.0E+4 3.2E+1 3.1


(25% increase in flow) 295 2.0E+5 2 5.0 4.4E+4 3.0E+1 3.2
300 1.9E+5 2 5.0 2.4E+4 2.0E+] 3.1
305 3.0E+4 0 >4.2 7.6E+3 3.7E+1 2.3
310 1.8E+4 0 >3.9 5.1E+3 5.0E+1 2.0
315 7.9E+3 0 >3.6 2.3E+3 6.0E+0 2.6
320 6.9E+3 0 >3.5 2.4E+3 J.6E+1 2.2
(continued)

223
Table B.4 (continued)
Ottawa pilot plant filter removal of C. parvum and B. subtilis

Date Type Seed C. parvum B. subtilis


of Duration (oocysts/L) (CFU/L)
Experiment at Sampling FI FE Log FI FE Log
(min) Removal Removal
6/22/99 Hydraulic Step 280 2.0E+5 4 4.7 2.6E+4 3.6E+1 2.9
(25% increase in flow) 295 2.2E+5 28 3.9 2.6E+4 1.9E+1 3.1
300 2.1E+5 14 4.2 9.1E+3 2.2E+1 2.6
305 3.8E+4 48 2.9 4.7E+3 1.7E+1 2.4
310 1.6E+4 76 2.3 1.1E+3 4.6E+1 1.4
315 6.4E+3 50 2.1 1.1E+3 2.0E+1 1.7
320 5.0E+3 22 2.4 l.OE+3 1.4E+1 1.9
360 3.7E+3 22 2.2 4.4E+2 7.8E+1 0.8
Note: All counts in italics are estimates because they are not in the statistically valid range of20-200 counts per
*
Filter influent Cryptosporidium data estimated by strip counts at 400X.
^on-detects (values of 0) were treated as 2 oocysts/L.
5 Non-detects (values of 0) were treated as 10 oocysts/L.
** Pump failure (for a few minutes) accounts for decreased concentration.
n Filter effluent Cryptosporidium data estimated by strip counts at 400X.
^ Substantial clumping was observed in filter influent Cryptosporidium samples.
Some sample loss occurred during processing.

224
Table B. 5
Summary of Ottawa pilot plant seeding data

Date Experiment Log Removal (Mean ± 1 Standard Deviation) Filter Effluent Value (Mean ± 1 Standard
Deviation)
C. parvum B. subtilis Particles Particles Turbidity
(#/mL) (NTU)
8/6/98 Stable Filter Operation 4.9 £ 0.21 * 3.2 ± 0.29 3.7 ± 2.9 0.02 ± 0.00
9/9/98 5.7 ± 0.06 * 3.8 ± 0.10 0.9 ± 0.2 0.02 ± 0.00
9/23/98 5.8 ± 0.03 * 2.8 ± 0.24 8.7 ± 5.6 0.03 ± 0.00
10/6/98 5.8 ± 0.15 * 4.6 ± 0.18 0.2 ± 0.1 0.02 ± 0.00
3/9/99 5.2 ± 0.38 2.1 ± 0.14 4.1 ± 0.10 0.4 ± 0.1 0.03 ± 0.00
5/31/99 5.6 ± 0.20 4.6 ± 0.05 3.7 ± 0.18 1.2 ± 0.6 0.03 ± 0.00
7/27/99 5.6 ± 0.02 4.5 ± 0.24 3.0 ± 0.22 5.1 ± 1.5 0.04 ± 0.00
1/19/00 5.3 ± 0.36 4.2 ± 0.01 * 4.8 ± 0.6 0.03 ± 0.00
Overall Average 5.5 ± 0.37 3.8 ± 1.07 3.6 ± 0.63 3.1 ± 3.5 0.03 ± 0.01
to 8/18/99 Stable Operation, No Coagulants in 5.6 ± 0.03 3.2 ± 0.11 4.1 ± 0.01 0.4 ± 0.0 0.03 ± 0.00
v\
Jar

7/20/99 Stable - Seeded at Rapid Mix 1.3 ± 0.93 1.1 ± 0.92 3.0 ± 0.10 5.7 ± 1.5 0.06 ± 0.01
10/27/98 Filter Ripening 5.0 ± 0.67 1.4 ± 0.84 3.1 ± 0.48 9.0 ± 12 0.07 ± 0.02
11/3/98 5.1 ± 0.73 2.1 ± 0.69 3.0 ± 0.46 21 ± 22 0.17 ± 0.09
11/10/98 5.0 ± 0.75 2.9 ± 1.93 2.6 ± 0.67 36 ± 61 0.16 ± 0.05
Overall A verage 5.7 ± 0.66 2.1 ± 1.36 2.9 ± 0.55 21.9 ± 37 0.13 ± 0.07
11/25/98 Breakthrough 1.8 ± 0.03 1.2 ± 0.16 1.5 ± 0.14 249 ± 74 0.94 ± 0.09
12/9/98 1.7 ± 0.06 1.0 ± 0.09 1.7 ± 0.05 140 ± 16 0.50 ± 0.04
1/13/99 1.5 ± 0.68 1.3 ± 0.18 3.0 ± 0.05 3.7 ± 0.4 0.61 ± 0.15
Overall Average 1.6 ± 0.38 1.2 ± 0.18 2.1 ± 0.71 131 ± 112 0.68 ± 0.21
1/21/99 Onset-of-breakthrough 3.0 ± 0.52 2.2 ± 0.21 3.3 ± 0.18 1.8 ± 0.7 0.06 ± 0.01
12/20/99 1.5 ± 0.20 0.5 ± 0.35 2.5 ± 0.46 32 ± 36 0.28 ± 0.03
12/22/99 1.4 ± 0.05 0.9 ± 0.08 * * 0.29 ± 0.02
Overall Average 2.0 ± 0.83 1.2 ± 0.79 2.9 ± 0.56 16.7 ± 28.2 0.21 ± 0.11
(continued)
Table B.5 (continued)
Summary of Ottawa pilot plant seeding data
Date Experiment Log Removal (Mean ± 1 Standard Deviation) Filter Effluent Value (Mean ± 1 Standard
Deviation)
C. parvum B. subtilis Particles Particles Turbidity
(#/mL) ' (NTU)
12/17/99 No Silicate in Plant/Jar 5.0 ± 0.39 3.5 ± 0.11 3.6 ± 0.09 3.6 ± 0.8 0.03 ± 0.00

6/29/99 No Coagulants in Plant (Coag. in Jar) 4.8 ± 0.02 2.5 ± 0.12 1.1 ± 0.02 438 ± 25 0.74 ± 0.02
7/13/99 5.0 ± 0.03 2.2 ± 0.18 1.1 ± 0.03 498 ± 34 0.69 ± 0.03

2/9/99 No Coagulants in Plant/Jar 2.1 ± 0.33 0.4 ± 0.29 1.6 ± 0.13 206 ± 50 0.76 ± 0.06
8/4/99 3.1 ± 0.20 0.2 ± 0.22 0.8 ± 0.03 565 ± 35 0.56 ± 0.01
Overall Average 2.6 ± 0.56 0.3 ± 0.28 1.2 ± 0.39 385 ± 196 0.66 ± 0.12

12/13/99 No Coagulants Since Backwash 0.3 ± 0.02 0.2 ± 0.19 0.3 ± 0.03 5172 ± 266 2.27 ± 0.00

0.09 ± 0.12 0.1 ± 0.06 * * *


8/24/99 No Coagulants in Plant/Jar, No Media
NJ
2/18/99 Suboptimal Coagulation 1.5 ± 0.24 0.9 ± 0.13 0.8 ± 0.02 588 ± 23 0.98 ± 0.00
3/23/99 3.3 ± 0.42 1.7 ± 0.25 1.7 ± 0.02 71 ± 3.7 0.93 ± 0.04
5/4/99 5.3 ± 0.17 0.0 ± 0.93 2.4 ± 0.04 21 ± 2.1 0.16 ± 0.03
Overall Average 3.4 ± 1. 66 1.2 ± 0.43 1.7 ± 0.71 227 ± 268 0.69 ± 0.39

4/8/99 Runoff 5.3 ± 0.32 * 4.0 ± 0.09 1.3 ± 0.2 0.04 ± 0.00
4/13/99 5.4 ± 0.18 * 3.9 ± 0.07 2.1 ± 0.3 0.03 ± 0.00
Overall Average 5.3 ± 0.24 * 4.0 ± 0.11 1.7 ± 0.5 0.03 ± 0. 01

6/7/99 Hydraulic Step 1.7 ± 2.52 1.4 ± 1.12 2.5 ± 1.20 105 ± 121 0.18 ± 0.13
6/15/99 4.3 ± 0.66 2.6 ± 0.48 2.7 ± 1.01 71 ± 124 0.05 ± 0.01
6/22/99 3.1 ± 1.02 2.1 ± 0.80 * * *
Overall Average 3.1 ± 1.78 2.1 ± 0.93 2.6 ± 1.06 87 ± 119 0.11 ± 0.11
* Not applicable.
Table B. 6
Experimental schedule at MWD pilot plant

Seeded Microorganism
Date Type of Experiment C. parvum G. lamblia B. sublilis £. coli MS-2 Comments
7/21/98 Lab recovery study y n n n n
8/26/98 Lab recovery study y y n n n
9/16/98 Lab recovery study y y n n n
1/5/99 Lab recovery study y y n n n
1/20/99 Lab recovery study y y n n n

8/4/98 No coagulants w/o media y y y n n

8/1 1/98 No coagulants w/media y n y y y Poor £. call ud MS-2 QA/QC


2/2/99 No coagulants w/media y n y y y
5/4/99 No coagulants w/media y n y y y
7/15/98 Stable filter operation y y y n n
7/28/98 Stable filter operation y y y n n
8/18/98 Stable filter operation y n y y y Poor £ coli QA/QC
9/22/98 Stable filter operation y y y y y Poor E. coli QA/QC
9/29/98 Stable filter operation n n y y y Poor MS-2 QA/QC
10/27/98 Stable filter operation y y y y y
1 1/24/98 Stable filter operation n n y y y No Cryplo or Gtardia
12/15/98 Stable filter operation y n y y y
2/9/99 Stable filter operation y n y y y
3/9/99 Stable filter operation y n y y y
3/16/99 Stable filter operation y n y y y SttittalMforfkt.paii* ;PoorCr>j»o QA/QC
4/27/99 Stable filter operation y n y y y
5/1 1/99 Stable filter operation y n y n n SttdeiaRMforSlu.palod

2/16/99 Suboptimal coagulation y n y y y Jar coegmlauj aajseedf^a Flforl kr.ptHoj ; Poor MS-2 QA/QC
2/23/99 Suboptimal coagulation y n y y y Jar coegitttttd and setdtJ a Flforl hr.perlod
3/2/99 Suboptimal coagulation y n y y y JarconsiilatilmiittMaFlfiirllir.ptrlot
3/23/99 Suboptimal coagulation y n y y y SfrJrJ al KMfar S kr. prHafoor Crypm QA/QC

11/17/98 End of run y y y y y


12/1/98 End of run y n y y y
12/8/98 End of run y n y y y
4/6/99 Hydraulic step y n y n y SteM for Ihr. pert*
4/13/99 Hydraulic step y n y n y Sailed for tkr.ptrioil
4/20/99 Hydraulic step y n y n y to**/.,.*,.,***

10/6/98 Ripening y n y y y UK teed & sampling; poS co« and MS-2 QA/QC
10/13/98 Ripening y y y y y
10/20/98 Ripening y n y y y
11/3/98 Ripening y n y y y
11/10/98 Ripening y n y y y

227
Table B. 7
MWD pilot plant influent water quality

Date Type of Source Temperature TOC UV254 Total Alkalinity pH


Experiment Water CO (mg/L) (abs/cm) (mg/L) Plant Influent Filter Influent
8/4/98 No coagulants, Blend 22* 2.7 0.043 108 7.8 -
no filter media

8/11/98 No coagulants, Blend 23* 2.7 0.048 107 7.8 _


2/2/99 with filter media CRW 13 2.8 0.038 125 8.1 8.1
5/4/99 Blend 15 2.7 0.048 113 7.9 8.1

7/15/98 Stable filter Blend 21* 2.8 0.048 109 7.8 .


7/28/98 operation Blend 21* 2.7 0.043 108 7.7 .
8/18/98 Blend 24* 2.6 0.046 107 7.7 _
9/22/98 Blend 25* 2.8 0.041 109 7.8 _
9/29/98 Blend 23* 2.6 0.048 110 8.1 .
10/27/98 CRW 21* 2.6 0.033 127 8.3 _
11/24/98 CRW 17* 2.6 . 129 8.2 _
12/15/98 CRW 15* 2.6 0.037 134 8.3 .
2/9/99 CRW 13 2.7 0.041 129 8.2 8.1
3/9/99 CRW 14 2.7 0.038 132 8.3 8.1
3/16/99 CRW 14 2.6 0.037 133 8.2 8.1
4/27/99 Blend 15 2.7 0.038 115 8.1 8.0
5/11/99 Blend 16 2.7 0.046 115 8.1 8.1

10/6/98 Ripening CRW 24* 2.8 0.035 124 8.1 .


10/13/98 CRW 23* 2.6 0.038 126 8.4 .
10/20/98 CRW 22* 2.5 0.035 127 8.2 .
11/3/98 CRW 20* 2.6 0.035 127 8.2 .
11/10/98 CRW 19* 2.6 0.036 128 8.2 -

11/17/98 End-of-run CRW 18* 2.6 0.038 129 8.3 .


12/1/98 CRW 17* 2.6 0.037 130 8.2 _
12/8/98 CRW 16* 2.7 0.037 128 8.3 -

2/16/99 Suboptimal CRW 13 2.9 0.037 128 8.2 8.2


2/23/99 coagulation CRW 13 2.7 0.040 131 8.2 8.3
3/2/99 CRW 14 2.6 0.039 132 8.3 8.2
3/23/99 CRW 14 2.6 0.027 133 8.2 8.2

4/6/99 Hydraulic Blend 14 2.7 0.047 114 8.4 8.3


4/13/99 step Blend 14 2.8 0.046 115 8.3 8.2
4/20/99 Blend 17 2.7 0.046 116 8.3 8.0
'Taken from F.E. Weymouth Filtration Plant database

228
Table B. 8
MWD pilot plant filter headloss, turbidity, and particle counts
Type Length Rate of Turbidity (NTU) Particle Counts >=2 \un (#/mL)
Date of of Run Headloss Median Median Log Removals Median Median Log Removals
Experiment (h) (ft/h) Plant Influent Filter Effluent Median 95th Percentile Plant Influent Filter Effluent Median 95th Percentile
8/4/98 No coagulants, . N/A -0.026 2.4 >1.0 O.39 <0.37 5300 7200 -0.13 -0.16
no filter media

8/1 1/98 No coagulants, N/A -0.030 2.0 0.84 0.37 0.32 3800 2600 0.17 0.10
2/2/99 with filter media N/A -0.0019 0.47 0.33 0.15 0.13 3100 1800 0.24 0.17
5/4/99 N/A 0.00078 0.80 0.51 0.20 0.18 5100 2600 0.30 0.28

7/15/98 Stable filter 71.9* 0.074 0.85 0.05 .3 .2 980 1.8 2.6 2.2
7/28/98 operation 55.3* 0.097 1.7 0.05 .6 .4 2400 4.6 2.7 2.5
8/18/98 65.6 0.083 1.6 0.05 .5 .5 3200 0.73 3.6 3.3
9/22/98 57.4* 0.065 1.2 0.04 .4 .4 3100 2.0 3.2 2.8
9/29/98 50.8* 0.075 1.4 0.05 .5 .4 4600 2.0 3.4 3.1
10/27/98 50.6* 0.037 0.94 0.05 .3 .3 3000 6.5 2.7 2.5
11/24/98 29.4 0.18 0.61 0.05 .1 .1 3100 29 2.0 1.9
12/15/98 53.6 0.098 0.58 0.05 .1 .0 2300 25 2.0 1.8
2/9/99 15.8* 0.078 0.54 0.05 .0 0.92 3300 2.0 3.2 2.2
3/9/99 33.6 0.16 0.48 0.05 .0 1.0 3100 2.2 3.2 2.9
4/27/99 45.1 0.12 0.81 0.05 .3 1.2 4800 27 2.3 2.1
3/16/99** 48.7* 0.077 0.53 0.05 .0 0.91 3200 2.1 3.2 2.5
5/1 1/99** 28.2 0.063 0.81 0.05 1.19 1.12 4600 8.9 2.7 2.5

10/6/98 Ripening 70.8* 0.036 1.3 0.05 1.3 1.2 3400 9.7 2.6 2.1
10/13/98 47.1* 0.050 0.96 0.05 1.3 1.3 3200 3.8 2.9 2.6
10/20/98 71.0* 0.050 1.0 0.04 1.4 1.3 3300 4.9 2!8 2.5
11/3/98 68.2 0.079 0.90 0.04 1.3 1.3 2900 3.3 2.9 1.6
11/10/98 47.0* 0.068 0.85 0.05 1.3 1.2 3800 8.8 2.6 2.5

11/17/98 End-of-run 75.7*** 0.080 0.75 0.05 1.2 1.2 3600 13 2.4 1.9
(LOH=6.69 ft.)
12/1/98 73.9*** 0.10 0.54 0.05 1.0 0.97 2800 23 2.1 1.9
(LOH=8 26ft.)
(continued)
Table B.8 (continued)
MWD pilot plant filter headloss, turbidity, and particle counts
Type Length Rate of Turbidity (NTU) Particle Counts >=2 \un (#/mL)
Date of of Run Headless Median Median Log Removals Median Median Log Removals
Experiment 00 (Mi) Plant Influent Filter Effluent Median 95th Percentile Plant Influent Filter Effluent Median 95th Percentile
12/8/98 End-of-run 67.8*** 0.11 0.65 0.05 1.1 1.1 3000 35 1.9 1.7
(LOH=7.87 ft.)
2/16/99 Suboptimal 71.9* 0.020 0.58 0.17 0.53 0.43 3800 480 0.90 0.82.
2/23/99 coagulation 74.2* 0.031 0.54 0.17 0.52 0.45 3500 450 0.89 0.75
3/2/99 55.4* 0.080 0.65 0.16 0.61 0.50 4200 420 1.0 0.80
3/23/99** 72.3* 0.012 0.39 0.14 0.33 0.11 2200 270 0.92 0.52

4/6/99 Hydraulic 43.6 0.064/0.19* 0.55 0.05 1.0 0.92 3300 15 2.3 2.0
4/13/99 step 45.6 0.069/0.17+ 0.55 0.05 1.0 0.97 3600 18 2.3 2.0
4/20/99 56.0 0.045/0. 14+ 0.61 0.05 1.1 1.0 3200 11 2.5 2.4
* Run terminated prematurely
** Seeded at rapid mix
to *** Run terminated at >6 ft headloss
° + Rate before hydraulic step / Rate after hydraulic step
Table B. 9
MWD pilot plant filter removal of turbidity and particles

Date Type Seed Length of Turbidity (NTU) Particles (#/mL)


of Duration Run PI FI FE2 Log >=2 jim
Experiment at Sampling at Sampling Removal
(min) (h) (PI to PI FE Log
FE2) Removal
8/4/98 No coagulants, 15 5.4 2.4 1.6 >1.0 <0.39 5200 7400 -0.15
no filter media 30 5.7 2.5 .6 >1.0 <0.39 5200 7400 -0.15
45 5.9 2.5 .6 >1.0 <0.40 5100 7400 -0.16
60 6.2 2.4 .6 >1.0 <0.39 5100 7400 -0.16

8/11/98 No coagulants, 15 1.0 1.9 .2 0.89 0.33 3800 2900 0.13


with filter 30 1.3 1.9 .2 0.89 0.34 3700 2900 0.11
media
45 1.5 1.9 .2 0.89 0.34 3700 2800 0.13
60 1.8 2.0 .2 0.87 0.36 3800 3100 0.09

2/2/99 15 4.8 0.47 0.32 0.34 0.15 3200 2000 0.20


30 5.0 0.47 0.32 0.34 0.14 3300 2000 0.20
45 5.3 0.47 0.32 0.34 0.14 3000 2100 0.16
60 5.5 0.46 0.32 0.34 0.13 3000 2100 0.16

5/4/99 15 4.9 0.81 0.55 0.49 0.22 5000 2600 0.29


30 5.1 0.85 0.56 0.45 0.28 5200 2600 0.30
45 5.4 0.84 0.55 0.42 0.31 5100 2600 0.30
60 5.6 0.83 0.55 0.49 0.23 5000 2500 0.30

7/15/98 Stable filter 15 4.6 0.93 0.61 0.05 1.3 1100 5.0 2.3
operation 30 4.8 0.94 0.61 0.05 1.3 1100 5.8 2.3
40 5.0 0.92 0.61 0.05 1.3 1100 7.8 2.1
50 5.2 0.89 0.62 0.05 1.2 1000 6.7 2.2

7/28/98 15 4.8 1.3 0.81 0.05 1.5 1800 5.1 2.6


30 5.0 1.2 0.82 0.05 1.4 1800 4.3 2.6
45 5.3 1.2 0.86 0.05 1.4 1800 4.2 2.6
55 5.4 1.2 0.85 0.05 1.4 1800 4.3 2.6

8/18/98 15 3.6 1.6 0.85 0.05 1.5 3300 1.5 3.3


30 3.9 1.7 0.83 0.05 1.5 3500 1.8 3.3
45 4.1 1.6 0.84 0.05 1.5 3200 1.4 3.4
60 4.4 1.6 0.84 0.05 1.5 3200 1.3 3.4

9/22/98 15 5.5 .2 0.95 0.05 1.4 3200 5.0 2.8


30 5.8 .1 0.95 0.05 1.4 3200 5.0 2.8
45 6.0 .2 0.95 0.05 1.4 3200 5.2 2.8
60 6.2 .1 0.93 0.05 1.4 3400 5.5 2.8

9/29/98 15 4.6 .2 0.90 0.05 1.4 4400 3.4 3.1


30 4.9 .3 0.93 0.05 1.4 4300 3.4 3.2
45 5.1 1.3 0.90 0.05 1.4 4300 2.7 3.4
60 5.3 1.3 0.90 0.05 1.4 4300 2.8 3.3

10/27/98 15 5.0 0.98 0.80 0.05 1.3 2900 6.6 2.6


30 5.3 0.99 0.79 0.05 1.3 3200 9.2 2.5
45 5.5 0.97 0.80 0.05 1.3 2900 11 2.4
60 5.8 1.0 0.80 0.05 1.3 3000 14 2.3
(continued)

231
Table B.9 (continued)
MWD pilot plant filter removal of turbidity and particles

Date Type Seed Length of Turbidity (NTU) Particles (#/mL)


of Duration Run PI FI FE2 Log >=2 |im
Experiment at Sampling at Sampling Removal
(min) (h) (PI to PI FE Log
FE2) Removal
11/24/98 15 4.1 0.60 0.57 0.05 1.1 3000 32 2.0
30 4.4 0.72 0.57 0.05 1.1 3400 34 2.0
45 4.6 0.64 0.59 0.05 1.1 3000 30 2.0
60 4.9 0.61 0.59 0.05 1.1 3100 33 2.0

12/15/98 15 3.5 0.55 0.53 0.05 1.0 2100 36 1.8


30 3.8 0.55 0.53 0.05 1.0 2100 35 1.8
45 4.0 0.54 0.53 0.05 1.0 2200 39 1.8
60 4.3 0.54 0.52 0.05 1.0 2100 38 1.7

2/9/99 15 5.3 0.53 0.53 0.06 1.0 3100 20 2.2


30 5.6 0.51 0.52 0.06 1.0 3000 20 2.2
45 5.8 0.57 0.50 0.06 1.0 3300 22 2.2
60 6.1 0.57 0.50 0.06 1.0 3400 23 2.2

3/9/99 15 5.2 0.48 0.51 0.05 1.0 3200 3.1 3.0


30 5.4 0.50 0.49 0.05 1.0 3300 3.8 2.9
45 5.7 0.50 0.49 0.05 1.0 3300 4.6 2.9
60 5.9 0.50 0.51 0.05 1.0 3200 4.0 2.9

4/27/99 15 5.2 0.77 0.62 0.05 1.2 4600 28 2.2


30 5.4 0.77 0.60 0.05 1.2 4600 28 2.2
45 5.7 0.77 0.62 0.05 1.2 4500 26 2.2
60 5.9 0.77 0.62 0.05 1.2 4500 26 2.2

3/16/99 Stable filter 255 4.2 0.52 0.39 0.06 0.97 2500 3.4 2.9
operation 270 4.5 0.46 0.39 0.06 0.92 2500 3.9 2.8
(Seeded at 285 4.7 0.46 0.39 0.06 0.92 2600 3.4 2.9
rapid mix)
300 5.0 0.46 0.39 0.06 0.92 2600 3.0 2.9

5/11/99 255 4.3 0.79 0.59 0.06 1.2 4400 10 2.6


270 4.5 0.80 0.59 0.06 1.2 4400 9.9 2.7
285 4.8 0.80 0.57 0.06 1.2 4400 8.9 2.7
300 5.0 0.80 0.56 0.05 1.2 4400 8.8 2.7

2/16/99 Suboptimal 15 3.3 0.62 0.41 0.19 0.5 4200 390 1.0
coagulation 30 3.5 0.59 0.39 0.17 0.5 4000 370 .0
45 3.8 0.58 0.39 0.17 0.5 3800 350 .0
60 4.0 0.55 0.40 0.17 0.5 3700 360 .0
2/23/99 15 5.3 0.59 0.40 0.17 0.6 4000 390 .0
30 5.6 0.60 0.42 0.16 0.6 4100 400 .0
45
60' 5.8 0.54 0.42 0.16 0.5 3600 420 0.9
6.1 0.54 0.43 0.16 0.5 3600 410 0.9
3/2/99 15 4.9 0.66 0.51 0.15 0.7 4400 210 1.3
30 5.1 0.64 0.51 0.14 0.7 4300 200 1.3
45 5.4 0.65 0.51 0.14 0.7 4300 190 1.4
60 5.6 0.64 0.49 0.13 0.7 4400 180 1.4
(continued)
232
Table B.9 (continued)
MWD pilot plant filter removal of turbidity and particles

Date Type Seed Length of Turbidity (NTU) Particles (#/mL)


of Duration Run PI FI • FE2 Log >=2 urn
Experiment at Sampling at Sampling Removal
(min) 00 (PI to PI FE Log
FE2) Removal
3/23/99 Suboptimal 255 4.2 0.37 0.31 0.18 0.3 2100 440 0.7
coagulation 270 4.5 0.37 0.29 0.18 0.3 2100 420 0.7
(Seeded at 285 4.7 0.39 0.29 0.17 0.3 2200 420 0.7
rapid mix)
300 5.0 0.37 0.29 0.17 0.3 2300 420 0.7

10/6/98 Ripening 15 0.40 0.76 0.82 0.07 1.0 3400 44 1.8


30 0.65 0.77 0.82 0.06 1.1 3300 31 1.9
45 0.90 0.69 0.80 0.06 1.1 3300 34 2.0
60 1.2 0.72 0.82 0.06 1.1 3500 24 2.0

10/13/98 10 0.20 0.99 0.86 0.16 0.8 3400 49 2.0


20 0.37 1.0 0.89 0.07 1.2 3400 29 2.2
40 0.70 0.98 0.86 0.06 1.2 3300 19 2.2
60 1.0 1.0 0.85 0.06 1.3 3300 18 2.3

10/20/98 10 0.18 1.0 0.82 0.17 0.8 3400 41 2.0


20 0.35 0.99 0.83 0.07 .1 3400 19 2.3
40 0.68 0.99 0.83 0.06 .3 3300 12 2.4
60 1.0 0.98 0.83 0.05 .3 3300 11 2.5

1 1/3/98 10 0.20 0.89 0.76 0.15 0.8 3300 36 2.0


20 0.37 0.89 0.79 0.06 .2 3100 13 2.4
40 0.70 0.88 0.77 0.06 .2 3000 11 2.4
60 1.0 0.88 0.75 0.05 .2 3000 12 2.4

11/10/98 10 0.20 0.87 0.82 0.14 0.8 3800 68 1.8


20 0.37 0.88 0.79 0.07 1.1 3700 22 2.2
40 0.70 0.86 0.80 0.06 1.2 3600 15 2.4
60 1.0 0.91 0.82 0.06 1.2 3900 15 2.4

11/17/98 End-of-run 15 74.9 0.70 0.65 0.05 .2 3300 7.9 2.6


30 75.2 0.70 0.66 0.05 .2 4200 7.9 2.7
45 75.4 0.70 0.66 0.05 .2 3600 7.8 2.7
60 75.7 0.69 0.67 0.05 .2 3700 8.0 2.7

12/1/98 15 73.1 0.53 0.57 0.06 .0 2600 17 2.2


30 73.4 0.54 0.56 0.06 .0 2600 17 2.2
45 73.6 0.53 0.56 0.06 1.0 3300 18 2.3
60 73.9 0.53 0.56 0.05 1.0 3000 20 2.2

12/8/98 15 67.1 0.65 0.62 0.05 1.1 2900 32 2.0


30 67.3 0.66 0.63 0.06 1.1 3000 29 2.0
45 67.6 0.69 0.62 0.05 1.1 3100 26 2.1
60 67.8 0.66 0.62 0.05 1.1 3100 27 2.1

4/6/99 Hydraulic 485 27.0 0.54 0.50 0.05 1.0 3200 16 2.3
step 500 27.3 0.54 0.46 0.05 1.0 3200 13 2.4
505 27.4 0.54 0.47 0.05 1.0 3200 14 2.4
520 27.6 0.54 0.48 0.05 1.0 3200 24 2.2
(continued)

233
Table B.9 (continued)
MWD pilot plant filter removal of turbidity and particles

Date Type Seed Length of Turbidity (NTU) Particles (#/mL)


of Duration Run PI FI FE2 Log >=2|im
Experiment at Sampling at Sampling Removal
(min) (h) (PI to PI FE Log
FE2) Removal
4/13/99 485 26.3 0.53 0.45 0.05 .0 3400 22 2.2
500 26.6 0.53 0.42 0.05 .0 3200 16 2.3
505 26.7 0.53 0.42 0.05 .0 3200 14 2.4
520 26.9 0.53 0.44 0.05 .0 3300 22 2.2
4/20/99 485 25.9 0.60 0.39 0.05 3000 11 2.4
500 26.1 0.60 0.39 0.05 3000 7.8 2.6
505 26.2 0.60 0.39 0.05 3000 7.5 2.6
520 26.5 0.58 0.39 0.05 3000 17 2.2

234
Table B. 10
MWD pilot plant filter removal of C. parvum and B. subtilis
Type Seed Duration C. parvum B. subtilis
Date of at Sampling (oocysts/L) (CFU/L)
Experiment (min) FI FE Log FI FE Log
Removal Removal
8/4/98 No coagulants, 15 259E+3 209E+3 0.09 49.0E+3 47.0E+3 0.018
no filter media 30 305E+3 332E+3 -0.04 48.0E+3 55.0E+3 -0.059
45 450.0E+3* 338E+3 0.12 47.0E+3 51.0E+3 -0.035
60 248E+3 325E+3 -0.12 52.0E+3 55.0E+3 -0.024

8/1 1/98 No coagulants, 15 360E+3 333E+3** 0.03 107E+3 55.0E+3 0.289


with filter media 30 339E+3 258E+3** 0.12 99.0E+3 72.0E+3 0.138
45 285E+3 251E+3** 0.06 88.0E+3 90.0E+3 -0.010
60 341E+3 47.0E+3* 0.86 85.0E+3 104E+3 -0.088

2/2/99 15 226E+3 205E+3 0.04 2.42E+06 2.42E+6 0.001


30 163E+3 198E+3 -0.08 2.24E+06 2.47E+6 -0.042
45 307E+3 180E+3 0.23 2.42E+06 2.39E+6 0.004
60 251E+3 200E+3 0.10 2.15E+06 2.70E+6 -0.100

5/4/99 15 214E+3 153E+3 0.14 3.73E+6 2.57E+6 0.163


30 144E+3 98.0E+3 0.17 3.57E+6 3.53E+6 0.004
45 273E+3 150E+3 0.26 2.73E+6 3.07E+6 -0.050
60 162E+3 112E+3 0.16 3.77E+6 3.83E+6 -0.008

7/15/98 Stable filter 15 94.8E+3 240E+0 2.6 82.0E+3 620E+0 2.1


operation 30 83.6E+3 190E+0 2.6 61.0E+3 480E+0 2.1
40 64.2E+3 190E+0 2.5 58.0E+3 520E+0 2.0
50 58.6E+3 120E+0 2.7 36.0E+3 530E+0 1.8

7/28/98 15 191E+3 84E+0 3.4 140E+3 290E+0 2.7


30 169E+3 76E+0 3.3 103E+3 280E+0 2.6
45 197E+3 76E+0 3.4 89.0E+3 290E+0 2.5
55 152E+3 86E+0 3.2 178E+3 150E+0 3.1

8/18/98 15 238E+3 30E+0 3.9 10.0E+3 80E+0 2.1


30 46.1E+3 16E+0 3.5 9.00E+3 30E+0 2.5
45 96.9E+3 10E+0 4.0 10.0E+3 60E+0 2.2
60 195E+3 2E+0 5.0 7.00E+3 30E+0 2.4

9/22/98 15 29.9E+3 6E+0 3.7 9.53E+3 577E+0 1.2


30 118E+3 26E+0 3.7 18.3E+3 120E+0 2.2
45 176E+3 18E+0 4.0 11.7E+3 77E+0 2.2
60 242E+3 30E+0 3.9 6.87E+3 73E+0 2.0

9/29/98 15 _ . _ 19.0E+3 27E+0 2.9


30 - - • 44.0E+3 40E+0 3.0
45 - - . 29.0E+3 10E+0 3.5
60 - - - 117E+3 27E+0 3.6

10/27/98 15 219E+3 98E+0 3.3 9.63E+3 60E+0 2.2


11/24/98 15 - . - 12.0E+3 57E+0 2.3
30 - - - 9.73E+3 1.63E+3 0.8
45 - - - 9.67E+3 20E+0 2.7
60 - - - 8.73E+3 33E+0 2.4
(continued)

235
Table B. 10 (continued)
MWD pilot plant C. parvum and B. subtilis data

Type Seed Duration C. parvum B. subtilis


Date of at Sampling (oocysts/L) (CFU/L)
Experiment (min) FI FE Log FI FE Log
Removal Removal
12/15/98 15 238E+3 190E+0 3.1 114E+3 1.68E+3 1.8
30 245E+3 284E+0 2.9 67.7E+3 1.24E+3 1.7
45 246E+3 310E+0 2.9 50.3E+3 833E+0 1.8
60 206E+3 304E+0 2.8 38.0E+3 793E-K) 1.7

2/9/99 15 626E+3 6.24E+3 2.0 229E+3 2.21E+3 2.0


30 544E+3 5.85E+3 2.0 111E+3 1.22E+3 2.0
45 588E+3 3.48E+3 2.2 83.0E+3 900E+0 2.0
60 644E+3 3.69E+3 2.2 56.0E+3 780E+0 1.9

3/9/99 15 183E+3 366E+0 2.7 139E+3 2.05E+3 1.8


30 154E+3 718E+0 2.3 103E+3 1.3E+3 1.9
45 209E+3 758E+0 2.4 70.0E+3 893E+0 1.9
60 189E+3 900E+0 2.3 48.3E+3 720E+0 1.8
4/27/99 15 119E+3 86.0E+0 3.1 185E+3 1.97E+3 2.0
30 157E+3 154E+0 3.0 127E+3 1.50E+3 1.9
45 136E+3 162E+0 2.9 93.3E+3 1.15E+3 1.9
60 57.8E+3 192E+0 2.5 69.7E+3 1.04E+3 1.8
3/16/99 Stable filter 255 2.28E+3 1.60E+3 0.2 68.0E+3 410E+0 2.2
operation 270 2.14E+3 1.70E+3 0.1 73.3E+3 467E+0 2.2
(Seeded at rapid mix) 285 1.90E+3 1.11E+3 0.2 65.0E+3 433E+0 2.2
300 2.16E+3 1.42E+3 0.2 65.7E+3 437E+0 2.2
5/11/99 255 4.47E+3 14.0E+0 2.5 63.7E+3 417E+0 2.2
270 3.96E+3 22.0E+0 2.3 66.7E+3 417E+0 2.2
285 5.06E+3 28.0E+0 2.3 70.3E+3 437E+0 2.2
300 6.00E+3 24.0E+0 2.4 65.3E+3 373E+0 2.2
2/16/99 Suboptimal 15 97.5E+3 13.1E+3 0.9 330E+3 86.7E+3 0.58
coagulation 30 154E+3 19.5E+3 0.9 223E+3 - -
45 261E+3 20.9E+3 1.1 230E+3 63.3E+3 0.56
60 123E+3 22.7E+3 0.7 127E+3 41.0E+3 0.49
2/23/99 15 255E+3 24.1E+3 1.0 543E+3 217E+3 0.40
30 240E+3 44.7E+3 0.7 400E+3 168E+3 0.38
45 145E+3 42.7E+3 0.5 407E+3 138E+3 0.47
60 297E+3 43.3E+3 0.8 397E+3 125E+3 0.50
3/2/99 15 122E+3 1.99E+3 1.8 573E+3 138E+3 0.62
30 91.9E+3 1.76E+3 1.7 373E+3 111E+3 0.53
45 48.6E+3 1.77E+3 1.4 317E+3 89.7E+3 0.55
60 129E+3 2.03E+3 1.8 303E+3 80.7E+3 0.58
3/23/99 Suboptimal 255 5.54E+3 3.63E+3 0.2 84.0E+3 39.7E+3 0.33
coagulation 270 7.10E+3 4.22E+3 0.2 82.0E+3 32.0E+3 0.41
(Seeded at rapid mix) 285 4.24E+3 4.53E+3 -0.03 82.3E+3 39.3E+3 0.32
300 5.80E+3 3.88E+3 0.2 78.7E+3 41.3E+3 0.28
(continued)

236
Table B. 10 (continued)
MWD pilot plant C. parvum and B. subtilis data

Type Seed Duratioii C. parvum B. subtilis


Date . of at Sampling (oocysts/L) (CFU/L)
Experiment (min) FI FE Log FI FE Log
Removal Removal
10/6/98 Ripening 15 1 10E+3 234E+0 2.7 16.8E+3 683E+0 1.4
30 123E+3 158E+0 2.9 13.8E+3 593E+0 1.4
45 79.1E+3 126E+0 2.8 12.4E+3 230E+0 1.7
60 192E+3 30E+0 3.8 10.1E+3 143E+0 1.8

10/13/98 10 210E+3 478E+0 2.6 22.1E+3 807E+0 1.4


20 112E+3 144E+0 2.9 21.8E+3 353E+0 1.8
40 96.8E+3 142E+0 2.8 20.0E+3 203E+0 2.0
60 73.8E+3 116E+0 2.8 9.37E+3 153E+0 1.8

10/20/98 10 269E+3 264E+0 3.0 8.97E+3 750E+0 1.1


20 281E+3 150E+0 3.3 12.2E+3 610E+0 1.3
40 162E+3 56E+0 3.5 11.1E+3 217E+0 1.7
60 220E+3 48E+0 3.7 5.03E+3 100E+0 1.7
1 1/3/98 10 34.5E+3 712E+0 1.7 5.63E+3 540E+0 1.0
20 151E+3 372E+0 2.6 4.73E+3 273E+0 1.2
40 52.8E+3 278E+0 2.3 2.43E+3 87E+0 1.4
60 247E+3 346E+0 2.9 2.37E+3 77E+0 1.5
11/10/98 10 37.0E+3 198E+0 2.3 5.37E+3 303E+0 1.2
20 63.9E+3 144E+0 2.6 4.90E+3 183E+0 1.4
40 90.3E+3 66E+0 3.1 2.33E+3 97E+0 1.4
60 231E+3 4E+0 4.8 2.37E+3 30E+0 1.9

11/17/98 End-of-run 15 130E+3 76E+0 3.2 4.60E+3 7E+0 2.8


30 302E+3 256E+0 3.1 9.80E+3 7E+0 3.2
45 88.3E+3 236E+0 2.6 3.70E+3 3E+0 3.0
60 242E+3 326E+0 2.9 5.90E+3 ND >2.8

12/1/98 15 95.7E+3 580E+0 2.2 1 14E+3 180E+0 2.8


30 113E+3 654E+0 2.2 90.3E+3 177E+0 2.7
45 239E+3 764E+0 2.5 71.7E+3 133E+0 2.7
60 118E+3 992E+0 2.1 47.3E+3 83E+0 2.8

12/8/98 15 145E+3 782E+0 2.3 85.0E+3 63E+0 3.1


30 285E+3 2.39E+3 2.1 78.7E+3 483E+0 2.2
45 225E+3 1.79E+3 2.1 51.7E+3 207E+0 2.4
60 103E+3 1.40E+3 1.9 39.7E+3 167E+0 2.4
4/6/99 Hydraulic 485 18.7E+3 404E+0 1.7 2.57E+3 33E+0 1.9
step 500 460E+0 6.00E+0 1.9 950E+0 10E+0 2.0
505 440E+0 40.0E+0 1.0 1.37E+3 12E+0 2.1
520 200E+0 74.0E+0 0.4 533E+0 3E+0 2.2

4/13/99 485 85.4E+3 350E+0 2.4 2.39E+3 124E+0 1.3


500 5.92E+3 36.0E+0 2.2 833E+0 12E+0 1.8
505 2.60E+3 30.0E+0 1.9 1.07E+3 3E+0 2.5
520 380E+0 52.0E+0 0.9 622E+0 1E+0 2.7
(continued)

237
TableB. 10 (continued)
MWD pilot plant C. parvum and B. subtilis data

Type Seed Duration C. parvum B. subtilis


Date of at Sampling (oocysts/L) (CFU/L)
Experiment (min) FI FE Log FI FE Log
Removal Removal
4/20/99 485 9.4E+3 6.00E+0 3.2 1.27E+3 13E+0 2.0
500 1.2E+3 ND 2.8 622E+0 1E+0 2.7
505 40.0E+0 4.00E+0 1.0 922E+0 2E+0 2.6
520 120E+0 ND 1.8 600E+0 1E+0 2.7
ND = None detected
* Estimated by strip counts @ 200X
** Estimated by field counts @ 400X

238
Table B. 11
MWD pilot plant filter removal of E. coli, MS-2 and G. lamblia
Type Seed Duration E. coli MS-2 G. lamblia
Date of at Sampling (CFU/L) (PFU/L) (cysts/L)
Experiment (min) FI FE Log FI FE Log FI FE Log
Removal Removal Removal
8/4/98 No coagulants, 15 8.23E+04 9.35E+04 -0.06
no filter media 30 1.19E+05 1.15E+05 0.01
45 1.40E+05 1.29E+05 0.04
60 1.55E+05 1.12E+05 0.14

8/11/98 No coagulants, 15 TNTC TNTC 1.13E+9 850E+6 0.12


with filter media 30 TNTC TNTC 1.10E+9 750E+6 0.17
45 TNTC TNTC 867E+6 TNTC
60 TNTC TNTC 1.13E+9 TNTC

2/2/99 - 15 7.2E+6 6.5E+6 0.04 207E+6 183E+6 0.054


30 7.8E+6 6.8E+6 0.06 217E+6 218E+6 -0.003
45 7.3E+6 6.3E+6 0.06 273E+6 217E+6 0.101
60 7.0E+6 6.6E+6 0.03 220E+6 260E+6 -0.073

5/4/99 15 27.3E+6 26.6E+6 0.01 110E+6 124E+6 -0.053


30 27.6E+6 26.3E+6 0.02 127E+6 146E+6 -0.060
45 25.1E+6 27.0E+6 -0.03 124E+6 146E+6 -0.072
60 28.3E+6 27.3E+6 0.02 142E+6 131E+6 0.035

7/15/98 Stable filter 15 1.01E+05 2 4.7


operation 30 1.14E+05 4 4.5
40 9.20E+04 0 >4.7
50 1.10E+05 2 4.7

7/28/98 15 1.56E+05 0 >4.9


30 7.43E+04 2 4.6
45 7.41E+04 0 >4.6
55 3.95E+04 4 4.0

8/18/98 15 ND 900E+0 8.33E+6 247E+3 1.5


30 ND 1.35E+3 7.67E+6 80.0E+3 2.0
45 ND 970E+0 7.33E+6 153E+3 1.7
60 ND 430E-K) - 10.3E+6 67.0E+3 2.2 - - -

9/22/98 15 2.18E-HS TNTC _ 12.0E+6 610E+3 1.3 1.32E+04 0 >3.8


30 2.07E+6 TNTC - 32.5E+6 443E+3 1.9 3.77E+04 0 >4.3
45 2.18E+6 TNTC - 59.0E+6 447E+3 2.1 9.02E+04 0 >4.7
60 5.34E+6 TNTC - 38.7E+6 380E+3 2.0 1.75E+05 0 >4.9

9/29/98 15 1.22E+6 427E+3 0.46 218E+6 12.7E+6 1.2*** _ _ .


30 2.12E+6 520E+3 0.61 281E+6 11.7E+6 1.4*** - - .
45 1.90E+6 363E+3 0.72 298E+6 13.0E+6 1.4*** - . .
60 1.27E+6 363E+3 0.54 294E+6 14.4E+6 1.3*** - - -

10/27/9 15 867E+3 367E+3 0.37 100E+6 9.7E+6 1.0 9.88E+04 0 >4.7

30 1.60E+6 313E+3 0.71 170E+6 13.0E+6 1.1 8.84E+04 0 >4.6


45 633E+3 347E+3 0.26 117E+6 14.7E+6 0.9 1.50E+03 0 >2.9
60 1.03E+6 300E+3 0.54 160E+6 14.0E+6 1.1 6.95E+04 0 >4.5
(continued)

239
Table B. 11 (continued)
MWD pilot plant filter removal of E. coli, MS-2 and G. lamblia
Type Seed Duration E. coli MS-2 G. lamblia
Date of at Sampling (CFU/L) (PFU/L) (cysts/L)
Experiment (min) FI FE Log FI FE Log FI FE Log
Removal Removal Removal
11/24/9 Stable filter 15 967E+3 313E+3 - 89.7E+6 4.63E+6 1.3 -

operation 30 1.37E+6 370E+3 _ 74.0E+6 7.07E+6 1.0 - .


(high FE E. coli 45 1.70E+6 370E+3 -- 87.0E+6 7.00E+6 1.1 -
background)
60 667E+3 447E+3 — 100E+6 6.63E+6 1.2 -

12/15/9 15 700E+3 210E+3 0.52 86.3E+6 6.27E+6 1.1 _


8
30 333E+3 233E+3 0.15 92.7E+6 8.37E+6 1.0 -
45 533E+3 187E+3 0.46 91.0E+6 6.83E+6 1.1 .
60 437E+3 243E+3 0.25 76.0E+6 7.37E+6 1.0 -

2/9/99 15 1.73E+6 197E+3 0.95 41.0E+6 4.27E+6 1.0 .


30 627E+3 223E+3 0.45 43.3E+6 6.17E+6 0.8 -
45 633E+3 263E+3 0.38 55.0E+6 5.83E+6 1.0 -
60 900E+3 220E+3 0.61 53.7E+6 4.87E+6 1.0 -

3/9/99 15 2.37E+6 233E+3 „ 21.0E+6 1.07E+6 1.3 _


(high FE E. coli 30 1.63E+6 267E+3 .. 19.7E+6 1.20E+6 1.2 -
background)
45 2.10E+6 400E+3 — 22.0E+6 1.17E+6 1.3 -
60 1.33E+6 767E+3 - 19.7E+6 1.03E+6 1.3 - .
3/16/99 255 63.3E+3 16.3E+3 0.59 2.63E+6 15.7E+3 2.2 .
(Seeded at rapid mix) 270 96.7E+3 16.3E+3 0.77 3.90E+6 23.7E+3 2.2 .
285 66.7E+3 19.0E+3 0.55 3.00E+6 16.7E+3 2.3 .
300 83.3E+3 14.3E+3 0.76 2.17E+6 22.7E+3 2.0 -

4/27/99 15 1.93E+6 470E+3 Q.61 91.3E+6 8.47E+6 1.0 _


30 1.57E+6 350E+3 0.65 80.0E+6 9.57E+6 0.9 .
45 867E+3 390E+3 0.35 100E+6 10.1E+6 1.0 .
60 1.37E+6 430E+3 0.50 94.7E+6 8.63E+6 1.0 -

5/11/99 255 _ _ „ _ . .
(Seeded at rapid mix) 270 - - - . • . .
285 - . . . . .
300 - - - - . -

10/6/98 Ripening 15 1.83E+6 1.90E+6 -0.02 ND ND ... .


30 2.43E+6 1.70E+6 0.16 ND ND . .
45 2.10E+6 1.33E+6 0.20 ND ND . .
60 2.60E+6 1.63E+6 0.20 ND ND . -

10/13/9 10 1.23E+6 320E+3 0.6 51.3E+6 7.47E+6 0.8 1.25E+05 104 3.1
8
20 1.53E+6 267E+3 0.8 54.3E+6 7.70E+6 0.8 7.19E+04 6 4.1
40 833E+3 317E+3 0.4 47.3E+6 79.7E+6 -0.2 8.16E+04 6 4.1
60 3.70E+6 257E+3 1.2 39.7E+6 101E+6 -0.4 5.14E+04 0 >4.4

10/20/9 10 1.53E+6 700E+3 0.3 22.3E+6 2.40E+6 1.0 .


8
20 1.87E+6 733E+3 0.4 19.3E+6 1.87EH-6 1.0 _
40 1.30E+6 l.OOE+6 0.1 15.3E+6 1.30E+6 1.1 .
60 1.87E+6 1.47E+6 0.1 9.67E+6 967E+3 1.0 -
(continued)

240
Table B. 11 (continued)
MWD pilot plant filter removal of E. coli, MS-2 and G. lamblia
Type Seed Duration E. coli MS-2 G. lamblia
Date of at Sampling (CFU/L) (PFU/L) (cysts/L)
Experiment (mjn) FI FE Log FI FE Log FI FE Log
Removal Removal Removal
1 1/3/98 10 1.43E+6 370E+3 0.6 76.7E+6 14.0E+6 0.7 .
20 933E+3 383E+3 0.4 113E+6 5.67E+6 1.3 -
40 1.03E+6 410E+3 0.4 163E+6 8.67E+6 1.3 -
60 733E+3 437E+3 0.2 143E+6 8.00E+6 1.3 -

11/10/9 10 600E+3 350E+3 0.2 100E+6 ll.OE+6 1.0 -

20 967E+3 407E+3 0.4 103E+6 17.0E+6 0.8 .


40 1.07E+6 460E+3 0.4 127E+6 10.3E+6 1.1 -
60 2.53E+6 503E+3 0.7 163E+6 9.00E+6 1.3 - -

11/17/9 End-of-run 15 480E+3 36.7E+3 1.1 76.0E+6 383E+3 2.3 1.05E+05 0 >4.7

30 447E+3 70.0E+3 0.8 84.7E+6 867E+3 2.0 1.92E+05 0 >5.0


45 1.43E+6 63.3E+3 1.4 94.0E+6 783E+3 2.1 1.07E+05 0 >4.7
60 280E+3 66.7E+3 0.6 94.0E+6 823E+3 2.1 1.47E+05 0 >4.9

12/1/98 15 633E+3 70.0E+3 __ 42.7E+6 353E+3 2.1 ,


(high FE E. coli 30 700E+3 127E+3 - 64.7E+6 647E+3 2.0 -
background)
45 400E+3 160E+3 — 56.0E+6 590E+3 2.0 -
60 1.07E+6 117E+3 - 50.3E+6 743E+3 1.8 -

12/8/98 15 367E+3 30.0E+3 „ 40.3E+6 287E+3 2.1 _


(highFE£ coli 30 933E+3 100E+3 - 57.3E+6 1.13E+6 1.7 -
background)
45 1.77E+6 76.7E+3 — 55.7E+6 1.07E+6 1.7 .
60 533E+3 123E+3 - 54.7E+6 1.09E+6 1.7 -

2/16/99 Suboptimal 15 1.43E+6 667E+3 0.3 ND ND - _


coagulation 30 1.40E+6 600E+3 0.4 ND ND . -
45 867E+3 467E+3 0.3 ND ND . -
60 1.37E+6 967E+3 0.2 ND ND . -

2/23/99 15 1.50E+6 1.27E+6 0.1 48.3E+6 6.77E+6 0.9 _


30 2.13E+6 933E+3 0.4 54.3E+6 9.67E+6 0.7 -
45 1.80E+6 1.13E+6 0.2 49.7E+6 9.93E+6 0.7 .
60 2.20E+6 1.10E+6 0.3 50.3E+6 8.77E+6 0.8 -

3/2/99 15 1.53E+6 467E+3 0.5 28.3E+6 3.17E+6 1.0 .


30 1.47E+6 667E+3 0.3 31.0E+6 3.97E+6 0.9 .
45 1.50E+6 850E+3 0.2 35.7E+6 3.53E+6 1.0 .
60 1.67E+6 433E+3 0.6 35.0E+6 4.93E+6 0.9 -

3/23/99 255 8.33E+3 5.67E+3 0.2 1.50E+6 967E+3 0.2 _


(Seeded at rapid mix) 270 11.3E+3 8.00E+3 0.2 2.37E+6 1.07E+6 0.3 -
285 15.3E+3 6.33E+3 0.4 2.40E+6 1.27E+6 0.3 .
300 11.7E+3 7.67E+3 0.2 2.07E+6 l.OOE+6 0.3 -

4/6/99 Hydraulic 485 . . 6.67E+6 273E+3 1.4 .


step 500 - - 900E+3 18.0E+3 1.7 .
505 . - 370E+3 14.0E+3 1.4 .
520 - - 130E+3 4.67E+3 1.4 -
(continued)

241
TableB. 11 (continued)
MWD pilot plant filter removal of E. coli, MS-2 and G. lamblia
Type Seed Duration E. coli MS-2 G. lamblia
Date of at Sampling (CFU/L) (PFU/L) (cysts/L)
Experiment (min) FI FE Log FI FE Log FI FE Log
Removal Removal Removal
4/13/99 485 . 2.93E+6 187E+3 1.2 . .
500 - 76.7E+3 3.67E+3 1.3 . -
505 - 120E+3 1.67E+3 1.9 - .
520 - 22.7E+3 l.OOE+3 1.4 - -

4/20/99 485 _ 147E+3 5.33E+3 1.4 _ .


500 - . 16.3E+3 l.OOE+3 1.2 . .
505 - 10.3E+3 667E+0 1.2 . .
520 - 2.50E+3 <1.00E+3 >0.4 - -
ND = None detected
TNTC = Too numerous to count
*** Questionable results

242
Table B. 12
Summary of MWD pilot plant seeding data
Date Type of Log Removals (Mean ± 1 Std. Dev.) Effluent Value (Mean ± 1 Std. Dev )
Experiment C. parvum B. subli/is £ co/i MS-2 G. lamblia Particles Turbiditv Particles
8/4/98 No coagulants. 0.02 ±0.11 -0.03 ± 0.03 0.03 ± 0.08 -0.2 ± 0.01 1.00 ±0.00 7400 ± 0.00
no filter media
8/11/98 No coagulants, 0.27 ± 0.40 0.08 ±0.1 7 0.14 ± 0.03 0.1 ± 0.02 0.88 ± 0.01 2925 ± 126
2/2/99 with filter media 0.07 ±0.13 -0.03 ± 0.05 0.05 ± 0.02 0.02 ± 0.07 0.2 ± 0.02 0.34 ± 0.00 2050 ± 58
5/4/99 0.18 ±0.05 0.03 ± 0.09 0.004 ± 0.02 -0.04 ± 0.05 0.3 ± 0.01 0.46 ± 0.04 2575 ± 50
Overall Average 0.17 ± 0.24 O.OJ ± ft// 0.03 ± O.OJ 0.02 ± 0.09 0.2 ± 0.08 0.45 ± ft/7 2400 ± 374

7/15/98 Stable filter 2.6 ± 0.07 2.0 ±0.13 4.6 ±0.13 2.2 ± 0.09 0.05 ± 0.00 6.3 ± 1.2
7/28/98 operation 3.3 ± 0.07 2.7 ± 0.26 4.5 ± 0.37 2.6 ± 0.04 0.05 ± 0.00 4.5 ± 0.4
8/18/98 4.1 ± 0.65 2.3 ±0.17 1.8 ±0.30 3.4 ± 0.04 0.05 ± 0.00 1.5 ± 0.2
9/22/98 3.8 ± 0.16 1.9 ± 0.46 1.8 ± 0.37 4.4 ± 0.49 2.8 ± 0.01 0.05 ± 0.00 5.2 ± 0.2
9/29/98 3.2 ± 0.37 0.6 ±0.11 1.3 ± 0.06 3.3 ±0.14 0.05 ± 0.00 3.1 ±0.4
10/27/98 3.2 ± 0.15 2.3 ±0.17 0.5 ± 0.20 1.0 ± 0.09 4.2 ± 0.88 2.5 ±0.14 0.05 ± 0.00 10 ± 3.2
1 1/24/98 - 2.1 ±0.86 . 1.1 ±0.11 2.0 ± 0.02 0.05 ± 0.00 32 ± 1.4
12/15/98 2.9 ±0.11 1.8 ± 0.06 0.3 ±0.17 1.1 ±0.06 1.8 ± 0.02 0.05 ± 0.00 37 ± 1.8
2/9/99 2.1 ±0.15 1.9 ± 0.07 0.6 ± 0.25 1.0 ± 0.08 2.2 ± 0.01 0.06 ± 0.00 21 ± 1.7
3/9/99 2.4 ± 0.18 1.9 ±0.04 1.3 ± 0.04 2.9 ± 0.07 0.05 ± 0.00 3.9 ± 0.6
4/27/99 2.9 ± 0.29 1.9 ±0.06 0.5 ± 0.14 1.0 ± 0.05 2.2 ± 0.02 0.05 ± 0.00 27 ± 1.2
Overall Average 3.0 ± 0.66 2.2 ± 0.12 0.1 ± 0.18 1.4 ± 0.44 4.4 ± 0.51 2.5 ± OJO ftftj ± 0.00 13.8 ± 12.6

03/16/99* Stable filter 0.2 ± 0.06 2.2 ± 0.02 0.7 ± 0.12 2.2 ±0.13 2.9 ± 0.05 0.06 ± 0.00 3.4 ± 0.4
05/1 1/99' operation 2.4 ± 0.12 2.2 ± 0.02 . . 2.7 ± 0.03 0.05 ± 0.00 9.5 ± 0.8
Overall Average AJ ± 1.17 2.2 ± 0.02 - • 2.S ± ft/2 0.05 ± ftOO 6.5 ± 3.3

2/16/99 Suboptimal 0.9 ±0.15 0.5 ± 0.05 0.3 ± 0.10 1.0 ±0.01 0.17 ± 0.01 368 ± 17
2/23/99 coagulation 0.8 ±0.21 0.4 ± 0.06 0.2 ± 0.13 0.8 ± 0.06 I.0±0.04 0.16 ± 0.00 405 ± 13
3/2/99 1.7 ±0.17 0.6 ± 0.04 0.4 ± 0.16 0.9 ± 0.07 1.3 ±0.03 0.14 ±0.01 195 ± 13
Overall Average ft« ± ft« 0.5 ± O.OS 0.3 ± 0.14 0.8 ±0.10 /./ ± ft/7 ft/6 ± 0.02 333 ± 96

03/23/99" Suboptimal 0.1 ±0.11 0.3 ± 0.05 0.2 ±0.11 0.3 ± 0.07 0.7 ± 0.02 0.17 ±0.00 425 ± 10
coagulation
10/6/98 Ripening 3.0 ± 0.52 1.6 ± 0.24 0.1 ± 0.10 _ 1.9 ±0.09 0.06 ± 0.00 33 ± 8.5
10/13/98 2.8 ±0.11 1.8 ± 0.23 0.7 ± 0.32 0.3 ± 0.67 3.9 ± 0.58 2.2 ±0.14 0.09 ± 0.05 29 ± 14
10/20/98 3.4 ± 0.28 1.4 ± 0.31 0.2 ± 0.15 1.0 ±0.04 2.3 ± 0.20 0.09 ± 0.06 21 ± 14
11/3/98 2.4 ± 0.51 1.3 ± 0.22 0.4 ±0.15 1.1 ±0.27 2.3 ± 0.23 0.08 ± 0.05 18 ± 12
11/10/98 3.2 ± 1. 10 1.5 ± 0.28 0.4 ± 0.20 1.0 ± 0.20 2.2 ± 0.30 0.08 ± 0.04 30 ±25
Overall Average 2.9 ± ft« 1.5 ± 0.28 0.4 ± 0.27 0.9 ± 0.49 2.2 ± 0.ZJ 0.0« ± 0.04 25 ± /5.J

11/17/98 End-of-run 2.9 ± 0.28 3.0 ± 0.18 1.0 ±0.32 2.1 ±0.13 4.8 ±0.12 2.7 ± 0.04 0.05 ± 0.00 7.9 ± 0.1
12/1/98 2.3 ±0.17 2.7 ± 0.04 . 2.0 ± 0.10 2.2 ± 0.04 0.05 ± 0.00 18 ± 1.5
12/8/98 2.1 ±0.16 2.5 ± 0.41 - 1.8 ± 0.22 2.0 ± 0.05 0.05 ± 0.00 28 ± 2.6
Overall Average 2.4 ± 0.43 2.7 ± 0.30 • 2.0 ± 0.19 2J ± 0.2« 0.05 ± 0.00 /«.0 ± «.P

4/6/99 Hydraulic 1.3 ± 0.66 2.0 ± 0.14 1.5 ±0.14 2.3 ±0.11 0.05 ± 0.00 17 ± 4.9
4/13/99 step 1.9 ± 0.68 2.1 ± 0.66 . 1.4 ± 0.29 2.3 ±0.10 0.05 ± 0.00 18 ± 3.9
4/20/99 2.2 ± 0.99 2.5 ± 0.37 - 1.1 ±0.46 2.5 ± 0.17 0.05 ± 0.00 1 1 ± 4.6
Overall Average /.« ± 0.82 2.2 ± 0.4« - 1.3 ± 0.35 2.J ± 0./5 ftft5 ± 0.00 15 ± 5.2
*Seeded at rapid mix

243
APPENDIX C

Clean Bed Filter Performance - The Johns Hopkins Experiments

245
Modeling Studies

Models have been developed in the literature for the efficiency of packed beds at the start
of a filtration run, for what is called the "clean bed efficiency". This approach is used here to
simulate the initial efficiency of a dual-media filter. Characteristics of this filter are identical to
the filters used in the laboratory-scale experiments at Johns Hopkins.

In these calculations a dual-media filter comprised of 45 cm of anthracite overlying 25


cm of sand is assumed, the same specifications as used in the filtration facility for the
experimental studies. The complete size distribution of the media used in this research has been
measured from the sieve analysis. The anthracite has an effective size of 0.96 mm and a
uniformity coefficient of 1.36; the effective size and uniformity coefficient of the sand are 0.38
mm and 1 .39, respectively.

The efficiency of a clean filter bed can be described by the following relationship (Yao et
al., 1971):

(C.I)

Here CL is the concentration of particles leaving a filter of length L, C0 is the concentration of


particles in the influent to the filter, dm is the size of the media in the bed, and /is the porosity of
the bed. The chemistry of the system is described by a, the sticking factor or attachment
probability that a suspended particle reaching a media grain in the bed will attach to it and be
removed from suspension. For a properly conditioned filter influent, a is expected to be 1 ; when
pretreatment chemistry is less than optimum, a is less than 1 and filter effectiveness is reduced.
Particle transport in the bed is described by TJT , a dimensionless mass transport coefficient
described in more detail subsequently.

Particles in water can be transported from a flowing suspension to the surfaces of the
media in a packed bed filter in three ways: by gravity or sedimentation, by fluid flow alone
(termed interception), and by diffusion or, more precisely, convective diffusion (Yao et al.,
1971). A dimensionless mass transport coefficient (^r) is often used; it is the ratio of the rate at
246
which particles contact a media grain by these three mechanisms to the rate at which they
approach it. The following useful expression describing TJT was developed by Rajagopalan and
Tien (1976).

rjT = 4A?N$ + AS N?0 N? +(3.38xlO-3 )^2 A^'4 (C.2)

Here As, Npe, NLO, NR, and NO are dimensionless groups representing the effects of bed
porosity, convective diffusion, dispersion forces, interception, and gravity, respectively. Details
are presented in Rajagopalan and Tien (1976) and summarized in Tobiason and O'Melia (1988).
The first set of terms on the right-hand-side of Equation C.2 describes transport by convective
diffusion; the second set is based primarily on transport by fluid flow (interception); the third set
refers to particle transport primarily by gravity settling. The dimensionless groups are defined as
follows:

(C.4)

NR = f- (C.6)

(C7)

where dm and dp are the diameters of the media grains and the suspended particles, respectively,
U is the filtration rate, vs is the settling velocity of the suspended particles, ju is the viscosity of
the water, Pp and p are the densities of the particles and the water, respectively, g is the gravity

247
acceleration, H is Hamaker's constant, D is the Brownian diffusion coefficient of the suspended

particles \D = kT/\3&pdp }\,k is Boltzmann's constant, and Tis the absolute temperature.

In the calculations presented in this report, a particle diameter (dp) of 4 um,


representative of Cryptosporidium cysts, is assumed. Other assumptions include/= 0.50 in the
anthracite layer and 0.46 in the sand layer, H = 10"20 J, T= 25 °C and pp = 1.055 g cm"3. Depths
of the anthracite and sand layers are 45 and 25 cm, respectively. Simulations of the effects of
filtration rate and pretreatment effectiveness on the removal efficiency of a clean dual-media bed
have been performed. Results are presented in Figure C. 1 to Figure C.3.

Calculations for the clean bed performance of the dual-media filter under standard
conditions (U=2 gpm/ft2) and with effective pretreatment (a = 1) are presented in Figure C.I.
This condition corresponds to the filtration experiments performed at conventional filtration rate
(U=2 gpm/ft2) with the alum dosages on or above optimum. Media size in the dual-media filter
is presented as a function of bed depth in Figure C.I (a); in these results it is assumed that the
media in the bed stratify by size and density after backwashing. The single collector efficiency
(TJT) calculated using Equation 2 is plotted as a function of bed depth in Figure C.I (b). Based on
these results, the predicted clean bed efficiency of the dual-media bed for removing
Cryptosporidiwn-size particles is plotted using Equation 1 in Figure C.I (c). Under these
conditions, selected for a typical filter operation with effective pretreatment, there are substantial
differences in the size distribution and the single collection efficiencies of the media in the bed
with depth. Nevertheless, most of the removal at the start of the run is accomplished in the
anthracite layer; the contribution of the sand layer to clean bed filter efficiency is
correspondingly small.

Calculations under different conditions yield different results and suggest different
conclusions. The effects of pretreatment chemistry are examined in Figure C.2 (a) and (b). Here
the effects of inadequate pretreatment (as characterized by a value of a = 0.1, indicating that 10
percent of the collisions between suspended particles and filter media result in attachment and
removal) are compared with the standard conventional or baseline case in which a = 1.0
presented previously in Figure C.I. Since pretreatment chemistry does not affect particle

248
transport in the filter bed, the single collector efficiency is not affected by this change (Figure
C.I (b) and Figure C.2 (a) are the same). However, inadequate pretreatment results in poor
performance of the anthracite layer, so that a significant concentration of particles penetrates
through this layer and reached the smaller sand layer (Figure C.2 (b)). Because of the smaller
size of the media in this layer, mass transport is sufficiently effective to compensate for poor
particle attachment, and the overall particle removal of the dual-media bed, while less than when
pretreatment is effective, is still predicted to be substantial. These results are confirmed by the
experimental studies in which the lower sand layer functioned as a multiple barrier when the
performance of the upper anthracite layer was inadequate. The sand layer, as indicated by
turbidity results, enabled effective removal of particles during the ripening and breakthrough
period while significant amount of particles passed through the anthracite layer.

Some effects of filtration rate on filter performance are presented in Figure C.2 (c) and
(d). Calculations are made for U = 6 gpm/ft2 and compared to the standard case with U = 2
gpm/ft2. Calculations are presented for both effective and inadequate pretreatment (a = 1.0 and
0.1, respectively). Because particle transport by sedimentation and by convective diffusion
depend strongly on filtration rate (Equations 2, 4, and 7) and transport by interception also has a
dependence on rate, although weak (Equations 2 and 5), the single collector efficiency at 6
gpm/ft2 is smaller than at 2 gpm/ft2 (compare Figure C.2 (a) and (c)). As a result, the sand layer
is expected to provide significant benefits when pretreatment is effective (compare Figure C.2
(b) and d at a = 1) as well as when pretreatment is inadequate (Figure C.2 (b) and (d) at a = 0.1).

Additional calculations for the effects of filtration rate on clean bed performance are
presented in Figure C.3. Calculations of the single collector efficiency as a function of depth
(Figure C.3 (a)) indicate, as expected, that rjr decreases as filtration rate increases. These effects
result in an increased loading of particles to the clean sand layer as U is increased (Figure C.3
(b)). Regardless of the filtration rate used, however, the clean bed removal efficiency of this
dual-media filter is very good. The experimental results from filtration studies conducted at both
conventional (4.9 m/h [2 gpm/ft2]) and high (14.7 m/h [6 gpm/ft2]) filtration rates are identical to
the model simulation results. The dual-media effluent turbidity and particle concentrations at
high filtration rate experiments were slightly higher when compared to conventional filtration
rate experiments, but still adequate.
249
The results of the calculations presented in Figure C.I to Figure C.3 indicate that
conventional mono-media filters can operate effectively at low filtration rates at the start of a
filtration run when pretreatment is effective and that a dual-media filter provides an effective
multiple barrier if pretreatment is ineffective. Conventional mono-media beds may have
inadequate performance at the start of a filter run when operated at high rates. Here again, a dual-
media filter provides improved filter performance when the filter is brought on line after
backwashing.

Summary

The modeling results for the initial or clean bed removal efficiency of a dual-media bed
indicate that these beds can function effectively at the start of a run at standard conditions (low
filtration rate) when chemical pretreatment is effective. Even under these conditions, the upper
anthracite layer provides inadequate removal, so that the lower sand layer acts as a multiple
barrier, proving efficient removal at the start of the run. Suboptimal pretreatment chemistry
reduces the clean bed removal efficiency of the dual-media beds. Under these conditions,
substantial penetration of the anthracite layer is predicted, even at low filtration rates. The sand
layer, with its smaller media size, again provides an effective multiple barrier at the start of the
run. Similar effects are predicted for the effects of filtration rate. When operated at higher
filtration rates, the clean bed removals of the anthracite layer are reduced and the lower sand
layer provides a larger contribution to the performance of the bed. All model simulation results
are confirmed and well agreed by the results from the filtration experiments performed at
identical conditions.

250
0.001 0.002 0.003 0.004 0.5
0 0 0

10 10 10 10 10 10

20 20 20 20 20 20

30 30 30 30 30 30

•s
D.
40 40 Q 40 40 ° 40 , ; 40
to

50 50 50 50 50 «. 50

60 60 60 • 60 60 <> 60
*

*
70 70 -*—
70 i——;——*————'————'————' 70 70 70
0 0.001 0.002 0.003 0.004 0 0.5
Media Size (mm) Single Collector Efficiency Filtrate Quality (C/C0)

(a) (b) (c)

Figure C. 1 Clean bed performance of the dual-media filter under standard conditions, U = 2gpm/ft2, a = 1.0. (a) Media size

(mm), (b) Single collector efficiency, (c) Filter efficiency(C/Co).


0 0.002 0.004 0 0.5 1 0 0.002 0.004 0 0.5 1
0 0 0 0 0 0 0
* 0 ° ** * 0 6 *
* o » * 0 »

? * 10 10 * 10 10 o » 10
10 . * 10 10
» 1 > * * o »
^ () • 4 3• •
20 20 20 20 20 - 20 20 20
* () • * 3 *
() • • (
^ > •
^ 30 30 _ 30 30 30 30 "s" 30 ' - • 30
• ^ () • * > *
U ^f 1 ,",
^ •s <, ^ •3 ^ •5 (( ^
&
« &
Q 40 -* 40 ° 40 ( ) » 40 ° 40 * 40 ° 40 ( > • 40

* * I ** » ^ • * **
to
* O •
* V3 * * •
50 * 50 50 < 50 50 * 50 50 fS* 50
• <> *
* • r»
+ 4 c>
60 60 60 - 60 60 60 60
* • 60 ( >
* 0 0=1.0 » 1> 0 0=1.0
* * 1^
* 0=0.1
1>
7n ———*——i—————— 7O 70 ( 70 70 ——*————i———————— 70 70 1 70
0 0.002 0.004 0 0.5 1 0 0.002 0.004 0 0.5 1
Single Collector Efficiency Filtrate Quality (C/C0) Single Collector Efficiency Filtrate Quality (C/C0)

(a) (b) (c) (d)


Figure C.2 Effects of pretreatment chemistry and filtration rate on clean bed filter performance, (a) Single collector efficiency, U =
2 gpm/ft2. (b) Filter efficiency(C/C0), U = 2 gpm/ft2. (c) Single collector efficiency, U = 6 gpm/ft2. (d) Filter efficiency(C/C0), U = 6
gpm/ft2.
0 0.001 0.002 0.003 0.004 o 0.2 0.4 0.6 0.8 1
0 |————————————————————————— nv v

cfcP n co
On a 2 gpm/ft2 n *o
• 6 gpm/ft2
10
O 10 gpm/ft2 10 10 O ^O 10
on MO

on »
20 20 20 20
on ID

on I3
30 30 30 30
on f
>
1
•a on •5 |3
° 40 on 40 40 , ) 40

on 0*^ n
o* n L?
¥
50 L o* n 50 50 ti 50
o> n
o» n I
o» a ?
60 on 60 60 n n 2 gpm/ft2 60
O^ Q 6 » 6 gpm/ft2
C^ CI ¥ O 10 gpm/ft2
o n
7 ft f^ n i i in in ph , i i 70
0 0.001 0.002 0.003 0.004 0 0.2 0.4 0.6 0.8 1
Single Collector Efficiency Filtrate Quality (C/C0)

(a) (b)

Figure C.3 Effects of filtration rate on clean bed filter efficiency, a = 1.0. (a) Single collector
efficiency, (b) Filter efficiency(C/Co).

253
APPENDIX D

Detailed Test Results - The Johns Hopkins Experiments

255
8II f«
— P
ih K>

QQ ,_ Settled UV254 Absorbance (cm'1 )


e $

o O
!-*5 >-+>
CO CO
s an>
£ a-

B.*
§:
8-
a
T3
ffi
I
o
CD

•a I

n>
»-t
2"
3-
i—*•
CL.

I
I \»CJ
II
-J o
o
n
V. lo

0.16 X ° O 9.0 mg/L


0.14 . x °° X 13.5 mg/L
x oo 00° °0 °oo o 0
o 0.12
o o ° o o
^ O O
8 0.10 x jp° (
1 0.08 var

a 0.06
;• x x x >
^ 0.04
0
X
0.02
<<A
nnn r* „. , ... __i, . . ——1_, _.., . , ... .. , ... , ,.
0.0 0.5 1.0 1.5 2.0 2.5 3.0 3.5 4.0

Time (hour)

Figure D.3 UVas4 absorbance of the dual-media effluents during a filter run (direct filtration
mode, filtration rate = 4.9m/h [2 gpm/ft2]).

0.8

0.7 * Anthracite
O Filtrate ** *
^ 0.6 » **

2 0.5

0.4

0.3

D 0.2

0.1
0 o 06 0 0 o ooo ooo oooooooo
0.0
9 12 15 18 21
Time (hour)

Figure D.4 UV254 absorbance of the anthracite and dual-media effluents during a filter run at
an alum dosage of 13.5 mg/L (direct filtration mode, filtration rate = 4.9 m/h [2 gpm/ft2]).

257
u.y

0.8
^»** * * » 6.8 mg/L
6 0.7 ^* » ^ » O 9.0 mg/L
o
o - A 11.3 mg/L
t^ff 0.6 ™
8 * X 13.5 mg/L
0.5
|
0.4
\
D 0.3 •

1 °'2 .
u.
•23

01 ^^^W^^B'Jl
&»«««* J? *."A K*,» .
>C A 8 A X
* X
*** X )
0.0 f————————'————————'————————J-————————'————————i————————1
0 2 4 6 8 10 12
Time (hour)

Figure D.5 Effect of alum dosage on the removal of UV254 absorbance by anthracite layer
during a filter run (direct filtration mode, filtration rate = 14.7 m/h [6 gpm/ft2]).

0.9 ————————————————————————————————————————————————————
ns
U.O

^ *» * + » 6.8 mg/L
gy
0.7 »** *
^
O 9.0 mg/L
O %* • All.3mg^
0.6
8 A X 13.5 mg/L
0.5
o •
1 0.4 • *
2 f
-
S 0.3
1 0.2 .
E r
o.i Ife._ x » x -
jJfflSY.Itfi A^°A2 ^A i A X X * * *
0.0 f————————i————————•———————j.———————i————————i————————
0246 8 10 12
Time (hour)

Figure D.6 Effect of alum dosage on the removal of UV254 absorbance by the entire dual-
media bed during a filter run (direct filtration mode, filtration rate = 14.7 m/h [6 gpm/ft2]).

258
10 12
Time (hour)

Figure D.7 Effect of alum dosage on the removal of DOC by anthracite layer during a filter
run (direct filtration mode, filtration rate = 14.7 m/h [6 gpm/ft2]).

1.2

1.0
** *** » » » 6.8 mg/L
O 9.0 mg/L
A 11.3 mg/L
0.8
X 13.5 mg/L

I 0.6
8Q
0.4

0.2 •*«

0.0
4 6 10 12
Time (hour)

Figure D.8 Effect of alum dosage on the removal of DOC by the entire dual-media bed during
a filter run (direct filtration mode, filtration rate = 14.7 m/h [6 gpm/ft2]).

259
Alum Dosage (mg/L)

Figure D.9 Jar test results for the effects of pH on settled turbidity (raw water: turbidity = 7
NTU, DOC =1.5 mg/L).

9 12 15
Alum Dosage (mg/L)

Figure D. 10 Jar test results for the effects of pH on filtered turbidity (raw water: turbidity = 7
NTU, DOC =1.5 mg/L).

260
0.140

6 9 12 15 18
Alum Dosage (mg/L)

Figure D.I 1 Jar test results for the effects of pH on settled UV254 (raw water: turbidity = 7
NTU,DOC=1.5mg/L).

0.140

6 9 12 15 18
Alum Dosage (mg/L)

Figure D. 12 Jar test results for the effects of pH on filtered UV254 (raw water: turbidity = 7
NTU,DOC= 1.5mg/L).

261
l.UU

0.90

0.80
E X * A
o 0.70 A^
^0 O alum = 6.8 mg/L, pH 6
y 0.60
J A alum = 9.0 mg/L, pH 8
o 0.50 t
1
3 0.40
•o 0.30
a 0.20 ^£0
E 0»vC$. OO ^ ^ ^ OO ^6 o o
0.10 > * o o o o o o o °«0<> o ° -
nnn 1
9 12 15 18 21 24
Time (hour)

Figure D.I 3 Effects of pH on the removal of UV254 absorbance by anthracite layer during a
filter run (direct filtration mode, filtration rate = 4.9 m/h [2 gpm/ft2]).

M . 4 . —— ——— • —— v-
V.7U

0.80

0.70
O
0
0.60 1 O alum = 6.8 mg/L, pH 6

i
TO
A
0.50 -
A alum = 9.0 mg/L, pH 8
|s
XI
A
< 0.40
S 4
D 0.30

1 0.20 r
£ k

f^So °°ooooooooooooo oooooo ooo <


0.00 •——————i——————i——————i——————i——————i——————i——————i——————
9 12 15 18 21 24
Time (hour)

Figure D. 14 Effects of pH on the removal of UV254 absorbance by the entire dual-media bed
during a filter run (direct filtration mode, filtration rate = 4.9 m/h [2 gpm/ft2]).

262
1.8

1.6

1.4
A A
1.2 A A A
"a, i.o A A A

0.6

0.4
O alum = 6.8 mg/L, pH 6
0.2 1
A alum = 9.0 mg/L, pH 8
0.0
0 9 12 15 18 21 24
Time (hour)

Figure D. 15 Effects of pH on the removal of DOC by anthracite layer during a filter run (direct
filtration mode, filtration rate = 4.9 m/h [2 gpm/ft2]).

A * A
A * A A,
A A
AA'

000°0

O alum = 6.8 mg/L, pH 6


A alum = 9.0 mg/L, pH 8

9 12 15 18 21 24
Time (hour)

Figure D. 16 Effects of pH on the removal of DOC by the entire dual-media bed during a filter
run (direct filtration mode, filtration rate = 4.9 m/h [2 gpm/ft2]).

263
10 20 30 40 50 60
Alum Dosage (mg/L)

Figure D. 17 Jar test results for the effects of alum dosage on settled turbidity (raw water:
turbidity = 7 NTU, DOC = 5 mg/L).

10 20 30 40 50 60
Alum Dosage (mg/L)

Figure D. 18 Jar test results for the effects of alum dosage on settled UV254 (raw water: turbidity
= 7 NTU, DOC = 5 mg/L).

264
10 20 30 40 50 60
Alum Dosage (mg/L)

Figure D. 19 Jar test results for the effects of alum dosage on filtered turbidity (raw water:
turbidity = 7 NTU, DOC = 5 mg/L).

10 20 30 40 60
Alum Dosage (mg/L)

Figure D.20 Jar test results for the effects of alum dosage on filtered UV254 (raw water:
turbidity = 7 NTU, DOC = 5 mg/L).

265
20 30 40 50 60
Alum Dosage (tng/L)

Figure D.21 Jar test results for the effects of alum dosage on dissolved organic carbon (raw
water: turbidity = 7 NTU, DOC = 5 mg/L).

U.Jl/

0.45 « * 27 mg/L .
/^
0.40 •^ * * * * 36 rag^
o t* * * * * * * * * A45mg/L
o 0.35 '^^
***f
8 0.30 ? *.*.**
*
1 0.25
3 0.20 .

| 0.15 £«Wte> ooo o°oo°o o ^ooooo ooooooo oooooo<


£ 0.10
E
0.05 -
o.on 1
3 6 9 12 15 18 21 24 27
Time (hour)

Figure D.22 Effects of alum dosage on the removal of UV254 absorbance by anthracite layer
during a filter run (conventional treatment mode, filtration rate = 4.9 m/h [2 gpm/ft2]).
266
»27 mg/L
0.40

0.35
•»V*« * *
O 36 mg/L
A 45 mg/L
* *
0.30

0.25

0.20

0.15 ooo o ° * o * o o o o o o o o oo<>oooo oo<>ooo


A A A AAAAAAAAAAA AAAAAAA,,
0.10

0.05

0.00
12 15 18 21 24 27
Time (hour)

Figure D.23 Effects of alum dosage on the removal of UV254 absorbance by the entire dual-
media bed during a filter run (conventional treatment mode, filtration rate = 4.9 m/h [2 gpm/ft2]).

i.o
0.9 * 27 mg/L
0.8

0.7 **' * *

I 0.6
£ 0.5
**o
O

A O
o * .

o
* • »
* •

o0<> o
o
8 0.4

0.3
A A A
0.2 A A A
«'O
0.1

0.0
0 3 6 9 12 15 18 21 24 27
Time (hour)

Figure D.24 Effects of alum dosage on the removal of DOC by anthracite layer during a filter
run (conventional treatment mode, filtration rate = 4.9 m/h [2 gpm/ft2]).

267
* 27 mg/L
O36mg/L
A 45 mg/L

o o
o o o o
A °
A A A A

0 3 6 9 12 15 18 21 24 27
Time (hour)

Figure D.25 Effects of alum dosage on the removal of DOC by the entire dual-media bed
during a filter run (conventional treatment mode, filtration rate = 4.9 m/h [2 gpm/ft2]).

0.25

^ 0.20
o * O O

1 0.15

<
A A A A A A
* 0.10

I 0.05 O 36 mg/L
U<
A 45 mg/L

0.00
0123456
Time (hour)

Figure D.26 Effects of alum dosage on the removal of UV254 absorbance by anthracite layer
during a filter run (conventional treatment mode, filtration rate = 14.7 m/h [6 gpm/ft2]).
268
0.18
o o <
0.16 0
o
o 0.14 ? •...•••
I•e 0.12 pyi>lttAAAA*A
0.10 •
* A A A A A A A \
1
s 0.08 k
D 0.06 *
is 0.04 . O Jo mgL
E
0.02 A45mgl
0.00 ' k ,.,..,,,
() 1 2 3 4 5 6 7 8 S
Time (hour)

Figure D.27 Effects of alum dosage on the removal of UV254 absorbance by the entire dual-
media bed during a filter run (conventional treatment mode, filtration rate = 14.7 m/h [6
gpm/ft2]).

1.2

O O
1.0
!> O
0.8

t 0.6 A A
A A A A
0.4

0.2 O36 mg/L


A 45 mg/L
0.0
0 1 4 5 8
Time (hour)

Figure D.28 Effects of alum dosage on the removal of DOC by anthracite layer during a filter
run (conventional treatment mode, filtration rate = 14.7 m/h [6 gpm/ft2]).

269
o

A o
A

0.2 O36mg/L

0.0
4 5
Time (hour)

Figure D.29 Effects of alum dosage on the removal of DOC by the entire dual-media bed
during a filter run (conventional treatment mode, filtration rate = 14.7 m/h [6 gpm/ft2]).

270
REFERENCES

Adams, M.H. 1959. Bacteriophages. New York: Interscience Publishers, Inc.

Al-Ani, M.Y., D.W. Hendricks, G.S. Logsdon, and C.P. Hibler. 1986. Removing Giardia Cysts
from Low Turbidity Waters by Rapid Rate Filtration. Jour. AWWA, 78(5):66-73.

Amirtharajah, A. 1985. The Interface Between Filtration and Backwashing. Wat. Res.,
19(5):581-588.

Amirtharajah, A. 1988. Some Theoretical and Conceptual Views of Filtration. Jour. AWWA,
80(12):36-46.

Amirtharajah, A. 1993. Optimum Backwashing of Filters with Air Scour: A Review. Water Sci.
Tech., 27(10): 195-211.

Amirtharajah, A., N. McNelly, G. Page, and J. McLeod. 1991. Optimum Backwash of Dual-
media Filters and GAC Filter Adsorbers with Air Scour. Denver, Colo.: AwwaRF and AWWA.

Amirtharajah, A., and D.P. Wetstein. 1980. Initial Degradation of Effluent Quality During
Filtration. Jour. AWWA, 72(9):518-524.

APHA, AWWA, WEF (American Public Health Association, American Water Works
Association, and Water Environment Federation). 1995. Standard Methods for the Examination
of Water and Wastewater. 19th ed. Washington, D.C.: APHA.

Arora, H., M.W. LeChevallier, and W.D. Norton. 1995. The Relationship Between Particle
Counts and Parasites: Full-Scale Experiences in Proc. AWWA Annual Conference. Denver,
Colo.: AWWA.

ASTM (American Society for Testing and Materials). 1993. Standards on Materials and
Environmental Microbiology, 2nd Edition. Philadelphia, PA.: American Society for Testing and
Materials.

Baudin, I., and J.-M. Lame. 1998. Assessment and Optimization of Clarification Process for
Cryptosporidium Removal in Proc. AWWA Water Quality Technology Conference. Denver,
Colo.: AWWA.

Becker, William C. 1995. The Impact of Ozonation on Coagulation: Model and Natural Water
Studies. Ph.D. Dissertation. The Johns Hopkins University, Baltimore, MD.

Billica, J.A., D. Gregory, and K.R. Gertig. 1999. A Pilot Plant Study to Compare the
Performance of Dual and Mixed Media Declining Rate Filters in Proc. AWWA Annual
Conference. Denver, Colo.: AWWA.

271
Black, A.P., F.B. Birkner, and J.J. Morgan, 1965. Destabilization of Dilute Clay Suspensions
with Labeled Polymers. Jour. AWWA, 57(12): 1547-1560.

Bukhari, Z. T.M. Hargy, J.R. Bolton, B. Dussert, and J.L. Clancy, 1998. Medium-Pressure UV
for Oocyst Inactivation. Jour. AWWA 91(3):86-94.

Campbell, A.T., L.J. Robertson, M.R. Snowball, and H.V. Smith. 1995. Inactivation of Oocysts
of Cryptosporidium parvum by Ultraviolet Irradiation. Wat. Res., 29(11):2583-2586.

Charles, G., J.M. Morgan, M. MacPhee, M. Kim, and D.W. Frederickson. 1995. Bench-scale
Parasite Spiking - An Alternative to Pilot-scale Giardia and Cryptosporidium Spiking
Investigations va.Proc.AWWA Water Quality Technology Conference. Denver, Colo.: AWWA.

Clancy, J.L., M.M. Marshall, I.E. Dyksen, E. Kawczynski, and K. Cams. 1996. Innovative
Electrotechnologies for the Inactivation of Cryptosporidium parvum Oocysts in Water in Proc.
AWWA Water Quality Technology Conference. Denver, Colo.: AWWA.

Cleasby, J.L., M.M. Williamson, and E.R. Baumann. 1963. Effect of Filtration Rate Changes on
Quality. Jour. AWWA, 55(6):869-881.

Coallier, J., M. Prevost, B. Barbeau, L. Boulos, and R. Desjardins. 1996. Full-Scale Physical and
Chemical Removals of a Fast Response and Economic Microbiological Indicator in Proc.
AWWA Water Quality Technology Conference. Denver, Colo.: AWWA.

Coffey, B.M., S. Liang, J.F. Green, and P.M. Huck. 1998. Quantifying Performance and
Robustness of Filters During Non-Steady State and Perturbed Conditions in Proc. AWWA Water
Quality Technology Conference. Denver, Colo.: AWWA.

Colton, J.F., P. Hillis, and C.S.B. Fitzpatrick. 1996. Filter Backwash and Start-Up Strategies for
Enhanced Paniculate Removal. Wat. Res. 30(10):2502-2507.

Conley, W.R. 1972. High Rate Filtration. Jour. AWWA, 64(3):203-206.

Conley, W.R., and K.Y. Hsiung. 1969. Test Program for Filter Evaluation at Hanford. Jour.
AWWA,52(2):2Q5.

Conley, W.R., and R.W. Pitman. 1960. Design and Application of Multimedia Filters. Jour.
AWWA,6l(2):91.

DiBernardo, L., and J.L. Cleasby. 1980. Declining-Rate Versus Constant-Rate Filtration. Jour.
Envir. Engrg., 106(EE6): 1023-1041.

Dugan, N.R., K.R. Fox, RJ. Miltner, D.A. Lytle, D.J. Williams, C.J. Parrett, C.M. Feld, and J.H.
Owens. 1999. Control of Cryptosporidium Oocysts by Steady-State Conventional Treatment in
Proc. AWWA Annual Conference. Denver, Colo.: AWWA.

272
Edzwald, J.K., and M.B. Kelley. 1998. Control of Cryptosporidium: From Reservoirs to
Clarifiers. Wat. Sci. TechnoL, 37(2): 1-8.

Edzwald, J.K., M.B. Kelley, H.J. Dunn, G.S. Kaminski, and J.P. Malley. 1996. Control of
Cryptosporidium by Coagulation, Flotation, and Filtration in Proc. AWWA Water Quality
Technology Conference. Denver, Colo.: AWWA.

Edzwald, J.K., J.E. Tobiason, L.M. Parento, M.B. Kelley, G.S. Kaminski, H.J. Dunn, and P.B.
Galant. 1999. Clarification and Filtration Performance for Removals of Giardia and
Cryptosporidium in Proc. AWWA Water Quality Technology Conference. Denver, Colo.:
AWWA.

Elimelech, M. 1991. Kinetics of Capture of Colloidal Particles in Packed Beds Under Attractive
Double Layer Interactions. Jour. Colloid Inter. Sci. , 146(2):337-352.

Emelko, M.B. 2001. Removal of Cryptosporidium parvum by Granular Media Filtration. Ph.D.
diss., Univ. of Waterloo, Waterloo, Ontario, Canada.

Emelko, M.B., P.M. Huck, and R.M. Slawson. 1999. Design and Operational Strategies for
Optimizing Cryptosporidium Removal by Filters in Proc. AWWA Water Quality Technology
Conference. Denver, Colo.: AWWA. •

Ferguson, D.W., M.J. McGuire, B. Koch, R.L. Wolfe, and E.M. Aieta. 1990. Comparing
PEROXONE and Ozone for Controlling Taste and Odor Compounds, Disinfection By-Products,
and Microorganisms. Jour. AWWA, 82(4):181-191.

Finch, G.R., E.K. Black, and L.L. Gytirek. 1994. Ozone and Chlorine Inactivation of
Cryptosporidium in Proc. AWWA Water Quality Technology Conference. Denver, Colo.:
AWWA.

Fitzpatrick, C.S.B., A.T. Campbell, and C.J. Cable. 1999. The Effect of Filtration Rate Changes
on Filtrate Particle Counts in Proc. AWWA Water Quality Technology Conference. Denver,
Colo.: AWWA.

Fox, K.R., N.R. Dugan, R.J. Miltner, D.A. Lytle, D.J. Williams, C.J. Parrett, C.M. Feld, J.H.
Owens, E.W. Rice, and C.H. Johnson. 1998. Comparative Removal of Cryptosporidium and
Surrogates in a Low Flow Pilot Plant System in Proc. AWWA Water Quality Technology
Conference. Denver, Colo.: AWWA.

Fox, K.R., and D.A. Lytle. 1996. Milwaukee's Crypto Outbreak: Investigation and
Recommendations. Jour. AWWA, 88(9):87-94.

Fuller, R.G., J. Steiner, and R. Butcher. 1995. Evaluation of Filter Performance to Remove
Giardia and Cryptosporidium City of Billings, Montana in Proc. A WWA Annual Conference.
Denver, Colo.: AWWA.

273
Ginn, T.M. Jr., A. Amirtharajah, and P.R. Karr. 1992. Effects of Particle Detachment in Granular
Media Filtration. Jour. AWWA, 84(2):66-76.

Gyiirek, L.L., L.R.J. Liyanage, M. Belosevic, and G.R. Finch. 1996. Disinfection of
Cryptosporidium parvum Using Single and Sequential Application of Ozone and Chlorine
Species in Proc. AWWA Water Quality Technology Conference. Denver, Colo.: AWWA.

Habibian, M.T., and C.R. O'Melia. 1975. Particles, Polymers, and Performance in Filtration.
Jour. Envir. Engrg., 101:567-583.

Hall, T., and B. Croll. 1996. The UK Approach to Cryptosporidium Control in Water Treatment
in Proc. AWWA Water Quality Technology Conference. Denver, Colo.: AWWA.

Hall, T., J. Pressdee, R. Gregory, and K. Murray. 1995. Cryptosporidium Removal During Water
Treatment Using Dissolved Air Flotation. Water Sci. Tech. 31(3-4): 125-135.

Hargesheimer, E.E., C.M. Lewis, and C.M. Yentsch. 1992. Evaluation of Particle Counting as a
Measure of Treatment Plant Performance. American Water Works Association Research
Foundation and AWWA, Denver, CO.

Hilmoe, D.J., and J.L. Cleasby. 1986. Comparing Constant-Rate and Declining-Rate Direct
Filtration of a Surface Water. Jour. AWWA, 78(12):26-34.

Horn, J.B., D.W. Hendricks, J.M. Scanlan, L.T. Rozelle, and W.C. Trnka, 1988. Removing
Giardia Cysts and Other Particles from Low Turbidity Waters Using Dual-Stage Filtration. Jour.
AWWA, 80:2:68.

Hudson, H.E. 1959. Declining-Rate Filtration. Jour. AWWA, 51(11): 1455-1463.

Huck, P.M., B.M. Coffey, M.B. Emelko and C.R. O'Melia, 2000. The Importance of
Coagulation for Removal of Cryptosporidium and Surrogates by Filtration. Proceedings, 9th
International Gothenburg Symposium on Chemical Treatment, Instanbul, Turkey pp. 191-200.
ISBN 3-540-67574-4.

Ives, K.J. 1982. Fundamentals of Filtration in Proc. Symposium on Water Filtration, European
Federation of Chemical Engineers, Antwerp, Belgium.

Ives, K.J., and I. Sholji. 1965. Research on Variables Affecting Filtration. Jour. Envir. Engrg. -
A5CE,91(SA4):1-18.

Jacangelo, J.G., S. Adham, and J.M. Laine. 1997. Membrane Filtration for Microbial Removal.
Denver, Colo.: AwwaRF and AWWA.

Jakubowski, W., S. Boutros, W. Faber, R. Payer, W. Ghiorse, M. LeChevallier, J. Rose, S.


Schaub, A. Singh, and M. Stewart. 1996. Environmental Methods for Cryptosporidium. Jour.
AWWA, 88(9):107-121.

274
Kelley, M.B., P.K. Warner, J.K. Brokaw, K.L. Barrett and SJ. Komisar. 1995. A Study of Two
U.S. Army Installation Drinking Water Sources and Treatment Systems for the Removal of
Giardia and Cryptosporidium in Proc. AWWA Water Quality Technology Conference. Denver,
Colo.: AWWA. :

Korich, D.G., J.R. Mead, M.S. Madore, N.A. Sinclair, and C.R. Sterling. 1990. Effects of Ozone,
Chlorine Dioxide, Chlorine, and Monochloramine on Cryptosporidium parvum Oocyst Viability.
Appl.andEnvir.Microbiol.,56(5y.U23-l428.

Lawler, D.F., J.L. Darby, and R.S. Gushing. 1995. Complexities of Filtration Dynamics:
Particles, Floes, and (Perhaps) Cysts in Proc. AWWA Annual Conference. Denver, Colo.:
AWWA.

LeChevallier, M.W., and W.D. Norton. 1992. Examining Relationships Between Particle Counts
and Giardia, Cryptosporidium, and Turbidity. Jour. AWWA, 84(12):54-60.

LeChevallier, M.W., W.D Norton, and R.G. Lee. 1991a. Giardia and Cryptosporidium spp. in
Filtered Drinking Water Supplies. Appl. and Envir. Microbiol., 57(9):2617-2621.

LeChevallier, M.W., W.D. Norton, and J.B. Rose. 1991b. Giardia and Cryptosporidium in
Water Supplies. Denver, Colo.: AwwaRF and AWWA.

Li, S.Y., J.A. Goodrich, J.H. Owens, G.E. Willeke, F.W. Schaefer III, and R.M. Clark. 1997.
Reliability of Surrogates for Determining Cryptosporidium Removal. Jour. AWWA, 89(5):90-99.

Logsdon, G.S., V.C. Thurman, E.S. Frindt, and J.G. Stoecker, 1985. Evaluating Sedimentation
and Various Filter Media for Removal of Giardia Cysts. Jour. AWWA, 77:2:61.

Logsdon, G.S. 1999. Effective Management and Operation of Coagulation and Filtration in Proc.
of the Seventh International Conference of the Israel Society for Ecology and Environmental
Quality Sciences, Jerusalem, Israel.

Logsdon, G.S., J.M. Symons, R.L. Hoye Jr., and M.M. Arozarena. 1981. Alternate Filtration
Methods for Removal of Giardia Cysts and Cyst Models. Jour. AWWA, 73(2): 111-118.

Lytle, D.A., and K.R. Fox. 1994. Particle Counting and Zeta Potential Measurements for
Optimizing Filtration Treatment Performance in Proc. AWWA Water Quality Technology
Conference. Denver, Colo.: AWWA.

Lytle, D.A., K.R. Fox, E.W. Rice, J. Owens, and C.H. Johnson. 1996. The Use of Aerobic Spore
Forming Bacteria for Evaluating Drinking Water Treatment Performance in Proc. AWWA
Annual Conference. Denver, Colo.: AWWA.

McDowell-Boyer, L.M. 1992. Chemical Mobilization of Micron-Sized Particles in Saturated


Porous Media Under Steady Flow Conditions. Environ. Sci. Technol., 26(3):586-593.

275
Montgomery, J.M., Consulting Engineers, Inc. 1985. Water Treatment Principles and Design.
New York: John Wiley & Sons.

Moran, D.C., M.C. Moran, R.S. Gushing, and D.F. Lawler. 1993a. Particle Behavior in Deep-
Bed Filtration: Part 1 - Ripening and Breakthrough. Jour. AWWA, 85:12:69-81.

Moran, M.C., D.C. Moran, R.S. Gushing, and D.F. Lawler. 1993b. Particle Behavior in Deep-
Bed Filtration: Part 2 - Particle Detachment. Jour. AWWA, 85(12):82-93.

Nieminski, E.G. 1994. Removing Giardia and Cryptosporidium Cysts Removal Through Direct
Filtration and Conventional Treatment in Proc. AWWA Annual Conference. Denver, Colo.:
AWWA.

Nieminski, E.G., and J.E. Ongerth. 1995. Removing Giardia and Cryptosporidium by
Conventional Treatment and Direct Filtration. Jour. AWWA, 87(9):96-106.

Nieminski, E.G., and W.D. Bellamy. 1998. Application of Pathogen Surrogates to Improve
Treatment Plant Performance in Proc. AWWA Water Quality Technology Conference. Denver,
Colo.: AWWA.

O'Melia, C.R. 1985. Particles, Pretreatment, and Performance in Water Filtration. Jour. Envir.
Engrg., 111(6):874-890.

O'Melia, C.R., and W. Ali. 1978. The Role of Retained Particles in Deep Filtration. Progr in
Water TechnoL, London, U.K., 10:(5/6): 167-182.

O'Melia, C.R., and D.K. Crapps. 1964. Some Chemical Aspects of Rapid Sand Filtration. Jour.
AWWA, 56(10)1326.

O'Melia, C.R., and W. Stumm. 1967. Theory of Water Filtration. Jour. AWWA, 59(11): 1393-
1412.

Ongerth, I.E., and J.P. Pecoraro. 1995. Removing Cryptosporidium Using Multimedia Filters.
Jour. AWWA, 87(12):83-89.

Ongerth, J.E., J. Riggs, and J. Crook. 1989. A Study of Water Treatment Practices for the
Removal of Giardia lamblia Cysts. Denver, CO.: AwwaRF and AWWA.

Oppenheimer, J.A., E.M. Aieta, R.R. Trussell, J.G. Jacangelo and I.N. Najm, 2000. Evaluation of
Cryptosporidium Inactivation in Natural Waters. Denver, CO.: AwwaRF and AWWA.

Owens, J.H., R.J. Miltner, F.W. Schaefer, and E.W. Rice. 1994. Pilot-scale Inactivation of
Cryptosporidium and Giardia in Proc. AWWA Water Quality Technology Conference. Denver,
Colo.: AWWA.

276
Patania, N.L., J.G. Jacangelo, L. Cummings, A. Wilczak, K. Riley, and J. Oppenheimer. 1995.
Optimization of Filtration for Cyst Removal. Denver, Colo.: AwwaRF and AWWA.

Plummer, J.D., J.K. Edzwald, and M.B. Kelley. 1995. Removing Cryptosporidium by Dissolved-
Air Flotation. Jour. AWWA, 87(9):85-95.

Proceedings of AWWA 2000 Water Quality Technology Conference. Copyright ©2000,


American Water Works Association

Rajagopalan, R., and C. Tien. 1976. Trajectory Analysis of Deep-Bed Filtration with the Sphere-
in-Cell Porous Media Model. Jour. AIChE, 22(3):523-533.

Raveendran, P., and A. Amirtharajah. 1995. Role of Short-Range Forces in Particle Detachment
During Filter Backwashing. Jour. Envir. Engrg., 121(12):860-868.

Rice, E.W., K.R. Fow, R.J. Miltner, D.A. Lytle, and C.H. Johnson. 1996. Evaluating Plant
Performance with Endospores. Jour. AWWA, 88(9): 122-130.

Rimer, A.E. 1968. Filtration Through a Tri-Media Filter. Jour. Envir. Engrg., 94(3):521-540.

Roebeck, <3.G., K.A. Dostal, and R.L. Woodward. 1964. Studies of Modifications in Water
Filtration. Jour. AWWA, 56:198.

Ryan, J.N., and P.M. Gschwend. 1994. Effects of Ionic Strength and Flow Rate on Colloid
Release: Relating Kinetics to Interface Potential Energy. Jour. Colloid and Interface ScL,

Scott, K.N., L.S. Palencia, R.P. Merlo, R.S. Yates, S. Liang, and R. DeLeon. 1997. Evaluation of
Cryptosporidium and Surrogate Removal Throughout a Full-Scale Treatment Plant. International
Symposium on Waterborne Cryptosporidium in Proc. AWWA International Symposium on
Waterborne Pathogens. Denver, Colo.: AWWA.

Sharma, M.M., H. Chamoun, D.S.H. Sita Rama Sarma, and R.S. Schechter. 1992. Factors
Controlling the Hydrodynamic Detachment of Particles from Surfaces. Jour. Colloid and
Interface Sci., 149(1):121-134.

Stanley, S.J., B.M. Coffey, and D.W. Rector. 1998. Feasibility Study to Develop an Artificial
Intelligence System for Optimization of Water Treatment Plant Operation. American Water
Works Association Research Foundation, Denver, CO.

Swaim, P.D., M.S. Heath, N.L. Patania, W.B. Wells, and R.R. Trussell. 1996. High-Rate Direct
Filtration for Giardia and Cryptosporidium Removal in Proc. Annual Conference. Denver,
Colo.: AWWA.

Swertfeger, J., D.H. Metz, J. DeMarco, A. Braghetta, and J.G. Jacangelo. 1999. Effect of Filter
Media on Cyst and Oocyst Removal. Jour. AWWA, 91(9):90-100.

277
Tate, C.H., and R.R. Trussell. 1978. The Use of Particle Counting in Developing Plant Design
Criteria. Jour. AWWA, 70(12):691-698.

Teunis, P.F.M., G. J. Medema, L, Kruidenier, and A. H. Havelaar. 1997. Assessment of the Risk
of Infection by Cryptosporidium or Giardia in Drinking Water from a Surface Water Source.
Wat. Res., 31(6): 1333-1346.

Timms, S., J.S. Slade, and C.R. Pricker. 1995. Removal of Cryptosporidium by Slow Sand
Filtration. Water Sci. Tech., 31(5-6):81-84.

Tobiason, I.E., and C.R. O'Melia. 1988. Physicochemical Aspects of Particle Removal in Depth
Filtration. Jour. AWWA, 80(12):54-64.

Trussell, R., A.R. Trussell, J.S. Lang, and C.H. Tate. 1980. Recent Developments in Filtration
System Design. Jour. AWWA, 72(12):705-710.

Tuepker, J.L., and C.A. Bauescher, Jr. 1968. Operation and Maintenance of Rapid Sand and
Mixed Media Filters in a Lime Softening Plant. Jour. AWWA, 60:1377.

USEPA (U.S. Environmental Protection Agency). 1990. Guidance Manual for Compliance with
the Filtration and Disinfection Requirements for Public Water Systems Using Surface Water
Sources. Office of Drinking Water, Washington, D.C.

USEPA (U.S. Environmental Protection Agency). 1991. Optimizing Water Treatment Plant
Performance Using the Composite Correction Program. Office of Research and Development:
Cincinnati, Ohio.

USEPA (U.S. Environmental Protection Agency). 1996. 1CR Microbiology Laboratory Manual.
Office of Research and Development, Washington, D.C.

USEPA (U.S. Environmental Protection Agency). 1998a. National Primary Drinking Water
Regulations: Disinfectants and Disinfection Byproducts; Final Rule. Fed. Reg., 63(241):69390-
69476.

USEPA (U.S. Environmental Protection Agency). 1998b. National Primary Drinking Water
Regulations: Interim Enhanced Surface Water Treatment; Final Rule. Fed. Reg., 63(241):69478-
69521.

West, T., P. Daniel, P. Meyerhofer, A. DeGraca, S. Leonard, and C. Gerba. 1994. Evaluation of
Cryptosporidium Removal Through High-Rate Filtration in Proc. AWWA Annual Conference.
Denver, Colo.: AWWA.

Wolfe, R.L., M.H. Stewart, S. Liang, and M.J. McGuire. 1989a. Disinfection of Model Indicator
Organisms in a Drinking Water Pilot Plant by Using PEROXONE. Appl. Environ. Microbiol.,
55(9):2230-2241.

278
Wolfe, R.L., M.H. Stewart, K.N. Scott, and M.J. McGuire. 1989b. Inactivation of Giardia muris
and Indicator Organisms Seeded in Surface Water Supplies by Ozone and PEROXONE.
Environ. Sci. & Technol., 23(6):744-745.

Yao, K.M., M.T. Habibian, and C.R. O'Melia. 1971. Water and Wastewater Filtration: Concepts
and Applications. Environ. Sci. Technol., 5:1105-1112.

Yates, R.S., J.F. Green, S. Liang, R.P. Merlo, and R. DeLeon. 1997a. Optimizing
Coagulation/Filtration Processes for Cryptosporidium Removal in Proc. of the International
Symposium on Waterborne Cryptosporidium. Denver, Colo.:AWWA.

Yates, R.S., J.F. Green, S. Liang, R.P. Merlo, and R. DeLeon. 1997b. Optimizing Direct
Filtration Processes for Cryptosporidium Removal in Proc. AWWA Water Quality Technology
Conference. Denver, Colo.: AWWA.

Yohe, T.L., J. Heichel, B. Stromberg, T.M. Getting, L. Zukus, and C. Ball. 1999. The Effect of
Low Uniformity Coefficient Anthracite on Dual-media Filtration in Proc. AWWA Annual
Conference. Denver, Colo.: AWWA.

Zhu, H., D.W. Smith, H. Zhou, and S.J. Stanley. 1996. Improving Removal of Turbidity Causing
Materials Using Polymers as a Filter Aid. Wat. Res., 30(1): 103-114.

279
LIST OF ABBREVIATIONS

Alum aluminum sulfate


ANOVA analysis of variance
APHA American Public Health Association
A/S anthracite/sand
ASCE American Society of Civil Engineers
ASTM American Society for Testing and Materials
ATCC American Type Culture Collection
AWWA American Water Works Association
AwwaRF Awwa Research Foundation
a significance level'

BSA bovine serum albumen


B. subtilis Bacillus subtilis
BW backwash
BWC chlorinated backwash

C carbon
°C degrees Celsius
CA California
CaCO3 calcium carbonate
CPU colony forming units
Cl chlorine
cm centimeter
cm" 1
per centimeter
Co. Company
cone. concentration
counts/mL counts per milliliter
C. parvum Cryptosporidium parvum
CRW Colorado River water
281
d day
d'1 inverse day
DAF dissolved air flotation
DOC dissolved organic carbon
D/DBP disinfectants/disinfection by-products (rule)
DPD N,N-diethyl-p-phenylene-diamine
DSW Dismal Swamp water

EBCT empty bed contact time


E. coli Escherichia coli
EDL electrical double layer
ES effective size
ESWTR Enhanced Surface Water Treatment Rule

FBC floe blanket clarification


FE filter effluent
FeCl3 ferric chloride
FI filter influent
ft foot
ft2 square feet

g gram
G Or Gyalue root-mean-square velocity gradient
GAC granular activated carbon
GC/ECD gas chromatograph electron capture detector
G. lamblia Giardia lamblia
gpm gallons per minute
gpm/ft2 gallons per minute per square foot

hour

282
HPC heterotrophic plate count

ID internal diameter
IFA immunofluorescence
in inch

JHU The Johns Hopkins University

k kilograms
kg/L kilograms per liter

L liter
L/mg-m liters per milligram per meter
L/min liters per minute
LVDW London-van der Waals (forces)

m meter
M molar
m/h meters per hour
m/min meters per minute
m/s2 meters per square second
MCL maximum contaminant level
mg milligram
mg/L milligrams per liter
mgal million gallons
min minute
mgd million gallons per day
#/mL number per milliliter
mL milliliter
mL/min milliliters per minute
ML/d megaliters per day

283
mm millimeters
MP Mannheim Plant
MRL minimum reporting level (usually 5 times the MDL)
MW molecular weight
MWD Metropolitan Water District of Southern California
"g/L micrograms per liter
uL microliters
urn micrometer (micron)

N normal
n.a. not available
n/a not applicable
Na2SO4 sodium thiosulfate
NC non-chlorinated
No. or no. number
NOM natural organic matter
NPOC nonpurgeable organic carbon
NTU nephelometric turbidity unit

OOP oxidation demonstration plant

particles/mL particles per milliliter


PC prechlorinated
pg/mL picograms per milliliter
PFU/L plaque forming units per liter
PI plant influent
Proc. Proceedings
pH negative log of proton (H+) activity

QA/QC quality assurance/quality control


R2A a proprietary agar used to culture aerobic, heterotrophic bacteria

284
R2 orr2
coefficient of determination (measure of closeness of fit)
RGF rapid gravity filtration
RNA ribonucleic acid
rpm revolutions per minute

SCADA supervisory control and data acquisition


scfm/ft2 standard cubic feet per minute per square foot
sec seconds
SPW State Project water
std. dev. standard deviation
SUVA specific ultraviolet absorbance
SWTR Surface Water Treatment Rule

IDS total dissolved solids


THM trihalomethane
THMFP trihalomethane formation potential
TOC total organic carbon

UC uniformity coefficient
US United States
USEPA United States Environmental Protection Agency
USP United States patent
UV ultraviolet
UV254 ultraviolet light absorbance at 254 nm wavelength
uw University of Waterloo

WEF Water Environment Federation

285
AWWA
Research
Foundation
- AdvmdngthtSdfnc* of Water*
6666 W. Quincy Avenue, Denver, CO 80235
(303)347-6100

1P-5C-90874-12/01-CM 9ll78T58"3 ll 2~iT700

You might also like