H
                United States Environmental
                Protection Agency
                Association off Metropolitan
                Water Agencies

                Final Report for
                Disinfection By-Products in
                United States Drinking Waters

                Volume 1  - Report
                November 1989
JNM
.ery
   Con
             Metropolitan Water District ol Southern Calik :

-------
UNITED STATES ENVIRONMENTAL PROTECTION AGENCY



  ASSOCIATION OF METROPOLITAN WATER AGENCIES
           DISINFECTION BY-PRODUCTS IN



          UNITED STATES DRINKING WATERS







                   FINAL REPORT
                   November 1989
      Metropolitan Water District of Southern California



      James M. Montgomery, Consulting Engineers, Inc.

-------
               PROJECT STAFF

Metropolitan Water District of Southern California

            Michael J. McGuire, Ph.D.
               Stuart W. Krasner
                Kevin M. Reagan
James M. Montgomery, Consulting Engineers, Inc.

             E. Marco Aieta, Ph.D.
           Joseph G. Jacangelo, Ph.D.
                Nancy L. Patania
                Karl M. Gramith

-------
                           ACKNOWLEDGEMENTS

This report has been made possible through the dedicated efforts of a large number of
people.   In particular, we  would like to acknowledge the the thirty-seven utilities that
participated in this study as well as the following individuals:

                 United States Environmental Protection Agency

                        James J. Westrick. Project Officer
                              Patricia Snyder Fair


                   Association of Metropolitan Water Agencies

                                   Ken Kirk
                               Diane Van De Hei
                    California Department of Health Services

                                  David Spath


                Metropolitan Water District of Southern California

                                 Russell Chinn
                                   Eric Crofts
                                  Melissa Dale
                                Cordelia Hwang
                                   Bart Koch
                                   Ching Kuo
                                 Fong-Yi  Lieu
                                   Ted  Lieu
                                Margaret Moylan
                                 Mary Romero
                                 Patricia Rottler
                                Warren Schimpff
                                Michael Sclimenti
                                Suzanne Teague
                                 Randy Whitney
                                Roniece Widman


               James M. Montgomery, Consulting Engineers,  Inc.

                                 Martha Frost
                                   Ulises Gil
                               Yolanda Gutierrez
                                   Seth Jelen
                                Judy Richardson
                                   Carol Tate
                                Rhodes Trussell
                              Jennifer Yarbrough

-------
                         TABLE OF CONTENTS


                                                                Number

                               VOLUME 1

EXECUTIVE SUMMARY

     Introduction	      ES-1

     Utility Selection	      ES-2

     Methodology	      ES-2

     Data Management and Analysis	      ES-2

     Baseline Sampling Results and Discussion	      ES-3

     Treatment Modification Studies -
      Results and Discussion	      ES-6
          Ozonation Studies	      ES-6
          Chlorine Dioxide Studies	      ES-8
          Coagulation Studies	      ES-9
          Granular Activated Carbon Study	     ES-10

     Summary and Conclusions	     ES-11

SECTION 1 - INTRODUCTION

     Project Background and Objectives	       1-1
          Project Objectives	       1-1
          Target Compounds	       1-2
          California Department of Health Services/
           California Public Health Foundation DBP Study	       1-2

     Project Description	       1-2
          Baseline Data Collection	       1-3
          Treatment Modification Studies	       1-3

SECTION 2 - UTILITY SELECTION

     Information Request	       2-1

     Selection Matrix	       2-1

     Process Trains of the Participating Utilities	       2-2

SECTION 3 - METHODOLOGY

     Sampling Procedure	       3-1

-------
                  TABLE OF CONTENTS (CONTINUED)
                                                                  Page
                                                                Number
    Analytical Methods	      3-2

    Simulated Distribution System Testing	      3-2

    Quality Assurance	      3-5

    On-Site Extraction Studies	      3-8

    Aldehyde Derivatizations/Extractions On-Site	      3-11

    Preservation of TOX Samples	      3-11

SECTION 4 - DATA MANAGEMENT AND ANALYSIS

    Database	      4-1

    Data Handling Protocol	      4-1

    Summary Statistics	      4-2

    Graphical  Presentation of Data	      4-4
          Box-and-Whisker Plots	      4-4
          Bar Charts	      4-5
          Star Symbol Plots	      4-5

SECTION 5 - BASELINE SAMPLING RESULTS AND DISCUSSION

    Overview of Baseline Data	      5-1

    Star Plot Analyses	      5-3

    Seasonal Variation	      5-4
          Influent Water Quality	      5-5
          Classes of DBP Compounds	      5-5

    Variation by Treatment Type	      5-7

    Variation by Source Water Type	      5-8

    Variation by Disinfection Scheme	      5-8
          Influent Parameters	      5-8
          Classes of DBP Compounds	      5-9

    DBP and Influent Parameter Correlations	      5-11
          Correlations with THMs	      5-12

-------
                  TABLE OF CONTENTS (CONTINUED)
                                                                  Page
                                                                Number
         Correlation with HAAs	      5-12
         Correlations with Influent Water Quality
           Parameters	      5-12
         Correlations with Bromide and Brominated DBPs	      5-13
         Correlations with Chloroform	      5-14
         Correlations with DCAA and TCAA	      5-15
         Correlations with Formaldehyde	      5-15
         Correlations with Total Organic Halide	      5-15

    Special Issues	      5-16
         Effect of pH	      5-16
         Effect of Temperature	      5-17
         Brominated DBPs	      5-18
         Aldehydes	      5-20
         Cyanogen Chloride Results	      5-21
         DBP Levels of Disinfection-Only Utilities	      5-22
         Removal of TOC During Treatment	      5-22
         Comparison of THM Levels from USEPA DBP Study
           and AWWARF THM Survey	      5-23
         Comparison of USEPA Study and CDHS DBP
           Study Results	      5-24

SECTION 6 - TREATMENT MODIFICATION STUDIES - RESULTS
             AND DISCUSSION

    Ozonation Studies	      6-1
         Utility 6	      6-1
         Utility 7	      6-3
         Utility 19	      6-4
         Utility 25	      6-6
         Utility 36	      6-8
         Discussion	      6-10

    Chlorine Dioxide Studies	      6-12
         Utility 16	      6-12
         Utility 37	      6-13
         Di scussion	      6-14

    Coagulation Studies	      6-15
         Utility 3	      6-15
         Utility 12	      6-17
         Discussion	      6-18

    Granular Activated Carbon Study	      6-19
         Process Description	      6-19
         GAC Contactor Operation	      6-20
         Treatment Study Description	      6-20
         Simulated Distribution System Testing	      6-21

-------
                TABLE OF CONTENTS (CONTINUED)
                                                              Page
                                                             Number
         Simulated Distribution System Testing
          Results and Discussion	      6-21
         DBF Levels Produced by Conventional Treatment:
          Results and Discussion	      6-26

SECTION 7 - SUMMARY AND CONCLUSIONS

    Baseline Results	      7-1

    Process Modification Results	      7-3

    Research Needs	      7-4

REFERENCES


                               VOLUME 2

APPENDIX A - Questionnaire

APPENDIX B - Sampling Instructions, Sample Information Sheet

APPENDIX C - Analytical/Quality Assurance Methods

APPENDIX D - Correlations

-------
                                 TABLES
                                                                  Following
Number                                                             Page
 ES-1     List of compounds targeted in study	   ES-I
 ES-2     Disinfection by-products in drinking water study:
            Utility selection matrix	   ES-2
 ES-3     Disinfection by-products in drinking water: Summary
            of baseline sampling median values	   ES-3
  1 -1      List of compounds targeted in study	    1-2
  2-1      Disinfection by-products in drinking water study:
            Utility selection matrix	    2-1
  3-1      Sampling kit contents	    3-1
  3-2     Sample holding times	    3-1
  3-3     Utility 11 GAC study: Day 13 DBF results	    3-3
  3-4     Utility 11 GAC study: Day 25 DBF results	    3-3
  3-5     Utility 11 treatment study, comparison of GAC
            effluent  SDS tests:  Effect of chlorine dose
            and pH on DBFs	    3-4
  3-6     Minimum reporting levels (MRLs)	    3-6
  3-7     DBF quality control limits	    3-7
  3-8     Utility 2:  Chlorinated filter effluent on-site
            and lab extraction	    3-8
  3-9     Utility 2:  Chloraminated clearwell effluent
            on-site and lab extraction	    3-8
 3-10     Utility 33: Chlorinated effluent on-site and
            lab extraction	    3-9
 3-11     Utility 6:  Chloraminated clearwell effluent
            on-site and lab extraction	   3-10
 3-12     Pentane -  extractable disinfection by-products:
            Matrix holding study	   3-10

-------
                             Tables (continued)
                                                                    Following
Number                                                              Page
 3-13     Utility 33: Aldehyde results on-site and lab
            extraction	  3-11
 3-14     Utility 6: Formaldehye results on-site and lab
            extraction	  3-11
 3-15     TOX preservation study	  3-12
 3-16     On-site TOX preservation studies	  3-12
  5-1     List of abbreviations used in DBF  study	    5-1
  5-2     Disinfection by-products in drinking water: Summary
            of baseline sampling median values	    5-1
  5-3     Comparison of seasonal DBF levels	    5-6
  5-4     Comparison of seasonal TTHMs and influent
            water quality	    5-7
  5-5     DBF concentrations at utility with seasonal change
            in bromide levels	  5-18
  5-6     DBF concentrations at inland utilities with high
            bromide levels	  5-19
  5-7     Utility 12: Influence of saltwater intrusion	  5-19
  5-8     Aldehyde levels in plant influents and effluents	  5-20
  5-9     Levels of aldehydes and THMs in  selected cleanvell
            effluents	  5-21
  6-1     Disinfection scenarios, plant scales and sources
            for utilities participating in study	    6-1
  6-2     Utility 6 treatment study	    6-1
  6-3     Utility 7 treatment study	    6-3
  6-4     Utility 19 treatment study	    6-5
  6-5     Utility 25 treatment study	    6-6
  6-6     Explanation of abbreviations used  in Utility 25
            DBF profiles and residence time  of distribution
            system sampling points	    6-7

-------
                             Tables (continued)
                                                                    Following
Number                                                              Page

  6-7     Utility 36 treatment study	    6-8
  6-8     Disinfection by-products concentrations before
            and after ozone addition at Utility 36	    6-8
  6-9     Utility 16 treatment study	  6-12
 6-10     Utility 37 treatment study	  6-13
 6-11     Utility 3 treatment study	  6-16
 6-12     Percent removal of total organic carbon and
            ultraviolet absorbance at 254 nm by various alum
            doses at Utility 3	  6-16
 6-13     Utility 12 treatment study	  6-17
 6-14     Percent removal of total organic carbon and
            ultraviolet absorbance at 254 nm by various alum
            doses at Utility 12	  6-17
 6-15     Concentrations of various disinfection by-products and
            total organic halide at varying alum doses at
            Utility 12	  6-18
 6-16     Utility II treatment study sampling plan	  6-20
 6-17     Utility 11 sampling locations	  6-20
 6-18     Utility II treatment study	  6-21
 6-19     Utility 11 treatment study TOX results (in //g/L)	  6-25
 6-20     Utility 11 treatment study comparison of SDS and
            distribution system results	  6-26

-------
                               FIGURES


                                                               Following
Number                                                           Page

ES-1      Guide to notched box-and whisker plots	   ES-3

ES-2      Disinfection by-product concentration by DBF class -
           four quarters	   ES-3

ES-3      Percent of sum of DBF class medians by DBF class -
           four quarters	   ES-3

ES-4      Influent total  organic carbon by quarter	   ES-3

ES-5      Influent total  organic carbon by source	   ES-3

ES-6      Trihalomethanes by quarter	   ES-3

ES-7      Trihalomethanes by influent temperature	   ES-3

ES-8      Trihalomethanes by disinfection scheme	   ES-4

ES-9      Correlations with trihalomethanes	   ES-4

ES-10    Correlations with influent chloride and bromide	   ES-4

ES-11     Correlations with influent bromide	   ES-5

ES-12     Correlations with dichloroacetic acid	   ES-5

ES-I3     Mean total organic carbon removal through
           filtering utilities' processes	   ES-5

ES-14     AWWARF 12-quarter vs.  USEPA 4-quarter means	   ES-5

ES-15     Effect of CI2,NH2C1 and O3,C12,NH2C1 on DBF
           formation (Utility 6)	   ES-6

ES-16     Effect of C12,  C12/NH3 & O3/NH3/C12 on DBF formation
           (Utility 7)	   ES-6

ES-17     Effect of various disinfection schemes  on
           disinfection  by-product formation at
           Utility 19	   ES-7

ES-18     Effects of NH2C1 and and O,/NH2C1 treatment on total
           THM formation  at Utility 25	   ES-7

-------
                           Figures (continued)
                                                                Following
Number                                                           Page
ES-19     Effect of various disinfection schemes on
           disinfection by-product formation at Utility 36	   ES-7
ES-20     Effect of Cl, and C1O2/C12 on DBF formation
           (Utility 16)	   ES-9
ES-2I     Effect of Cl, and C1O2 on DBF formation
           (Utility 37)	   ES-9
ES-22     Effect of alum dose on DBF formation (Utility 3,
           SDS)	   ES-9
ES-23     Effect of alum dose on DBF formation (Utility 12)	   ES-9
ES-24     TOC vs run time for GAC column influent and effluent
           Utility 11	  ES-10
ES-25     SDS TTHMs vs run time for GAC column influent and
           effluent, Utility  11	  ES-10
I -1       Structural formulas for disinfection by-products	    1-2
2-1       Utilities participating in DBF study	    2-2
2-2       Process trains of participating utilities	    2-2
4-1       Data-sampling protocol	    4-1
4-2       Guide to notched box-and-whisker plots	    4-4
4-3       Example of notched box-and-whisker plot  for
           three data sets	    4-5
4-4       Guide to bar charts using median, percentiles and
           minimum reporting level	    4-5
4-5       Example guide to star plots	    4-6
5-1       Disinfection by-product concentration by DBF class -
           four quarters	    5-1
5-2       Percent of sum of DBF class medians by DBF class -
           four quarters	    5-1
5-3       Percent of sum of non-THM DBF class medians by class -
           four quarters	    5-2

-------
                           Figures (continued)


                                                              Following
Number                                                         Page

5-4       Trihalomethane concentrations by THM compound -
           four quarters	    5-2

5-5       Percent of sum of THM compound medians by compound -
           four quarters	    5-2

5-6       Haloacetic acid concentrations by HAA compound
           four quarters	    5-2

5-7       Haloketone concentrations by HK compound - four
           quarters	    5-2

5-8       Haloacetonitrile concentrations by HAN compound -
           four quarters	    5-2

5-9       Miscellaneous DBF concentrations by compound -
           four quarters	    5-2

5-10      Aldehyde concentrations by compound -
           four quarters	    5-2

5-11      Guide to star symbol plots	    5-3

5-12      Plant influent by utility ID number - summer
           quarter	    5-3

5-13      Plant influent by source water type -
           summer quarter	    5-3

5-14      Plant influent by treatment type -
           summer quarter	    5-4

5-15      Plant influent by disinfection scheme -
           summer quarter	    5-4

5-16      Influent total organic carbon by quarter	    5-5

5-17      Influent UV absorbance by quarter	    5-5

5-18      Influent chloride by quarter	    5-5

5-19      Influent bromide by quarter	    5-5

5-20      XDBPsum by quarter	    5-5

5-21       Trinalomethanes by quarter	    5-5

5-22      Chloroform by quarter	    5-6

-------
                            Figures (continued)
                                                                 Following
Number                                                           Page
5-23      Bromodichloromethane by quarter	   5-6
5-24      Dibromochloromethane by quarter	   5-6
5-25      Bromoform by quarter	   5-6
5-26      Haloacetic acids by quarter	   5-6
5-27      Dichloroacetic acid by quarter	   5-6
5-28      Dibromoacetic acid by quarter	   5-6
5-29      Haloacetonitriles by quarter	   5-6
5-30      Haloketones by quarter	   5-6
5-31      Aldehydes by quarter	   5-6
5-32      Chloropicrin by quarter	   5-6
5-33      Cyanogen chloride by quarter	   5-6
5-34      Chloral hydrate by quarter	   5-6
5-35      Influent total organic carbon  by treatment	   5-6
5-36      Influent total organic carbon  by source	   5-8
5-37      Influent UV-254 absorbance by source	   5-8
5-38      Influent chloride by source	   5-8
5-39      Influent bromide by source	   5-8
5-40      Influent total organic carbon  by
           disinfection scheme	   5-8
5-41      Influent UV absorbance by disinfection scheme	   5-8
5-42      Influent chloride by disinfection scheme	    5-9
5-43      Influent chloride by disinfection
           scheme (w/o #10)	    5-9
5-44      Influent bromide by disinfeciton scheme	    5-9
5-45      Influent bromide by disinfection
           scheme (w/o #10)	    5-9

-------
                             Figures (continued)

                                                                  Following
 Number                                                             Page
 5-46      XDBPstim by disinfection scheme	    5-9
 5-47      Trihalomethanes by disinfection scheme	    5-9
 5-48      Haloacetic acids by disinfection scheme	   5-10
 5-49      Haloacetonitriles by disinfection scheme	   5-10
 5-50      Haloketones by disinfection scheme	   5-11
 5-51      Aldehydes by disinfection scheme	   5-11
 5-52      Chloropicrin by disinfection scheme	   5-11
 5-53      Chloral hydrate by disinfection scheme	   5-11
 5-54      Cyanogen chloride by disinfection  scheme	   5-1!
 5-55      Correlations with trihalomethanes	   5-12
 5-56      Correlations with haloacetic acids	   5-12
 5-57      Correlations with influent parameters	   5-12
 5-58      Correlations with influent parameters	   5-13
 5-59      Correlations with influent chloride and bromide	   5-13
 5-60      Correlations with influent bromide	   5-13
 5-61       Correlations with bromoform	   5-14
 5-62      Correlations with dibromoacetonitrile	   5-14
 5-63      Correlations with chloroform	   5-14
 5-64      Correlations with chloroform	   5-14
 5-65       Correlations with dichloroacetic acid	   5-15
 5-66      Correlations with trichloroacetic acid	   5-15
 5-67       Correlation  with formaldehyde	   5-15
5-68       Molar XDBPs versus molar TOX	   5-15
5-69       XDBPsum by effluent pH	   5-16

-------
                            Figures (continued)


                                                                 Following
Number                                                           Page

5-70      Trihalomethanes by effluent pH	   5-16

5-71      Haloacetonitriles by effluent pH	   5-16

5-72      Dichloroacetonitrile by effluent  pH	   5-16

5-73      1.1.1-trichloropropanone by effluent pH	   5-17

5-74      Sum of halogenated DBFs by influent temperature	   5-17

5-75      Trihalomethanes by influent temperature	   5-17

5-76      Haloacetonitriles by influent temperature	   5-17

5-77      Utility 12 disinfection by-products
            speciation	   5-18

5-78      Cyanogen chloride by final disinfectant	   5-21

5-79      Mean total organic carbon removal through filtering
            utilities processes	   5-22

5-80      Frequency distributions of THM survey data	   5-23

5-81      AWWARF 12-quarter vs USEPA 4-quarter means	   5-24

5-82      Total halogenated DBFs by study	   5-24

6-1       Schematic of Utility 6 treatment processes	    6-1

6-2       Effect of various disinfection schemes on
            disinfection by-product formation at Utility 6	    6-2

6-3       Effect of various disinfection schemes on
            disinfection by-product formation at Utility 6	    6-2

6-4       Effect of C12,NH2C1 and O3,C12,NH2C1 on DBF
            formation (Utility 6)	    6-2

6-5       Effect of C12,NH2C1 and O3.C12,NH2C1 on THM
            formation (Utility 6)	    6-2

6-6       Effect of C12,NH,CI and O3,C12,NH2C1 on HAA
            formation (Utility 6)	    6-2

6-7       Effect of C12,NH2C1 and O3,C12,NH2C1 on HAN
            formation (Utility 6)	    6-2

-------
                            Figures (continued)


                                                                  Following
Number                                                            Page

6-8       Effect of C12,NH2C1 and O3,C12,NH2C1 on HK
            formation (Utility 6)	    6-2

6-9       Effect of CI2,NH2C1 and O3,C12,NH2C1 on misc. DBF
            formation (Utility 6)	    6-2

6-10      Effect of seasonal temperatures on DBF formation
            (Utility 6)	    6-2

6-11      Effect of O3.C12,NH2C1 on aldehydes formation
            (Utility 6)	    6-2

6-12      Schematic of Utility 7 treatment processes	    6-4

6-13      Effect of various disinfection schemes on
            disinfection by-product formation at Utility 7	    6-4

6-14      Effect of various disinfection schemes on
            disinfection by-product formation at Utility 7	    6-4

6-15      Change in formaldehyde concentration thru
            process train, State project water	    6-4

6-16      Change in acetaldehyde concentration thru
            process train, State project water	    6-4

6-17      Effect of C12, C1;/NH3 & O3/NH3/C12 on DBF
            formation (Utility 7)	    6-4

6-18      Effect of C12, C1,/NH3 & O3/NH3/C12 on THM
            formation "(Utility 7)	    6-4

6-19      Effect of C12, CK/NH3 & O3/NH3/C12 on HAN
            formation (Utility 7)	    6-4

6-20      Effect of CI2, C1;/NH3 & O3/NH3/C12 on HK
            formation (Utility 7)	    6-4

6-21      Effect of C12, CU/NH3 & O3/NH3/C12 on HAA
            formation (Utility 7)	    6-4

6-22      Effect of Cl,, CK/NH3 & O3/NH3/C12 on misc. DBF
            formation "(Utility 7)	    6-4

6-23      Schematic of Utility 19 treatment processes	    6-4

6-24      Effect of various disinfection schemes on
            disinfection by-product formation at Utility 19	    6-5

-------
                           Figures (continued)


                                                                Following
Number                                                           Page

6-25      Effect of various disinfection schemes on
           disinfection by-product formation at Utility 19	    6-5

6-26      Effect of O3/C12 & C12 only on DBF  formation
           (Utility 19)	    6-5

6-27      Effect of O3/C12 & C\2 only on THM formation
           (Utility 19)	    6-5

6-28      Effect of O,/C12 & C12 only on HAA formation
           (Utility 19)	    6-5

6-29      Effect of O3/C12 & CI2 only on HAN formation
           (Utility 19)	    6-5

6-30      Effect of O,/C12 & C12 only on HK formation
           (Utility 19)	    6-5

6-31      Effect of O3/C1, & Cl, only on misc. DBF
           formation (Utility 19)	    6-5

6-32      Schematic of Utility 25 treatment processes	    6-6

6-33      Effect of various disinfection schemes on
           disinfection by-product formation at
           Utility 25	    6-6

6-34      Effect of various disinfection schemes on
           disinfection by-product formation at
           Utility 25	    6-6

6-35      Effects of NH2C1 and O3/NH2CI treatment on
           halogenated DBF formation, Utility 25	    6-7

6-36      Effects of NH2C1 and O,/NH2C1 treatment on total
           THM formation at Utility 25	    6-7

6-37      Effects of NH2C1 and O3/NH2C1 treatment on HAA
           formation at Utility 25	    6-7

6-38      Effects of NH2C1 and O3/NH2C1 treatment on DCAA
           formation at Utility 25	    6-7

6-39      Effects of NH2C1 and O3/NH2C1 treatment on
           chloropicrin formation at Utility 25	    6-7

6-40      Effects of NH2C1 and O3/NH2C1 treatment on HAN
           formation at Utility 25	    6-7

-------
                           Figures (continued)


                                                                Following
Number                                                          Page

6-41      Effects of NH2C1 and oyNH2Cl treatment on HK
           formation at Utility 25	    6-7

6-42      Effects of NH2C1 and O^/NH2C1 treatment on chloral
           hydrate formation Utility 25	    6-7

6-43      Effects of NH2C1 and CK/NH2C1 treatment on CNC1
           formation at Utility 25	    6-7

6-44      Effects of NH2C1 and O3/NH2C1 treatment an ALD
           formation at Utility 25	    6-7

6-45      Effects on  NH2C1 and O3/NH2CI treatment on
           formaldehyde formation at Utility 25 	    6-7

6-46      Schematic of Utility 36 treatment processes	    6-8

6-47      Effect of various disinfection schemes on
           disinfection by-product formation at Utility  36	    6-8

6-48      Effect of various disinfection schemes on
           disinfection by-product formation at Utility  36	    6-8

6-49      Effects of various treatments on halogenated
           DBF formation at Utility 36	    6-9

6-50      Effects of various treatments on THM formation
           at Utility 36	    6-9

6-51      Effects of various treatments on HAA formation at
           Utility 36	    6-9

6-52      Effects of various treatments on HAN formation at
           Utility 36	    6-9

6-53      Effects of various treatments on HK formation at
           Utility 36	    6-9

6-54      Effects of various treatments on CH formation at
           Utility 36	    6-9

6-55      Effects of various treatments on CHP formation at
           Utility 36	    6-9

6-56      Effects of various treatments on CNC1 formation
           at Utility 36	    6-9

-------
                            Figures (continued)
                                                                 Following
 Number                                                           Page

 6-57       Effects of various treatments on ALD formation at
            Utility 36	    6-9

 6-58       Effects of various treatments on formaldehyde
            formation at Utility 36	    6-9

 6-59       Percent change in DBF and TOX concentrations due
            to switch from chlorine only to ozone/chlorine
            treatment	  6-10

 6-60       Change in DBF concentrations due to switch from
            chlorine only to ozone/chlorine treatment	  6-10

 6-61       Percent change in DBF concentrations due to switch
            from chloramines only to ozone/chloramines
            treatments	  6-11

 6-62       Change in DBF concentrations due to switch from
            chloramines only to ozone/chloramines treatment	  6-11

 6-63       Percent change in DBF concentrations due to
            switch from chlorine only to ozone/chloramines
            treatment	  6-11

 6-64       Change in DBF concentrations due to switch from
            chlorine only to ozone/chloramines treatment	  6-11

 6-65       Schematic of utility  16 treatment processes	  6-12

 6-66       Effect of  C12 and C1O,/C12 on DBF formation
            (Utility  16)	  6-12

 6-67      Schematic of Utility 37 treatment processes	  6-13

 6-68      Effect of  Cl, and C1O2 on DBF formation
            (Utility  37)	!	  6-14

 6-69      Effect of  Cl, and CIO, on THM formation
            (Utility  37)	  6-14

 6-70      Effect of  CL and CIO, on HAN formation
            (Utility  37)	  6-14

 6-71      Effect of  C12 and CIO, on haloketone
            formation (Utility 37)	   6-14

6-72      Effect of CI2 and CIO, on haloacetic
            acid formation (Utility 37)	   6-14

-------
                           Figures (continued)


                                                                Following
Number                                                          Page

6-73      Effect of C12 and CIO, on misc. DBF
           formation (Utility 37)	  6-14

6-74      Schematic of Utility 3 treatment process	  6-15

6-75      Effect of alum dose on DBF formation (Utility 3,
           SDS)	  6-16

6-76      Effect of alum dose on THM formation (Utility 3,
           SDS)	  6-16

6-77      Effect of alum dose on HAA formation (Utility 3,
           SDS)	  6-16

6-78      Effect of alum dose on HK formation (Utility  3,
           SDS)	  6-16

6-79      Effect of alum dose on HAN formation (Utility 3,
           SDS	  6-16

6-80      Effect of alum dose on misc.  DBF formation (Utility 3,
           SDS)	  6-16

6-81      Schematic of Utility 12 treatment  process	  6-17

6-82      Effect of alum dose on DBF formation (Utility 12)	  6-17

6-83      Effect of alum dose on THM formation (Utility 12)	  6-17

6-84      Effect of alum dose on HAA formation (Utility 12)	  6-17

6-85      Effect of alum dose on HK formation (Utility  12)	  6-32

6-86      Effect of alum dose on HAN formation (Utility 12)	  6-17

6-87      Effect of alum dose on misc.  DBF formation
           (Utility 12)	  6-17

6-88      Effect of alum dose on ALD formation (Utility 12)	  6-17

6-89      Schematic of Utility 11 treatment  process	  6-19

6-90      TOC vs  run time for GAC
           column influent and effluent, Utility  11 	  6-21

6-91      TOC breakthrough profile at Utility 11	  6-21

-------
                           Figures (continued)
                                                              Following
 Number                                                        Page

 6-92      TOC breakthrough profiles for three
           demonstration-scale contactor runs at Utility 11 	  6-22

 6-93      SDS TTHMs vs run time for GAC
           column influent and effluent, Utility 11	  6-22

 6-94      SDS CHC13 vs run  time for GAC
           column influent and effluent, Utility 11	  6-24

 6-95      SDS CHBr3 vs run  time for GAC
           column influent and effluent, Utility 11 	  6-24

 6-96      SDS HAAs vs run time for GAC
           column influent and effluent, Utility 11 	  6-24

 6-97      SDS DCAA vs run  time for GAC
           column influent and effluent, Utility 11 	  6-24

 6-98      SDS DBAA vs run  time for GAC
           column influent and effluent, Utility 11 	  6-24

 6-99      SDS HANs vs run time for GAC
           column influent and effluent, Utility 11	  6-24

 6-100     SDS HKs vs run time for GAC
           column influent and effluent, Utility 11	   6-25

 6-101      SDS CH vs run time for GAC
           column influent and effluent, Utility 11	   6-25

 6-102      SDS CHP vs run  time for GAC
           column influent and effluent, Utility 11	   6-25

 6-103      Formation of TTHMs in distribution system
           (Utility II)	   6-25

6-104      Formation of DCAN in distribution system
           (Utility 11)	   6.27

6-105      Formation of CH  in distribution system
           (Utility 11)	   6_27

-------
                          LIST OF ABBREVIATIONS
 ACETAL       Acetaldehyde
 ALD          Aldehyde
 AMWA        Association of Metropolitan Water Agencies
 AWWARF      American Water Works Association Research Foundation
 BCAN         Bromochloroacetonitrile
 Br            Bromide
 C             Concentration
 C(,            Concentration at time zero
 CDHS         California Department of Health Services
 CE            Clearwell effluent
 CH            Chloral hydrate
 CHBr3         Bromoform
 CHBr2Cl        Dibromochloromethane
 CHBrCI,        Bromodichloromethane
 CHCI3         Chloroform
 CHP           Chloropicrin
 Cl             Chloride
 CI2            Chlorine
 CI,/NH3        Prechlorine/postammonia
 CI02           Chlorine dioxide
 cm '            Inverse centimeter
 /cm            Inverse centimeter
 CNCI          Cyanogen chloride
 CONV         Conventional treatment
 CP            Chlorophenol
 CPHF          California Public Health Foundation
 DBAA          Dibromoacetic acid
 DBAN          Dibromoacetonitrile
 DBP            Disinfection by-product
DCAA          Dichloroacetic acid
DCAN          Dichloroacetonitrile

-------
                        List of Abbreviations, Continued
 DCP
 U-DCP
 DF
 DIS
 EBCT
 FE
 FI
 FRM
 FS
 GAC
 GC
 GW
 HA
 HAA
 HAN
 HK
 HOBr
 ID
 JMM
 LAA
 LR
 MBAA
 MCAA
 MCL
 Metropolitan
//g/L
 mg/L
mgd
 MRL
 MS
n
 NA
ND
NH2CI
2,4-dichlorophenol
1,1 -dichloropropanone
Direct filtration
Disinfection only
Empty bed contact time
Filter effluent
Filter influent
Formaldehyde
Flowing stream
Granular activated carbon
Gas chromatograph
Groundwater
Hydrogen peroxide
Haloacetic acid
Haloacetonitrile
Haloketone
Hypobromous acid
Identification
James M. Montgomery, Consulting Engineers, Inc.
Los Angeles Aqueduct
Lake/reservoir
Monobromoacetic acid
Monochloroacetic acid
Maximum contaminant level
Metropolitan Water District of Southern California
Microgram per liter
Milligram per liter
Million gallons per day
Minimum reporting level
Mass spectrometer
Number of samples
Not analyzed
Not detected
Chloramines

-------
                        List of Abbreviations, Continued
nm            Nanometer
O3             Ozone
PCP           Pentachlorophenol
PI             Plant influent
r              Correlation coefficient
SDS           Simulated distribution system
SDWA         Safe Drinking Water Act
SFT           Softening
SPW           State Project Water
t              Time
TCAA         Trichloroacetic acid
TCAN         Trichloroacetonitrile
TCP           2,4,6-trichlorophenol
 1.1.1 -TCP      I.I.I -trichloropropanone
THM          Trihalomethane
TOC           Total  organic carbon
TOX           Total  organic halide
TTHM         Total  trihalomethanes
u              Number of utilities
USEPA        United States Environmental Protection Agency
UV            Ultraviolet
UV-254        Ultraviolet absorbance at 254 nanometers
XDBP         Halogenated disinfection by-product
XDBP         Sum of measured halogenated disinfection by-products

-------
Executive Summary

-------
                           EXECUTIVE SUMMARY
INTRODUCTION
The United  States Environmental  Protection Agency  (USEPA) will  be  developing
regulations to control disinfection by-products (DBFs) in drinking water as a result of
the 1986  amendments to the Safe Drinking Water Act.   Although the schedule for
promulgation of regulations under these amendments remains uncertain at this time, the
anticipated regulations require that the presence and  control of target DBFs be fully
understood.   Thus, in October, 1987, the Association of Metropolitan Water Agencies
(AMWA)  entered into a cooperative agreement with the USEPA to perform a study of
the formation and control of DBFs in full-scale drinking water systems.  The study was
performed by the  Metropolitan Water District of Southern California (Metropolitan)
and James M. Montgomery, Consulting Engineers, Inc. (JMM).

Specific objectives of the project  included determining  the occurrence  of  DBFs at 25
drinking water treatment facilities around  the nation; determining the seasonal nature of
DBF occurrence as a function of temperature, total  organic carbon (TOC), pH and
other  water quality parameters; and determining the  effect of changes in treatment
processes  and/or disinfectants  on  the production of DBFs at up to 10 drinking water
treatment  facilities. The study focused on the identification of DBFs as a function of
source water quality, water treatment process selection and operation,  and disinfection
processes  and chemicals.   The target DBF compounds for the study are listed in Table
ES-1.  Target compounds included trihalomethanes  (THMs), haloacetic  acids (HAAs),
haloacetonitriles (HANs).  haloketones (HKs),  aldehydes (ALDs), chloropicrin, chloral
hydrate, cyanogen chloride, and 2,4,6-trichlorophenol.  A companion study,  funded by
the California Department of Health Services (CDHS) through a grant to the California
Public Health Foundation  (CPHF), was  also  conducted by Metropolitan and  JMM.
This study  evaluated DBF production at 10 drinking  water  treatment  facilities in
California.

The first year of the two-year USEPA  study focused on establishing and verifying the
analytical  procedures used  in the study,  selecting utilities to participate in the  study,
developing DBF baseline  data through implementation  of a quarterly sampling program
at  the participating utilities, and conducting process modification studies at two utilities.
During the  second year  of the project,  baseline data collection  was completed and
process modification studies were conducted at  six utilities.

Baseline data collection for the 35 utilities participating in the combined  USEPA and
CDHS studies  involved  sampling  of clearwell effluents (after final disinfection but
before  distribution) on a quarterly basis for one year.  The first sampling quarter (mid-
March through  April,  1988) corresponded  to  the  spring season, and  subsequent
samplings corresponded  to the summer,  fall  and  winter seasons.   Utilities  were  sent
sampling kits containing  the sample bottles, detailed instructions for sampling,  and a
sample information sheet on which to record plant operating conditions  on the day of
sampling.
                                     ES-1

-------
                               TABLE ES-1

              LIST OF COMPOUNDS TARGETED IN STUDY
                               Compounds
                           Trihalomethanes

                               chloroform
                               bromodichloromethane
                               dibromochloromethane
                               bromoform

                           Haloacetonitriles

                               trichloroacetonitrile
                               dichloroacetonitrile
                               bromochloroacetonitrile
                               dibromoacetonitrile

                           Haloketones

                               1.1-dichloropropanone
                               l.U-trichloropropanone

                           Miscellaneous chloro-organics

                               chloropicrin
                               chloral hydrate
                               cyanogen chloride

                           Haloacetic acids

                               monochloroacetic acid
                               dichloroacetic acid
                               trichloroacetic acid
                               monobromoacetic acid
                               dibromoacetic acid

                           Chlorophenols

                               2.4-dichlorophenol*
                               2.4.6-trichlorophenol
                               pentachlorophenol*

                           Aldehydes

                               formaldehyde
                               acetaldehyde
* These chlorophenols were only analyzed for during the first sampling quarter.

-------
                              Executive Summary


 Treatment  modification studies were conducted to identify,  in a preliminary manner,
 the  impact of processes or  process modifications  on DBF production.   Thus, the
 following full-scale and/or pilot studies were conducted: five studies involving a change
 from chlorine or chloramines  to ozone/chlorine or ozone/chloramines;  two studies on
 alum coagulation for DBF precursor removal; two  studies involving a change from
 chlorine to chlorine dioxide/chlorine; and one study of granular activated carbon (GAC)
 for DBF precursor removal.

 UTILITY SELECTION

 An information request form was sent to 104 potential participants in  the study, and
 from the 78 responses received, utilities were selected to complete the matrix shown in
 Table ES-2.   The  matrix was divided into two  major categories, treatment type
 (conventional, direct filtration,  softening and disinfection only) and  source water type
 (groundwater,  lake/reservoir   and   flowing   stream).     Within   these   categories,
 geographical location and disinfectant  type were also considered.   In  addition, two
 other categories were utilized in developing the selection matrix:  population and THM
 level.

 METHODOLOGY

 Grab samples were collected at clearwell effluents (after final disinfection but before
 distribution)  for  DBF and  TOC analyses.   For those plants  which did not have
 clearwells.  samples were collected at specified points after final disinfection.  Additional
 analyses were performed on plant influents:  TOC, bromide, chloride and aldehydes.
 TOC was also  measured at  filter influents,  if filtration was employed  as  a treatment
 process.    Holding  studies performed  by  Metropolitan  were conducted   to identify
 preservatives and holding times for the DBF samples.

 The THM liquid/liquid extraction (LLE) gas chromatograph  (GC)  method was modified
 by Metropolitan to include THMs, HANs, HKs and chloropicrin.  Chloral  hydrate was
 analyzed by a separate LLE/GC method.   HAAs  and  2,4,6-trichlorophenol  were
 analyzed by  an acidic salted  LLE and GC.   Cyanogen chloride was  analyzed  by a
 purge-and-trap  gas  chromatograph/mass spectrometer  (GC/MS)  method.   Aldehydes
 were analyzed by a derivatization/extraction GC method.

 DATA MANAGEMENT AND ANALYSIS

 The large database developed during the study required a strict data handling protocol
 to ensure its accuracy and reliability.  Some of the elements of this protocol included
 use of signatures by  responsible  project team members, use  of data-sheet reference
 numbers for each individual project data sheet, and a tracking system for the status of
each group  of data.

 Nonparametric  statistical  methods  were used  to   analyze data  from   this study.
 Nonparametric methods do not require an assumed parametric distribution for the data,
and cases below the detection  limits can be incorporated more readily as compared to
parametric  methods.   A "five-number summary", including  the minimum  value,  25th
percentile,  median.  75th  percentile and maximum value of the data set, was used to
present a simple summary of the data.   Notched  box-and-whisker plots were used also

                                     ES-2

-------
                                 TABLE ES-2

       DISINFECTION BY-PRODUCTS IN DRINKING WATER STUDY

                       UTILITY SELECTION MATRIX
TREATMENT
GROUNDWATER
LAKE/RESERVOIR    FLOWING STREAM
CONVENTIONAL
Clermont Co., OH 1
Long Beach, CA 2
Norwich, CT* IB
MWD, CA -Mills 2a,5b
Arlington, TX* 2A
Hackensack, NJ 2e
MWD, CA -Weym. 2
San Francisco, CA 1
Big Spring, TX* 5
Shreveport, LA 6a,8b
Cape Girardeau, MO* 2A
Cincinnati, OH 1
Contra Costa WD, CA 2A
Sacramento, CA 1
Santa Clara Valley, CA 2
Newport News, VA IA
DIRECT
FILTRATION
                     East Bay MUD, CA I
                     Las Vegas, NV I
                     Little Rock, AR 1
                     Aurora, CO laA,5bA
                     Los Angeles DWP, CA 3
SOFTENING
Palm Beach Co., FL 2
Wausau, WI* 1
Minot, ND 1
Santa Monica, CA I
Macomb, IL* 1A
Galveston.TX 7c,4d
Louisville, KY 2
Ft. Meyers,  FL* I
Emporia,  KN* 2A
Omaha, NB 1A
DISINFECTION
ONLY
Mesa Consol., CA 3
North SkagitCo., WA* I
New York City, NY 1
Newark, NJ I
Note: Utilities participating in the California Public Health Foundation study are listed in bold type.
* Population under 50,000; all others over 50,000.
Key for chemical addition:

1  - chlorine only
2  - chlorine + chloramines
3  - ozone + chlorine
4 - chlorine + chlorine dioxide
5  - chloramines only
6 - chloramines + chlorine dioxide
7  - chlorine -f chloramines -I- chlorine dioxide
8  - ozone + chloramines
A - powdered  activated carbon
B - potassium permanganate
                           a - first quarter only
                           b - second through fourth quarters
                           c - first through third quarters
                           d - fourth quarter only
                           e - clearwell effluent sampled
                              before ammonia addition

-------
                              Executive Summary


used for presentation of project data.  This type of plot is presented and described in
Figure ES-1.

BASELINE SAMPLING RESULTS AND DISCUSSION

Table ES-3 summarizes the baseline median DBF values for each quarter as well as for
all  four  sampling  quarters combined.  It should be noted that these data  represent
clearwell effluent samples or samples collected at specified points after disinfection for
those plants without clearwells.  Some distribution system sampling was performed for
the process modification  studies (discussed  in  more  detail  below).  Results of the
process modification studies indicated that some  DBFs, such as THMs, increased in the
chlorinated distribution systems of some utilities,  while  their concentrations did not
change in the chloraminated  systems of other utilities.  In addition, it is important to
note that the disinfection practices of some of the participating utilities, such as the use
of chloramines.  are utilized  to meet the current TTHM regulation  and not to  meet
requirements of the proposed  Surface Water Treatment  Rule  (SWTR).   Thus,  some
utilities would produce different DBF  levels if their current disinfection  practices
required  modification in order to meet proposed concentration-time (CT)  requirements
of the SWTR.

Figure ES-2 is a summary of the four-quarter median concentrations of each DBF class
measured in this study. The median value of total  THMs (TTHMs)  was 36 /ug/L, and
the median  value  of HAAs  was 17 //g/L during  the four quarters of baseline  data
collection.  On a weight basis, THMs were  the largest class of DBFs detected in this
study (54.5  percent  of the total measured DBFs),  and HAAs were the second largest
fraction (25.4 percent of  the total)  (see Figure  ES-3).   A running annual average of
TTHMs  is  utilized to determine compliance with  the TTHM maximum contaminant
level (MCL) of 0.10 mg/L.  When the running  annual average (i.e., the mean of (he
four baseline  TTHM  values) was  computed for each utility,  the  35-utility  median
TTHM concentration was 39
TOC  analyses  were  performed on plant influent  samples after  the  first  sampling
quarter.  Figure ES-4 is a box-and-whiskers plot of the plant influent TOC data for the
summer, fall and winter quarters.   Note  that the "notches" in  the plots of the three
quarters overlap, indicating that there is not  a statistically significant difference, at  95
percent confidence, between the medians of any two quarters.  In Figure ES-5, influent
TOC values are plotted by source water type.  Again, there is no statistically significant
difference between the medians of any two sources.

Figure ES-6 is a plot of TTHMs by sampling quarter.  As would be expected,  TTHM
levels were highest in the summer quarter when water temperatures were the highest,
lower  in  the  fall, and  lowest in the winter and spring quarters,  although these
differences were not statistically significant at  a 95  percent confidence level.  Figure
ES-7  shows TTHM levels plotted  as  a function  of the  influent  water temperatures
measured  during baseline data collection.  In this  plot,  the  median TTHM level is
significantly higher in the highest  temperature range than in  the  lower three ranges,
illustrating the dependence of TTHM production on water temperature.

Figure ES-8 illustrates TTHM levels as a function of the disinfection scheme in  use  by
the utilities on  the day of  sample collection.  Only chlorination,  prechlorination/

                                     ES-3

-------
                      GUIDE TO NOTCHED BOX-AND-WHISKER PLOTS
                                                         D
                                            EXTREME OUTLIER
                                            any value outside 3 interquartile ranges
                                            measured from the 25 and 75 percentiles

                                            OUTLIER
                                            any value outside 1.5 interquartile ranges
                                            measured from the 25 and 75 percentiles

                                            MAXIMUM VALUE
                                            this is the largest value (excluding outliers)
INTERQUARTILE RANGE
contains the data between
the 25 and 75 percentiles
95% CONFIDENCE INTERVAL
 for the median (including outliers)
                                                                                75 PERCENTILE
                                                                                 75% of the data are less than or
                                                                                 equal to this value (including outliers)
MEDIAN
50% of the data are above or
below this value (including outliers)
        NOTE: Horizontal width of box is
        proportional to square root of sample size (N)
                                               1   -^—   25 PERCENTILE
                                                         25% of the data are less than or
                                                         equal to this value (including outliers)

                                               MINIMUM VALUE
                                                this is the lowest value (excluding outliers)
                                                  FIGURE ES-1

-------
                                  TABLE ES-3

          DISINFECTION BY-PRODUCTS IN DRINKING WATER

         SUMMARY OF BASELINE SAMPLING MEDIAN VALUES
                               1st         2nd        3rd         4th         All
Disinfection By-Products        Quarter    Quarter    Quarter     Quarter    Quarters
       (/fg/L)                 (Spring)   (Summer)    (Fall)      (Winter)   Combined


Trihalomethanes

   Chloroform                15          15          13           9.6         14
   Bromodichloromelhane       6.9        10          5.5         4.1         6.6
   Dibromochloromethane       2.6         4.5        3.8         2.7         3.6
   Bromoform                 0.33        0.57       0.88        0.51         0.57
   Total Trihalomethanes       34         44          40         30           36

Haloacetonitriles

   Trichloroacetonitrile        <0.012     <0.012     <0.029     <0.029      <0.029
   Dichloroacetonitrile          1.2         1.1         1.1         1.2          1.2
   Bromochloroacetonitrile       0.50        0.58        0.70        0.59         0.57
   Dibromoacetonitrile          0.54        0.48        0.51        0.46         0.50

   Total Haloacetonitriles        2.8         2.5         3.5         4.0          3.3

Haloketones

   I.l-Dichloropropanone        0.52        0.46        0.52        0.55
   1,1,1-Trichloropropanone       0.80        0.35        0.60        0.66

   Total Haloketones            1.4         0.94        1.0         1.8          1.2

Haloacetic adds

   Monochloroacetic acid      < 1.0         1.2       < 1.0
   Dichloroacettc acid           7.3         6.8         6.4
   Trichloroacetic acid          5.8         5.8         6.0
   Monobromoacetic acid      <0.5       <0.5       <0.5
   Dibromoacetic acid           0.9         1.5         1.4
   Total Haloacetic acids        18         20         21          13           17

-------
                                       Table ES-3

                     Disinfection By-Products In Drinking Water
                       Summary of Median Values, Continued

Aldehydes
Formaldehyde
Acetaldehyde
1st
Quarter
(Spring)

NA
NA
2nd
Quarter
(Summer)

5.1
2.7
3rd
Quarter
(Fall)

3.5
2.6
4th
Quarter
(Winter)

2.0
1.8
Ail
Quarters
Combined

3.6
2.2
     Total Aldehydes

  Miscellaneous
NA
6.9
5.5
4.2
5.7
Chloropicrin
Chloral hydrate
Cyanogen chloride
2 ,4,6-TrichIorophenol
Halogenated DBPsum
Total Organic Halide
Plant Influent Characteristics
Total Organic Carbon, mg/L
Ultraviolet absorbance, cm"1
Chloride, mg/L
Bromide, mg/L
0.16
1.8
0.45
<0.3
64
150

NA
NA
NA
NA
0.12
3.0
0.60
<0.4
82
180

2.9
0.11
28
0.07
0.10
2.2
0.65
<0.4
72
170

2.9
0.1 1
32
0.10
0.10
1.7
0.80
<0.4
58
175

3.2
0.13
23
0.07
0.12
2.1
0.60
<0.4
70
170

3.0
0.11
29
0.08
NA = Not Analyzed

Note (1):  Total class median values are not the sum of the medians of the individual compounds, but
          rather the medians of the sums of the compounds within that class.

Note (2):  The halogenated DBPsum median  values are not the  sum of the class medians for  all
          utilities, but rather the medians of the halogenated DBPsum values for .all utilities.  This
          value is only the sum of halogenated DBPs measured in this study.

-------

5
U
          Disinfection  By-Product  Concentration
                  by DBP Class    Four Quarters
7O

6O

50

40

3O

20

1O

 0
                                          THM
                                          HAN
                                          HK
                                          HAA
                                          ALD
                                          OP
                                          CH
                                          CNCI
                                          TCP
 Trihalome thanes
 Haloatstomtnies
 HBtokeTones
 Hotoacetic Acids
 Aldenydes
 Chtoropicrln
 Chloral Hy*ate
 Cyanogen Chloride
 TrlcNorocHenol
7S»1 I
                       VJSMWffA
                                                        -X--X--X-
            THM  HAN  HK  HAA  ALD   CHP   CH  CNCI  TCP

                               DBP  Class

                          FIGURE ES-2
Percent  of  Sum  of  DBP   Class  Medians
          By  DBP  Class    Four  Quarters
            HAA  25.4%
                                    ALD 8.5%
                                        MISC 5.O%
                                          HAN 4.9%

                                          HK  1.8%
                THM 5
ThM
HAN
I-K
HAA
ALD
MSC:
CHP
CH
CNCI
TCP
Tr ina tome rnanes
Haloacetom tr i tes
Haloketones
naioacetic Acids
Aioenyoes
OHoropicrin
Crvocai Hyoate
Cyarogeo Crtonde
Tricnloropnenol
                         FIGURE ES-3

-------
          Influent Total  Organic  Carbon
                    Bg Quarter
                            Influent Total Organic Carbon
                                       By Source
                         T
_J
a
E
0
.0

o
0
•-H
c
01
o
*J
0
                _L
 _L
                                                        •4
                  J
                   a
                   o
                  o
                   u
                  •H
                   c
                   a
                  o
                  •P
                   a
                                                                              T
                                                      T
                                                                    J_
               SUHHCR

                35
                         35
                      Quart i
UIHTER

 35
FS
33
It
   21
    7
Sourci
LR
51
17
                     RGURE ES-4
                                        FIGURE ES-5

-------
                   Trihalamethani

                      Bg  Quarter
                                         Trihalomethan«»

                                    Bg Influ«nt Temperature
                       T
    38*
T
J

O
•P
V

0
1.
h-
                               _L
                                                         M*
                      J

                      o
                      •P
                      M
                      E
                      0
                      ••4
                      L
                                                                            T
              SPRINQ

              35
                       35       35

                       Quarter
      UINTER

        35
8.8-18.1
                33.B-31.*
                                     21       1(7       'I*       27
                                     17       27       26       17
                                     Temperature Rang*  <°C)
                      FIGURE ES-6
                                                                          FIGURE ES-7

-------
                              Executive Summary


 postammoniation, and chloramination are shown since other disinfection schemes (such
 as preozonation/postchlorination) had very small sample sizes.  All other factors being
 equal, it  would be expected that the chlorinating utilities would produce  a higher
 median TTHM level than utilities using the other two  disinfection schemes.  However,
 Figure ES-8 indicates that the prechlorinating/postammoniating utilities had a higher
 median TTHM level than utilities employing the other two disinfection scenarios, and
 this  difference was  significant at a 95 percent confidence level.  The same trend was
 observed for the median level of the sum of halogenated DBFs (XDBP   ) measured in
 this  study.  The reasons for  this trend are not immediately apparent from the median
 influent TOC levels  for  the three  disinfection schemes, which  occurred within  the
 narrow range of  2.8  to  3.2  mg/L.   However, the  median values of  ultraviolet
 absorbance  at  254 nanometers  (UV-254)  for  the   prechlorinating/postammoniating
 utilities, indicate that these utilities had a higher UV-254 level than utilities employing
 either of the other two disinfection schemes, although the difference was not statistically
 significant at a 95  percent confidence level.  The  higher UV-254 levels may indicate
 higher levels of DBF precursors even though higher precursor levels were not reflected
 in the TOC  results.   Thus, the chlorinating utilities  participating in this  study may have
 been able to use free chlorine for oxidation/disinfection and still meet the TTHM MCL
 of  0.10  mg/L,  while  other   utilities  have  had   to  employ  free  chlorine  for
 oxidation/disinfection and then use ammonia at some point in their process trains in
 order to meet the TTHM MCL.

 In all. over  300 correlations were determined for the baseline data collected in this
 study. Correlations of non-THM DBFs with THMs were performed in order to explore
 the potential of using TTHMs as a surrogate for other DBFs  such as HAAs or HANs,
 since performing analyses for several different DBFs is beyond  the scope of a utility's
 routine monitoring  program.  Figure  ES-9 illustrates the correlations  of THMs with
 XDBPsum. the sum  of non-THM  DBFs, and HANs. As illustrated in the figure, there
 was  a strong correlation  between TTHMs and  XDBP   (r=0.96).    Since THMs
 represent the largest DBF fraction detected in this study, the data  were  re-evaluated by
 correlating TTHMs  with the sum of non-THM  DBFs  (XDBPs - THMs).   In this
 instance, r decreased to 0.76.   However,  this lower correlation  coefficient  does  not
 mean that THMs cannot be used as a surrogate or  predictor of the sum of all XDBPs.
 It should be noted,  though, that correlations between some classes of compounds were
 low  (for instance,  comparing TTHMs to  haloketones  yielded an r  of  only 0.06),
whereas for  TTHMs and HANs, r was equal to 0.78.

 Figure ES-10 illustrates the correlation of influent chloride levels with influent bromide.
There was a strong correlation  between these two  parameters (r=0.97), and  these
 results indicate that  chloride may be used as a predictor for bromide.  Using all of the
data points collected in the baseline sampling program (excluding three outlier values
from a utility with atypically  high bromide levels), linear regression analysis yielded  the
following equation:

                         [Br-]  = -0.0071 + 0.0034[C1-]

It is  significant to  note that high bromide levels  were detected  not only at utilities
impacted by tidal influences or saltwater intrusion, but at inland utilities as well.
                                     ES-4

-------
                  Trih«J.om«th«n««
              By Disinfection Schemi
J
a
3   «••
M

•
+J
tt
0
•H
L
                          I
                eta
              n -  78
              u -  21
CL2NH3
 dO
 1 I
II
 14
                Disinfection  Schsms
                    FIGURE ES-8

-------
       Correlations  with  Trihalomethanes
        XDBPsum  us  THMs
J
a
a
x
   400
   380
   280
   188
                     r-0.96
                             XDBPs-THMs  us  THMs
                                   168
                                   128
                                 a
                      j
                      a.
                      S
                      x
                                    40
          S8  188  158 288  288  300
               n-UO
              THMS
                                          r=0.76
                                60   100  150  288  250 300
                                     0=1^0
                                     THMS (ugxL)
        HANS us THM!
 -I
 a
 01
 
-------
           Correlations  with  Influent
               Chloride and  Bromide
         Br-  us  CL-
                                        CHBr3 us Br-
   2.5
J
\
a
a
a
   1.5
   8.5
                  r=0.97, 0.86*
                                 a
                                 o
                                    68
                                    48
                                    38
                                                           X '
                                        „«•
                                               ,  r=0.57, 0.69-
           280    480   688
            n-105, 102*
         Influ«nt Chiorid* 
-------
                               Executive Summary


 Figure ES-ll  shows the  relationship of influent  bromide  to  levels of chloropicrin,
 1,1,1-trichloropropanone (1,1,1 -TCP) and trichloroacetic acid (TCAA) measured during
 baseline data collection.  In each case, an exclusion relationship is demonstrated, i.e.,
 the presence of bromide  appeared to exclude  the presence  of the particular DBF.
 Correlations  of  the  four  THM  compounds  with dichloroacetic acid  (DCAA)  are
 illustrated in Figure ES-12.  The best correlation was found with chloroform (r=0.86).
 However,  as  the  THMs  shift to  the  more  brominated species, the  correlation
 coefficients decrease  until an  exclusion relationship  is observed  between DCAA and
 bromoform (r = -0.33).  This progression is consistent with the finding that bromide and
 various chlorinated DBFs were  related by  exclusion.

 Of the 35 utilities included in this study, only three employed ozone, yet almost all had
 detectable  levels of formaldehyde and acetaldehyde in the  clearwell  effluent samples.
 These aldehydes were also found in some plant influent samples, and chlorination alone
 was found to produce these compounds.

 Chloramines are recognized as  an effective control strategy for  THMs and other DBFs.
 However, for most waters studied in this project, concentrations of cyanogen chloride
 were  found  to  be  significantly  higher in  chloraminated waters  as  compared  to
 chlorinated  waters.  Moreover,  it was possible to statistically divide the  distribution  of
 cyanogen chloride  into two groups,  depending on whether the final disinfectant was
 chlorine or chloramines.

 Figure ES-13 illustrates the removal of TOC within the filtering plants included in the
 baseline sampling  program.    Overall,  TOC  removal within  these  plants  averaged
 approximately  24 percent.   It should  be  noted that  the  treatment  practices  of the
 utilities participating in the baseline sampling program most likely focused on turbidity
 control and  were not optimized with respect  to TOC removal.   Results of the two
 treatment modification  studies  on improved  coagulation  (described in more detail
 below) indicated that  these two utilities were able to achieve higher TOC removals than
 those indicated in Figure ES-13; however this ability may be source-water specific.

 The TTHM  data from  this study of 35  utilities was  compared to that of  the THM
 survey conducted by  the American Water Works Association Research  Foundation  in
 1987.  which  involved  727 utilities around  the  nation.  Since compliance  with the
 TTHM MCL of 0.10 mg/L is based on  a running annual average, mean  TTHM values
 were computed for each of the 35 utilities  in  this  study for the 4 sampling  quarters.
 The AWWARF survey utilized the means  of three years of quarterly data. The TTHM
 means for  bolh projects are plotted  in  Figure ES-14, which  illustrates that the two
 frequency distributions were very similar.

 TREATMENT MODIFICATION STUDIES - RESULTS AND  DISCUSSION

 Ozonation Studies

 Because of the increasing use of ozone in  the United States for  disinfection and control
 of DBFs, five treatment modification studies focused on  the use  of  ozone  at various
water treatment plants.
                                     ES-5

-------
Correlations  with  Influent Bromide
   CHP  us  Br-
                               1,1,1-TCP  us Br-
2

1.6
i
1.2
C
>«J
U
•4
* e.e
i
a
I

a





o


' a
«
J.
e.4 r •

t
fa r=-0.22, -0.25*
Uf •&*«,-. . "^

8

1
? *
s
y
L
§• *
L
0
£
U
•H
1.
H 2
Jj
A
*• ^

a
a


B

i

i
•

B

A i
•<* t^0.° r=-0.26, -0.35-
ffi-*E»-q.* S - N*V>1
               2.8  3
        n«10l», 101*
   Influent Browid*
                                     n-105, 102*
                                Influ.nt Bromid. (mgxL)
  TCAA us Br-
                                   CHC13 us Br-
au

i ,.
J
u
H 28
w
5
a
a
H
fi "
a

a
l'.
>,
i a
a
a
[• °
o
a
L ti •
1- I'- •
»• . r— 0.27, -0.35*
C
0 C^^^SEL! 	 2 	 i5j
158

128
i .

?
0
5 68
0
•^
5

38



8

a

' o
a

a
a
a
o

M
^ !

e
Jt\.m r— 0.25, -0.29*
«• °^-rt= . . v*>>'>>
8  8.5  1  1.5  3  2.5  3
         n-105, 102*
   Influent Bronid* (mg/L)


* Excludes indicated outliers
                            «  9.5  !
                                      1.5   2  2.5
                                     n-105, 102*
                               Influent Bromide (mgxL)
                  FIGURE ES-11

-------
<
u
J
a
a
       Correlations  with  Dichloroacetic Acid
          CHC13 us DCAA
    se
    68
    38
             •«i
    •/*:•  '•--
    °.f
                  r»0.86
                                   188
                                   68
                              CHCl2Br  us  DCAA
0

|
Q
                                   48
                                   28
   88
   »
   »
   28
  18   28   38   48   88

Oichloroccvtic Acid 
-------
(13
O
0)
DC
U
O
\-
(D
O
4^

-------
             AUUARF  12-Quarter  us.  USEPA

                   4-Quarter Means
    499
J

01
3
 in
 01
 c
 ID

4J
 01
 E
 0
-H
 10

•H
 L
    309
299
    let
                   AUIUIAW

                     727
                             USBPA

                              35
                         Study
                        FIGURE ES-14

-------
                               Executive Summary


 The study at Utility 6 involved  sampling before and after ozonation was incorporated
 into the treatment process of a 52-mgd plant treating a reservoir water.  A conventional
 treatment process was  used  before ozonation was installed, with chlorine added to the
 plant  influent,  flocculator influent,  and filter influent and effluent.   Ammonia was
 added after filtration.  When ozonation was installed (in combination with the rapid mix
 process)  the plant  was  modified to  direct filtration, with  flotation  and  skimming
 replacing  the  flocculation  and  sedimentation   processes.   As a  result  of ozone
 implementation, the plant was able to reduce chlorine doses and free chlorine contact
 time.   The  "before" and "after" samples (before and after ozonation was placed on-
 line) were collected approximately five months  apart.  Figure  ES-15 summarizes  the
 impact of ozonation on DBF levels measured in  the utility's distribution system after a
 residence time of approximately 7 hours. The figure illustrates that reductions of 56 to
 66 percent occurred for all XDBP classes after the implementation of ozonation, with
 the  exception of HKs which increased slightly.   Chloral hydrate,  chloropicrin and
 cyanogen chloride concentrations also decreased.  However, from this study,  it  was not
 possible to attribute the lower levels of most DBFs directly to ozone since changes in
 treatment process, and lower chlorine doses and shorter contact times were employed.
 It cannot be determined  if ozone caused a decrease  or modification of DBF precursor
 material or  if  lower  DBF  levels  can  be  attributed solely to the decreased use  of
 chlorine. From a full-scale perspective, however, it is notable that the application  of
 ozone decreased the dependence on  chlorine for oxidation and  disinfection, with the
 overall result of decreased DBF concentrations.

 Three different treatment  scenarios were studied at Utility 7, which operates a 400-mgd
 conventional  treatment  plant.  Chlorine-only (using chlorine doses of 2.3 mg/L at the
 plant influent and 1.1  mg/L at the filter influent) and prechlorination/postammoniation
 (chlorine doses  of 2.3  mg/L at the plant influent and 0.6 mg/L at  the filter influent;
 ammonia dose of 0.49  mg/L as ammonia-nitrogen at the filter effluent) were studied  at
 full scale and a  6-gpm pilot plant was utilized to  study preozonation/postchloramination
 (ozone dose  of 2.0 rng/L before rapid mix; and 0.5 mg/L ammonia  and  1.5 mg/L
 chlorine at the filter effluent).  Samples were collected for each scenario after 2-hour
 and  24-hour  simulated  distribution system (SDS)  tests were conducted.  In these tests,
 samples were dosed with disinfectants and  held under  conditions representative  of
 Utility 7's distribution system to provide an estimate of the levels of DBFs that would
 have been produced under realistic environmental conditions.  The effects  of ozonation
 on DBF levels after  the 24-hour SDS  test are illustrated in  Figure ES-16.  Preozonation
 followed by  concurrent addition of ammonia and chlorine after filtration decreased the
 levels  of THMs. HAAs,  HANs,  and chloral hydrate as compared to chlorine-only  or
 prechlorination/postammoniation  treatment.  Concentrations of 1,1,1-TCP were lowest
 for  ozone/chloramines treatment,  higher  for  prechlorination/postammoniation  and
 highest  for   chlorination   treatment,   while    1,1-dichloropropanone    (1,1-DCP)
 concentrations exhibited the opposite  trend.  Very little difference in chloropicrin levels
 were observed between the three treatment schemes; however,  all levels were less than
 I yug/L.  Cyanogen chloride  levels were highest for the ozone/chloramines  treatment
 after 2 hours of holding  time, but after 24 hours, there  was very little difference  in
 cyanogen chloride concentrations between the three  treatments.  Aldehydes were not
 measured at this utility in  this study.

 Utility 19 operates a large (600-mgd) preozonation/direct  filtration facility.   Chlorine-
only (1.8 mg/L of chlorine added to the plant influent and 0.3  mg/L added to the filter

                                      ES-6

-------
 _t
 •&»
 o
U

§5
                 Effect  Of  CI2.NH2CI  and  O3.CI2.NH2CI
                        on  DBP  Formation  (Utility  6)
         15O
         100
 50
           0
                                    CI2NH2CI
                                    (11/21/86)
                                ESS) O3.CI2.MH2a
                                    (5/15/89)
                                Dlst. Sys, Residence Time (t)
                                 t = 7 hrs
                                          • TCMM MOW l
                                          «-» MBAA b*low
                                                               NA
              XDBPsum   THM
                         HAN
HK
HAA
ALD
                                    DBP Class
                                 FIGURE ES-15


                Effect  Of CI2.  CI2/NH3  &  O3/NH3/CL2
                       on  DBP  Formation  (Utility  7)
         200
                                          ES3 CIS
CO
h_
*-•
$
U
5
U
Q
         150
1OO
          50
           0
      12

      XD8PSUH
                                Diet. Sys. Residence Time (t)
                                1: t = 2 hrs
                                2: t = 24 hrs
                                * Below
                                  12      12     12      12
                                  HAN        t-K       HAA      2.4-OCP

                                   DBP Class
                                   FIGURE ES-16

-------
                               Executive Summary


 effluent) and preozonation/postchlorination (ozone dose of 1.7 mg/L, and chlorine dose
 of 1.5 mg/L added to the filter  effluent)  treatments were  studied.   Samples were
 collected  at the  plant's clearwell effluent and in the distribution system at residence
 times of 4.3 and 11 hours.  Figure ES-17 shows  concentrations of DBFs after 11  hours
 of residence time in the distribution system.   Decreases of 13 //g/L and 8.7 /ug/L were
 observed  for TTHMs and HAAs, respectively, after implementation of ozonation with
 subsequent  chlorination.  A 2.3 //g/L increase in chloral  hydrate was observed.  After
 ozonation, HANs were decreased by 1.2 //g/L, and small increases were observed  for
 HKs and  chloropicrin.   The cyanogen chloride analysis was not  performed during  the
 ozonation trial,  and this compound was  only slightly detected during the chlorine-only
 trial.

 Sampling at Utility  25  was  conducted at full  scale at a 90-mgd conventional treatment
 facility.    Samples  were  collected  before  the  plant went on-line  with  two-stage
 ozonation.     Chloramines-only  treatment  was   utilized  before   ozonation   was
 implemented,  with  disinfectant addition  limited to the concurrent addition of ammonia
 (1.6 mg/L) and  chlorine (8.0 mg/L) at the rapid mix.   After the ozone system was
 placed on-line, ozone was applied to both the raw and settled water at 4.0 mg/L  per
 stage,  and chlorine (5.0 mg/L) and ammonia (1.0 mg/L) were added concurrently prior
 to filtration.  Samples  were collected at various  points in the plant:  2nd-stage ozone
 contactor influent (2OI), filter influent (FI), filter effluent  (FE), clearwell effluent  (CE);
 and at four distribution system locations:  at 4 to  5 hours of residence time (LI), 8 to 9
 hours of residence time (L2),  9 to 10 hours of residence time (L3), and 18 to 20  hours
 of residence time (L4).  Figure ES-18  presents  the profile of TTHM concentrations
 through the plant and distribution system.   The data show that  for  both treatment
 scenarios, THMs were formed immediately  after chlorine and ammonia addition and
 remained  stable  through the plant  and  into  the  distribution  system,  but  TTHM
 concentrations were significantly lower after  ozone  was  incorporated  into  the plant's
 treatment process.   The same trend was observed for levels of XDBPsum  and HAAs.
 The  unexpectedly high levels of DBFs  produced by chloramines-only  treatment at
 Utility 25 may have been due to poor mixing upon concurrent addition of chlorine and
 ammonia, and/or to the high  influent TOC concentration  of this  utility's source  water
 (7.7 mg/L).  Concentrations of chloropicrin and ALDs were substantially higher in  the
 plant  and distribution system  during the ozonation trial compared to the chloramines-
 only experiment.

 At Utility 36,  a  5-gpm  pilot plant was employed  to evaluate the effect  of five different
 treatment  scenarios  on the formation of  DBFs. Conventional treatment was employed
 with options to  add preozonation (2 mg/L), with  and  without hydrogen peroxide
 addition (0.67 mg/L), and free chlorine (6.5  mg/L) or chloramines (2.1 mg/L chlorine
 and 0.5 mg/L ammonia) as disinfectants in the rapid mix.  Samples were collected from
 the filter effluent and held for 24 hours at ambient temperature in  order to simulate the
 residence  time in a distribution system.  Figure ES-19 illustrates  the effect  of the five
 different disinfection schemes on various  DBF classes and  chloral hydrate.  The highest
 levels  of  TTHMs.   HAAs.  HANs and  chloral  hydrate   were  observed  with   those
 disinfection   schemes   which   employed   chlorine   as   the   final   disinfectant.
 Preozonation/postchlorination  appeared  to slightly increase the  level of TTHMs and
 slightly decrease  the concentration of HAAs;  chloral hydrate was increased.   Large
decreases were observed under any disinfection scheme which employed chloramines as
a disinfectant.  For example, as compared to chlorine-only  treatment, chloramines-only,

                                     ES-7

-------
EFFECT OF VARIOUS DISINFECTION SCHEMES ON
  DISINFECTION BY-PRODUCTS FORMATION
              AT UTILITY 19
                         NA: Not Analyzed
                FIGURE ES-17

-------
s
•i—'
(0
g
o
u
16
*—'

O
70




60




5O




40




30




20




10




 O
            Effects of  NH2CI  and  O3/NH2CI  Treatment

               on  Total  THM  Formation  at  Utility  25
            V777A NH2CI



            H^B O3/NH2CI
NA  = Not analyzed

*-**•  = No lime addition
                       -*-*-
            RAW  2OI    Fl    FE    CE    L1    L2    l_3   L4




                         Treatment &  Residence  Time


                                FIGURE ES-18
        EFFECT OF VARIOUS DISINFECTION SCHEMES ON

   DISINFECTION  BY—PRODUCT FORMATION AT UTILITY 36
                               FIGUREES-19

-------
                               Executive Summary


 o/one/chloramines and  ozone/hydrogen  peroxide/chloramines decreased TTHM levels
 by 96. 97 and 98 percent, respectively.

 At Utilities 19 and 36. the use of preozonation/postchlorination resulted in a shift to the
 brominated THMs and  HAAs.  Ozone can react with  bromide ions in the  raw water
 causing the formation of hypobromous acid (HOBr).   Reactions of HOBr with natural
 organic material  can  produce bromoform, dibromoacetic  acid, and  other brominated
 DBFs.  This  was found to occur even at a utility with  low levels of influent bromide,
 such  as  Utility   19,  which  typically  treats  water  having  an   influent  bromide
 concentration of only 0.04 mg/L.

 Ozonation was found to increase  aldehyde concentrations at  Utilities 6, 25  and 36.
 Aldehydes were not measured at Utilities 7 and  19 during the treatment modification
 studies.  However, subsequent pilot testing at Utility 7 indicated that  aldehydes were
 formed  in the ozone  contactors.   When the pilot  filters at this utility were operated
 without the upstream  addition of a secondary disinfectant, aldehyde concentrations in
 the  filter effluent  were lowered to  detection limits or below, suggesting that biological
 activity within the filter bed was responsible for aldehyde removal.  Utilities 6, 19, 25
 and  36 employed  secondary  disinfection prior to  filtration  and,  thus,   may  have
 precluded  or  reduced  the  potential  for  aldehyde  removal  within  their  filters.
 Furthermore,  when Utility  7 applied chloramines before the  pilot filters,  aldehyde
 concentrations were not reduced through the filtration step.

 Chlorine Dioxide Studies

 Chlorine dioxide  studies were conducted at two  utilities.   Both studies evaluated the
 combination  of chlorine dioxide/chlorine as a DBF control  method  compared to free
 chlorine.

 Utility 16  operates a 400-mgd direct  filtration treatment system with free chlorine for
 both preoxidation and final  disinfection,  but  periodically switches to chlorine dioxide
 preoxidation  to control THMs and  taste and odor.  Samples were collected in the plant
 and  at two distribution system locations (approximate residence times of 45 minutes and
 7 days).  On  the chlorine-only sample date, chlorine doses were 2.0 mg/L at the plant
 influent and  1.0 mg/L at the  filter  effluent.  For the chlorine dioxide test, the  chlorine
 dioxide dose was 0.5 mg/L at the plant influent and the chlorine dose was 1.9  mg/L at
 the filter effluent.   Figure  ES-20 illustrates levels of DBFs  measured in the distribution
 system.  There were very little differences observed  in DBF levels produced by the two
 different oxidation/disinfection schemes.

 Utility 37's 30-mgd treatment facility provides for two separate treatment trains, one of
which has provision for chlorine dioxide preoxidation, and one which employs free
chlorine.   Both treatment  trains have free chlorine as  a final disinfectant.   For this
study. 0.9 mg/L  of  chlorine dioxide was added to the  flocculator effluent of one
treatment train, and 2.25 mg/L of chlorine was added to the other treatment train for
preoxidation.  Although detectable levels of free chlorine were measured in the chlorine
dioxide treatment  train, and vice  versa, the  preoxidant  in one treatment  train was
predominantly chlorine dioxide (referred to as "chlorine dioxide" treatment), and in the
other  train, the preoxidant  was  predominantly chlorine  (referred  to as  "chlorine"
treatment).  Samples were  collected at the plant's sedimentation basin effluent, after a

                                      ES-8

-------
                               Executive Summary


 residence time of approximately 2.5 hours and before chlorine was added for residual
 disinfection.  Figure ES-21  shows the effect of the two different preoxidants on the sum
 of measured halogenated DBFs and the DBF classes.  From this figure, it is apparent
 that even in the relatively short detention time in the sedimentation basins, the use of
 chlorine dioxide preoxidation resulted in lower levels of all measured DBFs  compared
 to chlorine treatment.   XDBPSIM11 was  almost  50 percent  lower with  chlorine dioxide
 treatment.

 Coagulation Studies

 Coagulation studies were conducted at  two utilities  to evaluate the effect of coagulant
 dose on DBF formation.  These utilities were selected because they had the capabilities
 to adjust alum  doses at full scale without severely  compromising the quality of their
 finished water.

 Utility  3 operates a conventional  treatment facility  with a  10.5-mgd capacity.  Alum
 doses were  varied from low (10 mg/L) to medium (19 mg/L) to high (40 mg/L) for the
 study, and coagulation pH was held constant at pH 5.5.  Chlorine was not added until
 the  clearwell influent:  therefore, samples were  collected at  the  filter  effluent  and
 24-hour SDS tests were conducted  using a chlorine dose  of 3.5 mg/L.   Removal  of
 TOC increased from 25  to 50 percent in the filter influent as the alum dose  increased
 from 10 to  40 mg/L.  At the low alum dose,  a greater percent of TOC was  removed
 through filtration (20  percent) than  at the medium and  high  alum doses (15 and 9
 percent, respectively).  This was most likely due to the better settling characteristics of
 the floe at the higher doses.  Figure  ES-22 illustrates the effect  of increasing alum dose
 on XDBP    and the DBF classes.  In  general, DBFs decreased with  increasing alum
 dose.  XDBPsum was lowered from 150 to 94 /vg/L, and TTHMs from 86 to 55 /ug/L as
 alum doses increased from 10 to 40 mg/L.

 The study at Utility 12  was conducted at  a  conventional  treatment facility with a
 capacity of  72 mgd.   Chlorine  was added  at  two locations  in  the plant:  before the
 rapid mix (1.8 mg/L dose)  and before  filtration (1.3 mg/L dose).  The total  chlorine
 contact  time was approximately 100 minutes.   Ammonia was  added approximately 4
 minutes after filtration, prior to the clearwell.  Samples were collected at the plant
 influent, sedimentation basin effluent, filter effluent and clearwell effluent.  Alum doses
 were varied  from low (24.6 mg/L) to medium  (45.7 mg/L) to  high (73 mg/L) for the
 study.    Since the plant had no capability  to  control  pH  before  or  during  the
 sedimentation process,  the pH values decreased as the alum doses increased.   The  low
 alum dose removed 33 percent of the influent TOC, while  the  medium and high alum
 doses removed 47 and 46 percent,  respectively,  as measured in the filter effluent. Most
 of the TOC removal occurred in  the  sedimentation basins,  with little or no additional
 removal occurring through filtration.  Figure ES-23  illustrates the effect  of alum dose
 on DBF concentrations by class.  For XDBP  , concentrations decreased from 87 to 69
//g/L as the alum  dose increased from  25 to 75  mg/L.   For TTHMs, the levels
 decreased from 53  to  39 x/g/L.   In general,  individual  THMs and HAAs decreased
 slightly  with increasing alum dose; little or no change was observed for HKs, HANs or
 ALDs.  That greater DBF removal  was not observed with increasing alum dose  was due
to the utility's prechlorination practices.   Approximately 1.8 mg/L of free chlorine was
added to the raw water, with 75 minutes of contact time from the point of addition to


                                     ES-9

-------
 o
 s
 u
                    Effect  Of  CI2  and CIO2/CI2
                   on  DBP Formation (Utility  16)
6O.O
5O.O
40.0

3O.O
20.0
10.0
O/-\
.o

-
.
.

-



X



^
2
/
/
1
oe







2
PSLfl
?




£
/
1
n






^

TVtv/


7
/
/"
^
X
?
2

El
^^ CIO2/CI2
Dist. Sya Residence Time (t)
1: t = O.75 hrs
2.- t = 16O hrs
MA - Not Analyzed


.fl >- ll 11.
12 12 12 12
HAN HK HAA ALD
                              DBP Class
                           RGURE ES-20
_J
01
O

-------
_J

^

g
•*—'
10

                Effect  of  Alum  Dose
            on  DBP  Formation  (Utility  3.  SOS)
200
150 -
100 -
                                   Low Alum Dose
                                         Medium Alum Dose

                                         High  Alum  Dose
                                     Ixtaw »*».
                                ~ TON BCAH mo COM Mo» !>*»-•
                                •» MBAA and OBAA
                                • OfUA bM»r »*«.
     XDBPsum   THM
                              HAN
                                       HK
HAA
ALD
                          DBP Class

                        FIGURE ES-22
_j


s
*-t

0)
o
U
                Effect  of  Alum  Dose
              on  DBP  Formation  (Utility  12)
                                   Low  Alum Dose


                                   Medium Alum  Dose
                                   High Alum Dose
                               »  TCAN below M3L
                              **  ACBIataerryOo batow
                             »•»  Acataldehyda not analyzed
     XDBPsum   THM
                              HAN
                                       HK
HAA
ALD
                          DBP  Class

                        FIGURE ES-23

-------
                             Executive Summary


 the sedimentation basin effluent.  Consequently, DBF formation occurred before and
 during TOC removal processes.

 As discussed  previously,  mean  overall  TOC removal  through  the filtering plants
 participating in the baseline sampling program was 24 percent.  It is important to note
 that for those  utilities capable of increasing TOC removal by increasing applied alum
 doses, the enhanced precursor removal represents only an incremental increase over that
 achieved under normal operation. For example, at  Utility 12, TOC removals  of 24 to
 35 percent were observed from plant influent to clearwell effluent during baseline data
 collection (normal operation), and TOC removals up to 47  percent were achieved  by
 increasing alum doses in the treatment modification study.

 Granular Activated Carbon Study

 The granular activated carbon (GAC) study involved the collection of samples in Utility
 11's 235-mgd  conventional treatment facility and 2-mgd GAC demonstration plant over
 a  period  of approximately 4  months.  Unchlorinated water was diverted from  the
 conventional plant after  presedimentation (alum and polymer addition  followed  by
 lamella separation and 3-days of off-line storage), and through a rapid sand filter before
 application to  the GAC column.  The GAC was Filtrasorb 400 (12 x 40 mesh) and the
 column was operated with an empty bed contact time of 15 minutes.  Samples were
 collected at the GAC column influent and effluent and 3-day  SDS tests were performed
 using 4.5 mg/L  of  chlorine  in  order to evaluate  DBF production.   Samples were
 collected on GAC column Run Days 0.2, 13, 25, 54, 82 and 95.  The TOC  removal
 performance of the GAC column is  illustrated in Figure ES-24.  The GAC was  very
 effective  for TOC removal at Utility 11,  although research  at other  utilities indicates
 that this  technology  has site-specific benefits.  For the first  25 run days, the column
 effluent TOC remained the same as  that measured on  Run Day 0.2 (O.I mg/L), which
 represents the  non-adsorbable  fraction of the  TOC.   On Run Day 54, the  column
 effluent TOC had only increased to  0.2 mg/L, and it was not until the 82nd  run day
 that a substantial increase in column effluent  TOC was observed (0.6  mg/L).   The
 impact of TOC removal on levels of SDS THMs is illustrated in Figure ES-25.  This
 figure shows that SDS TTHMs were extremely low (approximately 6 //g/L or less)  in
 the column effluent for the first 54 run days,  indicating very  effective  removal of THM
 precursors. After 82 run days, SDS  TTHMs had increased to 34 //g/L and after 95 run
 days. SDS TTHMs were  47  /ug/L.   Comparing levels of SDS TTHMs in the column
 influent and effluent. GAC treatment led to reductions in SDS TTHM levels of >97,
 >92, >97 and  >96 percent on  each sampling day through  the 54th run day, and 71
 and 65 percent on the 82nd and 95th  run days, respectively.

 Despite the low  bromide  levels in Utility 11 's raw  water (less than or equal  to 0.06
mg/L), GAC effluent SDS samples   on the  82nd and 95th run  days had levels  of
bromoform and  dibromoacetic acid  that  exceeded  levels  in the  GAC influent SDS
samples.   The  increase in the percentage of brominated THMs and HAAs may  be due,
at  least in part, to the increased ratio of bromide to  precursor material after significant
levels of TOC have been removed in the GAC contactor.
                                   ES-10

-------
8
         O
               TOG  vs  Run  Time  for  GAG
             Column  Influent and Effluent.  Utility  1 1
            -+GAC Inf             A	A GAG Eff
              Immediate                Immediate
          	A	---- A
          _J_J	L I II I __l J	j_A. t..-l	1 t t
 20
40
6O
                             Run Days
                             FIGURE ES-24
80
                                                     .A
1OO
         SDS  TTHMs  vs  Run  Time  for  GAG
            Column  Influent  and  Effluent.  Utility  1 1
        +	+GAC  Inf             A	AGAC  Eff
             3-day SDS               3-day SDS
     2OO
     15O
     1OO
      5O
        O
             CI2 = 4.5 mg/L
             pH = 8.2
                   25 C
                '"•
20
                          4O
         6O
                            Run Days
                            FIGURE ES-23
                          ** All Tt-tM
                                              ..A-"
         80
                                      1OO

-------
                              Executive Summary


SUMMARY AND CONCLUSIONS

THMs. on a weight basis, represented the largest class of DBFs measured in this study
and  the  median TTHM  value for the  four quarters of baseline sampling of clearwell
effluents was 39 A/g/L (computed as the median  of running annual average TTHMs for
each individual utility). HAAs were the second largest fraction detected, and aldehydes
were the third largest.  Little difference was observed  in the concentrations of influent
water  quality parameters  and  concentrations of  DBFs on  a seasonal basis  when
considering the  overall  medians for all 35  participating utilities; however, seasonal
variations were observed for individual utilities.  In addition, levels of  XDBPsum and
TTHMs were found to depend on water temperature.

The  numerous correlations conducted for this study indicated that TTHMs correlated
well  with the sum of halogenated  DBFs measured in this study and with some DBF
classes  (e.g.,  HANs), while correlations  between  THMs  and other classes of DBF
compounds (e.g., HKs) were low; that  bromide levels could be predicted  from chloride
concentrations; and  that bromide present  in  raw waters impacted the  speciation  of
THMs, HANs and HAAs.   In  addition,  the  shift  in  speciation  to brominated  DBFs
occurred even in inland utilities not impacted by saltwater intrusion, and  in waters low
in bromide after preozonation or GAC treatment.

Ozone in conjunction with chlorine or chloramines  as  final disinfectants  was  generally
effective in lowering  concentrations of classes of halogenated  DBFs.   The  extent  to
which  halogenated  DBF levels were decreased  or  increased after implementation  of
ozonation depended primarily on the final  disinfectant which was employed.  Aldehyde
levels  were found to  increase upon ozonation.  although  it was found  in  one  case
possible  to remove aldehydes within the  treatment  plant when filters were  operated
without a residual disinfectant.

Chlorine dioxide was  effective for controlling DBFs at one utility, while at another, no
difference from DBF levels resulting from the utility's normal chlorination practices was
observed.  Coagulation was effective in removing DBF precursors at the two utilities
studied, as long as chlorine was not added before the DBF precursors were removed in
the coagulation, flocculation, sedimentation  and filtration processes.    GAC was an
effective  technique for controlling levels of DBFs by removing  precursors at the one
utility studied.
                                    ES-11

-------
Section 1
Introduction

-------
                                  SECTION 1

                               INTRODUCTION


 The  United  States  Environmental Protection  Agency  (USEPA)  will be  developing
 regulations to control disinfection  by-products (DBFs) in drinking  water as a result of
 the  1986  amendments  to  the Safe  Drinking  Water  Act  (SDWA).   Under these
 amendments, the USEPA is required to develop a priority list of chemicals that may be
 present in drinking water and to develop maximum contaminant levels (MCLs) for those
 compounds.  Included on this list  are disinfectants, trihalomethanes (THMs) and other
 DBPs.   Although  the  schedule  for  promulgation of  regulations under the  SDWA
 amendments  remains uncertain at this time,  the  USEPA  anticipates proposing DBP
 regulations in September, 1991 and finalizing the  regulations by September,  1992.  If
 the provisions of the SDWA are to be met within the required regulatory timetable, the
 presence and control of the target DBPs must be fully understood.

 PROJECT BACKGROUND AND OBJECTIVES

 In October.  1987. the Association of  Metropolitan Water  Agencies  (AMWA) entered
 into a cooperative agreement with the USEPA to develop information on the formation
 and  control of DBPs in full-scale drinking water treatment systems.  AMWA contracted
 with the Metropolitan Water District of Southern California (Metropolitan)  to perform
 the study.  Engineering services for the study were  provided by James M. Montgomery,
 Consulting Engineers, Inc. (JMM) through a subcontract with Metropolitan.

 Project Objectives

 The principal objective  of  the study  was  to collect data from  representative water
 utilities in the United States on the occurrence and control of DBPs  in drinking water.

 Specific objectives of the project included:

     o   Determine the  occurrence of DBPs at  25 drinking  water treatment facilities
         around the nation.  Facilities were  selected to provide a broad range of source
         water qualities and treatment processes.

     o   Determine the  seasonal  nature  of  the occurrence of DBPs as a function  of
         temperature,  total  organic  carbon  (TOC),  pH,  and other water quality
         parameters.

     o   Determine the effect of changes in  treatment processes and/or disinfectants on
         the  production of DBPs at bench, pilot and/or full scale at up to 10 drinking
         water treatment facilities.

The  study focused on  the identification  of  DBPs  expected in United States  drinking
waters as a function of  source water  quality, water treatment process selection  and
operation, and disinfection processes and chemicals. Previous studies or those currently
underway at  the  USEPA  have  been designed to  define the occurrence and  levels  of
DBPs in a broad  sampling of water treatment  systems.  This study, however, focused

                                      1-1

-------
                                  Introduction
 on the  relationships  hetween source  water quality,  processes  employed  at  water
 treatment facilities, and the level and frequency of occurrence of DBFs.

 Target Compounds

 The DBFs  of interest  to  the project  are  listed  in  Table 1-1, and include THMs,
 haloacetonitriles  (HANs). haloketones  (HKs), haloacetic  acids  (HAAs), chloropicrin,
 chloral hydrate, cyanogen chloride. 2,4.6-trichlorophenol, and aldehydes (ALDs). The
 chemical structure of each of the target compounds is illustrated in Figure 1-1.

 All of the  halogenated DBF compounds selected for study appear  on the  USEPA's
 Drinking Water Priority List of contaminants expected to be regulated  by  1991. The
 aldehydes (formaldehyde and acetaldehyde) were  added to the  study since ozonating
 utilities were  included in the study and these compounds  have  been identified  as by-
 products of the ozonation process (and speculated to be  chlorination  by-products as
 well).  In addition, the USEPA is considering the regulation of these  aldehydes as part
 of an ozone DBF priority list.

 California Department of Health Services/California Public Health Foundation DBF
 Study

 The USEPA  DBF project  was conducted in  conjunction  with  a  similar study being
 funded by a grant from the California Department of Health Services (CDHS) and the
 California Public Health Foundation (CPHF).   Metropolitan and JMM  also conducted
 the CDHS study.  The CDHS study involved 10 utilities around  the State of California
 selected  to  provide a  broad  range of source water qualities and treatment processes.
 The CDHS  study focused on utilities with source waters representative of supplies used
 by the majority of consumers in California.

 Conducting  the two studies simultaneously has been beneficial for both the USEPA and
 the CDHS.  Data for all 35 utilities involved in the combined studies has been available
 to both  funding  agencies by providing each  agency with copies  of progress  reports
 prepared  for  the  other agency's study.   In  addition, the combined  USEPA/CDHS
 studies provided the opportunity to study a larger number of utilities, allowing a more
 representative  selection of utilities.   Data in this  report  reflect results from  the  35
 utilities participating in the combined USEPA/CDHS studies.  A  separate report was
 prepared for  the  CDHS study (Metropolitan, 1989), focusing  on the  10 California
 utilities.

 PROJECT DESCRIPTION

 The USEPA project has been conducted over a period of two years.   The first year of
 the project  (October,   1987  through  September,  1988) focused on  establishing and
 verifying  the   analytical  procedures   at  Metropolitan's  Water Quality  Laboratory,
 selecting utilities to participate in  the study, developing  DBF  baseline data through
 implementation of a quarterly sampling program  at  the 25 utilities, and  performing
 process modification studies  at two utilities.  During the  second  year of the  project
 (October. 1988 through September, 1989), baseline data collection was  completed and
 process modification studies  were conducted at  six utilities.  Project status  has been
summarized in quarterly progress  reports submitted to the USEPA.

                                       1-2

-------
                                TABLE 1-1

              LIST OF COMPOUNDS TARGETED IN STUDY
                               Compound*?
                            Trihalomethanes

                               chloroform
                               bromodichloromethane
                               dibromochloromethane
                               bromoform

                            Haloacetonitriles

                               tricliloroacetonitrile
                               dichloroacetonitrile
                               bromochloroacetonitrile
                               dibromoacetonitrile

                            Haloketones

                               I .l-dichloropropanone
                               1,1, l-tricliloropropanone

                            Miscellaneous chloro-organics

                               chloropicrin
                               chloral hydrate
                               cyanogen chloride

                            Haloacetic acids

                               monochloroacetic acid
                               dichloroacetic acid
                               trichloroacetic acid
                               monobromoacetic acid
                               dibromoacetic acid

                            Chlorophenols

                               2.4-dichlorophenol*
                               2.4.6-trichlorophenol
                               pentachlorophenol*

                            Aldehydes

                               formaldehyde
                               acetaldehyde
* These chlorophenols were only analyzed for during the first sampling quarter.

-------
TmHALOMCTHAMES


Cl


Cl
1
- C - H
i
Cl
Cl
1
Cl - C - H
i
Br
Cl
i
Br - C - H
i
Br
Br
i
Br - C -
i
Br


H


  CHLOROFORM
DICHLOROBROMO-
    METHANE
DIBROMOCHLORO-
   METHANE
       BROMOFORM
HALOACETONITRILES
      Cl

   Cl - C - C S N
      i
      Cl
      Cl
       I
   Cl - C - C = N
       i
      H
      Br
       i
   Cl - C - C = N
       i
      H
           Br
           i
        Br - C - C = N


           H
  TRICHLORO-
  ACETONITRILE
   DICHLORO-
  ACETONITRILE
 BROMOCHLORO-
  ACETONITRILE
        DIBROMO-
      ACETONITRILE
HALOKETONE3
                  Cl  O  H
                  i   M  i
               CI-C-C-C-H
                  i      i
                  H     H
                           Cl
           O
           H
H
                        CI-C-C-C-H
                            i      i
                           Cl     H
            1.1 -DICHLOROPROPANONE
                    1,1,1-TRICHLOROPROPANONE
MISCELLANEOUS
           Cl
           i
        Cl - C - NOj
           i
           Cl
              Cl  H
              i   i
           Cl - C - C - OH
              i   i
              Cl OH
                 Cl - C 5 N
      CHLOROPICR1N
 (TRICHLORONITROMETHANE)
         CHLORAL HYDRATE
             CYANOGEN CHLORIDE
     Structural Formulas for Disinfection By-Products
                          FIGURE 1-1

-------
KALOACrnC ACIM
         Cl  O
          I  II
       H - C - C - OH

         H


    MONOCHLORO ACETIC
          ACID
      Cl O
       I  H
    Cl - C - C - OH
       i
      H


   DICHLOROACETIC
        AGIO
      Cl  O
      i   n
  Cl - C - C - OH
      i
      Cl

 TRICHLOROACETIC
      ACID
                  Br  O
                   i   n
                H - C - C - OH
                   i
                  H

              MONOBROMOACETIC
                    ACID
                     Br  O
                      i  n
                  Br - C - C - OH
                      i
                     H


                  DIBROMOACETIC
                      ACID
CHLOftOPHENOLS
              Cl


      Cl -/ O  V OH



    2,4-DICHLOROPHENOL
           Cl
   Cl Y O V OH


           Cl

2.4,6-TRICHLOROPHENOL
    Cl    Cl
      s  /

  Cl -/ O V OH

      )  (
    Cl    Cl

PENTACHLOROPHENOL
ALDEHYDES
                   H    H
                   I    I
               H - C  ~ C =O
                   I
                   H

                 ACETALDEHYDE
                H - C =: O
                FORMALDEHYDE
   Structural Formulas for Disinfection By-Products
                   FIGURE 1-1   (Continued)

-------
                                   Introduction


 For (he CDHS study,  (he  first  "year"  of Ihe project (February,  1988 through June,
 1988)  included  utility  selection,  baseline data  collection  at the  10  utilities,  and
 performance of process modification studies at two utilities.  During the second year of
 the study (July. 1988 through June, 1989), baseline data collection was completed.

 The major components of the  project,  the baseline data collection  and the process
 modification studies, are described in more detail  below.   The utility selection process
 is described in Section 2 of this report.

 Baseline Data Collection

 A sampling schedule for the 35  utilities participating in the USEPA and CDHS studies
 was  developed  which  provided  for sampling  of  clearwell effluents  (after  final
 disinfection but before  distribution) in the selected plants on a quarterly  basis for one
 year.   Ihe  first  sampling date was March  14,  1988 and  the  fourth sampling  was
 completed in February,  1989.   The first  sampling  quarter  (i.e..  mid-March through
 April)  corresponded  to the  spring  season.    The  subsequent   sampling  quarters
 corresponded to  the summer,   fall  and  winter seasons  (i.e.. July through  August.
 October through  November,  and  late  December  through  February,  respectively).
 Furthermore, the  utilities were sampled each quarter in the same order where possible.
 In  this  manner, each utility was sampled four times, each time  representing different
 seasonal conditions in terms of temperature and water quality.

 Utilities were sent coolers  containing the  sample  bottles, packing material,  blue  ice,
 detailed sampling instructions  and  a sample information  sheet.    Utility  personnel
 collected  the grab  samples at  the designated sampling  locations  on the designated
 sampling dates.  The utilities were asked to fill in the sample information sheet with
 plant operating information such as chemical dosages and locations of chemical addition
 on  the  day of sampling.   After sampling, the utilities returned the sampling kits via
 overnight mail to Metropolitan for analysis.

 The results of the  baseline data collection achieved  project  objectives  by  providing
 information on  the  occurrence  of  DBFs on a seasonal basis at  the selected utilities.
 These results are reported and discussed in detail in Section 5 of this report.

Treatment Modification Studies

Treatment  modification studies are described in detail in Section 6  of this report.  The
utilities at which these studies were conducted were selected based on the representative
nature of their normal treatment process, the flexibility offered  by their treatment plant
design or other facilities (i.e.,  pilot plant  facilities), and their willingness  to contribute
resources to the study.

The principal goal of conducting the treatment modification studies was to  identify, in a
preliminary manner, the processes  or process modifications having  the  greatest  impact
on  DBF production.  To achieve this goal, the following treatment modification studies
were conducted:

     o   Utility 3:  Full-scale study of effects  of alum dose on  DBF precursor removal
         by coagulation at pH 5.5.

                                       1-3

-------
                                   Introduction
     o    Utility 6:    Full-scale  study  of effects  of  changing  oxidant/disinfectant
          (chlprine/chlpramines and ozone/chloramines).
     o    Utility 7:  Pilot and full-scale study of effects of changing oxidant/disinfectant
          (chlorine only, chlorine/chloramines,  and ozone/chloramines).
     o    Utility II:  Demonstration-scale  study  of granular activated  carbon (GAC)
          adsorption for removal of DBF precursors.
     o    Utility 12:   Full-scale  study of  effects  of alum dose  on  DBF  precursor
          removal by coagulation.
     o    Utility 16:    Full-scale   study  of effects  of  changing  oxidant/disinfectant
          (chlorine only and chlorine dioxide/chlorine).
     o    Utility 19:    Full-scale   study  of effects  of  changing  oxidant/disinfectant
          (chlorine only and ozone/chlorine).
     o    Utility 25:    Full-scale   study  of effects  of  changing  oxidant/disinfectant
          (chloramines and  ozone/chloramines).
     o    Utility 36:   Pilot  study  of effects of changing  oxidant/disinfectant (chlorine
          only.    chloramines    only.   ozone/chlorine,    ozone/chloramines    and
          ozone/hydrogen peroxide/chloramines).
     o    Utility 37:    Full-scale  study  of  treating   parallel  trains  with  different
          preoxidants/disinfectants  (chlorine and chlorine dioxide).

Eight of  the  treatment  studies were performed  for  the USEPA study  and two  were
performed for  the CDHS  study.    Utilities  36  and  37 were selected  for process
modification  studies but did not participate in the baseline sampling  program.   The
results of all ten studies are reported and discussed in Section 6.
                                       1-4

-------
Section 2
Utility Selection

-------
                                   SECTION 2

                             UTILITY SELECTION


Twenty-five  utilities  around the  United  States were selected  for participation in the
USEPA  project.  The first task  of the  selection process was the  development  of a
criteria matrix  and  an information  request to be  filled  in  by potential participants.
Then the potential utilities were screened  and the utilities were selected for participation
based on  the criteria matrix developed for the study.

INFORMATION REQUEST

An information  request packet was sent to 104 potential participants in the study, along
with a cover  letter introducing  the study and providing some  background information
which  would familiarize  the  recipients  with  the  study and  emphasize  the study's
importance to  the  water  utility  industry.   The  letter  also  indicated  the level  of
commitment from each participant  that  would  be required for the  study.   In some
instances,  a  particular  plant  owned/operated  by  the  utility  was  identified  in  the
information request  and in other  cases where the utility had more than one treatment
plant,  they were asked to  respond for the plant producing the highest levels  of THMs.
Copies of the information  request form and cover letter are provided  in Appendix  A of
this report.

The information request was to be filled in  by each recipient, providing data  on the
plant's treatment processes, chemical doses and contact times, and the quality of the
plant's raw water, finished water  and distributed water.  Responses were received from
78 utilities, a  response rate of 75 percent.

Not  only  did the completed information requests  provide  the basis of the selection
process,  but as  the  study progressed, the  information was a valuable reference for
interpreting baseline data as they were collected.

SELECTION MATRIX

The 35 utilities  (25 for the USEPA study  and 10 for the CDHS  study) were selected for
participation in the study from those responding to the information  request based on the
following criteria:

    o     Utility willingness to participate,
    o     Disinfectant/oxidant(s) in use or available,
    o     Source water type.
    o     Source water quality.
    o     Geographical location.
    o     Treatment process configuration,
    o     Ability to alter or segregate treatment processes, and
    o     On-site pilot plant capabilities.

The resulting  matrix of utilities  is presented in  Table 2-1.   The matrix is divided  into
two major categories,  treatment  type (conventional,  direct filtration, softening  and

                                      2-1

-------
                                   TABLE 2-1

       DISINFECTION BY-PRODUCTS IN DRINKING WATER STUDY

                        UTILITY SELECTION MATRIX
TREATMENT
GROUNDWATER
LAKE/RESERVOIR    FLOWING STREAM
CONVENTIONAL
ClermontCo., OH I
Long Beach, CA 2
Norwich, CT* IB
MWD, CA -Mills 2",5h
Arlington, TX* 2A
Hackensack, NJ 2e
MWD, CA -Weym. 2
San Francisco, CA 1
Big Spring, TX*  5
Shreveporl, LA 6a,8b
Cape Girardean, MO* 2A
Cincinnati, OH I
Contra Costa WD, CA 2A
Sacramento, CA 1
Santa Clara Valley, CA 2
Newport News, VA IA
DIRECT
FILTRATION
                     East Bay MUD, CA 1
                     Las Vegas, NV  I
                     Little Rock, AR I
                     Aurora, CO l"A,5bA
                     Los Angeles DWP, CA 3
SOFTENING
Palm Beach Co., FL 2
Wausau. WI* I
Minot. ND I
Santa Monica, CA 1
Macomb, IL* IA
Galveston,TX 7c,4l1
Louisville, KY 2
Ft. Meyers, FL* I
Emporia, KN* 2A
Omaha, NB IA
DISINFECTION
ONLY
Mesa Consol., CA 3
North SkagitCo., WA* I
New York City, NY I
Newark, NJ 1
Note: Utilities participating in the California Public Health Foundation study are listed in bold type.
*    Population under 50.000; all others over 50,000.
Key for chemical addition:

I - chlorine only
2 - chlorine + chtoramines
3 - ozone + chlorine
4 - chlorine + chlorine dioxide
5 • chloramines only
6 - chloramines + chlorine dioxide
7 - chlorine 4- chloramines +
8 - ozone + chloramines
A - powdered activated carbon
B - potassium permanganate
      chlorine dioxide
                           " - first quarter only
                           h - second through fourth quarters
                           c - First through third quarters
                           *' - fourth quarter only
                           e - clear well effluent sampled
                              before ammonia addition

-------
                                 Utility Selection


disinfection  only)  and  source  water type  (groimdwater,  lake/reservoir,  and flowing
Stream).   Within  these categories,  geographical  location and  disinfectant  type  (free
chlorine,  chloramines,  chlorine  dioxide,  and/or  ozone)  were  also  considered.
population was also used as a category in developing the  selection  matrix  (<50,000
0nd >50.000). as  was TTHM level (<25 x/g/L, 25 to 50 /ug/L, and >50 //g/L).

The selection  process was  aided  by data  from the THM survey conducted for the
American   Water  Works  Association  Research  Foundation   (AWWARF)   in   1987
(McGuire  and  Meadow.  1988).   Using the AWWARF data  plus  the  questionnaire
responses,  the  selection process  was aimed at filling the utility  selection  matrix as
completely as possible based on the criteria described above.

To illustrate the nation-wide distribution of the selected utilities, a map showing  their
locations is presented in Figure 2-1.  It  should  be  noted that 10 of the  11  participating
utilities in  California were involved in the CDHS study.

PROCESS TRAINS OF THE PARTICIPATING UTILITIES

Simplified schematics of the process trains employed at  each  participating  utility are
shown in Figure 2-2 by utility  identification number.  The  figure indicates that of the
35 utilities participating in the combined USEPA and CDHS studies:

     o   Sixteen   utilized   a  conventional  treatment  process  with   coagulation,
         flocculation. sedimentation  and filtration;
     o   Five employed a direct filtration treatment process;
     o   Ten softened; and
     o   Four employed disinfection  only.

The schematics, in addition  to the  information provided in  Table  2-1,  indicate the
following disinfection schemes for the  participating  utilities  (during the majority  of
baseline  data collection):

     o   Nineteen  of the utilities  used  only free  chlorine  throughout  their  treatment
         trains  and as a final disinfectant;
     o   Nine utilities used free chlorine and provided for  some free chlorine contact
         time before  the addition of ammonia  (either  with or without further chlorine
         addition)  to form chloramines as the final  disinfectant;
     o   Three utilities used only chloramines (concurrent addition of  chlorine and
         ammonia, with no free  chlorine  contact time)  throughout their  treatment
         trains and as a final disinfectant;
     o   One utility employed chlorine  with subsequent ammonia addition to  form
         chloramines. in addition to chlorine dioxide;
     o   Three  utilities used ozone  (two  utilities  employed ozone and  free chlorine,
         and one utility utilized ozone, chloramines, and chlorine dioxide).

Several  utilities had changes in their disinfection  schemes during the course of the
baseline  sampling.     For  instance,  Utility  4  changed   from   prechlorination/
postammoniation to concurrent addition of chlorine and ammonia after the first quarter.
Utility 18  changed  from chlorine  to chloramines  after the first quarter.   Utility 25
switched  from chloramines  to ozone/chloramines after the  1st  quarter.  Utility 27, a

                                      2-2

-------
UTILITIES PARTICIPATING IN DBP STUDY
                                               PALM
                                               BEACH
              FIGURE 2-1

-------
  AERATION
RAPID MIX  FLOCCULATION  SEDIMENTATION   FILTRATION
Chlorine
                                    UTILITY  1
t


Aeration



v^ mum ic i


m^^
V '•-.''
                                                                       Fluoride
                                                                       Phosphate
                                     UTILITY 2
t
Polymer
t
Retention
	 , 	 te-
t


\

\.

N

V,



                                Polymer
                                                                Chlorine
                                                                        Chlorine'
                                                                       Ammonia
            t
          Alum
        Caustic Soda
    Potassium Permanganate
         Polymer
                                    UTILITY 3
                                                                Fluoride, Chlorine
                                                           Caustic Soda, Phosphate
                                   UTILITY  4
t
Ammonia
Chlorine
Alum
Polymer
t

d

o


                                                            Polymer
                                                            Chlorine
                                                  1
        * Configuration during last 3 quarters: during 1st quarter same as Utility 7
                                                              Sodium Hydroxide
                                     FIGURE 2-2

-------
                              RAPID MIX     FLOCCULATION  SEDIMENTATION   FILTRATION
   T
Powdered
Activated
 Carbon
 T
 Alum
Polymer
                                     UTILITY  5
                                                                         Fluoride
                                                                       Ammonia
                                                                 Sodium Hydroxide
                                     UTILITY 6
t
Alum
Polymer
Chlorine
I


o


o
                                                                 Chlorine -
                                        Chlorine                 Ammonia, Chlorine
                                                                    Caustic Soda
                               * Clearwell effluent sampled before ammonia addition
                                                               1
              t
            Chlorine
             Alum
            Polymer
                                     UTILITY 7
                                                                Ammonia
                                                             Sodium Hydroxide

           Chlorine
            Alum
                                     UTILITY 8
                                             Chlorine
                                             Polymer
                                                                    Chlorine
                                                                Sodium Hydroxide
                                                                    Fluoride
                                  FIGURE 2-2 (Continued)

-------
                  RAPID MIX   FLOCCULATION  SEDIMENTATION  FILTRATION
       1
      Lime
     Chlorine
     Polymer
      Alum
Powdered Activated
Carbon (Intermittent)
                                         UTILITY 9
                            Ammonia
                             Fluoride
                          Chlorine, Lime
     T
    Ammonia
    Chlorine
Lime (Intermittent)
     Alum
  1
O
o
                                       UTILITY 10
      Alum
     Polymer
                                        UTILITY  11
1 .

o
o



*v



s

\



Lamella
Settler

.
'\
                                                                                  Reservoir
                                                                                   Storage
                                                                                   (3 days)
      Chlorine
      Lime
      Fluoride
      Ferric Sulfate*
       I
Hydraulic
  Jump
                                                         ••'<••••„
        • Fciric Sulftte added intermittently (in use on 5/8/89.6/12/89 and 7/17/89)
                                   FIGURE 2-2 (Continued)

-------
                               RAPID MIX   FLOCCULATION  SEDIMENTATION   FILTRATION
          T
         Chlorine
          Alum
    Powdered Activated
    Carbon (Intermittent)
                                   UTILITY  12
o
o
                                       Lime, Fluoride
                                    Chlorine, Polymer
                                                                 Caustic Soda
                                                                   Ammonia
                                  UTILITY 13
I
Chlorine
Alum
f

O

O

                                                              Polymer
                                                            (Intermittent)
                                                                Chlorine
                                                                   Lime
Polymer
Chlorine
 Alum
                                   UTILITY 14
                                     Polymer
                                            Chlorine
                                            Ammonia
                               FIGURE 2-2 (Continued)

-------
                             RAPID MIX     FLOCCULATION  SEDIMENTATION  FILTRATION

              Alum
             Chlorine
             Fluoride
                                     UTILITY  15
                                                                      Chlorine
                                                                        Lime
                                                                                  1
                                     UTILITY 16
  t
Chlorine
T
                       Polymer
                     Ferric Chloride
                                                               Sodium Hydroxide
                                                                Sodium Silicate
                                                                   Chlorine
1
                                   UTILITY 17

            Chlorine
             Alum
o
o
                                                               Lime, Phosphate
                                                               Fluoride, Chlorine
                                                                                  1
                                  FIGURE 2-2 (Continued)

-------
   OZONE
      RAPID MIX  FLOCCULATION  SEDIMENTATION   FILTRATION
          Polymer, Fluoride
             Ammonia
                                  UTILITY  18
t
Polymer
Alum
Pr«i/H*»n"H Artivatft/1 farlvm
~1


o


1
                              Chlorine, Polymer
                           Powdered Activated Carbon.
                                 Ammonia
                                                            I
                                          Chlorine
                                          Ammonia
                                          Fluoride
                             * Used Chlorine only during 1st quarter
                                   UTILITY 19
1
000000
f\ f\ /\ S\
0000
0000000
000000000
feme imonae
Polymer
Chlorine
1
T »















\


X.

^s


s

Chlorine- intermittent
i
i
i
f


•
•
K ',',',
U^ t f
                                                                  Chlorine
Powdered Activated Carbon
     (Intermittent)
                         Aiu
                                  UTILITY 20
 urn
Lime
                               oo
                                                              Chlorine
                                                              Chlorine, Fluoride
                                                                Phosphate, Lime
                                                      1
                                 FIGURE 2-2 (Continued)

-------
AERATION
                                                                             FILTRATION
        t
      Chlorine
        Lime
      Polymer
                                  UTILITY 21
 Solids
Contact
Clarifier
Chlorine       Ammonia

  I            I
   __      Alum, Lime,
    V    Activated Silica
                                  UTILITY 22
                                  Solids
                                  Contact
                                  Clarifier
                                                                Fluoride.
                                                   Carbon Dioxide
                                                                       Chlorine
                                  UTILITY 23
              Lime
              Sodium Aluminate
              Polymer
                                    Solids
                                   Contact
                                   Clarifier
                                                               Fluoride
                                                              Phosphate
                                                               Chlorine"
                     Carbon Dioxide
                                                    1
           Chlorine
                                  UTILITY  24
Aeration





Ion
Exchange

                                                                     Caustic Soda
                                                                      Chlorine
                                                                        i
                                 FIGURE 2-2 (Continued)

-------
OZONE
RAPID MIX  FLOCCULATION  SEDIMENTATION  OZONATION
                FILTRATION
             Chlorine Dioxide
                                    UTILITY 25
pooooo
 0000
 0000 OCX)
 ooooooooo
    Chlorine**, Ammonia **
            Alum
           Polymer
                                               Chlorine
                                               Ammonia'
»*. " >l 1 > :
Mf.=^^^'


000000
0000
0000000
ooooooooo
                                         ;               Fluoride
                                        Lime     Chlorine Dioxide
                            * No Ozonation during 1st quarter
                            ** 1st quarter only.
                                    UTILITY 26
         Alum
        Polymer
         Lime
Powdered Activated
     (Intermittent)
        Solids
        Contact
       Clarifier
                                                                 Chlorine
Recarbon-
  ation
                                               Carbon Dioxide
                                                                     Chlorine
                                    UTILITY 27
         t          t
   Chlorine Dioxide     Lime
       Polymer    Ferric Sulfate
                  Solids
                  Contact
                 Clarifier
                                           Polymer
                                     Carbon Dioxide
                                           Chlorine
                          ' Ammonia off-line during 4th quarter
                                                            Phosphate, Fluoride
                                                              Chlorine Dioxide
                                                                    Ammonia
                                 FIGURE 2-2 (Continued)

-------
    PRESEDIMENTATION      RAPID MIX     FLOCCULATION  SEDIMENTATION    FILTRATION
                                      UTILITY 28
           1
 Alum
Polymer
                                                           Chlorine
/•?' ^^^'^
< "•*" ^MtV^

.
o
o



>ss_

S.

k>

s




Sf^^^
•v^AU-S,-, ^f-
W-^> ** »"*• ffv. •"<
                                                                               Fluoride
                                                                             Carbon Dioxide
                       Ammonia
                          Lime
                       Soda Ash
                                                                 Chlorine, Ammonia
                                     UTILITY 29
                                Solids
                                Contact
                                ClariFier
 Solids
Contact
Clarifier
 Alum
Polymer
Chlorine
                    Alum
                   Polymer
                 Lime, Chlorine
                     Ammonia
                     Lime. Alum
                                       UTILITY 30
  Polymer
  Chlorine
^"VvfS
I.

o
o
                                                         Ammonia
                                                         Chlorine
                                                         Fluoride
                                                      Carbon Dioxide
                                                                                I
         Carbon Dioxide

                       Powdered Activated Carbon
                                   FIGURE 2-2 (Continued)

-------
   OZONE
PRESEDEVtENTATION
SEDIMENTATION   FILTRATION
I
Polyr
l
ner
t

fe* : -•*,.
\j « »


Retention
Fluori
t
de, Chlorine —

'
-^-

Powdered Activated Carbon
     (Intermittent)
                             Polymer
                            Lime, Silica
                             Chlorine
         i
      000000
       0000
      0000000
       ooooooooo
                              UTILITY 32
     Chlorine
      i
                               UTILITY 33

                                 Chlorine
                                  i
                               UTILITY 34
                                  Chlorine
                                   i
                               UTILITY 35
                             Chlorine
                              Lime
                             FIGURE 2-2 (Continued)

-------
                                 Utility Selection


 prechlorine/postamrnonia/chlorine dioxide utility, had their ammonia off-line on the 4th
 quarter  sampling date.   Additionally, although  Utility  6 employed  prechlorination/
 poslammoniation. the  clearwell effluent samples were collected before the ammonia
 addition point, therefore it is classified as a chlorinating utility.

 In  all.  35 utilities were  each sampled four times during the course of baseline  data
 collection, yielding  140  total data  points.   Of the 140 disinfection  schemes in  use
 during  baseline data collection. 78  (or 56 percent) were free chlorine only,  40  (29
 percent) were prechlorination/postammoniation. 11 (8 percent) were chloramines only.
 8 (6 percent) were ozone/chlorine, and 3 (2  percent) were ozone/chloramines.

 The schematics indicate the following coagulants were used by the utilities:

     o    Seven of the utilities employed alum only;
     o    Ten used alum with a polymer as a coagulant aid;
     o    One utility used polymer as a primary coagulant;
     o    Two used ferric chloride and polymer; and
     o    Six utilities used no coagulant.

 Additionally, the  softening  utilities  used  a variety  of coagulants in  several different
 combinations:

     o    Lime plus polymer;
     o    Lime, alum and polymer (three utilities);
     o    Alum/polymer  in  one  solids  contact  clarifier  for  color  removal   and
          lime/polymer in a second unit for softening;
     o    Lime, alum and activated silica;
     o    Lime, sodium aluminate and polymer;
     o    Lime, ferric sulfate and polymer; and
     o    Lime.  alum, ferric sulfate.  activated silica and polymer.

 Eight of the 35 utilities utilized powdered activated carbon, at least on an intermittent
basis.   Only one  utility added  potassium  permanganate as  a preoxidant.   One utility
employed ion exchange for softening.
                                       2-3

-------
Section 3
Methodology

-------
                                  SECTION 3

                               METHODOLOGY


This section briefly describes the sampling and analytical methodology of the  study.
References  are  provided for further  information.   In  addition,  detailed  analytical
protocols are provided in Appendix C of this report.

SAMPLING PROCEDURE

The sampling instructions (Appendix B) included in each sampling kit provided detailed
information to the  utilities  on  how to properly sample for  the  study.   A  project
engineer contacted  each  utility to arrange the sampling date for each quarter.  At that
time,  if water  quality  conditions  or  treatment   plant  operations  were  deemed
unacceptable (non-representative), the sampling date was postponed. Typically,  on the
sampling date, all  sample collection was  completed by  a utility within  1  to 2  hours.
Evaluation  of plug flow conditions in each plant was not within the scope of the study.

Grab samples (see  Table 3-1) were collected by utility personnel at clearwell effluents
(after final  disinfection but  prior to distribution).   Starting with the second sampling
quarter.  TOC samples were  collected each sampling date at plant  influents and filter
influents.  Also starting  with that quarter, ultraviolet (UV) absorbance  measurements
were  made at all  three  sampling  points, the bromide and chloride levels of plant
influents were determined, and formaldehyde and  acetaldehyde analyses were made of
the clearwell effluents.  Some plant influents were analyzed for these aldehydes  during
the third sampling  quarter;  aldehyde determinations were made for all plant influents
during the  fourth sampling quarter.

Plant influent  samples  were collected at  the  head  of the  treatment  plants  and
corresponded  to  "raw"  water before the addition of disinfectant/oxidant, coagulant,
lime, etc.  This sampling location was important in enabling the characterization of the
DBF precursors  (i.e., as measured  by TOC and UV-254) prior  to the addition  of any
treatment chemicals.  In  addition, bromide measurements were made on plant influent
water to determine the influence of this ion on the  production of brommated  DBFs.
Since strong oxidants such as chlorine or ozone can convert bromide to hypobromous
acid, bromide samples were collected upstream of the preoxidation point.

Samples were  dechlorinated  and/or  preserved  as outlined  in  Table  3-1 and were
analyzed as  soon as possible within established  holding times.   Holding times are
described in Table  3-2.   The  dechlorination agents  and preservatives for the DBF
fractions were evaluated in holding studies to ensure that analyte concentrations  held to
+ /-  20  percent of their initial  values.  Total  organic halide  (TOX)  samples were
dechlorinated with a sodium sulfite solution.  Because sampling kits were prepared and
shipped  to  field  locations at  least  2 weeks  prior to sampling, the instability  of the
dechlorination solution required that TOX samples  be dechlorinated and preserved upon
receipt at  the laboratory.   Standard Methods (1985) establishes a 28-day maximum
storage  for  bromide sampledThis holding  time is  only applicable to unchlorinated
samples, as an unquenched  free chlorine residual  can lower the level of bromide.  In
this study,  bromide  samples were collected upstream of chlorine addition points.

                                      3-1

-------
                                      TABLE 3-1

                             SAMPLING KIT CONTENTS
Analytical
 Fraction
Sample Bottles
Number Size
Dechlorination Agent
and/or Preservative*
Plant influent sample bottles:
 Formaldehyde/acetaldehyde      3,    40 mL
 Bromide/chloride               1,    60 mL
 TOC/UV                      3,    60 mL

Filter influent sample bottles:
 TOC/UV                      3,    60 mL

Clearwell effluent sample bottles:
 PE-DBPs<                      3.    40 mL
 Chloral hydrate                3.    40 mL
 Haloacetic acids'1               4,    40 mL
 Cyanogen chloride              4,    40 mL
 Formaldehyde/acetaldehyde      3.    40 mL
 TOX                          2,   250 mL
 TOC/UV                      3.    60 mL

Additional bottles in kit:
 Travel blank--NH4CI            2,    40 mL
 Travel blank-ascorbic acid      2,    40 mL
 Travel blank—formaldehyde      2,    40 mL
                    None; HgCI2 solution + NH4C1 crystals'1
                    None
                    None"
                    None"
                    65 mg NH4CI crystals
                    20 mg ascorbic acid
                    65 mg NH4C1 crystals
                    20 mg ascorbic acid
                    None; HgCl2 solution +  NH4C1 crystals'1
                    None3
                    None8
                    65 mg NH4C1 crystals
                    20 mg ascorbic acid
                    None; HgC!2 solution +  NH4CI crystals'1
HFor the following analytical fractions, additional reagents are added after
 sampling as soon as possible after receipt at Metropolitan:
 TOC/UV: Acidify with reagent-grade phosphoric acid to pH < 2.
 TOX:    Dechlorinate and acidify, respectively, by addition of l/i mL of 50%
  sulfuric acid and 3 drops (utilizing a 3V4" pasteur pipette) of a fresh
  saturated sodium sulflte solution.

''No dechlorination agent and preservative used during second and third sampling
 quarter;  40 ^uL of a 10 mg/mL mercuric chloride solution plus 65 mg ammonium
 chloride  crystals used  during fourth sampling quarter.

cPE-DBPs (pentane-extractable disinfection by-products) are THMs. HANs.
 haloketones. and chloropicrin.
'Includes 2.4.6-trichlorophenol.

-------
                               TABLE 3-2

                       SAMPLE HOLDING TIMES
        Analytical               Sample Holding         Extract Holding
         Fraction                     Limit                  Limit
PE-DBPs                       Extract immediately           2 weeks
Chloral hydrate                       21 days                  15 days
Haloacetic acids                       9 days                  7 days
Cyanogen chloride              Analyze immediately        Not applicable
Formaldehyde/acetaldehyde       Extract immediately           14 days
TOX                                2 weeks              Not applicable
TOC/UV                            28 days              Not applicable
Bromide/chloride                     28 days              Not applicable

-------
                                 Methodology
ANALYTICAL METHODS
A  method  was  developed for  the  analysis of  THMs,  HANs,  haloketones.  and
chloropicrin that employed modified THM liquid/liquid extraction (LLE) (Koch et al.,
 1988).  For this analytical fraction,  pentane was used as the extraction solvent, sodium
sulfate was added  to improve the partitioning from the aqueous  phase to the solvent,
and a capillary gas chromatograph/electron capture  detector (GC/ECD) was utilized for
adequate resolution of the analytes.  The more polar chloral hydrate required a similar
LLE method, but methyl t-butyl ether was used as the extraction solvent (Krasner et al.,
 1989).  HAAs and 2,4,6-trichlorophenpl were analyzed by  an acidic, salted ether LLE,
and they required  esterification  with diazomethane  prior to GC/ECD analysis (Krasner
et  al.,  1989).    Cyanogen  chloride  was analyzed   by  a   purge-and-trap  gas
chromatograph/mass  spectrometer   (GC/MS)  method   (Krasner  et  al.,   1989).
Formaldehyde and acetaldehyde were  analyzed by  a derivatization/extraction  GC/ECD
method (Yamada and Somiya, 1989; Glaze et al., 1989b)

The TOC  measurements  were  performed  according  to  the   persulfate-ultraviolet
Standard Method  505B  (1985).    TOC samples  were  acidified with reagent-grade
phosphoric acid to a pH of less than  2 after receipt of the samples at Metropolitan's
laboratory.   This  preservation  technique was also used  on  UV samples.   UV was
measured  at  254 nanometers (nm)  with  a  UV-visible spectrophotometer (Lambda  5,
Perkin-Elmer Corp., Norwich, CT) and a 1-cm quartz cell.  The UV was also measured
at 800 nm to provide a correction  for the  presence of turbidity  or suspended solids.
Because some plant effluent samples were chloraminated, monochloramine (which has
UV absorbance at  243  nm) could have presented an interference problem.  However,
the acid preservation converted monochloramine to dichloramine in approximately 2
hours.   The  latter species  has a  UV  peak at  293  nm,  so it  did not  present an
interference problem.  To  evaluate the effect of the phosphoric acid on  the UV results,
some  unpreserved parallel effluent  samples  were  also  analyzed,  and these yielded
comparable results to the acidified samples.

Bromide and chloride analysis was conducted with an ion chromatograph (Model 2000,
Dionex Corp., Sunnyvale,  CA),  a  20-^uL or 50-//L sample loop, a  high-performance ion
chromatography analytical column (AS4A,  Dionex Corp.), an anion micromembrane
suppressor, and a  conductivity  detector.  The eluant was a solution of 2 millimolar
Na2CO3 and I millimolar NaHCO3.

The TOX  analysis was performed  using USEPA  Method 9020  from  SW 846 Test
Method for Evaluating Solid Waste, Physical/Chemical Methods (Second Edition).

SIMULATED DISTRIBUTION SYSTEM TESTING

Some samples in this study were analyzed using a simulated distribution system (SDS)
test protocol (Koch et al..  1989) developed at Metropolitan.  In this protocol, samples
are chlorinated  and  incubated under  conditions  which simulate actual  full-scale
conditions in  Metropolitan's distribution system. Typically, samples are chlorinated at
a level which will leave  a chlorine residual  of at  least  0.5 mg/L at the  end of the
incubation  period.   Firstly, this  mimics  typical  distribution  system  practices  of
providing some minimal chlorine residual to consumers, and secondly, some DBFs are
unstable when the chlorine residual has gone to zero.  Samples are buffered to a pH  of

                                     3-2

-------
                                  Methodology


 8.2. which is consistent with many utilities' practice of corrosion control.  Some DBFs
 will undergo some hydrolysis at this pH, as would be experienced in actual distribution
 systems  operating at  such  a pH.   The  samples are incubated  at  25°C, which  will
 produce  maximum  levels  of THMs and  other  DBFs,  as would  be experienced in
 summer  months.   The  samples are incubated for one to three days,  to provide a
 measure  of DBF levels that would  be  expected in  a distribution system with such a
 detention time.   Most importantly,  the use of standardized SDS conditions enables a
 consistent set of data to  be produced  that often  cannot be realized in  a distribution
 system which may be influenced by other confounding variables.  Also, the use of SDS
 testing allows evaluation of expected distribution system DBF levels when  studying pilot
 plant and demonstration-scale facilities (see Section 6).

 For the GAC study at Utility 11 (see Section 6), SDS tests were performed. The SDS
 test was  selected to provide a standard set of conditions  by which  to evaluate DBF
 precursor removal at Utility I I's GAC demonstration plant.  For  example, Utility  I I's
 ambient  water temperature  ranged from  11 to 27°C during the  course  of the  study.
 However, with SDS testing, all samples were incubated  at 25°C.   During the day 13
 and day  25  samplings, some additional  SDS tests were set up  at ambient  water
 temperature.  The results are shown in Tables 3-3 and 3-4.

 An SDS test  can  only mirror  a limited number of actual operating conditions.  For
 example, routine SDS  samples at Utility 11 received a 4.5 mg/L chlorine dose,  in order
 to provide adequate  chlorine for a 3-day incubation period. However, the actual plant
 dosed different amounts  of chlorine, depending on current demand  requirements.  In
 addition, the actual  plant applied  chlorine in three locations,  two before conventional
 filtration  and one after.  The pH of SDS samples are buffered to a pH of 8.2 to 8.3,
 whereas the pH  ranged from 8.4 to 8.8  in Utility 1 I's actual distribution system during
 the course of this study.

 For some DBFs, the distribution value is between that detected in the two SDS tests at
 Utility 11 (Tables  3-3  and 3-4). For example,  for the Day 13 testing, chloroform  was
 42.  31. and 55 //g/L in  the distribution,  ambient SDS, and standard SDS samples,
 respectively.  At that same time, 1.1,1-trichlpropropanone had values of 0.55,  1.2,  and
 0.28 x^g/L. respectively.   Since this ketone is an intermediate DBF (chloroform  is the
 stable  endproduct   of  1.1.1-trichloropropanone),   it appears   as   if  the   elevated
 temperature of the standard SDS  test drove the reaction of 1,1,1-trichloropropanone
 hydrolysis the furthest.  During the Day 25 testing, this ketone was detected at  levels of
 0.18. 1.5. and 0.36 /ug/L in the distribution, ambient SDS, and standard SDS samples,
 respectively.   The lowest  ketone value was found in the  distribution system where the
 elevated pH  (i.e.. 8.8) probably resulted in base-catalyzed hydrolysis of this DBF.

 In general, though, the SDS tests  yield  comparable DBF levels as that detected  in the
 actual distribution  system.   Unless noted. SDS tests were performed at  the  standard
 SDS condition of 25r>C.

 Another  SDS consideration was that the  GAC  effluent would  not have as high  a
chlorine demand as  the GAC influent.   Thus. GAC effluents were analyzed by  two
different SDS tests, one at a chlorine dose of 4.5 mg/L and another at 2.0 mg/L. The
 latter dose is a more realistic chlorine level for Utility 1 I's GAC effluent when TOC
 removal is high.  Since the chlorine dose may  influence  the level of DBF production,

                                      3-3

-------
                                    TABLE 3-3

                 UTILITY 11 GAC STUDY:  DAY 13 DBF RESULTS
Parameter
Temperature. °C
pH
Detention time/
incubation, days
Cl: dose*. mg/L
C\2 residual. mg/L
DBFs, x/g/L:
CHC1,
CHBrCI,
CHBr,Cl
CHBr*
TCAN
DCAN
BCAN
DBAN
1.1 -DCP
l.l.l-TCP
MCAA
DCAA
TCAA
MBAA
DBAA
Chloropicrin
Chloral hydrate
Distribution
System
15
8.7

3 (approx.)

0.7

42
15
6.0
0.49
<0.03
1.8
0.55
0.64
0.34
0.55
1.6
15
9.4
<0.5
1.4
0.60
9.9
Ambient SDS Standard SDS
GAC Inf.* SDS Blank GAC Inf.* SDS Blank
12
8.41

3
4.5
2.2

31
13
4.8
<0.91
<0.03
2.0
0.59

-------
                                    TABLE 3-4

                 UTILITY 11 GAC STUDY:  DAY 25 DBF RESULTS
Parameter
Temperature. °C
PH
Detention time/
incubation, days
Cl, dose*. mg/L
CI-, residual. mg/L
DBFs, uglL:
CHCI,
CHBrCI,
CHBrXl
CHBr;
TCAN
DCAN
BCAN
DBAN
I.I-DCP
I.I.I-TCP
MCAA
DCAA
TCAA
MBAA
DBAA
Chloropicrin
Chloral hydrate
Distribution
System
16
8.8

3 (approx.)
2.7
0.7

52
16
6.1
0.27
<0.03
0.75
0.49
0.20
0.25
0.18
1.7
17
8.4
<0.5
I.I
0.65
II
Ambient SDS Standard SDS
GAC Inf.* SDS Blank GAC Inf.* SDS Blank
14
8.22

3
4.5
0.90

33
14
5.0
<0.9I
<0.03
1.6
0.57

-------
                                 Methodology


the GAC effluent was dosed at the same concentration as the GAC influent in order that
comparisons could be made where the TOC level was the only variable.

A comparison of SDS data for the GAC  effluent is made for the  two SDS  protocols
used at Utility  11 (see Table  3-5).   The standard SDS conditions for the GAC influent
of a 4.5 mg/L chlorine dose and buffering to pH 8.2 were followed on one set of GAC
effluents.   The other set of  GAC  effluents received 2.0  mg/L chlorine and  were not
buffered (pH approximately 7.5).   On Day 82, there was 0.60 mg/L  of TOC  in the
GAC effluent,  and the chlorine demand was 0.7 and  1.4 mg/L for the 2.0 and 4.5
mg/L dosed tests, respectively.  On Day 95, there was 0.85 mg/L of TOC in the GAC
effluent, and the chlorine demand was  1.0 and 1.5 mg/L for the 2.0 and 4.5 mg/L
dosed  tests, respectively.   The difference in  chlorine  demands between the  two SDS
protocols may, in part, be related  to the differences in pH of the two tests.  On both
davs.  the  THMs  were higher in   the 4.5  mg/L  chlorine  dose  tests;  however, the
difference was  less  pronounced when there was higher TOC present (i.e., 38 and 47
u%IL TTHMs in  the 2.0 and 4.5 mg/L chlorine dose  tests,  respectively).  The  HANs
and the 1,1.1-TCP were lower in the 4.5 mg/L  chlorine dose tests.  As these  DBFs are
reactive intermediates, the higher  chlorine dose  and  pH probably  resulted in their
degradation (and resultant formation of other DBFs as stable endproducts).

SDS tests involve preparation of blanks (i.e., distilled water plus a bromide spike at the
level  occurring in the raw water).   There are low levels of  DBFs  detected in SDS
blanks. Whether these are formed  due to DBF precursors in the distilled water, SDS
bottle, reagents used, or a combination of these is  uncertain.  If the blank  levels are
due to precursors in the bottle  and/or reagents, then SDS blanks should be subtracted
from  the SDS sample results.  For  example,  the ambient SDS test performed on  Day
13 samples from  Utility 11  had 0.80 and 0.35 yug/L 1,1-dichloropropanone in the SDS
sample and blank, respectively.  If blank subtraction  is  performed,  the difference of
0.45 0g/L  compares well with the  0.34 //g/L found in the distribution sample.  Thus,
data reported in Section 6 reflect SDS blank subtraction.

SDS  blank subtraction was  performed  in  a  previous GAC study  performed  by
Metropolitan  (McGuire et al., 1989).  In the  latter study, SDS samples of virgin GAC
effluents each typically had levels of 1 to 2 /yg/L of dichloroacetic acid  (DCAA), when
different empty bed contact times were evaluated in parallel  (i.e., 7.5,  15, 30, and 60
minutes).   One explanation was that this DBF  formation was due  to a nonadsorbable
fraction of TOC.  However,  SDS  blanks  tended to have, within analytical error, the
same  values  for DCAA, which tended to support that these levels were  background
levels in all samples.   To resolve this issue for the  current study, minimum  reporting
levels (MRLs) for SDS testing were calculated  based upon the analysis of seven SDS
blank samples (see  discussion in Quality  Assurance section  below).  Thus,  relatively
high MRLs were obtained  for SDS  samples as contrasted to instantaneous samples  (see
Table  3-6).   SDS  data that were less  than these values  were  not  reported,  thus
precluding  the  need for SDS background  subtraction in many instances.   However,
when higher blank levels were detected, they were subtracted  from SDS  sample results.
For example, the standard  SDS tests at  Utility  11 for the Day 13  samples (see Table
3-3) for trichloroacetic acid (TCAA) in the GAC influent and  blank yielded values of 19
and 2.1 */g/L, respectively.   Thus, a blank-subtracted result  of 17 /ug/L was  reported,
As in  this  example,  the blank  subtraction  often resulted in  a number which did not
differ from  the uncorrected  value within analytical error.  The evaluation of SDS data is

                                     3-4

-------
                         TABLE 3-5

                UTILITY 11 TREATMENT STUDY
          COMPARISON OF GAC EFFLUENT SDS TESTS:
         EFFECT OF CHLORINE DOSE AND pH ON DBFs
                 Day 82 GAC Effl. SDS*    Day 95 GAC Effl. SDS*

Chlorine Dose:     2.0 mg/L    4.5 mg/L    2.0 mg/L    4.5 mg/L
pH
C12 residual, mg/L
DBFs, //g/L:
CHCI^
CHCUBr
CHClBr,
CHBr3
TCAN
DCAN
BCAN
DBAN
1.1 -DCP
U.I-TCP
MCAA
DCAA
TCAA
MBAA
DBAA
Chloropicrin
Chloral Hydrate
7.55
1.30

4.0
7.9
8.5
1.8
<0.029
0.40
0.47

-------
                                 Methodology


more complex, though, when SDS results approach  the same value as that detected in
SDS blanks.   For this reason,  MRLs are  set at higher values for SDS  testing, since
quantification can only be reliable when SDS values are significantly higher than blank
levels.

QUALITY ASSURANCE

Metropolitan is certified  by  the CDHS for  the analysis of THMs  and volatile organic
compounds.     Furthermore,   Metropolitan  obtained  permission   to  utilize  the
THM/HAN/haloketone/chloropicrin  method described above  for THM  compliance
monitoring.  For the list of DBFs under investigation, no officially-approved USEPA
methods  exist.   To validate all the DBF methods used for this study, Cincinnati tap
water and Metropolitan's Weymouth plant were both sampled as is and  a separate set
of bottles was spiked with  all the target-compound DBFs.  These samples were split
between  Metropolitan, Montgomery Laboratories  (Montgomery),  and  the Technical
Support Division (TSD) and the Environmental Monitoring Support Laboratory (EMSL)
of the USEPA (not all laboratories ran each analytical fraction).

Metropolitan's, Montgomery's, and  TSD's laboratories all  agreed on  THM,  HAN,
haloketone,  and chloropicrin results  (Fair,  1988a).   Metropolitan and TSD obtained
comparable results  for   chloral   hydrate  and   cyanogen  chloride  (Fair,  1988a).
Metropolitan and TSD had comparable TOC results,  and  Montgomery  and  TSD had
comparable TOX data (Fair,  1988a).   In addition,  Metropolitan,  TSD and EMSL, in
both  this  initial  cross-calibration  and  a  subsequent  one  which  also  included
2.4.6-trichlorophenol, obtained comparable HAA results (Fair,  1988a; Fair, I988b).

Finally, Metropolitan submitted a  16-point  Quality Assurance (QA) Project Plan for the
USEPA study, which was approved by the  USEPA  QA officer Audrey D.  Kroner
(McGuire,   1988).    The  QA  protocol   covers  accuracy,  precision,  independent
verification and the  use  of an internal standard.   Accuracy is dependent  on many
factors, but the most important is the calibration  curve.  Accuracy was  monitored  by
calculating  the  recoveries  of samples which  had  been  enhanced  with  known
concentrations of the compounds  of interest.  Precision is another parameter that is
dependent  on  more than one  factor.   The precision  of a method was  monitored  by
analyzing samples in duplicate and calculating the  difference between the two analyses.
Independent verification of a method was done by  interlaboratory calibration (discussed
above).  The internal standard was used to insure that there were  consistent  injections
into  the GC of samples and standards.

All of the  above mentioned parameters were important in assuring that good quality
data were produced.   It is important to note that  all portions of the QA program met
the established standards in order for an analysis  to be considered in control.

Initial calculations of the method detection  limits (MDLs) were made according to the
Code of  Federal Regulations 40 Part 136,  July 1,  1987.  A  set of 7 standards were
prepared in organic-pure water at 1 to 5 times the estimated detection limit.   Each
standard  was analyzed according  to  the method and  the standard deviation of the 7
replicate  measurements for each analyte was determined.  These MDLs were used as
minimum reporting levels (MRLs), except  where  the  instrumental detection  limit has
proved to be higher.  Often,  the MRLs corresponded to the lowest level standard on the

                                      3-5

-------
                                  Methodology


calibration curve.  The MRL for each analyte is shown in Table 3-6.  Since some DBFs
were detected in  the SDS blanks,  a higher MRL was used for SDS samples, the value
based on a calculation utilizing the results from 7 SDS blank samples.

At  the  beginning of this study,  many  MRLs corresponded to  the MDL-calculated
values,  even  though the calibration curves did not include a standard at that low of a
level.   Thus, MRLs were  either  raised to  correspond to the lowest level standard
analyzed  (e.g., chloroform  MDL  was 0.021  /yg/L,  but lowest level  standard run was
0.102) or the calibration curve was expanded  to include a lower  level standard at the
same level as the calculated MDL  (e.g.,  dichloroacetonitrile MDL was 0.025 /vg/L, but
lowest  level standard had initially  been  1 /7g/L; thus a 0.027 j/g/L level  standard was
added to  the curve).   As was discussed in the previous section on SDS  testing, SDS
samples require a higher MRL to remove the uncertainty in reporting  SDS data that are
comparable to those levels detected in SDS blanks.

Quantitation  was done  using an  external standard calibration curve.   Standards were
prepared  in  organic-pure water  spiked with the  appropriate DBFs and  extracted,
derivatized. or purged with  the samples.  The  extracted or purged  standards were used
to  compensate for  the varying  extraction  or purging  efficiencies  of  the different
compounds in the analysis.   Samples with analytes outside the range of the calibration
curve required re-analysis with an appropriate dilution.

An  internal  standard was spiked  into each  sample or extract.   The purpose of the
internal standard  was to monitor injections made into the GC.  A  sample injection was
deemed acceptable if the area counts of the  internal standard peak did not vary more
than +/-  10 percent from other samples which were extracted or  purged on the same
date.

Sample  spikes were analyzed to monitor the extraction or purging efficiency of specific
analytes in sample matrices.  This measured  the  accuracy of the method in  a  natural
matrix.  The spiked samples were  analyzed at a frequency of at least 10 percent of the
samples.  The samples were  spiked at the levels that are typically found in samples.
Data were entered into the  quality control (QC) table directly after the analysis.  The
QC charts were  reviewed by the  analyst and the immediate supervisor.   All spike
recoveries must fall  within the upper and lower control limits to be acceptable.  If a
spike recovery was not acceptable, then the  samples were re-analyzed from the point
where the last spike was in control and a sample was re-spiked and  re-analyzed only for
those analytes that were out of control.

Initial control limits were defined  by  calculating  the mean percent recovery from the
most recent 50 sample spike data  points. The 99 percent confidence interval is  +/-
three times the standard  deviation,  which defined the control limits.  Warning limits
were defined as +/- two times the standard deviation.  If a sample recovery was above
or below the warning limit this indicated there was a potential problem.  The problem
was determined and corrected before the analysis was  out of control.  Control limits
and warning  limits were re-calculated on a semiannual  basis using the most recent  50
spiked sample percent recovery values.   Data points that were out of control were not
included in the re-calculation of new control limits.
                                      3-6

-------
           TABLE 3-6
MINIMUM REPORTING LEVELS (MRLs)
Analyte
DBFs, //g/L:


Trihalomethanes
Chloroform
Bromodichloromethane
Dibromochloromethane
Bromoform
Haloacetonitriles
Trichloroacetonitrile
Dichloroacetonitrile
Bromochloroacetonitrile
Dibromoacetonitrile
Haloketones
1 . 1 -Dichloropropanone
1.1,1 -Trichloropropanone
Haloacetic Acids
Monochloroacetic acid
Dichloroacetic acid
Trichloroacetic acid
Monobromoacetic acid
Dibromoacetic acid
Chlorophenols
2 ,4-Dichlorophenol
2.4.6-Trichlorophenol
Pen t achlorophenol
Aldehydes
Formaldehyde
Acetaldehyde
Miscellaneous DBFs
Chloropicrin
Chloral hydrate
Cyanogen chloride
Minimum Reporting Levels
1st
Qtr


0.021
0.020
0.017
O.OJ3

0.012
0.025
0.035
0.076

0.030
0.013

1.0
0.6
0.6
0.5
0.6

2.0
0.3
0.4

NA
NA

0.010
0.2
0.02
2nd
Qtr


0.021
0.020
0.017
0.013

0.012
0.025
0.035
0.076

0.030
0.013

1.0
0.6
0.6
0.5
0.6

NA
0.4
NA

1.0
1.0

0.010
0.2
0.1
3rd
Qtr


0.102
0.103
0.108
0.101

0.029
0.027
0.027
0.028

0.026
0.029

1.0
0.6
0.6
0.5
0.6

NA
0.4
NA

1.0
1.0

0.026
0.050
0.1
4th
Qtr


0.102
0.103
0.108
0.101

0.029
0.027
0.027
0.028

0.026
0.029

1.0
0.6
0.6
0.5
0.6

NA
0.4
NA

1.0
1.0

0.026
0.050
0.1
Spring/
Summer
1989

0.102
0.103
0.108
0.101

0.029
0.027
0.04
0.08

0.026
0.029

1.0
0.6
0.6
0.5
0.6

NA
0.4
NA

3.0
3.0

0.026
0.050
0.2
ai,
SDS
Tests

0.52
0.22
0.48
0.91

0.03
0.04
0.19
1.20

0.09
0.03

1.7
2.1
1.5
0.5
0.6

NA
0.4
NA

1.7
1.8

0.04
0.43
O.I

-------
                              Table 3-6, Continued

                        Minimum Reporting Levels (MRLs)
Analyte


Other Analytes
Total organic carbon, mg/L
Total organic halide, //g/L
Ultraviolet absorbance. cm '
Chloride, mg/L
Bromide. mg/L
Minimum Reporting Levels
1st
Qtr


0.05
10
NA
NA
NA
2nd
Qtr


0.05
10
0.005
O.I
0.01
3rd
Qtr


0.05
10
0.001
0.1
0.01
4th
Qtr


0.05
10
0.001
O.I
0.01
Spring/
Summer
1989

0.05
10
0.001
O.I
0.01
SDS
Tests

NA
10
NA
NA
NA
NA  = Not Analyzed
SDS = Simulated Distribution System

-------
                                  Methodology


Prior to the calculation of initial control limits for spike  recoveries,  control limits of
+ /- 20 percent were used for all DBF fractions except for cyanogen chloride.  For the
latter analyte,  control limits from EPA Method 525.2 for volatile organic compounds
were preliminarily  used (i.e.,  +/- 40 percent; however, warning limits were set at +/-
20  percent). As the data in Table 3-7 indicate, calculated control limits were consistent
with  the  preliminary  assumptions.   For the  HAAs,  the use  of  +/- three standard
deviations  created  control limits  that were  relatively high due  to  some outlier values
affecting  the statistics.    In  fact, virtually all  spike  recoveries fell  within  +/-  two
standard deviations; therefore, the control limits for these analytes were based on the
latter statistics.

Sample duplicates  were  analyzed in order to monitor  the  precision  of  the method.
Duplicates were analyzed on randomly selected samples at a frequency of at least 10
percent of the samples.  Data were  entered into the QC table  after the analytical run
was completed.   The  QC charts were  reviewed by the  analysts  and the immediate
supervisor.   Control limits were determined by calculating the  range as  a function of
the  relative standard  deviation  (coefficient  of variation)  as  specified  in  Standard
Methods  proposed  Method  1020B.    The  normalized range  (R,,) was  calculated as
tollows:
                      + x2)/2J

     where x, and x2 are the duplicate values.

A mean normalized range (Rin) was calculated for 50 pairs of duplicate data points:
              n

     where n = number of duplicate pairs.

The variance (s2) of the normalized ranges was calculated:
     S* =  ~	
            n-1

The standard deviation (s) was calculated as the square root of the variance. The upper
and  lower control limits were defined as R,,, + 3s and zero, respectively.  All duplicates
must fall within the control limits to be acceptable.  If a duplicate  was not acceptable,
then the samples were re-analyzed from the point  where the last duplicates were in
control and a duplicate was re-analyzed only for those analytes that were out of control.
The upper warning  limit  was defined as R   + 2s.  If an  Rn was outside the warning
limit, this  indicated there  was a potential  problem.  The  problem was investigated
before the analysis  was  out of control.   Control limits  were recalculated on a semi-
annual basis using the most recent 50 points.  Data points that were out of control were
not included in the  recalculation of new control limits.
                                       3-7

-------
         TABLE 3-7



DBF QUALITY CONTROL LIMITS
Analyte
CHC13


CHCI2Br


CHCIBr,


CHBr,
J

TCAN


DCAN


BCAN


DBAN


I.I-DCP


Period
Covered
Preliminary
Initial
Recalculated
Preliminary
Initial
Recalculated
Preliminary
Initial
Recalculated
Preliminary
Initial
Recalculated
Preliminary
Initial
Recalculated
Preliminary
Initial
Recalculated
Preliminary
Initial
Recalculated
Preliminary
Initial
Recalculated
Preliminary
Initial
Recalculated
Spike Quality Control Limits
UCL UWL LWL LCL
Percent Recovery
120
121
119
120
123
119
120
125
126
120
120
120
120
122
116
120
126
134
120
131
141
120
134
135
120
123
126

113
111

116
112

117
117

113
112

114
110

119
124

122
129

124
124

116
119

83
81

86
84

86
82

87
81

77
85

91
84

87
82

82
84

88
87
80
75
73
80
79
77
80
78
73
80
81
73
80
68
78
80
84
74
80
78
71
80
71
74
80
81
79
Duplicate Limits
UCL UWL
Normalized Range
0.2
0.2
0.14
0.2
0.2
0.10
0.2
0.2
0.13
0.2
0.2
0.13
0.2
0.2
0.2*
0.2
0.2
0.21
0.2
0.2
0.20
0.2
0.2
0.19
0.2
0.2
0.20


0.10


0.076


0.099


0.095





0.15


0.15


0.14


0.15

-------
   Table 3-7, Continued
DBF Quality Control Limits
Analyte
I.I.I-TCP

Chloropicrin

Chi Hydrate

MCAA

DCAA

TCAA

MBAA

DBAA

TCP
CNCI
Period
Covered
Preliminary
Initial
Recalculated
Preliminary
Initial
Recalculated
Preliminary
Initial
Recalculated
Preliminary
Initial
Recalculated
Preliminary
Initial
Recalculated
Preliminary
Initial
Recalculated
Preliminary
Initial
Recalculated
Preliminary
Initial
Recalculated
Preliminary
Initial
Preliminary
Initial
Spike Quality Control Limits
UCL UWL LWL LCL
Percent Recovery
120
121 113 82
127 118 81
120
123 115 81
120 113 86
120
124 113 72
130 119 74
120
1200
1190
120
1290
1290
120
1370
1340
120
1220
1200
120
1210
1230
120
1330
140 120 80
138 126 75
80
75
72
80
73
79
80
62
63
80
870
760
80
740
740
80
810
780
80
820
820
80
820
740
80
700
60
63
Duplicate Limits
UCL UWL
Normalized Range
0.2
0.2
0.17
0.2
0.2
0.15
0.2
0.2
0.074
0.2
0.2
0.21
0.2
0.2
0.16
0.2
0.2
0.16
0.2
0.2
0.20
0.2
0.2
0.19
0.2
0.2*
0.2
0.28

0.13

0.11

0.057

0.16

0.12

0.12

0.15

0.15

0.21

-------
                               Table 3-7, Continued

                            DBF Quality Control Limits
UCL = Upper control limit
UWL = Upper warning limit
LWL = Lower warning  limit
LCL = Lower control limit
* Not enough points for calculation of control limits.
# The calculated control limits were relatively high due to some outlier values affecting the
statistics; therefore, the calculated warning limits were used as the control limits in order to
keep closer control over the analysis.

-------
                                  Methodology


 Prior to the calculation of initial control limits for the precision of replicate analyses,
 control limits of 0.2 for a normalized range were used for all DBF fractions.  As Table
 3-7  indicates,  calculated control  limits were determined  later in the study;  however,
 these values were consistent with initial assumptions.

 ON-SITE EXTRACTION STUDIES

 For  the  DBF studies, samples were  collected from  35 utilities  throughout the United
 States.   These samples  were collected  in  bottles containing a  dechlorination agent
 and/or preservative to  ensure the integrity of the sample during shipping and  storage
 prior to  analysis.  To evaluate how effective the preservation methods were in different
 matrices, two of Metropolitan's  laboratory staff extracted samples on-site  at Utility 2.
 In addition, split  samples were preserved and brought  back  to the  laboratory for
 analysis.   The latter  samples were stored at 4°C for 24 hours  to simulate typical
 overnight shipping  conditions.  Samples were analyzed in triplicate, both on-site and at
 the  laboratory, to  provide  adequate  statistics.    For the  Metropolitan  laboratory
 extractions,  the pentane-extractable  DBFs were  extracted upon  receipt  (after the
 24-hour  shipping period). The chloral hydrate and HAA samples were extracted in the
 laboratory within the established  holding period (see Table 3-2).

 Both the chlorinated filter effluent and chloraminated clearwell effluent were sampled in
 order  to evaluate the  effects  of  the  dechlorination  agents/preservatives  on  both
 disinfectants.   Ammonium chloride  is added to samples  to  convert free  chlorine to
 chloramines as  a means  of preserving pentane-extractable DBFs (i.e., THMs,  HANs,
 haloketones, and chloropicrin).   Since the clearwell effluent is already chloraminated,
 that  location was  sampled on-site both with and without the  addition  of ammonium
 chloride  for this  analytical fraction.   Other  analytical  fractions evaluated included
 chloral   hydrate (dechlorinated   with  ascorbic  acid) and  HAAs   (preserved  with
 ammonium  chloride).

 A concern in this special study was the effect of the dechlorination agents/preservatives
 on the pH  of samples.   The table below shows the  pH drop brought about by these
 reagents:
Sample               Actual      + Ammonium Chloride   + Ascorbic Acid

Filter effluent          7.48             7.27                   3.78
Clearwell effluent       7.55             7.16                   4.42

The data are presented in Tables 3-8 and 3-9; the findings are as follows:

     I.   The THMs, chloropicrin, choral hydrate, and the HAAs were stable to  + /- 20
         percent between extractions on-site and in the laboratory  in both samples.
         Note (in Table 3-9) that  mpnochloroacetic acid  was  27 percent higher in the
         clearwell sample analyzed in  the  laboratory.  However,  if one examines the
         number of significant  figures  in the results (i.e.. 2) and the relative standard
         deviation of the on-site results (i.e.. standard deviation divided by  the mean,


                                      3-8

-------
              TABLE 3-8



UTILITY 2: CHLORINATED FILTER EFFLUENT



     ON-SITE AND LAB EXTRACTION
On-Site Extraction

Compound
Chloroform
Bromodichloromethane
Dibromochloromethane
Bromoform
Trichloroacelonitrile
Dichloroacetonitrile
Bromochloroacetonitrile
Dibromoacetonitrile
1 . 1 -Dichloropropanone
I.I.I -Trichloropropanone
Monochloroacetic Acid
Dichloroacetic Acid
Trichloroacetic Acid
Monobromoacetic Acid
Dibromoacetic Acid
Chloropicrin
Chloral Hydrate
Mean*
Owg/L)
60.3
15.6
3.41
0.15
0.32
13.1
3.08
0.63
0.83
2.26
1.2
15.5
24.6
ND
0.6
0.10
7.02
Std.
Dev.
2.1
0.31
0.05
0.003
0.01
0.35
0.05
0.04
0.02
0.06
NA
0.67
0.61
NA
0
0.002
0.35
* Some mean data reported to three digits
however, data only
# Percent Stability =
NA = Not analyzed
ND = Not detected
good to two
100 x (mean


Lab Extraction
Mean*
(x/g/L)
58.6
14.6
3.00
0.13
0.14
10.9
2.03
0.24
0.66
1.91
1.3
14.2
23.3
ND
0.6
0.09
6.3
to enable percent
Std.
Dev.
3.2
0.73
0.25
0.007
0.006
0.47
0.17
0.05
0.06
0.11
O.I
0.45
1.0
NA
0
0.008
0.26
stability
Percent
Stability*
97
94
88
87
44
83
66
38
80
85
108
92
95
NA
100
90
90
calculations;
significant figures.
of lab extraction)/(mean of on-site extraction)









-------
                                    TABLE 3-9

              UTILITY 2: CHLORAMINATED CLEARWELL EFFLUENT

                         ON-SITE AND LAB EXTRACTION
On-Site Extraction
Compound
Chloroform
Bromodichloromethane
Dihromochloromethane
Bromoform
Trichloroacetonitrile
Dichloroacetonitrile
Bromochloroacetonilrile
Dibromoacetonitrile
1 . l-Dichloropropanone
1 . 1 . l-Trichloropropanone
Monochloroacetic Acid
Dichloroacetic Acid
Trichloroacetic Acid
Monobromoacetic Acid
Dibromoacetic Acid
Chloropicrin
Chloral Hydrate
Unpreserved*
Mean* Std.
0/g/L) Dev.
56.2
14.7
3.00
0.14
0.21
12.7
2.66
0.47
1.43
1.71
NA
NA
NA
NA
NA
0.13
NA
1.0
0.30
0.11
0.005
0.004
0.37
0.07
0.01
0.1 1
0.06
NA
NA
NA
NA
NA
0.04
NA
Preserved"
Mean*
0/g/L)
56.0
14.7
3.03
0.14
0.19
12.6
2.69
0.47
0.98
1.80
1.1
13.6
22.9
ND
0.5
0.12
6.74
Std.
Dev.
0.46
0.03
0.03
0.001
0.004
0.17
0.03
0.009
0.02
0.01
0.21
0.10
0.47
NA
0.05
0.00
0.07
Lab Extract.
Preserved"
Mean*
0/g/L)
49.8
13.0
2.47
0.12
0.097
9.90
1.84
0.18
0.58
1.46
1.4
13.7
21.7
ND
0.6
0.10
6.2
Std.
Dev.
3.4
1.1
0.29
0.008
0.005
0.62
0.23
0.21
0.04
0.14
0.12
0.06
0.35
NA
0
0.005
0.02
Percent
Stability**
89
89
82
86
51
79
68
38
59
81
127
101
95
NA
120
83
92
 *
**
NA
ND
Results from an unpreserved sample (i.e., no ammonium chloride).
Results from a preserved sample (ammonium chloride for all but chloral hydrate,
which was ascorbic acid preserved).
Some mean data reported  to three  digits to enable percent  stability  calculations:
however,  data only good to two significant figures.
% Stability =  100 x ((mean  of lab extraction)/(mean of preserved on-site extraction)!
Not analyzed
Not detected

-------
                                  Methodology


          thus  19 percent), the calculation of the stability is affected by the uncertainty
          in the former data.

     2.   Trichloroacetonitrile  (TCAN), as in  Metropolitan  water in previous studies
          (Koch et  al.. 1988). degraded to 44 to 51  percent of its initial value in 24
          hours.  In Metropolitan's previous studies, the other HANs  were stable to at
          least  80 percent of their initial  value;  however, the same situation was  not
          experienced  at  Utility  2.    Bromochloro-  and dibromoacetonitrile  (DBAN)
          degraded  to 66-68 and 38 percent, respectively, of their initial (on-site) values
          in 24 hours; although, dichloroacetonitrile (DCAN) did remain stable (i.e., to
          79-83 percent).  Note that DBAN experienced a high standard deviation in
          the clearwell samples.

     3.   Both   haloketones  were  stable  in  the  filter effluent  samples;  however,
          1,1-dichloropropanone (1,1-DCP) degraded to 59 percent of its on-site value
          in  24 hours  in the clearwell  sample.   Also note that clearwell samples
          collected  with and without  preservative yielded similar on-site results  for all
          pentane-extractable DBFs,  except for 1,1-DCP.  The unpreserved sample had
          1.43  /wg/L  1,1-DCP;  whereas,  the  ammonium-chloride-preserved  sample
          yielded an on-site 1,1-DCP  value of 0.98 //g/L.

To  continue  to evaluate  the preservation  techniques,  a  laboratory  analyst  from
Metropolitan performed on-site extractions at Utility 33. This utility was chosen for its
low alkalinity and lack of natural buffering capacity.  This matrix  is ideal for studying
the pH change and effects caused by the addition of these preservatives.  Samples that
were extracted  in the field included  the pentane-extractable DBFs and chloral hydrate.
The pH of the water was measured  in the field  as  is and at the  laboratory after field
preservation and overnight shipment.  The  results are as follows:

                           	pH
Sample               Actual      + Ammonium Chloride    + Ascorbic Acid

Effluent                6.96             6.31                   3.46

The data on the chlorinated effluent are presented  in Table  3-10; and the findings are
as follows:

     I.   There was little in the way of brominated DBPs, due to the low bromide level
         in the source water.  Thus, it was difficult to evaluate those data, as on-site
         results were  close to the minimum  reporting levels.  However, data for the
         other DBPs indicate that chloroform, bromodichloromethane, the haloketones.
         chloropicrin,  and  chloral hydrate were stable to  + /- 20 percent of their on-
         site values.

     2.   The major HAN present  was  DCAN;  although, the levels detected were less
         than 1.0 //g/L.  The laboratory result was higher by 0.24 //g/L.

During  the field  study, an  attempt  was made to analyze  some samples where the
chlorine residual was not quenched or converted to chloramines.  This utility practices

                                       3-9

-------
           TABLE 3-10
UTILITY 33: CHLORINATED EFFLUENT
   ON-SITE AND LAB EXTRATION
On-Site Extraction

Compound
Chloroform
Bromodichloromethane
Dibromochloromethane
Bromoform
Trichloroacetonitrile
Dichloroacetonitrile
Bromochloroacetonitrile
Dibromoacetonitrile
1 . 1-Dichloropropanone
I.I.I -Trichloropropanone
Chloropicrin
Chloral Hydrate
# Percent stability = 100
* On-site result too close
stability
** Tentative result due to
NA = Not analyzed
ND = Not detected
Mean
(x/g/L)
22
1.3
ND
0.16
ND
0.61
0.07**
0.09
1.8
3.3
0.50**
1.7
x [(mean
Std.
Dev.
1.1
0.06
NA
0.01
NA
0.02
0.0
0.12
0.04
0.14
0.02
0.04
Lab Extraction
Mean
(A/g/L)
23
1.4
ND

-------
                                 Methodology


disinfection only, so the effluent  sample is located, in time,  very close to  the point of
chlorine application.   Thus,  a high  free chlorine residual was present in  the  effluent
sample (i.e..  2.5 mg/L). The field laboratory where on-site extractions were performed
was  located  approximately two  hours away from  the effluent sample tap.   Thus,
unpreserved  samples  continued to react with free chlorine  yielding high  DBF levels
(e.g.. 62  /yg/L of  chloroform was  detected; whereas, preserved samples yielded 22
/ug/L).  These results emphasize  the need for a dechlorination agent or preservative to
stop the reaction of free chlorine in forming DBFs after a sample is collected.

One final utility (Utility 6) was  selected for an on-site study.   The pH results are as
follows:
Sample               Actual      + Ammonium Chloride   + Ascorbic Acid

ClearwelI  effluent       8.00              7.19                  4.66

At Utility 6. chloraminated clearwell effluent was sampled for pentane-extractable DBFs
and chloral hydrate.  In addition,  water was collected with no  preservatives.  The data
from Table 3-1 I indicate the following:

     1.    The THMs. chloropicrin, and chloral hydrate were stable to +/- 20 percent.

     2.    TCAN was not detected  in these samples.  The other HANs, unlike at Utility
          2, were all stable. The laboratory results were higher than the on-site values;
          however absolute differences were j<0.24 /ug/L for each HAN.

     3.    The haloketones  were stable between  the on-site and laboratory  analyses of
          preserved samples (j<0.25 /ug/L for each compound). As observed at Utility
          2. the unpreserved sample had 1.6 /ug/L 1,1-DCP, while the preserved sample
          had  0.91 //g/L.

Overall, the on-site experiments confirm the findings of Metropolitan's holding studies
(Koch  et  al.  1988;  Krasner et al,  1989), that the chosen dechlorination  agents  and
preservatives for the pentane-extractable DBFs, chloral  hydrate, and HAAs do  result in
laboratory results  representative   of the DBF  levels  at the   time  of  sampling.
Furthermore,  in continuing to evaluate the  stability of the pentane-extractable DBFs,
nine different  utilities' clearwell  effluents were sampled and  shipped  (overnight) to
Metropolitan.  They  were extracted  upon receipt, plus a second set of aliquots were
extracted  four days after the  initial sampling.  The results were essentially the same
(within  analytical error) on both days (see Table 3-12).  TCAN.  though, did degrade
between the  two  extraction  dates,  which is consistent with  previous results on  this
unstable HAN.  Similar experiments have been performed with the HAAs  and chloral
hydrate, and re-extractions have yielded comparable data.  More  holding studies need
to be  performed  in  other matrices in  order to  better define the validity of these
dechlorination  agents and preservatives.   The data, however,  do support the idea  that
samples can be shipped to the laboratory for analysis.
                                      3-10

-------
                 TABLE 3-11



UTILITY 6: CHLORAMINATED CLEARWELL EFFLUENT



         ON-SITE AND LAB EXTRACTION
On-Site Extractions
Unpreserved*
Mean Std.
Compound O^g/L) Dev.
Chloroform 6.9 0.17
Bromodichloromethane 6.4 0.21
Dibromochloromethane 3.5 0.12
Bromoform 0.45 0.01
Trichloroacetonttrile ND NA
Dichloroacetonitrile 0.81 0.09
Bromochloroacetonitrile 0.50 0.03
Dibromoacetonitrile 0.57 0.03
!.l-Dichloropropanone 1.6 0.17
l.l.l-Trichloropropanone 0.87 0.1
Chloropicrin 0.20 0.01
Chloral Hydrate 3.2 0.04
* Results from an unpreserved sample (i.e
Preserved"
Mean
(x*g/L)
6.5
6.2
3.4
0.43
ND
0.79
0.50
0.63
0.91
0.95
0.20
2.6
Std.
Dev.
0.06
0.06
0.1
O.I
NA
0.02
0.03
0.10
0.07
0,06
0
0.02
. , no ammonium
** Results from a preserved sample (ammonium chloride for
hvdrate. which was ascorbic acid preserved).
ft Percent stability = 100 x ({mean of lab
NA = Not analyzed
ND = Not detected

extraction)/(mean




Lab Extract.
Preserved"
Mean Std.
0/g/L) Dev.
7.2
6.6
3.8
0.43
ND
1.0
0.63
0.87
1.1
1.2
0.21
2.7
chloride).
all but chloral

0.4
0.4
0.36
0.03
NA
0.06
0.06
0.23
0.12
0.06
0.01
0.09



of preserved on -site




Percent
Stability'
III
106
1 12
100
NA
127
126
138
121
126
105
104



extraction))



-------
                                                TABLE 3-12

FOIMB-D
THACTABL
Utility/Martwr of
Days after Sailing CBClj CBCljBi
Utility 1
1 Day
4 Days
Utility 3
1 Day
4 Days
Utility 23
1 Day
4 Days
Utility 24
1 Day
4 Days
Utility 26
1 Day
4 Days
Utility 27
1 Day
4 Days
Utility 29
1 Day
4 Days
Utility 31
1 Day
4 Days
Utility 35
1 Day
4 Days
0.80
1.1
2.3
2.4
1.9
1.8
0.81
0.76
43
40
6.9
7.2
117
114
36
34
20
19
1.5
2.2
1.2
1.2
3.8
3.6
0.85
0.81
10
9.9
19
18
56
50
15
14
2.4
2.3
X DZSZHTE
CHClBr2
2.3
3.4
0.26
0.27
5.3
5.0
4.5
4.3
2.5
2.4
26
23
18
17
6.5
6.2
0.11
0.10
UJ.UB HI
CHBrj
1.3
2.0
0.080
0.074
2.3
2.1
23
21
0.10
0.097
7.2
7.2
0.76
0.74
0.61
0.62
HD
ND
-fiaXJUCXS: mXXIX HOUJlBG STBDI
Concentration (pq/L)
TCMT DCM BCMI DBftB
ND
HD
0.009
ND
HD
ND
ND
ND
0.017
ND
HD
HD
0.055
0.027
RD
HD
ND
ND
0.15
0.16
0.67
0.68
0.97
0.99
ND
HD
3.6
3.6
0.94
1.0
10
9.8
1.4
1.3
0.85
0.85
0.36
0.45
0.12
0.12
1.3
1.3
0.31
0.27
1.2
1.1
2.0
2.0
4.8
4.5
1.2
1.0
0.046
0.043
0.90
1.2
0.060
0.054
1.9
2.0
5.2
4.1
0.29
0.28
3.7
3.5
1.2
1.1
0.62
0.59
ND
ND
1,1-DCF
ND
0.11
1.1
1.0
0.26
0.25
ND
ND
0.52
0.51
0.41
0.48
1.5
1.5
0.35
0.33
1.1
1.1
1,1,1-TCP CHP
0.067
0.087
1.6
1.5
0.097
0.12
HD
ND
2.0
1.8
0.66
0.50
4.4
4.3
0.41
0.40
2.0
1.9
0.010
0.015
0.082
0.083
0.043
0.053
ND
ND
0.86
0.78
0.058
0.13
0.38
0.41
0.29
0.28
0.16
0.17
HD = Not detected

-------
                                 Methodology


ALDEHYDE DERIVATIZATIONS/EXTRACTIONS ON-SITE

When aldehyde analysis was added to the scope  of work,  a new preservation concern
developed.  Formaldehyde in particular is a very unstable  compound.  Thus, samples
were  analyzed  upon  receipt  at  the laboratory.   Another investigator  performed
derivatizations and extractions in the field for aldehyde analyses  of four utilities (Glaze
et al. 1989b); however, such an action is cost prohibitive when  35  utilities nationwide
are being analyzed on a quarterly basis.  Therefore, on-site evaluations were performed
at Utilities 33 and 6 for the aldehyde fraction.

A combination of two preservatives for aldehyde fractions was evaluated.   Mercuric
chloride was added as a biocide, in order to inactivate microorganisms capable of either
producing or degrading  aldehydes.   In addition,  ammonium chloride was  added to
convert  free chlorine to chloramines, as it was suspected that a free chlorine residual
could continue to form additional aldehydes. At Utility 33 (see  Table 3-13),  it is clear
that the  addition  of  mercuric chloride  was  essential  for  the influent  sample, as
unpreserved samples  had no detectable  levels  of the measured aldehydes after the
24-hour  shipping  period.    Levels of  aldehydes at this  utility, which  utilizes  free
chlorine, were low  (<:5  pgIL of each aldehyde), and the  coefficient of variation was
high for some of the analyses.

However, when Utility 6 implemented ozonation, it was expected that higher levels of
aldehydes would be produced, which would yield more conclusive results on the use of
preservatives  and the ability to ship aldehyde samples back for laboratory analysis.  In
this experiment,  unpreserved samples  on-site were compared to fully preserved samples
analyzed after a 24-hour shipping period.   As  Table 3-14  shows, formaldehyde levels
were  identical  between  on-site and  laboratory  analyses  (e.g., the ozone  contactor
effluent  had  16 and  15 yug/L of formaldehyde  for on-site  and laboratory analyses,
respectively).  Most of the  acetaldehyde results were  comparable as well (e.g., 7.4
versus 5.7  //g/L  for  the  plant  effluent  analyzed on-site  and  in  the laboratory,
respectively).    However,  the  on-site  extraction blank contained  3.2  /wg/L of
acetaldehyde, while none was detected in  the laboratory  extraction  blank,  so  it was
difficult to compare the two sets of results due to a possible acetaldehyde contamination
problem  on-site.    However, these  data  do indicate  that aldehyde samples  can be
preserved and shipped to the laboratory for analysis  of these analytes upon receipt.

PRESERVATION OF TOX  SAMPLES

For this  study,  total  organic  halide  (TOX)  samples  were collected  with  no
dechlorination agent  or  preservative in  the  field and  shipped  iced overnight to
Metropolitan's laboratory.   After receipt, the samples  were dechlorinated with a fresh
sodium  sulfite  solution  and  preserved  with  sulfuric  acid.    Several  studies  were
performed to  evaluate to what extent  TOX formation  occurred during the 24-hour
shipping period.  TOX bottles were shipped with (1) no preservative reagents and (2)
only sodium sulfite.  Contrary to expectations, the latter samples yielded TOX results
which were generally  much higher than  those dechlorinated  in the  laboratory.   For
example,  the TOX  results for Utility  33 were 680 /ug/L for a laboratory-dechlorinated
sample and  880 pgIL for the  sample shipped  with sulfite.   These data  suggested to
some  USEPA scientists (Sorrell and Brass.  1988) that the  sodium sulfite solution was
unstable and, perhaps,  samples collected in  bottles shipped with  this chemical were not

                                      3-11

-------
                                     TABLE 3-13

                          UTILITY 33:  ALDEHYDE RESULTS

                           ON-SITE AND LAB EXTRACTION
Location/Compound
   On-Site Extractions

Unpreserved      Preserved*

 Mean  Std.     Mean Std.
0/g/L)  Dev.    (jig/L) Dev.
  Laboratory Extractions

Unpreserved     Preserved*

 Mean  Std     Mean  Std.
0/g/L)  Dev.    (//g/L) Dev.
Influent/
Formaldehyde
Effluent/
Formaldehyde
Influent/
Acetaldehyde
Effluent/
Acetaldehyde

2.3
3.4

1.6

4.0

2.9
O.I

0.2

0.3

NA
1.8

NA

2.2

NA
0.1

NA

0.3

<1.0 NA
5.1 0.58


-------
                                   TABLE 3-14

                      UTILITY 6:  FORMALDEHYE RESULTS

                         ON-SITE AND LAB EXTRACTION




                         On-Site Extract.*         Lab Extract.*

Location
Plant Influent
O, Contactor Eff.
Plant Effluent
Mean
0/g/L)
0.63**
16
16
Std.
Dev.
0.06
0
0
Mean
0/g/L)
0.83**
15
16
Std.
Dev.
0.12
0.58
0.58
Percent
Stability*
NA**
94
100
 *     On-site samples contained no preservatives; laboratory samples contained mercuric
      chloride and ammonium chloride.
 #     Percent Stability =  100 x [(mean of lab extraction)/(mean of on site extraction)]
**    Levels detected below minimum reporting level of 1.0 /ug/L
NA = Not analyzed

-------
                                  Methodology


being  dechlorinated in  the field.   When  the  next set of samples was  received,  TOX
bottles shipped with the sulfite solution were analyzed for chlorine residual and, indeed,
a  residual remained.   Because  sampling kits were  prepared and  shipped to field
locations  at  least  two weeks prior to sampling, the  instability of the dechlorination
solution required that TOX samples be dechlorinated and preserved upon receipt at  the
laboratory.   Samples dechlorinated at Metropolitan's laboratory  were treated with a
fresh sodium sulfite solution each time in  order to ensure  that the samples  were
dechlorinated.

To test the importance  of timing of the  addition of preservation agents, the following
experiments   were  performed.     Metropolitan's  chlorinated   filter  effluent  and
chloraminated plant effluent  were sampled and stored at 4°C for 24 hours to simulate
overnight  shipping conditions.   Then the samples were  preserved with those reagents
not originally present in the  TOX bottle  at time of sampling.  These data (Table  3-15)
indicate that in  Metropolitan's matrix, results were the same when fully dechlorinated
and  preserved at the time  of sampling or  after 24 hours at 4°C.  However,  samples only
dechlorinated at the time of sampling  had a  slight  loss of TOX when  not acidified
immediately.

When  on-site  extractions were  performed   at  Utility  2,  another  experiment  was
performed with  TOX samples as well (Table 3-16).  Samples were evaluated  by on-site
dechlorination (with fresh sodium  sulfite) and preservation (with  sulfuric acid) versus
dechlorination/preservation at the  laboratory  (after  24-hour storage at  4°C).   The
chloraminated water had the  same TOX when dechlorinated/preserved on-site or in  the
laboratory.   However,  the  chlorinated water  did  appear to increase  by  16 percent
during shipment/storage with no dechlorination at time of sampling.

When  on-site analyses were conducted  at  Utility  33,  special  TOX sampling was
included (Table 3-16).  The sample dechlorinated and acidified in the field contained
390  /yg/L of TOX, while the sample dechlorinated/preserved 24 hours later contained
610  //g/L. There was a 56 percent increase in TOX during the  shipping period.  These
results reflect the fact that even during refrigerated shipping, the free  chlorine was able
to react with the sample matrix and produce additional  TOX during the 24-hour period.
Since Utility 33 practiced disinfection only and the sample tap was located near  the
point of chlorination. the large amount of free chlorine residual (i.e.,  2.5  mg/L) and
short contact time before  sampling  resulted in the sample containing sufficient chlorine
and  precursors to produce the additional TOX  during snipping.

Further TOX experiments were carried out at Utility 6 (Table  3-16).   Utility 6's filter
influent was  sampled,  since  it  represented a point  close to initial chlorination (at  the
time of this  sampling).   TOX was 60 /yg/L when dechlorinated on-site and  110 //g/L
when dechlorinated in the laboratory.  The filter effluent samples had a longer contact
time with  chlorine prior to sampling and there was a smaller difference in TOX results
between the  on-site  and laboratory-dechlorinated samples (77 //g/L on-site versus 110
/jg/L in the laboratory). The chloraminated plant effluent samples were less affected by
the timing of dechlorination  (mean results of 78 //g/L of TOX for on-site dechlorination
and  89 /yg/L for  the  laboratory dechlorination).  Samples  collected  from  Utility  6's
distribution system  (water temperature 6°C) had 76 to 89 /yg/L of TOX (dechlorinated
in  the  laboratory), suggesting that the TOX did not  increase over time in this utility's
cold, chloraminated distribution system.

                                     3-12

-------
      TABLE 3-15
TOX PRESERVATION STUDY
Preservatives at
Sample1 Time of Sampling
Filter Effluent2 Sulfite'/acid
" " None
Sulfite
Plant Effluent3 Sulfite/acid
None
Sulfite
Preservatives Added
after 24 hr @ 4°C
None
Sulfite/acid
Acid
None
Sulfite/acid
Acid
# Percent difference = 100 x [(mean TOX value - "control" mean
"control" mean TOX value]
where "control" sample = sample dechlorinated and acidified at
* Fresh sodium sulfite solution used.
1 Samples from Metropolitan's Weymouth filtration plant.
: Chlorinated water
3 Chloraminated water
TOX, //g/L
(Replicates)
170, 160
160, 170
130, 150
150, 150
130. 170
130, 140
TOX value)/
time of sampling.
Percent
Diff/
Control
0
-15%
Control
0
-10%


-------
                                     TABLE 3-16

                      ON-SITE TOX PRESERVATION STUDIES
Sample Location
Cl, Dose    Nli,  Residual
 (mg/L)   Dose     Cl,
         (mg/L)   (mg?L)
                                                         TOX,
                                                  On-Site*
                           MWDSC Lab*
                                           Percent
                                            Diff.*
Utility 2

Filler Effluent
Clearwell Effl.

Utility 33

Plant Effluent

Utility 6
   8.8
   NA
           0.5
2.67 total
2.2 total
2.5 free*
310. 320
330, 270
390
380, 350
310, 295
610
-f 16%
 + 1%
+56%
Filter Influent
Filter Effluent
Plant Effluent
Distrib. Loc. ff\
Distrib. Loc. #2
Distrib. Loc. #3
2 0.5 free
0.12 free
1 0.4 0.95 total
0.70 total
0.80 total
0.40 total
60, 61
76, 78
80, 75
NA
NA
NA
NO. 110
110, 110
98. 80
82
76
89
-1-82%
+43%
-f-15%



p   Replicate analyses when available,
*   Fresh sodium sulfite solution + acid added on-site.
ft   Sodium sulfite solution 4- acid added at MWDSC lab.
i   Percent difference =  100 x |(mean MWDSC lab TOX - mean on-site TOX)/
    mean on-site TOX))
X   Very short free CI2 contact time before sample site.
NA = Not analyzed.

-------
                                  Methodology


The TOX  experiments to date  suggest that for chloraminated  waters, dechlorination
after  receipts of samples  was not problematic.  However, the results  indicate  that
chlorinated  waters would continue to produce additional TOX, which appears to be
related  to  the  amount  of chlorination that occurred prior  to  sampling.   That is, if
samples were collected shortly after chlorination, TOX  values would be off by a high
degree, whereas  samples collected after  sufficient chlorine contact time had already
occurred in the treatment  plant  probably  would not continue to produce  a significant
amount of TOX during refrigerated shipping.  Since 14 of the 35 utilities  in this study
used chloramines as  a final disinfectant, those TOX data are probably representative of
values at the time of sampling.   The largest error in interpreting  the  TOX data  will
probably be for disinfection-only utilities;  however, only four of these were included in
the study.  The on-site TOX experiment at Utility 33, though, provides some data for
interpreting the TOX results of a disinfection-only utility.

Experiments on dechlorination of TOX samples are very difficult, in that specific DBFs
cannot be isolated for study as was performed for the pentane-extractable DBF fraction
(see discussion  above).  In fact,  Croue and Reckhow (1988) demonstrated that sodium
sulfite (the TOX  dechlorination agent) readily destroys many DBFs, including some of
particular health concern (including a reduction in the mutagenicity of a sample).  For
this study, it was found to be impractical to have the  individual utilities perform  the
proper dechlorination/preservation  at time  of sampling (e.g.,  the  sulfite solution is
unstable and has  been shown to be difficult to ship to participating  utilities).  For field
surveys, it  appears that the dechlorination and preservation of TOX samples requires
more  study.
                                     3-13

-------
Section 4
Data Management and
Analysis

-------
                                  SECTION 4

                    DATA MANAGEMENT AND ANALYSIS
This chapter discusses the organization of the database used in the study in terms of the
individual database files, the fields contained  in each, and their interrelationships; the
data handling and tracking protocol; and the statistical and graphical methods by which
data were analyzed and  summarized.

DATABASE

The conceptual design of the database involved two main data files.   The first data file,
called  UTILITIES, contained information about each participating utility.  Each utility
record (one per utility)  had fields to store information such as utility identification (ID)
number, name, water  source,  treatment  type,  population size,  geographic location,
coagulant type, and disinfectant type.   The second data file,  called RESULTS, stored
all  sampling results.  Each measured value was stored in a separate record.  The fields
of the record included  utility ID,  location of sampling  point, analysis code, sampling
quarter,  detection limit flag,  and  the  measured value of the analysis  compound or
parameter.  In addition  to the two main  data files, minor files were used to translate the
various abbreviation codes into their full-length descriptions.   This complete set of data
files and their interlinking key fields is referred to as a "relational database".  The
database was implemented  using the commercial software package dBASE HI PLUS™
(Ashton-Tate) on  an IBM compatible PC.

DATA HANDLING PROTOCOL

The basic elements of data handling protocol for this study were:   1) use of signatures
by  responsible team  members,  2) data-sheet  reference numbers for each individual
project data sheet, and  3) tracking the status  of each group of data. Figure 4-1 shows
the various protocol steps performed at Metropolitan and JMM.

The first component of  protocol involved  the sign-off procedures used by Metropolitan
in transferring data from laboratory analysts'  notebooks  to the data-entry sheets sent to
JMM.   Data sheets were signed or  initialed by the transcriber and subsequently by the
person performing a transcription check.  The second protocol step  took  place at JMM
when data was entered  into the database.   Every individual piece  of paper (data sheet)
was assigned a sequential number referred to as the Reference Number.  As data were
entered into  the  computer  data-entry  screens, the data-entry  clerk  typed  in  the
Reference Number for each screen/sheet (screens and sheets correspond one-to-one) of
data,  and  all data from  that  sheet were stored  in the database tagged with that
Reference Number.  The purpose of the Reference  Number is to  allow any data point
printed from the database into a report, table or graphics file to be traced  back to the
original paper data sheet from which it came, and from  there  back to the individual lab
analyst's notebook,  if necessary.  A Data Entry Status Log kept at  the computer work
area was continually  updated  to  prevent  use  of duplicate Reference Numbers.  The
data-entry clerk signed off on each data sheet as it was entered.
                                      4-1

-------
         MWD Inorganics and Organics Lab Analysts' Notebooks
                                I
               Lab analysts enter data onto data sheets
     JMM TOX Lab Reports
                Supervisor reviews data sheets for errors
          Data sheets forwarded to JMM for entry into database
Engineer transcribes data onto JMM
TOX Worksheets
 Data-entry clerk enters data sheets and records data-sheet Reference Numbers
                                I
      Engineer produces first verification printout and notes corrections
Engineer makes corrections in database and checks off corrections on printout
 Engineer produces final verification printout and checks against first printout
       Engineer files first printout and sends final printout to MWD
                                    DATA-SAMPLING PROTOCOL
                                             FIGURE 4-1

-------
                         Data Management and Analysis


The next protocol step was to create a data verification computer printout for the range
of newly-entered Reference Numbers.   A single dBASE software program was created
to produce the printout for all project data, referred to as the Verification Report.  The
output format  of the  report was  virtually identical to the original data-entry  sheets,
allowing rapid visual comparison of the printout to the originals. An engineer familiar
with the project  performed the  data-verification step.   The printout was compared to
the originals,  and any needed corrections were marked on the printout with a colored
pen.  As this verification took place,  the  engineer  signed off on each of the original
data sheets.

Following  verification,  the  engineer  went into the computer  database  files  with  the
dBASE data-file  editor and manually performed  the necessary corrections.   As each
correction was made, it was checked off on the verification printout using a second pen
of a different  color.  The final protocol step was the production of a final Verification
Report, which  was compared to the original, marked-up verification printout to  ensure
that  all  necessary corrections were made.  Completion of each of the  above steps was
duly noted on the Data Entry Status Log.

The  use of signatures by all responsible parties and the careful checking of data at each
step resulted in an extremely reliable database.

SUMMARY STATISTICS

The  data from this study consist of plant influent and filter influent (where applicable)
water quality  characteristics and  clearwell  effluent DBFs  from  35 water  utilities'
treatment plants.  As  described in preceding chapters, each plant was  sampled once
every quarter  for a total of four  quarters  (although not  all analytes were determined
every quarter). For each DBF compound or water quality parameter, it is of interest to
examine  and compare the set of values (one  per plant) for particular quarters, and  the
complete annual  set of values (from  all plants for all quarters sampled).  This large
amount of data created the need for a way to  summarize the various data sets.

For conveying  the essential  features of any set of data, it is desired to express  both a
measure  of central tendency and a measure of variability. The traditional parametric
approach  consists of  reporting  the  sample  mean and  sample  standard deviation.
However, this  approach suffers from several drawbacks:

I.   Assumed Distribution:  In the water quality field, it  is frequently  assumed that all
     water quality data is either normally or log-normally  distributed.  This assumption
     is perhaps based on a limited  understanding of statistics, and on studies involving a
     few  conventional water quality parameters or compounds  which were found to be
     adequately modeled by normal or log-normal  distributions.   Thus, a wholesale
     tendency  has emerged to apply parametric techniques based on those  distributions,
     frequently without testing  the validity of  the assumption.   It  is not always
     recognized  that  the  common practice  of  computing the mean  and  standard
     deviation  is rooted in the statistical methodology of estimation, in this context the
     estimation of the  population  parameters of a normal distribution from  a limited
     sample assumed to be drawn from  the  population.   In  association  with the
     computation  of the sample mean, the 95% confidence interval for the mean  is also
     frequently reported as a measure of the accuracy of the population estimate.

                                      4-2

-------
                         Data Management and Analysis


     Wilh  these classical statistical  methods (in the early  phase of this  project), the
     sample  means  of compounds  and class  totals  across the  35  utilities,  and their
     associated  95%  confidence  intervals  based  on  the  normal  distribution,  were
     computed.   Subsequent  testing of the individual  DBF compounds and chemical
     class totals indicated that they followed different distributions, and that neither the
     normal  nor log normal distribution seemed to apply in some cases.  Furthermore.
     upon  careful  reflection it was realized that  a major objective of the study was to
     explore, describe, and summarize  the sample data, not to estimate the parameters
     of an  underlying  distribution  for a larger population, such as the population of all
     United States water utilities,  for example.   Even if this were a major objective, it
     is  possible that Ihe non-randomized sample of 35 utilities  with different selected
     source waters and treatment practices would be inadequate to do so.

2.   Sensitivity to Outliers: A single or several extremely outlying data values can have
     a  substantial adverse  impact  on  both the  sample mean  and sample standard
     deviation.  Especially for exploratory  data analysis, it is often advantageous to use
     simpler  summaries based on sorting and counting which are  more resistant: that is.
     an arbitrary change in a  small part of the data set  can have only a small effect on
     the summary.

3.   Loss of Information:  Information about  the characteristics of  a  data set  can be
     obscured by reducing the data  to one or two summary numbers such as  the mean
     and standard  deviation.   For example, the minimum and maximum values cannot
     be discerned, and questions about how many of the values in the sample  are above
     or below a specified level cannot be answered.

4.   Censored data: Data  for compounds which  are reported as  below  detection limits
     (in this study. Minimum  Reporting Levels (MRLs)) complicate the issue of how to
     compute summary statistics such  as  the mean and standard deviation.   Various
     approaches for handling such  "left-censored"  data (below detection limits) are
     encountered in  practice,  including  treating these values as equal to zero, equal to
     the detection limit, or equal to one-half  the detection  limit.  However, each of
     these  techniques  bears conceptual  difficulties and objections (Helsel  and Cohn,
     1988).

Consideration of  the  above  discussion led to  a decision to  utilize  nonparametric
measures to analyze  the data from this study.  Nonparametric statistical  methods do not
require  an  assumed parametric distribution for  the data, and cases below the  detection
limits can  be incorporated.   In the area  of applied statistics  called exploratory data
analysis, (he  data in  a  sample  set are sorted in  ascending order of magnitude,  and then
characterized by counting up from the lowest value and determining percentages called
percenliles.   In particular, the  "five-number summary" (Tukey. 1977) is frequently
used to  summarize a data set.

The  five-number summary consists  of  the minimum value, 25th percentile, median.
75ill percentile,  and maximum value from the data set.  The "nth" percenlile is the
value below  which n  percent of  the data lie.   The most  familiar percentile is the
median,  the  50th percentile, above and  below  which  50  percent of the  data  lie.
Because the 25th.  50th, and 75th percentiles divide the data set into  four quarters, the
25th and 75th percentiles are frequently referred  to as  the  lower and upper quartiles,

                                      4-3

-------
                         Data Management and Analysis


 respectively.  The difference or spread between the lower and upper quartiles is known
 as ihe interquartile range, and it encompasses the middle 50 percent of the data values.
 Thus, in  nonparametric  statistics,  the median and interquartile range are analogous to
 the mean and one standard deviation above and below the mean in parametric statistics.
 They also provide much more resistant measures of central tendency and variability.
 For example, all the values below  the lower quartile including the minimum can change
 in  magnitude,  even  drastically,  without affecting  the  values  of  the median and
 interquartile range.

 The five-number summary gives five pieces of information about a data set instead of
 just one or two.  It also allows very useful but  simple  statements to  be  made, such as
 "75  percent of  the chloral hydrate data for the utilities sampled fell below 1.1 /jR/L."
 Lastly,  medians and other percentiles can still be determined for data sets containing
 censored data.  For example, for the following data set (n  = 9):

                <0.l. <0.l.  <0.l. <0.1. 0.12, 0.43, 1.1, 2.0,  3.5

 the lower  quartile is 
-------
                     GUIDE TO NOTCHED BOX-AND-WHISKER PLOTS
                                                       +

                                                        n
                                           EXTREME OUTLIER
                                           any value outside 3 interquartile ranges
                                           measured from the 25 and 75 percentiles

                                           OUTLIER
                                           any value outside 1.5 interquartile ranges
                                           measured from the 25 and 75 percentiles

                                           MAXIMUM VALUE
                                            this is the largest value (excluding outliers)
                                                                               75 PERCENTILE
                                                                                75% of the data are less than or
                                                                                equal to this value (including outliers)
INTERQUARTILE RANGE
contains the data between
the 25 and 75 percentiles
95% CONFIDENCE INTERVAL
 for the median (including outliers)
MEDIAN
50% of the data are above or
below this value (including outliers)
        NOTE: Horizontal width of box is
        proportional to square root of sample size (N)
                                                  -^-  25 PERCENTILE
                                                         25% of the data are less than or
                                                         equal to this value (including outliers)

                                                MINIMUM VALUE
                                                this is the lowest value (excluding outliers)
                                                     FIGURE 4-2

-------
                         Data Management and Analysis


outlier, and is excluded when determining the minimum and maximum for setting the
endpoints of the whiskers. For a perfectly symmetric normal distribution, the choice of
the  1.5 interquartile range cutoffs correspond to a probability of only 0.7 percent (p =
0.007. 0.0035 in each tail of the distribution) for a data point to lie outside those limits
(Hoaglin,  Mosteller, and  Tukey,  1983).  The  outlier criterion is only a guide  as to
which data points  should be more carefully scrutinized as potential outliers, not an
absolute rule.

Notched box-and-whisker plots portray the  95 percent confidence interval  about the
median as a notch-width in  the box.  This confidence interval  can be  used to test the
pairwise significance of differences  in medians  between  two data sets, which is  a
nonparametric analog of the classical t-test for the significance  of differences in  means
of normally distributed data  sets.  An example  is shown in Figure 4-3.  In comparing
the box-and-whisker plots for three data sets (A, B and C in Figure 4-3), the 95 percent
confidence interval  (notch-width) for one data set (A) would be  compared to that of the
second  data set (B), and then  to that  of the third data set (C).  If the 95 percent
confidence intervals of any two medians do  not overlap (medians for A and  B,  and A
and  C).  then the difference between those medians is  statistically significant at a 95
percent confidence  level.  If the 95 percent confidence intervals of the medians of any
two  data sets  overlap (B and C), then the  difference between those two medians is not
statistically significant at a 95 percent  confidence  level.  It should be noted  that this
type  of comparison is  pairwise only.  In other words, even if box-and-whisker plots
from three different data sets are  being compared, the 95 percent confidence intervals
for only two at  a  time  (A and  B,  A and  C, and B  and C) can  be compared for
determining whether or not there are  statistically significant differences between the two
at a  95 percent confidence  level.   A more detailed  discussion of  the 95 percent
confidence interval  about the median has been prepared by McGill, et al. (1978).

Bar  Charts

A derivative form of the box-and-whisker plot was also created and used for this  study.
As shown  in  Figure 4-4.  standard bar graphs were made with the height of the bar
equal to the median of the data set.  The interquartile range was then superimposed on
the bar. being drawn as an I-shaped "error bar" (more properly, "variability bar") with
the horizontal cross-members of the  "I" at the  lower and upper quartiles.  Thus, this
form of bar chart  represents a  "three-number summary",  omitting the minimum and
maximum  values.    These graphs were used in cases where  multiple compounds or
chemical class totals were portrayed  on  the same graph, for which differences in
medians  are not relevant.  Rather, they provide a relative indication of location (central
tendency) and spread (variability).  For  those DBFs that had relatively high MRL values
and  the 25th  percentile. median and  75th  percentile were less  than the MRL. the bar
chart might be interpreted as giving  a  misleading  representation of the concentration.
In such circumstances,  the bar was drawn solid black rather than cross-hatched.  For
purposes of comparison, the MRL is indicated on the bar chart in some figures  where
the median was greater  than the  MRL, as shown in  Figure 4-4.

Star  Symbol Plots

Another type of graphical presentation that was used for this study was the star symbol
plot  (STSC Inc..  1987).  The star symbol  plot is not part of the standard repertoire of

                                       4-5

-------
Example of Notched Box-and-Whisker Plot
           for Three Data Sets
                   B
                FIGURE 4-3

-------
                   GUIDE TO BAR CHARTS USING MEDIAN,
              PERCENTILES AND MINIMUM REPORTING LEVEL
INTERQUARTILE RANGE
contains the data between
the 25 and 75 percentiles
                                                                75 PERCENTILE
                                                                75% of the data are less than
                                                                or equal to this value
MEDIAN
50% of the data are above
or below this value
                                                                25 PERCENTILE
                                                                25% of the data are less than
                                                                or equal to this value

                                                                MINIMUM REPORTING LEVEL
                                                                lower limit of quantitation
                  NOTE - ASTERISKS ON CHARTS INDICTE THE FOLLOWING:
                     *  25 percentile < Minimum Reporting Level
                     ** 25 percentile and median < Minimum Reporting Level
                     *** 25 percentile, median, and 75 percentile < Minimum Reporting Level
                                       FIGURE 4-4

-------
                         Data Management and Analysis


exploratory data analysis and nonparametric statistics, as are box-and-whisker plots.  It
is. however, a unique way of visually portraying differences in multivariate (multiple
variable) "profiles" between entities.

A star symbol plot is used to  graphically display the relative magnitude of the values of
multiple variables for two  or  more  entities.   As  an  example,  consider the four
compounds which  make up  the regulated  quantity, total trihalomethanes as the four
variables to be portrayed.  This example is illustrated in  Figure  4-5.  On a star symbol
plot, the magnitude of each variable is plotted as a "ray" of a "star" emanating from a
common center point, starting with the first variable at a  "0 degrees" position, and with
the  remaining variables being spaced evenly around the  full 360 degrees of  a circle as
in trigonometry.   For the four trihalomethanes, a four-ray star results, with each  ray of
the  star at 0. 90.  180, and  270 degrees (right angles),  respectively.  For the general
case of "n" variables, n  rays  of the star will be plotted,  each separated by an angle of
360/n degrees.

Chapter  5  of this  report presents the actual  application of star symbol plots  to  this
study,  at which point this concept will be made more discernible.
                                      4-6

-------
   EXAMPLE GUIDE TO STAR PLOTS
CHBr3
              CHC13
CHBrC12
             CHBr2Cl
             FIGURE 4-5

-------
Section 5
Baseline Sampling Results
and Discussion

-------
                                 SECTION 5

            BASELINE SAMPLING RESULTS AND DISCUSSION


In this section,  results of the quarterly baseline sampling are presented and discussed.
Methods of data analysis and presentation were described in Section 4.  Table 5-1 lists
the abbreviations used in the figures in this section.

OVERVIEW OF BASELINE DATA

Figures 5-1  through  5-10 present  an overview of the results of the  baseline sampling
program for all  four quarters combined.  Table 5-2 shows the baseline median  DBF
values for each quarter as well as for all four sampling quarters combined.  It should be
noted that  a class total median value  is not the sum  of the  medians of individual
compounds, but  rather  the median value of the sum of  all the compounds within that
class.  To illustrate,  the median for a compound (chloroform,  for example) during  a
sampling quarter was the median of the 35 measured  values of chloroform for tnat
quarter.  For  each utility, the sum  of the  individual  compounds  within  each  class
(THMs. for example)  was computed,  and then the median of these 35 sums was
determined.  Finally, the sum of the individual halogenated DBF compounds measured
in this study (XDBPsum) was  computed, and then the median  of these 35 sums was
determined.

In examining  these  data, it  should be noted that  they  represent  clearwell effluent
samples.  Distribution system  sampling, which was performed for some of the process
modification studies  described  in  Section 6 of this  report, indicated that  some DBFs
(such as THMs)  increased in the chlorinated distribution systems  of some utilities, while
the  same  compounds  remained  stable  in chloraminated distribution systems.   In
addition, it  is important to note that  the disinfection  practices of some of  the 35
participating utilities, such as the use of chloramines,  were  employed  to meet the
current TTHM MCL of 0.10 mg/L, and not to meet the requirements of the proposed
Surface Water Treatment Rule (SWTR).  Thus, DBF levels at some utilities would most
likely be different if their current disinfection practices required modification in order to
meet proposed concentration-time (CT) requirements of the SWTR.

Figure 5-1  is a summary of the four-quarter median concentrations of each DBF class
(THMs, HANs,  HKs.  HAAs  and ALDs) and  the  miscellaneous compounds.   The
median value of  THMs  was 36 //g/L, and the median value of HAAs  was 17 j/g/L.  On
a weight basis,  ALDs were the next most  prevalent compound,  with a total median
concentration of 5.7 //g/L.

Figure 5-2  shows the contribution  of each of the median class totals  and miscellaneous
compounds to the total  concentration of DBFs measured  in this study.  On a weight
basis. THMs were the  largest  class of DBFs detected in this study, comprising 54.5
percent of  the  total measured DBFs.   The  second   largest  fraction  was  HAAs,
comprising  25.4 percent of the total.  The data indicate that the median level of THMs
was  approximately twice that  of HAAs.   Figure 5-2 also shows that the third largest
fraction detected was the ALDs (formaldehyde and acetaldehyde), which comprised 8.5
percent of  the  measured  DBFs.   These two  low  molecular weight aldehydes  were

                                     5-1

-------
                                TABLE 5-1

               LIST OF ABBREVIATIONS USED IN DBF STUDY
ABBREVIATION1
            DEFINITION
Disinfection By-Products
DBF
XDBPMItn

Haloacetic acids

HAA
MCAA
DCAA
TCAA
MBAA
DBAA

Haloketones

HK
l.t-DCP
U.I-TCP

Haloacetonitriles

HAN
DCAN
TCAN
BCAN
DBAN

Aldehydes

ALD
FRM
ACETAL
Disinfection by-product
Sum of measured halogenated disinfection by-products
Haloacetic Acid
Monochloroacetic Acid
Dichloroacetic Acid
Trichloroacetic Acid
Monobromoacetic Acid
Dibromoacetic Acid
Haloketone
1,1 -Dichloropropanone
1,1,1-Trichloropropanone
Haloacetonitrile
Dichloroacetonitrile
Trichloroacetonitrile
Bromochloroacetonitrile
Dibromoacetonitrile
Aldehydes
Formaldeyde
Acetaldehyde

-------
                                 TABLE 5-1

                        List of Abbreviations, Continued
ABBREVIATION1
            DEFINITION
Chlorophenols

CP
DCP
TCP
PCP
Chlorophenol
2,4-DichIorophenol
2,4,6-Trichlorophenol
Pentachlorophenol
Other Disinfection By-Products
CH
CHP
CNC1

Others

PI
FI
FE
CE
MRL
NA
THM
TOC
TOX
UV-254
Chloral Hydrate
Chloropicrin
Cyanogen Chloride
Plant Influent
Filter Influent
Filter Effluent
Clearwell Effluent
Minimum Reporting Level
Not analyzed
Trihalomethane
Total Organic Carbon
Total Organic Halogen
Ultraviolet light absorbance at 254 nanometers
'Disinfection by-products not listed are represented by their chemical formulas.

-------
                                   TABLE 5-2

           DISINFECTION BY-PRODUCTS IN DRINKING WATER

         SUMMARY OF BASELINE SAMPLING MEDIAN VALUES
                               1st         2nd        3rd         4th        All
Disinfection By-Products       Quarter    Quarter    Quarter     Quarter    Quarters
       (//g/L)                 (Spring)   (Summer)     (Fall)      (Winter)   Combined


Trihalomethanes

   Chloroform
   Bromodichloromethane
   Dihromochloromethanc
   Bromoform

   Total Trihalomethanes       34         44          40         30          36

Haloacetonitriles

   Trichloroacetonitrile         < 0.012     <0.012
   Dichloroacetonitrilc          1.2         1.1
   Bromochloroacetonitrile       0.50        0.58
   Dibromoacetonitrile          0.54        0.48

   Total Haloacetonitriles        2.8        2.5         3.5         4.0         3.3

Haloketones

   I.l-Dichloropropanone        0.52        0.46        0.52        0.55        0.52
   1,1,1-Trichloropropanone       0.80        0.35        Q.60        0.66        Q.60

   Total Haloketones            1.4        0.94        1.0         1.8         1.2

Haloacetic acids

   Monochloroacetic acid      < 1.0
   Dichloroacetic acid           7.3
   Trichloroacetic acid          5.8
   Monobromoacetic acid      <0.5
   Dibromoacetic acid           0.9

   Total Haloacetic acids        18          20         21          13          17

-------
                                        Table 5-2

                     Disinfection By-Products In Drinking Water
                        Summary of Median Values, Continued

Aldehydes
Formaldehyde
Acetaldehyde
1st
Quarter
(Spring)

NA
NA
2nd
Quarter
(Summer)

5.1
2.7
3rd
Quarter
(Fall)

3.5
2.6
4th
Quarter
(Winter)

2.0
1.8
All
Quarters
Combined

3.6
2.2
     Total Aldehydes

  Miscellaneous
NA
6.9
5.5
4.2
5.7
Chloropicrin
Chloral hydrate
Cyanogen chloride
2,4,6-Trichlorophenol
Halogenated DBPsum
Total Organic Halide
Plant Influent Characteristics
Total Organic Carbon, mg/L
Ultraviolet absorbance, cm"1
Chloride, mg/L
Bromide, mg/L
0.16
1.8
0.45
<0.3
64
150

NA
NA
NA
NA
0.12
3.0
0.60
<0.4
82
180

2.9
0.11
28
0.07
0.10
2.2
0.65
<0.4
72
170

2.9
0.11
32
0.10
0.10
1.7
0.80
<0.4
58
175

3.2
0.13
23
0.07
0.12
2.1
0.60
<0.4
70
170

3.0
0.11
29
0.08
NA = Not Analyzed

Note (I):  Total class median values are not the sum of the medians of the individual compounds, but
          rather the medians of the sums of the compounds within that class.

Note (2):  The halogenated  DBPsum median  values are not the  sum of the class medians for all
          utilities,  but rather the medians of the halogenated DBPsum values for all utilities.  This
          value is only the sum of halogenated DBFs measured in this study.

-------
I«


5
»
[TJ
B3
         Disinfection  By-Product  Concentration
                  by  DBP  Class     Four Quarters
7O


60


50


4O

30

20

10


o

THM TnhalomBthnnes
HAN Hotoocetonitriies
HK Hatohetonw











-







HAA Holoocetlc Acids
ALO AldoMyOa*
CHP Ctitoroptcrtn
CH Chloral Hydrate
CNCl Cyanogen CNortd»
TCP TrlcNoroofwool

>^/r,



w

Y/yfiffl/fc
mm.

7W. ,«-«. «— — ^
" '



PP
^^^^W
^Ui

            THM  HAN  HK  HAA  ALD   CHP   CH   CNCl  TCP


                               DBP  Class

                             FIGURE 3-1
     Percent  of  Sum  of   DBP  Class  Medians
               By  DBP Class     Four  Quarters
                 HAA 25.4%
                                         ALD 8.5%
                      THM 54.5%
                                             MISC 5,O%
                                               HAN 4.9%

                                               HK  1.8%
THVI   TrifiolometnanM
HAN   Haloacatontril**
HK    Haloketonea
HAA   Haloacetic Acid*
ALD   Aldafiydas
 MISC:
CHP   CrKoropic/in
CH    CMoral HyOrata
CNCl   Cyanogan Cttorida
TCP
                               FIGURE 5-2

-------
                    Baseline Sampling Results and Discussion


 initially discovered  as  by-products  of  ozonation  (Glaze  et al.,  1989a; Yamada and
 Somiya, 1989), and they also appear to  be by-products of chlorination.

 A  breakdown of the class medians of the non-THM DBFs measured  in this study is
 shown  in Figure 5-3 on a weight basis.  HAAs are by far the largest contributor to non-
 THM DBFs, comprising 56.8 percent of the total measured compounds.  The two next
 largest fractions are ALDs and HANs, comprising  19.0 and 11.0 percent of the total,
 respectively.  Chloral hydrate was the next largest fraction, comprising 6.9 percent  of
 the total measured non-THM compounds on a weight basis.

 Figure 5-4 shows the four-quarter median concentrations of the four THM compounds.
 Additionally,  Figure 5-5  shows the same data as a percentage of the total level  of
 THMs.  These figures indicate that chloroform was detected at the highest levels for the
 THM compounds, comprising 56.2 percent of the total. The four quarter median level
 of chloroform  was  14 />g/L.  Bromodichloromethane  and dibromochloromethane had
 median levels of 6.6 and 3.6 /vg/L, respectively.  The  median  level of  bromoform was
 0.57 fjg/L. representing 2.3 percent of the total THMs.
 Figure 5-6 presents the median HAA concentrations by compound for the four sampling
 quarters.   Trichloroacetic acid (TCAA) and dichloroacetic acid (DCAA) were the two
 HAAs detected in the highest concentrations, with median levels of 5.5 and  6.4 /i/g/L,
 respectively.  These results are consistent with those of other researchers, which have
 shown that the aqueous  chlorination  of  humic and fulvic  acids yielded TCAA and
 DCAA as  the major  chlorinated by-products (Quimby  et al.,  1980; Christman  et al.,
 1983; Miller and Uden,  1983; DeLeer et al.,  1985). Dibromoacetic acid (DBA A) was
 detected   at a  four-quarter  median  concentration  of  I.I   jjg/L.   The  median
 concentrations  of both monochloroacetic  acid  (MCAA)  and  monobromoacetic  acid
 (MBAA) were below their minimum reporting levels of 1.0 and 0.5 >ug/L, respectively.

 Four-quarter median concentrations of the HKs are presented  in Figure 5-7.  Both of
 these compounds had  median levels less than 1 /wg/L.

 Median  HAN  concentrations for  the  baseline  sampling  are shown in Figure  5-8.
 Dichloroacetonitrile was detected at the highest concentration among the HANs,  with a
 median  level of  1.2 //g/L.   Bromochloroacetonitrile (BCAN)  and  dibromoacetonitrile
 (DBAN)  had  median levels  of  0.57 and 0.50  /ug/L, respectively.   The median
 concentration of trichloroacetonitrile (TCAN) was below its minimum reporting level of
 0.029
Figure  5-9  presents  the  median  levels  of  cyanogen  chloride,  chloral  hydrate,
chloropicrin  and  2,4,6-trichlorophenol (TCP).    Chloral  hydrate  had the  highest
concentration  of these miscellaneous compounds (2.1 /yg/L), followed by  cyanogen
chloride  (0.60 //g/L).   The median concentration of  chloropicrin  was 0.12  yug/L.
During the first sampling quarter, TCP was  detected at low levels at  four of the 22
utilities sampled for this compound, yet it was not detected in any samples collected in
subsequent sampling  quarters.   Pentachlorophenol analyses  were conducted for  22
utilities during the first sampling quarter, but was not detected in any of the samples.
Analyses were also conducted  for 2,4-dichlorophenol at 12  utilities during  the first
sampling quarter.  This compound was only detected at  one utility, at a concentration
equal to the detection  limit (2 //g/L).

                                     5-2

-------
        Percent  of  Sum  of  Non-THM  DBP
         Class  Medians by  Class    Four Quarters
               ALD  19.0%
                                    HAN  1 1.O%
                   HAA 56.8%
                                          CH 6.9%
                                           HK 3.9%
                                            CNCI  2.O%
                                            CHP O.4%
                                    THM
                                    HAN
                                    HK
                                    HAA   H»IO*C»tlC Acids
                                    ALD   AI
-------
Percent  of  Gum  of  THM  Compound  Medians
              by  Compound     Four  Quarters
             CHBrCl2 26.9%
                                          CHBr2Cl  14.6%
                                             CHBr3 2.3%
                 CHCI3
                             RGUR65-5
              Haloacetic  Acid  Concentrations
               by  HAA  Compound     Four  Quarters
01
0)

0
I i
•1
I
a
n
Monochloroacetic Acid
Dichloroacetic Acid
Trichloroacetic Acid
Monobromoacetic Acid
Dibromoacetic Acid
MCA A
DCAA
TCAA
MBAA
DBAA
                                         25th parcentile value below
                                         25th parcantil* and nr»Oan
                                         values below
             MCAA     DCAA     TCAA


                             HAA Compound

                               FIGURE 5-6
                                                     DBAA

-------
 1

 §
 *-*
 to

 I
 n
 •6
         O
                    Haloketone  Concentrations

                  by HK  Compound  -  Four  Quarters
                1.1 -DCP     1.1 -Dichloropropanone

                1.1.1 -TCP     1,1.1 -Trichloropropanone
                     1.1-DCP
                                              1.1.1-TCP
                               HK Compound



                                FIGURE 5-7
                Haloacetonitnle  Concentrations

                 by  HAN  Compound - Four  Quarters
s
*-
05
o


<



(0
        O
               •*•*••*•
               FCAN
                                       TCAN   Tnchloroacetonitrile


                                       DCXXN   Dichloroacetonitnle


                                       BCAN   Bromochlofoacetonitrile

                                       DBAN   Oibromoacetonitnle




                                       *** 25th pefcentile. median, and 75th

                                          percentile values below
                           DCAN
                                        BCAN
                              HAN  Compound



                                FIGURE 5-8
DBAN

-------
-
5
o
g
U
a
c
ro
            Miscellaneous  DBP  Concentrations
                  by  Compound  - Four Quarters
                                  CNCI   Cyanogen Chloride
                                  CH    Chloral Hydrate
                                  CHP   Chloropicrin
                                  TCP   Tnchlorophenol
                                   *** 25th parcentue, median, and 75th
                                      percentile values below
1  -
        0
               CNCI
                                       CHP
                            MlSC Compound

                              FIGURE 5-9
                                           TCP
 0)
U
Q
 c
 5
 TD
                   Aldehyde  Concentrations
                   by Compound  -  Four  Quarters
                                   FRM =; Formaldehyde
                                   ACETAL =  Acetaldehyde
                                   •x-  25th percentile  value
                                      below MRL
                  Formaldehyde
                                          Acetaldehyde
                             ALD  Compound

                              FIGURE 5-10

-------
                    Baseline Sampling Results and Discussion


Median concentrations  of the ALDs are shown in Figure  5-10.   The four-quarter
median level of formaldehyde was 3.6 /t/g/L and the median level of acetaldehyde was
2.2/;g/L.

STAR PLOT ANALYSES

One of the stated objectives of  this study was  to  "determine  the  seasonal nature of
DBFs as a function  of temperature, total organic carbon,  pH, and other water quality
parameters; and to  determine the  effects  of changes in treatment processes and/or
disinfectants on the production of  DBFs".   Under  ideal  conditions, these objectives
could  be  approached  and  properly  treated   as  a  problem  in  formal statistical
experimental design.  The ideal conditions  would include a very  large  population of
United States utilities staffed and equipped for DBF sampling, with the investigator free
to perform any adjustments necessary at each plant to implement  strict experimental
controls in a  classical statistical level-response context.   Furthermore, approximately
equal numbers of utilities would be using  the same raw  water  source types, influent
water quality levels,  and oxidation/disinfection schemes. However, ideal conditions are
not  possible in field  studies  such  as this  one.   Rather,  the  group of  35 utilities
represented an  unbalanced mixture  of  differing source  waters,  raw water qualities,
treatments  and disinfection schemes.  The following paragraphs illustrate these  points
with the aid of star symbol plots.

Figure 5-11 shows the guide to the star plots used in this section. Using  data from the
summer sampling  quarter for  each  plant  in the study,  the raw water characteristics of
TOC and UV-254 absorbance were  plotted at 0  and  180 degrees to form what can be
termed the "organics axis", while raw water bromide and chloride levels were plotted at
90  and 270 degrees to  form the  "inorganics  axis".   It should  be noted  that the
magnitude  of each ray of the star plots is normalized to the utility with the highest level
of the parameter under consideration. Utility 10 had the highest chloride and  bromide
levels (640 mg/L and 3.00 mg/L, respectively) and Utility  29 had the highest TOC and
UV-254 levels (19.04 mg/L and 0.697 cnv1,  respectively).  However, because the levels
of these parameters were so  high  at these  two utilities,  relative differences  between
levels of these parameters for the other  utilities  tended to be minimized.  Thus,  these
two  utilities were  excluded from all star plots and the plots were normalized to the
utilities with the next highest levels  of these influent parameters, Utility 24 (115 mg/L
of chloride, 0.54 mg/L of bromide) and Utility 21 (10.57 mg/L of TOC, and UV-254
equal to 0.358 cnr1).

Figure 5-12 shows the star plots for the 33 utilities (excluding Utilities 10  and  29),
labeled by  utility number.  From the relative sizes and  various shapes of the "stars", it
is  immediately apparent that the raw water quality characteristics varied considerably
among the  participating utilities.

Figure 5-13 presents the  same star  plots, but with the utilities sorted by source water
type, each  source water type on a separate page  of the figure.  This plot  clearly begins
to illustrate the problem of trying to ascribe differences in DBF production to discrete
and  unrelated causal factors, in this case source water origin.   The star plots for the
groundwater sources are a good illustration of this point. The groundwater "stars" vary
widely in both  size and  shape;  and thus,  in the relative influences of bromide and
organic carbon, both of which have a major impact on the  level and speciation  of DBFs

                                       5-3

-------
       GUIDE TO STAR SYMBOL PLOTS
                             Bromide
UV Absorbance
TOC
                            Chloride
             Each "ray" of the star emanating from the center point
             represents the relative magnitude of that parameter
                     FIGURE 5-11

-------
         Plant Influent By Utility ID Number
                   Summer Quarter
                6
9
                11
13
                 FIGURE 5-12

-------
            Plant Influent By Utility ID Number
                       Summer Quarter
14
                 16
                                  16
                                                        17
18
                 19
                                   29
                                                        21
22
                 23
                                  24
25
                   FIGURE 5-12  (Continued)

-------
         Plant Influent By Utility ID Number
                    Summer Quarter
26
27
                                   28
31
                                   33
                                                    34
36
                 FIGURE 5-12 (Continued)

-------
        Plant Influent by Source Water Type
              Lake/Reservoir Sources
                   Summer Quarter
                                                 6
                                15
                                                 16
17
                18
                                                 26
                     FIGURE 5-13

-------
Plant Influent by Source Water Type
      Lake/Reservoir Sources
           Summer Quarter
        33
                        34
                                         36
         FIGURE 5-13 (Continued)

-------
       Plant Influent by Source Water Type
             Flowing Stream Sources
                   Summer Quarter
                                 12
                                                  13
14
                 19
                                                  28
                 31
              FIGURE 5-13 (Continued)

-------
       Plant Influent by Source Water Type
               Groundwater Sources
                   Summer Quarter
                                                22
23
                24
                                 32
                FIGURE 5-13 (Continued)

-------
                    Baseline Sampling Results and Discussion


 produced upon  oxidation/disinfection of that water.  For example. Utility 24 has a high
 level of bromide, while  the level  of  TOC at this utility is relatively low.  Utility 21
 treated a colored groundwater with a high concentration of TOC.  Finally, Utility  I's
 source  water was a groundwater low  in  both  inorganic and organic constituents. The
 plots from the flowing stream and lake/reservoir illustrate similar conditions.

 The  next  set of star  plots (Figure  5-14)  depicts  each utility  again,  but sorted  by
 treatment type.  Some  overall trends are apparent  in  these  plots.   For instance,  the
 direct filtration  and disinfection only  utilities  generally have relatively small  "stars",
 indicating  the presence  of relatively low levels of organics and inorganics, and hence
 the ability to employ less-extensive treatment systems for these source waters.  The fact
 that  the characteristics  of the source water  dictate  to a large degree the type of
 treatment required for that water serves to confound  an attempt to ascribe differences in
 clearwell effluent DBF levels  simply to treatment type.   It may  be misleading to
 consider concentrations of TTHMs, for instance, produced  by  plants  utilizing direct
 filtration or disinfection  only  and conclude  that THM  concentrations at these plants
 were caused solely by the treatment practices of the plants.

 Figure  5-15 presents utilities  labeled  by disinfection scheme.  In theory, this would
 seem to be the  most promising classification for identifying causal differences  in DBF
 production.   However, examination  of these star  plots reveals several  confounding
 factors.  For example, all  other parameters being equal, chlorine-only utilities would be
 expected to have higher TTHMs than  the chloraminating utilities.  However, the four-
 quarter     median     TTHM     value    was    actually     higher    for     the
 prechlorinating/postammoniating  utilities  (approximately  57  //g/L)   than  for   the
 chlorine-only  utilities (approximately  34 ^g/L).  (These findings will  be discussed in
 more detail later in  this section.)  This may be understood, in part, by considering  the
 widely  varying  raw water qualities of the  utilities  employing these two  disinfection
 scenarios,  as  shown  in  Figure  5-15.   While it appears that,  in  general,   the
 prechlorinating/postammoniatmg utilities may  have  a higher percentage  of relatively
 large "stars"  compared to the chlorinating utilities, the differences in these influent
 parameters  between  the  two disinfection schemes were  not  found to  be statistically
 significant  (as  will  be  discussed  later  in  this  section).   However,  it  may  be  the
 combination of high  organics  and  bromide levels  that  required four  of  the  10
 prechlorinating/postammoniating utilities to  use postammoniation for  THM  control.
 No overall trend in raw water quality  characteristics  is observed in the star plots of any
 of  the  disinfection  schemes.   Thus, characteristics such  as reactivity of precursor
 material, chlorine dose and contact time,  pH, water temperature, and precursor removal
 within the  treatment processes must also play  a role in the formation and speciation of
 DBFs.  The star plots presented here illustrate some of the difficulties arising from
 attempts to find discrete,  unrelated causal factors for  DBF production.

 SEASONAL VARIATION

 Figures 5-16 through 5-34 show the variation in the water quality parameters and DBFs
 measured in this study as a function  of sampling season.  The data are presented as
 notched box-and-whisker  plots, showing the  median, interquartile range, minimum and
 maximum values, outliers, and the 95  percent confidence intervals of the medians. The
number of samples (n) is also shown on each plot.   A more detailed discussion of this
type of data presentation was included  in Section 4 of this report.

                                       5-4

-------
Plant Influent By Treatment Type
 Conventional Utilities, Summer Quarter
         11
                          12
13
           FIGURE 5-14

-------
         Plant Influent By Treatment Type
          Conventional Utilities, Summer Quarter
14
               FIGURE 5-14 (Continued)

-------
          Plant Influent By Treatment Type
             Softening Utilities, Summer Quarter
21
                  22
                                   23
                                                     24
26
                  27
                                   28
31
                 FIGURE 5-14 (Continued)

-------
         Plant Influent By Treatment Type
         Direct Filtration Utilities, Summer Quarter
                 16
                                   17
                                                     18
19
                   FIGURE 5-14 (Continued)

-------
          Plant Influent By Treatment Type
          Disinfection-Only Utilities, Summer Quarter
32
                  33
34
                                                       35
                 FIGURE 5-14 (Continued)

-------
      Plant Influent By Disinfection Scheme
            Chlorine Only, Summer Quarter
                                  6
11
                 13
                                  15
                                                   16
17
                                  22
                                                   23
                    FIGURE 5-15

-------
         Plant Influent By Disinfection Scheme
               Chlorine Only, Summer Quarter
                 26
                                  31
                                                    33
34
                 36
                  FIGURE 5-15 (Continued)

-------
        Plant Influent By Disinfection Scheme
            Chlorine /Ammonia, Summer Quarter
12
                 14
                                                    27
28
                  30
                   FIGURE 5-15 (Continued)

-------
Plant Influent By Disinfection Scheme
            Summer Quarter
           Chloramines Only
                              18
             Ozone/Chlorine
      19
                             32
            Ozone/Chloramines
          FIGURE 5-15 (Continued)

-------
                    Baseline Sampling Results and Discussion


Influent Water Quality

The seasonal effects on influent water quality parameters measured in this study (TOC,
UV-254. chloride and bromide) are shown  in Figures 5-16 through 5-19.  Data from
only the summer, fall and winter quarters  are presented since these influent analyses
were not instituted until the second sampling quarter.

As  shown in Figure 5-16. median levels of TOC did not change appreciably by season,
ranging from 2,9 to 3.2 mg/L.  The three-quarter median influent TOC level was  3.0
mg/L.   There was not a significant  difference  in the medians of any  two sets  of
quarterly influent TOC data.  In other words, the 95 percent confidence interval for the
spring  quarter median was compared to that  of  the summer  quarter, then to the  fall
quarter.  The overlapping 95 percent confidence intervals indicate that the difference in
the median  TOC levels of any two quarters is not statistically  significant.   The "folded
over" box for the fall quarter data shown  in  Figure 5-16  results from  the 95 percent
confidence interval  of the median extending lower than the 25th percentile value.  The
notch width corresponds to the 95  percent  confidence interval of the median, and  the
horizontal line across the box corresponds to the 25th percentile value.

Figure 5-17, a plot  of UV-254 absorbance as a function of sampling quarter, indicates a
trend similar to the  influent TOC  levels.   Although the winter median  (0.13  cm1)
appears slightly  higher than the summer or fall medians (0.11 cnr1 for both quarters),
the  medians of the  summer and fall seasons are  within the 95 percent confidence
interval for  the median of the winter quarter,  indicating no statistical difference in  the
medians of any two seasons.

A plot of influent chloride levels by season is shown in Figure 5-18.  Again, there is
little seasonal variation indicated in  this figure.   The highest  median influent chloride
level was  32 mg/L, occurring  in the fall season.  The three-quarter median chloride
level was 29 mg/L.   Figure 5-19 shows the seasonal variation in influent bromide levels
on  a seasonal basis.  Quarterly  medians ranged from  0.07 to 0.10 mg/L  of bromide.
with a three-quarter  median  of 0.08 mg/L.  Note  the presence of  one outlying data
point in each quarter that occurs at substantially higher concentrations than  the 75th
percentiles and other outliers.  These outlying data points represent  Utility 10, which
had extremely high  influent bromide levels.  Bromide and its impact on  DBF formation
and speciation will be discussed in detail later in this section.

Classes of DBF Compounds

Figure 5-20 is  a  plot of the seasonal variation of  the  sum of halogenated DBFs
(XDBP  ) measured in this  study.  The median of the summer quarter  (82  //g/L) is
higher  than  the median of the winter quarter (58 */g/L); however, this difference is not
statistically  significant since the 95 percent confidence intervals  about the medians  of
both quarters overlap. The median values of XDBPS11I11 for the spring and  fall quarters
were 64 and 72 //g/L, respectively.

Figure  5-21  is a plot of TTHMs by sampling quarter.  As would  be expected based on
seasonal  temperature differences,  the highest  median TTHM level occurred in the
summer season and  the next highest in the fall.  For many utilities in  California and the
South,  the fall season can be  almost as warm as  the summer.   The  lowest  median

                                      5-5

-------
Influent Total Organic Carbon
         Bg Quartar
Influent UU Ab>orbancB
      By Quartar
am

J >•
a
6

C
0
SI *•
L
U
u
•H
c
a
o
— i
•
0 «
H


-1 	 1 	 1 	 1 	 1-
-
-
•
-
B

_
.
:
.


-
LJ LJ y^
> — < > — C / N
i i
...
9
u
\
N^ t.B
E
C

U)
01
-u
« «-*
fll
u
u
0
XI *-2

-------
Influent  Chloride By Quarter
                                                   Influent Bromide By  Quarter


7BI

67*
J
a
0 38*
TJ
•H
L
Q

L
0
19*


-i 	 1 	 1 	 ~ 	 1 n


•
•


-









i i i

4

3

J
a
81 *
•Q
•H
E
0
L
m


.
-r i i

:
-
•








•
-
"
iji i

i 	 i 	 — i 	 1 	 L— '
L_l 	 J 	 	 — i 	 ' 	 ' — •
SUV1ER FBLL WINTER
SUMMER POLL UIHTER 5UmEK
n= 35 35 35
n= 35 35 35
Quarter Quarter
            RGURE5-18
                                                               FIGURE 5-19

-------
                  XDBPsum  by Quarter
                                                      Trihalomethanei


                                                         Bg  Quarter
_J

n
D
E

N
a.
DO
a
x
                _L
_L
_L
                                                           -I

                                                           a
                                   •
                                   •

                                   r
                                   0
                                   -t

                                   i


                                   •H
                                   L
                                                                                   T
                                                                   T
                                                                                                    _L
               SPBIMO    SUMMER     POLL


                35        35       35


                         Quarter



                          FIGURE 5-20
                UINTER



                 35
                                         SPRINO



                                       n- 35
35       35

Quarter


FIGURE 5-21
               UINTEB




                 35

-------
                    Baseline Sampling Results and Discussion


TTHM level occurred in the winter. The impact of water temperature on the formation
of THMs and other DBFs will be discussed in detail  later in this section.  Figure 5-22
is a plot  of median chloroform levels by sampling  quarter.  There  are  no significant
differences between the medians of any two quarters  and only three outliers appear on
the  plot.    However,  plots  of  bromodichloromethane,  dibromochloromethane  and
bromoform  (Figures 5-23, 5-24 and 5-25, respectively)  indicate progressively  greater
numbers of outliers, although  there are no statistically significant  differences between
the medians of any two quarters for these compounds.   The increasing number of
outliers is most  likely  due to the presence  of high levels of brominated  THMs in
utilities with high influent bromide concentrations.  Note the extremely high levels of
bromoform  (over 50 /ug/L) of several of the  outliers in  Figure 5-25.  The  impact of
influent bromide levels on the production of  brominated DBFs is discussed in detail
later in this  section.

Following the  same trend as  median TTHM levels, median total  HAA levels  were
higher in summer (20 pg/L) and  fall (21 /ug/L)  than in winter  and spring (13  and 18
/t/g/L.  respectively), as  shown in Figure 5-26.   However,  there was  no statistical
difference between the medians  of any  two  quarters.   Figures 5-27 and 5-28 show
DCAA and  DBAA concentrations on a quarterly basis. Neither of these plots indicate a
significant difference between the medians of any two quarters.  However, following the
same trend  observed in the  plots of chloroform and bromoform,  the plot  of  DBAA
shows  many more outliers than the plot of DCAA.

In contrast  to  the  trend of increasing median concentrations with  increasing seasonal
temperatures seen in the  TTHM  and HAA plots. Figure  5-29 shows that the median
level  of total  HANs  was highest  in  winter and  lowest in summer,  although the
differences were not statistically significant. This same trend is observed  in the  plot of
quarterly  median HK  concentrations.  Figure  5-30.   The role of  HANs  and  HKs as
reactive  intermediates  rather  than stable  endproducts  of chlorination  reactions is
discussed  later  in this section.

Figure 5-31  illustrates the seasonal variation  in  aldehyde concentrations. Only three
quarters of data are presented since the ALD analysis was  not instituted until the  second
sampling quarter.  The plot indicates that the level of ALDs in the summer quarter (6.9
/ug/L) was higher than winter median (4.2 yug/L), but  the difference  was not statistically
significant at a  95 percent confidence level.

The  chloropicrin levels shown in  Figure 5-32 indicate that  there was  little seasonal
variation,  with  all four  quarterly medians falling within the range  0.10 to 0.16 //g/L.
Figure 5-33 indicates the same  lack of  significant  variation in  quarterly  cyanogen
chloride medians.   The  median concentrations of cyanogen chloride were within the
range 0.45  to 0.80 /yg/L for the four  sampling quarters.  The plot of chloral hydrate
concentrations shows no significant difference between the medians  of any two quarters
(Figure 5-34).

Although  Figures 5-20 through 5-34 do not  show  statistically significant differences
between the  median DBF concentrations  from season to season, seasonal variation in
DBF levels is demonstrated in Table 5-3.   In this table, DBF values for the 35  utilities
are presented for the 25th percentile, median and 75th percentile for the four sampling
quarters.  For TTHMs, the 25th  percentiles for the  summer and winter quarters were

                                      5-6

-------
                Chloroform  By Quarter
                                 Bramodichloramethani
                                       By  Quarter
a
E
a
o
a
u
                                                        \
                                                        a
                                                        ^
                  c
                  r
                  a
                  o
                  -H
                  U
                  •H
                  TJ
                  0
                  0
                  (D
                                                                                                I
              SPftlNO

               35
                        35       35
                        Quarter
UIMTER

 35
  SPRINO

n=   35
                                         35       35
                                         Quarter
WINTER

 35
                        FIGURE 5-22
                                                                                FIGURE 5-23

-------
Dibromochlorom«thani
     BU Quarter
                                                    Bromoform BU Quarter
••
J
\
n ••
11
•

*J
II
E «•
0
L
0
.-•
•^
0
0
E
0
L a.
a
•H
a

•

-T— 	 	 	 1 	 1 	 1 ' ^
-
_

m

,
_
-
" • * *

*
- .



- w 2 2 3
c 1
"

• •
S\
_J
0
^^

1 **
0
Q
E
Q
m

ai




i

n 	 -T 	 1 	 1 	 1 r-
-

•



h— • "
"
•

.
. .
1

.
=1= =±2 L2 = :
"l 1 1 _l 	 L
„_ SPRINQ «unrcn fMJ. UIHTER
SPHINQ ***** MU. UIKTEB
35 , 35
Quarter Quarter
        FIGURE 5-24
                                                            FIGURE 5-25

-------
                   Haloacetie Acid*


                      BU Quarter
 D
w
u
•H
u

-------
                Dibromoacetic Acid

                     Bg Quarter
                                   Haloacetonitrilee

                                       Bg Quarter
                              T
3
V

T)

U
u
•H
4J
II
U

a

o
              _L
                                      _L
                                                       D
                  •H
                  L
                  +J
                  •H

                  0
                  4J
                  •
                  U

                  0
                  H
                  II
                  z
                                                                             T
                                                T
             SPRING



              35
                      35       35
                       Quarter
UINTER


 35
  SPRINO


n=   35
           35      35

           Quarter
UINTER



 35
                       FIGURE 5-28
                                        FIGURE 5-29

-------
                     Haloketonea



                     Bg Quarter
                                                                        Aldehydes by  Quarter
O
0
+J
n
i
                               _L
                  j
                  \
                  n
                   in
                   o
                   •o
                   3
                   r
                   u
                   TJ
                                                                                              _L
              3PBINQ



            n= 35
       r«U-


35       35


Quarter




FIGURE 5-30
WINTER



 35
   SUMMER




n=   33
                                              35

                                          Quarter



                                          FIGURE 5-31
UINTEB





 33

-------
Chloropicrin By Quarter
                                              Cyanogen Chloride By  Quarter
a..

a


^
_J
\
a
3
^ l.B
C
•H
I.
U
•H
a.
0 »
i.
o

u

*.*

*

II iii
-
_


-










"

•
•
~

: zj M H M
i
"
13
^
J
\
a "
3
^

§|
.•0 .
L
0
^^
u .
c
ll
01
0
c «
o

2
*

-
_
-
-

—




_
:

_
•
-

•
.
-
•
; s i B 5
_J 	 1 	 1 	 1 	 1 	 L-
SPRING SUMMER PALI- UINTER
SPRING SUMMER P«U- UINTEB
n= 35 35 34 35 n= 20 35 34 35
Quarter Quarter
FIGURE 5-32 FIGURE 5-33

-------
Chloral Hydrate By Quarter
Influent Total Organic Carbon
         By  Tr«atm»nt

3*
a*

J
a
3 «
^
•
•
t
•D "
I

•H
9
i. 11
0
r
a
.

•
1 1 1 1 I
-
-

•

•~~ —
•
~
-
~
.



: :
™ •

	 _
; 0 H U i :
iii i

3B
J »•
\
n

c
0
J3 IB
•
0
U
.^
c
« It
a
L
O
•jj
4J
0 6
H

•
I I 1 I 1
•
•

•
.
'

~
.

1
— _

"
•

2' '_
_
*T S ^ :
i 11 i
SPRIHD SUMWR PMJ. WINTER CNU OFL 019 SPT
n- Ii8 15 12 30
n= 23 35 35 35 u- 16 5 A 10
Quarter Treatment
FIGURE 5-34 FIGURE 5-35

-------
                       TABLE 5-3

       COMPARISON OF SEASONAL DBF LEVELS
Quarter
  25th
Percentile
Median
  75th
Percentile
Spring
Summer
Fall
Winter
             Total Trihalomethanes (/ig/L)
 22
 28
 20
 15
 34
 44
 40
 30
 57
 76
 54
 48
              Sum of Haloacetic Acids Gt/g/L)
Spring
Summer
Fall
Winter
Spring
Summer
Fall
Winter
Summer
Fall
Winter
 10
 11
  9
 10
 17
 20
 21
 13
              Sum of Haloacetonitriles G/g/L)
   1.4
   1.6
   1.3
   1.3
  2.8
  2.5
  3.4
  4.1
                 Sum of Aldehydes G/g/L)
  4
  2
  2.3
  7
  5.5
  4
 42
 35
 3t
 26
  6.0
  7.0
  7.8
  6.0
  12
  9
  7

-------
                    Baseline Sampling Results and Discussion


28 and  15 ^g/L. respectively, indicating a decrease of almost 50 percent  from summer
to winter. The 75th percentile for TTHMs was 76 //g/L in the summer and 48 //g/L in
the winter,  indicating  a decrease  of almost  40 percent  from summer to winter.
Although the seasonal differences for HAAs,  HANs and ALDs were not  as substantial
as those observed for THMs, seasonal differences for these DBFs are apparent in the
table.   For  instance, the 75th percentile  for the sum of HAAs was 42 //g/L in the
spring,  and  decreased almost 40 percent to 26  /j g/L in the winter.   Furthermore, the
medians for HAAs were highest  in  the  summer and fall,  but the 75th percentile was
highest  in the spring.  From the data presented  in Table 5-3, it  is  clear that  seasonal
variations influenced levels of DBFs measured  over the sampling period.

Table 5-4 is a summary of clearwell effluent TTHMs and  influent  temperature,  TOC
and bromide levels  for all 35 utilities,  comparing data from the summer and winter
quarters.  This table  illustrates that seasonal variations in the reported parameters  were
different for  individual  utilities.    For  instance,  Utility  1 had  a  summer  water
temperature of  I8°C and a winter temperature of I4°C.  Bromide  and TOC levels at
Utility  I varied only slightly as well,  and this  lack of variability  is reflected in the
TTHM  levels (7.9 and  9.0 /ug/L  in the summer and winter quarters, respectively). At
Utility  20.   there  was essentially  no  seasonal  variation  in   TOC  and   bromide
concentrations, but water temperature varied  from 26 to 8°C from  summer to winter,
and TTHM  levels in the summer were significantly higher (76 and 24  //g/L in the
summer and winter  quarters,  respectively). At Utility  23,  the water  temperature was
similar  both quarters (10°C in the summer and 8°C in the winter); however, TOC and
bromide levels were  higher in the winter (4.49 mg/L and 0.58 mg/L, respectively) than
in the  summer (3.27  mg/L  and 0.44  mg/L,  respectively).   These  differences are
reflected in  the higher TTHMs in the winter versus the summer quarter (24 versus 3.8
/yg/L. respectively).   From the data presented in this  table,  it is apparent that  seasonal
variations in influent water quality and effluent DBFs do not exist uniformly for utilities
around  the nation, but rather for individual utilities.

VARIATION BY TREATMENT TYPE

Figure 5-35  shows the three-quarter influent TOC levels as a function of treatment type
(conventional, direct filtration, disinfection only,  and softening).  The  plot includes the
number of data points in each category (n) as well as the number of utilities within each
category (u).  There  is  no statistically significant difference between the median values
for conventional, direct filtration and disinfection only utilities. However, the softening
utilities  participating  in this study had a  significantly higher median  influent TOC than
either conventional or direct filtration utilities.  This is perhaps due  to the inclusion of
several utilities treating  highly-colored ground  and surface waters by softening.

Figure 5-35  is the only  plot of water quality parameters as a function of treatment type.
As discussed above  regarding the star plot analyses, levels of DBFs measured in this
limited  study of 35 utilities cannot be ascribed simply to discrete  causal factors such as
treatment type  and  source water type.   Thus,  plots of DBF levels as a function  of
treatment type are  relatively meaningless and  could be  misleading  if  taken out  of
context.   However,   some special  issues relating to DBF levels in  disinfection-only
utilities  are discussed later in this  section.
                                      5-7

-------
                            TABLE 5-4



   COMPARISON OF SEASONAL TTHMs AND INFLUENT WATER QUALITY
                  Summer Quarter
Winter Quarter
Utility
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
Effluent
TTHMs
0/g/L)
7.9
90
6.1
9.1
42
98
63
27
95
43
67
71
24
114
27
28
94
7.6
15
76
56
60
3.8
40
34
164
57
36
108
73
77
3.1
44
46
38
Influent Values
Temp.
18
30
28
23
30
24
21
20
31
28
27
23
19
22
19
15
30
17
20
26
25
6.6
10
22
NA
27
30
28
29
27
28
26
18
15
18
TOC
(mg/L)
0.75
2.4
3.2
2.8
5.4
4.5
2.8
1.6
3.6
4.6
2.1
2.6
1.1
2.6
1.3
2.7
3.0
1.8
1.8
7.4
11
5.5
3.3
0.60
7.1
5.2
2.9
2.4
19
3.6
3.0
2.6
3.2
2.6
3.1
Br
(mg/1)
0.07
0.12
0.02
0.27
0.12
0.06
0.15
0.02
0.06
3.0
0.07
0.41
1.0
3.5
0.01
0.05
0.01
0.02
0.10
0.04
0.18
0.02
0.44
0.54
0.22
0.19
0.33
0.09
0.16
0.10
0.05
0.07
0.01
0.01
0.01
Effluent
TTHMs
0/g/L)
9.0
60
6.1
12
32
33
30
12
40
47
28
85
26
76
22
36
29
0.7
5.9
24
48
38
24
43
9.1
98
36
16
259
66
57
0.8
25
53
20
Influent Values
Temp.
14
27
4.5
9.6
15
4.0
13
9.5
5.0
6.0
8.4
9.0
11
11
13
12
9.5
4.5
5.0
8.1
23
10
8.1
20
18
3.0
10
4.5
23
2.0
1.1
24
3.1
6.2
2.8
TOC
(mg/L)
0.96
2.5
3.2
2.7
5.5
4.9
2.7
1.7
4.5
5.3
3.0
2.8
1.5
3.9
1.4
2.7
3.2
1.9
1.5
6.9
11
4.0
4.5
0.74
6.9
3.7
3.8
2.6
17
20
3.2
2.5
3.5
3.3
3.5
Br
(mg/L)
0.08
0.12
0.02
0.47
0.13
0.07
0.07
0.02
0.10
2.8
0.06
0.79
0.01
0.32
0.01
0.07
0.01
0.03
0.04
0.03
0.17
0.02
0.58
0.44
0.14
0.68
0.35
0.04
0.21
0.11
0.05
0.06
0.01
0.03
0.01
NA = Not analyzed

-------
                   Baseline Sampling Results and Discussion
VARIATION BY SOURCE WATER TYPE

Figures 5-36 through 5-39 show the influent water quality parameters measured in this
study  as  a  function of source  water type  (flowing stream,  groundwater,  and
lake/reservoir).   The  105  TOC measurements performed for this study had a minimum
of 0.6 mg/L and a maximum of 19.9 mg/L.  As seen in Figure 5-36, there is very little
difference between the medians of any two source water types.  The highest TOC levels
occurring  in  this study were measured at Utility 29,  treating a highly-colored surface
water, and one anomalous reading  at Utility 30 in  the winter quarter.  The other two
influent TOC measurements at Utility 30 were less than 5 mg/L.  The 25th percentile
TOC level  for  the  groundwater sources was  the  lowest;  however, the  presence  of
colored groundwaters resulted in a median TOC value statistically comparable to that of
the other two source water types.

Influent  UV-254  absorbance by  source water  type  is illustrated  in Figure 5-37.
Although  the median UV-254 absorbance level is higher for the groundwater utilities
than  the  flowing  stream  or lake/reservoir utilities, the difference  is  not  statistically
significant.   The  outlier  points shown for  the flowing stream utilities are the three
quarterly  readings from Utility 29, reflecting  the high organic content of this utility's
source water.

The influent chloride levels plotted by source water  type in Figure 5-38 show very little
variability in chloride concentrations from source to source.   The influent  bromide data
plotted in Figure  5-39 indicate the same lack of variation  between the  source water
types. The  highest  outlier  points  on  both  plots represent Utility  10, treating  a
lake/reservoir source with extremely high chloride and bromide levels.   No  plots of
DBF concentrations  by   source water  type  are included  for the  reasons discussed
previously with  respect to the  star plot analyses and the  variation of DBF  data  by
treatment  type.

VARIATION BY DISINFECTION SCHEME

Figures 5-40 through 5-54 illustrate influent parameters and DBF levels  measured in
this study as a function of disinfection scheme (chlorine only, "C12"; prechlorination
and postammoniation. "CI2NH3"; and  chloramination, "NH2C1").    Results for  the
preozonation/postchlorination  and  preozonation/postchloramination  utilities  are  not
presented  here because the  very small  number of data points precluded  meaningful
observations about those data.  However, aldehyde data from the ozonating utilities will
be  discussed later  in  Section 5,  and further discussion  will be included  with the results
of the ozonation treatment studies in Section 6.

Influent Parameters

Figure 5-40 shows the overall influent TOC levels as a function of disinfection  scheme.
The medians show  very  little  variation  among utilities using the various disinfection
practices.  Although Figure 5-41. a plot of influent  UV-254  absorbance as a function of
disinfection  scheme, does  not show statistically significant differences between the three
categories, this figure does indicate that  prechlorinating/postammoniating utilities had
higher influent UV-254 levels than the  chlorinating or chloraminating utilities.  One

                                      5-8

-------
          Influ«nt Total Organic Carbon

                     By SourcB
                                                               Influ«nt UU-264 Abmorb.nci

                                                                         By Source
 SN  ••
 _l

 a
 E
 0
JJ


o

u
•H
c

a
L
O
+J
0
                                                    u
                                                    E
                                                    ID
                                                    01
II
U

a

o

a

-------
             Influent Chloride by Saurci
                                            Influent Bromide  by Source
a
•a
•H
L
a
•H
o
                                 \
                                 a
                                 E
                                 w
                                 •H
                                 E
                                 O
                                 DO
                                                                                   /IN
                  PS
                  33
                  11
   21
    7
Sourci
LR
51
17
33
ll
   ou
   21
    7
Source
LR
51
17
                         FIGURE 5-38
                                                                                  FIGURE 5-39

-------
Influent Total Organic Carbon

    By Disinfection Schema
Influent  UU Abaorbanci

By Disinfection Schemi

M

s-\
—1 >•
\
a
E
W

C
0
.Q IB
i.
•
0

U
•H
C
« '•
a
L.
O
	 i
•H
•
.u
0 B
^~

•

1 III
-
-
•
• —

"
.

.


- ' _

'
_
*
•

i
— _

~
-



i 	 ' 	 1 I 	 	 1 1 1
^_j rj n
/ — \ \ / v/
r i ^^ m
1
i i i i i

•.•
• -T
^%
E
U
\
v/ •.•

E
C

Tj I.B
ID
(VI

4J
• *.4

•
U
c
4 *.3
o
L
0
D i.a

-------
                    Baseline Sampling Results and Discussion


possible explanation for these results is that utilities with high UV-254 absorhance most
likely have high color  levels and require  free chlorine contact time  for color removal.
However, the high UV-254 levels also indicate the presence of dissolved organic matter
which could  function  as  THM precursors.  Thus, subsequent  ammonia  addition  is
required in order to control THMs.

Figures 5-42 and 5-43  illustrate overall influent chloride levels plotted with and without
the inclusion of Utility  10, respectively.  Because of the relatively small number of data
points in  the  chloraminating  category,  the inclusion  of Utility 10's extremely high
influent  chloride levels produces an oddly-shaped box-and-whisker  plot.   Whether or
not  Utility  10 is  included,  however,  the overlapping  95 percent confidence intervals
indicate  there is no statistical difference between the medians  of any two quarters.  In
Figures 5-44 and 5-45. influent bromide levels are plotted by  disinfection scheme both
with    and   without   the    inclusion    of   Utility    10.        In   both    plots,
prechlorinating/postammoniating utilities  have  a higher median  influent  bromide level
than chlorinating utilities, and  the difference is statistically significant at  a 95 percent
confidence level.

Classes of DBF Compounds

Figure 5-46 illustrates  levels of XDBP1lim by disinfection scheme.   The plot indicates
that the  prechlorinating/postammoniating  utilities have  a higher  median value  of
XDBPslim  (approximately  94  //g/L)  than  either  the  chlorinating   or  chloraminating
utilities (approximately 62 and 23 //g/L,  respectively),  and that  the median XDBPS(im
level  of the chlorinating  utilities is higher than  that  of the  chloraminating utilities.
These differences are statistically significant  at a 95 percent confidence level.

Figure 5-47, a plot of  overall median TTHM levels by disinfection scheme,  reflects the
same  trend  observed in  Figure   5-46.     The  median   TTHM  level  for  the
prechlorinating/postammoniating  utilities  (approximately  57  //g/L) was significantly
higher than that of the chlorinating and chloraminating utilities (approximately 34 and
12  //g/L. respectively), and the median TTHM  level of the chlorinating utilities was
substantially higher than that of the chloraminating utilities, although the difference was
not statistically significant.

The fact  that the prechlorinating/postammoniating utilities produced higher levels  of
XDBPSI1I)) than the  chlorinating utilities is contrary to an intuitive expectation.   While
the plots of influent water quality parameters did not indicate  significant differences in
TOC or  UV-254 levels  between the three disinfection schemes (Figures 5-40 and 5-41),
the median UV-254 level  was higher for  the prechlprinating/postammoniating  utilities
than for the other two  disinfection schemes. The  higher levels of UV-254  absorbance
for the prechlorinating/postammoniating  utilities  may  indicate  higher levels of  DBP
precursors (i.e.. higher humic  substances  content of the dissolved organic matter) than
in the influents of  utilities  using the  other disinfection schemes.   Thus,  the higher
median  level of XDBPslim  may  reflect  the higher concentrations of  DBP precursors  in
the prechlorinating/postammoniating utilities.

Another  factor contributing  to the unexpectedly high  levels of DBPs occurring at the
prechlorinating/postammoniating utilities may be the higher bromide  levels measured at
these  utilities.     The   data  plotted   in   Figure  5-45   indicated   that   the

                                       5-9

-------
D
M
•n
•H
r
o
                Influent Chloride


              By Disinfection Scheme
                Influent Chloride


        Bg Disinfection Scheme (w/o
                         T
                                                    _
                                                    \
                                                    a
L
0
-I


o
c
II
1
                                                    H
                                                                             T
                0-3
                                  NH2CI.
                                                                    CL2
                                                                            CL2NH3
                                                                                      NMSCL
                58        29        9

                20        1Q     .3
                Disinfection Scheme



                      FIGURE 5-42
               .  58       29         6

               .20       10         2

                Disinfection Scheme



                     FIGURE 5-43

-------
                Influent  Bromide


             By Disinfection Schemi
                 Influent  Bromide


        By  Disinfection Scheme 
-------
          XDBPsum By  Disinfection Schemi
                  Trihalomethenes


              By Disinfection  Schemi
X

Q   3»»
0.
s   -
•x.
                                                       -I


                                                       a
c

r
-P


o
                                                                                           T
                                    HH2CL.



                                     11
                  78        liO

                  21        11         
-------
                    Baseline Sampling Results and Discussion


prechlorinating/postammpniation utilities had  a  higher median  influent bromide  level
than the chlorinating  utilities, and  this difference was statistically significant  at  a 95
percent  confidence  level.    Research  by  Aizawa,   et  al.   (1989)   found   "...the
concentration of total  THM increased with the augmentation of bromide ions with the
same amount of chlorine dosage.  The increase in THMs is up to two times higher than
in the  absence  of bromide ions."   As discussed  later  in this section, utilities  treating
waters high  in  bromide  had relatively high concentrations  of bromoform  and other
brominated DBFs.   Since bromoform  is  the heaviest  of the  THM compounds,  its
presence can result in elevated TTHM levels.

In interpreting the DBF data presented as  a function of disinfection scheme, it should
be  noted that  baseline  data samples  for  this study were  collected at  the clearwell
effluents of the participating utilities.  Figure 2-2 illustrated the process trains of the
plants  at which samples  were collected.   Of the 11 prechlorinating/postammoniating
utilities participating in this study,  eight did not add ammonia  until  the filter  effluent
and  filtration was  the process  immediately upstream  of the clearwell.  Thus, these
utilities maintained  a  free chlorine  residual throughout most of the in-plant detention
time, and  levels of DBFs measured at  these utilities reflect the free chlorine contact
time.         The     differences     in     DBF     concentrations    between    the
prechlorinating/postammoniating utilities and the chlorinating utilities would be more
apparent  had   the  sampling  taken place within  the distribution  systems  of  the
participating utilities.

As  discussed previously  in the section entitled "Star Plot Analyses", all other factors
being equal, it would  be expected that the  chlorinating  utilities would have the highest
median TTHM  level compared  to  utilities  employing the  other  two disinfection
schemes.  Figures 5-46 and 5-47 highlight  the danger of oversimplifying the  influences
of causal factors on DBF  production.  Taken out of context, these two figures could
give the impression that  based  on median  levels,  prechlorinating/postammoniating
utilities could not meet a  revised  THM  standard  if it were lowered to 50 //g/L, while
utilities utilizing only free  chlorine would be able to meet this lower standard.  Such a
perception  would not consider  other potential causal  factors found in this  and other
DBF  studies,  such as  influent  TOC  levels, influent  UV-254 absorbance,  influent
bromide concentration, free chlorine contact time, pH,  and temperature,  among others.
A correct interpretation of the results presented in Figures 5-46 and 5-47 is that, of the
35     utilities    participating    in    this    study,     the    utilities    practicing
prechlorination/postammoniation  had higher median levels of halogenated   DBFs  and
TTHMs than the utilities employing chlorination or chloramination.  The UV-254 data
presented previously in  Figure 5-41  may  suggest that these  results are due  to  the
presence of  THM  and DBF  precursors in higher levels in  the source  waters of the
prechlorinating/postammoniating  utilities,  and  that  prechlorination/postammoniation
may be the most effective oxidation/disinfection method for such  source waters.

Figure  5-48  shows overall  HAA  concentrations  by  disinfection  scheme.    Both
chlorinating and prechlorinating/postammoniating  utilities had significantly higher HAA
levels     than     chloraminating      utilities.          The     chlorinating     and
prechlorinating/postammoniating utilities had median HAA levels of approximately 22
and 19 /yg/L. respectively,  while the chloraminating utilities had a median HAA  level of
only 8  /yg/L.
                                      5-10

-------
   H«lo«c«tic Acids
By Disinfection  Schemi
                                                   Heloecetonitriles
                                                By Disinfection  Schemi

IB!

/^
_J
\
a
j !»•
N-'
•
D
•H
U

-------
                    Baseline Sampling Results and Discussion


Overall HAN levels are shown in Figure 5-49 by disinfection scheme.  This plot reflects
the same trend seen  in  the plots  of XDBP,.im)  and TTHM levels  as  a function of
disinfection scheme;  that   is,  a  significantly  higher  HAN  concentration for  the
prechlorinating/postammoniating  utilities compared to  either  the  chlorinating  or
chloraminating utilities.   However, in contrast  to  the  plots of XDBPsum and TTHM
medians,  Figure  5-49 shows a number of outliers at high levels for the chlorinating
utilities, indicating that  some chlorinating utilities produced relatively  high levels of
HANs (in the approximate range of  12 to 21
Figure 5-50  is a plot  of overall  HK concentrations by disinfection scheme.  This plot
indicates  that the median level  of  HKs  was significantly  higher  for chlorinating and
prechlorinating/postammoniating utilities compared to chloraminating utilities.

Figure 5-5 1  shows  ALD levels  as  a function  of disinfection  scheme.   In this plot,
prechlorinating/postammoniating  utilities   produced  the  highest   median  ALD
concentration,  significantly  higher  than either  the  chlorinating or  chloraminating
utilities.     Influent   ALD  levels  and   ALD  production  as   a  function  of  the
oxidation/disinfection  schemes utilized by water  treatment facilities will be discussed
later in this section and in Section 6.

Overall chloropicrin and chloral hydrate  levels are  shown in  Figure  5-52 and  5-53,
respectively,  as a function of disinfection  scheme.   Both plots suggest the same  trend of
the highest median for chlorinating  utilities,  lower for prechlorinating/postammoniating
utilities, and  the lowest median for chloraminating  utilities.   For chloropicrin levels, the
differences between medians are significant for  the chlorinating and  chloraminating
utilities.   For chloral  hydrate, the  median level is significantly higher for chlorinating
and prechlorinating/postammoniating utilities compared to chloraminating utilities.

Figure 5-54  is a  plot of cyanogen  chloride data  for the three disinfection scenarios.
The  plot  indicates that the prechlorinating/postammoniating utilities had a significantly
higher  cyanogen  chloride  level  (approximately  2 /ug/L) compared  to  both  the
chlorinating and  chloraminating  utilities  (less  than 0.5 /i/g/L).  These results will  be
evaluated  in detail later in this section in the  "Special Issues" discussion.

DBF AND INFLUENT PARAMETER CORRELATIONS

The  objective  of this section  is  to examine correlations of various DBP  classes,
individual DBPs and various water  quality parameters.  Correlations were determined
between:

     o    DBP classes
     o    Individual DBPs
     o    Individual DBPs and influent water quality parameters
     o    Influent water quality parameters

In total,  over 300 correlations, as shown in Appendix  D. were determined.   Figures
5-55  to 5-68 present  a selected number of these.   Note that in some  figures,  the
correlations are re-evaluated without  including certain  outliers.   In some cases,  the
outlier is  an anomalous data point not observed in other samplings at the same utility.
In other  cases,   utilities  with  very high  concentrations of  the  parameter under

                                      5-11

-------
                                                               Aldahydsa Bg  Di»inf«ctian  Schams
               By DiHinfaction Schvmi
\
a
o
ii
o
-4

                                       3
                                       a
                                       TJ
                                       £
                                                       H
                                                       
-------
Chloropicrin  Bg  Disinfection Scheme
    Chloral Hgdrate
By Disinfection  Schema
...
1
^
1
\ ..a
a
3

c
•H
L • »
U

a.
0
L
0
H ••*
r
o


t.2
«
i




















•
-
j
1 	
n °
u =





































1
CL2
78
21




























































w
"

CC2HH3
II
	 1 	 r~
-
-

-

"
*


"



•








1 i
HH3CL
«
28
G
\
a
D »•


4J
i
L
TJ 18
31
X

^
fj
i. i.
0
H
£.
U
5

•
-i 	 r i i i_
_
•

— ~
-
"
.
.
_
•



•
•
_ —


-
-
l l
: r\ 2 ;
"l 1 	 1 	 L.
CLl Cl-2>*O NH2CL
11 n - 72 3* U
4 u • 2] 10 "i
         Disinfection Schem
               FIGURE 5-52
  Disinfection  Scheme
        FIGURE 5-53

-------
                Cyanogen Chloride

             By  Disinfection  Scheme
                         T
T
-I

a
3
•a
•H
L
0
H

a
n
a
O
               CL2
                                  KM30.
                71        32        n
                21        10         *
               Disinfection  Scheme


                      FIGURE 5-54

-------
                   Baseline Sampling Results and Discussion


consideration can create what may be an artificially  high correlation coefficient; thus a
more realistic value is obtained by treating those utilities as outliers.

Correlations with THMs

In this study, analyses were performed through four analytical fractions for a total of 19
individual halogenated  DBFs  (XDBPs).   Concerns have  been expressed as  to the
practicality  of performing  several  different  DBF analyses  in  a utility's  routine
monitoring  program,  and  the  question of  employing a  surrogate has  been  raised.
Figure 5-55 presents various correlations with  THMs.  There was a strong correlation
between TTHMs and the sum of XDBPs measured in this project (r=0.96). As THMs
represent the largest  DBF fraction  detected in  this study,  the data were reevaluated by
comparing  TTHMs to the sum  of  non-THM XDBPs.  In this instance, r decreased to
0.76.   However, the latter comparison does  not mean that THMs cannot be used  as a
surrogate or predictor of the  sum of all XDBPs.   It  should  be noted,  though,  that
correlations  between classes of compounds were often low (e.g., comparing TTHMs to
haloketones  yields an r of only 0.06).  Additionally,  some of the other individual DBFs
may require separate monitoring based on their health effects and their formation and
control relative to  THMs.  For example,  THM production can be minimized  with the
use of chloramines, whereas cyanogen chloride formation can be increased, as will be
discussed in detail later in this section.

Figure 5-55 also presents the correlation of HANs with THMs  (r=0.78).  The median
ratio of the sum of the HANs to THMs  for all   four quarters  was 0. II (the  25th
percentile was 0.073  and the 75th percentile  was 0.15). The linear regression  equation
for this relationship is:

                        IHANs) = 0.78 +  0.087[THMs]

In general, it appears that the concentrations  of HANs were approximately one-tenth of
the concentrations of THMs.   Another study  found that the concentration of DCAN
averaged about 10 percent of  the  THM concentration (Oliver,  1983).  The variations
from these  generalized relationships are probably due, at  least  in part, to the  different
effect of pH on these two DBF classes;  this is discussed  in greater detail  later in this
section.

Correlation with HAAs

Figure 5-56 presents various correlations with HAAs.  The correlation of HAAs  with
XDBPSUI)I was  0.87.  If THMs are not included in the correlation, r increases  to 0.98.
However.  HAAs  comprise approximately  57  percent of  the  non-THM  XDBPs  (see
Figure 5-3).   If both  HAAs and  THMs are subtracted from  the  XDBP    and  then
correlated with HAAs, r equals 0.77.  Furthermore, a correlation  of 0.74  is  found if
only HAAs are subtracted from XDBP     The latter two  correlations may be  useful in
helping to predict the sums of non-THM, non-HAA XDBPs.

Correlations with Influent Water Quality Parameters

The correlations of TOC with THMs and UV-254 absorbance, and  UV-254  absorbance
with THMs. are shown in  Figure 5-57.    Several  researchers have  investigated the

                                    5-12

-------
        Correlations  with Trihalom»thanes
-I
E



Q.


S
X
   308
   299
   188
        XDBPsum  us  THM«
       t
                      r=0.96
          sa   lea  isa see  asa  3ea

                n-1*»0
               THMS (ug/L)
       XDBPs-THMs us  THMs
H

§

s
X
                                     ise
                                      eo
                                      48
• '-I
O • B>
                    r=0.76
          se  iaa  isa aee  sse  3aa

                0=1^0

               THMS 
 J
    20
    16
    12
         HANS  us THMs
                     r»0.78
          sa  iaa  isa  sea  2sa 3aa

                n-1^0
               THMa (UQ/L)
                             FIGURE 5-55

-------
      Correlation* with  Haloacetic  Acids
       XDBPsum us HAA
     XDBPs-THMs  us  HAAs




aea

^
j
a
3
i 29e
a.
s
X

180




A

a

a
O






a
B
O
a
a
°.i >*.' '

5? & . '"f^ °

i»'f*'°
•&' r«0.87
Jjfca
fC~ 	 	 	 . 	
169



129
J
S
a
2

•
| 89
i
§
a
CO
o
X 48




9

a
0
o
a

a
!



%v°°
a o <* a
Bag ° OD

a«T» °
0 &** *
f^K
f r=0.98
9
L. 	 	 	 — 	 '
         28  48  68  89  199  128



             HAAS (ug/L)
        28   49  68   88  188 128

             n=1 kQ
            HAAS 
-------
       Correlations with  Influent Parameters
J
a
c
   300
   250
   see
   15B
   198
    58
           THMs us  TOO
           4   8    12   IB

                n-105
           Influent TOC 
                            28
               THMs  us UU  Abi
                                     380
                                     569
                                     209
                                     158
                                     188
                                      69
                 8.2
                       8.4
                            8.S
                                  8.8
                     n=102
                Influent UW-254 (/cm)
   8.8
§  a.e
\
U)
ru
§
   9'4
   8.2
        UU Abs us  TOC
                 r-0.79, 0.85*
           4    8   12   IB

               n-102,  101*
           Influent TOC  (mg/L)
28
        Excludes indicated outlier
                         HGURE 5-57

-------
                    Baseline Sampling Results and Discussion


 relationships  among  TOC,  UV-254  absorbance,  and  THM  formation  potential
 (THMFP)  (AWWA Organic  Contaminants  Committee,  1985; Amy  et al.,  1987;
 Chowdhury et al.. 1988). For this study, the correlation  between the UV-254 and TOC
 of plant influents was high (i.e., r=0.85, without outlier).  Determining this correlation
 by water source did not improve the correlation (Figure 5-58).   It  should be noted that
 neither TOC or UV-254  correlated well with  the TTHMs  of  the  plant  effluents.
 Specifically, during the  last three sampling  quarters,  the  correlation of TTHMs with
 influent TOC was 0.48: and for TTHMs with UV-254, r=0.41.  Because utilities apply
 chlorine  doses based on  numerous  considerations  (e.g., the chlorine demand of the
 water, disinfection   requirements,  taste and  odor  control),  in actuality  the TTHMs
 detected  do not  necessarily reflect  the THMFP of that water.   Also,  since a  large
 percentage of the utilities in this study used chloramines and several utilized ozone as a
 preoxidant. it is not surprising to find a poor correlation  between the influent TOC and
 the  effluent DBFs.   A mathematical  model accounting  for chlorine  dose,  TOC
 concentration, bromide level, temperature, pH  and  other parameters is currently being
 explored by other researchers to determine whether THM levels are predictable (Amy et
 al..  1987).

 Correlations with Bromide and Brominated  DBFs

 Figure 5-59 presents correlations of plant influent chloride with plant influent bromide,
 and  influent bromide with bromoform.  Influent chloride  had  a very high correlation
 with influent bromide (r=0.97). The three outliers  noted were from the same utility;  if
 these are  not included, the correlation coefficient decreases  to  0.86.  Because of the
 high correlations, chloride may be used as a predictor for bromide.  Using all the data,
 linear regression analysis yields the equations:

                  |Br  ] =  -0.0500  + 0.0044|Cl-j (with  outliers)

                 |Br |  = -0.0071  -I- 0.0034|CI]  (without outliers)

 The  correlation of influent bromide  with bromoform is also presented  in Figure 5-59.
 The  correlation  for  this  relationship was  0.57  unless  the noted outliers are excluded
 (r=0.69).

 Relationships of influent  bromide with chloropicrin, 1,1,1-TCP, TCAA and chloroform
 are presented in Figure 5-60.  In each case, an exclusion relationship is demonstrated,
 i.e.. the presence of bromide appeared to exclude the  presence of the particular DBP,
 and the inverse was also observed.  Mutual exclusion of bromide with these compounds
 is also demonstrated by their negative correlation coefficients.

 The  formation  of   chloropicrin  due  to  chlorination of:    1)  nitrogenous organic
 compounds, 2) non-nitrogenous organic substances  in  the presence of nitrites,  and 3)
 humic  substances, has  been well documented in the literature (Coleman et al.,  1976;
 Sayato et al.. 1982;  Duguet et al..  1984; Thibaud et al.,  1986).  Additionally, bromide
 has been shown by  many  researchers  to  play  an important  role  in the formation of
 trihalomethanes (Rook, 1976).   This  study has shown, however, that bromide may also
 play a significant role  in  precluding the formation  of certain other compounds.  The
exclusion relationship of bromide with chloropicrin  is consistent with that observed by
Thibaud,  et  al.  (1988).    These  researchers  showed  that  increasing  bromide

                                     5-13

-------
        Correlations  with  Influent  Parameters
               U«t»r Source i« F9
                                                   Uater Source ie QU
    a.a
 E   0.6
 u
 10
 nj
 C

 5

 *C   8.2

             00

           O  • •
                 r - 0.82, 0.90
             4    8    12    16    28

                  n -35, 3*» *
             Influent TOC (mg/L)
                                        8.5
                                        0.4
U
                                     ni


                                     3
                                         .1
                                                              0.8A
            3    6    9    12    15

                  n = 21

            Influent TOC (mg/L)
               Ueter Source im LR
ID
ru
    8.3
   8. 26
    8.2
   e. IB
    8. 1
   0.8E
                 f

                 I
                      r = 0.66
                   n = k6
             Influent TOC (mg/L)
        * Excludes Indicated Outlier
                                FIGURE 5-58

-------
        Correlations  with  Influent
             Chloride and  Bromide
2.5
J
V.
a
•H

0
L.
CD
1.5
e.s
       Br— us CL-
                                         CHBr3 us  Br—
                  r=0.97, 0.86*
                                  89
                                 60
                               a
                               a
                               a
                               o
                               n
                                 £8
                                             .  r=0.57, 0.69---
        208    486    688
         n=105, 102^
       Znfluant Chloride (tng/L)
                         688
                                                         2. 5
                                          0.5   1   1.5  2

                                             n-105, 102*
                                           Influwrt Bromid* (mgxL)
     Excludes indicated outliers
                         FIGURE 5-59

-------
         Correlations  with Influent  Bromide
J
   1.6
   1.2
   9'4
            CHP  us Br-
               r«-0.22, -0.25*
                  1.6
                         2.6
          0.6   1

               n-10^*, 101*
          Influent Bromide (mg/L>
1,1, 1-TCP  us Si—
                                   a  4
                                   a
                                   o
                                   -H

                                   u
                                   L
 &.°
        r>-0.26, -0.35^
   0.6  1  l.S   2   2.E

         n-105, 102*
   Influant Bromide (mg/L>
            TCAA us  Br—
     CHC13  us  Br-
40


S
? 30
w
TJ
• H
€

•H 20


a
B

.
a


i •
i |
i •

5 1 .
• L «
o E-.
* r.
£ 10 h
" Is' .
L O .
H F . r-0.27, -0.35*
^^•^0«« • ««
1SB


120

i
S
3
w "•
_
t
0
g 60
0
«^
5

a






1 a
o

*u
t*
i

ii1'
Ul .
30 »:

if
B\.' r=-0.25, -0.29'1-'
• Li '^tmAt • 	 > 	 i*-i
          0.6  1  1.6   2   2.6  3
                 n-105, 102*
           Influent Bromide (mg/L>
    8.6  1   1.6  2  2.5

          n-105,  102*
    Influent Bromide (mg/L>
          Excludes indicated outliers
                             FIGURE 5-60

-------
                   Baseline Sampling Results and Discussion


concentrations   led   to    a   decrease   in   the   production   of   chloropicrin
(trichtoronitromethane)  and  to  the  formation  of  brominated  halopicrins  (i.e.,
bromodichloronitromethane,  dibromochloronitromethane and  tribromonitromethane).
These halopicrins are  analogous  to the THMs and appear to be influenced by bromide
in a similar manner.

The  exclusion relationship  of bromide with 1,1,1-trichloropropanone is supported by
work performed  by Croue and Reckhow (1988).  These researchers showed that in the
presence of 0.4 mg of bromide per mg of carbon, the chlorinatiqn of a particular fulvic
acid  yielded  decreased  concentrations  of 1,1,1 -TCP.    Additionally,  several  new
brominated  compounds  were  detected  including  chloro-,  bromopropanones.   As
1,1.1 -TCP has been demonstrated to be an intermediate DBF  which can hydrolyze to
chloroform,  the  chloro-,   bromopropanones  may be  precursors  of the brominated
THMs.   Since bromide appears  to play  an important role in DBF formation, it is not
unusual that  other exclusion relationships  such  as  with TCAA  and chloroform were
observed, as shown in Figure 5-60.  A more detailed discussion of the effect of bromide
levels on DBF production is presented later  in this section.

The pattern  of exclusion relationships with  bromide and chorinated DBFs  is  similar to
that shown with bromoform and chlorinated DBFs.  Figure 5-61 presents correlations of
chloroform,  1.1.1-TCP, chloral  hydrate  and TCAA with bromoform.  In each case a
negative correlation was observed.   Koch  and  Krasner (1989)  observed  that chloral
hydrate  concentrations  were higher  in  two  water sources  low in  bromide when
compared to a third source water high in bromide.   Additionally, in this  study, when
bromoform was  correlated  with a brominated compound  such as dibromoacetonitrile
(Figure 5-62). a  fairly high correlation was achieved (r=0.88).   A similarly  high
correlation  was   observed   when  dibromoacetic   acid   was   correlated   with
dibromoacetonitrile (r=0.85).

Correlations with  Chloroform

In an   effort  to  evaluate  whether chloroform  could   be  employed as  a  surrogate
parameter for other  selected DBFs,  various  correlations  with  this compound were
determined.  Figure 5-63 shows  the correlations of chloroform with DCAA and TCAA.
Correlations with all utilities  participating in the study and with non-ozonating utilities
are presented  because it has  been shown that  ozonation can reduce  TCAA precursors
(Reckhow and Singer. 1984; Bruchet et  al., 1985;  Lykins  et  al.. 1986;  Dore et  al.,
1988).  Additionally,  it has been shown  conceptually that ozonation can shift the DBF
speciation from chloroform and TCAA to DCAA (Reckhow and Singer,  1985).  There
was very little difference observed whether or not ozonating utilities were included in
the correlation for either compound; however, only  three utilities  utilized ozone in the
baseline studies and.  thus,  may  not have impacted  the  overall correlations for the 35
utilities.  Correlations for  DCAA and TCAA  with chloroform when all utilities were
included were  0.85 and 0.80, respectively.

Correlations of  chloroform  with  chloral  hydrate,   1,1,1-TCP  and chloropicrin  are
presented in Figure 5-64.   The best correlation among the three compounds was  with
chloral  hydrate (r = 0.85).   The  poor correlation  with 1,1.1-TCP  (r=0.52)  may be
attributed to the  instability  of the compound.  Reckhow and  Singer (1985) showed that
1,1.1-TCP was an intermediate of chloroform.   They reported that  approximately  7.5

                                     5-14

-------
           Correlations  with  Bromoform
       CHC13 us CHBr3
    1,1,1-TCP us CHBr3
150


120

J
Q
90
8
L
0
w
g 60
O
£
U
38


A







O





r=-0.30
a
B
" 0 0 *
«. « « 	 4 	 IB - n 	 B-S 	 " 	
8
n
3

6
I
n
I
a. A
0








i
i
.
* 1
u f
t L"
H 2 f
3 | • „ r=-0.29
L" '•
F a A o ° ° «
e fc. 	 2 	 * 	 tf m * 	 • 	 9 	 1
         28    48    68
            n=1'*0
          Bromoform (ug/L)
                         88
          28    48    68
             n=1 ^0
          Bromarorm 
-------
     Correlations with  Dibromoac»tanitrile
        CHBr3  us  DBAN
    DBAA us DBAN
   80
   60
L
n
   29
                  r=0.88
                                28
                                "
                             TJ


                             I

                             0
            r-0.85
     0   3   6   9   12   IS


       Oibronoacvtonitril* 
e    3   6   9   12   15


  Dibromo«c«tonitril« (ug/L)
                       FIGURE 5-62

-------
             Correlations  with Chloroform
         DCAA us  CHC13

              All Utilities
         DCAA us  CHC13
                                            Utilities wXo Ozone
50

I *'
TJ
0 30
U
•H
4J

U 20
0
a
-4
r
u 10

a
a

* B

° . '•
a
a
B
<•• B
* . "


B* " °
. •%°.| "
if*'* B
"•':"" •

Ter*
£ " »." r-0.86
ee

\ 49
•D
'u 39
U

+J
1
u se
2
a
^ 19

0 0

°. '.
a
a
a
(0° a
ff -
B

.* •-
. -^'.B '
S» • *' B
//."" '

'° kir*
B*-Js- r=0.85
   60
   BO
   40
   30
U

I  -
•-4
I.  10
          39   69   99   129   160

              n=1^0
           Chloroform (ugXL)



         TCAA us  CHC13

              All Utilities
                    o.so
                                   "
                                 •a   49
                                 •H   39
k  2e
                                    10
          39    69   90   129

               n-UO
           Chlorororm (ug/U)
                            189
          39    69   99  128   158

               n=129
           Chloroform 
-------
            Correlations With Chloroform
   24
   20

J
\
a

2  16
4J


L
          CH us CHC13
                  r=0.85
          30   68    90   120   150

              n=128
           Chlorafarn (ugXL)
                          1,1,1-TCP us  CHC13
                       a.
                       a

                       a  4
                                u
                                L
                                h
                                                 ' -0.52
                                30   60    90   128   150

                                    n=1i*0
                                 Chloroform (ug/L)
J

a
L
a
-H
r
u
   1.6
   1.2
   0.4
          CHP  us  CHC13
   >  >

a*  •


••  *«
          30   60   90

               n-139
            Chlorafarm (ug/L)
                       120  150
                           HGURE 5-64

-------
                   Baseline Sampling Results and Discussion


percent of the chloroform  produced  following  72 hours  of  reaction time  between
chlorine  and  a  particular  fulvic  acid  at  pH   7 passed  through  the  1,1,1-TCP
intermediate.  This issue will be expanded upon later in Section 5.

Chloroform correlated  poorly with chloropicrin (r=0.49).  This may be due to several
factors,  including  the concentration  of nitrogenous  organic  compounds.    That
chloropicrin can be formed by nitro-compounds and amino acids  has been reported by
several researchers (Coleman et at.,  1976; Sayato et al., 1982; Thibaud et al.,  1986).
Additionally, chloropicrin  precursors can  be produced  during the oxidation of various
non-nitrogenous compounds, such as phenols, if nitrite is present (Duguet et al.,  1985).

Correlations with DCAA and TCAA

Correlations of the four trihalomethanes and DCAA are shown in  Figure 5-65.  The
best correlation was found with chloroform (r=0.86).  However,  as the THMs shift to
the more brominated species, the correlations decrease until an exclusion relationship is
observed between DCAA  and bromoform (r=-0.33).   This progression  is consistent
with the  bromide correlations discussed above, which showed that bromide and various
chlorinated DBFs were related by exclusion.

Figure 5-66 presents correlations with TCAA.  The correlation of DCAA with  TCAA
was 0.85, while that with  DCAN was 0.76. The figure also shows that the correlation
with 1,1,1-TCP was 0.77.  The latter correlation was better than that  discussed above
between  1.1.1-TCP and chloroform (r=0.52).  This observation  is  consistent  with the
fact that  at high pH. chloroform is stable while the other two compounds may undergo
hydrolysis.

Correlations with Formaldehyde

The correlation  of acetaldehyde  and  formaldehyde  is shown   in  Figure  5-67.    A
correlation of  0.64 was observed between the  two. Because it has been shown  in this
study and in others (Lykins et al., 1986;  Glaze et al.,  I989a) that ozonation produces
aldehydes during water treatment,  correlation of acetaldehyde with formaldehyde were
determined for only  non-ozonating  utilities.   Excluding the outlier,  the correlation
improved to 0.78.

Correlations with Total Organic Halide

Figure 5-68 presents the correlations of XDBPsum with TOX on a molar basis.  With all
utilities included  in the analysis,  the correlation  was 0.70.   However, if disinfection-
only utilities  are not  used  (since TOX  increased in  concentration between  time of
sampling to time  of receipt at Metropolitan's laboratory; see Section 3 of this report for
details), the correlation improved to 0.86, or 0.77 without the outliers.    When only
utilities which  employ chloramines as a final disinfectant are included in the correlation
(since TOX concentrations in chloraminated effluent appeared  to be least affected by
transit  time to Metropolitan's laboratory for preservation; see Section 3  for details), r is
equal to  0.78.  The correlations found in this study are not as high as those found by
Singer (1988)  who correlated the  THM formation potential  with the  TOX formation
potential of various raw waters (r=0.96, n = 60, reported on /ug/L  basis).  However, the
correlation  found  by  this researcher between TTHMs and  TOX on  finished  waters

                                      5-15

-------
      Correlations with  Dichloroacetic Acid


                                     CHCl2Br  us  DCAA
   169
   128
J
\
    68
    39
 CHC13 us DCAA
                  r=0.86
                               a
                               •H
                               Q
                                   68
                                   48
                                   28
                                            o
                                       On o
  18   28   38   48   58


Oichloro«c«tic Acid (ug/t)





CHBr2Cl  us DCAA
                                         18   28   38   48   58
                                                     .
                                        Dichloro«c«tic Acid (ugxl_)
                                       CHBrG us  DCAA
88

3
X
3 68
w
g
£
I 48
0
L
0
u
o
0 28
L.
a
,•4
Q

a



a



o
a a
a
a
e a
a
a* a.8
r-0.03
o •
Jo a g
ft *2^ ° * B °
^^ ^^ ° o* °8 f ° *
88



68
*»
^
J
L 40
a
a
!
o
20



a

a



a
"> 0

D.
1
a
J
°.
a
e
D°°' - r=-0.33
a B
a° o o
^r-»f?J^ -r.- - -
           18   28   38   48   58
                                  18   28   30   48   58



                                Oichloro»c«tic Acid (ug/L>
         Oichlaroacatic Acid Cug/L)
                          FIGURE 5-65

-------
  Correlations  with Trichloroacetic Acid
         DCAA  us  TCAA
1,l,1-TCP us  TCAA
68

J
3
TJ
U 38

-------
          Correlation  with Formaldehyde
                                         Non-0xon«ting Utilitii
   IB
   12
j
a
        .:-
       -%•/
                                    la
j
a
           IB     20    38

               n=95
           Formaldehyd" 
-------
I  '
t«d
og»n«



M
L

             Molar  XDBP»  uer«u«  Molar  TOX
               All Utilities
          -••»  *'.
                    r-0.70, 0.60*
              a   12
                           2«   24
                n-138.  136*
               Molar Vox
         Utilities using Chlaraninae


            As Final Disinfectant
                     r=0.78
      •    s   4   a    a   !•   12


               Molar TOX



      *  Excludes  Indicated Outliers
Net Diainfsetion-Oniu Utilities
                                    i
                                    TJ
                                    41
           r-0.86,0.77'1;
                                                                   24
        n- 122. 120*
        Molar TOX
                              FIGURE 5-68

-------
                   Baseline Sampling Results and Discussion


(r=0.89. n=!66) was  closer to the correlation found in  this study.  The difference in
correlations  between  the   formation  potentials  of  raw  water  and  instantaneous
measurements  on finished water is probably due to the variety of conditions with which
each plant treats its water.

SPECIAL ISSUES

In evaluating  the results of the baseline  data collection for this  study, a number of
issues were identified which could not  be  classified under the previous headings of this
section. These special  issues are discussed below.

Effect of pH

The  HAN and  HK  data  presented  in  Figures  5-29 and 5-30  tend  to support
relationships reported by other researchers in which HANs and HKs were found to be
reactive intermediates  rather than stable  endproducts such as  THMs.   For instance,
Reckhow and  Singer (1985) reported that  when a  fulvic  acid solution was chlorinated,
concentrations of DCAN and  1,1,1-trichloropropanone (1,1,1-TCP) declined  over time
after quickly reaching an initial peak while chloroform, TCAA and DCAA increased in
concentration  with increasing  contact  time.   These  researchers also found  the same
trend with increasing pH. demonstrating that DCAN and  1,1,1-TCP hydrolyzed at high
pH;  while chloroform concentrations increased with increasing pH, indicating that this
compound  is   a  product  of  the  hydrolysis reactions  of  DCAN  and  1,1,1-TCP.
Additionally.   Gurol. et  al. (1983) found that the presence of free chlorine greatly
increased the degradation rate of 1.1,1-TCP and the formation of chloroform.

In order to investigate the  impact of pH  on  the DBP concentrations measured in this
study, several  DBPs were plotted  as a function of clearwell  effluent pH.  In actuality, it
is the pH  (which varies) throughout the treatment process that impacts DBP formation,
but it is the final pH which determines the stability of the DBPs  and their fate in the
distribution system (i.e.. that received by consumers).   In these plots, clearwell effluent
pH values  from all  four quarters of baseline data collection were divided  into four
intervals  of varying  magnitude,  but with  an equal number  of measurements in each
interval.  (Usually, plots with  equal interval widths would be a preferable way to show
the data: however, because there were very few data points in the lowest  pH interval
and most of the data points were  in the highest pH interval, the plot was prepared with
varying interval widths and  equal  numbers of data points  in each interval.)  Figure 5-69
illustrates  the  levels of  XDBPfum plotted in this  manner.   No  significant  difference
between  XDBPsum values are  indicated by the  data  in  Figure  5-69.  The  problems
inherent in any attempts to ascribe differences in DBP  levels to  single causal factors
such as pH have been discussed previously in this section; and Figure 5-69 may indicate
the influence  of many  confounding factors.  However, it  may also be the case that pH
impacts the  levels  of individual DBP compounds  which  make up  the  parameter
XDBPsunr rather than influencing  the level  of XDBPsum itself.

To further explore this issue.  TTHMs  were plotted as a function  of clearwell effluent
pH  in  Figure  5-70.   As observed in  the plot of XDBP  , there are  no statistically
significant differences between the median levels of TTHMs at a 95 percent confidence
level.  HANs  are plotted as a function of pH in Figure 5-71.   This plot, in general,
indicates decreasing levels  of  HANs with  increasing pH.   When clearwell  effluent pH

                                     5-16

-------
                    XDBPsum  By Effluent  pH
J

a
E
3
«
a.
m
Q
X
                    13
                           T.BS-8.23


                               33

                               111
                                      O.24-8.7
8.11-9.78


   33

   13
                              pH Range


                              FIGURE 5-69
                                                                    J

                                                                    a
                                                                    3   «•
                    £
                    +J
                    II

                    0
                                                                    •H
                                                                     L
                                             Trihalomethanas


                                             By  Effluent  pH
                                                                                                  "T
                                                                                      4.6-7.BE
                                                                                        13
7.E6-8.23

   33
                                                  pH Range


                                                  FIGURE 5-70
                                                                                                                     \
                                                                                                                       J	1.
a.71-9.78

   33

   13

-------
                    Haloacetonitriles

                      BU Effluent  pH
                                                                               Dichloroacet.anit.rile

                                                                                  Bg Effluent  pH
                          T
_t

a
•H
L

•H



I   "
II
U

0
                                                                IB
                                                            J

                                                            a
                                                            •H
                                                            L
                                                            +J
                                                            •H

                                                            0
                                                            4J
                                                            II
                                                            U
                                                            •
                                                            0
                                                            L
                                                            0
                                                            H

                                                            U
                                                            •H
                                                            Q
                                                                            T
                                                                                      T
                    T
                                                                             J_
                        T.M-e.83
                                 •.24-V.7
                           33       ^
                           ft       f*
                          pH Range


                           FIGURE 5-71
                                           B.-71-S.7B

                                             33

                                             13
4.6-7.ft

  3<«
  13
                                                                                    T.ea-a.33
33       3*»
Hi       l"t
pH  Range


FIGURE 5-72
9.71-9.78

  33

  13

-------
                   Baseline Sampling Results and Discussion


values occurred  within  the  range 8.24  to 9.78, the  median  level of  HANs  was
approximately 2.5 to 3.5 /ug/L, whereas the HAN medians were 4.5 to 5.5  /ug/L in the
4.6 to 8.23 pH ranges.  The difference between the median HAN concentrations for the
7.56 to 8.23  pH  range and the 8.71 to 9.78 pH range is statistically significant at a 95
percent confidence level.  Thus,  Figure 5-71 tends to support the finding that HANs
hydrolyze at high pH.  When DCAN was plotted as a  function of clearwell effluent pH,
no statistically significant difference in median concentrations occurred  over  the four pH
ranges (Figure 5-72).  However, the 75th percentile value for the 4.6 to 7.55 pH range
(3.7 /yg/L) is approximately  twice the 75th percentile value for the 7.56  to 9.78 pH
ranges (1.6 to 2.0/ug/L).

A plot of 1.1,1-TCP  as a function  of pH is  shown  in  Figure 5-73.    This figure
illustrates  that   statistically   significant  differences   occurred   between  1,1,1-TCP
concentrations at  low pH (4.6 to  7.55) and at high pH (7.56 to 9.78).  Figure 5-73 is
thus an indication that hydrolysis of this haloketone occurred  at basic pH.  This same
trend was observed in  a plot of TCAA  as a  function  of pH.  The median TCAA
concentration was higher within the  pH range 4.6 to 7.55 than within the pH ranges
greater than  7.55; however, the differences  were  not significant  at a  95  percent
confidence level.   Miller and Uden (1983) reported that TCAA concentrations  declined
substantially as pH increased from 6 to 10 in the chlorination  of a fulvic acid solution.
Miller and Uden  (1983) also  found that chloral hydrate increased in concentration from
pH 4 to pH 7, and then decreased as pH  increased to 10.  These researchers  reported
that  the product  of chloral  hydrate  decomposition  at  elevated  pH  was chloroform.
However, chloral  hydrate results from  the  baseline  data of this study indicated that
there were no statistically significant differences between the median concentrations of
this compound occurring within the four clearwell effluent pH ranges.

Effect of Temperature

Seasonal effects on DBP levels were discussed previously in this section.  Plots of DBP
concentrations showed  some  indications of temperature-related effects, although some
seasonal impacts may be due to changes in the nature of naturally-occurring organics as
well.   The  impact of  temperature on THM  formation has  been  documented by  a
number of researchers  over  the  past decade, including Stevens, et  al.  (1976), who
found  that chloroform  concentrations  more  than doubled (from  approximately 100 to
over 225 jug/L) when incubation temperature was increased from 25  to 40°C in  raw
Ohio  River water  receiving 10 mg/L of chlorine.

In order  to  evaluate  temperature effects  on  DBP  levels measured in  this  study,
XDBP,lim, TTHMs and HANs were plotted as a function of influent water temperature.
Data from all four quarters of baseline data were divided into four temperature ranges
of equal magnitude.   Figure 5-74 illustrates that levels of  XDBP,um were  strongly
influenced by temperature.  Although there is very little difference in the  three lower
temperature  ranges (from 1.1  to  23.4°C), in the range 23.5 to 3I.O°C, XDBPS  was
significantly  higher at a 95  percent confidence level.   The  same trend is observed in
Figure 5-75.  a plot of TTHMs by temperature range.  In the three lower temperature
ranges, median TTHMs varied between approximately 25  and 35 /ug/L, while in  the
highest temperature range, the median TTHM concentration was over 65 /ug/L.  The
same  trend was observed for HANs as a function  of temperature, as  shown in Figure
5-76, although the differences were not statistically  significant.

                                     5-17

-------
J
\
a
o


a.
o

a.
o

o
H

U
•H
 •.

H

 *
                 1»1-Trichloropropanone


                     By Effluent pH
               _L
                  D
                  a.

                  a
                  x
4.e-7.ee

  34

  13
                      7.68-8.23


                        33
                                14

                        pH Range



                        FIGURE 5-73
e. 7i-9. 70


  "

   3
                                 Sum of Halogenated DBPi


                                 By Influent  Temperaturi
       e.6-ie,«   ie.i-33.4  aa.e-3i.8

         1,7       44      27

 17       27       26      17

Temperature Range (°C)



         FIGURE 5-74

-------
\
a
3   "•
•

+J

E
a
•H
t
K
                   Trihalomethanes

               By Influent.  Temperaturi
                                                            ai
                                                         J
                                                         a
                                                         •H
                                                         L
                                                         +J
                                                         •H
                                                         c
                                                         •
                                                         u
                                                         0
                                                         «
              i.i-«.•
            n-  2)
            u-  17
                      a.*-ie.i
                              1C. 1-23. 4
                                      23.S-31.

                                       27

                                       17
                       "       L°     n  ''
               Temperature Range  (UC)


                        FIGURE 5-75
                                                                          Haloacetonitrilee

                                                                      By Influent  Tamperaturi
i.i-a.B    ».e-i«.«    ie.i-23.4

  21       47       44
  17       27       26
27
17
                                                                       Temperature Range (°C)

                                                                               FIGURE 5-76

-------
                    Baseline Sampling Results and. Discussion
 Brominated DBFs
 During  the  first quarter  of  baseline sampling,  a high  correlation was found between
 dibromoacetic acid (DBAA) and dibromochloromethane (correlation  coefficient r  =
 0.91).   In  addition,  relatively high levels  of  the  measured brominated  DBFs were
 detected at  some  utilities.   These  findings suggested  that  the  influence  of bromide
 present  in the raw water  should be evaluated; therefore, bromide and chloride analyses
 of the  plant  influents were added  beginning  with the second quarter  of baseline
 sampling.

 Among  the 35 utilities in this project, bromide levels in the plant influents ranged from
 <0.01 to 3.00 mg/L.  High bromide  levels were found in each of the three types of
 source waters.  Figure 5-19  presented the raw water bromide levels at the 35 utilities
 for the latter three  sampling quarters.

 Table 5-5  shows DBF data for a utility with high bromide levels  (Utility 12).  Not only
 was there  a  shift in THMs to the more brominated species at high bromide levels, but
 the same situation  existed for the HANs and HAAs.  Dichloro- and trichloroacetic acid
 (DCAA  and TCAA).  which  are commonly found  in  other DBF  studies  (Uden  and
 Miller, 1983; Norwood et al., 1986) as well  as this one, were detected at low levels at
 Utility  12 when  bromide levels  were  high;  instead.   DBAA  was the major HAA
 detected.

 Additionally, there were seasonal shifts  in the raw water bromide concentration at  this
 utility.  Such shifts were observed in some utilities as a result  of drought conditions  and
 saltwater intrusion  problems.  During the summer of 1988,  0.41 mg/L bromide was
 detected  at Utility  12.  In the fall of 1988 and  winter of 1989,  higher bromide levels
 were  detected (0.78 to 0.79  mg/L).  As Table 5-5 shows, the change in distribution of
 brominated DBFs  was  consistent  with  the  change  in  bromide  level.   For  example,
 bromoform and dibromoacetonitrile (DBAN) represented (on a weight basis) 36 and 67
 percent of the  sum of their respective class fractions  in the summer, when the bromide
 level  was  lowest.    These   percentages increased  to approximately 66  and  85,
 respectively,  in the fall  and  winter, when the  bromide levels were higher.   Similar
 results have  been documented by other researchers, such as Lange and  Kawczynski
 (1978). when investigating the impact of bromide concentration on THM speciation.

 When bromide levels at Utility 12 increased from 0.41  to 0.78 mg/L from  summer to
 fall,  1988. there was a shift  to a higher percentage of brominated DBFs, as discussed
 above.  This  shift is illustrated in Figure 5-77.   However, this figure indicates that  the
 shift for  the HANs and  HAAs was  less dramatic  than  that for the  THMs,  as  the
 majority  of  HANs and  HAAs were already brominated during  the  summer, 1988
 sampling.  However, bromoform only represented 36 percent  of  the THMs  in summer
 sampling, while it  increased  significantly (to 68 percent) by the fall.   One possible
 explanation is that  the kinetics for the formation of  DBAN and  DBAA may  be faster
 than  that for bromoform.   If this is the case, this could explain why  the correlation
coefficients for  DBAN and DBAA versus bromoform  (i.e., 0.88 and 0.82. respectively)
are not higher,  as the formation of the fully brominated species for each  DBF class may
proceed to completion at different kinetic rates.
                                     5-18

-------
                       TABLE 5-5

           DBF CONCENTRATIONS AT UTILITY
      WITH SEASONAL CHANGE IN BROMIDE LEVELS
	 : 	

Component
PLANT
Total Organic Carbon
Chloride
Bromide

Summer
1988
INFLUENT
2.6
111
0.41
Utility 12*
Fall Winter
1988 1989
VALUES, mg/L
2.2 2.8
215 202
0.78 0.79

Summer*
1989

2.9
44
0.14
CLEARWELL EFFLUENT VALUES, /ig/L
Chloroform
Bromodichloromethane
Dibromochloromethane
Bromoform
Total Trihalomethanes
Trichloroacetonitrile
Dichloroacetonitrile
Bromochloroacetonitrile
Dihromoacetonitrile
Total Haloacetonitriles
Monochloroacetic Acid
Dichloroacetic Acid
Trichloroacetic Acid
Monobromoacetic Acid
Dibromoacetic Acid
Total Haloacetic Acids
1 . 1 -Dichloropropanone
1,1,1 -Trichloropropanone
Total Haloketones
Chloropicrin
Chloral Hydrate
Cyanogen Chloride
4.7
13
28
26
72
<0.012
0.74
1.6
4.6
6.9
<1.0
2.9
1.6
1.0
14
20
0.36
0.24
0.60
0.018
0.53
2.3
1.4 0.86
7.5 6.5
25 24
72 53
106 84
< 0.029 < 0.029
0.24 0.19
0.96 1.4
7.0 11
8.2 13
<1.0 
-------
  100
  90 -
  80
I 7°
CO

§ 60

CL
CQ —
Q 50
i 40
LU

O 30

LU
D_
  20
   10
                           SUMMER   FALL   SUMMER

                            1988    1988    1989
       UTILITY 12 DISINFECTION BYPRODUCTS SPECIATION
                              FIGURE 5-77

-------
                     Baseline Sampling Results and Discussion


 When the bromide  level  at Utility 12 decreased to 0.14 mg/L in the summer of 1989,
 chlorinated  DBFs   began  to   predominate.     As  would  be  expected  from  the
 bromide/chlorinated DBF  exclusion  relationships,  the  highest  levels of  chlorinated
 ketones.  chloropicrin  and  chloral hydrate were  detected  during  the low bromide
 sampling period.

 The  high bromide concentrations and  associated impacts on  DBF speciation  observed at
 Utility 12 were the result of saltwater intrusion.  However,  high levels of chloride and
 bromide were measured  in waters not limited to coastal origins with modern saltwater
 intrusion problems.   A mid-South utility (Utility 10), which  is located inland, derives it
 waters from  two lakes,  one  of which is  high in mineral  content.   In  dry years,
 evaporation results  in an  increase in salinity, and in particular, chloride levels.  During
 the three quarters that  chloride and bromide were measured, levels ranged  from 561 to
 680  mg/L  and 2.8  to  3.0 mg/L, respectively.  In  the Midwest, high bromide levels
 were detected in  waters  from two utilities (Utilities 23  and  26).  Utility 23 had 47 to
 152  mg/L  chloride  and 0.44 to  1.19 mg/L bromide.  Utility 26 had  69 to 251 mg/L
 chloride and 0.19 to 0.68 mg/L bromide. As  Table 5-6 indicates, these utilities have a
 higher level of brominated  DBFs than  chlorinated ones, due to the presence of high
 bromide  levels.   Thus,  the presence of brominated DBFs  is not restricted to  coastal
 areas, such as in California or Florida, experiencing saltwater intrusion problems in
 their source waters.  According to Standard Methods (1989), the bromide  content of
 groundwaters  and stream base-flows  can also be  affected  by connate water (ancient
 seawater that was trapped in sedimentary deposits at the time of geological  formation).
 In addition, industrial  and oil-field brine discharges can  contribute  to the  bromide in
 source waters.

 In the 35-utility study, there was a very  good correlation between bromide and chloride
 levels (correlation coefficient r  =  0.97), as  discussed previously in this section.  Since
 the levels of these ions were atypically high  at Utility 10. the data were re-examined by
 excluding the  levels detected at  that utility.  The correlation was still high  (r = 0,86).
 As discussed  previously,  an  equation for predicting bromide levels from chloride levels
 was determined, excluding the  outlier  points from Utility 10:

                          [Br  ] = -0.0071  +  0.00341C1)

 This  equation  was  then  applied to  the data  from Utility  12, where  the source of
 chloride  and  bromide is due  to saltwater intrusion problems,  and  the  results are
 reported  in Table  5-7.   The  measured and  predicted  bromide levels agreed to  10
 percent, on the average.   Furthermore, the concentrations of chloride and  bromide in
 seawater are  18.980 and 65 mg/L. respectively (Sverdrup  et al.. 1942).   From these
 data, if chloride and bromide were only from sea water diluted with unsalty freshwater.
 then:

                                |Br | = 0.0034[C1|

 Despite  the similarity of  the equations,  there will be more variability in the prediction
when more than modern  sea water intrusion is  involved.   However, since the correlation
holds in general for the  35 utilities,  one should be able to predict a relative level of
bromide in  waters where  only chloride measurements were made.


                                       5-19

-------
                          TABLE 5-6

           DBF CONCENTRATIONS AT INLAND UTILITIES
                 WITH HIGH BROMIDE LEVELS*
Mid-South
Component Util. #10
Midwest
Util. #23
Midwest
Util. #26
PLANT INFLUENT VALUES, mg/L
Total Organic Carbon
Chloride
Bromide
CLEARWELL
Chloroform
Bromodichloromethane
Dibromochloromethane
Bromoform
Total Trihalomethanes
Trichloroacetonitrile
Dichloroacetonitrile
Bromochloroacetonitrile
Dibromoacetonitrile
Total Haloacetonitriles
Monochloroacetic Acid
Dichloroacetic Acid
Trichloroacetic Acid
Monobromoacetic Acid
Dibromoacetic Acid
Total Haloacetic Acids
4.9
561
2.9
EFFLUENT
0.59
2.9
9.2
40
53
< 0.029
0.24
I.I
6.7
8.0

-------
                               TABLE 5-7

          UTILITY 12: INFLUENCE OF SALTWATER INTRUSION
                                      Bromide Level
Sampling Chloride
Period mg/L
Summer 1988
Fall 1988
Winter 1989
Summer 1989
1 II
215
202
44
Measured
mg/L
0.41
0.78
0.79
0.14
Predicted*
mg/L
0.37
0.72
0.68
0.14
Percent
Difference*
9.8
7.7
14
0.0
•Percent difference = 100 x (measured - predicted)/measured.

'Prediction based on relationship derived from this study:
 [Br-| = -0.0071 + 0.0034|C1]

-------
                    Baseline Sampling Results and Discussion


 As indicated above, the production of THMs in those utilities with high bromide levels
 can shift  to  the more brominated  species.   The same phenomenon was observed for
 HAAs and HANs.   It is fortunate that  many  brominated DBFs were included in the
 study: otherwise,  the DBF  levels  of some utilities  would  be misrepresented.   Yet,
 research has shown  that  other brominated species exist which were not included in the
 analytical methods for this DBF study.  GC/MS analysis revealed the presence of HAAs
 containing both bromine and chlorine atoms (Slocum et al., 1987) as is observed with
 THMs; however, analytical standards for these compounds did not exist commercially.
 Research  also indicated the formation of brominated  trihalonitromethanes in a manner
 similar to the production of chloropicrin  (trichloronitromethane) (Thibaud et al., 1988).
 These findings emphasize the fact  that brominated DBFs,  not just chlorinated DBFs,
 are important in chlorinated drinking water.

 Aldehydes

 Results of aldehyde analyses of cleanvell effluents were discussed previously in  this
 section.  When ALDs were added to the list of analytes after the second quarter, and it
 became clear that ALDs were occurring in the effluents of many of the 35 utilities (not
 exclusively ozonating utilities), the  issue of whether these compounds were originating
 in the plant  influents or within  the plant still needed to be resolved.  To determine if
 formaldehyde and acetaldehyde were produced  by the disinfectants/oxidants  used at the
 plants or  if these compounds originated from the source water (e.g., from a biogenic
 process),  all  35 utilities  were sampled at the plant influents (preserved with mercuric
 chloride) during the fourth quarter of baseline sampling.

 Table 5-8 shows the aldehyde levels in the plant influents and effluents.  Formaldehyde
 was found in  16 of the 34 influents analyzed, at levels of 1.2 to  13 /ug/L.  The median
 level of formaldehyde in plant  influents for all  34 utilities  was  <1.0 /ug/L,  and the
 median level for only the 16 utilities  where formaldehyde was  detected in the influents
 was 2.8 /ug/L.  Acetaldehyde was found  in 12 of the 33 influents analyzed,  at  levels of
 1.1 to 16 /ug/L.  The median level  of acetaldehyde in  plant influents for all 33 utilities
 was < 1.0 /ug/L. and the  median level for only the 12 utilities where acetaldehyde  was
 detected in the influents was  2.0 /ug/L.

 At  the three  ozone  plants, it  is  clear that  formaldehyde  was  a product of  the
 oxidation/disinfection process.   Acetaldehyde was  found in the effluents  of all three
 ozone  plants; however, it was detected at a higher  level in the plant influent of Utility
 25.   The  other  plants shown  in  Table 5-8 employed  either free chlorine  only or
 chloramination (primarily with pre-chlorination).  At some plants, no formaldehyde or
 acetaldehyde was detected in the plant influent,  but these aldehydes were  detected in
 the effluent.  Where these aldehydes were detected in the influent, they were either at a
very low level compared  to that  detected in the effluent (e.g.. Utility 29, a chlorinating
 utility, had  2.0 versus  8.0  /ug/L  formaldehyde  in  the  plant influent and  effluent,
 respectively)  or at a level comparable to that detected  in the effluent (e.g., Utility  14
 had  6.4 versus 4.1 /ug/L formaldehyde in the plant influent  and effluent, respectively).
 From these limited  data, formaldehyde and  acetaldehyde appear to be present  because
of a  combination of  the  effects of plant  disinfection processes  and influent water
quality, the combination varying from one utility to  another.
                                      5-20

-------
                              TABLE 5-8

      ALDEHYDE LEVELS IN PLANT INFLUENTS AND EFFLUENTS*
Utility
Number
4
10
18
2
5
7
9
12
14
21
28
30
1
3
6
8
11
13
15
16
17
20
22
23
24
26
29
31
33
34
35
27
25
19
32
Disinfection
Scheme
NH2CI
NH2C1
NH2C1
CI2, NH3
CI2, NH3
CI2, NH3
CI2, NH3
CI2, NH3
CI2, NH3
CI2. NH3
CI2, NH3
CI2, NH3
CI2
CI2
CI2
CI2
CI2
c!2

Cl,
CI2
CI2
CI2
CI2
CI2
CI2
CI2
CI2
CI2
CI2
CI2
CI2. CI02
O3. NH2CI
0,. CI2
03. CI2
Formaldehyde, A/g/L
Plant Clearwell
Influent Effluent
ND
1.7
1.2
NA
4.9
ND
3.9
ND
6.4
3.0
2.7
ND
ND
ND
13
ND
3.4
ND
ND
ND
ND
ND
1.2
ND
1.4
ND
2.0
ND
5.8
ND
ND
1.4
3.2
ND
1.4
ND
ND
ND
NA
ND
ND
8.7
2.0
4.1
6.9
3.5
4.3
ND
ND
7.6
NA
2.0
1.1
ND
1.6
ND
ND
4.3
ND
ND
2.1
8.0
3.3
3.6
1.8
2.1
4.6
19
7.5
21
Acetaldehyde, //g/L
Plant Clearwell
Influent Effluent
ND
ND
1.9
NA
4.5
ND
4.4
ND
2.4
10
1.7
1.3
ND
ND
NA
ND
ND
2.1
ND
ND
1.6
1.4
ND
ND
ND
ND
ND
ND
ND
ND
ND
1.1
16
ND
ND
1.5
1.2
1.5
NA
3.4
1.8
6.1
1.6
4.5
8.5
3.0
2.2
ND
ND
NA
NA
1.8
1.3
ND
2.0
ND
ND
2.2
1.0
1.4
2.1
4.6
2.6
1.3
2.4
1.5
1.9
5.5
3.9
2.1
 *  Fourth sampling quarter.
ND = Not detected (< 1.0 //g/L).
NA = Not analyzed; analytical problem with sample.

-------
                    Baseline Sampling Results and Discussion


Table 5-9 lists the aldehyde levels at three ozone plants during baseline sampling and at
four plants that utilized  only chlorination or prechlorination/postammoniation  but had
high formaldehyde levels (j>IO /t/g/L during at least one sampling).   The latter four
utilities had 10 //g/L or  more of formaldehyde at a time when they had approximately
 100 /vg/L (or more) TTHMs.   These limited data suggest that  when  a  chlorinating
utility's water quality and treatment practices produce a high level of THMs, they can
potentially produce  a high level of formaldehyde as well.   As  will  be discussed  in
Section 6 regarding treatment studies where ozonation was implemented, there can be a
 "tradeoff" of halogenated  DBFs for aldehydes in implementing ozonation.  Levels  of
halogenated and  non-halogenated  DBFs in the  clearwell  effluents and  distribution
systems of ozonation  plants will be  discussed in Section 6. In addition,  the issue of the
formation and  removal  of aldehydes in ozonation plants will  also be discussed  in
Section 6.

Cyanogen Chloride Results

Typically, chloramines are used as  a means of lowering THM levels in treated waters.
Of the 142 cyanogen chloride analyses conducted for this  study (35 utilities for three to
four quarters), 32 samples  represented prechlorination/postammoniation  and another  11
represented  chloramines only.    As  seen in Figure 5-54 (presented previously), the
prechlorinating/postammoniating utilities had a significantly higher cyanogen chloride
median concentration (approximately  2 //g/L)  compared  to  either the  chlorinating  or
chloraminating  utilities  (both  of   which  had  medians of  less  than  0.5  /t/g/L).
Additionally     (not     shown    in    Figure    5-54),    one     utility    with
preozonation/postchloramination  had   a  median  cyanogen   chloride   level   of
approximately 7 fJg/L.   Since  some of these disinfection  schemes represent small data
sets, making it difficult to  interpret  the data, Figure 5-78, a plot of cyanogen chloride
by  final  disinfectant  (regardless   of whether  prechlorination/postammoniation   or
chloramination  were  used)  was generated.   As  seen  in  the figure,  the utilities that
deliver chloraminated water were demonstrated to have  a statistically  higher  level  of
cyanogen chloride.

Research in Japan has  shown  that  cyanogen  chloride  was  sometimes  formed in the
presence of certain ami no acids and  hypochlorous acid,  but  was  always formed in the
presence of the amino acids tested when both hypochlorous acid and the  ammonium ion
(i.e.. chloramines) were present (Hirose et al.,  1988).   Other  research (Ohya and
Kanno. 1985) found that cyanogen  chloride was formed by the reaction of humic acid
with hypochlorous acid in  the  presence of the  ammonium ion.  It was found that the
amount of cyanogen chloride was at a maximum when the reaction mixture contained a
ratio of  8 to  9 ppm of chlorine to 1 ppm of ammonia (as  nitrogen), and that the
maximum yield of cyanogen chloride increased as increasing amounts of hypochlorous
acid were added.  Furthermore, these formation patterns were reproduced with three
raw waters from Japan.   These data imply that  cyanogen chloride  may be more readily
formed in chloraminated  systems.

In separate research  endeavors, N-chloroglycine was formed  as a result  of the reaction
of monochloramine with  glycine under conditions typical for drinking water (Margerum
and Gray. 1978).  and it  has been suggested that the formation of cyanogen chloride is
caused by the reaction of glycine with chlorine (Kopfler et al.,  1975). Studies have
demonstrated the formation of  organochloramines by the  use of inorganic chloramines

                                      5-21

-------
                                     TABLE 5-9



         LEVELS OF ALDEHYDES AND THMS IN SELECTED CLEARWELL EFFLUENTS
Utility
Formaldehyde, pgIL Acetaldehyde, fig/L
Summer Fall Wlnler Summer Foil Winter
1988 1988 1989 1988 1988 1989
TTHMs, //g/L
Summer Fall
1988 1988
Winter
J989
Chlorine or Chloramine Plants:
2
9
26
29
Ozone
19
25
32
10 8.2 NR* 4.2 4.4 NR
12 6.2 8.7 4.1 2.8 6.1
17 8.6 2.1 6.0 7.1 2.1
NR 13 8.0 4.3 5.2 4.6
Plants:
5.8 10 7.5 4.8 5.3 3.9
31 22 19 15 9.9 5.5
30 24 21 3.5 2.8 2.1
90
95
164
109

15
34
3.1
82
54
100
180

20
16
1.4
60
40
98
259

5.9
9.0
0.72
*NR = not reported; analytical problem wilh sample.

-------
                  Cyanogen  Chloride

                By Final Disinfectant
     IB
    14 —
    13
      r
 01
 T3
 •H
r
o


Q)
01
D

ID
3
O
    "
    !•

                   CHLOHXNB


                    78
                                CHLOAAMXNeS
                Final Disinfectant


                       FIGURE 5-78

-------
                        Baseline Sampling Results and Discussion


    in the treatment of water (Scully and Bempong, 1982).  The cyanogen chloride results
    presented above,  as well  as  the presence  of moderate  concentrations  of TOX  in
    chloraminated waters with  low levels of TTHMs.  indicate a need to  further identify
    chloramine by-products.

    DBF Levels of Disinfection-Only Utilities

    Figure  5-35. presented  previously, illustrated  influent TOC  levels as a  function of
    treatment type.  DBF levels as a function  of treatment type have  not  been discussed
    further  in this section for the reasons presented previously in Section 5, under the
    heading  "Star Plot Analyses".   However,  some findings of this study with respect to
    DBF levels occurring in disinfection-only utilities warrant further evaluation.

   The overall median concentration of TTHMs  for direct filtration and disinfection-only
   utilities  were the lowest of the four treatment  types, although the differences were not
   statistically  significant.   However, the  median level  of HAAs in disinfection-only
   utilities   (approximately   35   //g/L)  was   significantly  higher   than  that   of either
   conventional  or direct filtration utilities (approximately 17 and  9 /ug/L, respectively);
   and this difference  was significant at a 95 percent confidence level.  Additionally, the
   median concentration  of HAAs for softening utilities was only approximately 23 //g/L,
   although  this was  not significantly  lower  than the  HAA median for disinfection-only
   utilities.    This trend was even more pronounced  when  comparing the median HK
   concentrations of disinfection-only  utilities  (approximately  3.8  //g/L)  with those of
   conventional, direct filtration and  softening utilities (all of which occurred  within the
   approximate range  of  0.8  to 1.4 //g/L), and the differences were  statistically  significant
   at a 95 percent confidence level.

  Samples  from the  disinfection-only  utilities  were  collected  very  shortly   after  the
  addition of the disinfectant (in all  cases, free chlorine, although one disinfection-only
  utility also employed  preozonation).   There  was  very little free  chlorine contact time
  before sample collection since none of these  utilities had detention time in a clearwell
  after the addition of chlorine.

  The role of HKs as reactive intermediates rather than stable endproducts was discussed
  previously.   The HK data  from the disinfection-only utilities  seem to suggest that this
 class of  compounds  are  formed  very  rapidly  upon  chlorination, and there  was
 insufficient  contact time before sampling for the reaction to proceed further toward  the
 formation of chloroform.   A factor that may influence  HAA formation in disinfection-
 only utilities is that the disinfectant is  applied directly to the organic material present in
 the raw water,  before  the  precursor  material has been lowered  in  concentration, or
 altered in  molecular weight distribution or  other characteristics,  by the addition of
 treatment chemicals, filtration or other treatment  processes.

 Removal of TOC During Treatment

 Figure 5-79 illustrates the removal of TOC within the filtering plants included in this
 study.  The percent  removal of TOC from  the  plant influent to the filter influent  of
each  plant was calculated, and the mean value of  percent removal was plotted in the
figure.   In  a  similar manner,  the removal from  the filter influent  to  the clearwell
effluent,  and from the plant influent to the clearwell effluent were also calculated and

                                      5-22

-------
 O
 0)
DC
U
O
c
a
O
^
0)
CL
           Mean  Tota   Organic  Carbon   Hernova
                 Through Filtering  Utilities'  Processes
       40
       3O
20
10
        o
       PI    Plant Influent
       Fl    Filter Influent
       EFF   Clearwell  Effluent

       Mean  PI  TOG = 4.O mg/L
       Mean  EFF  TOG = 2.8 mg/L

           2O.9
                PI TO Fl
                         Fl TO EFF

                       FIGURE 5-79
   23.9
PI  TO EFF

-------
                       Baseline Sampling Results and Discussion


   are  shown  in the  figure.    The  plot indicates  that  most of the  TOC  removal
   (approximately 21 percent) occurred prior  to filtration, that is, by sedimentation, and
   only 5 percent of the remaining TOC was removed by filtration.  Overall, TOC removal
   within the filtering plants averaged approximately 24 percent.

   The  data shown in Figure 5-79 reflect the  TOC  removal achieved by  the participating
   utilities,  where much  of the  process  operation  is most likely  focused on  turbidity
   control.  In Section 6. results  are presented for two utilities that were able to achieve
   higher TOC removals than the mean values noted in Figure 5-79.  (It  should  be noted
   that  for one of the two  utilities which were capable of increasing TOC  removal  by
   increasing alum doses,  the enhanced  precursor removal  was only  an  incremental
   increase over that achieved in the baseline sampling.)  From the data collected for this
   study, it  is not possible to draw conclusions as to whether or not the levels of TOC
   removal illustrated in Figure 5-79 can  be improved under all or most circumstances by
   optimizing the  coagulation process.   Further research in  this  area is required, especially
   since  the TOC in some surface  waters does not  appear  amenable  to removal  by
   conventional treatment (Chadik and Amy,  1983).

  TOC removal by coagulation may be further optimized  by controlling coagulation pH,
  although many  utilities do not have the capability  to control pH at the rapid mix basin.
  Of  course, the  economics of  increased  chemical  consumption and associated sludge
  production, as well as increases in  total dissolved  solids  due  to acid and base addition,
  must  be weighed against the advantages of enhancing precursor removal by increasing
  coagulant  doses and adjusting the coagulation  pH.  The coagulation  studies will be
  discussed in detail in Section 6 of this report.

  Comparison of  THM Levels from USEPA DBF Study and AWWARF THM Survey

  The median  TTHM levels previously reported in  Table 5-2  were 34, 44, 40  and  30
  //g/L for the spring, summer and fall quarters of 1988 and the winter quarter of 1989.
  These data were compared to the TTHM values obtained in a survey of 727  utilities
  around  the  United States conducted  for  the  American Water  Works Association
  Research Foundation (AWWARF) in  1987 (McGuire and  Meadow, 1988).  The  median
  TTHM concentrations in the AWWARF survey for the spring, summer,  fall and winter
  seasons were 40, 44.  36 and 30 //g/L,  respectively.  (The AWWARF survey reflected
 more than 67 percent of the population represented by water utilities serving more than
  10.000 customers.)

 Because of the similarity of TTHM levels for the  DBF study and the  AWWARF survey,
 the data were further  evaluated.   Compliance with the THM  regulation  is based on a
 running annual average for each utility  (USEPA, 1979);  therefore,  mean  values were
 computed for each of the 35 utilities  for the  four sampling quarters.  The AWWARF
 survey  utilized the means of three years of quarterly data.  The means for both projects
 are plotted  on  a  frequency distribution curve in Figure 5-80. A log scale was used for
 the ordinate axis  in  order to compress the displayed range, not to imply a log-normal
 distribution  Visually, the 35-utility DBF study appears to represent  a THM frequency
 distribution  very   similar to that  of  the 727-utility  AWWARF survey.   The  major
 difference is in the data for low TTHM  levels (less than or  equal  to 25  //g/L).  As
 discussed in Section 2 of this report,  utility selection for this study attempted to achieve
a balance among  utilities which  had reported  low,  medium and high TTHM levels in

                                     5-23

-------
       FREQUENCY DISTRIBUTIONS OF
              THM SURVEY DATA
O)

CO
til
O
300

200



100



 50




 20



 10



 5
          USEPA/CDHS
          DBP SURVEY
        MEANS OF 4 QTRS
             (n = 35)
                     AWWARF
                    THM SURVEY
                      (n = 727)
        _L
        I  I  I  I I  1  |
I  I
        10    30  50   70    90 95   99    99.9  99.99

      PERCENT LESS THAN OR EQUAL TO GIVEN CONCENTRATION
                   FIGURE 5-80

-------
                    Baseline Sampling Results and Discussion


the AWWARF survey.  Thus, fewer utilities with very low TTHM levels were included
in the  DBF study  than  are  found nationwide, and this may  be the cause of  the
differences in the data below 25 /ug/L of TTHMs between  the two surveys.  A statistical
comparison of the  two  distributions, by means of a Kolmogorov-Smirnov test (Hoel,
1971). indicates  that the hypothesis that these samples are from the same distribution is
not rejected at a significance level of 0.01.

In addition, a notched box-and-whisker plot of the TTHM levels from the two surveys
is shown in Figure  5-81.  Both sets of survey results have median TTHM values of 39
//g/L. and  the 95 percent confidence intervals for the two medians overlap, indicating
that the two groups are statistically similar in terms of central tendency.  Furthermore,
their minima and maxima  are comparable, indicating a similarity in variability.

It should be noted  that  the AWWARF  survey reported distribution system THM  data,
whereas results of this study represent clearwell effluent THM levels. THM levels from
the clearwell effluents of utilities employing  chloramines as  a final disinfectant may
reflect more closely the  THM levels found in their distribution systems.

Comparison of USEPA  Study and CDHS DBF Study Results

Figure 5-82 illustrates the levels of XDBP   for the 25  utilities participating in  the
USEPA study  and the 10  utilities in the CDHS study, as  well as for the combined 35
utilities.  The four-quarter median level of XDBPsum was substantially  lower for  the
CDHS study,  approximately 48  //g/L. compared to approximately 75  /vg/L for  the
USEPA study, although the difference was not statistically significant  at a 95 percent
confidence  level.    Additionally,  the  75th  percentile values  from  each  study are
approximately  equal, as  are the ranges.
                                     5-24

-------
J
?
c
j
j
 I
•H
 L
            AUUARF 12-Quart«r  u». USEPA
                   4-Quart«r Maans
              Total Halaganated DBPi
                      By Study
                                                        a
E
M
Q
X
                                                                                  1
                                                                        _L
                                                                                                     _1_
                     727
                                 ,,-
                                 "
             n=
             u=
EPA
100
 25
                                                                                  40
                         Study

                       FIGURE 5-81
                          10
                        Study

                       FIGURE 5-82
ALL
140
 35

-------
Section 6
Treatment Modification
Studies • Results and
Discussion

-------
                                  SECTION 6

   TREATMENT MODIFICATION STUDIES - RESULTS AND DISCUSSION
 In  this section, results of the treatment modification  studies will  be  presented  and
 discussed.  A total  of ten studies were performed for the combined USEPA and CDHS
 DBF projects.   Five of these studies focused on me use of ozonation as a method of
 DBF control, two  investigated  coagulation for DBF precursor removal, two  studies
 evaluated chlorine dioxide for DBF control, and one involved the use of GAC to lower
 concentrations of DBF precursors.

 OZONATION STUDIES

 Because of the increasing use of ozone in the United States for disinfection and  control
 of  DBFs, five treatment  modification studies  focused  on  the use  of this oxidant at
 various water treatment  plants.   The  utilities which participated in this project were
 selected because ozonation processes were installed at their treatment plant or because
 they were able  to  provide a pilot  plant  which  could evaluate ozonation.   Table  6-1
 shows the  type of source water and  various disinfection schemes  employed at each
 utility in order to evaluate the effect of ozone on the formation of DBFs.

 Utility 6

 Utility 6 operates a plant  in  the eastern United States which had a capacity of 52 mgd
 at the time of the study.  Sampling at this plant was conducted before and after ozone
 was incorporated into the treatment process, which is presented schematically in Figure
 6-1.  As a conventional treatment plant, chlorine was added in four places:  to the  raw
 water, before flocculation, and before and after filtration.   Ammonia was also added
 after filtration to convert the  free chlorine to chloramines for residual disinfection.  As
 the  schematic  shows,  there  was  a change  to  direct  filtration when ozonation  was
 installed,  the  flocculation   and  sedimentation  basins having  been  converted   to
 flotation/skimming  tanks.    As a  result  of  changing   the  plant  configuration  and
 implementing ozone, the utility was able to employ lower doses of chlorine for  shorter
 contact times.   Under  this  configuration,  chlorine was applied  before  and after
 filtration, ammonia being added before  the  second  chlorination   point to produce
 chloramines for residual disinfection.  Therefore, free chlorine contact time was  limited
 to that required to prevent biological growth in the filters.

 Sampling locations  included  distribution  system  Location 3,  which  corresponds to a
 residence time of 7 hours.  Because of delays  in the switch  to ozone and  changes in
 water quality, the "before" and "after" samples (before and after the implementation of
 preozonation) were taken approximately four to six months  apart.   The first  "after"
 sample  was  collected  on March 13, 1989.   However, as indicated in Table 6-2,  the
 temperature  at Location  3 was  7°C lower as  compared to the November  21, 1988
 conventional  treatment sampling at that location.  The  raw  water TOC concentration
was  also lower in  March  compared to November,  Consequently, a second ozonated
 "after"  sample was collected on May  15,  1989 when  the  distribution system water
temperature  was comparable  to that of the November sampling.  Additionally, Table
6-2  shows that the samples collected on this sampling date and on November 21, 1988,

                                      6-1

-------
                                     TABLE 6-1

            DISINFECTION SCENARIOS, PLANT SCALES AND SOURCES

                     FOR UTILITIES PARTICIPATING IN STUDY
                    Source
Utility     Scale     Water
                    Disinfection Scenarios
                                                      Abbreviations
    19


    25
           Full    Reservoir
           Full/
           Pilot    Reservoir
Full     Flowing
        Stream
Full
Lake
1) Prechlorination. postammoniation           C12. NH3
2) Preozonation. postchlorination
     postammoniation                        O3. CI2, N

I) Pre- and postchlorination (full scale)        CI2
2) Prechlorination, postammoniation           C12, NH3
     (full scale)
3) Preozonation. postchloramination           O3, NH2C1
     (pilot scale)

1) Pre- and post chlorination                 CI2
2) Preozonation, postchlorination              O3, Cl,

1) Prechloramination only                    NH2C1
2) Pre- and ppstozonation,  post
     chloramination                          O3, NH:CI
3b Pilot Lake
1)
2)
3)
4)
5)
Prechlorination only
Prechloramination only
Preozonation. prechlorination
Preozonation. prechloramination
Preozonation with hydrogen peroxide
addition, prechloramination
CI2
NH2CI
03, CI2
03. NH,CI
03. H2d2.
NH:CI
Note:   The prefix  "pre-"  in disinfection  scenarios  denotes  addition prior to the rapid mix.  The
        prefix "post-" denotes addition directly before or after filtration.

-------
           OZONE
             RAPID MIX  FLOCCULATION  SEDIMENTATION   FILTRATION
 PRECHLORINATION / POSTAMMONIATION
                             Chlorine
                t
              Chlorine
               Alum
              Polymer
                                                           Chlorine
                                                Sodium Hydroxide
OZONE/CHLORINE / CHLORAMINES
                                                          Chlorine

 Alum
Pol \rner
000000
 0000
0000000
                                                   Ammonia
                                                   Chlorine      ^
                                                Sodium Hydroxide
         SCHEMATIC OF UTILITY 6 TREATMENT PROCESSES
                            FIGURE 6-1

-------
                                              TABLE 6-2

                                  UTILITY 6 TREATMENT STUDY
                                      Water Quality Parameters






TOC
(mg/L)
CONVENTIONAL (11/21/88
Raw Water
Location 3
4.25
2.84
UV
Absorbance


at 254 nm Chloride
(cm1)
DBF Sampling)
0.178
0.102
(mg/L)

43
NA


Bromide
(mg/L)

0.05
NA


TOX
0>g/L]

NA
240




pH
1

7
7


.8
.6
Free
Chlorine
Residual
(mg/L)

NA
NA
Total
Chlorine
Residual
(mg/L)

NA
>2.0


Temp.
(°C)

9
13
OZONE (5/15/89 DBF Sampling)
Raw
Location 3
4.28
3.60
OZONE (3/13/89 Aldehyde
Raw Water
Ozone Contactor
Effluent (OCE)
Filler Influent (FI)
Filter Effluent (FE)
Plant Effluent (PE)
Location 1 (LI)
Location 2 (L2)
Location 3 (L3)
NA = Not Analyzed
ND = Not Detected
Note: I. All compa
3.75

3.72
3.64
3.18
3.20
3.18
3.23
3.20


0.144
0.059
Profile)
0.137

0.093
0.092
0.070
0.069
0.071
0.072
0.070


risons of conventional
66
NA

66

NA
NA
NA
NA
NA
NA
NA


0.06
NA

0.07

NA
NA
NA
NA
NA
NA
NA


NA
100

NA

NA
60*
76*
80*
82
76
89


versus ozone treatment are
7
7

7

8
7
8
8
8
8
8


.5
.8

.7

.0
.6
.0
.0
.1
.1
.2


made
NA
ND

NA

NA
0.5
0.1
NA
NA
NA
NA


with
NA
2.0

NA

NA
1.0
0.5
1.0
0.7
0.8
0.4



14
15

6

5
5
5
5
7
6
6



         samples  collected on  11/21/88 and 5/15/89.  repectively.  The  data on
         these two sampling dates represent the water quality parameters associated
         with Figures 6-2 to 6-9.
      2.  Locations  represent  distribution  system  sampling  points in order  of
         increasing residence time.
*These TOX samples were dechlorinated and preserved in the field.

-------
                 Treatment Modification Results and Discussion


 had similar raw water TOC and pH.  Consequently, all comparison of "before" and
 "after" oronation samples were made using data from these two sampling dates.  Table
 6-2  shows that the total chlorine  residual at Location 3 with preozonation  (May  15,
 1989  sampling)  was  2.0  mg/L.    Before  ozone  was  implemented,  the  measured
 distribution system  residual  was greater than 2.0  mg/L; historical records showed that
 the  residual at this point was usually  2.8 mg/L.  The table also  indicates  that TOX
 decreased by 58 percent with the preozonation  disinfection scheme, despite the  higher
 concentration  of  TOC measured in the  distribution system.  However, it  should be
 noted  that  less  free  chlorine  and a shorter contact  time were employed  when
 preozonation was utilized.

 The effects of ozonation on  DBFs measured at Location 3 is presented in Figures 6-2 to
 6-9.   As shown  in Figures 6-2 through 6-4,  a reduction  of 56  to  66 percent was
 observed for all XDBP classes except  HKs, which  increased slightly from  3.6  to 4.4
 j/g/L.  Chloral hydrate, chloropicrin and cyanogen chloride decreased slightly.  Figures
 6-5. 6-6 and 6-7  show that  most of the individual THMs,  HAAs. and HANs decreased
 in concentration.   Because  of the low bromide levels in the raw water (0.05 to 0.06
 mg/L), there were no observed  increases in  concentrations of brominated DBFs.  The
 only individual DBF compound  to  increase  in concentration was l.l-DCP (1.4  to 2.8
 */g/L). as shown  in Figure  6-8.  Concentrations  of  the miscellaneous  compounds  are
 illustrated in Figure 6-9.

 The effect of water  temperature on formation of DBFs after ozonation was implemented
 is shown in Figure  6-10.  Samples were collected when the temperature was 6°C and
 approximately  2  months  later when the temperature had  increased   to I5°C.   The
 residence time at  the distribution sampling point was  7 hours. The  data show that DBF
 class totals increased only  slightly with the 9  degree increase in water temperature.

 From this study, it  is  not  possible to directly attribute the  reduction in  most  XDBPs to
 ozone  since lower  doses  of chlorine and shorter contact times were employed.   It
 cannot be  determined  if ozone  caused a decrease or modification of  DBF precursor
 material  or if lower DBF levels can be attributed solely to a decreased  use of chlorine.
 Examination of Figure 6-6 shows that the use of preozonation affected individual HAAs
 differently (50 percent reduction  in DCAA versus a 78 percent decrease  in TCAA). For
 this  study, the following order of increasing effect of preozonation on controlling  the
 formation of these DBFs was:

            I, I, I-TCP <  DCAA < TOXj< CHC13 < DCAN < TCAA

 According to bench-scale research  on preozonation/postchlorination  treatment of fulvic
 acids by Reckhow  and Singer  (1985),  the  following order of increasing  precursor
 destruction for nearly all of the investigated doses were:

            1.1,1-TCP <  DCAA 
-------
       EFFECT OF  VARIOUS DISINFECTION SCHEMES ON
  DISINFECTION  BY—PRODUCT FORMATION AT  UTILITY  6
                                 FIGURE 6-2
                OF  VARIOUS  DISINFECTION SCHEMES ON
DISINFECTION  BY —PRODUCT  FORMATION AT  UTILITY  6
                                                                    /•g/L
                                 FIGURE 6-3

-------
  c
  0)
 U
 a
 m
 a
                  Effect  Of  CI2.NH2CI  and  O3.CI2.NH2CI
                         on  DBP  Formation  (Utility  6)
          15O
          100
50
            o
                XDBPsum   THM
                                               CI2.MH2CI
                                               (1 V/21/88J
                                               O3.Q2MH2CI
                                               (5/15/69)

                                            Dist. Sys. Residence Time (t)
                                             t = 7 nrs
                                             * TCAN Mow
                                            »* IwdAA twlow t
                                                                  NA
                                    HAN
                                  HK
HAA
          40
                                                               ALD
                                      DBP  Class

                                  FIGURE 6-4
                 Effect  Of  CI2.NH2CI  and  O3.CI2.NH2C
                        on   THM  Formation  (Utility  6)
0
U
          30
          20
          10
                                   CI2.NH2CI
                                   (1 1/21/88)
                                   O3.CI2MH2CI
                                   (5/ 1 5/89)
                                          Dist. Sys. Residence T.me (t)

                                                t = 7 nrs
                  '"HC.I j
                   CHBfQ2
                                            CHBr2Cl
     CHBr3
                                   1 HM  Compound

                                 FIGURE 6-5

-------
 O)
 o
 c
 
-------
 oi
ro
i.
c
 O
 u
           4.O
           3.O
           2.0
           1.0
          O.O
                  Effect  Of  CI2.NH2CI   and  O3.CI2.NH2CI
                          on  HK  Formation  (Utility  6)
                        1.1-DCP
                                             CI2JSH2CI
                                             (1 1/21/88)
                                             03.02^20
                                             (5/15/89)
                                         Dist. Sys. Residence Time (tl
                                                t= 7 trs
                                                 1.1.1-TCP
                                    HK  Compound

                                   FIGURE 6-8
S1
O
U
          20
          15
          10
                 Effect  of  CI2.NH2CI  and O3.CI2.NH2CI
                    On  Misc. DBP  Formation  (Utility  6)
                   CUP
                                              » B«low MRL
                                                 CI2^H2CI
                                                 11 1/21/88)
                                                 O3.CI2AH2CI
                                                 (5/15/89)
                                             Disl System Residence Tim* (t)
                                                   t= 7 frs
                                 CH
                                            CNCI
TCP
                                  C'BP  Compound

                                  FIGURE 6-9

-------
g
•*-*
rtJ
(J

Q.
                   Effect  Of  Seasonal  Temperatures

                      on DBP  Formation  (Utility  6)
        150
         100
50
           o
                                           03.CI2MH2CI

                                           3/13/89 6 C

                                           O3.CI2M-I2CI

                                           5/15/89 IS C
                               Oist. Sys.

                                t = 7 hra



                               MA - Not Analyzed
                                                    Tinr» 
-------
                 Treatment Modification Results and Discussion


Stability of Aldehydes Through the Plant and Distribution System.  The effect of
ozone on  the formation of aldehydes is  presented in Figure 6-11.  No  formaldehyde or
acetaldehyde was detected in the plant  influent.  Immediately after ozonation, though,
concentrations of formaldehyde and acetaldehyde increased to approximately  15 and 5
//g/L.  respectively.    The  plant  and  distribution  system  profile  shows  that  these
compounds  were stable after formation.   As  noted above,  chlorine  is added in the
treatment  process before  filtration; a 0.5 mg/L free residual (1.0 mg/L total residual)
was  detected in the filter influent.   As  will  be discussed  in  more detail  below, the
maintenance of a disinfectant residual  through the filter most likely  precluded the
aldehyde-removal capabilities of the filter.

Utility 7

Three different treatment scenarios were  studied at Utility 7 (Figure 6-12), which is a
400-mgd   conventional   treatment   plant.     Chlorine-only  and   prechlorination/
postammoniation options  were studied at full scale. A chlorine dose of 2.3 mg/L was
applied at the plant influent.  Additional  chlorine was dosed at the filter influent:  I.I
mg/L during the chlorine-only test and 0.6 mg/L during the postammoniation option.
Chlorine-only DBF  samples  were  collected   from the  filter effluent  just  prior to
postammoniation;  the  prechlorination/postammoniation samples  were taken just  after
ammonia  (0.49 mg/L  as nitrogen)  was added to the  filter effluent.    There  were
approximately 4 hours  of free chlorine contact time under this scenario.   In order to
study preozonation at  this utility,  a 6-gpm pilot plant was employed.  The pilot  plant
followed similar treatment processes  as the full-scale plant but included ozone, which
was  applied at a dose of 2.0 mg/L.  The pilot filter effluent was chloraminated (0.5
mg/L ammonia as nitrogen.  1.5 mg/L chlorine) with no free chlorine contact time.

Samples for each scenario  were collected after 2  and 24  hours using the simulated
distribution  system (SDS) protocol.  The  SDS  test  was described in detail  in Section 3
of this  report.   In  these  tests, samples were  dosed with disinfectants and held  under
conditions  representative  of Utility 7's  distribution system  to  give an estimate of the
levels of DBFs that can be formed under realistic environmental conditions (Koch et al.,
1989).  For the SDS tests performed at  Utility  7, samples were incubated at 25°C  for 2
and  24 hours  to  simulate  samples from the  clearwell  effluent  and the  distribution
system, respectively.   The preozonated pilot-plant sample was buffered to pH 8.2 to
8.3.   The chloraminated filter effluent sample  had  received final pH adjustment in the
treatment  plant.   However,  the  chlorinated filter  effluent sample was  collected at a
point in the plant prior to final pH adjustment, so  the SDS sample was raised to  a pH
of 8.2  by the  addition of  a 1-percent sodium hydroxide solution.  The  latter two
samples  had sufficient  chlorine  or  chloramine  levels  for the  SDS testing.   The
preozonated pilot plant effluent required  a  1.5 mg/L  chloramine dose, with  ammonia
added prior to chlorine dosing.

The  chlorine  residuals  at  the  sampling  locations,  as  well as  other  water quality
parameters,  are shown in  Table 6-3.  The data indicate that the pH and temperature at
the sampling points for  all  the  disinfection schemes were similar.   The pilot  plant
received the same  raw  water as  the full-scale  plant, so the observed difference in
treated-water TOC levels was probably due to a slightly greater TOC removal during
the ozone-treatment trial.
                                      6-3

-------
   OZONE     RAPID MIX    FLOCCULATION  SEDIMENTATION
FILTRATION
 CHLORINE ONLY
 Polymer
 Chlorine
     T
   Chlorine
    Alum
   Polymer
 PRECHLORINATION/ POSTAMMONATION ONLY
                                                       Polymer
                                                       Chlorine
   T
  Chlorine
   Alum
  Polymer
                                                 Ammonia  —
                                              Sodium Hydroxide
OZONE / CHLORAMINES
            Alum
           Polymer
                                                          Polymer
000000
0000
0000000
oomocnoo
.

o
o



s

V

k

•s,







                                                   Ammonia
                                                   Chlorine
            SCHEMATIC OF UTILITY 7 TREATMENT PROCESSES

                              FIGURE 6-12

-------
                              TABLE 6-3

                     UTILITY 7 TREATMENT STUDY
                         Water Quality Parameters


TOC

-------
                 Treatment Modification Results and Discussion


Effects of ozonation on DBFs at  Utility  7  after the 24-hour SDS tests are shown in
Figures 6-13 and 6-14.   Preozonation followed by concurrent addition of ammonia  and
chlorine  after  filtration  decreased the levels  of  THMs,  HAAs, HANs,  and chloral
hydrate as compared to chlorination-only or prechlorination/postammoniation.  Very
little  difference in concentrations of  HKs,  cyanogen  chloride or  chloropicrin  was
observed between any of the treatments.

Aldehydes  were not analyzed  at  the time  of the study.   However,  subsequent pilot
testing at Utility 7 indicated  formation of aldehydes  in the ozone contactors, the  levels
of which (1)  remained  the same through  the filtration  step when  chlpramines were
applied upstream  of  the filters, and  (2) decreased  through  the filtration step  when
chloramines were  not added  until after filtration (Montgomery, 1989),  as shown in
Figures 6-15 and 6-16 for formaldehyde.  Since the ozone residual is short-lived, when
no chlorine or chloramines were added prior to filtration,  it is probable that biological
activity had  devejoped on  the filter media;  the data suggest that  this  activity  was
capable of removing  biodegradable  material produced  by ozonation, such as  the  two
aldehydes studied  here.  This  is consistent with other  studies which have shown  that
biologically operated  filters can decrease materials such as assimilable organic carbon
(Van der Kooij et al., 1982; Montgomery, 1989).

The use  of filtration with biological  activity is typical of many European ozone plants;
however, it is  atypical  in water treatment  plants in  the  United States.  As the plant
schematics  for the other ozonating utilities in this study  indicate (see Figures 6-1, 6-23,
6-32 and 6-46),  chlorine or chloramines  were added  after the ozone contactor  and
before the  filters, thus probably minimizing microbiological growths  in the filters.  An
important  finding of the  ozonation studies  conducted for this  project  is the large
increases observed in  aldehyde formation whenever ozonation is employed followed by
secondary disinfection prior to  filtration (see Figures 6-33 and 6-47).  The placement of
secondary  disinfection  with  respect  to filtration requires further study  in  order to
minimize aldehyde levels in the finished waters of ozonating plants.

Table  6-3  shows  that  24-hour SDS TOX  concentrations decreased as  free  chlorine
contact time decreased  in each successive treatment train, the lowest (38 /wg/L)  being
observed when ozonation/chloramination  (with  no  free  chlorine contact time)  was
tested.

Figures 6-17 to 6-22 present the  DBF data for the treatment modification study as a
function  of contact time.  Results from  SDS tests at 2 and 24 hours are shown. Under
the chlorine-only scenario, all DBFs increased with increasing residence time except for
the haloketones and chloropicrin.  Under the prechlorination/postammoniation  scheme,
all DBFs were relatively  stable  between  the 2-hour  and 24-hour  tests, except for
1.1,1-TCP   which  decreased  and  cyanogen  chloride  which  increased.     For  the
preozonation/postchloramination treatment scenario, most DBFs  remained  at or below I
to 2 //g/L each, except for cyanogen chloride,  which was detected at 2.3 to 3.4  vg/L.

Utility 19

This  utility,  which  was  studied   at  full   scale   (Figure   6-23),  is   a  600-mgd
preozonation/direct filtration   facility.    During  the  trial when only  chlorine  was
employed,  ozone was taken  offline  and pre-  and post-chlorine doses were applied to

                                      6-4

-------
            EFFECT OF VARIOUS DISINFECTION SCHEMES ON
      DISINFECTION  BY-PRODUCT FORMATION  AT  UTILITY
                                              FIGURE 6-13
                                                                       •Includes -1 .'lours of frw? i-hloi ii:.
                                                                       contact t Line l>etorcj ijostJJinnm it
         EFFECT OF  VARIOUS DISINFECTION SCHEMES  ON
   DISINFECTION  BY —PRODUCT FORMATION AT  UTILITY  7
*g/L
                                                                    *Fncludes 4  hours of free chlorine
                                                                    contact tune before [jostonrnonuit ion
                                           FIGURE 6-14

-------
 3

 9
 LU
 Q
QC
O
u.
           25
           20
15
           1O
               CHANGE  IN  FORMALDEHYDE  CONCENTRATION
                        THRU PROCESS TRAIN.  STATE PROJECT WATER
                               Reference:  Montgomery.  1989
            O
      GUIDE  TO FIGURES:

      PIA = plant influent
      OE3 = ozone  contactor  effluent
    .  PI  = Mlsr influent
      FE  - filter effluent
   Disinfectants
Upstream  of  Filters

+	1- 2  mg/L O3.
       no NH2C.I
A	A 2  mg/L O3.
        1.Smart- N-I2CI

O	O2  moA. 03.
                     PIA
                        OE3
                                 PL AM I  LOCATION

                                     FIGURE 6-15
LU
Q

a
O
LL
               CHANGE  IN  FORMALDEHYDE  CONCENTRATION
                       THRU PROCESS TRAM COLORADO  RIVER WATER
                               Reference: Montgomery,  1989
                GUIDE  TO  FIGURES:
                PIA = plant influent
                O63  • ozone contactor effluent
                Fl  = filter riflueni
                FE   =  tiller  effluent
                                                                    Disinfectants
                                                                 Upstream  of Filters

                                                                  +	1- 2  man. O3.
                                                                         no NH2CI
           10  -
            5  -
            O
                                                                             o-
                                                                      - A 2 mo/L O3.
                                                                         1.5moA_ N-CCI

                                                                      -O2 mr>L O3.
                                                                         1.5mp/l- N-GCI
                     PIA
                        OE3
                                 PLANT  LOCATION

                                    FIGURE 6-16

-------
                Effect  Of  CI2.  CI2/NH3  &  O3/NH3/CL2
                       on  DBP  Formation  (Utility  7)
"bi

o
HM  Compound
                                                   1    2
                                                   CH&3
                                   RGURE6-18

-------
                 Effect  Of  CI2.  CI2/NH3  &  O3/NH3/CI2
                        on  HAN  Formation (Uti ity  7)
0)
o
o
U
           O
                    CI2 only
                 rTTI O3/N-I3/CI2

                 Dist. Sys Residence Time (t)
                 1: t = 2 hrs
                 2: t = 24 hrs
                 » = Below MRL
                 1    2
                   TCAN
                      1     2

                       DC AN
                                               SCAN
                                   HAN Compound

                                    RGURE6-19
Ol
o
o
U
          1.5
          1.0
                 Effect  Of  CI2.  CI2/NH3  &  O3/NH3/CI2
                         on  HK  Formation (Utility  7)
O.5
          o.c
                     Dist. Sys. Residence Time (T)
                     1: t = 2 hrs
                     21  t = 24 hrs
                      I.I  DCP
                                   HK Compound

                                  FIGURE 6-20

-------
               Effect Ot  CI2. CI2/NH3  &  O3/NH3/CI2
                      on  HAA  Formation  (Utility  7)
1^

1O

3
i 8
*-*
(D
t 6
O
S
U 4
<
<
2


,%






_



~
|!
g |/ fte
g|/ ^J






;
j
^

J3
x








r

^
•'
1 »
12 12
MCAA DCAA

B??3 CI2 only
^ C.2/^3
EZ] 03^3/02
Oist. Sys. Residence Tune (0
I
'•
;
1
'/
\
\
^
'


;
f
j
\
/i ^
3_l
1: t = 2 nrs
2: t = 24 hrs p
%





^ ra^ |3,
d_ ^^1 ^fe






|
|
!

,
3 *







-T|
3
2 12 12
TCAA MBAA DBAA
                               HAA Compound

                             FIGURE 6-21
u
5
u
               Effect Of  CI2. CI2/NH3  &  O3/NH3/CI2
                  on  Misc.  DBP  Formation  (Utility  7)
         12
         10
          7
          2 -
               EZZ3 Q2 any
               IB CI2/M-O
DISI. Sy&. Rosidenca
1: t » 2 Hr»
2: t » 24 IYS
* - Below IvAL
                1     2
                 CHP
                              FIGURE 6-22

-------
OZONE
RAPID MEX   FLOCCULATION
FILTRATION
 CHLORINE ONLY
           T
        Ferric Chloride
          Polymer
          Chlorine
                                                       Chlorine
                                         1
 OZONE / CHLORINE
            Ferric Chloride
              Polymer
000000
0000
0000000
000000000
.

o
o
                            Chlorine -
                           Intermittent'
                                                         Chlorine
          SCHEMATIC OF UTILITY 19 TREATMENT PROCESSES
                              FIGURE 6-23
  * Chlorine was not added at this point on day of sampling.

-------
                Treatment Modification Results and Discussion


100 percent Los Angeles Aqueduct (LAA) water for five days with doses on the day of
sampling of 1.8 and 0.3  mg/L, respectively.   On the  fifth day, samples were collected
at  predetermined points:  at the clearwell  effluent and in the  distribution  system at
residence   times  of  4.3  and  11   hours.    When  preozonation  was  reinstated,
prechlorination  was stopped and an ozone  dose of  1.3 mg/L was  applied; a post-
chlorine dose of 1.0 mg/L was used.  However,  during this ozone/chlorine treatment,
the treatment plant was  using 78 percent LAA  water/22  percent State Project water
(SPW).   Therefore the  ozone/chlorine  treatment was  repeated when  the  plant  was
utilizing 100 percent LAA water.  At that time, an ozone dose  of 1.7 mg/L was applied
and a 1.5 mg/L post-chlorine  dose  was used.  The chlorine residuals at the  sampling
locations, as well as other water quality parameters, are shown in Table 6-4.  Free and
total chlorine  residuals and TOC  concentrations were similar in the "before"  and
"after" samplings when 100 percent LAA water was  treated.  The temperature for the
chlorine-only samples was slightly lower. The blended  water  had a higher TOC level,
since SPW  has a higher TOC than LAA water.

Figures  6-24 and  6-25 present concentrations of DBP classes  and the miscellaneous
DBPs.   All data are presented  after  11 hours of residence  time in the distribution
system and represent only the  100 percent LAA water samplings.  As shown  in Figure
6-24.  decreases of 13 yug/L and 8.7  /yg/L  were observed for TTHMs and  HAAs,
respectively, after  ozone implementation with subsequent chlorination.   A  2.3 j/g/L
increase in  chloral  hydrate was observed. The effect  of ozonation on HANs, HKs and
chloropicrin formation  is  shown  in  Figure  6-25.    After  ozone  addition,  the
concentration of HANs was reduced by  1.2 /ug/L, while small increases were observed
for HKs and chloropicrin.  Cyanogen chloride analysis was not performed during the
ozonation trial;  it was only detected  at levels equal to  or slightly above the MDL during
the chlorine-only  experiment.   Changes in  TOX  concentrations  before  and  after
ozonation are presented in Table 6-4.  For the 100  percent LAA water studies at the
11-hour residence  time samplings, a decrease in TOX  of 62  //g/L was observed after
ozonation was applied.

Figures 6-26 to 6-31  present the  data for the treatment modification study in relation to
residence time in the distribution system. Levels of XDBP classes and individual DBPs
are shown at the three different distribution  system residence times: 0 hours  (clearwell
effluent), and 4.3 and  11  hours.  In addition, ozone/chlorine  data are presented when
the treatment plant was using  100 percent LAA water versus 78 percent LAA water/22
percent  SPW.    In general, the data show that  all  DBPs increased with  increasing
residence time except for 1,1 -DCP and cyanogen chloride.

Figure 6-26 shows  that when only LAA water was used, decreases were observed in the
DBP class totals for THMs, HAAs and HANs with ozone implementation.  However, as
shown in Figures 6-27 to 6-29, the  di- and tri-brominated species of these DBP classes
increased with  ozonation.  Although  the  level of total HAAs decreased with ozone
implementation  by  approximately 50 percent at  the 11-hour residence time  sample
point, the level  of  DBAA increased  by more than  100 percent.  At Utility 19, bromide
was not  assayed at the  time  of the  study;  however,  0.04 mg/L was detected in a
subsequent  sampling of the same raw water source.

The change in source  water from 100  percent LAA water to a blend with  SPW further
shows the effect of bromide on concentrations of DBPs produced by ozonation followed

                                     6-5

-------
                                     TABLE 6-4

                         UTILITY 19 TREATMENT STUDY
                              Water Quality Parameters
Pre-
Ozone
TOC Dose
(mg/L) (mg/L)
Cleanvel] Eflluenl
O3/CI,M 1.50 1.3
oycu" 1.22 1.7
Cl;h 1.20 0.0
Distr. PI. 1
( 4.25 hours )
O3/CI:H 1.48 1.3
O,/CI,h 1.20 1.7
Cl,'1 1 .25 0.0
Distr. Pt. 2
( 1 1 hours )
O3/Cl:a 1.56 1.3
O3/Cl,h 1.21 1.7
Cl,h 1.28 0.0
Chlorine Free Total
Dose Chlorine Chlorine pH
Pre, Post TOX Residual Residual Temp.
(mg/L) 0/g/L) (mg/L) (mg/L) TO

0.0. 1.0 73 0.7 0.9 8.1 17.5
0.0. 1.5 68 1.1 1.2 8.1 18
1.8.0.3 140 1.0 .15 8.1 15.2


0.0. 1.0 100 0.68 0.71 7.7 16.8
0.0. 1.5 79 1.0 1.0 8.1 19.0
1.8.0.3 140 0.92 1.00 7.9 15.0


0.0, 1.0 92 0.55 0.68 7.8 16.8
0.0, 1.5 88 0.52 0.54 8.2 19.5
1.8,0.3 150 0.63 0.68 7.9 15.5
"Source water was 78 percent LA Aqueduct water and 22 percent State Project Water.
''Source water was 100 percent LA Aqueduct water.

-------
       EFFECT OF VARIOUS  DISINFECTION SCHEMES ON
 DISINFECTION  BY —PRODUCT  FORMATION  AT UTILITY  1  9
«g/L
                                   FIGURE 6-24
     EFFECT OF  VARIOUS DISINFECTION SCHEMES ON
DISINFECTION BY—PRODUCT FORMATION AT  UTILITY  19
                                                              -fe
                                                                    ug/L
                                                  NA: Not  Analyzed
                                FIGURE 6-25

-------
7O
s
4—•
w.
•*-•
§
u

Q
60  -


5O


4O


30


20


 10


 0
           Effect  of  O3/CI2  &  CI2  Only  on
                DBP Formation  (Utility  19)
                                        CI2. 1OO% LAA

                                        O3/CI2.1OOV. LAA

                                     P773 03/CL2. 78% LAA. 22* SPW
                                     Dist. Sys. Residence Time (t)

                                     1:1 = 0. Cleafwell Eff.
                                     2: t = 4.3 ITS
                                     3: t = 11 hrs
                                           L nrv
      1  2  3
      XDBPsum
                            1   2  3
                             THM
                              1  2  3
                               HAN
                            DBP Class
                        FIGURE 6-26
1   2  3
  HK
1  2  3
  HAA
_J
3
g
«-<
10
I

U

h-
20
 15
 10
  O
           Effect  of  O3/CI2  &  CI2  Only  on
                THM  Formation  (Utility  19)
       1   2
     CHO3
                            CI2 .  1OO% LAA

                            O3/CI2.  1OO% LAA

                        fTTTI O3/CI2 . 78% LAA. 22% SPW
                       Dist- Sys. Ftesidence Time (t)
                        1: t = 0. C1««rw«ll Bit.
                        2: t = 4.3 tra
                        3: t = 1 1 rrs
                                               1   2   3
                                              CHClBr2
                      1   2   3
                    CHBri',12
                          11 1M Compound
                        FIGURE 6-27
             2   3
           CHBr3

-------
 9


 g
 4-*
 ID
O

<
<
           10
            a
            o
                      Effect  of  O3/CI2  &  CI2  Only   on

                           HAA  Formation  (Utility   19)
        CI2.  1OO% LAA



        O3/CI2.  1OO% LAA



        O3/C.I2 . 78% LAA 22% SPW




      Dist Sys. Residence Time ttJ



       1  I = O. Ciearwell Eff.


       2: t = 4.3 hrs


       3: I = 1 1 hrs
                 1  2  3

               MCAA
                                     HAA Compound



                                  FIGURE 6-28
s
*-^

w
4-»


§
<
          2.5
          2.0
           1.5
          1.O
          0.5
          O.O
                      Effect  of  O3/CI2  &  CI2  Only  on

                           HAN  Formation  (Utility   19)
     CI2 .  100% LAA



     O3/CI2 .  lOOH LAA



     O3/CI2. 76% LAA 22% SPW

Dist. Sys. Residence Tune (t)


 1:  t =0. Claarwaii Erf


2:  t = 4.3 hrs


3:  t = 1 1 hrs


* = Below K/Ftt.
                 123       123       123

                  TCAN         DCAN           BCAN


                                    HAN  Compound


                                      FIGURE 6-29
               1   2   3

                 DBAN

-------
(J
         O.5 -
         0.0
                   Effect  of O3/CI2  &  CL2  Only  on
                         HK Formation  (Utility  19)
                          CI2.  10OK LAA

                          O3/CI2. 10OS LAA

                          O3/CI2. 78% LAA 22% SPW
                      Dial. Sys. RMKtonca Tim* (t)
                       1: t = O. Cl«w«»ali Eft
                       2: t = 4.3 hrs
                       3: t = 1 1
                                                1.1.1 ~  TCP
                                 HK  Compound
                               FIGURE 6-30
I
ro
u
          10
           8
           0
                   Effect  of O3/CI2 &  CI2  Only  on
                    Misc.  DBP  Formation  (Utility  19)
(%%%! CI2. 10CM LAA
^B O3/CI2. IOO% LAA


(7771 Q3/C12. 76% LAA 22S SPW

Oi»t Sy*. RMKfenca Tim* (1)
1: t « 0. Cl««rw4ll Eft.
2: t * A3 fr»
3: t • 11 rvt
* ~ BttlOW iLifil
o - Not Analyzed
-
* « «A..
1
^Bv
















^
!
>











^
^
/

\
'/
y,
/.
I
I
V









Ea°ra im0r7i «~°^,
                1    2
                CHP
 1    2   3
   CH
DBP  Compound
 FIGURE 6-31
1    2
  CMC I

-------
                 Treatment Modification Results and Discussion
by chlorination.  A comparison of the two shows that when SPW was employed, a shift
to increasing concentration of the brominated species occurred.  Although bromide was
not measured during this treatment  scenario, a subsequent  sample of a blend of 74
percent LAA water and 26 percent  SPW showed that the bromide level had increased to
0.1 mg/L.  These data are consistent with those discussed in Section 5 regarding Utility
12.  where  seasonal  shifts  in  bromide  concentrations promoted  increases in some
brominated  DBFs.  In addition, ozone can react with  bromide  ions in  the  raw water,
causing the  formation of hypobromous acid (HOBr) (Dore et al.,  1988).  Reactions of
HOBr with  natural  organic material can produce  bromoform  and other brominated
DBFs, as evidenced at Utility 19.

Utility 25

Sampling  at  Utility  25 was  conducted at full scale at a conventional treatment plant
with  a capacity of  90  mgd.    A  schematic  of the  plant  before  and  after  the
implementation  of ozonation  is presented in Figure 6-32.   Samples  were collected
before the plant went online with two-stage ozonation. During  this period, concurrent
addition of  ammonia (1.6 mg/L) and chlorine  (8.0 mg/L) was practiced at the rapid
mix; there were no other points of disinfectant addition.  Sampling of ozonated water
was conducted seven days after the plant switched to routine operation of ozonation to
insure that the distribution system  was well-flushed with ozonated water.  The  oxidant
was applied  to both raw and settled water at 4 mg/L per stage, and  chlorine  (5.0 mg/L)
and ammonia (1.0 mg/L) were added concurrently prior to filtration. Both before and
after ozone  implementation, samples were collected at various points in the plant and
distribution  system.   Water quality  parameters  of  the raw  water and at the  various
sampling points are shown in Table 6-5.

The effects of the treatment  modification on concentrations of DBFs for this utility are
shown in Figures 6-33  and  6-34.   In  these  plots,  data  from  Location 4   in  the
distribution system (residence time  of 18 to 20 hours) are employed.  The DBF classes
with the highest concentrations when chloramines-only were employed were TTHMs
and  HAAs.   It  should  be  noted,  though, that the total chlorine dose  and  residual
concentrations were higher during this sampling than during  the ozone study.  Despite
concurrent addition  of  chlorine and ammonia,  Figure  6-33  shows  that  under  the
chloramines-only scheme, the TTHM and HAA concentrations were relatively high,
both being 44 //g/L.  This is  in contrast to  the study at  Utility 36 which showed that
the chloramines-only scenario was very effective in  limiting the  formation of these
compounds.   The relatively high levels of these DBF classes at Utility 25 were probably
a result of some free chlorine contact time due to inadequate mixing  and/or the high
raw water TOC concentration (7.7 mg/L).

Concentrations  of TTHMs  and HAAs  were  reduced to  6.7 //g/L  and  15 ,t/g/L,
respectively, after the treatment modification  was implemented.  Concentrations of the
measured  aldehydes  increased  from 10 to 57 /yg/L  with  the switch to ozone.   Figure
6-34 shows that the ozone/chloramines treatment slightly decreased the concentration of
chloral hydrate and HANs as compared to chloramines only;   however,  small increases
were observed for HKs,  cyanogen chloride and chloropicrin.   Finally, Table 6-5 shows
that TOX decreased by 89 /wg/L when ozone treatment was implemented.
                                     6-6

-------
  OZONE
RAPID MIX   FLOCCULATION  SEDIMENTATION   OZONATION  FILTRATION
CHLORAMINES ONLY
     T
    Polymer
     Alum
    Chlorine
    Ammonia
                                     Lime
                                                 1
  OZONE / CHLORAMINES
       Polymer
       Alum
                                                        Chlorine
                                                        Ammonia
   1
000000
 0000
ooooooo
 000000000
                                          Lime

o
o




x

\

x

s,




^>V&£
•^:*\^."fr'\
]
i


000000
0000
ooooooo
000000000
                                             I
                                                 1
          SCHEMATIC OF UTILITY 25 TREATMENT PROCESSES
                            FIGURE 6-32

-------
                                            TABLE 6-5

                                UTILITY 25 TREATMENT STUDY
                                     Water Quality Parameters
uv
Absorbance
TOC at 254 nm
(mg/L) (cm1)
CONVENTIONAL
Raw
Filter Influent1
Filter Effluent1
Clearwell Effluent
Location 1
Location 2
Location 3
Location 4
OZONE
Raw
2nd O, Contactor Inf.
Filter Influent
Filler Effluent
Clearwetl Effluent
Location 1
Location 2
Location 3
Location 4

7.72
4.63
4.63
4.85
4.67
4.72
4.69
4.76

8.01
5.47
5.25
5.06
5.23
5.65
5.48
5.44
5.51

0.255
0.096
0.081
0.095
0.087
0.091
0.086
0.088

0.233
0.054
0.067
0.041
0.046
0.050
0.049
0.053
0.047
Chloride
(mg/L)

34
NA
NA
NA
NA
NA
NA
NA

34
NA
NA
NA
NA
NA
NA
NA
NA
Bromide
(mg/L)

0.22
NA
NA
NA
NA
NA
NA
NA

0.23
ND
ND
ND
ND
ND
ND
ND
ND
TOX
U/g/Lt

20
180
200
170
150
140
150
150

NA
29
57
65
67
64
56
64
61
Free
Chlorine
pH Residual
(mg/L)

8.3
7.3
8.1
9.0
9.0
8.9
8.9
9.0

NA
9.8
9.2
9.2
9.4
9.2
9.3
9.4
9.3

NA
NA
NA
NA
O.I
0.1
O.I
O.I

NA
NA
NA
NA
NA
NA
NA
NA
NA
Total
Chlorine
Residual
(mg/L)

NA
4.0
4.0
3.5
2.9
3.2
3.5
3.5

NA
ND
2.7
2.7
2.2
1.8
1.2
1.3
I.I
'No lime added
NA - Not Analyzed
ND - Not Detected
Note:  Temperature for all studies ranged from 26.0 to 30.0°C.
      Locations represent distribution system sampling points in order of increasing residence
      time.

-------
                 OF  VARIOUS DISINFECTION SCHEMES ON
 DISINFECTION  BY—PRODUCT  FORMATION  AT UTILITY
                                  FIGURE 6-33
     EFFECT OF  VARIOUS DISINFECTION SCHEMES Or J
DISINFECTION  BY—PRODUCT  FORMATION  AT UTILITY  2
                                  FIGURE 6-34

-------
                 Treatment Modification Results and Discussion


Stability of DBFs Through the Plant and Distribution System.  Figures 6-35 to 6-45
show a  profile  of  DBFs through the  treatment plant and  at  four  locations  in the
distribution system.  Explanations of abbreviations used in the figures and distribution
system residence times at the point of sampling are presented  in Table 6-6.  During the
sampling for conventional treatment,  the plant temporarily went offline and  lime was
not applied  immediately after startup.  The filter influent and  filter effluent samples
were collected after the plant startup, so these samples had abnormally  low pH values
(7.3 and 8.1.  respectively, as  compared  to 8.9  to  9.0 at the distribution  system
sampling points which represents water produced  when lime addition was underway) as
shown in Table 6-5.

Figures 6-35  to 6-39 present the DBF profiles for XDBPsum, THMs, HAAs, DCAA and
chloropicrin.  The data show that for both treatment scenarios, the DBFs were  formed
immediately after chlorine and ammonia addition and remained stable through the plant
and into the distribution system. During the ozonation study, some low  levels of DBFs
were  detected in the second-stage ozone influent.   It  is believed  that  some minimal
level of chlorine or chloramines was added in the treatment train prior to this sampling
point.

DBF profiles of HANs. HKs, and chloral hydrate are depicted in Figures 6-40 to  6-42.
In each  case, during  the chloramines-only  treatment  scenario, higher concentrations
were  found  in  the  filter influent and filter effluent than in the clearwell effluent or
distribution system.   As noted above, the pH values were 0.8  to  1.7 units  lower for
these two sampling  points during the chloramine-only study because of a temporary
discontinuation of lime addition.  The data  suggest that at the higher pH values these
compounds  are  unstable and  break down to other final products.  As  discussed in
Section  5,  other  researchers  have  reported that HANs  and  HKs  were reactive
intermediates rather than stable endproducts such as THMs  and HAAs (Reckhow and
Singer,  1985).   They also found  that with increasing  pH,  DCAN  and  1,1.1-TCP
underwent hydrolysis.  Likewise, Miller and  Uden (1983) found that chloral  hydrate
decomposed  at elevated  pH.   When ozonation was applied,  smaller concentrations of
these  compounds were detected,  but  similar  trends in relation  to  stability  were
observed.  In each case, concentrations in  high pH waters  decreased with increasing
residence time.

The cyanogen chloride profile (Figure 6-43) indicates that this compound was unstable
in Utility 25's distribution system.  In contrast, cyanogen chloride increased in Utility
6's distribution  system (e.g.,  from 4.7 /ug/L at  the clearwell effluent to 9.9 /ug/L at
location  3 during the chlorine/chloramines study).  Both utilities have  chloramines as
the final disinfectant; however,  the pH of Utility 6's distribution  system was 7.5 to 8.0.
These limited data imply that cyanogen chloride may be unstable at  a  pH of 9, as
evidenced at  Utility 25.

Figures 6-44 and 6-45 show the profiles for the aldehyde sum and for formaldehyde
only, respectively.   In the chloramines-only  scenario,  these compounds were  formed
after  disinfectant addition  and  remained  relatively  stable  through  the  plant and
distribution system.  In the ozone/chloramines treatment scenario, higher concentrations
of aldehydes were detected and a similar stability profile was observed.   It should also
be  noted that  these  data are  consistent  with  those  described  above at  Utility 6.


                                       6-7

-------
                            TABLE 6-6

     EXPLANATION OF ABRREVIATIONS USED IN UTILITY 25
 DBF PROFILES AND RESIDENCE TIME OF DISTRIBUTION SYSTEM
                        SAMPLING POINTS
Abbreviation
     Explanation
   RAW
   2OI
   FI
   FE
   CE
   LI
   L2
   L3
   L4
Plant Influent raw water
Second stage ozonation influent
Filter Influent
Filter Effluent
Clearwell Effluent
Location 1 in Distribution System (4-5 hours residence time)
Location 2 in Distribution System (8-9 hours residence time)
Location 3 in Distribution System (9-10 hours residence time)
Location 4 in Distribution System (18-20 hours residence time)

-------
O)



g

ro

c

?
o
U
 13
 OJ

 ro


 I
 0
 to
            htfects ol  NH2CI  and O3/NH2CI  Treatment

             c.in  Halogenated DBP Formation, Utility  25
       120
       100 -
                                             NA - Not analyzed

                                             *•*  = NO hme addition
            RAW  2OI   Fl   FE    CE    L1    l_2    L3
                        Treatment &  Residence Time

                               FIGURE 6-35
                                                         L4

o
U
m
o
            Effects  of NH2CI  and  O3/NH2CI  Treatment

              on Total  THM  Formation  at Utility  25
       70



       60



       50



       4O



       30
       2O -
       10 -
                NH2CI


                O3/NH2CI
NA  =  riot  analyzed

*•* = No lime  addition
                      •**•
           RAW ' 2Ol
                        Treatment & Residence Time


                              FIGURE 6-36

-------
 O)
 D


 g

 ro
 u
 o
 U
70



60



50



40



3O



20



10



 O
            Effects  of NH2CI  and  O3/NH2CI  Treatment

                   on  HAA  Formation  at  Utility  25
            V777A NH2CI


                 O3/NH2CI
 NA =  Not analyzed

*-*- = No lime addition


 * - Below  MRL
               NAN,


            RAW ' 2OI
                                        L2
                         Treatment & Residence Time


                                FIGURE 6-37
           L3
L4
0)
U

-------
 O

 8
 0
 c
 U
             Effects  of  NH2CI  and O3/NH2CI  Treatment

                on  Chloropicnn Formation  at  Utility  25
         0.6
         0.5 -
0.4 -
0.3 -
0.2 -
         0.1 -
         0.0
               NA = Not Analyzed
               * = Below MRL
                         Treatment &• Residence Time


                                FIGURE 6-39
Oi




O
O
5
U


<
I
             Effects  of  NH2CI  and  O3/NH2CI  Treatment

                   on HAN Formation at  Utility  25
                                            NA = Not analyzed


                                            *••* = No lime addition
                                    CE   L1    L2   L3
         0.0
     RAW   2OI   Fl
                         Treatment &• Residence  Time


                               FIGURE 6-40

-------

-------
 (TJ
O
0
u
         8.0
         6.0
         4.0
   2.0
         0.0
             Effects  of NH2CI  and  O3/NH2CI  Treatment
                   on CNCI Formation  at Utility 25
               W77X NH2CI

               BBB O3/NH2CI
                                         NA = Not analyzed
                                         » = Below N/I3L
                                         ** = No lime addition
                              -X--X-
RAW  2OI   Fl
                               FE    CE
                                    L1
               L2   L3
L4
                          Treatment & Residence Time

                                 FIGURE 6-43
o
ffl
k.
*^
0)
o
o
u
Q
            Effects of  NH2CI  and  O3/NH2CI  Treatment
                   on  ALD  Formation  at  Utility  25
100

 90

 80

 7O

 6O

 50

 4O

 30

 20

 10

  0
                NH2CI

                O3/MH2CI
NA =  Not  analyzed
* = Below MRL
•X--X- =  No lime addition
     = Formaldehyde not analyzed
            RAW  2OI
                        Treatment & Residence  Time

                               FIGURE 6-44

-------
O)
g
^->
ra
O

O
U


-------
                 Treatment Modification Results and Discussion
 Moreover, since disinfectant  was  applied  before  filtration,  there  was probably  no
 opportunity for biological degradation of these compounds in the filters.

 Utility 36

 At  Utility 36, a pilot plant with a flow rate of 5 gallons  per minute was employed to
 evaluate five different treatment  scenarios on the formation of DBFs.   Conventional
 treatment was employed with  options to add preozonation (2 mg/L), with and without
 hydrogen peroxide addition (0.67 mg/L),  and free chlorine (6.5 mg/L) or chloramines
 (2,1 mg/L chlorine  and 0.5 mg/L  ammonia) as disinfectants in the rapid mix.  The
 various treatment scenarios are shown in Figure 6-46.  Samples were collected from the
 filter effluent and held for 24 hours at ambient temperature in order  to simulate the
 retention  time  in a  distribution  system.   After  the 24-hour  incubation time,  samples
 were transferred to individual  sample bottles containing the appropriate dechlorination
 agents and preservatives.  Various water quality parameters are shown in Table 6-7.

 Table  6-8 shows the effect of ozonation on  the  individual DBF compounds produced
 with chlorination in  24-hour samples.  With  ozonation, chloroform decreased from  42
 to 37 //g/L.  However, increases were observed  for the brominated THMs,  resulting in
 an  increase  in TTHMs.  The level of bromide measured in the raw water at Utility  36
 was 320 fjg/L.  Dichloroacetic acid was essentially the same level, with  and without
 preozonation, while  trichloroacetic acid was reduced by 50 percent with preozonation.
 The varying effect  of  ozone on  HAA  precursors  is  consistent  with   bench-scale
 experiments performed by Reckhow and Singer (1985).   Other notable changes were
 observed for formaldehyde and acetaldehyde, which increased from 13 to 36 >ug/L and
 II  to  16 fjg/L, respectively, with ozonation.  Additionally,  chloral hydrate increased
 from 19 to 28 //g/L.   Figure 6-47 shows the effect  of the  five different  disinfection
 schemes on various  DBF classes  and chloral hydrate.  The highest levels of TTHMs,
 HAAs.  HANs and chloral  hydrate were observed with those  schemes which employed
 chlorine as a final disinfectant.  By comparison, large decreases were observed under
 any disinfection scheme which employed  chloramines.  For example, as compared to
 the  chlorine-only disinfection scenario,  chloramines-only,  ozone/chloramines. and
 ozone/hydrogen peroxide/chloramines reduced TTHM levels by 96, 97 and  98 percent,
 respectively.

 When  chloramines-only were  applied, only 6.1  fjg/L of  TTHMs were  formed.  This
 result is in contrast to that found during the  the study at  Utility 25, where fairly high
 levels  of TTHMs (44  fjg/L) were formed  when  chloramines-only were employed.
 Thus. Utility 36 data further support the probability that  there was some free chlorine
 contact time at Utility 25  although concurrent addition of ammonia and chlorine was
 practiced.

 Figure 6-47 shows that at Utility 36, chloramines only were as effective in reducing the
 levels    of   chlorinated   DBPs    as    ozone/chloramines   or   ozone/hydrogen
 peroxide/chloramines.     Similar  findings   are  shown  in  Table  6-7  for  TOX
concentrations.   Unlike the study at Utility 25, where the chloramine dose was  reduced
after  preozonation.  the  chloramine doses and  residuals at  Utility 36  were  similar
whether  or not  ozone  was employed.    Aldehydes  were  produced  in  greatest
concentrations under the disinfectant schemes which employed ozone.  The  level of the


                                      6-8

-------
    OZONE
RAPID MIX  FLOCCULATION  SEDIMENTATION  FILTRATION
CHLORINE ONLY
t
Chlorine
Ferric Sulfate
Polymer
I
O
O
— ^-



\

V

k,

x

	 ^.






 CHLORAMINES ONLY
*
Chlorine
Ammonia
Ferric Sulfate
Polymer
I
O
O




s.

X

\

\

^^


r — -


OZONE/CHLORINE
              Chlorine

t
000000
0000
0000000
000000000
Ferric Sulfate
Polymer
J .


O
O




\

X

\

s


— ^-





       SCHEMATIC OF UTILITY 36 TREATMENT PROCESSES
                          FIGURE 6-46

-------
     OZONE
                        RAPID MIX  FLOCCULATION SEDIMENTATION  FILTRATION
 OZONE/CHLORAMINES
"1
000000
0000
0000000
000000000

*
Chlorine

O
O




s

\

\

s,








            Ferric Sulfate
              Polymer
OZONE AND PEROXIDE / CHLORAMINES
             Hydrogen
             Peroxide
000000
0000
0000000
000000000

t
Chlorine
O
O
           Ferric Sulfate
             Polymer
       SCHEMATIC OF UTILITY 36 TREATMENT PROCESSES
                        FIGURE 6-46 (Continued)

-------
                                           TABLE 6-7

                               UTILITY 36 TREATMENT STUDY
                                    Water Quality Parameters
                                uv
                             Absorbance
                       TOC  at 254 nm   Chloride   Bromide    TOX
                      (mg/L)   (cm1)     (mg/L)    (mg/L)    
-------
                               TABLE 6-8

            DISINFECTION BY-PRODUCT CONCENTRATIONS

         BEFORE AND AFTER OZONE ADDITION AT UTILITY 36
Disinfection By-Products
Chlorine
  Only

 0/g/L)
                                                                   Ozone/
                                                                  Chlorine
Trihalomethanes
  Chloroform
  Bromodichloromethane
  Dibromochloromethane
  Bromoform
      Total Trihalomethanes

Haloacetic Acids
  Monochloroacetic Acid
  Dichloroacetic Acid
  Trichloroacetic Acid
  Monobromoacetic Acid
  Dibromoacetic Acid
      Total Haloacetic Acids

Haloketones
  I. I -Dichloropropanone
  I.I.I -Trichloropropanone
      Total Haloketones

Haloacetonitriles
  Trichloroacetonitrile
  Dichloroacetonitrile
  Bromochloroacetonitrile
  Dibromoacetonitrile
      Total Haloacetonitriles

Aldehydes
  Formaldehyde
  Acetaldehyde
      Total Aldehydes

Miscellaneous
  Chloropicrin
  Chloral Hydrate
  Cyanogen Chloride
  42
  50
  50
   9.5
 152
   4.0
  23
  22
   3.8
  II
  64
   0.24
   1.8
   2.0
 
-------
     EFFECT OF VARIOUS DISINFECTION SCHEMES  ON
DISINFECTION  BY —PRODUCT FORMATION AT  UTILITY
      EFFECT OF VARIOUS  DISINFECTION SCHEMES ON
DISINFECTION  BY-PRODUCT  FORMATION  AT UTILITY
                                 FIGURE 6-48

-------
                 Treatment Modification Results and Discussion


measured aldehydes for ozone/chlorine was  52 //g/L as contrasted to 24 /ug/L and 9.1
A/g/L for chlorine-only and chloramines-only schemes, respectively.

Figure 6-48 shows the effect of the five different treatments on HKs, cyanogen chloride
and  chloropicrin.   The data indicate that formation of cyanogen chloride was most
influenced by the final  disinfectant.   For those disinfection  schemes which employed
chloramines, cyanogen chloride levels were approximately 8 to  15 times greater than
those which employed  chlorine  as  a final disinfectant.   Additionally,  preozonation
resulted in a higher cyanogen chloride level than that produced by chloramines alone.

Chloropicrin concentrations were approximately  2  to  4  times greater  when the
disinfection scheme employed  preozonation.   Although these results were consistent
with those found by others (Becke et al., 1984; Hoigne and Bader, 1988), the levels  of
chloropicrin detected  were  low, ranging from 0.22 to 1.07 //g/L.  The cited studies
reported  increases  in chloropicrin in  systems where preozonation was followed  by
chlorine addition.  At Utility 36, the highest increase in chloropicrin concentration was
observed in the trial which employed ozone followed by chloramination.

The  use of hydrogen peroxide in  conjunction with ozone has been shown to enhance
the efficiency of ozone in  oxidizing some organic materials (Glaze et al.,  1987; Aieta et
al..  1988: Wallace et  al.,  1988).   In this study, peroxide was added at a ratio of 0.3  to
1 (hydrogen peroxide to ozone) on a weight basis; chloramines were employed as the
final disinfectant.  When compared to ozone/chloramine treatment, only small changes
were observed in DBF formation using the ozone/hydrogen peroxide combination.

Effects  of Holding Time  on DBFs.  Figures  6-49 to 6-58  show the effect of holding
time on levels of DBFs  which were measured in the pilot  plant effluent (t=0; the
residual disinfectant contact time through the  plant before filtration was approximately
one  hour) and after 24 hours.  The greatest increases in XDBPS(im or XDBP class total
concentrations over 24 hours were observed in the treatment  scenarios which employed
free  chlorine  as a final  disinfectant (Figures 6-49  to 6-53).   Similar results were
observed  for  chloral  hydrate  (Figure  6-54).    For  chloropicrin,  however, the
ozone/chloramines treatment scenario produced the greatest  increase over a 24 hour
period (Figure 6-55).  It should be noted, though, that the concentration even after the
24-hour holding time  was  low (1.07 jug/L).

As noted above, cyanogen  chloride  levels after 24  hours were 8  to  15 times greater
under treatment scenarios which employed chloramines as a final disinfectant.   Figure
6-56 shows that  levels  of this compound  increased  over the holding time under the
ozone/chloramines scenario, while it decreased under those which employed chlorine as
a final disinfectant.

Figure 6-57 and 6-58 show  that once aldehydes were formed, they remained stable over
24 hours.  This is consistent with the data found at Utility 6.  It should also be noted
that  since a disinfectant residual was carried through the filter,  there was probably no
significant biological  activity  to  lower  the aldehyde  levels  formed  from the various
disinfectants.
                                      6-9

-------
(0
t-
«-•
8

u
TJ
0)
                 Effects  of Various  Treatments on
             Halogenated  DBP  Formation  at Utility  36
      400
      3OO -
      2OO -
                                  G3S3 CL2

                                     MH3.CL2

                                     03.CL2

                                  17773 O3JSH3.CL2
      100 -
                t=0 hrs          t=24  hrs
                   Conventional  Filter  Media
                        Treatment & Residence Time

                              RGURE 5-49
5
«-•
(0
i

                 Effects  of Various  Treatments on
                    THM  Formation at Utility  36
      200
      150 -
      100 -
t-O hrs
                                t=24 hrs
                            Filter Media
                        Treatment. &  FHesidence Time
                              FIGURE 6-50

-------
o
ro
k_
•f-'

-------
ID
t_
4—'


0)
 u
^)



8
*-^
to
L.
*->


-------
 5
 4—'
 (0
 k_
 **
 
-------
00
Q
<
o»
U
*
>
I
m
b
u.
                 Effects of  Various Treatments  on
                    ALD Formation  at Utility  36
                t=O hrs          t=24 hrs
                 Conventional Filter Media
                        Treatment & Residence  Time
                             FIGURE 6-57


                 Effects of  Various Treatments  on
               Formaldehyde  Formation  at Utility 36
                t=0 hrs          t=24 hrs
                   Conventional  Filter Media
                        Treatment & Residence  Time
                              FIGURE 6-58

-------
                 Treatment Modification Results and Discussion
Discussion

As shown in the ozonation studies discussed above, the formation of DBFs is strongly
influenced by the particular disinfection  scheme employed at  the treatment plant.  In
order to meet the anticipated  Surface Water Treatment Rule disinfection requirements
as well  as   DBF  regulations,  many utilities  are  faced  with  selecting alternative
treatments.    Three  general  disinfection treatment  change  scenarios  for  utilities
considering ozonation include:

     1)   Those  which  currently employ only free  chlorine  for disinfection in their
         treatment  process and will switch  to ozone  and chlorine for primary and
         residual disinfection, respectively;

     2)   Those  which  currently employ only chloramines for disinfection,  but will
         switch  to ozonation followed by chloramination;

     3)   Those  which currently employ  only  free chlorine  in their  treatment and will
         switch  to ozonation followed by chloramination.

Figures 6-59 to 6-64 summarize results from this study based on the above DBF control
options.    All the  data  represented  in these  figures  correspond to the  terminal
distribution  system  points studied.   Changes in concentrations of TOX, XDBPSU11) and
other DBF classes except HKs are presented as percent change.  Individual compounds
and  HKs are  shown as actual changes in concentration.  Changes were calculated by
comparing concentrations  of DBFs before  and  after ozonation  was added  to  the
treatment process.

Modification  from Chlorine to Ozone/Chlorine.  Figures 6-59 and 6-60 present data
for utilities  where  treatment plants  were modified from  employing only chlorine for
disinfection  to ozone/chlorine  treatment.  At  Utility  19, preozonation decreased the
concentration of TOX and most classes of the measured DBFs.  In contrast, a 7 percent
increase in XDBP   was observed at Utility 36; this was  primarily due  to the observed
increase in  TTHM concentration.   That ozonation can increase  TTHMs  has been
observed in  other studies (Riley et al.,  1978;  Lawrence,  1977; Umphres et ah, 1979).
Increased levels  of THMs after  preozonation may  be  attributed  to  an  increase  in
precursor materials by  formation  of:    I)  m-hydroxy  aromatic  compounds; or  2)
secondary precursors, i.e.,  aliphatic  carbonyl compounds  (Glaze et al.,  1982); or to an
increase in brominated THMs due  to bromide ion in the raw water (Dore et al., 1988).

Figures 6-24 and 6-25. and Table 6-8. show  the individual DBF concentrations which
were measured during the  modification  studies at Utilities  19 and 36.  As mentioned
previously, the data show  decreases in chloroform for the  two utilities and  a general
shift to the brominated  THMs after  ozonation.  This was  particularly evident at Utility
36 where the increase in TTHMs after ozone addition was due primarily to increases in
dibromochloromethane  and bromoform.   These changes  were greater  than  the upper
control  limits of  the  respective compounds,  i.e..  the differences were  not due  to
analytical variability.  As discussed in Section 5.  levels of brominated THMs in treated
water have been shown to  be associated  with bromide concentrations in the raw water.
The  same shift to the brominated species is seen to varying extents upon examination of

                                      6-10

-------
                Percent Change in DBP Concentrations c*je to Switch  from
                    CHLORINE  ONLY  to OZONE/CHLORINE  Treatment
                   UTILITY  19
                                                    UTILITY  36
TJ
O
a

CO

o
u
0)
"c
If}
a
         TOX
    XDBPsum
        TTHM
         HAA
         HAN
         ALD
                                         NA
- 10O
                         -50
                                        0
50
1OO
ISO
                        % Change  in  DBP and TOX Concentrations
                                       FIGURE 6-59
                    Change  in DBP  Concentrations  due  to  Switch from
                    CHLORINE  ONLY  to OZONE/CHLORINE Treatment
                   UTILITY  19
                                               V777A UTILITY  36
y
•p
o
a
i
CO
c
c
c
in
LJ
           HK
         CHP
           CH
         '-"NCI
                                                r-JA

                                                (O)
                                              O
                                                                               1O
                            Change  in  DBP  Concentration  (ug/L)
                                      FIGURE 6-60

-------
                Treatment Modification Results and Discussion


the concentrations of dibromoacetic acid.   Although the levels of total HA As decreased
by approximately 50 and  14  percent at Utilities 19 and 36, respectively, the level of
dibromoacetic acid increased by more than 100 percent at Utility 19, and by  18 percent
at  Utility 36.   Ozone  can react with bromide  ions  in  the  raw water causing the
formation of hypobromous acid (HOBr) (Dore et al., 1988).  Reactions of HOBr with
natural organic matter can produce bromoform and other brominated DBFs.  When
preozonation  and   postchlorination   is   practiced,   competition  exists   between
hypochlorous acid  and HOBr for organic  matter, leading to varying concentrations of
chlorinated and brominated DBFs (Dore et al.,  1988).

In this study, HAAs represented the second largest fraction of halogenated DBFs on a
weight basis.  The data presented in Figure 6-28 and Table 6-8 confirm work by other
researchers who found DCAA and TCAA to be major by-products of chlorination of
humic and fulvic acids (Quimby et al.,  1980; Christman et  al.,  1983; Miller and Uden,
1983: DeLeer et al.. 1985).  At Utility 19, chloroform and TCAA were decreased by
68 and 82 percent after preozonation and postchlorination.  DCAA was decreased by
50 percent,  but  remained  the largest contributor to the HAA fraction.   At  Utility 36,
the TCAA concentration  was  decreased  by 41  percent while DCAA decreased only
slightly.   Dore, et al. (1988) showed that ozonation followed by chlorination of fulvic
acids  reduced TCAA levels but did  not  greatly affect  DCAA.  Reckhow and  Singer
(1985) demonstrated decreases in  chloroform  and TCAA  precursors after similar
treatment while  DCAA  precursors remained largely unchanged.   Moreover, the latter
researchers  predicted  the  destruction  of dichloroacetonitrile  precursors  and  an
enhancement of 1,1,1-TCP.  In this  study, ozone addition resulted in a  decrease in
HANs and a small increase in HKs at Utilities 19 and 36, respectively.

At  Utility 36.  19  //g/L of chloral  hydrate  were detected after the chlorination-only
treatment,  and  at  Utility  19,  6.3  //g/L.   After ozonation and  chlorination, those
concentrations were increased to 28 and 8.6 //g/L, respectively.

Modification from  Chloramines  to  Ozone/Chloramines.   Four utilities  modified
treatment  from  chloramines-only  or  prechlorination/postammoniation  to  ozone/
chloramines.   Figures 6-61 and 6-62 show that, in each  study, the modification  was
effective  in  reducing levels of  all classes of halogenated  DBF compounds except  for
HKs.  However, increases  in this DBF class were less than 0.7 //g/L in all cases.  Small
increases were observed for chloropicrin  and cyanogen chloride in two of the studies,
while chloral hydrate decreased in three of them.  The most  significant increase was
observed  in  the concentrations of  aldehydes.   Preozonation  increased the sum of
formaldehyde and acetaldehyde  over 300 percent in the utilities where these compounds
were measured.

Modification from Chlorine to Ozone/Chloramines.  Figures 6-63 and 6-64 present a
summary of the  results of the two plants which modified treatment from chlorine-only
as  a  disinfectant   to  ozone/chloramines.    Decreases of at least  80  percent in
concentrations  of  TOX.  XDBPslim  and  sums  of  all  halogenated classes  of DBF
compounds except for HKs were observed.  Aldehydes increased by 67 percent and
cyanogen chloride by 2.9 //g/L  for the  treatment modification study  at Utility  36.  The
greatest decrease among the other compounds was detected in concentrations of chloral
hydrate,  which was  reduced by approximately 9 and  15  //g/L at Utilities 7 and 36,
respectively.

                                     6-11

-------
o

CL
I
>
CD


O
4->
O
0)
M—
c
in

Q
                Percent  Change in DBP Concentrations due to Switch from
                CHLORAMINES ONLY to OZC>E/CHLORAMINES  Treatment
           UTILITY
         TOX
    XDBPsum
         THM
         HA A
         HAN
          ALD
            -1OO
                             -50
                                                                                 100
                         % Change  in  DBP  and  TOX  Concentrations

             •»   Includes 4 hours of free chlorine  contact  time before postammoniation

                                       HGURE 6-61
                     Change in DBP Concentrations due to Switch from
                 CHLORAMINES ONLY to OZOE/OLORAMINES  Treatment
o
3
o
a.
 i

CD

g

o
0)
 Q
         UTILITY
           HK
          CHP
                                                 UTILITY  36   77777A UTILITY  6
                       V/////////////////7//,
         CNCI
                              Change  in  DBP  Concentration (ug/L)

              *   Includes 4 hours of free chlorine  contact  time before postammoniation

                                         FIGURE 6-62

-------
               Percent Change in DBP Concentrations  due to Switch from

                  CHLORINE ONLY  to  OZONE/CHLORAMINES treatment
                   UTILITY  7
                                                   UTILITY 36
 0

a
 i

m
u
Q
         TOX
    XDBPsum
       TTHM
         HAA
         HAN
         ALD
           -100
                            -5O
            0
                50
                     1OO
                       % Change  in DBP  and  TOX Concentrations


                                      FIGURE ^63
                   Change  in DBP Concentrations due to  Switch from

                 CHLORINE ONLY to OZONE/CHLORAMINES  Treatment
                   UTILITY  7
                                                   UTILITY  36
y

E

a
i

m


g
•*-•
u


"c


Q
          HK
         CHP
          CH
        CNCI
                                                        (0)
           -20
                      -15
-1O
-5
0
                           Change in DBP  Concentration (ug/L)


                                     FIGURE 6-64
                                                                            10

-------
                 Treatment Modification Results and Discussion
CHLORINE DIOXIDE STUDIES

Chlorine dioxide studies  were conducted  at  Utility  16 and Utility 37.   Both utilities
employed chlorine dioxide as a preoxidant, and used chlorine as a final disinfectant.

Utility 16

Utility  16  operates  a  large (400  mgd) direct filtration treatment system  for a  low-
organics reservoir water in the western United States.   The plant  utilizes free chlorine
for both  preoxidation  and final  disinfection  for  most of  the  year,  but periodically
switches to chlorine dioxide preoxidation  to  control THMs and taste and odor.  The
plant utilized chlorine-only treatment for all  quarterly baseline samples collected for
this  study.   The plant was  sampled  on November 21.  1988 for  the  chlorine-only
treatment,  and on March 21,  1989 for the chlorine dioxide/chlorine treatment.  Figure
6-65 illustrates the treatment trains employed on the two sampling days.

Samples were  collected at the plant  influent,  filter influent, clearwell effluent,  and at
two  distribution  system locations  (approximate residence times of 45 minutes and  7
days).   DBF analyses were conducted on the clearwell effluent and distribution  system
samples only.  On the chlorine-only sample date, the  plant was operating with a pre-
chlorine dose of 2 mg/L (applied  to the plant influent) and a post-chlorine dose of I
mg/L (applied  to the filter effluent).  On the chlorine dioxide/chlorine sample date, the
plant was  operating with  a chlorine dioxide  dose of 0.5 mg/L (plant influent) and a
chlorine dose of 1.9 mg/L (filter effluent).

An emergency  situation occurred on the March sampling date.  A chlorine storage tank
developed  a  leaky valve,  requiring that the tank be emptied.  The contents  of the tank
were  discharged  into  the  water  being treated in  the  plant, resulting in a  significant
amount of free chlorine in addition to the chlorine dioxide.  Thus, the in-p!ant samples
were  unusable for  purposes  of this  study.    However,  the  two  distribution  system
samples were collected prior  to  the chlorine leak problem  and the results  from these
samples are reported in Table 6-9 and illustrated in Figure 6-66.

Figure 6-66  indicates very little difference in  the levels of DBFs produced  by the two
different oxidation/disinfection schemes.  In fact, at distribution system Location 2, the
chlorine dioxide/chlorine treatment produced slightly higher levels  of XDBP    THMs,
HANs and HKs than the chlorine-only treatment.  With both  chlorine dioxide/chlorine
and chlorine-only treatment, levels of all DBFs increased with increasing  residence time
in the distribution system.

Chlorine dioxide  residual, and  chlorite and  chlorate levels  were  measured  at  the
laboratory  of Dr. Gilbert Gordon at  Miami   University in  Oxford, Ohio.   Preserved
samples were  shipped  overnight  to  Dr. Gordon  on both  the  November and  March
sampling dates.  Dr. Gordon's laboratory utilized a flow injection analysis method to
analyze for chlorine dioxide,  chlorite and chlorate (Gordon et al., 1989; Themelis et
al.. 1989).  Chlorine dioxide residual levels are reported in Table 6-9.  Neither chlorite
nor chlorate  were detected in  either of the chlorine-only samples.  However, for the
chlorine dioxide/chlorine samples, 0.30 mg/L of  chlorite  and 0.14 mg/L  of chlorate
were detected at  Location I.   Also, 0.09 mg/L of chlorite and  0.28  mg/L  of chlorate
were detected at Location 2.

                                      6-12

-------
                                 RAPID MIX     FLOCCULATION       FILTRATION
  CHLORINE ONLY
     Chlorine
                     Ferric Chloride
                                                   Zinc Orthophosphate
  CHLORINE DIOXIDE / CHLORINE
Chlorine Dioxide
                   Ferric Chloride
                                                  Zinc Orthophosphate
           SCHEMATIC OF UTILITY 16 TREATMENT PROCESSES
                               FIGURE 6-65

-------
                                         TABLE 6-9

                             UTILITY 16 TREATMENT STUDY
                                  WaCer Quality Parameters



uv
TOC Absorbance


(mg/L)
(cm1)
Chlorine Chlorine
Residunl Dioxide
pH Temp. (mg/L) Residual
(°C) Free
Total
(mg/L)
Chlorite
(mg/L)
Chlorate
(mg/L)
TOX
(PHIL)
CHLORINE ONLY



C
I



Rau Water
Location 1
Location 2
2.69
2.64
2.62
0.098
0.083
0.082
7.72 14.8 NA
7.59 15.1 0.6
7.65 16.5 0.9
NA
0.9
1.5
ND
ND
ND
ND
ND
ND
ND
ND
ND

-------
o>
ro

c

o

5
u

a
CD
Q
                    Effect  Of  CI2  and OO2/CI2

                    on  DBP Formation  (Utility  16)
/UL>
60.0
50.O
40.O

30.O

20.O
10.O
O.O
-

-

-
-


/
/
/
's
'',








T
^
/
^
^
^
^








7]
^
^
^




7
\
'/
'',
\
\
/
m^
%% CIO2/CI2
Dist. Sys. Residence Time (t)
1: t = 0.75 hrs
2: t = 16O hrs
NA - Not Analyzed



|-|
mWi *^**A SHI ilL
12 12 12 12 12 12
xCGP'jtm THM HftN hK HAA ALD
                                UBP Class
                              RGURE6-66

-------
                 Treatment Modification Results and Discussion
Utility 37
Utility 37 did not  participate in the quarterly baseline sampling program for this study.
hut was selected for a treatment study because the plant's configuration and operation
offered a  unique opportunity for a side-by-side comparison of DBF levels produced by
chlorine dioxide and chlorine-only preoxidation.  This Southeastern utility treats a  low-
organics  river water with a conventional  treatment process.   The plant typically treats
approximately 30 mgd.

The configuration  of the plant  provides for two separate treatment  trains.  In 1981. the
plant  began seasonal use of chlorine dioxide preoxidation (added following flocculation)
in one of the trains for THM control.  The other train has chlorine preoxidation.  Free
chlorine is used for residua! disinfection in both trains; however, samples were collected
before the addition of chlorine  for residual disinfection.  Schematic diagrams of the two
treatment trains are shown in Figure 6-67.

The chlorine  dioxide  dose was approximately 0.9  mg/L and the chlorine dose was
approximately  2.25 mg/L for  preoxidation.  However,  the  chlorine dioxide generator
product could not be analyzed directly;  it could only be analyzed  after addition of the
generator product to the flocculation  basin effluent.  Because chlorine dioxide reacts
very  rapidly,  the  actual  dose was  most  likely higher  than 0.9  mg/L,  although the
sample was taken  immediately  after addition 01 the generator product to the flocculated
water.  To further illustrate  this point,  plant  operations personnel reported that the
applied chlorine dose  was  2.25 mg/L; however,  the total chlorine residual measured
immediately after application  of the chlorine  to  the flocculated water was  only  1.15
mg/L.

All samples for this study were collected on May 19, 1989.  Water quality parameters
for this utility  are reported in Table 6-10.  On the  sampling date, it was  determined
that there was some degree of mixing between  the two treatment trains, since low levels
of chlorine  dioxide were  measured in  the chlorine-only  treatment  train and vice versa.
Additionally,  although the chlorine dioxide  generator  was  operated  to produce the
lowest practical levels of free chlorine in the generator product, detectable levels of free
chlorine  were  measured  immediately  after addition  of  the  generator product to the
flocculated  water  (0.2 mg/L).   Thus,  the  preoxidant in  one  treatment  train  was
predominantly  chlorine dioxide (referred  to as the "chlorine dioxide" treatment train).
and the preoxidant in the other train was  predominantly free chlorine (referred to as the
"chlorine" treatment train).

Samples  were  collected at  the  plant influent and sedimentation basin effluent.   DBF
analyses   were  performed on  the  sedimentation   basin  effluent  samples.    The
sedimentation basins had  a detention time of approximately 2.5 hours.  Following the
sedimentation basins, there was mixing of the two  trains in the filter influent flume.

Chlorine dioxide,  chlorite and chlorate measurements were  performed on-site  by  flow
injection analysis.  Chlorite levels were  1.16  mg/L in  the  chlorine dioxide treatment
train  and 0.07 in  the  chlorine train (measured at the sedimentation basin effluent).
Chlorate concentrations were 0.53 and 0.22 mg/L in  the chlorine  dioxide and chlorine
treatment  trains, respectively (sedimentation basin effluent).


                                      6-13

-------
       RAPID MIX/FLOCCULATION   SEDIMENTATION
FILTRATION
CHLORINE
                      Chlorine
          Fluoride
          Chlorine
                                Solids
                                Contact
                                Clarifier
                                        Lime
                                     (Not Operational)
CHLORINE DIOXIDE / CHLORINE
                                Solids
                                Contact
                                Clarifier
                                            Lime
         SCHEMATIC OF UTILITY 37 TREATMENT PROCESSES
                              FIGURE 6-67

-------
                                        TABLE 6-10

                             UTILITY 37 TREATMENT STUDY
                                  Water Quality Parameters
                           UV                     Chlorine
                       Absorbance                  Residual
                 TOC   at 254 nm   pH   Temp       (mg/L)
                (mg/L)   (cm ')          (°C)    Free    Total
Chlorine
Dioxide
Residual Chlorite  Chlorate  TOX
 (mg/L)  (mg/L)   (mg/L)   (//g/L)
    Raw Water      1.01     0.050     7.0   15      ND     ND      NA      NA      NA    NA

CHLORINE

    Scd Basin
     Effluent       1.03     0,023     6.7   15      0.2    0.35     0.19      0.07     0.22     140

CHLORINE DIOXIDE

    Sed Basin
     Effluent       1.11     0.028     6.8   15      0.05   0.06     0.21      1.16     0.53     69
NA  - Not Analyzed
ND  - Not Detected

-------
                 Treatment Modification Results and Discussion


Results of the DBF sampling are presented in Figures 6-68 through 6-73.   Figure 6-68
shows the effect of the  two different preoxidants on  XDBPsum and the DBF classes.
From this figure,  it is apparent that  even in the relatively short detention time in the
sedimentation basins, the use of chlorine dioxide preoxidation  resulted  in  lower levels
of all  measured  DBFs  compared to  chlorine treatment.   XDBP    was  almost 50%
lower  with  chlorine dioxide treatment.   However,  the  figure indicates that chlorine
dioxide treatment produced detectable levels of DBFs, most likely due, at least in part,
to the presence of free chlorine in the chlorine dioxide generator product.

Figure 6-69 illustrates THM levels resulting  from the two  preoxidation  schemes. The
advantages of chlorine dioxide as a THM control method are apparent, since levels of
chloroform, bromodichloromethane and dibromochloromethane were 52,  75 and  74
percent  lower,  respectively,  with  the  chlorine  dioxide  treatment  compared to the
chlorine-only treatment.

HAN formation was also substantially  lower with the chlorine dioxide treatment, as
illustrated in Figure 6-70. The concentration of DCAN was over 60 percent lower with
chlorine dioxide preoxidation (1.0 /ug/L) compared  to free chlorine preoxidation (2.9
//g/L).   Figure 6-71 shows HK  levels  produced by  the  two preoxidation schemes.
Again,  chlorine dioxide  treatment produced  significantly  lower levels of the two HK
compounds.  The concentration of I, I, I-TCP was significantly lower for  the chlorine
dioxide treatment (0.2 /vg/L) as compared to the chlorine treatment (1.6  //g/L).  As this
ketone is a  chloroform precursor, the previously  discussed lower TTHM concentration
with the use of chlorine dioxide treatment is consistent.

HAA levels are plotted in Figure 6-72.   MCAA concentrations  were approximately the
same for the  two treatments and DC A A levels  were  only 28 percent lower  for the
chlorine dioxide treatment.  However, the concentration of TCAA was over 70 percent
lower with chlorine dioxide treatment.  Figure 6-73 shows the levels of miscellaneous
compounds  for  the  two treatments.   Very  little difference in the concentrations of
chloropicrin and  cyanogen chloride was observed.  Chloral  hydrate was  significantly
lower with chlorine dioxide treatment (1.8 //g/L) compared to  chlorine treatment (4.2
//g/L).

Discussion

Use  of chlorine  dioxide  has been identified as  a  treatment  technology  capable  of
enabling utilities to lower THM levels.   Chlorine dioxide preoxidation allows chlorine
addition  to  be delayed  until after coagulation,  sedimentation and filtration  processes
have  removed  precursors to some extent.   Chlorine  dioxide  does  not form THMs,
although  a  small  amount  of  TOX  may  be  formed  (Chow and Roberts,  1981;
Fleischacker and Randtke.  1983; Werdehoff and Singer,  1986; Lykins  and Griese,
1986).  Additionally, some  researchers  have  found that chlorine dioxide is capable of
lowering the concentration of THM and TOX precursors by oxidation (Rav-Acha et al.,
1985: Werdehoff and Singer,  1986; Lykins and Griese,  1986).  Thus, chlorine dioxide
preoxidation in combination with free chlorine residual disinfection would be expected
to produce lower levels of DBFs.

Although  the  results of the  study at  Utility  16 were  inconclusive with  respect  to
reductions in  levels  of  DBFs produced  by chlorine dioxide preoxJdation, Utility 37

                                     6-14

-------
_!
O
o
U
a
m
Q
                 Effect  of  C'2  AND  CIO2
                  «;>n  L'BP  Formation  (Utility  3i')
            XDBP
                    THM
                              DbP Class

                              FIGURE 6-68
                                            HA A
                                                    ALD
5
*-•
(0
k_
4-*
0)

u

i
       40
       30
       20
       10
                  Effect  of  CI2  and CIO2
                  on THM Formation (Utility 37)
              CHCI3
                                           CI2

                                           CIO2
                                          Below  MRL
                        CHBrCll
CHBr2Cl
                                                 CHBr3
                             THM Compound

                             FIGURE 6-69

-------
o

-t-*

ro
V.
(J


Z
        4.O
        3.0 -
        1.O -
        o.o
                  Effect  of  CI2  and CIO2

                  on HAN  Formation  (Utility  37)
               TCAN
DCAN
BCAN
                            I-IAN Compound



                             FIGURE 6-70
8
*-*
(0
*-•


1
o
(J
                  Effect  of  CI2  and  CIO2 .

                on Haloketone Formation (Utility  37)
        0.5 -
        0.0
                   1.1-DCP
               1.1,1-TCP
                             HK Compound



                           FIGURE 6-71

-------
 Oi
 s
 o
 5
 u
       30
       20
       10
        O
               Effect  of  CI2  and CIO2  on

              Haloacetic Acid Formation  (Utility  37)
                                        CI2



                                        CI02
                                   *  Below MRL
            MCAA
DCAA
TCAA
                           HA A Compound



                          FIGURE 6-72
MB A A
fl



I





O
                Effect  of CI2  and  CIO2

              on Misc. DBP Formation (Utility 37)
             CHP
              CNCl
                          UBP Compound



                          FIGURE 6-73
                TCP

-------
                 Treatment Modification Results and Discussion
 results indicated  that chlorine dioxide preoxidation produced  substantially lower levels
 of many DBFs, even with the presence of free chlorine  in the generator product and
 with some  degree of mixing between the two process trains.

 The chlorine  dioxide generator product was optimized for this study to contain  the
 lowest practical levels of free chlorine.  It is interesting to  note that routine operation
 of the generator at Utility  37  resulted in  approximately  equal amounts of chlorine
 dioxide and free  chlorine in the generator product (before generator optimization, 1.60
 mg/L of chlorine dioxide and 1.56 mg/L of total chlorine  were measured upon addition
 of the generator  product to the flocculated water).  The  plant superintendent reported
 that  the generator is  operated  in  this way to avoid  "kerosene"  and "bleach" odor
 complaints  from consumers  that occurred  when  the level  of free chlorine  in  the
 generator product was minimized.   Household odors associated with chlorine dioxide
 treatment,  including "kerosene" and "chlorinous", have been investigated  by Hoehn, et
 al.  (1989).  This chlorine dioxide  study  reports on DBF levels  produced when  the
 operation of the chlorine dioxide  generator was  optimized and does not necessarily
 reflect the  levels experienced under routine operations at Utility 37.

 Chlorite  and chlorate  are inorganic by-products of  chlorine  dioxide.  Currently,  the
 USEPA  recommends that the combined  residuals  of chlorine  dioxide,  chlorite and
 chlorate  not exceed 1.0 mg/L in distribution systems (USEPA, 1983).   Additionally,
 chlorine  dioxide, chlorite and chlorate  are among the disinfectants and DBFs targeted
 by the USEPA's Drinking Water Priority List (Federal Register,  1988).

 Levels of chlorite and chlorate in the distribution system samples of Utility 16 were low
 and  the  distributed  water easily met  the  recommended  level of 1.0 mg/L combined
 chlorine  dioxide,  chlorite and chlorate residual. Levels of chlorine dioxide, chlorite and
 chlorate  were not measured  in the distribution system of Utility 37, but  based  on the
 levels measured in the sedimentation basin effluent, this utility may not have been able
 to meet the 1.0 mg/L recommended limit under the conditions employed for this study.

 COAGULATION STUDIES

 Coagulation studies  were conducted in order to study the effect of coagulant dose on
 DBF  precursor removal.   Utilities 3  and 12 were  selected  because they  were able to
 adjust  their alum dose at  full scale  without  severely compromising the quality of the
 finished water.  Each applied low, medium  and high doses of aluminum  sulfate, after
 which  samples  were collected  and  analyzed for  DBFs and  various water quality
 parameters.

 Utility 3

 Utility 3  is a conventional  treatment plant in  the eastern United States with a 10.5-mgd
 capacity.   Figure 6-74 presents  a process  schematic  of the  plant and shows the three
 doses of alum which were  applied during the study.   Each alum dose was applied on a
 different  day; a period of at least 48 hours passed between initiating each  alum dose to
 assure adequate flushing of the plant before  collecting samples.  Samples of the plant
 influent raw water,  filter influent and  filter effluent were  collected  and  analyzed  for
TOC  and UV-254 absorbance.   Turbidity measurements were made on-site at these
 three  locations.   Bromide and chloride were also  measured  in  the plant influent.

                                      6-15

-------
  RAPID MIX   FLOCCULATION  SEDIMENTATION  FILTRATION
ALUM COAGULATION
              Caustic Soda. Alum
            Potassium Permanganate
                                                          Chlorine
                                       Polyphosphate, Fluoride
                                       Caustic Soda, Chlorine
To
Distribution
                             Alum Dosages (ppm)

                                  Low = 10
                                Medium = 19
                                  High = 40
      SCHEMATIC OF UTILITY 3 TREATMENT PROCESS
                               FIGURE 6-74

-------
                 Treatment Modification Results and Discussion
Chlorine is normally applied at two places in the plant, at the entrances to the clearwell
and to the distribution system.  However, during this study,  samples from  the filter
effluent were collected and 24-hour SDS tests were conducted.  Thus, no chlorine was
applied in the plant  to the water that was sampled.  A 3.5 mg/L dose of chlorine was
applied for  the  SDS testing.   A  description  of the protocol  for  the SDS test was
provided in Section 3 of this report.

Chlorine residuals,  as well as other  water quality parameters, are presented in Table
6-11.   The table shows that the pH values for each sampling point in the plant were
similar  for all three alum doses  (5.5  to 5.6 at the filter influents and  effluents).
Additionally, chloride and bromide levels were low, being detected at 5 mg/L and 0.01
to 0.02 mg/L, respectively.  Influent turbidity ranged from  1.1 to 1.6 NTU, with filter
effluent values of 0.21  NTU for the  low alum  test and 0.04 to 0.06  NTU  for the high
and medium alum tests.

Table 6-11 also  shows that the influent TOC concentrations ranged between 2.97 and
3.11 mg/L and the UV-254 absorbance between  0.120 and 0.133 cnr1.   The percent
removals of these  water quality parameters at each alum dose are presented in Table
6-12.   Percent removals of TOC increased from 25 to 50 percent in the filter influent
water as the alum dose increased from 10 to 40 mg/L.  At the low alum dose, a greater
percent of TOC was removed through filtration (20 percent) than at the medium and
high alum doses (15 and  9 percent,  respectively); this was probably due to the better
settling characteristics of the floe at these doses.

A pattern similar  to TOC removal  was observed  for UV-254  absorbance.  Percent
removal ranged from 39 to 69 percent through flocculation/sedimentation at the various
alum doses.  An additional 9 to 26 percent removal was obtained  through filtration.

Figure 6-75 presents the  effect of increasing alum dose on XDBPSIII11 and DBP classes.
In general, DBPs  decreased with increasing alum dose.   XDBPsum was lowered from
150 to 94 yug/L, and THMs from 86 to 55 //g/L as alum doses increased from  10 to 40
mg/L.  Figure 6-76 shows that chloroform was the predominant THM; this is consistent
with the low  bromide levels found  in the plant  influent.  Chloroform concentrations
were reduced from 74 to 45 x/g/L as alum doses increased.  Changes in coagulant dose
had little effect upon the  other chlorinated THMs, although these were detected only at
levels less than 11 /sg/L.

The effect of alum dose on other individual DBPs is presented in Figures 6-77 to 6-80.
DCAA  and TCAA  followed  patterns  similar to  chloroform  (Figure 6-77) as  did
1.1.1-TCP  (Figure 6-78), DCAN (Figure 6-79) and  chloral hydrate  (Figure 6-80).  It
should be noted that some DBPs may have undergone hydrolysis since the SDS  tests
were conducted at  pH 8.2; however,  this effect would have  been  consistent for all alum
doses tested.

As part of the two coagulation studies, residual aluminum was  measured  in the plant
influent and in the actual distribution system at a residence time of approximately 24
hours.   Table 6-11 shows that levels in the  plant  influent were similar on  the three
sampling dates (46 to 50 /ug/L)  and that the  levels in  the distribution system were
higher (75 to  124 //g/L),  although  residual aluminum was the lowest in the distribution
system sample with the highest alum  dose.

                                      6-16

-------
                                    TABLE 6-11

                          UTILITY 3 TREATMENT STUDY
                              Water Quality Parameters
Sampling TOC
Location (mg/L)
Plant Influent
Low Alum
Mecl Alum
Hi Alum
Filter Influent
Low Alum
Meil Alum
Hi Alum
Filler Effluent
Low Alum
Med Alum
Hi Alum
24 Hr. Simulated
Low Alum
Med Alum
Hi Alum

3.11
3.08
2.97

2.32
1.87
1.48

1.71
1.41
1.22
UV
absorbince Chloride
(cm"1) (mg/L)

0 122
0.133
0.120

0.074
0.047
0.037

0.043
0.030
0.026

5
5
5

NA
NA
NA

NA
NA
NA
Bromide
(mg/L)

0.02
0.01
0.02

NA
NA
NA

NA
NA
NA
Turbidity
(NTU)

1.32
1,62
1.09

1.10
0.68
0.44

0.21
0.06
0.04
Aluminum
tf/g/L)

47
30
46

NA
NA
NA

NA
NA
NA
TOX
U/g/L)

NA
NA
NA

NA
NA
NA

NA
NA
NA
Free
Chlorine
Residual
(mg/L)

ND
NA
ND

ND
ND
ND

ND
NA
ND
pH

6.4
fi.3
6.4

5.5
5.5
5.5

5.5
5.6
5.5
Temp.
<°C)

14.5
16.0
14.5

14.0
16.0
13.5

14.5
16.0
13.5
Distribution System Test
NA
NA
NA
NA
NA
NA
NA
NA
NA
NA
NA
NA
NA
NA
NA
12<
871
75*
240
190
150
1.2
1.3
1.9
8.2
8.2
8.2
25
25
25
NA  =  Not Analyzed
ND  =  Not Delected
*    =  Aluminum data from actual plant distribution samples.

-------
                           TABLE 6-12


          PERCENT REMOVAL OF TOTAL ORGANIC CARBON

           AND ULTRAVIOLET ABSORBANCE AT 254 nm BY

                VARIOUS ALUM DOSES AT UTILITY 3
                      PLANT INFLUENT          PLANT INFLUENT
                    TO FILTER INFLUENT      TO FILTER EFFLUENT
                      (Percent Removal)          (Percent Removal)
Total Organic Carbon

    Alum Dose
       Low                  25                      45
       Medium               39                      54
       High                  50                      59

UV-254 Absorbance

    Alum Dose
       Low                  39                      65
       Medium               65                      77
       High                  69                      78

-------
 8
 u
      200
      150 -
      1OO -
       50 -
                     Effect  of  Alum  Dose
                 on DBP  Formation  (Utility 3,  SDS)
                            
-------
_J

3
U
       1O -
       0
                    Effect  of  Alum  Dose
                on HAA Formation (Utility 3. SDS)
                                         Low Alum Dose

                                         Medium Alum Dose

                                         High Alum Dose
            MCAA
                      DCAA
                            HAA Compound

                            RGURE6-77
o»
ra

I
o
U
     1.0O
     0.75
     0.50
     0.25
     0.00
                    Effect  of  Alum  Dose
                 on HK  Formation  (Utility 3,  SDS)
                Low Alum Dose

                Medium Alurn Dose

                High Alum Dose
                  1.1-DCP
                                         1.1,1-TCP
                            HK Compound

                           FIGURE 6-78

-------
o>
(0
U
       3 -
       O
                    Effect  of  Alum  Dose
                on HAN  Formation  (Utility  3.  SDS)
            -**•*•
              TCAN
                                     Low Alum Dose

                                     Medium  Alum Dose
                                     High Alum Dose
                                 * Below MRL
DCAN
BCAN
                            HAN Compound

                             FIGURE 6-79
                                                 DBAN
 o>
 
-------
                 Treatment Modification Results and Discussion
 Utility 12

 Utility 12 is a conventional  treatment plant in the western United States with a 72-mgd
 capacity.   Figure  6-81  presents a process schematic of the plant and shows the three
 doses of alum  applied  during the study.  Chlorine was  added at two locations  in  the
 plant:   before  the rapid mix (1.8  mg/L dose) and before filtration (1.3 mg/L  dose).
 The total chlorine contact time was approximately 100 minutes.  Ammonia was added
 approximately 4 minutes after filtration, prior to the clearwell, in order  to convert  the
 free chlorine  residual  to chloramines.   Chlorine  residuals and other  water  quality
 parameters are  shown in Table 6-13.

 Samples of the raw water,  sedimentation basin  effluent, filter effluent  and clearwell
 effluent  were  collected and analyzed for  all  DBFs.   Sampling  was  conducted a
 minimum of three plant detention times after alum doses were changed.

 Table 6-13 presents various  water quality parameters for each alum dose  applied. Free
 and total chlorine  residuals  and temperatures were similar throughout the study.  Since
 the plant had no capability  to control pH before or during  the sedimentation process,
 the  pH values  decreased  as alum  dose increased.   It is important to  note that  the
 bromide levels were 0.14 to 0.15 mg/L during the treatment modification study.  These
 bromide levels  are in contrast to the quarterly sampling,  in which 0.41 mg/L and 0.79
 mg/L were detected during  the summer, 1988 and winter, 1989 quarters, respectively.
 The effect  of elevated levels of bromide on DBFs  as  compared to  the concentrations
 found during the  treatment  study was discussed in more detail in Section 5  of this
 report.

 Residual aluminum was measured in the plant influent and clearwell  effluent.   Table
 6-13  shows that  increasing the  alum  dose  did  not  increase  the  residual aluminum
 leaving  the  plant.   In fact,  removals of 71  to 76 percent of influent aluminum were
 observed.  These results are  in contrast to that observed at Utility 3, indicating a need
 to  better understand  aluminum  chemistry  in  relation  to the coagulation process.
 Influent turbidity  ranged from  II to  17 NTU, with clearwell  effluent values of 0.14
 NTU for the low alum test and 0.06 to 0.07  NTU for the medium and high alum  tests.

 Table 6-14 shows  that the low alum treatment removed 33 percent of the influent TOC,
 while the medium and  high alum doses removed 47  and 46 percent, respectively, as
 measured in the filter effluent.  It is important to note, however, that most of the TOC
 removal  occurred  in  the  sedimentation basins,  with  little  or  no additional removal
 having occurred through filtration.  Removal  of UV-254 absorbance ranged from 76 to
 86 percent  from plant influent to filter effluent.

 Figures  6-82 to 6-88 present the DBF data  from the  clearwell effluent  in relation to
 alum dose.  Figure 6-82 shows the effect of alum dose on DBF  concentrations by class.
 For XDBP  ,  concentrations decreased from  87 to 69 //g/L as the alum dose increased
 from 25 to 75  mg/L; for THMs, the levels decreased from 53 to 39 /ug/L.  Little or no
 change  was observed  for  the other  DBF  classes.    Figures  6-83  to  6-88 present
 individual DBF  concentrations  by DBF class and the miscellaneous compounds.   In
 general, individual THMs and HAAs decreased slightly  with increasing alum dose; little
or no change was  observed  for individual HKs, HANs and ALDs, or for chloropicrin,
chloral hydrate, and cyanogen chloride.

                                      6-17

-------
    RAPID MIX
FLOCCULATION
SEDIMENTATION
FILTRATION
  ALUMJCOAGULATION
 Alum
Chlorine
                               Fluoride, Polymer
                                Lime, Chlorine
                                          Sodium Hydroxide
                                             Ammonia
                             Alum Dosages ( ppm)

                                 Low = 24.6
                                Medium = 45.7
                                  High = 73
          SCHEMATIC OF UTILITY 12 TREATMENT PROCESS
                              FIGURE 6-81

-------
                                        TABLE 6-13
                            UTILITY 12 TREATMENT STUDY
                                 Water Quality Parameters
  Sampling
  Location
         UV
      Absorbftncc
TOC  at 234 nm
(mg/L)
(cm-1)
Chloride
 (mg/L)
                         Free     Toul
                        Chlorine  Chlorine
Bromide Turbidity  Aluminum  Residual  Residual
(mg/L)   (NTU)    OWg/L)   (mg/L)    (mg/L)
                                                          pH  Temp.
Plant Influent
Inw Alum
Mod. Alum
High Alum
Scdimcntnlion
Basin Effluent
Ijow Alum
Mcd. Alum
High Alum
Filter Effluent
Low Alum
Mod. Alum
High Alum
2
2
2


2
1
1

1
1
1
.XX
.92
.76


.07
.55
.42

.92
54
.48
0.123
0.123
0.121


0.051
0.032
0.027

0.030
0.021
0.017
50
44
49


NA
NA
NA

NA
NA
NA
0.15
0.14
0.15


NA
NA
NA

NA
NA
NA
11.0
17.0
13.5


2.2
1.2
1.7

O.J3
0.04
0,04
450
350
370


NA
NA
NA

NA
NA
NA
ND
ND
ND


0.1
0.1
0.1

1.0
1.0
0.8
ND
ND
ND


0.2
0.2
0.5

1.2
1.1
1.0
8.6
8.5
8.6


7.4
7.0
6.7

7.4
7.4
6.9
25
23
23


23
22
22

23
22
22
Clcarwcll Effluent
Low Alum
Mcd. Alum
High Alum
2.
1.
1.
00
57
42
0.041
0.025
0.022
NA
NA
NA
NA
NA
NA
0.14
0.06
0.07
110
94
107
0.1
0.1
0.1
1.0
1.0
1.0
8.6
8.7
8.6
23
22
22
NA = Not An«ly7fil
ND = None Detected

-------
                           TABLE 6-14


          PERCENT REMOVAL OF TOTAL ORGANIC CARBON
           AND ULTRAVIOLET ABSORBANCE AT 254 nm BY
                VARIOUS ALUM DOSES AT UTILITY 12
                     PLANT INFLUENT          PLANT INFLUENT
                    TO FILTER INFLUENT      TO FILTER EFFLUENT
                      (Percent Removal)           (Percent Removal)


Total Organic Carbon

    Alum Dose
       Low                  28                       33
       Medium               47                       47
       High                  49                       46

UV-254 Absorbance

    Alum Dose
       Low                  59                       76
       Medium               74                       83
       High                  78                       86

-------
 
-------
 o»
 0)
 o

 b

 C
       20
       10 -
                     'Effect  of  Alum  Dose

                   on HAA Formation (Utility  12)
                                         Low Alum Dose




                                          Medium Alum Dose




                                         High Alum Dose
                            I iAA  Compoui ul




                            FIGURE 6-84
                                                  DBA A
Ol
o
o
     2.00
                    Effect  of  Alum Dose

                   on  HK  Formation  (Utility  12)
                 Medium Alum Dose
     O.25 h
     o.oo
                  1.1-DCP
1.1,1-TCP
                            HK Compound


                           FIGURE 6-85

-------
(0
o
U
                    Effect  of  Alum  Dose

                  on  HAN  Formation (Utility  12)
                  Medium Alum Dose
             » Balow MinMTun Reporting Level
              TCAN
                  DCAN
BCAN
                            HAN Compound



                            FIGURE 6-86
DBAN

8
u
3





2
       0
                    Effect   of  Alum  Dose

               on  Misc.  DBP Formation (Utility   12)
                CNCI
                                     Low Alum Dose




                                     Medium Alum Dose




                                     High Alum Dose
                        CH




                    DBP Compound



                    FIGURE 6-87
          CHP

-------
o»



5
•«-^
a



-------
                 Treatment Modification Results and Discussion
That  greater  DBF  removal  was  not  observed  with increasing  alum  dose  was
undoubtedly due to the utility's prechlorination practices.  Approximately 1.8 mg/L of
free chlorine was added to the raw water, with 75  minutes of contact time from the
point of addition to the sedimentation basin effluent.  Free chlorine residuals measured
in the basin effluent for all three trials were 0.1 mg/L.  Consequently, DBF formation
occurred before and during TOC removal processes, as demonstrated by  the similar
concentrations of THMs detected  in  the  sedimentation basin effluent (Table  6-15).
However,  from the sedimentation basin effluent to the clearwell effluent,  an additional
18 /yg/L of THMs formed at the low alum dose  while  only 9 //g/L were produced at
high alum dose.  A similar trend, although to a lesser extent, was observed for HAAs.

Table 6-15  also shows that  THMs increased  from the  plant influent  to  the clearwell
effluent.   However, other classes of DBF compounds  such as  HKs,  and  to a lesser
extent. HANs. decreased from the filter effluent to the clearwell effluent; similar results
were observed for TOX. Since Utility  12 adds sodium hydroxide to increase the pH for
corrosion  control,  the observed decreases were probably due to hydrolysis of these
compounds.  These results are consistent  with those observed  by Reckhow  and  Singer
(1984) who showed  that TOX and chloroform exhibited opposite  pH dependencies.
However,  as noted in Section 3, problems with pn-site dechlorination  of TOX samples
may have  accounted for additional TOX formation during shipping of the  chlorinated
in-plant samples (this should not have been a problem for the chloraminated clearwell
effluent  sample).  Additionally. Reckhow and Singer (1985) showed  the instability of
1.1.1-TCP at high pH, as did Oliver (1983) for DCAN.

Discussion

Several studies have focused upon  the use of alum coagulation for removal of THMs
and  other  organic precursor materials from  water  (Kavanaugh. 1978;  Babcock and
Singer,  1979;  Semmens and Field,  1980; Dempsey  et al.,  1984; Hubel and Edzwald.
1987).  The optimum  pH for coagulation of humic substances with alum has been
reported to  be between  5 and 6 (Kavanaugh, 1978; Babcock and Singer  1979).  As
noted above, studies at Utility 3 were conducted at pH  5.5 to 5.6.  At Utility 12, the
pH ranged from 6.7 to 7.4 in the sedimentation basin.

Using a 40  mg/L dose of alum. Utility  3 was able to  achieve a TOC removal of 50
percent in the sedimentation basin  effluent (filter influent).  At this sampling point at
Utility 12  for a 73 mg/L alum dose, TOC was reduced by 49  percent.  These percent
removals are consistent with  those reported by Singer (1988), which ranged  from 32 to
66 percent.  In general, increasing alum dose decreased the levels of DBFs that were
produced.    At Utility 3. the order of increasing  percent removal of DBF and TOX
precursors. TOC.  and UV-254 absorbance  at the  high alum dose versus the low alum
dose (reference dose) was as follows:

        TCAA > UV-254. DCAA  > DCAN, TOX, THMs > HKs, TOC

These removals were similar to those reported by  Reckhow and Singer  (1984) on Black
Lake. North Carolina treated water for UV-254, and TOX and DBF precursors:

         UV-254  > DCAN, TCAA > TOX, THM,  DCAA, > 1,1,1-TCP


                                    6-18

-------
                               TABLE 6-15

 CONCENTRATIONS 0>g/L) OF VARIOUS DISINFECTION BY-PRODUCTS AND TOTAL
          ORGANIC HALIDE AT VARYING ALUM DOSES AT UTILITY 12
DBF
Trihalo-
meihancs

Haloacctic
Acids

'Haloaccto-
nitriles

Halokciones


Chloropicrin


Chloral
Hydrate

Cyanogen
Chloride

Total Organic
Halide

Alum
Dose
low
medium
high
low
medium
high
low
medium
high
low
medium
high
low
medium
high
low
medium
high
low
medium
high
low
medium
high
Plant
Influent
0.33
0.13
0.16
ND
ND
ND
ND
ND
ND
ND
ND
ND
ND
ND
ND
ND
ND
ND
ND
ND
ND
18
18
16
Sedimentation
Basin
Influent
35
34
30
16
18
17
4.9
5.0
4.9
2.5
2.7
3.0
0.059
0.068
0.070
2.6
2.8
2.6
2.0
2.1
2.2
140
110
120
Filter
Effluent
45
43
36
27
26
23
6.1
6.1
6.0
3.0
3.0
3.2
0.088
0.094
0.10
3.6
3.6
3.1
2.7
2.9
3.3
210
170
170
Cleat-well
Effluent
53
48
39
25
25
17
5.1
5.0
4.8
1.4
1.5
1.6
0.088
0.10
0.094
3.4
3.3
3.0
3.5
3.7
3.8
140
120
110
ND = Not Detected

-------
                 Treatment Modification Results and Discussion
It was not possible to assess removals of these compounds for Utility  12 since the plant
influent  was   chlorinated.    It  is  important  to  note   that  Utility  12  practices
prechlorination for disinfection  and taste-and-odor control.   Thus, the advantages of
optimized coagulation can be lost if a suitable preoxidant/disinfectant is  not available.

Utilities 3  and 12 showed greater TOC removal (39 to 50 percent, from  plant influent
to filter influent) at the medium and high alum doses as compared  to the  average TOC
removal by the filtering utilities participating in this  study  (21  percent).   It should be
noted, however, that the treatment practices by the various utilities  surveyed focused on
turbidity,  rather than on TOC removal.  It  is  not possible to determine  whether high
TOC removals, as observed at Utilities 3  and 12, would result at other utilities by
improving  the coagulation process.  Such an  assessment would need to be  conducted on
a case by case basis, since some surface waters may not be amenable  to greater TOC
removals by conventional treatment.

Assessing costs associated with increased alum doses  was not part  of the scope of this
project.  However, applying higher alum doses for greater DBF precursor removal will.
in  many  cases,  increase  the  concentrations  of  total dissolved  solids  due  to  pH
adjustment for coagulation.   This is particularly the  case for well  buffered  waters.
Moreover, the amount of chemical sludge produced due to  elevated alum doses will be
increased.  Therefore, costs associated  with  additional  sludge handling, treatment  and
disposal should be considered.

GRANULAR  ACTIVATED CARBON STUDY

The GAC study  conducted  at Utility   11  was unique among the  treatment  studies
conducted  for this project, in  that samples were collected during a  GAC contactor  run.
rather than collecting one or two sets of samples  at different treatment conditions.  In
all. six sets of samples were collected over a period of almost four months.

Process Description

Utility 11 operates a large  (235 mgd) conventional treatment facility for a low-organics
river water.   This Midwestern  utility also operates  demonstration facilities  for GAC
adsorption and regeneration at  the  site of the full-scale plant.  Numerous  pilot  and
demonstration-scale  studies have  been  conducted at  this utility  in  recent  years to
evaluate the performance and economic feasibility of GAC for removing  trace  organic
compounds from the plant's source water, a river water receiving industrial discharges
and  susceptible to spill  events.  Studies at this  utility  have also  evaluated  the feasibility
of GAC  treatment for  removing THM  and TOX precursors (McGuire  et al.. 1989:
DeMarcoet al., 1981:  Clark,  1987; Miller and  Hartman, 1982).

A schematic diagram of the full-scale and demonstration facilities is presented  in Figure
6-89.  The full-scale plant's process train includes rapid mix with alum or alum  and
polymer addition:  flocculation; lamella separation;  reservoir storage for  3 days; addition
of  lime,  fluoride and  intermittent  addition of ferric  sulfate to  a  hydraulic jump;
sedimentation  and filtration.   Free chlorine is added at the hydraulic jump, and at the
filter influent  and effluent.   For the GAC demonstration facilities,  a sidestream  was
diverted after  presedimentation to a rapid sand filter  (identical  to the full-scale plant's
filters) and then to two separate GAC contactors, each  with a  hydraulic  capacity of I

                                      6-19

-------
     RAPID MIX    FLOCCULATION   SEDIMENTATION   FILTRATION
PRETREATMENT:
Alum
Polymer
1

T























V



\



\



\





^






Settler



\ Reservoir /
^\ Storaee /
\ (3da>si /

CONVENTIONAL PLANT:
                                                                       -
 Chlorine
 Lime
 Fluoride
 Feme Sulfate'
DEMONSTRATION  PLANT:

                            T
                           Fluoride
                           Ferric Sulfate'
              * Ferric Sulfate added intermittently (in use on 5/8/89,6/12/89 and 7/17/89)

               SCHEMATIC OF UTILITY 11 TREATMENT PROCESS
                                  FIGURE 6-89

-------
                 Treatment Modification Results and Discussion
mgd.   The GAC facilities  received unchlorinated water and  no chlorine was added
during  treatment in  the demonstration plant until after GAC treatment (GAC  effluent
samples for this study were collected before the addition of chlorine).

The GAC contactors have a diameter of 11 feet.  The contactor sampled for  this study
had a  15-foot carbon depth  and was operated at an empty bed contact time (EBCT) of
 15  minutes.  The carbon was Calgon Filtrasorb 400 (12x40 mesh).

GAC Contactor Operation

Previous studies at this utility have found that dioxins were formed upon regeneration
of GAC if chlorinated water had been applied to the carbon, due to the combustion of
adsorbed chlorinated compounds (DeMarco  et al.. 1988).  Thus, recent operation of the
GAC contactors at  this  utility involves  somewhat intermittent operation  to  avoid
application of chlorinated water to the GAC.

At  Utility II, the rapid sand filter  upstream of the GAC contactors is backwashed with
chlorinated (finished plant) water.   Filter run lengths varied during the GAC contactor
run  from  approximately  12  to 50 hours.   When backwash was  initiated,  the GAC
contactor  was taken  off-line.  The sand  filter was backwashed and then flushed with
unchlorinated  water  until such time as residual chlorine was undetectable  in the filter
effluent.  For this reason, each time the dual  media filter  was backwashed,  the GAC
contactor  was off-line for  periods  of  1.5  to as much as  4  hours.   Thus,  the GAC
contactor  was operated somewhat  intermittently, and the "run time"  of the contactor
did not coincide with "calendar time".  Although the sampling period  for  this study
was April  10.  1989  to July 31. 1989, a period of 113 calendar days, the actual run
time of the GAC  contactor  was 95 days.   One  "run day"  of the GAC  contactor was
defined by the utility as the time required to pass I million gallons of water through the
contactor at the  I -mgd nominal hydraulic loading rate of the contactor.

Treatment Study Description

The objectives of the treatment study at Utility  11 were twofold:

     o   To evaluate the effectiveness of GAC over time  in removing  DBF precursors;
         and
     o   To determine DBF levels produced by  conventional treatment as  practiced in
         the full-scale plant.

The first objective was achieved by sampling the GAC column influent and  effluent and
performing SDS tests on those samples.  Immediate DBF samples were also collected at
the  GAC  influent  and effluent as  a control (no chlorine added), to determine if the
source water  contributed  to measured levels  of DBFs.    The  second objective was
achieved by collection of DBF samples in  the conventional plant and at one distribution
system location.

Table 6-16  summarizes the  sampling  plan for  the  treatment study.  The  sampling
locations are  defined  in  Table  6-17.   Six  sample sets were  collected  during the
treatment study.  Samples were collected at the conventional and demonstration plants
on  Mondays.   Bottles were filled for immediate DBF analyses, as well  as for SDS

                                     6-20

-------
                                                  TABLE 6-16

                                          UTILITY 11 TREATMENT STUDY
                                                 SAMPLING PLAN
Sample
Location

1.
2.
3.
4.
5.
6.
7.
8.
9.
10

Raw
Al
A2
Bl
B2
B21
B22
B3
B31
.B33
Sample
Type

Immediate
Immediate
Immediate
Immediate
Immediate
SDS 4.5
SDS 4.5 tab.
Immediate
SDS 4.5
SDS 2.0
pH

X
X
X
X
X
X
X
X
X
X
Temp.

X
X
X
X
X
X
X
X
X
X
Analyses
C12 TOC/UV

X
X
X
X
X
X
X
X
X
X

X
X
X
X
X


X


Cl-/Br~ DBPs ALDs Comments
(incl. Only
ALDs)
X X

X

X
X
X
X
X
X
            11.A3
            12.Blank
Immediate
 SDS 4.5
Collected 3 days after
  collection of
  Samples 1-10
Buffered to pH 8.2,
  addition of Br~ to
  level measured in
  raw water
Note: DBF analyses  include THMs, HANs, HKs, HAAs, CH, CHP, CNCl  and ALDs.
1.    SDS protocol:  Cl2 dose  = 4.5 mg/L, pH = 8.2, Temp. = 25°C, Time = 3 days.
2.    SDS protocol:  Clz dose  = 4.5 mg/L, pH = 8.2, Temp. = Temp, of clearwell  effluent  at  time  of  sampling,
        Time =  3 days.
}.    SDS protocol:  Cl 2 dose  = 2.0 mg/L, pH . ambient, Temp.  =  25°C, Time = 3  days.

-------
                                TABLE 6-17
                    UTILITY 11 SAMPLING LOCATIONS
ABBREVIATION  LOCATION
        Raw       Plant Influent
        AI         Full-Scale Plant. Filter Influent (After primary and secondary
                  sedimentation, and prechlorination)
        A2         Full-Scale Plant. Clearwell Effluent
        A3         Full-Scale Plant, Distribution System (Detention time = 3 days)
        Bl         Demonstration Plant, Filter Influent  (After primary
                  sedimentation, no chlorination)
        B2         Demonstration Plant, Filter Effluent/GAC Contactor Influent
        B3         Demonstration Plant, GAC Contactor Effluent

-------
                 Treatment Modification Results and Discussion


 testing of the unchlorinated GAC influent and effluent, and were shipped overnight to
 Metropolitan for analysis and set-up of the 3-day SDS tests.  On Thursdays following
 the  Monday sampling  dates, the distribution  system was sampled.   A  distribution
 system sampling point was selected which had an approximate residence time of 3 days,
 determined  by fluoride tracer studies (conducted by Utility 11) and checked monthly
 (by  Utility 11) by comparing simulated distribution system THM and actual distribution
 system THM  sample results.   Thus,  the water sampled  at  this point on a given
 Thursday was the same water  leaving the  plant's  clearwell  effluent the  previous
 Monday.  In this way. formation of DBFs in the distribution system could be evaluated,
 although not on a strictly controlled basis.

 Simulated Distribution System Testing

 The SDS  test protocol was described in Section 3.  For this study, all SDS testing was
 performed at  Metropolitan according to Metropolitan's SDS protocol, which had  not
 been developed specifically to simulate Utility II 's  distribution system.  The protocol
 called for buffering to  pH 8.2,  and incubating at 25°C for three days.   The applied
 chlorine dose of 4.5 mg/L was determined  adequate to insure a chlorine residual of at
 least 0.5 mg/L after  3 days of incubation.  Two separate chlorine demand tests were
 conducted on  Utility  11 's unchlorinated  GAC  column influent water, one in mid-
 February  and the other in  early April. 1989. before start-up  of the GAC column.
 Although the GAC column was expected to  remove a substantial amount of the chlorine
 demand, especially at the beginning of the column run. it was determined that the dose
 of 4.5 mg/L would be applied to both the GAC column influent and effluent in order to
 provide a basis of comparison for the SDS DBFs in the column influent and effluent.
 In practice,  the chlorine  dose required by the GAC  column effluent is significantly
 lower than that required by the column influent.

 An additional SDS sample was run on  the GAC column effluent at identical conditions
 as the other  contactor effluent SDS  samples,  except the chlorine dose was 2.0 mg/L
 and  the samples  were run at ambient pH.  These  samples were  run  to approximate
 operating  conditions anticipated by Utility 11  when future full-scale GAC facilities are
 on-line (i.e.. lower chlorine dose and no lime addition) (see Section 3 of this report).
 On  two sampling dates, a third  SDS protocol was used,  in which  a GAC influent
 sample was incubated at the temperature of Utility 11's clearwell effluent at the time of
 sample collection. Results of this test provided a basis of comparison between actual
 and  SDS data (see Section 3 of this report).

Simulated Distribution System Testing Results and  Discussion

Water quality parameters for the immediate  samples collected on all six sampling dates
are reported  in Table 6-18.  As  reported in  the  table. TOC  levels in the  raw water
 ranged from 1.75 to 2.83 mg/L, and TOC levels in the GAC column influent (sampling
 point "B2")  fell within the range  1.37  to 2.31 mg/L.  The TOC removal performance
of the GAC  column is plotted in  Figure 6-90.  TOC breakthrough, defined by Utility
 11 as  a TOC concentration of 1.0  mg/L  in  the combined effluent of on-line GAC
contactors (only one of which was sampled for this study), did not occur within the
95-day sampling period.  The GAC was very effective  for TOC removal.  For the first
25 run days,  the contactor effluent TOC remained the same as that measured on Run
Day  0.2 (0.1 mg/L), which represents the non-adsorbable fraction of the TOC.  After
                                     6-21

-------
         TABLE 6-18

UTILITY 11 TREATMENT STUDY
    Water Quality Parameters
Sampling TOC
Location (mg/L)
Absorbance
at 254 nm
(cm1)
Free
Chlorine
Residual
(mg/L)
Total
Chlorine
Residual
(mg/L)
pH
Temp.
(°C)
Run Day 0.2 (4/10/89)
Raw
Al
A2
A3 (4/13/89)
Bl
B2
B3
2.83
.88
.77
.78
.88
.80
0.10
0.175
0.072
0.063
0.062
0.074
0.070
0.032
NA
2.08
1.60
0.9
NA
NA
NA
NA
2.30
1.78
NA
NA
NA
NA
7.6
8.2
8.6
8.6
7.0
7.1
10.0
10
16
9
11
9
12
9.4
Run Day 13 (4/24/89)
Raw
Al
A2
A3 (4/27/89)
Bl
B2
B3
Run Day 25 (5/8/89)
Raw
Al
A2
A3 (5/11/89)
Bl
B2
B3
2.10
.54
.51
.52
.87
.52
0.10

.75
.59
.56
.49
.60
.37
0.10
0.148
0.050
0.048
0.021
0.058
0,058
0.029

0.161
0.094
0.024
0.022
0.087
0.042
0.003
NA
2.23
1.66
0.7
NA
NA
NA

ND
2.10
1.64
0.7
ND
NA
ND
NA
2.41
1.75
0.8
NA
NA
NA

ND
2.30
1.88
0.8
ND
NA
ND
7.6
8.5
8.5
8.7
7.3
7.3
7.3

7.6
9.0
8.9
8.8
7.2
NA
6.8
14
12
12
15
12
12
13

15
16
14
16
15
16
15
Run Day 54 (6/12/89)
Raw
Al
A2
A3 (6/15/89)
Bl
B2
B3
2.47
2.24
2.15
2.00
2.14
2.24
0.20
0.116
0.066
0.034
0.042
0.058
0.054
0.003
ND
1.94
1.53
0.3
ND
ND
ND
ND
2.23
1.68
0.4
ND
ND
ND
7.7
8.6
8.4
8.6
7.4
7.5
7.2
23
23
23
22
23
23
23

-------
                           TABLE 6-18 (Continued)

                       UTILITY 11 TREATMENT STUDY
                           Water Quality Parameters


Sampling
Location
Run Day 82
Raw
Al
A2


TOC
(mg/L)
(7/17/89)
2.42
2.22
2.08
A3 (7/20/89) 1 .95
B1
B2
B3
Run Day 95
Raw
Al
A2
A3 (8/
Bl
B2
B3
2.15
2.16
0.60
(7/31/89)
2.59
2.39
2.13
3/89) 2.17
2.30
2.31
0.85

Absorbance
at 254 nm
(cm1)

0.090
0.066
0.032
0.036
0.056
0.056
0.009

0.098
0.115
0.035
0.038
0.064
0.060
0.015
Free
Chlorine
Residual
(mg/L)

ND
1.58
2.08
0.4
ND
ND
ND

NA
1.71
2.05
0.3
NA
NA
NA
Total
Chlorine
Residual
(mg/L)

ND
1.82
2.48
0.5
ND
ND
ND

NA
1.96
2.18
0.4
NA
NA
NA


pH


7.6
8.7
8.4
8.4
7.5
7.4
7.2

7.7
8.7
8.4
8.5
7.6
7.4
7.2


Temp.
("0

27
27
27
26
27
27
27

27
27
27
NA
27
27
27
NA = Not Analyzed
ND = Not Detected

-------
U
              TOG  vs  Run  Time  for  GAG
            Column  Influent  and  Effluent. Utility  1 1
         +—+GAC mf             A	AGAC Eff
             Immediate                Immediate
                                  -A"
                           4G       6O

                             Run Days
                            RGURE 6-90
                                              .-A''
                                      8O
                           100
        1.0
                 TOG  Breakthrough  Profile
                           At  utility  1 1
u
u
u
o
O.6
        O.4
        O.2
        0.0
           0
           2O
4O       60

  Run Days

  FIGURE 6-91
80
                                                       100

-------
                Treatment Modification Results and Discussion


another 29 run days, contactor effluent TOC had only increased to 0.2 mg/L.  It was
not until the sampling on  the 82nd run day that a substantial increase in contactor
effluent TOC  was  measured (0.6 mg/L).   A further  13  run days had increased  the
contactor  effluent   TOC  to  0.85  mg/L,  showing evidence  of  imminent  TOC
breakthrough.

These TOC removal results are consistent with published data from previous GAC runs
at Utility 11's demonstration facilities.   DeMarco,  et al.  (1981) reported a non-
adsorbable TOC level  of approximately 0.2 mg/L in an example plot of a GAC run at
Utility 11  (16-minute EBCT).  For the same run, the column effluent TOC was within
the range of 0.8 to 1.2 mg/L after 95 days, while column influent TOC values  ranged
from  approximately 1.3 to 2.8 mg/L during the first 95  days of column operation.
Additionally.  Lykins. et al. (1988) reported non-adsorbable TOC levels of 0.2  from a
GAC  run at  Manchester.  New  Hampshire (23-minute EBCT),  and 0.5 mg/L from
Jefferson Parish. Louisiana (20-minute EBCT).

Figure 6-91 is the TOC breakthrough profile for Utility II, plotted in the form C/C0 as
a function of  run  time.  Additionally, Figure 6-92 shows breakthrough  profiles from
three  previous GAC column runs  at Utility II  (15-minute EBCT) published by Miller
and Hartman (1982).  In comparing Figures 6-91 and  6-92, even though the shapes of
the curves are different, the values of C/C0 at approximately 90 to 100 days range from
approximately 0.4  to 0.6 in the  breakthrough profiles shown in  Figure 6-92, and  the
value  of C/C0 at 95 days is approximately  0.4 in Figure 6-91.  It is interesting to note
that  the operation of the  GAC columns for the  data in  Figure 6-92  included
prechlorination upstream  of the  columns,  which  prevented  or  at least reduced  the
potential for microbiological activity within the GAC bed.  The curve shown in  Figure
6-91 was produced from data without chlorination upstream of the GAC column.

The TOC removal  results from Utility  11  are favorable compared to  published  results
from  other  utilities.    For  example,  in  an  extensive  pilot  study  conducted   by
Metropolitan  (McGuire et al., 1989),  influent TOC values for Colorado River water
ranged from approximately 2.0 to 3.1 mg/L (only slightly higher than the influent TOC
range for Utility  II).   However,  using a pilot GAC column  with  an  EBCT of  15
minutes and  12x40 mesh  carbon,  a column effluent TOC of 0.85 mg/L occurred after
only 30 run days, and  after 95 run days, the column effluent TOC was 1.7 mg/L.

TOC  removal  results from  GAC studies at other utilities occur between the favorable
results achieved at  Utility 11 on the one hand, and the relatively unfavorable results of
Metropolitan  on the other.  A pilot study was conducted at Jefferson Parish, Louisiana
using Mississippi  River water and  l-mgd GAC contactors with a 20-minute  EBCT.
Results  of this study,  reported by Lykins, et al.  (1988) and McGuire, et  al. (1989).
indicated that TOC was steadily removed by the GAC columns for up to  160 days, with
C/C0  continuing to increase during that time.  Influent TOC concentrations generally
ranged  from  approximately 2.5  to  3.5 mg/L.   After  120  to  160  days, C/C0 was
reported in the range  of 0.6 to 0.7 in  three separate column runs at  Jefferson  Parish.
At Manchester, New Hampshire,  pilot study results indicted that after approximately 35
days.  C/C0 had increased to approximately 0.7 and fluctuated around that value  for the
remainder of the 130-day column run (Lykins et al., 1988).  Influent TOC for the lake
water used in this  study generally ranged  from 2.2 to 2.8 m|/L.  Results of  a pilot
study  conducted at Philadelphia,  Pennsylvania, on  Delaware  River water was reported

                                     6-22

-------
    1.0

    0.9

    0.8

    0.7
             1O
 CONTACTOR A, 15 MM. EBCT
 SEPT. is, ieao TO DEC. so, ioeo
 CONTACTOR C, 15 MR EBCT
 AUG. 11, 1080 TO NOV. 12, 198O
 CONTACTOR O, 15 MUt EBCT
 SEPT. 1, 1080 TO DEC. 16, 1080
2O    3O    4O    SO    6O
    DAYS OF COLUMN OPERATION
                                                70
                                                      80
                                                                 1OO
                       TOC Breakthrough Profiles for Three
                       Demonstration-Scale Contactor RUM
                        •I Utility II (McGulre et •!., 1989>.


                             FIGURE 6-92
           SDS  TTHMs   vs  Run   Time   for  GAG
              Column  Influent and  Effluent, Utility  1 1
          +—+ GAC Inf               A	A GAG Eff
               3-day SDS                  3-day  SDS
i-
      2OO
       15O
       1OO
       5O
                CIS = 4,5
                pH =  3.2
                Terrp. = 25 C
                                  OO3
                                  All
                                                              .-A
           .-.••.• r:',". ".",".• ri-.Qr -.--.-1-1-1 -r TV-.-•.•-."•:
          0
  20
40
60
80
100
                                  Run  Days

                                  nr.l IRE 6-93

-------
                Treatment Modification Results and Discussion


by McGuire. et al (1989).  In this study, influent TOC concentrations ranged from  1.86
to 2.94 mg/L  and  the  columns were operated at an EBCT of 15 minutes.   C/C0
increased steadily over the first 65 days of the column run to approximately 0.4,  then
increased rapidly after  that time until C/C0  reached 0.6 after 85 days.   Thus,  it is
apparent that  the effectiveness  of GAC for  DBF precursor  removal, and hence the
feasibility of GAC as a  DBF control  method, is site-specific, with Utility  11's results
being favorable compared to those of some other utilities.


It was beyond  the scope of the treatment study at Utility 11 to completely characterize
TOC  and SDS DBF removal  through the  entire duration of the GAC  contactor  run.
Rather,  the objectives of this study focused  on TOC and SDS DBF removal  in the early
breakthrough stage.  Thus, as seen in Figures 6-90 and 6-91, sampling did not extend
into the steady state  region of the TOC breakthrough curve.

GAC  treatment was  also effective in removing UV-254, as is apparent in examining the
data presented in Table 6-18.  The initial UV-254 data appear anomalous, in that GAC
effluents had 0.03 cnr1  UV-254 values during the first two samplings and  0.003  cm'
for the next two samples, even though TOC in the GAC effluent remained  low  (0.1 to
0.2 mg/L)  during this period.   However, as  TOC concentrations in the  GAC  effluent
increased on  Run Days 54, 82 and 95  (GAC  effluent TOC of 0.20, 0.60 and  0.85
mg/L, respectively),  the  UV-254 steadily increased from 0.003 to 0.009 and 0.015  cm
•', respectively.

The impact of GAC treatment  on levels of  SDS  DBFs is illustrated in Figures  6-93
through 6-102.  The GAC contactor influent and effluent SDS values plotted  in these
figures were measured  in the  samples dosed  at 4.5 mg/L of chlorine, buffered to pH
8.2. and incubated for  3 days at 25°C.  As shown in Figure  6-93, SDS  TTHMs  were
extremely low (approximately  6 /ug/L or less) in the contactor effluent for  the first  54
run days,  indicating very effective removal of  THM precursors.  After 82 run days.
SDS  TTHMs  had increased to 34 /ug/L and after 95 run days, SDS TTHMs increased
to 47 ug/L.  SDS TTHMs in the contactor influent were 104. 80,  74,  136,  116 and
134 in  the  0.2. 13. 25. 54. 82 and 95 run day  samples,  respectively.  Thus, GAC
treatment led to reductions in SDS TTHM levels of >97,  >92, >97 and >96 percent
on the first four sampling days, respectively; and  71 and 65  percent on  the 82nd and
95th run days, respectively.

DeMarco, et al. (1981)  reported similar simulated  distribution system THM results for
a previous  GAC contactor run at  Utility 11  (16-minute EBCT).  The SDS procedure
used by these  researchers included adding sufficient chlorine to produce  an initial free
chlorine residual of 2.5 mg/L, buffering to  pH 8.2 and incubating for three days at the
temperature of the full-scale plant's finished water.  After 95 days of column operation,
column  effluent SDS  TTHMs  were  less  than  30 //g/L,  while column  influent  SDS
TTHMs ranged from approximately 60 to 160 //g/L during the first 95 days of column
operation.

As was the  case for TOC removal, the effectiveness of THM precursor removal by GAC
is also site specific.  For example,  SDS THM data  published by McGuire, et al. (1989)
from  GAC  pilot studies conducted by Metropolitan on Colorado  River water  indicate
much  less effective THM control than was  achieved at Utility 11.  Metropolitan's  SDS

                                    6-23

-------
                Treatment Modification Results and Discussion


conditions for the latter testing included adding 1 mg/L of chlorine, buffering to pH 8.2
and incubating  for five days at  25°C.  For a I5-minute  EBCT (12x40 mesh  carbon),
SDS TTHMs in the column effluent were approximately equal to the SDS TTHM levels
in the column influent (40 to 52  jjg/L) after only 30 days of column operation.

Figures 6-94 and 6-95 illustrate the levels of chloroform and bromoform at Utility  11 in
the 3-day SDS. 5 (4.5 mg/L of chlorine, buffered  to pH 8.2, 25°C) GAC contactor
influent and effluent  samples.   SDS chloroform concentrations in the GAC influent
occurred  within the range  52 to 100 //g/L.  AH of the SDS chloroform  levels in the
GAC  effluent  samples were less  than 2 */g/L through  the first 54 run days,  then
increased to 8.9 and  14 */g/L on the 82nd and 95th run days, respectively.  Thus,  GAC
treatment was very  effective for lowering the levels of chloroform in the SDS samples.

Figure 6-95 illustrates SDS4 5  bromoform  levels in  the GAC  influent  and  effluent
samples.   Bromoform was  not detected in any of the contactor influent SDS  samples.
and was  not detected in  the contactor  effluent samples for the  first 54 run  days.
However,  after 82  run days, bromoform was detected in the contactor effluent  SDS
samples,  at 2.0 and 2.8 //g/L on the 82nd and 95th  run days, respectively. These data
indicate that GAC caused a shift from the chlorinated species to the brominated species
in the column  effluent samples.  This is reported by  Utility  II to be  a typical, though
unexplained, phenomenon at their GAC facilities. To further illustrate this occurrence.
on the 82nd and 95th run days,  chloroform comprised 70 and 72 percent, respectively.
of the SDS TTHMs in the  contactor influent; while  chloroform only comprised 30 and
26 percent of the SDS TTHMs  in  the contactor effluent  on  those sampling days. On
those  same  sampling days, bromoform  comprised less  than 1 percent  of  the  SDS
TTHMs in the contactor influent,  but 6 percent of the contactor effluent SDS TTHMs
were contributed by bromoform.

Similar trends  to those observed for SDS THMs were found for SDS HAAs.  Figure
6-96 shows the SDS4, results for HAAs in the GAC column  influent and effluent. For
the first 25 run days, no HAAs were detected in the contactor effluent SDS  samples.
although  the contactor  influent SDS samples  contained HAA levels ranging  from
approximately  25 to 40 yug/L during that time.  SDS HAA levels gradually increased in
the contactor effluent samples over the subsequent sampling dates, although  the  GAC
remained  very  effective at lowering SDS HAA concentrations  over the  95  days of
sampling.  The shift from chlorinated to brominated species in the GAC  contactor
effluent SDS samples is seen in Figures 6-97 and 6-98. DCAA in the  contactor effluent
SDS samples remains significantly lower  than in the  contactor influent SDS  samples.
However.  DBAA concentrations in the contactor influent SDS samples remained low
throughout the sampling period (less than or equal  to  1.8  /ug/L), while SDS DBAA
levels in  the effluent samples (3.0 and 4.6  /ug/L  on the 82nd  and 95th run  days.
respectively) exceeded those in the influent SDS samples after 82 run days.  In  the SDS
samples from  the  82nd and 95th run days, DBAA accounted for  approximately  3
percent of the  total measured HAAs in the  GAC influent, but accounted  for 26  to 29
percent of the total measured HAAs in the contactor effluent.

Raw water bromide levels  were  0.03, 0.05, 
-------
Ol
rn
U

U
         SDS CHCI3  vs  Run  Time  For  GAG
            Column  Influent and Effluent, Utility  1 1
        +—+GAC mf             A	AGAC  Eff
             3-day SDS              3-day SDS
     15O
     100
      so -
            CI2 = 4.5 mo/L
            pH = 8.2
            Temp. = 25 C
                                         * Below MRL
        0
4O
6O
80
1OO
                           Run Days
                           FIGURE 6-94


         SDS  CHBr3 vs  Run  Time  For  GAG
           Column Influent and  Effluent.  Utility  1 1
        +—+GAC  Inf            A	AGAC Eff
            3-day SDS              3-day SDS
^
3
--K
.3
n 2
U
1
0,
C
CI2 - 4.5 man. » influent below K/WL
f*"1 = 8-2 »* influont »na «ffkj«ot t»iow M=«.
Temp. = 25 C
.A
,-'
x
,-'
-
•** -X--X- -*--* -X-J(-/' -)<- -K-






) 20 4O 6O 80 10O
Run Days
FIGURE 6-95

-------
          SDS  HAAs  vs  Run  Time  for  GAG
            Column  Influent  and  Effluent.  Utility  1 1
        +	+GAC  Inf             A	A GAG  Eff
             3-day SDS               3-day SDS
1
?
o
rti
c
o
0
O
<
I




«u
70
60
50

40
30
20

10
0
t
c


CI2 = 4.5 mg/L * AM ^j^ IMOW ^^g
pH = 8.2 *•* MCAA, TCAA. and I*BAA below MRLs
Tenx>. = 25 C **•» KfiAA balow k/«_ +
+/^^
S~~~~~
,/
* X^
- +-^-X
-
•jr^r-*r _ , - £i

) 2O 40 60 8O 1C
Run Days
FIGURE 6-96









DO


5
ra
o
o
O
<
<
O
Q
          SDS  DCAA  vs  Run  Time  for  GAG
            Column Influent  and Effluent,  Utility  1 1
        +—+GAC Inf              A	A GAG Eff
             3-day SDS               3-day SDS
      30
      20
      10
       0
             CI2 = 4.5 mg/L
             pH = 8.2
             Temp. = 25 C
                                       * Below MR.
          *    *     *	A	
          j . j. j . I -L .ijd^. .i. J. J. i. i^y—*.-y l'1' i ' i  It- ' • • ' 1 ' ' ' '-*—*--

                  2O       40       60


                              Run Days

                             FIGURE 6-97
                                              SO
                                                    ...A
100

-------
 o>
 0)
0
           SDS  DBAA  vs  Run   Time  for  GAG
             Column Influent and Effluent,  Utility   1 1
         +—+GAC  inf
              3-day SDS
        6 r
        0
             CI2 = 45 mg/L
             pH = 8.2
             Temp. = 25 C
.^-..J...! .1 .I£L ... J.J.,.1^1



0        2O
                         A	AGAC  Eff
                              3-day SDS
                                      * Below
                           4O
                             6O
                             Run Days

                             FIGURE 6-98
          8O
                                              1OO
o»

-------
                Treatment Modification Results and Discussion


influent bromide.  This low level of bromide did not lead to bromoform production and
caused only low levels of DBAA formation in the column  influent SDS samples.   The
shift in speciation  from chlorinated to brominated  compounds in the GAC contactor
effluent may be due to the  following:

    o   Because bromide  is not  adsorbed by GAC, it is possible  that the increased
         percentage of brominated THMs and HAAs in the column effluent is due to
         the increased ratio of bromide to precursor material after significant levels of
         TOC have been  removed in  the GAC contactor.  Bench-scale TOC dilution
         experiments conducted at Metropolitan (McGuire et al., 1989) indicated that
         as the TOC of a  water was diluted while the bromide concentration was held
         constant,  the speciation of THMs shifted toward the brominated compounds.

    o   Some researchers  have  found that  the selectivity  of precursors for either
         bromination  or  chlorination  reactions may  be  a  function  of  precursor
         molecular weight  (Schnoor et al..  1979),  and the average of which  is most
         likely altered significantly by adsorption within the column.  However,  other
         researchers have  found no relation between THM speciation and  molecular
         weight (Glaze et al.. 1980).

Removal of SDS HANs by GAC treatment is  plotted  in Figure 6-99.  Very low levels
of HANs were observed  in the  GAC contactor effluent SDS samples, with  HAN
concentrations  showing evidence of slightly increasing toward the end of the sampling
period.    SDS HK  removal  is  illustrated  in  Figure 6-100.   Overall,  very  low
concentrations  of HKs were found in the GAC contactor effluent SDS samples.

SDS chloral hydrate levels  are plotted in Figure 6-101 for the GAC column influent and
effluent.  Contactor influent levels are relatively high, in the range of 9.5  to 22 //g/L.
The GAC was very effective in removing SDS chloral hydrate for the first 54  days of
column operation,  with concentrations at  or near  detection limits for the first  four
samples.   However, concentrations  gradually  increased  in the next two  contactor
effluent SDS samples, with a level of 4.0 //g/L measured on the 95th run day.  As will
be discussed below, chloral hydrate levels in Utility  I I's distribution system were also
relatively high.

SDS chloropicrin concentrations in the contactor influent  and effluent are  plotted in
Figure 6-102.  Levels of this compound in the GAC column effluent  SDS samples were
at  or  near  detection  limits throughout  the sampling  period.  Cyanogen  chloride and
2.4.6-trichlorophenol were not detected in any GAC column influent or effluent SDS4  ?
samples.

Table  6-19 summarizes  TOX  results  for  the six  sampling  dates  at all  sampling
locations.  Because the GAC facilities received unchlorinated water, data from the GAC
column influent (B2) indicate that some TOX was present in the raw water (12 to 26
//g/L). which would be expected due to the industrial nature of the raw water source.
On Run Days 0.2 through 82, the 12 to 16  /ug/L of TOX  in the GAC column  influent
was  removed  (or lowered  in concentration) by  GAC treatment, since  no TOX was
detected  (detection  limit equals 10 //g/L) in the GAC effluent on those days.  However,
on the 95th run day, 26 //g/L of TOX  was measured in the unchlorinated GAC column
influent and 21  //g/L of TOX was  measured in the unchlorinated GAC column effluent,

                                     6-25

-------
 u
 0)
 ro
ro
p
£1
u
               SDS  HKs   vs  Run  Time  for   GAG
                Column  Influent  and Effluent.  Utility  1 1
             +—+GAC  inf
                  3-day SDS
                              A	AGAC Eft
                                   3-day  SDS
          1.0
          0.8
          0.6
          0.4
          0.2
          O.O
                 CI2 = 4.5 mo/L
                 pH - 6.2
                 Tennp. - 25 C
                                      *  1.1.1-TCP below HFL.
                                     *•»  Botn i-K» batow ivH_s
                                    «»»  1.1 -OCP Mow I*RL
                  .-A.
                        •A*..
                                                          ...-A
                                                     •A"
             )    10   2O   3O   4O   50   6O   7O   SO   9O   10O

                                  Run Days
                                 FIGURE 5-100


             SDS  CH  vs  Run  Time  for   GAG
              Column  Influent and  Effluent. Utility  1  1
             - GAG Inf               ---A--- GAC Eff
               3-day SDS                3-day  SDS
       30
               CI2 = 4.5 mo/l_
               PH = 8.2
               Tenrp. = 25 C
                                        * Below N/R_
       20
10
                                                 	A'f"
         U
                              4O
                                        6O
                                           ao
                                                             10O
                                Run Days

                              FIGURE 6-101

-------
y
a
o
o
6
            SDS  CHP  vs  Run   Time  for  GAG
             Column  Influent and Effluent, Utility   1 1
          +	+GAC Inf              A	AGAC  Eff
              3-day SDS                3-day SDS
             CI2 = 45 mg/L
             pH = 8.2
             Tetrp. = 25 C
                                              » Beio*
                            . I . I .1. .'. -t. J- 1 . I ^ ' J_J.J.*.t.L.«-
                             4 O        6O

                               Run Days
                              FIGURE 6-102
       -'-'•tfr
        SO
                                                         A
1OO
Ol
      20O
                      Formation  of  TTHMs
                  In  Distribution  System  (Utility  1 1)
              Clearwell
              Effluent
Dist System
T = 3 days
           4/10/89  4/24/89  5/8/89   6/12/89 7/17/89  7/31/89
                          Sampling Date (Week of)
                             FIGURE 6- 103

-------
                               TABLE 6-19

                     UTILITY 11  TREATMENT STUDY
                          TOX RESULTS (in //g/L)
Run Day          0.2        13         25        54        82        95
Sampling Date   4/10/89    4/24/89    5/08/89    6/12/89    7/17/89    7/31/89


Location/
 Condition

A2/lmmed.1          140       110       150       250       260       290
A3/Immed.2          180       150       170       240       270       341
B2/Immed.            16        12        13        16         14        26
B2/SDS4.53         220       170       180       270       270       300
B3/Immed.          < 10      < 10      
-------
                Treatment Modification Results and Discussion


indicating that the halogenated organic compounds present in the raw water were not
removed by GAC treatment.

As indicated in Table 6-19, the SDS45 TOX concentrations in the GAC column influent
samples  ranged  from  170 to  300 jt/g/L,  while  concentrations  in  the  GAC effluent
samples  ranged from  only 10  to  86  //g/L.  SDS^5 TOX levels in the GAC effluent
remained below  20 x/g/L in  samples  collected during the first 54 run days, but more
than doubled on each successive sampling date after 54 run days.

A comparison of SDS data with actual in-plant and distribution system data is made in
Table 6-20 for three of the  six sampling days for the levels of chloroform, TTHMs,
DCAA.  HAASU1I1 and TOX. The demonstration plant filter effluent (B2) SDS, 5 was run
to provide a controlled basis of comparison with the SDS45 GAC column effluent (B3)
samples  as discussed  in the previous paragraphs.   The distribution system sampling
point (A3)  had an approximate residence time of 3 days, but actual in-plant chlorine
doses varied between 2.7 and 4.8  mg/L, temperatures in the distribution system varied
between  11 and  27°C. and  distribution system  pH ranged  from 8.4 to 8.8 over the
95-day sampling  period.

In comparing the DBFs measured in  the B2 SDS and A3 samples,  varying results are
seen. For the samples collected the week  of 4/10/89, there are  substantial differences
in the levels of chloroform and TTHMs between the B2 SDS4 5 and A3 samples.  These
differences were  probably  due, in  part, to the difference in temperatures (A3 was I I°C
and  B2  was  incubated at 25°C).   However, for the samples collected the week of
7/31/89, the SDS and actual conditions were very similar, and the DBF data shown in
Table 6-20 for that week  agree to 22 percent or better.   The data  presented in Table
6-20  illustrate  the importance of  utilizing a  standardized  SDS  protocol for the
evaluation of DBF precursor removal  by GAC at  Utility 11.  By  conducting SDS tests.
variables such as pH. temperature, chlorine dose and holding time  were held constant
over the  sampling period so that the only independent variable was TOC.

DBF Levels Produced by Conventional Treatment: Results and Discussion

Although the primary  objective of the treatment  study at Utility  11 was to evaluate
removal  of DBF precursors  by GAC adsorption, the sampling program at this utility
offered an opportunity to investigate levels  of DBFs produced in conventional treatment
and in a chlorinated distribution system, and thus expand upon the DBF data collected
under the baseline sampling program.

Figure 6-103 illustrates TTHM levels measured  in Utility I I's  clearwell effluent and
distribution system.  As noted previously,  the clearwell effluent  (A2) was sampled on
Monday  of each sampling week, and the distribution system  sampling  point (A3),
having an approximate residence  time of  3  days  in the system, was sampled  3 days
later. This figure illustrates  two important points.  First, the seasonal change in THM
production from  spring to summer conditions is  apparent as the water temperature at
A2 increased from  9 to 27°C. Additionally, the chlorine demand of the water increased
over the  sampling period,  since applied chlorine doses increased  from 2.7 to 4.8 mg/L
in the plant,  while free chlorine  residuals at the  distribution system sampling point
decreased from  0.9 to 0.3 mg/L  from the 4/13/89 to the 8/3/89 samples (see Table
6-18).  TTHMs measured  at A2 increased  from 32 to 84 //g/L from the early April to

                                    6-26

-------
                                                  TABLE 6-20

                                          UTILITY 11 TREATMENT STUDY
                               COMPARISON OF SDS AND DISTRIBUTION SYSTEM RESULTS
Location
4/10/89
A3 (4/13/89)
B2/SDS 4.5
5/8/89
A3 (5/11/89)
B2/SDS 4.5
7/31/89
A3 (8/3/89)
B2/SDS 4.5
Free C12
Temp. pH C12 Dose Residual TOC CHC13 TTHMs DCAA HAAsu. TOX
°C (mg/L) 
-------
                 Treatment Modification Results and Discussion
the late July samples.  Increased  formation of THMs in the warm weather months was
observed in the baseline data collected for this study and DBF production was found to
be  strongly influenced by water temperature (see Section 5).  It is apparent  that the
increase  at  Utility 11 is  caused by factors such as the  higher water temperatures and
higher TOC levels.  TOC levels at A2 were 1.77, 1.51,  1.56. 2.15, 2.08 and 2.13
mg/L on the 6  sampling dates (in chronological order), indicating an increase from
April to  July.  It  is interesting to note, however, that UV-254 values at A2 exhibit a
different  trend than  TOC, with Uy-254 equal to 0.063, 0.048, 0.024,  0.034,  0.032
and 0.038 cm ' on successive sampling days from  early April to late July.

The second important issue illustrated in Figure 6-103 is the production of DBFs in the
distribution  system of  Utility II.   TTHMs  measured at A3 were  substantially  higher
than levels measured at A2 on all six sampling dates.  For the first two sampling dates,
TTHM  concentrations  were  approximately  double after 3  days of retention  in the
distribution  system compared to  the clearwell effluent TTHMs.  For the six  samples
collected for  this  study,  increases of 29, 36, 33, 57,  42 and 60 /ug/L  of  TTHMs
occurred in  the distribution system.

Figure 6-104  shows  the fate  of DCAN in  Utility  I I's distribution system.  In contrast
to  the increased  levels  of  TTHMs observed  in the  previous  figure, DCAN  levels
decreased with residence  time in the distribution system, with the decrease increasing in
magnitude  over  the  six sampling dates.   This figure highlights an  issue  discussed in
Section 5.  where  results of this study confirmed the  results of other researchers who
found that  DCAN was a  reactive  intermediate rather than a stable  endproduct  of the
chlorinalion of natural  organic matter.  DCAN is known to undergo a base-catalyzed
decomposition (Trehy and Bieber, 1981), and the hydrolysis rate is significantly  higher
at pH 8.5 than at  neutral pH (Croue and  Reckhow, 1988) (the pH  at A3  ranged from
8.4 to 8.8.).

Chloral hydrate  levels at Utility 11 measured during the baseline data collection were
relatively high compared to other utilities, exceeding the 75th percentile values  in the
summer and fall samples. The chloral hydrate concentrations measured in  Utility i  I's
clearwell  effluent  samples  in  April  through July, 1989 are consistent  with  those
recorded  in the summer quarter baseline sample in 1988.  Figure 6-105 illustrates the
concentrations of chloral  hydrate at A2 and A3 for the six sampling dates.  In the first
four samples, chloral hydrate levels approximately doubled from A2 to A3, however the
increase  measured at A3 was  significantly  lower  in the  last two  samples.   In the
6/12/89 sample,  the  highest level of chloral  hydrate was measured (20 /ug/L),  and this
compound was the fourth most prevalent of the measured DBF compounds on a weight
basis, exceeded only by chloroform (94 /ug/L), bromodichloromethane (26 /ug/L), and
DCAA (26 //g/L).
                                     6-27

-------
en
O
Q
                       Formation  of   DCAN

                  In  Distribution System  (Utility   11)

              Clearwell                HH Dist System
              Effluent                     T  =  3 days
            4/10/89 4/24/89  5/8/89  6/12/89  7/17/89 7/31/89
                          Sampling Date (Week  of)

                             FIGURE 6-104
 0)
 4—
 ro
 ti
 ro
 b
 n
 U
                         Formation  of  CH
                  In Distribution  System (Utility  11)
               Clearwell
               Effluent
Dist System
T = 3 days
            4/10/89 4/24/89  5/8/89  6/12/89  7/17/89  7/31/89


                          Sampling Date (Week  of)

                             FIGURE 6-105

-------
Section 7
Summary and Conclusions

-------
                                 SECTION 7

                      SUMMARY AND CONCLUSIONS
In this  section,  major study results  are  summarized and conclusions based on  study
results are drawn.  Additionally, research  needs arising from study results are discussed.

BASELINE RESULTS

The  median  TTHM  value  for (he  four quarters of  baseline sampling  at clearwell
effluents was 39 ^g/L (computed on  a running annual average  basis for each individual
utility).  The THMs represented the largest class of DBFs measured in this study (on a
weight basis), comprising approximately 54 percent of  the measured compounds.  The
second  largest  DBF  fraction measured  in  this  study  was  HAAs,  with a  median
concentration of 17 /vg/L. which represents approximately 25  percent  of the measured
compounds.  Thus, the median level  of TTHMs was approximately  twice that of HAAs.
The  third  largest  fraction  detected  was the aldehydes, comprising approximately 9
percent  of the measured compounds,  with a median concentration of 5.7 //g/L.

Little difference was  observed in the overall median concentrations  of influent  water
quality  parameters and  concentrations of DBF  compounds in  clearwell  effluents  from
quarter  to quarter  in  the baseline samples.  Because of the nationwide distribution of
participating  utilities and the wide variations in seasonal weather conditions and water
temperatures, "season" or  "sampling quarter" proved  to  be a  somewhat  arbitrary
method  of categorizing the data.  Water temperature, however, was  found to have  a
significant  impact  on  concentrations of DBF  compounds.    Levels  of XDBPsnni  and
THMs were  found to be higher in the highest temperature  range  when the data were
sorted into four equal temperature ranges.  The differences  in DBF levels between the
highest  temperature  range  and  any  of  the  three  lower ranges  were  statistically
significant  at a 95 percent  confidence level.  In addition, plant effluent pH was  found
to strongly impact  a number of DBFs which are unstable and hydrolyze at high  pH.

Raw  water quality  characteristics  varied  considerably  among the  35  participating
utilities.  Relative  concentrations of  TOC, UV-254 absorbance. chloride and  bromide
indicated that classification of the utilities and their DBF levels by source water type.
treatment  type and disinfection scheme was too simplistic to account  for other factors
having considerable impact  on DBF production.

Classification of the  utilities  by disinfection scheme  indicated  that influent parameters
varied  considerably   among  the utilities  within each category.   Prechlorinating/
poslammoniating  utilities  were  found  to have  a higher  median  influent   bromide
concentration  than  chlorine-only  utilities,  with the  difference  being  statistically
significant at a 95  percent confidence level.  Significant differences were also observed
between median concentrations of some DBFs as a function  of disinfection  scheme.
Median  levels of XDBPMm, and  TTHMs for  the prechlorinating/postammoniating
utilities  (approximately  94  and 57 //g/L. respectively)  were significantly higher than
those of either chlorinating (62 and  34 j/g/L. respectively)  or chloraminating utilities


                                      7-1

-------
                          Summary and Conclusions


(2.1 and 12 //g/L. respectively).  The same trend was observed in median concentrations
of HANs. aldehydes and cyanogen chloride, and the differences were also statistically
significant  at  a  95  percent  confidence   level.     However,  the  prechlorinating/
postammoniating utilities treated water with the highest median level of UV-254 (0.15
cm1 as compared to 0.10 cm1 for chlorine-only and chloraminating utilities).  Since
UV-254 may be an indicator of THM formation  potential,  the former  utilities may
require postammoniation to minimize further THM formation in their systems.

A  strong correlation was found  between TTHMs and the sum of measured halogenated
DBFs (r=0.96).  However, since TTHMs represented over 50 percent of XDBPsum. the
correlation coefficient for TTHMs and the sum of non-THM  DBFs was determined and
found to be 0.76.   Some of the other DBF classes correlated well with TTHMs (e.g..
HANs). while others correlated poorly with TTHMs (e.g., HKs).

HAAs  correlated strongly  (r = 0.98) with the  sum of non-THM XDBPs.   When both
HA As  and TTHMs were subtracted  from  XDBP    and were  then correlated with
HAAs. r equaled 0.77.  Fn addition,  a correlation of 0.74 resulted between HAAs and
XDBPslim minus HAAs.  The  latter two correlations may  be useful in helping to predict
the sums of non-THM.  non-HAA XDBPs.

The correlation between the UV-254 absorbance and TOC of plant influent samples was
strong  (r=0.85). although neither UV-254 nor TOC  correlated well with the TTHMs of
plant effluents.  Influent chloride correlated strongly with influent  bromide (r=0.97);
thus, chloride may be  used  as  a predictor for bromide.   Exclusion  relationships were
found  for  bromide with  chloropicrin,  1.1,1-TCP,  TCAA,  chloroform,  and other
chlorinated DBFs: that  is.  the presence of bromide appeared  to exclude the presence of
the particular DBF. and the inverse was also observed.

Bromide present in the source water  was  found to  shift the distribution  of THMs,
HANs  and HAAs to the more brominated species.  High bromide levels were  observed
not only in utilities susceptible to tidal influences and saltwater intrusion,  but  at inland
utilities as well.

Of the 35 utilities  included in this study, only three  employed ozone, yet  almost all 35
had detectable  levels of formaldehyde and acetaldehyde.  These aldehydes were found
in  some plant  influent samples, and  chlorination  alone  was found  to produce these
compounds.

Chloramines are recognized as an effective control strategy for THMs and other DBFs.
However,  for most waters studied  in this project, cyanogen chloride was found to be
preferentially produced in chloraminated water.  The distribution of cyanogen chloride
could  statistically  be  divided  into  two groups,  depending on  whether  the final
disinfectant was chlorine or chloramines.

The TOC removal within filtering plants included  in the baseline  sampling program
averaged approximately 24 percent.  It should be noted  that the treatment practices of
the utilities participating  in  the  baseline sampling  most likely focused on  turbidity
control and were not optimized with respect to TOC removal.


                                      7-2

-------
                           Summary and Conclusions
The TTHM  data  from this study of 35  utilities  was compared to that of the THM
survey conducted  by the  American Water Works  Association Research Foundation  in
 1987.  which  involved  727 utilities  around the  nation.   Based on  running annual
averages, frequency distribution  curves for the two  sets of data were similar.   The
hypothesis that the data from  the two studies were from the same distribution was not
rejected at a significance level  of 0.01, using a Kolmogorov-Smirnov test.

PROCESS MODIFICATION  RESULTS

Ozone in conjunction with chlorine or chloramines as final  disinfectants was generally
effective in lowering concentrations of classes  of  halogenated DBFs, except lor  HKs.
However, observed increases in HKswere equal  to or less than 1.0 //g/L.  The extent  to
which  DBF  levels were  decreased  or increased  after  implementation of ozonation
depended primarily on the final disinfectant which was employed.  Where comparable,
ozonation  followed by chloramination   was   more  effective  in  reducing  levels  of
halogenated DBF classes and chloral hydrate than ozone  followed by chlorination.  For
all ozone studies, ozone addition resulted in decreased TOX concentrations.

Although the  use  of ozone and  chlorine or chloramines appeared  to  be effective  in
minimizing the formation of compounds in the major halogenated DBF classes, shifts  to
greater  concentrations  of the  brominated  species were observed for THMs and HAAs
when free chlorine was used  as  a final  disinfectant.  Increases in chloropicrin  were
observed in some cases where ozone was  used first, regardless of the final disinfectant;
however,  these increases  were always less than 1.0 //g/L.   When chloramines  were
employed after ozonation. increases in cyanogen chloride concentrations were observed
at some utilities.
Aldehyde concentrations  increased substantially  whenever  ozonation  was  employed.
These  increases  ranged  from  67  to  459  percent,  depending  on  the  treatment
modification  implemented.  However, at one utility, the use of filtration to which a
disinfectant was not applied, indicated  that these aldehydes  could  be  removed.  It is
important to  note that free chlorine also produced aldehydes, although to a lesser extent
than ozone.

At  one utility  studied, chlorine dioxide  preoxidation with  free  chlorine for residual
disinfection was not found to lower concentrations of DBFs when compared to chlorine-
only oxidation/disinfection.  However, at another utility, chlorine dioxide preoxidation
was  found  to  be  an  effective   control method  for  DBFs  compared to  chlorine
preoxidation. even though free  chlorine was detected in the  chlorine dioxide generator
product.   Use  of  chlorine dioxide preoxidation  at this utility led to decreases of
approximately  50  percent in  levels of  XDBPsum,  TTHMS, HANs,  HKs and  HAAs
compared to free chlorine preoxidation.

At two utilities  studied, increasing  alum dose increased removal of DBF precursors and
resulted in lower concentrations of DBFs. However, at one  utility,  chlorine was added
before  the removal of TOC in the coagulation, flocculation, sedimentation and filtration
processes, resulting  in less effective DBF control.


                                      7-3

-------
                           Summary and Conclusions
 At  one utility. GAC was very effective in removing  TOC during the initial 54 clays of
 column operation.   After  the  54th run clay. TOC  in the column  effluent  increased
 steadily.   Levels of  DBFs in  column  influent and effluent  samples  which  were
 chlorinated and held under strictly controlled conditions indicated that  DBFs followed
 the  same trend as TOC:  that  is. very low  levels  for the first  54 days of column
 operation, followed  by a  steady  increase.   Concentrations of individual THMs and
 HAAs in the GAC  filter effluent indicated  that GAC treatment  caused  a  shift  from
 chlorinated to brominated species.

 Although this study  focused on  DBF concentrations  and control of DBFs  by  treatment
 modifications, it should be noted that  in full-scale applications,  DBF control strategies
 must be evaluated not only on their effectiveness for limiting concentrations  of DBFs,
 but on their economic and operational feasibility as well. Evaluation of these aspects of
 DBF control  at full scale were beyond the scope of this study.

 RESEARCH NEEDS

 Several  areas  warranting   further  research   are  apparent  from the  study results.
 Additional research on  the effectiveness of currently accepted DBF control  methods,
 such  as  chloramination  and  ozonation,   is  justified  due  to  the  detection  of
 chloramination by-products,  such as cyanogen  chloride, and ozonation  by-products,
 such as formaldehyde, in this and other studies.  Additionally, this study focused on the
 occurrence and control of DBFs, rather than the  economic and operational feasibility of
 the DBF control strategies investigated, subjects of critical importance for utilities  faced
 with  meeting  future DBF regulations.   Consequently, operation  and  maintenance
 considerations and associated costs are subjects in need of further investigation.

 Another area of needed effort is the development  of water quality data for utilities
 around  the nation for modeling purposes, especially  TOC and bromide  data.   In  order
 to develop a background of information on the  present and projected future  treatment
capabilities of the nation's utilities  with  respect to DBFs  and  DBF precursors, it  is
essential that  TOC  be more routinely measured.  The feasibility of monitoring  plant
performance  based  on  TOC, instead  of a parameter  such  as  turbidity, in order to
control  DBF  production also warrants investigation.

Treatment  modification  studies at  two utilities also demonstrated that incremental
increases  in  TOC removal could  be accomplished by  increasing alum doses.   More
research needs to be focused on enhancing TOC  removal by increased coagulant doses
and  on overall optimization of the  coagulation  process.  Moreover, data from such
research needs to be developed on  a nationwide basis.

This  study  highlighted  the  role  of  bromide  in the formation of  DBFs and the
importance of brominated analogs, such as brominated HAAs and possibly brominated
picrins.  Further investigation 01 brominated DBF compounds  is required for continued
progress in the characterization of compounds contributing to TOX.

A great deal  of effort in this project  was devoted to development of analytical methods
tor  the compounds of  interest, and effective sampling and  preservation techniques.

                                      7-4

-------
                           Summary and Conclusions


Further research  is needed  to expand the list  of analytes investigated in  future studies
and protect valuable data by improving sample preservation methods.

This study focused on  levels  of DBFs produced in full-scale  drinking water treatment
facilities.   In evaluating DBP production within plants and distribution systems, bench-
scale  (simulated  distribution  system) research was  also conducted  for  this study to
provide a controlled  environment from  which comparisons between various treatments
could be made.   In this report,  results of a great deal of laboratory research from other
studies  were compare  with the results of this  study.   However, further research  is
required to more fully  apply  the  results of strictly  controlled laboratory research to
"real world" applications in full-scale water treatment facilities.  Operation of full-scale
facilities is not easily compared to laboratory experiments, in that raw water quality and
plant  operation  may vary  from day to day  or  season to season, and  oxidants  and
disinfectants are  added  at  multiple points with  various contact  times  and degrees of
mixing, among  other  differences.   Further research  in  these  areas  will  increase
understanding of the presence and control of DBFs in  drinking water.
                                       7-5

-------
References

-------
                             REFERENCES
Aieta.  E.M.; Reagan, K.M.; Lang, J.S.; McReynplds, L.; Kang, J.W.; & Glaze,
W.H.   Advanced  Oxidation Processes for Treating  Groundwater Contaminated
With TCE and PCE:  Pilot-Scale Evaluations. Jour. AWWA, 5:64 (1988).

Aizawa.  T.;  Magara,   Y.;  &  Musashi,  M.   Effect  of Bromide  Ions   on
Trihalomethane (THM) Formation in Water.  Aqua. 38:165 (1989).

Amy. G.L.; Chadik,  P.A.; & Chowdhury, Z.K. Developing  Models for Predicting
Trihalomethane Formation Potential and Kinetics.   Jour. AWWA. 79:7:89 (1987).

Babcock. D.B.:  & Singer. P.C.   Chlorination and  Coagulation  of  Humic and
Fulvic  Acids. Jour. AWWA, 71:3 (1979).

Becke. C.:  Maier. D.;  & Sontheimer,  H. Herkunft  von Trichlornitromethan  in
Trinkwasser.  Vom Wasser. 62:125 (1984).

Bruchet,  A.; Tsutsumi,  Y.;  Duguet, J.P.;  & Mallevialle,  J. Characterization  of
Total  Halogenated Compounds During  Water Treatment Processes.  In:  Water
Chlorination: Chemistry. Environmental  Impact and  Health Effects,  Vol.5. R.L.
Jolley.  et al.. eds. Lewis Publ.. Inc.. Chelsea, MI  (1985).

Chadik. P.A.: & Amy, G.L.  Removing Trihalomethane Precursors from Various
Natural Waters by Metal Coagulants.  Jour. AWWA, 75:10 (1983).

Chow.  B.M.; & Roberts.  P.V.   Halogenated Byproduct Formation by CIO-, and
CI2.  Journal. Environ. Engrg. Div. - ASCE,  107:EE4  (1981).

Chowdhury. Z.K.; Cline, G.C.; Wiesner. M.R.;  & Hiltebrand, D.J.   Monitoring
of Flocculation  Using  UV-Absorbance.   Proceedings,  AWWA  Water Quality
Technology Conference, St. Louis. MO (Nov. 13-17, 1988).

Christman.  R.F.; Norwood,  D.L.; Milington,  D.S.;  Johnson,  J.D.;  & Stevens.
A. A.   Identity and Yields of Major  Halogenated  Products of Aquatic  Fulvic Acid
Chlorination. Environ. Sci. & Technol.,  17:10:625 (1983).

Clark.  R.M.   Evaluating the  Cost and  Performance of  Field-Scale  Granular
Activated Carbon Systems. Environ. Sci. Technol.. 21:6 (1987).

Coleman. W.E.:  Lingg.  R.D.: Melton R.G.:  & Kopfler, F.C.  The Occurrence of
Volatile  Organics  in   Five  Drinking  Water   Supplies   Using GC/MS.   In:
Identification and Analysis of Organic Pollutants  in Water.  L.H. Keith, ed.  Ann
Arbor Sci. Publ.  Inc.. Ann Arbor" Ml (1976).

Croue. J.P.: & Reckhow. D.A.  The Destruction  of Chlorination Byproducts with
Sulfite. Presented at  the  AWWA  Annual Conference, Orlando,  FL (June  19-23,
1988).

-------
 DeLeer, E.W.B.; Damste, J.S.S.; Erkelens, C.; & De Galan, L.   Identification of
 Intermediates Leading  to  Chloroform and  C-4 Diacids in the Chlorination of
 Humic Acid.  Environ. Sci. Technol.,  19:6:512(1985).

 DeMarco,  J.; Miller, R.; & Hartman, D.J.  Discovery and Elimination of Dioxins
 From a Carbon Reactivation Process. Jour. AWWA, 80:3 (1988).

 DeMarco.  J.: Stevens. A.A.; & Hartman. D.J.   Application of Organic Analysis
 for  Evaluation of Granular  Activated Carbon  Performance  in  Drinking  Water
 Treatment.  In:  Advances jn the Identification & Analysis of Organic Pollutants in
 Water.  Vol.  2.  L.  H. Keith, ed.   Ann  Arbor  Sci. Publ.  Inc.T Ann Arbor,  MI
 (T9TT).

 Dempsey.  B.A.;  Ganho. R.M.; & O'Melia, C.R.   The Coagulation of Humic
 Substances by Means of Aluminum Salts.  Jour. AWWA, 76:4 (1984).

 Dore.  M.;  Merlet. N.:  Legube.  B.; & Croue, J.P.  Interactions  Between Ozone,
 Halogens and Organic Compounds.  Ozone Sci. & Engrg., 10:53 (1988).

 Duguet,  J.P.; Tsutsmui, Y; Buchet, A.; Mallevialle, J.   Chloropicrin in Potable
 Water: Conditions of Formation and Production During Treatment Processes.  In
 Water Chlorination:  Environmental Impact and Health Effects Vol.5, R.L. Jolley,
 et al..  eds.  Ann Arbor Sci. Publ., Inc.. Ann  Arbor, MI (1985).

 Fair, P.S.  (Chemist. Technical Support Division, U.S. Environmental Protection
 Agency.  Cincinnati. OH) Personal communication (May, 1988a).

 Fair. P.S.  Comparison  of Two Haloacetic Acid  Methods in  Use at MWDSC.  In-
 house   memorandum.   Drinking  Water  Quality  Assessment   Branch,   U.S.
 Environmental Protection Agency. Cincinnati, OH (Nov. 28,  1988b).

 Federal Register. National  Primary Drinking Water Regulations; Substitution of
 Contaminants and Drinking Water Priority List of Additional  Substances Which
 May Require Regulation Under the Safe Drinking Water Act. 53 FR 1892 (January
 22/1988).

 Fleischacker.  S.J.; & Randtke,  S.J.  Formation  of Organic  Chlorine  in Public
 Water Supplies.  Jour. AWWA, 75:3 (1983).

 GJaze, W.H.. et al.   Ozone as a Disinfectant and Oxidant in Water Treatment.  In:
 Disinfection By-Products:   Current Perspectives, AWWA, Denver, CO (in press,
 I989a).

 Glaze.   W.H.:  Koga.  M.;  & Cancilla,  D.     Ozonation   Byproducts.    2.
 Improvement of an  Aqueous-Phase Derivatization  Method  for the Detection of
 Formaldehyde and  Other Carbonyl  Compounds  Formed  by the Ozonation of
 Drinking Water.  Environ. Sci. & Technol.. 23:7:838 (1989b).

Glaze.  W.H.:  Rang.  J.W.;  & Chapin, D.H.   The Chemistry of Water Treatment
Processes Involving Ozone.  Hydrogen Peroxide and Ultraviolet Radiation.  Ozone
Sci. &  Engrg.. 9:335 (1987).                                          	

-------
 Glaze.  W.H.; Peyton, G.R.;  Lin, S.; Huang. R.Y.; & Burleson, J.L.  Destruction
 of Pollutants in Water with Ozone in Combination with Ultraviolet Radiation.  2.
 Natural Trihalomethane Precursors.  Environ. Sci. & Technol.,  16:454  (1982).

 Glaze,  W.H.;  Saleh, F.Y.;  &  Kinstley,  W.   Characterization of  Nonvolatile
 Halogenated  Compounds  Formed  During  Water  Chlorination.    In:  Water
 Chlorination: Environmental Impact and Health Effects, Vol. 3. R.L. Jolley, et al.
 eds.  Ann Arbor Sci.  Publ., Inc..  Ann Arbor, MI (1980).

 Gordon.  G.;   Yoshino.  K.:  Themelis.  D.G.;  Wood. D.W.;  &  Pacey,  G.E.
 Utilization  of  Kinetic   Based  Flow  Injection   Analysis   Methods  for  the
 Determination of Chlorine  and Oxychlorine Species.  Anal. Chem. Acta (in press,
 1989).                                            	

 Gurol.  M.D.: Wowk.  A.: Myers.  S.: and Suffet. l.H.  Kinetics and Mechanism of
 Halofprm Formation:   Chloroform  Formation from  Trichloroacetone. In: Water
 Chlorination: Environmental Impact and Health Effects, Vol. 4, R.L.  Jolley, et al.,
 eds. Ann Arbor Sci. Publ.,  Inc.! Ann Arbor, MI (1983).

 Helsel.  D.R.:  & Cohn,  T.A.  Estimation  of Descriptive  Statistics  for  Multiple
 Censored Water Quality Data. Water Resources Research, 24:12:1997 (1988).

 Hirose.  Y.. et  al.  Formation of Cyanogen Chloride by the Reaction of Amino
 Acids   with  Hypochlorous  Acid in  the   Presence  of  the  Ammonium   Ion.
 Chemosphere. 14:11/12:1717(1988).
Hoaglin.  D.C.:  Mosteller.  F.;  &  Tukey.  J.W.   Understanding
Exploratory Data Analysis.  John Wiley and Sons,  Inc., New York. N
 Robust  and
Y (1983).
 Hoehn. R.C.,  et  al.   Household Odors Associated  with  the Use  of  Chlorine
 Dioxide  During Drinking Water Treatment.   Presented at  the AWWA Annual
 Conference  Sunday  Seminar "Identification and  Treatment of Taste and  Odor
 Compounds". Los  Angeles, CA  (June 18. 1989).

 Hoel.  P.G.   Introduction to  Mathematical Statistics.  John Wiley & Sons. Inc..
 New York. NY (\97\).

 Hoigne. J.:  & Bader. H.  The Formation of Trichloronitromethane (Chloropicrin)
 and  Chloroform in a Combined Ozonation/Chlorination Treatment  of  Drinking
 Water. Wat. Res.. 22:3:313 (1988).

Hubel.  R.E.:  &  Edzwald.  J.K.   Removing  Trihalomethane  Precursors  by
Coagulation. Jour.  AWWA. 79:7 (1987).

 Kavanaugh.   M.C.      Modified  Coagulation   for   Improved   Removal  of
Trihalomethane Precursors. Jour. AWWA, 70:11 (1978).

 Koch.  B.; & Krasner. S.W.  The Occurrence of  Disinfection  By-Products in a
Distribution  System.   Proceedings. AWWA Annual  Conference,  Los Angeles, CA
(June 18-22, 1989).

-------
Koch.  B.; Krasner, S.W.; & Sclimenti, M.J.   A Simulated Distribution System
Disinfection  By-product Method.  Proceedings. AWWA Water Quality Technology
Conference,  Philadelphia, PA (Nov.  12-16, T989).

Koch.  B.. et al.  Analysis of Halogenated Disinfection By-products by Capillary
Chromatography.   Proceedings,  AWWA  Water Quality  Technology  Conference,
St. Louis. MO (Nov. 13 17,  1988).

Kopfler.  F.C.. et al.  Human Exposure to Water Pollutants.  Presented at  169th
ACS Natl. Mtg.. Philadelphia, PA (April, 1975).

Krasner.  S.W.: Sclimenti. M.J.; & Hwang, CJ. Experiences with Implementing a
Laboratory Program to Sample and Analyze  for Disinfection By-products  in a
National  Study.    In:  Disinfection  By-products:  Current  Perspectives,  AWWA.
Denver. CO  (in press.
Lange. A.L.: &  Kawczynski.  E.  Controlling Organics: The Contra Costa County
Water District Experience. Journal. AWWA. 70:1 1  (1978).

Lawrence. J.  The Oxidation of Some Haloform Precursors with Ozone.  Presented
at  the Third International Symposium on Ozone Technology,  International Ozone
Institute. Paris (May,  1977).

Lykins.  B.W.. Jr.: Clark, R.M.; & Adams, J.Q.  Granular Activated Carbon for
Controlling THMs.  Jour. AWWA,  80:5 (1988).

Lykins.  B.W..   Jr.:  Koffskey.  W.;  &  Miller,  R.G.   Chemical  Products and
Toxicologic Effects of Disinfection.  Jour. AWWA. 78: 1 1 :66  (1986).

Lykins.  B.W.. Jr.: & Griese. M.H.  Using Chlorine Dioxide for Trihalomethane
Control. Jour. AWWA. 78:6 (1986).

McGill. R.:  Tukey.  J.W.:  and W.A.  Larsen.   Variations  of  Box Plots. The
American Statistician. 32:1:12 (1978).

Margerum. D.W.: & Gray. E.T.. Jr.  Chlorination and the Formation of N-Ch!oro
Compounds  in   Water Treatment.    Presented  at  the 175th  ACS NatT.  Mtg.,
Anaheim. CA (March. 1978).

McGuire.  M.J..  et  al.   Optimization and  Economic  Evaluation of  Granular
Activated  Carbon for Organic  Removal.  American Water  Works  Association
Research Foundation, Denver, CO (1989).

McGuire.  M.J.:  & Meadow. R.G.   AWWARF Trihalomethane Survey.   Jour.
AWWA. 80:1:61  (1988).

McGuire.  M.J.   Quality  Assurance  Project Plan:   Disinfection By-Products  in
Drinking Water.   In-house  report. Water  Quality Division,  Metropolitan Water
District of Southern California, Los Angeles, CA (June 15, 1988).

Metropolitan  Water  District of Southern California;  & Montgomery, James M.,
Consulting Engineers. Inc.  Disinfection By-Products in California Drinking Water,
Final Report to  the California Department of Health Services (July, 1989).

-------
 Miller,  J.W.; & Uden, P.C.  Characterization of Nonvolatile Aqueous Products of
 Humic Substances.  Environ. Sci.  & Techno!..  17:3:150(1983).

 Miller,  R.; &  Hartman,  D.J.   Feasibility Study  of  Granular  Activated Carbon
 Adsorption and On-Site Regeneration. EPA-600/S2-82-087 (1982).

 Montgomery.  James  M.  Consulting Engineers,  Inc.   Evaluation of Ozone and
 PEROXONE for Disinfection and Control of Disinfection By-Products  and Taste
 and Odor  Compounds, Phase IV -  Experiment  3, Project Status  Report to the
 Metropolitan Water District of Southern California (September. 1989).

 Norwood. D. L.. et al.  Using  Isotope Dilution  Mass Spectrometry to  Determine
 Aqueous Trichloroacetic Acid.  Jour.  AWWA. 78:4:175 (1986).

 Ohya. T.; & Kanno, S.    Formation of  Cyanide Ion or  Cyanogen Chloride
 Through the Cleavage of Aromatic Rings by Nitrous Acid or  Chloride.  VIII.  On
 the  Reaction  of  Humic Acid  with  Hypochlorous  Acid in  the  Presence  of
 Ammonium Ion.  Chemosphere. 14:11/12:1717(1985).

 Oliver.  B.C. Dihalocetonitriles in Drinking Water:   Algae  and  Fulvic  Acid  as
 Precursors.  Environ.  Sci.  & Techonol.. 17:2:80(1983).

 Organic  Contaminants  Committee,  AWWA Research  Division.    Non-Specific
 Organic  Analysis  for  Water   Treatment  Process  Control  and  Monitoring.
 [Proceedings.  AWWA Seminar  on  Non-Specific  Organic  Analysis  for Water
 Treatment Process Control and Monitoring (1985).

 Quimby. B.D.:  Delaney.  M.F.: Uden. P.C.; &  Barnes, R.M.  Determination  of
 the Aqueous Chlorination  Products of Humic Substances by Gas Chromatography
 with Microwave  Emission Detection.  Anal. Chem.. 52:259 (1980).

 Rav-Acha.  C.: Serri.  A.:  Choshen, E.;  & Limpni. B.  Disinfection of Drinking
 Water Rich in Bromide with Chlorine and Chlorine Dioxide, while  Minimizing the
 Formation  of Undesirable By-products.  Wat.  Sci. Tech.. Vol. 17,  pp. 611-621
 (1985).

 Reckhow.  D.A.; &  Singer.  P.C.   Mechanisms of Organic Halide  Formation
 During  Fulvic Acid Chlorination and  Implications  with Respect to Preozonation.
 In: Water Chlorination: Chemistry, Environmental Impact and  Health Effects,  Vol.
 5. R.L.  Jolley. et al..  eds. Lewis Publ.. Inc..  Chelsea, MI (1985).

 Reckhow. D.A.: &  Singer. P.C.  The Removal of Organic Halide Precursors by
 Pre-ozonation and Alum Coagulation.  Jour. AWWA,  76:4:151  (1984).

 Riley. T.L.;  Mancy.  K.H.; & Boettner,  E.A.  The  Effect of Pre-ozonation on
 Chloroform  Production  in  the  Chlorine  Disinfection  Process.    In:  Water
Chlorination;  Environmental Impact and Health Effects, Vol.  2. R.L. Jolley. et al..
eds. Ann Arbor Sci. Publ., Inc.. Ann  Arbor MI (1978).

-------
 Rook. J.  Haloforms in Drinking Water.  Jour. AWWA, 68:168 (1976).

 Sayato. Y.; Nakamora, K.; & Matsui, S. Mechanism of Formation of Chloroform
 and Chloropicrin  by Chlorination of Humic Acid.   Suishitsu Qdaku  Kenkyu 3,
 127-134 (1982).

 Schnoor,  J.L., et al.  Trihalomethane Yields as a Function of Precursor Molecular
 Weight.   Envir. Sci. & Technol.,  13:1134(1979).

 Scully. F.E.; &  Bempong,  M.A.   Organic N-Chloramines in Drinking Water:
 Chemistry and Toxicology. Environ. Health  Persp.. 46:111 (1982).

 Semmens. M.J.:  &  Field.  T.K.  Coagulation: Experiences in Organics Removal.
 Jour. AWWA.  72:8 (1980).

 Singer. P.C.   Alternative  Oxidant  and  Disinfectant Treatment Strategies  for
 Controlling Trihalomethane Formation.  EPA/600/2-88/044 (1988).

 Slocum, C. J.: Moore. L. A.; &  Stevens, A. A.  Analytical Criteria for Detection
 of  Disinfection Byproducts.    Proceedings.  AWWA Water  Quality  Technology
 Conference. Baltimore. MD (November  15-20,  1987).

 Sorrell. K.: & Brass. H.  (U.S. Environmental Protection Agency, Cincinnati. OH)
 Personal communication (1988).

 STSC. Inc. Statgraphics User's Guide. STSC. Inc.. Rockville. MD  (1987).

 Standard  Methods for the  Examination  of Water and Wastewater. 17th Edition.
 APHA. AWWA. WPCF.  Washington. D.C. (1989).

 Standard  Methods for the Examination  of Water And Wastewater, 16th Edition.
 APHA. AWWA. WPCF.  Washington, D.C. (1985).

 Stevens.  A.A.; Slocum.  C.J.;  Seeger. D.R.; & Robeck, G.G.   Chlorination of
 Organics in Drinking Water.  Jour. AWWA, 68:11 (1976).

 Sverdmp.  Johnson & Flemming.  The Oceans, Prentice-Hall, Inc.,  New York. NY
 (1942).

 Themelis.   D.G.:  Wood.  D.W.; & Gordon,  G.  Determination  of Sub-mg  L1
 Levels of  Chlorite Ion and  Chlorate Ion  by Using  a Flow  Injection  System.
 Anal.Chem. Acta. (in press. 1989).

Thibaud. T.; De Laat. J.; &  Dore.  M.  Effects of Bromide Concentration on the
 Production of Chloropicrin  during Chlorination of Surface Waters.  Formation of
 Brominated Trihalonitromethanes.  Wat.  Res.. 22:3:381 (1988).

Thibaud.  H.:  Merlet. N.; &  Dore. M.   Formation de  Chloropicrine  lors de la
Chloration de Quelques  Composes  Organiques   Nitres.    Incidence  d'une
Preozonation.  Envir. Technol. Lett. 7, 163-176 (1986).

-------
Trehey.  M.L.;   &  Bieber,   T.I.    Detection  and  Quantitative  Analysis  of
Dihaloacetonitriles  in   Chlorinated  Natural  Waters.    In:   Advances  in  the
Identification &  Analysis of Organic Pollutants in Water,  Vol. 2, L.H. Keith, ed.
Ann Arbor Sci. Publ. Inc.. Ann Arbor, MI (1981).

Tukey. J.W.  Exploratory Data Analysis.  Addison-Wesley, Reading, MA (1977).

Uden. P. C.; & Miller. J. W.  Chlorinated Acids and  Chloral in Drinking Water.
Jour. AWWA. 75:10:524 (1983).

Umphres.  M.D..  et  al.   The Effects  of  Preozonation  on  the  Formation of
Trihalomethanes. Ozonews. 6:3 Part 2 (1979).

United States Environmental  Protection  Agency.  Trihalomethanes in Drinking
Water (Sampling. Analysis,  Monitoring  and  Compliance).   EPA 570/9-83-002.
USEPA (1983).

United  States  Environmental  Protection  Agency.    National  Interim  Primary
Drinking Water  Regulations;  Control  of Trihalomethanes in  Drinking  Water.
Federal Register. 44:231:68624-68707 (Nov. 29,  1979).

Van der Kooij.  K.: Fisser, A.; & Hijnen W.A.M. Determining the Concentration
of Easily Assimilable  Organic Carbon in Drinking  Water.  Jour. AWWA, 74:10
(1982).

Wallace. J.L.; Vahadi. B.: Fernandes, J.B.; & Boyden. B.H.  The Combination of
Ozone/Hydrogen   Peroxide   and   Ozone/UV   Radiation   for  Reduction   of
Trihalomethane  Formation  Potential in Surface Water.  Ozone  Sci. & Engrg.,
10:103 (1988).

Werdehoff, K.S.; & Singer, P.C.  Effects of Chlorine  Dioxide on Trihalomethane
and Total Organic Halide Formation Potentials and on  the Formation of Inorganic
By-Products.  Proceedings. AWWA Annual Conference, Denver, CO (June 22-26.
1986).

Yamada,  H.; &  Somiya, I.    The  Determination of Carbonyl Compounds In
Ozonated Water by the PFBOA Method.  Ozone  Sci.&  Engrg. 11:2  (1989).

-------