WATER TREATMENT PLANT SIMULATION PROGRAM
VERSION 1.21
USER'S MANUAL
DRINKING WATER TECHNOLOGY BRANCH
DRINKING WATER STANDARDS DIVISION
OFFICE OF GROUND WATER AND DRINKING WATER
UNITED STATES ENVIRONMENTAL PROTECTION AGENCY
WASHINGTON, D.C.
JUNE 1992
Prepared by:
MALCOLM PIRNIE, INC.
One International Boulevard
Mahwah, New Jersey 07495-0018
2 Corporate Park Drive
P.O. Box 751
White Plains, New York 10602
-------
WATER TREATMENT PLANT SIMULATION PROGRAM
VERSION 121
USER'S MANUAL
WORK ASSIGNMENT NO. 1-3
AMENDMENT NO. 3
CONTRACT NO. 68-CO-0062
JUNE 1992
Prepared fon
USEPA - Office of Ground Water
and Drinking Water
Technology Transfer
Prepared by:
MALCOLM PIRNIE, INC.
One International Boulevard
Mahwah, New Jersey 07495-0018
2 Corporate Park Drive
P.O. Box 751
White Plains, New York 10602
-------
TABLE OF CONTENTS
Page
1.0 INTRODUCTION 1-1
1.1 Background 1-1
1.2 Purpose of This Manual 1-2
1.3 Manual Organization 1-3
1.4 PC System Requirements 1-3
2.0 DATA DEVELOPMENT AND INPUT 2-1
2.1 Main Menu Components 2-1
2.2 Simulating the Treatment Plant 2-4
2.3 Creating an Example Process Train 2-5
2.3.1 Create Process Train 2-6
2.3.2 Unit Process Parameters 2-6
2.3.3 Modify Process Train 2-7
2.3.4 Run Model 2-7
3.0 INTERPRETING MODEL OUTPUT 3-1
3.1 THM Prediction 3-1
3.2 HAA Prediction 3-2
3.3 Determination of Inactivation Ratio 3-3
3.4 Prediction of Other Parameters 3-6
REFERENCES
LIST OF TABLES
Table Following
No. Description Page
2-1 Input Data for Example 2 2-5
2-2 Output File for Example 2 2-5
RECYCLED PAPER
-------
TABLE OF CONTENTS (Continued)
LIST OF FIGURES
Figure Following
No. Description Page
2-1 Main Menu 2-1
2-2 Interaction for Various Process Units 2-4
2-3 Process Flow Schematic for WTP 2-4
2-4 Algorithm for WTP Simulation Model 2-4
2-5 Process Flow Schematic for WTP 2-6
2-6 Flow Chart for WTP Program 2-6
2-7 Process Train Menu 2-6
2-8 Process Train Menu 2-6
2-9 Raw Water Parameters 2-6
2-10 Process Train 2-7
2-11 Predicted Water Quality Profile 2-8
LIST OF APPENDICES
Appendix Description
B Results of Model Verification Efforts
ii
RECYCLED PAPER
-------
FORWARD
This User's Manual for Version 1.21 of the WTP Simulation Program has been prepared
to provide a basic understanding of 1) how to operate the program, and 2) the underlying
assumptions and equations that are used to calculate the removal of natural organic matter
(NOM) and the formation of disinfection by-products (DBPs). This manual represents the
first public release of the program.
The WTP Simulation Model was developed for the United States Environmental Protection
Agency (USEPA) by Malcolm Pirnie, Inc. in support of the Disinfectant/Disinfection By-
products (D/DBP) Rule. It was prepared with the understanding that the predictions may
reflect the central tendency for treatment. It is not to be construed that the results from the
model will necessarily be applicable to individual raw water quality and treatment effects at
unique municipalities and agencies. This model does not replace sound engineering
judgement using site-specific treatability data to evaluate the best manner in which to
address the potential requirements of the D/DBP Rule for an individual application.
It is understood that one limitation of the model is the extent of the data base available to
verify model predictions. In a desire to systematically improve the overall predictive
capability, the intent of this release is to solicit public comment on the usefulness and
relative accuracy of the predictions on a case-by-case basis. To this end, a questionnaire is
provided in Appendix C of this document that requests information on treatment process
characteristics and DBP formation from individual utilities.
The USEPA encourages constructive comments on methods to improve the model, results
from verification efforts, and utility data that can be used to enhance the overall
applicability of the model. Please forward comments to:
Barbara M. Wysock
Water Supply Technology Branch
Technical Support Division
Office of Ground Water and Drinking Water
United States Environmental Protection Agency
Cincinnati, Ohio 45268
The guidance provided herein may be of educational value to a wide variety of individuals in
the water treatment industry, but each individual must adapt the results to fit their own practice.
The USEPA and Malcolm Pimie, Inc. shall not be liable for any direct, indirect, consequential,
or incidental damages resulting from the use of the WTP model.
RECYCLED PAPER
-------
1.0 INTRODUCTION
1.1 BACKGROUND
The water treatment plant simulation computer program (WTP) was developed for
the U.S. Environmental Protection Agency (USEPA) by Malcolm Pirnie, Inc., in support of
the Disinfectant/Disinfection By-Products (D/DBP) Rule. The primary purpose of the
program is to simulate the removal of natural organics matter (NOM), the formation of
disinfection by-products (DBPs), and disinfection levels in water treatment plants and
distribution systems, based upon specified inputs including raw water quality, treatment
process characteristics and chemical dosages. At this time, only trihalomethane (THM) and
haloacetic acid (HAA) formation can be simulated. The model equations and format were
developed primarily with funding from USEPA, with some support from the Metropolitan
Water District of Southern California and the D/DBP Technical Advisory Workgroup
(TAW) for developing specific predictive equations.
WTP was developed to assist in determining the method by which treatment
processes can be implemented to achieve the required levels of disinfection while
maintaining compliance with potential requirements of the D/DBP Rule. The basic
modeling approach includes estimation of:
¦ NOM removal by individual unit processes;
¦ disinfectant decay based upon demands exerted by NOM and other sources;
and
¦ DBP formation based upon water quality throughout the treatment plant and
in the distribution system.
The predictive equations for water quality are described in Appendix A.
The model simulates DBP formation under given treatment conditions and permits
the user to evaluate the effects of changes in these conditions on the projected disinfectant
decay and DBP formation. By using the model under different treatment scenarios, the user
can gain an understanding of how the input variables affect disinfection and DBP formation.
The WTP model was used by the USEPA to assist in evaluating treatment strategies
for providing adequate disinfection, while controlling the formation of DBPs, using
simulated water qualities representative of surface waters throughout the country. A copy
1 - J
RECYCLED PAPER
-------
of a paper describing this effort, as a part of the Regulatory Impact Analysis (RIA) for the
D/DBP Rule, is provided at the end of this manual.
The model is not intended as a replacement for treatability testing to evaluate the
effectiveness of various processes on disinfectant decay and DBP formation in specific water
supplies, but does provide a useful tool for evaluating the potential effect of different unit
processes on the interrelationships between many of the new and forthcoming regulations.
Users of the program should be familiar with water treatment plant operation, as well as
procedures and methodologies used to disinfect water and control DBP formation. WTP,
like any computer program, can not replace sound engineering judgement where input and
output interpretation is required. Further, the technical adequacy of the output is primarily
a function of the extent and quality of plant-specific data input, and the extent to which an
individual application can be accurately simulated by predictive equations that are based
upon the central tendency for treatment.
12 PURPOSE OF THIS MANUAL
This manual is intended to guide the user in operating WTP and to assist in the
preparation of information necessary to execute the program. The manual provides a step-
by-step guide for operation, and describes how to utilize and interpret the program output.
The manual includes the following components:
¦ Instructions for using the computer program;
¦ A description of the equations used in the program (Appendix A);
¦ Results of model verification efforts (Appendix B);
¦ Water treatment plant data questionnaire (Appendix C);
¦ Copy of a technical paper entitled "Simulation of Compliance Choices for the
Disinfection By-Products Regulatory Impact Analysis"; and
¦ A 5.25-inch floppy disk of the executable version of WTP.
It must be stressed that the model is largely empirical in nature. It can not be used
as the sole tool for "full-scale" or "real-time" decisions for individual public water supplies.
l -2
RECYCLED PAPER
-------
13 MANUAL ORGANIZATION
This manual assumes that WTP users have a working knowledge of water treatment
plants. This basic understanding is necessary to provide meaningful input data to the
program and correctly interpret the output.
It is not necessary for the user to have any programming knowledge or extensive
computer experience. WTP operates through a user-friendly prompting program. It is
assumed, however, that the user is familiar with fundamental computer operating systems.
This manual will not address functions such as loading disks or connecting a printer.
Operating system information of this type is usually contained in the users manual for a
given computer system along with other fundamental computer operations.
In addition to this introductory chapter, this user's manual contains two other
chapters:
¦ Chapter 2 describes the information needed to run WTP and how the data
should be input. The components of the main menu are described and
explained, a diagram of a typical treatment plant is developed as an example,
data input options are outlined, and a general description of how to use the
program is provided.
¦ Chapter 3 provides guidance for interpretation of the output from the WTP
program.
Appendix A offers a description of the equations used in the program. Appendix B
provides a summary of model verification efforts, and Appendix C contains a questionnaire
for generating additional input data based upon actual water treatment plant operations.
1.4 PC SYSTEM REQUIREMENTS
The source code for this simulation program is written in the computer language C.
The executable version of this program can be run on an IBM PC or compatible hardware.
The computer must be equipped with:
¦ DOS 3.1 or later versions of DOS;
¦ 8086/286/386/486 processor (a math coprocessor is not required);
¦ A minimum of 640 kilobyte random access memory (RAM); and
i - 3
RECYCLED PAPER
-------
~ Graphics- capability (CGA, MCGA, EGA, EGA64, EGA-MONO, Here,
ATT400, VGA, PC3270, or IBM 8514).
l -4
RECYCLED PAPER
-------
2.0 DATA DEVELOPMENT AND INPUT
This chapter describes how to develop and enter data to WTP. It includes sections
explaining the features of the main menu, how to develop a simulated model of the specific
plant being analyzed, and how to enter the proper information to activate the program.
2.1 MAIN MENU COMPONENTS
The functions of the main menu are illustrated in Figure 2-1, and are described below.
To select a particular function, the user must move the arrow keys until the desired function
is selected. Alternately, pressing the capitalized letter shown for a function will select that
function. At the completion of each function, the user is returned to the main menu.
¦ Open process train: Retrieves a process train and unit process data previously
entered and saved to disk by the user.
Submenu 1 - File Name: The user is presented with a list of existing file names
with .WTP extensions and the user is prompted to enter the name of an
existing file (without the ".WTP" extension). If the file name entered by the
user exists, the unit process data are loaded into memory and the file name
becomes the new working file. The user can optionally return to the main
menu without entering a file name by pressing ENTER.
Submenu 2 - List Process Train: A listing of the process train is presented to
the user, and the user is given the option to list more detailed unit process
data. Following this listing the user is returned to the main menu.
¦ New process train: Allows the user to enter a new process train with new unit
process data, and replaces any previous train and data with the new train and
data. An example of creating a new process train is illustrated in Section 2.3.
Submenu 1 - File Name: Before any existing data are replaced in the memory,
the user is presented with a list of existing file names and prompted to enter
a new file name. At this point the user can optionally return to the main menu
and not replace data in the memory by pressing ENTER without entering a
new file name. After a new file name is entered and verified by the user,
existing data are replaced in the memory and the new file name becomes the
working file.
2 - 1
RECYCLED PAPER
-------
U.S. Environmental Protection Agency
Water Treatment Plant Model
Main Menu
IQpen
process
trainl
I Run nodel 1
Mew
process
trainl
1 Thn & disinfection I
ISaue
process
trainl
1 Delete data file I
IMod i f y
process
trainl
1 Quit 1
List
process
trainl
Highlight Desired Function and Press ENTER
Uersion 1.21, June 1992
-------
Submenu 2 - Process Train: Following the new file name menu, the user is
presented with a second menu to enter the new process train. Unit processes
can be added and deleted from the process train in this menu. When the user
is satisfied with the sequence of unit processes in the process train menu, the
user selects "Done - Process Train Complete."
Submenu 3 - Unit Process Data: After the process train is complete, the user
is presented with a series of menus to enter specific data for each unit process.
Each menu has default values for each unit process parameter which may need
to be changed to meet the needs of the user application.
The process train and unit process data are saved to a disk file with the name'
entered in Submenu 1 with a .WTP extension. The user is then returned to the
main menu.
¦ Save process train: Saves process train data and unit process data to a disk
file. This function can also be used to make a duplicate copy of the data.
Submenu - File Name: The user is presented with a list of existing data file
names and prompted to enter an existing file name or new file name. Entering
a new file name will change the working file name and save the data in the new
file, thus making a duplicate of the process train data. Pressing ENTER
without entering a file name will save data in the current working file. Existing
data files are over-written.
¦ Modify process train: Allows the user to view and optionally change the
process train and unit process data. An example of modifying a process train
is illustrated in Section 2.3.
Submenu 1 - Process Train: The first menu presents the process train and
prompts the user to enter a specific unit process. Entering a number which
corresponds to a unit process will bring up Submenu 2. Entering a negative
number in Submenu 1 will delete the unit process from the process train.
Entering a zero or pressing ENTER without entering a number will return to
the main menu. Unit processes can not be inserted into an existing process
train.
Submenu 2 - Unit Process Data: This menu displays the unit process data and
allows the user to change any parameter.
Submenu 3 - Keep Changes: At this point the user has the option to keep the
changes or restore the original data. Answering "Yes" saves the changes to the
current working disk file. Answering "No" restores the original data.
Note: If the user wishes to have an original file and modified file then "Save
process train" should be used before "Modify process train" to make a duplicate
copy of the working file.
2-2
RECYCLED PAPER
-------
¦ List process train: Lists the process train and unit process data currently in
memory.
The user is given the option to view the data on the CRT or down load the
data to a printer. An example input data screen is shown in Table 2-1 in
Section 2.2.
¦ Run model: Executes the water treatment plant model.
Submenu - Output Device: The user is prompted to select between viewing the
model output on the CRT, printing the output to the system printer, or saving
the output to a disk file. An example output screen or file is shown in Table
2-2 in Section 2.2
If the output is viewed on the CRT, a series of output screens is displayed.
After each screen, the user is prompted to press ENTER to continue or press
ESC to return to the main menu.
If the output is saved in a disk file the user is prompted to enter a file name.
The output file name will be given an .OUT extension.
¦ THM and Disinfection: Executes the water treatment plant model with a
summary screen illustrating selected parameters from the output file. This
menu does not have any submenu.
¦ Delete data file: Allows the user to remove data files created by program.
Submenu 1 - Data File Type: Two types of data files are created by WTP. The
process train/unit process data files all contain a .WTP extension while the
model output files contain an .OUT extension. The user is prompted to enter
which type of file is to be deleted.
Submenu 2 - File Name: A list of all existing files of the type selected in
Submenu 1 is displayed and the user is prompted to enter a file name to be
deleted.
¦ Quit: Exits the program and returns to DOS.
2-3
RECYCLED PAPER
-------
22 SIMULATING THE TREATMENT PLANT
Before WTP can be executed, a simulated version of the treatment plant process train
must be developed. Data specific to this plant must also be collected for input when
creating the simulated plant.
WTP is an interactive computer program that consists of a main program that acts
as a manager for the number of plant simulation subroutines created for the input, output,
and manipulation of data. A conceptual schematic of program inputs/outputs are shown
in Figure 2-2. The program algorithm (steps the program follows), for a typical simulated
process train illustrated in Figure 2-3 is shown on Figure 2-4.
The executable version of the computer program is interactive and menu-driven. The
main menu functions permit the user to direct the program to:
¦ Create an input file;
¦ Modify an input file;
¦ Save input/output files;
¦ Perform water treatment plant simulation runs; and
¦ Print input/output files.
An input file for the simulation program consists of the following:
¦ Source type (surface water or groundwater)
¦ Organic Raw Water Quality Parameters
TOC; and
UV-absorbance at 254 nm.
¦ Inorganic Raw Water Quality Parameters
Bromide concentration;
Alkalinity concentration;
Total and calcium hardness concentration; and
Ammonia nitrogen concentration.
2-4
RECYCLED PAPER
-------
FIGURE 2-2
INTERACTION FOR VARIOUS PROCESS UNITS
FOR WATER TREATMENT PLANT SIMULATION MODEL
( Raw Water
\. Quality )
( Process?
(^Characteristics/
WTP
MODEL
/THM/HAA FormatlonX
V Module J
( CT/lnactlvatlon \
I Module J
Other DBP
' Formation Modules
Treated
Water
Quality
-------
FIGURE 2-3
PROCESS FLOW SCHEMA TIC FOR WTP '
Alum
Chlorine
Flocculation &
Sedimentation
Caustic
nitration
Clear well
-------
FIGURE 2-4
INPUT FILE
¦ Water Quality
¦ Process Characteristics
ALGORITHM FOR WTP
SIMULA T/ON MODEL
¦ WQ at end
of each unit process
-THM/CI^NHpi
^Distribution system
-------
¦ Water Treatment Process Characteristics
Type of unit process;
Plant flow at average and peak hour conditions;
Baffling characteristics and detention times; and
Chemical dosages.
¦ Other raw water quality parameters
Giardia cyst concentration;
- pH;
Turbidity; and
Average and minimum temperature.
The output file from a water treatment plant simulation run contains information for
all the input parameters such as the raw water quality and the treatment plant process
characteristics. In addition, the output contains information for calculated concentrations
of the following parameters at the end of each of the unit processes for the simulated water
treatment plant:
¦ Organic and inorganic water quality;
¦ Disinfectant residuals;
¦ DBPs (THMs and HAAs) formed; and
¦ Percentage of CT achieved (based upon the CT numbers provided in the
USEPA Guidance Manual for Compliance With the Filtration and Disinfection
Requirements for Public Water Systems Using Surface Water Sources.
An example of input and output files for the program are presented in Tables 2-1
and 2-2, respectively.
23 CREATING AN EXAMPLE PROCESS TRAIN
This section outlines how to create a process train, enter process parameters, and run
the model. A typical process train is developed as an example.
2-5
RECYCLED PAPER
-------
TABLE 2-1
INPUT DATA FOR EXAMPLE 2
Raw_water
PH
. 7.5
Temperature
.15.0 Degrees Celsius
Annual Mln Temperature
. 0.5 Degrees Celsius
Total Organic Carbon
. 3.0 mg/L
UV Absorbance at 254 .nm
. 0.10 /cm
Bromide
. 0.10 mg/L
Alkalinity
.80 mg/L as Calcium Carbonate
Calcium Hardness
.80 mg/L as Calcium Carbonate
Total Hardness
.100 mg/L as Calcium Carbonate
Ammonia
. 0.05 mg/L as N
Turbidity
. 1.5 NTU
Giardia
. 2.0 cysts/100 L
Peak Hourly Flow
.20.0 M6D
Flow Rate
.10.0 MGD
Surface Water as defined by SWTR.
. Yes
Alum_Coagulation
Alum Dose
.10.0 mg/L as A12(S04)3*14H20
Flow Rate
.10.0 MGD
Surface Water as defined by SWTR.
.Yes
Theoretical Detention Time
.270.0 minutes
Ratio of Mean Detention Time to
Theoretical Detention Time....
. 1.0
Ratio of t(10) to
Theoretical Detention Time....
. 0.5
Chlorine
Chlorine Dose
. 4.0 mg/L as C12
Filtration
Theoretical Detention Time
.15.0 minutes
Ratio of Mean Detention Time to
Theoretical Detention Time....
. 1.0
Ratio of t(10) to
Theoretical Detention Time....
. 0.5
Basin
Theoretical Detention Time
.60.0 minutes
Ratio of Mean Detention Time to
Theoretical Detention Time....
. 1.0
Ratio of t(10) to
Theoretical Detention Time....
.0.5
Caustic
Caustic Dose
.11.0 mg/L as HaOH
Distribution
Average Residence Time
. 3.0 days
Maximum Residence Time
. 7.0 days
HAA Equations (H-Haas, T-TAW)
.H *
' Two sets of equations are available for HAA prediction as discussed in Section 3.2.
-------
TABLE 2-2
OUTPUT FILE FOR EXAMPLE 2
CHC13
CHBrC12
CHBr2Cl
CHBr3
TTHMs
Location
(ug/L)
(ug/L)
(ug/L)
(ug/L)
(ug/L)
Raw Water
0.0
0.0
0.0
0.0
0.0
Alum Addition
Basin Ef£luent
0.0
0.0
0.0
0.0
0.0
Chlorine
Filtered water
5.7
3.6
1.1
0.3
10.7
Basin Effluent
8.8
5.7
1.6
0.4
16.5
Caustic
Average Tap
28.4
19.0
5.3
0.6
53.3
End of System
35.8
24.0
6.7
0.7
67.2
Press ENTER to
continue
or Esc
to return to Main Menu.
HAAs Under Average Conditions Using Haas HAA Equations
MCAA
DCAA
TCAA
MBAA
DBAA
THAAs
Location
(ug/L)
(ug/L)
(ug/L)
(ug/L)
(ug/L)
(ug/L)
WTP Effluent...
1.0
7 . 7
4.3
0.3
0.9
14.2
Average Tap
1.0
10.6
8.9
0.4
1.2
22.1
End of System..
1.0
11.5
11.1
0.5
1.4
25.5
Predicted water
Quality
Profile
at Minimum Temperature
and Peak
Hour Flow
Temp
PH
C12
NH2C1
Inactivation
Location
(C)
(-)
(mg/L)
(mg/L)
Ratio
Raw Water
0.5
7.5
0.0
0.0
0.0
Alum Addition..
Basin Effluent.
0.5
7.2
0.0
0.0
0.0
Chlorine
Filtered Water.
0.5
7.1
3.0
0.0
0.1
Basin Effluent.
0.5
7.1
2.9
0.0
0.4
Caustic
Average Tap.. ..
0.5
8.0
1.0
0.0
0.4
End of System. .
0.5
8.1
0.3
0.0
0.4
-------
OUTPUT FILE FOR EXAMPLE 2 (CON'T)
Predicted Water
Quality
Profile
for EXAMPLE2
.WTP Under
Average
Conditions
PH
TOC
UV-254
Alk
Temp
C12
NH2C1
NH3-N
Location
(-)
(mg/L)
(1/cm)
(mg/L) (C)
(mg/L)
(mg/L)
(mg/L)
Raw Water
7.5
3.0
0.100
80
15.0
0.0
0.0
0.1
Alum Addition..
Basin Effluent.
7.2
2.3
0.048
75
15.0
0.0
0.0
0.1
Chlorine
Filtered Water.
7.1
2.3
0.048
72
15.0
3.0
0.0
0.0
Basin Effluent.
7.0
2.3
0.048
71
15.0
2.9
0.0
0.0
Caustic
Average Tap.. . .
8.0
2.3
0.048
84
15.0
1.0
0.0
0.0
End of System. .
8.1
2.3
0.048
83
15.0
0.3
0.0
0.0
Press ENTER to
continue
or Esc
to return to
Main Menu.
Bromide
Ca Hard
Mg Hard
Inactivation
Solids
Location
(mg/L)
(mg/L)
(mg/L)
Ratio
(mg/L)
Raw Water
0.10
80
20
0.0
0.0
Alum Addition..
Basin Effluent.
0.10
80
20
0.0
5.7
Chlorine
Filtered Water.
0.10
80
20
0.5
5.7
Basin Effluent.
0.10
80
20
2.3
5.7
Caustic
Average Tap
0.10
80
20
2.3
5.7
End of System..
0.10
78
20
2.3
5.7
Press ENTER to
continue
or Esc
to return to
Main Menu.
-------
A unit process flow diagram, detention times, and raw water quality data are shown
in Figure 2-5. This information is needed to create the process train and enter data. To
better understand the development of this process train, it is recommended that the unit
process components of the process train be arranged in a sequential block diagram, as
illustrated in Figure 2-6.
2J.1 Create Process Train
Once the information has been collected and organized, the following procedures can.
be followed to create the process train and operate the program:
¦ Step 1 - Insert the program diskette, type WTP, and press "Enter". The main
menu will be displayed on the screen.
¦ Step 2 - At the main menu, select Nw prwess train and press "Enter".
¦ Step 3 - Type the name of the plant to be simulated and press "Enter". This
displays the screen shown in Figure 2-7. The screen offers unit process options
the user can select to create your process train.
¦ Step 4 - Highlight each unit process desired and press "Enter" to construct the
process train. The options selected move to a list at the left side of the screen
as in Figure 2-8. When the user has completed choosing options, the unit
processes selected should match those in the block diagram in Figure 2-6.
¦ Step 5 - Select "Done - Process Train Complete" to finalize the process train.
The simulated plant is now created.
232 Unit Process Parameters
After the simulated plant has been created, the screen illustrated in the top of Figure
2-9 will appear. This, and a series of similar screens for each unit process selected when
constructing the simulated plant, will prompt the user for specific information unique to this
plant design, flow, and source water. The following sequence of steps can be used to enter
the specified information for each unit process:
¦ Step 1 - Enter the requested information at the data entry point marked by the
blinking cursor. Press "Enter". If the default value shown left of the arrow is
correct, press "Enter" to proceed to the next parameter.
¦ Step 2 - Enter the requested information for each subsequent prompt. Press
"Enter".
2-6
RECYCLED PAPER
-------
FIGURE 2-5
PROCESS FLOW SCHEMATIC FOR WTP
Alum = 10 mg/L
Chlorin* s 4 mg/L
Flocculatlon &
Sadlmantatlon
(270 mln)
t(mean)/t(theorstlcal) e 1.0
t(10)/t (theoretical) = 0.5
Caustic = 11 mg/L
nitration
(15 mln)
t(m«an)A(theoretlcal) = 1.0
t(10)/t(thaor«tlcal) = 0.5
Claar wall
(60 mln)
t(maan)/t(theoratlcal) « 1.0
t(10)A(thaoratical) = 0.5
-------
FIGURE 2-6
All diagrams start with raw
water (given).
RAW
-------
FIGURE 2-7
Neu Process Train
Unit Process Train:
Raw Uater
Available Unit Processes
|ft luw coagulationl
|Chl<
II ron coagulationl Ian won i a "
Isulfuric acid
[caust ic
|L int
Isoda ash
GAC
IMenbranes
F iltrat ion
|Basin~
Permanganate
Icarbon dioxide
ID istr ibut ion
IDone
— Process Train Conplete |
lOops
— Delete Last Entry 1
Select
Unit Process & Press ENTER
-------
FIGURE 2-8
Unit Process Train:
Raw Mater
§AV?a??i3iifSi00
Bas in
Ki?fi?fco
Basin.
Caustic
Distribution
Neu Process Train
Available Unit Processes
Ifllun coagulation] IChlorine
II ron coagulationl lawwonia "
Isulfuric acid | IF i It rat ion
Icaust ic
[Basin
ILiwe
Isoda ash
[GftC
perwanganate
Icarbon dioxide
IDistr ibut ion
IMenbranes
IDone
— Process Train Conplete I
Pops
— Delete Last Entry |
ICance 1
— Return to Main Henu(Esc)|
Select
Unit Process ft Press ENTER
-------
FIGURE 2-9
a) Initial Screen
= RAW WATER PARAMETERS ——
pH value: 7.7 —>
Type new value or press ENTER for no change
b) Completed Screen
===== RAW WATER PARAMETERS
pH value
7.7 —>
7.5
Temperature (Celsius)
18.0 —>
15.0
Annual Min. Temp. (Celsius)
0.5 ~>
0.5
TOC (mg/L)
3.00 —>
UV-254 (1/cm)
0.10 -->
Bromide (mg/L)
0.10 —>
Alkalinity (mg/L as CaC03)
50.00 —>
80
Calcium Hardness (mg/L as CaC03)
100.00 —>
80
Total Hardness (mg/L as CaC03)
120.00 —>
100
Ammonia (mg/L as N)
0.01 —>
.05
Turbidity (NTU)
2.00 —>
1.5
Giardia (cysts/100 L)
1.0 —>
2.0
Peak hour flow rate (MGD)
5.00 —>
20
Daily average flow rate (MGD)
2.00 —>
10
Surface Water as defined SWTR (Y/N):
Y -->
are the data correct (Y or N)
= Type new value or press ENTER for no change
-------
¦ Step 3 - After all the process information has been entered for a specific
process, the program will prompt: "are the data correct?" The initial screen
with filled data is shown in the bottom of Figure 2-9. Enter "y[es]n if the
information is correct. Enter "n[o]" if you need to change data, then enter
"y[es]" after the corrections have been made.
The program automatically proceeds to the next unit process in the simulated plant.
The above steps are duplicated. After all unit process data have been entered, the
simulated plant model can be run or modified from the main menu.
233 Modify Process Train
The following sequence can be used to modify the input parameters in the created
or retrieved process train:
¦ Step 1 - Return to main menu and highlight Of^ f^ocess train. Press "Enter".
¦ Step 2 - Type in the name assigned to the simulated process train. Press
"Enter". Type "y[esj" to review the unit process data. Type "n[o]" to return to
the main menu.
¦ Step 3 - Highlight MbdHy process train and press "Enter".
¦ Step 4 - Enter the number for the unit process to be modified from the ordered
processes in the simulated train on the screen (Figure 2-10). Press "Enter".
(Unit processes can be deleted by entering a negative number at this step.)
¦ Step 5 - Modify the input parameters for the selected unit process in the same
way the process unit parameters were entered.
¦ Step 6 - Enter "y[es]" or "n[o]M for modification status.
¦ Step 7 - Enter "y[es]B or "n[o]" to select the option to edit another process data
module. A "y[esj" entry invokes the process train modification menu.
¦ Step 8 - Enter "y[es]" to save the changes.
23.4 Run Model
¦ Step 1 - Return to main menu and highlight Open process train. Press "Enter".
¦ Step 2 - Type in the name assigned to the simulated process train. Press
"Enter". Type "y[es]" to see the unit process data. Type "n[o]" to return to
main menu.
2-7
RECYCLED PAPER
-------
FIGURE 2-10
===^== Modify Process Train ==
Current Working File: EXAMPLE2.WTP
0:Exit this Menu
1:Raw_Water
2:Alum_Coagulation
3:Basin
4:Chlorine
5:Filtration
6:Basin
7:Caustic
8:Distribution
Enter number to modify unit process data or
enter negative number to delete unit process or
enter 0 to exit this menu with option to save changes
>
Note: A Unit Process can be deleted, or the unit process data can be
modified, but a unit process cannot be inserted into an existing
process train.
-------
¦ Step 3 - At main menu, highlight Eup model and press "Enter". Type "c[rt]H to
send output to CRT; type "p[rinterj" to send output to printer; type "f[ilej" to
send output to a file. An example of the output is shown in Table 2-2.
Alternately, to see a summary of selected parameters, at Step 3 select
piid disinfection and press "Enter." This will send the selected output only to
the CRT. An example of the summary screen is shown in Figure 2-11.
These procedures and illustrative figures provide some direction on the operation of
WTP. Main menu options not specifically explained are described in Section 2.1, and are
selected by highlighting and entering their function title.
2 - 8
RECYCLED PAPER
-------
FIGURE 2-11
Predicted Water Quality Profile
PH
TOC
C12
NH2C1
IR at Min. T
THMs
Location
(-)
(mg/L)
(mg/L)
(mg/L)
& Peak Hr. Q
(ug/L)
Raw Water
3.0
0.0
0.0
0.00
0.0
Basin Effluent....
2.3
0.0
0.0
0.00
0.0
Filtered Water....
2.3
3.0
0.0
0.08
10.7
Basin Effluent....
.7.0
2.3
2.9
0.0
0.40
16.5
Average Tap Water.
.8.0
2.3
1.0
0.0
0.40
53.3
tend of System
.8.1
2.3
0.3
0.0
0.40
67.2
Press ENTER to return to Main Menu.
-------
3.0 INTERPRETING MODEL OUTPUT
This chapter provides a description of the output of the WTP program. An example
output file was presented in Table 2-2, and each of the parameters in the output file is
discussed in this chapter. The output generated after the "Run Model" command contains
the full output from the simulation exercise. This can be directed to the printer, disk or
screen. The output generated after the "THM and Disinfection" command contains a
summary of selected parameters displayed on the computer screen.
3.1 THM PREDICTION
The program predicts THM formation after chlorination and chloramination. The
program does not predict THM formation strictly from ozonation (in the absence of
chlorination and chloramination).
The program predicts TTHM formation using a single equation for total THM
(TTHM) concentration. The individual THM components are predicted using four
equations for the concentration of each individual THM. These proportions are then applied
to the TTHM concentration to produce the individual concentrations. The equations are
described in detail in Appendix A.
TTHMs presented in the output file represent the concentration predicted by one
TTHM equation. These concentrations also appear on the computer screen after the "THM
and Disinfection" command is selected. The proportion of each individual THM
concentration to the sum of the four THMs is determined from four individual THM
predictive equations.
An example considers a treatment process that predicts the following concentrations
of individual THMs:
50 /ig/L chloroform (CHC13)
25 /ig/L bromodichloromethane (CHBrCl2)
20 /ig/L dibromochloromethane (CHBr2Cl)
5 up/L bromoform (CHBr3)
100 /ig/L TTHM by summing individual species
3 -1
RECYCLED PAPER
-------
In this example, the proportion of chloroform to the TTHM concentration is 50/100, or
0.50. The proportions for the other THMs are determined in a similar manner.
If the single equation predicts a TTHM concentration of 80 /*g/L, then the program
will predict the following concentrations for the individual THMs:
40 /tg/L chloroform (CHC13)
20 ftg/L bromodichloromethane (CHBrCl2)
12 /tg/L dibromochloromethane (CHBr2Cl)
4 ^g/L bromoform (CHBr3)
80 /ig/L TTHM from single TTHM equation
Therefore, the individual THM concentrations associated with the TTHM value of 80 /xg/L
would be presented in the output file.
The reason for performing the analyses in this manner is that the equation for TTHM
has been determined to be more accurate than the sum of the individual species, based upon
verification analyses. Because the initial efforts using the model focused upon the impact
of different TTHM regulatory scenarios, the accuracy of the TTHM prediction was more
important than that for the individual predictions.
3.2 HAA PREDICTION
The program has the capability of predicting the concentrations of five individual
haloacetic acids (HAAs) using two different sets of equations. One set of equations is
referred to as the "Haas" HAA equations and the other is referred to as the "TAW" HAA
equations. A complete background and description of these equations is presented in
Section A.3 in Appendix A. The user is given the opportunity to select which set of
equations to use in the program, and the output file will present the results using the
appropriate equations.
The major difference between the Haas and TAW HAA equations is the method used
to calculate the concentrations at various stages of treatment. The TAW equations predict
HAA concentrations on a cumulative basis which accounts for HAA formation through
sequential treatment processes. In other words, the concentrations at a given point in the
treatment process is calculated by summing the concentrations predicted through each
treatment process.
3-2
RECYCLED PAPER
-------
The Haas equations predict HAA concentrations on a "static" basis. The statistical
correlations predict individual HAA levels based on concentrations of individual THM
concentrations and, therefore, do not account for the cumulative effect of changing water
quality parameters on the final levels of HAAs formed.
Because of the method used by the Haas equations to calculate HAA concentrations,
the predicted levels of HAAs may decrease from the plant effluent to the average
distribution system, under some treatment scenarios. For example, if the pH in the
distribution system is > 7.5, the Haas equations will predict low levels of HAAs. If low pH
values exist during treatment (e.g.< 6.0 during coagulation), however, the Haas HAA
equations may predict HAA levels at the plant effluent higher than those predicted at the
average distribution system.
The TAW equations do not demonstrate this trend. Using the above scenario, the
TAW equations would predict a level of HAA formation through the treatment plant based
on the predicted water quality parameters. If the pH is increased entering the distribution
system, the formation rate will decrease, but the concentrations of HAAs at the average
distribution system will always be greater than or equal to the plant effluent concentration.
3.3 DETERMINATION OF INACTIVATION RATIO
The inactivation ratio is used to evaluate whether a system meets disinfection
requirements for surface waters or ground waters. For surface water systems (or ground
water systems under the influence of surface water), the Surface Water Treatment Rule
(SWTR) requires a 3-log (99.9 percent) removal/inactivation of Giardia cysts and a 4-log
(99.99 percent) removal/inactivation of viruses. For ground water systems (not under the
influence of surface waters) it is anticipated that the Ground Water Disinfection Rule
(GWDR) may require a 4-log removal/inactivation of viruses. In the WTP model, the type
of source water (i.e., surface or ground water) is specified in the input file.
For surface water systems using coagulation and filtration, the model provides a
2.5-log removal credit for Giardia and a 2.0-log removal for viruses. Therefore, the required
inactivations for such systems are 0.5-log for Giardia and 2.0-log for viruses. For ground
water systems using coagulation and filtration, the model provides a 2.0 log removal of
viruses and therefore the required virus inactivation for such systems is assumed as 2.0-log.
3 - 3
RECYCLED PAPER
-------
Although the SWTR currently requires surface waters (or ground waters under the
influence of surface water) to achieve a minimum 3-log removal/inactivation of Giardia and
a 4.0-log removal/inactivation of viruses, the USEPA recommends that utilities achieve
greater inactivations depending on the Giardia concentration in the raw water. According
to the SWTR, the recommended levels of removal/inactivation are based on the raw water
Giardia concentrations as follows:
Daily Average Giardia Cyst Recommended Giardia Recommended Virus
Concentration/100 L Removal/Inactivation Removal/inactivation
For surface waters, the program establishes the recommended level of Giardia
removal/inactivation based on the raw water concentration of Giardia in the input file. For
example, if a Giardia concentration in the range of 1 to 10 cysts is input, the
removal/inactivation requirement will be 4-log for Giardia and 5-log for viruses. The log
removal credit through filtration for Giardia and viruses is similar to that discussed above.
Therefore, a system with a required 4-log and 5-log removal/inactivation for Giardia and
viruses, respectively, would be required to provide a 1.5-log inactivation of Giardia and a
3-log inactivation of viruses.
The inactivation ratio is defined as the level of inactivation (calculated as CT)
achieved through a given process divided by the required amount of inactivation for Giardia
or viruses from the SWTR or GWDR. If the value of the inactivation ratio at the treatment
plant effluent (representing the first customer) is equal to or greater than 1.0, the system
meets the disinfection requirements. For example, if the required CT value to meet a
required level of inactivation is 100, and the calculated CT value through a given treatment
process is 80, the resulting inactivation ratio for that process is 80/100, or 0.80. A complete
1 - Recent surveys (i.e., Le Chevallier, et al.. 1991) of the occurrence of Giardia in
surface water supplies indicate that Giardia concentrations in some supplies exceed the
maximum range (10-100) specified by USEPA in the SWTR. Therefore, linear extrapolation
of CT the criteria were made to include 6- and 7- log removal/inactivations for higher raw
water Giardia concentrations.
< 1
1-10
10-100
100-1,000'
1,000-10,000'
3-log
4-log
5-log
6-log
7-log
4-log
5-log
6-log
7-log
8-log
3-4
RECYCLED PAPER
-------
description of the inactivation ratio algorithm (for Giardia and viruses) is presented in
Appendix A.
It is important to note that when free chlorine is used as the primary disinfectant in
a surface water system using coagulation and filtration, a 0.5-log inactivation of Giardia will
provide greater than 2.0-log inactivation of viruses. Similarly, for a non-filtering surface
water, a 3.0-log inactivation of Giardia will provide a greater than 4.0-log inactivation of
viruses. As a result, when free chlorine is used as the primary disinfectant, the level of
inactivation required for Giardia is always greater than the corresponding level of
inactivation required for viruses.
If chloramines are used as the primary disinfectant in a surface water treated with
coagulation and filtration, however, a 0.5-log inactivation of Giardia will not provide greater
than 2.0-log inactivation of viruses. In this case, the model will still set the required level
of inactivation based on Giardia inactivation, not upon viruses. Thus, the inactivation ratio
would provide an inaccurate description of the system's disinfection requirements. In
subsequent versions, the model will use the higher of the two levels of inactivation for either
Giardia or viruses as the required inactivation.
It should also be noted that different types of filtration (i.e., direct, slow sand and
diatomaceous earth) can provide different removal of Giardia cysts and viruses than
coagulation and filtration systems. The current version of the model, however, does not
account for different removals that may be associated with other types of filtration systems.
It is intended that subsequent versions of the model will address this issue.
Because the WTP model does not yet contain equations for the decay of ozone or
chlorine dioxide, inactivation ratios are not calculated for these two disinfectants. When
these equations are added to the model, it will be possible to calculate inactivation ratios
from the use of these disinfectants.
The output file for a given modeled treatment system presents the inactivation ratio
under two different scenarios. The first scenario describes average temperature and average
flow conditions while the second describes minimum temperature and peak hourly flow
conditions. The second scenario represents the most stringent disinfection conditions and,
therefore, represents the conditions under which plants would most likely design their
treatment systems to meet the disinfection requirements. The inactivation ratios visually
displayed on the summary screen after the "THM and Disinfection" command is selected,
are the inactivation ratios predicted under the minimum temperature/peak flow conditions.
3-5
RECYCLED PAPER
-------
3.4 PREDICTION OF OTHER PARAMETERS
The following describes the remaining parameters in the output from a simulation
run:
¦ gH - The predicted pH value at the effluent of a given unit process.
¦ TOC - The predicted TOC concentration (in mg/L) at the effluent of a given
unit process.
¦ UV-254 - The predicted UV-254 value (in 1/cm) at the effluent of a given unit
process.
¦ Alk - The predicted alkalinity (in mg/L as calcium carbonate) at the effluent of
a given unit process.
¦ Temp - The average temperature (in °C) through a given unit process. Changes
in temperature through a process train are not calculated in this version of the
program.
¦ C12 - The predicted free chlorine concentration (in mg/L) at the effluent of a
given unit process.
¦ NH2C1 - The predicted combined chlorine concentration (in mg/L) at the
effluent of a given unit process.
¦ NH3-N - The predicted concentration of ammonia (in mg/L as N) at the
effluent of a given unit process.
¦ Bromide - The predicted bromide concentration (in mg/L) through a given unit
process. (Note: bromide removal is not calculated in any unit process because
of the lack of information on the fate of bromide in water treatment).
¦ Ca Hard - The predicted calcium hardness (in mg/L as calcium carbonate) in
the effluent of a given unit process. Calcium hardness is removed when the
solubility of calcium carbonate is exceeded.
¦ Mp Hard - The predicted magnesium hardness (in mg/L as calcium carbonate)
in the effluent of a given unit process. Magnesium hardness is removed when
the solubility of magnesium hydroxide is exceeded.
¦ Solids - The predicted concentration of solids (in mg/L) produced by the
addition of either alum or ferric chloride or by the precipitation of calcium
carbonate or magnesium hydroxide. The rate of solids production (lbs/day) can
be obtained by multiplying this result by the plant flow rate.
3-6
RECYCLED PAPER
-------
REFERENCES
G. L. Amy, B. C. Alleman and C. B. Cluff (1990). "Removal of Dissolved Organic Matter
by NanofUtration." J. Env. Eng.. 116(1), p. 200.
G.L. Amy, J.H. Greenfield, and WJ. Cooper (1990). "Organic Halide Formation During
Water Treatment Under Free Chlorine Versus Chloramination Conditions." in Water
Chlorination: Chemistry. Environmental Impact and Health Effects. Vol. 6.; R.L. Jolley,
et al„ eds; Lewis Publishers, Chelsa, MI.
G. L. Amy, P. A. Chadik and Z. K. Chowdhuiy (1987). "Developing Models for Predicting
Trihalomethane Formation Potential and Kinetics." J. AWWA. 79(7), p. 89.
B. Batchelor (1989). "A Kinetic Model for Formation of Disinfection By-Products.", Internal
USEPA Report.
B. Batchelor, G. Fusilier and E. H. Murray (1987). "Developing Haloform Formation
Potential Tests." J. AWWA. 79(1), p. 50.
R.J. Bull and F.C. Kopfler (1991). Health Effects of Disinfectants and Disinfection By-
products. AWWA Research Foundation, Denver, CO.
P.A. Chadik and G.L. Amy (1983). "Removing Trihalomethane Precursors from Various
Natural Waters by Metal Coagulants." Journal AWWA. 75(10), p532.
T. J. Christ and J. D. Dietz (1988). "Influence of Bromide Ion Upon Trihalomethane
Formation and Speciation." Proc. 1988 AWWA Nat'l Conference, Orlando, FL.
R.M. Clark and S. Regli (1991). "The Basis for Giardia CT Values in the Surface Water
Treatment Rule; Inactivation by Chlorine." In: Guidance Manual for Compliance with the
Filtration and Disinfection Requirements for Public Water Systems Using Surface Water
Supplies; USEPA, Washington, D.C.
R. M. Clark (1987). "Modeling TOC Removal by GAC: The General Logistic Function."
J. AWWA. 79(1), p. 33.
R. M. Clark, J. M. Symons and J. C. Ireland (1986). "Evaluating Field Scale GAC Systems
for Drinking Water." J. Env. Eng.. 112(4), p. 744.
J. Corollo Engineers (1989). "Pilot Study Final Report; Union Hills Water Treatment Plant
Water Quality Enhancement Study." Prepared for the City of Phoenix, AZ.
K.G. Denbigh and J.C.R. Turner (1971). Chemical Reactor Theory: An Introduction. 2nd
Edition, Cambridge University Press, Cambridge, UK.
H. Dharmarajah, N.L. Patania, J.G. Jacangelo, and E.M. Aieta (1991) "Empirical Modeling
of Chlorine and Chloramine Residual Decay." in Water Quality for the new Decade. 1991
Annual Conference Proceedings, AWWA.
R -1
RECYCLED PAPER
-------
J. K. Edzwald, W. C. Becker and K. L. Wattier (1985). "Surrogate Parameters for
Monitoring Organic Matter and THM Precursors." J. AWWA. 77(4), p. 122.
J. K. Edzwald (1984). Removal of Trihalomethane Precursors bv Direct Filtration and
Conventional Treatment. USEPA Municipal Environmental Research Laboratory, Rept.
No. EPA-600/2-84/068, NTIS Publ. No. PB84-163278.
B. A. Engerholm and G. L. Amy (1983). "A Predictive Model for Chloroform Formation
from Humic Acid." J. AWWA. 75(8), p. 418.
P. Fair (1990). Letter to S. Regli, USEPA.
G.W. Harrington, Z.K. Chowdhury, and D.M. Owen (1991). "Integral Water Treatment
Plant Model: A Computer Model to Simulate Organics Removal and Trihalomethane
Formation." in Water Quality for the New Decade. 1991 Annual Conference Proceedings,
AWWA.
S.A. Hubbs and G.C. Holdren (1986). Chloro-Organic Water Quality Changes Resulting
from Modification of Water Treatment Practices. AWWA Research Foundation, Denver,
CO.
R.E. Hubel and J.K. Edzwald (1987). "Removing Trihalomethane Precursors by
Coagulation." Journal AWWA. 79(7), p98.
H. Hudson (1981). Water Clarification Processes: Practical Design and Evaluation. Van
Nostrand Reinhold, New York, NY.
M.C. Kavanaugh (1978). "Modified Coagulation for Improved Removal of Trihalomethane
Precursors." Journal AWWA. 70(11), pl63.
D.G. Kleinbaum and L.L. Kupper (1978). Applied Regression Analysis and Other
Multivariate Methods. Duxbury Press, Boston, MA.
W.R. Knocke, S. West and R.C. Hoehn (1986). "Effects of Low Temperature on the
Removal of Trihalomethane Precursors by Coagulation." Journal AWWA. 78(4), pl89.
S. W. Krasner (1991). Internal Correspondence to Distribution, February 1, 1991.
M. W. Le Chavallier, W.D. Norton and R. G. Lee (1991) "Occurrence of Giardia and
Cryptosporidium in Surface Water Supplies". Applied and Environmental Microbiology.
57(9), p. 2610.
O. Levenspiel (1972). Chemical Reaction Engineering. 2nd Edition, John Wiley & Sons,
New York, NY.
B. W. Lykins, R. M. Clark and J. Q. Adams (1988). "Granular Activated Carbon for
Controlling THMs." J. AWWA. 80(5), p. 85.
R -2
-------
Malcolm Pimie, Inc. and Metropolitan Water District of Southern California (1991).
Evaluation of SDWA Impacts on Metropolitan Member Agencies.
Malcolm Pirnie, Inc. and City of San Diego (1990). Water Quality Report.
Malcolm Pirnie, Inc. and City of Phoenix (1989). Water Quality Master Plan.
James M. Montgomery Consulting Engineers and AWWA (1992). Mathematical Modeling
of the Formation of THMs and HAAs in Chlorinated Natural Waters.
James M. Montgomery Consulting Engineers and Metropolitan Water District of Southern
California (1989). Disinfection Bv-Products in U.S. Drinking Waters. United States
Environmental Protection Agency and Association of Metropolitan Water Agencies;
Cincinnati, OH and Washington, DC.
J.C. Morris and R.A. Isaac (1985). "A Critical Review of Kinetic and Thermodynamic
Constants for the Aqueous Chlorine-Ammonia System." in Water Chlorination: Chemistry.
Environmental Impact and Health Effects. Vol 4; lewis Publishers, Chelsa MI.
R. G. Quails and J. D. Johnson (1983). "Kinetics of the Short-Term Consumption of
Chlorine by Fulvic Acid." Environ. Sci. Technol.. 17(11), p. 692.
M.J. Semmens and T.K. Field (1980). "Coagulation: Experiences in Organics Removal."
Journal AWWA. 72(8), p476.
P. C. Singer (1988). Alternative Oxidant and Disinfectant Treatment Strategies for
Controlling Trihalomethane Formation. USEPA Risk Reduction Engineering Laboratory,
Rept. No. EPA/600/2-88/044, NTIS Publ. No. PB88-238928.
P. C. Singer and S. D. Chang (1989). "Correlations Between Trihalomethanes and Total
Organic Halides Formed During Water Treatment." J. AWWA. 81(8), p. 61.
W. Stumm and J. J. Morgan (1981). Aquatic Chemistry: An Introduction Emphasizing
Chemical Equilibria in Natural Waters. John Wiley and Sons, New York, NY.
J. M. Symons (1991). Correspondence to D. Owen, March 1, 1991.
J. S. Taylor, D. M. Thompson and J. K. Carswell (1987). "Applying Membrane Processes
to Groundwater Sources for Trihalomethane Precursor Control." J. AWWA. 79(8), p. 72.
J. S. Taylor, L. A. Mulford, W. M. Barrett, S. J. Duranceau and D. K. Smith (1989). Cost
and Performance of Membranes for Organic Control in Small Systems: Flagler Beach and
Punta Gorda. Florida. USEPA Risk Reduction Engineering Laboratory.
S. M. Teefy and P. C. Singer (1990). "Performance and Analysis of Tracer Tests to
Determine Compliance with the SWTR." J. AWWA. 82(12), p. 88.
R - 3
RECYCLED PAPER
-------
United States Environmental Protection Agency (1991). Guidance Manual for Compliance
with the Filtration and Disinfection Requirements for Public Water Systems Using Surface
Water Supplies.
J.S. Young and P.C. Singer (1979). "Chloroform Formation in Public Water Supplies: A
Case Study." Journal AWWA. 71(12), p87.
R -4
-------
APPENDIX A
DESCRIPTION OF MODEL EQUATIONS
-------
A.1 INTRODUCTION
This appendix presents the basis of a model that simulates disinfection by-product
(DBP) formation and removal of natural organic matter (NOM) in water treatment plants.
The purpose of the model is to:
¦ determine DBP levels that can be achieved by existing treatment technologies,
given the requirements for microbiological safety;
¦ identify those technologies that may be considered best available technology
(BAT) for DBP control; and
¦ provide a tool that will assess the impacts of new regulations on DBP formation
in existing treatment plants.
The basic modeling approach begins with the estimation of DBP precursor removal
by individual process units in the process train of interest. The fate of applied disinfectant
through the treatment process train is analyzed and the concentration of the disinfectant at
the beginning and end of a process unit is determined. The final step involves the
calculation of DBP formation based on water quality through the process train.
At this time, the model is only capable of simulating THM and HAA formation
because of the lack of information on the kinetics of formation and decay of other DBPs.
At this time, the formation of only trihalomethanes (THMs) and haloacetic acids (HAAs)
can be predicted. The formation of THMs have been studied much more extensively than
HAAs, and therefore the database for THM formation is considerably more robust.
Insufficient data are available to develop equations to predict other DBPs at this time.
This appendix summarizes the development of equations for simulating:
¦ total and individual THM concentrations;
¦ total and individual HAA concentrations;
¦ removal of total organic carbon (TOC) and ultraviolet absorbance at 254 nm
(UV-254);
¦ changes in alkalinity and pH; and
¦ decay of chlorine and chloramines.
a -1
RECYCLED PAPER
-------
A2 EQUATIONS FOR THM FORMATION
A2.1 Published Models for THM Formation
Numerous studies have used linear regression techniques to correlate trihalomethane
formation potential (THMFP) with TOC and UV-254. While these results showed good
correlations, general use of these regression equations is limited because they do not include
parameters such as chlorine dose, pH, temperature and time (Edzwald, et al.. 1985; Singer
and Chang, 1989; Lykins, et al.. 1988; Batchelor, et al.. 1987).
Additional studies have taken a more thorough approach for modeling kinetics of
THM formation. Engerholm and Amy (1983) developed a model that simulates the rate
of chloroform formation from pH, temperature, TOC and the ratio between chlorine dose
and TOC. The following equation was developed from chlorination of a peat soil humic
acid extract:
CHCl} = kjk/TOC)
0.95
[TOC I
\0.28
(A-l)
where CHC13 is the chloroform concentration in /ig/L, TOC is the total organic carbon
concentration in mg/L, Cl2 is the chlorine dose in mg/L and t is the reaction time in hours.
The value of k, is pH dependent while the values of z and k2 are temperature dependent.
Amy, Chadik and Chowdhury (1987) studied chlorination of 13 natural waters from
various locations throughout the United States. These chlorination studies produced the
following equation:
THM = 0.00309[(TOC)(UV-254)J0U°(Cl2)04O9(t)0265(T)106(pH -2.6)071$(Br +1)0036 (A-2)
where THM is the TTHM concentration in /imole/L, TOC is the total organic carbon
concentration in mg/L, UV-254 is the absorbance of ultraviolet light at a 254 nm wavelength
(in cm"1), Cl2 is the chlorine dose in mg/L, t is the reaction time in hours, T is the
temperature in degrees Centigrade and Br is the bromide concentration in mg/L. Raw
A-2
RECYCLED PAPER
-------
water samples were filtered through a 0.45 /*m filter. Therefore, the TOC used in the model
actually could be considered DOC by operational definition.
A third model, developed by Christ and Dietz (1988), was based on chlorination
studies involving two natural waters from Florida. The following equation was generated:
THM = 0.06Sf7TOQft9JO(>^,^(7)0652fC//^Br)aMV)a2^ (A"3)
where THM is the TTHM concentration in /ig/L, TOC is the total organic carbon
concentration in mg/L, T is the temperature in degrees Centigrade, Cl2 is the chlorine dose
in mg/L, Br is the bromide concentration in mg/L and t is the reaction time in hours.
Table A-l shows a comparison of test conditions for each of the three equations
shown above. This comparison demonstrates that Equation A-2 was based upon the most
thorough database incorporating the largest number of observations and widest range of
parameter values.
A fourth model has been developed for the USEPA that is based on a mechanistic
framework rather than an empirical framework (Batchelor, 1989). This model simulates the
formation and, where applicable, decay of four THMs, five non-THM DBPs and total
organic halogen (TOX). Simulations are based on TOC, pH, time, bromide and chlorine
dose. Temperature was not considered due to lack of data. Parameter values in the model
were based on results observed for water from the Ohio River at Cincinnati, Ohio.
Additional tests conducted on water from the Edisto River in South Carolina showed that
the parameter values developed for Ohio River water were not applicable to Edisto River
water. Tests should be conducted to develop parameter values from a group of waters so
that the model can be used to describe the general case. Temperature should be a variable
in these tests.
Equation A-2 was selected as the basis for THM formation in the WTP program.
A22 Refinement of Equation A-2
One limitation associated with Equation A-2 is its simulation of TTHM formation in
/imoles/L. The molecular weights of the four THM species vary significantly (119 for
CHClj to 252 for CHBr3) and, therefore, conversion from /xmoles/L to /ig/L is difficult. In
order to convert the TTHM concentrations calculated by Equation A-2 from /tmole/L to
A - 3
RECYCLED PAPER
-------
TABLE A-1
COMPARISON OF TRIHALOMETHANE FORMATION MODELS
Authors
Published
in
Number
of
Natural
Waters
Tested
Number
of
Data
Points
r*
Range of Values Tested
TOC
(mg/L)
UV-254
(1/cm)
Chlorine
Dose
(mg/L)
Bromide
(mg/L)
PH
Temperature
CO
Time
(hr)
Engerholm and Amy
1983
01
648
0.99
2.5 - 10.0
N.U. 3
5-60
N.U.
5.5 - 8.5
10-35
1 - 96
Amy, Chadik and Chowdhury
1987
13
1090
0.90
3.0- 13.8
0.063 - 0.489
1.5-69
0.01 - 1.245
4.6 - 9.8
10-30
0.1 - 168
Christ and Dietz
1988
2
180
N.R. *
CNJ
1
CO
CO
N.U. 3
8-25
0.087 - 0.500
7-9
15-30
2-144
Notes:
1 Humic acid extracted from a peat soil was used during this effort.
2 N.R. - Not Reported taba-i.wki
3 N.U. « Not Used in this study. 06/07/92
-------
/tg/L, the database used to develop Equation A-2 was used to develop a model which
describes the average molecular weight of the THM species formed. Using a stepwise
variable selection procedure, the following equation was developed with multiple linear
regression:
AMW = 105.3(Br +lfM{UV-254)-<>™ (A_4>
where AMW is the average molecular weight of the trihalomethane species formed. The
results calculated by Equations A-2 and A-4 can be multiplied together to calculate TTHM
formation in ng/l*.
A23 Development of Models for Formation of THM Species
The database used to develop Equation A-2 was used to formulate new empirical
equations for individual THM species. The relationship between the various THM species
and the controlling variables such as pH, temperature, chlorine dosage, TOC and bromide
exhibit nonlinear behavior. An appropriate empirical relationship for such a function can
be of the following form:
Y = AfXjNX/fXJ (A-5)
where A, a, b and c are empirical constants; X„ X2 and X3 are independent variables and
Y is the dependent variable. This relationship can be linearized by taking logarithms of
both sides of the above equation. The resulting equation therefore becomes:
ln(y) = ln(/4) + aln(Xj) + b\n(XJ + cln(Xj) (A-6)
Multiple regression analysis can be performed to correlate ln(Y) with a linear combination
of the independent variables. This analysis determines the intercept, ln(A), and the slopes
for the independent variables (a, b and c). These constants can then be used to describe
the equation shown in Equation A-5.
a - 4
RECYCLED PAPER
-------
As a first step of the modeling effort, dependent and independent variables were
defined as shown in Table A-2. The experiments showed that THM formation from the
chlorination of NOM virtually stops at pH 2.6 (Amy, et al.. 1987). Therefore, 2.6 was
subtracted from pH so that THM formation is not calculated at a pH value of 2.6. Similarly,
the variable Br + 1 was used to ensure no formation of brominated THM species when
bromide is not present in the water (the natural log of 1.0 is 0). The definition of X« could
be improved by adding stoichiometric coefficients to Fe, Mn, and NH3. This was not
determined, however, until after the modeling effort and this improvement has not yet been
tested.
The development of an appropriate regression equation consisted of selecting the
most significant variable forms from the list above and then performing a stepwise multiple
linear regression analysis using the selected variable forms. The selection of the appropriate
variable forms was done by developing a Pearson's Correlation matrix (Table A-3) for all
of the above variable forms on the entire data base. A Pearson's Correlation matrix shows
the correlation coefficients among all the variables in the matrix. The importance of a
particular variable in the regression equation is shown by the correlation coefficient,
considering the variable as a single predictor. A high correlation between two independent
variables indicates that if one is selected, the addition of the other will not improve the
significance of the regression.
In a stepwise regression analysis, the most significant independent variable describing
the dependent variable is taken into consideration first. Variables are added one at a time
according to the highest remaining correlation coefficient after the previously selected
variable is removed. Addition of independent variables increases the overall correlation
coefficient, but as the degrees of freedom decrease (as a result of increasing the number of
variables), the significance of the regression equation (portrayed by the F value) decreases.
In selecting independent variable forms for describing chloroform (Y,) formation, only
one occurrence of each of the controlling variables was desired. Table A-2 shows that the
correlation coefficient between Yl and X3 is higher (r = 0.603) than that between Y, and
either X, (r = 0.557), X2 (r = 0.589) or X4 (r = 0.303). Therefore, X3 was selected as the
form of the independent variable incorporating TOC and UV-254. Because X3 includes both
UV-254 and TOC, no other variables were desired for describing these precursors.
Similarly, X5 was selected instead of to represent chlorine dose.
A -5
RECYCLED PAPER
-------
TABLE A-2
VARIABLES DEFINED IN DEVELOPMENT OF TRIHALOMETHANE SPECIES
Dependent Variables
, I CHC1, ^
Yl ~ H 119 .38 j
Y2 = ln|
r CHBrCl2
k 163.83
• ln(
CHBr2Cl>
208.28,
r. - in( CHBr> |
4 \ 252.73 J
where CHCI3 is the chloroform concentration in pg/L, CHBrCI2 is the
dichlorobromomethane concentration in vg/L, CHBr2CI is the chlorodibromomethane
concentration in jjg/L and CHBr3 is the bromoform concentration in fjg/L The constants
used in the above equations are the molecular weights of the respective species.
Independent Variables
X1 = In (TOC)
X. = In {UV-254)
X3 = In [ (UV-254) (TOC)]
*4 =
= m/ uv~254\
\ TOC j
X5 = In(Cl2 Dose)
Xn = In (t)
X9 = In ipH-2.6)
X6 = In (Cl2 Dose-Fe-Mn-NH3)
xR ? ln(r)
Jf10 = ln(Br + l)
Xxl = In (Br)
*12 =
= In/———\
\ UV-254}
*13 -
-ln(.
Br
TOC
where TOC is the total organic carbon concentration in mg/L, UV-254 is the UV
absorbance at 254 nm in cm'1, Cl2 Dose is the free chlorine dose in mg/L, Fe is the iron
concentration in mg/L, Mn is the manganese concentration in mg/L, NH3 is the ammonia
concentration in mg/L, t is the reaction time in hours, T is the temperature in degrees
centigrade, pH is the pH of reaction and Br is the bromide ion concentration in mg/L.
TABA-2.WP
06/07/92
-------
TABLE A-3
PEARSON'S CORRELATION MATRIX FOR THM EQUATION VARIABLES
Variable
Variable
Name
Definition
X1
X2
X3
X4
X5
X6
X7
X8
X9
X10
X11
X12
X13
X1
Ln(TOC)
1.000
0.818
0.938
0.133
0.556
0.549
-0.006
-0.013
0.009
0.091
0.089
-0.260
-0.248
X2
Ln(UV254)
0.818
1.000
0.967
0.679
0.486
0.469
-0.009
-0.003
-0.238
0.041
-0.033
-0.445
-0.305
X3
Ln(UV254*TOC)
0.938
0.967
1.000
0.467
0.541
0.527
-0.008
-0.007
-0.139
0.066
0.020
-0.384
-0.294
X4
Ln(UV254/TOC)
0.133
0.679
0.467
1.000
0.128
0.107
-0.007
0.011
-0.421
-0.046
-0.170
-0.434
-0.207
X5
Ln(CI2 dose)
0.556
0.486
0.541
0.128
1.000
0.997
0.115
0.266
0.191
0.278
0.265
0.035
0.072
X6
Ln(CI2 Dose - Fe - Mn - NH3)
0.549
0.469
0.527
0.107
0.997
1.000
0.117
0.270
0.195
0.271
0.249
0.028
0.059
X7
Ln(Reaction Time)
-0.006
-0.009
-0.008
-0.007
0.115
0.117
1.000
0.036
0.043
0.041
0.035
0.035
0.037
X8
Ln(temperature)
-0.013
-0.003
-0.007
0.011
0.266
0.270
0.036
1.000
0.563
0.248
0.207
0.186
0.205
X9
Ln(pH - 2.6)
0.009
-0.238
-0.139
-0.421
0.191
0.195
0.043
0.563
1.000
0.255
0.294
0.362
0.283
X10
Ln(Br + 1.0)
0.091
0.041
0.066
-0.046
0.278
0.271
0.041
0.248
0.255
1.000
0.876
0.768
0.822
X11
Ln(Br)
0.089
-0.033
0.020
-0.170
0.265
0.249
0.035
0.207
0.294
0.876
1.000
0.910
0.943
X12
Ln(Br/UV254)
-0.260
-0.445
-0.384
-0.434
0.035
0.028
0.035
0.186
0.362
0.768
0.910
1.000
0.972
X13
Ln(Br/TOC)
-0.248
-0.305
-0.294
-0.207
0.072
0.059
0.037
0.205
0.283
0.822
0.943
0.972
1.000
Y1
Ln(CHCI3/119.38)
0.557
0.589
0.603
0.303
0.607
0.605
0.547
0.322
0.113
-0.101
-0.118
-0.351
0.301
Y2
Ln(CHCI2Br/163.8)
0.261
0.159
0.212
-0.059
0.527
0.513
0.494
0.377
0.379
0.611
0.715
0.575
0.608
Y3
Ln(CHCIBr2/208.3)
-0.027
-0.166
-0.112
-0.252
0.250
0.233
0.222
0.260
0.400
0.769
0.921
0.894
0.905
Y4
Ln(CHBr3/252.7)
-0.131
-0.212
-0.186
-0.198
0.121
0.108
0.118
0.196
0.374
0.850
0.823
0.825
0.844
TABA-3.WK1
06/07/92
-------
Variables X10 to X13 define the variable forms for bromine incorporation. However,
the selection of X3, which contains UV-254 and TOC, makes it less desirable to select X12
or X13 since these forms contain UV-254 or TOC. Selection X12 or X13 would result in a
multi-colinearity problem with X3. Y, had a slightly higher correlation coefficient for X„
(r = -0.118) than X10 (r = -0.101), however, the choice of variable X10 ensures that the
formation of chloroform will continue in the absence of bromide. A similar approach was
followed to select independent variables for the other THM species.
The elimination of independent variables which are highly correlated with each other
is important in avoiding multi-colinearity. This problem is minimized by using the
correlation matrix. In this effort, however, the elimination of all such variables was not
possible. For example, X5 is highly correlated with Xj (r = 0.56) because the chlorine dose
used in the development of the database was based on the TOC concentration. Both of
these variables are of prime importance in THM formation and, therefore, were considered
in the equation in spite of the high correlation between them.
The stepwise multiple linear regression analysis described above was performed on
the data base, resulting in the following equations:
CHCl3 = 0.21S[(UV-254)(TOC)]0616(Cl2 Dose)039l(t)0:l65(T)l ls(j)H-2.6)0SO0(Br *l)""3 (A-7)
[r = 0.941, adj r2 = 0.884, SEE = 0.313, F = 713 (a < 0.001), n = 561]
CHBrCl2 = 0.863[(l/K-254)(rOC)]0177(C/2 Dose)°™(t)°*\T)0™(pH-2.6f ™(Br)om (A-8)
[r = 0.945, adj r2 = 0.892, SEE = 0.360, F = 768 (a < 0.001), n = 561]
CHBr2Cl = 2.57|-^^j^,M(C/2 Dose^^itf^iTf^tpH-2.6)l3\Br)^ (A-9)
[r = 0.962, adj r2 = 0.926, SEE = 0.657, F = 1160 (a < 0.001), n = 561]
CHBr3 = 61 A(UV-254)om(Cl2 Dose)^"\t)on\T)*^(pH-2.6)lM^^n (A-10)
[r = 0.810, adj r2 = 0.652, SEE = 1.454, F = 176 (a < 0.001), n = 561]
A -6
RECYCLED PAPER
-------
The number of data points (n), the correlation coefficient (r), the adjusted coefficient of
determination (adj r2), the standard error of estimate (SEE) and the F-statistic at
significance level of a are shown beneath each equation.
An equation developed by regression analysis has strong descriptive capabilities if it
shows a high correlation coefficient. The F values for Equations A-7 through A-10 are
significant at the 99.9 percent confidence level as shown by a less than 0.001. The
correlation coefficients, however, are highest for chloroform and gradually decrease with
increasing bromine incorporation. This indicates a better quality of fit for the chloroform
equation and relatively worse qualities of fit for the brominated species.
The use of Equations A-7 through A-10 is limited in the following respects:
¦ the equations were based on THM formation in untreated waters and the
applicability of these equations to treated waters is unknown;
¦ the database used for developing the equations had waters primarily low in
bromide, so the use of these equations for high bromide waters may be limited;
¦ the equations were based on data collected from thirteen waters (Amy, et. al.,
1987) and, therefore, represent the best generalized method of simulating THM
formation. While a generalized approach is needed for the purposes of this
document, THM formation rates are site-specific due to variation in NOM
characteristics. Users of these equations are, therefore, encouraged to develop
site-specific parameter values instead of using the general parameter values
shown; and
¦ bromine incorporation during THM formation is a function of chlorine dosage.
Excess chlorination does not result in bromine incorporation as high as would
be realized under more moderate chlorination schemes. The database used for
developing the equations contained experiments with higher chlorine dosages
than would normally be expected in water treatment. Therefore, the equations
are expected to underpredict bromine incorporation.
The same limitations apply to the use of Equations A-2 and A-4.
\2A Calculation of Trihalomethane Formation Through Multiple Treatment
Processes
The calculation of trihalomethane formation in water treatment plants would be
relatively simple if water quality conditions were constant from the point of chlorination to
the end of the distribution system. This calculation would involve the use of Equations A-2
A-7
RECYCLED PAPER
-------
and A-4 for total trihalomethanes and the use of Equations A-7 through A-10 for the
individual trihalomethane species.
However, water quality conditions are not normally constant from the point of
chlorine application to the end of the distribution system. An example of this is shown on
Figure A-l, where chlorine is applied at time to prior to Basin 1. The removal of NOM at
time t, produces TOC and UV-254 levels in Basin 2 that are less than those in Basin 1. The
THM formation in these basins is calculated as follows:
¦ TOC„ UV-254, and t, are input into Equations A-2, A-4 and A-7 through A-10 to
calculate the THM concentrations at t,.
¦ TOQ, UV-2542 and t, are input into Equations A-2, A-4 and A-7 through A-10. The
results are subtracted from the results obtained when TOQ, UV-2542 and tj are input
into Equations A-2, A-4 and A-7 through A-10. These subtractions produce an
estimate of THM formation in Basin 2.
¦ The estimate of THM formation in Basin 2 is added to the estimate of THM
formation in Basin 1. A graphical representation of this procedure is shown in Figure
A-l.
Similar procedures are followed for changes in other variables such as pH.
The calculation of THM formation in systems employing two points of chlorination
is also an involved process. An example of this calculation is shown on Figure A-2. The
THM formation in these basins is calculated as follows:
¦ Equations A-2, A-4 and A-7 through A-10 are used to calculate the THM
concentrations at t,.
¦ After calculating THM concentrations at t„ the cumulative time was reset to zero and
Equations A-2, A-4 and A-7 through A-10 are used to calculate THM formation in
Basin 2. The cumulative time was not reset to zero in the example shown on Figure
A-l.
¦ The estimate of THM formation in Basin 2 is added to the estimate of THM
formation in Basin 1. A graphical representation of this procedure is shown in Figure
A-2.
This calculation method is necessary because THM formation is based only on the
chlorine dose and not on the chlorine residual. This is another limitation of the THM
formation models.
a - 8
RECYCLED PAPER
-------
FIGURE A-1
CALCULATION OF TRIHALOMETHANE FORMATION
Chlorine
Basin
1
1
Basin
2
1
TOC(2) < TOC(1)
UV-254(2) < UV-254(1)
PH(2) = PH(1)
Br(2) = Br(1)
T (2) = T(1)
*2
C
o
ca
*->
c
0
O
c
o
O
CD
C
(0
o
E
o
(0
Time
-------
FIGURE A-2
CALCULATION OF TRIHALOMETHANE FORMATION
FOR TWO POINTS OF CHLORINATION
Chlorine Chlorine
Basin
1
1 ^
TOC(2)
UV-254(2)
PH(2)
Br(2)
T(2)
TOC(1)
UV-254(1)
PH(1)
Br(1)
T(1)
c
0
1
+->
c
Q)
O
C
o
O
0)
c
cfl
a>
E
o
cd
sz
Time
-------
A25 Trihalomethane Formation in the Presence of Chloramines
The formation of trihalomethanes is possible in the presence of a chloramine residual
and has been observed (Hubbs and Holdren, 1986). This formation is possible for two
reasons:
¦ The reaction between free chlorine and ammonia to produce monochloramine is a
reversible reaction (Morris and Isaac, 1985). Therefore, free chlorine and
monochloramine can coexist in an equilibrium state. The ability of free chlorine
present in such a system to form trihalomethanes depends on the rate of the forward
reaction relative to the rate of the reverse reaction.
¦ Mixing conditions in the treatment plant are not sufficient to bring free chlorine and
ammonia into contact with each other instantaneously. The presence of free chlorine
for this limited period of time may result in trihalomethane formation.
Studies have not been conducted to evaluate the conditions under which
trihalomethane formation is observed during chloramination. Based on the nature of the
reversible reaction between free chlorine and ammonia, however, it can be speculated that
free chlorine is present in higher concentrations when the chlorine to ammonia ratio is
increased. Also, because the rate of the reaction between free chlorine and ammonia is
optimal at pH 8.2, it can be speculated that free chlorine is present at higher concentrations
when pH conditions deviate from pH 8.2.
Even with this speculation, there are insufficient data available to develop a model
of trihalomethane formation during chloramination. Nevertheless, because trihalomethane
formation is observed during chloramination, the model must account for it. A review of
the data produced by Hubbs and Holdren (1986) indicates that trihalomethane formation
observed during chloramination was approximately 10 percent of the formation observed
during free chlorination. Bull and Kopfler (1991) used an estimate developed by Amy, et al.
(1990) that trihalomethane formation in chloraminated waters "would approximate 20
percent of that observed if the same waters were chlorinated." Based on this review, the
model assumes that the rate of trihalomethane formation during chloramination is 20
percent of the rate of trihalomethane formation during chlorination.
a -9
RECYCLED PAPER
-------
A3 EQUATIONS FOR HAA FORMATION
A3.1 Equations Based on Utility Survey
A nationwide survey was conducted to evaluate DBP concentrations in treatment
plant effluents (JMM and Metropolitan, 1989). During this survey, finished water samples
were collected from 35 utilities nationwide and analyzed for THMs and other DBPs. Four
samples were collected from each of the participating utilities representing four seasons.
These data were statistically analyzed at the Illinois Institute of Technology to develop
correlations between THMs and other DBPs.
The general relationship between HAA and THM concentrations took the following
form (Haas, 1990):
f(DBP) = A + B(Ln CHCL,) + C(Ln CHCl2Br) + D(Ln CHClBr2) + E(TOC) +
F(pH) + G(CHBr3 Class) (A-ll)
Where:
A, B, C, D, and E are coefficients having the general format:
|C,|
Coefficient = (Constant) + |XiX2x3| | Cj |
|C,|
Where:
| x, x2 x31 = A vector of descriptor variables used to classify the disinfection category;
and
IQ|
| Cj | = A vector of constants associated with each independent variable.
10,1
In this vector of coefficients, the first row (C,) is multiplied by the value of x„ the second
row (Cj) is multiplied by the value of x2 and the third row (Q) is multiplied by the value of
x3. The sum of these three products would be the numerical value of the coefficient. In
other words, whenever a column vector of coefficients appears in one of the regression
equations, it is multiplied by the row vector | x, x2 x3 | to determine the numerical value of
the coefficient.
The values for x„ x2 and x3 are determined from the following table and are a function
of the disinfection practices of a given treatment plant:
a -10
RECYCLED PAPER
-------
Descriptor Variables by Disinfection Type
Disinfection Group
x3
Free Chlorine
1
0
0
Post-Ammoniation
0
1
0
Concurrent
0
0
1
Other
-1
-1
-1
Because many of the reported CHBr3 concentrations were below detection limit, the .
regression analysis used a variable called "Br3 Class". This variable was based on coding into
classes spaced on a logarithmic scale. In this manner, the large number of bromoform
observations below detection limits could be included in the regression without biasing the
results. The values of Br3 Class are listed in Table A-4. As shown in the table, all
observations having a bromoform concentration less than 0.101 /ig/L (the detection-limit
used for the survey samples) were assigned a value of 1 for Br3 Class. All observations
having a bromoform concentration greater than 0.101 ng/L but less than 0.126 ^g/L were
assigned a value of 2 for Br3 Class, and so on. The class boundaries increased by a factor
of 1.25 until all observations were included, with the highest value of Br3 Class
corresponding to the highest observed bromoform concentration (81.6 ng/L) in the 35-
Utility DBP Study data base.
The constants developed during this regression analysis are listed in Table A-5 for the
various HAA species. The example below illustrates the use of this model to predict the
concentration of DCAA for a utility in which the only disinfectant employed was free
chlorine.
Given: [CHC1J = 16.69 fig/L
[CHBrJ = 7.21 /ig/L (Br3 Class = 21)
[TOC] = 4.01 mg/L
pH = 7.63
Further: For free chlorine, x, = 1, x2 = 0, and x3 =0.
10.668651
(DCAA)"3 = 1.33006 + |100| j 0.64799 [ + 0.32686 Ln(CHCl3) - 0.00887 (Br3 Class)
11.248041
10.171661 - 10.147851
+ TOC 11001 j 0.01770 j + pH |100| j-0.081431 (A-12)
|-0.01272| - j 0.156251
a- n
RECYCLED PAPER
-------
TABLE A-4
VALUES OF THE INDEPENDENT VARIABLE "Br, CLASS"
Upper Class
Value of
Number of
Boundary for
Variable
Observations
CHBr3 (fjg/L)
"Br3 Class"
in Class
0.101
1
46
0.126
2
4
0.158
3
2
0.197
4
2
0.247
5
4
0.308
6
1
0.385
7
3
0.482
8
3
0.602
9
7
0.753
10
5
0.941
11
5
1.176
12
9
1.470
13
6
1.837
14
3
2.296
15
4
2.871
16
3
3.588
17
1
4.485
18
1
5.607
19
1
7.008
20
4
8.760
21
3
10.950
22
1
13.688
23
0
17.110
24
5
21.388
25
2
26.734
26
3
33.418
27
3
41.773
28
5
52.216
29
0
65.270
30
3
81.587
31
1
TBLA-4.ABG
06/07/92
-------
TABLE A-5
CONSTANTS FOR DBP REGRESSION ANALYSIS
DCAAA(1/3)
ln(TCAA)
In(MBAA)
In(DBAA)
In(MCAA)
CONSTANT
C1
C2
C3
1.33006
0.66865
0.64799
1.24804
2.47537
-2.05187
-1.27803
-0.63410
LCHC13
L BrC12
L Br2C1
TOC
PH
Br3 Class
LC13*T0C
Br3Class*T0C
LBr2C1*pH
0.32686
-0.00887
1.07748
-0.42038
0.08289
0.07401
0.13276
0.20800
C1 LC13
C2 LC13
C3 LC13
-0.10960
-0.32806
0.29479
C1 LBrC12
C2 LBrC12
C3 LBrC12
0.29841
0.31497
-0.88464
0.13349
0.44408
0.10853
C1 LBr2C1
C2 LBr2C1
C3 LBr2Cl
-0.24591
-0.22422
0.53450
C1 TOC
C2T0C
C3 TOC
0.17166
0.01770
-0.01272
0.06587
0.00956
-0.18197
C1 pH
C2 pH
C3 pH
-0.14785
-0.08143
-0.15625
-0.00733
0.10439
-0.07219
C1 Br3Class
C2 Br3Class
C3 Br3Class
-0.02820
-0.05424
-0.00925
YBLA-S.ABg
06/07/92
-------
Simplification of the above equation results in:
(DCAA)1/3 = 1.33006 + 0.66865 + (0.32686) (LnCHCl3) +
(-0.00887)(Br, Class) + (0.17166)(TOC) + (-0.14785)(pH),
and
DCAA = 12.1 ng/L (predicted)
A3 2 Equations Based on Laboratory Studies
The American Water Works Association (A WW A) Technical Advisory Workgroup
(TAW) for D/DBP sponsored a laboratory study to examine the formation of THMs and
HAAs as a result of chlorination. Data used for developing HAA predictive equations were
derived from two experimental studies as described below:
¦ In the D/DBP TAW database, kinetic experiments were conducted on raw water
samples from eight (8) utilities across the country. Each water sample was subjected
to a single dose of chlorine and a single incubation temperature of 20°C. The
chlorine dose was selected to maintain a free residual of 0.2 to 0.5 mg/L at the end
of the 96-hour reaction time. Samples were collected at ten different reaction times
over the 96-hour period and analyzed for HAAs. A series of additional experiments
were conducted on two of the eight water samples to briefly investigate the impact
of pH, chlorine dose, and temperature on the reaction kinetics. The entire database
contains 172 observations on HAA formation.
¦ In the D/DBP TAW coagulation database, raw water samples from eighteen utilities
were coagulated at different coagulant doses and at different coagulation pH values.
The treated water was filtered through a 8.0 um filter paper. The filtered water
sample were pH adjusted to 8.0, and was chlorinated at a 1:1 chlorine to TOC ratio,
and were incubated at 20°C for sixteen hours. Formation of HAAs was measured at
the end of the 16-hour reaction time. This database contained 71 observations on
HAA formation.
These laboratory observations on HAA formation were used to develop regression
equations correlating various HAA species with the water quality parameters. The
equations for MCAA, DCAA, TCAA, MBAA, and DBAA are shown in Tables A-6 to A-10.
These tables also include the boundary conditions relevant to each of the equations
(JMM, 1992).
A .3 3 Calculation of HAA Formation
Calculation of HAA formation in the model utilizes one of the two sets of empirical
equations outlined above. The user is given a choice to select which set of equations to use.
A - 12
RECYCLED PAPER
-------
TABLE A-6
MCAA EQUATION
TIME > 12 hours:
MCAA = 1.634 • (TOC)0753 • (BR+0.01 J"0 085 • (PH)"1124 •
(CL2DOSE) • (TIME)0,300
BOUNDARY CONDITIONS
STANDARD
PARAMETER
MIN
MAX
MEAN
DEVIATION
CL2DOSE/Br
9.8
819.9
196.4
211.3
CL2DOSE/TOC
1.0
2.3
1.5
0.4
TOC (mg/L)
2.8
11.0
4.3
1.8
UV-254 (cm1)
0.050
0.382
0.127
0.073
BR (mg/L)
0.01
0.43
0.14
0.14
PH
5.6
9.0
7.5
0.8
CL2DOSE (mg/L)
3.0
25.3
6.9
4.9
TIME (hr)
15.8
105.0
55.0
27.5
TEMP (°C)
13.0
20.0
19.4
1.9
MCAA fcyg/L)
1.2
22.0
6.0
4.8
TBLA-S.ABG
06/07/92
-------
TABLE A-7
DCAA EQUATION
DCAA = 0.605 • (TOC)0-291 • (UV-254)0 726 • (BR+0.01 J"0'568 •
(CL2DOSE)0 • (TIME)0-239 • (TEMP)0665
BOUNDARY CONDITIONS
STANDARD
PARAMETER
MIN
MAX
MEAN
DEVIATION
CL2DOSE/Br
9.8
819.9
202.3
214.3
CL2DOSE/TOC
1.0
2.3
1.6
0.4
TOC (mg/L)
2.8
11.0
4.4
1.9
UV-254 (cm1)
0.050
0.382
0.131
0.077
BR (mg/L)
0.01
0.43
0.13
0.14
PH
5.6
9.0
7.5
0.8
CL2DOSE (mg/L)
3.04
25.30
7.3
5.2
TIME (hr)
0.1
105.0
27.4
32.6
TEMP (°C)
13.0
20.0
19.5
1.8
DCAA (pq/L)
1.9
251.0
36.6
38.4
TBLA-7.ABG
06/07/92
-------
TABLE A-8
TCAA EQUATION
TCAA = 87.182 • (TOC)0 355 • (UV-254)0901 • (Bl
R + 0.01)-0-679 •
(PH) • (CL2DOSE)
°'881 • (TIME)0"2*
BOUNDARY CONDITIONS
STANDARD
PARAMETER
MIN
MAX
MEAN
DEVIATION
CL2DOSE/Br
9.8
819.9
202.3
214.3
CL2DOSE/TOC
1.0
2.3
1.6
0.4
TOC (mg/L)
2.8
11.0
4.4
1.9
UV-254 (cm1)
0.050
0.382
0.131
0.077
BR (mg/L)
0.01
0.43
0.13
0.14
PH
5.6
9.0
7.5
0.8
CL2DOSE (mg/L)
3.04
25.30
7.3
5.2
TIME (hr)
0.1
105.0
27.4
32.6
TEMP (°C)
13.0
20.0
19.5
1.8
TCAA (j/g/L)
1.7
485.0
72.6
90.1
fBU-fi.ABff
06/07/92
-------
TABLE A-9
MBAA EQUATION
MBAA = 0.176 • fTOC)1,664 • (UV-254)"0'624 • (BR)0'795 •
(PH) • (TIME)0145
• (TEMP)0450
BOUNDARY CONDITIONS
STANDARD
PARAMETER
MIN
MAX
MEAN
DEVIATION
CL2DOSE/Br
9.8
192.0
39.0
50.4
CL2DOSE/TOC
1.0
2.0
1.4
0.4
TOC (mg/L)
3.0
5.9
3.8
0.9
UV-254 (cm'1)
0.050
0.110
0.077
0.018
BR (mg/L)
0.05
0.43
0.24
0.13
PH
7.0
9.0
8.1
0.5
CL2DOSE (mg/L)
3.0
10.3
5.4
2.5
TIME (hr)
0.1
103.5
31.3
33.6
TEMP (°C)
13.0
20.0
19.4
2.0
MBAA (j/g/L)
0.5
4.6
1.8
0.9
T6U-8.AB5
06/07/92
-------
TABLE A-10
DBAA EQUATION
DBAA = 84.940 • (TOC)-0620 • (UV-254)0651 • (BR)1073 •
(CL2DOSE)-0-2®0 • (TIME)0120 • (TEMP)0 6®7
BOUNDARY CONDITIONS
PARAMETER
MIN
MAX
MEAN
STANDARD
DEVIATION
CL2DOSE/Br
9.8
280.0
41.1
59.9
CL2DOSE/TOC
1.0
2.0
1.4
0.4
TOC (mg/L)
3.0
5.9
3.7
0.9
UV-254 (crrT1)
0.050
0.170
0.080
0.025
BR (mg/L)
0.02
0.43
0.24
0.13
PH
5.6
9.0
8.1
0.6
CL2DOSE (mg/L)
3.0
10.3
5.2
2.4
TIME (hr)
0.1
103.5
27.6
32.8
TEMP (°C)
13.0
20.0
19.4
2.0
DBAA (jjg/L)
0.6
31.0
11.0
7.9
TBLA-IO.ABG
06/07/92
-------
The application of the equations presented in Section A.3.2 to the treatment plant
parameters for calculation of HAA concentrations follow the same techniques described in
Section A.2.4 for THMs. The application of the equations presented in Section A.3.2 to the
treatment plant parameters is based on statistical correlations and illustrated in
Section A.3.2.
A.4 EQUATIONS FOR REMOVAL OF NATURAL ORGANIC MATTER
A.4.1 Alum Coagulation, Flocculation, Clarification and Filtration
Many investigators have evaluated the impacts of alum dose and coagulation pH on
the removal of natural organic matter (NOM) at a bench-scale level (Kavanaugh, 1978;
Young and Singer, 1979; Semmens and Field, 1980; Chadik and Amy, 1983; Knocke, et al..
1986; Hubel and Edzwald, 1987). In general, these studies have found that the removal of
NOM can be optimized by maintaining certain pH ranges during coagulation, flocculation
and clarification.
Despite extensive bench-scale testing, a complete field-scale assessment of pH and
coagulant dose impacts has not been published to-date. Therefore, an effort was made to
develop a model that described the removal of TOC and UV-254 in alum coagulation,
flocculation, clarification and filtration plants. TOC and UV-254 were selected to quantify
the NOM so that compatibility with the THM formation model was assured.
Removal of NOM by alum coagulation was based on full-scale process data collected
in three studies (Montgomery and Metropolitan, 1989; Singer, 1988; and Edzwald, 1984).
The complete field-scale database includes 45 data points obtained from the 17 treatment
plants shown on Figure A-3. Characteristics of the database are shown in Table A-ll and
indicate that the database covers a wide range of raw water qualities and treatment
conditions. Median TOC and UV-254 removals in this database were 26 and 68 percent,
respectively.
A multiple regression analysis was employed to analyze the influence of raw water
TOC (TOCq), alum dose, coagulation pH (pHc), temperature and chloride on finished water
TOC (TOCr) in these 17 treatment plants. Because alum dose, TOC and chloride data were
lognormally distributed, the natural logarithm of these variables was used for this analysis.
Neither temperature nor chloride had a statistically significant impact on TOCf in this data
A - 13
RECYCLED PAPER
-------
flgursU
ALUM COAGULATION TREATMENT PLANT LOCATIONS
. "\
> (.
¦' \
/
Contra Costa County. CA «Sacramento, CA
San Francisco, CA#El ^ CA .AuroraiCO
Santa Clara Valley,'CA
\ r' ' j ;
MWD - Mills WTP, CA«mwD - Weymouth W?P, CA
~Canton, NY
t. a Oneida, NY
i ^Norwich, CT
• Ha c ken sack, NJ
•vj 7
•Newport News, VA
I • Little Rock, AR
~' """ *•«., .* 1 <
Big Spring, TX® •Arlington, TX
i Monroe, NC
REFERENCES
~ Edzwald, 1984 ¦ Singer, 1988 • JMM and MWD, 1989
O
c
31
m
>
-------
TABLE A-11
CHARACTERISTICS OF FIELD-SCALE ALUM COAGULATION DATABASE
Distribution
Range
Median
Raw Water Quality
TOC (mg/L)
lognormal
1.11 -12.1
2.78
UV-254 (cm1)
lognormal
0.019-0.84
0.111
PH
normal
6.3 - 9.0
7.7
Alkalinity (mg/L as CaC03)
lognormal
5.8 - 149
52
Hardness (mg/L as CaCOa)
lognormal
10-550
93
Temperature (°C)
normal
2-31
18
Turbidity (NTU)
lognormal
0.4 - 26
2.4
Chloride (mg/L)
lognormal
1 -640
20
Finished Water Quality
TOC (mg/L)
UV-254 (crrr1)
Turbidity (NTU)
lognormal
lognormal
lognormal
0.78 - 6.3
0.002 - 0.095
0.04 - 0.61
1.98
0.037
0.16
Coagulation pH
normal
5.5 - 8.0
7.15
Alum Dose
lognormal
1.5-55
11
0313\549\tbl«\taba-11. wp
06/07/92
-------
base. The apparent lack of significance between temperature and TOC removal is in
agreement with the results of bench-scale studies performed by Knocke, et al. (1986).
The regression analysis was conducted a second time without temperature and
chloride. Because both the intercept and slope of the relationship between ln(alum dose)
and ln(TOCf) are likely to depend on ln(TOQ,) and pHj, several interactive variables were
also included in the analysis (Kleinbaum and Kupper, 1978). Without the interactive
variables, the resulting model would have indicated that the slope of the relationship
between ln(alum dose) and ln(TOCf) was independent of pHc and ln(TOCo). The variables
examined in this analysis were those shown in Table A-12.
A stepwise variable selection procedure was used to select the most statistically
significant of the independent variables and include them in an equation describing the
performance of the 17 treatment plants in the data base. The resulting equation and its
associated statistics are as follows:
In(TOC) = -0.16 + 1.16[ln(7t>C0)]
-0.45[ln(a/«w dose)] /A
-0.070[ln(rOC0)]tln(aiMm dose)] 1 '
+0.057(pHc)\\n(alum dose)]
[r2 = 0.968, adj r2 = 0.965, SEE = 0.087, F = 297 (a < 0.001), n = 44]
An analysis was performed to determine the differences between finished water TOC
levels predicted by Equation A-13 and those observed in the 17 treatment plants. Figure
A-4 shows that 60 and 90 percent of the predicted finished water TOC levels were within
0.2 and 0.4 mg/L, respectively, of the observed concentration. In addition, 90 percent of the
predicted values differed from observed values by less than 12 percent.
A sensitivity analysis of Equation A-13 is shown in Figures A-5 and A-6. These
figures indicate the following:
¦ TOC removal increases with increasing alum dose, however, the magnitude of the
increase becomes smaller as alum dose increases. This is consistent with observations
described in the literature.
¦ TOC removal increases with decreasing coagulation pH down to pH 5.5. This is also
in agreement with the literature. The range of coagulation pH levels in the database,
however, did not allow for the determination of an optimal coagulation pH.
A -14
RECYCLED PAPER
-------
TABLE A-12
VARIABLES DEFINED IN DEVELOPMENT OF NOM REMOVAL BY ALUM COAGULATION
Dependent Variable:
Independent Variables:
ln(TOC,)
ln(TOC0)
In (alum dose)
pHe
ln(TOCo) * ln(alum dose)
ln(TOCo) * pHe
In (alum dose) * pHe
ln(TOC0) * ln(alum dose)
pHe
Notes: TOCf = Filtered water total organic carbon in mg/L
TOC0 = Raw water total organic carbon in mg/L
alum dose = alum dose in mg/L as AI2(S04)3*14 H20
pHc = pH of coagulation
0313\549\tbls\taba-12.wp
06/07/92
-------
ftgunM
RESIDUAL PROBABILITY PLOT
99.9
99
111
O
cc
hi
CL
ID
o
95
80
LU
> 50
20
0.1
-0.6 -0.5 -0.4 -0.3 -0.2 -0.1 0 0.1 0.2 0.3 0.4 0.5 0.6 0.7
(OBSERVED EFFLUENT TOC) - (PREDICTED EFFLUENT TOC)
>
-------
figuraOB
IMPACT OF COAGULATION pH ON TOC REMOVAL BY
ALUM COAGULATION/FILTRATION
ALUM DOSE (mg/l)
0
c
3J
m
>
01
-------
•gunOS
100%
80%
O
LU
>
o
2
LU
CC
O
o
60%
40%
20%
0%
IMPACT OF RAW WATER TOC ON TOC REMOVAL
BY ALUM COAGULATION / FILTRATION
10
20 30 40
ALUM DOSE (mg/l)
50
60
>
6>
-------
¦ At alum doses greater than 10 mg/L, TOC removal increases with increasing raw
water TOC levels.
¦ At pH 5.5, the maximum possible removal of TOC is approximately 55 percent for
a water with a median TOC of 2.8 mg/L. Because the curves shown on Figure A-5
were developed from a regression analysis, the maximum possible removal of 55
percent is the expected average. Some utilities will observe greater removals while
others will observe lower removals.
A database developed by USEPA's Technical Support Division (Fair, 1990) was used
to develop a relationship for UV-254 removal. Using similar techniques to those used for
TOC removal, the following equation was developed:
\n(UV-254} = -4.64 + 0.879[ln( t/K-25^]
-0.185[ln(tf/u/M dose)] (A-14)
+ 0.564(pi/t)
[r2 = 0.733, adj r2 = 0.702, SEE = 0.299, F = 23.8 (a < 0.001), n = 30]
where UV-2540 is the raw water UV-254 in cm"1 and UV-254f is the filtered water UV-254
in cm'1. The basis for selecting the USEPA database for this equation is described
elsewhere (Harrington, et al.. 1991).
A lower quality of fit was obtained for Equation A-14 when compared with Equation
A-13. This lower quality of fit can be expected because the removal of UV-254 may also
occur as a direct result of oxidation processes conducted during the coagulation, flocculation,
clarification and filtration processes. An evaluation of this phenomenon has not been
conducted to date.
A.4.2 Ferric Coagulation, Flocculation, Clarification and Filtration
Unlike the equations for alum coagulation (based solely on full-scale observations)
the equations for ferric coagulation were based on bench and pilot studies. An empirical
equation for NOM removal by ferric salt coagulation was based on data collected from
several bench and pilot scale studies conducted by Malcolm Pirnie, Inc (1989, 1990). and
from one study conducted by John Carollo Engineers (1989). The following raw water
sources and geographical locations were represented:
a -15
RECYCLED PAPER
-------
¦ San Diego, California (140 observations);
¦ Central Arizona Project Water, Arizona (46 observations);
¦ Clermont, Ohio (9 observations);
¦ Akron, Ohio (9 observations);
¦ Elizabethtown, New Jersey (6 observations); and
¦ Omaha, Nebraska (5 observations).
Summary statistics of this database are shown in Table A-13. The ability of the ferric
coagulation equations to simulate NOM removal for the general case is not expected to be
as good as the alum coagulation equations because fewer waters were included in the
analysis and because the results may be skewed by the large number of observations from
two sources.
Using similar techniques to those used for developing the alum coagulation equations,
the following equations were developed for ferric salt coagulation:
log10(7£>9 = 0.316+0.89inog10(7OC0)]
-0.018[log10(FeCiJ dose)]
(A-15)
[adj r2 = 0.742, F = 200 (a < 0.001), n = 213]
log l0(UV-254} = 0.228+ 1.025[log w{UV-254j\
-O.OSSflogj^FeC/j dose)]
(A-16)
[adj r2 = 0.796, F = 247 (a < 0.001), n = 195]
where FeCl3 dose is the ferric salt dose in mg/L as FeCl3 • 6 H20.
A -16
RECYCLED PAPER
-------
TABLE A-13
SUMMARY STATISTICS FOR FERRIC CHLORIDE DATABASE
Standard
Parameter
Mimimum
Maximum
Mean
Deviation
Raw Water DOC (mg/L)
2.7
9.0
4.8
1.7
Raw Water UV-254 (1 /cm)
0.044
0.720
0.080
0.08
Raw Water pH
6.0
10.8
8.1
0.7
FeCI3 Dose (mg/L)
0.3
75
15.8
14.8
Treated Water DOC (mg/L)
1.9
8.9
3.9
1.3
Treated Water UV-254 (1/cm)
0.010
0.420
0.050
0.04
Treated Water pH
5.6
10.8
7.4
0.9
j:\0313\549Ubls\tob»-13.wp
06/07/92
-------
A.4J Precipitative Softening, Clarification and Filtration
Full-scale data, including all of the parameters required for modeling, have been
compiled for only twelve precipitative softening plants. TOC, UV-254, lime dose and total
hardness values were available for most of the treatment plants. Seven of the plants
reported calcium and magnesium hardness values in addition to total hardness.
Nine field-scale observations were available for these seven plants (Table A-14). Four
observations were collected from four plants in the database developed by USEPA's
Technical Support Division (Fair, 1990), three observations were collected from one plant
in the database developed by Metropolitan and Montgomery (1989), and two observations
were collected from two plants recently evaluated by Malcolm Pirnie, Inc. (1991)
Using similar techniques to those used for developing the coagulation equations, the
following equations were developed for precipitative softening:
WJOC) = lWflndOCo)]
-0.0862[ln(ACfl)] (A*17)
- 0.0482[ln(AAfg)]
[r2 = 0.984, adj r2 = 0.978, SEE = 0.217, F = 120 (a < 0.0001), n = 9]
In (UV-254} = 0.923MUV-254
-0.0758[ln(ACa)] (A"18>
-0.198[ln(AA/g)]
[r2 = 0.989, adj r2 = 0.986, SEE = 0.366, F = 183 (a < 0.0001), n = 9]
where
ACa = Raw Water Calcium Hardness (mg/L as CaCO})
+ Calcium Hardness from Lime Addition (mg/L as CaC03)
- Finished Water Calcium Hardness (mg/L as CaC03)
and
AMg = Raw Water Magnesium Hardness (mg/L as CaC03)
- Finished Water Magnesium Hardness (mg/L as CaC03)
A - 17
RECYCLED PAPER
-------
TABLE A-14
CHARACTERISTICS OF FIELD-SCALE
PRECIPITATIVE SOFTENING DATABASE
Range Median
Raw Water Quality
TOC (mg/L) 2.2 -19.0 4.6
UV-254 (cm1) 0.030 - 0.697 0.123
pH 6.9 - 8.4 7.8
Alkalinity (mg/L as CaC03) 68 - 295 182
Calcium Hardness (mg/L as CaC03) 64 - 300 168
Magnesium Hardness (mg/L as CaC03) 0- 183 28
Total Hardness (mg/L as CaC03) 98 ¦ 360 247
Temperature (°C) 9 -29 23
Turbidity (NTU) 0.1-10 2.2
Finished Water Quality
TOC (mg/L) 1.7-9.5 2.9
UV-254 (cm ') 0.010 - 0.232 0.060
Removals
Calcium Hardness (mg/L as CaC03) 100 - 602 262
Magnesium Hardness (mg/L as CaC03) 0-90 16
Total Hardness (mg/L as CaC03) 104 - 644 346
Softening pH 8.9-11.3 10.2
Lime Dose (mg/L as CaC03) 109 - 394 228
Soda Dose (mg/L as Na2C03) 0-9.4 0
0313\549\tbls\taba-14. wp
06/07/92
-------
Because of the lack of data available for the development of the above equations, an
extensive survey of field scale precipitative softening plants is recommended.
A.4.4 Granular Activated Carbon
TOC breakthrough curves for alum coagulated, settled and filtered water were
described by the general logistic function (Clark, et al.. 1986; Clark, 1987). This function
is represented by the following equation:
where TOCf is the finished water TOC, TOQ is the influent TOC, 1/n is the slope of the
Freundlich isotherm and t indicates the time since the beginning of the adsorption process.
The two constants of the logistic function, A and r, can be determined by the procedure
discussed by Clark (1987).
The logistic function parameters, A and r, are a function of empty bed contact time
(EBCT), the type of GAC used and water quality factors such as pH and ionic strength. A
modified form of this equation is not developed at this time to account for these other
factors. The existing database on granular activated carbon (GAC) adsorption of NOM
should be evaluated to determine if Equation A-19 can be modified to reflect a general case.
In the meantime, the model assumes that the Jefferson Parish, Louisiana case is
representative of the general case, because this appears to be a median case when compared
with results from other utilities. For instance, TOC removal was higher at Jefferson Parish
than at Manchester, NH, Miami, FL and Metropolitan. TOC removal was lower at
Jefferson Parish, however, than at Philadelphia, PA, Cincinnati, OH and Shreveport, LA.
The parameters reported for the Jefferson Parish case are as follows:
(A-19)
1 +Ae'rt
r = 0.0743(£flC7)
-0.429
(A-20)
A = 0.757(EBCT)'35
(A-21)
A - 18
RECYCLED PAVER
-------
- = 0.316
n
(A-22)
The model uses Equations A-19 through A-22 to calculate a running average TOC removal
based on the EBCT and the regeneration frequency. Running average removals are of
interest because USEPA plans to regulate DBPs on a running annual average basis. The
model, therefore, is not intended to simulate the dynamics of TOC removal by this process
but, rather, simulate average water quality conditions within the treatment plant being
considered. In the algorithm, the running average TOC is calculated as follows:
ET.
TOC,
(JOCJ
,11-1
1 +Ae'rt
#1-1
(A-23)
RF
where RF is the regeneration frequency.
Because a sufficient database was not available to develop a separate equation for
UV-254 removal, the model assumes that UV-254 removal is equivalent to TOC removal
on a percentage basis. Because UV-254 is a surrogate measure of the more adsorbable
humic, non-polar fraction of organic material, this assumption is conservative. UV-254
removal would be expected to exceed that of TOC, which is comprised of both non-polar
and polar (less adsorbable) organic material.
A.4.5 Membrane Processes
Removal of NOM by membrane systems was based on bench scale process evaluations
performed by Taylor, et al. (1987 and 1989) and Amy, et al. (1990). Relevant water quality
characteristics from these studies are shown in Table A-15. Data from these three studies
were combined and analyzed for relationships between nominal molecular weight cutoff
(MWC) and removal of natural organic parameters. The best relationship obtained for
dissolved organic carbon (DOC) removal was:
[adj r2 = 0.886, n = 44]
A - 19
RECYCLED PAPER
-------
TABLE A-15
CHARACTERISTICS OF BENCH-SCALE,
CONTINUOUS FLOW MEMBRANE STUDIES
Range Median
Raw Water Quality
TOC (mg/L)
3.4 - 25.1
8.9
PH
5.8 - 7.9
7.2
Alkalinity (mg/L as CaC03)
20 - 338
97
Total Hardness (mg/L as CaC03)
152 - 352
231
Permeate Water Quality"
TOC (mg/L)
0.2 - 16.2
2.0
PH
5.9 - 7.7
6.8
Alkalinity (mg/L as CaC03)
7-343
48
Total Hardness (mg/L as CaC03)
4 - 365
122
Operating Parameters
Molecular Weight Cutoff
50 - 40,000
400
Operating Pressure (psi)
50 - 190
130
Recovery (%)
45-86
76
Product Flux (gpd/sf)
9-57
26
0313\549\tbl»\tab»-15.wp
06/07/92
-------
(A-24)
where DOCf and DOQ, are the product water and feed water DOC levels, respectively.
This equation had an adjusted R2 of 0.886 and was based on 44 observations from 6
different waters. Data for UV-254 removal were not reported in either study and, therefore,
the model assumes that UV-254 and DOC are removed to the same extent in membrane
systems.
A.5 EQUATIONS FOR ALKALINITY AND pH ADJUSTMENT
A.5.1 General
The addition of alum in drinking water treatment consumes alkalinity and,
consequently, depresses the pH. Alkalinity is defined by the following expression (Stumm
and Morgan, 1981):
Where [HC03 ] is the molar concentration of the bicarbonate ion, [C032] is the molar
concentration of the carbonate ion, [OH ] is the molar concentration of the hydroxide ion
and [H+] is the molar concentration of the hydrogen ion. The concentrations of the
carbonate and bicarbonate ions are pH dependent and may be defined as:
Alkalinity = [HCO3~] + 2[CO,2"] + [OH ] - [/T]
(A-25)
[HCOj ] - oc i Cf COj
(A-26)
[COj ] - fltjCj.COj
(A-27)
A-20
RECYCLED PAPER
-------
Where:
cTCOj = [H2co3] - [HCO;] + [CO]-]
(A-28)
a
(A-29)
[H+]2 * *,[//~] *
(A-30)
[H*]2 * KJH+] +
In the above equations, [H2C03] is the molar concentration of dissolved carbon dioxide
(carbonic acid), while K, and K2 are the acidity constants for carbonic acid and bicarbonate
ion, respectively. At 25°C and an ionic strength of zero, the value for K, is 10"4 3 while the
value for K2 is 10'10 3.
The model assumes that K, and K2 are temperature dependent as follows (Stumm and
Morgan, 1981):
where AH0 is the standard enthalpy change of the dissociation of carbonic acid to
bicarbonate, T° is the standard temperature of 298°K (25°C) and K,° is the value of the
equilibrium constant at T°. Solving Equation A-31 for K, produces the following:
Equations A-31 and A-32 assume that AH° is independent of temperature.
Using Equation A-32, the following equations were developed for K, and K2:
(A-31)
(A-32)
A - 21
RECYCLED PAPER
-------
K{ = exp
7700
mole
8.314
'K-mole t
f 1 -1]
^ 298.15° T)
14.5
(A-33)
Kj = exp'
14900
mole
8.314
'Kmole t
f 1 -1]
\ 298.15°^ rj
- 23.7
(A-34)
Equations A-33 and A-34 require the use of degrees Kelvin for the temperature, T. Degrees
Kelvin can be calculated by adding 273.15 to the temperature determined as degrees"
Centigrade.
The concentration of hydroxide ion, [OH"], is calculated as follows:
[OH] = (A-35)
[JH
where Kv is the ion product of water and is 10"M at 25°C. Like K, and K* K» is
temperature dependent. The model uses an empirical equation described in Stumm and
Morgan (1981) to calculate K„ at temperatures other than 25°C (298.15°K) as follows:
lo8io(^J = "44^°" + 6 0875 - 0.017067" (A-36)
where T is the temperature in degrees Kelvin.
Based on Equations A-25 through A-30 and on Equation A-35, Equation A-25 can be
rewritten as follows:
A - 22
RECYCLED PAPER
-------
Alkalinity = (Oj + 2a2)CTCOj + [0H~] - [i/+]
(A-37)
" [HI
As shown by this equation, alkalinity is dependent on [H+], Cr.coj and several equilibrium
constants. The equilibrium constants are dependent on temperature. Therefore, any change
in alkalinity, C-tcoj or temperature will produce a change in [H+]. Because pH is equal to
-log10[H+], any change in alkalinity, CtiC0, or temperature will produce a change in pH.
Electroneutrality also requires that alkalinity be defined by the following expression:
where CB is the equivalent concentration of all positively charged ions except hydrogen and
CA is the equivalent concentration of all negatively charged ions except hydroxide,
bicarbonate and carbonate.
Equation A-38 can also be written as follows:
Alkalinity = CB - CA
(A-38)
Alkalinity = C'B + 2 [Ca2*] + [CaOH*]
+ 2 [Mg2*] + IMgOH1 + [NH; ]
- C'A - [OCI-]
(A-39)
where:
[CaOH'] =
[Ca2*Wa,*
Co2*-CaOH'
(A-40)
[HI
A -23
RECYCLED PAPER
-------
[afgoin =
[HI
(A-41)
^Ca1' "CaOH' = exP
-72320-
= exp
mole
[8.314—I—1(7)
{ mole-°K)
(A-42)
and
^Mgu-MgOH' ~ exP
-65180-
exp
mole
(8.314 -—](T)
{ mole- K)
(A-43)
Using Equations A-40 and A-41, Equation A-39 can be rewritten as follows:
Alkalinity = c'B + [Ca2*]
[Mg2*]
'2 + ^Ca2'"CaOH' \
[H*] )
' v
2 + ^Ug1' -MgOH*
IH1
+ m;] - c'A - [ocn
(A-44)
The concentration of dissolved calcium is assumed to be the following:
ct.o = tCfl21 + [CaOH*] * [Ca(OH)2(eqJ (A-45)
A-24
RECYCLED PAPER
-------
where:
[Ca^*\Krj.
»«*W
(A-46)
and
V-ctoio.^ ~ exp(~^~)
exp
•159800-
(A-47)
mole
(8.314—-—\l)
V mole°Kj
Combining Equations A-40, A-45 and A-46, the following relationship is obtained:
r
2*1 _ *~T,Ca
[Ca2l =
J + KCa2* «• CaOH* + KCc*'-.CajO^, (A-48)
[HI [H*f
The concentration of dissolved magnesium is assumed to be the following:
CTMg = [A/g2t] * [MgOH+] * [Mg{OH)Uaq) (A-49)
where:
[Mg(OH)2{aqJ = Wg ]K"£~"S(OHh"1) (A_50)
and
Combining Equations A-40, A-49 and A-50, the following relationship is obtained:
A -25
RECYCLED PAPER
-------
Ktig* afa€) = eXP
= exp
-159760-
mole
(8.314——\j)
( molefiKf
(A-51)
IMg1^ =
'TMg_
j + KMg*-MgOH' + S(0H)2M
[#1
[H T
(A-52)
The concentration of dissolved free chlorine is assumed to be the following:
CTOa = [HOCl\ + [OCr] (A-53)
where:
[ifocq = Igcnun (A.54)
and
K
^Hoci-ocr
*Hoci-oci' ~ exP
13800
mole
8.314
5Kmole
1
298.15° A-
-i)
- 17.5
(A-55)
Combining Equations A-53 and A-54, the following relationship is obtained:
[OCT] =
1 . [HI (A-56>
*Hoci-ocr
The concentration of dissolved ammonia is assumed to be the following:
A -26
RECYCLED PAPER
-------
CTw} = [NH; ] + [NHJ
(A-57)
where:
[NHJ =
[NH^.
«h4 -«w,
[HI
(A-58)
and
knh; -nH) " exP
52210
mole
8.314
5 K-mole,
( 1 -1)
^298.15° AT rj
- 21.4
(A-S9)
Combining Equations A-57 and A-58, the following relationship is obtained:
m;] =
1 +
^NH'-NH,
[HI
(A-60)
Combining Equations A-44, A-48, A-52, A-56 and A-60, the following expression may
be written:
The model calculates [H+] based on the equality of Equations A-37 and A-61. As shown
by these equations, the model assumes that a change in [H+] occurs when a change occurs
in CA\ CB\ Ctq,, Cj-mj, Ct NH3, Ctoci. C,-^ or temperature. The calculation method is
described below.
Calcium Carbonate Solubility
The calcium ion and the carbonate ion are assumed to be in equilibrium with calcium
carbonate when water quality conditions cause calcium carbonate to precipitate from
solution. Under these circumstances, the molar quantity of calcium ions leaving the aqueous
A -27
RECYCLED PAPER
-------
Alkalinity = c'B - C'A
'T.Ca
2 +
Ca —CaOH'
UH
j + ^Ca1' —CaOH* +
m2
[/T]
'TMg
' If \
2 + Mg1' •'MgOH'
[HI
J + KM^-UgOH' + Kus1'-u8(Oirh(ai)
[//I [H*f
'TMi
'T.OCI
j + KNH:^NUy j +
[#+]
[HI
^uoct-ocr
(A-61)
phase is equivalent to the molar quantity of carbonate ions leaving the aqueous phase.
Therefore, the model assumes the following:
Ccoj>pt ~ ^cOyppt (A-62)
where ^ is the concentration of precipitated calcium in equilibrium with the aqueous
phase and Cc03ppt is the concentration of precipitated carbonate in equilibrium with the
aqueous phase. These concentrations are related to dissolved phase concentrations and
Equation A-62 can be rewritten as follows:
Cc " CT,Ca ~ Ccos ~ ^r.co, (A-63)
where is the sum of the aqueous phase calcium concentration (C^) and the
concentration of precipitated calcium in equilibrium with the aqueous phase. is the
sum of the aqueous phase carbonate concentration (Crco,) and the concentration of
precipitated carbonate in equilibrium with the aqueous phase.
The solubility of calcium and carbonate ions in equilibrium with calcium carbonate
is given by the following expression:
A -28
RECYCLED PAPER
-------
[C-'*][COh - KfjCaCO,
(A-64)
where
-12530
j»,CaCOj
= exp
mote
8.314
1
^298.15°*
-i)
- 19.1
(A-65)
'Kmole t
Based on Equations A-27, A-29 and A-44, Equation A-64 can be rewritten as follows:
Y WTjaH \
M pr? * *,[*•]~ K&) (A"66)
[#1 J
Solving Equation A-66 for CriCa yields the following:
xo.CoCOj
-r.Ca
J + ^Ca2'-CaOH' + ^Ca2' -Ca(OH^M
IH~] [J/*]2
'7, CO,
*1*2
^[JT]2 + KJ/T] +
(A-67)
Solving Equation A-63 for Cr.coi and substituting Equation A-67 for CrfQl yields the
following:
^Tjcoy ~ Cccty CCO
VCoCO,
J + ^Ca2' "CaOH' + *Ca*' w
[#*]
~l2
(A-68)
'r.co,
^CH2 + *,[#1 * K^j
A-29
RECYCLED PAPER
-------
Multiplying each term in Equation A-68 by Ct>co, yields:
0 = ^r.co, " *t>COj(CCOj ~ CJ
- K.
to.CaCO)
j + -CcOH' + ^Ca^-CtKPB^M
[HI [HI1
[h*]1 * *,[/n +
(A-69)
Using the quadratic formula to solve for Q- co, yields:
where
and
t.co,
-b + \/ b2 - 4ac
2a
(A-70)
a = l
(A-71)
b = C- - C™
"Co CO,
(A-72)
c = -A".
to,CaCOs
j + ^Ce1' "CaOH" +
Kca*'-
ra un2
[/T]2 + *,[#~] +
(A-73)
The model calculates from Equations A-70 through A-73 when
A -30
RECYCLED PAPER
-------
Y Wrxo, '
, + + Ko>--<*<*>,„, IWT2 - KJfll *
un wV J
(A-74)
After calculating Crcon the model calculates C^q, from Equation A-63. The calculated
values of Q-^oj and C, a are substituted into the alkalinity expressions in Equations A-37
and A-61.
The model uses Equation A-63 and Equations A-70 through A-73 only when the
aqueous phase is considered to be in equilibrium with calcium carbonate precipitate. If the
precipitate is removed by a treatment process and pH conditions are changed such that the
aqueous phase is undersaturated with respect to calcium carbonate precipitate, the model
assumes that the aqueous phase is no longer in equilibrium with calcium carbonate
precipitate.
Magnesium Hydroxide Solubility
The magnesium ion and the hydroxide ion are assumed to be in equilibrium with
magnesium hydroxide when water quality conditions cause magnesium hydroxide to
precipitate from solution. The solubility of magnesium and hydroxide ions in equilibrium
with magnesium hydroxide is given by the following expression:
[Mg2*] . K
[H*]2 ' "WOHh.
(A-75)
where
-113960
+ 38.8
(A-76)
8.314
O
Kmole
Based on Equation A-52, Equation A-75 can be rewritten as follows:
K
toMgiOl0j
(A-77)
[JT]2- IH1KU,
A - 31
RECYCLED PAPER
-------
Solving Equation A-77 for yields the following:
^TMg ~ * [H*]KUgi. ^UgOH. + KUgt. (A-78)
The model calculates from Equation A-78 when
Ksomaob^ * 2 (A"?9)
in j +1/1 -ug(pm2W
The model uses Equation A-78 only when the aqueous phase is considered to be in
equilibrium with magnesium hydroxide precipitate. If the precipitate is removed by a
treatment process and pH conditions are changed such that the aqueous phase is
undersaturated with respect to magnesium hydroxide precipitate, the model assumes that
the aqueous phase is no longer in equilibrium with magnesium hydroxide precipitate.
Calculating pH Changes
An iterative procedure is used to calculate the change in pH associated with chemical -
addition. This iterative procedure uses a bisection method to search for the concentration
of hydrogen ions, [H+], that allows the alkalinity calculated by Equation A-37 to be
equivalent to the alkalinity calculated by Equation A-61. The flowchart shown in Figure A-7
provides a detailed description of this procedure.
Open System Versus Closed System
The calculation of pH changes in the model is based only on equilibrium
considerations. This is an important limitation of the model because the kinetics of some
processes, such as calcium carbonate precipitation and carbon dioxide dissolution, may be
important.
A more realistic model would account for the kinetics of carbon dioxide transfer from
the air into water by using an appropriate mass transfer model with the appropriate mass
transfer coefficients. As an alternative to using such a kinetic model, two equilibrium
models were considered for the water treatment plant model described in this paper. One
A - 32
RECYCLED PAPER
-------
FIGURE A-7
-------
equilibrium model, referred to as the open system model, assumes that the concentration
of carbon dioxide dissolved in the water in treatment plant basins is in equilibrium with the
concentration of carbon dioxide in the atmosphere above the water. Hie other equilibrium
model, referred to as the closed system model, assumes no exchange of carbon dioxide
between the water and the atmosphere. The water treatment plant model described in this
paper assumes that the closed system model more closely approximates actual conditions
in a water treatment plant and distribution system than the open system model. This
assumption needs to be verified and the possibility of including the kinetics of carbon
dioxide dissolution into the water treatment plant model needs further consideration,
particularly for the unit processes in a water treatment plant.
A£2 Alum Coagulation, Flocculation, Clarification and Filtration
When alum is added to water, it dissociates according to the following equation:
For the purposes of this model, all of the sulfate ions are assumed to remain dissociated and
all of the aluminum ions are assumed to form insoluble aluminum hydroxide, Al(OH)3(s).
Therefore, when one mole of alum is added to one liter of water, the quantity (CV - CA')
decreases by six equivalents per liter and the alkalinity correspondingly decreases by six
equivalents per liter (see Equation A-61). The assumption that all aluminum ions form
insoluble aluminum hydroxide becomes less valid at pH levels farther from the pH of
minimum solubility (pH = 5.9 at 25°C). The iterative procedure described above is used
to obtain a new value for [H+] and, hence, pH.
AS3 Ferric Coagulation, Flocculation, Clarification and Filtration
When ferric sulfate is added to water, it dissociates according to the following
equation:
Aip0Jf\4H20 - 2 Al3* + 3 SO]' + 14 Hp
A-80)
Fe2(S0J3»H20 - 2Feyt * 3SO]~ * Hp
(A-81)
A -33
RECYCLED PAPER
-------
For the purposes of this model, all of the sulfate ions are assumed to remain dissociated and
all of the ferric ions are assumed to form insoluble ferric hydroxide, Fe(OH)3. Therefore,
when one mole of ferric sulfate is added to one liter of water, the quantity (Cy - CA')
decreases by six equivalents per liter and the alkalinity correspondingly decreases by six
equivalents per liter (see Equation A-61). The assumption that all ferric ions form insoluble
ferric hydroxide becomes less valid at pH levels farther from the pH of minimum solubility
(pH = 9 at 25°C). The iterative procedure described above is used to obtain a new value
for [H+] and, hence, pH. Also, the model does not account for any excess acidity that might
accompany commercial ferric sulfate.
When ferric chloride is added to water, it dissociates according to the following
equation:
For the purposes of this model, all of the chloride ions are assumed to remain dissociated
and all of the ferric ions are assumed to form insoluble ferric hydroxide, Fe(OH)3.
Therefore, when one mole of ferric chloride is added to one liter of water, the quantity (C„'
- CA') decreases by three equivalents per liter and the alkalinity correspondingly decreases
by three equivalents per liter (see Equation A*61). The assumption that all ferric ions form
insoluble ferric hydroxide becomes less valid at pH levels farther from the pH of minimum
solubility (pH = 9.0 at 25°C). The iterative procedure described above is used to obtain a
new value for [H+] and, hence, pH. Also, the model does not account for any excess acidity
that might accompany commercial ferric chloride.
A.5.4 Precipitative Softening, Clarification and Filtration
When added to water, lime dissociates according to the following expression:
Therefore, when one mole of lime is added to one liter of water, C^q, increases by one mole
per liter (see Equations A-74 and A-61). If the condition described by Equation A-74 is met
FeCl3»6 H20 - Fe3* * 3 C/" + 6H20
(A-82)
Ca(OH)Ut) - Ca2* + 2 OH~
(A-83)
A - 34
RECYCLED PAPER
-------
(this is the objective in precipitative softening), C^co, is recalculated with Equations A-70
through A-73 and Cx>Ql is recalculated with Equation A-63. If the condition described by
Equation A-74 is not met, however, C^coj does not change and Cj c, increases by one mole
per liter (see Equation A-61). The iterative procedure described above is used to obtain
a new value for [H+] and, hence, pH.
A.5.5 Chlorine Addition
When chlorine gas is added to water, it reacts according to the following equation:
d2 (g) + Hi° - Hocl * H* + CI (A-84)
For the purposes of this model, all of the chloride ions are assumed to remain dissociated.
Therefore, when one mole of chlorine gas is added to one liter of water, Cr.oa increases by
one mole per liter and the quantity (CB' - CA') decreases by one equivalent per liter (see
Equation A-61). The iterative procedure described above is used to obtain a new value for
[H+] and, hence, pH.
A.5.6 Sodium Hypochlorite Addition
When sodium hypochlorite is added to water, it reacts according to the following
equation:
NaOCl - No* * OCl- (A-8S)
For the purposes of this model, all of the sodium ions are assumed to remain dissociated.
Therefore, when one mole of sodium hypochlorite is added to one liter of water, Ct qc
increases by one mole per liter and the quantity (CB' - CA') increases by one equivalent per
liter (see Equation A-61). The iterative procedure described above is used to obtain a new
value for [H+] and, hence, pH.
A - 35
RECYCLED PAPER
-------
A.5.7 Potassium Permanganate Addition
When potassium permanganate is added to water, it dissociates according to the
following equation:
KMn04 - JT + MnO; (A"86)
For the purposes of this model, all of the potassium ions are assumed to remain dissociated
and all of the permanganate ions are assumed to react and form insoluble manganese
dioxide as shown by the following equation:
MnO; + AH* + 3«r" - Mn01 * 2H20 (A"87>
The assumption that all permanganate ions form insoluble manganese dioxide becomes less
valid at pH levels farther from the pH of minimum solubility.
When one mole of potassium permanganate is added to one liter of water, the net
effect of Equations A-86 and A-87 is to increase the quantity (CB' - Cx') by four equivalents
per liter. This net effect is only true if the oxidized product remains in solution. For
instance, if the permanganate ion oxidizes ferrous ions to ferric ions, the ferric ions can
precipitate as ferric hydroxide. The net effect of this precipitation in concert with Equations
A-86 and A-87 is to increase the quantity (CB' - CA') by only one equivalent per liter rather
than four equivalents per liter. The model presently assumes that the quantity (C„' - CA')
increases by one equivalent per liter when one mole of potassium permanganate is added
to one liter of water. Under most water treatment conditions, potassium permanganate is
added in quantities that do not change alkalinity to the degree that are effected by
coagulants. Therefore, this assumption is not expected to be critical in most cases.
A.5.8 Sulfuric Acid Addition
Sulfuric acid is typically used to decrease pH and alkalinity in drinking water
treatment. When added to water, sulfuric acid dissociates according to the following
equation:
A-36
RECYCLED PAPER
-------
H2SOa -2H* + sol'
(A-88)
Sulfate ions are assumed to remain completely dissociated (this assumption is valid for pH
greater than 2 and when the solubility limit of calcium sulfate is not exceeded). Thus, when
one mole of sulfuric acid is added to one liter of water, the quantity (C^' - CA') is decreased
by two equivalents per liter (see~Equation A-61). The iterative procedure described above
is used to obtain a new value for [H+] and, hence, pH.
XS.9 Sodium Hydroxide (Caustic) Addition
Sodium hydroxide is used to increase pH and alkalinity in drinking water treatment
When added to water, sodium hydroxide dissociates according to the following equation:
Sodium ions are assumed to remain completely dissociated. Thus, when one mole of
sodium hydroxide is added to one liter of water, the quantity (CB' - CA') is increased by one
equivalent per liter (see Equation A-61). The iterative procedure described above is used
to obtain a new value for [H+] and, hence, pH.
Lime is also added to increase alkalinity and pH levels in drinking water treatment.
When added to water, lime dissociates according to the following expression:
Therefore, when one mole of lime is added to one liter of water, C^q, increases by one mole
per liter (see Equations A-74 and A-61). If the condition described by Equation A-74 is met
(this is not the objective when using lime for pH adjustment), Cjcoj is recalculated with
Equations A-70 through A-73 and Q. q, is recalculated with Equation A-63. If the condition
described by Equation A-74 is not met, however, Ctcoi does not change and C^q, increases
NaOH - Na* * OH~
(A-89)
A.5.10 Calcium Hydroxide (Lime) Addition
Ca(OH)m - Ca2* + 2 OtT
(A-90)
A - 37
RECYCLED PAPER
-------
by one mole per liter (see Equation A-61). The iterative procedure described above is used
to obtain a new value for [H+] and, hence, pH.
A.5.11 Sodium Carbonate (Soda Ash) Addition
Soda ash is another chemical added to increase alkalinity and pH levels in drinking
water treatment. When added to water, soda ash dissociates according to the following
expression:
NafOj -2Na+ + CO]
(A-91)
Therefore, when one mole of soda ash is added to one liter of water, the quantity (CB' -
CA') is increased by two equivalents per liter and C^co, is increased by one mole per liter
(see Equations A-74 and A-61). If the condition described by Equation A*74 is met, CtiC0,
is recalculated with Equations A-70 through A-73 and Ctq, is recalculated with Equation
A-63. The iterative procedure described above is used to obtain a new value for [H+] and,
hence, pH.
A.5.12 Carbon Dioxide Addition
Carbon dioxide is added to lower pH levels and increase the carbonate concentration
in some precipitative softening systems. When one mole of carbon dioxide is added to one
liter of water, Ct
-------
Therefore, when one mole of ammonium hydroxide is added to one liter of water,
increases by one mole per liter (see Equation A-61). The iterative procedure described
above is used to obtain a new value for [H+] and, hence, pH.
A.5.14 Membrane Processes
Membrane processes can alter the alkalinity of a given water through the rejection
of specific ions. Changes in Ct q,, C,- ^ C^co, and (CB* - CA') through a membrane process
determine the alkalinity and pH of the permeate water. In an effort to predict pH changes
. through membrane processes, removals of hardness, alkalinity and Cx reported by Taylor,
et al. (1987,1989) were studied to determine their relationships with molecular weight cutoff
(MWC), recovery and operating pressure.
The best relationship obtained for hardness removal was:
In
THn
TH.
f)
= 13.1 -0.0568 (Recovery)
-0.261[ln(WC)Jln(77/o)]
[adj r2 = 0.876, n = 25]
(A-93)
where THq is the feed water total hardness in mg/L as calcium carbonate, THf is the
product water total hardness in mg/L as calcium carbonate, MWC is the molecular weight
cutoff, and Recovery is the ratio between the product water flow rate and the feed water
flow rate in percent. The model assumes that the percent removals of Cj q, and are
equivalent to the percent removal of total hardness as calculated by Equation A-93.
The best relationship obtained for alkalinity removal was:
In
ALKq)
-1
ALKf
\ f)
= 7.50 - 0.421 [ln(MWC)Jln(AIJC0)]
+ 0.00338[ln( MWC)\]n{ALK0)]Pressure)
- 0.0O0U5(PressureXRecoveryl]n(MWC)]
(A-94)
[adj r2 = 0.893, n = 17]
A-39
RECYCLED PAPER
-------
where ALK<, is the feed water alkalinity in mg/L as calcium carbonate, ALK, is the product
water alkalinity in mg/L as calcium carbonate, MWC is the molecular weight cutoff,
Pressure is the operating pressure in lb/in2 and Recovery is the ratio between the product
water flow rate and the feed water flow rate in percent.
The best relationship obtained for removal of total carbonates was:
In —^ -1 = 5.88 - O.QM5\\\n(MWC)\Pressure)
il'WJ ] (A-M)
- O.Om[in[MWC)jRecovery)
+ 0.000082(Pressure X Recovery )[ln (MTTC)]
[adj r2 = 0.783, n = 15]
where Cjcoi.o is the feed water total carbonate concentration in mg/L as calcium carbonate,
Cr ccn.f is the product water total carbonate concentration in mg/L as calcium carbonate,
MWC is the molecular weight cutoff, Pressure is the operating pressure in lb/in2 and
Recovery is the ratio between the product water flow rate and the feed water flow rate in
percent.
The changes in alkalinity and Cjcqj calculated by Equations A-94 and A-95 are used
in combination with Equation A-37 to calculate changes in pH. This iterative procedure
uses a bisection method to search for the concentration of hydrogen ions, [H+], that allows
the alkalinity calculated by Equation A-37 to be equivalent to the alkalinity calculated by
Equation A-94. Once the new value of [H+] is obtained, Equation A-61 is used in
combination with the new values of and 0,^ (calculated by Equation A-93) to
calculate a new value for the quantity (CB' - CA').
A.6 CHLORINE AND CHLORAMINE DECAY
A.6.1 Background
Rate equations for calculating the decay of chlorine and chloramines were developed
by Dharmarajah, et al. (19911 through funding provided by the AWWA D/DBP TAW. The
experimental design consisted of collecting raw water samples from 16 utilities nationwide
and observing the decay of chlorine and chloramines under laboratory conditions. The
A -40
RECYCLED PAPER
-------
equations which resulted from this effort are described in the following sections. The
equations for chloramine decay were modified for use in the WTP program as described in
Section A.6.3.
A.62 Chlorine Decay
Experimental observations suggested that the free chlorine residual decayed rapidly
for the first 5 hours after the addition of chlorine. The decay rate decreased for the
remainder of the experiment (120 hours). The decay during the first 5 hours was modeled .
as a second order reaction with respect to chlorine concentration while the decay after 5
hours was modeled as a first order reaction. The decay constants were correlated with raw
water quality parameters by multiple, stepwise regression analyses similar to those described
above. The following were the final equations for decay of free chlorine:
For (CMn:(TOQ £ 1.0 mg Cl/m? TOC
For 0 hours < t < 5 hours
1 1 + kxt (A-96)
(c/*), iaA
where (Cl2), is calculated from:
ra0
ln[(C/2)o - (CZ2)i - 7.6(NH3-N)] = - 0.620 ~ 0.522 jln ^
+ 0.302[ln( l/K—254)]
+ 0.842[ln(7OC)]
(A-97)
and (Cl2)0 is the applied chlorine dose.
For 5 hours < t < 120 hours
A-41
RECYCLED PAPER
-------
(«4
«p(-V)
(A-98)
where:
(«.),-
4 (M|
(A-99)
For rCU„:fTOa < 1.0 mg Cl/me TOC
For Q hours < t < 120 hours
(clA
exp(-V)
(A-100)
Where (Cl2), is calculated by Equation A-97. The difference between the applied chlorine
dose, (Cl2)o, and (Cl2), is considered the instantaneous chlorine demand.
The equations used to determine the chlorine decay rate constants, k„ k2 and k3, are
as follows:
ta[(cMj(*i)] ="244 -
(^)o
TOC
+ 0.799[In (1/7-254)]
+ 0A22(pH)
(A-101)
ln^) = - 1.67 + 1.00[In (1/7-254)]
+ 2.73 [In (7TOC)]
A -42
RECYCLED PAPER
-------
ln(*2) = - 2.31 - 2.12
+ 1.27[ln(l/K-254)]
- 0.842[ln(7TOC)]
+ OAl\(pH)
(A-102)
The boundary conditions for these experiments were:
- (C12)0:(TOC) = 0.5 to 4.0
¦ TOC = 2.0 to 13.9 mg/L
¦ UV-254 = 0.049 to 0.489 cm"1
¦ pH = 6.4 to 8.4
¦ (Cl2)0 = 1.0 to 41.6 mg/L
Temperature was not varied in these experiments and bromide spiking was not performed
(ambient bromide concentrations were not reported).
A.6.3 Chloramine Decay
Dharmarajah, et al. (1991), also developed rate equations to describe the decay of
chloramine concentration. These equations, however, were found to predict chloramine
formation under certain water quality conditions, particularly when a water with an alkalinity
greater than 150 mg/L as calcium carbonate was chloraminated with a chlorine to ammonia
ratio of 4:1. The prediction of chloramine formation occurred both within and outside the
boundary conditions of the chloramine decay experiments. Because the prediction of
chloramine formation is not desirable, USEPA funded Malcolm Pirnie to develop a new set
of equations from the data collected by Dharmarajah, et al. (1991). The revised set of
equations is as follows:
For 0 £ t ^ 10 hours
The decay over this time period was modeled as a generalized m* order reaction with
a rate constant k,. The differential form of the equation is as follows:
A -43
RECYCLED PAPER
-------
(A-104)
The integrated form of the equation is as follows:
[NH20\ = ~ (m - l)*^
(A-105)
Statistical analyses of the experimental data yielded the following correlations for the order
and the rate constant:
The low quality of fit for Equation A-106 indicates that the reaction order is not a strong
function of initial chloramine concentration for the initial portion of the decay curve.
For t > 10 hours
The decay over this time period was modeled as a first order reaction with a rate
constant k2. The residual chloramine concentration, [NH2C1]10, calculated by the above
equations at ten hours is used as the initial concentration for this set of equations. The
differential form of the equation is as follows:
m = \9n2\mfii
|0.0977
(A-106)
[r = 0.23, r2 = 0.05, adj r2 = 0.03, F = 2.08]
kx = (1.86 x 10"12 )[NH2C[\-^(TOC)1 M(t/F-254)3 83[ff+]101 (A-107)
[r = 0.91, r2 = 0.83, adj r2 = 0.81, F = 44.9]
d[NH2Cl\
d(t - 10)
= -kJLNH20\
(A-108)
A-44
RECYCLED PAPER
-------
The integrated form of the equation is as follows:
[NH2a\ = [iV/TjCi] 10exp[-*j(f - 10)]
(A-109)
Statistical analyses of the experimental data yielded the following correlation for the rate
constant:
fcj = 0.424(7100"<,187(£/K-254)°-376[/T+]0-227 (A-110)
[r = 0.74, r2 = 0.55, adj r2 = 0.51, F = 16.9]
The boundary values for the chloramine decay experiments were:
¦ Cl2:TOC = 0.5, 1.0 and 3.0
¦ C12:NHj.= 2.0 to 4.0 (42 of 54 tests were conducted at C12:NH3 = 4.0)
¦ TOC = 1.3 to 10.5 mg/L
- UV-254 = 0.039 to 0.401 cm"1
> NTU = 0.22 to 31.6 NTU
¦ Alkalinity = 7 to 290 mg/L as CaC03
¦ pH = 6.5 to 8.6
¦ Cl2 Dose = 0.7 to 31.5 mg/L
Temperature was not varied in these experiments and bromide spiking was not performed
(ambient bromide concentrations were not reported).
A.6.4 Modeling Chlorine and Chloramine Decay in Treatment Plants
Basins and filters are modeled as a number of continuous flow, stirred tank reactors
(CFSTRs) connected in series. A complete description of this model is provided elsewhere
(Denbigh and Turner, 1971; Levenspiel, 1972; Teefy and Singer, 1990), however, some key
points will be discussed here.
For a single CFSTR, the effluent concentration of a solute is determined with the
following expression:
A -45
RECYCLED PAPER
-------
C, + tr
(A-lll)
where Cf is the final concentration of the solute within and leaving the CFSTR, Q is the
initial concentration of the solute entering the CFSTR, t is the mean residence time in the
CFSTR and r is the reaction rate. Hie reaction rate is greater than zero for solute
formation and is less than zero for solute decay. The theoretical residence time in the
CFSTR is calculated by dividing the volume of the CFSTR by the volumetric flow rate of
the fluid moving through the CFSTR.
For a set of N CFSTRs connected in series, the final concentration of a solute leaving
the N* CFSTR is given by the following set of equations:
C„ = Cu * tf j (A-112)
'2J
'1/
(A-113)
/ ~ ^14 + hrl
C-Nj = ^N-l/
(A-114)
(A-115)
^n/ = + Vn (A-116)
where CN>f is the final concentration of the solute within and leaving the N* CFSTR, CN>i is
the initial concentration of the solute entering the N* CFSTR, tN is the mean residence time
in the N4 CFSTR and rN is the reaction rate in the N4 CFSTR. Equation A-112 can be
used to calculate Clif from known values of Cu, t, and r,. Once C1>f is calculated, Equation
A-113 is used to obtain Qti. Equation A-114 can then be used to calculate C^f from known
values of t2 and r2. The process is repeated through Equation A-116 until CN f is calculated.
A - 46
RECYCLED PAPER
-------
In Equations A-112 through A-116, the reaction rate is given by the differential
equation appropriate to the reaction of interest. For a first order decay process, rN is given
by the following expression:
rN = ~kCNJ
(A-117)
Integrated forms of this differential equation are shown above as Equations A-98, A-100 and
A-109. For a second order decay process, rN is given by the following expression:
An integrated form of this differential equation is shown above as Equation A-98. For an
m* order decay process, rN is given by the following expression:
An integrated form of this differential equation is shown above as Equation A-105.
In order to use a set of equations such as those defined by Equations A-112 through
A-116, one must determine the number of CFSTRs connected in series that best simulates
the basin or filter of interest. This number is determined by using a theoretical "F curve"
which describes the time dependent concentration of a tracer solute leaving the set of
CFSTRs in a step dose tracer study (for a more complete discussion, see Denbigh and
Turner, 1971; Levenspiel, 1972; or Teefy and Singer, 1990). The theoretical F curve is
defined by the following expression:
(A-118)
(A-119)
A -47
RECYCLED PAPER
-------
_ ZM
cu
= 1 - exp
ldal + >!L+±
> 1 r» 2!
{Nt)
•+••• + ¦
(N-l)\
(C
(A-120)
where N is the number of CFSTRs connected in series, tm is the mean residence time for
an entire series of equally sized CFSTRs and t is the time of interest.
The Surface Water Treatment Rule (SWTR), finalized by USEPA in 1989, has placed
great importance on the time at which F is equal to 0.1 (USEPA, 1989). This time is
frequently referred to as t10 because it represents the time required for the effluent tracer
concentration to be 10 percent of the influent tracer concentration in a step tracer study.
At a time equal to t10) the following holds:
0.1 = 1 - exp
-Nt
10
1 Nti0 1 1
2 1
f"'l 0
2!l
(JV-1)!
\N-1
(A-121)
Using this equation, the theoretical value of t,0:tm can be calculated for any value of N by
using an iteration procedure. The results of such an analysis are presented in Figure A-8
and are compared with data obtained from actual treatment plant basins and filters (Teefy
and Singer, 1990).
A given basin or filter is simulated as N equally sized CFSTRs based on user input
values for t,^, t10:ttbco and tm:ttlieo where tihoo is the theoretical residence time of the basin or
filter as given by:
<*.. -1
where V is the volume of water in the basin or filter and Q is the volumetric flow rate
through the basin or filter. The volume of water in a filter must not include the volume of
the filter medium or media. The ratios t^:^ and tm:t1heo can be determined by field-scale
A-48
RECYCLED PAPER
-------
tracer studies or estimated from results reported elsewhere (USEPA, 1991; Teefy and
Singer, 1990).
Because N can only take on integer values, a continuous distribution of does not
theoretically exist. However, actual tracer studies will not likely find the specific t10:tm ratios
shown on Figure A-8. Therefore, the selected value of N is based on the t10:tm ranges listed
in Table A-16.
Once the value of N is selected from Table A-16, Equations A-112 through A-116 are
used in conjunction with the appropriate rate equations to calculate chlorine and chloramine
decay in the basin or filter of interest.
A.7 DETERMINATION OF INACTIVATION RATIO
A.7.1 Background
The Surface Water Treatment Rule (SWTR) requires specific levels of disinfection
for surface water and ground water sources. For surface water systems (or ground water
systems under the direct influence of surface water), the minimum disinfection requirement
is based on the removal and/or inactivation of Giardia. For ground water sources, the
disinfection requirement is based on the removal and/or inactivation of viruses. The value
of the inactivation ratio determines whether a system meet the inactivation requirements
of the SWTR.
A.72 Calculating Inactivation Ratios
The inactivation ratio is defined as follows:
IR = IRX + //Jj + //?, + (A-123)
where:
^ CT value achieved through processi (A-124)
' CT value needed for inactivation requirement
The achieved CT value is calculated as follows:
A -49
RECYCLED PAPER
-------
PERFORMANCE OF CFSTRs IN SERIES
1.0
0.8
c
cd
a>
E
0.6
0.4
0.2
0.0
Observed
Theoretical
(Teefy & Singer, 1990)
~
A
r
3 ffi
C
^ 2
S
0
3
*
L
I
^ c
3
\ L
k
m
4 6 8
NUMBER OF CFSTRs IN SERIES
10
-------
TABLE A-16
DETERMINATION OF THE NUMBER OF CFSTRs IN SERIES
t,0lL
_N_
0.000 to 0.186
1
0.186 to 0.317
2
0.317 to 0.402
3
0.402 to 0.461
4
0.461 to 0.506
5
0.506 to 0.540
6
0.540 to 0.569
7
0.569 to 0.593
8
0.593 to 0.613
9
0.613 to 0.630
10
0.630 to 0.645
11
0.645 to 0.659
12
0.659 to 0.671
13
0.671 to 0.682
14
0.682 to 0.691
15
0.691 to 0.700
16
0.700 to 0.708
17
0.708 to 0.716
18
0.716 to 0.723
19
0.723 to 0.729
20
0.729 to 0.735
21
0.735 to 0.741
22
0.741 to 0.746
23
0.746 to 0.751
24
0.751 to 1.000
25
-------
^Achieved ~ ~~~
(A-125)
^thto
where Residual is the disinfectant residual (in mg/L) at the end of the contact time, ^ is
the theoretical contact time, and t10 is the detention time corresponding to the time for
which 90 percent of the water has been in contact with the residual concentration.
The required CT values for different levels of inactivation and specific treatment
conditions (i.e. temperature, pH and disinfectant residual) are obtained from the tables in
the SWTR Guidance Manual (USEPA, 1991). Tables A-17 to A-18 present the CT
requirements for ground water systems using chlorine or chloramines, respectively, as the
primary disinfectant. Table A-19 presents the CT requirements for surface water systems
using chloramines as the primary disinfectant. For surface water systems using chlorine as
the primary disinfectant, the required CT value can be estimated from the following
equation (Clark and Regli, 1991):
where Cl2 is the chlorine residual at the end of the contact time (measured as free chlorine
in mg/L), T is the temperature throughout the contact time, pH is the pH throughout the
contact time, N is the number of Giardia cysts remaining after the contact time and N0 is
the number of Giardia cysts prior to the contact time. Equation A-126 calculates values of
CTreq.(1 that are within 10 percent of the values listed in the SWTR Guidance Manual.
Equation A-126 is only valid for temperatures between 0.5 and 5°C. To calculate the
CT,.,.,, at higher temperatures, the value for CT^.,, is halved for every 10°C increment
increase in temperature. For example, the CT^.j at 10°C is one-half the CT,.,.,, value at
0.5°C and the CT^.j at 15°C is one-half the CT^.j value at 5°C (all other conditions equal).
The required log inactivation is given as follows:
CTnq,d = 0.36(C/2)015(pi/)2e9(D"0'15
(A-126)
A - 50
RECYCLED PAPER
-------
TABLE A-17
CT VALUES FOR INACIWATION
OF VIRUSES BY FREE CHLORINE
Log Inactivation
2,0 3.0 4.0
PH pH pH
Temoerature <°C)
£2
10
10
Ifl
0.5
6
45
9
66
12
90
5
4
30
6
44
8
60
10
3
22
4
33
6
45
15
2
15
3
22
4
30
20
1
11
2
16
3
22
25
1
7
1
11
2
15
-------
TABLE A-18
CT VALUES FOR INACTIVATION
OF VIRUSES BY CHLORAMINE
Temperature (°C)
Inactivation
< = 1
-15-
20
25
2-log
1,243
857
643
428
321
214
3-log
2,063
1,423
1,067
712
534
356
4-log
2,883
1,988
1,491
994
746
497
Note: CT values apply for systems using combined chlorine where chlorine is added
prior to ammonia in the treatment sequence.
-------
TABLE A-19
CT VALUES FOR INACTIVATION
OF GIARDIA CYSTS BY CHLORAMINE pH 6-9
Temperature (°C)
Inactivation
< = 1
_5_
_lfl_
15
20
25
0.5-log
635
365
310
250
185
125
1-log
1,270
735
615
500
370
250
1.5-Iog
1,900
1,100
930
750
550
375
2-Iog
2,535
1,470
1,230
1,000
735
500
2.5-log
3,170
1,830
1,540
1,250
915
625
3-log
3,800
2,200
1,850
1,500
1,100
750
-------
, . . . , (n\ (A-127)
log inactivation = -log 101 — 1
Although the SWTR currently requires surface waters to achieve a minimum 3-log
removal/inactivation of Giardia. the USEPA recommended that utilities achieve greater
inactivations depending on the Giardia concentration in the raw water. According to the
SWTR, the recommended levels of removal/inactivation are based on the raw water Giardia
concentrations as follows:
Daily Average Recommended Giardia Recommended Virus
Giardia Cyst Removal/inactivation Removal/inactivation
Concentration/ 100L
< 1
3-log
4-log
1-10
4-log
5-log
10-100
5-log
6-log
100-1,ooo1
6-log
7-log
1,000-10,000'
7-log
8-log
For surface waters, the program sets the required level of Giardia removal/inactivation
based on the raw water concentration of Giardia. Therefore, if a Giardia concentration in
the range of 1 to 10 cysts/100 L or 10 to 100 cysts/100 L is input, the removal/inactivation
requirement will be 4 logs (99.99 percent) or 5 logs (99.999 percent), respectively.
Surface water systems (or ground waters under the influence of surface waters) with
coagulation and filtration receive a 2.5-log removal credit for Giardia. Therefore, a surface
water plant with a 3-log (99.9%) removal/inactivation requirement would only be required
to achieve a 0.5-log inactivation.
It is anticipated that all ground water systems will be required to achieve a 4-log
removal/inactivation of viruses. According to the SWTR, a 2.0-log removal credit is given
1 Recent surveys (Le Chevallier, et al.. 1991) of the occurrence of Giardia in surface
water supplies indicate that Giardia concentrations in some supplies exceed the maximum
range (10-100) specified by the USEPA in the SWTR. Therefore, linear extrapolation of
the CT criteria were made to include 6- and 7- log removal/inactivations for higher raw
water Giardia concentrations.
A - 51
RECYCLED PAPER
-------
for systems with coagulation and filtration. Therefore, a ground water plant with
coagulation/filtration would be required to achieve a 2.0-log inactivation of viruses, while
an unfiltered ground water would be required to achieve a 4.0-log inactivation of viruses
The flowchart shown in Figure A-9 provides a detailed description of how the model
calculates the inactivation ratio through a given treatment process (i.e., a basin or filter).
A -52
RECYCLED PAPER
-------
FIGURE A-9
ALGORITHM FOR CALCULATING INACTIVATION RATIOS
-------
APPENDIX B
RESULTS OF MODEL VERIFICATION EFFORTS
RECYCLED PAPER
-------
B.1 INTRODUCTION
The model verification process presented in this Appendix is based upon an
evaluation of SDWA impacts on the member agencies of the Metropolitan Water District
of Southern California (Metropolitan). The overall project, completed in June 1990,
included the following major elements:
¦ modifying the model by incorporating new equations for chlorine and
chloramine decay, and individual THM species formation;
¦ performing a preliminary model verification using a 35 Utility Survey
(Montgomery and Metropolitan, 1989);
¦ collecting data from member agency plants for l)further model verification, and
for 2)providing baseline conditions for sequential process modifications to meet
projected scenarios resulting from the SDWA Amendments; and
¦ evaluating the process and economic impacts of the SDWA Amendments on
the member agency plants.
This effort was funded solely by Metropolitan and, accordingly, the validation effort
presented here includes only Metropolitan's member agency plants. This member agency
database is not Hkety representative of all drinking water supplies in the United States.
Therefore, it is strongly recommended that the model be further verified with a database that is
more representative of nationwide treatment plants. Nonetheless, this work is the most
complete verification to-date, and for that reason, is described here.
B2 VERIFICATION OF MODEL PREDICTIONS
B.2.1 Introduction
Verification was performed in October 1990 to determine the relative accuracy of
predictions not only for THM formation, but also for NOM surrogates (i.e. TOC, UV-254)
that impact the prediction of THM formation. It is important to note that the modeling
effort described here is based solely on the prediction of THMs, and does not include other
DBPs (e.g. haloacetic acids) that may be regulated as a part of the D/DBP Rule.
In the subsequent discussion, verification is a term used when comparing predicted
and measured results with fixed equations described in Appendix A. Calibration defines
modifying the predictive equations using measured data. When modeling, it is inappropriate
B-1
RECYCLED PAPER
-------
to verify results with measurements used to calibrate the model, because this results in an
"internal" data set that biases the comparison of predicted and measured values.
Predicted and measured results were compared using two data sets:
¦ a nationwide survey of 35 utilities (Montgomery and Metropolitan, 1989); and
¦ a survey of member agencies of the Metropolitan Water District of Southern
California (Malcolm Pirnie and Metropolitan, 1991).
The 35 Utility Survey was selected initially because these plants would have the most
complete data available for comparing measured and predicted values. The THM model
had not been compared at that time with an external database, and the 35 Utility Survey
provided the opportunity to identify limitations in model calculations and predictions on a
preliminary basis. Upon review of the preliminary verification results from the 35 utility
data set, model limitations were identified, and in some instances, relationships were
recalibrated with results from the 35 utility data set.
While the preliminary verification with the 35 utility data set was being performed,
data were being collected from Metropolitan's member agency treatment plants. As both
data and model limitations became evident through preliminary verification, the data
collection from the member agency plants was focused on ensuring that critical values (e.g.
unit process detention times) were collected. The model was then verified using the
member agency data.
This section summarizes:
¦ preliminary verification results from the 35 utility database;
¦ identification of model limitations; and
¦ verification results from the member agency treatment plants.
B J PRELIMINARY VERIFICATION RESULTS FROM THE 35 UTILITY DATABASE
BJ.l Database Compilation
The 35 Utility Survey was initially performed to determine the spatial and temporal
distribution of DBPs in treated drinking waters throughout the U.S. Treated water samples
were collected over four seasons (the first season, spring, did not include a complete
database for modeling purposes) in calendar years 1988 and 1989 and were analyzed for a
B - 2
-------
host of physical and chemical parameters, of greatest interest being the DBPs other than
THMs. THMs, haloaectonitrites (HANs), haloketones (HKs), chloropicrin, HAAs and
cyanogen chloride were measured for all of the utilities, and some additional DBPs that are
expected to occur based upon individual treatment processes employed also were measured
(e.g. chlorite and chlorate for utilities using chlorine dioxide).
In a meeting among Malcolm Pimie, Inc., Metropolitan staff, and James M.
Montgomery Consulting Engineers, Inc. (JMM, consultant to the D/DBP TAW), it was
determined that the 35 Utility Survey would be a reasonable external data set for initial
verification efforts. As such, compilation of additional data necessary to run the model
became a portion of the scope of work to be performed by JMM for the D/DBP TAW.
Malcolm Pirnie, Inc. provided JMM with the data input requirements for the model,
so that JMM could determine the extent of data gaps in the survey and update the database
as necessary. Over a three month period, JMM contacted the 35 utilities, updated the
database as required, and transmitted those results to Malcolm Pirnie, Inc.; Upon
completion, JMM and Malcolm Pirnie, Inc. reviewed the results and determined those
utilities which had data sets that would be most applicable for verification based upon:
¦ the type of plant operation (i.e. continuous or batch treatment);
¦ the nature of the treatment processes and the ability to simulate those
processes with the existing model (e.g. chlorine dioxide is currently not an
available unit process in the model, and the database for lime softening had not
been compiled at the time of model verification); and
¦ the degree of confidence in the information provided.
Of the original 35 utilities, 20 were selected for model verification. Table B-l
summarizes the coagulation type, primary disinfectant type and location, and residual
disinfectant type for the utilities used in the preliminary verification step. The table is
arranged in order by primary and secondary disinfectant. Plants using the same
disinfectants are listed in order of no treatment, coagulation with polymers, alum
coagulation, and ferric coagulation. Throughout the presentation of the preliminary
verification results, the plants are ordered in this manner.
B32 Model Assumptions
The raw water quality and operational data collected from the data compilation effort
were used as the basis for developing the input files for the individual utility treatment
B - 3
RECYCLED PATER
-------
TABLE B-1
LIST OF UTILmES IN 35-PLANT SURVEY
Utility
No.
Primary Disinfection
Residual
Disinfectant
Coagu
ation/So
ftening
Type
Location
Alum
Iron
Polymer
Lime (1)
1
CI2
Post-Air
CI2
24
CI2
Pre & Post
CI2
33
CI2
Only trtmnt
CI2
34
CI2
Only trtmnt
CI2
3
CI2
Post
CI2
X
8
CI2
Pre
CI2
X
15
CI2
Pre & Post
CI2
X
X
17
CI2
Pre & Post
CI2
X
X
22
CI2
Post
CI2
X
X
35
CI2
Pre
CI2
X
2
CI2
Post
NH2CI
X
5
CI2
Pre
NH2CI
X
6
CI2
Pre & Post
NH2CI
X
7
CI2
Pre
NH2CI
X
14
CI2
Pre & Post
NH2CI
X
18
CI2
Pre & Post
NH2CI
X
12
CI2
Pre
NH2CI
X
X
13
CI2
Pre & Post
NH2CI
X
X
4
NH2CI
Pre
NH2CI
X
10
NH2CI
Pre
NH2CI
X
Notes: 1. For some utilities lime is added for pH adjustment. TABB-1 WK1
06/07/92
-------
plants. As a part of the data gathering effort, baffling characteristics for clearwells were
requested. None of the utilities, however, had performed tracer tests to determine actual
detention times in the unit processes. Therefore, it was assumed that t^ : t^_^, for
basins and filters were 0.7 and 0.9, respectively. These ratios were based upon studies
performed to assess these ratios for specific baffling characteristics and basin configurations
(Hudson, 1981; Malcolm Pirnie, Inc. 1990) and are required when verifying actual plant
performance.
B33 Preliminary Verification Results
The preliminary verification was performed for complete water quality data sets for
summer, fall and winter seasons. Measured and predicted values were compared for :
TOC;
pH;
¦ chlorine residual; and
TTHMs.
BJ.4 Comparison of TOC Predictions
Table B-2 summarizes predicted and measured TOC removals for the utilities
evaluated. In general, the predicted and measured results compared well; predicted values
were within 10 percent of the measured values 70 percent of the time (n=60). It was noted
that the finished water TOC measured at Utility 33 in the summer and fall was up to 40
percent less than the raw water, although there is no coagulation at this Utility. The model
did not predict any removal of TOC under these conditions.
The good comparison between the predicted and measured values, to some extent,
could be expected. The 35 Utility Survey was a subset of the database used to prepare the
predictive equations for TOC removal (see Section A.4.1 in Appendix A) and, therefore, the
model was not verified with a completely external database in this preliminary step.
Verification with external data is described in Section B.4.2 and elsewhere (Harrington, et
aL, 1991).
B -4
RECYCLED PAPER
-------
TABLE B-2
COMPARISON OF MEASURED AND PREDICTED TOC;
35 UTILITY SURVEY
Utility
No.
TOTAL ORGANIC CARBON CONCENTRATION (mg/L)
Summer 1988
Fall 1988
Winter 1988
Raw
Water
Finished Water
Raw
Water
Finished Water
Raw
Water
Finished Water
Measured
Predicted
Measured
Predicted
Measured
Predicted
1
0.8
0.8
0.8
1.1
0.8
1.1
1.0
0.9
1.0
24
0.6
0.7
0.6
0.7
0.6
0.7
0.7
0.7
0.7
33
5.4
3.1
5.4
5.2
3.9
5.2
3.5
3.5
3.5
34
2.6
2.6
2.6
2.9
2.8
2.9
3.3
3.3
3.3
3
3.2
1.6
1.8
2.7
1.7
1.6
3.2
1.9
2.0
8
1.6
1.4
1.4
1.5
1.4
1.3
1.7
1.4
1.4
15
1.3
1.1
1.2
1.2
1.1
1.2
1.4
1.0
17
3.0
2.0
1.9
2.8
2.0
1.9
3.2
1.9
2.0
22
5.5
3.8
3.8
4.4
3.6
3.0
4.0
3.4
2.6
35
3.1
3.1
3.1
3.0
3.1
3.0
3.5
3.4
3.5
2
2.4
2.3
2.2
2.9
2.2
2.8
2.5
2.0
2.3
5
5.4
3.8
4.4
5.3
4.2
4.2
5.5
4.4
4.0
6
4.5
3.7
3.2
3.9
3.1
2.8
4.9
2.9
2.9
7
2.8
2.6
2.6
2.8
2.7
2.5
2.7
2.6
2.7
14
2.6
1.9
1.8
2.9
2.7
2.2
3.9
2.4
2.5
18
1.8
1.4
1.4
2.7
2.4
2.5
1.9
1.7
1.7
12
2.6
1.7
2.1
2.2
1.7
1.5
2.8
1.8
1.9
13
1.1
0.8
0.8
1.5
1.1
1.1
1.5
1.0
1.1
4
2.8
2.5
2.6
2.7
2.4
2.5
2.7
2.4
2.5
10
4.6
3.8
3.5
4.9
3.7
3.9
5.3
4.2
4.0
Notes: 1. Blank cells indicate missing data.
TABB-2 WK1
06/07/92
-------
B.3.5 Comparison of pH Predictions
Table B-3 compares measured and predicted values for pH over the 3 seasons. The
pH was not predicted reliably under the following circumstances:
¦ The raw water for Utility 3 has low alkalinity, and the model did not predict
the measured finished water pH of between 4.6 and 4.8. The finished water
sample was collected prior to pH adjustment before distribution at this utility
while the model was used to predict pH after this adjustment;
¦ For utilities with no treatment other than free chlorination (Utilities 1, 24, 33,
and 34), the extent of pH depression predicted by the model was dependent
upon the alkalinity and the chlorine dose. The measured values did not always
reflect this relationship. For example, the measured finished water pH for
Utility 33 was 6.8 (raw water pH 8.2), while the predicted value from the model
in this buffered water remained at 8.2. The example cited here indicates that
the predicted values from the model more closely approximate expectations
based upon water chemistry; and
¦ Finished water pH in plants that add lime (Utilities 12, 13, 15, 17, 22, and 35)
were not predicted as well as in plants that do not add lime. This is a direct
result of the preliminary nature of the lime softening unit process in the model,
at the time of verification and that recarbonation is not included in the current
model version. Also, the kinetics of CaO dissolution and C02 exchange are
slow when lime is added. Therefore, the predicted values for pH were
consistently lower than the measured values.
For the remaining utilities, the pH was generally predicted within 0.5 units. The
above factors, however, indicated that it was important to document exactly where pH
adjustment was performed in the plant, and that lime softening was a process that could not
be evaluated accurately at the time of model verification.
B3.6 Comparison of Measured and Predicted Chlorine Residual
Table B-4 summarizes the comparison of measured and predicted values for free and
total chlorine residual from the 35 Utility Survey. This verification work was performed
prior to the development of the chloramine decay equations shown in Section A.6. The
verification effort identified the following:
¦ free chlorine decay (hence, residual) was predicted better than chloramine
decay. In some instances, chloramine residual increased through the treatment
system, as a result of water quality parameters that were outside of the
boundary values used during the chloramine modeling effort (see Utility 7);
and
B - 5
RECYCLED PAPER
-------
TABLE B-3
COMPARISON OF MEASURED AND PREDICTED pH;
35 UTILITY SURVEY
Utility
No.
PH
Summer 1988
Fall 1988
Winter 1988
Raw
Water
Finished Water
Raw
Water
Finished Water
Raw
Water
Finished Water
Measured
Predicted
Measured
Predicted
Measured
Predicted
1
7.4
7.4
7.3
7.2
7.5
7.2
8.3
7.7
7.1
24
7.1
7.6
6.8
7.2
8.2
6.9
7.2
7.5
6.8
33
8.2
6.8
8.2
7.3
6.9
6.6
7.4
7.0
6.7
34
. 7.3
7.1
7.2
6.7
6.7
6.7
7.1
NA
6.9
3
6.3
4.6
6.7
6.7
4.8
7.4
6.5
4.8
7.4
8
7.7
8.2
9.0
7.6
8.5
9.1
7.9
9.0
9.3
15
9.0
9.0
7.2
9.0
9.1
7.7
9.1
17
7.0
8.7
4.7
6.5
8.7
5.8
6.9
9.1
5.3
22
7.7
8.7
6.9
7.8
8.5
7.0
7.5
8.3
6.8
35
7.0
8.0
6.9
7.2
8.5
7.1
7.3
8.6
7.1
2
8.6
7.5
7.3
8.6
7.6
7.3
8.4
7.6
7.4
5
8.2
8.1
8.3
8.1
8.2
8.3
6.9
8.1
6.9
6
7.6
8.1
7.8
7.2
7.0
7.5
7.2
6.9
7.3
7
8.2
8.2
8.6
8.3
8.0
8.7
8.4
8.2
8.4
14
7.8
7.2
6.8
7.8
7.2
7.2
7.8
7.1
6.8
18
7.9
7.2
7.3
8.4
7.9
7.1
7.4
7.8
7.3
12
8.2
8.4
7.5
8.1
8.6
8.1
8.0
8.5
8.6
13
7.2
8.6
7.4
7.1
7.1
7.5
7.6
8.9
7.8
4
8.4
8.5
7.9
7.8
8.5
8.3
8.0
8.4
8.5
10
7.9
7.3
7.4
8.3
7.2
7.2
8.1
NA
7.7
Notes: 1. Blank cells indicate missing values. tabb-swki
-------
TABLE B-4
COMPARISON OF MEASURED AND PREDICTED CHLORINE RESIDUAL;
35 UTILITY SURVEY
Utility
No.
Summer 1988
Fall 1988
Winter 1988
Measured
Predicted
Measured
Predicted
Measured
Predicted
Free
Total
Free
Total
Free
Total
Free
Total
Free
Total
Free
Total
1
1.0
1.1
0.8
0.8
0.8
0.9
0.9
0.9
1.2
1.3
1.2
1.2
24
0.6
0.7
1.9
1.9
0.8
0.8
1.3
1.3
0.6
0.7
1.7
1.7
33
3.4
3.6
3.7
3.7
2.7
3.1
2.1
2.1
2.5
2.8
2.4
2.4
34
0.7
0.9
1.3
1.3
0.6
0.7
1.0
1.0
0.7
0.7
0.2
0.2
3
0.7
1.7
2.2
2.2
0.6
1.3
1.2
1.2
0.6
1.1
1.2
1.2
8
0.8
1.3
1.3
0.8
0.9
1.1
1.1
0.7
0.9
1.2
1.2
15
0.3
0.3
0.3
0.3
0.3
0.4
0.4
0.4
0.4
17
1.0
1.0
4.1
4.1
1.0
1.0
1.0
1.0
1.1
1.1
1.3
1.3
22
0.6
2.2
2.2
0.4
0.8
0.8
0.3
1.1
1.1
35
1.1
1.2
1.3
1.3
1.2
1.3
0.9
0.9
1.1
1.0
1.0
2
0.0
1.9
0.0
2.3
0.0
2.0
0.0
1.6
0.0
2.2
0.0
1.8
5
0.2
2.5
0.0
3.7
0.5
2.5
0.0
3.6
1.2
2.5
0.0
3.6
6
1.5
3.0
0.5
0.5
0.9
1.3
0.0
18.1
0.6
1.0
0.0
3.2
7
0.0
1.5
0.0
19.5
0.0
1.5
0.0
24.8
0.0
1.5
0.0
19.8
14
0.1
1.4
0.2
0.2
0.2
1.8
0.0
3.6
0.2
1.6
0.0
3.8
18
0.0
1.1
0.0
1.0
0.0
1.3
0.0
1.2
0.1
1.3
0.0
1.0
12
0.1
1.1
0.0
0.6
0.1
0.7
0.0
0.2
0.2
0.9
0.0
0.0
13
0.5
0.5
1.2
1.2
0.3
0.4
1.2
1.2
0.3
0.3
1.4
1.4
4
0.0
1.5
0.0
2.0
0.0
1.5
0.0
2.0
0.0
1.6
0.0
2.7
10
0.1
3.3
0.0
6.2
0.1
3.3
0.0
6.5
0.1
4.5
0.0
6.5
Notes: 1. Blank cells indicate missing values.
-------
¦ upon review with JMM, it appeared that there may have been some confusion
in the measured values reported for some utilities.
The THM formation equations are based upon chlorine dose, not residual. From a
modeling standpoint, however, the prediction of chlorine and chloramine decay are
important. A desired disinfectant residual in the distribution system is selected during
alternative treatment evaluation, and the decay determines the dosage at the plant. This
dosage impacts THM formation.
BJ.7 Comparison of Measured and Predicted THMs
Table B-5 summarizes the measured and predicted THM values for individual
compounds, as well as the TTHMs, in the summer sampling effort. Because THM
formation increases with temperature, the summer results are expected to exhibit the highest
THM concentrations. Higher concentrations of THMs during summer are also expected to
result in larger absolute differences between the measured and the predicted concentrations
of THMs. The following are noted:
¦ in general, predicted values for TTHMs are lower than measured values. This
may result from the difficulties that arise in predicting actual free chlorine
contact times in treatment processes, which strongly influence the predicted
THM formation. JMM indicated that accessing accurate information on
flowrates at the time of the sampling was difficult, and this parameter, along
with the actualitheoretical detention time ratios, strongly impact the prediction
of THMs; and
¦ predictions for the brominated species become poorer when the bromide
concentration in the raw water increases and the extent of bromine
incorporation increases .
A review of the sensitivity of THM predictions to variations in the independent
variables indicated that the estimated detention times have a significant effect on predicted
THMs. Because the water quality data were collected over four quarters in 1988, JMM
attempted to access accurate operational information for the dates when the samples were
collected. In a meeting between Metropolitan and Malcolm Pirnie, Inc. it was decided that
the preliminary verification had served its purpose and that it was not necessary for JMM
to attempt to refine the operational input values. Model limitations and sensitivities had
been determined, and a more accurate database incorporating water quality data and
B - 6
RECYCLED PAPER
-------
TABLE B-5
COMPARISON OF MEASURED AND PREDICTED THM;
35 UTILITY SURVEY: SUMMER 1988
Utility
No.
THM (ug/L)
CHCI3
CH
CI2Br
CH
CIBr2
C
HBr3
Total
Meas.
Pred.
Meas.
Pred.
Meas.
Pred.
Meas.
Pred.
Meas.
Pred.
1
0.9
2.2
2.0
3.4
3.1
1.7
1.9
0.8
7.9
8.1
24
0.4
0.5
4.2
34.8
39.9
0.0
33
41.8
15.4
2.2
0.8
0.0
0.0
0.0
0.0
44.0
16.2
34
36.0
25.0
9.0
2.1
1.2
0.0
0.1
0.0
46.3
27.1
3
4.1
7.7
1.7
1.5
0.3
0.1
0.0
0.0
6.1
9.3
8
18.1
6.5
7.0
3.3
2.0
0.5
0.1
0.1
27.2
10.4
15
25.0
1.9
1.9
0.6
0.1
0.0
0.0
0.0
27.0
2.5
17
85.0
15.3
5.0
1.9
0.2
0.0
0.0
0.0
90.2
17.2
20
64.2
16.2
10.7
3.4
1.1
0.1
0.0
0.0
76.0
19.7
22
56.1
29.2
4.2
2.5
0.2
0.0
0.1
0.0
60.6
31.7
35
34.7
15.6
2.9
0.9
0.1
0.0
0.0
0.0
37.7
16.5
2
66.7
81.0
18.1
29.8
4.5
7.2
0.3
2.5
89.6
120.5
5
15.4
17.5
15.0
8.5
10.2
3.2
1.5
0.1
42.1
29.3
6
70.3
48.3
22.0
15.0
5.2
2.3
0.2
0.2
97.7
65.8
7
12.0
15.2
20.5
16.3
24.3
13.4
6.2
1.2
63.0
46.1
14
13.2
10.3
36.7
31.3
48.8
75.1
15.7
12.0
114.4
128.7
18
5.1
0.9
1.6
0.3
0.2
0.0
0.7
0.0
7.6
1.2
12
4.7
12.6
28.3
25.8
71.4
0.0
13
20.1
2.0
3.1
0.9
0.3
0.1
0.0
0.0
23.5
3.0
4
2.7
5.7
3.5
8.9
2.4
14.8
0.5
1.5
9.1
30.9
10
1.0
3.8
8.6
29.8
43.2
0.0
Notes: 1. Blank cells indicate missing values. tabb-s wki
06/07192
-------
operational practices needed to be developed for verification. The ongoing member agency
data collection effort provided potential for this database.
The following subsections describe first a summary of model limitations, and then
verification with the member agency database.
B.4 VERIFICATION RESULTS FROM THE MEMBER AGENCY DATABASE
B.4.1 Database Compilation
After the model limitations were understood and addressed to the extent permitted
by the scope of work, the improved model was tested against the member agency database.
The experience gathered from the preliminary verification with the 35 Utility Survey and
a better understanding of the model data needs at the time of member agency data
collection helped create a more complete database. This effort included visits to some of
the member agency plants to gather supplemental samples. For these reasons, it was hoped
that the database enhancement, along with the improved model from the preliminary
verification with the 35 Utility Survey, would improve the overall model prediction. A
complete set of input parameters were available for 19 of the member agency plants.
As described previously, the contact time with free chlorine has a strong impact on
THM prediction. Malcolm Pirnie, Inc. had performed tracer studies at two of the member
agency plants as part of another study (Malcolm Pirnie, Inc. 1990). Because many of the
member agencies had similar basin designs, these previous studies were useful in the overall
evaluation.
B.42 Verification Results
Comparisons were made for the following parameters:
¦ TOC concentration;
¦ UV absorbance; and
¦ TTHM concentrations.
Comparison of Measured and Predicted TOC
Figure B-l illustrates measured and predicted values for TOC in treated water from
the 19 member agency plants evaluated, for both alum and ferric chloride coagulation.
There are two relationships presented on this figure:
B-7
RECYCLED PAPER
-------
FIGURE B-1
Statistical Comparison of Measured and Predicted TOO Values
Metropolian Member Agency Plants
25
-20 to-15 -15 to-10 -10 to-5 -5 toO 0to5 5 to 10 10 to 15
Deviation (Percent)
100
^ 80
#
c
V
3
CT
©
U.
P
60
« 40
3
E
3
o
20
11
13
15
17
19
Deviation (Absolute Value, Percent)
-------
¦ a cumulative frequency showing the absolute value of the percent deviation
between measured and predicted values; and
¦ a histogram illustrating the relative frequency of percent deviation between
measured and predicted values. In this relationship, negative values indicate
that the model predicted lower TOC values than were measured (overpredicted
removals), and positive values indicate that the predicted TOC concentrations
were higher than measured (underpredicted removals).
These results indicate that approximately 65 percent of the predictions were within
10 percent of the measured values. All of the predictions were within 19 percent of the
measured values. For the member agency plants evaluated, the model appears to
overpredict TOC removal (70 percent showed higher removal and 30 percent showed lower
removal than observed). The central tendency was to underpredict finished water TOC by
5 to 10 percent.
From a standpoint of predicting THM formation, overpredicting TOC removal will
translate to lower predicted TTHMs. Considering the variation in treatment processes, the
use of two different coagulants employing separate equations for predicting removal, and
the analytical error associated with sample analyses, the predicted values compare well with
measured values.
Comparison of Measured and Predicted UV Absorbance
Figure B-2 illustrates the comparison for measured and predicted UV-absorbance at
254 nm. In this relationship, UV-254 data were available for only 19 of the 32 member
agency plants. This comparison indicates:
¦ approximately 50 percent of the predicted values deviate less than 20 percent
from the measured values;
¦ the model tends to predict higher UV-254 values (i.e. underpredicts removals),
as evidenced by the relative frequency of positive percent deviation in the
histogram (56 percent showed lower removal and 44 percent showed higher
removal than measured); and
¦ the central tendency of the model was to overpredict finished water UV-254 by
10 percent.
The spread of the deviation between predicted and measured values is greater than
for the TOC prediction, in this instance towards underprediction of UV-254 reduction. One
B - 8
RECYCLED PAPER
-------
FIGURE B-2
Statistical Comparison of Measured and Predicted UV-254 Values
Metropolitan Member Agency Plants
25
Deviation (Percent)
100
80
&
§
| 60
u.
I
t 40
E
3
o
20
2.5 7.5 12.5 17.5 22.5 27.5 32.5 37.5 42.5 >45
Deviation (Absolute Value, Percent)
I
-------
factor that might contribute to the larger spread in UV-254 predictions is that chlorination
is known to impact UV-254 values without any change in TOC concentration.
The underprediction of reduction in UV-254 corresponds to a relatively higher
prediction for TTHMs. This offsets the trend discussed previously for overprediction of
TOC removal.
Comparison of Measured and Predicted Values for TlHM
Measured and predicted values of TlHM were compared for the following cases:
¦ simulated distribution system for 19 member agency data set; and
¦ a modified data set of SDS-THM for plants treating raw waters with < 0.30
mg/L bromide (17 of the 19 plants).
Figure 6-3 illustrates a comparison of measured and predicted simulated distribution
system (SDS) values for TTHM. The relationship indicates that:
¦ 50 percent of the predicted values deviate 20 percent or less from the
measured values;
¦ 80 percent of the predicted values deviate 40 percent or less from the
measured values;
¦ the remaining 20 percent of the values deviate between 40 and 85 percent of
the measured values; and
¦ there is a general trend to underpredict TTHM formation. This
underprediction could not be traced to any one dependent variable, such as
consistently predicting lower pH values than measured (thereby underpredicting
TTHMs).
¦ the central tendency was to underpredict TTHMs by 20 to 30 percent.
Considering the compounding errors in predicting dependent variables such as TOC
and pH, in this level of accuracy may be expected. From the verification work performed
previously, however, the member agency data set was reviewed to determine how the
prediction might be improved. Figure B-4 illustrates a comparison of the cumulative
frequency of deviation in predicted and measured values for two data subsets:
¦ all SDS observations for TTHM; and
B -9
RECYCLED PAPER
-------
FIGURE B-3
Statistical Comparison of Measured and Predicted SDS-THM
Metropolian Member Agency Plants
Predlcted-THM • Meaaured-THM
Deviation ° xioo
Measured-THM
LT-50 -80to-40 -40to-30 -30to-20 -20to-10 -lOtoO OtolO 10to20 20to30 30to40 40to50 QT50
Deviation (Percent)
100
80
&
?
c
0)
3
l
3
E
3
o
60 -
40
20
15 25
35 45 55 65 75
Deviation (Absolute Value, Percent)
85 95 GT100
-------
FIGURE B-4
Evaluation of SDS-THM Predictions
Effect of Bromide on Prediction
100
_ 80
£
» 60
f
LL
•| 40
CO
3
E
« 20
c
0
3
CT
©
l
3
E
3
o
AiiSDS
Observations
100
^ 80
Bromide
< 0.3mg/L
60
40
20
S 15 25
Deviation ¦
Prodlcted-THM - Moasured-THM
Measured-THM
X100
15 25 35 45 55 65 75
Deviation (Absolute Value, Percent)
85
95 GT100
35 45 55 65 75
Deviation (Absolute Value, Percent)
85
95 GT100
-------
SDS-TTHM for the member agency plants treating raw water with bromide
concentrations < 0.30 mg/L.
The results from this comparison illustrate improvement in prediction. For the data
set with bromide concentrations < 0.30 mg/L bromide:
¦ approximately 50 percent of the predicted values deviate less than 20 percent
from the measured values;
¦ 80 percent of the predicted values deviate less than 35 percent of the measured
values; and
¦ all of the values deviate less than approximately 50 percent of the
measurements, compared to 85 percent of the predictions within 50 percent
when all data were included.
This indicates that the bromide containing waters (Br'> 0.30mg/L) were responsible for the
most significant deviation in TTHM prediction.
B.4J Summary of Observations
The degree of prediction outlined above represents the best that can be performed
with the model at this time. Model predictions for SDS-THM were expected to improve
when using the Metropolitan member agency database as compared to the 35 Utility Survey,
because of greater confidence in the actual disinfectant contact times for the member agency
plants. However, the member agency database is a small subset of utility water qualities
throughout the country, and does not necessarily represent the central tendency for all
waters in the United States.
At this time, the major limitation is in the prediction of brominated THM formation
for high bromide waters. This can be attributed to an understanding that the THM
formation equations are based upon a database (Amy et al, 1987) that contains:
¦ limited values for higher bromide waters. Bromide spiking was performed, but
the impact of higher bromide on increased formation of brominated species
cannot be predicted with certainty; and
¦ THM measurements performed at a Cl2:TOC ratio of 3 to 1. Recent studies
(Krasner, 1991; Symons, 1991) indicate that at lower Cl2:TOC ratios (e.g. 1 to
1.5), which are more typical of plant operation and provide a chlorine residual
of 0.5 to 1.0 mg/L after a specified holding time (24 to 72 hours), bromine
incorporation increases.
B -10
RECYCLED PAPER
-------
Together, these factors indicate that the model may underpredict SDS-THM
measurements for member agency plants with bromide concentrations exceeding 0.3 mg/L
and Cl2:TOC dosages between 1.0 and 1.5.
B - 11
RECYCLED PAPER
-------
APPENDIX C
SAMPLE SURVEY QUESTIONNAIRE
-------
TABLE C-1
SAMPLE SURVEY QUESTIONNAIRE
Utility Name:
Utility Address:
Contact person:
Phone #:
FAX#:
Water Treatment Plant name and address:
Source of raw water:
Yes/No % of Total
Supply
Groundwater
Lake/Reservoir
Purchased (from other utility):
Total = 100%
TABC-1.MC
06/07/93
Design hydraulic capacity:
Average Flow:
Range of flow rates:
Peak Hourly Flow:
RECYCLED PAPER
-------
TABLE C-1 (continued)
Schematic of the Plant Processes Including the Locations of Chemical Addition
Please attach a schematic drawing of your plant.
Unit Process Basins
Dimensions (feet) Baffling Characteristics fves/no)
Width or
Unit Process Length Diameter Depth Inlet Outlet Intermediate
Rapid Mix
Flocculation
Sedimentation
Filtration
Clear Well
On-Site Storage
Please note any other basins or large pipelines/conduits in the treatment
process train or any other dimensions which will affect water detention time:
Have any tracer studies conaucted to determine the actual detention times?
If yes, please attach or explain results.
Please state any nonconventional treatment process operations,
such as operating the plant in a batch or semi-batch mode, non-
continuous operations, etc..
RECYCLED PAPES
-------
TABLE C-1 (continued)
Chemical Addition
Annual
Range of
Chemical
Average Dose
Dosages (ppm)
(ppm)
Minimum
Average
Maximum
Pretreatment
Rapid mix
Flocculation
Sedimentation
Filtration
Disinfection
Softening
Recarbonation
Other
-------
TABLE C-1 (continued)
Advanced Unit Treatment Processes
Does the treatment plant use:
Yes/No
If yes:
Granular activated carbon?
Empty bed contact time =
Regeneration frequency =
Aeration or Air Stripping?
Air-to-water ratio =
Diffused, mechanical, or
packed tower?
Ion Exchange?
Media type is:
Contact time =
Regeneration frequency =
Membranes?
Membrane type =
Moleculer weight cutoff =
Recovery =
Water Quality
Raw Water
Finished Water
Annual
Average
Range
Annual
Average
Range
PH
Alkalinity (mg/L as CaC03)
Turbidity (NTU)
Color (Standard Color Units)
Total Organic Carbon (mg/L, ppm)
Temperature (C)
UV-Absorbance (1/cm)
Bromide (mg/L or ppm)
Chloride (mg/L or ppm)
Calcium Hardness (mg/L as CaC03)
Total Hardness (mg/L as CaC03)
Ammonia (mg/L)
RECYCLED PAPER
-------
TABLE C-1 (continued)
Disinfection and Disinfection Bv-Products
Disinfectant Residual in the Treatment Plant
After
Rapid Mix
After
Settling
After
Filters
Filtered
Water
Free Chlorine (ppm)
Total Chlorine (ppm)
(Chloramines)
Chlorine Dioxide (ppm)
Ozone (ppm)
Disinfectant Residual in the Distribution System
Free Chlorine (ppm)
Combined Chlorine (ppm)
Chlorine Dioxide (ppm)
Trihalomethanes
Total (ppb)
Chloroform (ppb)
Dichlorobromomethane (ppb)
Dibromochloromethane (ppb)
Bromoform (ppb)
Haloacetic Acids
Total (ppb)
Monochloroacetic Acid (ppb)
Dichloroacetic Acid (ppb)
Trichloroacetic Acid (ppb)
Monobromoacetic Acid (ppb)
Dibromoacetic Acid (ppb)
Other DBPs (if any measured)
Finished Water
Annual Range
Average
Distribution System
Annual
Average
Range
Estimated Detention Time in
the Distribution System (days)
------- |