EPA814/B-96-003
April 1996
ICR Manual for Bench- and
Pilot-Scale Treatment Studies
Technical Support Division
Office of Ground Water and Drinking Water
U.S. Environmental Protection Agency
Cincinnati, OH 45268
-------
Foreword
The Information Collection Rule (ICR) for Public Water Systems (Subpart M of the
National Primary Drinking Water Regulations, § 141.141(e)) requires public water systems
that meet certain applicability criteria to conduct disinfection byproduct (DBF) precursor
removal studies, referred to as treatment studies. These treatment studies are intended to
provide cost and performance data on granular activated carbon (GAG) and membrane
processes for meeting the DBF regulations. The purpose of the "ICR Manual for Bench- and
Pilot-Scale Treatment Studies" is to provide the information necessary to conduct these
studies.
This manual is referenced in the final ICR regulation and contains the specific
requirements for conducting treatment studies. The document is divided into three parts.
Part 1 summarizes the ICR regulation, including criteria to determine applicability to the
treatment study requirement and includes applications for treatment study options, and
provides general guidelines for conducting the treatment studies.
* Part 2 details the requirements of the bench- and pilot-scale GAC studies.
Part 3 details the requirements of the bench- and pilot-scale membrane studies.
Each of these three Parts is a self contained document, and all references to specific table
numbers, figure numbers or section numbers refer to the tables, figures and sections within the
Part of the document in which the reference is made unless explicitly stated otherwise. For
example, a reference to Table 3-1 hi Part 1 refers to Table 3-1 at the end of Part 1, while a
reference to Table 3-1 hi Part 3 refers to Table 3-1 at the end of Part 3.
DISCLAIMER
Mention of trade names or commercial products does not constitute endorsement or
recommendation for use.
-------
Acknowledgements
Jeffery Q. Adams; USEPA, Office of Research and Development, Water Supply and Water
Resources Division
Harish Arora, Ph.D.; American Water Works Service Company
Blake Atkins; USEPA, Region 6
Phyllis A. Branson; USEPA, Office of Ground Water and Drinking Water, Technical Support
Division
Shankar Chellam; Montgomery Watson
Shiao-Shing Chen; University of Central Florida
Sarah C. Clark; City of Austin, Water and Wastewater Utility
David A. Cornwell, Ph.D.; Environmental Engineering & Technology, Inc.
Laura Cummings; Montgomery Watson
Michael D. Cummins; USEPA, Office of Ground Water and Drinking Water, Technical
Support Division
Thomas R. Grubbs; USEPA, Office of Ground Water and Drinking Water, Drinking Water
Standards Division
Thomas Holdsworth; USEPA, Office of Research and Development, NRMRL
Joseph Jacangelo, Ph.D.; Montgomery Watson
Kay Koppmeier; USEPA, Office of Ground Water and Drinking Water, Technical Support
Division
Stuart W. Krasner; Metropolitan Water District of Southern California
loop C. Kraithof, Ph.D.; KIWA,N.V.
Sun Liang, Ph.D.; Metropolitan Water District of Southern California
Richard J. Lieberman; USEPA, Office of Ground Water and Drinking Water, Technical
Support Division
James C. Lozier; CH2M Hill
Wendy Marshall; USEPA, Region 10
Roy Martinez; AWWARF
Michael J. McGuire, Ph.D.; McGuire Environmental Consultants, Inc.
Edward G. Means, III; Metropolitan Water District of Southern California
Richard J. Miltner; USEPA, Office of Research and Development, Water Supply and Water
Resources Division
Eva C. Nieminski, Ph.D.; State of Utah, Department of Environmental Quality
Douglas M. Owen; Malcolm Pirnie, Inc.
Franklyn Pogge; Water and Pollution Control Department, City of Kansas City, Missouri
Brian L. Ramaley; City of Newport News, Department of Public Utilities
Alan Roberson; American Water Works Association
Michael R. Schock; USEPA, Office of Research and Development, Water Supply and Water
Resources Division
Thomas Speth; USEPA, Office of Research and Development, Water Supply and Water
Resources Division
James B. Walasek; USEPA, Office of Ground Water and Drinking Water, Technical Support
Division
Jeannie K. Wiginton; City of Austin, Water and Wastewater Utility
m
-------
-------
Parti
General Requirements And Guidelines For Conducting Precursor
Removal Studies
James J. Westrick
Director, Technical Support Division
Office of Ground Water and Drinking Water
United States Environmental Protection Agency
Steven C. Allgeier
Research Fellow
Oak Ridge Institute for Science and Education
Sponsored by
Technical Support Division
Office of Ground Water and Drinking Water
United States Environmental Protection Agency
-------
-------
Contents
List of Tables 1-iv
1.0 Introduction 1-1
2.0 Background 1-3
3.0 ICR Requirements 1-5
3.1 Applicability Monitoring . 1-5
3.2 Criteria Under Which No Treatment Studies Are Required 1-6
3.3 Criteria For Determining Individual Treatment Study Requirements 1-8
3.4 Joint Treatment Studies 1-9
3.5 Criteria Under Which An Alternative To Conducting A Treatment Study Is
Allowed 1-10
3.6 Criteria For A Common Water Resource 1-11
4.0 General Guidelines 1-13
4.1 Precursor Removal Technology 1-13
4.2 Influent To The Treatment Studies 1-14
4.3 GAC Studies 1-15
4.4 Membrane Studies 1-16
4.5 Analytical Methods 1-17
4.6 Chlorination Procedures 1-17
5.0 Applications And Reports 1-19
5.1 Application Materials And Dealines 1-19
5.2 Treatment Study Option Review Criteria 1-20
5.2.1 Individual Studies 1-20
5.2.2 Treatment Study Avoidance 1-20
5.2.3 Single Study For Multiple Plants Owned By A Single PWS 1-21
5.2.4 Joint Treatment Studies 1-21
5.2.5 Treatment Study Buyout 1-22
5.2.6 Grandfathering Previous Treatment Studies 1-23
5.3 Review Criteria For A Common Source Designation 1-23
5.3.1 Rivers And Streams 1-24
5.3.2 Lakes, Reservoirs And Ground Waters Under The Direct Influence 1-24
5.3.3 Ground Water Aquifers 1-24
5.4 Review Criteria For Study Concept Forms 1-24
5.5 Treatment Study Progress Reports 1-25
5.6 Final Treatment Study Report 1-25
l-iii
-------
List of Tables
Table 3-1 Treatment Study Applicability Monitoring Requirements For Plants
Table 3-2 Monitoring Required Across Full-Scale GAC Or Membrane Processes
Table 3-3 Joint Study Requirement For Plants With A Population Served < 500,000
Table 3-4 Joint Study Requirement For Plants With A Population Served ;>500,000
Table 4-1 Factors For Consideration In Selecting Treatment Study Technology
Table 4-2 Summary Of Treatment Study Requirements For GAC And Membrane,
Pilot-Scale And Bench-Scale Systems
Table 5-1 General Public Water System And Plant Information
Table 5-2 Treatment Study Applicability Data Reporting Form
Table 5-3 Information Required With ICR Treatment Study Option Applications
Table 5-4 Important Deadlines Related To The ICR Treatment Studies
Table 5-5 Treatment Study Avoidance Application
Table 5-6 Application For Multiple Plants Owned By A Single PWS And Operating
On A Common Source To Conduct A Single Treatment Study
Table 5-7 Joint Treatment Study Application
Table 5-8 Treatment Study Buyout Application
Table 5-9 Grandfathered Treatment Study Application
Table 5-10 Common Source Designation Application
Table 5-11 Study Concept Form
Table 5-12 Treatment Study Progress Report
1-iv
-------
1.0 Introduction
This part of the manual contains information on the ICR requirements for conducting
treatment studies, including information on how plants may avoid conducting studies, conduct
joint studies, or participate in funding a cooperative research program. General guidelines and
requirements for the treatment studies and information common to both GAC and membrane
studies will be provided in Section 4.0. Section 5.0 outlines the application process for the
various treatment study options and provides the necessary application forms.
This manual has been incorporated by reference in the final ICR regulation, and the
detailed requirements for conducting studies are found hi this manual rather than in the Federal
Register Notice. These detailed requirements are enforceable, and public water systems must
comply; however, because of the many site-specific considerations involved in conducting
appropriate treatment studies, some flexibility is provided hi the protocols.
l-l
-------
-------
2.0 Background
The Information Collection Rule (ICR) was a result of the negotiated rule-making process
(reg-neg) which was used to reach agreement on a regulatory course of action for control of
microbial contaminants and disinfection byproducts (DBFs). The committee which negotiated
the rule-making strategy decided that it is necessary to obtain information on the potential
impact of future requirements to reduce the level of disinfection byproduct precursors (i.e.,
natural organic material measured as total organic carbon) on the cost of drinking water
treatment in the U.S. In order to accomplish this objective, a requirement to conduct
treatment studies is included in the ICR. The committee also felt that the treatment study
requirement would accelerate local acquisition of information to assess the feasibility of
advanced treatment to reduce the levels of DBF precursors.
1-3
-------
-------
3.0 ICR Requirements
3.1 Applicability Monitoring
All public water systems (PWS) which use surface water or ground water under the direct
influence of surface water and serve more than 100,000 persons, and PWSs which use only
ground water and serve more than 50,000 persons shall conduct treatment study applicability
monitoring at the following plants:
All treatment plants which serve over 100,000 persons (categories A & B in Tables 1 and
2 of § 141.141(b) of the ICR).
The largest treatment plant owned by a PWS which serves over 100,000 persons if no
single plant operated by the PWS serves over 100,000 persons (categories C & D in Tables
1 and 2 of § 141.141(b) of the ICR).
The largest ground water plant owned by a PWS, with a PWS ground water population
served of 50,000 to 99,999 persons (category G hi Tables 1 and 2 of § 141.141(b) of the
ICR).
The population served includes both wholesale and retail populations. Appendix A to
§141.141(a) of the ICR describes the procedure to calculate the populations served by PWSs
and treatment plants.
Ground water plants include multiple wells hi the same aquifer with no other treatment
than chlorination, as described under § 141.141(a)(3) of the ICR, as well as central treatment
plants such as softening plants treating water from one or more wells.
Plants that use only purchased finished water that is re-disinfected prior to distribution do
not have to conduct treatment study applicability monitoring and are exempt from the
treatment study requirement.
Treatment study applicability monitoring is conducted to determine: (1) if the treatment
plant precursor levels are low enough to avoid the treatment study requirement, and/or (2) if
two or more treatment plants qualify for a common source designation as described hi Section
3.6. The treatment study applicability monitoring requirements summarized in Table 3-1 are:
Twelve (12) consecutive monthly total organic carbon (TOC) samples taken from the plant
influent for surface water plants or, twelve (12) consecutive monthly TOC samples taken
from the finished water for plants treating only ground water. Treatment plant influent is
defined as water that represents the water quality challenge to a particular plant, and
finished water is defined as water that does not undergo further treatment by a plant other
than maintenance of a disinfectant residual.
Twelve (12) consecutive monthly total organic halide (TOX) samples evaluated under
uniform formation conditions (UFCTOX) may be required for a common source
designation. UFCTOX samples are taken from the plant influent for surface water plants
and from the finished water for plants treating only ground water.
1-5
-------
* If a surface water or ground water plant is using only chlorine as the primary and residual
disinfectant, quarterly trihalomethane (THM4) and haloacetic acid (HAAS) distribution
system samples can be measured to determine treatment study applicability. Each quarter,
samples must be collected from the four following distribution system points: one sample
point representative of the maximum residence time for the treatment plant and three
sample locations representative of the average residence time hi the distribution system for
the treatment plant.
All treatment study applicability monitoring shall be conducted using the methods and the
mandatory quality control procedures contained hi the "DBP/ICR Analytical Methods Manual"
(EPA 814-B-96-002). Additionally, the TOC analyses shall be conducted by laboratories that
have received approval from EPA to perform TOC analysis for compliance with this rule.
Although not a requirement, it is recommended that EPA approved laboratories also analyze
the UFCTOX, THM4 and HAAS samples collected during treatment study applicability
monitoring.
This treatment study applicability monitoring must begin no later than three (3) months
after the date of publication of the final rule hi the Federal Register. Specifically, monthly
sampling of TOC, and UFCTOX if required for a common source designation, must begin no
later than three (3) months after the date of publication of the ICR. If distribution system DBP
samples are to be used to demonstrate treatment study applicability, then quarterly sampling of
THM4 and HAAS distribution system samples must begin between three (3) and six (6)
months after the date of publication of the ICR. These results must be submitted to the ICR
Treatment Studies Coordinator, at the address listed hi Section 5.0, no later than seventeen
months after publication of the final rule in the Federal Register.
3.2 Criteria Under Which No Treatment Studies Are Required
TOC: Treatment plants using surface water are excused from conducting treatment studies
if they do not exceed an annual average TOC of 4.0 mg/L hi the treatment plant influent,
based on the twelve monthly TOC samples. Treatment plants using only ground water not
under the direct influence of surface water are excused from conducting treatment studies if
they do not exceed an annual average TOC of 2.0 mg/L in the finished water, based on the
twelve montihly TOC samples.
DBP: Treatment plants that use only chlorine as both the primary and residual disinfectant
and have, as an annual average of four quarterly averages, levels of less than 40 /^g/L for
THM4 and less than 30 //g/L for HAAS need not conduct treatment studies. Quarterly
averages are the arithmetic averages of the four distribution system samples (i.e., one sample
point representative of the maximum residence tune for the treatment plant and three sample
locations representative of the average residence time in the distribution system for the
treatment plant).
1-6
-------
Full-scale GAC Or Membrane Treatment
For a treatment plant that already uses full-scale GAC or membrane technology capable of
achieving precursor removal, a PWS shall conduct the monitoring listed in Table 3-2 and
submit full-scale plant data, as per § 141.142 of the ICR, ensuring that the GAC or membrane
processes are included in the process train being monitored. GAC capable of removing
precursors is defined hi the regulation as GAC with an empty bed contact tune (EBCT) of 15
minutes or greater with a time between carbon reactivation or replacement not more than nine
months. In special cases, GAC plants which operate outside the above specifications of EBCT
and reactivation frequency may submit a request to avoid treatment studies, including data
demonstrating effective precursor removal. The purpose of this requirement is to not excuse
plants from conducting treatment studies which technically have "full-scale GAC treatment",
but may have very limited precursor removal. For treatment plants to be considered to have
membrane technology to achieve precursor removal, the plant shall use nanofiltration or
reverse osmosis membranes. These types of membranes have been demonstrated to be
effective for the removal of precursor materials. Membrane plants using technologies other
than nanofiltration or reverse osmosis may submit a request to avoid treatment studies,
including data demonstrating effective precursor removal.
Grandfathered Studies
A PWS that has conducted precursor removal studies using granular activated carbon or
membrane technology (nanofiltration or reverse osmosis) may use the results of those studies
in fulfillment of the ICR treatment study requirement if all the following conditions are met:
The study was conducted using analytical methods described in "DBP/ICR Analytical
Methods Manual," EPA 814-B-96-002.
The study was conducted using a protocol similar to one of those specified in this manual.
The study data meets the requirements of the ICR and is supplied to EPA.
Request To Avoid Treatment Studies
A PWS that believes it qualifies to avoid the treatment study requirement under any of the
conditions listed in Section 3.2 shall submit an application according to the instructions listed
in Section 5.2.2 (and Section 5.2.6 for grandfathered studies). The PWS will be notified in
writing of EPA's decision regarding these applications.
TOC - An application to avoid treatment studies on the basis of TOC data shall be
requested no later than eighteen (18) months after publication of the final rule hi the Federal
Register, showing the annual average TOC concentration less than the treatment studies trigger
levels described in this section.
DBP - If the TOC trigger is exceeded but the distribution system DBPs are less than the
levels specified in this section, the PWS shall submit an application on that basis as soon as the
final DBP results are available, but no later than eighteen (18) months after publication of the
final rule in the Federal Register.
1-7
-------
Full-scale GAG or membrane treatment - If the PWS wishes to avoid the treatment studies
requirement on the basis of operating full-scale GAG or membrane processes, the PWS shall
submit an application, including any full-scale or pilot-scale data showing precursor removal,
not later than eighteen (18) months after final rule publication.
Grandfathered Studies - If a PWS believes it qualifies to avoid the requirements for a
treatment study under the grandfathered study provisions, it shall submit an application to EPA
not later than nine (9) months after publication of the final rule. The application must include
a description of the study, the equipment used, the experimental protocol, the analytical
methods and any reports resulting from the study. If the grandfathered study is approved then
specific information will be requested, including cost information, which must be submitted no
later than eighteen (18) months after publication of the final rule in the Federal Register.
3.3 Criteria For Determining Individual Treatment Study Requirements
The reg-neg committee agreed that granular activated carbon and membrane treatment
were the technologies most likely to be effective for precursor removal. The committee also
had determined that both bench-scale studies and pilot-scale studies should be allowed,
recognizing that pilot-scale studies provide more relevant data, but that bench-scale studies
should be less expensive to conduct. The treatment objective for all studies should be the
achievement of levels of disinfection byproducts less than 40 y.g/L and 30 ,ug/L for THM4 and
HAAS, respectively when free chlorine is used as the disinfectant.
Plants required to conduct the monitoring described in Section 3.1 that do not qualify to
avoid the treatment studies must conduct individual treatment studies or joint studies, or
contribute funds to a cooperative research effort. The scale of the required treatment study
(i.e., bench-scale or pilot-scale) is based on both system and plant size as described in Table 2,
and footnote 4 to Table 2, of § 141.141(b)(2) of the ICR rule. The required studies for plants
that do not meet the avoidance criteria are summarized here.
All plants that serve 500,000 persons or more shall conduct a pilot-scale study (categories
A & B in Tables 1 and 2 of § 141.141(b) of the ICR).
All plants that serve 100,000 to 499,999 persons shall conduct either a pilot-scale or a
bench-scale study (categories A & B in Tables 1 and 2 of § 141.141(b) of the ICR).
If a PWS serves more than 100,000 persons, but does not own a single plant that serves
more than 100,000 persons, then the largest plant owned by that system shall conduct
either a pilot-scale or a bench-scale study (categories C & D hi Tables 1 and 2 of
§141.141(b)oftheICR).
If a PWS serves 50,000 to 99,999 persons with 50,000 persons or more served by ground
water, then the largest plant owned by that system shall conduct either a pilot-scale or a
bench-scale study (category G in Tables 1 and 2 of § 141.141(b) of the ICR).
1-8
-------
A PWS with multiple plants operating on a demonstrated common source, as defined in
Section 3.6, is not required to conduct more than one treatment study for those plants. A
PWS must apply for this option according to the instruction in Section 5.2.3 no later than
eighteen (18) months after promulgation of the ICR. If both pilot-scale and bench-scale
treatment studies would otherwise be required for treatment plants operating on a common
source, the PWS shall conduct a pilot-scale study.
Plants required to conduct individual treatment studies shall submit a study concept form
and a study plan to EPA no later than eighteen (18) months after publication of the final rule in
the Federal Register. The study concept form and requirements for the study plan are
described hi Section 5.4. The study plan will be reviewed by EPA, and feedback on the
design will be provided if necessary.
3.4 Joint Treatment Studies
Public comment in response to the proposed ICR rule was hi favor of allowing systems to
cooperate hi conducting joint studies. Therefore, the final rule allows plants which treat water
from a common water resource, as defined hi Section 3.6, and which have similar treatment to
join together hi conducting studies. Similar treatment means that, for example, softening
plants may not conduct joint studies with conventional plants.
The miiiimum number and types of studies to be conducted during joint studies are
described hi the final regulation and hi Tables 3-3 and 3-4. These minimum requirements
were developed in order to maintain some degree of equity in the treatment study requirement
while providing some benefit to systems conducting joint studies. Only plants hi the same
population served category (i.e., < 500,000 persons or ^ 500,000 persons) will be allowed to
join together to conduct joint studies. The maximum number of plants allowed to join together
to conduct joint studies in the < 500,000 size category is six; and for the ;> 500,000 size
category, no more than three plants may join together to conduct a joint study.
An approved grandfathered study may not be used as a joint study unless all the common
source requirements, as defined hi Section 3.6, and the requirements hi Table 3-3 or 3-4 are
met, and the PWS which conducted the study submits written concurrence.
Joint studies should be conducted to maximize the benefit to the participating plants. In
most cases, when joint studies are conducted, multiple studies are required. The systems
should use that opportunity to develop information appropriate for their situations. For
example, if it is not obvious whether GAC or membrane technology would be the technology
of choice, the joint study could include a comparison of the two. If, on the other hand,
membrane technology would have a clear advantage over GAC for a situation where two
studies are required, the joint study might include two separate studies of membrane
technology, designed to evaluate different membranes, membrane configurations or
pretreatments; or conducted at two different locations to provide hands-on experience at each
plant.
1-9
-------
Request To Conduct Joint Treatment Studies
PWSs that wish to conduct joint treatment studies shall submit a letter of intent to EPA no
later than twelve (12) months after publication of the final rule in the Federal Register. All
participating PWSs should understand that the letter of intent is not binding. If the results of
the full twelve months of monitoring show that a plant may avoid conducting studies because a
surface water plant influent TOC is less than 4.0 mg/L or a ground water plant finished water
TOC is less than 2.0 mg/L or distribution system THM4 and HAAS are lower than 40 /ag/L
and 30 //g/L, respectively, the system may avoid conducting a study under one of those
exclusions. Likewise, if the full twelve months of monitoring result hi TOC or UFCTOX data
which no longer support the common source designation, the systems will not be allowed to
conduct joint studies. Also, one or more of the systems may simply change its mind, and
decide not to participate in the joint study. The purpose of the letter of intent is to provide
early notification to EPA of the interest of the participating PWSs and allow EPA to begin
reviewing the supporting information.
Once all of the supporting data has been obtained from the twelve months of applicability
monitoring, the cooperative of plants shall submit an application to conduct a joint study,
according to the instructions hi Section 5.2.4, no later than eighteen (18) months after
publication of the final rule in the Federal Register. EPA will review this application and
notify the plant cooperative whether the joint study is approved or disapproved.
3.5 Criteria Under Which An Alternative To Conducting A Treatment Study Is
Allowed
In lieu of conducting the required treatment study, a PWS may apply to contribute funds to
a Disinfection Byproducts/Microbial Research Fund (termed the buyout option). The PWS
selecting this option must use a common water resource, as described in Section 3.6, on which
a plant or cooperative of plants is conducting a study. A treatment plant serving 500,000
persons or more cannot buy out unless a plant serving 500,000 persons or more is conducting
a pilot-scale study on the common source. A treatment plant serving fewer than 500,000
persons can buy out if either a bench-scale or a pilot-scale study is being conducted on the
common source. An approved grandfathered study can be used as justification for contributing
to the cooperative research effort.
A PWS selecting this alternative for a treatment plant serving a population of 500,000 or
more shall contribute $300,000. A PWS choosing this option for a treatment plant serving
fewer than 500,000 persons shall contribute $100,000. The funds shall be contributed to the
Disinfection Byproducts/Microbial Research Fund, to be administered by the American Water
Works Association Research Foundation (AWWARF) under the direction of an independent
research council, for use hi a dedicated cooperative research program related to disinfectants,
disinfection byproducts, and enhanced surface water treatment.
Request To Buy Out Of Treatment Studies
A PWS that believes it qualifies to buy out of the treatment study requirement under the
alternative described in the preceding paragraph shall submit a letter of intent no later than
1-10
-------
twelve (12) months after the publication of the final rule in the Federal Register expressing its
intention to contribute funds to the Disinfection Byproducts/Microbial Research Fund. The
letter shall identify the other treatment plant(s) using the common water resource which will be
conducting a study.
After all of the supporting data has been obtained from the twelve months of applicability
monitoring, the PWS must submit an application to buy out of the treatment study requirement
according to the instructions hi section 5.2.5. This application must be submitted no later than
eighteen (18) months after publication of the final rule hi the Federal Register. Approval
cannot be final until EPA has confirmed that a treatment study is being conducted by another
plant on the common source. EPA will notify the PWS if it can avoid the study by
contributing funds to the Disinfection Byproducts/Microbial Research Fund. Information will
be provided in the notification of approval on the mechanism for contributing the funds to the
research fund. The PWS shall make the contribution no later than 90 days after notification by
EPA that the buy out application is approved.
3.6 Criteria For A Common Water Resource
Three treatment study options require that cooperating plants demonstrate that they operate
on a common water source: (1) a PWS with multiple plants operating on a common source
intending to conduct a single study, (2) a cooperative of treatment plants intending to conduct
a joint treatment study, and (3) a plant intending to buy out of the treatment study requirement.
A PWS or a cooperative of PWSs applying for one of these treatment study options shall
submit an application for a common source designation according to the instructions listed in
Section 5.3. This section describes the requirements for a common source designation.
Common River Source
Treatment plants on the same river are considered to have a common source if:
(1) each cooperating plant intake is no more than 20 river miles from all other intakes, and the
mean influent TOC of each cooperating plant is within 10% of the average of the mean TOCs
of all the cooperating plants; or (2) the intake of all cooperating plant are between 20 and 200
river miles apart, and the mean influent UFCTOX of each cooperating plant is within 10% of
the average of the mean UFCTOXs of all the cooperating plants. The mean TOC or
UFCTOX is calculated from the twelve consecutive months of monitoring to determine
treatment studies applicability as described hi Section 3.1.
Common Lake, Reservoir Or Ground Water Source Under the Direct Influence
Treatment plants using the same lake, reservoir or ground water resource under the direct
influence of surface water are considered to have a common resource if the mean influent TOC
of each cooperating plant is within 10% of the average of the mean TOCs of all the
cooperating plants. The mean TOC is calculated from the twelve consecutive months of
monitoring to determine treatment studies applicability as described in Section 3.1.
1-11
-------
Common Ground Water Resource
Treatment plants with intakes from a single aquifer are considered to be using a common
source if the mean finished water TOC of each cooperating plant is within 10% of the average
of the mean TOCs of all the cooperating plants. The mean TOC is calculated from the twelve
consecutive months of monitoring to determine treatment studies applicability as described hi
Section 3.1.
1-12
-------
Table 3-1 Treatment Study Applicability Monitoring Requirements For Plants
Sampling Point
Plant Influent for Surface Water Plants
Plant Effluent for Ground Water Plants
Four Distribution System Samples for
Both Surface and Ground Water Plants
Monthly Analyses
TOC and UFCTOX1
TOC and UFCTOX1
Quarterly Analyses
THM4 and HAA52
1 UFCTOX analysis, defined as total organic halides evaluated under uniform formation conditions, is only
required for treatment plants using a common river source with intakes between 20 and 200 river miles apart
and applying for a common source designation.
2 Treatment plants that use only chlorine as the primary and residual disinfectant may monitor THM4 and
HAAS in the distribution system to determine applicability.
Table 3-2 Monitoring Required Across Full-Scale GAC Or Membrane Processes
Sampling Point
Before GAC or membranes
After GAC or membranes
Monthly Analyses1
pH, Alkalinity, Turbidity, Temperature,
Calcium and Total Hardness, TOC and UV254
pH, Alkalinity, Turbidity, Temperature,
Calcium and Total Hardness, TOC and UV254
Sampling Point
Before GAC or membranes if
disinfectant is applied at any point in the
treatment plant prior to these processes
After GAC or membranes if disinfectant
is applied at any point in the treatment
plant prior to these processes
Quarterly Analyses2
THM4, HAA6, HAN, CP, HK, CH, TOX
THM4, HAA6, HAN, CP, HK, CH, TOX
1 TOC: total organic carbon. UV^: absorbance of ultraviolet light at 254 nanometers.
2 THM4: trihalomethane (four). HAA6: haloacetic acids (six). HAN: haloacetonitriles. CP: chloropicrin.
HK: haloketones. CH: chloral hydrate. TOX total organic halide. For THM4, HAA6, HAN, and HK,
analytical results for individual analytes shall be reported.
-------
Table 3-3 Joint Study Requirement For Plants With A Population Served < SOOjOOO1
No of Plants
2
3
4
5
6
Minimum Studies to be Conducted
1 pilot (GAG or membrane)
1 pilot and 1 bench (GAG or membrane)
2 pilots (GAG and/or membrane)
2 pilots (GAG and/or membrane),! bench (GAG or membrane)
2 pilots and 2 bench (GAG and/or membrane)
1 Each treatment plant must serve a population less than 500,000.
Table 3-4 Joint Study Requirement For Plants With A Population Served
No. of Plants
2
3
Minimum Studies to be Conducted
1 pilot (GAG or membrane), 2 bench (GAG and/or membrane)
2 pilots (GAG and/or membrane)
1 Each treatment plant must serve a population of 500,000 or more.
-------
4.0 General Guidelines
4.1 Precursor Removal Technology
The reg-neg committee agreed that granular activated carbon and pressure-driven
membrane processes held the most promise for widespread application for DBF precursor
removal in the U.S., and some of the pros and cons associated with each technology are
summarized in Table 4-1. Both technologies have been used to a varying extent and for
various purposes for some time, and there were sufficient data available to indicate that each
technology has the capability to achieve relatively low levels of precursors under certain
circumstances. There was widespread experience hi conducting both bench-scale and pilot-
scale testing of GAC within the drinking water community but less familiarity with testing
membrane technology. The reg-neg committee used a volunteer group of technical experts to
craft a tentative strategy for precursor removal testing in the ICR proposal. The outcome was
that there should be allowed either bench-scale or pilot-scale testing of either GAC or
pressure-driven membrane technology, specifically either reverse osmosis or nanofiltration.
The treatment study requirements for both technologies using either bench- or pilot-scale
systems are summarized in Table 4-2.
Pilot testing of GAC has been done for decades in the chemical process industry, for
wastewater reclamation and drinking water treatment. An effective bench-scale procedure for
testing GAC performance, called the rapid small-scale column test (RSSCT), has been used
successfully by a number of researchers and has provided results comparable to longer pilot
tests. This manual contains procedures for conducting GAC pilot tests and RSSCTs.
There is less experience in the water industry with testing membrane technology, and thus
less confidence hi specifying a testing technique for the ICR. Therefore, two approaches to
bench-scale testing of membrane technology are presented to allow systems opting for
membrane bench studies a choice which will provide at least one technique suitable for local
circumstances. One approach, termed the rapid bench-scale membrane test (RBSMT), uses a
small sheet of membrane material and a small volume of water to allow the PWS the option of
conducting membrane testing off-site. The second approach, the single element bench-scale
test (SEBST), uses a single membrane element similar to those which would be used in large
numbers in a full-scale system. This approach requires testing on-site but may provide better
data. Likewise, considerable flexibility is provided for pilot studies. Pilot membrane systems
must be designed as a 2-1 array and operated at a recovery of at least 75%. Since membrane
pilot studies represent a significant investment, the PWS conducting the test should do so in a
way which provides the most appropriate data for its use while, concurrently meeting EPA's
need for data from which estimates of national impact of future regulations can be made.
The treatment goals for the ICR treatment studies are effluent levels of disinfection
byproducts no greater than those proposed as Stage 2 DBF MCLs in the Federal Register,
Vol.59, No. 145, Friday, July 29, 1994. Stage 2 proposed MCLs were set at 40 /j.g/L for
THM4 and 30 //g/L for HAAS as an annual average. These studies will provide information
1-13
-------
concerning the feasible level of DBFs that can be achieved, which will be used in a second
negotiated rule-making process to determine the final MCLs.
4.2 Influent To The Treatment Studies
The ICR specifies that the precursor removal testing be done on the plant water after it has
passed through any full-scale treatment processes currently in place to remove precursors,
such as coagulation/sedimentation and filtration. However, one of the guiding principles hi
considering future regulations for control of DBFs is that removal of DBF precursors should
be maximized before contact with disinfectants, especially chlorine. Therefore, the ICR also
requires that feed water to the treatment studies be taken from a point in the treatment plant
before any application of oxidant or disinfectant that could form chlorinated byproducts. If a
chlorine-based oxidant is added prior to full-scale processes which remove precursors, then the
full-scale treatment should be simulated on the bench- or pilot-scale prior to introduction to the
test system. For example, if a water plant routinely adds chlorine to the rapid mix unit, and a
system intends to conduct pilot testing of either membranes or GAC, the system must collect
the water for the pilot test ahead of the rapid mix unit and treat the non-chlorinated water with
pilot unit processes which simulate the full-scale processes currently in place. For bench-scale
testing, it may be possible to alter the point of disinfectant addition for a short enough time to
obtain a batch of water from the appropriate point hi the treatment process; if not, a small
volume of water from a point before disinfectant addition could be obtained and batch-treated
to simulate the full-scale processes.
There may be cases where the simulation of some full-scale treatment process may not be
appropriate or necessary. For example, since both the RSSCT and the RBSMT use cartridge
filtration as a routine test pretreatment step, it would not be necessary to simulate granular
media filtration for those procedures. One of the primary concerns in membrane processes is
pretreatment to minimize fouling or loss of productivity. Therefore, it may not be appropriate
to require that water applied to a membrane system undergo exactly the same processes as the
current full-scale system, if there are good reasons to treat the water differently.
Multiple bench-scale tests or long-term pilot studies are generally required to assess
seasonal variation or, if seasonal variation is not significant (as is the case for most ground
waters), to assess other factors impacting the performance of the precursor removal
technology. For example, one of the major factors impacting membrane performance is
pretreatment. Therefore, if seasonal variation is not an issue and the water to be treated is not
a high-quality ground water, systems may wish to conduct studies under various pretreatment
conditions and would be allowed to do so. If seasonal variation does occur, the tests should be
conducted to account for it; and for bench-scale tests, the test conditions should be as
consistent as possible over all four quarterly tests, with differing results indicating seasonal
differences.
Seasonal variability of source waters will not be explicitly defined due to the variety of
factors that could lead to significant seasonal variation. One of the most important parameters
that can be used to assess the seasonal variability of a water is the TOC concentration, and if
1-14
-------
the TOG varies by more than a factor of two over the course of a year, then it is recommended
that seasonal variability be evaluated. Another parameter that can be used to assess seasonal
variability is the concentration of DBF precursors assessed under simulated distribution system
(SDS) conditions. SDS-DBPs are not only affected by the precursor concentrations, but also
the conditions of the SDS test including temperature, pH and chlorine residual. Thus, SDS-
DBPs may be a better indicator of seasonal variability than TOG for the purposes of the ICR.
GAG performance is sensitive to both the concentration and nature of organic precursors in the
feed water, and if these parameters vary by more than a factor of two over one year, the effect
of seasonal variability on performance should be evaluated. The performance of membrane
processes is less sensitive to the concentration of DBF precursors in the feed water; however,
seasonal variations in feed water temperature and the concentration of inorganic solutes can
have a significant impact on productivity and cost.
The PWS conducting the study must identify the water quality parameters that vary in a
given source, and use judgement to decide if these parameters have a significant impact on the
ability of the process under investigation to control DBPs. For membrane processes, the effect
of seasonal varibility on water production and the rate of membrane fouling must also be
considered when assessing the impact of seasonal variability on performance.
4.3 GAC Studies
GAG has limitations for economical precursor removal. If high levels of TOG (greater
than 6 to 8 mg/L) are applied to GAC systems with EBCTs in the range of the ICR testing
procedure, early breakthrough may occur and the ICR treatment goal of less than 40 ug/L for
THM4 and 30 jxg/L for HAAS may not be economical. PWSs with high levels of precursors
should be aware that GAC without some additional pretreatment, such as enhanced coagulation
or pH control, may not be a good candidate for precursor removal.
Pilot Studies
Details of the pilot study procedure are provided in Part 2 of this manual. Two EBCTs,
10 minutes and 20 minutes, shall be evaluated for the pilot studies. Each EBCT run will be
terminated when either 70% breakthrough or steady-state removal of precursors is achieved,
as described in Part 2. If either of these criteria is met for the 20-minute contact time prior to
4000 hours run tune, a second run at both EBCTs shall be conducted following the same
sampling requirements. In all cases, the maximum run length for the pilot-scale study (one or
two runs) is 8000 hours. The pilot study should be timed to account for seasonal variations in
water quality. If seasonal variation is not a factor and the first 20 minute EBCT terminates
before 4000 hours run time, the second run can be used to obtain data on other factors of
interest. Factors important in GAC treatment include pretreatment, such as enhanced
coagulation or pH adjustment, carbon type and EBCT.
Precursor removal performance is characterized by conducting simulated distribution
system (SDS) chlorination experiments on the GAC influent and effluent along with other
analyses. The SDS conditions are the average conditions in the distribution system served by
the plant conducting the study. The chlorination of samples for SDS analysis is described hi
1-15
-------
Part 1, Section 4.6, and also in Part 2, Section 6.0, where special requirements for GAC
testing are included.
Bench-scale Studies
The bench-scale GAC test is the RSSCT, as described hi Part 2. This test shall be run to
simulate full-scale EBCTs of 10 minutes and 20 minutes. The RSSCT runs shall be conducted
quarterly to account for seasonal variation or in the case of insignificant seasonal variation, to
examine other factors as with the pilot studies. If seasonal variation is not significant, the
PWS may conduct the four runs at 10- and 20-minute EBCTs at any tune. The ICR requires
that if the first quarterly RSSCT run is terminated because of early breakthrough, as specified
in Part 2, within 20 full-scale-equivalent days for the 10-minute EBCT test or 30 full-scale-
equivalent days for the 20-minute EBCT test, the system shall switch to membrane bench tests,
conducting the remaining three quarterly tests with only one membrane.
4.4 Membrane Studies
Membrane processes such as nanofiltration and reverse osmosis have been shown to
achieve very high removals of DBF precursors that would enable utilities treating feed waters
with TOC concentrations greater than 10 mg/L to meet the ICR treatment goal of 40 ug/L for
THM4 and 30 jig/L for HAAS when free chlorine is used as both the primary and residual
disinfectant. Membranes treating ground waters have demonstrated excellent control of DBFs
while maintaining acceptable productivity. However, a limited number of studies on surface
waters indicate that severe membrane fouling can occur when treating these waters. This may
require that surface waters undergo extensive pretreatment, such as enhanced coagulation or
microfiltration, prior to membrane treatment.
Pilot Studies
Part 3 of this manual describes the requirements of the membrane pilot studies. The
evaluation of only one membrane type under one set of operating conditions is required during
a pilot study. The pilot study shall be run continuously over a period of one year, with
allowances for down-tune due to membrane cleaning, maintenance or other reasons. The
pilot-scale run tune shall be no less than 6600 hours, which represents approximately 75% of
one calendar year. A pilot system must use standard elements at least 2.5 inches in diameter
by 40 niches in length. This size requirement is for membranes in spiral-wound
configurations, but standard hollow-fiber elements can also be used, although hollow-fiber
technology is not recommended for surface waters. The system must consist of at least two
stages, with a minimum of two pressure vessels in the first stage and one pressure vessel in the
second stage (i.e., a 2-1 array), and each pressure vessel must contain at least three membrane
elements.
Membrane performance is to be assessed hi terms of productivity and permeate water
quality including precursor removal. Precursor removal is measured by conducting SDS
chlorination experiments on the membrane feed and permeate along with other analyses. The
SDS conditions are the average conditions of the distribution system served by the plant
conducting the study.
1-16
-------
Bench-scale Studies
Two approaches for bench-scale membrane testing are described in Part 3, the RBSMT and
the SEBST. Additionally, there are two options provided for conducting SEBSTs: (1)
quarterly studies and (2) a yearlong study. The RBSMT and quarterly SEBST studies require
that two membranes be evaluated four times over the course of one year to account for
seasonal variation, or hi the case of hisignificant seasonal variation, to examine other factors.
If seasonal variation is not significant, the PWS may conduct the four runs on two membranes
at any tune. Some variables that a PWS may wish to investigate include pretreatment,
additional membrane types and operating parameters such as flux and recovery. The yearlong
SEBST study requires that one membrane be evaluated for a run time of at least 6600 hours
over a one year period.
Three options for bench-scale membrane testing are provided to allow some flexibility hi
meeting the ICR requirements. The RBSMT can be run off-site and offers a great deal of
operational flexibility, while a single element test may provide better data but must be
conducted on-site and would require long-term operator attention and a continuous supply of
treated, unchlorinated feed water. The long-term SEBST study provides the most flux data,
but only for one membrane type. Regardless of the approach selected, title membranes
investigated must be evaluated with respect to productivity and permeate quality including
precursor removal as assessed under SDS conditions.
4.5 Analytical Methods
A list of approved analytical methods for the treatment studies is provided in Table 7 of the
ICR rule. These methods and mandatory quality control procedures are contained in the
"DBP/ICR Analytical Methods Manual" (EPA 814-B-96-002). One method not described in
this manual is the measurement of conductivity or total dissolved solids (TDS) with either a
conductivity probe or a TDS meter. TDS or conductivity measurements are commonly used to
assess the rejection characteristics of a membrane. Method 2510 B (Standard Methods, 18th
ed., 1992) describes the procedure for measuring conductivity with a probe. Measurement of
TDS is similar to the measurement of conductivity; the main difference is that a TDS meter is
calibrated to read conductivity as TDS in mg/L using a specific conversion factor.
Some of the data sheets hi Parts 2 and 3 contain spaces for all nine of the HAA species;
however, only the following six HAAs are required: monochloroacetic acid, dichloroacetic
acid, trichloroacetic acid, monobromoacetic acid, dibromoacetic acid and bromochloroacetic
acid. These six HAAs make up HAA6, and the first five HAAs in this list make up HAAS. If
additional HAAs are measured, then they should be reported.
4.6 Chlorination Procedures
SDS Test Procedure
As described hi § 141.144(b)(3) of the ICR rule, samples collected during the treatment
studies must be chlorinated under site-specific SDS conditions to evaluate the formation of
THM4, HAA6 and TOX.
1-17
-------
The SDS conditions that must be selected are incubation tune, free chlorine residual at the
end of the incubation period, pH during incubation, and the temperature during incubation.
These conditions should be representative of the average distribution system conditions at the
time of the test, and the same set of conditions must be used for the influent sample and the
corresponding effluent sample for each experiment. For the bench-scale studies, it is required
that the same pH, incubation tune, free chlorine residual and temperature be used for all SDS
samples during a quarter; however, different SDS conditions can be used during different
quarters to simulate the current distribution system conditions. If chlorine is not used as the
final disinfectant in practice, then the target free chlorine residual should be set at 1.0 to 0.5
mg/L under the SDS pH, incubation time and temperature.
The chlorine dose required to achieve the target chlorine residual can be determined by
first conducting a demand study with the water sample. The SDS incubation tune, temperature
and pH should be used during the demand study, but the chlorine dose should be varied. The
chlorine residual can be plotted as a function of chlorine dose, and this relationship can be
used to determine the chlorine dose required to achieve the target chlorine residual. Since the
TOG of a water can vary over the course of a run (e.g., the GAC effluent), the chlorine
demand will also vary. To achieve the same free chlorine residual in these samples, the
chlorine dose must be varied according to the demand.
The target pH can be difficult to achieve for poorly buffered samples, and it may be
necessary to buffer these waters to achieve the desired pH at the end of the incubation period.
UFC Test Procedure
The uniform formation conditions (UFC) procedure is a standardized set of representative
chlorination conditions:
Incubation tune:
Incubation temperature:
Buffered pH:
24-hour chlorine residual:
24 ± 1 hours
20.0 ± 1.0°C
8.0 ± 0.2
1.0±0.4mgCl2/L
Some ICR monitoring requirements include the evaluation of TOX under UFC conditions
(i.e., UFCTOX) to demonstrate a common source for some types of surface water resources.
The standard operating procedure for the UFC test is included hi Appendix 2-B located at the
end of Part 2 of this manual.
1-18
-------
Table 4-1 Factors For Consideration In Selecting Treatment Study Technology
Technology
Pros
Cons
Membranes
(Nanofiltration and
Reverse Osmosis)
-Can remove pathogens and DBF
precursors to very low levels.
-Can remove inorganic contaminants.
-Fouling can limit the
application of membrane
technology for surface waters.
-The concentrate waste stream
requires disposal and/or
treatment.
GAC
-Typically less expensive than
membrane technology.
-A variety of finished water qualities
are available through blending.
-Generally not feasible with
GAC influent TOC > 6 to 8
mg/L without additional
pretreatment.
-Does not remove pathogens.
Table 4-2 Summary Of Treatment Study Requirements For GAC And Membrane, Pilot-
Scale And Bench-Scale Systems
Technology
Bench-Scale
Pilot-Scale
Membranes
(Nanofiltration and
Reverse Osmosis)
The three options are:
1.) RBSMT studies evaluating two
membranes at four recoveries
over four quarters.
2.) Single element studies evaluating
two membranes at 75% recovery
over four quarters.
3.) Long-term single element studies
evaluating one membrane at 75%
recovery over at least 6600 hours.
-A minimum configuration of:
two pressure vessels hi the
first stage and one pressure
vessel in the second stage.
-At least three elements per
pressure vessel.
-A minimum 2.5" x 40"
element size.
-A minimum recovery of 75%.
-Minimum 6600 hour runtime.
GAC
-RSSCT studies evaluating two
EBCTs (10 and 20 minutes)
conducted quarterly.
-Operated until 70% breakthrough.
-At least two inch diameter
columns evaluating 10 and 20
minute EBCTs run until 70%
breakthrough.
-If the 70% breakthrough
occurs prior to 4000 hours
runtime for the 20 minute
EBCT, a second pilot run
must be conducted.
-------
-------
5.0 Applications And Reports
A PWS must submit the following for each plant required to conduct treatment study
applicability monitoring according to the schedule listed in Table 5-4: (1) the treatment study
applicability data and (2) an application for one of the treatment study options. This section
includes application materials for the various treatment study options, which include:
1. Individual treatment studies
2. Treatment study avoidance
3. Single study by multiple plants operating on a common source and owned by the same
PWS (abr., single study for multiple plants)
4. Joint treatment studies
5. Treatment study buyout
6. Grandfathering previous studies
All data, applications and reports relevant to the treatment studies shall be submitted to the
following address:
U.S. EPA
Technical Support Division
ICR Treatment Studies Coordinator
26 West Martin Luther King Drive
Cincinnati, OH 45268
5.1 Application Materials And Deadlines
This section includes nine forms (Tables 5-1, 5-2, and 5-5 thru 5-11) that are to be used in
the application process for various treatment study options. Every correspondence between
the plant or PWS and EPA must include the general information sheet shown in Table 5-1.
All plants required to conduct treatment study applicability monitoring shall submit their
treatment study applicability data on the form shown in Table 5-2, no later than seventeen (17)
months after promulgation of the ICR. This applicability data is required to review most
applications.
Table 5-3 summarizes the tables and additional information required with each application.
All applications shall consist of a cover letter, a general information form (Table 5-1) and at
least one additional table. The cover letter must be signed by the official contact person of the
applying PWS (or each PWS involved in a joint study). This letter should include any special
circumstances or information to be considered during the review of the application. Only one
application can be submitted at a time, and multiple applications will be returned to the sender
without consideration.
Table 5-4 summarizes the deadlines for the applicability data, treatment study option
applications, start of studies, quarterly progress reports, and the final treatment study report.
The deadlines are listed relative to the date of rule publication since the ICR was not
1-19
-------
promulgated at the time this manual was finalized. However, a blank column is provided hi
Table 5-4 so that the actual dates can be entered into the table when available.
EPA will provide technical assistance through the Safe Drinking Water Hotline at
1-800-426-4791. If the Hotline is unable to answer a question, it will be forwarded to a
technical assistance group that will provide an answer.
5.2 Treatment Study Option Review Criteria
The general criteria by which applications for each treatment study option will be reviewed
is presented here. These criteria may not be inclusive of all situations, but are intended to
provide an indication of the necessary requirements to qualify for each of the treatment study
options.
5.2.1 Individual Studies
The individual study requirements are based on plant size and are summarized as follows:
Plants serving 500,000 persons or more must conduct a pilot-scale treatment study.
Plants serving fewer than 500,000 persons must conduct either a bench-scale or a pilot-
scale treatment study.
A study concept form, Table 5-11, and study plan shall be submitted for each study and
will be reviewed according to the criteria described in Section 5.4.
5.2.2 Treatment Study Avoidance
A plant intending to avoid the treatment study requirement on the basis of TOC, DBPs or
full-scale membrane or GAC technology shall submit the application hi Table 5-5 along with
the general information form, Table 5-1. This section describes the criteria that must be met
to avoid the requirement of conducting a treatment study.
Plants can avoid treatment studies on the basis of TOC.
* Treatment plants using surface waters or ground waters under the direct influence of
surface water with a yearly average influent TOC concentration less than 4.0 mg/L,
based on monthly applicability monitoring, do not have a treatment study requirement.
Treatment plants using only ground waters with a yearly average finished water TOC
concentration less than 2.0 mg/L, based on monthly applicability monitoring, do not
have a treatment study requirement.
Or
Plants can avoid treatment studies on the basis of THM4 and HAAS.
Plants must be using only free chlorine as the primary and residual disinfectant AND
the mean of four quarterly average THM4 concentrations must be less than 40 ug/L
AND the mean of four quarterly average HAAS concentrations must be less than 30
1-20
-------
The quarterly average THM4 and HAAS are determined from four distribution system
samples as described in Sections 3.1 and 3.2
Or
Plants can avoid treatment studies if they use full-scale membrane or GAC.
Full-scale GAC technology is defined as GAC with an empty bed contact time of at
least 15 minutes and a reactivation or replacement frequency of no less than 9 months.
Full-scale membrane technology is defined as nanofiltration or reverse osmosis capable
of removing DBF precursors.
Plants using full-scale GAC or membranes that do not meet the above criteria may
apply for treatment study avoidance if data demonstrating effective precursor removal
is included with the application.
A plant using full-scale membrane or GAC technology must conduct the monitoring
and submit full-scale plant data as described in Section 3.2 ensuring that GAC or
membranes technology is included in the treatment train being monitored.
5.2.3 Single Study For Multiple Plants Owned By A Single PWS
A PWS operating multiple plants on a demonstrated common source may apply to conduct
a single treatment study for all plants using that common source. The PWS shall submit the
application in Table 5-6 along with a common source designation application, Table 5-10.
Additionally a general information form, Table 5-1, must be submitted for each plant involved
hi the single study for multiple plants, and all information shall be included in a single
application packet. The criteria that must be met for approval of this treatment study option
are:
All cooperating plants must demonstrate that they are owned by a single PWS.
All cooperating plants must demonstrate that they operate on a common source according
to the criteria described in Section 5.3.
If the largest plant owned by the system and operating on a common source serves 500,000
persons or more, then a single pilot-scale study must be conducted by the PWS at that
source.
If the largest plant owned by the system and operating on a common source serves under
500,000 persons, then a single bench-scale or pilot-scale study must be conducted by the
PWS at that source.
« A study concept form, Table 5-11, and study plan must be submitted for each study and
will be reviewed according to the criteria described in Section 5.4.
5.2.4 Joint Treatment Studies
Multiple plants owned by different PWSs operating on a demonstrated common source may
apply to conduct joint treatment studies. The cooperative of plants shall submit the application
in Table 5-7 along with a common source designation application, Table 5-10. Additionally, a
general information form, Table 5-1, must be submitted for each plant involved hi the joint
study. Only a single joint study application, containing all of the necessary information, shall
1-21
-------
be submitted by the cooperative of plants. In order for a joint treatment study to be approved,
the following criteria must be met:
All cooperating plants must demonstrate that they are in the same size category (i.e., all
plants serve either^ 500,000 persons or < 500,000 persons).
All cooperating plants must demonstrate that they operate on a common source according
to the criteria described in Section 5.3.
All cooperating plants must demonstrate that they use similar treatment.
The number and type of studies to be conducted must be consistent with the rules outlined
in Section 3.4.
A study concept form, Table 5-11, and study plan must be submitted for each study and
will be reviewed according to the criteria described in Section 5.4.
An approved grandfathered study cannot be used as a joint treatment study unless all of the
joint study criteria are met, and the PWS submits written concurrence of the PWS which
conducted the study.
PWSs intending to conduct a joint study should notify EPA of this intent no later than
twelve (12) months after ICR promulgation. The joint study letter of intent should include the
same information required hi the actual application, including all available applicability data
and common source data. However, it should be understood that this letter is not binding, and
the PWS can pursue other options prior to the deadline for treatment study option applications
(i.e., eighteen months after publication of the ICR). The purpose of this letter is to allow EPA
to start reviewing the available information; however, EPA cannot make a final decision until
all supporting data has been submitted.
5.2.5 Treatment Study Buyout
Plants may apply to contribute to a DBP/Microbial Research Fund hi lieu of conducting a
treatment study if the criteria described hi this section are met. A PWS seeking to buy out of
the treatment study requirement for a plant(s) shall submit the application hi Table 5-8 along
with a common source designation application, Table 5-10. Additionally, a general
information form, Table 5-1, must be submitted by the plant applying for the buyout and for
the plant operating on the common source which is conducting a treatment study. This section
describes the criteria that must be met for a buyout to be approved.
Plants must demonstrate that they operate on a common source according to the criteria
described in Section 5.3.
At least one plant operating on the common source must conduct a treatment study.
Plants serving 500,000 persons or more must contribute $300,000 to the cooperative
research fund.
Plants serving fewer than 500,000 persons must contribute $100,000 to the cooperative
research fund.
* A plant serving 500,000 persons or more will only be permitted to buy out if another
common source plant serving 500,000 persons or more is conducting a pilot-scale study.
A plant serving fewer than 500,000 persons will be permitted to buy out if another
common source plant is conducting a bench-scale or a pilot-scale study.
1-22
-------
An approved grandfathered study can be used by other common source systems to buy out
of a treatment study.
PWSs intending to buy out of the treatment study requirement should notify EPA of this
intent no later than twelve (12) months after ICR promulgation. The buyout letter of intent
should include the same information required hi the actual application, including all available
applicability data and common source data. However, it should be understood that this letter
is not binding, and the PWS can pursue other options prior to the deadline for treatment study
option applications (i.e., eighteen months after publication of the ICR). The purpose of this
letter is to allow EPA to start reviewing the available information; however, EPA cannot make
a final decision until all supporting data has been submitted.
5.2.6 Grandfathering Previous Treatment Studies
A plant intending to grandfather a previous study to meet the treatment study requirement
shall submit the application in Table 5-9 along with the general information form, Table 5-1.
Additionally, the PWS must provide documentation to demonstrate the grandfathered study
meets the following requirements:
The study was conducted using the analytical methods described in "DBP/ICR Analytical
Methods Manual," EPA 814-B-96-002.
Both HAA6 and THM4 were evaluated using free chlorine and under SDS conditions
representative of the distribution system conditions experienced hi the full-scale plant
during the study, as described in Section 4.6.
The study was conducted using a protocol similar to one of the methods described in Part 2
or 3 of this manual.
A bench-scale study cannot be grandfathered to meet a pilot-scale study requirement.
Additionally, a plant must be able to provide the following information eighteen (18)
months after publication of the ICR rule:
The process performance data specified hi Part 2 or Part 3 of this manual in a format
specified by EPA.
The cost information requested hi Part 2 or Part 3 of this manual.
5.3 Review Criteria For A Common Source Designation
In order to conduct a single study for multiple plants (Section 5.2.3), conduct a joint
treatment study (Section 5.2.4) or buy out of a treatment study (Section 5.2.5), the cooperating
plants must demonstrate that they operate on a common source. Table 5-10 is an application
for a common source designation. Only one common source application shall be submitted for
the group of cooperating plants; however, a general information form, Table 5-1, for each of
the cooperating plants must be included with the common source application. This section
describes the criteria for determining a common source designation for various water sources.
1-23
-------
5.3.1 Rivers And Streams
Treatment plants on the same river are considered to have a common source if:
Each cooperating plant intake is no more than 20 river miles from all other intakes, and the
mean treatment plant influent TOC of each of the cooperating plants is within 10% of the
average of the mean TOC of all cooperating plants. (The mean is calculated from twelve
consecutive months of monitoring).
Or
* The intakes of all cooperating plants are between 20 and 200 river miles apart and the
mean treatment plant influent UFCTOX of each of the cooperating plants is within 10% of
the average of the mean UFCTOX of all cooperating plants. (The mean is calculated from
twelve consecutive months of monitoring).
5.3.2 Lakes, Reservoirs And Ground Waters Under The Direct Influence
Treatment plants using the same lake, reservoir or ground water under the direct influence
are considered to have a common source if:
The mean treatment plant influent TOC of each of the cooperating plants is within 10% of
the average of the mean TOC of all cooperating plants. (The mean is calculated from
twelve consecutive months of monitoring).
5.3.3 Ground Water Aquifers
Treatment plants using ground water are considered to have a common resource if:
The wells supplying water to the plants are developed in the same aquifer AND the mean
treatment plant finished water TOC of each of the cooperating plants is within 10% of the
average of the mean TOC of all cooperating plants. (The mean is calculated from twelve
consecutive months of monitoring).
5.4 Review Criteria For Study Concept Forms
All plants conducting individual treatment studies (Section 5.2.1), single studies for
multiple plants (Section 5.2.3) or joint treatment studies (Section 5.2.4), shall submit a study
concept form for each study to be conducted. Table 5-11 must be filled out for each study,
and a brief study plan (usually not more than two pages of text and two pages of figures) shall
accompany each study concept form. Each plant involved in the study must also submit Table
5-1. These forms and study plans will be reviewed by EPA primarily to insure that the study
is consistent with the requirements of the rule, and the study plans will also be reviewed for
technical soundness. This section describes the information that should be included in the
study concept form and study plan:
* The technology to be investigated during the study, GAC or membrane technology.
The scale of the study, pilot or bench.
1-24
-------
First point at which a chlorine-based disinfectant is added in the plant.
Point at which water to be used in the treatment study will be sampled.
A brief description of all pretreatment processes to be used in the treatment study (e.g.,
full-scale alum coagulation at a range of doses from 20 to 40 mg/L, full-scale
sedimentation, full-scale sand filtration at a 2.5 gpm/ft2 filtration rate, bench-scale acid
addition to pH » 5.0 and bench-scale cartridge filtration).
An estimate of the TOC concentration of the treatment study feed water after all
pretreatment processes.
The seasonal variability of the source water, and an estimate of how many tests over a one
year period would be required to evaluate the impact of these seasonal variations on GAC
or membrane.
If membrane studies are to be conducted, the study description should include the
procedure to be used (i.e., RBSMT, SEBST or a pilot system), and the model number,
manufacturer and molecular weight cutoff of all membranes being investigated.
If GAC studies are to be conducted, the study description should include the carbon type(s)
being investigated, the carbon particle diameter to be investigated and the column
diameter.
If quarterly bench-scale studies are to be conducted, the study description should include
the number of runs that will be used to assess seasonal variation. If this is less than four
runs, the planned experiments for the remaining quarterly runs should be described.
5.5 Treatment Study Progress Reports
In order to insure that the treatment studies are progressing a timely manner and to identify
the potential for delays, quarterly correspondence will occur between plants conducting studies
and EPA. The PWS, or cooperative of PWSs, should submit a progress report for each
treatment study being conducted. The one page progress report in Table 5-12 should be
submitted quarterly at the following times: one month, four months, seven months and ten
months after the start of the treatment studies. This report requests information such as the
number of bench-scale studies completed, the cumulative run time for pilot-scale studies, any
unplanned down-time, the number of primary and duplicate DBF samples collected to date and
other information to help EPA track the progress of the treatment studies.
5.6 Final Treatment Study Report
One complete data report shall be submitted no later than thirty-eight (38) months after
publication of the final rule in the Federal Register. Specific data reporting requirements for
the treatment studies are described in Parts 2 and 3 of this manual for GAC and membrane
studies, respectively. Data sheets are included in each part to demonstrate the required data
elements. Prior to the deadline to commence treatment studies, a computer diskette containing
data collection software will be provided to the Official ICR Contact at all PWSs conducting
treatment studies. This software shall be used to record and report most of the treatment study
data and shall be submitted to EPA along with the final treatment study report.
1-25
-------
In addition to the computer file, the plant must submit a summary report describing the
experimental design, analytical and experimental methods, significant results, process
performance and/or cost analyses (if conducted), a QA/QC summary, problems encountered
during the study, and any other information relevant to the treatment studies that is not
included in the computer data file. A plant required to conduct a treatment study must also
report design information on any full-, pilot- or bench-scale processes that precede the
advanced treatment process under investigation. This design information should include
process schematics, chemical doses and concentrations, hydraulic detention times, overflow
rates, loading rates, flow rates and any other critical design parameters.
1-26
-------
Table 5-1 General Public Water System And Plant Information (page 1 of 2)
Public Water System Information
Utility name
PWSID#[
WIDB# (optional) |_
PWS combined population served
PWS ground water population served
Official Contact Person
Name
Mailing address
Phone#
FAX#
E-mail address
ICR Contact Person
Name
Mailing address
Phone #
FAX#
E-mail address
Treatment Plant Information
Plant name
Plant ICR #
Plant combined population served
Plant ground water population served
Plant surface water population served
Plant Contact Person
Name
Mailing address
Phone #
FAX#
E-mail address
-------
Table 5-1 General Public Water System And Plant Information (page 2 of 2))
Treatment Plant Type (Select Appropriate
Plant Type ID
2
3
4
5
6
7
8
9
10
11
12
13
14
15
T7
Plant Type Code
DF
ILF
SSF
SOFT
TS/SOFT
CS/SOFT
STS
CPTS
DIS/GW
OTHER/GW
UNFILT/SW
MEMBRANE
DIS/WHSALE
OTH
PURCHASED
Plant Type
Conventional Filtration
Direct Filtration
In-Line Filtration
Slow Sand Filtration
Softening
Iwo Stage Softening
Coagulation/Sedimentation/Softening
Split Treatment/Softening
Complex Parallel Train/Softening
Disinfection Only/Ground Water
Other Groundwater (describe below)
Unfiltered Surface Water
Membrane Treatment
Disinfection of Wholesale Supply
Other (describe below)
i^urcnased Finished Water Without Re-Disinfection
Box
Describe other:
-------
rt
=tfc
2
00
*$
OJD
a
1
S
I
&
Q
2
2
US
1
f
oa
-M
"S
^
CA
1
1
t-l
a 4) 4) «
g a a c«
B ||^
«! ^ +j 4J
PI y
H i j§ «e
*?
«0
1
S
-*
1
o
s
m
1
S
cs
1
o
s
1 I
43
1
1 TOC monitoring is to be conducted on the treatment plant influent for surface waters and on the finished water for ground waters.
ability Monitoring '
CTOX (\ig/L) Applic
S
4)
|
cT
j
T t
r t
|
0
1
ON
1
oo
0
I
i
1
Month 4
i
o
1
T 1
1
O
2 UFCTOX monitoring may be required if a treatment plant is trying to qualify for a common source designation.
3 UFCTOX monitorhlg is to be conducted on the treatment plant influent for surface waters and on the finished water for ground waters.
& 4>
g g>
T* *-<
3 >:
O/ ca
m
a
bO
> co
< co
-J Q
&
b
S
a
(N
I
-------
Table 5-2 Treatment Study Applicability Data Reporting Form (page 2 of 2)
Information for Approved ICR Laboratory Conducting TOC Analysis
Laboratory name I " ~ I
ICR Laboratory ID Code^
Official Laboratory Contact Person
Name
Mailing address
Phone #
FAX#
E-mail address
-------
Table 5-3 Information Required With ICR Treatment Study Option Applications
[Application [Required tables
(Additional information
Applicability
data1
Individual study
Treatment
study avoidance
Single study for
multiple plants2
Joint treatment
study2
Buyout2
Grandfathered
treatment study
Treatment
study progress
reports
Table 5-1, Table 5-2
Table 5-1, Table 5-11 and
study plan
Table 5-1, Table 5-5
Tables 5-1, Table 5-6, Table 5-
10, Table 5-11 and study plan
Tables 5-1, Table 5-7, Table 5-
10, Tables 5-11 and study
plans
Tables 5-1, Table 5-8, Table 5-
10
Table 5-1, Table 5-9
Table 5-1, Table 5-12
None.
A cover letter.
A cover letter and data
demonstrating the use of full-scale
GAC, NF or RO if applicable.
A cover letter.
A cover letter and data
demonstrating that the plants
operate similar treatment trains.
A cover letter.
A cover letter and data
demonstrating that the study meets
the requirements of the ICR.
None.
1 Note the deadline for the treatment study applicability data, Table 5-2, is one month prior to all treatment
study option applications except for the grandfathered study application.
2 Note that Table 5-1 must be submitted for each plant involved these treatment study options.
-------
Table 5-4 Important Deadlines Related To The ICR Treatment Studies
Action
Start treatment study applicability monitoring
Submit grandfathered study application
Submit joint study letter of intent
Submit buyout letter of intent
Submit treatment study applicability data form
Submit treatment study avoidance application
Submit treatment study concept form
Submit single study/multiple plants application
Submit joint treatment study application
Submit buyout application
Submit grandfathered study data
Deadline for buyout payment
Start collecting data from the treatment studies
Submit first treatment study progress report
Submit second treatment study progress report
Submit third treatment study progress report
Submit fourth treatment study progress report
Submit final treatment study report
Deadline (no later than)
3 mos. after pub.
9 mos. after pub.
12 mos. after pub.
12 mos. after pub.
17 mos. after pub.
18 mos. after pub.
18 mos. after pub.
18 mos. after pub.
18 mos. after pub.
18 mos. after pub.
18 mos. after pub.
90 days after approval
23 mos. after pub.
24 mos. after pub.
27 mos. after pub.
30 mos. after pub.
33 mos. after pub.
38 mos. after pub.
Date
-------
Table 5-5 Treatment Study Avoidance Application
Utility name
Plant name
PWSID #
Plant ICR #
Application for avoidance is based on (check one of the following boxes):
(1) Average influent TOC is below 4.0 mg/L for a surface water plant1
(2) Average finished water TOC is below 2.0 mg/L for a ground water plant1
(3) Average distribution system THM4 < 40 ug/L and HAAS < 30 ug/L2
(4) The treatment plant is using full-scale GAC as described in the ICR3
(5) The treatment plant is using full-scale membrane treatment (NF or RO)3
1 The average TOC concentrations are calculated from the 12 months of applicability monitoring
2 The average DBF concentrations are calculated from the 12 months of applicability monitoring, and
the plant must be using only chlorine as the primary and residual disinfectant.
3 Include the full-scale plant data required under Section 141.142(a) of the ICR rule, demonstrating
the use of full-scale GAC or membrane technology.
-------
Table 5-6 Application For Multiple Plants Owned By A Single PWS And Operating On A
Common Source To Conduct A Single Treatment Study1
Plants Applying to Conduct a Single Common Source Study2
Plant name
Plant ICR #
PWSID #
Plant population served
Proposed Studies to be Conducted3'4
Technology to be investigated
(GAG or membranes)
Scale of study
(pilot or bench)
1 Only one application accompanied by one common source application, Table 5-10, should be submitted.
2 Each cooperating plant must submit Table 5-1 with this application.
3 A study concept form, Table 5-11, must be submitted for each proposed study to be conducted during
the common source study.
4 Only one treatment study is required for multiple plants owned by a single PWS and operating on
a common source. If the largest plant serves at least 500,000 persons, then the single study for multiple
plants must be a pilot-scale study.
-------
Table 5-7 Joint Treatment Study Application1
Plants Applying to Conduct the Joint Study2
Plant name
Plant ICR #
PWSID #
Plant population served
Proposed Studies to be Conducted3'4
Technology to be investigated
(GAC or membranes)
Scale of study
(pilot or bench)
1 Only one application accompanied by one common source application, Table 5-10, should be submitted.
2 Each cooperating plant must submit Table 5-1 with this application.
3 A study concept form, Table 5-11, must be submitted for each proposed study to be conducted during
the joint treatment study.
4 The number and type of studies required for a joint study is determined from the size and number of
cooperating plants as described hi Section 141.141 (e) of the ICR rule and Tables 3-2 and 3-3 of this manual.
-------
Table 5-8 Treatment Study Buyout Application1
Utility name
Plant name
PWSID #
Plant ICR #
Plant combined population served[
Proposed buyout fee2
Information for the Common Source Treatment Plant Conducting a Study3'4
Plant name
Plant ICR # I
Plant combined population served
Plant ground water population served
Plant surface water population served
PWSID#|
Official contact person
Name
Mailing address
Phone #
FAX#
E-mail address
ICR contact person
Name
Mailing address
Phone #
FAX#
E-mail address
1 One common source application, Table 5-10, should be submitted with this application.
2 The amount of funds to be contributed to the research fund is based on the total
plant population served and is described in Section 141.141(e) of the ICR rule.
3 At least one plant operating on a demonstrated common source must conduct a
treatment study for a buyout to be approved.
4 A plant serving fewer than 500,000 persons can buy out if a pilot-scale study is being
conducted on the common source; however, a plant serving greater than 500,000 persons
can not buy out if a smaller plant is conducting a bench-scale study on the common source.
-------
Table 5-9 Grandfathered Treatment Study Application
Utility name
Plant name
PWSID #
Plant ICR #
Plant combined population served [
What technology was investigated?
What was the scale of the study according to ICR?
What chlorination conditions were used for
DBP samples?
At what point in the full-scale plant was water
collected for the study?
Where is the first point that chlorine is added in
the full-scale plant?
Was the treatment study influent collected prior
to the addition of chlorine-based oxidants?
What was the duration of the study?
Study Description and Supporting Information
Attach a brief (no more than five page) description of the study which should include: (1) the
experimental procedure used, (2) the analytical methods used, (3) all processes used prior to
membranes or GAC, and (4) the operating and design parameters used in the study.
Also attach any reports resulting from the study that would help to demonstrate that the study
was consistent with the requirements of the ICR.
-------
es
0
s
1
a
^Q
f
Common Source D
Source water name
?
0>
1
fc"*
"*§
'S
CO
0
.S5
Q
1
o
CO
0
1
Source water type
ts Applying for a C
1
P-l
o
^
'-4-1
O
i *!
D "^
^9 cti
0 -S3
CO ^_>
0) en
1 1
g
"Si)
D ?3
« *^ s«*'
X X
J O O
>
X
UFCTO
4-*
*&
! > \ ' S 60
i 9 S 1
§ § 1 £
~ » 1 g.
1 i
^53 K
c^i C
^^ ^
x~> X-
-H « 4_I rt
3 -1 i 8
3 a & -S
5 cs1 §3 S
2 c 1 °
5 i^ k>
& ft? ^3
S £ §
m *~< *i<-
s 3 a
g |-^
- a i *
^ j i «
i s ^ ^
C3 ,U " .G
H tSi M IH
."- S ,3 -3
g ^ "ro
-§ a .98
S ^ | 1
3 & 'p «
* M
-------
Table 5-11 Study Concept Form1
General Study Information
Is this an individual or a joint study?
Will GAC or membranes be investigated?
Is this a pilot- or a bench-scale study?
At what point in the full-scale plant will water be
collected for the study?
Where is the first point that chlorine is added in
the full-scale plant?
Will the treatment study influent be collected
prior to the addition of chlorine based oxidants?
What is the average TOC concentration of the
treatment study influent?
How many tests will be required to evaluate
seasonal variability?
GAG Study Information
Carbon type and manufacturer to be investigated
Carbon particle diameter
Carbon column diameter
Membrane Study Information
Procedure to be used (RBSMT, SEBST, pilot)
Element size to be investigated
Model number and manufacturer of membrane #1
Molecular weight cutoff of membrane #1
Model number and manufacturer of membrane #2
Molecular weight cutoff of membrane #2
Study Plan
Attach a brief study plan (usually not more than two pages of text and two pages of figures)
which should include the equipment to be used, pretreatment to be used prior to GAC or
membranes, design parameters, operating parameters, whether or not seasonal variability need
to be evaluated and if seasonal variability can be evaluated in fewer than four quarters,
the parameters that will be investigated in lieu of seasonal variability.
1 One study concept form must be submitted for each study to be conducted.
-------
Table 5-12 Treatment Study Progress Report
General Information
Date:
Plant ICR #
Type of study (e.g., individual, joint, multiple plant)
Scale of study (i.e., pilot or bench)
Technology (i.e., membranes or GAC)
What was the study start date?
What is the anticipated finish date?
What is the anticipated date for the final report?
Pilot Study Information
What is the cumulative runtime?
What is the cumulative downtime?
Is the system currently operating?
Bench Study Information
How many quarterly runs have been completed?
Is seasonal variability being investigated?
GAC Study Information
Has headless been a problem?
What is the runtime to Stage 2 DBF breakthrough?
What is the anticipated runtime to 70% breakthrough?
Membrane Study Information
Has membrane fouling been a problem?
What is the nature of the foulant?
What cleaning procedure has been effective?
What is the cleaning frequency for the system?
Have the proposed Stage 2 MCLs been obtainable?
Sampling Information
Number of primary DBF samples collected to date?
Number of duplicate DBF samples collected to date?
-------
Part 2
Granular Activated Carbon Precursor Removal Studies
R. Scott Summers
Department of Civil and Environmental Engineering
University of Cincinnati
Stuart M. Hooper
Department of Civil and Environmental Engineering
University of Cincinnati
Seongho Hong
Department of Civil and Environmental Engineering
University of Cincinnati
-------
Notice
This Section was prepared for the EPA's Office of Ground Water and Drinking Water
under the guidance of the Drinking Water Technology Branch, Drinking Water Standards
Division (small purchase order #4W-1951-MASA). Members of the Technologies Working
Group who convened to provide technical support to the Regulatory Negotiating Committee
for the D/DBP Rule played a significant role in the development of the treatment study
approach, and hi some instances hi the preparation of the text for this Section.
2-ii
-------
Contents
Page
List of Figures 2-iv
List of Tables 2-v
1.0 Introduction 2-1
2.0 Background 2-3
2.1 Adsorption of Natural Organic Matter and
Disinfection Byproduct Precursors 2-3
2.2 Prediction of Field-Scale GAG Performance 2-5
3.0 Treatment Study Influent 2-11
4.0 Pilot-Scale GAG Test Protocol 2-13
4.1 Design 2-13
4.2 Operation 2-15
4.3 Sampling 2-16
5.0 Bench-Scale GAG Test Protocol 2-21
5.1 Design 2-21
5.2 Operation 2-26
5.3 Sampling 2-28
5.4 Design Example 2-31
6.0 Simulated Distribution System Chlorination Conditions 2-35
6.1 Selection of SDS Chlorination Conditions 2-35
6.2 Uniform Formation Conditions Approach 2-37
7.0 Cost Estimations 2-39
8.0 References 2-41
Appendix 2-A - Sampling Activated Carbon 2-45
Appendix 2-B - Uniform Formation Conditions For DBP Formation 2-47
Appendix 2-C - AWWA Standard For Granular Activated Carbon 2-49
2-iii
-------
Figures
2-1. Description of NOM breakthrough terminology
2-2. DOC breakthrough for three treated waters
2-3. Correlation between influent TOC and bed volumes to 50% TOC breakthrough
2-4. Effect of optimized coagulation on TOC breakthrough for Harsha Lake water
2-5. TOC and TTHMFP breakthrough for Palm Beach groundwater
2-6. TOC and TTHMFP breakthrough for Burnett Woods pond water
2-7. TOC and TTHMFP breakthrough for ozonated Ohio River water
2-8. THM species breakthrough behavior for ozonated Ohio River water
2-9. Application of constant diffusivity design (X=0) to DOC breakthrough of Fuhrberg
NOM
2-10. Application of proportional diffusivity design (X=1) to DOC breakthrough of
Fuhrberg NOM
2-11. RSSCT reproducibility with Ohio River water
2-12. UV absorbance (SAC) breakthrough comparison for Fuhrberg NOM
2-13. TOC breakthrough comparison for Ohio River water and a backwashed filter adsorber
2-14. SDS-TOX breakthrough comparison for Palm Beach GW
2-15. TTHMFP breakthrough comparison for Ohio River water and a backwashed filter
adsorber
2-16. SDS-HAA5 breakthrough comparison for Ohio River water
2-17. CHClBr2FP breakthrough comparison for Ohio River water and a backwashed filter
adsorber
2-18. SDS-BCAA breakthrough comparison for Ohio River water
3-1. Full-scale plant sample points for bench/pilot scale GAG tests
4-1. Pilot plant set-up
4-2. Typical GAG breakthrough behavior for natural organic matter
4-3. Pilot plant breakthrough and sampling
5-1. Impact of Reynolds number on breakthrough
5-2. RSSCT set-up
5-3. RSSCT breakthrough and sampling
6-1. Correlation between chlorine demand and TOC for two waters
2-iv
-------
Tables
2-1. Summary of RSSCT Application to NOM and DBF Formation Control
3-1. Sample Point and Required Pretreatment for GAC Tests (see Figure 3-1)
4-0. Sample of GAC Pilot-scale Systems
4-1. Pilot Plant Design Parameters
4-2. Estimated Sample Times for Pilot-plant GAC Column Operation
4-3. GAC Pilot-plant Sampling
4:4. Pilot Plant Results
5-0. Sampling of GAC Bench-scale Systems
5-1. RSSCT Proportional Difrasivity (X= 1) Design Equations
5-2. RSSCT Design for Full-scale 8x30 Mesh GAC (dLC = 1.6 mm)
5-3. RSSCT Design for Full-scale 12x40 Mesh GAC (dLC =1.1 mm)
5-4. Estimated Sample Times for RSSCT Operation
5-5. RSSCT Design Parameters
5-6. RSSCT Sampling - EBCTLC = 10 min
5-7. RSSCT Sampling - EBCTLC = 20 min
5-8. RSSCT Results - EBCTLC = 10 min
5-9. RSSCT Results - EBCTLC = 20 min
6-1. Dilution Study Parameters
7-1. Input Data for Cost Model
2-v
-------
-------
1.0 Introduction
The Information Collection Requirements (ICR) for Public Water Systems (Subpart M of
the National Primary Drinking Water Regulations), § 141.141(e) requires certain water
systems to conduct disinfection byproduct (DBF) precursor removal studies, termed herein as
treatment studies. The public water systems affected by this requirement are defined hi the
ICR rule, § 141.141(b) and the treatment studies are defined hi § 141.144 and Part 1 of this
document.
The objectives of the treatment studies are to generate representative process performance
data to be used for the development of treatment cost estimates at different levels of organic
DBF control. The two appropriate candidate technologies to be investigated by treatment
studies are granular activated carbon (GAC) adsorption and membrane separation processes.
Both processes have been shown to be effective for the removal of DBF precursors. The
treatment studies can be conducted with bench-scale and/or pilot-scale systems using effluent
from the full-scale treatment processes already hi place that remove DBF precursors.
The purpose of this part of the ICR Manual for Bench- and Pilot-Scale Treatment Studies
is to provide guidance hi conducting pilot- and bench-scale GAC treatment studies to meet the
requirements of the ICR rule. The target audience of this document are the parties that will
conduct or oversee these treatment studies, including water treatment plant operators,
scientists, and engineers, utility managers, and consulting engineers.
Based on the information presented hi Sections 2.0 (Figure 2-3), 4.0 and 5.0 it is
anticipated that waters having TOC values above 6 to 8 mg/L hi the GAC influent, will have
very early breakthrough and yield very short GAC run times to achieve DBF control at the
Stage 1 and 2 levels of the proposed Disinfectants/Disinfection Byproducts Rule. These short
run times will increase the operational costs of GAC treatment. Thus, it is suggested that
utilities with treated water TOC values above 6 to 8 mg/L: (a) evaluate DBF precursor
removal options such as enhanced/optimized coagulation or biological filtration prior to GAC
treatment, (b) pH control during GAC operation, or (c) run treatment studies with
membranes, as described in § 141.144 of the ICR and Part 3 of this document instead of
GAC.
This treatment studies manual is referenced in the ICR rule and contains specific
requirements of the rule. The ICR requirements listed in this manual include analytical and
reporting requirements for each type of study. This manual also provides guidance to assist in
setting-up and conducting these studies, and this guidance should not be interpreted as specific
requirements. In preparing this manual, an effort has been made to distinguish between specific
ICR requirements and guidance.
2-1
-------
-------
2.0 Background
2.1 Adsorption Of Natural Organic Matter And Disinfection Byproduct Precursors
Granular activated carbon can be used for adsorbing natural organic matter (NOM) and
and therefore, limiting the formation of disinfection byproducts. The effectiveness of the
GAC adsorption process is impacted by NOM characteristics which may be a function of the
origin and pretreatment of the source water. Specifically, NOM characteristics such as initial
concentration, molecular size and hydrophobicity affect the adsorbability, as does pH, and
pretreatment processes such as coagulation and ozonation.
For drinking water purposes, NOM is typically quantified by total organic carbon (TOC)
or dissolved organic carbon (DOC). After sedimentation and filtration there is usually very
little difference between TOC and DOC and they are often used interchangeably. UV
absorbance, commonly measured at a wavelength of 254 nm, can also be used to assess a
fraction of the NOM. The component of NOM that reacts to form DBFs, termed DBF
precursors, can be assessed indirectly by measuring the amount of DBFs formed under
specific disinfection conditions of dose, time, temperature and pH. Formation potential (FP)
conditions are used to approximate the maximum DBF formation, while simulated distribution
system (SDS) and uniform formation conditions (UFC) are used to represent the level of
DBFs formed under conditions encountered in practice.
A breakthrough curve that represents the adsorption behavior of NOM is shown in Figure
2-1. To facilitate comparisons of breakthrough behavior of waters with different influent
(initial) concentrations, the effluent concentration, c, is normalized by the influent
concentration, c0. The amount of water treated is expressed as operation time, or as
throughput in number of bed volumes (BV) treated at a given empty bed contact time
(EBCT):
BV= t/EBCT (2.1)
where t is the operation tune or
BV= treated water volume/Vb (2.2)
where Vb is the GAC bed volume. Expressing the operation time or amount of water treated
as throughput in bed volumes treated, facilitates comparisons of results from GAC adsorbers
with different EBCTs.
For most natural waters the influent concentration varies during a short GAC run,
typically by 10 to 20% as shown in Figure 2-1, where the instantaneous influent
concentration is normalized by the average influent concentration. However, for surface
waters the influent concentration can vary by as much as a factor of two to three when the
GAC run is long enough to cover seasonal trends. The immediate breakthrough is the
effluent fraction that is present in the first sample. If this level remains constant for any
2-3
-------
length of time, as shown in Figure 2-1, then the immediate breakthrough represents the
nonadsorbable NOM fraction. Typically, the nonabsorbable fraction is 5 to 20% of the
influent NOM as measured by DOC, and 0 to 10% as measured by UV absorbance. Initial
breakthrough is the point at which the effluent concentration increases beyond the
nonabsorbable fraction. It characterizes the first portion of the breakthrough of the
adsorbable fraction. Pseudo steady-state represents a section in the breakthrough curve in
which little if any changes occur in the effluent concentration at a constant influent
concentration. NOM removal in this region is due to biodegradation and/or slow adsorption.
The DOC breakthrough of three treated waters with different influent concentrations and
by GAG columns with different EBCTs is shown hi Figure 2-2. All three waters display an
immediate breakthrough of about 5 to 10%, but the breakthrough behavior thereafter is very
different: earlier exhaustion of the GAC occurs with increasing influent DOC concentration.
The mid-point, or 50%, breakthrough occurs after 2000, 4000 and 12,000 bed volumes for
Palm Beach ground water, Mississippi River water and Ohio River water, respectively. For
Mississippi River water, a plateau is reached after 12,000 bed volumes. Sampling was not
continued long enough to establish a plateau for the other two waters. This range of
adsorption behavior represents that found hi practice.
To ascertain whether differences hi influent TOC concentrations impact breakthrough
behavior, the number of bed volumes to 50% breakthrough (BV50) was correlated to the
influent concentration for data from 23 TOC breakthrough curves. The correlation is shown
in Figure 2-3 (Summers et al., 1994b) and a best-fit curve yields the following equation:
BVSO = 21,700-TOC -13 (2.3)
These data are for bituminous coal based GACs and are from 16 different water sources from
around the world, including river waters, lake/pond waters, and ground waters. As can be
seen, breakthrough behavior is highly correlated, rz = 0.88, to influent concentration. While
the differences hi the NOM adsorbability of the sources waters should not be ignored, the
influent concentration gives a good estimate of the effectiveness of GAC.
Many drinking water utilities will be optimizing their coagulation processes to meet the
enhanced coagulation requirements of the proposed Disinfectants/Disinfection Byproduct
Rule. To achieve better removal of DOC by coagulation lower pH values are often used. If
the GAC system is used after coagulation then the lower pH values will also enhance the
removal of TOC by GAC as shown hi Figure 2-4 (Hooper et al., 1995). The increased
effectiveness of GAC after enhanced coagulation is due to both lower pH and lower influent
TOC values, as well as changes hi the molecular size and hydrophobicity of the NOM.
Semmens et al. (1986a; 1986b) have also shown the positive impact of lower pH and
increased coagulant dose on GAC performance for Mississippi River water. The
breakthrough data from GAC columns treating enhanced coagulated waters at low pH, below
7.0, were not utilized in the development of the correlation shown in Figure 2-3.
2-4
-------
Of particular importance is the use of GAC to remove NOM for the control of DBF
formation. The breakthrough behavior of TOC and total trihalomethane formation potential
(TTHMFP) for three different waters is shown in Figures 2-5, 2-6, and 2-7. The
breakthrough behavior of TOC seems to be a good, but somewhat conservative indicator for
that of TTHMFP. The rapid small-scale column test (RSSCT) was utilized in these studies.
The scaling factor (SF) is the ratio of the GAC particle size used hi the full-scale column to
that in the RSSCT and establishes the relationship between the EBCT of the full-scale system
(EBCTLC) and the RSSCT. These relationships are defined hi Section 5.0.
The formation of the four different trihalomethane (THM) species after GAC adsorption is
shown hi Figure 2-8 for ozonated Ohio River water. In the influent water the formation
potential is highest for chloroform. However, after GAC treatment chloroform and
dichlorobromomethane are formed at the same level and dibromochloromethane and
bromoform are formed at proportionally higher levels compared to chloroform. This shift in
the relative THM speciation to the more brominated species is thought to be due to shifts in
the bromide to TOC ratios (Summers et al. 1993). Since bromide is not removed by GAC,
the bromide to TOC ratios in the effluent at the beginning of the GAC run are high, and this
leads to the formation of proportionally more brominated THM species after chlorination. As
the TOC breaks through the bromide to TOC ratio approaches that of the influent and
chloroform formation will again dominate.
2.2 Prediction Of Field-Scale GAC Performance
A variety of methodologies have been applied for the prediction or simulation of the
breakthrough behavior of full-scale GAC columns. These include pilot-plant columns,
numerical modeling using data from equilibrium and kinetic lab tests, non-scaled mini-
column tests and the RSSCT. Of these pilot-plant columns and the RSSCT are the most
commonly used. The two dominant factors that control the breakthrough in GAC columns
are the adsorption capacity (extent of adsorption) and the rate of adsorption (adsorption
kinetics). These two parameters must be accurately assessed by any method that successfully
emulates full-scale GAC behavior. Characterizing the different GAC performance
methodologies in terms of operation tune and numerical modeling requirements is useful.
Under most circumstances, minimizing the operation time of the evaluation system and
reducing the modeling component will lead to lower costs, but reliability and accuracy must
not be sacrificed. For evaluation methodologies utilizing column contactors, the GAC
particle size and the column length or EBCT are indicators of the operation tune.
Consideration of the above listed criteria has led to the use of pilot columns and RSSCTs as
the most commonly chosen methods of GAC performance evaluation.
Pilot columns which utilize the same GAC and influent water as the full-scale system have
been shown to be accurate and reliable predictors of the breakthrough behavior in full-scale
columns both in terms of the capacity and rate of adsorption (Wood and DeMarco 1980,
DeMarco et al. 1983, DeMarco and Brodtman 1984, Chrobak, Kelleher, and Suffet 1985,
and Speth et al. 1989). They also have the advantage of not requiring the use of numerical
models for data interpretation. Because they must utilize column lengths and GAC particle
2-5
-------
sizes that are the same as full-scale columns, pilot columns have the same operation time as
the full-scale system. This equivalent operation time, coupled with the capital investment
required for the pilot system, can lead to costs higher than other methods of evaluation. Pilot
columns have the distinct advantage over other methods hi that the biodegradability of the
compound of interest can be assessed, as can the impact of seasonal variability, if the columns
are operated on a long-term basis.
Another predictive approach to breakthrough behavior is to assess the equilibrium
adsorption capacity by the bottle-point method, commonly referred to as the isotherm test,
and to estimate the adsorption kinetics with completely mixed batch reactors or differential
column batch reactors. Data from these tests are used as input to a fixed-bed numerical
model for predicting the breakthrough. Limited success has been achieved using this
approach to emulate full-scale systems due to inaccuracies hi the assessment of both the
adsorption capacity and kinetics hi complex mixtures, such as those hi natural waters
(Sontheimer, Crittenden and Summers, 1988). This approach is attractive due to the use of
small GAC particle sizes, which leads to short operation tunes. However, its utility is limited
due to the constraints of requiring a numerical model for the evaluation of both kinetic
parameters and adsorption capacity. More importantly, this approach does not adequately
assess any long-term changes hi the adsorption capacity or kinetics due to the presence of a
background matrix nor does it assess the impact of biological activity on NOM removal.
Mini-columns utilize small particle and column sizes to shorten the operation tune to a
small fraction of that hi the full-scale or pilot-scale column. When they are not scaled to the
dimensions of a full-scale adsorber, mini-columns are limited to the assessment of GAC
adsorption capacity. These columns are usually operated with particle sizes of 0.05 to 0.1
mm, column lengths of 2 to 5 cm, and flow rates which yield operation tunes of a few hours.
While only the capacity data can be directly compared to full-scale data, this approach is very
useful for the relative comparison of different GACs. Thus, they are a good tool for
selectively screening GACs to determine the most efficient GAC prior to any additional
studies, such as pilot columns. An advantage this approach holds over batch isotherm tests is
that the competitive interactions between adsorbing compounds are assessed under dynamic
conditions similar to the full-scale system (Summers and Crittenden, 1989).
To compare mini-column results to those from the full-scale system, kinetic parameters
which characterize the external and internal mass transfer and a numerical model are
required. In this approach mini-columns are used to assess the adsorption capacity and short
fixed-bed test results are used to yield the external and internal mass transfer coefficients.
These parameter values are then used hi diffusion-adsorption models to predict the
breakthrough behavior.
Small-scale columns are mioi-columns hi which similitude to full-scale GAC adsorbers
has been maintained. The relationship between the particle size, column length or EBCT,
hydraulic loading and operation tune of the small- and large-scale columns (pilot or full) is
determined through the use of dimensional analysis. Successful application of the small-scale
columns produces breakthrough curves which are equivalent to those of a full- or pilot-scale
2-6
-------
adsorber. The method was pioneered by Prick (1982) and further developed and extensively
applied by Crittenden and co-workers (Crittenden, Berrigan, and Hand 1986; Crittenden et al.
1987, 1989 and 1991: Hineline, Crittenden, and Hand 1987) leading to the rapid small-scale
column test.
The RSSCT has a number of advantages over the other methods used to predict or
simulate full-scale GAC performance. Depending on the conditions, this method can be
conducted in less than one percent to 15 percent of the time that is required for a pilot- or
full-scale column study. Because of the short operation time of the RSSCT, several column
runs can be conducted in the tune required to complete one pilot column run. This allows for
the optimization of design and pretreatment options. The volume of influent water required is
normally small enough that an adequate volume of water can be transported to a laboratory,
thus eliminating the need for a field study and the associated costs. As compared to the
predictive model approach, separate experimental efforts to evaluate the adsorption capacity
and kinetics are not required nor is the use of numerical or analytical models.
A potential drawback of the RSSCT is the possible difficulty in obtaining a representative
batch of influent water. Since there is often natural variability hi the influent water quality,
the use of a batch influent may yield misleading results as compared to the long term
operation of full- or pilot-scale systems, which embody influent water variations. Thus, the
selection of a representative sample to serve as the influent to the RSSCT is paramount. The
effect of seasonal water quality variations can be assessed by conducting these tests several
tunes a year. Furthermore, organic matter removal hi a GAC column due to long-term
biodegradation is not simulated by the RSSCT, since the microorganisms may not have
enough time to acclimate.
Another consideration in using the RSSCT is the dependence of intraparticle diffusivity on
GAC particle size. The design equations were originally developed with the assumption of
constant diffusivity (CD). However, a number of small column tests and batch kinetic tests
have shown that the diffusion coefficient decreases proportionally with decreasing particle
size and a proportional diffusivity (PD) design approach has been developed. Both
approaches and a review of the method are described hi Summers and Crittenden (1989) and
Crittenden et al. (1991).
A summary of the application of RSSCTs to predict the field-scale control of NOM and
DBF formation is shown in Table 2-1. About thirty comparisons of RSSCT results to those
of pilot- or full- scale columns have been made for the adsorption of organic matter as
measured by TOC, DOC or UV absorbance.
Summers et al. (1989) investigated both the PD- and CD-designed small-scale columns
with four parallel columns of different particle sizes. In this study, NOM that had been
extracted from a ground water (Fuhrberg, Germany) was added to tap water, which had been
previously treated by GAC to remove adsorbable compounds. In addition to characterization
by DOC, UV-absorbance at 254 nm was also measured and reported hi this part of this
manual as the spectral absorption coefficient (SAC). As shown*hi Figure 2-9, the CD-
2-7
-------
designed columns yielded earlier breakthrough as particle size decreased. The results from
the PD-designed columns well-predicted the breakthrough from the pilot column, dp =1.58
mm, as shown in Figure 2-10.
The results from three parallel RSSCTs with the same batch influent source (Ohio River
water) are shown in Figure 2-11 (Namuduri, 1990). As can be seen, the reproducibiliry of
this test is very good. The use of UV absorbance to characterize the adsorption of the
extracted ground water NOM also proved successful, as indicated by the SAC results in
Figure 2-12.
In the study by Wallace et al. (1988) and McGuire et al. (1989) five different raw water
sources were used: Colorado River, California State Project, Ohio River, Mississippi River,
and Delaware River. For these waters, the PD design yielded good comparisons between the
DOC results of the small-scale and the pilot columns. Good predictability of DOC results
was also found for Lake Gaillard water at two EBCTs (Malcolm Pirnie, 1990).
Summers et al. (1992) investigated the use of RSSCTs with three different GAC types and
three water sources: Delaware River, Palm Beach Florida groundwater and Ohio River. The
PD design was used in all cases. The RSSCT adequately predicted pilot- or full- scale results
in all nine cases for the breakthrough of TOC and UV absorbing substances. In all of the
above cases biodegradation of organic matter did not preclude the use of the RSSCT. The
application of the RSSCT to predict the TOC breakthrough behavior of a filter adsorber that
was regularly backwashed is shown in Figure 2-13 (Summers et al., 1994a). The pilot-scale
breakthrough was well-predicted by the non-backwashed RSSCT, except in the initial
breakthrough section. Although the RSSCT data shown is from a column that was not
backwashed, a parallel study with backwashed RSSCTs showed little impact of backwashing
(Hong etaL, 1994).
Only a few studies, Summers et al. (1992, 1994a, 1995), Cummings and Summers (1994)
and Metz, Summers and DeMarco (1993), have also investigated the use of RSSCTs for the
prediction of DBP control. These are summarized in Table 2-1 (Studies 5 thru 8).
Cummings and Summers (1994) found the RSSCT to well predict the pilot-scale breakthrough
behavior of THM and TOX precursors hi a Florida groundwater for three different GACs.
An example for SDS-TOX is shown hi Figure 2-14. Some problems were encountered for
THM species that were formed at concentrations below 5 ug/L. TTHM formation potential
results from a filter adsorber are shown hi Figure 2-15. Like that found for TOC for this
water source (see Figure 2-13), the RSSCT did not well-predict the initial breakthrough
results, but the immediate breakthrough and mid-point breakthrough were well-predicted.
Metz, Summers and DeMarco (1993) used RSSCTs to predict the full-scale control of non-
THM DBFs, as well as THM and TOX in Ohio River water with reactivated GAC. The
RSSCT predicted slightly earlier breakthrough in the initial breakthrough portion of the
breakthrough curve for formed HAAS and chloral hydrate, as shown in Figure 2-16 for
HAAS. The RSSCT results from individual formed DBP species can also be compared to
those from field-scale GAC columns. Examples are shown hi Figure 2-17 for
dibromochloromethane and inJFigure 2-18 for bromochloroacetic acid (BCAA). The results
2-8
-------
from Figure 2-18 are from a current study in which the applicability of RSSCTs to predict
field-scale behavior is being evaluated for four waters and six parameters as listed in Table 2-
1 (Summers et al., 1995). One important conclusion of that study is that obtaining a
representative sample for the RSSCT is critical to the success of the RSSCT for predicting
field-scale breakthrough behavior.
While limited to nine raw water sources and 30 direct comparisons to field adsorbers, the
RSSCT using the PD design approach has been shown to be an effective method for
predicting NOM breakthrough in pilot- or full-scale GAG columns. Unlike other predictive
tools, the RSSCT does not require numerical modeling nor additional capacity or kinetic tests,
yet can be completed in a small fraction of the time and costs of a pilot-scale study.
Therefore, the RSSCT may be very useful hi gathering representative information that will
allow for the development of treatment cost estimates for different levels of organic DBF
control by GAG.
2-9
-------
-------
1.2
1.0
2 0.8
50
0)
g 0.6
8
0)
.y 0.4
-------
5 30000
I
{{f 10000
I
h-
vp
3
8
TJ
CD 1000
i 1 1 1 1 r
16 source waters
Bituminous coal based GAC -
BVSO = 21,700TOC0-1'3
r2 = 0.88
1
2 3 4 56789 10
Influent TOG, TOC0 (mg/L)
Figure 2-3 Correlation between influent TOG and bed volumes to 50% TOC break-
through
3.0
2.5
2.0
I"
1.0
. TOC
c0 (mg/L)
Conventional treatment influent 3.0
Optimized coagulation influent 2.2
Conventional
treatment
50 100 150
Scaled operation time (days)
200
Figure 2-4 Effect of optimized coagulation on TOC breakthrough for Harsha Lake
water
-------
£
o
§
g
1000
2000 3000 4000
Throughput, BV (bed volumes)
5000
6000
Figure 2-5 TOO and TTHMFP breakthrough for Palm Beach groundwater
1.0
0.8
o
2 °-6
I
8 0.4
8
a
0.2
-------
1.0
.cp 0.8
o
.1
I °-6
o
I
CO
I
0.4
0.2
0.0
Ozonated Ohio River water
ITOC0-1.75mg/L
OTTHMFP0= 133 ng/L
O O'
8 min
GAC: F400
5000 10000
Throughput, BV (bed volumes)
15000
Figure 2-7 TOC and TTHMFP breakthrough for ozonated Ohio River water
CHCl3=49.7ng/L
CHCIjBr- 35.7 ng/L
CHClBr2- 23.7 ng/L
5000 10000 15000
Throughput, B V (bed volumes)
20000
Figure 2-8 THM species breakthrough behavior for ozonated Ohio River water
-------
if
.1
I
§
I
T3
(D
^
OJ
o
1.0
0.9
0.8
0.7
0.6
0.5
0.4
0.3
0.2
0.1
0.0
Constant Diffusivity Design
/1A
Run d(mm) I (m) v(m/h) C0(mg/L)
4D 1.58
3C 0.85
1A 0.26
1.3
0.7
0.25
5.0
9.3
31.2
8.03
8.72
8.93
GAC: F100
0
1000 2000 3000
Throughput, BV (bed volumes)
4000
5000
Figure 2-9 Application of constant diffusivity design (X = 0) to DOC breakthrough
of Fuhrberg NOM (Summers et al., 1989)
o
8
o
o
0)
.N
1.0
0.9
0.8
0.7
0.6
0.5
0.4
0.3
0.2
0.1
0.0
'_ Proportional Diffusivity Design Ds = f(d )
3A
/:
Run d(mm) I (m) v(m/h) C0 (mg/L)
1A 0.26
3A 0.85
4A 1.58
0.25
0.25
0.25
31.2
9.3
5.0
8.93
8.72
8.03
GAC: F100 -
1000 2000 3000 4000
Throughput, BV (bed volumes)
5000
Figure 2-10 Application of proportional diffusivity design (X = 1) to DOC break-
through of Fuhrberg NOM (Summers et al., 1989)
-------
Jl.VI
O"
1> 1.5<
u
g"
J3
£ 1.0
c
0)
o
c:
8
8 °-5
I
0.0
1 1 1 1 I 1 1 1
"
^
o
.
-
~
-
.
-
^
mBB
-*
",,,,!,,,
1 1 ' ' ' ' 1 ' ' ' ' 1 ' ' ' ' 1 ' ' '
EBCTLC=15min
GAG: F400
SF= 5.5
o o
A A"
^ A"A A A A
*
V
H mM ^~
HV
""A A
. _
>A4*« *
L o Influent
A RSSCT 1
RSSCT 2
RSSCT 3
i 1 , , , , 1 , , , , 1 , , , , 1 , , ,
'
_
-
_
-
-
_
_
-
~
-
-
,
10
50
20 30 40
Operation time, t (days)
Figure 2-11 RSSCT reproducibility with Ohio River water (Namuduri, 1990)
60
'Proportional Diffusivity Design Ds=f(dp)
1000 2000 3000 4000
Throughput, BV (bed volumes)
5000
Figure 2-12 UV absorbance (SAC) breakthrough comparison for Fuhrbere NOM
(Summers etai, 1989)
-------
Pilot-scale, C0= 1.74mg/L
OPilot-scale breakthrough
n RSSCT breakthrough, C0= 1.39 mg/L
5000 10000 15000
Throughput, BV (bed volumes)
Figure 2-13 TOC breakthrough comparison for Ohio River water and
a backwashed filter adsorber (Summers et al., 1994a)
20000
EBCTLC= 15 min
GAC: HD4000
SF= 5.5
Pilot-scale, C0= 511 jig/L
Pilot-scale breakthrough
RSSCT breakthrough, C0= 713
t t I I I I 1 I I I I I I I I I t t I t I II 1 1
Ill 1
20 30 40 50
Scaled operation time (days)
60
70
Figure 2-14 SDS-TOX breakthrough comparison for Palm Beach GW
(Cummings and Summers, 1994)
-------
1.4
1.2
c" 1-°
o
S 0.8
I
o 0.6
J 0.4
0.2
0.0
TTHMFP
8 min
GAC: F400
SF= 8.0
Pilot-scale, C0= 133 ng/L
OPilot-scale breakthrough
Q RSSCT breakthrough, C0= 112 jig/L
0 5000 10000 15000
Throughput, BV (bed volumes)
Figure 2-15 TTHMFP breakthrough comparison for Ohio River water and
a backwashed filter adsorber (Summers et al., 1994)
20000
1.0
£ 0.8
I
I 0.6
8
§ o-4
N
1=
2 0.2
iii|iiir
SDS-HAA5
-i i i i
i i i r
-iiir
Field scale breakthrough, c0= 40 jxg/L
RSSCT breakthrough, c0= 53 ug/L
EBCTLC= 15 min
GAC: F400 R.
SF=19.0
0.0 I'''i-
_ii i i
i i i i
i t
_ii i i
0 50 100 150 200
Scaled operation time (days)
Figure 2-16 SDS-HAA5 breakthrough comparison for Ohio River Water
(Metz, Summers, andDeMarco, 1993)
250
-------
Pilot-scale, C0= 20.9 ug/L
Pilot-scale breakthrough
RSSCT breakthrough, GO= 23.7 ug/L
1
20000
5000 10000 15000
Throughput, BV (bed volumes)
Figure 2-17 CHClBr2FP breakthrough comparison for Ohio River water and a
backwashed filter adsorber
0°
o
1
8
o
o
1
ro
Full-scale influent 13.0
RSSCT influent 14.4
Full-scale effluent
RSSCT effluent
0.0
25
175
50 75 100 125 150
Scaled operation time (days)
Figure 2-18 SDS-BCAA breakthrough comparison for Ohio River water
200
-------
Table 2-1 Summary of RSSCT application to NOM and DBF formation control.
Study
1
2
2
2
2
2
3
4
5
6
7
8
8
8
8
Water Source
Fuhrberg NOM
Colorado River
State Project, CA
Ohio River
Mississippi River
Delaware River
Lake Gaillard, CT
Delaware River
Palm Beach GW
Ohio River
Ohio River
Mississippi River
Ohio River
Passaic River, NJ
Lake Gaillard, CT
Treatment
extracted
conv
conv
conv
conv
conv
direct fil
C>3/conv
softened/
O3/biofil
conv
C>3/conv
conv/C>3
conv
conv
direct fil/
03
TOC
(mg/L)
8.0 - 8.9
2.1
2.2
2.2
2.7
2.7
1.8
2.5
6.6
1.8
1.4
2.8
2.1
3.2
1.5
EBCTLC
(min)
1.6, 3.0, 16
15, 30, 60
15
15
20
15
5.4, 15
15 (3)*
15 (3)*
15 (3)*
8
6.2
15
20
15
dsc
(mm)
0.26, 0.85
0.21
0.21
0.21
0.21
0.21
0.21
0.20
0.20
0.05
0.20
0.11
0.08
0.11
0.20
Parameters
TOC, UVA
TOC
TOC
TOC
TOC
TOC
TOC
TOC
TOC, UVA,
TOX, THM
TOC, UVA, TOX,
THM, HAA, CH
TOC, UVA, THM
TOC, UVA, TOX,
THM, HAA, CH
TOC, UVA, TOX,
THM, HAA, CH
TOC, UVA, TOX,
THM, HAA, CH
TOC, UVA, TOX,
THM, HAA, CH
1) Summers et al., 1989
3) MP/SCCRWA, 1990
5) Cummings and Summers, 1994
7) Summers et al., 1994a
*threc GAG types
2) Wallace etal., 1988
4) Summers et al., 1992
6) Metz, Summers, andDeMarco, 1993
8) Summers et al., 1995
-------
3.0 Treatment Study Influent
The ICR Rule, § 141.144(b), states that "The treatment study shall be conducted with the
effluent from treatment processes already in place that remove disinfection byproduct
precursors and TOC." The intent of the treatment study is to assess the additional removal of
DBF precursors by GAC. However, the influent to the GAC must be taken prior to the first
point of use of oxidants or disinfectants that would form chlorinated DBFs, § 141.144(b).
This is to insure that the DBF precursors are not reacted prior to the GAC treatment and that
oxidant residuals are not present hi the GAC influent, which would lead to oxidation of the
GAC. If the use of these oxidants or disinfectants precedes any full-scale treatment process
that removes DBF precursors, then bench- or pilot-scale treatment processes that represent
these full-scale treatment processes are required to generate treated water that has not been
exposed to oxidants or disinfectants that form chlorinated DBFs for use in the GAC treatment
study.
This presumes that the most appropriate location in the treatment tram for the advanced
precursor removal process under study is after the in-place treatment processes. However,
there may be cases where the simulation of some full-scale treatment process may not be
appropriate or necessary. For example, since the RSSCT uses cartridge filtration as a routine
test pretreatment step, it would not be necessary to simulate granular media filtration for
those procedures. Also, utilities may investigate the optimization of existing treatment
processes, like enhanced/optimized coagulation, or the addition of new DBF precursor
removal processes, like biological filtration prior to the GAC treatment studies.
A schematic diagram of three possible scenarios is shown hi Figure 3-1 and explained in
Table 3-1 for a full-scale conventional treatment system. The sample points for the bench- or
pilot-scale systems must precede the addition point of the chlorine based oxidant or
disinfectant. For example, if pilot-plant studies are to be conducted and the
oxidant/disinfectant is added at Point A, then water for the treatment study must be taken at
Point I and a continuous flow pilot plant with coagulation, sedimentation and filtration
processes must be used prior to the GAC columns. These pilot plant processes must produce
a water quality that represents that from the full-scale system. It is critical that the pilot plant
system be cleaned or in use long enough that no leaching from or adsorption to the materials
of construction occur, since this would affect the TOC or DBF formation. If the
oxidant/disinfectant is added at Point B, then pilot-scale continuous-flow filtration must be
provided prior to the GAC columns.
If the bench-scale RSSCT is used for the GAC treatment study, then batch bench-scale
coagulation/fiocculation, sedimentation and filtration can be used if treatment study water is
taken at Point I. Like that for the pilot-scale system, these bench-scale batch processes must
produce a water quality that represents that from the full-scale system. As will be shown hi
Section 5.0, the RSSCT requires 200 to 400 liters (50 to 100 gallons) per run for most
situations. This volume of water can be batch coagulated, flocculated and settled in a
laboratory using clean 55-gallon drums or other appropriately sized clean vessels. It is
Critical that the vessels be clean and no leaching from or adsorption to the vessel walls
2-11
-------
occurs, that would affect the TOC or the DBF formation. Because of stability problems
associated with storing ozonated waters, and acclimation times for biotreatment, plants that
want to investigate ozonation directly before GAG application should do so at the pilot scale.
If a bench-scale GAG treatment study is to be used for a plant with post-ozonation and no
biotreatment, then a batch of water should be taken prior to ozonation and ozonated and
biotreated at the bench-scale, prior to the bench-scale GAG run.
RSSCT columns are often prone to the development of rapid pressure (headloss) buildup
as small particle sizes of GAG are used. Membrane or cartridge filtration of the RSSCT
influent diminishes the rapid pressure buildup to an acceptable level. Membranes, such as
glass fiber filters and cartridge filters with 1.0 jam nominal pore openings, have been shown
to be effective in reducing headloss. In all cases the filtration system must be extensively
cleaned to insure that no organic matter is released into the treatment study water. This
should be checked using distilled water with a low organic matter content. This membrane or
cartridge filtration can also serve as the bench-scale filtration treatment process that emulates
the full-scale filtration process, if the treatment study water was taken at Points I or II.
Figure 3-1 illustrates the location of the treatment study influent water for a conventional
treatment system, i.e., coagulation, fiocculation, sedimentation and filtration. If other
treatment processes that remove the DBF precursors or affect the formation of DBFs are used
in the full-scale plant and the oxidant/disinfectant addition precedes them, then bench- or
pilot-scale treatment processes that represent these full-scale treatment processes are required
prior to the GAG treatment study. These processes include, but are not limited to, softening,
recarbonation, aeration, ozonation, powdered activated carbon addition and pH adjustment.
As discussed in Section 5.0, it is critical that the batch of water taken from the full-scale
plant for use as the RSSCT influent be representative of the water quality of the season of
interest. The batch influent water should be immediately evaluated, on-site if possible, for
TOC, UV254, pH, alkalinity, hardness (total and calcium), ammonia and bromide, prior to use
as the RSSCT influent. The dissolved oxygen levels should also be checked prior to running
the RSSCT as extremely high or low deviations from the field-scale value may affect TOC
adsorption.
Part 1, Section 4.7 requires that design information for all full-, pilot- or bench-scale
pretreatment processes that precede the bench/pilot systems be reported. Any costs incurred
due to the addition of pretreatment processes to an existing plant should be estimated and
reported. In addition the point in the full-scale treatment train in which water for the pilot- or
bench-scale treatment systems should be noted. The flow rate for all pilot-scale processes
preceding the pilot-scale GAG adsorber should be reported. The flow rate for continuous
systems or the volume for batch systems for all bench-scale processes preceding, the bench-
scale GAG adsorber should be reported.
2-12
-------
Point of chlorine
based disinfection
Raw water
Sample point
* I
Coagulation
Sedimentation
B
II
Filtration
-*- III
Figure 3-1 Full-scale plant sample points for bench/pilot scale GAC tests
Table 3-1. Sample point and required pretreatment for GAC tests (See Figure 3-1)
Addition point of
chlorine based
chemical
A- before
coagulation
sedimentation
B- after
sedimentation
C- after filtration
Required sample
point1
I
II
in
Required pilot
plant continuous
flow pretreatment2
coagulation
sedimentation
media filtration
media filtration
none
Required bench
scale batch
pretreatment
coagulation2
sedimentation2
1.0 ^irn membrane
filtration3
1.0 jim membrane
filtration3
1.0 pm membrane
filtration3
1) In all cases samples for GAC testing must be taken before the addition point of
chlorine-based chemicals
2) Pilot and bench-scale pretreatment should be designed to simulate the full-scale
treatment processes. All full-scale processes that impact the adsorbability of organic
matter should be simulated by pilot and bench-scale processes, e.g., pH adjustment,
aeration, softening, recarbonation, and ozonation.
3)Filtration with membranes or cartridges with 1.0 urn nominal pore openings is to be
used prior to all RSSCTs. This use simulates full-scale media filtration for purposes of
RSSCT pretreatment and minimizes pressure build-up in the RSSCT column.
-------
-------
4.0 Pilot-Scale GAC Test Protocol
4.1 Design
As discussed in Section 2.0, well designed and operated pilot-scale GAC columns have
been found to yield results that are comparable to those of full-scale GAC systems. The GAC
used in the pilot columns should be representative of that to be used in the full-scale columns
with respect to particle size and activity. The GAC selected for use in the pilot column must
meet the AWWA Standard for Granular Activated Carbon ANSI/AWWA B604-90 (Appendix
2-C). This standard describes the sampling and characterization of the GAC. A
representative batch of GAC shall be obtained from the manufacturer and care should be
taken in choosing a sample from this batch for the pilot column, as it will contain a range of
particle sizes. The smaller particles in a batch will tend to yield better removal performance
because the mass transfer kinetics are faster and the capacity may be higher. However,
smaller particles will lead to faster buildup of headless. A sample tube, riffle splitter, sample
reducer or coning and quartering technique can be used to obtain a representative sample
(Randtke and Snoeyink, 1983). The coning and quartering technique is described in
Appendix 2-A.
Typically, 8x30 or 12x40 mesh (US Std.) sized GAC is used in water treatment, although
custom sized GAC can be obtained. These GACs have average particle diameters (APDs) of
about 1.6 and 1.1 mm, respectively. To minimize the chances of channeling hi a GAC
column, the inner column diameter should be at least 25 tunes greater than the APD of the
GAC. The ICR Rule specifies the use of columns with a minimum inner diameter of 2.0
niches (51 mm). The ICR Rule requires that GAC columns with EBCTs of 10 and 20
minutes be tested. The pilot testing shall include the water quality parameters listed in Table
4-0.
An example of a typical pilot plant set-up that can be used to gather the required
information is shown in Figure 4.1. A filter with inert media, e.g., sand, anthracite or
garnet, is also shown in Figure 4-1, although if the treatment study water is taken at Point III
(Figure 3-1) pilot-scale filtration is not required. If the GAC column inner diameter is 6
inches or less, then all GAC column values and piping material hi contact with the water shall
be limited to glass, Teflon or stainless steel. Glass columns are recommended so that the
height of the GAC or filter media can be easily monitored during operation and backwashing.
A plastic safety shield should be used for glass GAC columns. If the GAC column inner
diameter is greater than 6 inches or the cross sectional surface area is greater than 0.2 square
feet, then other material of construction can be used. However, the column's piping and
valves must be cleaned or in use long enough that no leaching from or adsorption to the
materials of construction occur, since this would affect the TOC or DBF formation. This
lack of impact must be demonstrated by a blank run with no GAC in the column. Provisions
should be made for backwashing the filter and the GAC columns. Appropriate valving and a
non-chlorinated backwash water supply are needed. The columns should be at least 60%
longer than the media or GAC bed depth to allow for the expansion of the bed during
backwashing without loss of media. Pressure gauges should be provided on the lines at the
2-13
-------
top of each column to allow pressure buildup monitoring. Pressure relief valves should be
added to the top of each column. Screens with openings, slightly smaller than the smallest
media size should be placed at the bottom of each column or hi the tubing/pipe connectors at
the bottom of the columns. The screen should retain 70 mesh (US Std.) particles. This
prevents carry-over of the particles to the next column or into the effluent.
A flow control valve should be placed after the last column. By controlling the flow at
this point, the columns will always remain under positive pressure during the run. This
prevents the buildup of ah* pockets in the column. The tubing/pipe connecting the columns
should be sized to carry the appropriate flow. For two inch diameter columns, 1/4 or 3/8 inch
tubing is sufficient. The effluent line should break to the atmosphere at a level higher than
the top of the GAG bed. This will prevent the water level from falling below the top of the
GAG bed if flow to the columns stops and siphoning occurs.
Figure 4-1 shows two 10 min EBCT columns hi series. Configured in this manner data
from both the 10 and 20 mm EBCT columns can be gathered at once by sampling at the
designated points. Sampling of columns hi series should always begin at the last sample point
and proceed upstream to minimize flow variations. The two columns could also be arranged
in parallel with one containing enough GAC for a 10 min EBCT and the other enough for a
20 min EBCT. There are advantages and disadvantages of both series and parallel operation.
Series operation requires less equipment, but is less flexible with respect to operation
compared to parallel operation.
It is also possible to utilize sequential operation hi which a study is first conducted with
one of the two EBCTs columns prior to testing with the second EBCT column. Compared to
series or parallel operation, sequential operation requires the least amount of equipment, but
takes the longest tune to complete and requires the influent to be monitored for both runs,
thus doubling the analytical costs of influent monitoring.
A representative hydraulic loading rate, also termed superficial velocity, v, should be
used:
v = Q/A (4.1)
where Q is the volumetric flow rate and A is the column cross sectional area. A value for v
hi the range of 2 to 6 (galloris/min)/ft2 (5 to 15 m/hr) shall be used. Higher values of v yield
longer bed lengths, 1, as
1 = v-EBCT (4.2)
The EBCT can also be defined as
EBCT = Vb/Q (4.3)
2-14
-------
where Vb is the GAC bed volume. Values of v in this range will yield bed depths of 2.8 to
8.2 ft (0.83 to 2.5 m) and 5.6 to 16.4 ft (1.7 to 5.0 m) for EBCTs of 10 and 20 min,
respectively.
4.2 Operation
When loading the GAC into the columns care should be taken to reach the desired design
bed depth and avoid entrapping air in the GAC beds. Based on the required bed volume of
GAC
Vb = l-A (4.4)
for the 10 or 20 min EBCT columns and the manufacturers supplied dry bed (apparent)
density (mass of GAC/ Vb), the mass or weight of GAC required to yield a 10 or 20 min
EBCT can be calculated: mass = Vb x bed density. An amount of representative GAC that is
20% more than this mass should be carefully weighed and noted. From this amount of GAC
the GAC columns will be filled. The amount of GAC hi the column will be calculated by the
difference between the original amount and that remaining after filling.
The empty GAC column should be valved off and filled with non-chlorinated filtered
water to a level of about 50% of the design GAC bed depth. The dry GAC should be slowly
and carefully added to the column to a depth of about 5 to 10% greater than the design bed
depth. Allow the GAC to remain hi the bed to become 'wet' with unchlorinated filtered
water. This can take up to 12 hours for most GACs, but low density GACs may take longer
than 24 hours. The GAC bed should then be carefully backwashed with unchlorinated
filtered water to remove only the fines. The bed should be initially expanded by 10%, and
held at this level until most of the very fine particles are washed out. The bed can then be
slowly (over a 5 minute period) expanded to a 40% level. It is important to ensure that only
the fines are being backwashed out of the column. Backwash the GAC at this level for about
20 min. Slowly (over a 30 sec period) turn the backwash supply water off and allow the
GAC to settle. The settled bed depth should be about 5% longer than the design bed depth.
If it is not, add more GAC to the column to achieve this bed depth. A longer initial bed
depth is needed as the GAC will compact during operation. Allow the GAC to sit submerged
hi the unchlorinated filtered water overnight before start-up, to allow the air in the pores to
dissolve.
Prior to start-up, backwash the GAC at 40% bed expansion for 5 min. Again check the
settled bed depth. Weigh the amount of GAC remaining and subtract this amount from the
initial amount to determine the weight of GAC added to the columns. Report the mass of
GAC added to the columns in Table 4-1. This approach does not directly account for the
mass of fines backwashed out of the column, however, they are typically a small amount
compared to the total mass of GAC hi the column.
Open the appropriate valves to send the water stream to the GAC columns. A pump to
feed the water to the GAC or filter/GAC system may be needed if the influent water is not
2-15
-------
under enough pressure to overcome the headless in the GAG or filter/GAC system. The flow
rate should be set to yield EBCTs of 10 and 20 min using the flow control valve at the end of
the system.
Q = Vb/EBCT = (l'A)/EBCT
(4.5)
Use the valves at the top of the column (backwash water waste valves), see Figure 4-1, to
release any air trapped in the system.
After 12 to 18 hours of operation check the GAG bed depth levels, as the bed should
reach most of its compaction during this period. Readjust the flow rate to yield the 10 and 20
min EBCT based on this measured bed depth, with a margin of error of +/- 5%. If series
operation is used and only one column needs adjustment, add more GAG or remove excess
GAG and adjust the flow. Weigh the amount of GAG added or taken from the columns and
appropriately adjust the reported mass of GAG used hi each column. If more is added, the
bed must be backwashed and run for at least 4 hours before sampling. If GAG is removed
from the column, it must be dried at 100°C for at least 24 hours prior to weighing. Take the
first samples after about. 24 hours of operation.
The flow rate should be maintained to within 5% of that needed to produce the 10 and 20
min EBCT. Flow rates should be checked daily and adjusted to within this tolerance.
Unusually long periods of no flow to the columns (longer than 1 to 2 hours per day with a
maximum of 7 hours per week) should be accounted for by not including it hi the cumulative
operation time. Therefore, operation tune may be shorter than clock time.
If proper filtration at the full- or pilot-scale is maintained, then the GAG columns should
not need to be backwashed during operation. A clean-bed pressure drop of 0.05 to 0.15 psi
per ft of bed depth can be expected. High values of v and low values of the GAG particle
size will yield higher headloss. If the buildup of headloss becomes excessive, greater than 10
psi (or some other site determined value) above the clean bed value, then the GAG columns
should be backwashed. Expansion of the bed volume by 25 % for 20 min should be adequate.
Often the GAG at the top of the bed will bind together causing the GAG to rise as an intact
segment. This problem can be overcome by opening the top of the column and rodding the
bed, i.e., disturbing the top 2 to 4 niches (5 to 10 cm) of the bed with a rod. This problem
occurs less often with columns with diameters greater than 4 inches. Slowly turn the
backwash supply water off (30 sec period) and allow the GAG to settle. The water used for
backwashing should be GAG effluent that has been collected prior to backwashing and stored
off-line.
4.3 Sampling
A wide range hi breakthrough behavior like that shown hi Figure 4-2 is expected for GAG
columns with EBCTs in the 10 to 20 min range. This range in adsorption behavior makes
sampling the GAG column effluent difficult, as the goal is to characterize the breakthrough
behavior. Figure 4-3 depicts a sampling scenario which will meet the sample frequency
2-16
-------
requirements listed in Table 4-0 and yield a well characterized breakthrough curve. A
minimum of 15 influent samples and 15 effluent samples at both 10 and 20 mm EBCTs are
required. If the 10 and 20 min EBCT runs are being operated at the same time, either in
series as shown in Figure 4-1, or in parallel, then the 15 influent samples should be taken at
the same time as 20 mm EBCT effluent samples. If the runs for the 10 and 20 min EBCT are
being made at different times, then at least 15 influent samples need to be taken over the
course of both runs. These influent samples need to be taken at the same tune as the effluent
samples.
Three duplicate influent samples need to be taken over the course of the run. They should
be taken at approximately equally spaced intervals through the run. The first duplicate
influent sample should be taken with the third or fourth influent sample, the second duplicate
influent sample should be taken with the seventh or eighth influent sample and the third
duplicate influent sample should be taken with the eleventh, twelfth or thirteenth influent
sample.
As stated in Table 4-0 and illustrated hi Figure 4-3, effluent samples should be taken after
the first day and then at 3 % to 7% increments of the average influent TOC. The average
influent TOC is defined as the running average of the influent TOC at the tune of sampling.
This approach requires a relatively quick turn around time for TOC analysis. The intent is to
yield a good assessment of the breakthrough with a minunum number of samples. More
frequent effluent monitoring for TOC may be necessary hi order to sample for the other
analytes at the 3% to 7% increments of the average influent TOC.
Three duplicate effluent samples need to be taken over the course of the run for each
EBCT tested. They should be taken at approximately equally spaced intervals through the
run. The first duplicate effluent sample should be taken with the third or fourth effluent
sample, the second duplicate effluent sample should be taken with the seventh or eighth
effluent sample and the third duplicate effluent sample should be taken with the eleventh,
twelfth or thirteenth effluent sample.
The breakthrough of UV absorbance at a wavelength of 254 nm, UV^, often parallels
that of TOC and may be used to estimate the sampling tunes. However, UV-absorbing
substances are normally better removed by GAG and UV254 breakthrough lags behind that of
TOC. Thus, the relationship between UV254 and TOC hi the GAG effluent needs to be first
established. This requires that both TOC and UV254 data were collected for a previous GAG
run. Thus, if this monitoring approach is to be used, prior experience with GAG treatment of
the site-specific water is needed.
If on-site TOC or fast turn-around of TOC measurement, or previous GAG experience are
not available, a sampling plan may be estimated from the influent TOC by the following
procedure. In Figure 2-3 a correlation was presented that related the number of bed volumes
to 50 percent TOC breakthrough, BVSO, to the influent TOC concentration, TOC0, for
bituminous coal based GACs. This relationship can be expressed by the following equation
(Equation 2.3):
2-17
-------
BVSO = 21,700-TOC0-13 (4.6)
The time in days to 50 percent TOC breakthrough, tso, can be estimated by the following
equation for an EBCT in minutes:
tso = BVSo-EBCT/1440 (4.7)
Based on the shape of previous TOC breakthrough curves, a 1-9-4-1 sample plan is
recommended for the 15 effluent samples. The first sample is taken after one day, nine
additional samples are taken at regular tune intervals up to the 50 percent breakthrough, four
samples are taken at regular tune intervals after 50 percent breakthrough, and one sample is
taken at the end of the run. The sample time interval hi days, t^, for the first half of the
TOC breakthrough can be estimated from the following:
^ = ^0-1)79 (4.8)
Since TOC breakthrough curves are not symmetrical and tend to have a lower slope after 50
percent breakthrough, the sample time interval after 50 percent breakthrough should be
estimated as 50 percent longer than t;0.
Table 4-2 presents a general sample plan and a sample plan for ten examples with TOC0
values of 2.0, 2.5, 3.0, 4.0, and 6.0 mg/L and EBCTs of 10 and 20 minutes. Figure 2-3 and
Equations 4.6, 4.7 and 4.8 are utilized for this method of estimating a sample plan. This
approach is based on a data set limited to 16 water sources and one general GAC type, and
may not be valid in all cases. If the pH value of the water to be treated is 7.0 or below then
the BV50 value may be 10 to 40 percent higher than the value given by Equation 4.6, based on
the work of Hooper et al. (1995).
It is strongly suggested that on-site TOC or fast turn-around TOC measurements be used
whenever possible to characterize the breakthrough and accurately determine the sample
times.
Both the 10 and 20 minute EBCTs of the pilot study shall be run until: a) the effluent
TOC concentration is 70% of the average influent TOC concentration on two consecutive
TOC sample dates that are at least two weeks apart, or b) 50% TOC breakthrough occurs and
a plateau is reached in which the effluent TOC concentration does not increase over 1440
hours by more than 10% of the average influent TOC concentration. The 10 minute EBCT
run will take about half of the tune required for the 20 minute EBCT. If the 10 and 20
minute EBCT contactors are operated in series, then operation of the 10 minute EBCT
contactor must continue until the 20 minute EBCT run is terminated. However, the 10
minute EBCT contactor does not need to be sampled after 70% breakthrough or a plateau is
achieved.
2-18
-------
It is strongly recommended that the pilot-scale columns be run continuously. Down time
that cannot be avoided should not be included in the cumulative run time.
If either the 70% breakthrough or the plateau criteria is met for the 20 minute EBCT prior
to 4000 hours run time, a second run shall be conducted following the same sampling
requirements. In all cases, the maximum run length for the pilot-scale study (one or two
runs) is 8000 hours. The pilot-scale testing should be of sufficient duration and appropriate
tuning to capture seasonal variations in water quality. If seasonal variation is not significant
(e.g. as is the case for most ground waters), other factors, such as pretreatment, carbon type,
etc. can be evaluated.
The pilot-plant design parameters, sampling times and results should be reported in
accordance with Tables 4-1, 4-3 and 4-4 of part 2. Samples should be taken according to the
procedures described in the "ICR Sampling Manual" (EPA 814-B-96-001). The approved
analytical methods for the analytes listed in Table 4-0 are listed in Table 7 § 141.142 of the
ICR Rule, described in "DBP/ICR Analytical Methods Manual" (EPA 814-B-96-002) and
must be used by all systems conducting treatment studies. Laboratory QA/QC plans listed in
this manual must also be followed. Guidance for the simulated distribution system (SDS) test
and the chlorine demand test are given in Section 6.0 of this document.
2-19
-------
-------
Backwash
water waste
Effluent
Flow
control
valve
Backwash
water
supply
Figure 4-1. Pilot plant set-up
30
60
180
210
90 120 150
Operation time (days)
Figure 4-2 Typical GAC breakthrough behavior for natural organic matter
240
-------
30
60
90 120 150
Operation time (days)
Figure 4-3 Pilot plant breakthrough and sampling
180
210
240
-------
Table 4-0. Sampling of 6AC Pilot-scale Systems
Sampling Point
GAG Influent
GAG Effluent <§
EBCT=10min
GAG Effluent <§
EBCT=20 min
Analyses
pH, alkalinity; turbidity,
temperature, total &
calcium hardness, ammonia,
bromide, TOG and UV254.
SDS1 for THM4, HAA6,
TOX, and chlorine demand.
pH, turbidity, temperature,
ammonia2, TOG and UV254.
SDS1 for THM4, HAA6,
TOX, and chlorine demand.
pH, turbidity, temperature,
ammonia2, TOG and UV^.
SDS1 for THM4, HAA6,
TOX, and chlorine demand.
Sample Frequency3-4
A minimum of 15 samples
taken at the same time as
the samples for GAG
effluent5.
A minimum of 15 samples.
One after one day, and
thereafter at 3% to 7%
increments of the average
influent TOG.
A muiimum of 15 samples.
One after one day, and
thereafter at 3% to 7%
increments of the average
influent TOG.
1 - SDS conditions are defined in Part 1, Section 4.6 of this document. Additional
guidance is found hi Section 6.0 of this Part.
2 - If present in the influent.
3 - More frequent effluent monitoring may be necessary hi order to predict the 3 % to
7% increments of average influent TOG.
4 - Three duplicate samples are required for the influent and the effluent at each EBCT.
5 - If columns for EBCT=10 min and EBCT=20 min are run simultaneously, then
influent samples should be taken at the same sample frequency as that for GAG
effluent at EBCT=20 min.
-------
Table 4-1. PILOT PLANT DESIGN PARAMETERS
Utility name and address
Utility ID number
Contact person
Contact phone number
Contact FAX number
GAC type and manufacturer
GAC mesh size
EBCT= 10 min
Bed depth, 1
Volumetric flowrate, Q
Superficial velocity, v
Mass (dry) of GAC
EBCT- 20 min
Bed depth, 1
Volumetric flowrate, Q
Superficial velocity, v
Mass (dry) of GAC
Design and operation comments
mm or
_ml/min or.
m/hr or
mm or
_ml/min or.
m/hr or
US std mesh
inches
. gal/hr
_gpm/ft2
grams
inches
.gal/hr
_gpm/ft2
grams
-------
!
a
(U
a
CO
I
(N
4
a
O
I
IT)
(N
II
O
gs
^
ffl
W
a
H 'S
u
CQ °
- ^
.a
s
03 o
W '-'
u
O
O
,
PJ (S)
'-
pq
PL)
PL] oo
J
J J
J J
CO
co a
3
rt
-------
Table 4-3. GAG PILOT PLANT SAMPLING
Utility name and address
Utility ID number
Contact person
Contact phone number
Contact FAX number
GAG Influent1 (A group)
pH, alk, turb, temp, TH, CaH, NH4-N,
Br, TOC, UV and SDS
Sample Al
Sample A2
Sample A3
Sample A4
Sample A5
Sample A6
Sample A7
Sample A8
Sample A9
Sample A10
Sample Al 1
Sample A12
Sample A13
Sample A14
Sample Al 5
Sample Dl -I
Sample D2-I
Sample D3-I
Date
(M/D/Y)
Time
Operation
time (day)
1.0
1) Sample influent at the same time as EBCT= 20 min effluent if EBCT= 10 min and
EBCT= 20 min are being run at the same time. If EBCT= 10 min and EBCT= 20 min
are being run at different times then the GAC influent of both runs need to be
monitored
Sampling comments
-------
GAG PILOT PLANT SAMPLING (page 2) Utility ID #
GAC Effluent EBCT= 10 min (B group) pH,
turb, temp, NH4-N2, TOC, UV, & SDS
Sample B 1
Sample B2
Sample B3
Sample B4
Sample B5
Sample B6
Sample B7
Sample B8
Sample B9
Sample BIO
Sample Bll
Sample B12
Sample B 13
SampleB 14
Sample B15
Sample D 1-10
Sample D2- 10
Sample D3- 10
GAC Effluent EBCT= 20 min (C group)
pH, turb, temp, NHU-N2, TOC, UV, & SDS
Sample Cl
Sample C2
Sample C3
Sample C4
Sample C5
Sample C6
Sample C7
Sample C8
Sample C9
Sample CIO
Sample Cll
Sample C12
Sample C 13
Sample C 14
Sample CIS
Sample D 1-20
Sample D2-20
Sample D3-20
Date
(M/D/Y)
Time
Operation
time (day)
1.0
1.0
2) Sample for NH4-N only if present in influent
-------
Table 4-4. PILOT PLANT RESULTS
Utility name and address
Utility ID number
Contact person
Contact phone number
Contact FAX number
GAC INFLUENT
Group A
Sample Al
Sample A2
Sample A3
Sample A4
Sample A5
Sample A6
Sample A7
Sample A8
Sample A9
Sample A10
Sample Al 1
Sample A12
Sample A13
Sample A14
'Sample Al 5
Sample Dl-I
Sample D2-I
Sample D3-I
alk
TH
CaH
NH4-N
turb
Units: alk, TH, CaH- mg/L as CaCO3; NH4-N- mg/L; turb- ntu
GAC influent results comments
-------
PILOT PLANT RESULTS (page 2)
Utility ID #
GAC INFLUENT
Group A
Sample Al
Sample A2
Sample A3
Sample A4
Sample A5
Sample A6
Sample A7
Sample A8
Sample A9
Sample A10
Sample Al 1
Sample A12
Sample A13
Sample A14
Sample Al 5
Sample Dl-I
Sample D2-I
Sample D3-I
Group A-SDS
Sample Al
Sample A2
Sample A3
Sample A4
Sample A5
Sample A6
Sample A7
Sample A8
Sample A9
Sample A10
Sample Al 1
Sample A12
Sample A13
Sample A14
Sample A15
Sample Dl-I
Sample D2-I
Sample D3-I
PH
Cl dose
temp.
Cl res.
Br
CD
TOC
temp.
uv
pH
H. time
Units: TOC, Cl dose, Cl res., CD- mg/L; UV- cm'1; temp- °C; Br- fj.g/L; H. time- hrs
-------
PILOT PLANT RESULTS (page 3) Utility ID
GAC INFLUENT
#
Group A-SDS
Sample Al
Sample A2
Sample A3
Sample A4
Sample A5
Sample A6
Sample A7
Sample A8
Sample A9
Sample A10
Sample Al 1
Sample Al 2
Sample A13
Sample A14
Sample Al 5
Sample Dl-I
Sample D2-I
Sample D3-I
Group A-SDS
Sample Al
Sample A2
Sample A3
Sample A4
Sample AS
Sample A6
Sample A7
Sample A8
Sample A9
Sample A10
Sample Al 1
Sample A12
Sample A13
Sample A14
Sample A15
Sample Dl-I
Sample D2-I
Sample D3-I
TOX
HAA6
THM4
MCAA
DCAA
CHC13
TCAA
MBAA
CHBrCl2
DBAA
CHBr2Cl
BCAA
TBAA
CHBr3
CDBAA
DCBAA
Units: SDS- jag/L
-------
PILOT PLANT RESULTS (page 4)
Utility ID #
GAC EFFLUENT - EBCT= 10 mm
Group B
Sample B 1
Sample B2
Sample B3
Sample B4
Sample B5
Sample B6
Sample B7
Sample B8
Sample B9
Sample BIO
Sample Bll
Sample B 12
Sample B 13
Sample B 14
Sample B 15
Sample D 1-10
Sample D2- 10
Sample D3-10
Group B-SDS
Sample Bl
Sample B2
Sample B3
Sample B4
Sample B5
Sample B6
Sample B7
Sample B8
Sample B9
Sample BIO
Sample Bll
Sample B12
Sample B 13
SampleB 14
Sample B15
Sample D 1-10
Sample D2-10
Sample D3- 10
NH4-N2
Cl dose
PH
CIres.
turb.
CD
temp.
temp.
TOC
pH
uv
H. time
Units: NH4-N, TOC, Cl dose, Cl res., CD- mg/L; turb- ntu; UV- cnr^temp- °C;
H. time-hrs 2) Sample for NH4-N only if present in influent
-------
PILOT PLANT RESULTS (page 5) Utility ID #
GAC EFFLUENT - EBCT= 10 min
Group B-SDS
Sample Bl
Sample B2
Sample B3
Sample B4
Sample B5
Sample B6
Sample B7
Sample B8
Sample B9
Sample BIO
Sample Bll
Sample B12
Sample B13
Sample B 14
Sample B 15
Sample Dl-10
Sample D2-10
Sample D3-10
Group B-SDS
Sample Bl
Sample B2
Sample B3
Sample B4
Sample B5
Sample B6
Sample B7
Sample B8
Sample B9
Sample BIO
Sample Bll
SampleB 12
Sample B13
Sample B 14
Sample B 15
Sample Dl-10
Sample D2-10
Sample D3-10
TOX
THM4
CHC13
CHBrCl2
DBAA
CHBr2Cl
BCAA
TBAA
CHBr3
CDBAA
DCBAA
Units: SDS- jig/L
-------
PILOT PLANT RESULTS (page 6)
Utility ID #
GAC EFFLUENT - EBCT= 20 min
Group C
Sample Cl
Sample C2
Sample C3
Sample C4
Sample C5
Sample C6
Sample C7
Sample C8
Sample C9
Sample CIO
Sample Cll
Sample C12
Sample C13
Sample C14
Sample CIS
Sample Dl-20
Sample D2-20
Sample D3-20
Group C-SDS
Sample Cl
Sample C2
Sample C3
Sample C4
Sample C5
Sample C6
Sample C7
Sample C8
Sample C9
Sample CIO
Sample Cll
Sample C 12
Sample C 13
Sample C14
Sampled 5
Sample Dl-20
Sample D2-20
Sample D3-20
NH4-N2
Cl dose
pH
Cl res.
turb.
CD
temp.
temp.
TOC
pH
uv
H. time
Units: NH4-N, TOC, Cl dose, Cl res., CD- mg/L; turb- ntu; UV- cnr^temp- °C;
H. time-hrs 2) Sample for NH4-N only if present in influent
-------
PILOT PLANT RESULTS (page 7) Utility ID #
GAC EFFLUENT - EBCT= 20 min
Group C-SDS
Sample Cl
Sample C2
Sample C3
Sample C4
Sample C5
Sample C6
Sample C7
Sample C8
Sample C9
Sample CIO
Sample Cl 1
Sample C12
Sample CIS
Sample C14
Sample CIS
Sample Dl-20
Sample D2-20
Sample D3-20
Group C-SDS
Sample Cl
Sample C2
Sample C3
Sample C4
Sample C5
Sample C6
Sample C7
Sample C8
Sample C9
Sample CIO
Sample Cll
Sample C12
Sample CIS
Sample C14
Sample CIS
Sample Dl-20
Sample D2-20
Sample D3-20
TOX
THM4
CHC13
MBAA
CHBrCl2
DBAA
CHBr2Cl
BCAA
TBAA
CHBr3
CDBAA
DCBAA
Units: SDS- pg/L
-------
5.0 Bench-Scale GAC Test Protocol
5.1 Design
Scaling equations, which are used to design the RSSCT, have been developed based on
dimensional analysis to maintain similitude to the full-scale GAC column (Prick, 1982;
Crittenden, Berrigan, and Hand 1986; Crittenden et al. 1987, 1989 and 1991; Hineline,
Crittenden, and Hand 1987). The critical RSSCT design parameters are the EBCT and the
hydraulic loading rate or superficial velocity (v). Also of interest is the operation time (t).
The scaling relationship is a function of the carbon particle size used hi the small-(RSSCT)
and large-(pilot or full) scale columns. The scaling factor (SF) can be defined as the ratio of
particle diameters (d) of the large particle column (LC) to that of small particle column (SC)
SF = dLC/dsc (5.1)
The dependency of intraparticle diffusivity, as expressed by the diffusion coefficient (D),
on particle size can be written as
DSC = (dsc/dLC)x'DLC = SFX-DLC (5.2)
where X is the diffusivity factor. The RSSCT was originally developed with the assumption
that intraparticle diffusivity did not change with particle size. Thus, the design equations
were initially based on an assumption of constant diffusivity (CD), with X=0 in Equation
5.2.
A number of small column tests and batch kinetic tests have shown that the diffusion
coefficient decreases proportionally with decreasing particle size and a proportional
diffusivity (PD) design approach has been developed. In fact, several studies have shown a
linear relationship between D and d, especially for NOM as measured by TOC and UV2S4,
(Benz, 1989; Summers and Crittenden, 1989; Crittenden et al., 1991; Summers et al., 1994a;
1995). This linear dependency is characterized by setting X=l in Equation 5.2 and leads to
the proportional diffusivity design approach. The EBCT and operation tune of the RSSCT
are both directly related to that of the large-scale column by the ratio of the small to large
particle diameters, or the inverse of SF.
EBCTSC = (dsc/dLC)-EBCTLC = EBCTLC/SF (5.3)
tsc = (dsc/dLC)-tLC = tLC/SF (5.4)
The similitude velocity hi the RSSCT, v*sc, is also directly related to the velocity in the
large-scale column by the scaling factor or the ratio of the large to small particle diameters.
v*sc = (dLc/dscVvLc = vLC-SF (5.5)
2-21
-------
The design of a RSSCT based on Equations 5.3 and 5.5 will yield a RSSCT column with the
same length as the large-scale column.
Isc = v*sc-EBCTsc = (vLC-SF)-(EBCTLC/SF) (5.6)
Isc = vLC-EBCTLC = 1LC (5.7)
However, RSSCTs designed with such long lengths will likely produce excessive headless
and are difficult to operate at the bench-scale level.
To shorten the RSSCT length, the dominance of internal mass transfer over external mass
transfer is utilized. The design velocity of the RSSCT, vsc, can be reduced to a value below
that of the similitude velocity as long as it is above the minimum velocity at which the
RSSCT can be operated such that internal mass transfer still dominates. This minimum
velocity is defined using the minimum Reynolds number, ReSCiITlin, which ranges from 0.02 to
0.13, depending on the molecular weight of the compound (Crittenden et al. 1987). The lack
of impact of Resc on the breakthrough curve is shown in Figure 5-1 (Summers et al. 1994a).
The removal of natural organic matter, as measured by TOC, by a GAC column with full-
scale GAG particles (12x40 mesh), a length of 1.2 m (3.9 ft), an EBCT of 10 minutes and
RCLC of 5 was modeled. Reducing the Reynolds number from 5 hi the large-scale column to
0.05 in the RSSCT with a dsc of 0.11 mm had very little impact on the modeled breakthrough
curve.
Using a Resc hi the RSSCT that is lower than that in the full-scale system allows for the
use of a lower vsc in the RSSCT. As shown hi Equation 5.6, lower vsc values will result in
shorter columns and lower flow rates, both of which are desirable for bench-scale operation.
The design equation for the RSSCT hydraulic loading thus becomes
vsc = SF-vLC-(ReSC)min/ReLC) (5.8)
Values of ReSCimin ranging from 1.0 to 0.5 have been successfully used and are recommended.
For the example modeled in Figure 5-1, the scaling factor yields a EBCTSC of 1.0 mm.
Using a ReSCimin of 0.5 hi the RSSCT yields a vsc of 7.2 m/hr, the same as that in the pilot
column. A vsc of 7.2 m/hr and an EBCTSC of 1.0 min produces a column depth of 12 cm
(4.7 in.), as compared to 120 cm (47 in.) hi the large-scale column or hi me RSSCT if the
similitude velocity, v*s, were used.
Based on the above discussion, it is recommended that the equations summarized in Table
5-1 be used for the RSSCT design when NOM is the target component. These equations are
based on proportional diffusivity and have been found to yield results that are comparable to
those of pilot- and full-scale systems (Table 2-1).
The two most commonly used sizes of GAC for drinking water treatment are the 8x30 and
12x40 US Std. mesh with apparent particle diameter (APD), dLC, of 1.6 and 1.1 mm,
respectively. The 8x30 mesh GAC is often used hi filter adsorbers, while the 12x40 mesh
2-22
-------
GAC is commonly used in post-filter adsorbers. The design of several RSSCTs based on
these two large particle size GACs are shown in Tables 5-2 and 5-3. These tables show the
EBCTSC and lsc values needed in the RSSGT to yield designs that are the equivalent of 10 and
20 min full-scale EBCTS. Also shown hi Tables 5-2 and 5-3 are the volumetric flow rate
(Qsc) and minimum volume of influent water required for the RSSCT (Vsc) for two RSSCT
columns with inner column diameters (DCSC) of 8.0 and 11.0 mm. These two column sizes
with lengths that accommodate the RSSCT-GAC bed depths are commercially available.
The RSSCT designs shown in Tables 5-2 and 5-3 were generated using a proportional
diffusivity design based on the equations in Table 5-1 with a Resc min of 0.5. A temperature
of 20°C, which yields a kinematic viscosity of l.OxlO'6 m2/s, and a porosity hi the large-scale
column of 0.45 were assumed. Since the performance of the RSSCT is not sensitive to the
Resc,min» as shown in Figure 5-1, precise values for the viscosity and porosity are not needed
for the calculation of vsc.
For both the 8x30 and 12x40 mesh full-scale particles, RSSCTs have been designed with
four different particle sizes, dsc: 60x100 (200 jan), 100x200 (112 /on), 170x230 (76 fim) and
230x325 (54 /mi) mesh size. Other particle sizes can be used hi the RSSCTs, but previous
runs have been made with dsc values hi this range. Smaller dsc values will yield shorter run
tunes, but may yield excessive headless buildup, while larger dsc values will yield longer run
tunes. The operation (run) tune of the RSSCT, tsc, is dependent on the scaling factor and the
run tune hi the large scale system, tLC. For purposes of these designs a total operation tune,
tc7. value of 180 days was assumed. A conservative tLCT value, longer than needed, should
be chosen, as it sets the volume of water needed to run the RSSCT, Vsc, which is sampled
prior to the start of the RSSCT. RSSCT operation tunes range from 6 to 23 days for the
RSSCTs representing the 8x30 GAC and 9 to 33 days for that representing the 12x40 GAC.
Without any prior GAC adsorption experience with the water to be treated, it is difficult
to estimate the length of a full-scale adsorber run, tLCT. One method to estimate tLCT and the
required influent volume for the RSSCT, Vsc, is to use the correlation presented hi Figure 2-3
(Equation 2.3). It relates the number of bed volumes to 50 percent TOC breakthrough, BVSO,
to the influent TOC concentration, TOC0, for bituminous coal based GACs. This
relationship can be expressed by the following equation:
= 21,700-TOC0-13 (5.9)
The tune in days to 50 percent TOC breakthrough, tso, for the large-scale column can be
estimated by the following equation for an EBCT hi minutes :
tso = BV50-EBCTLC/1440 (5.10)
where BV50 is calculated by Equation 5.9. If the total operation tune is defined as the tune to
achieve greater than 75% breakthrough or reach a pseudo-steady-state plateau, then it can be
estimated as twice that o
2-23
-------
tLCT = 2-tso (5.11)
The required influent volume for the RSSCT can be estimated from BV50 assuming that
the total volume required is twice the volume to 50% breakthrough:
Vsc = 2-BV50-EBCTLC-QsC/SF= Qsc-tLCT/SF= Qsc-tscT (5.12)
The volume actually taken for the RSSCT influent should be at least 30 percent larger than
this value to allow for influent sampling and as precautionary measure due to the
heterogeneous nature of the adsorption of NOM. If a nonbituminious coal based GAC is to
be used, then use Equations 5.9 through 5.12 and multiple Vsc by a factor of 2 to yield the
volume of water taken for the RSSCT influent. This will very likely be a very conservative,
more volume than needed, value."
As shown by Equation 5.9 and Figure 2-3, higher influent TOC values lead to earlier
breakthrough. Very early breakthrough can lead to a situation where the total effluent
volume treated is less than the volume of water needed for the analyses required to monitor
the RSSCT breakthrough. To insure that an adequate sample volume is present hi the
effluent, it is recommended that a RSSCT column with a DCSC of 11 mm or greater be used
for waters with a GAC influent TOC value greater than 4 mg/L or for waters with a high
level of poorly adsorbing NOM.
The RSSCT designs outlined in Tables 5-2 and 5-3 are not inclusive of all acceptable
designs, but are intended to represent the experience gathered to-date with RSSCTs and
natural organic matter removal (Table 2-1). However, the proportional diffusivity design
equations shown in Table 5-1 should be used.
A schematic of the equipment configuration for the RSSCT is shown in Figure 5-2. All
components of the bench-scale system shall be of glass, Teflon or stainless steel construction.
One-quarter inch Teflon tubing, rated for appropriate system pressures, should be used
throughout the system. A positive displacement feed pump with controls for stroke frequency
and stroke volume should be used. It is very important that the pump for the system be
selected based on the appropriate system pressures and flow rates. It is recommended that
one pump be used for each RSSCT (10 and 20 mm EBCT) when operated hi parallel. To
dampen the pulse created during pumping with a positive displacement feed pump, a 0.5 L
stainless steel air-tight container (pulse dampener) is placed at the top of the system. A gas
sample cylinder is well-suited for this purpose. During start-up of the RSSCT this container
is not filled with water and the compressibility of the air in the container serves to dampen the
pulsing action of the pump.
The batch of RSSCT influent water should be prefiltered with membrane or cartridge
filters with nominal pore openings of about 1.0 um. This will reduce the buildup of
excessive headloss during the RSSCT run. The membrane or cartridge filters should be
thoroughly flushed with low organic content water and the TOC of the filter effluent should
be measured to ensure that no TOC is being leached from the membrane or filter. As a
2-24
-------
precautionary step, an optional column containing a filter media, e.g. glass beads, can be
used as a pre-filter, as shown in Figure 5-2, to eliminate excessive headless build-up in the
carbon columns.
Typically, glass columns with inner diameters (ID) of 4 to 15 mm are used. Standard
chromatography columns with IDs of 8 to 15 mm are available. One-quarter inch outer
diameter glass tubes with 4 mm ID have also been used, but are not recommended due to
excessive pressure build-up. The ratio of column inner diameter to carbon diameter should
be 25 or greater. Multiple columns used in parallel (not shown) should each be equipped
with a flow control valve to accommodate flow rate adjustments for the individual columns.
In addition, the influent line to each column should have a shut-off valve, to prevent flow
interruptions to the other columns in case a column needs to be taken off-line during
operation.
Series operation of RSSCTs cannot be used as the sample volume requirements of the
intermediate sample point, EBCT = 10 min, are so large that they significantly affect the
breakthrough and the volume of water treated by the second column.
Before the column is packed with the GAC, a series of three stainless steel screens,
coarse, fine and coarse, should be placed at the bottom of the inside of the column as shown
in Figure 5-2. The screens serve to contain the carbon in the column. The fine screen should
have openings that are smaller than the smallest carbon particles. The coarse screens have
openings that are only slightly larger. For example, for 60x100 GAC, 60 mesh screens are
used for the coarse screens and a 200-mesh screen is used for the fine screen. A depth of 1 to
2 cm of glass beads, which have a larger mesh size than the fine screen but a smaller mesh
size than the carbon, can be used to replace the top coarse screen. Glass wool has also been
used in place of the glass beads, although its use may cause a faster headloss buildup. The
glass beads, or glass wool, serve as an additional support for the carbon bed. An additional
screen, with a mesh size smaller than the carbon, should be placed on top of the glass beads
or glass wool. Care should be taken in handling the screens, glass beads and especially the
glass wool if used. They should all be thoroughly cleaned prior to use, so that no detectable
organic carbon leaches from them..
A flow control valve should be placed after the RSSCT column, except when a positive
displacement pump is used, in which case a flow restricter should be used at this point. In
either case, by controlling or restricting the flow at this point, the columns will always remain
under positive pressure during the run. This prevents the buildup of air pockets in the
column. The effluent line should break to the atmosphere at a level higher than the top of the
GAC bed, as shown in Figure 5-2. This will prevent the water level from falling below the
top of the GAC bed if flow to the columns stops and siphoning occurs. In addition, the
influent line should be airtight and securely attached to the reservoir to prevent pumping air
into the columns.
2-25
-------
The granular activated carbon selected foyise in the RSSCT must meet the AWWA
Standard for Granular Activated Carbon ANSI/AWWA B604-90 (Appendix 2-C). This
standard covers the sampling and characterization of the GAC.
A representative sample of the carbon is taken from the carbon stock, as discussed in
Section 4.0 and Appendix 2-A. This sample is ground to an appropriate mesh size for the
RSSCT. It is very important that all of the carbon sample be ground such that the entire
amount passes the upper sieve of the desired mesh size range. Not grinding the entire amount
of GAC sample so that it to passes the upper screen size may lead to a ground GAC sample
which has a higher adsorption capacity than the unground full-scale GAC. The use of very
small sized carbon, 400 mesh (< 0.038 ptm), leads to excessive headless in the column.
Typical mesh sizes used in RSSCTs are 60 x 100 (200 /urn APD), 100 x 200 (112 jan APD)
and 170 x 230 (76 /-im APD). Smaller particles sizes, such as 200 x 325 (60 jum APD), 230 x
325 (54 fj.m APD) or 200 x 400 (57 /on APD) mesh sizes, can be used if the system is
designed for high pressures (> 50 psi).
After sieving, the ground carbon should be washed with purified water. Deaerated, low
organic concentration, distilled water, termed laboratory clean water (LCW), should be used.
Step-wise decantation in a beaker can be Used to wash the carbon. Pour the wash water over
the ground GAC, stir the slurry until the GAC is wet, allow the GAC to settle for 1 to 2
minutes and then decant. The amount of washing depends on the carbon-to-wash vessel
volume ratio. For ratios of 0.1 or less, 50 to 100 decantations may be necessary. Care
should be taken to thoroughly wash the carbon to avoid headless buildup in the column
caused by fines. An ultrasonic bath may be cautiously used to better remove the carbon
fines. The carbon can be placed in a beaker filled with low organic content water and
sonicated for 10 seconds at a low setting followed by decantation. This process can be
repeated once followed by step-wise decantation until all fines are removed. Excessive
sonication of the carbon will result in the formation of additional fines and must be avoided.
After washing, the ground carbon is dried overnight to a constant weight at a temperature of
80°C. The temperature should be increased to 100°C for 4 hours. The carbon is then
weighed again and if the weight is more than 5% different than the previous weight, then
drying at 100°C should be continued to a constant weight. High temperature drying leads to
the formation of additional fines. The dried carbon is then transferred to a clean bottle,
capped and stored in a desiccator until ready for use.
5.2 Operation
The bed density of the ground GAC should be assessed by measuring the dry weight of
ground GAC per unit bed volume. This can be done by precisely weighing about 2 grams of
ground GAC and adding it to a 5 ml or 10 ml calibrated graduated cylinder and determining
the bed volume of GAC. The calibration of the graduated cylinder should be verified using a
precise volume from a suitable pipet. If the volume measurements differ by 5% or greater,
the graduated cylinder should be recalibrated. The cylinder should be vibrated or tapped by
hand to allow the ground GAC to compact. The bed density, psc, is the GAC (dry) weight
divided by the GAC bed volume.
2-26
-------
An appropriate mass of dried carbon, msc, is calculated based on the RSSCT column
diameter, DCSC, required bed depth, lsc, and oft the ground GAG bed density.
msc = Isc-psc-[7t(DCsc)2/4] (5.13)
After weighing this carbon mass, the carbon is 'prewetted1 by placing it into an Erlenmeyer
flask and adding LCW to a level of about one inch over the carbon surface. The ground GAC
is then deaerated by applying a vacuum for at least 15 minutes. The vacuum will help speed
up the removal of air which may be trapped inside the pores of the carbon particles.
Removing the trapped air prior to the column operation will prevent diffusion hindrance of
the inward diffusing organic adsorbate. The carbon is easier to deaerate if it is allowed to sit
overnight, 16 to 24 fir, hi LCW, prior to deaeration.
After deaeration remove the excess LCW, so that a ground GAC slurry exists that can be
transferred into the column with a laboratory spatula. The prewetted and deaerated carbon is
then packed into the column as a slurry. The column should be first filled with LCW to a
level of 25% of the GAC bed depth. The column should be tapped, very gently, during the
addition of the GAC slurry to pack the carbon particles as the column is filled. The carbon
bed should be completely submerged during and after the packing process. LCW should be
used during the packing and testing of the columns. The integrity of the RSSCT system
should be tested for leaks, air pockets or immediate headless buildup with LCW by opening
the appropriate valves and feed the LCW to the system for about 10 minutes. Resolve any
problems that are identified. A plastic safety shield should be used around all glass GAC
columns.
The reservoir is then filled with the water to be evaluated. The columns are disconnected
from the feed system and the feed system from the reservoir to the carbon columns is purged
of ah- and the LCW with the feed solution. The column(s) are placed into position and the
flow rate adjusted to the desired level. The flow rate should be maintained to within 5% of
that needed to produce the equivalent of the 10- and 20- nun full-scale EBCT. Flow rates
should be checked at least twice a day and adjusted to within this tolerance. Unusually long
periods of no flow to the columns (longer than 0.5 hour per day) should be accounted for by
not including it in the cumulative operation tune. Therefore, operation tune may be shorter
than clock time.
The first 20 minutes of flow should be wasted, and the first sample should be taken after
1.0 hour. Samples are taken as specified in the next section, and flow and pressure
measurements are recorded at regular intervals. During operation, the pressure should be
closely monitored. Significant increases suggest a need to change to a clean filter and/or
drain the dampener which becomes less effective as it accumulates a large amount of water.
In severe cases of pressure build-up, the RSSCT column may be taken off-line and the top 0.5
cm of the GAC bed stirred to break apart large lumps that form and are responsible for
excessive headloss.
2-27
-------
5.3 Sampling
*
A wide range of breakthrough behavior like that shown in Figure 4-2 is expected for GAC
columns with EBCTs hi the 10 to 20 min range. This range of adsorption behavior coupled
with the short RSSCT run times makes sampling the RSSCT column effluent even more
difficult than that for the pilot column. Figure 5-3 depicts a sampling scenario which will
meet the sample frequency requirements listed in Table 5-0.
The first step is to sample a batch of water that is to be used as the influent to the
RSSCT. Section 3.0 discusses sample point location and pretreatment. It is imperative that
an adequate volume and a representative sample of the water to be GAC treated be taken.
Minimum required volumes for the RSSCT are listed hi Tables 5-2 and 5-3 for an anticipated
full-scale run time of 180 days. Longer run times may occur, especially for low TOC (< 3
mg/L) or low pH (< 7.3) influent waters at a 20 min EBCT. Thus, larger volumes of water
should be gathered for conditions which will lead to long run tunes.
For bituminous coal based-GAC and pH values above 7.3, Equation 5.12 can be used to
estimate Vsc. The volume actually taken for the RSSCT influent should be at least 30 percent
larger than this value to allow for influent sampling and as precautionary measure due to the
heterogeneous nature of the adsorption of NOM. Only one batch of influent water per run is
gathered so its representativeness is critical to the success of the RSSCT hi evaluating field-
scale performance.
The two approaches for operation of the required two RSSCTs, 10 and 20 minute
equivalent EBCTs, are parallel and sequential operation. In parallel operation both columns
are run at the same tune and the influent batch of water must be of sufficient volume to
satisfy both columns. In sequential operation one RSSCT is run and sampled first, followed
by the other RSSCT. This requires two batches of influent water, each of which needs to be
checked for representativeness and to be sampled during operation according to Table 5-0.
As stated earlier, series operation of two columns with EBCTs of 10 min cannot be used due
to sampling limitations.
Prior to the operation of the RSSCT, the water quality of the batch influent water should
be immediately evaluated on-site for TOC, UV^, pH, alkalinity, hardness (total and
calcium), ammonia and bromide. The intent of this preliminary water quality evaluation is to
quickly ascertain the representativeness of this water without waiting for off-site laboratory
analysis. The representativeness of this water should be confirmed by comparison with
previous water quality data. In many cases, UV^ and bromide analyses cannot be conducted
on-site, in which case they should not be used as a check of representativeness. Once the
representativeness of this water has been confirmed, the water should be prefiltered with a
membrane or cartridge filter, as described hi Section 5.2. If possible this batch of GAC
influent water should be stored at 4°C. If this is not possible then the water should be stored
at the lowest possible temperature. In all cases an aliquot of the stored water should be
brought to laboratory room temperature prior to use for the RSSCT. The aliquot should be of
sufficient volume to last for one or two days as the RSSCT influent. Depending on the
2-28
-------
temperature of storage and aliquot volulme, it can take several hours to bring the water to
room temperature. Furthermore, as the pH of the batch influent water may change during
storage, the influent pH of each aliquot should be measured and adjusted if necessary. For
some waters, it may be necessary to readjust the pH of the aliquot as it is used.
Sampling is described hi Table 5-0 and hi the following paragraphs. A minimum of two
samples for alkalinity, total and calcium hardness, ammonia and bromide should be taken
from the batch influent to the RSSCT; one at the startup and one midway through the run. A
minimum of three samples for pH, temperature, TOC, UV^ and SDS for THMs, HAA6,
TOX, and chlorine demand should be taken from the batch influent to the RSSCT: one at the
start-up, one midway through the run and one at the end. A minimum of 12 effluent samples
and three duplicate effluent samples at both 10 and 20 min full-scale equivalent EBCTs are
required.
Effluent samples should be taken after the first hour and then at 5% to 8% increments of
the average influent TOC. The average influent TOC is defined as, the running average of the
influent TOC at the tune of sampling. This approach requires a very quick turn around time
for TOC analysis. The intent is to yield a good assessment of the breakthrough with a
minimum number of samples. More frequent effluent monitoring for TOC may be necessary
hi order to sample for the other analytes at the 5% to 8% increment of the average influent
TOC.
Three duplicate samples of the RSSCT effluent should be taken at different tunes on the
breakthrough curve. One should be taken with the third or fourth sample, one with the
seventh or eighth sample and one with the tenth or eleventh sample. Because of the unsteady
state nature of GAG adsorption, i.e., the effluent concentration increases with operation tune,
the duplicate sample, if taken after the normal sample, may yield different results, as the tune
to collect a sample can be quite long. To overcome this problem, a sample volume
sufficiently large to satisfy both the normal and duplicate sample should be collected in one
container and then split hi two samples.
The breakthrough of UV254 often parallels that of TOC and may be used to estimate the
sampling tunes. However, UV-absorbing substances are normally better removed by GAG
and their breakthrough lags behind that of TOC. Thus, the relationship between UV254 and
TOC in the GAC effluent needs to be established. This requires the use of TOC and UV254
data from a previous GAC run. Thus, if this monitoring approach is to be used, prior
experience with GAC treatment of the site-specific water is needed.
If on-site TOC analysis or if a fast turn-around of TOC measurement results is not
available, a sampling plan may be estimated by the following procedure. The tune in days to
50 percent TOC breakthrough, t;0, for the large-scale column can be estimated by Equation
5.10 for an EBCT hi minutes. Based on the shape of previous TOC breakthrough curves a 1-
7-3-1 sample plan is recommended for the 12 effluent samples. The first sample is taken
after one hour of RSSCT operation, seven additional samples are taken at regular tune
intervals through the 50 percent breakthrough, three samples are taken at regular tune
2-29
-------
intervals after 50 percent breakthrough and one sample is taken at the end of the run. The
sample time interval hi large scale operation days, t^, for the first half of the TOC
breakthrough can be estimated from the following:
(5-14)
Since TOC breakthrough curves are not symmetrical and tend to have a lower slope after 50
percent breakthrough, the sample time interval after 50 percent breakthrough should be
estimated as 50 percent longer than t^ (1.5 tjnt).
Table 5-4 presents a general sample plan and sample plans for ten examples with TOC0
values of 2, 2.5, 3, 4 and 6 mg/L and EBCTLCs of 10 and 20 minutes. To calculate the actual
sample tunes for the RSSCT the sample time values (scaled days) in Table 5-4 are divided by
the scaling factor for the specific RSSCT. The exception is the first sample, which is
sampled after one hour of RSSCT operation for all cases. Figure 2-3 and Equations 5.9, 5.10
and 5.14 are utilized in this method of estimating a sample plan. This approach is based on a
data set limited to 17 water sources and one general GAC type, and may not be valid in all
cases. Based on the work of Hooper et al. (1995), for a given TOC if the pH value of the
water to be treated is 7.0 or below, then the BV50 value may be 10 to 40 percent higher than
that calculated by Equation 5.9.
It is strongly suggested that on-site TOC or fast turn-around TOC measurement be used
whenever possible to characterize the breakthrough and accurately determine the sample
times.
§ 141.144(b)(l)(i) of the ICR Rule states that both the 10 and 20 minute EBCT RSSCTs
shall be run until (a) the effluent TOC concentration is 70% of the average influent TOC
concentration on two consecutive sample dates that are at least two full-scale equivalent weeks
apart, or (b) after 50% breakthrough a plateau is reached in which the effluent concentration
does not increase over a two full-scale equivalent month period by more than 10% of the
average influent. In all cases, the maximum run length is one full-scale equivalent year.
RSSCTs at both EBCTs shall be conducted quarterly over one year in order to capture the
seasonal variation in water quality and adsorption behavior. Thus, a total of four RSSCTs at
each EBCT are required. If after completion of the first quarter RSSCTs it is found that the
effluent TOC reaches 70% of the average influent TOC within 20 full-scale equivalent days
on the EBCT = 10 nun test or within 30 full-scale equivalent days on the EBCT = 20 min
test, then the last three quarterly tests shall be conducted using bench-scale testing with only
one membrane, as described in Part 3 of this manual.
Based on the information used to generate Figure 2-3 it is anticipated that waters that have
both TOC values above 6 mg/L and pH values above 7 in the GAC influent will have very
early breakthrough and yield times to 70% breakthrough below those limiting values
discussed above. Thus, it is recommended that utilities with treated water TOC values above
6 mg/L evaluate optimized pretreatment processes prior to GAC as described in Section 1.0
2-30
-------
or run treatment studies with membranes, as described in Part 3.0, instead of GAG. Waters
with low pH, less than 7, may yield longer 70% breakthrough times and thus, low pH waters
with TOG values in the 6 to 8 mg/L range may yield acceptable breakthrough times.
The RSSCT design parameters, sampling times and results should be reported in
accordance with Tables 5-5 through 5-9. Samples should be taken according to the
procedures described in the "ICR Sampling Manual" (EPA 814-B-96-001). The approved
analytical methods for the analytes listed in Table 5-0 are listed in Table 7 § 141.142 of the
ICR Rule. These methods and laboratory QA/QC plans are described hi "DBP/ICR
Analytical Methods Manual" and must be used by all systems conducting treatment studies.
Guidance for the simulated distribution system (SDS) test and the chlorine demand test are
given in Section 6.0 of this document.
5.4 Design Example
To illustrate the design of an RSSCT the following example is provided. The design
equations are provided hi Table 5-1 and hi Sections 5.1 and 5.2.
Example Problem: Design a RSSCT with a 20 minute EBCT for a water with a GAG influent
TOG concentration of 3.0 mg/L.
1) Assume that the GAG column will be placed after a conventional filter; termed post-filter
GAG contactor. Therefore, a 12x40 mesh GAG (dLC = 1.1 mm) will likely be utilized.
2) A 100x200 mesh GAG (dsc = 0.112 mm) is selected for use hi the RSSCT. This particle
size has been widely used and is large enough not to cause problems with excessive pressure
build-up and small enough to yield short RSSCT run tunes. Using Equation 5.1, this
combination of particle sizes yields the following SF
SF = dLC/dsc = 1.1 mm / 0.112 mm = 9.82
3) The EBCT of the RSSCT can be calculated using Equation 5.3
EBCTSC = EBCTLC/SF = 20 min/9.82 = 2.04 min
4) Assuming a Resc min of 0.5, a bed porosity of 0.45 and a temperature of 20°C, which yields
a kinematic viscosity of 1.0-10'6 m2/s, Equation 5.8 hi the form presented hi Table 5-1,
yields the following vsc,
vsc = SF-vLC-(Resc>min/ReLC) = Resc,min-vLC-eLC/dsc
= 0.5-MO-6(m2/s)-0.45/ 0.112 (mm)- MO3 (mm/m)
= 0.00201 m/s = 0.120 m/min = 12.0 cm/min
= 7.23 m/h
5) The length of the RSSCT bed can be calculated from
2-31
-------
lsc = vsc-EBCTsc = 0.120 m/min 2.04 min = 0.245 m
= 24.5 cm
6) Assuming a RSSCT column with an inner diameter of 8.0 mm (0.8 cm), yields the
following flow rate, Qsc,
QSC = vsc-7i;-(DCsc)2/4 = 12.0 cm/min it (0.8 cm)2/4
= 6.06 cnrVmin
= 6.06 ml/min
7) The mass of GAC needed for the RSSCT can be calculated with Equation 5.13 once the
density of the 100x200 mesh GAC has been calculated, as described in Section 5.2. A typical
value of psc is 0.5 g/cm3,
msc = lsc'Psc-[TC(DCsc)2/4]
= 24.5 cm 0.5 g/cm3 -[7t(0.8 cm)2/4]
= 6.16g
8) To calculate the total run time of the RSSCT, use Equations 5.9 (valid for bituminous coal
based GACs) and 5.10 and the influent TOC to yield the bed volumes and tune to 50%
breakthrough,
BV50 = 21,700-TOCo-1-3 = 21.700-3.fr1-3
= 5200 bed volumes
tso = BV50-EBCTLC/1440 = 5200-20 min 71440
= 72.2 days
Assuming that the total run time is twice that to 50% breakthrough (Equation 5.11) yields
tLCT = 2-tso = 2-72.2 days
= 144 days
and using Equation 5.4 the RSSCT run time is
tscT = tLCT/SF = 144 days / 9.82
= 14.7 days
9) The minimum volume of influent water needed for the RSSCT can be calculated from
Equation 5.12
Vsc = Qsc'tsc7 = 6-06 ml/min 14.7 days -1440 mm/day
= 128,000 ml
= 128 L
2-32
-------
The volume actually sampled should be at least 30% larger than Vsc,
Total GAG influent volume = 1.3-VSC = 1.3-128 L
= 166 L
A summary of the design for this example is provided below, as are the design values for
a RSSCT with a 10 min EBCT (same influent TOG, 3 mg/L; particle size, dsc = 0.112 mm;
and column diameter, DCSC = 0.8 cm, as the 20 minute EBCT system).
Parameter EBCT=20 min EBCT=10 min
SF 9.82 9.82
EBCTSC, min 2.04 1.02
vsc, m/h 7.23 7.23
Isc, cm 24.5 12.3
Qsc, ml/min 6.06 6.06
msc, g 6.16 3.08
BV50, bed volumes 5200 5200
tso, days 72.2 36.1
tLCT, days 144 72.2
tscT, days 14.7 7.4
Vsc, L 128 64
Total influent
volume, L 166 84
The sampling times can be estimated from Table 5-4 for a bituminous coal based GAG.
2-33
-------
-------
§
1
1.0-
0.8-
0.6-
0.0-
Reynolds number
o0.05RSSCT
° 0.1RSSCT
*0.5 RSSCT
*1.0 RSSCT
»5 full-scale
I
1000
2000 3000 4000
RSSCT operation time (min.)
Figure 5-1. Impact of Reynolds number on breakthrough
DOC0= 6.3 rag/L
VLC= 7.2 m/hr
dLC"l.l mm
EBCTj^ 10 min
dj^O.ll mm
EBCT^l.O min
SF=10
5000
6000
Pressure gauge
Column
Coarse screen
Fine screen
Coarse screen
Column
end piece
Tubing
Figure 5-2 RSSCT set-up
-------
1.0
0.9-
4? O-8-
ef 0.7-
| 0.5-
I 0.4-
.§ 0.3-
I 0.2-
^ 0.1-
0.0'
"hour 1
0 30 60 90 120 150
Operation time (hours)
Figure 5-3 RSSCT breakthrough and sampling
180
210
240
-------
Table 5-0. Sampling of GAC Bench-scale Systems
Sampling Point
GAC Influent
GAC Influent
GAC Effluent @
EBCT=10 min (scaled)
GAC Effluent @
EBCT=20 min (scaled)
Analyses
Alkalinity, total & calcium
hardness, ammonia and
bromide.
pH, turbidity, temperature,
TOCandUV^. SDS1 for
THM4, HAA6, TOX, and
chlorine demand.
pH, temperature, TOC and
UV254. SDS1forTHM4,
HAA6, TOX, and chlorine
demand.
pH, temperature, TOC and
UV^. SDS'forTHlVM,
HAA6, TOX, and chlorine
demand.
Sample Frequency2
Two samples per batch of
influent evenly spaced over
the RSSCT run.
Three samples per batch of
influent evenly spaced over
the RSSCT run.
A minimum of 12 samples.
One after one hour, and
thereafter at 5% to 8%
increments of the average
influent TOC.
A minimum of 12 samples.
One after one hour, and
thereafter at 5% to 8%
increments of the average
influent TOC.
1 - SDS conditions are defined hi Part 1, Section 4.6 of this document. Additional
guidance is found hi Section 6.0 of this Part.
2 - Three duplicate effluent samples are required at each EBCT.
-------
Table 5-1. RSSCT proportional diffusivity (X=l) design equations
bed volume bed depth 1
= - - = - ; =
flowrate velocity v
flowrate _ Q
cross sectional area TiDC2 / 4
VSC ""
T>
Re =
us
(^*'*'^SCtmin \ _
n RP '~'
aSC X^LC
*SC
_v
sc ~ vsc
"sc = Qsc*sc
Notation:
Subscripts:
d= particle diameter (L)
D= intrapaticle diffusion coefficient (L2/T)
DC= column diameter (L)
EBCT= empty bed contact time (T)
1= bed depth (L)
Q= volumetric flowrate (L3/T)
Re= Reynolds number (-)
SF= scaling factor (-)
t=runtime(T)
v= hydraulic loading or superficial velocity (L/T)
V= minimum required water volume (L3)
s= bed porosity (-)
{4== dynamic viscosity (M/L-T)
pw= density of water (M/L3)
u= kinematic viscosity (L2/T)
LC= large particle column
SC= small particle column
Superscript:
T: total
-------
Table 5-2. RSSCT design for full-scale 8x30 mesh GAC (dLC= 1.6 mm)
mesh size (US std.)
dsc (mm)
SF(-)
vsc(m/h)
tsc (days)1
60x100
0.200
8.00
4.05
22.5
100x200
0.112
14.3
7.23
12.6
170x230
0.076
21.0
10.7
8.55
230x325
0.054
29.6
15.0
6.08
EBCTLc= 10 min
EBCTsc(min)
lsc (mm)
1.25
84.4
0.70
84.4
0.48
84.4
0.34
84.4
EBCTLc= 20 min
EBCTsc(min)
lsc (mm)
DCSC= 8.0 mm
Qsc (ml/min)
Vsc (liters)1
2.50
169
1.40
169
0.95
169
0.68
169
3.39
110
6.06
110
8.93
110
12.6
110
DCsc=11.0mm
Qsc (ml/min)
Vsc (liters)1
6.41
208
11.5
208
16.9
208
23.8
208
Assumptions: ReSC)rain= 0.5, 0^= 1.0xlO-« m2/s (T= 20°C), 8^= 0.45
1) tLC= 180 days
-------
Table 5-3. RSSCT design for full-scale 12x40 mesh GAC (dLC= 1.1 mm)
mesh size (US std.)
dscCmm)
SF(-)
vsc(m/h)
tsc(days)'
60x100
0.200
5.50
4.05
32.7
100x200
0.112
9.82
7.23
18.3
170x230
0.076
14.5
10.7
12.4
230x325
0.054
20.4
15.0
8.84
EBCTLc= 10 min
EBCTsc(min)
lsc (mm)
1.82
123
1.02
123
0.69
123
0.49
123
EBCTLC= 20 min
EBCTSC (min)
lsc(mm)
3.64
245
2.04
245
1.38
245
0.98
245
DCSC- 8.0 mm
Qsc (ml/min)
Vsc (liters)'
3.39
160
6.06
160
8.93
160
12.6
160
DCsc=11.0mm
. Qsc (ml/min)
Vsc (liters)^
6.41
302
11.5
302
16.9
302
23.8
302
Assumptions: Resc>min= 0.5, ULC= l.OxlO-6 m2/s (T= 20°C), SLC= 0.45
1) t^= 180 days
-------
H
G
co
&
CO
1
I
*
co"
G
co
P
J
CO
I
o
vd
II
G
g
I
o
I
o
CO
O
I
c
'
H '
U «
«
O
@2
H -a
G o
03 o
a ^
11 a
H '3
G s
QQ O
H -*
II .d
G »
DO o
H ^
PQ
B
QQ O
H ^
CO
H-
"a,,
CO
r^ 3
^oO
r
M-l 10 rf>
in" ^ in § 12L
CO
I
.9
13
Ed4 °
s i
C^ CQ
800
bp >-i
m
g
r
t"1
co
0)
S & O
S H
N S 3
_J co o
CN
w
o>
-
8
a
J3
a
co
II 3
& -3
9
-------
Table 5-5. RSSCT DESIGN PARAMETERS
Utility name and address
Run No.
Utility ID number
Contact person
Phone number
FAX number
GAC type and manufacturer
Original GAC mesh size
Diameter of full-scale GAC, d^
EBCTLC= 10 min
Diameter of GAC used in RSSCT, dsc
Scaling factor, SF
EBCTSC
Minimum Reynolds number, Resc nlin
Kinematic viscosity, u
Bed porosity, SLC
Superficial velocity, vsc
Bed depth, lsc
RSSCT column inner diameter, DCSC
Bed (apparent) density (dry) of GAC, psc
Mass (dry) of GAC, msc
Volumetric flow rate,
= 20 min
Diameter of GAC used in RSSCT, dsc
Scaling factor, SF
EBCTSC
Minimum Reynolds number, ReSQmfa
Kinematic viscosity, u
Bed porosity, e^
Superficial velocity, vsc
Bed depth, lsc
RSSCT column inner diameter, DCSC
Bed (apparent) density (dry) of GAC, psc
Mass (dry) of GAC, msc
Volumetric flow rate, Qsc
US std mesh
mm
mm
min
. m2/s
m/hr
mm
mm
_g/ml
grams
ml/min
mm
_(-)
min
.(-)
m2/s
.'(-)
.m/hr
mm
mm
_g/ml
grams
ml/min
-------
Table 5-6. RSSCT SAMPLING - EBCTLC= 10 min
Utility name and address
Run No.
Utility ID number
Contact person
Contact phone number
Contact FAX number
GAC Influent (A group)
alk, TH, CaH, NH4-N, Br
Sample Al-10
Sample A2- 10
GAC Influent (B group)
pH, turb, temp, TOC, UV, and SDS
Sample B 1-10
Sample B2-10
Sample B3-10
GAC Effluent (C group)
pH, temp, TOC, UV, and SDS
Sample Cl-10
Sample C2-10
Sample C3-10
Sample C4-10
Sample C5-10
Sample C6-10
Sample C7-10
Sample C8-10
Sample C9-10
Sample CIO- 10
Sampled 1-10
Sampled 2- 10
Sample Dl-10
Sample D2- 10
Sample D3-10
Date
(M/D/Y)
Time
Operation
time (hr)
0
0
1
-------
Table 5-7. RSSCT SAMPLING - EBCTLC= 20 min
Utility name and address
Run No.
Utility ID number
Contact person
Contact phone number
Contact FAX number
GAC Influent (A group)
alk, TH, CaH, NHU-N, Br
Sample Al-20
Sample A2-20
GAC Influent (B group)
pH, turb, temp, TOC, UV, and SDS
SampleB 1-20
Sample B2-20
Sample B3-20
GAC Effluent (C group)
pH, temp, TOC, UV, and SDS
Sample Cl-20
Sample C2-20
Sample C3-20
Sample C4-20
Sample C5-20
Sample C6-20
Sample C7-20
Sample C8-20
Sample C9-20
Sample C 10-20
Sampled 1-20
Sample C 12-20
Sample D 1-20
Sample D2-20
Sample D3-20
Date
(M/D/Y)
Time
Operation
time (hr)
0
0
1
-------
Table 5-8. RSSCT RESULTS - EBCTLC= 10 min
Utility name and address
Run No.
Utility ID number
Contact person
Contact phone number
Contact FAX number
GAC Influent (Batch)
Group A
Sample Al-10
Sample A2- 10
Group B
Sample Bl-10
Sample B2- 10
Sample B3-10
Group B-SDS
Sample Bl-10
Sample B2- 10
Sample B3-10
Group B-SDS
Sample Bl-10
Sample B2-10
Sample B3-10
Group B-SDS
Sample Bl-10
Sample B2-10
Sample B3-10
alk
pH
Cl dose
TOX
HAA6
TH
temp.
Cl res.
THM4
MCAA
DCAA
CaH
turb.
CD
CHC13
TCAA
MBAA
NH4-N
TOC
temp.
CHBrCl2
DBAA
Br
UV
PH
CHBr2Cl
BCAA
TBAA
H. time
CHBr3
CDBAA DCBAA
Units: alk, TH, CaH- mg/L as CaCO3; NH4-N, TOC, Cl dose, Cl res., CD- mg/L;
turb- ntu; UV- cm"1; temp- °C; Br and SDS- ug/L; H. time- hrs
-------
RSSCT RESULTS - EBCTLC= 10 min (page 2)
Run No.
GAC EFFLUENT
Utility ID #
Group C
Sample Cl-10
Sample C2-10
Sample C3-10
Sample C4- 10
Sample C5-10
Sample C6-10
Sample C7-10
Sample C8-10
Sample C9-10
Sample C10-10
Sampled 1-10
Sample C12-10
Sample D 1-10
Sample D2-10
Sample D3-10
Group C-SDS
Sample Cl-10
Sample C2-10
Sample C3-10
Sample C4-10
Sample C5-10
Sample C6-10
Sample C7-10
Sample C8-10
Sample C9-10
Sample C10-10
Sampled 1-10
Sample C12-10
Sample D 1-10
Sample D2-10
Sample D3-10
PH
Cl dose
temp.
Clres.
CD
TOC
temp.
UV
PH
H. time
Units: TOC, Cl dose, Cl res., CD- mg/L; turb- ntu; UV- crrr1;
temp- °C; H. time- hrs
-------
RSSCT RESULTS - EBCTLC= 10 min (page 3) Run No..
Utility ID # _
GAC Effluent
Group C-SDS
Sample Cl-10
Sample C2-10
Sample C3-10
Sample C4-10
Sample C5-10
Sample C6-10
Sample C7-10
Sample C8-10
Sample C9-10
Sample CIO- 10
Sampled 1-10
Sample C12-10
Sample Dl-10
Sample D2-10
Sample D3- 10
Group C-SDS
Sample Cl-10
Sample C2-10
Sample C3-10
Sample C4-10
Sample C5-10
Sample C6-10
Sample C7-10
Sample C8-10
Sample C9-10
Sample CIO- 10
Sampled 1-10
Sampled 2- 10
Sample Dl-10
Sample D2- 10
Sample D3-10
TOX
HAA6
THM4
MCAA
DCAA
CHC13
=
TCAA
MBAA
CHBrCl2
DBAA
CHBr2Cl
BCAA
TBAA
CHBr3
CDBAA
DCBAA
Units: SDS- ug/L
-------
Table 5-9. RSSCT RESULTS - EBCTLC= 20 min
Utility name and address
Run No._
Utility ID number
Contact person
Contact phone number
Contact FAX number
GAC Influent (Batch)
Group A
Sample Al-20
Sample A2-20
Group B
Sample Bl-20
Sample B2-20
Sample B3-20
Group B-SDS
Sample Bl-20
Sample B2-20
Sample B3-20
Group B-SDS
Sample Bl-20
Sample B2-20
Sample B3-20
Group B-SDS
Sample Bl-20
Sample B2-20
Sample B3-20
alk
pH
Cl dose
TOX
HAA6
TH
temp.
Clres.
THM4
MCAA
DCAA
CaH
0
turb.
CD
CHC13
TCAA
MBAA
NH4-N
TOC
temp.
CHBrCl2
DBAA
Br
UV
pH
CHBr2Cl
BCAA
TBAA
H. time
CHBr3
CDBAA DCBAA
Units: alk, TH, CaH- mg/L as CaCO3; NH4-N, TOC, Cl dose, Cl res., CD- mg/L;
turb- ntu; UV- cm'1; temp- °C; Br and SDS- fig/L; H. time- hrs
-------
RSSCT RESULTS - EBCTLC= 20 min (page 2) Run No..
Utility ID # _
GAC EFFLUENT
Group C
Sample Cl-20
Sample C2-20
Sample C3-20
Sample C4-20
Sample C5-20
Sample C6-20
Sample G7-20
Sample C8-20
Sample C9-20
Sample C 10-20
Sampled 1-20
Sample C 12-20
Sample Dl-20
Sample D2-20
Sample D3 -20
Group C-SDS
Sample Cl-20
Sample C2-20
Sample C3-20
Sample C4-20
Sample C5-20
Sample C6-20
Sample C7-20
Sample C8-20
Sample C9-20
Sample C 10-20
Sampled 1-20
Sample C 12-20
Sample Dl-20
Sample D2-20
Sample D3-20
PH
Cl dose
temp.
Cl res.
CD
TOC
temp.
UV
pH
H. time
Units: TOC, Cl dose, Cl res., CD- mg/L; turb- ntu; UV- cm'1;
temp- °C; H. time- hrs
-------
RSSCT RESULTS - EBCTLC= 20 min (page 3) Run No..
Utility ID # _
GAC Effluent
Group C-SDS
Sample Cl-20
Sample C2-20
Sample C3-20
Sample C4-20
Sample C5-20
Sample C6-20
Sample C7-20
Sample C8-20
Sample C9-20
Sample C10-20
Sampled 1-20
Sample C12-20
Sample Dl-20
Sample D2-20
Sample D3-20
Group C-SDS
Sample Cl-20
Sample C2-20
Sample C3-20
Sample C4-20
Sample C5-20
Sample C6-20
Sample C7-20
Sample C8-20
Sample C9-20
Sample C10-20
Sampled 1-20
Sample C12-20
Sample Dl-20
Sample D2-20
Sample D3-20
TOX
THM4
CHC13
CHBrCl2
DBAA
CHBr2Cl
BCAA
TBAA
CHBr3
CDBAA
DCBAA
Units: SDS- ug/L
-------
6.0 Simulated Distribution System Chlorination Conditions
According to ICR Rule § 141.144(b)(3), samples taken to monitor the pilot- and bench-
scale GAC breakthrough that will be analyzed for the formation of THM4, HAA6, and TOX,
and for chlorine demand should be chlorinated under site-specific simulated distribution
system (SDS) conditions. The SDS conditions of incubation time, temperature, pH, and
chlorine residual should be representative of actual plant and distribution system conditions at
the time of the GAC study. However, the selection and control of these conditions can be
complex, especially for samples from the GAC effluent.
6.1 Selection Of SDS Chlorination Conditions
The chlorine demand (chlorine dose subtracted from chlorine residual, based on Standard
Method 2350: Oxidant Demand/ Requirement) in the GAC effluent increases throughout the
breakthrough curve because the amount of organic matter exiting the GAC column increases
with time. A constant chlorine dose for all samples would yield decreasing chlorine residual
concentrations as samples in the beginning of the breakthrough curve will have much lower
chlorine demands than later samples. This Chlorination approach would not be representative
of full-scale GAC operation, where the chlorine dose would be varied to yield a CT or set
distribution system residual. Therefore, a constant chlorine residual approach is
recommended, whereby the chlorine dose is adjusted according to the demand of each
sample, with the goal of obtaining a target chlorine residual concentration after the specified
incubation period. The target residual should be that which represents the residual
concentration in the distribution system at a representative residence time. As will be
discussed later, a chlorine demand study can be used to determine the chlorine dose required
to yield the target chlorine residual.
The SDS incubation time should be set to match a residence time hi the distribution
system that represents average conditions for the specific distribution system at the tune of the
GAC study. The SDS incubation temperature should be chosen hi the same manner.
Depending on the alkalinity and chlorine dose, the pH of the sample can decrease after
Chlorination. Significant changes hi pH have been shown to affect DBF formation (Summers
et al. 1994c; Hooper et al. 1994). For many utilities the pH of the water will be adjusted
prior to distribution, often for corrosion control. The SDS pH should reflect this final
distribution system pH. As will be discussed later, SDS samples may be buffered prior to
Chlorination to maintain the SDS pH.
For the RSSCTs it is recommended that the SDS conditions of incubation time,
temperature, pH and chlorine residual chosen remain constant through the duration of any
given RSSCT run, although changes can be made hi the conditions chosen for each quarterly
RSSCT run to reflect seasonal variability. For pilot-scale studies the SDS conditions of
incubation tune, pH and chlorine residual should remain constant throughout the run. While
it is recognized that these conditions can change hi a distribution system over the course of a
GAC run, complications hi interpreting and utilizing the SDS-DBP data dictate that constant
2-35
-------
SDS conditions be used for these parameters. Temperature variations in the distribution
system, however, can be incorporated into the SDS conditions of a pilot-scale run. The
chlorination conditions used for each sample should be recorded in Table 4-4 for pilot-scale
studies or Tables 5-8 and 5-9 for bench-scale studies.
Some difficulty may be encountered when attempting to achieve target SDS chlorine
residuals for GAG effluent samples, as the unsteady-state behavior of GAG is reflected in
chlorine demand. This difficulty is heightened by the presence of inorganic compounds
which may exert a significant chlorine demand, and are not removed by GAG. If inorganic
demand is significant, then it may account for a large fraction of the total chlorine demand
present at the beginning of the breakthrough curve, when organic chlorine demand is well
removed by the GAG. With breakthrough, organic demand increases, while inorganic
chlorine demand remains constant, thus diminishing the effect of inorganic demand on total
demand.
For GAG effluent samples, chlorine demand usually correlates well with TOG
concentration, and this relationship can be utilized to aid hi predicting chlorine demand,
without directly accounting for inorganic demand. Figure 6-1 is a plot of chlorine demand
against TOG for Mississippi River (New Orleans, LA) water and Harsha Lake (sw Ohio)
water. For Mississippi River water, a linear curve fit has a positive y-axis intercept of 0.3,
indicating an inorganic chlorine demand of approximately 0.3 mg C12/L. Harsha Lake water
does not show an appreciable inorganic chlorine demand, as the linear curve fit passes
through the origin.
If prior experience with GAG relating chlorine demand to TOG is not available, a method
has been developed that simulates breakthrough conditions to obtain a relationship between
chlorine demand and TOG, without requiring the operation of a separate GAG column. The
method is termed a dilution study, and is based on diluting the GAG influent water to several
intermediate TOG concentrations and investigating the chlorine demand of these dilution
samples. However, while TOG is diluted, the inorganic background should not be affected
by the procedure. This is accomplished by diluting the GAG influent with water taken from
the GAG effluent very early in GAG operation, so that natural organic matter removal is .
maximized and inorganic constituents are conserved.
The following outlines the dilution study procedure. Two aliquots of water are needed:
one from the GAG influent, and one from the GAG effluent (20 minute EBCT, if possible)
taken as early as possible hi the study, so that natural organic matter removal is maximized.
The aliquot from the GAG effluent is termed the dilution aliquot. The two aliquots are
systematically mixed to form seven dilution samples with varying composition, as outlined in
Table 6-1. The volume of each dilution sample should be sufficient for TOG and UV254
analysis and to chlorinate at three doses for chlorine demand analysis. Note that regardless of
sample size, the total volume of the dilution aliquot required is 50% more than the required
volume of the GAG influent. After mixing, each of the seven dilution samples are analyzed
for TOG and UV^. Three chlorine doses for each dilution sample are determined by
multiplying the measured TOG, TOCds, by the respective chlorine demand (CD) to TOG ratio
2-36
-------
(CD:TOC) listed in Table 6-1 and adding to these values the chlorine residual required for the
SDS test.
Chlorine dose = TOCds_(CD:TOC) + target chlorine residual
Therefore, the chlorine dose for each dilution sample is bracketed with the goal of achieving a
residual in one of the three samples that is near the target chlorine residual. Note that at low
TOC concentrations, dilution samples 1 to 3, a wide range of chlorine demand to TOC ratios
are used, since the inorganic demands may dominate.
All SDS conditions of pH, temperature, and incubation time that will be followed during
the GAC study should be used for dilution study chlorination. The chlorine demand
calculated from the dose that yielded a residual nearest to the target residual for each dilution
is used to generate a plot of chlorine demand against TOC. The relationship found should be
similar in form to those shown hi Figure 6-1.
The results of the dilution study can now be used to estimate the chlorine demand of each
GAC effluent sample, after TOC analysis. It is recommended that a small aliquot from each
effluent sample be chlorinated at a dose based on the dilution study results and analyzed only
for chlorine demand prior to chlorination for DBF analysis. Adjustments hi the chlorine dose
can then be made if needed, since the adsorbability or chlorine reactivity can naturally change
with tune.
6.2 Uniform Formation Conditions Approach
A standardized approach to representative chlorination conditions has been developed
(Summers et al., 1996; Summers et al., 1994c; Hooper et al., 1994) and has resulted in a set
of uniform formation conditions (UFC) for chlorination:
Incubation tune: 24 + 1 hours
Incubation temperature: 20.0 ± 1.0°C
pH: 8.0 ± 0.2
24-hour chlorine residual: 1.0 + 0.4 mg C12/L
These conditions are based on average conditions reported hi the AWWA Water Industry
Database, where the average mean residence time hi the distribution system was reported as
1.3 days and the average mean chlorine residual hi the distribution system was reported as
0.9 mg C12/L. The UFC test pH was selected to represent the impact of the Lead and Copper
Rule on distribution system pH. The incubation temperature was standardized so that
variations in ambient laboratory temperatures would not affect DBF formation.
The UFC test may be used as described, or specific parameters may be changed as
necessary to reflect specific distribution system conditions. For example, the incubation
temperature for chlorination of an RSSCT operated during the whiter could be set at 10 °C to
reflect seasonal conditions. The pH could be buffered to a different value, or allowed to
2-37
-------
remain unbuffered to represent ambient conditions. The intent of the UFC test or a modified
version of the UFC test is to provide constant chlorination conditions to allow for an accurate
assessment of the breakthrough patterns of DBF precursors. The chlorination conditions
(chlorine dose, chlorine residual, pH, incubation temperature, and incubation time) used for
each sample should be recorded hi Table 4-4 for pilot-scale studies or Tables 5-8 and 5-9 for
bench-scale studies. The proposed standard operating procedure for the UFC test is included
in Appendix 2-B.
2-38
-------
O
O)
2.5
,0
1.5
(0
I 1.0
0)
.o
O 0.5
0.0
Mississippi River water 0.953
Harsha Lake water 0.961
0.0 0.5 1.0 1.5 2.0 2.5
TOG (mg/L)
Figure 6-1 Correlation between chlorine demand and TOC for two waters
3.0
Table 6-1 Dilution study parameters
Sample
number
1
2
3
4
5
6
7
Influent water
0%
10%
20%
35%
50%
70%
100%
Dilution
water
100%
90%
80%
65%
50%
30%
0%
CD:TOC (mg Cl2/mg C)
Low
0.5
0.5
0.5
0.3
0.3
0.3
0.3
Target
2.5
2.0
1.5
1.3
1.0
1.0
1.0
High
5.0
.3.5
2.5
2.3
1.7
1.7
1.7
-------
-------
7.0 Cost Estimations
Using the breakthrough data from the pilot- and bench-scale DBF precursor removal
studies, estimations of the run time to different effluent criteria will be made. These effluent
criteria include, but are not limited to, the D/DBP Rule Stage 2 'place holders' of 40 and 30
jttg/L for THM4 and HAAS, respectively. These run times can be used to calculate a carbon
usage rate, GAC mass per volume of water treated (kg/m3 or lbs/1000 gal). Using the carbon
usage rate, costs estimates to achieve different effluent criteria can be made for different
system sizes under several design options, e.g. on-site versus off-site reactivation or steel
pressure vessels versus concrete gravity contactors. Cost estimates will be based on the use
of GAC contactors placed after existing filters, termed post-filter adsorbers.
To better estimate the national costs for different effluent criteria, assessments of site-
specific costs are needed. Site-specific cost information is requested from each utility
performing pilot- or bench-scale DBF precursor removal studies. The requested cost
information is listed in Table 7-1, along with example parameter values (Clark and Adams,
1991). If the specific value for a requested parameter is not available, then enter NA as its
value in the table. In addition to the standard cost parameters, estimates of the costs of
modifying the existing treatment plant to accommodate the addition of GAC post-filter
adsorbers is requested.
2-39
-------
-------
Table 7-1 Input data for cost model
M
;3
o
OH
CO
m.
1
"H,
tS
Cost Parametel
0
T-H
^
CAPITAL RECOVERY INTEREST RATE (
8
CAPITAL RECOVERY PERIOD (YEARS)
>n
P
CO
O
U
§
P
L>
D
ONSTR
OVERHEAD & PROFIT FACTOR (% OF C
m
P
CO
O
O
55
0
P
u
1
CO
55
SPECIAL SITEWORK FACTOR (% OF CO
o
T-H
P
CO
0
O
g
6
£
B
CO
§
o
fe
CONSTRUCTION CONTINGENCIES (% 0
o
P
CO
0
O
55
O
P
STRUC'
ENGINEERING FEE FACTOR (% OF CON
^
^
t
m
ON
&
|
cn"
ON
i i
-.j
K
w
cq
CO
CQ
ENR CONSTRUCTION COST INDEX (CCI
-
ON
vo
rn
&
^
o1
o
^H
ON
1
PRODUCERS PRICE INDEX (PPI BASE Y]
m
T-H
LABOR RATE + FRINGE ($/MANHOUR)
o
§
LABOR OVERHEAD FACTOR (% OF LAE
oo
O
o
ELECTRIC RATE ($/KWH)
ON
oo
O
FUEL OIL RATE ($/GALLON)
>n
>n
o
0
D
L)
CO
5
i-J
p
55
c5
PROCESS WATER RATE ($71000 GAL)
VI
I
O
1
so
§
O
s
MODIFICATIONS TO EXISTING PLANT (
COST)
-------
-------
8.0 References
Benz, M. 1989. Doctoral Dissertation. University of Karlsruhe, Karlsruhe, Germany.
Chrobak, R.S., D.L. Kelleher and I.H. Suffet. 1985. Full-Scale GAC Adsorption
Performance Compared to Pilot-Plant Predictions. In Proc. of the AWWA Annual Conference.
Washington, D.C.: AWWA.
Clark, R.M. and J.Q. Adams. 1991. E.P.A.'s Drinking Water and Groundwater Remediation
Cost Evaluation: Granular Activated Carbon.
Lewis Publishers.
Crittenden, J.C., P.S. Reddy, H. Arora, J. Trynoski, D.W. Hand, D.L. Perram and R.S.
Summers. 1991. Predicting GAC Performance With Rapid Small-Scale Column Tests. Jour.
AWWA, 83(l):77-87.
Crittenden, J.C., P.S. Reddy, D.W Hand and H. Arora. 1989. Prediction of GAC
Performance Using Rapid Small-Scale Column Tests. AWWARF/AWWA.
Crittenden, J.C., J.K. Berrigan, Jr., D.W. Hand and B.W. Lykins. 1987. Design of Rapid
2Fixed Bed Adsorption Tests for Non-Constant Diffusivities. Jour. Environmental
Engineering, Amer. Society of Civil Engineers, 113(2):243-259.
Crittenden, J.C., J.K. Berrigan and D.W. Hand. 1986. Design of Rapid Small-Scale
Adsorption Tests for a Constant Surface Diffusivity. Jour. Water Pollution Control Fed.,
58(4):312-319.
Cummings, L. and R.S. Summers. 1994. Prediction of Field-scale GAC Control of DBF
Formation Using RSSCTs. Jour. AWWA, 86(6): 88-97.
DeMarco, J., and N. Brodtmann. 1984. Prediction of Full-Scale Plant Performance From
Pilot Columns. In Adsorption Techniques in Drinking Water Treatment. Edited by P.V.
Roberts, R.S. Summers and S. Regli. USEPA 570/9-84-005. Washington, D.C.: Government
Printing Office.
DeMarco, J., R. Miller, D. David and C. Cole. 1983. Experiences hi Operating a Full-Scale
Granular Activated Carbon System With On-site Reactivation. In Treatment of Water by
Granular Activated Carbon. Edited by M.J. McGuire and I.H. Suffet. Advanced in
Chemistry Series 202. Washington, D.C.: American Chemical Society.
Frick, B.R. 1982. Theoretische Betrachtungen zu den Problemen des Scale-up von
Aktivkohlefestbettadsorbern, Heft 20, Karlsruhe, Germany, Engler-Bunte-Institut University
of Karlsruhe.
2-41
-------
Hineline, D.W., J.C. Crittenden and D.W. Hand. 1987. The Use of Rapid Small-Scale
Column Tests to Predict Full-Scale Adsorber Capacity and Performance. In Proc. of the
AWWA Annual Conference. Kansas City, MO: AWWA.
Hooper S., R.S. Summers, H. Shukairy and D. Owen. 1994. Development of a New Test for
the Assessment of Disinfection By-Product Formation: Uniform Formation Conditions; In
Proc. of the AWWA Water Quality Technology Conference, San Francisco, CA: AWWA.
Hooper S., G. Solarik, S. Hong, R.S. Summers and D. Owen. 1995. The Impact of
Optimized Coagulation on Granular Activated Carbon Adsorption of Natural Organic Matter
and Disinfection By-Product Control. In Proc. of the AWWA Annual Conference, Anaheim,
CA: AWWA.
Malcolm Pirnie. 1990. Report to South Central Connecticut Regional Water Authority,
Granular Activated Carbon Pilot Study Program.
McGuire, M.J., M.K. Davis, S. Liang, C.H. Tate, E.M. Aieta, I.E. Wallace, D.R. Wilkes,
J.C. Crittenden, and K. Vaith. 1989. Optimization and Economic Evaluation of Granular
Activated Carbon for Organic Removal. AWWARF, Denver CO.
Metz, D.H., R.S. Summers and J. DeMarco. 1993. The Assessment of Preozonation,
Biotreatment and GAC Adsorption of DBP Precursors and Ozone DBPs Using the Rapid
Small Scale Column Test. In Proc. of the AWWA Water Quality Technology Conference.
Miami, FL: AWWA.
Namuduri, P. 1990. MS Thesis, University of Cincinnati, Cincinnati, OH.
Randtke, S.J. and V.L. Snoeyink. 1983. Evaluating GAC Adsorptive Capacity. Jour. AWWA,
75(8):406-413.
Semmens, M.J. et al. 1986a. Influence of pH on the Removal of Organics by Granular
Activated Carbon. Jour. AWWA, 78(5):89.
Semmens, MJ. et al. 1986b. Influence of Coagulation on the Removal of Organics by
Granular Activated Carbon. Jour. AWWA, 78(8):80.
Sontheimer, H., J.C. Crittenden, and R.S. Summers. 1988. Activated Carbon for Water
Treatment, 2nd ed., DVGW-Forschungsstelle, University of Karlsruhe, West Germany.
Speth, T.F., B.W. Lykins and RJ. Miltner. 1989. The Use of Pilot Columns for Predicting
Full-Scale GAC Performance, In Proc. AWWARF/USEPA Conference for the Design and Use
of Granular Activated Carbon: Practical Aspects. Cincinnati, OH: AWWA.
2-42
-------
Summers, R.S., M. Benz, P. Hermann and H. Sontheimer. 1989. Adsorption of Dissolved
Organic Carbon by Activated Carbon: Scale-Up, Modeling, and Desorption. Presented at the
AWWA Annual Conf., June 19-22, at Los Angeles, CA.
Summers, R.S. and J.C. Crittenden. 1989. The Use of Mini-columns for Predicting Full-scale
GAC Performance. In Proc. AWWARF/USEPA Conference for the Design and Use of
Granular Activated Carbon: Practical Aspects. Cincinnati, OH: AWWA.
Summers, R.S., L. Cummings, J. DeMarco, D. Hartman, D. Metz, E.W. Howe, B. Macleod
andM. Simpson. 1992. Standardized Protocol for the Evaluation of GAC.: AWWARF,
Denver, CO.
Summers, R.S., M. Benz, H.M. Shukairy and L. Cummings 1993. Effect of Separation
Processes on the Formation of Brominated Trihalomethanes. Jour. AWWA, 85(l):88-95.
Summers, R.S., S. Hooper, S. Hong and G. Solarik. 1994a. The Use of RSSCTs to Predict
GAC Field-scale Performance. In Proc. of the AWWA Annual Conference. New York, NY:
AWWA.
Summers, R.S., S. Hong, S. Hooper and G. Solarik. 1994b. Adsorption of Natural Organic
Matter and Disinfection By-Product Precursors. In Proc. of the AWWA Annual Conference.
New York, NY: AWWA.
Summers R.S., S.M. Hooper, H. Shukairy and D. Owen. 1994c. Development of Uniform
Formation Conditions for the Assessment of Disinfection By-Product Formation. In Proc. of
the AWWA Annual Conference, New York, NY: AWWA. AWWA NYC.
Summers, R.S., S.M. Hooper, G. Solarik, D.M. Owen and S. Hong. 1995. Bench-Scale
Evaluation of GAC for NOM Control. Jour. AWWA, 87(8):69-80.
Summers, R.S., S.M. Hooper, H.M. Shukairy, G. Solarik and D.M. Owen. 1996.
Assessing DBP Formation with the Uniform Formation Conditions Test. Accepted for
publication in Jour. AWWA, 88(6).
Wallace, I.E., E.M. Aieta, C.H. Tate, J.C. Crittenden, M.J. McGuire, andM.K. Davis.
1988. The Application of the Rapid Small-Scale Column Test to Model Organic Removal by
Granular Activated Carbon. In Proc. of the AWWA Annual Conference. Orlando, FL:
AWWA.
Wood, P.R. and J. DeMarco. 1980. Removing Total Organic and Trihalomethane Precursor
Substances. In Activated Carbon Adsorption of Organics from the Aqueous Phase, Vol.11.
Edited by M.J. McGuire and I.H. Suffet. Ann Arbor, Mich.:Ann Arbor Science.
2-43
-------
-------
Appendix 2-A
Sampling Activated Carbon
A sample tube, riffle splitter, sample reducer or coning and quartering technique should
be used to obtain a representative sample of the activated carbon. The first three of the above
mentioned techniques require mechanical devices, are only applicable for granular particles
and can not handle small sample sizes. The coning and quartering technique does not have
these limitations and therefore, can be applied to both GAG and ground GAG. However, it is
labor intensive especially for obtaining very small samples from large samples. The coning
and quartering procedure is illustrated in Figure A-l and described below.
1) The total volume of the as-received GAG is taken from the original container and placed
into a cone shape pile, scoop-by-scoop. Each scoop is added to the center of the pile and
allowed to flow evenly in all directions (a).
2) The pile is then evenly flattened from above to form a shallow cylinder of uniform
thickness (b).
3) This cylinder of carbon is then evenly divided into pie-shaped quarters as shown hi Figure
A-l (c and d).
4) Two opposite quarters are removed (e).
5) The remaining two opposite quarters are piled into a cone again by taking scoops from
alternate quarters (f).
6) Steps 2 through 5 are repeated until the desired sample size is obtained.
2-45
-------
-------
c)
e)
f)
Figure A-l Coning and Quartering Sampling Technique
-------
-------
Appendix 2-B
Uniform Formation Conditions (UFC) For DBF Formation
Proposed Standard Operating Procedure
Uniform Formation Conditions:
pH: 8.0 + 0.2
temperature: 20.0 ± 1.0°C
incubation time: 24 ± 1 hr
chlorine residual: 1.0 + 0.4 mg/L as free chlorine after 24 hr
Preliminary Study:
A 24-hour chlorine demand study on the water sample may be required before
dosing under UFC to determine the applied dose that will yield a chlorine residual of
1.0 mg/L after 24 hours (procedure described below).
Materials:
chlorine demand-free glassware
pH 8.0 borate buffer
pH 8.0 combined hypochlorite/buffer dosing solution
Methods:
Chlorine demand-free glassware:
Incubation bottles (amber, with TFE-faced caps): soak hi detergent (Fisher FL-
70, 4%) at least overnight, rinse 4x with hot tap water, 2x with DI water. Place hi 10-
20 mg/L chlorine solution (made with DI water) for at least 24 hours. Rinse 4x with
DI water and then l-2x with laboratory clean water; dry hi 140°C oven at least
overnight. Store dosing pipettes hi 50 mg/L C12 (made with laboratory clean water).
Rinse 3x with dosing solution prior to use, and return pipettes to storage hi chlorine
solution after use.
pH 8.0 borate buffer:
Before dosing, water samples are buffered to pH 8.0 with 2 mL/L borate
buffer: l.OM boric acid (ACS grade) and 0.26M sodium hydroxide (ACS grade) in
boiled laboratory clean water (RO/IX/GAC).
pH 8.0 combined hypochlorite/buffer dosing solution:
A combined hypochlorite/buffer solution (based on method described hi Koch et
al., "A Simulated Distribution System Trihalomethane Formation Potential Method,"
1987 AWWA WQTC) is made by buffering the hypochlorite solution to pH 8.0 with
pH 6.7 borate buffer.
2-47
-------
- pH 6.7 borate buffer: 1.OM boric acid (ACS grade) and 0.11M sodium
hydroxide (ACS grade) in boiled laboratory clean water (RO/IX/GAC).
- add pH 6.7 borate buffer to chlorine solution (1000-4000 mg C12/L) to yield a
pH 8.0 dosing solution. (A 4-5:1 volume ratio of pH 11.2 hypochlorite solution
to pH 6.7 borate buffer has been found to yield a pH 8.0 combined
hypochlorite/buffer solution, with an approximately 20% drop in chlorine
strength.)
The dosing solution (combined OC1 /buffer) chlorine strength should allow for a dosing
volume of < 0.5% of the water sample volume (e.g. 2.5 mL dosing solution in 1.0 L
bottle).
Preliminary study:
Perform a 24-hour chlorine demand study (buffered at pH 8.0 and incubated in
the dark at 20°C as described hi the dosing procedure) using a series of three chlorine
doses based on C12:TOC ratios of 1.2:1, 1.8:1, and 2.5:1, after adjusting for inorganic
demand. From the results of these tests, the chlorine dose for UFC is selected to yield
a 24-hour residual of 1.0 mg/L free chlorine.
Dosing procedure:
1. Add 2.0 mL/L pH 8.0 borate buffer to water sample
2. Adjust to pH 8.0 with H2SO4/NaOH (if necessary)
3. Fill incubation bottle 3/4 full with buffered water sample
4. Dose with combined hypochlorite/buffer solution holding pipette Just above
water surface
5. Cap bottle, invert twice
6. Fill to top with buffered water sample and cap headspace free
7. Invert 10 times
8. Incubate hi dark at 20.0°C for 24 hours
9. After incubation period, measure chlorine residual, pH, and sample for DBPs
2-48
-------
Appendix 2-C
AWWA Standard For Granular Activated Carbon
ANSI/AWWA B604-90
Contact AWWA, 6666 West Quincy Ave., Denver, CO 80235
(303) 794-7711 for information about revisions to this standard.
2-49
-------
-------
American Water Works Association
ANSI/AWWA B604-90
(Revision of AWWA B604-74)
AWWA STANDARD
FOR
GRANULAR ACTIVATED CARBON
[AMERICAN NATIONAL]
ISTANDARDI
Effective date: Feb. 1, 1991.
First edition approved by AWWA Board of Directors Jan. 28, 1974.
This edition approved June 17, 1990.
Approved by American National Standards Institute, Inc., Nov. 13, 1990.
AMERICAN WATER WORKS ASSOCIATION
6666 West Quincy Avenue, Denver, Colorado 80235
-------
AWWA Standard
This document is an American Water Works Association (AWWA) standard. It is not a specification.
AWWA standards describe minimum requirements and do not contain all of the engineering and
administrative information normally contained in specifications. The AWWA standards usually con-
tain options that must be evaluated by the user of the standard. Until each optional feature is
specified by the user, the product or service is not fully defined. AWWA publication of a standard
does not constitute endorsement of any product or product type, nor does AWWA test, certify, or
approve any product. The use of AWWA standards is entirely voluntary. AWWA standards are
intended to represent a consensus of the water supply industry that the product described will
provide satisfactory service. When AWWA revises or withdraws this standard, an official notice of
action will be placed on the first page of the classified advertising section of Journal AWWA. The
action becomes effective on the first day of the month following the month of Journal AWWA publi-
cation of the official notice.
American National Standard
An American National Standard implies a consensus of those substantially concerned with its scope
and provisions. An American National Standard is intended as a guide to aid the manufacturer, the
consumer, and the general public. The existence of an American National Standard does not in any
respect preclude anyone, whether he has approved the standard or not, from manufacturing,
marketing, purchasing, or using products, processes, or procedures not conforming to the standard.
American National Standards are subject to periodic review, and users are cautioned to obtain the
latest editions. Producers of goods made in conformity with an American National Standard are
encouraged to state on their own responsibility in advertising and promotional materials or on tags
or labels that the goods are produced in conformity with particular American National Standards.
CAUTION NOTICE: The American National Standards Institute (ANSI) approval date on the front
cover of this standard indicates completion of the ANSI approval process. This American National
Standard may be revised or withdrawn at any time. ANSI procedures require that action be taken
to reaffirm, revise, or withdraw this standard no later than five years from the date of publication.
Purchasers of American National Standards may receive current information on all standards by
calling or writing the American National Standards Institute, Inc., 1430 Broadway New York NY
10018 (212) 354-3300.
Copyright © 1991 by American Water Works Association
Printed in USA
11
-------
Committee Personnel
The AWWA Standards Committee on Activated Carbon, Powdered and
Granular, which reviewed and approved this standard, had the following personnel
at the time of approval:
Joseph A. Bella, Chair
Alan F. Hess, Vice-Chair
James L. Fisher, Secretary
Consumer Members
J.A. Bella, Passaic Valley Water Commission, Little Falls, N.J. (AWWA)
D. J. Hartman, Cincinnati Water Works, Cincinnati, Ohio (AWWA)
A.F. Hess, Regional Water Authority, New Haven, Conn. (AWWA)
F.J. Holdren, West Virginia Water Company, Charleston, W. Va. (AWWA)
W.R. Inhoffer,* Passaic Valley Water Commission, Clifton, N.J. (AWWA)
E.D. Mullen, Elizabethtown Water Company, Bound Brook, N.J. (AWWA)
M. J. Pickel, Philadelphia Water Department, Philadelphia, Pa. (AWWA)
H.L. Plowman Jr., Philadelphia Suburban Water Company,
BrynMawr, Pa. (AWWA)
C.E. Stringer, Dallas Water Utilities, Sunnyvale, Texas (AWWA)
General Interest Members
E.E. Baruth.t Standards Engineer Liaison, AWWA, Denver, Colo. (AWWA)
J.C. Crittenden, Michigan Technological University, Houghton, Mich. (AWWA)
L.L. Harms, Black & Veatch Engineers, Kansas City, Mo. (AWWA)
R.A. Hyde, WRc Swindon, Wiltshire, United Kingdom (AWWA)
C.R. James, J.M. Montgomery Consulting Engineers,
Walnut Creek, Calif. (AWWA)
B.H. Kornegay, Engineering-Science, Inc., Fairfax, Va. (AWWA)
Wolfgang Kuhn, Universitat Karlsruhe, Karlsruhe, Germany (AWWA)
R.G. Lee, American Water Works Service Company, Inc.,
Voorhees, N. J. (AWWA)
J.C. Mallevialle, Lyonnaise des Eaux, Le Pecq, France (AWWA)
S.J. Medlar, Camp, Dresser & McKee, Inc., Edison, N.J. (NEWWA)
Richard Miltner, US Environmental Protection Agency,
Cincinnati, Ohio (USEPA)
R.G. Saterdal, CH2M Hill, Inc., Denver, Colo. (AWWA)
Paul Schorr, State of New Jersey, Trenton, N. J. " (AWWA)
I.H. Suffet, Drexel University, Philadelphia, Pa. (AWWA)
W.J. Weber Jr., University of Michigan, Ann Arbor, Mich. (AWWA)
*Alternate
tLiaison, nonvoting
in
-------
Producer Members
D.A. Ainsworth, Cameron Carbon, Inc., Baltimore, Md.
R.W. Edwards,* American Norit Company, Inc., Jacksonville, Fla.
J.L. Fisher, Calgon Carbon Corporation, Pittsburgh, Pa.
J.R. Graham, Westates Carbon, Inc., Los Angeles, Calif.
M.L. Massey, Westvaco Corporation, Covington, Va.
T.N. McFerrin,* ActiCarb, Dunnellon, Fla.
J.S. Neulight,* Calgon Carbon Corporation, Bridgewater, N.J.
G. Parker, ActiCarb, Dunnellon, Fla.
R.J. Potwora, ATOCHEM North America Inc., Pryor, Okla.
D.O. Hester, American Norit Company, Inc., Marshall, Texas
(AWWA)
(AWWA)
(AWWA)
(AWWA)
(AWWA)
(AWWA)
(AWWA)
(AWWA)
(AWWA)
(AWWA)
*Alternate
IV
-------
Contents
SEC. PAGE
Foreword
I History of Standard..... vi
II Introductory Information vi
III Acceptance vii
IV Information Regarding Use
of This Standard viii
V Modification to Standard viii
VI Major Revisions viii
Standard
1 General
1.1 Scope 1
1.2 Definitions 1
1.3 Affidavit of Compliance 2
1.4 Basis for Shipment,
Acceptance, and Rejection 2
2 Materials
2.1 Physical Requirements 3
2.2 Performance Criteria 4
2.3 Impurities 4
3 Sampling, Packaging and
Shipping, and Marking
3.1 Sampling 4
3.2 Packaging and Shipping 5
3.3 Marking 5
4 Testing Methods
4.1 Samples 6
4.2 Testing Period 6
4.3 Moisture 6
4.4 Apparent Density 6
4.5 Particle-Size Distribution 7
4.6 Abrasion Resistance 9
4.7 Test Method for Iodine Number 15
4.8 Surface Area Determination 15
4.9 Pore Volume Determination 15
4.10 Water Extractables Test 15
SEC.
Appendixes
A Bibliography..
PAGE
16
Adsorptive Capacity Tests
B.1
B.1.1
B.1.2
B.I .3
Tannin Adsorption Test
Stock Tannic Acid Solution
500 mg/L 18
Test Procedure 18
Determination of Tannin in
Effluent 18
B.2 Phenol Adsorption Test
B.2.1 Reagents 20
B.2.2 Standardization of Reagents 20
B.2.3 Test Procedure 21
Figures
1 Apparent-Density Test
Apparatus... 7
2 Stirring Abrasion Unit 10
3 Testing Pan Assembly for
Ro-Tap Abrasion Test 12
4 Abrasion Testing Pan for Ro-Tap
Abrasion Test 12
Tables
1 US Standard Sieves and
Opening Sizes 8
2 Sieving Apparatus Required for
Stirring Abrasion Test. 9
3 Recommended Particle Sieve
Sizes 13
4 Di Values for Ro-Tap Abrasion
Test 15
B.I Standard Curve of Tannin
Dilution 19
-------
Foreword
This foreword is for information only and is not a part ofANSI/AWWA B604.
I. History of Standard. The first edition of AWWA B604, Standard for
Granular Activated Carbon, was approved by the AWWA Board of Directors on
Jan. 28,1974. This is the first revision to AWWA B604.
The purpose of ANSI/AWWA B604 is to provide a standard for use in prepar-
ing purchase specifications for the purchase of granular activated carbon to be used
as adsorption media for the treatment of municipal and industrial water supplies.
Powdered activated carbon is covered in ANSI/AWWA B600, Standard for Powdered
Activated Carbon.
This standard does not cover the design of carbon handling facilities or adsorp-
tion processes. Design information may be found in the Journal AWWA and in other
publications, some of which are listed in the bibliography (appendix A) to this
standard.
II. Introductory Information.
Description. Activated carbon is a form of carbon that is activated by a care-
fully controlled oxidation process to develop a porous carbon structure with a surface
area greater than 500 m2/g. This surface area gives the activated carbon the adsorp-
tive capacity to adsorb dissolved organic materials, many of which are taste- and
odor-causing substances in water.
The major raw materials used in the manufacture of granular activated car-
bons are peat, bituminous coal, coconut shell, and lignite. After preliminary process-
ing, these materials are heated to a high temperature and reacted with steam to
develop the extensive internal pore structure required for adsorption. Subsequent
processing includes crushing, grading, screening, and packaging.
Water treatment with granular activated carbon is usually accomplished by
percolating the water to be treated through fixed-adsorption beds of granular acti-
vated carbon. The granular activated carbon may be crushed and screened to any
particle size, but typical sizes used for water treatment range from No. 8 to No. 50
US standard sieve sizes.
Source of supply. Activated carbon to be used in water treatment should be
obtained from manufacturers regularly engaged in the production of activated car-
bon found satisfactory for service in the water treatment field.
Caution in handling and storage. Activated carbon will readily adsorb oxygen
from the air, creating an acute oxygen depletion hazard in confined areas.
Appropriate safety measures for oxygen-deficient atmospheres should be strictly
adhered to when entering enclosed or partially enclosed areas containing activated
carbon.
In storing activated carbon, precautions must be taken to avoid direct contact
with strong oxidizing agents, such as chlorine, hypochlorites, potassium permanga-
nate, ozone, and peroxide.
Mixing carbon with hydrocarbons (such as oils, gasoline, diesel fuel, grease,
paint thinners, and so forth) may cause spontaneous combustion. Therefore,
activated carbon must be kept separated from hydrocarbon storage or spills.
Particle-size distribution. Granular activated carbons extending over a wide
range of size distributions have given satisfactory results. The proper size distribu-
tion for a particular application cannot be specified without consideration of the
VI
-------
nature of the water to be treated, the particular treatment process in question, and
other controlling factors.
In general, coarser particle-size distributions will permit higher hydraulic load-
ing and backwash rates but will exhibit somewhat lower adsorption rates for certain
specific organic chemicals.
The particle-size distribution for a particular grade of activated carbon is
usually specified by particle-size range, effective-size range, and maximum unifor-
mity coefficient. Commonly manufactured particle-size ranges for granular activated
carbons, expressed in limiting US standard sieve sizes, include 8 x 16, 8 x 20, 8 x
30, 10 x 30, 12 x 40, 14 x 40, 20 x 40, and 20 x 50 with effective sizes ranging
from 0.35 mm to 1.30 mm. The purchaser should specify a particular-size range and
an effective-size range meeting site-specific requirements, which may include one or
more of the manufactured size ranges previously enumerated. In general, the unifor-
mity coefficient for granular activated carbon should not exceed 2.1 after backwash-
ing in the filter.
Abrasion resistance. Granular activated carbons used for municipal water
treatment are exposed to a variety of external forces during shipping, loading into
adsorption beds, and backwashing. These forces can cause activated carbon granule
crushing on impact, granule-to-granule abrasion, and the generation of undesirable
fines. Because of difficulty in devising a test that simulates the various handling
conditions that may be encountered, the industry has not yet agreed on any one
standard test for predicting activated carbon durability.
Two tests, the stirring abrasion test and the Ro-Tap abrasion test, have been
included in this standard for measuring granular activated carbon durability. It is
recognized that differences in bulk density and other physical properties of the
various manufactured activated carbons, which might not be related to durability,
influence the results obtained in using these tests. For this reason, it is current
practice to use the stirring abrasion test for lignite-based granular activated carbons
and the Ro-Tap abrasion test for bituminous-based granular activated carbons.
Adsorptive capacity. The optimum method for determining the effectiveness of
a granular activated carbon is by using water from the particular plant in question
for the test. Other tests have been developed that give an indication of a granular
activated carbon's performance under specific conditions. These tests use a very high
concentration of adsorbate to reduce the amount of time required to complete the
test. Various producers of activated carbon suggest different adsorbates to give an
index jof a carbon's performance. Examples are phenol, tannin, iodine, and molasses.
Phenol adsorption is an index of a carbon's ability to remove some types of chemical
taste and odor; tannin is representative of organic compounds added to water by
decayed vegetation; and iodine adsorption is an index of the total surface area of a
carbon. Iodine and molasses adsorption are often used to show if a carbon is
activated. An iodine adsorption test is included in this standard. Information on
tannin and phenol adsorptive capacity tests may be found in appendix B to this
standard for those purchasers who want to include these requirements in their
specifications.
in. Acceptance. Government legislative and regulatory bodies at national
and state or provincial levels promulgate rules that may control the use of products
described in ANSI/AWWA B604. AWWA does not obtain or provide information
about all of the actual or proposed regulations in the many involved jurisdictions.
The user of this standard is cautioned to determine that the use of products
described in this standard conforms to all applicable laws and regulations.
\Q1
-------
Questions concerning laws and regulations should be referred to the appropriate
regulatory agency.
Consensus standards have been developed for direct and indirect additives
from products that come in contact with potable water. Manufactured products
covered by ANSI/AWWA B604 eventually may be required to be certified to meet
those standards. Questions regarding additives should be referred to the appropriate
state regulatory agency.
IV. Information Regarding Use of This Standard. When purchasing acti-
vated carbon under the provisions of this standard, the purchaser should provide
specifications covering the following:
1. Standard usedthat is, ANSI/AWWA B604-90, Standard for Granular
Activated Carbon.
2. Quantity of granular activated carbon to be purchased in cubic feet (cubic
metres), backwashed, drained, and in placeor by weight. Activated carbon
intended for immediate placement in an adsorption bed is typically purchased by
volume, backwashed, drained, and in placeor by weight. Makeup activated carbon
or other activated carbon intended for subsequent placement is purchased on a
volume or weight basis.
3. Whether an affidavit of compliance is required (Sec. 1.3).
4. Reference sample and acceptance method (Sec. 1.4.1 and Sec. 1.4.4).
5. Particle-size range, effective size, and uniformity coefficient, if other than
that specified (Sec. 2.1).
6. Special adsorptive capacity tests (Sec. 2.2.1 and Sec. 2.2.2).
7. Provisions for reaching agreement on sampling technique (Sec. 3.1.1).
8. Method of packaging and shipping (Sec. 3.2).
9. If shipment is to be in bulk: type of rail car or hopper truck (Sec. 3.2.4);
and whether bulk shipments are to be accompanied by weight certificates of certified
weighers (Sec. 3.2.5).
V. Modification to Standard. Any modification of the provisions, defini-
tions, or terminology in this standard must be provided in the purchaser's
specifications.
VL Major Revisions. The following revisions were incorporated in this edi-
tion of ANSI/AWWA B604:
1. The standard has been revised to conform with the current style and con-
tent of AWWA standards.
2. An "Acceptance" section has been added in the Foreword.
3. The "Definitions" section has been revised and expanded.
4. All references to "contractor" have been changed to "supplier."
5. The "Basis for Shipment, Acceptance, and Rejection" section has been
revised.
6. Impurities information from the Water Chemicals Codex* has been added.
7. The minimum apparent density has been decreased to 0.25 g/cc
(Sec. 2.1.2).
8. Minimum values for surface area and pore volume have been added
(Sec. 2.1.7 and Sec. 2.1.8).
*Water Chemicals Codex, National Academy Press, 2101 Constitution Ave NW
Washington, DC 20418. ' ' ''
vm
-------
9. A maximum value for water-soluble ash (Sec. 2.1.9) and a range of effec-
tive sizes (Sec. 2.1.4) have been added.
10. The iodine test procedure from the 1974 edition has been deleted. This
edition of ANSI/AWWA B604 includes ASTM* D4607-86, Standard Test Method for
Determination of Iodine Number of Activated Carbon. Minor changes in other test
methods have also been made.
*American Society for Testing and Materials, 1916 Race St., Philadelphia, PA 19103.
IX
-------
-------
American Water Works Association
ANSI/AWWA B604-90
(Revision of AWWA B604-74)
AWWA STANDARD FOR
GRANULAR ACTIVATED CARBON
SECTION 1: GENERAL
Sec. 1.1 Scope
This standard covers granular and extruded activated carbon for use as adsorp-
tion media in the treatment of municipal and industrial water supplies.
Sec. 1.2 Definitions
The following definitions shall apply to this standard:
1.2.1 Activated carbon: A family of carbonaceous substances manufactured by
processes that develop internal porosity, thereby creating adsorptive properties.
1.2.2 Adsorption: A process in which fluid molecules are concentrated on a
surface by chemical or physical forces or both.
1.2.3 Effective size: That size opening that will just pass 10 percent of a repre-
sentative sample of a filter material; that is, if the size distribution of the particles
is such that 10 percent of a sample is finer than 0.45 mm, the filter material has an
effective size of 0.45 mm.
1.2.4 Extruded activated carbon: A form of granular activated carbon in which
the particles are uniform cylinders or pellets in shape. Effective size and uniformity
coefficient are not applicable for extruded carbons.
1.2.5 Manufacturer: The party that manufactures, fabricates, or produces
materials or products.
1.2.6 Purchaser: The person, company, .or organization that purchases any
materials or work to be performed.
1.2.7 Supplier: The party who supplies materials or services. A supplier may
or may not be the manufacturer.
1.2.8 Uniformity coefficient: A ratio of the size opening that will just pass
60 percent of a representative sample of the filter material divided by that opening
that will just pass 10 percent of the same sample.
-------
2 AWWA B604-90
Sec. 1.3 Affidavit of Compliance
When requested by the purchaser, the supplier shall provide an affidavit of
compliance stating that the activated carbon furnished complies with the applicable
provisions of this standard and the purchaser's specifications.
Sec. 1.4 Basis for Shipment, Acceptance, and Rejection
1.4.1 Reference sample. When requested, a representative sample of the
granular activated carbon shall be submitted to the purchaser for acceptance before
shipment. The sample must be submitted in clean, vaporproof containers, clearly
marked with the address of the supplier, and identified with the lot number of the
contents. A duplicate sample shall be tested by the supplier and a certified test
report shall be submitted to the purchaser with the purchaser's sample, showing
compliance with the requirements of the purchaser's specifications, along with a
statement certifying that the material for shipment is equal in quality to the sample
submitted.
1.4.2 Authorization for shipment. The purchaser may authorize shipment on
the basis of the supplier's certification of quality, or may test the reference sample
submitted by the supplier to confirm compliance before shipment is authorized.
1.4.3 Sampling and testing after delivery of shipment. The purchaser may
elect to collect a representative sample of the material after delivery. The procedure
used shall be in accordance with Sec. 3.1. One of the three sample portions taken
may be tested to determine compliance with the purchaser's specifications.
1.4.4 Acceptance. The purchaser may elect to accept the granular activated
carbon on the basis of (1) the supplier's certified test report and an accompanying
affidavit of compliance indicating the product proposed for use complies with this
standard and with the purchaser's specifications with no exceptions; (2) the
supplier's certified test report completed by a qualified third-party testing laboratory
approved by the purchaser and an accompanying affidavit of compliance; (3) the
purchaser's own testing of the reference sample submitted by the supplier and the
required affidavit of compliance; or (4) the purchaser's own testing of the repre-
sentative sample, collected according to Sec. 3.1 after receipt of shipment, showing
compliance with this standard and the purchaser's specifications. (See note in
Sec. 2.1.1.)
1.4.5 Notice of nonconformance. If the granular activated carbon delivered
does not meet the requirements of this standard or the purchaser's specifications, a
notice of nonconformance must be provided by the purchaser to the supplier within
15 working days* after receipt of the shipment at the point of destination. The
results of the purchaser's test shall prevail unless the supplier notifies the pur-
chaser within five working days of the notice of nonconformance that a retest is
desired. On receipt of the request for a retest, the purchaser shall forward to the
supplier one of the sealed samples taken according to Sec. 3.1. In the event the
results obtained by the supplier on retesting do not agree with the test results
obtained by the purchaser, the other sealed sample shall be forwarded, unopened,
for analysis to a referee laboratory agreed on by both parties. The results of the
referee's analysis shall be accepted as final. The cost of the referee's analysis shall
be paid for by the supplier if the material does not meet the requirements of this
*If testing for the removal of a specific challenge compound is required by the purchaser's
specifications, adequate time must be allowed for conformance testing.
-------
GRANULAR ACTIVATED CARBON 3
standard and the purchaser's specifications, or by the purchaser if it does meet the
requirements of this standard and the purchaser's specifications.
1.4.6 Removal or price adjustment. If the material does not meet the require-
ments of this standard and the purchaser's specifications, the supplier shall remove
it from the premises of the purchaser and replace it with a like amount of satisfac-
tory granular activated carbon or make a price adjustment acceptable to the
purchaser.
SECTION 2; MATERIALS
Sec. 2.1 Physical Requirements
2.1.1 Moisture. The moisture content of granular activated carbon shall not
exceed 8 percent, by weight, of the listed container contents as packaged or at the
time of shipment by the supplier in the case of a bulk shipment. The moisture
content shall be determined according to Sec. 4.3.
NOTE: Because ambient conditions may be beyond the control of the supplier,
the moisture content of activated carbon may increase during bulk shipment. A
moisture content exceeding 8 percent is permitted in the reference sample that is
collected after receipt of shipment (Sec. 1.4.4 and Sec. 1.4.5).
2.1.2 Apparent density.* The apparent density of the activated carbon shall be
not less than 0.25 g/cc as determined according to Sec. 4.4.
2.1.3 Particle-size distribution. Particle-size distribution shall be determined
according to Sec. 4.5. The particle-size range of the granular activated carbon shall
be as specified by the purchaser. Not greater than 15 percent of the activated carbon
shall be retained on the maximum-size sieve, and not greater than 5 percent of the
activated carbon shall pass the minimum-size sieve.
2.1.4 Effective size. The effective size of the granular activated carbon shall be
within the limits specified by the purchaser. A range from .35 mm to 1.5 mm is
available. This parameter does not apply to extruded carbons.
2.1.5 Uniformity coefficient. .Unless otherwise specified by the purchaser,
granular activated carbon shall have a uniformity coefficient not greater than 2.1
after backwashing and draining in the filter. This parameter does not apply to
extruded carbons.
2.1.6 Abrasion resistance. The retention of average particle size of granular
activated carbon shall not be less than 70 percent as determined by either the stir-
ring abrasion test or the Ro-Tap abrasion test, according to Sec. 4.6.
2.1.7 Surface area. The surface area of granular activated carbon shall not be
less than 500 m2/g as determined by the Nitrogen BET Surface Area Test, according
to Sec. 4.8.
2.1.8 Pore volume. The total pore volume of granular activated carbon shall
not be less than 0.80 cc/g as determined according to Sec. 4.9.
2.1.9 Water-soluble ash. The water-soluble ash shall not exceed 4 percent as
determined according to Sec. 4.10, Water Extractables Test.
*Backwash and drained density (in pounds) is calculated by multiplying apparent density
by 62.43 times 0.85 (for bituminous-based activated carbons).
-------
4 AWWAB604-90
Sec. 2.2 Performance Criteria
2.2.1 Adsorptive capacityiodine number. The iodine number of the granular
activated carbon shall not be less than 500 mg/g carbon as determined according to
Sec. 4.7. (See Foreword, Adsorptive Capacity, for discussion of other tests to deter-
mine special adsorptive characteristics for color, taste, and odor and specific
organics removal. These special test procedures should be specified if deemed
advisable, and the purchaser's specifications should allow adequate time to complete
testing and confirmation analysis.)
2.2.2 Additional adsorptive capacity tests. If the purchaser desires to use addi-
tional adsorptive capacity tests to measure adsorptive capacity, the purchaser shall
notify the supplier which type of shipment sampling will be required (Sec. 1.4.4).
Sec. 2.3 Impurities
2.3.1 General impurities. The granular activated carbon supplied according to
this standard shall contain no substances in quantities capable of producing delete-
rious or injurious effects on the health of those consuming water that has been
properly treated with granular activated carbon.
2.3.2 Specific impurity limits. The granular activated carbon shall not contain
specific impurities in excess of the limits listed in the Water Chemicals Codex.*
SECTION 3: SAMPLING, PACKAGING AND
SHIPPING, AND MARKING
Sec. 3.1 Sampling
3.1.1 Sampling location. If the purchaser elects to accept the material on the
basis as required in Sec. 1.4.4(4), samples shall be taken at the point of destination.
The technique of sample collection shall be agreed on by both the supplier and the
purchaser before shipment.
3.1.2 Mechanical sampling. If the granular activated carbon is handled by
conveyor or elevator or shipped in bulk, a mechanical sampling arrangement may be
used.
3.1.3 Package sampling. If the material is packaged, 5 percent of the packages
shall be sampled. No sample shall be taken from a broken package. If the packaged
material is shipped in carload quantities, one package from each lot number should
be selected for sampling, with a minimum of 20 bags sampled from each carload.
3.1.4 Sampling tube. Carbon may be sampled, by the use of a sampling tube
of at least % in. (19 mm) diameter, from bulk carload shipments or from packages.
When taking samples from packages, the sampling tube shall be extended the full
length of the package to obtain a representative sample. It should be noted that it is
virtually impossible to avoid particle fracture when using a sampling tube. Extreme
care should be taken to minimize the effect of this on particle-size distribution. Sam-
pling bulk shipping containers after shipment from the manufacturer will be subject
*Water Chemicals Codex, National Academy Press, 2101 Constitution Ave., N.W.,
Washington, DC 20418.
-------
GRANULAR ACTIVATED CARBON 5
to error caused by stratification and compaction during shipment. Extreme care
should be exercised in sampling bulk shipping containers after shipment.
3.1.5 Sample size. The gross sample shall be sealed in airtight, moistureproof
containers. Each sample container shall be labeled to identify it, and the label shall
be signed by the sampler. The gross sample shall be divided using one of the follow-
ing methods:
3.1.5.1 Mix thoroughly and divide to provide three 1-lb (0.45 kg) samples, or
3.1.5.2 Pour through a sample riffler. Repeat as necessary using the split por-
tions to provide three 1-lb (0.45 kg) samples.
Sec. 3.2 Packaging and Shipping
3.2.1 Containers. Granular activated carbon shall be in packages acceptable to
the US Department of Transportation (USDOT) and shall contain from 35 to 150 Ib
(16 to 68 kg) in each package, or other quantity as agreed on between the purchaser
and the supplier.
3.2.2 Package shipments. Paper bag packages used in shipments of activated
carbon in less than carload lots shall be protected by an outer package of a resistant
nature, to avoid tearing the bags. Complete protection from weather shall be
provided for the individual packages or by the conveyance.
3.2.3 Tolerances. The net dry weight of the packages shall not deviate from
the recorded weight by more than 5 percent, plus or minus. Objections to the weight
of the material received shall be based on a certified unit weight of not less than
10 percent of the packages shipped, which are selected at random from the entire
shipment.
3.2.4 Bulk shipments. Bulk shipments of activated carbon shall be in clean
cars or trucks with tight closures to avoid loss and contamination of the material in
transit. The interior of the cars or trucks shall be clean and free from dirt,
corrosionscale, and other sources of contamination. Shipments in open-top hopper
bottom cars are acceptable only with adequate provision for covering the material
and keeping it contained and protected during shipment. The type of rail car or
hopper truck shall be agreed on by. the supplier and the purchaser before shipment.
The criteria for choosing a car or truck are the type of handling equipment and the
unloading facilities at the destination.
3.2.5 Weight certification (bulk). Bulk shipments shall be accompanied by
weight certificates of certified weighers, if specified by the purchaser; or the weights
may be checked by certified weighers for the purchaser on delivery.
Sec. 3.3 Marking
Each shipment of the material shall carry with it some means of identification.
3.3.1 Packaged material. Each container of granular activated carbon shall
have marked legibly on it the net weight of the contents, the name of the manufac-
turer, the lot number, a brand name, if any, and shall bear other markings as
required by applicable regulations and laws.*
3.3.2 Bulk material. When shipped in bulk, the information required in
Sec. 3.3.1 for packaged material shall accompany the bill of lading.
*Because of frequent changes in these regulations, their specific provisions should not be
included in the purchaser's specifications.
-------
6 AWWA B604-90
3.3.3 Conformance with standard. (Optional) Containers may bear the state-
ment: "This material meets the requirements of AWWA B604, Standard for
Granular Activated Carbon," provided that the requirements of this standard are
met, and the material is not of a different quality in separate agreement between
the supplier and the purchaser.
SECTION 4; TESTING METHODS
Sec. 4.1 Samples
If the purchaser elects to accept the material on the basis specified in
Sec. 1.4.4(3), samples shall-be taken from each shipment of granular activated car-
bon according to Sec. 3.1. The sample delivered to the laboratory shall be divided to
provide approximately 1 Ib (0.45 kg). After thorough mixing, this sample should be
stored in an airtight container and weighed out of it rapidly to avoid change in
moisture content.
Sec. 4.2 Testing Period
The laboratory examination of a sample shall be complete in time to meet the
requirements of Sec. 1.4.5 for notification of the supplier in the event tests reveal
that the material does not comply with this standard or the purchaser's
specifications.
Sec. 4.3 Moisture
4.3.1 Procedure. In a tared weighing bottle, accurately weigh approximately
2 g of the sample. Dry in a drying oven at 140°C for 2 h or 110°C for 3 h; then cool
in a desiccator and weigh rapidly.
4.3,2 Calculation.
loss of weight , n_ - .
" x 100 = % moisture
weight of sample
Sec. 4.4 Apparent Density
4.4.1 General. The apparent density of a carbon is the weight in grams per
cubic centimetres (g/cc) of the carbon in air. Carbons should have the density deter-
mined on an "as-received" basis with corrections made for moisture content.
4.4.2 Apparatus. Testing apparatus shall be as shown in Figure 1. Reservoir
and feed funnels are glass or metal. The metal vibrator is 26-gauge galvanized sheet
metal. A balance having a sensitivity of 0.1 g is required.
4.4.3 Procedure.
1. Carefully place a representative sample of the carbon into the reservoir
funnel. If the material prematurely flows into the graduated cylinder, return the
material to the reservoir funnel.
2. Add the sample to the cylinder by the vibrator feeder at a uniform rate not
less than 0.75 mL/s nor greater than 1.0 mL/s up to the 100-mL mark. Adjust the
rate by changing the slope of the metal vibrator or raising or lowering the reservoir
-------
GRANULAR ACTIVATED CARBON 7
Ring Stand
Reservoir Funnel
Clamped to Ring Stand
Metal Vibrator
k76.2 mm»j
(Not to Scale)
Assembly of Apparatus
Door Bell "Buzzer"
" (For 10V, 60-Hz Service)
.Feed Funnel
Clamped to Ring Stand
^ Switch
([SPST] Bat-Handle Toggle)
^100-mL ASTM Graduated
Cylinder
Transformer
Primary Volts 115
Volt Ampere 10
Secondary Volts 6/12/18
Frequency 50-60 Hz
4
-------
8 AWWA B604-90
SievesUS standard sieves, 8 in. in diameter and 2 in. high
Bottom receiver pan8 in. diameter, full height
Top sieve cover8 in. diameter
Balancetop loader with sensitivity of 0.1 g
Brushsoft brass-wire brush
4.5.3 Procedure.
1. Assemble the sieves to be used on the bottom receiver pan in order of
increasing sieve opening size from bottom to top. The smallest and largest opening
size sieves should correspond to the limiting sizes for the grade of carbon specified;
such as for 12 x 40 carbon, use sieve numbers 12, 14, 16, 20, 30, and 40. US stan-
dard sieves and opening sizes are tabulated in Table 1.
2. Mix the sample by passing the material through the riffle and recombining
twice.
3. Carefully reduce the mixed sample by repeated passes through the riffle to
obtain a test sample of 100 ± 5 g. No more than 5.0 g of activated carbon may be
added or taken from the test sample without additional riffling.
4. Transfer the weighed sample to the top sieve. Install the sieve cover and
sieve shaker cover and place the assembly on the sieve shaker.
5. Allow the sieve assembly to shake for 3 min ±3 s with the hammer
operating.
6. Remove the sieve assembly from the sieve shaker and quantitatively trans-
fer the activated carbon retained on the top sieve to a tared balance pan and weigh
to the nearest 0.1 g. Repeat this procedure for material retained on each subsequent
sieve and the bottom receiver pan. Lightly brush the material from each sieve to
free particles held in the screen.
7. Add the weights of each sieve fraction, and if the sum deviates more than
2.0 g from the test sample weight, repeat the analyses.
4.5.4 Percentage retained on each sieve.
(sieve fraction weight) (100) = % retained Qn ea
-------
GRANULAR ACTIVATED CARBON 9
4.5.5 Effective size and uniformity coefficient.
a. From the percentage retained on each sieve, calculate the cumulative per-
centage passing each sieve. The cumulative percentage passing a sieve is the sum of
all the percentages retained on subsequent (smaller) sieves plus the percentage
retained on the pan.
b. Using probability x logarithmic paper or semilogarithmic paper, plot the
sieve opening in millimetres on the ordinate or vertical scale versus the cumulative
percentage passing each sieve on the abscissa or horizontal scale.
c. The effective size is the sieve opening in millimetres at which 10 percent of
the material passes on the cumulative percentage passing scale.
d. The uniformity coefficient is determined by dividing the millimetre opening
at which 60 percent passes by the millimetre opening at which 10 percent passes.
Sec. 4.6 Abrasion Resistance
4.6.1 General. Determine abrasion resistance either by the stirring abrasion
test or by the Ro-Tap abrasion test as follows.
4.6.2 Stirring abrasion test. The stirring abrasion test measures percentage
retention of the average particle size in the carbon after abrading the carbon by the
action of a T-shaped stirrer in a specially fabricated abrasion unit. This test is used
to measure abrasion resistance of lignite- and petroleum-coke-base granular
activated carbons.
4.6.2.1 Sieving apparatusstirring abrasion test.
Sieve shaker, electrically driven, equipped with automatic timersimilar to
Ro-Tap
SievesUS standard sieves, 8 in. diameter and 2 in. high; Table 2 indicates
sieves that are required
Bottom receiver pan8 in. diameter, full height
Top sieve cover8 in. diameter
Balancetop loader with sensitivity of 0.1 g
Brushsoft brass-wire brush
4.6.2.2 Stirring abrasion unit. The abrasion unit is detailed in Figure 2. The
apparatus includes a T-shaped stirrer made from l/S-in. metal rod that is driven at
855 ± 15 rpm. The stirrer and cylinder may be made of any suitable material; for
Table 2 Sieving Apparatus Required for Stirring Abrasion Test
' ' "" " Tin.11.. _ .u.l.a,,,.. umm,m,:,m«fmfff
JJS Standard Sieve Opening Average Opening
Sieve Number mm mm (p.)
" ~ " 2^36 " ~
12 1.70 2.03
16 1.18 1.44
20 0.850 1.02
40 0.425 0.64
50 0.300 036
70 0.212 0.26
0.15
-------
10 AWWA B604-90
S
Ball Bearings
(New Departun
77R8)
-££!
,c Ends
CD Turned-
3 True
2
4-in. V-Pulley
Thick-Walled Brass Tubing
1 Va-in. OD x Vi-in. Wall
Silver Soldered
Supporting Frame
Slotted for Belt- .
Tension Adjustment
Va-in.
Brass Plate
Brass Tubing
4.00-in. ID
Clearance,
Bottom and Ends
0.500 + 0.010 in.
Figure 2 Stirring abrasion unit.
example, steel, stainless steel, or brass. The absence of burrs and rough welds is
absolutely necessary. The T-bar stirrer should be replaced when the length of the
cross bar is 0.02 in. less than the designed size or when the hemispherical ends
show signs of serious wear; that is, when the length of the cross bar is more than
0.025 in. from the designed size. Such wear will show on the leading edge of the
T-bar stirrer.
4.6.2.3 Procedurestirring abrasion test.
1. Place a No. 8 sieve on top of a No. 70 sieve on the sieve shaker. Screen
sufficient granular activated carbon sample to obtain 250-300 mL of 8 x 70 mesh
carbon by shaking portions of the activated carbon on the sieve shaker for exactly
3 min ± 2 s with the hammer operating. Discard the material retained on the No. 8
sieve and the material passing the No. 70 sieve.
2. Place the 250-300-mL portion of granular activated carbon on the top
screen of a nest of US standard sieves, numbers 12, 16, 20, 40, 50, and 70; and
shake on the sieve shaker for 15 min ± 10 s with the hammer operating.
3. Remove the sieve assembly from the sieve shaker and quantitatively trans-
fer the carbon retained on the top sieve to a tared balance pan and weigh to the
nearest 0.1 g. Repeat this procedure for material retained on each subsequent sieve
and the bottom receiver pan. The material should be lightly brushed from each sieve
to free particles held in the screen. Record the weight of each sieve fraction and the
total weight of carbon recovered.
4. Recombine and blend by tumbling the sieve fractions very gently in a
quart fruit jar or similar container and place the carbon in the abrasion unit.
Operate the abrasion unit for 1 h ± 1 min.
-------
GRANULAR ACTIVATED CARBON 11
5. Remove the carbon from the abrasion unit and repeat the screening on a
nest of US standard sieves, numbers 12,16, 20, 40, 50, and 70, as in step 2. Use the
same sieve shaker as was used for the initial sieve analysis. Record the weight of
each sieve fraction and the total weight of carbon recovered.
4.6.2.4 Calculationsstirring abrasion test. Calculate the average particle
size before and after stirring by using the following equation:
summation of (Wt x Di)
-i^avg =
Where:
_
avg ~ summation of (Wj)
= the average particle size, in millimetres
Wi = the weight of a sieve fraction, in grams
Dt = the opening in millimetres that corresponds to the average of the
openings in the two sieves that enclose that mesh fraction (See
Table 2.)
Calculate the percentage retention of average particle size; adjust to 1 mm
original particle size by using the following equation:
% retention/millimetre = (100) [l - (°rignal P«vg - final J«yg) 1
[ (original Davg)2 J
Report the value obtained as the percentage retention of particle size from the
stirring abrasion test.
4.6.3 Ro-Tap abrasion test. The Ro-Tap abrasion test measures the percentage
retention of original average particle size by the resistance of the particles to the
action of steel balls in the Ro-Tap machine. This test is used to measure abrasion
resistance of bituminous coal and petroleum-coke-base granular activated carbons.
4.6.3.1 Sieving apparatusRo-Tap abrasion test.
Sample splittersimilar to Jones riffler
Ro-Tapsieve shaker, electrically driven, equipped with automatic timer
SievesUS standard sieves, 8 in. diameter and 2 in. high
Bottom receiver pan8 in. diameter, full height
Top sieve cover8 in. diameter
Balancetop loader with sensitivity of 0.1 g
Brushsoft brass-wire brush
4.6.3.2 Testing pan assembly. The abrasion pan assembly is detailed in
Figure 3. The assembly consists of a Ro-Tap lid with cork insert, a half-height blank
pan, a specially fabricated abrasion testing pan, and a bottom receiver pan. The
abrasion testing pan is detailed in Figure 4. Ten l/S-in. (12.7-mm) diameter and ten
%-in. (19-mm) diameter smooth steel balls will also be required. The steel balls will
be placed in the testing pan together with the carbon sample to be tested for the
abrasion test.
4.6.3.3 ProcedureRo-Tap abrasion test.
1. Assemble the sieves to be used on the bottom receiver pan in order of
increasing sieve opening from bottom to top. Suggested sieve sizes to be used with
various particle-size ranges are given in Table 3.
-------
12 AWWA B604-90
-Cork
Lid
r :-{- -\ 3
Cf E=.=^=J f
\ Half-Height Blank Pan |
Abrasion Testing Pan
Bottom Receiver Pan.
Figure 3 Testing pan assembly for Ro-Tap abrasion test (not to scale).
1111/16 in.
2,in-_L_
14-Gauge Sheet Brass
Brown and Sharp
Standard
Figure 4 Abrasion testing pan for Ro-Tap abrasion test (not to scale).
-------
GRANULAR ACTIVATED CARBON" 13
Table 3 Recommended Particle Sieve Sizes
Particle-Size Range
8 x 16
8 x 20
8 x 30
. 10 x 30
12 x 40
14 x 40
20 x 40
20 x 50
US Standard Sieve Sizes
8, 12, 16, pan
8, 12, 16, 20, pan
8, 12, 16, 20, 30, pan
10, 12, 16, 20, 30, pan
12, 16, 20, 30, 40, pan
14, 16, 20, 30, 40, pan
20, 30, 40, pan
20, 30, 40, 50, pan
2. Mix the sample by passing the material through the riffle and recombining
twice.
3. Carefully reduce the mixed sample by repeated passes through the riffle so
as to obtain a test sample of 100 ± 5 g. Do not add to or take from the sample more
than 5.0 g of carbon without additional riffling.
4. Transfer the weighed sample to the top sieve.
5. Install the sieve cover and Ro-Tap cover and place the assembly on the
Ro-Tap sieve shaker.
6. Allow the sieve assembly to shake for 10 min ±10 s with the hammer
operating.
7. Prepare the abrasion testing pan and count the steel balls to ensure that
ten l/6-in. (12.7-mm) and ten 3/4-in. (19-mm) diameter smooth steel balls are con-
tained in the pan.
8. Remove the sieve assembly from the Ro-Tap and quantitatively transfer
the carbon retained on the top sieve to a tared balance pan and weigh to the nearest
0.1 g; then transfer to the abrasion testing pan. Repeat this procedure for material
retained on each subsequent sieve and the bottom receiver pan. The material should
be lightly brushed from each sieve to free particles held in the screen. Record the
weight of each sieve fraction and the total weight of carbon recovered.
9. After the sieve fractions have been weighed and recombined in the
abrasion testing pan, place the testing pan assembly on the Ro-Tap sieve shaker.
The testing pan assembly must be level and fit snugly on the Ro-Tap.
10. Allow the testing pan assembly to shake for 20 min ± 2 s with the hammer
operating. The time is critical, and if the automatic timer is not capable of the
specified accuracy, the sieve shaker should be manually controlled and timed with a
stopwatch.
11. Remove the abrasion pan from the Ro-Tap and quantitatively transfer the
contents to the original set of sieves. A large-opening sieve may be temporarily
nested into the top sieve to remove the steel balls from the carbon, or the balls may
be removed by hand.
12. Repeat steps 5, 6, and 8 using the same Ro-Tap as was used for the initial
sieve analysis:. However, after this second sieve analysis, discard the individual
screen fractions after weighing. Repeat the analysis if the sum of either sieve
analysis deviates by more than 2.0 g from the test sample weight.
-------
14 AWWA B604-90
4.6.3.4 Calculations Ro-Tap abrasion test.
(a) Calculate the original and final average particle size by using the following
equation:
_
Davg ~
summation of (Wix Z>0
summation of (Wi)
Where:
Davg
W»
= the average particle size, in millimetres
= the weight of a sieve fraction, in grams
= the opening in millimetres that corresponds to the average of the
openings in the two sieves that enclose that mesh fraction
Material caught on the pan is not considered in calculating the average particle
diameter. Values for Di are given in Table 4.
(b) Calculate average particle size; example calculation using a 12 x 30 mesh
material.
US Standard
Sieve No.
+12
12 x 16
16 x 20
20 x 30
+30
Retained
percent
1.5
25.0
50.0
22.5
1.0
100.0
Dav
Average
Opening
Dt
mm
2.03f
1.44
1.02
0.725
0.00
106.3 l Q63
g ~ 100.0
Average*
3.0
36.0
51.0
16.3
0.0
106.3
(c) Calculate the percentage retention of average particle size by using the
following equation:
final Davg * nA
retention, percentage = original p *vg * 100
Report the value obtained as the percentage retention of average particle size from
the Ro-Tap Abrasion Test.
*The 2.03 factor was used for material remaining on the No. 12 sieve because it was
assumed this material would pass through a No. 8 sieve (generally the next larger sieve in
the square root of two series).
tWeighted.
-------
GRANULAR ACTIVATED CARBON 15
Table 4 Di Values for Ro-Tap Abrasion Test
US Standard
Sieve Numbers
6 x 8
8 x 10
8 x 12
10 x 12
12 x 14
12 x 16
14 x 16
16 x 20
20 x 30
30 x 40
40 x 50
Average Opening
(Di) mm
2.86
2.18
2.03
1.85
1.55
1.44
1.29
1.02
0.725
0.513
0.360
Sec. 4.7 Test Method for Iodine Number
The procedure for determining the iodine number of activated carbon is ASTM*
D4607, Standard Test Method for Determination of Iodine Number of Activated
Carbon.
Sec. 4.8 Surface Area Determination
The procedure for determining the surface area of activated carbon is the
Nitrogen BET Surface Area Test. The reference procedure to perform this test is
ASTM D3037, Standard Test Methods for Carbon BlackSurface Area by Nitrogen
Adsorption.
Sec. 4.9 Pore Volume Determination
The procedure for the determination of the total pore volume of an activated
carbon employs a combination of mercury and helium displacement techniques. The
reference procedure to perform this test is ASTM C699, Standard Methods for
Chemical, Mass Spectrometric, and Spectrochemical Analysis of, and Physical Tests
on, Beryllium Oxide Powder.
Sec. 4.10 Water Extractables Test
The method used for determining the water extractable content of activated
carbon is found in the Food Chemicals Codex'f procedure, under the category of
Carbon, Activated.
*American Society for Testing and Materials, 1916 Race St., Philadelphia, PA 19103.
Chemicals Codex, National Academy Press, 2101 Constitution Ave NW
Washington, DC 20418. ., . .,
-------
APPENDIX A
Bibliography
This appendix is for information only and is not a part ofANSI/AWWA B604.
CARPENTER, F.G. Development of a New Test for the Abrasion Hardness of Bone Char. Proc.
5th Tech. Sess. Bone Char. 1957. National Bureau of Standards Supplement to Miscellaneous
Publication 240 (Apr. 3,1967).
CHEREMISINOFF, P.N. & ELLERBUSCH, F. Carbon Adsorption Handbook, Ann Arbor Sci. Publ.,
Inc., Ann Arbor, Mich. (1978).
CLARK, R.M. Modeling TOG Removal by GAG. The General Logistic Function. Journal AWWA
79:33-37 (1987).
COUGHLIN, R.W. & EZRA, F.S. Role of Surface Acidity in the Adsorption of Organic Pollutants
on the Surface of Carbon. Environmental Science & Technology. 2:4:291-297 (1968).
CRITTENDEN, J.C.; HAND, D.W.; ARORA, H.; & LYKINS, B.W. JR. Design Considerations for GAG
Treatment of Organic Chemicals. Journal AWWA 79:1:74-82 (1987).
CRITTENDEN, J.C.; LUFT, P.J.; HAND, D.W.; & FRIEDMAN, G. Prediction of Fixed-Bed Adsorber
Removal of Organics in Unknown Mixtures. J. Environ. Eng. Div., Am. Soc. Civ. Eng.
113:3:486-498 (1987).
Design and Use of GAGPractical Aspects, AWWA Research Foundation. May 9-10, 1989.
Cincinnati, Ohio. American Water Works Assoc. Conference Proceedings, Denver, (1989).
DoBBS, R.A. & COHEN, J.M. Carbon Adsorption Isotherms for Toxic Organics. EPA Rpt.
EPA600/8-80-023. USEPA, Cincinnati, Ohio (1980).
. Treatments of Organic Compounds in Drinking Water. EPA Rpt. EPA600/8-83-019.
USEPA, Cincinnati, Ohio (1983).
Food Chemicals Codex. Com. on Specifications, Food Chemicals Codex; Com. on Food Protec-
tion, Nat. Res. Council. Nat. Acad. Press, Washington, D.C. (1988).
HASHIMOTO, K; MIURA, K; YOSHIKAWA, F.; & IMAI, I. Change in Pore Structure of Car-
bonaceous Materials During Activation and Adsorption Performance of Activated Carbon, Ind.
Eng. Chem. Process Dec. Dev. 18:1:72-80 (1979).
HASSLER, J.W. Activated Carbon, Leonard Hill, London (1967).
. Activated Carbon, Chem. Publ. Co., New York (1963).
. The History of Taste and Odor Control. Journal AWWA 33:2124-2152 (1941).
HERZING, D.R.; SNOEYINK, V.L.; & WOOD, N.F. Activated Carbon Adsorption of the Odorous
Compounds 2-Methylisoborneol and Geosmin. Journal AWWA 69:4:223-228 (1977).
Identification and Treatment of Taste and Odor in Drinking Water, edited by J. Mallevialle.
AWWA Research Foundation, American Water Works Assoc., Denver (1987).
JUHOLA, A.J. Manufacture, Pore Structure and Application of Activated Carbons Part 1.
Kemia-Kemi 11:543-551 (1977).
. Iodine Adsorption and Structure of Activated Carbons, Carbon 13:437-^42 (1975).
JUNTGEN, H. Manufacture and Properties of Activated Carbon. In: Translation of Reports on
Special Problems of Water Technology, EPA Rpt. EPA600/9-76-030, Ofce. Res. Devel.,
USEPA, Cincinnati, Ohio, (1976a).
16
-------
GRANULAR ACTIVATED CARBON 17
KIM, B.R. & SNOEYINK, V.L. The Monochloramine-Activated Carbon Reaction: A Mathematical
Model Solved Using the Orthogonal Collocation Method on Finite Elements. In: Activated
Carbon Adsorption ofOrganics from the Aqueous Phase, Vol. 1, (I.H. Suffet and MJ. McGuire,
editors). Ann Arbor Sci. Publ. Inc., Ann Arbor, Mich. (1980a).
. The Monochloramine-GAC Reaction in Adsorption Systems. Journal AWWA 72:488
(1980b).
KIM, B.R.; SNOEYINK, V.L.; & SCHMITZ, RA. Removal of Dichloramine arid Ammonia by
Granular Carbon. Journal Water Pollution Control Federation 50:122-133 (1978a).
KIWA-AWWARF. Activated Carbon in Drinking Water Technology. AWWA Research Foun-
dation, Denver (1983).
MANTELL, C.L. Adsorption, McGraw-Hill Book Company, New York, London (1951).
McGuiRE, M.J. The Optimization of Water Treatment Unit Processes for Removal of Trace
Organic Compounds with an Emphasis on the Adsorption Mechanism. Ph.D. thesis. Drexel
University, Philadelphia (1977).
McGuiRE, M.J. & SUFFET, I.H. Activated Carbon Adsorption of Organics From the Aqueous
Phase. 2 vols. Ann Arbor Sci. Publ., Inc., Ann Arbor, Mich. (1980).
. Treatment of Water by Granular Activated Carbon. Advances in Chemistry Series,
no. 202, American Chemical Society, Washington, D.C. (1983).
Organic Pollutants in Water: Sampling, Analysis and Toxicity Testing, edited by M.
Malaiyandi, no. 214. American Chemical Society, Advances in Chemistry Series, Washington,
D.C. (1987).
ROBECK, G.G.; DOSTAL, K.A.; COHEN, J.M.; & KREISSL, J.F. Effectiveness of Water Treatment
Processes in Pesticide Removal. Journal AWWA 57:2:181-200 (1965).
SNOEYINK, V.L. & SUIDAN, M.T.. Dechlorination by Activated Carbon and Other Reducing
Agents. In: Disinfection of Water and Wastewater, J.D. Johnson (ed.). Ann Arbor Sci. Publ.
Inc., Ann Arbor, Mich. (1975).
SUIDAN, M.T.; CHACEY, K.A.; & CROSS, W.H. Pulsating Bed Activated Carbon Dechlorination,
J. Environ. Eng. Div., Amer. Soc. Civ. Eng. 104:1223 (1978).
SUIDAN, M.T., SNOEYINK, V.L. & SCHMITZ, R.A. Reduction of Aqueous Free Chlorine With
Granular Activated CarbonpH and Temperature Effects.. Environ. Sci. Technol. 11:785
(1977a).
. Reduction of Aqueous HOC1 With Activated Carbon, J. Environ. Eng. Div., Amer. Soc.
Civ. Eng. 103:677 (1977b).
SUMMERS, R.S.; FUCHS, P.; & SONTHEIMER, H. The Fate and Removal of Radioactive Iodine in
the Aquatic Environment. In: Influences of Aquatic Humic Substances on Fate and Treatment
of Pollutants, edited by P. MacCarthy and I.H. Suffet, no. 219. Advances in Chemistry Series,
American Chemical Society, Washington, D.C. (1988b).
SUMMERS, R.S. Activated Carbon Adsorption of Humic Substances: Effect of Molecular Size
and Heterodispersity, Ph.D. dissertation, Dept. of Civil Engineering, Stanford University
Stanford, Calif. (1986).
Water Treatment Plant Design. ASCE, AWWA, & CSSE. New York (1989).
WEBER, W.J. JR. Physiochemical Processes for Water Quality Control. Wiley, Inter-Science
New York (1972).
WHITE, G.C. Handbook ofChlorination, Van Nostrand Reinhold Co., Cincinnati, Ohio (1972).
-------
APPENDIX B
Adsorptive Capacity Tests
This appendix is for information only and is not a part ofANSI/AWWA B604.
These tests were added because they are referred to in the Adsorptive Capacity
section of the foreword, and they could aid some purchasers in the analysis of
granular carbon used for water treatment.
SECTION B.I TANNIN ADSORPTION TEST
Sec. B.I.I Stock Tannic Acid Solution500 mg/L
Dissolve exactly 1.00 g of NF grade tannic acid in distilled water and dilute to
2 L in a volumetric flask. Use tannic acid similar to Merck & Company, NF catalog
number 04541 or equivalent.
Sec. B.1.2 Test Procedure
A piece of 0.75-in. (19-mm) inside diameter (ID) tubing* (glass or acrylic) about
7 in. (175 mm) long is fitted at the bottom with a one-hole rubber stopper, a short
piece of rubber tubing, and an adjustable hose clamp. A piece of 80-mesh screen is
used in the bottom of the tube to support the carbon. The tube is marked to indicate
a carbon volume of 32 mL. A weighed amount of granular carbon is added to the
tube to approximately the 32-mL mark, then gently washed upflow to remove any
fines. If needed, additional carbon is added to the 32-mL mark, and the column is
backwashed again. After backwashing, the water level is allowed to fill to the top of
the carbon bed. Care should be taken so the carbon in the column is completely
submerged at all times; otherwise, channeling will occur through the center of the
bed.
One litre of 500-mg/L tannic acid solution is passed downflow through the
column at the rate of 15 mL/min (assuming 0.75-in. [19-mm] tubing is used), and
the entire effluent is collected in a receiving flask. If a suitable pump is not avail-
able, the tannic acid solution can be fed from a separatory funnel held above the
column by adjusting the stopcock to give a flow of 15 mL/min.
The effluent is mixed and the concentration of tannin remaining is determined
by either UV absorbance at 275 mum by evaporation of a portion of the mixed
filtrate or by the standard AWWA color method (see Sec. Bl.3.3) for the tannin
analysis.
Sec. B.I .3 Determination of Tannin in Effluent
B.l.3.1 UV absorbance. A sample of mixed effluent and of the original
500-mg/L tannin feed is read on a UV spectrophotometer at 275 mum (maximum
*A larger diameter tube can be used with appropriate adjustments of carbon volume, flow
rate, and total volume of tannic acid solution used.
18
-------
GRANULAR ACTIVATED CARBON 19
Table B.I Standard Curve of Tannin Dilution
500 mg/L Distilled Water
Tanninmg/L TanninmL mL
5 1 99
25 5 95
50 10 90
100 20 80
200 40 60
300 60 40
400 80 20
absorbance peak). Samples are diluted with distilled water until a direct instrument
scale reading can be obtained and corrected for dilution. A standard curve is
prepared by diluting the 500-mg/L tannin feed as shown above in Table B.I.
The optical density of each above dilution at 275 mum is plotted against milli-
grams per litre tannin. From the standard curve, the milligrams per litre tannin in
each effluent sample is determined.
Calculation:
percentage tannin adsorbed
_ lonfi _ effluent tannin (in milligrams per litre) 1
[ influent tannin (in milligrams per litre) J
weight tannin adsorbed (in grams)
100 g carbon
_ percentage tannin adsorbed x 5g x litres used
weight carbon in column (in grams)
B.I.3.2 Evaporation of mixed effluent. Exactly 200 g of mixed effluent and a
200-g feed sample are evaporated in a 100°C convection oven to dryness and each
residue weighed on an analytical balance to the nearest milligram.
Calculation:
percentage tannin adsorbed = 100 [ 1 - fluent residue (in grams) 1
[_ influent residue (in grams) J
weight tannin adsorbed (in grams)
100 g carbon
_ percentage tannin adsorbed x 5 g x litres used
weight carbon in column (in grams) X
B.l.3.3 AWWA-APHA-WPCF method. The residual mg/L tannin is determined
colorimetrically for the mixed effluent and feed using the AWWA-APHA-WPCF
method for tannin and lignin. Reagents and apparatus required are given in method
-------
20 AWWA B604-90
number 5550 Standard Methods for the Examination of Water and Wastewater.*
Calculations are made in the same manner as previously described.
SECTION B.2 PHENOL ADSORPTION TEST
Sec. B.2.1 Reagents
a. Stock phenol solution5000 mg/L. Dissolve 5.0 g reagent-grade phenol in
distilled water and dilute to 1 L. Phenol should be weighed in a glass weighing dish.
Rinse the dish several times with distilled water to ensure transfer of all phenol
from the dish into the solution. Standardize. If the concentration of phenol is more
than ± 200 mg/L, either dilute with distilled water or add more phenol to get the
desired concentration. After two weeks, this solution should be discarded and fresh
solution prepared. (Reagent-grade phenol should be stored in a refrigerator.)
b. Sodium thiosulfate solution0.12V. Dissolve 25.0 g reagent-grade sodium
thiosulfate and 1.0 g reagent-grade sodium carbonate (as a preservative) in boiled
distilled water and make up to 1 L. Store in a brown bottle. Standardize.
c. Potassium bromate-4>romide solution0.12V. Dissolve 2.784 g of reagent-
grade potassium bromate and 10.0 g reagent-grade potassium bromide
(bromate-free) in distilled water and make up to 1 L. Store in a brown bottle.
d. Potassium biniodate solution0.17V. Dissolve 3.2499 g potassium biniodate,
primary standard, in distilled water and make up to I L.
e. Potassium iodide solution12.5 percent. Dissolve 25 g reagent-grade potas-
sium iodide in 175 mL distilled water. Store in a brown bottle. (Discard when it
develops a yellow color.)
f. Starch solution. Dissolve 5.0 g soluble potato powder starch and 1.25 g
reagent-grade salicylic acid in 50 mL distilled water. Add the dissolved starch and
salicylic acid slowly, while stirring, to 950 mL boiling distilled water. Rinse the
beaker with some of the hot starch solution to ensure removal of all the starch.
Sec. B.2.2 Standardization of Reagents
a. Sodium thiosulfate solution. Add 100 mL distilled water; 4 mL concen-
trated, reagent-grade hydrochloric acid; and 8 mL, 12.5 percent potassium iodide
solution to a 500-mL iodine flask and mix. Rinse down sides of the flask with dis-
tilled water. Using a transfer pipette, add 25 mL 0.17V potassium biniodate solution
to the flask. Mix and allow to stand for 3 min. Titrate with the 0.17V sodium thiosul-
fate solution using starch solution as the indicator.
Calculation:
phenol concentration (in grams per litre)
_ [(millilitres bromate-bromide x NF) - (millilitres thiosulfate x NF)]
~~ millilitres phenol solution titrated
x 15.685
^Standard Methods for the Examination of Water and Wastewater. APHA, AWWA, WPCF.
AWWA, Denver (17th ed., 1989).
-------
GRANULAR ACTIVATED CARBON 21
b. Stock phenol solution. Pipette 25 mL stock phenol solution into a 500-mL
iodine flask and add 15 mL concentrated, reagent-grade hydrochloric acid. Titrate
with the potassium bromate-bromide solution to a slight yellow color. (For
5000 mg/L phenol concentration, it will require about 8090 mL of solution to
produce the yellow color.) Shake the flask and allow to stand for 3 min. Add 8 mL of
12.5 percent potassium iodide solution, shake, and allow to stand for 3 min. Titrate
the liberated iodine with the standardized 0.1N sodium thiosulfate solution, using
the starch solution as indicator.
Calculation:
normality sodium thiosulfate solution
millilitres potassium biniodate x NFbiniodate
millilitres sodium thiosulfate solution used
Sec. B.2.3 Test Procedure
a. Phenol adsorption. A piece of 0.75-in. (19-mm) ID tubing (glass or acrylic)
about 7 in. (175 mm) long is fitted at the bottom with a one-hole rubber stopper, a
short piece of rubber tubing, and an adjustable hose clamp. A piece of 80-mesh
screen is used in the bottom of the tube to support the carbon. The tube is marked
at such a height as to indicate a carbon volume of 32 mL. The granular carbon is
added to the tube to about the 32-mL mark, then gently washed upflow to remove
any fines. If needed, additional carbon is added to the 32-mL mark, and the column
is backwashed again. After backwashing, the water level is allowed to fall to the top
of the carbon bed.
One litre of phenol solution at 5000 mg/L is passed downflow through the
column at the rate of 15 mL/min, and the entire effluent is collected in a receiving
flask.
The effluent is mixed and the concentration of phenol determined.
b. Determination of phenol in effluent. A 25-mL aliquot of the mixed effluent
is placed in a 500-mL iodine flask. The same procedure and calculation used to
determine the concentration of the stock solution in B.2.2(b) is used to determine
the phenol in effluent.
Calculation:
percentage phenol adsorbed
influent phenol (in milligrams per litre) J
., ftn f., effluent phenol (in milligrams per litre) ]
= 1UU 1 T~Zr^ rr. ^rr2 c . I
weight phenol adsorbed (in grams)
100 g carbon
_ percentage phenol adsorbed x 5 g x litres used
weight carbon in column (in grams)
-------
-------
Part 3
Membrane Precursor Removal Studies
Steven C. Allgeier
Research Fellow
Oak Ridge Institute for Science and Education
Sponsored by
Technical Support Division
Office of Ground Water and Drinking Water
United States Environmental Protection Agency
Luke A. Mulford
Department of Civil and Environmental Engineering
University of Central Florida
James S. Taylor
Department of Civil and Environmental Engineering
University of Central Florida
-------
Notice
This section was prepared for the EPA's Office of Ground Water and Drinking Water
through contract with the American Water Works Association Research Foundation, Project
Number 170, and KIWA, N.V.
3-ii
-------
Contents
Page
List of Figures 3-v
List of Tables 3-vi
List of Symbols '. 3-ix
1.0 Introduction 3-1
1.1 Organization Of Part 3 3-1
1.2 , ICR Requirements For Membrane Studies 3-1
2.0 Background Information 3-3
2.1 Fundamentals Of Membrane Processes 3-3
2.1.1 Terminology And Concepts 3-3
2.1.2 Membrane System Design 3-4
2.2 Membrane Technology To Meet Drinking Water Treatment Objectives . . 3-6
2.3 Literature Review 3-9
2.4 Mechanisms And Control Of Membrane Fouling 3-14
2.4.1 Scaling 3-14
2.4.2 Colloidal, Biological And Chemical Fouling 3-15
2.5 Fouling Indices 3-15
2.5.1 Silt Density Index 3-16
2.5.2 Modified Fouling Index . 3-16
2.5.3 Mini Plugging Factor Index 3-17
2.5.4 Index Guidelines 3-17
2.6 Membrane System Design And Operational Factors For Precursor Removal 3-17
2.6.1 Membrane Selection 3-18
2.6.2 Pretreatment 3-20
2.6.3 Impact Of Operating Parameters On Performance 3-20
3.0 Membrane Test Systems 3-23
3.1 The Rapid Bench-Scale Membrane Test 3-23
3.1.1 General Description And ICR Requirements 3-23
3.1.2 Advantages And Limitations Of The RBSMT . . . 3-24
3.2 Single Element Bench-Scale Tests 3-25
3.2.1 General Description And ICR Requirements 3-25
3.2.2 Advantages And Limitations Of SEBSTs 3-26
3.3 Pilot-Scale Testing 3-26
3.3.1 General Description And ICR Requirements 3-26
3.3.2 Advantages And Limitations Of Pilot Systems 3-27
4.0 Rapid Bench-Scale Membrane Test 3-29
4.1 RBSMT System 3-29
4.1.1 RBSMT Definitions 3-29
4.1.2 System Design 3-29
4.1.3 System Start-up 3-32
4.1.4 System Shut-down 3-34
3-iii
-------
4.1.5 System Cleaning 3.34
4.2 RBSMT Procedure 3-35
4.2.1 Obtaining Membrane Samples And Manufacturer Data 3-35
4.2.2 Selecting Operating Parameters 3-36
4.2.3 Membrane Pretreatment And Treatment Study Influent 3-40
4.2.4 Steps Of The RBSMT Procedure 3-42
4.2.5 Monitoring And Sampling Requirements 3-44
4.2.6 Seasonal (Quarterly) Variation 3-45
4.2.7 Data Sheets 3-46
4.2.8 Interpretation Of The Results 3-48
5.0 Single Element Bench-Scale Test 3-55
5.1 SEBST System 3-55
5.1.1 SEBST Requirements And Options 3-55
5.1.2 System Design 3-56
5.1.3 Continuous-flow Pretreatment 3-61
5.1.4 System Start-up 3-62
5.1.5 System Shut-down 3-64
5.1.6 Membrane Cleaning And Preservation 3-65
5.2 SEBST Procedure 3-66
5.2.1 Selecting Operating Parameters 3-66
5.2.2 Monitoring Requirements . . . '. 3-67
5.2.3 Sampling Requirements 3-67
5.2.4 Data Sheets 3-68
5.2.5 Membrane Productivity 3-70
6.0 Pilot-Scale Evaluation Of Membrane Processes 3-73
6.1 Design Considerations 3-73
6.1.1 Minimum Requirements For A Pilot-Scale Study 3-73
6.1.2 Pilot Plant Description 3-74
6.1.3 Sampling And Monitoring Locations 3-76
6.1.4 Design Calculations 3-78
6.1.5 Design Example 3-82
6.1.6 Pretreatment 3-87
6.2 System Operation 3-88
6.2.1 System Start-up 3-88
6.2.2 System Shut-down 3-90
6.2.3 Membrane Cleaning And Preservation 3-90
6.2.4 Membrane Productivity 3-92
6.3 Sampling Requirements 3-96
6.3.1 Daily System Monitoring 3-96
6.3.2 Biweekly System Monitoring 3-97
6.3.3 Additional Monitoring And Reporting 3-97
6.3.4 Pressure Vessel Or Element Monitoring (Optional) 3-98
6.3.5 Data Sheets 3-98
7.0 Cost Survey 3-101
8.0 References 3-103
Appendix 3-A: Water Quality Prediction For Multi-Stage Systems 3-107
3-iv
-------
Figures
Figure 2-1 Basic Diagram Of A Membrane Separation Process
Figure 2-2 Conventional NF/RO Systems Showing Pretreatment, Membrane Filtration And
Post-treatment
Figure 2-3 Two Stage (One Array) Membrane System
Figure 2-4 Definition Curve For The Modified Fouling Index
Figure 2-5 Typical Curve For The Mini Plugging Factor Index
Figure 4-1 Definition Sketch Of A Tangential-Flow Membrane-Cell With Recycle
Figure 4-2 Bench-Scale Tangential-Flow Membrane System With Recycle
Figure 4-3 Schematic Of The Tangential-Flow Flat-Sheet Cell Used In The RBSMT
Figure 4-4 Set-Up For Membrane Cleaning Procedure
Figure 4-5 Flow Chart For The Quarterly RBSMT Membrane Studies
Figure 4-6 MTCW as a function of time for a Fluid Systems TFCS membrane treating sand
filtered Ohio River water from the Cincinnati Water Works
Figure 4-7 Long-term MTCW study for a FilmTec NF-90 membrane treating sand filtered
Ohio River water from the Cincinnati Water Works
Figure 4-8 Membrane characterization curves for the FilmTec NF-90 membrane
treating sand filtered Ohio River water from the Cincinnati Water Works
Figure 4-9 Rejection of TOC, UV^ and conductivity with tune for the FihnTec NF-90
treating sand filtered Ohio River water from the Cincinnati Water Works
Figure 4-10 Bulk and feed rejections of TOC, UV254 and conductivity as a function of
recovery for an NF-90 treating sand filtered Ohio River water
Figure 4-11 Rejections summary for an NF-90 membrane treating sand filtered Ohio River
water
Figure 5-1 Schematic Of A Single Element Bench-Scale Unit
Figure 5-2 Temperature Normalized MTCW Plotted As A Function Of Tune
Figure 6-1 Flow Diagram Of A Two-Stage Pilot Plant Showing Sampling And
Monitoring Locations By Number
Figure 6-2 Two-Stage Pilot Plant Membrane Productivity
3-v
-------
Tables
Table 2-1 Terminology Used In RO/NF Membrane Processes
Table 2-2 Definitions Commonly Used In Membrane Processes
Table 2-3 Characteristics Of Membrane Processes
Table 2-4 Summary Of Membrane Applications For Regulated Contaminants
Table 2-5 Fouling Index Approximations For RO/NF
Table 2-6 Membrane Selection Characteristics
Table 2-7 Advanced Pretreatment For Various Fouling Mechanisms
Table 2-8 Effect Of Independent Variables On Solute Permeate Concentration By Mass
Transfer Mechanism (T Increase, I Decrease, O No Effect)
Table 3-1 Summary Of The RBSMT-ICR Requirements
Table 4-1
Table 4-2
Table 4-3
Table 4-4
Table 4-5
Table 4-6
Table 4-7
Table 4-8
Table 4-9
Table 4-10
Table 4-11
Table 4-12
Table 4-13
Table 4-14
Table 4-15
Table 4-16
Table 4-17a
Table 4-17b
Table 5-1
Table 5-2
Table 5-3
Table 5-4
Table 5-5
Table 5-6
Table 5-7
Definitions Used In The RBSMT Procedure
Experimental Matrix For The RBSMT-ICR Membrane Studies
Recommended Minimum Monitoring Frequencies For The RBSMT
RBSMT Water Quality Monitoring Requirements
Membrane Characteristics As Reported By The Manufacturer
RBSMT Design Parameters
Membrane Pretreatment Data
Membrane Feed Water Quality (After Membrane Pretreatment)
Membrane Setting Data (Operation With Laboratory Clean Water)
Membrane Performance Data Monitored During Operation With The
Pretreated Test Water
Membrane Permeate Water Quality For Run ID#1
Duplicate Analysis Of The Membrane Permeate Water Quality For Run ID#1
Membrane Permeate Water Quality For Run ID#2
Membrane Permeate Water Quality For Run ID#3
Membrane Permeate Water Quality For Run ID#4
Composite Concentrate Water Quality Parameters And Mass Balance
Closure Errors
Blending Calculations To Determine The Fraction Of Flow That Requires NF
Treatment To Meet The Stage I DBF Regulations
Blending Calculations To Determine The Fraction Of Flow That Requires NF
Treatment To Meet The Stage IIDBP Regulations
Example Design Parameters For Single Element Units
Design Characteristics For The Membrane Used In The Design Example
SEBST Routine Monitoring Requirements
SEBST Water Quality Monitoring Requirements
Short-term Membrane Productivity Study
Membrane Characteristics As Reported By The Manufacturer
Membrane Pretreatment Data
3-vi
-------
Table 5-8
Table 5-9
Table 5-10
Table 5-11
Table 5-12
Table 5-13
Table 5-14
Table 5-15a
Table 5-15b
Table 6-1
Table 6-2
Table 6-3
Table 6-4
Table 6-5
Table 6-6
Table 6-7
Table 6-8
Table 6-9
Table 6-10
Table 6-11
Table 6-12
Table 6-13
Table 6-14
Table 6-15
Table 6-16
Table 6-17
Table 6-18
Table 6-19a
Table 6-19b
Table 7-1
Table 7-2
Table 7-3
Membrane Operational And Performance Data Monitoring With Time
Membrane Feed And Permeate Water Quality For Week One
Membrane Feed And Permeate Water Quality For Week Two
Membrane Feed And Permeate Water Quality For Week Three
Membrane Feed And Permeate Water Quality For Week Four
Duplicate Analysis Of Membrane Feed And Permeate Water Quality For Week
Concentrate Water Quality Parameters And Mass Balance Closure Errors
Blending Calculations To Determine The Fraction Of Flow That Requires NF
Treatment To Meet The Stage I DBP Regulations
Blending Calculations To Determine The Fraction Of Flow That Requires NF
Treatment To Meet The Stage H DBP Regulations
Two-Stage Membrane Pilot Plant Numerical Identification Code
Minimum Monitoring Matrix For A Two-Stage Pilot Plant
Daily Operations Log Sheet For A Two-Stage Membrane Pilot Plant
RO And NF Membrane Pilot Plant Design Criteria
Osmotic Pressure Estimates For Stages 1 And 2 Of A 2-1 Membrane System
With Three Elements Per Pressure Vessel, Operated At 75% Recovery
Example Design Criteria For A Nanofiltration Pilot Plant
Pilot Plant Design Parameters For The Example Problem
Example Of Time-Dependent Membrane Productivity For A Two-Stage Pilot
Plant
Membrane Characteristics As Reported By The Manufacturer
Membrane Pretreatment Data
Membrane Operational And Performance Data Monitoring With Time
Membrane Pilot System Biweekly Water Quality Parameters For Week Two
Membrane Pilot System Biweekly Water Quality Parameters For Week Four
Membrane Pilot System Biweekly Water Quality Parameters For Week Six
Membrane Pilot System Biweekly Water Quality Parameters For Week
Duplicate Analysis Of Membrane Pilot System Biweekly Water Quality
Parameters For Week Ten
Duplicate Analysis Of Membrane Pilot System Biweekly Water Quality
Parameters For Week Twenty
Duplicate Analysis Of Membrane Pilot System Biweekly Water Quality
Parameters For Week
Blending Calculations To Determine The Fraction Of Flow That Requires NF
Treatment To Meet The Stage I DBP Regulations
Blending Calculations To Determine The Fraction Of Flow That Requires NF
Treatment To Meet The Stage II DBP Regulations
Design Parameters For Cost Analysis
Estimated Building Area Requirements
Construction Cost Information
3-vii
-------
Table 7-4 O&M Cost Information
Table 7-5 Cost And Dose For Chemicals Required For Membrane Treatment
Table 7-6 Example Pretreatment Scheme And Concentrate Disposal Method
3-viii
-------
A
B (subscript)
b (subscript)
BCAA
C
C (subscript)
CA
CaCO3
CF
cfs
C12
D-DBP
DBAA
DBF
DCAA
dMTCV/ dt
DV
e (subscript)
EDR
fps
F (subscript)
Fs
F
GAG
gfd
HAAs
HAAS
HAA6
I (subscript)
ICR
PVD
LCell
MBAA
MCAA
MCL
MF
MFI
MPFI
MW
MWCO
Symbols
Membrane area
Bulk solution
Blended water quality parameters (permeate:feed blend)
Bromochloroacetic acid
Concentration
Concentrate
Cellulose acetate
Calcium carbonate
Cleaning frequency
Cubic feet per second
Chlorine
Disinfectants-Disinfection byproducts
Dibromoacetic acid
Disinfection byproduct
Dichloroacetic acid
Time rate of decline in the MTC^
Design value
Element
Electrodialysis reversal
Feet per second
Feed
Solute flux
Water flux
Fw normalized to the average yearly water temperature
Granular activated carbon
Gallons per day
Gallons per day per square foot
Haloacetic acids
Sum of 5 HAAs: MCAA, DCAA, TCAA, MBAA & DBAA
Sum of 6 HAAs: HAAS & BCAA
Influent
Information Collection Rule
Polyvinyl derivative
Active length of membrane area in the bench-scale cell
Monobromoacetic acid
Monochloroacetic acid
Maximum contaminant level
Microfiltration
Modified fouling index
Mini plugging factor index
Molecular weight
Molecular weight cutoff
Water mass transfer coefficient
3-ix
-------
o) Baseline MTCW
Ne Number of elements in a pressure vessel
Nv Number of pressure vessels in a stage
NDP Net driving pressure
NF Nanofiltration
NOM Natural organic matter
P Pressure
p (subscript) Permeate
Q Flow rate
QT Total product flow rate (sum of by-pass and permeate flows)
r Recycle ratio
R Recovery
R (subscript) Recycle
RB Bulk rejection
RF Feed rejection
RBSMT Rapid bench-scale membrane test
ReJTso Manufacturer reported TDS rejection
RO Reverse osmosis
RPD Relative percent difference
s (subscript) Stage
SDI Silt density index
SEBST Single element bench-scale test
sys (subscript)System
T Feed spacer thickness
T°C Ambient temperature in degrees Celsius
Tavg°C Average yearly water temperature in degrees Celsius
TCAA Trichlorpacetic acid
TDS Total dissolved solids
TFC Thin-film composite
THMs Trihalomethanes
THM4 The sum of four trihalomethanes
TOC Total organic carbon
TOX Total organic halides
UF Ultrafiltration
USEPA United States Environmental Protection Agency
UV2S4 Ultraviolet absorbance (wave length = 254 nm)
vc Cross-flow velocity
wccll Acitve width of membrane in the bench-scale cell
wcicmcnt Scroll width of an element
W (subscript) Waste
ATI Osmotic pressure gradient
APe The pressure drop through a membrane element
AP$ The pressure drop through the stage hardware
Q Acceptable fractional loss in the MTCW prior to cleaning
r Stage flow weighted factor
3-x
-------
1.0 Introduction
The Information Collection Rule (ICR) for Public Water Systems (PWSs) (Subpart M of
the National Primary Drinking Water Regulations) will require PWSs to conduct disinfection
byproduct (DBP) precursor removal studies if they meet the applicability criteria described hi
§ 141.141 (e) of the ICR. The DBP precursor removal studies, referred to as treatment
studies, are described hi § 141.144 of the ICR. Part 1 of this manual provides a summary of
the ICR treatment studies requirements.
Treatment studies must be performed on either membrane separations (i.e., nanofiltration
or reverse osmosis), or granular activated carbon (GAC) using either bench-scale or pilot-scale
procedures. Both of these technologies are effective for the removal of DBP precursors, and
are potentially capable of meeting the maximum contaminant levels (MCLs) stated in the
proposed Disinfectants/Disinfection Byproducts (D/DBP) Rule. Stage 1 of this rule establishes
the MCLs for the sum of four trihalomethanes (THM4) and the sum of five haloacetic acids
(HAA5) at 80 ug/L and 60 |ag/L, respectively. The place holders for the Stage 2 MCLs are 40
ja,g/L and 30 jig/L for THM4 and HAAS, respectively. The objective of these treatment studies is
to generate cost and performance data for membrane and GAC processes used to meet the
proposed Stage 2 DBP MCLs when free chlorine is used as the disinfectant.
1.1 Organization Of Part 3
The purpose of this part of the manual is to provide guidance for the pilot- and bench-scale
evaluation of membrane processes in accordance with the requirements of the ICR. This
document provides guidance on three procedures for evaluating membrane performance, a flat-
sheet test (the rapid bench-scale membrane test), a single element bench-scale approach, and a
pilot-scale approach. Section 2.0 provides an overview of concepts and terminology used in
nanofiltration. This section also sites the results of several membrane studies and provides a
general indication of membrane performance in water treatment.
Section 3.0 presents an overview of the three procedures described in this manual: the rapid
bench-scale membrane test (RBSMT), a single element bench-scale test (SEBST), and pilot-scale
testing; and describes the advantages and limitations of each approach. Sections 4.0, 5.0 and 6.0
describe the details and specific requirements of the RBSMT, SEBST and pilot-scale testing,
respectively. Section 7.0 requests specific cost information to develop cost estimates for this
technology.
1.2 ICR Requirements For Membrane Studies
The ICR requires that treatment studies be conducted with the effluent from treatment
processes already in place that remove DBP precursors but prior to the addition of chlorine-based
oxidants that form halogenated DBFs. If chlorine-based oxidants are added prior to full-scale
processes that remove DBP precursors, then bench- or pilot-scale treatment processes must be
used to simulate these full-scale processes. These processes may be optimized, and additional
processes employed, in order to provide an acceptable feed water for the membrane studies.
3-1
-------
Judgment should be exercised when selecting the point at which water is sampled for the study
and the treatment processes used prior to membrane separations.
The ICR has provisions for both quarterly studies using the RBSMT or SEBST procedures
and long-term SEBST studies. For the quarterly bench-scale studies, the ICR requires that two
different membranes with manufacturer reported molecular weight cutoffs less than 1000 Daltons
be evaluated quarterly over one year to evaluate the impact of seasonal variation on membrane
performance. If seasonal variation is not expected to be significant, then four tests must be run to
evaluate the impact of other parameters. For the long-term SEBST studies, the ICR requires that
one membrane with a manufacturer reported molecular weight cutoff less than 1000 Daltons be
evaluated continuously for one year.
The ICR requires that pilot-scale systems be designed as a staged array of elements to yield
information on fouling, pretreatment requirements, cleaning requirements and permeate quality.
Only one membrane type with a manufacturer reported molecular weight cutoff less than 1000
Daltons must be evaluated during pilot-scale studies. The pilot-studies should be run
continuously over a one year period for at least 6600 hours run time, regardless of seasonal
variation.
This treatment studies manual is referenced in the ICR rule and contains specific
requirements of the rule. The ICR requirements listed in this manual include sampling and
reporting requirements for each type of study. This manual also provides guidance to assist in
setting-up and conducting these studies, and this guidance should not be interpreted as specific
requirements. Guidance is provided on selecting membranes, selecting pretreatment processes,
and on the design of bench- and pilot-scale systems. In preparing this manual, an effort has been
made to distinguish between specific ICR requirements and guidance.
3-2
-------
2.0 Background Information
The growth of drinking water regulations, improved water quality analysis and advances in
membrane technology have created applications for membranes that will change accepted
drinking water treatment technology. Public notification of outbreaks of waterborne pathogens
following scientific disclosure and reporting by national and local news media has raised the
public and political awareness of drinking water quality to a point where drinking water
suppliers must use the best available technology to meet drinking water regulations (Fox and
Lytle, 1994; Fox and Lytle 1993). The resulting financial liability is justification enough for
using the most reliable technology to produce drinking water. Membrane technology is clearly
among the leading technologies for producing high quality drinking water and can be used to
meet current and future drinking water regulations hi the United States.
2.1 Fundamentals Of Membrane Processes
2.1.1 Terminology And Concepts
The basic definition of a synthetic membrane is a barrier which separates two phases and
restricts the transport of various chemical species hi a specific manner (Porter, 1990). Under
this broad definition, membranes can vary widely hi composition and characteristics; they can
be polymeric or ceramic, charged or neutral, homogeneous or heterogeneous, and symmetric,
asymmetric or thin-film composite (TFC). This wide range of characteristics yields a variety
of membranes with different separation capabilities ranging from microfiltration (MF)
membranes that remove suspended solids to nanofiltration (NF) and reverse osmosis (RO)
membranes that remove molecules and ions. NF membranes are ideal for the control of
disinfection byproducts (DBPs) as they can remove greater than 90% of the organic matter
while operating at substantially lower pressures than RO membranes.
The resistance to transport imposed by the membrane barrier is termed the membrane
permeability, which can be subdivided into water permeability and solute permeability (Gekas,
1988). Membrane permeability is a function of membrane properties such as the molecular
weight cutoff (MWCO), charge, hydrophobicity and chemical structure; as well as
environmental factors such as the properties and concentrations of individual solutes, pH, ionic
strength and temperature. Of these properties, the MWCO is one of the most important;
however, this is not an absolute cutoff since many parameters affect solute rejection.
A schematic of a membrane separation process is shown hi Figure 2-1, and Tables 2-1 and
2-2 list basic terminology and definitions associated with membrane processes that will be used
throughout this document. The system shown in Figure 2-1 is a tangential-flow system since
the feed stream flows tangential to the membrane surface while permeation occurs
perpendicular to the direction of feed flow. As a result of water permeation, the solution on
the feed side of the membrane becomes concentrated with rejected solutes and is termed the
concentrate stream when it exits the membrane system. Between the inlet and concentrate
outlet, the solution on the high pressure side of the membrane will be referred to as the bulk
solution or the feed/concentrate stream. This system employs recycle to increase productivity
by returning some of the concentrate to the feed side of the membrane. In Figure 2-1 and
3-3
-------
Table 2-2, Q refers to the volumetric flow rate, C refers to the concentration of a constituent,
P refers to the pressure and Ait refers to the osmotic pressure gradient; while the subscripts F,
I, C, p, W, R, and B refer to the feed, influent, concentrate, permeate, waste, recycle and
bulk, respectively.
The fluid velocity tangential to the membrane surface is defined as the cross-flow velocity,
vc, and will decrease hi the direction of flow due to the loss of feed flow as permeate. The
cross-flow velocity is reported hi units of feet per second (fps). Recovery is the ratio of
permeate flow to feed flow, and is a measure of the amount of product obtained per unit of
feed water applied to the system. The system recovery, R,,ys, is the total permeate flow divided
by the total feed flow and defines the overall productivity of the system. The recovery of
individual stages can also be calculated by dividing the stage permeate flow rate by the influent
flow rate entering that stage. The permeate flow rate per unit of membrane area is the water
flux, Fw, and is reported hi gallons per day per square foot of active membrane area (gfd);
while the mass flow rate of a solute through the membrane per unit area is the solute flux, Fs.
The water flux per unit of net driving pressure is the water mass transfer coefficient1,
MTC,y. The MTCW normalizes the flux with respect to pressure and is a good indicator of
membrane productivity. This parameter can be determined from a membrane characterization
curve which is a plot of water flux as a function of net driving pressure. Characterization
curves should be linear and pass through the origin under typical operating conditions
encountered during drinking water treatment. The slope of the characterization curve is the
MTCW. Characterization curves are used to check the validity of calculating the MTC^ by
dividing the flux by the net driving pressure.
In order to quantify the removal of solutes in a membrane system, the concept of rejection
is employed; however, the rejection of solutes can be defined hi three ways (Rosa and Pinho,
1994). The intrinsic rejection, R;, is the concentration of solute rejected with respect to the
solute concentration at the membrane-solution interface. The bulk rejection, RB, is the
concentration of solute rejected with respect to the solute concentration hi bulk solution, CB.
The feed rejection, RF, is the concentration of solute rejected with respect to the feed solute
concentration, CF. RF is the most useful index for the practitioner, since it indicates removal
with respect to feed water quality. Both feed and bulk rejections will be used to quantify
removal hi these bench- and pilot-scale studies. Unless specifically stated, the term
"rejection" will refer to feed rejection throughout this document.
2.1.2 Membrane System Design
An understanding of the fundamentals of membrane applications is necessary to conduct
and interpret bench- or pilot-scale membrane studies. A conventional NF or RO membrane
system is defined as three separate sub-systems consisting of pretreatment, membrane filtration
and post-treatment. A conventional RO/NF membrane system is shown in Figure 2-2.
1 Strictly speaking, a mass transfer coefficient should relate a mass flow rate to a driving
force. The rational for using a volume flow rate to define a mass transfer coefficient is that the
the volume flow rate is related to the mass flow rate through the density of water.
3-4
-------
The purpose of pretreatment in these systems is to control membrane fouling or flux loss.
Scaling is a type of fouling caused by the precipitation of an inorganic salt on the membrane
surface and is controlled by the addition of an acid or antiscalant to either change the
speciation of the anion or complex the metal involved in the formation of the precipitating salt.
Typical limiting salts are CaCO3 or CaSO4. Other mechanisms that reduce productivity in
membrane processes are colloidal fouling, biological fouling and chemical fouling.
The membrane process follows pretreatment and is where the majority of contaminants are
removed. If the membrane surface is scaled or fouled then the productivity of the membrane
system is reduced, and eventually the membranes must be chemically cleaned to restore
productivity. Typical cleaning frequencies for RO/NF systems operating on ground waters are
3 months to 2 years and average about 6 months (Taylor et al., 1990). Surface water systems
using additional pretreatment are termed integrated membrane systems (IMSs) and have only
been piloted. The cleaning frequencies of surface water IMSs can be as low as 1 to 2 weeks
resulting in more difficult plant operation.
Post-treatment consists of disinfection at a minimum and any other unit operations that are
necessary before final distribution. Typical post-treatment unit operations are disinfection,
aeration, stabilization and storage. Aeration may be required to strip dissolved gases such as
hydrogen sulfide. Since membrane permeate can be corrosive, stabilization may be required
to produce a non-corrosive product. Alkalinity recovery is an effective process for recovering
dissolved inorganic carbon (DIG) in the permeate. Alkalinity can be recovered by lowering
the pH prior to membrane filtration in order to convert alkalinity to CO2, and then raising the
pH of the permeate in a closed system to recover dissolved CO2 as alkalinity. Blending
membrane permeate with by-passed feed water can be another means of stabalizing the product
stream; however, this practice negates the benefit of the membrane treatment system acting as
a barrier to pathogens.
In addition to post-treatment, the concentrate stream from membrane processes must be
treated and/or disposed of. Some concentrate disposal methods that may be viable include
deep well injection, ocean discharge, discharge to sanitary sewers, land application and
evaporation ponds; however, the most effective concentrate disposal method will depend on
the concentrate water quality, local regulations and site specific factors. Additional
information about concentrate disposal options can be found in Membrane Concentrate
Disposal. 1993.
Membrane systems are configured in arrays which consists of stages. A typical two-stage,
2-1 membrane array is shown in Figure 2-3. The first stage shown in Figure 2-3 consists of
two pressure vessels in parallel, and the second stage consists of one pressure vessel. Each
pressure vessel typically contains from one to seven membrane elements. The most common
membrane elements used for drinking water production are spiral wound membrane elements
which have an 8" diameter, are 40" long and have 340 to 400 ft2 of active membrane area.
Membrane elements used for drinking water production can also be manufactured in a hollow-
fiber configuration. If each of the pressure vessels in Figure 2-3 contained six 8" by 40"
elements and were producing water at an average flux of 15 gfd, then each element would
3-5
-------
produce 6000 gpd; 72,000 gpd would be produced from the first stage and 36,000 gpd would
be produced from the second stage. The total permeate produced by the system would be
108,000 gpd.
This combination of stages is called a tapered or "Christmas tree" array and is used to
increase recovery of the overall membrane system while maintaining acceptable hydraulic
conditions within each element. As shown in Figure 2-3, the concentrate stream from the first
stage is used as the feed stream to the second stage. By passing the concentrate from up-
stream stages through membranes in down-stream stages additional permeate is produced
resulting in an increase in the system recovery.
Hydraulics are important within the membrane elements and a minimum cross-flow
velocity is recommended to avoid an excessive build-up of ions at the membrane/solution
interface (a phenomena known as concentration polarization). For 8" elements, a flow of 12
to 30 gpm is typically sought in the last element in the pressure vessel to avoid concentration
polarization. If 4" elements are used, a flow of 3 to 6 gpm is typically sought in the last
element in the pressure vessel to avoid concentration polarization. The tapered design is used
to decrease the number of parallel elements in down stream stages in order to maintain an
acceptable cross-flow velocity. Staging and control of concentration polarization must be
considered in pilot-scale or full-scale applications.
2.2 Membrane Technology To Meet Drinking Water Treatment Objectives
Reverse osmosis, nanofiltration, electro-dialysis reversal (EDR), ultrafiltration (UP) and
microfiltration are the five membrane processes which have major application to the
production of drinking water (Taylor et al., 1990). The basic characteristics of these
processes are shown in Table 2-3. Although many factors affect solute rejection by these
processes, a general understanding of drinking water application can be achieved by
associating the minimum size of solute rejected by a membrane process with regulated
contaminants.
One manner of interpreting the characteristics shown in Table 2-3 is to correctly assume
that each membrane process has the capability of rejecting solutes that are larger than the size
shown hi the exclusion column. As shown hi Table 2-3, regulated drinking water solutes can
be simplified into categories of pathogen, organic solutes and inorganic solutes. Pathogens are
subdivided into cysts, bacteria and viruses. Organic solutes are subdivided into disinfection
byproduct precursors (DBPP) and synthetic organic compounds (SOCs). Inorganic solutes
include general parameters such as total dissolved solids (TDS), total hardness, heavy metals
and other inorganic contaminants. Three mechanisms of solute rejection are referenced: size
exclusion (sieving), diffusion and charge repulsion.
EDR processes are capable of removing ions as small as 0.0001 //m, but a charge is
required. Consequently EDR processes are limited to treatment of charged contaminants and
are ineffective for the removal of pathogens and most organic solutes. RO and NF processes
are capable of removing many organic contaminants by sieving, and many inorganic and
3-6
-------
organic contaminants by diffusion. If the membrane seals are not compromised, RO and NF
processes have the potential to remove all pathogens. RO and NF processes have the broadest
span of treatment capability. UF and MF are sieving controlled and can remove most if not all
pathogens from drinking water if the MWCO is tight enough. Consequently these processes
are ideal for removing turbidity and microbiological contaminants, and they are ideal for
treating the majority of drinking water sources in the United States. The application of
UF/MF processes and conventional treatment is similar in that both processes are used to
remove particulates from drinking water; however, UF and MF have a significant advantage
over conventional processes in that a membrane is a solid film which provides a static barrier
to pathogens which are simply too big to pass the membrane. Conventional processes are
more dynamic since a solid must be formed and contact must occur between the floe and the
particle, followed by floe growth and finally sedimentation before effective pathogen removal
can be achieved. Although filtration is another barrier used hi conventional treatment,
numerous studies have shown that high settled turbidities can result hi pathogenic
breakthrough hi the filtered water.
The surface water treatment rule (SWTR) has been in effect since 1993 and was developed
to reduce the potential for pathogenic contamination of drinking water. The SWTR requires
that all waters under the direct influence of surface water achieve a minimum of 3-log removal
of Giarda cysts and 4-log removal of enteric viruses. The USEPA has recommended that well
operated coagulation, sedimentation and filtration systems be given 2.5 log removal credit for
Giarda and 3 log removal of viruses. Disinfection is required by the SWTR to meet a CT
(mg/L disinfectant x disinfectant contact tune) standard. The required CT depends on the type
of disinfectant used, the pathogen and other water quality parameters. The expected CT
requirements for Cryptosporidium will have a major impact on surface water treatment and
may result in the large-scale use of ozone throughout the United States. The enhanced SWTR
is expected to link the required log removal of pathogens, including cryptosporidium, to the
source water quality. Studies have shown that the log removal of all pathogens achieved by
hollow-fiber membrane processes is controlled by the pathogen concentration hi the feed water
and that membranes can achieve complete pathogen removal if the cutoff is tight enough
(Jacangelo et al., 1991). The future ground water disinfection rule (GWDR) may require
some level of microorganism removal or inactivation. Pressure driven membrane processes
have the potential to remove all pathogens regardless of source and should be able to satisfy
the primary inactivation/removal requirements of the GWDR. The least costly membrane
processes, UF and MF, are ideally suited for pathogen removal and have been shown to
consistently remove all pathogens from drinking water, with the exception of viruses which are
removed by UF but not MF. Tighter RO and NF membranes should achieve the same degree
of removal if the brine seals are not compromised. However, all of these membrane processes
should be followed by disinfection to ensure acceptable distribution system water quality.
The improving methods of detection for biological pathogens have shown that
contamination by water-borne, disease-causing agents may well be the most serious problem hi
providing safe drinking water. Specific contamination by Salmonella, Legionella, E. coli and
Cryptosporidium has been documented in the United States. Indications are that the majority
of incidents of water-borne disease are unidentified and unreported. Consequently, the need to
3-7
-------
use the best treatment technology and maintain distribution system water quality has been
emphasized in regulatory action. These biological contaminants should be partially if not
totally rejected by most pressure driven membrane processes. NF and RO processes also
remove much of the disinfectant demand and nutrients which can contribute to loss of integrity
in the distribution system.
Inorganic compounds (lOCs), volatile organic compounds (VOCs) and SOCs are regulated
by the amendments to the SDWA. The IOC, SOC and VOC regulations involve eighty-three
specific compounds which have varying susceptibility to removal by membrane processes;
however, some generalizations can be made. No commercially available membrane process is
being used for VOC removal since VOCs are uncharged, have low molecular weights (MW),
and pass through membranes with water.
MF and UF membranes don't have a low enough MWCO to reject SOCs or lOCs. RO
and NF can achieve significant SOC rejection because the MWCO of these membranes is
sufficiently low that many SOCs cannot pass or are diffusion controlled (Duranceau and
Taylor, 1993). This is also the case with lOCs or radionuclides like uranium and radium.
Although RO and NF have been shown to be among the most promising processes for SOC
and IOC removal, not all SOCs or lOCs can be rejected by these processes. RO and NF use
both sieving and diffusion mechanisms to reject SOCs and lOCs from drinking water, and
rejection will increase as the MW and charge of the contaminant increases. Arsenic and
sulfates will be removed by EDR, RO and NF but not by UF and MF. Typically, charged
solutes and solutes with MWs greater than 200 are highly rejected by RO and NF. UF and
MF can remove SOCs if powdered activated carbon (PAC) is used for SOC adsorption and UF
or MF are used for PAC removal. EDR is a viable means of removing any charged solutes
from drinking water but is not effective for the removal of uncharged contaminants such as
pathogens, SOCs or natural organic material.
The lead and copper rule has been finalized since 1991 and is intended to control the lead
and copper concentrations hi drinking water. Since corrosion occurs in the distribution
system, the chemistry of the finished water is very important, and corrosion is significantly
affected by inorganic solutes such as sulfate, sodium, chloride and bicarbonate. Whenever the
inorganic matrix of a finished water is altered there is a potential for water quality problems hi
the distribution system as equilibrium is reestablished between the finished water and the
distribution system materials. This can result hi unacceptable levels of lead and cadmium,
aesthetic problems such as a metallic taste or red-brown stains, and increased operating costs
due to infrastructure damage (Schock, 1990). UF and MF do not affect corrosivity because
ions are not removed by these processes; however, RO and NF do remove inorganic solutes
from water, and thus can impact the corrosivity of a finished water.
Post-treatment processes are designed to stabilize NF and RO product streams by partially
reminalizing the permeate to minimize corrosion in the distribution system, and some research
has indicated that post-treated product water from membranes can be less corrosive than
conventionally treated finished water for new pipe materials (Taylor et al., 1992). Alkalinity
recovery is a post-treatment process used to pass DIG from the feed stream to the permeate
3-8
-------
stream during membrane treatment. The amount of DIG removed by a membrane process will
depend on the carbonate speciation of the feed water which is a function of the feed water pH.
Since many membrane processes are operated at a reduced pH to control scaling, much of the
DIG can pass through the membrane as aqueous CO2 or carbonic acid. Base is then added to
the permeate to reconvert aqueous CO2 or carbonic acid to bicarbonate or carbonate species.
Since other water quality parameters in addition to alkalinity and DIG can affect the
corrosivity of the water, additional post-treatment steps may be required for corrosion control.
Guidance on corrosion and corrosion control is readily available through the USEPA Lead and
Copper rule Guidance Manual Volume II: Corrosion Control Treatment. 1992; Lead Control
Strategies. 1990; and Internal Corrosion of Water Distribution Systems. 1985.
The final rule for discussion is the Disinfectants-Disinfection byproducts (D-DBP) rule.
EDR has not been shown to be effective for DBF control, and UF and MF have pores which
are too large to reject significant amounts of DBF precursors. The existing DBF MCL is 100
yUg/L for the sum of four trihalomethanes (THM4) but Stage 1 of the D/DBP rule has proposed
MCLs of 80 //g/L for THM4 and 60 ,ug/L for the sum of five haloacetic acids (HAA5) which
could be reduced further to 40 /ug/L for THM4 and 30 /^g/L for HAAS after the turn of the
century.
UF and MF can be used to control DBFs and TOG if these processes are preceded by
coagulation. As with the SOC application, they replace sedimentation and filtration in
conventional treatment and are limited to the 50% to 75% TOG removal that can be achieved
by coagulation (Taylor et al., 1986). RO and NF can achieve greater than 90% TOG removal
and are unmatched for effective TOG and DBF precursor removal in drinking water treatment
(Taylor et al., 1990). Moreover, these processes reject so much of the disinfectant demand
that even though the disinfectant dose is greatly reduced, maximum distribution system
integrity can be achieved because the residual persists longer. A final major advantage of
membranes is the removal of contaminants without producing any oxidation byproducts. All
oxidants, including chloramines, produce disinfection byproducts which may be regulated in
the future.
A summary of the applicability of membrane processes for meeting some of the regulations
described above is provided in Table 2-4.
2.3 Literature Review
The literature describing DBF precursor removal by NF or RO is not extensive. However,
work has been done on surface and ground waters that has shown NF and RO to be effective
processes for the control of DBFs by removing the organic precursors. The following
literature is selected to provide an overview of DBF precursor removal by these membrane
processes.
Membrane material was shown to affect the rejection of TOG in membrane investigations
(McCarty and Aieta, 1983). Cellulose acetate (CA) membranes were found to reject only 50%
of the dissolved organic carbon (DOC), whereas polyamide (PA) membranes were found to
3-9
-------
reject greater than 90% of the DOC for the same source and operating conditions. Even
though the chemical characteristics of TOC are not well understood, this investigation showed
that the surface chemistry of the membrane film affected TOC mass transfer through the film.
Bench scale experiments conducted at the University of Central Florida have shown that
trihalomethane formation potential (THMFP) of the potable water supply (Lake Washington)
for Melbourne, Florida could not be controlled to less than 100 //g/L by aluminum-, iron- or
magnesium-based coagulation. Since chlorine was the only disinfectant used, THM4 in the
distribution system was greater than 100 ^g/L. Solute fractionation of raw and coagulated
Lake Washington water demonstrated that the THMFP precursors were organic species with
nominal molecular weights varying from less than 1000 to over. 10,000. Coagulation was
shown to remove most of the THMFP precursors with a MW above 10,000; however, the
remaining THMFP precursors with a MW of less than 1000 resulted in THMFP
concentrations which exceeded the MCL when only chlorine was used for disinfection. This
work indicated that removal of organic matter with nominal MWs of less than 1000 was
necessary to meet the THM4 MCL (Fouroozi, 1980) indicating that RO and NF processes with
a MWCO less than 1000 Daltons would be effective for controlling DBFs.
Research has been conducted which demonstrated THM precursors could be removed from
surface and ground waters with high THMFP by membrane processes. Membrane filtration of
a highly organic ground water at the Village of Golf (VOG) water treatment plant was
conducted with different membranes hi short-term (2-6 hr) single element studies, and the
study demonstrated that 10% to 90% of the THMFP precursors were removed from the water
over a MWCO range of 10,000 to 500. No significant removal of THMFP precursors could
be realized at lower MWCOs (Taylor et al., 1987). However, softening of VOG water to
concentrations attainable by lime-softening required a membrane with a MWCO of 300.
Identical results for THM precursor removal and softening were observed for the Acme
Improvement District (AID) ground water and for the Caloosahatchee River surface water
(Taylor et al., 1986). These waters were the potable water supplies for AID and Lee County,
Florida, respectively. Based on these studies, a FilmTec NF 50 membrane was selected for
long-term pilot plant operation because the membrane could be operated at lower pressures
(100 psi) and normal recoveries (70% to 80%) while controlling THMFP to less than 100
A conventional nanofiltration pilot plant consisting of acid feed for scaling control,
cartridge filtration and membrane filtration was used to investigate NF for THMFP control at
the ground water sites. The pilot plant was operated for 300 to 600 hours at each site. These
studies found that THMFP could be controlled to less than 100 //g/L while maintaining
consistent production at the highly organic ground water sites. A single 4" x 40" NF element
preceded by acid addition and cartridge filtration was used to investigate NF at the surface
water site. Results from the surface water study indicated that while NF could attain the same
high water quality as observed at the ground water sites, consistent productivity could not be
maintained with conventional pretreatment. Significant organic fouling occurred at the surface
water site, resulting in a rate of flux loss near 0.40 ft/d2 which was more than three orders of
magnitude higher than the fouling rate at the ground water sites. These results showed that
3-10
-------
ground water treatment by NF was feasible, but that surface water treatment by NF would
require additional pretreatment and more frequent cleaning.
In operational studies at the surface water site (Lee County) and the ground water sites
(VOG and AID), the permeate concentrations of inorganic solutes were observed to increase as
the net driving pressure decreased or as system recovery increased. This phenomenon would
be expected when solute mass transfer hi a membrane process is diffusion controlled. The
corresponding permeate concentrations of organic solutes were observed to be independent of
pressure and recovery. This phenomenon would be expected when solute mass transfer in a
membrane process is controlled by sieving (Taylor et al., 1986).
A long-term investigation of THMFP control was conducted at the Flagler Beach, Florida
water treatment plant (a ground water site) and at the Punta Gorda, Florida water treatment
plant (a surface water site). Several membranes were investigated for use hi short-term studies
and found to control THMFP and provide softening. A MWCO of 300 was found to control
THMFP and to provide softening to less than 100 mg/L as CaCO3. These pilot studies were
conducted for approximately one year at each site. A 15,000 gpd membrane pilot plant using
FilmTec NF 70 membranes was operated for one year at Flagler Beach, Florida, and
controlled THMFP to less than 100 jug/L while maintaining consistent production (Taylor et
al., 1990). The Flagler Beach ground water is highly organic with a DOC concentration of 11
mg/L and a raw water THMFP hi excess of 600 y
Another yearlong pilot study was conducted using the same membrane pilot plant to treat
the highly organic surface water at Punta Gorda, Florida. Results of short-term membrane
selection studies demonstrated the same 300 MWCO requirement to achieve THM4 control
and softening as was observed at previous sites at Flager Beach, AID, VOG and Lee County,
Florida. Advanced pretreatment processes of: (1) sand filtration; (2) alum coagulation and
sand filtration; (3) dispersant addition coupled with the two previously noted pretreatment
processes; (4) decreased pressure; and (5) increased cross-flow velocity were investigated.
Although consistent control of permeate THMFP was achieved at the surface water site, the
flux loss was again severe for all conditions tested. Membrane cleaning was conducted on 20
occasions in an attempt to sustain a water flux of 10 gfd. The operating system at the ground
water site required prefiltration and acidification in order to sustain a flux of over 15 gfd with
semiannual cleaning. These studies found, as had previously been shown, that the surface
water was more difficult to treat by membrane processes than the ground water, but high water
quality was produced at both sites. The work was the first successful demonstration of long-
term, consistent production for THM control by membranes using a highly organic source
(Taylor etal., 1990).
The removal of DBP precursors by membrane processes was found to be sieving controlled
in these studies. Furthermore, over 90% of the DOC was removed by membranes with
manufacturer reported MWCOs of 300 to 500 from both surface water and ground water
supplies. Membranes with lower MWCOs removed very little additional DBP precursors.
The THM4 and HAA5 concentrations in the permeate averaged 15 and 4 Mg/L, respectively,
which represented more than a 90% rejection of DBP precursors. Membranes treating surface
3-11
-------
water supplies had to be cleaned every two weeks to maintain production, whereas ground
water membrane systems required cleaning no more frequently than every six months. These
results have found that membrane systems can be used effectively to remove natural organic
contaminants from ground water supplies (Jones et al., 1992).
The results from a pesticide study using nanofiltration indicated the removal of uncharged
organic contaminants such as pesticides and naturally occurring dissolved organic compounds
by RO or NF was controlled by sieving (Duranceau and Taylor, 1990). This investigation
clearly showed that both charge and molecular weight increased solute rejection by
nanofiltration but that charge was far more significant to solute rejection than molecular weight
for the solutes investigated. These investigations also found that for this water, increasing
cross-flow velocity across the membrane surface within the range recommended by the
manufacturer had no effect on the rate of fouling.
Work by Lozier and Carlson hi 1991 and 1992 investigated the treatment of Dismal
Swamp water hi Virginia using CA and polyvinyl derivative (PVD) membranes and found that
TOC removal did increase with increasing recovery for the CA membranes and did not vary
with recovery for the PVD membranes. Hence the mass transfer mechanism for the TOC hi
this work was controlled by the membrane surface characteristics. These results demonstrate
that generalizations about TOC removal hi a membrane process cannot be made.
In 1993, Kohl et al. found that microfiltration could be used to reduce the fouling of both
CA and polyamide membranes used to treat wastewater tertiary effluent that was being used as
a recharge source for a recreational water body. RO cleaning frequencies exceeded one month
for this integrated membrane system. This study and others indicate that surface water fouling
is dependent on the source and membrane surface characteristics.
In a study by Lahie, et al. (1993), the brominated THMFP was found to increase in the
permeate from membrane processes as the MWCO increased. In this study, surface waters
were processed through membranes with varying MWCOs, and bromide spiking studies were
conducted to determine the effect of increased bromide concentrations and MWCO on DBF
formation. NF was found to control brominated DBFs but pretreatment was necessary, and
the concentration of brominated DBFs increased as the bromide concentration increased.
Membrane processes were found to be the most effective processes for removing color and
DBF precursors from a low DOC ground water hi the western United States (Lo Tan and
Amy, 1991). The THMFP was reduced from 182 pg/L to 39 pg/L and the HAAFP from 77
^ug/L to below the detection limit by a membrane with a 300 MWCO. Although membranes
were found to be the most effective process for DBF control, concern was expressed about
pretreatment requirements for this source.
Different performance for different types of nanofiltration membranes have been observed
in membrane pilot studies on a highly organic California ground water (Fu et al., 1994). A
membrane selection study was conducted comparing eight NF membranes, followed by a one
year pilot study which compared a softening and a high permeability NF membrane. The
3-12
-------
selection study found that all the membranes rejected more than 85% of the TOC but that
some of the membranes passed more alkalinity to the permeate. The eight membranes were
classified into a softening group that rejected more than 80% of the alkalinity and a high
permeability group that rejected less than 40% of the alkalinity. The cited advantages of the
high permeability membrane were a higher water mass transfer coefficient, higher finished
water alkalinity and run times in excess of six months between cleanings. This study is
significant in that it demonstrated that a high permeability membrane could operate at lower
pressures, a lower scaling potential, a greater potential recovery, achieve a concentrate TDS
concentration low enough to permit sewer discharge and still achieve high TOC rejection.
A bench-scale procedure, termed the rapid bench-scale membrane test (RBSMT), has been
developed and tested using several different waters and membrane types (Allgeier and
Summers, 1995). The RBSMT can be used to generate data quickly from several different
films. Mesh feed spacers and cross-flow velocities representative of those used in practice are
used to approximate the flow conditions in a actual spiral-wound element. The results of this
study indicated that membranes could be used to meet the requirements of the D-DBP rule
except for a Florida ground water based on the proposed Stage 2 MCLs, and that TOC mass
transfer occurred by diffusion and/or convection. The water quality results were similar to
those of other investigators, but this was one of the first studies that indicated TOC mass
transfer was not controlled by a sieving mechanism. This finding is significant since diffusive
and convective transport of DBF precursors may result hi water recovery being limited by
DBF formation in the finished water.
Existing literature on DBF control by membrane technology has shown that highly organic
ground waters can be successfully treated by nanofiltration systems using conventional
pretreatment and thin-film composite membranes with MWCOs less than 500. Cleaning
frequencies for membrane systems treating ground waters are typically projected at 3 months
to 2 years and have been demonstrated in large-scale pilot plant studies. Membrane systems
treating highly organic surface waters have demonstrated successful TOC removal and DBF
control very similar to ground water systems; however, several pilot studies on United States
surface waters have indicated that cleaning frequencies of two weeks or less will be required to
maintain consistent production. Although a pilot study treating a surface water with a TOC
concentration of 10 mg/L showed cleaning frequencies of greater than two weeks for thin-film
composite RO membranes and greater than two months for CA RO membranes.
There is a need to emphasize research on surface water systems as opposed to ground
water systems for productivity or fouling control. However, the successful application of both
conventional RO membranes and a high permeability NF membrane to surface waters has
shown that high TOC rejection could be achieved while maintaining an acceptable
productivity. Membrane developments by manufacturers are occurring rapidly which
emphasizes the need for bench- and pilot-scale testing of membrane systems for the control of
organic contaminants in drinking water systems.
3-13
-------
2.4 Mechanisms And Control Of Membrane Fouling
There are different mechanisms that can reduce the productivity hi a membrane process
including scaling, colloidal fouling, biological fouling and organic or chemical fouling, and
these fouling mechanisms require different control strategies.
2.4.1 Scaling
Scaling occurs when a sparingly soluble salt is concentrated to its solubility limit and
precipitates onto the feed side of the membrane. A diffusion controlled membrane process
will naturally concentrate salts on the feed side of the membrane, and if excessive water is
passed through the membrane, this concentration process will continue until a salt precipitates
and sealing occurs. Scaling will reduce membrane productivity and consequently recovery is
limited by the allowable concentration just before the limiting salt precipitates. The limiting
salt and maximum recovery can be determined from the solubility products of potential *
limiting salts and the feed stream water quality. Ionic strength must also be considered in
these calculations.
Scaling control is essential hi RO/NF membrane processes. Control of scaling within the
membrane element requires identification of a limiting salt, acid addition and possibly the
addition of an antiscalant. Calcium carbonate scaling is commonly controlled by sulfuric acid
addition; however, other salts such as sulfates can also be the limiting salt. Commercially
available antiscalants can be used to control scaling by complexing the metal ions in the feed
stream and preventing precipitation. Equilibrium constants for these antiscalants are not
available which prohibits direct calculation of the required dose. However, manufacturer
provided computer programs are available for estimating the required antiscalant dose for a
given recovery, water quality and membrane.
To determine the potential for calcium carbonate precipitation, the Langlier Saturation
Index (LSI) should be calculated using an estimate of the concentrate water quality. CaCO3
scaling can be controlled by lowering the pH until the LSI is negative. A conservative
approximation for the maximum allowable pH to achieve scaling control for softening
membranes is given by Equation 2.1.
pH for CaCO3 scaling control < -log[2.4 x 10'u x CH^ED x ALK^n! (2. 1)
(for softening membranes)
where CHj,-EED is the feed calcium hardness hi mg/L as CaCO3 and ALK^ED is the feed
alkalinity in mg/L as CaCO3. For membranes that do not remove hardness, the maximum
allowable pH for scale control can be estimated as:
pH for CaCO3 scaling control < -log[8 x 10'13 x CH^D x ALK^ED! (2.2)
(for non-softening membranes)
These approximations account for the increase in the concentration of the alkalinity and
calcium on the feed side of the membrane, but do not account for concentration polarization
3-14
-------
and actual salt concentrations at the membrane surface and should be checked against more
detailed calculations or a manufacturer's computer program. Also, the LSI only indicates the
potential for calcium carbonate scaling, and the potential for precipitation of other salts must
be checked through consultation or computer programs.
2.4.2 Colloidal, Biological And Chemical Fouling
Colloidal fouling is defined as fouling by particles that exist in the influent stream to the
membrane elements. The particles or colloids are smaller than the exclusion size of the
cartridge filter that is commonly used for pretreatment in a membrane process. The typical
pore size of the cartridge filter used in pretreatment is 1 to 20 //m. The colloids can build-up
on the surface of the membrane and form a cake which is eventually compressed and reduces
flow through the membrane. The resulting phenomena is similar to models that have been
used to describe cake filtration in vacuum filters. Initially during cake formation and build-up,
productivity is not diminished. When the cake collapses or compresses, the resistance to
filtration is increased and the cake must be removed. RO and NF membranes, unlike MF or
UF membranes, cannot be back-washed, and chemical cleaning is required to remove the cake.
Colloidal fouling can be reduced by additional pretreatment such as MF and UF. The
common fouling indices, which are described in Section 2.5, are indicators of the potential for
colloidal fouling.
Biological fouling is defined as biological growth in the membrane element that results in a
reduction in membrane productivity or an increase in the pressure drop through an element.
Unfortunately, no reliable methods have been demonstrated for the prediction of biofouling.
Biological growth can occur hi the feed spacers or on the membrane surface. Biological
growth will always occur in a membrane system as it is practically impossible to maintain
sterile conditions hi a plant environment; however, this growth will not always result hi a
significant productivity loss hi a membrane process. It should be possible to control biological
fouling with additional or advanced pretreatment processes such as disinfection, MF or UF.
Chemical fouling is defined as the interaction of dissolved solutes hi the feed stream with
the membrane surface that results in a reduction hi membrane productivity. Examples of such
interaction would be the adsorption of polysaccharides on the surface of a thin-film composite
membrane. Chemical interaction between solutes and the membrane surface will always occur
to some degree, but membrane productivity may not be reduced. In general, fouling by
organic matter can be reduced by removing a portion of the TOC prior to membrane filtration
through processes such as enhanced coagulation.
2.5 Fouling Indices
Fouling indices can be quickly developed from simple filtration tests and are used to
qualitatively estimate pretreatment requirements. The silt density index (SDI), modified
fouling index (MFI) and mini plugging factor index (MPFI) are the most common fouling
indices, and are determined by monitoring flow through a commercially available Millipore
test apparatus. The water must be passed through a 0.45 um Millipore filter with a 47 mm
diameter at 30 psi to determine any of the indices. Because of the effects of different filters,
3-15
-------
only 47 mm diameter, 0.45 jam pore size Millipore filters should be used to generate accurate
index measurements. The tune required to complete data collection for these tests varies from
15 minutes to 2 hours depending on the fouling nature of the water. Although similar data is
collected for each index, there are significant differences among these indices. These filtration
indices are primarily indicators of the potential for particulate or colloidal fouling.
2.5.1 Silt Density Index
The SDI (ASTM D-4189-82) is the most widely used fouling index and is shown hi
Equation 2.3. The Millipore test apparatus is used to determine two tune intervals, the tunes
to collect an initial 500 mL and a final 500 mL. The tune between these two sample collection
periods is 5, 10 or 15 minutes. The 15 minute interval is used unless the water is so highly
fouling that the filter plugs before the 15 minute interval is realized. In this case, the tune
between the initial and final sample collection is decreased until a final 500 mL sample can be
collected.
SDI=-
where tt is the tune to collect the initial 500 mL of sample, tf is the time to collect the final 500
mL of sample and tj is the total running tune of the test.
The SDI is a static measurement of resistance which is determined by samples taken at the
beginning and the end of the test. The SDI does not measure the rate of change of resistance
during the test and is the least sensitive fouling index.
2.5.2 Modified Fouling Index
The MFI is determined using the same equipment and procedure used to determine the
SDI, except that the volume is recorded every 30 seconds over a 15 minute filtration period
(Schippers and Verdouw, 1980). The development of the MFI is consistent with Darcy's Law
in that the thickness of the cake layer formed on the membrane surface is assumed to be
directly proportional to the filtrate volume. The total resistance is the sum of the filter and
cake resistance. The MFI is derived hi Equations 2.4 to 2.6 and is defined graphically in
Figure 2-4 as the slope of an inverse flow verses cumulative volume curve.
dt
AJM 2APA2 (2.5)
= a + MFI x V
(2.6)
3-16
-------
where Rf is the resistance of the filter, Rk is the resistance of the cake, I is a measure of
fouling potential V is the volume of filtrate (L), Q is the average flow (L/s), and a is a
constant.
Based on the work of Schippers and Verdouw (1980), Morris determined an instantaneous
MFI by calculating the ratio of flow over volume in 30 second increments to increase the
sensitivity of the test (Morris, 1990). Typically, the cake formation, cake build up and cake
compaction or failure can be seen hi three distinct regions on a MFI plot, as shown in Figure
2-4. The regions corresponding to blocking filtration and cake filtration represent productive
operation, whereas compaction would be indicative of the end of a productive cycle.
2.5.3 Mini Plugging Factor Index
The MPFI is defined as the ratio of flow to tune and is shown hi Figure 2-5 for the same
tap water as was used to determine the MFI (Figure 2-4). The MPFI is stated mathematically
in Equation 2.7.
Q' = a + MPFIxt (2.7)
where Q' is the flow at 30 second increments, t is the tune of operation and a is a constant.
The MPFI curve ideally shows regions of blocking filtration, cake filtration and cake
compaction as does the MFI curve. The MPFI is actually the change hi the MTCV as a
function of tune since membrane pressure and area are constant. The MPFI would seemingly
be the best indicator of membrane fouling as the change hi the MTCW with time is the exact
measurement of productivity decline. However since there is very little flow when fouling
occurs, it is very difficult to collect flow and time data that accurately reflects fouling.
2.5.4 Index Guidelines
Some approximations for the feed water fouling indices required to control membrane
fouling are given hi Table 2-5. These numbers are only approximations and do not replace the
need for pilot or bench studies. Pretreatment requirements cannot be determined for most
installation without a pilot study unless actual plant operating data can be obtained on a very
similar water. Pilot studies have been omitted hi the design of some brackish water reverse
osmosis plants using waters that have TOC concentrations and SDI values less than 1.
However, membrane manufacturers typically permit SDIs hi the range of 3 to 4 for spiral-
wound elements.
2.6 Membrane System Design And Operational Factors For Precursor Removal
Membrane systems are affected by the membrane type selected and the conditions under
which the system is operated. Conventional pretreatment is necessary in all systems, and
advanced pretreatment may be necessary in some systems, to control fouling. Following
membrane treatment, post-treatment and concentrate disposal may be required. All of these
factors must be considered to develop an accurate cost estimate for membrane treatment. In
some instances, blending of membrane permeate with raw, partially treated or other process
3-17
-------
treated streams may be possible and should be evaluated for water quality and cost impact.
However, any pathogen control is compromised if blending is used.
2.6.1 Membrane Selection
Membranes can be selected using many different parameters. The primary parameters for
membrane selection are percent NaCl rejection, the MWCO, the water mass transfer
coefficient, resistance to fouling mechanisms of significance and water quality.
Membrane systems can be categorized according to the level of total dissolved solids
(TDS) in the raw water. The suggested levels of TDS classification would be (A) < 1000
mg/L, (B) between 1000 mg/L and 10,000 mg/L and (C) > 10,000 mg/L. Level A systems
are essentially freshwater systems and could be treated by a NF membrane system unless the
majority of the TDS is composed of monovalent ions. Level B systems are brackish water
systems and could be treated by low pressure RO or EDR membrane systems. Level C
systems are salt water systems and require high pressure RO. Membrane application programs
are available from membrane manufacturers or consulting engineers and can be used to
determine the potential recovery and general finished water quality. Consequently the first
step hi membrane selection would be based on raw water quality parameters such as TDS.
Manufacturers generally categorize membranes by percent NaCl rejection or MWCO.
Percent NaCl rejection will only be applicable for membranes selected for treatment of (B)
brackish or (C) salt waters. These waters are generally not highly organic but will require
membranes capable of rejecting greater than 97% NaCl which should be the primary basis for
selection. However, if membranes are used specifically for TOC or precursor control then a
high flux membrane with a higher MWCO can be utilized, Current research has shown that a
MWCO of 1000 Daltons or less is required to achieve greater than 70% TOC rejection and
control DBFs to less than 80 /ng/L THM4 and 60 /ug/L HAAS. If softening is desired then a
MWCO of 300 or less is usually required. Pilot studies in Florida have shown that a
membrane with a MWCO of 500 only removed 30% of the hardness whereas a membrane
with a MWCO of 300 removed 60% of the hardness. However, these are general statements,
and investigators considering membranes for TOC and DBF control can evaluate membranes
with varying MWCOs from 1000 to 100 Daltons. Not all manufacturers measure the MWCO
of membranes, and the method of measuring MWCO varies among manufacturers; thus, the
MWCO is not an absolute performance index.
The water quality of any source must be characterized and the treatment objectives
identified for membranes to be selected! The water quality characterization will be based on
fundamental measurements of fouling indices, inorganic contaminants, organic contaminants
and biological water quality:
The SDI, MFI or MPFI of the raw water should be determined to estimate pretreatment
requirements.
3-18
-------
Typical inorganic water quality parameters include turbidity, TDS, total hardness, calcium
hardness, Fe, Mn, Si, Sr, Ba, Na, Cl, SO4, Br, NO3 and other inorganic solutes that are
regulated drinking water parameters.
Typical organic water quality parameters include TOC; simulated distribution system
(SDS) samples for THM4, HAA6 and TOX; and other organic solutes such as pesticides or
other SOCs that are regulated and may be applicable to the particular drinking water
source. UV254 is a useful surrogate parameter for TOC and DBF precursors.
Typical biological water quality parameters include standard plate count, heterotrophic
plate count, biodegradable dissolved organic carbon, assimilable organic carbon, E-coli,
particle count and other biological parameters that may be applicable to a given drinking
water source.
It is possible that a given source may have one contaminant that will control membrane
selection. Such a contaminant could be nitrates, chlorides, or bromides; however, this
contaminant would be identified in the water quality analysis of the raw water and the ability
of the membrane to reject this contaminant should be considered when membranes are selected
for testing.
The second category for membrane selection will be drinking water source. If the source
is a ground water, then the chemical and biological water quality will allow a relatively
accurate estimation of the potential hindrances to membrane production. Biological activity
and foulants are typically low in ground water systems, and a ground water would typically be
successfully treated by conventional membrane systems; however, some contaminants might
require advanced pretreatment for ground water systems. For example, high concentrations of
Fe and Mn would indicate that the system must either be kept chemically anaerobic or that the
Fe and Mn must be removed by advanced pretreatment prior to the NF or RO membrane
system. Surface water systems will likely require advanced pretreatment or an integrated
membrane system to maintain productivity. If a surface water is being treated, then
membranes that are resistant to biological fouling, biological degradation, particulate fouling
and chemical fouling should be considered. Cellulose acetate membranes are more susceptible
to biological degradation relative to thin-film composite membranes; however, cellulose
acetate membranes are less costly than other membranes. Membranes that can accommodate
high cross-flow velocities or that have feed spacers that produce hydraulic turbulence and are
designed to minimize particulate fouling would be desirable in some applications.
Membranes can be selected on the basis of membrane surface characteristics. Many
manufacturers categorize membranes with respect to the hydrophobic or hydrophilic nature of
the surface. Hydrophilic surfaces would be expected to pass water with the least resistance
and be the least susceptible to chemical fouling but may be more susceptible to biological
fouling. Membrane manufacturers can rank their own products relative to hydrophobicity, but
there is no common ranking among manufacturers.
3-19
-------
A summary of general membrane selection criteria in shown in Table 2-6. Membrane
selection should be discussed with membrane manufacturers and consultants. Additionally,
membrane screening studies can be conducted to quickly evaluate several membranes and
choose the most appropriate film for a given application. The information presented herein
provides general guidelines for membrane selection.
Cost information for pilot studies is difficult to obtain because of the way in which
membranes are typically purchased. Typically, membranes are bid to a set of specifications
supplied by a consultant working for a utility. Membrane manufacturers choose to bid or not
bid following a review of the bid criteria. In many cases there are several membranes which
can be successfully used to treat a given water.
2.6.2 Pretreatment
Scaling control is defined as conventional pretreatment and has been presented earlier.
Advanced pretreatment processes may be necessary to treat surface waters prior to membrane
processes and can be classified by the type of foulant that they remove. For example
pretreatment by UF or MF would be expected to greatly reduce the tendency for biofouling or
fouling by particles larger than 0.01 jam or 0.2 um, respectively. Categories of pretreatment
are shown in Table 2-7 for control of various fouling mechanisms.
Advanced pretreatment would be unit operations that precede scaling control and cartridge
filtration. Examples of advanced pretreatment would be coagulation, oxidation followed by
green sand filtration, ground water recharge, continuous microfiltration and GAC filtration.
Any other unit operations that precede conventional pretreatment are advanced pretreatment by
definition. In some pretreatment unit operations, such as alum coagulation, the feed water is
saturated with a salt such as aluminum hydroxide. In such instances the solubility of such salts
must be accounted for in the feed water stream to avoid precipitation onto the membrane.
Finally advanced pretreatment is a costly addition to a RO/NF membrane process (i.e., the
addition of coagulation, sedimentation and filtration prior to a NF membrane process could
easily increase the cost of treatment by $1/1000 gal or more).
This information is only a guide, and there are many more pretreatment processes than
shown hi Table 2-7. Additionally, pretreatment processes can be successfully combined to
achieve other water treatment objectives while reducing the rate of membrane fouling.
2.6.3 Impact Of Operating Parameters On Performance
The effect of six independent variables on permeate water quality can be considered as
shown in Table 2-8. These parameters are feed stream concentration (CF), recovery (R), net
driving pressure (NDP), solute mass transfer coefficient (Ks), recycle ratio (r) and the water
mass transfer coefficient (MTCW). The effects of these operating parameters on the permeate
stream solute concentration are summarized hi Table 2-8 for sieving and diffusion controlled
mass transfer mechanisms. Table 2-8 should be read as all other variables are constant with
the exception of the noted variable. For example, the terms CFt CpT indicate an increase in
the permeate concentration due to an increase in the feed stream concentration, while RT CpO
indicates no change in the permeate concentration with increasing recovery. Decreasing the
3-20
-------
feed stream concentration results in lower permeate concentrations in both diffusion and
sieving controlled processes; consequently, pretreatment to reduce the feed stream
concentration may be an option for decreasing the permeate stream concentration. If the net
driving pressure (NDP) is increased and all other variables are held constant, then the
permeate concentration will decrease for diffusion controlled species and remain constant for
sieving controlled species. If recovery is increased and all other variables are held constant,
then the permeate concentration will increase for diffusion controlled species and remain
constant for sieving controlled species. These effects may be hard to realize if an existing
membrane array is considered, for it is impossible to increase pressure without increasing
recovery in such an environment.
The maximum obtainable recovery for a system can be controlled either by the
precipitation of a limiting salt or by a diffusion controlled contaminant. For example, assume
that a recovery of 90% could be achieved without reaching the solubility limit of a sparingly
soluble salt, but bench or pilot studies showed that TOC was diffusion controlled. If the
permeate TOC was not low enough to meet the treatment objectives (i.e., the proposed Stage 2
DBF MCLs) at a recovery of 90%, then the system recovery would have to be decreased until
an acceptable permeate TOC was achieved. This is possible with any species and can be
established through bench or pilot studies.
Membrane characteristics such as the solute and solvent mass transfer coefficients will also
affect permeate quality. An increase in the solute mass transfer coefficient will always lead to
an increase in the permeate concentration regardless of the transport mechanism. In general,
the solute mass transfer coefficient will decrease with decreasing MWCO, but this will not
always be the case. The impact of the water mass transfer coefficient on permeate
concentration is identical to the impact of NDP on permeate concentration.
3-21
-------
-------
PRETREATED
FEED(F)WTER INFLUENT (I)
Q, C, P,
PERMEATE (p)
Figure 2-1 Basic Diagram Of A Membrane Separation Process
Pretreatment
Membrane Filtration
Post-Treatment
(jartr
Acid/Antiscalant
Addition
lugej
- -,
filtration
>
]
/ A
3ermeate
Stabilization
DisMfe'Ctoir
Storag^:^--
;Disp
Distribution
Figure 2-2 Conventional NF/RO Systems Showing Pretreatment, Membrane
Filtration And Post-treatment
-------
p-sys
Qp-sys = Qp-1 + Qp-2
Figure 2-3 Two Stage (One Array) Membrane System
^*r
5 ..
4 ,.
3
2
1
Blocking Filtration
Cake Failui
Cake Filtration (slope in this region Is the MFI)
-+-
H-
-t-
-+-
4 6 8 10 12 14
Volume (L)
16
Figure 2-4 Definition Curve For The Modified Fouling Index
-------
0.03
0.025
0.02 --
"at
£, 0.015
-------
Table 2-1 Terminology Used In RO/NF Membrane Processes
Array or Train
Bulk rejection
Bulk solution
Concentrate
Conventional
RO/NF Process
Feed
Feed rejection or
Rejection
Flux
Fouling
High Recovery
Array
Influent
Mass Transfer
Coefficient (MTC)
Membrane
element
Permeate or
Product
Productivity
Pressure vessel
Raw
Recovery
Scaling
Solute
Solvent
Stage or Bank
System Arrays
Waste
Multiple interconnected stages in series.
Percent solute concentration retained by the membrane relative to
the bulk stream concentration.
The solution on the high pressure side of the membrane that has
a water quality between that of the influent and concentrate
streams.
One of the membrane output streams that has a poorer water
quality than the feed stream.
A treatment system consisting of acid or antiscalant addition for scaling
control, cartridge filtration, RO/NF membrane filtration, aeration,
chlorination and corrosion control.
Input stream to the membrane array after pretreatment.
Percent solute concentration retained by the membrane relative to the
feed stream concentration.
Mass (Ib/ft2day) or volume (gal/ftMay, gsfd, gfd) rate of transfer through
membrane surface.
Reduction of productivity measured by a decrease in the temperature
normalized MTC^.
Arrays where the concentrate stream from preceding arrays is fed to
down-stream arrays to increase recovery.
Input stream to the membrane array after the recycle stream has been
blended with the feed stream. If there is no concentrate recycle then the
feed and influent streams are identical.
Mass or volume unit transfer through membrane based on driving force
(gfd/psi).
A single membrane unit containing a bound group of spiral wound
or hollow-fiber membranes to provide a nominal surface area.
One of the membrane output streams that has a better water quality
than the feed stream.
The efficiency with which a membrane system produces permeate
over time.
A single tube that contains several membrane elements in series.
Input stream to the membrane process prior to any pretreatment.
The ratio of permeate flow to feed flow.
Precipitation of solids onto the membrane surface due to solute
concentrations on the feed side of the membrane (i.e. in bulk solution).
Dissolved constituent in any process stream.
Liquid containing solutes; usually water.
Parallel pressure vessels.
Several arrays that produce the required plant flow.
The concentrate stream that exits the membrane system and must
be treated or disposed.
-------
Table 2-2 Definitions Commonly Used In Membrane Processes
Parameter
Recovery
Cross-flow velocity
Total scroll width
Feed spacer thickness
Active area
Water flux
Solute flux
Influent concentration
Bulk concentration
Bulk rejection
Symbol
R
vc
w
T
A
FW
Fs
Q
CB
RB
Definition
(Qp/Qp) x 100%
Qi/(w x T)
Width of flow channel in element or cell
Thickness of flow channel hi element or
Membrane area available for permeation
Qp/A
CpxFw
(QFxCF+(QI-QF)xCc)/QI
(Q + Cc)/2
[(CB - Cp)/CB] x 100%
Feed rejection RF
Osmotic pressure gradient (psi) ATI
Total dissolved solids TDS
Net driving pressure NDP
Concentration gradient AC
Water mass transfer coefficient M.1CW
Solute mass transfer coefficient Ks
[(CF - Cp)/CF] x 100%
(TDSB-TDSp)/100
Total dissolved solids hi mg/L
- Pp - ATT
FVNDP
FS/AC
-------
Table 2-3 Characteristics Of Membrane Processes
Process Mechanism Exclusion
EDR C 0.0001 fj.m
RO S, D 0.0001 ^m
NF S, D 0.001 toa
UF S 0.001 too.
MF S 0.01 urn
Regulated solutes rejected by process
Pathogens Organics Inorganics
None None Most
C, B, V DBPPs, SOCs Most
C, B, V DBPPs, SOCs Some
C, B, V None None
C, B None None
Mechanism: C=charge, S-size exclusion, D=diffusion
Pathogens: C=cysts, B=bacteria, V=viruses
Organics: DBPPs= disinfection by-product precursors, SOCs= Synthetic Organic
Compounds
Table 2-4 Summary Of Membrane Applications For Regulated Contaminants
Rule
SWTR/ESWTR*
(Giardia/Crypto)
IOC
SOC
Radionuclides
DBP precursors
GWDR**
(bacteria/viruses)
Arsenic
Sulfates**
Membrane Process
EDR
no
yes
no
yes (not Rn)
no
no
yes
yes
RO/NF
yes
yes
yes
yes (not Rn)
yes
yes
yes
yes
UF
yes
no
Poss. (w/PAC)
no
Poss. (w/PAC or
coagulants)
yes
no
no
MF
yes
no
Poss. (w/PAC)
no
Poss. (w/PAC or
coagulants)
bacteria yes
viruses no
no
no
*: proposed regulation
**: to be proposed in the future
-------
Table 2-5 Fouling Index Approximations For RO/NF
Fouling Index
MFI
MPFI
SDI
Range
0 - 2 s/L2
0-IOs/L2
0-3x10-5L/s2
0-1.5x10-4L/s2
0-2
0-4
Application
Reverse Osmosis
Nanofiltration
Reverse Osmosis
Nanofiltration
Reverse Osmosis
Nanofiltration
Table 2-6 Membrane Selection Characteristics
Raw water
IDS mg/L
>1 0,000
10,000-1,000
<1.000
Source
Water
GW/SW
GW/SW
GW/SW
% NaCl
Rejection
>97%
>95%
NA
MWCO
Daltons
<100
<150
1,000-300
% DOC
Rejection
80%+
80%+
70%+
% Hardness
Removal
100%
100%
0% to 70%
Membrane characteristic
Hydrophobic
Hydrophilic
Spacer type
Resistance to Fouling Mechanism
Scaling
NA
NA
NA
Plugging
NA
NA
variable
Biofilm
yes
no
variable
Chemical
no
yes
NA
-------
Table 2-7 Advanced Pretreatment For Various Fouling Mechanisms
Advanced Pretreatment
Process
Coag.-Sed-Fil
Bank Filtration
Soil Filtration
Slow Sand Filtration
GAG Filtration
Microfiltration
Ultrafiltration
Ion-Exchange
Oxidation-Filtration
Fouling Mechanism
Scaling
no
no
no
no
no
no
no
yes
possible
Plugging
yes
yes
yes
yes
no
yes
yes
no
yes
Biofilm
no
yes
yes
yes
possible
yes
yes
no
possible
Chemical
possible
yes
yes
yes
yes
no
no
no
possible
Table 2-8 Effect Of Independent Variable On Solute Permeate Concentration By Mass
Transfer Mechanism (IIncrease,! Decrease, 0 No Effect)
Mechanism
Diffusion
Sieving
Independent Variable
CFT CPT
Cpl Cpl
CFT CPT
CFI Cpl
RtCpt
RICPI
RTCpO
RiCpO
NDPt Cpl
NDPI Cpt
NDPT CPO
NDPI CP0
MTCVt Cpl
MTCV1 Cpt
MTCVt CPO
M1XVI CPO
Kst CpT
Ksl Cpl
Kst Cpt
Ksl Cpl
rtCpr
rICpl
rtCpO
rICpO
-------
3.0 Membrane Test Systems
This section describes three procedures for evaluating membranes that can be used to meet
the ICR. Although any ,of the procedures can be used to meet the ICR requirements (with the
exception of plants serving more than 500,000 people, which must conduct pilot studies), each
procedure does provide different data or a different level of detail. The advantages,
limitations and ICR requirements of each approach are described here in order to help utilities
decide on the most appropriate procedure for their individual circumstances.
The three systems described in this section provide different information and judgment
must be exercised when interpreting the results of these studies. The rapid bench-scale
membrane test is the most flexible system but is farthest removed from a full-scale plant,
single element bench-scale tests sacrifice some flexibility in order to evaluate a spiral-wound
element, and pilot systems have the least flexibility but can be scaled to simulate full-scale
performance. All three procedures provide important information, and in an integrated
membrane study (i.e., a study used to develop data sufficient for full-scale design) information
is obtained from a progression of studies starting with bench-scale studies to select a
membrane for use in pilot-scale studies where detailed design information is developed.
3.1 The Rapid Bench-Scale Membrane Test
3.1.1 General Description And ICR Requirements
The rapid bench-scale membrane test (RBSMT) is a systematic bench-scale procedure for
the evaluation of membranes used in a spiral-wound configuration. The test uses a tangential-
flow cell with mesh feed spacers and permeate carriers to approximate a differential element of
a full-scale, spiral-wound module. The cell is operated hi a range of cross-flow velocities used
in practice to simulate the hydrodynamic conditions within a full-scale element. Concentrate
recycle is used to increase the recovery of the bench system, producing a permeate quality
more representative of practice than that obtained at lower recoveries. Operating at a high
recovery also greatly reduces the test water volume requirements.
Although dead-end stirred cells have been used to evaluate membrane performance in the
past, these dead-end cells can not be used for bench-scale ICR experiments. Since these cells
are operated without a concentrate waste, the concentration of rejected solutes at the
membrane surface will never reach steady-state. Also, these systems are not operated hi a
tangential flow and are operated without the mesh feed spacers used hi full-scale spiral-wound
elements. These operating conditions are not representative of conditions hi full-scale spiral-
wound elements, and therefore cannot be used hi an ICR treatment study.
To meet ICR requirements, four RBSMT runs must be conducted with two different
membranes with manufacturer reported MWCOs less than 1000 Daltons. These two
membranes must be evaluated each quarter for four quarters, resulting hi a total of thirty-two
runs over one year. The purpose of the four RBSMT runs is to evaluate the impact of
recovery on permeate quality, and two membranes are evaluated in order to compare their
performance. Quarterly experiments are required to evaluate the impact of seasonal variation
3-23
-------
on membrane performance. These requirements are summarized in Table 3-1, and the detailed
requirements are described in Section 4.2.
If seasonal variation can be evaluated in less than four quarters, or if seasonal variation is
not significant, then the remaining quarterly experiments can be used to investigate other
parameters. Possible investigations include additional membrane types, different
pretreatments, different operating parameters and long-term RBSMT studies. These options
are discussed hi Section 4.2.
3.1.2 Advantages And Limitations Of The RBSMT
The RBSMT has several advantages over larger systems. The small test water volume
requirement (e.g., 200 liters) of the RBSMT allows the procedure to be conducted off-site.
Furthermore, the influent to the membrane study may be batch pretreated eliminating the
potential need for costly, continuous-flow pretreatment. The small membrane area
requirement for this test allows several membranes to be quickly evaluated at a minimal cost
since most manufacturers will provide a sample of membrane material at no charge. The
RBSMT also maintains a great deal of operational flexibility allowing the recovery, pressure
and cross-flow velocity to be independently varied. Thus, this procedure can be used to
investigate the impact of operating parameters on membrane performance.
The RBSMT also has some important limitations. This short-term, bench-scale test does
not provide data on long-term fouling and performance, and the duration of the test will
usually be too short to demonstrate problems due to biofouling. Also, the small test water
volumes will not capture variations that exist hi natural waters; however, this can be addressed
by collecting a sample representative of the average water quality for a given source and
season, and conducting multiple runs to capture seasonal variations. Finally, membrane
variability can impact results if precautions are not taken to insure that the membrane sample
is representative of the product being tested.
The RBSMT is a new procedure and only a limited amount of verification data is available
at this time. This limited verification data indicates that the RBSMT can provide reasonable
estimates of: (1) initial membrane productivity (i.e., the productivity after a few days of
operation) within 10% of the initial productivity observed in pilot studies, (2) solute rejections
within 2% to 20% of rejections observed hi pilot studies, (3) cleaning frequencies within 40%
of those observed on the pilot-scale, (4) the potential for severe and rapid membrane fouling,
and (5) concentrate water quality (Allgeier et al., 1996). This verification work is continuing,
and will further define the applicability and limitations of this test.
These advantages and limitations must be appreciated when using this procedure. The
flexibility of this test, along with the small test water volume and membrane area requirements
make this procedure ideal as a screening tool prior to a pilot-scale investigation. The RBSMT
can be used to compare different membranes, pretreatments and other operating conditions.
Even if a pilot-scale study is planned for the ICR, the information provided by the RBSMT
approach could be useful when designing the pilot study. If only bench-scale tests are to be
3-24
-------
conducted, the results should be interpreted conservatively as the RBSMT is not a substitute
for a thorough pilot-scale study and does not provide sufficient data for full-scale plant design.
3.2 Single Element Bench-Scale Tests
3.2.1 General Description And ICR Requirements
A single element bench-scale test (SEBST) uses a single membrane element in a continuous
flow mode. Thus, this approach is actually a pilot-scale procedure, but will be considered a
bench-scale test for the purposes of the ICR. The SEBST employs concentrate recycle to
increase the recovery to 75% while the system is being operated at a cross-flow velocity and
pressure within the manufacturer's specifications. The minimum element size to be used in a
SEBST is a 2.5" diameter by a 40" length (2.5" x 40") element. Larger elements can be used,
and standard 4" x 40" elements used in practice are commonly used in SEBSTs. The
performance of 4" x 40" elements may better represent full-scale operation than smaller
elements, and it is typically easier to control the concentrate waste flow rate, and thus the
recovery, for a 4" x 40" element. The SEBST can also use hollow-fiber nanofiltration or
reverse osmosis elements to meet the requirements of the ICR bench-scale membrane studies;
however, hollow-fiber technology is not recommended for surface waters.
All SEBSTs must be run at a recovery of 75 + 5% and at operating parameters, such as
pressure, flux and influent flow rate, within the manufacturer's specifications. The high
recovery is used to challenge the membrane, since permeate quality can decrease with
increasing recovery. Each SEBST must be run continuously for four weeks with allowances
for down-time due to membrane cleaning and minor maintenance.
One option to meet the ICR using the SEBST requires that two different membranes with
manufacturer reported MWCOs less than 1000 Daltons be evaluated each quarter for four
quarters. Two membranes are evaluated hi order to compare their performance. It is
recommended that the two membranes be evaluated simultaneously in parallel, which would
require two separate systems; however, one membrane can be evaluated after the other using a
single system. The quarterly experiments are required to evaluate the impact of seasonal
variation on membrane performance. The impact of recovery does not need to be evaluated
each quarter, as is the case for the RBSMT studies; however, four permeate SDS-DBP
samples must be collected from each membrane each quarter. The detailed requirements of
the SEBST are described in Section 5.0.
If seasonal variation can be evaluated in fewer than four quarters, or if seasonal variation
is not significant, then the remaining quarterly experiments can be used to investigate other
parameters. Possible investigations include, additional membrane types, different
pretreatments and different operating parameters. These options are discussed in Section 5.2.
There is also an option to conduct long-term SEBST studies. Long-term SEBST studies
must be run for a one year period, and only one membrane with a MWCO less than 1000
Daltons needs to be evaluated. The sampling requirements for long-term SEBST studies are
similar to the requirements for the pilot studies. This option is provided to allow small
3-25
-------
utilities to develop data from a long-term study without incurring the expense of a large pilot
system; however, this study is classified as a bench-scale study for the purposes of the ICR.
3.2.2 Advantages And Limitations Of SEBSTs
With respect to scalability, a SEBST falls between the RBSMT and larger pilot-scale
systems. The use of a full-scale, spiral-wound element may provide better data than the
tangential-flow cell used in the RBSMT but provides less information than a pilot-scale study.
A single element bench-scale test is ideal for conducting long-term studies, which can provide
a better estimate of the fouling rate and required cleaning frequency than short-term bench-
scale studies. Similar to the RBSMT, the SEBST uses concentrate recycle to obtain a recovery
more representative of practice. However, the use of recycle does not capture the detail of
pilot systems which provide data on inter-stage performance.
The use of concentrate recycle in the SEBST provides some operational flexibility allowing
the recovery, cross-flow velocity and pressure to be varied; however, it does not facilitate the
rapid evaluation of several different membrane types or non-destructive membrane autopsies,
while the RBSMT maintains these advantages.
A potential limitation of the SEBST is that it must be conducted on-site due to the high
flow requirements relative to the RBSMT. Thus, any pretreatment required for the membrane
system must be continuous-flow, as opposed to batch pretreatment which can be used for the
RBSMT studies.
These advantages and limitations must be appreciated when using this procedure. Since
this procedure is an on-site test, it is ideal for conducting long-term performance studies;
however, this also requires continuous pretreatment processes and staff to monitor the system.
Single element tests have been performed for several years, and currently there is more
confidence in the data produced by SEBSTs than the data produced by the relatively new
RBSMT procedure. However, the correlation between single element performance and pilot-
scale performance has not been sufficiently demonstrated to make the SEBST a substitute for a
thorough pilot-scale study, and this procedure does not provide sufficient data for full-scale
plant design. Thus, SEBST results must be used conservatively. In an integrated membrane
study, a single element test would follow initial RBSMT studies, and be used to demonstrate
sustained performance and provide an estimate of the fouling rate and cleaning frequency.
3.3 Pilot-Scale Testing
3.3.1 General Description And ICR Requirements
Pilot-scale systems typically consist of multiple, staged elements similar to full-scale
plants. Thus, properly designed pilot systems provide the most accurate simulation of full-
scale membrane performance.
Pilot-scale systems can vary widely in their complexity and cost, but minimum constraints
must be placed on pilot-scale systems to be used in the ICR studies. The system must consist
of at least two stages with at least two pressure vessels in the first stage and one pressure
3-26
-------
vessel in the second stage (i.e., a 2-1 array); furthermore, each pressure vessel shall contain
no fewer than three elements. The system must be designed to achieve a recovery of at least
75%, while meeting the design specifications set forth by the membrane manufacturer, such as
the minimum and maximum flow rate to an element and the design pressure range.
Concentrate recycle may be used if high cross-flow velocities are required to minimize
fouling. The minimum element size that can be used is a 2.5" x 40" module, and larger
elements such as 4" x 40", 8" x 40" or 4" x 60" elements can also be used. Although not
required, it is strongly suggested that the smallest standard elements used in practice (i.e., 4" x
40" elements for the spiral-wound configuration) be used in the pilot system in order to
minimize costs but provide appropriate design data.
The ICR requires the evaluation of only one membrane type, so caution must be used when
selecting the membrane for investigation. It is strongly recommended that multiple membrane
types be evaluated using the RBSMT prior to selecting one for pilot testing. This preliminary
membrane screening should be conducted several months prior to the deadline to begin the
pilot-scale treatment study. The membrane investigated must have a manufacturer reported
MWCO less than 1000 Daltons. The pilot study shall be run continuously over a period of
one year, with allowances for down-time due to membrane cleaning, maintenance or other
reasons. The pilot-scale run time should be no less than 6600 hours which represents
approximately 75% of a calendar year. In this manner, the sustained performance of the
system will be demonstrated, and the potential impact of seasonal variation will be realized.
The sampling and analytical requirements for pilot-scale studies are described in Section
6.0. Some inter-stage monitoring is required for pilot studies in order to demonstrate any
differences hi stage performance.
Depending on the size and complexity of a pilot system, the costs can become substantial.
For this reason, a pilot study should be carefully planned to insure that it will meet the
objectives of the PWS as well as the ICR. Bench-scale studies such as the RBSMT or single
element tests can demonstrate the ability of a membrane to meet a treatment objective or
identify the need for additional pretreatment. A pilot-scale investigation is the final step in an
integrated membrane study, conducted after bench-scale pre-studies have provided enough
information to select an appropriate membrane and pretreatment scheme. For this reason, it is
strongly suggested that some sort of pre-study be conducted prior to a pilot study. The large
investment for a thorough pilot study should not be made without some preliminary
information obtained from a pre-study.
3.3.2 Advantages And Limitations Of Pilot Systems
Pilot-scale systems potentially provide the most accurate data and the most detail. If a
pilot system is properly scaled, it can provide data sufficient for full-scale plant design.
However, the objective of the ICR is not to generate detailed design data, but to develop data
sufficient to estimate costs and demonstrate technical feasibility. Thus, a pilot system that
meets the minimum requirements of a 2-1 array operated at 75% recovery may not be
sufficient to generate design data, and caution must be exercised when applying the results of
pilot-scale studies.
3-27
-------
The size and complexity of pilot systems limits their flexibility. Operational parameters
can only be varied within a small range, but this is somewhat countered by the ability to
monitor performance between stages. Also, the evaluation of different membrane types on the
pilot-scale can be very expensive, and it is recommended that membrane screening be done
with one of the bench-scale procedures described previously. An alternative that requires a
moderate additional investment is to operate one or more additional membrane types in single
element systems parallel to the staged pilot system. These parallel, single-element tests afford
a comparison of front end system performance for different membrane types.
The limitations of pilot systems are not detrimental to an integrated membrane study.
Flexibility is required in bench-scale studies where the details of a pilot study can be worked
out. By the time a pilot study is started, the membrane type should be selected, a range of
acceptable operating parameters chosen, adequate pretreatment schemes identified and a basic
design decided upon. The pilot study is then used to demonstrate sustained performance and
develop additional information such as cleaning frequency, stage performance, pressure losses
through a system, membrane replacement frequency and other design parameters.
3-28
-------
Table 3-1 Summary Of The RBSMT-ICR Requirements
First quarter
Membrane 1
Membrane 2
Runl
Recovery = 70%
Runl
Recovery = 70%
Run 2
Recovery = 90%
Run 2
Recovery = 90%
Run3
Recovery = 50%
Run3
Recovery = 50%
Run 4
Recovery = 30%
Run 4
Recovery = 30%
Second quarter
Membrane 1
Membrane 2
Runl
Recovery = 70%
Runl
Recovery = 70%
Run 2
Recovery = 90%
Run 2
Recovery = 90%
Run3
Recovery = 50%
Run3
Recovery = 50%
Run 4
Recovery = 30%
Run 4
Recovery = 30%
Third quarter
Membrane 1
Membrane 2
Runl
Recovery = 70%
Runl
Recovery = 70%
Run 2
Recovery = 90%
. Run 2
Recovery = 90%
RunS
Recovery = 50%
RunS
Recovery = 50%
Run 4
Recovery = 30%
Run 4
Recovery = 30%
Fourth quarter
Membrane 1
Membrane 2
Runl
Recovery = 70%
Runl
Recovery = 70%
Run 2
Recovery = 90%
Run 2
Recovery = 90%
RunS
Recovery = 50%
RunS
Recovery = 50%
Run 4
Recovery = 30%
Run 4
Recovery = 30%
-------
-------
4.0 Rapid Bench-Scale Membrane Test
4.1 RBSMT System
This section describes the system used in the RBSMT procedure and is divided into the
following subsections: RBSMT Definitions, System Design, System Start-up, System Shut-
down, and System Cleaning.
4.1.1 RBSMT Definitions
The RBSMT procedure is centered around a tangential-flow cell with concentrate recycle.
A schematic of the RBSMT system is shown in Figure 4-1 using the same terminology listed
hi Tables 2-1 and 2-2 and Figure 2-1. Table 4-1 lists nomenclature specific to this test, and
many of these terms are used in the design equations presented hi Section 4.2.2.
This is a tangential-flow system since the feed stream flows tangential to the membrane
surface, while permeation occurs perpendicular to the direction of feed/concentrate flow. The
cross-flow velocity is controlled by the influent flow rate (Q^, which is the combined feed
flow (QF) and recycle flow (QR). The concentration of this influent stream (Q) is higher than
the feed concentration (CF) but lower than the concentrate concentration (Cc); furthermore, the
influent concentration increases as the system recovery is increased. As the influent stream
flows through the membrane cell, the bulk solution is concentrated until it exits the cell as the
concentrate stream. The average concentration of the bulk solution (CB) is calculated as the
average of the influent (Q) and concentrate (Cc) concentrations.
4.1.2 System Design
A schematic of a typical system for use with the RBSMT procedure is shown hi Figure
4-2. This system was designed with consideration to full-scale parameters such as recovery,
pressure and cross-flow velocity. The central element of the system is a tangential-flow
membrane cell operated with a feed spacer and permeate carrier typical of those used in full-
scale, spiral-wound elements. A typical tangential-flow cell is shown hi Figure 4-3. The bulk
solution flows through the cell-body bottom, tangential to the membrane surface and through
the feed spacer, making the unit hydrodynamically similar to a spiral-wound element. After
permeating the membrane, the permeate flows through the permeate carrier to a central
collection channel where it exits the cell. A small area (24 hi2 typical) membrane sheet is
sandwiched between the two halves of the cell-body and placed into a cell-holder (not shown).
The cell-holder seals the cell body either pneumatically or mechanically. The membrane cell
should be capable of operating up to pressures of at least 50 psi for high flux membranes;
however, pressures as high as 125 psi may be required for lose RO membranes or if the
system is to be operated in a constant flux mode with varying pressure. It is recommended
that the cell-body be constructed of a chemical resistant material such as stainless steel or
plastic. Cells that meet these criteria are readily available from manufacturers.
The dimensions of the active membrane area (i.e., the area available for permeation) are
shown in Figure 4-3 (in the cell-body bottom) and must be determined to calculate some
operating parameters. The length of active membrane area (Lcell) is measured parallel to the
3-29
-------
direction of feed/concentrate flow and the width of active membrane area (wcell) is measured
perpendicular to the direction of feed/concentrate flow. The active membrane area is defined
as that portion within the sealing surface of the cell, and will typically be identical to the
dimensions of the mesh feed spacer.
The system can be operated hi either a constant pressure or a constant flux mode.
Constant flux operation is more typical of full-scale operation, but will cause the pressure to
increase during operation, and the system must designed to allow the pressure to vary. If
constant pressure operation is used, the pressure can be maintained by setting a pressure relief
valve, located just downstream of the feed pump, at the desired feed pressure.
This system is equipped with a recycle loop so that a portion of the concentrate can be
recycled, enabling the system to be operated at a representative recovery and cross-flow
velocity. Other important features are described below by numbered paragraphs which
correspond to the boxed numbers in Figure 4-2.
1. The feed pump should be capable of providing pressures at least 10% greater than the
maximum estimated operating pressure (e.g., 50 to 130 psi for NF membranes) and a
range of flow rates anticipated during operation (e.g., 0.16 gph to 3.1 gph for a 24 hi2
membrane). The pump does not need to be sized exactly since some of the pump
effluent can be returned to the feed tank via the by-pass line, or a variable speed drive
can be used to obtain the desired flow rate. (If constant flux operation is to be used, a
variable speed drive may be required since by-pass may prevent the pressure from
increasing as the membrane fouls). Pulseless flow is also desired since membrane
performance can be affected by rapid pressure fluctuations. Rotary vane and gear
pumps are capable of meeting these requirements. However, if a rotary vane pump
with graphite vanes is used, it may be necessary to use a 1 jam in-line filter just
downstream of the pump to prevent graphite from depositing on the membrane.
2. An adjustable pressure relief valve is placed just downstream of the feed pump to
control the system pressure and minimize the risk of over-pressurizing the system. For
constant pressure operation this valve is set to maintain the desired feed pressure, and
only the permeate water flux is allowed to vary with time. An alternate mode of
operation will allow the system to operate at a constant flux and varying pressure. If
the system is designed to operate at high pressures (e.g., 130 psi), the pressure relief
valve can be set to crack at a high pressure and the system operated in a constant flux
mode. The advantage of operating at a constant flux is that it is more representative of
full-scale operation. The disadvantage is that a high pressure system may be more
expensive.
3. A flow meter can be placed downstream of the feed pump but upstream of the recycle
loop to monitor the feed flow entering the system. This flow meter is optional since
the actual feed flow rate will be calculated by adding the volumetrically measured
permeate and waste flow rates; however, this flow meter may be useful when setting
the operating conditions during an experiment.
3-30
-------
4. A flow meter must be placed downstream of the junction where the feed flow is
blended with concentrate recycle. This flow meter measures the influent flow rate to
the membrane cell which is used to calculate the cross-flow velocity.
5. Gauges should be placed just upstream of the membrane cell to monitor the temperature
and pressure of the influent stream. The temperature gauge should be able to read
values 20°C above ambient temperature. The feed pressure gauge should be able to
read values at least 10% higher than the maximum estimated operating pressure.
Accurate temperatures are required to correct the water fluxes to a common
temperature, and accurate pressures are required to calculate the water mass transfer
coefficient which normalizes flux with respect to pressure.
6. The requirements of the tangential-flow cell have been described earlier, and several
manufacturers provide acceptable cells.
7. For some tangential-flow cells, a compressed gas will be required to pneumatically seal
the system. The gas should be non-combustible, such as nitrogen or argon.
8. A pressure gauge on the permeate line is optional since the permeate pressure will be
close to atmospheric pressure.
9. An optional flow meter placed on the permeate line is useful for setting operating
parameters; however, the flow rate must be measured with a graduated cylinder and
stopwatch to confirm the system settings and obtain accurate flow measurements.
10. A pressure gauge should be placed on the concentrate line and should have the same
pressure range as the feed pressure gauge.
11. The concentrate stream should be split with a union tee so that a portion of the
concentrate stream can be wasted while the remainder of the concentrate flow is
recycled.
12. A fine needle valve should be used to control the waste flow rate which will control the
system recovery.
13. An optional flow meter placed on the concentrate waste line is useful for setting
operating parameters; however, the flow rate must be measured with a graduated
cylinder and stopwatch to confirm the system settings and obtain accurate flow
measurements.
14. A recycle pump should be placed in the recycle line to boost the pressure of the recycle
stream. The pump only needs to provide a differential pressure of about 5 to 30 psi,
but should be capable of providing flow rates up to 0.4 gpm. This pump should
produce pulseless flow and should not shed particles, since these particles would
accumulate in the recycle loop. A gear pump with Teflon or stainless steel gears is
,3-31
-------
suitable for this application. Either a variable speed pump or a needle valve (at
location #15) can be used to control the recycle flow rate.
15. A needle valve will be required to control the recycle flow rate if a variable speed
pump is not used at location #14.
16. A check valve is required at this location to prevent feed flow from by-passing the
membrane cell and going directly to waste.
It is recommended that all parts of the system hi contact with water be constructed of
stainless steel, Teflon or glass. The tubing used hi the system can be 1/4" or 3/8" stainless
steel or Teflon that is rated for the system pressures. The cost of the system will be reduced
by bending the tubing instead of using unnecessary fittings such as elbows.
The system design presented hi this section does not have to be followed exactly, but
presents the critical elements of the system: concentrate recycle, a tangential-flow cell using
mesh spacers, and a separate pump for feed and recycle flows. Any design that utilizes
tangential-flow through mesh spacers at representative cross-flow velocities, recoveries,
pressures and fluxes can be used.
4.1.3 System Start-up
The following section will describe the general steps to follow during initialization and
start-up for the RBSMT system described hi Section 4.1.2. The numbers in parentheses refer
to components labeled by the boxed numbers in Figure 4-2.
1. Select a membrane type and obtain a membrane sample or a small membrane element
that can by cut apart, and the manufacturer's specification sheet for that membrane.
2. Conduct a thorough analysis of the source water to evaluate the inorganic chemical
matrix and determine limiting salts by calculation or manufacturer computer program.
3. Conduct fouling index tests, calculate indices and determine the potential for fouling
problems. If a fouling problem is indicated, additional pretreatment alternatives may
need to be evaluated.
4. Select a membrane flux rate consistent with the fouling potential of the water.
5. Calculate or consult manufacturer computer programs for feed water acid and/or
antiscalant dose.
6. Cut out a membrane sheet using a template to ensure a proper fit within the cell. A
straight edge and utility knife are suitable for this task. Care must be taken when
cutting the membrane to avoid damage to the film. Membranes are sometimes stored
in a preservative that can be harmful if ingested, so caution should be used when
working with new membranes. After the initial wetting of a new membrane, it should
3-32
-------
not be allowed to dry out and should be stored in water, or a solution recommended by
the membrane manufacturer, and kept at 4°C.
7. Cut out a permeate carrier and feed spacer for use in the membrane cell. These mesh
spacers must fit snugly within the cell cavities without extending beyond the boundaries
of the cavities. These spacers can be obtained from membrane manufacturers and
reused indefinitely as long as they are not damaged. Each membrane type should be
used with its corresponding spacers when possible; however, most cells will be limited
to a specific spacer thickness, such as 34 mils (0.0028 ft).
Place the feed spacer hi the cell-body bottom and place the membrane over the feed
spacer. The active side of the membrane, typically the shiny side, should face the feed
spacer. Avoid touching the active side of the membrane.
Wet the permeate carrier with laboratory clean water, and place it in the cell-body top.
Assemble the cell-body. The adhesive forces between the water and the cell-body will
keep the permeate carrier in place when the cell-body top is turned over.
Place the cell-body into the cell-holder and open the valve on top of the cell-holder to
pressurize the pneumatic ram that seals the cell-body.
Attach the concentrate, influent and permeate cell fittings to the corresponding system
connections.
Open the pressure relief (#2) and recovery (#12) valves so that upon start-up the system
pressure will be low, and water will flow freely to waste.
Place the feed line into the feed tank and turn on the feed pump (#1).
If operating hi a constant pressure mode, adjust the pressure relief valve (#2) until the
feed pressure is close to the desired setting. If constant flux operation is used, set the
pump feed rate to achieve the desired flux, and set the pressure relief valve to crack at
a pressure below the maximum operating pressure for the system.
16. Turn on the recycle pump (#14).
17. Slowly adjust the recovery valve (#12) until the waste flow rate is close to the desired
setting.
18. Adjust the flow rate on the recycle pump (#14) or the needle valve (#15) until the
desired influent flow rate to the cell is achieved.
8.
9.
10.
11.
12.
13.
14.
15.
3-33
-------
19. Repeat steps 15, 17, and 18 until the desired waste flow rate, feed pressure (or
permeate flux), and influent flow rate are achieved. Throughout the run, these valves
and variable speed controllers can be manipulated to maintain the desired settings.
20. Take initial readings, but wait one hour to take any permeate water quality samples.
4.1.4 System Shut-down
1. Take final readings and samples.
2. Open the recovery valve (#12).
3. Slowly open the pressure relief valve (#2) to relieve the system pressure.
4. Turn off the recycle pump (#14).
5. Turn off the feed pump (#1). If the membrane is to be cleaned, follow the procedure
discussed hi section 4.1.5 at this point.
6. When the system pressure is zero, disconnect the concentrate, permeate and influent
lines from the membrane cell.
7. Slowly release the pressure in the pneumatic cell-holder and remove the cell.
8. Carefully remove the membrane from the cell and store it in an appropriate solution at
4°C.
4.1.5 System Cleaning
After the membrane has been operated with a test water, the MTCW will decrease due to
fouling. Some of this lost productivity can be recovered by cleaning the membrane with
agents recommended by the manufacturer. Membranes are usually cleaned when the
temperature-normalized MTCW has decreased by 10 to 15% from the baseline at the beginning
of the study or the baseline established after the most recent cleaning.
Since membranes are made from many different materials, membrane manufacturers
specify chemicals, chemical strengths, temperatures and pH values for cleaning solutions.
Membrane compatibility with a specific cleaning solution must be verified with the membrane
manufacturer to avoid damage to the film. There are two basic categories of membrane
cleaning solutions, alkaline and acidic solutions. In general, alkaline solutions such as a 0.1%
solution of sodium EDTA and a 0.1 % solution of sodium hydroxide are effective for removing
organic and biological fouling agents. Alkaline solutions are typically used in conjunction
with surfactants and detergents such as sodium lauryl sulfate or Triton-X. Acidic solutions
such as a 0.5% solution of phosphoric acid are effective for removing inorganic foulants.
When the exact nature of the foulant is not known, the manufacturer's recommendation for an
appropriate cleaning procedure should be solicited.
3-34
-------
In practice, spiral-wound elements are cleaned in their housings by recirculating the
cleaning solution. To simulate this practice, the membrane in the bench-scale cell should be
cleaned within the cell-holder. Figure 4-4 shows how the membrane can be cleaned while in
the sealed (pressurized) cell-holder. With the system pumps off, the influent and effluent lines
for the recycle pump and the influent, concentrate and permeate lines on the membrane cell
should all be disconnected. Flexible tubing should be attached to the influent, concentrate and
permeate fittings on the membrane cell. The tubing on the influent port of the membrane cell
should be connected to the discharge side of the recycle pump. Tubing from the suction side
of the recycle pump should be placed into a reservoir containing the cleaning solution (a 1 liter
volume should be sufficient). The free ends of the concentrate and permeate lines should be
placed into the cleaning reservoir so that the cleaning solution can be completely recycled.
The recycle pump should then be operated at a high cross-flow velocity (e.g., 0.5 to 1 fps)
for a period of time recommended by the manufacturer which could range from 0.5 to 24
hours. Membranes may also be soaked hi a cleaning solution for a period of time, and for
highly fouled membranes, some manufacturers will recommend heating the cleaning solution.
After the cleaning is complete, the membrane cell should be flushed with laboratory clean
water. All chemicals should be disposed of in a safe and approved manner.
4.2 RBSMT Procedure
This section describes the RBSMT procedure with respect to the requirements of the ICR.
Figure 4-5 presents the steps of the RBSMT studies to meet the ICR requirements. This
section is divided into the following subsections: Obtaining Membrane Samples And
Manufacturer Data, Selecting Operating Parameters, Membrane Pretreatment And Treatment
Study Influent, Steps Of The RBSMT Procedure, Monitoring And Sampling Requirements,
Seasonal (Quarterly) Variation, Data Sheets, and Interpretation Of The Results. The data
sheets, referenced in this section as Tables 4-5 through 4-17, are described hi Section 4.2.7.
Utilities electing to conduct RBSMT experiments will be provided with data collection
software that can be used to record and report the data elements presented in these data sheets.
The requirements of the ICR when using the RBSMT procedure are: (1) at least two
membranes must be evaluated to compare their performance, (2) four recoveries must be
investigated for each membrane to evaluate the impact of recovery on permeate quality, and
(3) four sets of quarterly experiments on each membrane must be conducted to evaluate the
impact of seasonal variation on membrane performance, or if a source is not subject to
significant seasonal variation, to evaluate the effects of other parameters on membrane
performance.
4.2.1 Obtaining Membrane Samples And Manufacturer Data
According to § 141.144(b)(l)(ii) of the ICR rule, a minimum of two different membrane
types with manufacturer reported MWCOs less than 1000 Daltons shall be investigated during
RBSMT studies. The membranes should be selected in accordance with the following
treatment objectives: (1) removal of organic matter to levels that would enable the plant to
meet the proposed Stage 2 DBP MCLs when free chlorine is used as the disinfectant, (2) a
3-35
-------
high M1CW per unit cost of membrane material, (3) minimal pre- and post-treatment
requirements, (4) minimization of unnecessary contaminant removal (e.g., removal of total
hardness when softening is not a treatment objective), and (5) minimization of concentrate
disposal costs. Section 2.6.1 provides additional guidance on membrane selection.
Membrane manufacturers will be able to suggest appropriate membranes to meet the
desired treatment objectives. The manufacturer should also be able to assist in determining the
pretreatment requirements and appropriate cleaning procedures. Many manufacturers will
provide a few square feet of membrane material at no charge. The membrane area should be
large enough to produce at least four sheets for use in the bench-scale membrane cell. Mesh
feed spacers and permeate carriers should be requested at a size sufficient for the membrane
cell being used.
Membrane variability is an important consideration when conducting bench-scale studies
with a small membrane area. Membranes of the same type and from the same manufacturer
can vary significantly from batch to batch making scale-up difficult, and the most significant
variations are typically with respect to water flux. If pilot or full-scale investigations are to
follow bench-scale studies, then all membranes should be obtained from the same batch.
Membrane variation can be assessed on the bench-scale by evaluating the MTCW of the
membrane with laboratory clean water, or the manufacturer's standard testing solution, and
comparing this value to a range specified by the manufacturer. In this manner the flux
characteristics of the membrane sample can be compared to the average value reported by the
manufacturer.
The use of a representative membrane sample is critical for obtaining useful data from the
RBSMT. Sample representativeness can be verified by comparing the observed clean-water
MTCV and the rejection of an easily measured parameter, such as total dissolved solids, with
the manufacturer's specifications. The MTCW values must be normalized to a common
temperature for an accurate comparison. If a membrane sheet is outside of the manufacturer's
specifications, then a new sheet should be cut from a different area of the membrane sample.
If the entire sample is outside of specifications, then a new sample should be requested from
the manufacturer.
The information that should be obtained from the manufacturer is listed in the data sheet
presented hi Table 4-5. This information will be used to design the bench-scale membrane
studies.
4.2.2 Selecting Operating Parameters
There are several operating parameters that must be selected for the RBSMT. These
include pressure, flux, cross-flow velocity and recovery. Table 4-2 presents four sets of
operating conditions designed to achieve various simulations: (1) a conservative average
system recovery for a full-scale plant, 70%; (2) the recovery in the final stage of a full-scale
plant, 90%; (3) an average recovery for a full-scale plant, 50%; and (4) the first stage of a
full-scale plant, 30%. If the precipitation of a sparingly soluble salt limits the recovery, then
the highest obtainable recovery should be investigated for the second simulation. The
3-36
-------
pressure/flux and cross-flow velocity should be held constant over these four runs to isolate
the effect of recovery on permeate quality. These four sets of conditions must be evaluated for
each membrane during each quarter.
The only specific values listed in Table 4-2 are the recoveries. The design cross-flow
velocity (or minimum flow rate to an element), the design flux and pressure must be obtained
from the manufacturer to complete the experimental matrix.
Typical pressures in nanofiltration range from 50 psi to 125 psi, and typical fluxes range
from 10 gfd to 20 gfd but should not exceed 15 gfd for a high fouling surface water. The
design flux and pressure are interdependent, and either the flux or the pressure should be held
constant over the study. In either case the data will be normalized to a common measure of
productivity, the temperature-normalized MTCW.
Typical cross-flow velocities range from 0.15 to 1.0 fps, and the design cross-flow velocity
or minimum feed flow rate to an element can be obtained from the manufacturer. Typical
minimum influent flow rates to a 4" x 40" element range from 3 to 6 gpm. Lower cross-flow
velocities will typically lead to a greater degree of fouling and produce more conservative
data. The required influent flow rate to the membrane cell can be calculated from the design
cross-flow velocity and the following equation:
Ql-cell = Vc-aesiga X (T X Wcell) (4.1)
where Q^, is the influent flow rate to the bench-scale membrane cell required to achieve the
design cross-flow velocity, vc.design is the design cross-flow velocity obtained from the
manufacturer, T is the feed spacer thickness obtained from the manufacturer, and wcel, is the
active width of membrane hi the membrane cell. Any consistent set of units can be used with
this equation.
Most manufacturers will specify a minimum influent flow rate to be applied to the element
instead of a design cross-flow velocity. The required influent flow rate to the membrane cell
can be related to the influent flow rate to the element by the following equation:
Ql-cell - Ql-element x
welement (4.2)
where Clement & the minimum influent flow rate to the element as specified by the
manufacturer, and welement is the total scroll width of the element as specified by the
manufacturer.
A representative recovery is more difficult to obtain since the recovery increases in the
direction of flow in a full-scale plant. Furthermore, downstream stages typically produce a
smaller fraction of the total permeate flow than the lead stage. One approach to reduce these
variable recoveries to an average system recovery is by weighting the recoveries of each stage
3-37
-------
of a full-scale plant with the percentage of the total permeate flow produced by that stage.
Representative average system recoveries range from 40% to 60%. Thus, the 50% recovery
in Table 4-2 is intended to produce an average permeate quality for a large-scale system while
the 70% recovery is designed to produce a more conservative estimate of the average permeate
quality.
The osmotic pressure gradient must be estimated in order to calculate the flux. Equation
4.3 can be used to approximate the osmotic pressure gradient from the TDS concentration in
the feed water and other system parameters:
A« = 0.01XTDSFX(1 - Rx(l -
/ (1 - R)
(4.3)
where is ATE an estimate of the osmotic pressure gradient in psi, 0.01 is the factor to convert
from mg/L of TDS to osmotic pressure in psi, TDSF is the TDS concentration in the feed
water in mg/L, R is the fractional recovery and RCJTDS is the TDS rejection of the membrane
as reported by the manufacturer.
The operational parameters listed in Table 4-2 can be used to estimate the appropriate flow
rates for each simulation. For constant pressure operation, the initial permeate water flux can
be estimated from the feed pressure, an estimate of the osmotic pressure gradient and the
MTCjy using Equation 4.4; however, this flux will decrease as the membrane fouls. If the
system is to be operated hi a constant flux mode, a design flux is selected and Equation 4.4 is
rearranged to solve for the initial feed pressure; however, this pressure will increase as the
membrane fouls:
w = MTCWX (PF - ATI)
(4.4)
where Fw is the permeate water flux hi gfd, MTCW is the water mass transfer coefficient in
gfd/psi and PF is the feed pressure in psi.
The permeate flow rate is calculated using the water flux and the active area of the
membrane sheet used in the bench-scale cell:
QD =
(4.5)
where Qp is the permeate flow rate and Acell is the active area of membrane in the bench-scale
cell.
The feed flow rate is estimated from the permeate flow rate and the fractional recovery
using Equation 4.6:
(4.6)
where QF is the feed flow rate.
3-38
-------
The concentrate waste flow rate is the difference between the feed and the permeate flow
rates as shown in Equation 4.7:
Qw = QF-Qp (4.7)
where Qw is the concentrate waste flow rate.
The flow estimates calculated according to Equations 4.5 through 4.7 are only intended to
provide a starting point for the simulation. The flows will have to be modified during the
actual experiments to maintain the desired operating conditions. Table 4-6 should be used to
report the specific values of the experimental matrix.
The following example demonstrates the use of these equations to design an RBSMT study.
For the purposes of this example, assume the following information has been obtained from
the cell manufacturer and the membrane manufacturer. Also, this system will be operated in a
constant pressure mode at a feed pressure of 85 psi, and at a recovery of 0.70.
wcell = 0.33ft
Acdl = 0.167ft2
T = 0.0025ft
Vc-design = 0.30 fpS
welement = 12ft
Ql-element = 4 gpm
MTCW = 0.20gfd/psi
TDSF = 300mg/L
RejTDS = 0.90
The minimum required influent flow rate to the cell can be calculated according to either
Equation 4.1 or 4.2.
Qi-ceii = vc.designxTxwcel, = 0.3 fpsxO.0025 ftxO.33 ft
= 0.0002475 ft3/s
= 420 mL/min
Ql-cell = Ql-element><(wceIl/Welement) = 4 gpmX(0.33 ft / 12 ft)
= 0.11 gpm
= 420 mL/min
Thus, both approaches yield the same answer, and either Equation 4.1 or 4.2 can be used
depending on the available information.
Equation 4.3 can be used to approximate the osmotic pressure gradient for the system
which will then be used to estimate the flux.
3-39
-------
ATI = 0.01 XTDSFX(1 - RX(1 - RejTDS)) / (1 - R)
= 0.01X300 mg/Lx(l - 0.70X(l - 0.90)) / (1 - 0.70)
= 9psi
Equation 4.4 can be used to calculate the permeate flux. For this calculation it will be
assumed that the temperature at which the MTCW was determined is close to the temperature at
which the RBSMT studies will be conducted. (If this is not the case, then the MTCV should
be corrected to the temperature at which the study will be conducted using an appropriate
temperature correction equation, such as Equation 4.11.)
Fw = MTCVx(PF - ATT) = 0.20 gfd/psix(85 psi - 9 psi)
= 15.2 gfd
The permeate flux and the active area of the membrane sheet can be used with Equation
4.5 to calculate the permeate flow rate.
Qp = FwxAcell = 15.2 gfdxO.167 ft2
= 2.53 gpd
= 6.7 mL/min
The feed flow rate is then calculated from Equation 4.6 and the system recovery.
QF = Qp / R = 6.7 mL/min / 0.70
= 9.6 mL/min
The concentrate waste flow rate is calculated by subtracting the permeate flow rate from
the feed flow rate using Equation 4.7.
Qw = QF~ QP = 9.6 mL/ min - 6.7 mL/min
= 2.9 mL/min
These calculations have established the system operation parameters that need to be set for
an experiment. The feed pressure will be set at 85 psi using the pressure relief valve. The
influent flow rate will be set at 420 mL/min using the controller on the recycle pump or a
needle valve hi the recycle line. The concentrate waste flow rate will be set at 2.9 mL/min
with the needle valve on the waste line. The remaining parameters will be set by these three
operating parameters. The system will need to be adjusted over the course of a run to
maintain the desired recovery, cross-flow velocity and pressure.
4.2.3 Membrane Pretreatment And Treatment Study Influent
The requirements of the treatment study influent are described in Part 1, Section 4.0 of this
manual. Typically, 150 liters of influent will be sufficient to evaluate a single membrane;
thus, a 300 liter batch of water can be used to conduct a set of quarterly experiments. A better
estimate of the feed water volume requirements is provided by Equation 4.8.
3-40
-------
V = 3.785 xFVxAuflXKVRj) + (VR^ + (t3/R3) + (t4/RJ] (4.8)
where V is the test water volume requirement (liters) and ^ is the number of days during
which the cell is operated at a recovery of R,,. The recoveries that are listed in Table 4-2 are:
R! = 0.70, R2 = 0.90, R3 = 0.50, and R4 = 0.30. Reasonable estimates of the times are: ^
= 3.5 days and tj = t3 = t4 = 0.5 days. Using these values, Equation 4.8 reduces to:
V = 3.785xFwxAceI1x8.22 (4.9)
This equation provides a quick estimate of the feed water volume requirements; however,
Equation 4.8 should be used when there are significant deviations from the assumed tunes or
recoveries.
The treatment study feed water must be representative of the source under investigation.
For example, experiments on an anaerobic ground water should be conducted in a way to
maintain anaerobic conditions. This is important as the introduction of oxygen into an
anaerobic ground water can increase fouling. Anaerobic conditions can be maintained by
conducting the test on-site, or shipping and handling the water in a manner that does not
introduce oxygen into the sample.
As discussed in Section 2.0, appropriate pretreatment must be used prior to a membrane
process to prevent excessive flux loss due to fouling. Pretreatment for membrane processes
commonly includes chemical addition to prevent inorganic precipitation and cartridge filtration
to reduce colloidal fouling. The most common chemical pretreatment is sulfuric acid addition
to reduce the pH to prevent calcium carbonate precipitation. In some cases hydrochloric acid
is substituted for sulfuric acid. The potential for calcium carbonate precipitation can be
checked by calculating the Langlier Saturation Index for the concentrate stream. Some
applications may require the addition of a chemical antiscalant to control such inorganic
precipitants as calcium sulfate, barium sulfate or strontium sulfate. The use of an antiscalant
can often eliminate the need for acid addition all together, and many antiscalants can prevent
scaling in systems with a concentrate LSI <; +1.5. The required chemical doses are
determined by limiting salt calculations or manufacturer computer programs, both of which
typically require a preliminary and comprehensive chemical analysis of the raw water. The
required chemical dose will typically increase with increasing recovery, and this must be
considered when conducting the RBSMT experiments over a range of recoveries from 30% to
90% . Additionally, the fouling potential of the feed water should be evaluated using one or
more of the methods presented in Section 2.5.
The feed water must also be passed through a cartridge filter to remove larger suspended
solids or colloidal material. Polypropylene cartridge filters with a size exclusion of 5 um are
acceptable for membrane pretreatment.
For feed waters which have excessive fouling as indicated by fouling indices or other
factors, the water can be pretreated by advanced processes which may include enhanced
coagulation and sand filtration with reduced pH. An example of this would be collecting the
3-41
-------
membrane feed water from a conventional treatment plant after sand filtration and prior to the
addition of any chlorine-based disinfectant. If alum is used as the coagulant, the pH of the
feed water may need to be reduced to around 4.5 (or just above the minimum operational pH
as specified by the manufacturer) to ensure the solubility of aluminum hydroxide. However,
operating at a low pH may increase fouling by high molecular weight organic matter.
Additional advanced pretreatment schemes may include microfiltration to control particulate or
microbial fouling.
As mentioned in Section 3.1, one advantage of the RBSMT procedure is that batch
pretreatment can be used, eliminating the need for continuous-flow pretreatment systems and
allowing for greater flexibility in the pretreatment processes investigated. A large reservoir of
water can be batch coagulated, settled, pumped through a cartridge filter and adjusted to the
proper pH to provide a pretreated influent for the bench-scale membrane studies. Advanced
pretreatment such as microfiltration, ultrafiltration, enhanced coagulation, or
ozonation/biofiltration can also be applied in a batch mode. In any case, the choice of
pretreatment is left up to the discretion of the utility as long as the influent sample meets the
requirements of the ICR as discussed in Section 4.2 of Part 1 of this document. The final
treatment study report must include design information for all full-, pilot- or bench-scale
processes preceding membrane treatment and any costs associated with pretreatment processes
not currently used in the full-scale plant. Table 4-7 requests information about foulants and
membrane pretreatment used during the RBSMT experiments.
After the feed sample has received appropriate pretreatment, it can be analyzed for the
water quality parameters listed in Table 4-8.
4.2.4 Steps Of The RBSMT Procedure
Once the membrane, pretreatment and operating parameters have been selected, the runs
can be started. The experiments should be performed in the order listed in Table 4-2, starting
with run #1. Running at a high recovery, 70%, during the first run minimizes the test water
volume requirements, and evaluating the recoveries hi neither ascending nor descending order
partially randomizes the experiments. Below is a step-by-step procedure for conducting a set
of RBSMT experiments.
1. Operate the membrane with laboratory clean water1'2 until the change in the MTCW over
a 12 hour period is less than 4%. This period of operation with laboratory clean water is
referred to as setting, and the cumulative run time should be set at 0:00 at the start of
'An example of laboratory clean water suitable for setting is deionized water with a TOC
concentration below 0.2 mg/L.
2As an alternative to laboratory clean water, the test solution used by the manufacturer
(i.e., a 2000 mg/L of MgSO4 dissolved in deionized water) can be used during setting. Using the
manufacturer's standard testing solution allows direct comparison of the flux and rejection
characteristics of the membrane sample with manufacturer specifications. The setting process is
not affected by most salt solutions used by manufacturers for product evaluation.
3-42
-------
setting. The operating parameters used during setting should be identical to the operating
parameters to be used in the first experiment. If a manufacturer's testing solution is
being used during setting, then the standard testing parameters, specified by the
manufacturer, should be used. The important point in this step is to obtain a MTCW after
the clean water flux has stabilized so that decline observed in the MTCW during operation
with the test water can be attributed to water quality and fouling and not to membrane
setting. The stable MTCW obtained at this point can be compared to the value reported
by the manufacturer to determine the representativeness of the membrane sheet. The
operating parameters monitored during setting should be reported in Table 4-9.
2. After the test water has been pretreated and has equilibrated with room temperature, the
test water can be applied to the membrane. The cumulative run time should be reset to
0:00 at the start of operation with the test water. When the test water is not being used,
it should be kept in cold storage. During operation with the test water the permeate
UV254 and TDS should be monitored with time. TOC can be monitored instead of, or in
addition to UV254 if desired; however, it is usually easier and less expensive to analyze
UV254. The concentrate UV254 and TDS should also be sampled with time but less
frequently than the permeate samples. A reasonable sampling frequency is three times a
day (i.e., once every eight hours) for the permeate samples and once a day for the
concentrate samples.
3. Permeate quality can vary during the beginning of a RBSMT run, and composite
permeate and composite concentrate samples should not be collected until stable permeate
quality has been achieved. Use real time measurements of permeate UV254 as an
indicator of stable solute rejections. Permeate conductivity (or TDS) has been shown to
follow the same temporal trends as permeate UV254 and TOC for a softening membrane
(Allgeier and Summers, 1995). Thus, permeate conductivity (or TDS) can also be used
as an indicator of stable permeate quality for softening membranes, but this approach
may not work for NF membranes with TDS rejections lower than approximately 30%,
and the permeate UV254 may need to be used for these membranes. The data obtained
during operation with the test water should be reported in Table 4-10 for all four runs.
4. Once the change in the permeate water quality (UV254 or TDS) over a 10 hour period is
less than 3%, or within the variability of the analysis, begin collecting one-gallon
(approximate volume) permeate and concentrate samples for a complete water quality
analysis. The permeate sample should be collected in a clean glass container, such as a
one-gallon jug. The first experiment (at 70% recovery) should be run for at least 78
hours beyond setting to insure stable performance.
5. The one-gallon permeate sample should be analyzed for pH, alkalinity, TDS, turbidity,
temperature, total and calcium hardness, bromide, TOC, UV254, and SDS for THM4,
HAA6, TOX and chlorine demand. If the SDS sample cannot be immediately
chlorinated then it should be stored at 4°C until chlorination. It may be advantageous to
collect two one-gallon permeate samples, so that the experiment will not need to be
repeated if there is an error during the SDS test. The permeate water quality and
3-43
-------
chlorination conditions should be reported in Tables 4-11 through 4-15 for run IDs 1
through 4.
6. The one-gallon concentrate sample should be analyzed for TOC, UV254, TDS, turbidity,
pH, alkalinity, total hardness and calcium hardness. These concentrate water quality
parameters are reported in Table 4-16 and can be used to check mass balance closure
errors as described in Section 4.2.8.
7. After collection of one-gallon permeate and concentrate samples at the first set of
operating conditions has been completed, operate the membrane system at the next set of
operating parameters and repeat steps 2 through 6 for each set of conditions. The
membrane does not need to be operated with laboratory clean water between sets of
operating conditions. Since the membrane has achieved stable performance during the
run at 70% recovery, the remaining three recoveries can all be evaluated in 24 to 48
hours (i.e., 8 to 16 hours for each of the three remaining runs). The permeate UV254 (or
TDS for softening membranes) should be monitored at 20 to 30 minute intervals over the
first 2 hours to insure stable permeate quality (less than a 2% change over 1 hour) before
collecting one-gallon permeate and composite concentrate samples; the system should
stabilize within one to two hours.
8. After the fourth experiment, clean the membrane according to the procedure described in
Section 4.1.5, and reevaluate the MTCW at the final recovery investigated (e.g., 30%)
using the test water. The data from this step indicates the relative amounts of reversible
and irreversible fouling. The data collected for the cleaned membrane should be reported
in Table 4-10.
The KBSMT does not need to be run continuously and can be shut-down during short
periods when it cannot be monitored; however, the membrane should remain sealed in the cell
during these intermittent shut-down periods. When possible, it is recommended that the test
be run continuously.
4.2.5 Monitoring And Sampling Requirements
Operating parameters must be monitored to assess membrane performance and the
parameters and recommended monitoring frequencies are summarized in Table 4-3. These
operating parameters include: permeate, waste and influent flow rates; influent, concentrate
and permeate pressures; and the influent temperature. These parameters should be monitored
approximately every four hours with more frequent monitoring at the beginning of the run
(i.e., during the first 12 hours) and less frequent monitoring at the end of the run.
In addition to these monitoring requirements, Table 4-3 also lists monitoring requirements
for TDS, pH and UV24S. The membrane permeate must be sampled for TDS, pH and UV254
approximately three times each day. The TDS and UV254 measurements are used to monitor
for stable permeate quality. As stated earlier, at least one of these parameters should be
evaluated in real tune to verify attainment of stable performance when collection of the one-
3-44
-------
gallon permeate sample should commence. These parameters should also be monitored for the
feed and concentrate, but less frequently (i.e., once each day).
Table 4-4 summarizes the water quality parameters that must be sampled during RBSMT
experiments. The analyses for the feed to the membrane system should be conducted on the
feed water after it has received appropriate membrane pretreatment and the results reported in
Table 4-8. The membrane feed should be sampled twice, once immediately before the
experiments and once at the end of the experiments, for the following parameters: alkalinity,
TDS, total and calcium hardness, bromide, pH, turbidity, temperature, TOG, UV254, and SDS
for THM4, HAA6, TOX, and chlorine demand. If enough feed water is generated to conduct
the experiments on both membranes, then the sampling requirements for the feed only need to
be met for the single batch (i.e., the feed water would only have to be analyzed for the water
quality parameters in Table 4-4 twice in one quarter). In this case, analytes should be
evaluated prior to the first experiment with the first membrane and after the final experiment
with the second membrane.
A one-gallon permeate sample should be collected for each run (i.e., at each recovery)
with each membrane for the following analyses: pH, alkalinity, TDS, turbidity, temperature,
total and calcium hardness, bromide, TOC, UV254, and SDS for THM4, HAA6, TOX, and
chlorine demand. Duplicate analyses are required for the permeate sample from the run at
70% recovery (run ID# 1 in Table 4-2) for each membrane. The analytical results for the
membrane permeate should be reported in Tables 4-11 to 4-15 for run IDs 1, ldupiicate, 2, 3 and
4, respectively.
A one-gallon concentrate sample should be collected for each run and analyzed for TOC,
UV254, TDS, turbidity, pH, alkalinity, total hardness and calcium hardness. Data from these
composite concentrate samples should be reported in Table 4-16.
Sampling, in terms of holding times, preservation, and sampling techniques, should be
conducted in accordance with the "ICR Sampling Manual" (EPA 814-B-96-001). Approved
methods for analysis are listed in Table 7, § 141.142 of the ICR Rule. The analyses for the
treatment studies must be conducted according to the analytical and quality control procedures
contained in the "DBP/ICR Analytical Methods Manual" (EPA 814-B-96-002).
4.2.6 Seasonal (Quarterly) Variation
In order to evaluate the performance of membranes treating surface waters under the range
of conditions anticipated over the course of a year, bench-scale membrane studies must be
performed quarterly. Furthermore, it is critical that each quarterly batch of treatment study
feed water be representative of the season being evaluated.
The two membranes selected for the study must be evaluated each quarter. The sample of
each membrane type obtained from manufacturers should be large enough to yield at least four
membrane sheets for use in the bench-scale membrane cell. Each quarterly study should be
conducted with a fresh membrane sheet from the same membrane sample. The use of a fresh
membrane insures that fouling from previous runs does not affect membrane productivity in
3-45
-------
following runs. The use of sheets from the same membrane sample minimizes the effects of
membrane variation over the four seasons. The characteristics of the membrane will not
change over a period of one year if the membranes are kept in cold storage and kept wet after
the initial wetting. (If the membrane sample is not received in a moist condition, then it can
be kept in dry storage until it is used.)
The variability of some source waters may be captured in fewer than four quarters. For
example, many ground waters do not exhibit significant seasonal variations. In general,
seasonal variation only needs to be evaluated when it is significant; however, four sets of
experiments evaluating at least two membranes must be performed to meet the ICR
requirements. These four sets of experiments can be used to evaluate any variables of interest
such as additional membranes, different pretreatments or different operating parameters. For
example, during the second quarter the experimental matrix in Table 4-2 could be repeated for
the same two membranes using different pretreatment. In a following quarter, two different
membranes could be evaluated according to the conditions listed in Table 4-2. Different
operational parameters could also be investigated. For example, the recovery could be held
constant and four fluxes or cross-flow velocities could be evaluated during a quarter.
Another option that may provide additional productivity data is a long-term RBSMT study.
These studies should be conducted for at least 480 hours with cleaning performed when the
MTC,y drops by a pre-determined percentage, such as 15 %. A utility electing to perform a
long-term bench study must evaluate two membrane at only one recovery, 70%, during that
quarter and sample the permeate for the analytes listed in Table 4-4 four tunes over the run for
each membrane. The analyses on one of the four permeate sample sets must be duplicated.
The feed must be sampled twice over each membrane run for the analytes listed in Table 4-4.
Also, monitoring of pH, TDS and UV^ can be reduced to once per day for the permeate and
once every other day for the feed and concentrate. However, flow rates, pressures and
temperatures should still be monitored six tunes per day.
If the water is not subject to seasonal variations, then the RBSMT experiments can be
performed in any convenient tune frame as long as they are completed within a year of the
starting date.
4.2.7 Datasheets
This section describes the data sheets that can be used to record the appropriate data from
the RBSMT procedure for the ICR. Corresponding data collection software will be sent to
utilities electing to conduct RBSMT experiments after the plant submits a study concept form
to EPA.
The data sheet in Table 4-5 is used to report the characteristics of each membrane used for
the ICR membrane studies. These are the membrane characteristics as reported by the
manufacturer, and this information will be required for the experimental design reported in
Table 4-6. Some of this information may not be available, but the data sheet should be filled
out as completely as possible. The area and cost of an 8" x 40" element are requested for use
in the cost analysis.
3-46
-------
The data sheet in Table 4-6 is used to develop and report the experimental design for each
membrane. The corresponding spreadsheet file uses input parameters to estimate the required
flow rates for each set of conditions. These flow rates are only intended to provide a starting
point and will need to be adjusted to achieve the desired simulation. The matrix in this data
sheet can be modified for the 2nd, 3rd, and 4th quarters if seasonal variation is not significant
as discussed in Section 4.2.6.
The data sheet in Table 4-7 requests information on the foulants in the feed water and the
pretreatment processes used to control fouling. The first section of this table requests
concentrations of foulants and fouling indices for the feed water. Only the foulants and indices
relevant to the water being investigated need to be measured and reported. The blank rows
should be used to report additional foulants that were measured but not listed in this table.
This information should be used as a guide to selecting appropriate pretreatment processes.
During the run, the relevant fouling indices and inorganic solutes should be periodically
measured to insure that pretreatment processes are performing properly and that the proper
chemical doses are being applied to the feed water.
The second half of this table requests information about the pretreatment processes used
prior to nanofiltration. All pretreatment processes used should be reported here including
processes in the existing full-scale treatment tram, upstream of the point where the feed to the
bench-scale system is collected; and processes that are added specifically for membrane
pretreatment. Existing full-scale treatment processes used as membrane pretreatment should
be marked with an "E", modifications to processes in the existing plant treatment train (e.g.,
an increase in the coagulant dose) should be indicated by an "M" and pretreatment processes
used in addition to the existing treatment train (e.g., acid or antiscalant addition) should be
indicated with an "A". This table can be used to provide some of the pretreatment
information required in the final treatment study report.
The data sheet in Table 4-8 is used to report the feed water quality after membrane
pretreatment. These water quality parameters must be evaluated twice for each batch of feed
water.
Table 4-9 contains the parameters that should be monitored and reported during setting
with laboratory clean water. Water fluxes at ambient temperature and normalized to the
average yearly water temperature experienced at the plant should be reported. The recovery
can be calculated by a spreadsheet; however, the recovery should also be manually calculated
during the run to insure that the desired simulation is achieved. The cumulative time should
be set at 0:00 at the start of setting, and reset at 0:00 at the start of operation with the test
water and continued over the four runs. Down time for the system must be subtracted from
the cumulative tune (i.e., by stopping the timer when the system is turned off).
Table 4-10 contains the parameters that should be monitored and reported during operation
with the test water. The time at which collection of the one-gallon permeate and composite
concentrate samples begins should be indicated in this table. Temperature normalized fluxes,
the MTCV, and both feed and bulk rejections will be calculated by the data collection software.
3-47
-------
The feed rejection is calculated from the feed and permeate concentrations; thus, the average
feed values from Table 4-8 must be entered hi the appropriate cells hi Table 4-10. To
calculate the bulk rejection, the permeate and feed concentrations and the recovery are used to
first estimate the bulk concentration which is then used to calculate the bulk rejection.
Spreadsheet calculations can also be used to determine the mass balance closure errors for all
sample sets that include permeate, feed and concentrate values.
The water quality parameters for the one-gallon permeate samples from runs 1, 2, 3 and 4
should be reported hi Tables 4-11, 4-13, 4-14 and 4-15, respectively. The analyses for run ID
1 should be duplicated and reported hi Table 4-12.
Results from the analyses of the one-gallon concentrate samples for all four runs should be
reported hi Table 4-16 along with corresponding composite permeate and average feed
concentrations. The concentrate concentrations can be calculated from the feed and permeate
concentrations and the recovery. The calculated and measured concentrate values are then
used to calculate the mass balance closure errors.
Since permeate quality often exceeds the treatment objectives, permeate can be blended
with by-passed feed water. This can substantially reduce the required membrane area and
post-treatment requirements and thus the cost. By calculating the water quality for different
blending ratios, costs for different levels of treatment can be estimated. Tables 4-17 (a and b)
use the average feed and the permeate water quality parameters evaluated at each recovery to
calculate the amount of flow that must be treated by membranes to achieve the Stage 1 MCLs
(Table 4-17a) and the proposed Stage 2 MCLs (Table 4-17b). Water quality parameters that
impact post-treatment requirements are also calculated for the blended waters. In these two
tables, QT is the total blended flow, and the ratio QP/QT is the fraction of flow that must be
treated by membranes to meet the treatment objectives. This ratio is calculated for both
THM4 and HAAS as either of these water quality parameters can control the blend ratio. The
subscript b hi these tables denotes blended water quality.
4.2.8 Interpretation Of The Results
The parameters measured during an RBSMT run need to be transformed into standard
parameters such as feed and bulk rejections and water mass transfer coefficients. These
calculated parameters can then be plotted to facilitate data interpretation. This section presents
an approach for plotting and interpreting the data generated hi these experiments along with
examples.
Calculating the temperature-normalized MTCW. The permeate flow rates, influent (feed)
and concentrate pressures and the influent temperature are used to calculate the temperature-
normalized MTC^r which is used as a standard measure of membrane productivity. The
permeate flux is calculated by dividing the permeate flow rate by the active area of membrane
hi the cell.
= QP / A(
cell
(4.10)
3-48
-------
Since water fluxes are sensitive to temperature, the flux values must be normalized to a
common temperature. For the purposes of developing an average cost estimate, the average
yearly water temperature experienced at the plant conducting the study will be used. Different
membranes use different temperature correction factors and equations, but an equation that
works well for most membranes is given below.
Fw(Tavg0C) = Fw(T°C)xl.03(ravg°-T°) (4.11)
where Fw(Tavg°C) is the flux corrected to the average yearly water temperature experienced
at the plant conducting the study, FW(T°C) is the flux measured at ambient temperature, T° is
the temperature at which the flux was measured in °C, and Tavg° is the average yearly water
temperature experienced at the plant conducting the study in °C.
In order to determine the net driving pressure (NDP) that will be used to calculate the
MTCW, the osmotic pressure gradient must first be estimated from the influent, concentrate
and permeate TDS values using Equation 4. 12. An osmotic pressure gradient must be
calculated for each recovery evaluated since the influent and permeate TDS concentrations will
increase with increasing recovery.
ATI = ([(TDSj+TDScVll-TDSpJX (0.01 psi/mg/L) (4.12)
where ATI is the osmotic pressure gradient in psi, TDSj is the influent TDS concentration in
mg/L, TDSC is the concentrate TDS concentration in mg/L and TDSp is the permeate TDS
concentration in mg/L.
The influent TDS concentration is calculated from a mass balance using measured feed and
concentrate TDS concentrations according to Equation 4.13.
TDS, = (QF X TDSF + (Q! - QF) X TDSc) / Qx (4.13)
where QF is the feed flow rate, TDSF is the feed TDS concentration hi mg/L and Qx is the
influent flow rate.
Once the osmotic pressure gradient has been estimated, the net driving pressure is
calculated according to Equation 4.14.
NDP = [(P^Pc^l-Pp-Arc (4.14)
where NDP is the net driving pressure, Pt is the influent pressure, Pc is the concentrate
pressure and Pp is the permeate pressure which will usually equal zero in RBSMT
experiments.
The temperature-normalized MTCW is determined by dividing the temperature-normalized-
flux by the net driving pressure according to Equation 4.15.
3-49
-------
MTCw(Tavg0C) =
/ NDP
(4.15)
where MTCw(Tavg0C) is the temperature-normalized MTC
'W
MTCW as a Junction of time. After the temperature-normalized MTCV has been calculated
for each data set, the data can be plotted as a function of operating tune as shown in Figure 4-
6 for a Fluid Systems TFCS membrane treating Ohio River water. During setting, the MTCV
was observed to increase slightly over the first few hours after which it leveled out to a value
of 0.219 gfd/psi, shown as point a hi Figure 4-6. This value is within 12% of the
manufacturer's reported value of 0.193 gfd/psi. If the MTCV at the end of setting is outside of
an acceptable range reported by the manufacturer, it would be prudent to obtain another
membrane sample for testing.
In this case, the MTCV increased slightly during setting; however, other membranes have
exhibited a sharp decline hi the MTCW during setting. Whether a membrane exhibits an
increase or a decrease hi the MTCV during setting is not important. The purpose of setting is
to obtain a stable MTCw with clean water for comparison with the manufacturer's value and to
serve as a baseline for the assessment of flux loss.
The MTCV curve during operation with the test water can be divided into two sections, a
rapid initial MTCV decline followed by a more moderate MTCV decline. The inflection point
where the slope of the MTCW curve rapidly changes (point b hi Figure 4-6) can serve as a
baseline for membrane performance treating a specific test water. In Figure 4-6, the value of
this baseline MTCV is 0.211 gfd/psi. The second portion of the MTCV curve can be
approximated as a line, and the slope of this line can be used as an estimate of the rate of
MTCV decline. This rate of MTCV decline can be used to estimate the required cleaning
frequency according to Equation 4.16.
CF =
_QxMTCy(o)
dMTCV/dt
(4.16)
where CF is the cleaning frequency, Q is the acceptable fractional loss in the MTCV Pri°r to
cleaning (e.g., 0.15), MTCW(0) is the baseline MTCV during operation with the test water, and
dMTCVdt is the rate of MTCV decline determined from a linear regression of the flux data.
If the r2 value of the regression line is less than approximately 0.90, the fit may not accurately
predict the rate of MTCV decline. In this case other models can be used to predict the rate of
MTCV decline, such as a linear regression for a plot of the log of tune versus the log of the
MTCV
For this experiment, dMTCydt was -0.0023 gfd/psi/day, and using 0.15 for Q, the
cleaning frequency is calculated to be 14 days. This cleaning frequency is excessive, but it is
likely that many surface waters will exhibit high fouling rates and require frequent cleaning
without advanced pretreatment. This calculation also assumes a constant rate of flux decline
which may not be a valid assumption over the life of a membrane.
3-50
-------
After 100 hours of operation, the membrane was chemically cleaned first with a solution of
HC1 at pH = 2.0 followed by a solution of NaOH at pH = 10.5. The MTCW after cleaning is
indicated as point c in Figure 4-6. Immediately prior to cleaning, the MTC^had declined 5%
from the baseline MTCW, and cleaning restored 50% of the flux lost relative to the baseline
The long-term flux study with the FilmTec NF-90 membrane treating Ohio River water
shown in Figure 4-7 shows the continued fouling experienced by a membrane treating a
conventionally treated surface water. It also demonstrates that the rate of MTCW decline can
decrease with tune, and long-term studies may provide a better estimate of the rate of MTCW
decline.
At this time, the rate of fouling and the cleaning frequency determined according to the
RBSMT has not been verified with sufficient pilot-scale data. However, this work is in
progress, and a significant amount of verification data will be amassed and published prior to
commencement of the treatment studies. Preliminary verification data indicates that the
baseline MTCW and rate of MTCW decline obtained using the RBSMT may be reasonable
estimates of single element or pilot-scale performance. Until the relationship between single
element and bench-scale MTCW data has been established, the fouling rate and cleaning
frequency must only be used on a relative basis to compare different bench-scale studies and
should not be inferred to predict single element performance.
Membrane characterization curves. Under most operating conditions used in practice, the
permeate flux is directly proportional to the net driving pressure, and the proportionality
constant is the MTCW. However, it is possible for this relationship to fail under extreme
conditions. At very high net driving pressures, a limiting flux is approached and the linear
relationship between flux and the net driving pressure is no longer valid.
One method for verifying the validity of this relationship is to conduct a membrane
characterization study. In a characterization study, the flux is evaluated at several pressures
using the test water, recovery and cross-flow velocity under investigation. The pressures must
encompass the operating pressure and cover a wide range (e.g., 40 psi to 95 psi). The
permeate flux plotted as a function of net driving pressure should exhibit a linear relationship
passing through the origin. The slope of this line is the MTCW, and if the relationship is linear
over the entire range, than the MTCW can be calculated by dividing the flux by NDP.
The four characterization curves shown hi Figure 4-8 demonstrate a linear relationship.
The membrane was characterized when it was new, after setting, after it was fouled and after
it was cleaned. In every case the linear relationship between flux and NDP was shown to be
valid as will be the case under most reasonable operating conditions.
Rejection as a junction of time. Permeate UV^ and TDS are monitored as a function of
time to insure stable performance before permeate and concentrate samples are collected for a
complete water quality analysis. Using the average feed water quality, the feed rejection of
UV254 and TDS can be calculated and plotted as a function of tune of operation with the test
3-51
-------
water. An example of this type of data is shown in Figure 4-9 (in this figure conductivity
rejection is plotted instead of TDS rejection).
For this experiment, the rejection of TOC, UV254 and conductivity increased with time due
to an increase hi the mass transfer resistance of the system. After 34 hours, the change in
permeate conductivity was less than 3% per 10 hours; thus, collection of one-gallon permeate
and concentrate samples could commence at this point. However, the first run, at 70%
recovery, should be continued for at least 78 hours to insure stable performance and to obtain
a more accurate slope from the MTCW curve.
For this softening membrane, the temporal rejection trends were similar for all three
parameters, and real tune measurements of any of these parameters could have been used to
monitor for stable performance. However, TDS rejection may not follow TOC and UV254
rejection for some membranes. In these cases, either UV^ or TOC rejection must be used to
monitor system performance.
Rejection as a function of recovery. For each set of experiments with each membrane,
permeate quality is evaluated at four recoveries. The average feed water quality parameters
can be used to calculate the feed and bulk rejections for each recovery, and these rejections
can be plotted as a function of recovery as shown in Figure 4-10.
In this figure, only TOC, UV^ and conductivity rejections are plotted. The feed rejection
of all three parameters decreased with increasing recovery demonstrating that permeate quality
can deteriorate at higher recoveries (as could be experienced in the down-stream stages of a
full-scale plant), fc cases where the solute is rejected on the basis of size (i.e., sieving), the
permeate quality will be independent of recovery. This has been shown to be the case for
TOC in some natural waters treated by softening membranes. However, the impact of
recovery on membrane performance will be a function of solute and membrane characteristics
and must be evaluated on a site specific basis.
Increasing the recovery in a recycle system will increase the bulk concentration in the
membrane feed channel. The bulk rejection normalizes for the effect of recovery since the
bulk concentration is used to calculate the rejection. In Figure 4-10, the bulk rejection of all
three parameters remained relatively constant with recovery. This indicates that the membrane
was performing consistently, and the decrease hi permeate quality was due to the increase in
the bulk concentration.
Rejection summary chart. Figure 4-11 is a bar chart showing the feed rejection of several
water quality parameters. The feed rejection at 50% recovery is used since this is a reasonable
estimate of the average system recovery for a full-scale plant. These charts present a quick
summary of membrane performance and can be used to make rough comparisons; however,
summary tables are more useful for assessing detailed membrane performance to determine if
the treatment objectives have been met.
3-52
-------
Mass balance closure errors. Any data set that contains values for feed, permeate, and
concentrate water quality parameters can be used to check the mass balance on the system.
The mass balance closure error can be used to assess sampling and analytical techniques.
Mass balance closure errors on common parameters (e.g., TOC, UV254, TDS, alkalinity,
hardness, etc.) are typically less than a few percent.
The mass balance closure error is calculated in two steps. First the concentrate
concentration is calculated from the feed and permeate concentrations and the fractional
recovery using Equation 4.17.
'C(calc)
CF-(CpxR)
(1-R)
(4.17)
where CC(calc) is the calculated concentrate concentration. Next the mass balance closure error
is determined by comparing calculated and measured concentrate values:
Err0rMB =
C
c(calc)
C(meas)
(4.18)
where ErrorMB is the mass balance closure error expressed as a percentage and CC(meas) is the
measured concentrate concentration.
Permeate/feed blending. When the permeate quality exceeds the treatment objectives, feed
can be blended with permeate to reduce the required membrane area and possibly minimize
post-treatment requirements. The required permeate flow to total flow ratio (QP/QT) can be
calculated from the feed and permeate concentrations and the treatment goal. These
calculations can be performed automatically in spreadsheets for Tables 4-17a and 4-17b for the
Stage 1 and proposed Stage 2 DBF MCLs, respectively. These calculations are only valid if
the permeate DBFs are below 90% of the THM4 / HAAS MCLs (i.e., 72 / 54 for Stage 1 and
36 / 27 for Stage 2). Since either THM4 or HAAS can control, the user must compare the
Qp/QT flow ratios for each case and choose the higher of the two ratios as the controlling case.
The blended concentrations of other water quality parameters such as alkalinity and hardness
can also be calculated % a spreadsheet.
The concentration of various water quality parameters can be plotted as a function of the
Qp/QT flow ratio to show the range of water qualities obtainable through blending.
3-53
-------
-------
QR Cc
QF
Membrane
ATI
Qw
Cc C<
QP
Pn
Figure 4-1 Definition Sketch Of A Tangential-Flow Membrane-Cell With Recycle
-------
Waste
1-7-1 Compressed
Gas
Concentrate
Permeate
Water
Tank
Feed
LEGEND
Needle
Valve
Temperature
Gage
Check
Valve
Recycle
Pump
ifIc item described in the text
Pressure
Gage
Flowmeter
Figure 4-2 Bench-Scale Tangential-Flow Membrane System With Recycle
-------
(3*' Permeate Outlet
Cell-Body Top
Permeate Carrier
Membrane
"O" Rings
Cell-Body Bottom
Concentrate Outlet
Feed Inlet
Figure 4-3 Schematic Of The Tangential-Flow Flat-Sheet Cell Used In The RBSMT
-------
Cleaning
solution
reservoir
Concentrate line
Membrane cell
Figure 4-4 Set-Up For Membrane Cleaning Procedure
-------
1
Select two membranes
Collect influent to the treatment
study
4
Pretreat the influent to control
membrane fouling
Analyze influent
sample for seasonal
representativeness
Analyze the pretreated influent for
the ICR requirements
Conduct experiemnts with the first
membrane
16
Conduct experiments with the
second membrane
8
ICR sample
10
ICR sample
12
ICR sample
14
ICR sample
15
Clean membrane I
18
ICR sample
20
ICR sample
22
ICR sample
24
ICR sample
25
Clean membrane
Figure 4-5 Flow Chart For The Quarterly RBSMT Membrane Studies
-------
Membrane setting i
0.30
0.25 -
ts- 0.20
8.
s
«s
0.15 -
0.10 -
0.05 -
0.00
Operation with Ohio River Water
-0.0023
gfd/psi/day
Membrane chemically cleaned
R = 40%
P = 83 psi ~
vc = 0.33 fps
10 20 30 40 50 60 70 80 90 100 110
Time (hours)
Figure 4-6 MTC^as a function of time for a Fluid Systems TFCS membrane
treating sand filtered Ohio River water from the Cincinnati Water Works
Membrane setting |
0.5
w
t
O)
Operation with Ohio River Water
0.4 -
0.3 -
0.2-
0.1 -
0.0
-0.0175
gfd/psi/day
-0.0130
gfd/psi/day
Membrane chemically cleaned
R = 20% "
P = 86 psi
vc = 0.60 fps
50
I
100
I
150
I
200
250
I
300
350
Time (hours)
Figure 4-7 Long-term MTCW study for a FilmTec NF-90 membrane treating
sand filtered Ohio River water from the Cincinnati Water Works
-------
a. New membrane
MTCIV= 0.365 gfd/psi
b. After setting
0.367 gfd/psi
End of run
MTC,,,= 0.259 gfd/psi
After cleaning
MTC..,= 0.327 gfd/psi
10
40 50 60
Pressure (psi)
90
100
Figure 4-8 Membrane characterization curves for the FilmTec NF-90 membrane
treating sand filtered Ohio River water from the Cincinnati Water Works
100
c
.g
t3
-------
vO
C
2
.2,
5T
CC
90 -
80 -
70 -
60 -
.
50 -
40 -
.
30 -
20 -
10 -
o
^=======SP====Q ° _i
* " __JlT^r=- =^r& _^
^ ^^=»
* -^ -
>* J
-
-
-
BB BF
o TOC
D uv254
A A Conductivity
1 I ' I ' I ' I ' I ' I ' I ' I ' I '
0 10 20 30 40 50 60 70 80 90 1C
Recovery (%)
Figure 4-10 Bulk and feed rejections of TOC, UV254 and conductivity as a function
of recovery for an NF-90 treating sand filtered Ohio River water
(D
100
90 -
80 -
70 -
60 -
50 -
40 -
30 -
20 -
10 -
0
10
1
O
T3
>.
O
Q
m
Ji
£
o
O
Figure 4-11 Rejections summary for an NF-90 membrane treating sand filtered
Ohio River water
-------
Table 4-1 Definitions Used In The RBSMT Procedure
Symbol
cell
wcell
Acell
T
w
eiement
V
Q
CF
Definition
Active length of membrane in the cell
Active width of membrane in the cell
Active area of membrane in the cell
Spacer thickness
Design cross-flow velocity (minimum)
Minimum influent flow rate to the cell to achieve vc.design
Length of one membrane envelope in a full-scale element
Width of all envelopes hi a full-scale element (total scroll width)
Minimum influent flow rate to a full-scale membrane element
Estimate of the volume requirements for a RBSMT study
Acceptable fractional loss hi the MTCW prior to cleaning
Cleaning frequency
Table 4-2 Experimental Matrix For The RBSMT-ICR Membrane Studies
ID#
1
2
3
4
Simulation
Conservative system average
Final stage of a full-scale NF plant
System average for full-scale plant
First stage of a full-scale NF plant
Recovery
70%
90%
50%
30%
Cross-flow
velocity
DV
DV
DV
DV
Pressure
( or Flux)
DV
DV
DV
DV
DV: Design values obtained from the membrane manufacturer (see Table 4-5)
-------
Table 4-3 Recommended Minimum Monitoring Frequencies For The RBSMT
Routine RBSMT Study Monitoring Requirements
Parameter
Flow
Pressure
Temperature
TDS
PH
UV254
Feed
none
none
none
IxD
IxD
IxD
Permeate
6xD
none
none
3xD
3xD
3xD
Concentrate
6xD
6xD
none
IxD
IxD
IxD
Influent
6xD
6xD
6xD
none
none
none
Recycle
none
none
none
none
none
none
IxD - once per 24 hours
3xD - three time per 24 hours
6xD - six times per 24 hours
Table 4-4 RBSMT Water Quality Monitoring Requirements
Water Quality Parameters To Be Evaluated At Each Recovery
Parameter
pH
Total Hardness
Calcium Hardness
Alkalinity
Total Dissolved Solids
Turbidity
Total Organic Carbon
UV254
Bromide
SDS - THM4
SDS-HAA6
SDS - TOX
SDS - C12 demand
Feed
TPR
TPR
TPR
TPR
TPR
TPR
TPR
TPR
TPR
TPR
TPR
TPR
TPR
Permeate
FTPR
FTPR
FTPR
FTPR
FTPR
FTPR
FTPR
FTPR
FTPR
FTPR
FTPR
FTPR
FTPR
Concentrate
FTPR
FTPR
FTPR
FTPR
FTPR
FTPR
FTPR
FTPR
none
none
none
none
none
TPR - twice per run.
FTPR - five times'per run (i.e., once at each recovery and one duplicate).
-------
Table 4-5 Membrane Characteristics As Reported By The Manufacturer
Utility name and address
ICR plant number
Phone number
Contact person
FAX number
Characteristics for a standard 8" x 40" element
Membrane manufacturer
Membrane module model number
Active membrane area of an equivalent 8" x 40" element
Purchase price for an equivalent 8" x 40 " element ($)
Molecular weight cutoff (Daltons)
Membrane material / construction
Membrane hydrophobicity (circle one)
Membrane charge (circle one)
Design pressure (psi)
Design flux at the design pressure (gfd)
Variability of design flux (%)
MTCV (gfd/psi)
Standard testing recovery (%)
Standard testing pH
Standard testing temperature (°C)
Design cross-flow velocity (fbs)
Maximum flow rate to the element (gpm)
Minimum flow rate to the element (gpm)
Required feed flow to permeate flow rate ratio
Maximum element recovery (%)
Rejection of reference solute and conditions
of test (e.g. solute type and concentration)
Variability of rejection of reference solute (%)
Spacer thickness (ft)
Scroll width (ft)
Acceptable range of operating pressures
Acceptable range of operating pH values
Typical pressure drop across a single element
Maximum permissible SDI
Maximum permissible turbidity (ntu)
Chlorine/oxidant tolerance
Suggested cleaning procedures
Hydrophilic Hydrophobic
Negative Neutral Positive
Note: Some of this information may not be available, but this table should be filled out as completely
as possible for each membrane tested.
-------
J>
|
5
[j
^
i O
-J m*
03
H
s
L-, O
3 xUl'
3 eo
S 'g
Ja> S <3 T3 a
2^-29 fc
8 5 B 2
o 3 |L 3 o n
ca c ir a S U
ro " CJ o CO
" -2 a 2 gj
a> CX o .Q -a
£» a §
3 <3 -|
H^ S
"S
cu
«
1 Membrane Elei
mbrane area (ft2)
0 §
TT >=<
>>
^
cu
1
O
^
a
s
Q
i
55
o>
'ca
o
55
I
cn
w
S
u
p
/"N
H
^
o
V3
§ S
a .1
Minimum flow rate
ale System Character
U
S
g
i
1
i
60
'1
&
5-
g
"5
o>
S
s~
43
i
1
0>
1
s
08
.£?
S
ca
4^
a
a
'C
cu
B.
H
-S
:a|
/ N
S
S1
-i
*s
^s
> c^
5
?f'i
' \\1-'-
* V
WP
w
' :
s
-
;
f;;
-
-
0
o\
CN
'
f
^'" *,
'^4
*
'
0
<«>
: -
; -
U
U
o
en
-
Notes: + The values in the shaded cells can be calculated by a spreadsheet.
* For experunental conditions see Table 4-2.
** The flow rates calculated in the above table are approximate values intended to provide
a startuig point for the experiments. The waste and feed flow rates will need to be
adjusted to obtain the target recovery based on the actual permeate flow rate.
-------
Table 4-7 Membrane Pretreatment Data
Utility name and address
ICR plant number Phone #
Contact person FAX #
Membrane trade name
Foulants and fouling indices of the feed water prior to pretreatment1
Alkalinity (mg CaCO3/L)
Ca Hardness (mg CaCO3/L)
LSI
Dissolved iron (mg/L)
Total iron (mg/L)
Dissolved aluminum (mg/L)
Total aluminum (mg/L)
Fluoride (mg/L)
Phosphate (mg/L)
Sulfate (mg/L)
Calcium (mg/L)
Barium (mg/L)
Strontium (mg/L)
Reactive silica (mg/L as SiO2)
Turbidity (ntu)
SDI
MFI
MPFI
1: Only those foulants and fouling indices relevant to the water being tested need to be evaluated.
Additional foulants and indices can be listed in the blank rows or on an attached sheet.
Pretreatment processes used prior to nanofiltration or reverse osmosis2
Pre-filter exclusion size (^im)
Type of acid used
Acid concentration (units)
mL of acid per L of feed
Type of antiscalant used
Antiscalant concentration (units)
mL of antiscalant per L of feed
Type of coagulant used
Coagulant dose (mg/L)
Type of polymer used during coag.
Polymer dose (mg/L)
2: Use an "E" to indicate a pretreatment process that is currently part of the plant treatment train, an "M" to
indicate a modification to a process that is currently part of the plant treatment train, and an "A" to
indicate an addition to the current treatment train.
Additional pretreatment processes, such as MF, can be listed in the blank rows or on an attached sheet.
-------
Table 4-8 Membrane Feed Water Quality (After Membrane Pretreatment)
Utility name and address
ICR plant number
Phone number
Membrane trade name
Contact person
FAX number
Water quality parameter | Units
Sampling date
Sampling time
Alkalinity
Total dissolved solids
Total hardness
Calcium hardness
Bromide
pH
Turbidity
Temperature
Total organic carbon
UV254
MM/DD/YY
hh:mm
mg/L as CaCO3
mg/L
mg/L as CaCO3
mg/L as CaCO3
ug/L
ntu
°C
mg/L
cm"1
1st sample
2nd sample
Average
RPD
SDS samples
Chlorine dose
Chlorine residual
Chlorine demand
Chlorination temperature
Chlorination pH
Incubation time
TOX
CHC13
CHBrCl2
CHBr2Cl
CHBr3
THM4
MCAA'
DCAA'
TCAA'
MBAA'
DBAA'
BCAA'
TBAA
CDBAA
DCBAA
HAA6
mg/L
mg/L
mg/L
°C
hours
ug/L
ug/L
ug/L
ug/L
ug/L
ug/L
ug/L
ug/L
ug/L
Ug/L
ug/L
ug/L
ug/L
ug/L
ug/L
ug/L
Note: RPD is the relative percent difference between the two samples.
* These six species make up HAA6.
The other three HAA species should be reported if measured.
-------
I-
P-I
£
B
2
o
I
I
en
s
1
CO
e
|
O
S
5
o
ja
s
T3
s
v
£
>
<
?
s
.g
o
ra
x>
i
CJ M
Q>
&
"8
1
a
I
G
'S
a
B
2
y
a
en
c
s
o
u
J
A
itored wit
B
0
S
i
s -8 1
IS 2
g. o «
(V "3
y -| §
^ 2
If
H
(D
is
3
B
(5
Q
!-
-
,
"
.,
"
*
'
~
.
'
-
-
-
,
,,.
-
'
-
cs
CN
i^
u cd
££
^ S
Js
t.l
L) rr(
1 -a
5 -0
anbecal
is define
O (t)
. o
§ "
M Iz;
i
-------
Table 4-10 Membrane Performance Data Monitored During Operation With The Pretreated Test Water (page 1)
Utility name and address
ICR plant number
Membrane trade name
Phone number
FAX number
Contact person
Operational parameters monitored with time during operation with the test water
Date
Time
(hr)
Cumulative
time
(hr)
PI
(psi)
PC
(psi)
Influent
Temp.
CO
Qi
(mL/min)
Qp
(mL/min)
Qw
(mL/min)
QF
(mL/min)
'
'--
ff
s s
ff f f
,
>
Recovery
(decimal)
ff f f f
, #
<
f f ffftt ff f
fff "''
"
ff
f
f
' f f
s
f f
Notes: Shaded cells can be calculated by a spreadsheet, but should also be manually calculated during the run.
Nomenclature is defined in Figure 2-1 and Table 2-2.
-------
Table 4-10 Membrane performance data monitored during operation with the pretreated test water (page 2)
Utility name & address
ICR plant number
Membrane trade name
Phone number
FAX number
Contact person
Water quality parameters monitored with time during operation with the test water
Date
Time
(hr)
Cumulative
time
(hr)
PH
Feed
CF
Permeate
CD
Concentrate
cc
TDS(mg/L)
Feed
CF
Permeate
CD
Concentrate
Cc
Notes: Nomenclature is defined in Figure 2-1 and Table 2-2.
-------
Table 4-10 Membrane performance data monitored during operation with the pretreated test water (page 3)
Utility name & address Phone number
FAX number
ICR plant number
Membrane trade name
Contact person
Water quality parameters monitored with time during operation with the test water
Date
Time
On)
Cumulative
time
(hr)
UV254(cm-1)
Feed
CF
Permeate
Cp
Concentrate
Cc
TOC (mg/L), optional analysis
Feed
CF
Permeate
Cp
Concentrate
Cc
Notes: Nomenclature is defined in Figure 2-1 and Table 2-2.
-------
Table 4-11 Membrane Permeate Water Quality For Run ED# 1
Utility name and address
ICR plant number
Contact person
Membrane trade name
Qp (mL/min)[
Phone #
FAX#
; ' >
| |QF(mL/min)| ((^(mL/min)! |
Water quality parameter
Sampling date
Sampling time
Alkalinity
Total dissolved solids
Total hardness
Calcium hardness
Bromide
PH
Turbidity
Temperature
Total organic carbon
UV254
Units
MM/DD/YY
hh:mm
mg/L as CaCO3
mg/L
mg/L as CaCO3
mg/L as CaCO3
ug/L
ntu
°C
mg/L
cm"
ft
*^F(averaee)
CD
RF
« , , , ,
-
V.
-.-.
V.,
s'ss
RB
..
"'"
"
'
'
--
SDS samples
Chlorine dose
Chlorine residual
Chlorine demand
Chlorination temperature
Chlorination pH
Incubation time
TOX
CHC13
CHBrCl2
CHBr2Cl
CHBr3
THM4
MCAA*
DCAA*
TCAA*
MBAA"
DBAA"
BCAA*
TBAA
CDBAA
DCBAA
HAA6
mg/L
mg/L
mg/L
°C
hours
ug/L
ug/L
ug/L
ug/L
Hg/L
ug/L
ug/L
ug/L
ug/L
ug/L
ug/L
ug/L
ug/L
ug/L
ug/L
ug/L
'"'',''
*
V
'
-
-
'"'
~ ,
. ,%
^
/ , / f "
, -
s
'
>
^
/
* s
%%
* These six species make up HAA6. The other three HAA species should be reported if measured.
Shaded cells can be calculated by a spreadsheet.
-------
Table 4-12 Duplicate Analysis Of The Membrane Permeate Water Quality For Run ID# 1
Utility name and address
ICR plant number
Contact person
Membrane trade name
Qp (mL/min)[
Phone #
FAX#
| |QF(mL/min)| iQjCmL/min)! |
Water quality parameter
Sampling date
Sampling time
Alkalinity
Total dissolved solids
Total hardness
Calcium hardness
Bromide
pH
Turbidity
Temperature
Total organic carbon
UV254
Units
MM/DD/YY
hh:mm
mg/L as CaCO3
mg/L
mg/L as CaCO3
mg/L as CaCO3
ug/L
ntu
°C
mg/L
cm"1
*--F(average)
CD
RF
< *
,
'
f ff fff f ff
RB
/ '^
"""
V
$*:
SDS samples
Chlorine dose
Chlorine residual
Chlorine demand
Chlorination temperature
Chlorination pH
Incubation time
TOX
CHC13
CHBrCl2
CHBr2Cl
CHBr3
THM4
MCAA"
DCAA'
TCAA'
MBAA'
DBAA'
BCAA'
TBAA
CDBAA
DCBAA
HAA6
mg/L
mg/L
mg/L
°C
hours
ug/L
ug/L
ug/L
ug/L
ug/L
ug/L
ug/L
ug/L
ug/L
ug/L
ug/L
ug/L
ug/L
ug/L
ug/L
ug/L
., *
f f
'
f
'
"
'
* These six species make up HAA6. The other three HAA species should be reported if measured.
Shaded cells can be calculated by a spreadsheet.
-------
Table 4-13 Membrane Permeate Water Quality For Run ID# 2
Utility name and address
ICR plant number
Contact person
Membrane trade name
Qp (mL/min)[
Phone #
FAX#
|QF (mL/min) |
|Qj (mL/min) |
Water quality parameter
Sampling date
Sampling time
Alkalinity
Total dissolved solids
Total hardness
Calcium hardness
Bromide
PH
Turbidity
Temperature
Total organic carbon
UV254
Units
MM/DD/YY
hh:mm
mg/L as CaCO3
mg/L
mg/LasCaCO3
mg/L as CaCO3
ug/L
ntu
°C
mg/L
cm"1
/-<
*-'F(averaee>
CD
RF
.
'
: %
s
RB
-
_, SSS
"
f
-
SDS samples
Chlorine dose
Chlorine residual
Chlorine demand
Chlorination temperature
Chlorination pH
Incubation time
TOX
CHC13
CHBrCl2
CHBr3Cl
CHBr3
THM4
MCAA"
DCAA*
TCAA*
MBAA"
DBAA*
BCAA*
TBAA
CDBAA
DCBAA
HAA6
mg/L
mg/L
mg/L
°C
hours
ug/L
ug/L
ug/L
ug/L
ug/L
ug/L
ug/L
ug/L
ug/L
ug/L
ug/L
ug/L
ug/L
ug/L
ug/L
ug/L
"
..
.
s
"
..
%
-
--
ss
^
-
-
* These six species make up HAA6. The other three HAA species should be reported if measured.
Shaded cells can be calculated by a spreadsheet.
-------
Table 4-14 Membrane Permeate Water Quality For Run ID# 3
Utility name and address __.
ICR plant number
Contact person
Membrane trade name
Qp (mL/min)[
Phone #
FAX#"
|QF(mL/min)
|QI(mL/min)|
Water quality parameter
Sampling date
Sampling time
Alkalinity
Total dissolved solids
Total hardness
Calcium hardness
Bromide
PH
Turbidity
Temperature
Total organic carbon
UV254
Units
MM/DD/YY
hh:mm
mg/LasCaCO3
mg/L
mg/LasCaCO3
mg/L as CaCO3
Fg/L,
-------
Table 4-15 Membrane Permeate Water Quality For Run BD# 4
Utility name and address
ICR plant number
Contact person
Membrane trade name
Qp (mL/min)L
Phone #
FAX#"
|QF(mL/min)|
|Q,(mL/min)|
Water quality parameter
Sampling date
Sampling time
Alkalinity
Total dissolved solids
Total hardness
Calcium hardness
Bromide
PH
Turbidity
Temperature
Total organic carbon
UV254
Units
MM/DD/YY
hh:mm
mg/L as CaCO3
mg/L
mg/L as CaCO3
mg/L as CaCO3
ug/L
ntu
°C
mg/L
cm"1
/-i
*-^F{average)
cp
^-...-
RF
«,
««
,«
%
-
s
RB
.
v...
-
.^
SDS samples
Chlorine dose
Chlorine residual
Chlorine demand
Chlorination temperature
Chlorination pH
Incubation time
TOX
CHC13
CHBrCl2
CHBr2Cl
CHBr3
THM4
MCAA*
DCAA*
TCAA"
MBAA*
DBAA"
BCAA*
TBAA
CDBAA
DCBAA
HAA6
mg/L
mg/L
mg/L
°C
hours
u-g/L
ug/L
Rg/L
jig/L
Mg/L
Mg/L
ug/L
ug/L
ug/L
ug/L
ug/L
Mg/L
ug/L
jig/L-
ug/L
^g/L
s%% V
^ %
1 1 1"! i il i 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1
; »
-
f f
f
s ,
" "
"" " "
ts VlS ""
s^'
,
>,, *
r nir "\
::,
s-.
.«
S
'
'
* These six species make up HAA6. The other three HAA species should be reported if measured.
Shaded cells can be calculated by a spreadsheet.
-------
I
Table 4-16 Composite Concentrate Water Quality Parameters And Mass Balance Closure Errors
Utility name and address
ICR plant number Contact person
Phone number
FAX number
Membrane trade name
Composite sample for run ID#1
Parameter
PH
Turbidity
TDS
TOG
UV^
Alkalinity
Total hardness
Calcium hardness
Units
ntu
mg/L
mg/L
cm"1
mg CaCO3/L
mg CaCO3/L
mg CaCO3/L
Date Time Recovery (decimal))
fi
*-F(avcrage)
Cp
Cc(meas)
1 Ccfcafc)
' '
* f '
^^^TM^)
..
..'
'
; % " %-
: , ,
Composite sample for run ID#2 Date Time Recovery (decimal)
Parameter
pH
Turbidity
TDS
TOC
UV^
Alkalinity
Total hardness
Calcium hardness
Units
ntu
mg/L
mg/L
cm"1
mgCaCO3/L
mg CaCO3/L
mg CaCO3/L
^Ffaveraee)
Cp
Cc(meas)
Ccfcalc)
"
' ,, ,' , '""*&
:
^rrorM|(%)
: v. ^ -,
-
: ::V, s
"
Composite sample for run BD#3 Date Time Recovery (decimal)
Parameter
pH
Turbidity
TDS
TOC
UVa*
Alkalinity
Total hardness
Calcium hardness
Units
ntu
mg/L
mg/L
cm"1
mg CaCO3/L
mg CaCO3/L
mg CaCO3/L
fi
M?(average)
Cp
Cctmeas)
Ccfcalc)
* :
/ '
ErrorM^%)
''; ,
' <
..
Composite sample for run ID#4 Date Time Recovery (decimal)
Parameter
PH
Turbidity
TDS
TOC
UV^
Alkalinity
Total hardness
Calcium hardness
Units
ntu
mg/L
mg/L
cm"1
mg CaCO3/L
mg CaCO3/L
mgCaCO3/L
*"F(avcrage)
Cp
Cc(meas)
^^^2j^l
-
f
;
>
- --J
^^^«2^
, ,
> f ft f
-------
Table 4-17a Blending Calculations To Determine The Fraction Of Flow That Requires NF
Treatment To Meet The Stage I DBF Regulations
Utility name and address
ICR plant number
Membrane trade name
THM4 Controls
HAAS Controls
Phone number
FAX number
Contact person
Parameter
THM4F, ng/L
THM4p, ng/L
HAA5F, |j.g/L
HAA5P, ug/L
AlkF, mg/L CaCO3
Alkp, mg/L CaCO3
T-HdF, mg/L CaCO3
T-Hdp, mg/L CaCO3
Ca-HdF, mg/L CaCO3
Ca-Hdp, mg/L CaCO3
RUN ID #
1
2
3
4
Qp/QT(THM4),%
Alkb, mg/L CaCO3
T-Hdb, mg/L CaCO3
Ca-Hdb, mg/L CaCO3
THM4b, ng/L
HAA5b, ug/L
"
.
% s
-
vws Cs :
;
,
- -
%
'
, , :
' .,. ,
,
f ;
,, \
' ,
" :
" ' !
Qp/QT(HAA5),%
Alkb, mg/L CaCO3
T-Hdb, mg/L CaCO3
Ca-Hdb, mg/L CaCOS
THM4b, ug/L
HAA5b, ug/L
^
~
-
^
%
-. f f
s %
""
-'
:
% ;
i
Notes: In the first section of this table, the feed and permeate concentrations are entered for each run.
The spreadsheet uses these values to calculate the flow that must be treated to meet the stage I
DBF MCLs with a 10% factor of safety (i.e.72/54 ug/L - THM4/HAA5). The higher of the two
Qp/Qt flow ratios (i.e. permeate flow to total flow) based on either THM4 or HAAS controls the design.
The blended water quality parameters are also calculated for the feed / permeate blends.
If the permeate quality does not meet the MCLs prior to blending, then these calculations are meaningless.
-------
Table 4-17b Blending Calculations To Determine The Fraction Of Flow That Requires NF
Treatment To Meet The Stage H DBF Regulations
Utility name and address
ICR plant number
Membrane trade name
Phone number
FAX number
Contact person
Parameter
THM4F, ng/L
THM4DJ figff,
HAA5F, ng/L
HAAS^g/L
AlkF, mg/L CaCO3
Alkp,mg/LCaCO3
T-HdF, mg/L CaCO3
T-HdB, mg/L CaCO3
Ca-Hdf,mg/LCaCO3
Ca-Hdp, mg/L CaCO3
RTJNID#
1
2
3
4
THM4 Controls
Qn/QT(THM4)J%
Alkb, mg/L CaCO3
T-Hdb, mg/L CaCO3
Ca-Hdb, mg/L CaCOS
THM4b, ng/L
HAA5b, lig/L
v ^y\>,v-.
^j. *» <**,
4s ' ' ' 'i'SSKs.^-.
v, - . tve^v- . ,
s*" 's^" v
s ""
"^V^sV *^
" %
; %
i s
. fff f ' < <
\ s
:,;,
- .
>^ *» '
f
f :
"* *
HAAS Controls
Qp/QT(HAA5)5%
Alkb, mg/L CaCO3
T-Hdb, mg/L CaCO3
Ca-Hdb, mg/L CaCOS
THM4b, ng/L
HAA5bJ ^g^L
-------
5.0 Single Element Bench-Scale Test
5.1 SEBST System
The following section presents the general design, equipment and procedures to be used
with the single element bench-scale test (SEBST). Section 5.1 is divided into the following
sub-sections: SEBST Requirements And Options, System Design, Continuous-flow
Pretreatment, System Start-up, System Shut-down, and Membrane Cleaning And Preservation.
5.1.1 SEBST Requirements And Options
The SEBST is a bench-scale procedure, and can only be used to meet the requirements of
the ICR by plants serving under 500,000 persons (with the exception of some joint study
options as described hi Part 1). Two options are available to plants using the SEBST
procedure (1) quarterly tests on two membrane types or (2) a yearlong study on a single
membrane type.
General SEBST Requirements
The following list of requirements applies to both the quarterly and yearlong SEBST
options:
The minimum element size to be evaluated is a 2.5" x 40" element. If hollow-fiber
technology is to be investigated, the smallest standard production size hollow-fiber element
should be investigated.
The experiments must be run at a recovery of 75% ±5% using concentrate recycle. If the
precipitation of a sparingly soluble salt limits the recovery of the system, then the highest
obtainable recovery should be used after appropriate pretreatment steps have been taken.
Other operating parameters, such as pressure, flux, influent flow rate, temperature and pH
must be within the membrane manufacturer's specifications.
Only membranes with manufacturer reported molecular weight cutoffs less than 1000
Daltons may be investigated.
The monitoring and sampling requirements for the SEBST are described in Sections 5.2.2
and 5.2.3, respectively.
Quarterly Studies On Two Membranes
Plants may elect to conduct four quarterly studies on two different membranes using the
SEBST procedure. The requirements for this option are as follows:
The study will evaluate two membranes for no less than four weeks per membrane each
quarter of one year with allowances for down-tune due to membrane cleaning and minor
maintenance. Thus, each membrane must be evaluated for approximately 112 days over
the course of one year. This will enable the impact of seasonal variation on membrane
performance to be evaluated.
3-55
-------
The two membranes may be evaluated simultaneously using two parallel systems or they
may be evaluated sequentially using a single SEBST system. However, the membranes
cannot be evaluated in series.
The same two membrane elements should be used for all four quarters when possible.
However, the membranes must be properly cleaned and stored according to the
manufacturer's specifications between quarterly runs. This insures that fouling from
previous runs does not affect membrane productivity hi following runs. If the MTCW can
not be restored to at least 80% of its initial value after cleaning, or if the rejection changes
by more than 15% during operation, the membrane may need to be replaced during
subsequent runs.
The purpose of quarterly studies is to evaluate the impact of seasonal variation on
membrane performance; however, the variability of some source waters may be captured hi
fewer than four quarters. For example, many ground waters do not exhibit significant
seasonal variation. In general, seasonal variation only needs to be evaluated when it is
significant, but four quarterly runs evaluating at least two membranes must be performed to
meet the ICR requirements. These four runs can be used to evaluate any variables of interest
such as additional membranes, different pretreatments or different operating parameters. For
example, during the second quarter the same two membranes can be evaluated using different
pretreatment. In a following quarter, two different membranes could be evaluated using the
optimal pretreatment scheme as determined during the previous two quarters. Different
operational parameters could also be investigated.
If the water is not subject to seasonal variations, then the quarterly SEBST runs can be
performed in any convenient tune frame as long as they are completed within a year of the
starting date.
Yearlong Study On A Single Membrane
As an alternative to quarterly studies, the utility may evaluate a single element
continuously for one year using the SEBST. This option requires that a single membrane type
be run continuously over a period of one year, with allowances for down-tune due to
membrane cleaning, maintenance or other reasons. The run time should be no less than 6600
hours which represents approximately 75% of a calendar year. Since only one membrane is to
be investigated, it is recommended that a preliminary evaluation be conducted to select an
appropriate membrane type. A membrane could be selected through a membrane selection
study, a literature review, discussions with manufacturers or based on experience with
membranes applied to a similar water source. The RBSMT procedure presented in Section 4.0
could be used to rapidly screen a number of membranes prior to a yearlong SEBST study.
5.1.2 System Design
The following section will present the guidelines and equipment to be used hi the design
and construction of single element bench-scale units. These guidelines are general, and
individual equipment for a specific design should be sized and verified for compatibility with
3-56
-------
individual membrane specifications. The terminology and definitions used in this section are
described in Tables 2-1 and 2-2 and Figure 2-1.
The minimum element size to be evaluated is 2.5" x 40". The only other common size that
is available for evaluation on a bench-scale would be a 4" x 40" element; however, larger
elements such as 8" x 40" elements can also be evaluated. The 2.5" x 40" element will
typically have an active membrane area of 17 to 23 ft2 and will require an influent flow rate of
approximately 0.75 to 1.5 gpm. The 4" x 40" element will have an active membrane area of
70 to 90 ft2 and will require an influent flow rate of 3 to 6 gpm.
The flow diagram of a SEBST unit showing its major components, system control points
and sampling locations is shown hi Figure 5-1. The basic design depicted hi the flow diagram
would apply to either the 2.5" x 40" or 4" x 40" membrane element, but the flow meters,
booster pump and pressure vessel would need to be sized based on the dimensions of the
element used. Most 2.5" x 40" elements are manufactured hi similar lengths and are
compatible with standard 2.5" x 40" pressure vessels. However, 4" x 40" element lengths are
not necessarily standard between manufacturers, and a product water tube adapter is typically
needed for length compatibility with standard 4" x 40" pressure vessels. Table 5-1 presents
approximate ranges for pumps and measurement devices to be used on a SEBST system.
Feed water is pretreated to prevent membrane fouling. As shown hi the flow diagram, the
chemically treated feed water is supplied to the prefilter at a pressure (20 to 40 psi) sufficient
to push the water through the prefilter and prevent the booster pump from cavitating. The
pretreated water can be combined with recycled concentrate water just after the prefilter on the
suction side of the booster pump. The combined feed and concentrate recycle flow is termed
the influent stream, and the influent flow rate sets the cross-flow velocity of the system. The
filtered water then enters the booster pump. Two pumps types that can be used as booster
pumps are positive displacement pumps or multi-stage centrifugal pumps, and the choice of an
appropriate pump depends on the pressure and influent flow rate requirements. The booster
pump must be sized to meet the membrane manufacturer's pressure and flow requirements for
the specific membrane element used; however, a pump recirculate line and valve can be used
to adjust the feed pressure. The pump recirculate line transfers influent water from the high
pressure side of the booster pump to the suction side of the pump. In this manner, the feed
pressure can be controlled with the recirculate valve allowing the pump to be slightly
oversized, providing some flexibility.
The pressurized influent stream flows into the membrane pressure vessel and enters
through the end of the spiral-wound membrane element. The influent stream is separated into
a permeate stream which results from passage through the membrane film, and a concentrated
stream (i.e., bulk stream) which flows across the membrane surface. The permeate is
collected into a center collection tube and passed out of the membrane pressure vessel. The
permeate flow rate is measured with a flow meter, and discharged through an outlet tube at
atmospheric pressure. The permeate flow rate should also be measured with a graduated
container and a stopwatch to verify and correct flow meter readings.
3-57
-------
The concentrate flow can be recycled or wasted. The concentrate waste is passed through
a flow meter, a valve and then discharged at atmospheric pressure. The concentrate waste
flow rate should also be measured with a graduated container and a stopwatch. The flow
meter and valve on the concentrate waste line are used to set the system recovery. A portion
of the concentrate stream is returned to the suction side of the booster pump and recycled to
obtain a system recovery of 75% and to meet manufacturer specifications for a rninimum
influent flow rate to an element or a maximum element recovery (i.e., single-pass recovery).
The recycle flow rate is controlled with a valve and in-line flow meter.
The following procedure was used to develop these equipment and size ranges for a
recovery of 75%. The design at 15% recovery is presented in Table 5-1 to demonstrate a
single-pass system (i.e., no recycle), but all SEBST studies are to be conducted at 75% ±5%
recovery or the highest obtainable recovery if the design is controlled by the precipitation of a
limiting salt after adequate pretreatment steps have been taken.
1. Obtain the membrane manufacturer's specification sheet for the specific membrane film
and element size to be used. The equipment presented hi Table 5-1 was sized based on
nanofiltration membrane specifications of: a flux of 15 gfd, maximum operating
pressure of 250 psi, and membrane surface areas of 20 ft2 for the 2.5" x 40" element
and 70 ft2 for the 4" x 40" element. Table 5-2 presents the membrane manufacturer
information used for this design.
2. The design permeate flux must be selected on the basis of feed water quality and
manufacturer specifications. Typically, a flux of 10 to 20 gfd is reasonable for a
ground water, and fluxes from 10 to 15 gfd or lower are recommended for surface
waters. In this example a ground water is being treated, so a flux of 15 gfd will be
used as the design flux. Once the design flux is selected, the permeate flow rate can be
calculated using Equation 5.1.
Qp = ₯w X Ae (5.1)
where Qp is the permeate flow rate Fw is the design flux and Ae is the active membrane
area of the element.
For the 4" x 40" element the permeate flow rate is calculated as:
Qp = Fw X Ae = 15 gfd X 70 ft2 = 1,050 gpd = 0.73 gpm
3. The feed flow rate is calculated from the permeate flow rate and the recovery using
Equation 5.2.
QF = Qp/R (5.2)
where QF is the feed flow rate and R is the fractional recovery.
3-58
-------
For the 4" x 40" element at a recovery of 75% the feed flow rate is calculated as:
QF = Qp / R = 0.73 gpm / 0.75 = 0.97 gpm
4. The concentrate waste flow rate is determined by subtracting the permeate flow from
the feed flow using Equation 5.3.
QW = QF-QP (5.3)
where Qw is the concentrate waste flow rate.
For the 4" x 40" element at 75% recovery, the concentrate waste flow rate is
calculated as:
Qw = QF - QP = 0.97 gpm - 0.73 gpm = 0.24 gpm
5. The feed flow rate of 0.97 gpm is below the manufacturer's minimum flow rate of 4
gpm per element. To obtain an acceptable influent flow rate at this recovery,
concentrate recycle must be employed. The required recycle flow rate is calculated
from the desired influent flow rate and the feed flow rate according to Equation 5.4.
QR = Qi-QF (5.4)
where QR is the recycle flow rate and Q, is the influent flow rate. The system can be
designed for any influent flow rate within the manufacturer's specifications.
In this example, for a single 4" x 40" element, the manufacturer's recommended single-
pass recovery is 15%; thus, the required influent flow rate can be calculated by dividing the
permeate flow rate by the single-pass recovery (0.73 gpm / 0.15 = 4.9 gpm). This influent
flow rate of 4.9 gpm is within the manufacturer's specifications (a muiimum flow rate of 4
gpm and a maximum flow rate of 16 gpm), and the recycle flow rate is calculated as:
QR = Qi - QF = 4.9 gpm - 0.97 gpm = 3.93 gpm
6. The recycle ratio can be calculated by dividing the recycle flow rate by the feed flow
rate according to Equation 5.5.
r = QR / QF (5.5)
where r is the recycle ratio.
3-59
-------
In this example, the recycle ratio is calculated as:
r = QR / QF= 3.93 gpm / 0.97 gpm = 4
7. In order to estimate the required feed pressure, first the osmotic pressure gradient must
be estimated. This requires that the TDS concentration be estimated hi the waste,
permeate and influent streams using Equations 5.6 through 5.8.
TDSW = TDSFX[1 + r - R + (RxRejTOS)] / [1 + r - R - (rxRxRejTOS)] (5.6)
TDSp = (QF X TDSF - Qw X TDSw) / Qp (5.7)
TDSj = (QFxTDSF + QnXTDSw) / Q, (5.8)
where TDSp, TDSF, TDSw and TDSX are the TDS concentrations hi mg/L hi the
permeate, feed, waste and influent streams, respectively; and RejTOS is the
manufacturer reported TDS rejection expressed as a decimal fraction.
In this example, the feed TDS concentration is 300 mg/L and the manufacturer reported
TDS rejection is 0.70. The TDS concentrations hi the waste, permeate and influent streams
are:
TDSW = TDSFX[1 + r - R + (RxRejTOS)] / [1 + r - R - (rXRxRejTDS)] =
300 mg/Lx[l + 4 - 0.75 + (0.75X0.70)] / [1 + 4 - 0.75 - (4x0.75x0.70)] =
666 mg/L
TDSp = (QpXTDSp - QWXTDSW) / Qp=
(0.97 gpmXSOO mg/L - 0.24 gpmX666 mg/L) / 0.73 gpm = 180 mg/L
TDSj = (QpXTDSp + QRXTDSW) / Qj =
(0.97 gpmXSOO mg/L + 3.93 gpmX666 mg/L) / 4.9 gpm = 593 mg/L
8. The TDS concentrations calculated hi the preceding step can now be used to estimate
the osmotic pressure gradient using Equation 5.9.
ATI = [(CTDSr + TDSW) / 2) - TDSp] X 0.01 (5.9)
where An is an estimate of the average osmotic pressure gradient hi psi, and 0.01 is the
factor to convert TDS (mg/L) to pressure (psi).
In this example, the osmotic pressure gradient is calculated as:
ATI = [((TDS! + TDSW) / 2) - TDSp] X 0.01 =
[((593 mg/L + 666 mg/L) / 2) - 180 mg/L] X 0.01 psi per mg/L = 4.5 psi
3-60
-------
9.
The net driving pressure is estimated from the design permeate flux and the water mass
transfer coefficient using Equation 5.10. (Note that the MCTW may have to be
corrected to the temperature at which the study will be conducted using Equation 5.16).
w
/MTC,
(5.10)
where Fw is the permeate flux, MTCV is the water mass transfer coefficient for the
membrane under investigation and NDP is the net driving pressure for the specific
membrane and flux.
In this example, the net driving pressure is calculated as:
NDP = Fw I MTCV = 15 gfd / (0.20 gfd / psi) = 75 psi
10. The required feed pressure can be calculated by summing the osmotic pressure
gradient, the NDP and any additional losses in the system.
PF = NDP + ATI + AP,,
(5.11)
.where PF is the feed pressure that must be supplied by the booster pressure pump and
APloss is the summation of additional pressure losses that occur through the membrane
element and system hardware such as pipes, valves, flow meters, etc.
For this example, assuming system losses of 7 psi, the feed pressure requirement is
calculated as:
PF = NDP + ATT + APloss = 75 psi + 4.5 psi + 7 psi = 86.5 psi
Thus, the high pressure pump should be sized for optimal operation near 87 psi and at a
flow rate of 5 gpm. However, the pump should be able to operate over a range of pressures
from 80 to 100 psi so that the membrane can be operated in a constant flux mode.
11. From these calculations, flow meters can be sized to measure these flows at the
approximate mid-range of the gauges. Table 5-1 presents typical flow gauge ranges for
this particular design for both 2.5" and 4" diameter elements. In most single element
units, the permeate and concentrate flows are added to determine the feed flow rate;
thus it is only necessary to have a recycle flow meter since the permeate and
concentrate flows are measured directly by timed volumetric displacement.
5.1.3 Continuous-flow Pretreatment
Pretreatment for membrane processes commonly includes chemical addition to prevent
inorganic precipitation and cartridge filtration to reduce colloidal fouling. The most common
chemical pretreatment is sulfuric acid addition to reduce the pH to prevent calcium carbonate
precipitation. In some cases hydrochloric acid is substituted for sulfuric acid. The potential
for calcium carbonate precipitation can be checked by calculating the Langlier Saturation Index
3-61
-------
for the concentrate stream. Some applications may require the addition of a chemical
antiscalant to control such inorganic precipitants as calcium sulfate, barium sulfate or
strontium sulfate. The use of an antiscalant can eliminate the need for acid addition in some
cases, and many antiscalants can prevent scaling in systems with a concentrate LSI ^ +1.5.
The required chemical doses are determined by limiting salt calculations or manufacturer
computer programs, both of which typically require a preliminary and comprehensive chemical
analysis of the raw water. Additionally, the fouling potential of the feed water should be
evaluated using one or more of the methods presented in Section 2.5.
Since they have relatively small flows, the concentrated chemicals used for pretreatment
are generally diluted into 20 to 50 gallon containers before injection into the feed stream. The
injection point for pretreatment chemicals shown hi Figure 5-1 was chosen so that the cartridge
filter and booster pump will assist in mixing the pretreatment chemicals. In some cases an in-
line static mixer may be used before the cartridge filter to insure proper mixing.
After chemical addition, the water is passed through a cartridge filter to remove larger
suspended solids or colloidal material, mix the pretreatment chemicals and protect the booster
pump and membrane element from sand or other foreign materials. Polypropylene cartridge
filters with a size exclusion of 5 um are acceptable for membrane pretreatment. The filter
cartridges should be cleaned or replaced when the pressure drop across the prefilter increases
by a predetermined percentage (e.g., 50%).
For feed waters which have excessive fouling as indicated by fouling indices or other
factors, advanced pretreatment will need to be incorporated into the system. Advanced
pretreatment may include enhanced coagulation and sand filtration with reduced pH. An
example of this would be operating the membrane system using water from a conventional
treatment plant after sand filtration and prior to the addition of any oxidant or disinfectant. If
alum is used as the coagulant, the pH of the feed water may need to be reduced to around 4.0
(or just above the minimum operational pH as specified by the manufacturer) to ensure the
solubility of aluminum hydroxide. However, operating at a low pH may increase fouling by
high molecular weight organic matter. Additional advanced pretreatment schemes may include
microfiltration to control particulate or microbial fouling.
5.1.4 System Start-up
The following section will describe the general steps to follow during initialization and
start-up of a single element test unit.
1. Select a membrane type and obtain the appropriate element size and the manufacturer's
specification sheet for that element.
2. Conduct a thorough analysis of the source water to evaluate the inorganic chemical
matrix and determine limiting salts by calculation or manufacturer computer program.
3-62
-------
3. Conduct fouling index tests on the feed water, calculate indices and determine the
potential for fouling problems. If a fouling problem is indicated, additional
pretreatment alternatives may need to be evaluated.
4. Select a membrane flux rate consistent with the fouling potential of the water and the
manufacturer's recommendations.
5. Calculate or consult manufacturer computer programs for feed water acid and/or
antiscalant dose. Calculate the dilution needed for the chemical feed tank capacity and
the chemical pump feed rate.
6. Load the element into the pressure vessel and carefully secure the end caps.
7. Flush the feed line upstream of the first stage of the membrane system, but downstream
of any pretreatment processes to remove debris and other contaminants. Flushing
should be continued for several minutes.
8. Connect the feed line of membrane unit to a pressurized (20 to 40 psi) source water
transmission line and open feed valve to allow water to enter the membrane prefilter.
9. Open all valves including permeate, concentrate waste, concentrate recycle, and pump
recirculate valves. This is to allow water to flow through the system, displace any
trapped air and flow to waste upon start-up.
10. Verify that the feed water is at the prefilter and release trapped air by depressing the
bleed button located at the top of the prefilter (if available).
11. Verify that there are proper and secure connections for the chemical feed system, and
energize the chemical feed pump. If the unit is going to be operated unattended for
significant periods of time, some consideration should be given to controls to shut-
down the unit if chemical dosing is interrupted.
12. Connect the power cord of the membrane unit to a compatible power supply and
energize the booster pump. Verify that water is passing through the unit by observing
the waste flow meter. Adjust the concentrate waste valve so that the concentrate flow
does not exceed the manufacturer's maximum recommended flow rate per element.
13. While continuously monitoring the feed pressure to prevent exceeding the unit or
membrane manufacturer's specifications, slowly close the concentrate waste valve to
set the system recovery at 75%.
14. While continuously monitoring the feed pressure to prevent exceeding the unit or
membrane manufacturer's specifications, slowly close the concentrate recycle valve to
obtain the selected recycle and influent flow rates.
3-63
-------
15. Slowly close the pump reckculate valve to set the permeate flow at the design flux rate.
16. Adjust the concentrate waste, concentrate recycle and pump recirculate valves to set the
recovery, influent flow rate and flux rate at the desired values. These parameters must
be consistent with the requirements of the ICR and the specifications of the
manufacturer.
17. Confirm the proper feed rate for acid by measuring pH or antiscalant by calibrating the
chemical pump feed rate.
*
18. Use the data sheet in Table 5-8 to record the initial system conditions, including the
feed, concentrate and permeate flow rates and pressures, the recycle flow rate, and the
influent temperature.
5.1.5 System Shut-down
The following section will describe the general steps to follow during shut-down of a
single element test unit. The single element studies should be conducted with the objective of
operating continuously from start-up to finish; however, continuous operation is not a
requirement as long as the run time meets the requirements of the ICR. The system may be
shut-down for brief periods (i.e., not to exceed 48 hours) if it cannot be monitored or if the
feed stream from the plant needs to be shut-off. If unit operation is interrupted for more than
24 hours, the element should be flushed with feed water once per day for 30 minutes to
minimize biological growth.
1. Use the data sheet in Table 5-8 to record the final system conditions, including the
feed, concentrate and permeate flow rates and pressures, the recycle flow rate, and the
influent temperature.
2. Collect approximately 20 gallons of permeate to use for membrane preservation and
cleaning solutions.
3. Open the waste, recycle, pump reckculate and permeate valves to fall open.
4. Turn the membrane booster pump to the off position.
5. Turn the chemical feed pumps to the off position.
6. Close the valve from the feed water source.
7. If the system is to be shut-down for more than 24 hours but less than one week, the
previously collected permeate water should be used to flush the feed/concentrate side of
the membrane. If the membrane is to be shut-down for longer than one week, it should
be preserved according to the manufacturer's recommended procedure.
3-64
-------
8. The permeate valve should not be closed during shut-down in order in order to avoid
osmotic flow from the permeate to the feed side of the membrane. This "reverse flow"
can result hi membrane damage.
5.1.6 Membrane Cleaning And Preservation
The following section will provide general procedures to be followed during the cleaning
and preservation of membranes used hi the SEBST. Membranes are usually cleaned when the
temperature-normalized MTC^has decreased by 10 to 15% from the baseline at the beginning
of the study or the baseline established after the most recent cleaning. It should be noted that
the MTCW has to be normalized for temperature, since a drop hi temperature will cause an
increase hi the net driving pressure required to maintain a constant flux. The equation used to
normalize the MTCW to a common temperature (i.e., the average yearly water temperature at
the plant conducting the study) is presented hi Section 5.2.5 as Equation 5.16.
Membrane cleaning frequencies greater than once per month have been suggested to be
limiting because of the associated cost and lack of automation. However, there is no reason
that membrane cleaning cannot be highly automated and made a routine part of membrane
plant operation. Unfortunately, that technology is not widely used today, and the impact of
cleaning frequency must be considered hi the overall cost and performance of a membrane
process.
Since membranes are made from many different materials, membrane manufacturers
specify chemicals, chemical strengths, temperatures and pH values for cleaning solutions.
Membrane compatibility with a specific cleaning solution must be verified with the membrane
manufacturer to avoid damage to the film. There are two basic categories of membrane
cleaning solutions, alkaline and acidic solutions. In general, alkaline solutions such as a 0.1%
solution of sodium EDTA and a 0.1% solution of sodium hydroxide are effective for removing
organic and biological fouling agents. Alkaline solutions are typically used hi conjunction
with surfactants and detergents such as sodium lauryl sulfate or Triton-X. Acidic solutions
such as 0.5% phosphoric acid are effective for removing inorganic foulants. When the exact
nature of the foulant is not known, the manufacturer's recommendation for an appropriate
cleaning procedure should be solicited.
During membrane cleaning, the single element unit is operated as a closed loop system.
The feed line from the suction side of the booster pump is connected or submerged hi the
cleaning solution tank along with the permeate and waste discharge lines. With the permeate
and concentrate waste valves completely open, the cleaning solution is recirculated through the
pressure vessel. In situations were the membranes are highly fouled, the cleaning solution
may need to be pre-heated and the membranes may need to be soaked hi the pre-heated
cleaning solution for 1 to 24 hours. It is also noted that some manufacturers require low
pressures during cleaning which might require a separate low pressure pump for cleaning;
most manufacturers limit the pressure to 60 psi during cleaning. Once the cleaning cycle has
been completed, the membranes should be flushed with previously collected permeate water.
Permeate should also be used to flush the cleaning tank and system between cleanings with
acidic and alkaline solutions.
3-65
-------
To preserve the membranes for storage, follow membrane manufacturer specifications for
type and strength of preservative. Membranes should always be cleaned prior to preservation
and storage. Use previously collected membrane permeate and mix with preservative
chemicals to achieve the proper concentration. Connect a suction line from the booster pump
to the preservative solution tank, the same tank used for cleaning can be used for membrane
preservation solutions, and energize the booster pump long enough to replace the feed and
concentrate water left from previous operation with the preservative solution. Turn off the
booster pump and close all valves to trap the preservative solution within the membrane
pressure vessel. If the membrane is to be stored separately from the pressure vessel, then it
may be wetted in the preservative solution and placed into a sealed plastic bag.
Handle, dispose and store all chemicals used in the membrane study in a safe and approved
method.
5.2 SEBST Procedure
The following section will present the procedures for monitoring membrane performance
during SEBST runs. It will present the parameters and frequency of monitoring as well as
sheets for recording data. This section will also present an example of the methodology used
to calculate and predict membrane productivity. The section is divided into the following
subsections: Selecting Operating Parameters, Monitoring Requirements, Sampling
Requirements, Data Sheets, and Membrane Productivity.
5.2.1 Selecting Operating Parameters
There are several operating parameters which must be selected and maintained to conduct a
successful membrane study, and Section 5.1.2 describes the procedure for calculating these
operating parameters. The recovery must be set and maintained at 75 % ± 5 %. A lower
recovery can be used if the recovery is limited by the precipitation of a sparingly soluble salt
after appropriate pretreatment steps have been taken. Membrane manufacturers recommend a
minimum influent flow rate (or cross-flow velocity) to avoid excessive ion build up at the
membrane film surface (concentration polarization). For 4" diameter membranes, this equates
to an influent flow rate of 3 to 6 gpm. The concentrate waste valve is set to achieve a
recovery of 75% while the concentrate recycle valve must be set to produce a flow which
combined with the feed flow maintains the desired influent flow rate. Additionally, there are
manufacturer specifications for a recommended water flux range, a maximum allowable
influent flow rate to an element and a maximum allowable pressure. The water flux range
usually falls between 10 and 25 gfd. Lower flux (< 15 gfd) and lower single-pass recoveries
(< 10% per element) have achieved increased water productivity hi some surface water pilot
studies (Taylor et al., 1990; Taylor et al., 1992). Although the system recovery is set at 75 ±
5%, the single-pass recovery can be reduced to 10 to 15% by recycling a portion of the
concentrate stream.
Pretreatment chemical addition and the appropriate feed pH must also be constantly
maintained to prevent inorganic fouling. In addition, the membrane manufacturer may list
additional specifications that must be met to prevent damage to the membrane film or element.
3-66
-------
For example, most thin-film composite membranes cannot tolerate any chlorine-based
disinfectants.
5.2.2 Monitoring Requirements
The flux, net driving pressure and temperature must be monitored during the entire study
to calculate the MTCW and evaluate the rate of MTCW decline. Since the flux and recovery are
maintained at constant levels, fouling will be identified with an increase in net driving
pressure, and the drop in productivity can be quantified by calculating the MTCW.
Additionally, some general water quality parameters must be routinely monitored to assess
variations hi permeate quality.
The sites in Figure 5-1 that need to be monitored are the feed stream before joining with
the recycle stream, as well as the permeate, concentrate waste and concentrate recycle streams.
The parameters that must be monitored at these sites are pressure, flow rate, temperature, pH
and TDS. The feed and permeate TOC and/or UV^ may also be measured and reported as
part of routine monitoring if desired. The sensor for monitoring temperature should be
positioned to monitor the influent stream just before it enters the pressure vessel.
These parameters must be monitored daily at the sites listed in Table 5-3 for both quarterly
and yearlong SEBST studies. In addition to monitoring and recording the gauge readings, the
permeate and concentrate flow rates should be measured directly to calibrate or verify the flow
meters. Table 5-8 presents a data sheet for recording the SEBST monitoring data.
5.2.3 Sampling Requirements
The feed, permeate and concentrate streams should be sampled at the locations shown in
Figure 5-1. Table 5-4 presents the analyses to be conducted on the samples collected from
each site for both the quarterly and yearlong SEBST studies. The THM4, HAA6 and TOX
samples should be formed under SDS conditions that represent the average distribution system
conditions at the plant. The chlorine dose and SDS conditions should be reported along with
the results of the DBF analyses.
The parameters listed in Table 5-4 must be sampled weekly from each membrane for the
quarterly studies. For yearlong studies, the analytes listed in Table 5-4 must be sampled
biweekly.
The analyses for the feed should be conducted after it has received appropriate membrane
pretreatment, but prior to joining with any recycled concentrate. The feed should be sampled
far enough upstream of the junction where the feed joins the recycled concentrate to prevent
any back-flow of concentrate into the feed sample. In some cases a check valve may be
required to prevent back-flow of concentrate into the feed sample. If the two membranes are
evaluated concurrently during quarterly studies, then the feed only needs to be sampled once
per week for the analytes listed in Table 5-4.
Sampling, in terms of holding times, preservation, and sampling techniques, should be
conducted in accordance with the "ICR Sampling Manual" (EPA 814-B-96-001). Approved
3-67
-------
methods for analysis are listed in Table 7, § 141.142 of the ICR Rule, and the analyses for the
treatment studies must be conducted according to the analytical and quality control procedures
contained in the "DBP/ICR Analytical Methods Manual" (EPA 814-B-96-002).
5.2.4 Datasheets
This section describes data sheets that can be used to record the appropriate data from the
SEBST procedure for the ICR. Corresponding data collection software will be sent to utilities
electing to conduct SEBST experiments after the plant submits a study concept form to EPA.
Some of these data sheets will have to be modified for the long-term SEBST.
The data sheet hi Table 5-6 is used to report the characteristics of each membrane used
during the treatment study. These are the membrane characteristics as reported by the
manufacturer, and although some of this information may not be available, the data sheet
should be filled out as completely as possible. The area and cost of an 8" x 40" element are
requested for use in the cost analysis.
The data sheet in Table 5-7 requests information on the foulants hi the feed water and the
pretreatment processes used to control fouling. The first section of this table requests
concentrations of foulants and fouling indices for the feed water. Only the foulants and indices
relevant to the water being investigated need to be measured and reported. The blank rows
should be used to report additional foulants that were measured but not listed hi this table.
This information should be used as a guide to selecting appropriate pretreatment processes.
During the run, the relevant fouling indices and water quality parameters should be
periodically measured to insure that pretreatment processes are performing properly and that
the proper chemical doses are being applied to the feed water.
The second half of this table requests information about the pretreatment processes used
prior to nanofiltration. All pretreatment processes used should be reported here including
processes hi the existing full-scale treatment tram, upstream of the feed to the SEBST system;
and processes that are added specifically for membrane pretreatment. Existing full-scale
treatment processes used as membrane pretreatment should be marked with an "E",
modifications to processes hi the existing plant treatment train (e.g., an increase in the
coagulant dose) should be indicated by an "M" and pretreatment processes used in addition to
the existing treatment tram (e.g., acid or antiscalant addition) should be indicated with an
"A". This table can be used to provide some of the pretreatment information required hi the
final treatment study report.
Table 5-8 contains the parameters that should be monitored with tune for each membrane.
Temperature normalized fluxes, the MTC^, and both feed and bulk rejections will be
calculated by the data collection software. The feed rejection is calculated from the feed and
permeate concentrations. To calculate the bulk rejection, the permeate and feed concentrations
and the recovery are used to first estimate the bulk concentration, which is then used to
calculate the bulk rejection. The cumulative run tune hi this data sheet should be set at 0:00 at
the start of the run and continued over each four week run or over the yearlong study. Down
3-68
-------
time for the system must be subtracted from the cumulative run time (i.e., by stopping the
timer when the system is turned off).
The water quality parameters listed in Table 5-4 must be measured once each week for
each membrane during the quarterly studies, and this data should be reported in Tables 5-9
through 5-12. For each membrane, the analyses for one set of weekly samples must be
duplicated each quarter and reported in Table 5-13.
For the yearlong SEBST studies, the water quality parameters listed in Table 5-4 must be
sampled biweekly and reported on data sheet similar to Tables 5-9 through 5-12. The date of
sample collection should be listed at the top of the data sheet along with the week number
(e.g., week 2, week 4, week 6, ... week 52). Thus at least twenty (20), but no more than
twenty-six (26), sample sets are required for both the feed and the permeate. The analysis on
every fifth set of samples (feed and permeate) should be duplicated, resulting in four to five
sets of duplicate analyses over the course of the yearlong study.
Results from analysis of the concentrate samples sets should be reported hi Table 5-14
along with corresponding permeate and feed concentrations. Table 5-14 will need to be
extended to include twenty (20) to twenty-six (26) sample sets for the yearlong study. These
complete data sets can be used to calculate the mass balance closure errors which can be used
to assess sampling and analytical techniques. Mass balance closure errors on common
parameters (e.g., TOC, UV254, TDS, alkalinity, hardness, etc.) are typically less than a few
percent.
The mass balance closure error is calculated in two steps. First the concentrate
concentration is calculated from the feed and permeate concentrations and the fractional
recovery using Equation 5.12.
C - CF ~ (C" X R) (5.12)
C(calc)~ (1-R)
where Cc(calc) is the calculated concentrate concentration. Next the mass balance closure error
is determined by comparing calculated and measured concentrate values:
Cc(meas) - CC(calc)
x lUUvo
CC(meas) (5.13)
where Error^H is the mass balance closure error expressed as a percentage and CC(meas) is the
measured concentrate concentration.
Since permeate quality often exceeds the treatment objectives, permeate can be blended
with by-passed feed water. This can substantially reduce the required membrane area and
post-treatment requirements, and thus the cost. However, any removal of pathogens that the
3-69
-------
membrane is capable of will be negated by blending permeate and feed water, unless the
pathogens were removed or inactivated by a process upstream of the membrane process. By
calculating the water quality for different blending ratios, costs for different levels of treatment
can be estimated. Tables 5-15 (a and b) use the average feed and permeate water quality
parameters to calculate the amount of flow that must be treated by membranes to achieve the
Stage 1 MCLs (Table 5-15a) and the proposed Stage 2 MCLs (Table 5-15b). Water quality
parameters that impact post-treatment requirements are also calculated for the blended waters.
In these two tables, QT is the total blended flow, and the ratio QP/QT is the fraction of flow that
must be treated by membranes to meet the treatment objectives. This ratio is calculated for
both THM4 and HAAS as either of these water quality parameters can control the blend ratio.
The subscript b in these tables denotes blended water quality.
5.2.5 Membrane Productivity
Membrane productivity is assessed by the MTC,y decline with time of operation. All
membranes foul during operation, and constant production is achieved in full-scale membrane
plants by increasing pressure. To develop the cleaning frequency or rate of fouling for a
SEBST, the following procedure is presented:
The MTC,y is calculated by using the following equations and data recorded from a short-
term single element bench-scale study shown hi Table 5-5. Using the definitions for permeate
flux and the water mass transfer coefficient from Table 2-2:
Vw = Qp/Ae = MTCV X NDP (5.14)
Rearranging the equation for water flux and solving for MTCW:
MTCV = ?w I NDP (5.15)
The flux of water is calculated using a permeate flow of 784 gpd and single 4" x 40"
element membrane area of 70 ft2:
Fw = QP/Ae= (784 gpd) / (70 ft2) = 11.2 gfd
The flux is normalized to a common temperature. Equation 5.16 can be used to normalize
the flux to the average yearly water temperature experienced at the plant conducting the study
if a manufacture does not specify a temperature correction equation.
F^Tavg'C) = FTV(T0C)xl.03 ^'-^ (5.16)
where Fw(Tavg0C) is the flux corrected to the average yearly water temperature at the plant
conducting the study, FW(T0C) is the flux measured at ambient temperature, T° is the
temperature at which the flux was measured in °C, and Tavg° is the average yearly water
temperature at the plant conducting the study hi °C.
3-70
-------
The temperature correction equation was not used in this particular case since the
temperature at which the study was performed was equivalent to the average temperature of
the source.
In order to determine the NDP, the osmotic pressure gradient must first be estimated from
the influent, concentrate waste and permeate TDS values using Equation 5.9 which was first
presented hi Section 5.1.2.
For this example TDSp is 100 mg/L, TDSj is 200 mg/L and TDSW is 500 mg/L.
ATT = ([q±>SI+.TDSw)/2]-TDSp)X( 1 psi / 100 mg/L) =
[(200 mg/L+500 mg/L)/2]-100 mg/L)(lpsi / 100 mg/L TDS) = 2.5 psi
The net driving pressure is calculated according to Equation 5.17.
-PP-ATI (5.17)
where NDP is the net driving pressure, P, is the influent pressure, Pc is the concentrate
pressure and Pp is the permeate pressure.
For this example, Pz is 45 psi, Pc is 35 psi and Pp is 5.5 psi.
NDP = [(Pj+Pc^J-Pp-ATt = [(45psi+35psi)/2]-5.5psi-2.5psi = 32psi
Using these results the MTCW can be calculated from Equation 5.14 as:
MTCV = FW/NDP= 11.2 gfd / 32 psi = 0.35 gfd/psi
The MTCV is calculated for each set of operational data and plotted versus the cumulative
time of operation. A spreadsheet incorporating these equations can calculate the net driving
pressure, the temperature normalized flux and the MTCW using the data recorded in Table 5-8.
The data summarized in Table 5-5 is presented graphically hi Figure 5-2. The slope of a
linear least squares fit of the plotted data is the change hi the MTCW with time and this slope is
used to predict the rate of fouling and the required cleaning frequency. From the data
summarized hi Table 5-5 and presented graphically hi Figure 5-2, the linear regression line
calculated from a spreadsheet is:
MTCV = - 0.0005 X tune + 0.355
3-71
-------
Equation 5.18 can be used to estimate the cleaning frequency from the rate of MTCW
decline and the acceptable loss in the MTCW.
CF =
dMTCV/dt (5.18)
where CF is the cleaning frequency, Q is the acceptable fractional loss in the MTCW prior to
cleaning (e.g., 0.15), MTCW(o) is the baseline MTCW established at the start of the run or after
the most recent cleaning, and dMTC^/dt is the rate of MTCW decline determined from a linear
regression of the flux data.
The data used to develop Figure 5-2 came from an actual short-term pilot study. Note the
baseline MTCW is 0.35 gfd/psi, and the acceptable 15 % MTCW loss is 0.054 gfd/psi. The
calculated cleaning frequency using the linear rate of decline is 105 days or 3.5 months. The
rate of MTCW decline in this example was estimated from a linear regression which worked
well for this data. However, the rate of MTCW decline may not always follow a linear model,
and the best model of MTCW decline over tune of production should be used to estimate the
rate of MTCV decline. In general, if the r2 value of the regression line is less than
approximately 0.90, the fit may not accurately predict the rate of MTCW decline. In this case,
another model for MTCW decline should be used. For example a linear regression for a plot of
the log of time versus the log of the MTCW may be used to predict the rate of MTCW decline.
A CF of 105 days is not considered an unreasonable rate of cleaning for a membrane plant
treating a highly organic ground water. This SEBST study showed that chemical fouling was
not prohibitive. Furthermore, this bench study showed that this source is treatable by a
conventional membrane process and that a more extensive pilot study could be undertaken to
develop design data. However, due to its short time of operation, this study did not measure
the potential for biofouling which is one of the reasons a more thorough pilot study is justified.
This point must be appreciated when interpreting data from short-term studies, since
biofouling may control the cleaning frequency in some situations.
3-72
-------
Acid/Antiscalant Sample
Temperature
Pump , & Pressure
Recirculate
Pressure & Flow
Gauge; Sample
Pressure & Flow
Gauge; Sample
Recycle
Figure 5-1 Schematic Of A Single Element Bench-Scale Unit
0.5
0.45.
0.4.
0.35,
f 0.3.
i> 0.25.
f=! 0.2.
0.15 .
0.1 .
0.05.
0.
*
MTCiv = -0.0005*time + 0.3549
' - ...
0 2 4 6 8 10 12 14
Time (days)
Figure 5-2 Temperature Normalized MFCW Plotted As A Function Of Tune
-------
Table 5-1 Example Design Parameters For Single Element Units
Recovery
Element size
Feed
Flow (gpm)
Pump Flow (gpm)
Pressure (jpsi)
Permeate
Flow (gpm)
Pressure (psi)
Concentrate
Flow (gpm)
Pressure (psi)
Recycle
Flow (gpm)
75% (with recycle)
2.5" x 40"
0.28
1.39
87
0.21
0
0.07
80
1.11
4"x40"
0.97
4.9
87
0.73
0
0.24
80
3.96
15% (single-pass)
2.5" x 40"
1.39
1.39
87
0.21
0
1.18
80
0
4" x 40"
4.9
4.9
87
0.73
0
4.2
80
0
Table 5-2 Design Characteristics For The Membrane Used In The Design Example
Parameter
Manufacturer
Membrane trade name
Element size
Membrane material / construction
Continuous operational pH range
Short-term cleaning pH range
Molecular weight cutoff, Daltons
Membrane area, ft2
Maximum operating temperature, °C
Maximum flow to element, gpm
Minimum flow to element, gpm
Maximum pressure, psi
Design pressure, psi
Design flux, gfd
Single element recovery, %
MTCW(T°C), gfd/psi
Max. pressure drop across element, psi
TDS rejection, %
Value
**:{::)::{:********************
*************************
4 inch diameter by 40 inch length
PVD / thin-film composite
3 to 9
1 to 11
500
70
35
16
4
250
80
15
15
0.20 at (20°C)
10
70
-------
Table 5-3 SEBST Routine Monitoring Requirements
Quarterly SEBST Study Monitoring Requirements
Parameter
Flow
Pressure
Temperature
TDS
pH
Feed
none
none
none
D
D
Permeate
D
D
none
D
D
Concentrate
D
D
none
D
D
Influent
none
D
D
none
none
Recycle
D
none
none
none
none
Yearlong SEBST Study Monitoring Requirements
Parameter
Flow
Pressure
Temperature
TDS
pH
Feed
none
none
none
D
D
Permeate
D
D
none
D
D .
Concentrate
D
D
none
D
D
Influent
none
D
D
none
none
Recycle
D
none
none
none
none
D - daily (once per shift)
-------
Table 5-4 SEBST Water Quality Monitoring Requirements
Quarterly SEBST Study Sampling Requirements
Parameter
pH
Total Hardness
Calcium Hardness
Alkalinity
Total Dissolved Solids
Turbidity
Total Organic Carbon
UV2S4
Bromide
SDS - THM4
SDS - HAA6
SDS-TOX
SDS - C12 demand
Feed
W
W
W
W
W
W
W
W
W
W
W
W
W
Permeate
W
W
W
W
W
W
W
W
W
W
W
W
W
Concentrate
W
W
W
W
W
W
W
W
none
none
none
none
none
Yearlong SEBST Study Sampling Requirements
Parameter
pH
Total Hardness
Calcium Hardness
Alkalinity
Total Dissolved Solids
Turbidity
Total Organic Carbon
UV2S4
Bromide
SDS-THM4
SDS-HAA6
SDS-TOX
SDS - C12 demand
Feed
BW
BW
BW
BW
BW
BW
BW
BW
BW
BW
BW
BW
BW
Permeate
BW
BW
BW
BW
BW
BW
BW
BW
BW
BW
BW
BW
BW
Concentrate
BW
BW
BW
BW
BW
BW
BW
BW
none
none
none
none
none
W - weekly
BW - biweekly
-------
Table 5-5 Short-term Membrane Productivity Study
Time
Days
0.00
0.99
1.98
2.99
3.94
4.96
5.91
6.90
8.99
10.02
11.04
12.02
13.02
14.02
NDP
(psi)
32
31
33
32
31
31
31
32
30
30
30
30
30
30
Recovery
(%)
15
15
14
15
15
15
16
14
15
15
15
14
14
11
Flux
Cera)
11.20
11.16
11.22
11.20
11.16
11.16
11.16
11.20
10.50
10.50
10.50
10.50
10.20
10.50
MTCW
(gfd/psi)
0.35
0.36
0.34
0.35
0.36
0.36
0.36
0.35
0.35
. 0.35
0.35
0.35
0.34
0.35
-------
Table 5-6 Membrane Characteristics As Reported By The Manufacturer
Utility name and address
ICR plant number
Phone number
Contact person
FAX number
Characteristics of the membrane element used in the study
Membrane manufacturer
Membrane module model number
Size of element used in study (e.g. 4" x 40")
Active membrane area of element used in study
Active membrane area of an equivalent 8" x 40" element
Purchase price for an equivalent 8" x 40 " element ($)
Molecular weight cutoff (Daltons)
Membrane material / construction
Membrane hydrophobicity (circle one)
Membrane charge (circle one)
Design pressure (psi)
Design flux at the design pressure (gfd)
Variability of design flux (%)
MTCV (gfd/psi)
Standard testing recovery (%)
Standard testing pH
Standard testing temperature (°C)
Design cross-flow velocity (fps)
Maximum flow rate to the element (gpm)
Minimum flow rate to the element (gpm)
Required feed flow to permeate flow rate ratio
Maximum element recovery (%)
Rejection of reference solute and conditions
of test (e.g. solute type and concentration)
Variability of rejection of reference solute (%)
Spacer thickness (ft)
Scroll width (ft)
Acceptable range of operating pressures
Acceptable range of operating pH values
Typical pressure drop across a single element
Maximum permissible SDI
Maximum permissible turbidity (ntu)
Chlorine/oxidant tolerance
Suggested cleaning procedures
Hydrophilic Hydrophobic
Negative Neutral Positive
Note: Some of this information may not be available, but this table should be filled out as completely
as possible for each membrane tested.
-------
Table 5-7 Membrane Pretreatment Data
Utility name and address
ICR plant number
Contact person
Membrane trade name
Phone #
FAX#"
Foulants and fouling indices of the feed water prior to pretreatment1
Alkalinity
Ca Hardness (mg CaCO3/L)
LSI
Dissolved iron (mg/L)
Total iron (mg/L)
Dissolved aluminum (nig/L)
Total aluminum (mg/L)
Fluoride (mg/L)
Phosphate (mg/L)
Sulfate (nig/L)
Calcium (mg/L)
Barium (mg/L)
Strontium (mg/L)
Reactive silica (mg/L as SiO2)
Turbidity (ntu)
SDI
MFI
MPFI
1: Only those foulants and fouling indices relevant to the water being tested need to be evaluated.
Additional foulants and indices can be listed in the blank rows or on an attached sheet.
Pretreatment processes used prior to nanofiltration or reverse osmosis2
Pre-filter exclusion size (um)
Type of acid used
Acid concentration (units)
mL of acid per L of feed
Type of antiscalant used
Antiscalant concentration (units)
mL of antiscalant per L of feed
Type of coagulant used
Coagulant dose (mg/L)
Type of polymer used during coag.
Polymer dose (mg/L)
2: Use an "E" to indicate a pretreatment process that is currently part of the plant treatment train, an "M" to
indicate a modification to a process that is currently part of the plant treatment train, and an "A" to
indicate an addition to the current treatment train.
Additional pretreatment processes, such as MF, can be listed in the blank rows or on an attached sheet.
-------
Table 5-8 Membrane Operational And Performance Data Monitoring With Time (page 1)
Utility name and address
Phone number
FAX number
ICR plant number
Membrane trade name
Contact person
Operational parameters monitored with time
Date
Time
(hr)
Cumulative
time
(hr)
PI
(psi)
PC
(psi)
PP
(psi)
Influent
Temp
(°C)
QR
(gpm)
QP
(gpm)
Qw
(gpm)
QF
(gpm)
'
f
f >
_,
"
/*
,
"
' -
"'
'
.,
'
* j^-A..
.», '
,
Recovery
(decimal)
tS"tSs
,
' /%
::.
»
A
Notes: Shaded cells can be calculated by a spreadsheet, but should also be manually calculated during the run.
Nomenclature is defined in Figure 2-1 and Table 2-2.
-------
Table 5-8 Membrane Operational And Performance Data Monitoring With Time (page 2)
Utility name & address Phone number
FAX number
ICR plant number
Membrane trade name
Contact person
Water quality parameters monitored with time
Date
Time
(hr)
Cumulative
time
(hr)
PH
Feed
CF
Permeate
cp
Concentrate
cc
TDS (mg/L)
Feed
CF
Permeate
cp
-
Concentrate
Cc
Notes: Nomenclature is defined in Figure 2-1 and Table 2-2.
-------
Table 5-8 Membrane Operational And Performance Data Monitoring With Time (page 3)
Utility name & address Phone number
FAX number
ICR plant number
Membrane trade name
Contact person
Water quality parameters monitored with time
Date
Time
(hr)
Cumulative
time
(hr)
UV254 (cm"1), optional analysis
Feed
CF
Permeate
CD
Concentrate
Cc
TOC (mg/L), optional analysis
Feed
CF
Permeate
CD
Concentrate
Cc
Notes: Nomenclature is defined in Figure 2-1 and Table 2-2.
-------
Table 5-9 Membrane Feed And Permeate Water Quality For Week One
Utility name and address
ICR plant number
Contact person
Membrane trade name
QP (gpm)[
Phone #
FAX#
1 1 QF(gpm)| |
Qi(gpm)| |
Water quality parameter
Sampling date
Sampling time
Alkalinity
Total dissolved solids
Total hardness
Calcium hardness
Bromide
PH
Turbidity
Temperature
Total organic carbon
UV254
SDS samples
Chlorine dose
Chlorine residual
Chlorine demand
Chlorination temperature
Chlorination pH
Incubation time
TOX
CHC13
CHBrCl2
CHBr2Cl
CHBr3
THM4
MCAA"
DCAA"
TCAA"
MBAA"
DBAA*
BCAA*
TBAA
CDBAA
DCBAA
HAA6
Units
MM/DD/YY
hh:mm
mg/L as CaCO3
mg/L
mg/L as CaCO3
mg/L as CaCO3
ug/L
ntu
°C
mg/L
cm"
mg/L
mg/L
mg/L
°C
hours
Ug/L
ug/L
Ug/L
Ug/L
ug/L
ug/L
ug/L
Ug/L
ug/L
ug/L
ug/L
ug/L
ug/L
ug/L
ug/L
ug/L
CF
CB
RF
%%%
"? .
,ff
::v.
,
>,;; ;:
f.V, fS
v,-S:
S S SSSSS f
V.-. %%
"
^ v.%
ffffff f s
"""7- " -
'"
,.,....
-
^
«
ff f
"
""""
RB
%
"
ff ff f
- '
^ s
S V, S
...
-
s
---'
w
.f-, V
, ,
III 1 II II ll'Mlfl
f ff f
,
~ %%%
"
,
V.JM ''
* These six species make up HAA6. The other three HAA species should be reported if measured.
Shaded cells can be calculated by a spreadsheet.
-------
Table 5-10 Membrane Feed And Permeate Water Quality For Week Two
Utility name and address
ICR plant number
Contact person
Membrane trade name
Qp(gpm)[
Phone #
FAX#"
QF
Qi(gpm)|
Water quality parameter
Sampling date
Sampling time
Alkalinity
Total dissolved solids
Total hardness
Calcium hardness
Bromide
pH
Turbidity
Temperature
Total organic carbon
UV254
Units
MM/DD/YY
hh:mm
mg/L as CaCO3
mg/L
mg/L as CaCO3
mg/L as CaCO3
Hg/L
ntu
°C
mg/L
cm"
CF
CP
RF
^
' '
; -
RB
...
-
'"-y
", *
"
"-' '"'Z
SDS samples
Chlorine dose
Chlorine residual
Chlorine demand
Chlorination temperature
Chlorination pH
Incubation time
TOX
CHC13
CHBrCl2
CHBr2Cl
CHBr3
THM4
MCAA'
DCAA'
TCAA'
MBAA"
DBAA'
BCAA*
TBAA
CDBAA
DCBAA
HAA6
mg/L
mg/L
mg/L
°C
hours
Hg/L
jig/L
ug/L
Hg/L
Hg/L
Hg/L
Hg/L
Hg/L
Hg/L
Hg/L
ug/L
Hg/L
Hg/L
ug/L
ng/L
ng/L
. / y %'''
>
' ',''4,1
. < '' *
, ,
fffr
"' 'yf.
-
, ,
' J'f,
, " "''''f
; "
'
->',. ,
'
f
* These six species make up HAA6. The other three HAA species should be reported if measured.
Shaded cells can be calculated by a spreadsheet.
-------
Table 5-11 Membrane Feed And Permeate Water Quality For Week Three
Utility name and address
ICR plant number
Contact person
Membrane trade name
QP(gpm)[
Phone #
FAX#
1 1 QF(gpm)l
1
Qi (gpm)| |
Water quality parameter
Sampling date
Sampling time
Alkalinity
Total dissolved solids
Total hardness
Calcium hardness
Bromide
pH
Turbidity
Temperature
Total organic carbon
UV254
Units
MM/DD/YY
hh:mm
mg/L as CaCO3
mg/L
mg/L as CaCO3
mg/L as CaCO3
Hg/L
ntu
°C
mg/L
cm"1
CF
CD | RF
-
S t t
s
* "f
RB
'
"
"
-
^
,v*v . .
SDS samples
Chlorine dose
Chlorine residual
Chlorine demand
Chlorination temperature
Chlorination pH
Incubation time
TOX
CHC13
CHBrCl2
CHBr2Cl
CHBr3
THM4
MCAA*
DCAA*
TCAA*
MBAA*
DBAA"
BCAA*
TBAA
CDBAA
DCBAA
HAA6
mg/L
mg/L
mg/L
°C
hours
ug/L
ug/L
ug/L
Ug/L
Ug/L
ug/L
Ug/L
Ug/L
ug/L
ug/L
ug/L
ug/L
ug/L
ug/L
ug/L
ug/L
*
-
"v.
-
.v.
V*
-
^
'
-.-.
-
'
f f
fffS f
"
,
V,-. -. %
.'' "
%
"
\ % ,
ft f? ^
s
.
--
ffW f
ff
" * ff
s "*" ffft
* These six species make up HAA6. The other three HAA species should be reported if measured.
Shaded cells can be calculated by a spreadsheet.
-------
Table 5-12 Membrane Feed And Permeate Water Quality For Week Four
Utility name and address
ICR plant number
Contact person
Membrane trade name
QP(gpm)[
Phone #
FAX#
QF(gpm)| |
Qi(gpm)|
Water quality parameter
Sampling date
Sampling time
Alkalinity
Total dissolved solids
Total hardness
Calcium hardness
Bromide
pH
Turbidity
Temperature
Total organic carbon
UV2J4
Units
MM/DD/YY
hh:mm
mg/L as CaCO3
mg/L
mg/L as CaCO3
mg/L as CaCO3
ug/L
ntu
°C
mg/L
cm"1
CF
cp
RF
" *
-
... "i ''
«
RB
S-',ff f
''
SDS samples
Chlorine dose
Chlorine residual
Chlorine demand
Chlorination temperature
Chlorination pH
Incubation time
TOX
CHC13
CHBrCl2
CHBr2Cl
CHBr3
THM4
MCAA"
DCAA'
TCAA'
MBAA'
DBAA"
BCAA'
TBAA
CDBAA
DCBAA
HAA6
mg/L
mg/L
mg/L
°C
hours
ug/L
ug/L
ug/L
ug/L
ug/L
ug/L
ug/L
ug/L
ug/L
ug/L
ug/L
ug/L
ug/L
ug/L
ug/L
ug/L
* ;
f
-------
Table 5-13 Duplicate Analysis Of Membrane Feed And Permeate Water Quality For Week
Week for which water quality analyses are being duplicated
Utility name and address
ICR plant number
Contact person
Membrane trade name
QP(gpm)[
Phone #
FAX#
1 QF(gpm)| |
Qi (gpm)
Water quality parameter
Sampling date
Sampling time
Alkalinity
Total dissolved solids
Total hardness
Calcium hardness
Bromide
PH
Turbidity
Temperature
Total organic carbon
UV254
Units
MM/DD/YY
hh:mm
mg/L as CaCO3
mg/L
mg/L as CaCO3
mg/L as CaCO3
ug/L
ntu
°C
mg/L
cm"1
CF
. CD
RF
^ ,, _.
: "f
:
s«
s %
RB
s
««
--
*"
f SS f .
r>*>
SDS samples
Chlorine dose
Chlorine residual
Chlorine demand
Chlorination temperature
Chlorination pH
Incubation time
TOX
CHC13
CHBrCl2
CHBr2Cl
CHBr3
THM4
MCAA*
DCAA*
TCAA"
MBAA*
DBAA*
BCAA*
TBAA
CDBAA
DCBAA
HAA6
mg/L
mg/L
mg/L
°C
hours
ug/L
ug/L
ug/L
ug/L
Hg/L
Hg/L
ug/L
ug/L
Hg/L
Ug/L
Hg/L
Hg/L
ug/L
ug/L
ug/L
ligfL
' f f
, , ,
f f
-
-
-
'
-'
..
s
,
_, _,
f f "
,,, ^ ;
^
v
-
--
%
,f
.
-.%%
, ,
,
"
* These six species make up HAA6. The other three HAA species should be reported if measured.
Shaded cells can be calculated by a spreadsheet.
-------
Table 5-14 Concentrate Water Quality Parameters And Mass Balance Closure Errors
Utility name and address
ICR plant number
Phone number
Membrane trade name
Concentrate sample from week 1
Concentrate sample from week 2
Date
CF
CP
Contact person
FAX number
Time
Cc(meas)
Fractional recovery
Ccfcaw I ErrorMB(%)
'" ' "A ,
-,-, -, -
' ,..,
" f ss
i
Date Time Fractional recovery
Parameter
TDS
TOC
UV^
Alkalinity
Total hardness
Calcium hardness
Units
mg/L
mg/L
cm"1
mg CaCO3/L
mg CaCO3/L
mg CaCO3/L
CF
CD
CC(n,ras) | CC(ailc) | Error^ (%)
- >$;-,
f f
'
-
'
', ,
1 1* f
Concentrate sample from week 3
Date
Time
Fractional recovery^
Parameter
TDS
TOC
UV2J4
Alkalinity
Total hardness
Calcium hardness
Units
mg/L
mg/L
cm"1
mg CaCO3/L
mg CaCO3/L
mg CaCO3/L
CF
CD
Ccfmem)
Ccfcalc)
,
'
** f
ErrorMB (%)
f f '
f
yss
'
Concentrate sample from week 4
Date
Time
Fractional recovery^
Parameter
TDS
TOC
UV2M
Alkalinity
Total hardness
Calcium hardness
Units
mg/L
mg/L
cm"1
mg CaCO3/L
mg CaCO3/L
mg CaCO3/L
CF
CD
Ccfmeaj)
CC(caic) I ErrorMB (%)
*
-------
Table 5-15a Blending Calculations To Determine The Fraction Of Flow That Requires NF
Treatment To Meet The Stage I DBF Regulations
Utility name and address
ICR plant number
Membrane trade name
THM4 Controls
HAAS Controls
Phone number
FAX number
Contact person
Parameter
THM4F, ug/L
THM4P, ug/L
HAA5F, ug/L
HAA5P, ng/L
AlkF, mg/L CaCO3
Alkp, mg/L CaCO3
T-HdF, mg/L CaCO3
T-Hdp, mg/L CaCO3
Ca-HdF, mg/L CaCO3
Ca-Hdp, mg/L CaCO3
RUN ID #
1
2
3
4
Qp/QT(THM4),%
Alkb, mg/L CaCO3
T-Hdb, mg/L CaCO3
Ca-Hdb, mg/L CaCO3
THM4b, ug/L
HAA5b, ng/L
--
fff^ s
,
-
ft/ -
S %.
..
" ' '
f f f f_
'.;; . ;
" :
" i
Qp/QT(HAA5),%
Alkb, mg/L CaCO3
T-Hdb, mg/L CaCO3
Ca-Hdb, mg/L CaCOS
THM4b, ngfl,
HAA5b, |ag/L
^j ^ :
^ «^sv - ,v
"
^ ««
. % %
-
"" '
- -
ssv
.. ^
s ss
' ' i
"
i
-
Notes: In the first section of this table, the feed and permeate concentrations are entered for each run.
The spreadsheetuses these values to calculate the flow that must be treated to meet the stage I
DBF MCLs with a 10% factor of safety (i.e.72/54 ug/L - THM4/HAA5). The higher of the two
QP/Q! flow ratios (i.e. permeate flow to total flow) based on either THM4 or HAAS controls the design.
The blended water quality parameters are also calculated for the feed / permeate blends.
If the permeate quality does not meet the MCLs prior to blending, then these calculations are meaningless.
-------
Table 5-lSb Blending Calculations To Determine The Fraction Of Flow That Requires NF
Treatment To Meet The Stage H DBF Regulations
Utility name and address
ICR plant number
Membrane trade name
Phone number
FAX number
Contact person
Parameter
THM4F, jtg/L
THM4D) ng/L
HAA5F, ng/L
HAA5D, ng/L
Alkp, mg/L CaCOa
Alkp, mg/L CaCO3
T-Hdp, mg/L CaCO3
T-Hdp, mg/L CaCO3
Ca-HdF, mg/L CaCO3
Ca-Hdp, mg/L CaCO3
RUN ID #
1
2
3
4
TBDV14 Controls
Qp/QT(THM4),%
Alkb, mg/L CaCO3
T-Hdb,mg/LCaCO3
Ca-Hdb, mg/L CaCOS
THM4b, ng/L
HAA5b, ng/L
*- ':/'! *
\,x rxx'"-"'
" " " *.,', ^
-,\ \ -------c?
"$--^%>>Sr'
: «,,*"*', ^ ^ '
Vi °v v> "* v* s . f
, > s '
, ' f^f^f""
' ".' , '"'", '
'' ftf f tft'tf
', '
-
" *" ' "*
' ' :' '','
'
",- :f ' , ".
' ~:
f jf *^%Vt' ;
' f / X
- w * \
HAAS Controls
Q,/QT(HAA5),%
Alkb, mg/L CaCO3
T-Hdb> mg/L CaCO3
Ca-Hdb, mg/L CaCOS
THM4bJ ^g^L
HAA5b) ng/L
^- *
^> C^ -v -^
v ^^ << -Juv^ j-V
-^J't^n^'
^ ^fC-i^c-
^ ^ ..^v":":":^!^
, .', "* '^ ,., '?""
: : - v;
: ' s f f
~,^'"- " 'V*
; -. "4-'#J
: ' \ ' '
: f t :w.s^:
, ' -A
"
' , ' '
f " :
' ' i
I
ffff '
* :
/ :
Notes: In the first section of this table, the feed and permeate concentrations are entered for each run.
The spreadsheet uses these values to calculate the flow that must be treated to meet the stage II
DBF MCLs with a 10% factor of safety (i.e.36/27 |ig/L - THM4/HAA5). The higher of the two
Qp/Qi flow ratios (i.e. permeate flow to total flow) based on either THM4 or HAA5 controls the design.
The blended water quality parameters are also calculated for the feed / permeate blends.
If the permeate quality does not meet the MCLs prior to blending, then these calculations are meaningless.
-------
6.0 Pilot-Scale Evaluation Of Membrane Processes
6.1 Design Considerations
The design of membrane pilot systems typically follows the same guidelines used hi the
design of full-scale systems, making the results of a pilot study directly scalable to full-scale
operation. Thus, the results of a properly designed pilot-study can be used to design a full-
scale plant; however, factors such as membrane variability and process upsets will lead to
some variation between the performance of full-scale and pilot-scale systems.
The design of membrane pilot systems can vary depending on the objectives of the study
and the availability of resources. A pilot study is also a learning experience, and operating
conditions within the pilot plant will need to be varied to optimize performance. For these
reasons a single, specific protocol for pilot-scale membrane studies will not be provided.
Instead, general guidelines for conducting a pilot-scale study to evaluate membrane
performance will be described in this section. Since this document is rule referenced, it also
contains specific requirements of the pilot-scale membrane studies for compliance with the
ICR.
6.1.1 Minimum Requirements For A Pilot-Scale Study
The objectives of a pilot-scale membrane study are to demonstrate sustained performance
and to develop an estimate of the costs associated with membrane technology used to remove
DBF precursors from source waters. Additional objectives, such as pre-qualification of
membrane products and production of a representative concentrate stream water quality for
evaluating concentrate disposal options may be incorporated based on site-specific
considerations.
The ICR requires that only a single membrane type be investigated during a pilot study,
and for this reason the membrane should be carefully selected. Although not an ICR
requirement for pilot-scale membrane studies, it is recommended that utilities use a bench-
scale procedure to screen at least two membranes prior to the pilot-scale study. A utility
interested hi pursuing membrane technology beyond the requirements of the ICR may want to
consider evaluating additional membrane types hi single element studies parallel to the pilot
system. This is especially important when a utility wants to promote competitive membrane
procurement during bidding of full-scale facilities construction.
The pilot study shall be run continuously over a period of one year, with allowances for
down-tune due to membrane cleaning, maintenance or other reasons. The pilot-scale run tune
shall be no less than 6600 hours which represents approximately 75% of a calendar year. If a
membrane system fails over the one year period, a new membrane run shall be started as soon
as possible after the failure. In general, a failure is defined as (1) an irreversible decrease hi
system productivity to an unacceptable level or (2) an irreversible degradation hi permeate
quality to an unacceptable level when the membrane was originally demonstrated to meet the
treatment objective. Irreversible implies that attempts, primarily chemical cleaning or
3-73
-------
modifications to pretreatment processes, consistent with the manufacturer's recommendations
were made to correct the problem but were unsuccessful.
The membrane pilot study should be redesigned to prevent the second run from failing, and
the same membrane type used in the first study should be used in the second study unless it
can be demonstrated that the membrane type was responsible for failure of the first run.
^ Flexibility will be permitted in the design of a membrane pilot system so that it can be
tailored to the specific objectives of a utility; however, certain minimum requirements must be
imposed to insure that the objectives of the ICR are met. The pilot system shall consist of
standard spiral-wound elements no smaller than 2.5" diameter by 40" long. The pilot plant
must, as a minimum configuration, be composed of a two stage array with two pressure
vessels in the first stage and one pressure vessel in the second stage. Each pressure vessel
shall contain at least three spiral-wound membrane elements. This would require at least nine
membrane elements for the pilot system. The pilot plant shall be designed to achieve a system
recovery of at least 75% while meeting membrane manufacturer's specifications for pressure,
flux, minimum and maximum influent flow rates, etc. Concentrate recycle can be used to
meet the minimum membrane element flow specification for pilot systems. If high recoveries
are feasible for a given water supply, (80 to 95%), serious consideration should be given to
the use of additional stages or concentrate recycle to achieve these recoveries during piloting.
Such operation will provide greater confidence with respect to achieving water quality
objectives. Membrane manufacturers also specify a maximum feed flow rate to be applied to a
single element. Exceeding this maximum flow rate may result in damage to an element or
seals which could result in a deterioration hi performance. Any design that meets these
requirements can be used for the ICR.
Hollow fiber membranes are also used in NF applications, but are less common and not
recommended for surface waters. If hollow-fiber technology is being investigated for the ICR,
then standard-production, hollow-fiber elements should be used hi the pilot system. The
system must be staged to achieve a system recovery of at least 75% while being operated at a
pressure, cross-flow velocity and feed flow rate within the manufacturer's specifications.
Finally, there are specific sampling and analytical requirements that must be met for the
ICR pilot studies which are described in Section 6.3.
6.1.2 Pilot Plant Description
Pilot membrane systems are typically designed as a staged array of elements similar to the
design of full-scale membrane plants. Staging is used to increase the system recovery by
feeding the concentrate from previous stages to downstream stages. A minimum cross-flow
velocity must be maintained through all stages in order to minimize concentration polarization
and inorganic precipitation at the membrane surface. Since some of the flow exits the system
as permeate in upstream stages, the area available for flow must be decreased in downstream
stages to maintain the desired cross-flow velocity. This is accomplished by reducing the
number of parallel pressure vessels hi downstream stages. Since the cross-sectional area
available for flow is difficult to quantify for membrane elements, manufacturers typically
3-74
-------
specify a minimum feed flow to a single element instead of a minimum cross-flow velocity.
The feed flow rate to the last element in a pressure vessel in a pressure vessel must be greater
than this minimum flow rate to a single element specified by the manufacturer.
The minimum acceptable feed flow per element is directly related to the specified recovery
and flux for a single element. For a 4" x 40" element the specified single-element-recovery
can vary from 5% to 20%, and a typical flux is 15 gfd for low fouling waters and may be
reduced to less than 10 gfd for high fouling waters. Assuming a membrane area of 70 ft2, a
permeate flow of 0.73 gpm, and a 15% single-element-recovery, the feed flow rate is 4.9 gpm
and the concentrate flow is 4.2 gpm per element. Concentrate recycle may be used to increase
the recovery while maintaining an acceptable flow rate through the system.
Figure 6-1 is a flow diagram of a two-stage membrane pilot plant which could be used to
meet the ICR requirements for a pilot-scale membrane system. This flow diagram includes a
numbering system to identify selected monitoring and/or sampling points, and Table 6-1
presents a description of these monitoring sites. This particular pilot plant design is skid
mounted and occupies a space of approximately 5 ft tall x 4 ft wide x 14 ft long, and a single
electrical cable provides the necessary power from a 220 volt, three phase source.
A low pressure centrifugal pump is used to increase the feed stream pressure to
approximately 40 psi hi order to maintain sufficient pressure through the cartridge filters to
prevent high pressure pump cavitation. The chemicals used for pretreatment are added
between the feed water pump and the prefilter. Mixing is provided by the prefilter and static
mixers. From the prefilter the feed stream is directed to a high pressure pump, such as a
multi-stage centrifugal pump, which provides the necessary pressure for the membrane
process.
The two-stage membrane system consists of three pressure vessels in a 2-1 array. Pressure
vessels one and two are in the first stage and pressure vessel three is the second stage. Each
pressure vessel houses three 4" x 40" spiral-wound elements. For this pilot plant the total
number of membrane elements used is nine and the total membrane area is 630 ft2 at 70 ft2 per
element.
The pressurized influent stream is directed through a manifold to pressure vessels one and
two hi the first stage. The permeate water is collected hi the connected product water tubes of
each element and routed to a permeate water collection manifold at the end of the pressure
vessel. The concentrate from pressure vessels one and two of stage one is combined and
routed to the influent end of pressure vessel three in stage two. The permeate water from
stage two is routed to the permeate collection manifold and combined with the permeate from
stage one. Most systems typically use one flow meter and one valve to measure and control
the combined permeate stream from stage one. However, valves or flow controlling devices
may be required on the permeate discharge line from each pressure vessel of the first stage in
order to balance the permeate flows and maintain equal fluxes hi each pressure vessel. If there
are no anomalies between the pressure vessel configuration or individual elements of stage
one, then the differences between the parallel pressure vessels should be insignificant. After
3-75
-------
stage 2 the concentrate waste is routed through a valve and flow meter before being discharged
as waste. The ability to recycle a portion of the concentrate back to the suction side of the
high pressure pump is included hi this design and may be necessary to maintain the desired
flow rate through all membrane elements.
The system utilizes an electrical control panel which monitors feed pressure, TDS of the
feed and permeate streams, feed temperature and hours of high pressure pump operation. If
the pressure between the prefilter and the high pressure pump drops to a level of
approximately 10 psi, the system will automatically shut down to protect the high pressure
pump. In addition, a pH monitor and chemical feed pump controller can be added to control
the influent pH and shut-down the system to prevent damage to the membranes if chemical
feed is interrupted.
In some cases excessive hydraulic losses through the mesh feed spacers and system
plumbing result in an excessive decrease in the feed pressure through a system. This results in
a lower permeate flux and an under utilization of available membrane area in subsequent
stages. In order to maintain the design permeate flux throughout a system, inter-stage
pumping may be required. Pumps can be placed between stages to boost the pressure of the
influent stream to the next stage. Booster pumps must have a capacity that can handle the total
concentrate flow from the upstream stage while providing the desired differential pressure.
Inter-stage pumping also provides the flexibility to recycle around an individual stage.
Major problems associated with the design of membrane systems are excessive flux or
permeate production in the first stage, inadequate inter-stage feed pressure, and inadequate
feed flow to the last element hi a pressure vessel. Excessive flux.or permeate flow from a
stage can be controlled by incorporating valves on the permeate lines from the stage or
pressure vessels. This restricts the flow, creates a back-pressure and allows the flux for each
stage to be controlled, there-by balancing the stage production. If permeate back-pressure is
used, it must be measured and used to calculate the net driving pressure. Inadequate inter-
stage feed pressure can be increased with the use of niter-stage booster pumps. Inadequate
flow to the last element in any stage or pressure vessel can be increased by incorporating
concentrate recycle around the system or around a single stage to increase the influent flow
rate.
6.1.3 Sampling And Monitoring Locations
The following section summarizes the locations and frequencies at which selected operating
and water quality parameters should be monitored for pilot-scale membrane studies. Section
6.3 provides an explanation of the monitoring requirements as well as formats for recording
and reporting the data. Figure 6-1 and Table 6-1 describe the location of selected monitoring
sites for pilot-scale membrane studies. Table 6-2 presents a matrix showing the location and
frequency at which operating and water quality parameters must be monitored for pilot-scale
studies, and Table 6-3 can be used to record daily readings of flow rates, pressures,
temperatures, pH and TDS concentrations.
3-76
-------
The sampling and monitoring requirements described in this section are only intended to
meet the requirements of the ICR and may not be adequate to develop .the information
necessary for the design of a full-scale plant. For example, if recoveries higher than 75% are
achievable for a given water source, then the design recovery proposed for a full-scale plant is
suggested for use hi the pilot-scale study in order to produce the most representative
concentrate and permeate water quality data. This is of critical importance when investigating
concentrate water quality in order to assess concentrate disposal options. In addition,
permeate water quality parameters of particular concern to the treatment objectives of a
specific utility are suggested for consideration hi the membrane investigation.
Temperature and pH
Temperature and pH measurements must be monitored at least once per shift each day of
operation. The temperature should be measured using an in-line temperature gauge located at
the influent to the system (i.e., location #4 hi Figure 6-1). The pH of the feed water, the
pretreated feed stream, and the system concentrate waste stream should also be measured as
part of the shift monitoring.
Flow rates
Flow rates are measured by reading in-line flow meters at the following points: the
permeate flow from each stage, the permeate flow from the system, the concentrate waste flow
rate from the system, and the concentrate recycle flow rate. These monitoring points are listed
as locations 12, 13, 14, 15 and 16 hi Table 6-2. All other flow rates in the system can be
determined from these flow measurements. The system permeate and system waste flows
should also be measured by manually timed volumetric displacement. These manual
measurements are used to verify and calibrate the flow meters.
Additional flow meters can be placed in the pilot plant to monitor the performance of
individual pressure vessels if desired. For the pilot plant shown hi Figure 6-1 and described hi
Section 6.1.2, panel mounted flow meters with an approximate range of 2 to 20 gpm were
used to measure the influent flow rate to each pressure vessel. Flow meters with a range of 1
to 10 gpm were used to measure the concentrate flows from each pressure vessel and the
system permeate and concentrate recycle flow rates. Flow meters with a range of
approximately 0.5 to 5 gpm were used to measure permeate flow from each pressure vessel
and the concentrate waste flow rate from the system. The ranges of the flow meters will vary
with the size and type of membranes.
Pressure
The locations and frequency of pressure measurement are also shown in Table 6-2. The
gauges should have a range of approximately 0 to 200 psi for influent and concentrate lines
and 0 to 60 psi for permeate lines. The gauges should be liquid filled and calibrated before
installation onto a common front panel of the pilot plant.
Hours of operation
The membrane pilot plant should be designed to operate on a continuous basis. The actual
hours of operation for the unit are recorded by an hour meter mounted on the front panel,
3-77
-------
which should record only when the pilot plant is operating. This is accomplished by tying the
hour meter operation to the high pressure pump motor's operation. If the pilot plant is shut-
down due to mechanical failure or cleaning, the down-time should be recorded on a daily
operating log sheet similar to the one presented hi Table 6-3, along with a description of the
event. The tune to complete any maintenance as well as the restart tune should also be
recorded.
6.1.4 Design Calculations
The design of the membrane pilot plant shown hi Figure 6-1 is based on criteria presented
in Table 6-4.
The first step in determining the operating pressure and feed stream flow of a membrane
pilot plant is to determine the permeate produced by each element by using the flux and
membrane area as shown hi Equation 6.1.
Q^ = ₯wx A£ (6.1)
where Q,^ is the permeate flow rate produced by a single element, Fw is the design flux and
AC is the area of a single element.
The permeate flow from each pressure vessel can be determined by multiplying the number
of elements per pressure vessel by the permeate flow produced by a single element using
Equation 6.2.
QP-V = Ne X Q^ (6.2)
where Qp.v is the permeate flow rate produced by a pressure vessel and Ne is the number of
elements hi one pressure vessel.
The permeate flow from each stage is calculated by multiplying the number of pressure
vessels in each stage by the flow produced by a single pressure vessel using Equation 6.3.
QP-S = Nv x Qp_v (6.3)
where Qp^ is the permeate flow rate produced by a stage and Nv is the number of pressure
vessel hi the stage.
The stage permeate flows can be added together to calculate the permeate flow produced
by the system using Equation 6.4.
Qp-sys = Qp-s(l) + Qp-s(2) + ... + Qp.s(n) (6.4)
where Q^s is the permeate flow rate produced by the system, which is equal to the sum of the
permeate flows from all stages hi a system with n stages, Qp.s(i) is the permeate flow rate
3-78
-------
produced by stage one, Qp.s(2) is the permeate flow rate produced by stage two, and Qp.s(n) is
the permeate flow rate produced by stage n.
The feed flow rate to a pressure vessel, a stage, or the system can now be estimated by
dividing the associated permeate flow by the associated recovery as shown in Equation 6.5.
where QM is the feed stream flow to either a pressure vessel (QF.V), a stage (QF.S), or the
system (QF_sys); Qp.# is the permeate flow produced by the pressure vessel (Qp.v), stage (Qp.s) or
system (Qp.^) and R# is the fractional recovery for either the pressure vessel (RJ, stage
or system
The concentrate waste flow rate from the system is calculated as the difference between the
feed flow rate to the system and the permeate flow produced by the system as shown in
Equation 6.6.
Qw-sys = Qp-sys " Qp-sys (6.6)
where Qw.sys is the concentrate waste flow rate from the system.
The manufacturer's recommended minimum and maximum element flow rates can be used
to determine if recycle is necessary to maintain the minimum required flow rate through the
final element hi a pressure vessel, or if the feed flow rate to the front end of a pressure vessel
must be reduced. In order to determine if recycle is required, the feed flow rate to the last
element in each stage must be calculated according to Equation 6.7. The last element is used
because this element experiences the lowest feed flow hi a pressure vessel.
(QF-s)end = QF-S - Nv X Qp^ x (Ne - 1) (6.7)
where (QpJend is the feed flow rate to the last element in a stage, NT is the number of pressure
vessels in a stage, QF.S is the feed flow rate to the respective stage, Ne is the number of
elements in the respective pressure vessel, and Qp^ is the permeate flow produced per element.
The mhiimum required recycle flow for a stage is determined from the manufacturer's
recommended minimum flow rate and the flow entering the last element hi a stage as shown hi
Equation 6.8. If calculated recycle flow rate is negative, then the flow into the last element is
greater than the minimum required flow and recycle is not required. This calculation must be
made for each stage hi the system, and the recycle flow rate used hi the system must provide
adequate flow hi all stages (i.e., the system recycle flow rate must be equal to or greater than
the largest minimum recycle flow rate calculated according to Equation 6.8). The most critical
element is typically the last element in the first stage.
3-79
-------
(QR-^min = Nv X (Qp.^ - (Qi-Xd (6.8)
where (QRJm[n is the minimum required recycle flow rate for a stage and (QFJrain is the
manufacturer's recommended minimum feed flow rate to a single element.
Once an acceptable system recycle flow rate has been calculated, the influent flow rate to a
pressure vessel, stage or the system can be determined by adding the corresponding feed flow
rate, calculated from Equation 6.5, to the recycle flow rate as shown in Equation 6.9.
QI-* = QF-# + Qn-sys (6.9)
where QM is the influent flow rate to a pressure vessel (Qj_v), a stage (Q,.s) or the system
If recycle is not used, the influent flow rate is equal to the feed flow rate.
The flow rate entering each stage must be compared to the maximum allowable feed flow
rate for each stage, which is calculated from the manufacturer's recommended maximum flow
rate to an element and the number of pressure vessels in a stage using Equation 6. 10. If the
influent flow rate entering any stage exceeds the maximum allowable flow rate for that stage
then the influent flow rate must be reduced.
(QlXax = (QF-e)maxXNv (6.10)
where (Qr.^max is the maximum allowable influent flow rate to a stage and (QF_e)max is the
maximum allowable feed flow rate to a single element as specified by the manufacturer.
The maximum recycle flow rate for each stage can be calculated as the difference between
the maximum allowable influent flow rate to the stage and the feed flow rate to the stage using
Equation 6.11. The maximum allowable flow rates must be calculated for each stage and the
lowest value is the maximum allowable recycle flow rate for the system.
x = (QlJmax ' QF-S (6.11)
where (QR^)max is the maximum allowable recycle flow rate for a stage.
The concentrate flow rate from a pressure vessel, a stage or system is calculated from the
corresponding influent flow rate, the corresponding permeate flow rate and the system recycle
flow rate as shown in Equation 6.12.
Qc-# = QF-# + Qn-sys - QP-# (6.12)
where Qc^, is the concentrate flow rate from either a pressure vessel (Qc.v), a stage (Qc.s) or
the system (
The recycle ratio is calculated as the ratio of the recycle flow rate to the feed flow rate
according to Equation 6.13.
3-80
-------
r = QR-sys / QF-sys (6.13)
where r is the recycle ratio.
The required feed stream pressure can be determined once the osmotic pressure gradient
for each stage is estimated and hydraulic losses through stage hardware and membrane
elements are accounted for. The osmotic pressure gradient is estimated from the TDS of the
influent, waste and permeate streams as shown in Equation 6.14.
ATCS = [((TDSJ + TDSc) / 2) - TDSJ X 0.01 (6.14)
where A7ts is an estimate of the average osmotic pressure gradient for a stage in psi, TDSi is
the influent TDS concentration in mg/L, TDSC is the concentrate TDS concentration in mg/L,
TDSp is the permeate TDS concentration hi mg/L and 0.01 is an approximation factor for
converting TDS (mg/L) to pressure (psi).
To determine the approximate osmotic pressure for each stage, the TDS of all flow streams
must be predicted, using a manufacturer reported TDS rejection. This method uses a
sequential iterative mass balance approach around each element of a stage within a system.
Appendix 3-A presents the equations used to predict the TDS concentration of the influent,
permeate and concentrate for each stage. These equations incorporate recycle around the
system (not individual stages), recovery and the manufacturer reported TDS rejection. It
assumes an equal number of elements per pressure vessel with constant and equal per element
flux.
Since the equations presented hi Appendix 3-A can be cumbersome, they were solved for a
typical pilot system that could be used in the ICR (i.e., a two stage system with three elements
per pressure vessel operated at 75% recovery). The resulting osmotic pressures are presented
hi Table 6-5. This table presents osmotic pressures for each stage of a two stage system for
five different recycle ratios, seven different manufacturer reported TDS rejections and four
different feed TDS concentrations. To use this table, select a recycle ratio, manufacturer
reported TDS rejection and feed TDS concentration representative of the testing conditions to
be used during the study. Next, use these parameters to read the stage one and stage two
osmotic pressures from Table 6-5.
The results presented in this table were calculated for a two stage system similar to the one
described hi Section 6.1.2. Systems that differ significantly from the 2-1 array described hi
this section may need to use the equations is Appendix 3-A or a manufacturer's computer
program to estimate the osmotic pressure for use in the design calculations.
Once the osmotic pressure has been estimated, the feed stream pressure that must be
supplied by the high pressure pump can be estimated. The flux equation shown hi Equation
6.15 can be rearranged to solve for the required feed stream pressure as shown hi Equation
6.16. In words Equation 6.16 is: (the required system feed pressure) is equal to (the pressure
required for permeation) plus (the pressure of the permeate stream) plus (pressure losses
3-81
-------
through the membranes and stage hardware) plus (the osmotic pressure gradient). The losses
through the elements and stage piping cumulate throughout the system; therefore, the losses in
each stage must be a summation of all losses in the preceding stages in addition to the losses in
mat stage.
J?w = MTCW x fr, x [{(2 x PF - (2 x i- l)x L)/2}- Pp - ATI.] (6.15)
PF =
x £r' x (i ~ °*5)) + (gri
T| = Qp-sO/Qp-sys (6.17)
L = APS + NexAPe (6.18)
where i is the stage number in a system consisting of n stages; r, is the stage flow weighted
factor (defined in Equation 6.17 as the permeate flow produced by stage i divided the permeate
flow produced by the system); PF is the required system feed pressure; Pp is the system
permeate pressure; AT^ is the osmotic pressure gradient associated with stage i; L is the
pressure loss term associated with the stage hardware and elements; APS is pressure loss term
associated with stage hardware; APe is pressure loss term associated with membrane elements;
and Nc is the number of elements in a pressure vessel.
6.1.5 Design Example
A design for a nanofiltration pilot plant is shown to clarify system design procedure
described in Section 6.1.4. Basic assumptions used in this design are shown hi Table 6-6.
Note the recovery per stage is based on the influent flow to and permeate flow from each
individual stage.
The permeate produced per element, pressure vessel, stage and by the system can be
determined from the average flux, number of elements per pressure vessel and the system
configuration using Equations 6.1 through 6.4.
Equation 6.1 is used to calculate the permeate flow produced by a single element.
QP-C = FW X Ae = 15 gfd X 70 ft2 = 1,050 gpd per element
Equation 6.2 is used to calculate the permeate flow produced by a single pressure vessel.
QP-V = Ne x Qp.e = 3 X 1,050 gpd = 3,150 gpd per pressure vessel
Equation 6.3 is used to calculate the permeate flow produced by each stage.
Qp-S(D = Nv(i, X Qp.v = 2 x 3,150 gpd = 6,300 gpd from stage one
3-82
-------
QP-S (2) = Nv(2) x Qp.v = 1 x 3,150 gpd = 3,150 gpd from stage two
Equation 6.4 is used to calculate the permeate flow produced by the system.
Qp-sys = Qp-s(i) + Qp-s(2) = 6,300 gpd + 3,150 gpd = 9,450 gpd
Equation 6.5 is used to calculate the feed flow to the system based on a design system
recovery of 75 %.
Qp-sys = Qp-sys / Rsys = 9,450 gpd / 0.75 = 12,600 gpd
Equation 6.6 is used to calculate the concentrate waste flow rate from the system, and it is
this waste flow rate along with the system permeate flow rate that controls the recovery of the
system.
Qw-sys = Qp-sys - Qp-sys = 12,600 gpd - 9,450 gpd = 3,150 gpd
The feed flow rate to the first stage is identical to the feed flow rate to the system.
,ys = 12,600 gpd
QF-S(D - Q
The feed flow rate to the second stage is calculated using Equation 6.5 (or alternatively,
the difference between the feed flow to the first stage and the permeate flow from the first
stage).
as
QF-S (2) = Qp-s(2) / RS (2) = 3,150 gpd / 0.50 = 6,300 gpd
In order to determine if there is adequate feed flow going to each element in a stage, the
feed flow rate to the last element in each stage must be calculated using Equation 6.7.
The feed flow rate to the last element in each of the two parallel pressure vessels of stage
one is calculated as:
(QF-s(i))end = QF-S(D - Nv (1) x Qp.e x (Ne (1) -1) = 12,600 gpd - 2X1,050 gpd x (3-1)
= 8,400 gpd
The feed flow rate to the last element hi the single pressure vessel of stage two is
calculated as:
(Qp-s (2))end = QF-S (2) - Nv (2) XQp.e X (Ne (2) - 1) = 6,300 gpd -1x1,050 gpd X (3-1)
= 4,200 gpd
Equation 6.8 can be used to determine the minimum required concentrate recycle flow
rate. Recycle will be required if the flow rate to the last element in any pressure vessel is less
3-83
-------
than the minimum flow rate to a single element, specified by the manufacturer as 4 gpm or
5,760 in this example.
The minimum requked recycle flow rate for stage one is calculated as:
(QR-S(i))min = N^yXCQpJ^ -(QF-s(i))end = 2x5,760 gpd - 8,400 gpd = 3,120 gpd
The minimum requked recycle flow rate for stage two is calculated as:
(QR.s(2))min = Nv(2)X(QFJmin-(Qp.s(2))end = 1X5,760 gpd-4,200 gpd = 1,560 gpd
Since the recycle flow rates calculated according to Equation 6.8 are positive, concentrate
recycle is requked. Furthermore, the recycle flow rate requked for stage one is greater than
the recycle flow rate requked for stage two, and thus controls the design. The minimum
recycle flow rate for the system is:
(QR-sys)min = 3,120 gpd
The minimum influent flow rate to the system, which is equal to the influent flow rate to
stage one, is calculated by adding the feed flow rate to the minimum recycle flow rate for the
system using Equation 6.9.
Ql-,ys = Ql-s(1) = QF-sys + (QR-sys)mm = 12,600 gpd + 3,120 gpd = 15,720 gpd
The corresponding influent flow rate to stage two can also be calculated using Equation
6.9.
Ql-s(2) = QF-s(2) + (QR-sys)min = 6,300 gpd + 3,120 gpd = 9,420 gpd
The maximum allowable influent flow rate to each stage must be calculated using Equation
6.10 and the maximum allowable flow rate to a single element, specified by the manufacturer
as 16 gpm or 23,040 in this example.
The maximum allowable influent flow rate to stage one is:
(Qi-s(i))max = (Qp.e)maxXNv(1) = 23,040 gpdx2 = 46,080
The maximum allowable influent flow rate to stage two is:
(Ql-s(2))max = (QF-e)maxXNv(2) = 23,040 gpdXl = 23,040
Next, the influent flow rate to each stage must be compared to the maximum allowable
flow rate to each stage.
Qi-s(i) = 15,720 gpd < (QF.s(1))max = 46,080 gpd
3-84
-------
Qi-s(2) = 9,420 gpd < (QF.s(2))max =23,040 gpd
Since the influent flow rates to each stage are below the corresponding maximum allowable
flow rates to each stage, this design is acceptable.
In some cases, a recycle flow rate larger than the minimum recycle flow rate for the
system, (QR.sys)min, may be desired to mmimize fouling. The maximum allowable recycle flow
rate for the system must be determined by calculating the maximum allowable recycle flow
rates for each stage according to Equation 6.11.
(QR-s(i))max = (Qi-s (i))max - QF-S (i) = 46,080 gpd - 12,600 gpd = 33,480 gpd
(QR-s(2))max = (Qi-s (2))max - QF-S a) = 23,040 gpd - 6,300 gpd = 16,740 gpd
The maximum allowable recycle flow rate for stage two is lower than that for stage one.
Thus, the maximum allowable recycle flow rate for the system is the maximum recycle flow
rate for stage two, 16,740 gpd, since a recycle flow rate greater than this value would result in
a flow rate in stage two outside of the manufacturer's specifications.
The recycle ratio can calculated for both the maximum and minimum recycle flow
requirements hi this example using Equation 6. 13.
s = 16,740 gpd / 12,600 gpd = 1.33
fmin = (QR-sysWQp-sys = 3,120 gpd / 12,600 gpd = 0.25
The minimum concentrate flow rates from each stage and the system can be calculated
using Equation 6.12 using the minimum system recycle flow rate, 3,120 gpd:
Qc-s(i) = QF-S(I) + (QR-sys)mm - Qp-s a) = 12,600 gpd + 3,120 gpd - 6,300 gpd = 9,420 gpd
Qc-s(2) = QF-S (2) + (QR-sys)min - Qp-s (2) = 6,300 gpd + 3,120 gpd - 3,150 gpd = 6,270 gpd
Qc-syS = Qp-sys + (QR-sys)min - Qp-sys = 12,600 gpd + 3,120 gpd - 9,450 gpd = 6,270 gpd
The maximum concentrate flow rates are calculated using Equation 6. 12 and the maximum
system recycle flow rate, 16,740 gpd:
Qc-S(i) = Qp-s(i) + (QR-sys)max - QP-S (i) = 12,600 gpd + 16,740 gpd - 6,300 gpd = 23,040 gpd
Qc-s(2) = QF-S (2) + (QR-Sys)max - QP-s (2) = 6,300 gpd + 16,740 gpd - 3,150 gpd = 19,890 gpd
Qc-sys = Qp-sys + (QR-sys)max - QP-sys = 12,600 gpd + 16,740 gpd - 9,450 gpd = 19,890 gpd
3-85
-------
Note that the concentrate flow rate from the system is always equal to the sum of the
concentrate waste flow rate and the concentrate recycle flow rate.
For this design example, the feed flow rate is 12,600 gpd with a minimum recycle needed
of 3,120 gpd producing a minimum influent flow to the system of 15,720 gpd. If higher
recycle flow rates are to be tested, the maximum concentrate flow that should be recycled is
16,740 gpd, which corresponds to a total maximum influent flow to the system of 29,340 gpd.
Once the system flows have been calculated, the pressure requirements can be determined
for the membrane system using the basic flux equation on a flow weighted basis and
accounting for the entrance and exit pressure losses, element pressure losses and osmotic
pressure.
Using a feed TDS of 300 mg/L, an element TDS rejection of 0.70, a recovery of 75 % and
a recycle ratio of 0.25 the approximate osmotic pressure for each stage was selected from
Table 6-5. The values for the osmotic pressure can be interpolated between the recycle ratios
of 0 and 1 and rounded to the nearest 1 psi. For stage one the AT^ is 3 psi and for stage two
the A,itz is 5 psi. Equations 6.16 through 6.18 are used to calculate the pressure required to
drive the membrane process.
Using Equation 6.18 the mechanical losses are calculated:
L = APS + NeXAPe = 5 psi + 3 X3 psi = 14 psi
Using Equation 6.17 the stage flow weighted factor for stage one and two are calculated as
0.67 and 0.33, respectively:
I\ = QP,s(i) / Qp-sys = 6,300 gpd / 9,450 gpd = 0.67
T2 = Qp,s(2) / Qp-sys = 3,150 gpd / 9,450 gpd = 0.33
Using Equation 6.16 the feed pressure is calculated by breaking up the equation for clarity,
solving for each part and then solving for PF. A permeate stream pressure of 30 psi will be
assumed for this example.
PF = (Fw I MTCW)+ (Pp)+ [L x EFj x (i - 0.5)) + [zFi x Arc^
= (15 gfd / 0.30 gfd/psi) = 50 psi
PP = 30 psi
3-86
-------
L x t r, x (i - 0.5) = 14 psi x [0.67 x (1 - 0.5) + 0.33 x (2 - 0.5)] = 11.6 psi
, x ATT; = (0.67 x 3 psi) + (0.33 x 5 psi) = 3.7 psi
PF = 50 psi + 30 psi + 11.6 psi +3.7 psi = 95 psi
All of the parameters necessary for a system design have been calculated or estimated, and
a summary of this design is presented in Table 6-7. The flows can be used to size flow meters
and system plumbing, and the pressures can be used to select pressure gauges. When sizing
the flow meters for the stage influent and stage concentrate lines, the flow contribution from
the recycle stream must be accounted for. In this design the high pressure pump should be
selected so that optimum operation occurs at approximately 95 psi at a flow rate of
approximately 15,7200 to 29,340 gpd. However, the pump should be capable of operating
over a range of pressures from 80 to 110 psi so that the pressure can be varied to maintain a
constant flux during operation.
6.1.6 Pretreatment
Pretreatment for membrane processes commonly includes chemical addition to prevent
inorganic precipitation and cartridge filtration to reduce colloidal fouling. The most common
chemical pretreatment is sulfuric acid addition to reduce the pH to prevent calcium carbonate
precipitation. In some cases hydrochloric acid is substituted for sulfuric acid. The potential
for calcium carbonate precipitation can be checked by calculating the Langlier Saturation Index
for the concentrate stream. Some applications may require the addition of a chemical
antiscalant to control such inorganic precipitants as calcium sulfate, barium sulfate or
strontium sulfate. The use of an antiscalant can often eliminate the need for acid addition all
together, and many antiscalants can prevent scaling hi systems with a concentrate LSI <. +1.5.
The required chemical doses are determined by limiting salt calculations or manufacturer
computer programs, both of which typically require a preliminary and comprehensive chemical
analysis of the raw water. Additionally, the fouling potential of the feed water should be
evaluated using one or more of the methods presented hi Section 2.5.
Multi-stage pilot systems with fouling indices within an acceptable range should require
minimal pretreatment equipment. Generally, concentrated sulfuric acid is pumped through an
injection port connected directly to the feed water line between the feed water pump and the
prefilter. The injection point for pretreatment chemicals is typically chosen so the prefilter
and booster pump will assist in mixing the chemicals with the feed water, as shown in Figure
6-1. In some cases an in-line static mixer may be used to insure proper mixing. A chemical
feed pump made of resistant materials is used to inject the acid from storage barrels. If
3-87
-------
antiscalants are evaluated, they are generally highly concentrated and must be diluted into 30
to 50 gallon containers before injection into the feed water stream.
After chemical addition, the water is passed through a cartridge filter to remove larger
suspended solids or colloidal material and protect the booster pump and membrane elements
from sand or other foreign materials. Polypropylene cartridge filters with a size exclusion of 5
jim are acceptable for membrane pretreatment. The filter cartridges should be cleaned or
replaced when the pressure drop across the prefilters increases by a predetermined percentage
(e.g., 50%).
For feed waters which have excessive fouling as indicated by fouling indices or other
factors, advanced pretreatment will need to be incorporated into the system. Advanced
pretreatment may include enhanced coagulation and sand filtration with reduced pH. An
example of this would be operating the membrane system using water from a conventional
treatment plant after sand filtration and prior to the addition of any oxidant or disinfectant. If
alum is used as the coagulant, the pH of the feed water may need to be reduced to around 4.5
(or just above the minimum operational pH as specified by the manufacturer) to ensure the
solubility of aluminum hydroxide. However, operating at a low pH may increase fouling by
high molecular weight organic matter. Additional advanced pretreatment schemes may include
microfiltration to control particulate or microbial fouling.
6.2 System Operation
6.2.1 System Start-up
If the proper sequence is not followed during start-up of a membrane system, the
membranes could be damaged by overfeeding, over pressurizing, or hydraulic shock.
Although the specific start-up procedure will vary from system to system, a typical
initialization and start-up sequence is provided hi this section.
1. Select a membrane type and obtain the appropriate element size and the manufacturer's
specification sheet for that element.
2. Conduct a thorough analysis of the source water to evaluate the inorganic chemical
matrix and determine limiting salts by calculation or manufacturer computer program.
3. Conduct fouling index tests on the feed water, calculate indices and determine the
potential for fouling problems. If a fouling problem is indicated, additional
pretreatment alternatives may need to be evaluated.
4. Select a membrane flux rate consistent with the fouling potential of the water and
recommendations of the manufacturer.
5. Calculate or consult manufacturer computer programs for feed water acid and/or
antiscalant dose. Calculate the dilution needed for the chemical feed tank capacity and
the chemical pump feed rate.
3-88
-------
6. Load the elements into the pressure vessels, coupling elements in a common pressure
vessel, and carefully secure the end caps.
7. Flush the influent line upstream of the first stage of the membrane system, but down-
stream of any pretreatment processes to remove debris and other contaminants.
Flushing should be continued for several minutes.
8. Connect the feed line of membrane unit to a pressurized (20 to 40 psi) source water
transmission line and open feed valve to allow water to enter the membrane prefilter.
9. Open all valves, including permeate, concentrate waste, concentrate recycle, and pump
recirculate valves. This is to allow water to flow through the system, displace any
trapped air and flow to waste upon start-up.
10. Verify that the feed water is at the prefilter and release trapped air by depressing the
bleed button located at the top of the prefilter (if available).
11. Verify that there are proper and secure connections for the chemical feed system, and
energize the chemical feed pump. If the unit is going to be operated unattended for
significant periods of time, some consideration should be given to controls to shut-
down the unit if chemical dosing is interrupted.
12. Allow water to flow through the system to waste in order to purge trapped air from the
system. When all of the air has been purged from the system, slowly increase the
system pressure by opening the feed valve and turn on the high pressure pump. Adjust
the concentrate waste valve so that the concentrate flow does not exceed the
manufacturer's maximum recommended flow rate per element.
13. Continue to open the feed valve to increase the system pressure and feed flow rate until
the design permeate flux is achieved.
14. Making sure that the recirculate valves on the inter-stage pumps are open, turn on the
inter-stage pumps.
15. Slowly close the concentrate waste valve on the final stage until the desired concentrate
waste flow rate is achieved while making sure not to overpressurize the first stage.
16. While monitoring the pressure, slowly close the concentrate recycle valve (if used) to
obtain the proper recycle flow rate.
17. Adjust the recirculate valves on the inter-stage pumps to achieve the desired stage
pressures and fluxes.
3-89
-------
19.
20.
Repeat steps 13, 15, 16 and 17 as necessary until the desired flows and fluxes are
achieved.
Confirm the proper feed rate for acid by measuring pH or antiscalant by calibrating the
chemical pump feed rate.
After one hour of operation, take an initial set of readings (i.e., flow rates, pressures,
influent temperature, pH and TDS) to insure that the system is performing according to
specifications.
6.2.2 System Shut-down
The following section will describe the general steps to follow during shut-down of a pilot
system. Pilot-scale studies should be operated continuously; however, the system will need to
be periodically shut-down for cleaning and other maintenance. The system may be shut-down
for brief periods (i.e., not to exceed 48 hours) if it cannot be monitored or if the feed stream
from the plant needs to be shut-off. If unit operation is interrupted for more than 24 hours,
the system should be flushed with feed water once per day for 30 minutes to minimize
biological growth.
1. Record the system readings and collect any samples necessary.
2. Collect 50 to 500 gallons of permeate to use for membrane preservation and cleaning
solutions.
3. Open the recirculate valves on the niter-stage pumps and turn off the pumps.
4. Open the concentrate waste, recycle and permeate valves.
5. Slowly close the feed valve, and turn off the high pressure pump before completely
closing the feed valve.
6. Turn the chemical feed pumps to the off position.
7. Close the valve from the feed water source.
8. If the system is to be shut-down for more than 24 hours but less than one week, the
previously collected permeate water should be used to flush the feed/concentrate side of
the membrane. If the membrane is to be shut-down for longer than one week, it should
be preserved according to the manufacturer's recommended procedure.
6.2.3 Membrane Cleaning And Preservation
The following section will provide general procedures to be followed during the cleaning
and preservation of membranes used during the operation of a membrane pilot plant.
Membranes are usually cleaned when the temperature-normalized MTCW has decreased by 10
to 15% from the baseline at the beginning of the study or the baseline established after the
3-90
-------
most recent cleaning. Membranes should also be cleaned if the pressure drop across a
pressure vessel increases by more than 20%. It should be noted that the MTCW has to be
normalized for temperature, since a drop in temperature will cause an increase in the net
driving pressure required to maintain a constant flux. The equation used to normalize the
MTCW to a common temperature (i.e., the average yearly water temperature experienced at the
plant) is presented in Section 6.2.4 as Equation 6.21. Membrane cleaning frequencies greater
than once per month have been suggested to be limiting because of the associated cost and lack
of automation. However, there is no reason to believe that membrane cleaning cannot be
highly automated and made a routine part of membrane plant operation. Unfortunately, that
technology does not exist today, and the impact of cleaning frequency must be considered in
the overall cost and performance of a membrane process.
Since membranes are made from many different materials, membrane manufacturers
specify chemicals, chemical strengths, temperatures and pH values for cleaning solutions.
Membrane compatibility with a specific cleaning solution must be verified with the membrane
manufacturer to avoid damage to the film. There are two basic categories of membrane
cleaning solutions, alkaline and acidic solutions. In general, alkaline solutions such as a 0.1 %
solution of sodium EDTA and a 0.1% solution of sodium hydroxide are effective for removing
organic and biological fouling agents, and are typically used in conjunction with detergents
such as sodium lauryl sulfate or Triton-X. Acidic solutions such as 0.5% phosphoric acid are
effective for removing inorganic foulants. When the exact nature of the foulant is not known,
the manufacturer's recommendation for an appropriate cleaning procedure should be solicited.
To clean the membranes, the pilot plant is generally modified to form a closed loop
system. The pilot plant is isolated from the raw water source, and the pretreatment chemical
pumps are turned off. The cleaning solution can be made in a 50 to 100 gallon chemical
resistant tank using the previously collected permeate. In some cases pre-heating the cleaning
solution may be necessary to properly clean fouled membranes. A hose is connected to the
feed line before the prefilters and the cleaning solution is pumped throughout the pilot plant
using a low pressure pump. The first flush of cleaning solution from the membrane pilot plant
generally contains the highest concentration of foulants and is wasted for the first five to ten
minutes of the cleaning procedure. The cleaning solution is then pumped through the pilot
plant to fill the membrane pressure vessels after which the pumps can be turned off to allow
the membranes to soak for one to several hours. The cleaning solution used during soaking is
flushed out of the system and replaced by fresh cleaning solution. After this second flush, the
remaining cleaning solution is directed back to the cleaning tank so it can be recirculated
through the membrane system for one to two hours. This procedure can be repeated with a
new cleaning solution if needed. Once the cleaning cycle has been completed, the membranes
should be flushed with previously collected permeate water.
To preserve the membranes for storage, follow membrane manufacturer specifications for
type and strength of preservative. Membranes should always be cleaned prior to preservation
and storage. Use previously collected membrane permeate to dilute the preservative chemicals
to the proper concentration. Connect a suction line from the booster pump to the preservative
solution tank. The same tank used for cleaning can be used for membrane preservation
3-91
-------
solutions. Energize the booster pump long enough to replace the feed and concentrate water
left from previous operation with the preservative solution. Turn off the booster pump and
close all valves to trap the preservative solution within the membrane pressure vessel. If the
membranes are to be stored separately from the pressure vessels, then they may be wetted in
the preservative solution and placed into sealed plastic bags.
Handle, dispose and store all chemicals used in the membrane study in a safe and approved
method.
6.2.4 Membrane Productivity
Membrane productivity is assessed by the rate of MTCW decline with time of operation.
All membranes foul during operation and constant production is achieved hi full-scale
membrane plants by increasing pressure to maintain a constant flux, and membrane pilot plants
should be operated in the same fashion.
The following section presents the procedure for determining the fouling rates for each
stage, and a flow weighted approach to determine the overall fouling rate for the system. This
procedure could also be used to determine the fouling rates for individual pressure vessels.
Determination of the fouling rate and cleaning frequency for a multi-stage membrane pilot
system is presented here.
The MTCW is calculated by using the following equations, and data recorded from a pilot-
scale study shown in Table 6-8. Using the definitions for permeate flux and the water mass
transfer coefficient from Table 2-2:
Fw = Qp/A = MTCVX NDP (6.19)
Rearranging the equation for water flux and solving for MTCW:
MTCV = Fjy/NDP (6.20)
The flux of water for each stage is first calculated using the permeate flow from each stage
and dividing by the membrane area associated with the each stage. For this example stage one
contained six 4" x 40" membrane elements with an area of 70 ft2 per element and produced
5,628 gpd of permeate at the initial data recording. Stage two contained three membrane
elements and produced 2814 gpd of permeate.
The flux for stage one is:
Fw = Qp/A = (5,628 gpd)/(6 elements x 70 ft2 per element) = 13.4 gfd
The flux for stage two is:
pw = Qp/A = (2,814 gpd)/(3 elements x 70 ft2 per element) = 13.4 gfd
3-92
-------
The flux is normalized to a common temperature so that water production can be compared
on an equivalent basis. Equation 6.21 can be used to normalize the flux to the average yearly
water temperature experienced at the plant conducting the study. If a manufacturer specifies a
temperature correction equation, then the specific equation should be used instead of Equation
6.21.
F^Tavg°C) = Fw(T°C)Xl.03rravg°-T°) (6.21)
where F^(Tavg0C) is the flux corrected to the average yearly water temperature experienced
at the plant conducting the study, FW(T°C) is the flux measured at ambient temperature, T° is
the temperature at which the flux was measured in °C, and Tavg° is the average yearly water
temperature at the plant conducting the study in °C.
The temperature correction equation was not used hi this particular case since the feed was
a ground water source with a constant temperature of 20 °C
In order to determine the NDP, the osmotic pressure gradient must first be estimated from
the feed, concentrate and permeate TDS values using Equation 6.14.
ATTS = ([(TDSJ, + TDSc) / 2] - TDSp) X (1 psi / 100 mg/L)
For this example and stage one: TDSP = 60 mg/L, TDSj = 300 mg/L and TDSC =
500 mg/L; for stage two: TDSp = 100 mg/L, TDS! = 500 mg/L and TDSC = 700 mg/L.
The osmotic pressure gradient for stage one is:
ATI! = ([(300 mg/L + 500 mg/L)/2] - 60 mg/L) X (1 psi/100 mg/L) = 3.4 psi
The osmotic pressure gradient for stage two is:
A7i2 = ([(500 mg/L + 700 mg/L)/2] -100 mg/L) X (1 psi/100 mg/L) = 5.0 psi
The net driving pressure is calculated according to Equation 6.22.
NDP = [(?,, + PC) / 2] - Pp - ATI (6.22)
where NDP is the net driving pressure, Px is the influent pressure, Pc is the concentrate
pressure and Pp is the permeate pressure.
For this example, the pressures for stage one are P! = 154 psi, Pc = 100 psi and Pp = 90
psi, and the pressures for stage two are Pj = 100 psi, Pc = 54 psi and Pp = 30 psi.
The NDP for stage one is:
NDP = [(154 psi + 100 psi) / 2] - 90 psi - 3.5 psi = 33.5 psi
3-93
-------
The NDP for stage two is:
NDP = [(100 psi + 54 psi) / 2] - 30 psi - 5.0 psi = 42 psi
Using these results, the MTCV can be calculated from Equation 6.20 as:
The MTCV for stage one is:
MTCV(1) = Fwm I NDP(1) = 13.4 gfd / 33.5 psi = 0.40 gfd/psi
The MTCV for stage two is:
= FW(2) I NDP{2) = 13.4 gfd / 42 psi = 0.32 gfd/psi
The MTCV for the system is calculated by flow weighting the MTCV f°r eacn sta§e using
the respective permeate flow of each stage:
MTCVsys = [(MTCV(1) X Qp_s(1))+(MTCV(2) X Qp.s(2))] / (Qp_s(1) + Qp.s(2))
Using the preceding results:
MTCVsys = [(0.40gfd/psix5628gpd)+(0.32gfd/psix2814gpd)]/(5628gpd + 2814gpd)
MTCVsys = 0.37 gfd/psi
The MTCV for each stage and for the system is calculated for each set of operational data
and plotted as a function of cumulative operating tune. A spreadsheet incorporating these
equations can be used to calculate the net driving pressure, the normalized flux and the MTCV
from the data recorded hi the daily operations log, Table 6-3. For the preceduig example the
system recovery, flux, MTCV, NDP, and run tune are summarized in Table 6-8 for each set of
operational data collected during the operation of a two-stage pilot system.
The data summarized hi Table 6-8 is presented graphically in Figure 6-2. The slope of a
linear least squares fit of the plotted data is the change hi the MTCV with tune for each stage
and the system. The slope of these lines is used to predict the rate of fouling and the required
cleaning frequency. The linear regression lines for stage one, stage two and the system were
calculated from the data plotted in Figure 6-2.
Linear regression for the stage one flux data:
= -0.0004 x tune + 0.379
3-94
-------
Linear regression for the stage two flux data:
MTCV = 0.00002 x time + 0.312
Linear regression for the system flux data:
MTCW = -0.0003 x time + 0.357
The slope of the flux curve for stage one is negative indicating a decline hi the flux during
operation, while the slope of the flux curve for stage two is positive indicating no decline over
the course of this study. One possible explanation is that colloidal fouling occurred hi the first
stage.
Equation 6.23 can be used to estimate the cleaning frequency from the rate of MTCV
decline and the acceptable loss in the MTCW.
Q x
dMTCV/dt
(6.23)
where CF is the cleaning frequency, Q is the acceptable fractional loss in the MTCV prior to
cleaning (e.g., 0.15), MTCW(0) is the design MTCV during operation with the source water,
and dMTC^/dt is the rate of MTCW decline determined from a linear regression of the flux
data.
A linear model was used to estimate the rate of MTCW decline in this example. The best
model to fit the data generated during the membrane studies should be used to estimate MTC^
decline, and this model may or may not be linear. In some situations, a linear regression for a
plot of the log of tune versus the log of the MTCV may be used to predict the rate of MTCV
decline.
Using equation 6.23 with a baseline MTCW of 0.35 gfd/psi and a 15% decline before
cleaning, the following cleaning frequencies are calculated:
The cleaning frequency for stage one:
CF = (0.15 x 0.35 gfd/psi) / 0.0004 gfd/psi/day = 131 days
The cleaning frequency for stage two:
CF = (0.15 x 0.35 gfd/psi) / -0.00002 gfd/psi/day =not defined
The cleaning frequency for stage two is not defined since the rate of MTCV decline was
negative (i.e., the MTCV was increasing with time).
3-95
-------
The cleaning frequency for the system:
CF = (0.15 x 0.35 gfd/psi) / 0.0003 gfd/psi/day = 175 days
These results indicate that the system will need to be cleaned approximately every 175 days
or 6 months when the system MTCW has declined by 15 %. This is not an excessive cleaning
frequency and this study, conducted on a ground water with a high TOC concentration, shows
that membrane processes are a viable option for the control of DBF formation.
The cleaning frequency in this example is based on a linear model of the system MTCW
decline. The linear least fit equation was chosen to model this data because it followed a
linear trend which is similar to information published in reports of long-term (greater than one
year) studies which used linear least squares models of MTCW decline to estimate cleaning
frequencies (Taylor et al., 1986, 1990, 1992). However, the model which best fits the data
should be used to predict cleaning frequencies. The decision to use the system MTCW decline
instead of the second stage MTCW decline was made in this case, because over-all water
production was deemed most important. There may be cases were the decline of an individual
stage may determine the time to stop production for cleaning.
6.3 Sampling Requirements
The following section describes the monitoring and sampling requirements for the pilot
membrane studies, and data sheets that present the reporting requirements are described in
Section 6.3.5. Unless explicitly stated otherwise, the monitoring and sampling requirements
listed in this section are requirements of the ICR.
6.3.1 Daily System Monitoring
The system must be monitored at least once each shift during each day of operation to
insure that it is functioning properly. The operating parameters that must be monitored during
each shift are summarized in Table 6-2 and include flow rates, direct flow measurements,
pressures, temperatures, pH and TDS concentrations. These monitoring requirements are
specific to a two-stage, 2-1 system. If a different configuration is used, monitoring should be
conducted at the following points.
The flow rate must be monitored for the permeate stream from each stage, the permeate
stream from the system, the concentrate waste stream from the system and the concentrate
recycle stream.
The flow must be directly measured for the system permeate stream and the system
concentrate waste stream.
The pressure must be measured for the influent to the cartridge filters, the effluent from
the cartridge filters, the influent stream to each stage, the permeate flow from each stage,
the permeate flow from the system, the concentrate waste stream from the system, and
before and after any inter-stage pumps used in the system.
3-96
-------
The temperature must be measured for the system influent stream and after any inter-stage
pumps used in the system.
The pH must be measured for the feed water prior to acid addition, the feed water after
mixing with the pretreatment chemicals, the influent stream to the system, and the
concentrate waste stream from the system.
The TDS concentration must be measured for the feed water prior to acid addition, the
feed water after mixing with the pretreatment chemicals, the influent to each stage, the
permeate flow from each stage, the permeate flow from the system, and the concentrate
waste stream from the system.
Table 6-2 only lists the required monitoring points. Monitoring at additional points may
be appropriate in some situations. It may also be appropriate to monitor additional water
quality parameters specific to the source water or treatment objectives.
An example of a daily operations log that can be used to record this data is shown in Table
6-3, but this data should be summarized in a data sheet similar to Table 6-11.
6.3.2 Biweekly System Monitoring
Biweekly system monitoring must include analysis of the folio whig parameters: TOC,
UV254, pH, alkalinity, total hardness, calcium hardness and turbidity. Each of these analyses
will be conducted on the feed to the system (after cartridge filtration but prior to joining with
the concentrate recycle stream), the influent to each stage, the permeate from each stage, the
system permeate and the concentrate waste from the system.
/*
~"; "* " v
Biweekly sampling must also include collection of samples for the following analyses:
bromide and SDS for THM4, HAA6, TOX and chlorine demand. Each of these analyses must
be conducted on the feed to the system (after cartridge filtration but prior to joining with the
concentrate recycle stream) and the system permeate.'.
The results from these biweekly analyses can be summarized hi data sheets similar to those
shown in Tables 6-12 through 6-15. A total of at least twenty (20), but no more than twenty-
six (26), biweekly sample sets will be collected over the course of a yearlong pilot-scale study.
The analyses on every fifth set of biweekly samples should be duplicated, resulting in four to
five sets of duplicate analyses over the course of the yearlong study. The results from the
duplicate analyses can be reported on data sheets similar to Tables 6-16 through 6-18.
6.3.3 Additional Monitoring And Reporting
In addition to the above monitoring and reporting requirements, any additional information
that would help to assess membrane performance should be summarized and included in the
ICR report. Some examples include:
3-97
-------
The date and time of membrane cleaning and a brief description of the cleaning procedure
(i.e., cleaning agent, volume of cleaning solution, duration of cleaning, etc.).
Process upsets that could affect performance (e.g., pretreatment failure, a major change in
water quality, operator error, etc.).
Replacement of a membrane element or any other system component.
Any change in the system operating parameters.
Any time that the system is off for more than two hours.
6.3.4 Pressure Vessel Or Element Monitoring (Optional)
Single pressure vessels or elements do not need to be monitored for the ICR as this level of
analysis would be cumbersome. However, when there is a deterioration in permeate quality or
productivity within a stage, it may be advantageous to isolate a single pressure vessel or
element. In this manner, failed elements can be identified and replaced. The flexibility to
isolate single pressure vessels can be realized by adding flow meters, pressure gauges and
sample ports to each pressure vessel. Single elements may need to be evaluated in a single
module apparatus, such as the system described hi Section 5.0.
6.3.5 Datasheets
This section includes data sheets that can be used to record the appropriate data from the
pilot-scale studies for the ICR. Corresponding data collection software will be sent to utilities
electing to conduct pilot-scale experiments after the plant submits a study concept form to
EPA.
The data sheet in Table 6-9 is used to report the characteristics of the membrane used
during the ICR treatment studies. These are the membrane characteristics as reported by the
manufacturer, and although some of this information may not be available, the data sheet
should be filled out as completely as possible. The area and cost of an 8" x 40" element are
requested for use hi the cost analysis.
The data sheet hi Table 6-10 requests information on the foulants in the feed water and the
pretreatment processes used to control fouling. The first section of this table requests
concentrations of foulants and fouling indices for the feed water. Only the foulants and indices
relevant to the water being investigated need to be measured and reported. The blank rows
should be used to report additional foulants that were measured but not listed hi this table.
This information should be used as a guide to selecting appropriate pretreatment processes.
During the run, the relevant fouling indices and water quality parameters should be
periodically measured to insure that pretreatment processes are performing properly and that
the proper chemical doses are being applied to the feed water.
The second half of this table requests information about the pretreatment processes used
prior to NF or RO. All pretreatment processes used should be reported here including
3-98
-------
processes in the existing full-scale treatment train, upstream of the point where the feed to the
pilot-scale system is tapped; and processes that were added specifically for membrane
pretreatment. Existing full-scale treatment processes used as membrane pretreatment should be
marked with an "E", modifications to processes in the existing plant treatment train (e.g., an
increase in the coagulant dose) should be indicated by an "M" and pretreatment processes used
in addition to the existing treatment tram (e.g., acid or antiscalant addition) should be
indicated with an "A". This table can be used to provide some of the pretreatment
information required in the final treatment study report.
Table 6-11 can be used to summarize the daily operating parameters that are recorded in
the daily operations log, Table 6-3. Temperature normalized fluxes, the MTCW, and both feed
and bulk rejections will be calculated by the data collection software. The feed rejection is
calculated from the feed and permeate concentrations. To calculate the bulk rejection, the
permeate and feed concentrations and the recovery are used to estimate the bulk concentration
which is then used to calculate the bulk rejection. The cumulative time in this data sheet
should be set at 0:00 at the start of the study and continued over the yearlong study. Down-
time for the system must be subtracted from the cumulative time (i.e., by stopping the timer
when the system is turned off).
Results from the analyses of biweekly water quality samples should be reported on data
sheets similar to Tables 6-12 through 6-15. The date on which the samples were collected
should be listed at the top of the data sheet along with the week of the run (e.g., week 2, week
4, week 6, ... week 52). At least twenty (20), but no more than twenty-six (26), biweekly
sample sets will be collected over the course of a yearlong pilot-scale study. The analyses on
every fifth set of biweekly samples should be duplicated, resulting in four to five sets of
duplicate analyses over the course of the yearlong study. The results from duplicate analyses
can be reported on data sheets similar to Tables 6-16 through 6-18.
The analyses for alkalinity, TDS, total hardness, calcium hardness, turbidity, TOC, and
UV254 are conducted on the cartridge filtered feed water, the system permeate and the system
concentrate waste; thus mass balance closure errors can be calculated for these parameters.
Mass balance closure errors for these analytes are typically less than a few percent.
The mass balance closure error is calculated in two steps. First the concentrate
concentration is calculated from the prefiltered feed and system permeate concentrations and
the fractional system recovery using Equation 6.24.
CF-(CpxR)
where CC(calc) is the calculated concentrate concentration. Next the mass balance closure error
is determined by comparing calculated and measured concentrate values:
3-99
-------
ErrorMB = Cc(meas) Cc(calc) x 100%
CC(meas) (6.25)
where ErrorRffi is the mass balance closure error expressed as a percentage and CC(mcas) is the
measured concentrate concentration.
Since permeate quality often exceeds the treatment objectives, the system permeate can be
blended with by-passed feed water. This can substantially reduce the required membrane area
and post-treatment requirements, and thus the cost. However, any removal of pathogens that
the membrane is capable of will be negated by blending permeate and by-passed feed water,
unless the pathogens were removed or inactivated by a process upstream of the membrane
process. By calculating the water quality for different blending ratios, costs for different
levels of treatment can be estimated. Tables 6-19 (a and b) use the average prefiltered feed
and system permeate water quality parameters to calculate the amount of flow that must be
treated by membranes to achieve the Stage 1 MCLs (Table 6-19a) and the proposed Stage 2
MCLs (Table 6-19b). Water quality parameters that impact post-treatment requirements are
also calculated for the blended waters. In these two tables, QT is the total blended flow, and
the ratio Qp/Qr is the fraction of flow that must be treated by membranes to meet the treatment
objectives. This ratio is calculated for both the THM4 and HAAS MCLs as either can control
the blend ratio. The subscript b in these tables denotes blended water quality.
3-100
-------
System or stage recycle (optional) 16
Feed
pump
Stage One
Cartridge
filtration
Acid or antiscalant
addition
pump
h j
ssure
_
5
rH h
?L>1
....fe L
* 1
-^
1
9
10
r
i Int
JA
i
1
er-stage pump
(optional)
Stage Two
|
i
I
-rtri hr>
i
2
_J "»
13
r h.
15
Figure 6-1 Flow Diagram Of A Two-Stage Pilot Plant Showing Sampling And
Monitoring Locations By Number
. Stage 1
- Linear (Stage 2)
. Stage 2
-Linear (Stage 1)
. System
- Linear (System)
0.50
Stage 1 MTCw = -0.0004 x t + 0.3794
Stage 2 MTCw = 0.00002 x t + 0.3123
System MTCw = -0.0003 x t + 0.3569
Time (days)
Figure 6-2 Two-Stage Pilot Plant Membrane Productivity
-------
Table 6-1 Two-Stage Membrane Pilot Plant Numerical Identification Code
Number
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
Description
Feed water
Acidified feed water
Cartridge-filtered feed water
Influent to stage 1
Influent to pressure vessel 1, stage 1
Influent to pressure vessel 2, stage 1
Permeate from pressure vessel 1, stage 1
Permeate from pressure vessel 2, stage 1
Concentrate from pressure vessel 1, stage 1
Concentrate from pressure vessel 2, stage 1
Influent to stage 2
Permeate from stage 1
Permeate from stage 2
System concentrate waste
System permeate
Concentrate recycle
Table 6-2 Minimum Monitoring Matrix For A Two-Stage Pilot Plant
Parameter
How rate
Pressure
Temperature
Direct flow measurement
TDS
PH
Total Hardness
Calcium Hardness
Alkalinity
Turbidity
UV^
TOC
Bromide
SDS - THM4
SDS - HAA6
SDS - TOX
SDS - Chlorine demand
Location Number
1
D
D
2
D
3
D
D
D
B
B
B
B
B
B
B
B
B
B
B
4*
D
D
D
D
B
B
B
B
B
B
5
6
7
8
9
10
11
D
D
B
B
B
B
B
B
B
12
D
D
D
B
B
B
B
B
B
B
13
D
D
D
B
B
B
B
B
B
B
14
D
D
D
D
D
B
B
B
B
B
B
15
D
D
D
D
B
B
B
B
B
B
B
B
B
B
B
B
16
D
D - daily (once per shift)
B - biweekly
*: The water quality parameters at location 4 only need to be monitored if concentrate recycle is used.
-------
Table 6-3 Daily Operations Log Sheet For A Two-Stage Membrane Pilot Plant
Operators initials
Date
Time
Clock Counter
Time on
Time off
Pressure Measurements (psi)
2 Acidified feed water
3 Cartridge-filtered feed
4 Influent to stage 1
11 Influent to stage 2
12 Permeate stage 1
13 Permeate stage 2
14 Concentrate stage 2
15 System permeate
Flow Rate Measurements (gpm)
12 Permeate stage 1
13 Permeate stage 2
14 System waste
15 System permeate
16 Concentrate recycle
Directly Measured Flows (gpm)
14 System concentrate waste
15 System permeate
Temperature Measurements (°C)
4 Influent to stage 1
11 Influent to stage 2
TDS Measurements (mg/L)
1 Feed water
3 Cartridge-filtered feed
4 System influent
11 Influent to stage 2
12 Permeate stage 1
13 Permeate stage 2
14 System waste
15 System permeate
pH Measurements
1 Feed water
3 Cartridge-filtered feed
4 Influent to stage 1
14 System waste
Reasons For Down Time:
Notes:
Time To Complete Maintenance:
-------
Table 6-4 RO And NF Membrane Pilot Plant Design Criteria
Parameter
Number of stages
Number of pressure vessels per stage
Number of elements per pressure vessel
Recovery per stage, %
Recovery for system, %
Design flux, gfd
Surface area per element, ft"'
MTCiV, gfd/psi
Acceptable range of feed flow rates
through an element, gpm
Pressure loss per element, psi
Pressure loss in stage entrance and exit,
psi
Feed stream TDS, mg/L
TDS rejection, %
Design Criteria
At least 2 stages are required for the ICR, but
1 or 3 is not unusual in pilot plants.
At least a 2-1 array is required for ICR pilot
plants.
At least 3 elements per pressure vessel are
required for the ICR but 1 to 6 is possible.
Restricted by minimum flow into element,
typically 30% to 50% for a two-stage pilot
plant.
At least 75% for the ICR pilot studies, but may
be restricted by a limiting salt.
Typically 15 gfd for RO/NF ground water
systems but can range from 10 to 20 gfd.
Restricted by element diameter, typically 20 ftz
for a 2.5" element diameter and 80 ft2 for a
4.0" element diameter.
Restricted by membrane resistance, typically
0.20 to 0.50 gfd/psi for a NF element, and
0.10 to 0.25 gfd/psi for a RO element.
Specified by the manufacturer, typically 0.5 to
4 gpm for a 2.5" element and 3 to 16 gpm for
a 4" element.
Hydraulic losses due to flow across the element
surface and through the feed spacer, estimated
at 3 psi per element.
Hydraulic losses due to flow in and out of the
inter-stage piping, estimated at 5 psi per stage.
Typically a function of the raw water source.
Provided by manufacturer or testing, typically
30 to 90% for NF and 95 to 99% for RO.
-------
Table 6-5 Osmotic Pressure Estimates For Stages 1 And 2 Of A 2-1 Membrane System
With Three Elements Per Pressure Vessel, Operated At 75% Recovery
Stage 1
RCJTDS
0.5
0.6
0.7
0.8
0.9
0.95
0.98
TDSF = 1000 mg/L (Rsys = 75%)
r = 0 r = l r = 2 r = 3 r = 5
ATT (psi) ATI (psi) ATC (psi) ATC (psi) ATT (psi)
6.7 7.6 7.8 7.8 7.9
8.2 10.1 10.4 10.5 10.7
9.8 13.1 13.7 14.0 14.3
11.4 17.0 18.1 18.6 19.1
13.2 22.1 24.2 25.1 26.0
14.1 25.3 28.1 29.4 30.7
14.6 27.5 30.9 32.5 34.0
TDSp = 500 mg/L (Rsys = 75%)
RCJTDS
0.5
0.6
0.7
0.8
0.9
0.95
0.98
r = 0 r = l r = 2 r = 3 r = 5
ATC (psi) ATI (psi) ATC (psi) ATI (psi) ATC (psi)
3.3 3.8 3.9 3.9 3.9
4.1 5.0 5.2 5.3 5.3
4.9 6.6 6.9 7.0 7.1
5.7 8.5 9.1 9.3 9.6
6.6 11.0 12.1 12.6 13.0
7.0 12.7 14.1 14.7 15.3
7.3 13.8 15.5 16.3 17.0
TDSp = 300 mg/L (Rsys = 75%)
ReJTDS
0.5
0.6
0.7
0.8
0.9
0.95
0.98
r = 0 r=l r = 2 r = 3 r = 5
ATC (psi) ATI (psi) ATC (psi) ATC (psi) ATC (psi)
2.0 2.3 2.3 2.4 2.4
2.5 3.0 3.1 3.2 3.2
2.9 3.9 4.1 4.2 4.3
3.4 5.1 5.4 5.6 5.7
4.0 6.6 7.2 7.5 7.8
4.2 7.6 8.4 8.8 9.2
4.4 8.3 9.3 9.8 10.2
TDSp = 200 mg/L (Rsys = 75%)
RCJTDS
0.5
0.6
0.7
0.8
0.9
0.95
0.98
r = 0 r = l r = 2 r = 3 r = 5
ATC (psi) ATC (psi) ATC (psi) ATC (psi) ATC (psi)
1.3 1.5 1.6 1.6 1.6
1.6 2.0 2.1 2.1 2.1
2.0 2.6 2.7 2.8 2.9
2.3 3.4 3.6 3.7 3.8
2.6 4.4 4.8 5.0 5.2
2.8 5.1 5.6 5.9 6.1
2.9 5.5 6.2 6.5 6.8
Stage 2
TDSF = 1000 mg/L (Rsvs = 75%)
ReJTDS
0.5
0.6
0.7
0.8
Q.9
0.95
0.98
r = 0 r = 1 r = 2 r = 3 r = 5
ATC (psi) ATC (psi) ATC (psi) ATC (psi) ATC (psi)
9.6 8.5 8.3 8.2 8.1
12.6 11.5 11.3 11.2 11.1
16.2 15.3 15.1 15.0 14.9
20.2 20.3 20.3 20.2 20.1
24.8 27.1 27.4 27.5 27.6
27.3 31.4 32.1 32.4 32.7
28.9 34.5 35.5 35.9 36.3
TDSF = 500 mg/L (Rsys = 75%)
ReJTDs
0.5
0.6
0.7
0.8
0.9
0.95
0.98
r = 0 r = l r = 2 r = 3 r = 5
ATC (psi) ATC (psi) ATC (psi) ATC (psi) ATC (psi)
4.8 4.2 4.1 4.1 4.1
6.3 5.7 5.6 5.6 5.5
8.1 7.7 7.6 7.5 7.5
10.1 10.2 10.1 10.1 10.1
12.4 13.6 13.7 13.8 13.8
13.7 15.7 16.1 16.2 16.3
14.5 17.2 17.7 18.0 18.2
TDSF = 300 mg/L (Rsys = 75%)
RCJTDS
0.5
0.6
0.7
0.8
0.9
0.95
0.98
r = 0 r = l r = 2 r = 3 r = 5
ATC (psi) ATC (psi) ATC (psi) ATC (psi) ATC (psi)
2.9 2.5 2.5 2.5 2.4
3.8 3.4 3.4 3.4 3.3
4.8 4.6 4.5 4.5 4.5
6.1 6.1 6.1 6.1 6.0
7.4 8.1 8.2 8.3 8.3
8.2 9.4 9.6 9.7 9.8
8.7 10.3 10.6 10.8 10.9
TDSF = 200 mg/L (Rsys = 75%)
RSJTDS
0.5
0.6
0.7
0.8
0.9
0.95
0.98
r = 0 r=l r=2 r = 3 r = 5
ATC (psi) ATC (psi) ATC (psi) ATC (psi) ATC (psi)
1.9 1.7 1.7 1.6 1.6
2.5 2.3 2.3 2.2 2.2
3.2 3.1 3.0 3.0 3.0
4.0 4.1 4.1 4.0 4.0
5.0 5.4 5.5 5.5 5.5
5.5 6.3 6.4 6.5 6.5
5.8 6.9 7.1 7.2 7.3
TDSF = the TDS concentration in the feed water.
ReJTos = me manufacturer reported TDS rejection, expressed as a decimal fraction.
r = the recycle ratio (QR/QF)
-------
Table 6-6 Example Design Criteria For A Nanofiltration Pilot Plant
Parameter
Number of stages
Number of pressure vessels in stage 1
Number of pressure vessels in stage 2
Number of elements per pressure vessel
Recovery per stage, %
Recovery for system, %
Design flux, gfd
Surface area per 4" x 40" element, ft^
MTCW, gfd/psi
Maximum flow rate to an element, gpm
Minimum flow rate to an element, gpm
Pressure loss per element, psi
Pressure loss in stage entrance and exit, psi
Feed stream TDS, mg/L
TDS rejection, %
Value
2
2
1
3
50
75
15
70
0.30
16
4
3
5
300
70
-------
Table 6-7 Pilot Plant Design Parameters For The Example Problem
Location #l
3
4
5
6
7
8
9
10
11
12
13
14**
14
15
16
Description
Cartridge-filtered feed water
Influent to system / stage 1
Influent to pressure vessel 1, stage 1
Influent to pressure vessel 2, stage 1
Permeate from pressure vessel 1, stage 1
Permeate from pressure vessel 2, stage 1
Concentrate from pressure vessel 1, stage 1
Concentrate from pressure vessel 2, stage 1
Influent to stage 2
Permeate from stage 1
Permeate from stage 2
Concentrate from stage 2
System concentrate waste
System permeate
System concentrate recycle
Flow rate* (gpm)
8.8
10.9 to 20.4
5.4 to 10.2
5.4 to 10.2
2.2
2.2
3.2 to 8.0
3. 2 to 8.0
6.4 to 16.0
4.4
2.2
4.2 to 13. 8
2.2
6.6
2.2 to 11. 6
Pressure (psi)
20 to 30
95
95
95
30
30
81
81
81
30
30
67
0
30
67
1: Refer to Table 6-1 and Figure 6-1 for an explanation of the location numbers.
*: A range of flow rates is provided to account for different recycle flow rates.
**: This location represents the total concentrate flow leaving stage 2 (i.e., recycle plus waste)
-------
Table 6-8 Example Of Time-Dependent Membrane Productivity For A Two-Stage Pilot Plant
Runtime
(days)
0.0
1.0
2.0
2.8
3.8
6.7
7.7
8.7
9.6
13.6
14.4
15.4
16.3
17.4
20.2
22.2
23.2
24.1
27.0
27.8
28.8
30.8
33.7
34.7
35.6
36.7
40.5
42.4
44.2
47,2
48.1
49.1
50.1
51.1
54.0
54.9
56.9
57.9
60.8
61.5
62.7
63.5
64.6
67.4
63.4
70.4
71.3
74.1
76.1
77.2
78.1
82.0
82.0
83.1
84.0
86.9
System
Recovery
(%)
75
74
75
76
77
76
76
76
76
75
76
76
74
76
76
77
75
76
76
76
75
76
76
76
76
76
76
76
76
75
75
75
76
75
75
75
76
75
73
76
75
75
75
75
75
75
75
75
74
75
74
74
76
76
75
75
NDP
Stage 1
(psi)
34
37
36
35
36
37
38
38
37
38
36
37
38
37
38
37
38
36
37
37
37
37
37
37
37
37
38
38
37
39
38
38
38
38
38
37
38
38
38
38
37
38
38
38
39
38
38
38
38
38
38
39
38
38
38
38
NDP
Stage 2
(psi)
42
43
42
44
45
45
45
45
45
45
45
45
42
43
48
45
47
45
50
45
47
45
47
46
47
44
45
47
44
43
45
46
45
45
45
45
44
46
45
43
44
45
46
44
44
44
45
44
44
44
45
44
44
45
44
46
NPD
System
(psi)
36
38
38
38
39
39
40
40
39
40
39
39
39
38
41
39
40
39
40
39
40
39
40
39
40
39
40
41
39
40
40
40
40
40
40
40
40
40
40
39
39
40
40
40
40
40
40
40
39
40
40
41
40
40
40
40
Flux
Stage 1
(gfd)
13.4
13.8
13.6
13.6
14.2
14.2
14.5
13.6
13.4
13.0
13.6
13.6
13.0
14.0
13.8
14.0
13.4
13.6
13.8
13.6
13.3
14.1
13.9
13.7
13.6
13.4
13.8
13.3
13.6
13.4
13.4
13.4
13.4
13.3
13.2
13.4
13.6
13.2
13.0
13.7
13.1
13.2
13.0
13.2
13.1
13.2
13.0
12.6
12.9
13.0
13.0
13.3
14.1
13.5
13.2
13.6
Flux
Stage 2
(gfd)
13.4
12.6
13.0
13.6
14.2
14.2
14.2
13.6
14.0
13.0
13.9
14.2
13.0
14.0
14.4
14.6
14.0
14.8
14.4
13.9
14.8
13.8
14.8
14.6
14.2
14.0
14.4
14.2
14.2
14.0
14.0
14.0
14.0
14.2
13.8
14.0
14.2
13.8
13.6
14.0
14.0
13.8
13.6
13.8
14.0
13.8
13.6
13.8
13.8
13.6
13.6
13.6
14.1
14.4
14.1
13.9
Flux
System
(gfd)
13.4
13.4
13.4
13.6
14.2
14.2
14.4
13.6
13.6
13.0
13.7
13.8
13.0
14.0
14.0
14.2
13.6
14.0
14.0
13.7
13.8
14.0
14.2
14.0
13.8
13.6
14.0
13.6
13.8
13.6
13.6
13.6
13.6
13.6
13.4
13.6
13.8
13.4
13.2
13.8
13.4
13.4
13.2
13.4
13.4
13.4
13.2
13.0
13.2
13.2
13.2
13.4
14.1
13.8
13.5
13.7
MTCV
Stage 1
(gfd/psi)
0.40
0.38
0.38
0.39
0.39
0.39
0.39
0.36
0.37
0.34
0.38
0.37
0.35
0.38
0.37
0.38
0.36
0.37
0.38
0.37
0.36
0.38
0.38
0.37
0.37
0.37
0.37
0.35
0.37
0.35
0.36
0.36
0.35
0.35
0.35
0.36
0.36
0.35
0.34
0.36
0.35
0.35
0.34
0.35
0.34
0.35
0.34
0.33
0.34
0.34
0.34
0.34
0.37
0.36
0.35
0.36
MTCV
Stage 2
(gfd/psi)
0.32
0.30
0.31
0.31
0.32
0.32
0.32
0.30
0.31
0.29
0.31
0.32
0.31
0.33
0.30
0.32
0.30
0.33
0.29
0.31
0.32
0.31
0.32
0.32
0.31
0.32
0.32
0.31
0.33
0.33
0.31
0.31
0.31
0.32
0.31
0.31
0.32
0.30
0.31
0.33
0.32
0.31
0.30
0.31
0.32
0.31
0.30
0.32
0.32
0.31
0.30
0.31
0.32
0.32
0.32
0.31
MTCV
System
(gfd/psi)
0.37
0.35
'0.36
0.36
0.37
0.37
0.36
0.34
0.35
0.33
0.35
0.35
0.33
0.36
0.34
0.36
0.34
0.36
0.35
0.35
0.34
0.36
0.36
0.35
0.35
0.35
0.35
0.33
0.35
0.34
0.34
0.34
0.34
0.34
0.33
0.34
0.35
0.33
0.33
0.35
0.34
0.33
0.33
0.34
0.33
0.34
0.33
0.33
0.33
0.33
0.33
0.33
0.35
0.34
0.34
0.34
-------
Table 6-9 Membrane Characteristics As Reported By The Manufacturer
Utility name and address
ICR plant number
Phone number
Contact person
FAX number
Characteristics of the membrane elements used in the study
Membrane manufacturer
Membrane module model number
Size of element used in study (e.g. 4" x 40")
Active membrane area of element used in study
Active membrane area of an equivalent 8" x 40" element
Purchase price for an equivalent 8" x 40 " element ($)
Molecular weight cutoff (Daltons)
Membrane material / construction
Membrane hydrophobicity (circle one)
Membrane charge (circle one)
Design pressure (psi)
Design flux at the design pressure (gfd)
Variability of design flux (%)
MTCV (gfd/psi)
Standard testing recovery (%)
Standard testing pH
Standard testing temperature (°C)
Design cross-flow velocity (fps)
Maximum flow rate to the element (gpm)
Minimum flow rate to the element (gpm)
Required feed flow to permeate flow rate ratio
Maximum element recovery (%)
Rejection of reference solute and conditions
of test (e.g. solute type and concentration)
Variability of rejection of reference solute (%)
Spacer thickness (ft)
Scroll width (ft)
Acceptable range of operating pressures
Acceptable range of operating pH values
Typical pressure drop across a single element
Maximum permissible SDI
Maximum permissible turbidity (ntu)
Chlorine/oxidant tolerance
Suggested cleaning procedures
Hydrophilic Hydrophobic
Negative Neutral Positive
Note: Some of this information may not be available, but this table should be filled out as completely
as possible for each membrane tested.
-------
Table 6-10 Membrane Pretreatment Data
Utility name and address
ICR plant number
Contact person
Membrane trade name
Phone #
FAX#"
Foulants and fouling indices of the feed water prior to pretreatment1
Alkalinity (mg CaCO3/L)
Ca Hardness (mg CaCO3/L)
LSI
Dissolved iron (mg/L)
Total iron (mg/L)
Dissolved aluminum (mg/L)
Total aluminum (mg/L)
Fluoride (mg/L)
Phosphate (mg/L)
Sulfate (mg/L)
Calcium (mg/L)
Barium (mg/L)
Strontium (mg/L)
Reactive silica (mg/L as SiO2)
Turbidity (ntu)
SDI
MFI
MPFI
1: Only those foulants and fouling indices relevant to the water being tested need to be evaluated.
Additional foulants and indices can be listed in the blank rows or on an attached sheet.
Pretreatment processes used prior to nanofiltration or reverse osmosis2
Pre-filter exclusion size (um)
Type of acid used
Acid concentration (units)
mL of acid per L of feed
Type of antiscalant used
Antiscalant concentration (units)
mL of antiscalant per L of feed
Type of coagulant used
Coagulant dose (mg/L)
Type of polymer used during coag.
Polymer dose (mg/L)
2: Use an "B" to indicate a pretreatment process that is currently part of the plant treatment train, an "M" to
indicate a modification to a process that is currently part of the plant treatment train, and an "A" to
indicate an addition to the current treatment train.
Additional pretreatment processes, such as MF, can be listed in the blank rows or on an attached sheet.
-------
O
i-<
4)
1
1
PH
1
O)
I
a a
o
u
!
5
-o
S
§
I
3
CO* £
p B
t^ o
CSS
«*
a
A
-
JJ
cy
i
u
-------
ge 2 of 3
With
I
1
3
t-4 Ut
5^
S x
*s
Uti
o
S
tf
'
1
a
CJ
.§ 1 0
« § '^
i .g §
| S |
o M 1
ei Qi
"5! "7
*c5 T"H
a
M
m
z
0
@
<
!vi
I!
P 1
O ^
K*
04
*s
^.^
£ a
*s
a ^,
S u
Q_> O
"I
-I
u
>
^ e -C-
I 1 ^
U
B t?
^ ^
Q
-------
s
I
ith time
f
1
1
t5
If
C/J
0
3-
1
CO
t\
r!
co* t;
B I
fi
* I
-"I
/^s
ft1"1 'w
PH Q
& '-^
a o
<& ^
*S
*I
>
3
"^ § G.
U
1 1
I
-------
Table 6-12 Membrane Pilot System Biweekly Water Quality Parameters For Week Two
Utility name and address
ICR plant number
Contact person
Membrane trade name
Phone #
FAX#"
Parameter
Sampling date
Sampling time
Alkalinity
TDS
Total hardness
Calcium hardness
Bromide
PH
Turbidity
Temperature
TOC
UV^
SDS samples
Chlorine dose
Chlorine residual
Chlorine demand
Chlorination temp.
Chtorination pH
Incubation time
TOX
CHClj
CHBrCt2
CHBr}Cl
CHBr}
THM4
MCAA'
DCAA'
TCAA'
MBAA '
DBAA'
BCAA'
TBAA
CDBAA
DCBAA
HAA6
Units
MM/DD/YY
hh:mm
mgCaCO3/L
mg/L
mgCaCO3/L
mgCaCO3/L
Hg/L
ntu
°C
mg/L
cm"1
Cp-sys
^p-syj
CW.sy!
...
Ci-s(l)
Q-s (2)
Cp-sd)
Cp-s (2)
mg/L
mg/L
mg/L
°C
_
hours
Hg/L
VS^
Hg/L
Hg/L
Hg/L
Hg/L
Hg/L
Hg/L
Hg/L
Hg/L
Hg/L
Hg/L
Hg/L
Hg/L
Ug/L
Hg/L
...
...
...
.
___ *
...
...
...
...
...
...
...
...
...
...
...
...
...
...
...
...
...
...
...
...
...
...
...
...
...
...
...
...
...
...
...
...
...
...
...
...
...
...
...
...
...
These six species make up HAA6. The other three HAA species should be reported if measured.
Sampling points are defined in Figure 6-1 and Table 6-2: CF.sys = system feed after prefiltration,
Cp-jy, " system permeate, C\v.iyi = system waste, C^Q = influent to stage i, and Cp.sys = permeate from stage i
-------
Table 6-13 Membrane Pilot System Biweekly Water Quality Parameters For Week Four
Utility name and address
ICR plant number
Contact person
Membrane trade name
Phone #
FAX#"
Parameter
Sampling date
Sampling time
Alkalinity
TDS
Total hardness
Calcium hardness
Bromide
PH
Turbidity
Temperature
TOC
UV254
SDS samples
Chlorine dose
Chlorine residual
Chlorine demand
Chlorination temp.
Chlorination pH
Incubation time
TOX
CHC13
CHBrCl2
CHBr2Cl
CHBr3
THM4
MCAA'
DCAA"
TCAA *
MBAA'
DBAA'
BCAA
TBAA
CDBAA
DCBAA
HAA6
Units
MM/DD/YY
hh:mm
mg CaCO3 / L
. mg/L
mg CaCO3 / L
mg CaCO3 / L
ug/L
.
ntu
°C
mg/L
cm '
Cp-sys
^p-sys
C\V-sys
...
Q[-s (l)
GI_S m
Cp.s(i)
cps(2)
mg/L
mg/L
mg/L
°C
hours
ug/L
ug/L
ug/L
ug/L
ug/L
ug/L
Ug/L
ug/L
ug/L
ug/L
ug/L
ug/L
ug/L
ug/L
ug/L
ug/L
...
.
.
...
...
...
...
--
...
...
...
...
...
...
...
-
__.
...
...
* Ihese six species make up HAA6. The other three HAA species should be reported if measured.
Sampling points are defined in Figure 6-1 and Table 6-2: CF.sys = system feed after prefiltration, '
Cp-syS = system permeate, Cw.sys = system waste, CVs& = influent to stage i, and Cp.sys = permeate from stage i
-------
Table 6-14 Membrane Pilot System Biweekly Water Quality Parameters For Week Six
Utility name and address
ICR plant number
Contact person
Membrane trade name
Phone #
FAX#"
Parameter
Sampling date
Sampling time
Alkalinity
TDS
Total hardness
Calcium hardness
Bromide
pH
Turbidity
Temperature
TOC
UVjM
SDS samples
Chlorine dose
Chlorine residual
Chlorine demand
Chlorination temp.
Chlorination pH
Incubation time
TOX
CHC13
CHBrCl3
CHBr2Cl
CHBr3
THM4
MCAA'
DCAA'
TCAA'
MBA A'
DBA A'
BCAA'
TBAA
CDBAA
DCBAA
HAA6
Units
MM/DD/YY
hh:mm
mgCaCO3/L
mg/L
mgCaCO3/L
mgCaCO3/L
ug/L
ntu
°C
mg/L
cm
Cp-sys
Cp-sys
Cw-sys
Ci-s(l)
...
Cj-s (2)
...
CP-S(1)
...
Cp-s (2)
...
mg/L
mg/L
mg/L
°C
hours
ug/L
ug/L
ug/L
ug/L
ug/L
ug/L
ug/L
ug/L
ug/L
ug/L
ug/L
ug/L
ug/L
ug/L
ug/L
ug/L
...
...
...
...
...
...
...
...
...
...
...
...
...
...
...
...
... -
...
...
...
...
...
...
...
...
...
...
...
...
...
...
...
...
...
...
...
...
...
...
...
...
...
...
...
...
...
...
...
...
...
...
...
...
...
...
...
...
...
...
...
...
...
...
...
...
...
...
...
...
...
...
...
...
.
...
...
...
...
...
...
...
...
...
...
...
...
...
* These six species make up HAA6. The other three HAA species should be reported if measured.
Sampling points are defined in Figure 6-1 and Table 6-2: CF.sys = system feed after prefiltration,
Cp.jy, - system permeate, Cw.syj = system waste, C^ = influent to stage i, and Cp.sys = permeate from stage i
-------
Table 6-15 Membrane Pilot System Biweekly Water Quality Parameters For Week
Utility name and address
ICR plant number
Contact person
Membrane trade name
Phone #
FAX#"
Parameter
Sampling date
Sampling time
Alkalinity
TDS
Total hardness
Calcium hardness
Bromide
pH
Turbidity
Temperature
TOC
UV254
SDS samples
Chlorine dose
Chlorine residual
Chlorine demand
Chlorination temp.
Chlorination pH
Incubation time
TOX
CHC13
CHBrCl2
CHBr2Cl
CHBr3
THM4
MCAA*
DCAA"
TCAA'
MBAA'
DBAA'
BCAA '
TBAA
CDBAA
DCBAA
HAA6
Units
MM/DD/YY
hh:mm
mg CaCO3 / L
mg/L
mg CaCO3 / L
mg CaCO3 / L
Ug/L
...
ntu
°C
mg/L
cm"1
Cp-sys
ctsys
C\V-sys
...
CI-S
-------
Table 6-16 Duplicate Analysis Of Membrane Pilot System Biweekly Water Quality Parameters For Week Ten
Utility name and address
ICR plant number
Contact person
Membrane trade name
Phone #
FAX#"
Parameter
Sampling date
Sampling time
Alkalinity
TDS
Total hardness
Calcium hardness
Bromide
PH
Turbidity
Temperature
TOG
UVa4
SDS samples
Chlorine dose
Chlorine residual
Chlorine demand
Chlorination temp.
Chlorination pH
Incubation time
TOX
CUCli
CHBrClj
CHBr2Cl
CHBr}
THM4
MCA A'
DCAA'
TCAA'
MB A A'
DBA A'
BCAA'
TBAA
CDBAA
DCBAA
HAA6
Units
MM/DD/YY
hh:mra
mgCaCO3/L
mg/L
mgCaCO3/L
mgCaCO3/L
ug/L
_
ntu
°C
mg/L
--I
cm
Cp.jyj
f
*-p-sys
Cw-sys
...
Q-s (i)
...
Q-s (2)
...
Cp-s (1)
Cp-s (2)
mg/L
mg/L
mg/L
°C
hours
Hg/L
ug/L
Ug/L
Ug/L
Hg/L
Ug/L
ug/L
ug/L
Ug/L
Ug/L
ug/L
ug/L
ug/L
ug/L
ug/L
ug/L
...
...
...
...
...
...
...
...
...
...
...
...
...
...
...
...
...
...
...
...
.
...
...
...
...
...
...
...
...
...
...
...
...
...
...
...
...
...
...
...
...
...
...
...
...
...
...
...
...
...
* These six species make up HAA6. The other three HAA species should be reported if measured.
Sampling points are defined in Figure 6-1 and Table 6-2: CF.sys = system feed after prefiltration,
Cp.,y, = system permeate, Cw^yj = system waste,
influent to stage i, and Cp.sys = permeate from stage i
-------
Table 6-17 Duplicate Analysis Of Membrane Pilot System Biweekly Water Quality Parameters For Week Twenty
Utility name and address
ICR plant number
Contact person
Membrane trade name
Phone #
FAX#"
Sampling date
Sampling time
Alkalinity
TDS
Total hardness
Calcium hardness
Bromide
pH
Turbidity
Temperature
TOC
UV
2S4
SDS samples
Chlorine dose
Chlorine residual
Chlorine demand
Chlorination temp.
Chlorination pH
Incubation time
TOX
CHC13
CHBrCl2
CHBr2Cl
CHBr3
THM4
MCAA
DCAA
TCAA
MBAA
DBAA
BCAA
TBAA
CDBAA
DCBAA
HAA6
hh:mm
mg CaCO3 / L
mg/L
mg CaCO3 / L
mg CaCO3 / L
Hg/L
ntu
mg/L
p-sys
mg/L
mg/L
mg/L
°C
hours
Hg/L
Hg/L
Hg/L
Hg/L
ug/L
ug/L
These six species make up HAA6. The other three HAA species should be reported if measured.
Sampling points are defined in Figure 6-1 and Table 6-2: CF.sys = system feed after prefiltration,
cp-sys = system permeate, Cw.sys = system waste, C,.s(i) = influent to stage i, and Cp.sys = permeate from stage i
-------
Table 6-18 Duplicate Analysis Of Membrane Pilot System Biweekly Water Quality Parameters For Week _
Utility name and address
ICR plant number
Contact person
Membrane trade name
Phone #
FAX#"
Parameter
Sampling dote
Sampling time
Alkalinity
TDS
Total hardness
Calcium hardness
Bromide
pH
Turbidity
Temperature
TOC
UVK4
SDS samples
Chlorine dose
Chlorine residual
Chlorine demand
Chlorination temp.
Chlorination pH
Incubation time
TOX
CHC13
CHBrCl,
CHBr2Cl
CHBr3
THM4
MCAA '
DCAA'
TCAA'
MBAA '
DBAA '
BCAA'
TBAA
CDBAA
DCBAA
HAA6
Units
MM/DD/YY
hh:mm
mgCaCO3/L
mg/L
mgCaCO3/L
mgCaCO3/L
ug/L
ntu
°C
mg/L
cm"'
Cr-sy,
Cp.^
Cw-sys
Ci.s(i)
...
Q-s(2)
...
Cp-s(l)
...
Cp.s(2)
...
mg/L
mg/L
mg/L
°C
hours
ug/L
ug/L
ug/L
ug/L
ug/L
ug/L
ug/L
ug/L
ug/L
ug/L
ug/L
ug/L
ug/L
Ug/L
ug/L
ug/L
...
...
...
...
...
...
...
...
...
...
...
...
...
...
...
...
...
...
...
...
...
...
...
...
...
...
...
...
...
...
...
...
...
...
...
...
...
...
...
...
...
...
...
...
...
...
...
...
...
...
...
...
...
...
...
...
...
...
...
...
...
...
...
...
...
...
...
...
...
...
...
...
...
...
...
...
...
...
...
...
...
...
...
...
...
...
* These six species make up HAA6. The other three HAA species should be reported if measured.
Sampling points are defined in Figure 6-1 and Table 6-2: CF.5ys = system feed after prefiltration,
CP-IJJ m system permeate, Cw.sys = system waste, Q.S(j) = influent to stage i, and Cp.sys = permeate from stage i
-------
Table 6-19a Blending Calculations To Determine The Fraction Of Flow That Requires NF
Treatment To Meet The Stage I DBF Regulations
Utility name and address
ICR plant number
Membrane trade name
THM4 Controls
HAAS Controls
Phone number
FAX number
Contact person
Parameter
THM4F,^ig/L
THM4p, jigflL
HAA5F, |ig/L
HAA5p, ^ig/L
AlkF, mg/L CaCO3
Alkp, mg/L CaCO3
T-HdF, mg/L CaCO3
T-Hdp, mg/L CaCO3
Ca-HdF, mg/L CaCO3
Ca-Hdp, mg/L CaCO3
WEEK#
2
4
6
8
..*
Qp/Qx (THM4), %
Alkb, mg/L CaCO3
T-Hdb, mg/L CaCO3
Ca-Hdb, mg/L CaCO3
, THM4b, ng/L
HAA5b, ^ig/L
"t
-
.
"
-. ytt ff
, ^
'-'-
._ % %
s ,
""-
-"''
f s
'
"*
f
.
-
^
*
s
Qp/QT(HAA5),%
Alkb, mg/L CaCO3
T-Hdb, mg/L CaCO3
Ca-Hdb, mg/L CaCO3
THM4b, ^g/L
HAA5b, ng/L
' '
% % %%
^
' - ""'"
^ >, ;
^
--
' '
s
" %
" f *
-.-,
' '
-
Notes: In the first section of this table, the feed and permeate concentrations are entered for each run.
The spreadsheet uses these values to calculate the flow that must be treated to meet the stage I
DBF MCLs with a 10% factor of safely (i.e.72/54 |ng/L - THM4/HAA5). The higher of the two
QP/Q! flow ratios (i.e. permeate flow to total flow) based on either THM4 or HAA5 controls the design.
The blended water quality parameters are also calculated for the feed / permeate blends.
If the permeate quality does not meet the MCLs prior to blending, then these calculations are meaningless.
-------
Table 6-19b Blending Calculations To Determine The Fraction Of Flow That Requires NF
Treatment To Meet The Stage II DBF Regulations
Utility name and address
ICR plant number
Membrane trade name
Phone number
FAX number
Contact person
Parameter
THM4F, ng/L
THM4D, jag/L
HAA5F, jig/L
HAA5D, ligfL
AlkF, mg/L CaCO3
AlkD, mg/L CaCOs
T-HdF, mg/L CaCO3
T-HdD, mg/L CaCO3
Ca-HdF, mg/L CaCO3
Ca-Hdp, mg/L CaCO3
WEEK#
2
4
6
8
...
THM4 Controls
Q,/QT(THM4),%
Alkb, mg/L CaCO3
T-Hdb, mg/L CaCO3
Ca-Hdb) mg/L CaCC-3
THM4b, ng/L
HAA5b, ^g/L
-,-,
^ '
s * "* % s
s >
> "
s
"
-
^s ,
< :
f i S-
, , ,
-
-
'
'
HAAS Controls
(VQT(HAA5),%
Alkb, mg/L CaCO3
T-Hdb, mg/L CaCO3
Ca-Hdb, mg/L CaCO3
THM4bJ Mg/L
HAA5b> lig/L
s
s S
- »V *'*<
*
\
'-
:
-"
-
4'
, ,,
Notes: In the first section of this table, the feed and permeate concentrations are entered for each run.
The spreadsheet uses these values to calculate the flow that must be treated to meet the stage II
DBF MCLs with a 10% factor of safety (i.e.36/27 |ig/L - THM4/HAA5). The higher of the two
Qp/Qi flow ratios (i.e. permeate flow to total flow) based on either THM4 or HAA5 controls the design.
The blended water quality parameters are also calculated for the feed / permeate blends.
If the permeate quality does not meet the MCLs prior to blending, then these calculations are meaningless.
-------
7.0 Cost Survey
The cost of a membrane system is generally divided into construction costs and operating
and maintenance costs. The largest components of construction costs are the membranes,
membrane skids and the building; and the largest components of O&M costs include
electricity, labor, membrane replacement and chemical costs (Suratt, 1991). However, the
cost of pretreatment and concentrate disposal can be significant components of both
construction and O&M costs.
In order to develop an estimate of the cost of membrane treatment for the control of DBFs,
the ICR requires the utility to provide estimates of several cost parameters. These parameters'
include design information such as the MTCW, cleaning frequency and the permeate quality,
and site-specific parameters such as the local cost of electricity and labor. Cost parameters'
must be provided for each membrane tested.
An assumption in this cost analysis is that a membrane process is being added to an
existing treatment facility. Therefore, some of the cost parameters requested in this section
may not be applicable. For example, some utilities may need to acquire additional land to
construct a membrane facility, while others may be in possession of the required land. If a
cost is not incurred, then zero (0) should be entered for that cost in the tables described in this
section.
The design parameters necessary for the cost analysis will be obtained from bench- or
pilot-scale membrane studies, and Table 7-1 lists these parameters along with example values.
The "total required plant production" is the demand that must be met by the treatment plant.
In a nanofiltration plant, the total production consist of the membrane train capacity and the
flow that by-passes the membrane train. If by-pass is not used, then the total required plant
capacity is equivalent to the membrane tram capacity. It should be noted that the required feed
flow rate to a membrane system is significantly higher than the membrane tram capacity and
can be calculated by dividing the membrane train capacity by the fractional recovery. The
average THM4 and HAAS permeate concentrations requested hi Table 7-1 should be averaged
over all of the data collected during the study for a specific membrane. The average plant feed
water temperature should reflect the yearly average water temperature experienced hi the full-
scale plant. This average temperature is used to normalize the MTCW to a common
temperature. The average temperature-normalized MTCW over the course of the study should
be reported along with minimum and maximum values. The range of acceptable operating
pressures for a specific membrane is also requested and will typically range from zero to the
maximum allowable pressure. The cleaning frequency must be obtained from an analysis of
the flux data collected over the study. The feed TDS and TDS rejection should be averaged
over all of the data collected during the study for a specific membrane.
Table 7-2 lists estimates of the building area requirements for a membrane facility. More
accurate site-specific information can be used to estimate the building area requirements if
available. The estimated building area will be used as one of the construction cost estimate
parameters listed in Table 7-3. In Table 7-2, the area estimates for the "membrane process
3-101
-------
equipment," "electrical room," and "chemical rooms" are based on the membrane train
capacity. Thus, the total area for each of these three rooms is calculated by multiplying the
area estimate by the membrane tram capacity. For example, the building area for the
membrane process equipment for a 10 mgd membrane tram would be 9500 ft2. The area
estimates for the other five rooms are independent of the membrane train capacity, and the
area for the generator room would be 400 ft2 for either a 10 or a 20 mgd plant:
The information required to make a construction cost estimate is listed in Table 7-3, and
the information required to make an O&M cost estimate is listed in Table 7-4. Example
values are included in these tables and should only be used as default values if site-specific
estimates are unavailable. As stated earlier, if a cost is not incurred then zero (0) should be
entered for that cost. The last item hi Table 7-3 asks if odor control would be required for the
product water. In most cases, odor hi membrane permeate is due to hydrogen sulfide which
would need to be stripped from the product prior to distribution.
The costs associated with chemicals must be estimated to develop an accurate O&M cost
projection. Table 7-5 requests information on chemicals used specifically for membrane
processes and the associated costs and dosages. Chemicals used in membrane processes
include pretreatment and post-treatment chemicals as well as cleaning agents. The cost of
chlorine is also requested since the low chlorine demand of membrane permeate can result hi a
significantly lower dose.
The cost of membrane pretreatment and concentrate disposal can be substantial additions to
the cost of a membrane plant. However, due to the variety of pretreatment and concentrate
disposal processes, it is impossible to generate a set of standard cost parameters. For the
purpose of the ICR, the plant is asked to report a pretreatment scheme and concentrate
disposal method which could be used at the specific plant. An example of a pretreatment
scheme and concentrate disposal method is shown in Table 7-6. In this example, pretreatment
consists of enhanced coagulation with a 50 mg/L increase in the alum dose along with
cartridge filtration and acid addition to a pH of 4.0, and the concentrate stream is assumed to
have a low TDS concentration < 1500 mg/L and can be discharged to the local sewer system
at a cost of $0.50 per 1000 gallons. If more accurate cost data for pretreatment and
concentrate disposal is available, then it should be provided to EPA.
3-102
-------
Table 7-1 Design Parameters For Cost Analysis
Utility name and address
ICR plant number
Phone number
Membrane trade name
Contact person
FAX number
Manufacturer
Design Parameter
Total required plant production (mgd;
Permeate to total flow ratio
By-pass flow rate (mgd)
Required membrane train capacity (mgd)
Average THM4 permeate concentration (ug/L)
Average HAAS permeate concentration (pg/L)
Average THM4 feed concentration (|J.g/L)
Average HAAS feed concentration (ug/L)
Average plant feed water temperature (°C)
Average temperature-normalized MTCW (gfd/psi)
Maximum temperature-normalized MTCW (gfd/psi)
Minimum temperature-normalized MTCW (gfd/psi)
Range of acceptable operating pressures (psi)
Average cleaning frequency (days)
Average feed TDS (mg/L)
Average TDS rejection (%)
Example values
10
0.8
2
8
21
9
102
90
18.3
0.18
0.25
0.16
0 to 225
60
150
30
Specific utility values
-------
Table 7-2 Estimated Building Area Requirements
Utility name and address
ICR plant number
Phone number
Membrane trade name
Contact person
FAX number
Manufacturer
Building area
Membrane process equipment (ft2 per mgd)
Electrical room (ft2 per mgd)
Chemical rooms (ft2 per mgd)
Control room (ft2)
Generator (ft2)
Transformer vault (ft2)
Offices and reception (ft2)
Bathrooms (ft2)
Area estimate
950
125
175
350
400
500
variable
250
Total area
-------
Table 7-3 Construction Cost Information
Utility name and address
ICR plant number
Phone number
Membrane trade name
Contact person
FAX number
Manufacturer
Cost Parameter
. Capital recovery interest rate (%)
Capital recovery period (years)
Overhead and profit factor (% of construction cost)
Special site-work factor (% of construction cost)
Construction contingencies (% of construction cost)
Engineering fee factor (% of construction cost)
Contract mobilization, insurance and bonds (% oi
construction costs]
ENR construction cost index (CCI base year 1913)
(date)
Producers price index (PPI base year 1967 = 100)
(date)
Building area requirements (ft2), from Table 7-2
. Building costs ($/ft2)
Land area requirements (ft2)
Land costs ($/ft2)
Cost of a standard 8"x 40" membrane element ($)
Area of a standard 8"x 40" membrane element (ft2)
Would sulfide concentrations necessitate odor control
(ves / no)
Example values
10
20
5
5
10
10
5
4965 (May 92)
326 (May 92)
14,500
100
20,000
0
1000
400
Specific utility values
-------
Table 7-4 O&M Cost Information
Utility name and address
ICR plant number
Phone number
Membrane trade name
Contact person
FAX number
Manufacturer
Cost Parameter
Labor rate + fringe ($/personnel-hour)
Labor overhead factor (% of labor)
Number of O&M personnel hours per week
Electric rate ($/kWh)
Membrane replacement frequency (%/year)
SDS chlorine demand of membrane feed (mg/L)
SDS chlorine demand of membrane permeate (mg/L)
Example values
15
10
80
0.086
12
8.0
1.5
Specific utility values
-------
Table 7-5 Cost And Dose For Chemicals Required For Membrane Treatment
Utility name and address
ICR plant number
Phone number
Membrane trade name
Contact person
FAX number
Manufacturer
Chemical1
Chlorine
Sulfuric acid
Alum
Hydrochloric acid
Antiscalant2
Caustic
Sodium Hydroxide
Phosphoric acid
Use for chemical
Disinfectant
Pretreatment
Pretreatment
Pretreatment
Pretreatment
Post-treatment
Membrane cleaning
Membrane cleaning
Chemical dose
Bulk chemical cost
1: Information for cleaning chemicals and pretreatment chemicals (such as alum) should also
be provided in this table. For cleaning agents, the concentration of the cleaning solution
used to clean the membranes should be reported as the chemical dose.
2: Report the product name and manufacturer of the specific antiscalant used.
-------
Table 7-6 Example Pretreatment Scheme And Concentrate Disposal Method
Pretreatment scheme
Conventional treatment train in existing plant (process data supplied separately)
50 mg/L increase in the alum dose during coagulation
5 um cartridge filtration
5 mL/gal of concentrated sulfuric acid to adjust the pH of the feed stream to 4.0
Concentrate disposal method
Discharge low TDS concentrate to local sewer at a cost of $0.50 per 1000 gallons
-------
8.0 References
Allgeier, S.C. and Summers, R.S., 1995. "Evaluating NF for DBF control with the
RBSMT," Jour. AWWA. 87:3:87.
Allgeier, S.C., Gusses, A.M., Speth, T.F., Westrick, JJ. and Summers, R.S., 1996.
"Verification of the Rapid Bench-Scale Membrane Test and ICR Requirements." Presented at
the AWWA GAG and Membrane Workshop, Cincinnati, OH.
American Society for Testing and Materials, 1982. "ASTM Test Method for Silt Density
Index, ASTM Test Designation D-4189-82." Philadelphia, PA.
Duranceau, S.J. and Taylor, J.S., 1990. "Investigation and Modeling of Membrane Mass
Transfer." In Proceedings National Water Improvement Supply Association.
Duranceau, S.J. and Taylor, J.S., 1993. "Solute Charge and Molecular Weight Modeling for
Prediction of Solute Mass Transfer Coefficients." In Proceedings AWWA Membrane
Technology Conference, Baltimore, Md.
Environmental Protection Agency, Monitoring Requirements for Public Drinking Water
Supplies; Proposed Rule, 40 CFR Part 141, Federal Register. 59(26), February 1994
Fouroozi, J., 1980. "Nominal Molecular Weight Distribution of Color, TOC, TTHM
Precursors and Acid Strength in a Highly Organic Potable Water Source." M.S. Thesis,
University of Central Florida, Orlando, FL.
Fox, K.R. and Lytle, D.A., 1993. "Cryptosporidiosis Outbreak: Investigation and
Recommendations." In Proceedings AWWA Water Quality Technology Conference, Miami,
FL.
Fox, K.R. and Lytle, D.A., 1994. "Cryptosporidium and the Milwaukee Incident." In
Proceedings National Conference on Environmental Engineering, Boulder, CO.
Fu, P., Ruiz H., Thompson, K., and Spangenber, C., 1994. "Selecting Membranes for
Removing NOM and DBF Precursors," Jour. AWWA. 86:12:55.
Gekas, V., 1988. "Terminology for Pressure-Driven Membrane Operations," Desalination.
68:77.
Hofman, J.A.M.H., Kruithof, J.C., Noij, Th.H.M. and Schippers, J.C., 1993. "Removal of
Pesticides and Other Contaminants with Nanofiltration" (hi Dutch), H2O. 26.
Internal Corrosion of Water Distribution Systems. 1985. AWARF/DVGW-Forschungsstelle
Cooperative Research Report, American Water Works Association Research Foundation,
Denver, Colo.
3-103
-------
Jacangelo, J.G., Lain6, J.M., Carns, K. E., Cummings, E. W. and Mallevaille, J., 1991.
"Low-Pressure Membrane Filtration for Removing Giardia and Microbial Indicators," Jour.
AWWAT 83:9:97.
Jones, P.A., Taylor, J.S., Morris, K.A., and Mulford, L.A., 1992. "Control of DBP
Precursors by Nanofiltration," Jour. AWWA. 84:12:104.
Kohl, H.R., McKim, T., Smith, D., Bedford, D. and Lozier, J.C., 1993. "Evaluating
Reverse Osmosis Membrane Performance on Secondary Effluent Pretreated by Membrane
Microfiltration." In Proceedings AWWA Annual Conference, San Antonio, TX.
Laine", J.M., Jacangelo, J. G., Cummings, E. W., Carns, K. E. and Mallevaille, J., 1993.
"Influence of Bromide on Low-Pressure Membrane Filtration for Controlling DBFs in Surface
Waters," Jour. AWWA. 85:6:87.
Lead Control Strategies. 1990. American Water Works Association Research Foundation,
Denver, Colo.
Lo Tan and Amy, G., 1991. "Comparing Ozonation and Membrane Separation for Color
Removal and Disinfection By-Product Control," Jour. AWWA. 83:5:74.
Lozier, J.C. and Carlson, M., 1991. "Organics Removal From Eastern U.S. Surface Waters
Using Ultra-Low Pressure Membranes." In Proceedings AWWA Membrane Technology
Conference, Orlando, Fla.
Lozier, J.C. and Carlson, M., 1992. "Nanofiltration Treatment of a Highly Organic Surface
Water." In Proceedings AWWA Annual Conference, Vancouver, British Columbia, Canada.
McCarty, P.L. and Aieta, E. M., 1983. "Chemical Indicators and Surrogate Parameters for
Water Treatment." In Proceedings of the 1983 AWWA Annual Conference, Denver, CO.
Membrane Concentrate Disposal. 1993. American Water Works Association Research
Foundation, Denver, Colo.
Morris, K.M., 1990. "Predicting Fouling in Membrane Separation Processes." M.S. Thesis,
University of Central Florida, Orlando, PL.
Porter, M.C., 1988. "Membrane Filtration." Handbook of Separation Techniques for
Chemical Engineers 2nd ed.. (P.A. Schweitzwer, editor in chief), McGraw-Hill Book
Company, New York.
Rosa, M.J. and Pinho, N., 1994. "Separation of Organic Solutes by Membrane Pressure-
Driven Processes," L_Merrjb_rjflieJ>cL., 89:235.
3-104
-------
Schippers, J.C. and Verdouw, J., 1980. "The Modified Fouling Index, A Method of
Determining the Fouling Characteristics of Water," Desalination. 32:137.
Schock, M.R., 1990. "Internal Corrosion and Deposition Control" in Water Quality and
Treatment. (F.W. Pontius, technical editor), AWWA and McGraw-Hill, Inc., New York.
Suratt, S.B., 1991. "Estimating the Costs of Membrane Water Treatment Plants." In
Proceedings AWWA Membrane Technology Conference, Orlando, Fla.
Taylor, J.S., Thompson, D.M., Snyder, B.R., Less, J. and Mulford, L.A., 1986. Cost and
Performance Evaluation of In-Plant Trihalomethane Control Techniques. EPA/600/S2-
85/138. Cincinnati, OH, USEPA Water Engineering Research Laboratory.
Taylor, J.S., Thompson, D.M. and Carswell, J.K., 1987. "Applying Membrane Processes to
Groundwater Sources for Trihalomethane Precursor Control," Jour. AWWA. 79:8:72.
Taylor, J.S., Mulford, L.A., Barrett, W.M., Duranceau, S.J. and Smith, O.K., 1990. Cost
and Performance of Membrane Processes for Organic Control on Small Systems.
EPA/600/S2-89/022. Crncinnati, OH, USEPA Water Engineering Research Laboratory.
Taylor, J.S., Reiss, C.R., Jones, P.S., Morris, K.E., Lyn, T.L., Smith, O.K., Mulford, L.A.
and Duranceau, S.J., 1992. Reduction of Disinfection By-Product Precursors by
Nanoffltration. EPA/600/SR-92/023. Cincinnati, OH, USEPA Water Engineering Research
Laboratory.
Taylor, J.S., Yousef, Y.A., Mulford, L.A. and Lyn, T.L., 1992. "Corrosion Control of
Finished Water from a Membrane Process." In Proceedings of the 1992 AWWA Annual
British Columbia, Canada.
U.S. Environmental Protection Agency, 1992. Lead and Copper Rule Guidance Manual
Volume II: Corrosion Control Treatment. EPA 811-B-92-002.
3-105
-------
-------
Appendix 3-A
Water Quality Prediction For Multi-Stage Systems
Define:
(1) RejTDS = TDS rejection per element.
(2) TDSP = TDSIx(l-RejTDS)
(3) R = System recovery = Qp/Qp
(4) r = Recycle ratio = QR/QF
(5) Ne = Number of elements per pressure vessel.
(6) Nv.s(1) = Number of pressure vessels per stage (i through j), Nv_s(0)=0.
TO
(7) Re = recovery for each element =
(8) TDSi = TDSw(0) = Feed water quality into first stage.
(1) Final Concentration Quality
(i)
where,
i=j k=Ne
n n
i=l k=l
1 + r - N.R.SN,,,,, / N,_ - kR, + R, x Re J
r-NeRe£Nv_ l/Nv_Si -kRe
(2)
(2) Final Permeate Quality
TDSp =|TDSF-^l-NeReZNv_sjTDSw | / [NeReZNv_Si
or
(3)
TDS =
R
(4)
3-107
-------
(3) Feed Water Quality
TDS, = TDSF + -^-TDS
1-f-r F 1 + r
(5)
(4) Water Quality in each stage
(a) Concentration
TDSw(i) = B x TDSW(M)
where,
B = 'fl
k-l
[(l + r - NCRCENV_SM) / Nv_Si j - kRe + Re x ReJTDS
«M,/N_J-kR.
(b) Permeate
((l + r-N^i
NR
Nv_SM /Nv_Sj -NeReTDS
w(i)
NeRe
'U.S. GOVERNMENT PRINTING OFFICE: 1996-750-001/41002
(6)
(7)
(8)
3-108
------- |