a. i
United States
Environmental Protection
Agency
Office of Water
Washington DC 20460
EPA-440/4-85-032
September 1985
Technical
Support Document
for Water Quality-based
Toxics Control
-------
Technical Support Document for
Water Quality-based Toxics Control
o
U.S. Environmental Protection Agency
Region 5, Library (PL-12J)
77 West Jackson Boulevard, 12th Floor
Chicago, IL 60604-3590
September, 1985
Office of Water Enforcement and Permits
Office of Water Regulations and Standards
U.S. Environmental Protection Agency
Washington, D.C. 20460
-------
Foreword
The U. S. Environmental Protection Agency (EPA) and the State pollution control
agencies have been charged with enforcing the laws regarding pollution of the nat-
ural environment. Environmental pollution is considered an urgent and continuing
problem and, consequently, the laws grant considerable discretion to the control
authorities to define environmental goals and develop the means to attain them.
Establishing environmental tolerance levels and incorporating them in a decision
making process entails a considerable amount of scientific knowledge and judg-
ment. One area where scientific knowledge is rapidly changing concerns the dis-
charge of toxic pollutants to the Nation's surface waters.
This document provides technical guidance for assessing and regulating the dis-
charge of toxic substances to the waters of the United States. It was issued in sup-
port of a recent EPA policy initiative involving the application of biological and
chemical assessment techniques to control toxic pollution. The recommendations
contained in this document are not mandatory and are intended to be suggestions
for approaching problems which tend to be complex and site-specific.
This document is expected to be revised periodically to reflect advances in this rapid-
ly evolving area. Comments from users will be welcomed.
l/o.
James M. Conlon, Acting Director Rebecca W. Hanmer, Director
Office of Water Regulations Office of Water Enforcement
and Standards and Permits
in
-------
Contents
Foreword jji
Contents v
Acknowledgment ix
Executive Summary xi
Water Quality Standards and Criteria xi
Effluent Characterization xi
Health Hazard Assessment xi
Exposure and Wasteload Allocation xii
Permit Requirements xii
Compliance Monitoring xii
Case Example xii
List of Abbreviations xiii
Glossary xv
Introduction xvii
Purpose xvii
Organization of the Technical Support Document xvii
1. Approaches to Water Quality-based Toxics Control 1
The Whole-effluent Approach 1
The Chemical-specific Approach 1
Advantages and Disadvantages of the Approaches 2
Issues Involved in the Toxicity-based Toxics Control Process 2
References 7
2. Water Quality Standards 9
Water Quality Standards 9
Mixing Zones 9
Magnitude, Duration, and Frequency 10
Magnitude for Single Chemicals 10
Magnitude for Whole-effluent Toxicity 10
Duration 11
Frequency 12
References 12
3. Effluent Characterization 13
Tiered Testing 14
Use of Toxicity Testing in Multiple-source Discharge Situations 21
Use of Effluent Characterization Data and Decision Criteria 22
Bioaccumulation 22
References 24
4. Human Health Hazard Assessment 25
Overview 25
Background 25
Bioaccumulation 26
-------
Status of Chemical vs. Biological Approaches 26
Priority Setting 27
Assessment and Control 28
References 30
5. Exposure and Wasteload Allocation 31
Introduction 31
Part 1 — The Mixing Zone and Toxic Wasteload Allocations 31
Part 2 — Wasteload Allocation Modeling 38
References 48
6. Permit Requirements 51
Overview 51
Basic Principles of Effluent Variability 51
Permit Limit Derivation 52
Ensuring Consistency with the WLA 52
Expressing Limitations 56
Toxicity Reduction Evaluation and Indicator Limits 57
Other Permit Requirements 58
References 59
7. Compliance Monitoring 63
Overview 63
Self-monitoring Requirements 63
Inspections 63
Enforcement 64
References 64)
8. Case Example 65
Site Description 65
Criteria Determination (Section 2) 65
Effluent Characterization (Section 3) 66
Exposure Assessment (Section 5) 76
Permit Limit Derivation (Section 6) 72
Appendix A. Policy for the Development of Water Quality-based
Permit Limitations for Toxic Pollutants A-1
Appendix B. Sampling B-1
Sampling B-2
Sampling Methods B-2
Sampling Frequency B-3
References B-4
Appendix C. Ambient Toxicity Testing and Data Analysis C-1
Ambient Toxicity Analysis C-2
Procedures C-2
Selecting Sampling Stations C-2
Analysis of Ambient Toxicity Measurement C-3
Appendix D. Duration and Frequency D-1
Duration D-2
Frequency D~4
Considerations for Setting Frequency for Minor Stresses D-4
Considerations for Setting Frequency for Areas under Major
Stresses and Sensitive Communities D-4
Recommendations for Duration and Frequency D-6
References D-7
VI
-------
Appendix E. Lognormal Distribution and Permit Limit Derivations E-1
Introduction E-2
Example Data E-2
Normal Distribution E-2
Probability and Log-probability Plots E-3
Lognormal Distribution E-4
Daily Permit Limit E-5
Monthly Permit Limit E-5
Consistency of Limits E-6
Derivation of Water Quality-based Limits E-6
Correlated Data E-8
Estimating ne from Correlated Data E-9
References E-9
VII
-------
Acknowledgment
The preparation of this document was a collaborative
effort by many individuals. The principal authors were
Rick Brandes and Bruce Newton of the Permits Division,
Elizabeth Southerland of the Monitoring and Data
Support Division, and Maurice Owens of the Office of
Analysis and Evaluation. Individuals listed below con-
tributed in the preparation of this document and their
efforts are greatly appreciated.
EFFLUENT CHARACTERIZATION WORKGROUP
Rick Brandes, Workgroup Chairman, U.S. EPA
Permits Division
Harold Bergman, University of Wyoming
Richard Kimerle, Monsanto Corporation
Donald Mount, U.S. EPA Environmental Research
Laboratory—Duluth
Jim Plafkin, U.S. EPA Monitoring and Data Support
Division
Lee Tebo, U.S. EPA Region IV
WASTELOAD ALLOCATION WORKGROUP
Jim Plafkin, Workgroup Chairman, U.S. EPA
Monitoring and Data Support Division
Kenneth Dickson, North Texas State University
Dominic DiToro, HydroQual, Inc.
Gordon Loewengart, Allied Chemical Corporation
Alan Maki, Exxon Corporation
EXPOSURE ASSESSMENT WORKGROUP
Mark Morris, Workgroup Chairman, U.S. EPA
Monitoring and Data Support Division
Thomas Barnwell, U.S. EPA Environmental Research
Laboratory—Athens
Charles Delos, U.S. EPA Monitoring and Data
Support Division
Lee Dunbar, Connecticut Department of Environ-
mental Protection
Paul Freedman, Limno-Tech Inc.
Tom Gallager, HydroQual, Inc.
Larry Roesner, Camp, Dresser, and McKee, Inc.
John Pawlow, U.S. EPA Criteria and Standards
Division
Elizabeth Southerland, U.S. EPA Monitoring and
Data Support Division
Nelson Thomas, U.S. EPA Environmental Research
Laboratory—Duluth
HUMAN HEALTH WORKGROUP
Richard Bull, U.S. EPA Environmental Research
Laboratory—Cincinnati
Lymon Condie, U.S. EPA Environmental Research
Laboratory—Cincinnati
Frank Gostomski, U.S. EPA Criteria and Standards
Division
Thomas Purcell, U.S. EPA Criteria and Standards
Division
CASE EXAMPLES WORKGROUP
Bruce Newton, Workgroup Chairman, U.S. EPA
Permits Division
Thomas Fikslin, U.S. EPA Region II
Marshall Hyatt, U.S. EPA Region IV
Robert McGhee, U.S. EPA Region IV
Walter Redmon, U.S. EPA Region V
Andrew Yasinsac, South Carolina Department of
Health
IX
-------
Executive Summary
The Technical Support Document (TSD) for Water
Quality-based Toxics Control presents recommenda-
tions to regulatory authorities when they are faced with
the task of controlling the discharge of toxic pollutants
to the nation's waters. It provides guidance for each
step in the water quality-based toxics control process
from screening to compliance monitoring. It also details
the current water quality criteria recommended by the
U.S. Environmental Protection Agency (EPA) and how
those criteria should be applied. The following is a brief
synopsis of the guidance provided in the TSD.
Water Quality Standards and Criteria
Where specific numerical criteria for a chemical or bi-
ological parameter (such as toxicity) are absent, com-
pliance with the standards must be based on the
general narrative criteria and on protection of the desig-
nated use (see National Policy, Appendix A). For many
pollutants, EPA-recommended criteria may be used, or
data on the toxicological effects of the pollutant may
be found in the literature or requested of a discharger.
EPA's recommended criteria for whole-effluent toxici-
ty are as follows. To protect aquatic uses against chron-
ic effects, the ambient toxicity should not exceed 1.0
chronic toxic unit (TUC) to the most sensitive of at
least three test species. For acute effects, the ambient
toxicity should not exceed 0.3 acute toxic units (TUa)
to the most sensitive of at least three test species. If
only two species are tested and used to evaluate com-
pliance with the criteria, the criteria listed above should
be divided by 10. This will account for the additional
uncertainty involved with using fewer than the recom-
mended three species.
Aquatic impacts are a function not only of magnitude,
but also of duration and frequency with which criteria
are exceeded. EPA's recommended criteria for both in-
dividual toxicants and whole-effluent toxicity are speci-
fied as two numbers: the criteria continuous concen
tration (CCC), applied as a four-day average concentra-
tion; and the criteria maximum concentration (CMC),
applied as a one-hour average concentration. The fre-
quency with which criteria are allowed to be exceed-
ed depends on site factors as explained in the text.
Effluent Characterization
A screening analysis is recommended to identify poten-
tial problem discharges. Because effluent toxicity and
chemistry are often unknown, the screening triggers
(or decision criteria) are based mostly on effluent dilu-
tion in the receiving water. Extensive data collection is
not mandatory in order to establish permit limitations.
The reasons for collecting additional data are: to evalu-
ate the degree of a water quality impact; to develop
variability information if a dynamic model will be used
for a wasteload allocation (WLA); and to test several
test species in order to identify the most sensitive, thus
avoiding the need to use a species sensitivity factor and
identifying a species for compliance monitoring pur
poses.
If additional assessment is desired, it is recommend-
ed that a tiered testing program be followed. At each
tier, effluent data and exposure data are combined and
compared to triggers using uncertainty factors to ac-
count for insufficient data. If no impact is projected, no
further assessment or control is required. If impact is
projected, regulatory authorities may require additional
data to be collected to eliminate the use of uncertainty
factors or may institute controls. The uncertainty fac-
tors and their associated data requirements are ex-
plained in Section 3.
Health Hazard Assessment
Health effects from toxic pollutants are divided into two
groups. Pollutants that are carcinogenic or mutagenic
are termed genotoxic. These pollutants are presumed
to have no safe exposure level (the "no threshold"
model), but incremental risk levels can be determined
based on the potency of the chemical. All other health
effects are termed target organ effects. Pollutants that
cause target organ effects are presumed to have safe
exposure levels. Exposure criteria or risk estimates are
generally developed using animal studies. It is gener-
ally impractical to develop such quantitative effects
data for complex mixtures such as effluents.
For effluent discharges, it is difficult to develop a site-
specific risk assessment because the pollutants dis-
charged are unknown, the degree of human exposure
is unknown, and (even if these were known) toxicolo-
gy information is not available for the vast majority of
chemicals. Several short term biological tests have
been developed, but these tests are difficult or impos-
sible to translate into potential health hazards, and they
cannot quantify the degree of toxicity to humans.
Therefore, short term tests are recommended only for
screening and priority setting. If screening indicates a
XI
-------
potential health hazard, the risk assessment should be
based on chemical characterization, available toxico-
logical data and structure activity relationships, and an
exposure assessment of the target population.
Exposure and Wasteload Allocation
In order to determine an effluent composition that will
protect aquatic organisms, the exposure of the aquat-
ic community to the effluent must be determined. Ex-
posure assessment includes an analysis of how much
of the water body is subject to the criteria being ex-
ceeded, for how long, and how frequently. The first step
is to evaluate the effluent plume dispersion. If mixing
is not rapid and complete, the wasteload allocation
must be based on a mixing zone analysis (where State
standards allow a mixing zone). Means to assess dilu-
tion at the edge of a mixing zone are described in
Section 5.
If mixing is rapid and complete, there are several models
that can be used to assess exposure. The simplest is
a steady state model that assumes the toxicant or tox-
icity does not decay. This model uses a mass balance
equation of the following form:
concentration in effluent = concentration in
dilution factor receiving water
where the inputs are steady state estimates usually
designed to reflect some "worst case." Deriving an
acceptable effluent concentration to protect water
quality simply involves rearranging the equation as
follows:
criteria concentration x dilution factor =
maximum acceptable concentration in effluent
Steady state models work on the assumption that the
effluent concentration is steady and that the duration
and frequency with which criteria are exceeded can be
reflected entirely by selecting a critical flow condition
in the receiving water of appropriate duration and fre-
quency (see Appendix D).
However, the assumption that effluent concentration
is steady state does not hold for most effluents. A sec-
ond problem is that the acceptable effluent concentra-
tion derived from a steady state model does not include
any information about acceptable effluent variability.
Therefore, an interim lexicologically based steady state
modeling approach is recommended to provide some
variability information along with the acceptable ef-
fluent concentration projections.
The best means of modeling exposure is to use com-
puter models that incorporate the variability of the
individual inputs (such as effluent flow and concentra-
tion, receiving water flow, temperature, background
concentration, etc.). These models are termed dynamic
models and are much more accurate than steady state
models in reflecting or predicting exposure provided
adequate data exist. The acceptable effluent condition
derived using these models is expressed as the effluent
long term average (LTA) and variance, which greatly
simplifies derivation of permit limits. Three dynamic
modeling approaches are described along with instruc-
tions for their use.
Permit Requirements
The requirements of a WLA must be incorporated in the
wastewater discharge permit. In many cases permit
limits will be different than the WLA to reflect differ-
ent assumptions and means of expressing effluent
quality. Three types of WLAs are identified, and recom-
mendations are provided for deriving permit limits to
properly enforce each type of WLA. Other permit-
related issues such as data collection, limit expression,
and compliance schedules are discussed.
Compliance Monitoring
For important parameters, at least once per week
monitoring is recommended. A phased monitoring ap-
proach is discussed in which monitoring frequency is
reduced after continuing compliance is demonstrated.
Unique considerations for enforcement of water
quality-based permits are discussed.
Case Example
A multiple source, "integrated" approach case exam-
ple based on data from several toxic discharge sites is
provided. The decisions made regarding screening, ef-
fluent characterization, exposure assessment, and per-
mit limit derivation are described.
XII
-------
List of Abbreviations
fj.g microgram LTA
fim micrometers MEXAMS
AA atomic absorption mg
ACR acute-chronic ratio MGD
ADI acceptable daily intake MICH
BAT best available technology ml
BCF bioconcentration factor MS
BCT best conventional technology MSDS
BOD biological oxygen demand Ni
BPJ best professional judgment NLM
BPT best practicable technology NOEL
CAG Carcinogen Assessment Group (EPA) NPDES
CCC criterion continuous concentration
Cd cadmium OECD
CETIS Complex Effluent Toxicity Information
System ORD
cfs cubic feet per second POTW
CHNTRN Channel Transport Model SAR
CIS Chemical Information System SERATRA
CMC criterion maximum concentration SLSA
CN cyanide STP
Cr+6 chromium TMDL
CTAP Chemical Transport and Analysis Program TOC
Cu copper TODAM
CV coefficient of variation
DMR discharge monitoring report TOXIWASP
DO dissolved oxygen TOXIC
DOC dissolved organic carbon
EPA U.S. Environmental Protection Agency TSCA
ERL Environmental Research Laboratory (EPA) TSS
EXAMS Exposure Analysis Modeling System TUa
FDA Food and Drug Administration TUC
FETRA Finite Element Transport Model USGS
FIFRA Federal Insecticide, Fungicide, and WASTOX
Rodenticide Act WHO
g grams WLA
GC gas chromatography Zn
HPLC high-pressure liquid chromatography
HSPF Hydrologic Simulation Program - FORTRAN
IHS in-house software
IWC instream waste concentration
kg kilograms
I liter
LC50 lethal concentration killing 50% of exposed
organisms
long term average
Metals Exposure Analysis Modeling System
milligrams
million gallons per day
Michigan River Model
milliliter
mass spectroscopy
Monitoring and Data Support Division (EPA)
nickel
National Library of Medicine
no observable effect level
National Pollutant Discharge Elimination
System
Organization of Economic Cooperation
Development
Office of Research and Development (EPA)
publicly owned treatment works
structure-activity relationships
Sediment Contaminant Transport Model
Simplified Lake/Stream Analysis
sewage treatment plant
total maximum daily load
total organic carbon
Transport One-Dimensional Degradation and
Migration Model
Chemical Transport and Fate Model
Toxic Organic Transport and
Bioaccumulation Model
Toxic Substances Control Act
total suspended solids
acute toxic unit
chronic toxic unit
U.S. Geological Survey
Estuary and Stream Quality Model
World Health Organization
wasteload allocation
zinc
XIII
-------
Glossary
acute involving a stimulus severe enough to rapidly induce
a response; in toxicity tests, a response observed in 96
hours or less typically is considered acute. An acute effect
is not always measured in terms of lethality; it can meas-
ure a variety of effects. Note that acute means short, not
mortality.
acute-chronic ratio (ACR) the ratio of the acute toxicity
(expressed as an LC50) of an effluent or a toxicant to its
chronic toxicity (expressed as an NOEL). It is used as a
factor for estimating chronic toxicity on the basis of acute
toxicity data.
additivity the characteristic property of a mixture of toxi-
cants that exhibits a cumulative toxic effect equal to the
arithmetic sum of the effects of the individual toxicants.
ambient toxicity toxicity manifested by a sample collected
from an aquatic receiving system.
antagonism the characteristic property of a mixture of tox-
icants that exhibits a less-than-additive cumulative toxic
effect.
bioaccumulation uptake and retention of substances by an
organism from its surrounding medium and from food.
bioassay a test used to evaluate the relative potency of a
chemical by comparing its effect on a living organism with
the effect of a standard preparation on the same type of
organism. Bioassays are frequently used in the pharmaceu-
tical industry to evaluate the potency of vitamins and drugs.
"Bioassay" and "toxicity test" are not synonymous.
bioconcentration uptake of substances from the sur-
rounding medium through gill membranes or other exter-
nal body surfaces.
bioavailability the property of a toxicant that governs its
effect on exposed organisms. A reduced bioavailability
would have a reduced toxic effect.
chronic involving a stimulus that lingers or continues for a
relatively long period of time, often one-tenth of the life
span or more. Chronic should be considered a relative term
depending on the life span of an organism. A chronic ef-
fect can be lethality, growth, reduced reproduction, etc.
Chronic means long.
conservative pollutant a pollutant that is persistent and not
subject to decay or transformation.
continuous stimulation model a fate and transport model
that uses timeseries input data to predict receiving water
quality concentrations in the same chronological order as
that of the input variables.
criteria continuous concentration (CCC) the EPA national
water quality criteria recommendation for the highest in-
stream concentration of a toxicant or an effluent to which
organisms can be exposed indefinitely without causing un-
acceptable effect.
criteria maximum concentration (CMC) the EPA national
water quality criteria recommendation for the highest in-
stream concentration of a toxicant or an effluent to which
organisms can be exposed for a brief period of time with-
out causing mortality.
design flow the critical flow used for steady state wasteload
allocation modeling.
critical life stage the period of time in an organism's lifespan
in which it is the most susceptible to adverse effect caused
by exposure to toxicants, usually during early development
(egg, embryo, larvae). Chronic toxicity tests are often run
on critical life stages to replace long duration, life cycle tests
since the toxic effect occurs during the critical life stage.
diversity the number and abundance of species in a
specified location.
duration the period of time over which the instream concen-
tration is averaged for comparison with criteria concentra-
tions. This specification limits the duration of concentra-
tions above the criteria.
effluent biomonitoring the measurement of the biological
effects of effluents (such as toxicity, biostimulation, and
bioaccumulation).
Final Acute Value (FAV) an acute toxicity limit in which the
measured acute toxicity expressed as an LC50 of an ef-
fluent or a toxicant is adjusted by a factor (0.3 is recom-
mended) to eliminate mortality.
frequency how often criteria can be exceeded without un-
acceptably affecting the community.
LC50 the toxicant concentration killing 50% of exposed
organisms at a specific time of observation.
lognormal probabilistic dilution model a dilution model that
calculates the probability distribution of receiving water
quality concentrations from the lognormal probability dis-
tributions of the input variables.
magnitude how much of a pollutant (or pollutant parameter
such as toxicity), expressed as a concentration or toxic unit
is allowable.
Monte Carlo simulation a stochastic modeling technique that
involves the random selection of sets of input data for use
in repetitive model runs in order to predict the probability
distributions of receiving water quality concentrations.
No Observed Effect Level (NOEL) the highest measured con-
tinuous concentration of an effluent or a toxicant that
causes no observed effect on a test organism.
permit averaging period the duration of time over which a
permit limit is calculated — day(s), week, or month.
persistence that property of a toxicant or an effluent that is
a measurement of the duration of its effect. A persistent
toxicant or toxicity maintains effect after mixing, degrad-
ing slowly. A non-persistent toxicant or toxicity may have
xv
-------
a quickly reduced effect after mixing as degradation proc-
esses such as volatilization, photolysis, etc., transform the
chemical.
plug flow sampling a monitoring procedure that follows the
same slug of wastewater throughout its transport in the
receiving water. Water quality samples are collected at
receiving water stations, tributary inflows, and point source
discharges only when a dye slug or tracer passes that point.
probability a number expressing the likelihood of occurrence
of a specific event, such as the ratio of the number of out-
comes that will produce a given event to the total number
of possible outcomes.
probability distribution a mathematical representation of the
probabilities that a given variable will have various values.
recurrence interval the average number of years within
which a variable will be less than or equal to a specified
value. This term is synonymous with return period.
steady state model a fate and transport model that uses con-
stant values of input variables to predict constant values
of receiving water quality concentrations.
STORET EPA's computerized water quality data base that in-
cludes physical, chemical, and biological data measured in
waterbodies throughout the United States.
sublethal involving stimulus below the level that causes
death.
synergism the characteristic property of a mixture of toxi-
cants that exhibits a greater-than-additive cumulative toxic
effect.
total maximum daily load (TMDL) the total allowable pollu-
tant load to a receiving water such that any additional load-
ing will produce a violation of water quality standards.
toxicity test the means to determine the toxicity of a
chemical or an effluent using living organisms. A toxicity
test measures the degree of response of an exposed test
organism to a specific chemical or effluent.
toxic unit acute (TUa) the reciprocal of the effluent dilution
that causes the acute effect by the end of the acute
exposure period.
toxic unit chronic (TUC) the reciprocal of the effluent dilu-
tion that causes no unacceptable effect on the test organ-
isms by the end of the chronic exposure period.
uncertainty factors factors used in the adjustment of tox-
icity data to account for unknown variations. Where toxic-
ity is measured on only one test species, other species may
exhibit more sensitivity to that effluent. An uncertainty fac-
tor would adjust measured toxicity upward and downward
to cover the sensitivity range of other, potentially more or
less sensitive species.
wasteload allocation (WLA) the portion of a receiving
water's total maximum daily pollutant load that is allocated
to one of its existing or future point sources of pollution.
whole-effluent toxicity the aggregate toxic effect of an
effluent measured directly with a toxicity test.
XVI
-------
Introduction
Purpose
The purpose of this Technical Support Document (TSD)
for Water Quality-Based Toxics Control is to provide
procedural recommendations for identifying, analyzing,
and controlling adverse water quality impacts caused
by the discharge of toxic pollutants to the waters of the
United States. The guidance supplied in this document
is meant to support implementation of the U.S. Environ-
mental Protection Agency's (EPA) national "Policy for
the Development of Water Quality-Based Permit Limi-
tations for Toxic Pollutants" (Appendix A).
The recommendations presented in this document are
not binding. They have been developed as the most
practical recommendations for the current state of
knowledge. The field of assessment of water quality
impacts from toxic pollutants is evolving and, therefore,
recommended techniques can be expected to evolve
also. EPA is currently developing detailed guidance
related to several aspects of the control of toxic pollu-
tants. These efforts are referenced at appropriate
points in the document.
The need for assessment and control of toxic pollutants
should not be underestimated. The EPA national poli-
cy statement was issued because of evidence that
many industrial and municipal sources continue to dis-
charge toxic substances in amounts significant to the
waterbody. Data collected through the EPA's ongoing
Complex Effluent Toxicity Testing Program indicate that
increased attention should be directed toward munici-
pal point sources treating industrial wastewaters.
Development of techniques to assess and control tox-
ics has focused primarily on individual toxicants. Ana-
lytical methods and fate and transport models are
established to assess the effects of the discharge of
specific chemicals. Toxicity, as measured through tox-
icity testing of effluents, has not been widely used as
a control parameter in the National Pollutant Discharge
Elimination System (NPDES) permits program. Now,
EPA's national policy recommends an integrated ap-
proach to controlling toxic pollutants using either or
both of these approaches, as appropriate.
Organization of the Technical Support
Document
Section 1 discusses the advantages and disadvantages
of the two approaches and also includes a discussion
of certain issues associated with the assessment and
control of toxic pollutants.
The relationship of subsequent sections of the docu-
ment to the process of water quality management is
schematically presented in Figure 1. Each section is
noted according to the aspect of the process it covers.
Section 2, Water Quality Criteria and Standards, con-
tains a brief discussion of water quality standards and
the derivation of ambient criteria to attain and maintain
standards.
Section 3, Effluent Characterization, describes proce-
dures that provide the data needed to make decisions
regarding toxic impact. A two-tiered assessment ap-
proach is recommended. In Tier 1 (Screening), a proc-
ess for cost-effective screening is described. In Tier 2
(Definitive Data Generation), a data collection process
is described in which increasingly more complete data
are generated in stages using uncertainty factors to
judge whether or not an effluent exceeds a dilution-
based margin of safety.
Section 4, Human Health Hazard Assessment, de-
scribes procedures for assessing potential human
health hazards associated with effluents. Some test-
ing procedures for measuring bioaccumulation are also
presented.
Section 5, Exposure Assessment, describes proce-
dures for estimating exposure to toxicants and calculat-
ing effluent requirements based on the ambient criteria
and exposure. Exposure is expressed in terms of con-
centration, area, duration (how long will an exposure
level exist), and frequency (how often will an exposure
level be experienced).
The analytical techniques are presented in order of
complexity, from simple dilution evaluations to more
advanced modeling techniques and probability ana-
lyses. A discussion of mixing zones is included.
Section 6, Permit Requirements, describes methods to
develop water quality-based NPDES permits. Available
permitting options for toxics control are described in-
cluding procedures for the establishment of Toxicity
Reduction Evaluations by individual municipal and in-
dustrial dischargers.
Section 7, Compliance Monitoring, provides recom-
mendations for establishing monitoring requirements
in NPDES permits.
XVII
-------
Set permit limits directly
Define water quality
objectives, criteria, and standards
Establish priority
water bodies
Section 2
Section 4
Screen for individual chemicals
including potential bioaccumulative,
carcinogenic, or mutagenic
chemicals
Collect definitive data
for specific chemicals
Screen for effluent
toxicity
Collect definitive data
for effluent toxicity
Section 3
Evaluate exposure
(critical flow, fate modeling,
and mixing) and calculate
wasteload allocation
Define required discharge
characteristics by the
wasteload allocation
Derive permit
requirements
Section 5
Evaluate toxicity reduction
Investigate indicator
parameters
Section 6
Final permit with
monitoring requirements
Figure 1. Overview of the water quality-based toxics control process.
Section 7
Section 8, Case Example, illustrates how the recom-
mended procedures from each preceding section are
applied in a toxic water quality impact situation. A mul-
tiple source toxic impact site is described, and effluent
characterization, exposure assessment, permitting,
and monitoring requirements for the site are detailed
step by step. The purpose of this example is to tie the
recommendations contained in each section together
to provide an overall picture of the water quality-based
toxics control process.
There are five appendices to this document. Several of
these are more detailed discussions of specific subjects
referenced in the text of the document. Others provide
data or examples of different techniques.
XVIII
-------
1. Approaches to Water Quality-based Toxics Control
Two approaches to the assessment and control of tox-
ics are presented in this document: the whole-effluent
approach and the chemical-specific approach.
The Whole-effluent Approach
The whole-effluent, or toxicity-based, approach to tox-
ics control involves the use of toxicity tests to meas-
ure the toxicity of industrial and municipal wastewater
discharges. A toxicity test is a relatively simple labora-
tory procedure. An effluent sample is collected and
diluted in test chambers; usually the dilutions used are
100%, 30%, 10%, 3%, 1%, and a control. Often the
receiving water is used to dilute the effluent because
it more closely simulates effluent/receiving water inter-
actions. Test organisms are placed in these test cham-
bers for specified periods of time. At various points
during the testing period, the number of organisms af-
fected in each chamber is counted (the endpoint can
be mortality, lower fecundity, reduced growth rates,
etc.) and the lowest effluent concentration that causes
that endpoint is calculated. This endpoint concentra-
tion becomes a quantified measure of the concentra-
tion that would cause instream impact if exceeded for
a particular period of time. It is usually stated either as
an LC50 (the effluent concentration at which 50% of
the test organisms are killed) or a No Observed Effect
Level or NOEL (the highest effluent concentration at
which no unacceptable effect will occur even at con-
tinuous exposure).
The measurement of whole-effluent toxicity can then
be used to limit the discharge of toxicants in an effluent.
Toxicity itself is used as the effluent parameter. The tox-
icants creating that toxicity need not be specifically
identified or controlled where the effluent's toxicity is
limited. An analogy between effluent toxicity and bi-
ochemical oxygen demand (BOD) can be drawn. Both
are measurements of a biological effect. Both can be
quantified. In neither case are the causative agents of
the biological effect specifically identified. Thus,
whole-effluent toxicity is like BOD in that it is a useful
parameter for reducing an undesirable effect caused by
the discharge of a complex mixture of waste materials.
Permit limits can be expressed as LC50s or NOELs (e.g.,
the effluent shall exhibit an LC50 or NOEL of greater
than 10%). The proper derivation of such toxicity-
based permit limits is described in Section 6. Due to the
procedures involved in deriving the toxicity limits and
since toxicity involves an inverse relationship (the lower
the LC50 or NOEL the higher the toxicity of the ef-
fluent), it is better to translate concentration-based
toxicity measurements into toxic units (TUs). In this
way, the problem of the inverse relationship is over-
come and the permit limit derivation process is better
served. A toxic unit is simply 100 divided by the toxic-
ity measured:
100
TU =
LC50 or NOEL
(where LC50 or NOEL is expressed as
percent effluent in dilution water)
An effluent with a toxicity of 10% is an effluent con-
taining 10 TUs. A very important aspect of toxic units
is that there are two different types depending on
whether acute or chronic toxicity is measured. The
proper expression of toxic units is TUa and TUC. TUa is
the measurement of acute toxicity units, and TUC is a
measurement of chronic toxicity units (see the glossary
for a definition of these terms). They are not the same
measurement and should not be used interchangeably.
Acute and chronic TUs make it easy to quantify accept-
able instream toxicity. By setting numerical water qual-
ity criteria for acute and chronic toxicity, regulatory
agencies can quantify narrative State standards such
as the "no toxics in toxic amounts" standard common
to all State water quality standards. The recommended
numerical whole-effluent toxicity criteria are described
in Section 2.
The Chemical-specific Approach
The chemical-specific approach to toxics control in-
volves the use of laboratory-generated water quality
criteria or State standards to limit specific toxicants
directly. The toxicity analysis of those chemicals is
done in a comprehensive testing program that, unlike
whole-effluent testing, attempts to consider a range of
toxic endpoints including human health impact and bio-
accumulation. Once a criterion is developed, the num-
ber is applied, through an exposure analysis, as a permit
limit to ensure that the level of that toxicant is not ex-
ceeded after discharge. Fate modeling of the toxicant
to estimate its behavior after discharge can be an im-
portant step in establishing water quality-based permit
limits.
-------
Advantages and Disadvantages of the
Approaches
The principal advantages of chemical-specific tech-
niques are:
• Treatment systems are more easily designed to
meet chemical requirements because more treat-
ability data are available and treatment engineers
and permit writers are more familiar with the
procedures.
• The fate of a pollutant can be measured through
modeling.
• Chemical analyses, in simple cases, can be less ex-
pensive than toxicity testing.
• Specific problem chemicals such as carcinogens
or bioaccumulative chemicals can be directly
limited.
The principal disadvantages of chemical-specific tech-
niques are:
• All toxicants in complex wastewaters may not be
known and, therefore, control requirements for all
potential toxicants cannot be set.
• It is not always clear which is the toxic com-
pound^) in the mixture.
• It can become expensive to measure individual tox-
icants, particularly where many are present in the
mixture. Organic chemicals can, in particular, be
costly to measure.
• The bioavailability of the toxicants at the discharge
site are not assessed, and the interactions be-
tween toxicants (e.g., additivity, antagonism) are
not measured or accounted for.
The principal advantages of the whole-effluent ap-
proach are:
• The aggregate toxicity of all constituents in a com-
plex effluent is measured, and toxic effect can be
limited by limiting one parameter — effluent
toxicity.
• The bioavailability of the toxic constituents is as-
sessed, and the effects of interactions of consti-
tuents are measured.
The principal disadvantages of the whole-effluent ap-
proach are:
• Effluent toxicity treatability data are lacking, and
treatment engineers and permit writers are un-
familiar with the procedures.
• Where there are chemical/physical conditions (pH
changes, photolysis, etc.) present that act on tox-
icants in such a way as to "release" toxicity down-
stream, such toxicity may not be measured in the
effluent.
• Properties of specific chemicals in complex ef-
fluents (such as bioaccumulation, carcinogenici-
ty) are not assessed.
Since each approach can complement the other, this
document has been written to integrate them. This
document focuses on the whole-effluent approach as
a toxics control mechanism because it is a relatively
new process when compared to the chemical-specific
approach.
Experience has shown that certain issues can cause
unnecessary problems and delays in toxics control be-
cause of a misunderstanding of their relative impor-
tance or technical merit in a water quality-based
permitting situation. This is particularly true for the
whole-effluent toxicity testing approach. The follow-
ing discussion focuses on issues involved in the appli-
cation of this approach.
Issues Involved in the Toxicity-based
Toxics Control Process
Correlation of Effluent Toxicity
Measurements to Actual Instream Impact
EPA has devoted much attention to this issue. Consider-
able criticism has been directed at the use of effluent
toxicity tests to assess adverse water quality impact
because it had not been shown that levels of effluent
toxicity measured in aquatic toxicity tests correlated
directly to instream toxicity and corresponding ecosys-
tem degradation.
Research conducted specifically to address this criti-
cism, EPA's Complex Effluent Toxicity Testing Program,
has shown that at each of the eight sites investigated
to date (Lima, Ohio; Circleville, Ohio; Birmingham, Al-
abama; Baltimore, Maryland; Waterbury, Connecticut;
Enid, Oklahoma; Charleston, West Virginia; and Wheel-
ing, West Virginia), when exposure and potential chem-
ical interactions are assessed, effluent toxicity can be
correlated directly to impact in receiving waters [1-8].
This relationship is proving valid in both "unimpacted"
sites (Circleville) and "impacted" sites (the other sites
listed above).
There is a growing consensus among aquatic biologists
that an evaluation of effluent toxicity, when adequately
related to instream conditions, can give a valid assess-
ment of receiving system impact [9].
This is also true in the application of chemical-specific
water quality criteria for toxicants. Provided exposure
is adequately assessed, measured instream concentra-
tions of individual chemicals exceeding established
safe concentrations can be considered to cause unac-
ceptable impact to the receiving system.
Toxicity Test Method Precision
Considerable concern has been expressed about the
precision and reproducibility of toxicity tests because
it was believed that the tests themselves exhibited a
large inherent variability. Research into this issue by var-
ious groups has shown that toxicity test methods,
where properly followed, exhibit an acceptable range
-------
of variability [10-18]. This variability is similar to test
method variabilities exhibited by accepted analytical
procedures for individual chemicals.
On-site vs. Off-site Toxicity Testing
Comparisons of toxicity data between tests conduct-
ed on-site to tests conducted off-site on samples
shipped to Environmental Research Laboratory (ERL)-
Duluth via airfreight have shown little variation. For
many effluents, on-site or off-site test data do not ap-
pear to be significantly different. The major consider-
ation is practicality.
Flow-through toxicity tests require a continuously
pumped sample. If it is determined that flow-through
tests are needed, on-site testing may be mandatory.
Cost should be weighed against data needs to make
this determination. If it is not considered important to
the analysis of toxic impact, off-site testing (which is
cheaper and can result in the generation of more data)
is as acceptable as on-site testing.
Flow-through vs. Static and Renewal
Toxicity Testing
A flow-through toxicity test is often conducted using
a diluter system and a continuous feed of effluent and
dilution water. A static toxicity test is conducted in test
chambers, without a serial diluter delivery system, into
which effluent and diluent are added manually. Only
one test sample is added at the beginning of the test.
A renewal toxicity test uses the same delivery system
as that of a static test but the test solutions are
changed on a pre-determined schedule.
There are several factors to consider in making the
choice of test system: what type of toxicity is being
measured (is the wastewater highly variable or not;
continuous or intermittent discharge?), how much data
need to be obtained (variable effluents may require
more data), and expense. Each testing method meas-
ures effluent toxicity. However, on-line continuous
flow-through testing can catch "peaks" of toxicity
(should they occur during the testing period) in varia-
ble effluents whereas static or renewal tests have a
lesser probability of doing so.
Cost is also a factor. Flow-through tests are more re-
source intensive and require complex delivery systems.
Consequently, less data can be generated than with
static or renewal testing. Where more data at less cost
are desirable, static or renewal testing is probably more
appropriate.
Variability
There are three important sources of variability in a
water quality impact analysis:
• Effluent variability caused by changes in the com-
position of the effluent.
•Exposure variability caused by changes in flow
rates of both effluent and receiving water. There
are also variable receiving water parameters that
are independent of flow, such as background toxi-
cant levels, pH, suspended solids, hardness, and
temperature, that can be important in assessing
impact.
• Species sensitivity variability caused by the differ-
ences in response to toxicants between species.
Effluent Variability
Effluent variability can be handled by a proper sampling
and testing procedure. Different sampling techniques
will result in the measurement of different aspects of
toxic impact. For example, composite samples tend to
smooth out peaks of toxicity but do stand a better prob-
ability of sampling those periods of peak toxicity. Grab
sampling will measure peak toxicity but only if the sam-
ple is taken during that period of peak toxicity. Since
this period may be very short, there can be a low prob-
ability of catching that peak. The probability will in-
crease by requiring analysis of more grab samples.
Sampling measurements should be tailored to that tox-
icity measurement of importance and the need to de-
sign testing that accounts for effluent variability.
Section 3 describes recommendations for testing fre-
quency designed to assess variable effluents. Appen-
dix B details suggested sampling procedures.
Exposure Variability
Exposure variability is a complex factor that can be ad-
dressed in two ways. A simpler, more easily applied ap-
proach is to assume a steady state exposure condition
(usually an estimate of presumed "worst case" ex-
posure using a low critical receiving water flow and a
typical effluent flow).
A second method is to attempt to estimate or actually
measure the variable exposure state at the discharge
site. This requires statistical analysis and some form of
dynamic modeling. Section 5 of this document de-
scribes appropriate exposure assessment procedures.
Species Sensitivity Variability
Different species exhibit different sensitivities to toxi-
cants. Often several orders of magnitude difference ex-
ist between the least sensitive and the most sensitive
species when they are exposed to a particular toxicant.
This range varies greatly and can be narrow or wide de-
pending on the toxicant involved. Therefore, generali-
zations about sensitive and non-sensitive species when
considering effluent toxicity are difficult to make.
Trout are considered sensitive organisms requiring high
quality water. This generalization is almost universally
accepted. When dealing with toxicity, however, this
generalization may be inaccurate. Trout are very sensi-
tive to oxygen depletion but may be relatively insensi-
tive to certain toxicants. Daphnia magna, a very
sensitive test species when exposed to many toxicants,
-------
is relatively insensitive to exposure to the pesticide en-
drin. Bluegills are very resistant to metals, particularly
copper. Conversely, they are in the sensitive range of
test species for organophosphate pesticides. Figures
1-1, 1-2, and 1-3 graphically show these differences in
species sensitivities to hexavalent chromium, dieldrin,
and an effluent from a publicly owned treatment works
(POTW), respectively. The wide range between sensi-
tivities for the different test species is shown. Compar-
ing the figures shows that the fish and the invertebrates
shift relative sensitivities to the toxicants. The fish are
less sensitive to chromium but more sensitive to diel-
drin. For the cladocerans, the reverse is true. A compar-
ison of species sensitivities to individual toxicants has
recently been published [19].
The primary goal in establishing a toxicity testing re-
quirement for a discharger is to find a sensitive test spe-
cies. The indigenous biota exposed to that effluent in
the receiving water are likely to exhibit a lesser sensi-
tivity to that effluent than the sensitive test species and
will be protected so long as that test species' no effect
level is not exceeded.
Since the measured toxicity of an effluent will be
caused by unknown toxic constituents, the relative
sensitivities of the test species will also be unknown.
Therefore, proper effluent toxicity analysis requires an
assessment of a range of sensitivities of test species
to that effluent. A knowledge of the range is necessary
so that the regulatory authority can identify a test spe-
cies among the different species tested that is truly
sensitive relative to the indigenous biota. The only way
to assess the range of sensitivities is to test a number
of different species. To determine how many species
to test, cost must be balanced against reducing scien-
tific uncertainty.
Analysis of species sensitivity ranges found in the na-
tional water quality criteria documents indicates that
if tests are conducted on three particular species
(Daphnia magna, Pimephales promelas, and Lepomis
macrochirus), the most sensitive of the three will have
an LC50 within one order of magnitude of the most
sensitive of all species tested [20]. This was found to
be true for 71 of the 73 priority pollutants tested with
four or more species.
To best balance cost with toxicological certainty, EPA
recommends three species be considered the number
of test species needed to eliminate a sufficient portion
of the uncertainty for this factor.
Often, regulatory agencies have been requiring testing
on resident species under the assumption that such
tests are needed to assess impact to local biota. How-
ever, standard test species, which are easier to test
than resident species (based on availability, etc.) will be
representative of expected impact on resident species
in most cases so long as the range of sensitivities is as-
sessed and a sensitive test species is identified. Resi-
Q.
-------
concentration using an available acute toxicity data
point. The most widely applied ACRs are 20 and 100;
that is, chronic toxicity concentration is one-twentieth
to one one-hundredth of the acute toxicity level. Com-
monly, 20 has been used for non-persistent toxicants
and 100 has been used for persistent toxicants. These
numbers have been used for both chemical-specific
and whole-effluent approaches to impact assessment.
This parameter is a source of variability in assessing
toxic water quality impact because the ACR varies both
between species and, for any one species, between
different toxicants. Therefore, to account for such
differences, the regulatory authority must apply some
ACR to an effluent when faced with a limited amount
of toxicity data.
Research on acute and chronic toxicity for effluents
shows that ACRs for whole effluents are more often
around 10 [1-8]. When dealing with effluent toxicity,
EPA recommends regulatory agencies use 10 as an
ACR, in the absence of specific information.
This value can be used both to extrapolate to chronic
concentrations from acute toxicity data and to set per-
mit limits limiting chronic toxicity where chronic tox-
icity is not directly measured.
Of course, where acute and chronic toxicity data are
available, the ACR can be directly calculated for that
specific effluent. With this value, more cost-effective
monitoring can be conducted using acute tests only.
However, if the effluent's constituents vary, so will the
ACR.
Behavior of Toxicants and Toxicity
after Discharge
An understanding of the behavior of both single toxi-
cants and whole-effluent toxicity after discharge may
be necessary in the application of water quality-based
toxics controls. Evaluating the combined effect of in-
teracting toxic discharges may also be important in
multiple discharge situations. When evaluating the in-
stream behavior of toxicants and toxicity, certain fac-
tors become important.
• Persistence — How long, to what extent in terms
of area, and at what level does effluent toxicity or
the toxicity of a single toxicant persist after dis-
charge? Is it reasonable to assume that the persist-
ence of both individual toxic chemicals and
effluent toxicity is conservative?
• Additivity, antagonism, and synergism — When
toxicants or effluents with toxic properties mix in
the receiving water, what is their combined toxic
effect?
These factors are discussed below.
Persistence
As soon as an effluent mixes with receiving water, its
properties begin to change. The rate of change of tox-
icity is a measure of the persistence. In most cases, the
level of toxicity instream will either remain relatively
constant (until further diluted by some additional
source), or drop as decay processes (photodecompo-
sition, microbial degradation) or compartmentalization
processes (sediment deposition, volatilization) occur.
On-site toxicity testing has indicated that the toxicants
causing toxicity measured at discharge sites tend to be
persistent. There may be little near field degradation of
measured effluent toxicity. Effluent toxicity does ex-
hibit far field decay. Typical patterns of progressive
downstream decreasing toxicity (similar to BOD decay)
have been observed in a number of discharge situations
[1-8].
Since mixing zones designated by State standards are
usually restrictive enough to render toxicity assess-
ments essentially near field situations, toxicity degra-
dation may not be important in these permitting
situations. Toxicity can be considered conservative
(non-degrading) in these cases. However, if a permit-
tee believes that his effluent's toxic effect is non-
persistent, several measurement techniques can be
employed to assess changes of toxicity after discharge.
• The toxicity test itself, when performed with dilu-
tion water from immediately upstream, is an ana-
logue of the mixing processes taking place in the
receiving water. Rapid chemical reactions that oc-
cur in the mixing zone can be expected to occur
in the dilution procedure. The effects of varying
physical/chemical conditions on persistence can-
not be predicted from these results, however. Test-
ing should be performed during critical low flow
periods.
• Ambient toxicity testing, as detailed in Appendix
C, measures the instream interactions of effluent
and receiving water and can be used to assess per-
sistence.
• Artificial "aging" of an effluent by holding the
sample on the shelf for a specific period of time is
another technique that can be used to estimate
persistence.
One disadvantage of the chemical-specific approach
is that the bioavailability of the toxicant after discharge
is not measured. However, persistence can be modeled
and the persistence of specific toxicants can be ac-
counted for in making impact predictions and setting
controls.
Additivity, Antagonism, and Synergism
ADDITIVITY
The issue of additivity can be stated as follows: if mul-
tiple effluents are discharged to a receiving water, is the
toxicity additive in some way? Since each effluent is
composed of individual toxic substances, the mixture
of effluents in a receiving water (assuming conserva-
tive behavior for the moment) produces a mixture of
-------
these individual pollutants. The concentrations of each
can be calculated from mass balance. Thus, the ques-
tion of additivity of toxic effluents is the same as the
question of additivity of the toxicity of mixtures of
chemicals. An extensive review of this question is avail-
able [21]. The Chapter 11 summary section is relevant:
Examination of available data using this [additivity]
model shows that for mixtures of toxicants found in
sewage and industrial effluents, the joint acutely le-
thal toxicity to fish and other aquatic organisms is
close to that predicted assuming simple addition of
the proportional contribution from each toxicant. The
observed median value for the joint effect of these
toxicants on fish is 0.95 of that predicted; the cor-
responding collective value for sewage effluents, riv-
er waters and a few industrial wastes, based on
toxicity of their contituents, is 0.85, while that for
pesticides is 1.3. The less than additive effect of
commonly occurring toxicants in some mixtures may
be partly attributable to small fractions of their
respective LC50s having little or no additional
effects.
Figure 1-4 illustrates Alabaster and Lloyd's data sum-
mary. Over 90% of the data, from the 5 to 95 percen-
tile normalized for "relative" acute toxicity, are within
a factor of two of confirming the additivity model. Thus,
for acute toxicity at least, mixtures of effluents will
probably exhibit additivity in terms of relative acute
toxicity.
Little information appears to be available on the addi-
tivity of chronic toxicity. For the growth of fish, Alabas-
ter and Lloyd conclude:
On the other hand, in the few studies on the growth
of fish, the joint effect of toxicants has been consis-
tently less than additive which suggests that as con-
centrations of toxicants are reduced towards the
levels of no effect, their potential for addition is also
reduced. There appear to be no marked and consis-
tent differences between the response of species to
mixtures of toxicants.
10
I 20
3
.D
!•=
S 50
E
(J
80
90
95
99
99.
N
\
\
(a)
O
\
\
Note:
(a) represents effluent having a high content of industrial
wastes including pesticides.
Legend:
• Constituents of sewage and industrial wastes
• • River waters
Sewage effluents
Gas liquor
A Drilling fluid
- — - Pesticides and other substances
s
r • River
I 0 Sewa
1 A Gasli
L A nrillin
• ' L.
0.4 124
Times as Toxic as Predicted
(from ITU)
Figure 1-4. Data summary on additivity from Alabaster and Lloyd [19].
6
-------
Additivity appears to supply a reasonable estimate of
the combined effects of individual acutely toxic agents.
Toxicities approaching the NOEL do not appear additive,
however. This suggests that additivity will produce an
upper bound of the actual chronic toxicity.
ANTAGONISM
Cases in which one effluent ameliorated the toxicity of
another effluent (antagonism) have been observed. If
practical, a testing procedure designed to measure
such interaction can be employed. A description of
such a procedure is found in Section 3, under the head-
ing Multiple Source Discharge Situations.
SYNERGISM
Under certain conditions, synergism, a greater-than-
additive increase in toxicity upon mixing, can theoret-
ically occur. However, field studies involving effluent
toxicity and laboratory experiments with specific
chemicals imply that synergism is an extremely rare
phenomenon (it has not been observed during on-site
effluent toxicity studies) and may not be an important
factor in the toxicological assessment of effluents.
It should be noted that ambient toxicity tests can give
some indication of the presence or absence of each of
these factors.
References
1. Mount, D., N. Thomas, M. Barbour, T. Norberg,
T. Roush, and R. Brandes. 1984. Effluent and
Ambient Toxicity Testing and Instream Commu-
nity Response on the Ottawa River, Lima, Ohio.
Permits Division, Washington, D.C., Office of Re-
search and Development, Duluth, MN,
EPA-600/2-84-080, August, 1984.
2. Mount, D. I., and T. J. Norberg-King, (editors).
1985. Validity of Effluent and Ambient Toxicity
Tests for Predicting Biological Impact, Scippo
Creek, Circleville, Ohio. U.S. Environmental Pro-
tection Agency, EPA/600/3-85/044, June,
1985.
3. Mount, D. I., et al. (editors). (In press). Validity of
Effluent and Ambient Toxicity Tests for Predict-
ing Biological Impact, Five Mile Creek, Birming-
ham, Alabama. U.S. Environmental Protection
Agency, EPA 600/in preparation.
4. Mount, D. I., et al. (editors). (In press). Validity of
Effluent and Ambient Toxicity Tests for Predict-
ing Biological Impact, Back River, Baltimore Har-
bor, Maryland. U.S. Environmental Protection
Agency, EPA 600/in preparation.
5. Mount, D. I., et al. (editors). (In press). Validity of
Effluent and Ambient Toxicity Tests for Predict-
ing Biological Impact, Naugatuck River, Water-
bury, Connecticut. U.S. Environmental
Protection Agency, EPA 600/in preparation.
6. Mount, D. I., et al. (editors). (In press). Validity
of Effluent and Ambient Toxicity Tests for
Predicting Biological Impact, Skeleton Creek,
Enid, Oklahoma. U.S. Environmental Protection
Agency, EPA 600/in preparation.
7. Mount, D. I., et al. (editors). (In press). Validity
of Effluent and Ambient Toxicity Tests for
Predicting Biological Impact, Ohio River, Wheel-
ing, West Virginia. U.S. Environmental Protection
Agency, EPA 600/in preparation.
8. Mount, D. I., et al. (editors). (In press). Validity
of Effluent and Ambient Toxicity Tests for
Predicting Biological Impact, Kanawha River,
Charleston, West Virginia. U.S. Environmental
Protection Agency, EPA 600/in preparation.
9. Bergman, H., R. Kimerle, and A. W. Maki (editors).
1985. Environmental Hazard Assessment of Ef-
fluents. Pergamon Press, Inc., Elmsford, NY.
10. Nebeker, A. 1982. Evaluation of a Daphnia magna
Renewal Life-cycle Test Method with Silver and
Endosulfan. Water Research. Vol. 16, pp
739-744.
11. Grothe, D., and R. Kimerle. 1985. Inter- and Intra-
laboratory Variability in Daphnia magna. Effluent
Toxicity Test Results. Env. Tox. and Chem. Vol
4(2), pp. 189-192.
12. Qureshi, A. D., K. W. Flood, S. R. Thompson, S. M.
Junhurst, C. S. Inniss, and D. A. Rokosh. 1982.
Comparison of a Luminescent Bacterial Test
with Other Bioassays for Determining Toxicity of
Pure Compounds and Complex Effluents, pp.
179-195. In J. G. Pearson et al. (eds.), Aquatic
Toxicology Hazard Assessment: Fifth Confer-
ence, ASTM STP 766. American Society for
Testing and Materials, Philadelphia, PA.
13. Strosher, M. T. 1984. A Comparison of Biological
Testing Methods in Association with Chemical
Analyses to Evaluate Toxicity of Waste Drilling
Fluids in Alberta. Volume 1. Canadian Petroleum
Association, Calgary, Alberta.
14. Schimmel, S. C. 1981. Results: Interlaboratory
Comparison — Acute Toxicity Tests Using Estu-
arine Animals. EPA-600/4-81-003.
15. Peltier, W., and C. I. Weber. 1985. Methods for
Measuring the Acute Toxicity of Effluents to
Aquatic Organisms. 3rd edition. Office of Re-
search and Development, Cincinnati, OH
EPA-600/4-85-013. April, 1985.
16. Horning, W., and C. I. Weber. 1985. Methods for
Measuring the Chronic Toxicity of Effluents to
Aquatic Organisms. Office of Research and De-
velopment, Cincinnati, OH, EPA-600/4-85-014.
September, 1985.
17. U.S. Environmental Protection Agency. 1982. Pes-
ticide Assessment Guidelines. Office of Pesti-
cide Programs, Washington, D.C., EPA/9-82-018
through 028.
-------
18. U.S. Environmental Protection Agency. 1982. Toxic
Substances Test Guidelines. Office of Toxic Sub-
stances, Washinton, D.C., EPA/6-82-001
through 003.
19. LeBlanc, G. A. 1984. Interspecies Relationships in
Acute Toxicity of Chemicals to Aquatic Organ-
isms. Env. Tox. and Chem. 3(1), pp. 47-60.
20. Kimerle, R. A., A. F. Werner, and W. J. Adams.
1984. Aquatic Hazard Evaluation Principles Ap-
plied to the Development of Water Quality Criter-
ia. In Aquatic Toxicology and Hazard
Assessment; Seventh Symposium, ASTM STP
854. R. D. Cardwell, R. Purdy, and R. C. Bahner
(eds.) American Society for Testing and Materi-
als, Philadelphia.
21. Alabaster, J. S., and R. Lloyd (editors). 1982.
Water Quality Criteria for Fish. 2nd edition. But-
terworths, London, 361 pp.
-------
2. Water Quality Standards
The starting point for a water quality-based control pro-
gram is the determination of the level of water quality
adopted for the waterbody. The following discussion
describes standards and provides guidance on narra-
tive and numerical criteria that, when met, will achieve
water quality standards. It should be noted that the wa-
ter quality criteria recommendations presented in the
TSD are based directly on the national water quality
criteria (including averaging periods). These recom-
mendations are subject to change if and when the
Agency determines any changes, improvements, or up-
dates are needed.
Water Quality Standards
Water quality standards are adopted by States (or,
where necessary, promulgated by EPA) to define desig-
nated uses and ambient characteristics of receiving
waters that must be maintained to allow those uses.
These standards must be met in order to discharge
wastewaters legally. Standards are composed of three
parts: 1) an antidegradation statement; 2) designated
uses for individual waterbodies; and 3) narrative and/or
numerical criteria. Some criteria apply State-wide,
whereas others are specific to particular designated
uses or waterbodies.
In recent years EPA has sought to reinforce the central
function of water quality standards and to emphasize
the process for establishing and reviewing standards.
The new water quality standards regulation (see 40
CFR 131; 48 FR 51400, November 8,1983) and the Wa-
ter Quality Standards Handbook [1] implement sub-
stantial changes in the standards program. To support
certain standards actions. States must assess the at-
tainability of uses (e.g., whether natural or irreversible
man-made factors preclude achieving a designated
use)[2]. States are also required to establish new criter-
ia to control toxic materials and detail in their standards
or planning documents how criteria will be implement-
ed. Flexibility has been allowed in establishing chemi-
cal criteria different from the national criteria in order
to account for local site-specific factors. Several steps
have also been taken to better involve the public in de-
cision making.
An important feature of the water quality standards
regulation is an antidegradation requirement. If a desig-
nated use is currently being attained, the waterbody
may not be classified for a less stringent use. For ex-
ample, if a waterbody is being used for fishing and pota-
ble water supply, the waterbody must be classified for
those or more stringent uses regardless of the current
classification or development pressures. Also, if water
quality is better than necessary to maintain aquatic and
recreation uses, that level of water quality must be
maintained unless the State meets the conditions dis-
cussed in the Water Quality Standards Regulation.
Mixing Zones
Many State water quality standards allow a zone of
mixing in which less stringent criteria apply than ap-
ply to the rest of the waterbody. The rationale is that
a small area of degradation can exist without causing
adverse effects to the overall water body. The criteria
that apply to mixing zones vary from State to State. In
some States there are explicit requirements for water
quality within mixing zones (such as no acute toxicity,
floating materials, or deposit-forming solids). In other
States there are no requirements or the requirements
are ambiguous. EPA's policy is described in the Hand-
book [1] and discussed in detail in Section 5.
The allowable size of mixing zones also varies by State.
Most States specify that the zone must not be as wide
as the stream in order to allow a zone of passage for
fish. Very few States specify the allowable length.
Usually, the size of the mixing zone is determined on
a case-by-case basis taking into account the critical re-
source areas that need to be protected.
It is important to note that mixing zones should be
evaluated and used for regulation in cases where mix-
ing is not complete within a short distance of the out-
fall. In the majority of cases involving conventional
pollutants, mixing has been assumed to be complete
since the impact of these pollutants occurs down-
stream. The regulatory authority has conducted the
evaluation of the discharge and calculation of permit
limits based on effluent dilution in the full stream flow.
Evaluation and control based on a mixing zone has
usually been limited to situations where mixing is
known to be poor (e.g., shore-hugging plumes and dis-
charges to large rivers, lakes, and estuaries). However,
if mixing is assumed to be rapid and complete when it
is not, a toxic discharge that appears to meet standards
may cause zones of chronic toxicity that extend for
miles. Therefore, regulatory agencies should carefully
evaluate mixing. Methods to evaluate dispersion and
set mixing zones are included in Section 5.
-------
Magnitude, Duration, and Frequency
Criteria are specifications of water quality designed to
ensure protection of the designated use. EPA criteria are
developed as national recommendations to assist
States in developing their standards and to assist in
interpreting narrative standards. EPA criteria consist of
three components:
• Magnitude — How much of a pollutant (or pollu-
tant parameter such as toxicity), expressed as a
concentration, is allowable.
• Duration — The period of time over which the in-
stream concentration is averaged for comparison
with criteria concentrations. This specification
limits the duration of concentrations above the
criteria.
• Frequency — How often criteria can be exceeded
without unacceptably affecting the community.
For each parameter, specifications for these compo-
nents are developed for two levels of effect. These lev-
els of effect are typically acute effects and chronic
effects (criteria for dissolved oxygen are further sub-
divided). Thus, a criterion will consist of at least six
specifications: recommended magnitude, duration, and
frequency values for both acute and chronic levels of
protection.
Magnitude for Single Chemicals
Water quality criteria for aquatic life contain two ex-
pressions of allowable magnitude: a criterion maximum
concentration (CMC) to protect against acute (short
term) effects; and a criterion continuous concentration
(CCC) to protect against chronic (long term) effects.
The two concentrations for the priority pollutants are
presented in EPA's criteria documents. These criteria
generally apply after mixing.
Most State standards include numerical criteria for only
a few individual toxic chemicals. Generally, evaluation
and control of toxic pollutants is based on maintenance
of the designated use and the narrative criterion pro-
hibiting toxic substances in toxic amounts.
The adverse effects of concern will depend on the
designated use and the chemical. Human health haz-
ards, bioaccumulation of chemicals in aquatic organ-
isms, toxicity to these organisms, and aesthetic factors
may be important. Available information on the toxic ef-
fects of the chemical is used when standards do not
include specific numerical criteria. Such information
can include EPA criteria documents, published literature
reports, or studies conducted by the discharger.
Water quality-based controls may be based directly on
the State's technical determination of what concentra-
tion of a specific pollutant meets the State's narrative
"free from" toxics criterion. Although the new water
quality standards regulation requires that the State's
process for implementing its narrative criterion be
described in the State standards, there is no require-
ment that this concentration be adopted as a numeri-
cal criterion in State water quality standards prior to use
in developing water quality-based controls.
The Water Quality Standards Regulations allow States
to develop numerical criteria or modify EPA's recom-
mended criteria to account for site-specific factors. In
cases where additional toxicological data are needed
to modify or develop criteria, the discharger may be re-
quired to generate the data. Guidance on modifying na-
tional criteria is found in the Handbook [1]. When a
criterion must be developed for a chemical for which
a national criterion has not been established, the
regulatory authority should refer to the Guidelines for
Deriving Criteria for Aquatic Life and Human Health
(see 45 FR 79341: November 28, 1980). Revisions to
the guideline for deriving aquatic life criteria have been
finalized (see 50 FR 30784, July 29, 1985).
Magnitude for Whole-effluent Toxicity
Criteria for toxicity in current State standards range
from the narrative prohibition (e.g., no discharge of toxic
chemicals in toxic amounts) to detailed requirements
that specify the test species and the allowable toxici-
ty level. State standards often include the application
factors (to derive a "safe" toxicity value from an acute
toxicity value) that were recommended in the 1972
"Blue Book" guidance (i.e., 0.05 for nonpersistent
wastes and 0.01 for persistent wastes)[3].
EPA's recommended magnitudes for whole effluent
toxicity are as follows. Again, two expressions of allow-
able magnitude are used: a CMC to protect against
acute (short term) effects; and a CCC to protect against
chronic (long term) effects.
For acute protection, the CMC should not exceed 0.3
acute toxic unit (TUa) as measured by the most sensi-
tive of at least three test species. As explained in the
previous section, the selection of test species is not
critical provided species from ecologically diverse taxa
are used (e.g., a fish, an invertebrate, and a plant). The
factor of 0.3 is used to adjust the typical LC50 endpoint
of an acute toxicity test (50% mortality) to an LC1 val-
ue (virtually no mortality). Specifically, a factor of 0.3
was found to include 91 % of observed LC50 to LC1 ra-
tios in 496 effluent toxicity tests as illustrated in Fig-
ure 2-1. This figure presents effluent toxicity data from
many years of toxicity testing of both industrial and mu-
nicipal effluents by the Environmental Services Divi-
sion, U.S. EPA Region IV, Athens, Georgia.
For chronic protection, the CCC should not exceed 1.0
chronic toxic unit (TUC) to the most sensitive of at
least three test species. The selection of test organisms
is as described above.
For both the acute and chronic magnitude, if fewer than
three species are tested, the test results are divided by
10
-------
10 to account for species sensitivity differences (see
Section 1 for discussion). Thus, if only two species are
used to measure the toxicity of the receiving water or
effluent discharges, the CCC would be 0.1 TUC.
The following example illustrates the use of these mag-
nitudes for toxicity. Assume that an effluent discharge
is the only source of toxicity to a water body and that
the available dilution is 30 to 1 (dilution factor = 30).
The basic relation between criteria concentrations, di-
lution, and allowable effluent quality can be stated as
follows:
allowable effluent concentration x 1/dilution factor
< criteria concentrations
Using this equation with the criteria, the effluent acute
toxicity should be less or equal to nine TUa (or
LC50>11% effluent) and the effluent chronic toxicity
should be less than 30 TUC (or NOEL > 3.3% ef-
fluent).
As with single chemicals, water quality-based controls
may be based directly on the State's technical deter-
mination of what level of toxicity meets the State nar-
rative toxics criterion. Again, the process should be
described in the State standards but this level of tox-
icity need not be adopted as a numerical criterion prior
to use.
Duration
The quality of an ambient water typically varies in re-
sponse to variations of effluent quality, stream flow,
and other factors. Organisms in the receiving water are
not experiencing constant, steady exposure but rath-
er are experiencing fluctuating exposure. Short periods
of exposure to high concentrations of toxicants can
have adverse effects. Thus, EPA's criteria restrict the
maximum time period over which exposure is to be
averaged, thereby limiting the magnitude and duration
of exposure.
For the CMC, the maximum period in which to average
the exposure is one hour. Thus, the one-hour average
exposure must be less than the CMC to protect against
acute effect. In practice, one-day periods are the
shortest periods for which wasteload allocation (WLA)
modelers and enforcement personnel have adequate
data. Compliance with the duration criterion can be en-
sured by paying particular attention to short term ef-
fluent variability and requiring measures to control
variability (e.g., installation of equalization basins) when
needed.
For the CCC, the maximum period in which to average
the exposure is four days. Some pollutants may have
magnitude criteria that correspond to different periods.
The toxicity criterion for chronic effects is specified as
a four-day average. Thus, the four-day average ex-
posure must be equal to or less than the CCC.
These specifications for duration apply to both
130 r-
120
110
100
90
80
& 70
c
ZJ
of 60
u.
50
40
30
20
10
-
-
-
-
-
-
_
_
4
12
125
67
40
29
89
86
42
2
0- .11- .21- .31- .41- .51- .61- .71- .81- .91-
.10 20 .30 .40 .50 .60 .70 80 .90 1.0
LC1/LC50 Ratio
Figure 2-1. LC1 to LC50 ratios for effluent toxicity tests.
chemical-specific and whole-effluent toxicity criteria.
Many people have erroneously assumed that, because
many chronic toxicity tests are 28 or 30 days in length,
the CCC was meant to be used as a 30-day average.
However, the duration of a toxicity test has nothing to
do with the critical period of exposure to concentra-
tions greater than the criteria. Many chronic toxicity
tests are of a one-year or longer duration, yet this does
not lead to the establishment of an averaging period of
one year's duration. Obviously, if a one-year averaging
period were used, the CCC could theoretically be ex-
ceeded for six months, a duration more than long
enough to cause an unacceptable chronic effect in a
waterbody.
The toxicity tests used to establish the national criter-
ia numbers are conducted using steady state exposure
techniques. They do not account for fluctuating ex-
posures such as those experienced by the instream bi-
ota due to effluent variability. What is pertinent is that
as the period of averaging increases, so too does the
period of time the exposure concentrations can be
above the criterion concentration without exceeding
the average. The question is, how long can the ex-
posure concentration be above the criterion concentra-
tion without significantly affecting the endpoint of the
test (e.g., survival, growth, or reproduction)? This is the
significant consideration involved in setting duration
11
-------
criteria. The period of time above the criteria concen-
tration must be limited. The four-day duration recom-
mended by EPA specifies the period of time over which
to average exposure. This duration time was derived to
limit the duration above the CCC.
Frequency
The application of criteria involves predictive modeling
of ambient conditions based on the assimilative capac-
ity of the receiving water and sources of pollution. One
essential element in this modeling is an assessment of
the projected frequency with which criteria will be ex-
ceeded. There are two reasons why this frequency is
a factor in water quality criteria. First, it is statistically
impossible to project that criteria will never be exceed-
ed (unless pollution sources are removed entirely). Sec-
ond, ecological communities are able to recover from
stresses. Since WLA modeling requires a specification
for frequency, it is important that ecologically sound
recommendations be provided.
The frequency with which criteria are allowed to be ex-
ceeded depends on site-specific factors. To implement
the criteria, site specific factors as described in Appen-
dix D must be taken into account. EPA's recommenda-
tions for f requencyare once in three years for both the
CMC and the CCC. These recommendations apply to
both chemical-specific and whole-effluent approaches.
As explained in Appendix D, field studies indicate that
many discharge situations are affected both by predict-
able and measurable discharges of toxicants and by un-
predictable spills of toxic substances. In most cases the
dischargers were unaware that spills were occurring.
These spills are a second source of stress for the com-
munity and decrease recovery potential. An aggressive
program to minimize, contain, and treat spills should be
in place at any plant where the potential for spills ex-
ists before using the three-year frequency criteria.
Detailed guidance on derivation of design flows for
steady state modeling will be provided by EPA. Until this
detailed guidance is issued, the 1Q10 and 7Q10 should
be used as an interim design flow for stressed systems
and the 1Q5 and 7Q5 should be used for unstressed
systems, where conditions discussed in Appendix D
are met and when dynamic modeling is not undertaken.
References
1. U.S. Environmental Protection Agency. 1984.
Water Quality Standards Handbook. Office of
Water Regulations and Standards (WH-585),
Washington, D.C.
2. U.S. Environmental Protection Agency. 1984. Tech-
nical Support Document for Conducting Use
Attainability Studies. Office of Water Regula-
tions and Standards (WH-585), Washington,
D.C.
3. U.S. Environmental Protection Agency. 1972.
Water Quality Criteria. EPA-R3-73-033.
12
-------
3. Effluent Characterization
The purpose of effluent characterization is to provide
the data needed to determine whether or not the dis-
charge of toxic materials causes adverse impact to the
biota of the receiving system.
This section of the Technical Support Document dis-
cusses specific measurement techniques and overall
analytical procedures to be followed to achieve this
objective. The specific procedures employed at any one
site should be tailored to that site to account for its
unique characteristics. Aspects of the procedures dis-
cussed below are based on conclusions reached at the
Fifth Pellston Environmental Workshop held in August,
1982, at Cody, Wyoming [1].
Two general measurement techniques can be employed
in effluent characterization; whole-effluent toxicity
analysis and chemical-specific analysis. Whole-effluent
toxicity testing is most appropriate in the following
situations:
• Where an effluent's constituents are not com-
pletely known or where a complex mixture of po-
tentially additive, antagonistic, or synergistic toxic
pollutants are discharged.
• Where more than one discharger is located in a
specific area and the potential exists for effluent
mixing and additive toxic effect.
• Where chemical-specific evaluation is impractical
due to a lack of information about the toxic effects
of a chemical or a lack of the resources required
to model the chemical(s) present.
Chemical-specific analysis is most appropriate in these
discharge situations:
• Where an effluent contains only one or several well
quantified pollutants for which toxicological data
are available or can be generated, or where one or
more exists as a large percentage of the waste
flow.
• Where bioaccumulative, carcinogenic, teratogenic,
or mutagenic chemicals are suspected in the
effluent and specific limits for each of these chem-
icals may be required.
• Where potentially toxic constituents of an effluent
may be masked, but cause delayed toxicity when
acted on by downstream processes.
EPA's recently issued policy (Appendix A) specifically
suggests the use of an integrated strategy for toxics
control. There will be discharge situations where both
toxicity analysis and chemical-specific analysis should
be used. An example of such a situation would be
where a known toxicant is discharged at very high
levels in a complex effluent known to cause a toxic
effect. A limit would be set for that toxicant specifically.
For example, suppose ammonia is discharged by a pub-
licly owned treatment works (POTW) at toxic levels.
Following the policy, a limit for ammonia would be set
independently to reduce these levels. An additional
whole-effluent toxicity limit would be set to control the
toxicity of the other toxicants in the effluent when it
is suspected continued toxicity will exist after the limit
on ammonia is implemented. An example of such an
approach is described in Section 8.
Toxic impact is characterized by measuring effluent tox-
icity or concentrations of individual chemicals and
comparing these measurements to the exposure con-
centrations of that effluent or individual chemical in the
receiving water. The potential for impact is minimal
where the exposure concentration is less than the toxic
effect concentration and maximal where the exposure
concentration approaches or exceeds the toxic effect
concentration. A simple equation can be used to illus-
trate this:
IWC < NOEL
This is stated as "the Instream Waste Concentration
(IWC) of an effluent or a toxicant must be less than or
equal to the No Observed Effect Level (NOEL)." A NOEL
is the highest concentration of a toxicant or an effluent
that, if maintained over time, is expected to cause no
biological impact. An IWC is the concentration of a tox-
icant or an effluent in the receiving water after mixing.
The most appropriate or allowable mixing is determined
by the regulatory authority. For single chemicals, the
NOEL is the chronic water quality criterion number. For
whole effluents, the NOEL is the measured or esti-
mated chronic toxicity concentration as determined by
the appropriate toxicity test on a sensitive test species.
This section provides recommendations for generating
the data to determine the NOEL. Section 5, Exposure
Assessment, provides recommendations for more
detailed assessment of the IWC and the calculation of
wasteload allocations for establishing permit limits.
Section 6 describes the actual calculation of permit
limits.
Before presenting EPA's effluent characterization test-
ing recommendations, an important principle should be
discussed. It is not necessary to generate any effluent
13
-------
toxicity data to set toxicity-basedpermit limits. Regula-
tory authorities do not need to require toxicity tests of
a permittee nor do they need to generate the data them-
selves to limit effluent toxicity. Just as a water quality-
based permit limit can be set for a single toxicant
(based on the available dilution and the water quality
criterion or State standard) in the absence of effluent
data, so too can a toxicity limit be set in the absence
of effluent toxicity data. Again, the available dilution
and the water quality criterion for toxicity, as described
in Section 2, can be used to establish a toxicity-based
permit limit. The basic relationship is expressed as:
allowable effluent toxicity x 1/dilution factor
< criterion concentration
where effluent toxicity and the criterion are expressed
in toxic units and the dilution factor is the factor by
which the effluent is diluted in the receiving water (e.g.,
one gallon effluent into nine gallons receiving water
yields a dilution factor of 10). Since both the criterion
for toxicity and the dilution factor are known, the value
for allowable effluent toxicity can be calculated as
follows:
allowable effluent toxicity 4.8%.
In this case, the State can set a permit limit of 21 TUC
without requiring effluent toxicity testing. This does
not imply, however, that toxicity testing is not neces-
sary to ensure compliance with the permit limit. Tox-
icity monitoring would be used to determine if the
discharger was meeting this limit.
For multiple sources, two additional considerations are
involved. First, toxicity additivity must be assumed.
Second, the flows must be summed and allowable tox-
icity allotted to the individual sources using some
option for establishing toxicity wasteload allocations
(WLAs) (see Section 5). The relationship for multiple
sources is expressed as:
(QeTe)
0-
criterion
where: Qe = effluent flow
Te = effluent toxicity
Qs = stream flow at the point of reference
(below the last source)
An example of the use of this equation is in Section 8,
Case Example.
Of course, there are advantages and disadvantages to
whichever route is taken (i.e., setting limits directly or
engaging in effluent characterization). For setting per-
mit limits directly without effluent characterization, the
advantages and disadvantages are as follows:
• Advantages
— The cost of data generation prior to permit issu-
ance is eliminated.
— The permit limit is quickly derived and permit
issuance is not delayed.
• Disadvantages
— Uncertainty as to whether or not a real toxicity
problem exists because no data are available.
— Exposure must be simplified so that only a
steady state exposure condition can be used.
This can be overprotective or underprotective
depending on the design flow utilized.
— The permittee may object to the limit since no
data are available to show toxicity is an actual
problem.
Where the regulatory authority decides to proceed
directly to effluent characterization, the advantages
and disadvantages are as follows:
• Advantages
— Will indicate whether or not a real toxics prob-
lem exists.
— Can generate the data needed to assess
exposure in a more sophisticated manner. A
sensitive test species is identified and variabil-
ity is assessed.
— The permittee is less likely to object to subse-
quent limits because a sound basis for the re-
quirement is established.
• Disadvantages
— The cost of testing can be high.
— The testing and analysis could delay permit
insurance.
Given these advantages and disadvantages, the regula-
tory authority must decide which course to take on a
site-specific basis.
Tiered Testing
The most cost-effective way to generate the data
necessary to assess impact is to follow a "tiered" test-
ing procedure. The concept of tiered testing is similar
to that used under the Toxic Substances Control Act
(TSCA) when testing a new chemical product for
potential hazard. Testing begins with short, inexpensive
screening tests and progresses to more and more
definitive data generation techniques.
The tiered approach integrates effluent effect meas-
urements and simplified exposure assumptions into
various stages of analysis and arrives at one of three
decisions: 1) apparently safe because of a large mar-
gin between exposure (IWC) and effects (NOEL),
14
-------
2) unacceptable hazard because the exposure concen-
tration exceeds the NOEL, or 3) additional data are
needed to resolve more clearly the marginal difference
between exposure and effects. This exposure/effect
level comparison enables a regulatory authority to
determine where no further analysis is needed (appar-
ently safe), where limit derivation should begin (unac-
ceptable hazard), or whether there is a need for more
complete effluent characterization (additional data
needed).
Uncertainty Associated with Data Bases
All effects testing and exposure assessment
parameters, for both effluent toxicity and individual
chemicals, have some degree of uncertainty associated
with them. The more limited the test data available, the
larger the uncertainty factor becomes.
The least amount of uncertainty exists for an effluent's
impact on the receiving water when a complete data
base is available on acute and chronic toxicity on many
indigenous species, when there is a clear understand-
ing of ecosystem species composition and functional
processes, and when actual measured exposure con-
centrations are available for all chemicals during sea-
sonal changes and dilution capacities. The uncertainty
associated with this ideal situation would be minimal.
No manipulation of the data using adjustment factors
or safety factors would be needed. Obviously, resource
constraints generally prohibit the development of such
ideal data bases at NPDES discharge sites.
The current method for calculating a national water
quality critierion for a specific toxicant utilizes labora-
tory acute and chronic testing on a variety of species
to arrive at the criterion maximum concentration (CMC)
and the criterion continuous concentration (CCC).
These concentrations, if not exceeded, are expected to
protect aquatic life (see 50 FR 30784, July 29, 1985)
and thus have uncertainty factors of one. Uncertainty
is minimized when chemical-specific limits are applied
because a large amount of data has been generated on
each toxicant.
For the purpose of evaluating which effluents need
additional testing, each tier of data should be judged
in light of the uncertainty of the data through the use
of uncertainty factors. Table 3-1 lists recommended
uncertainty factors. Uncertainty factors are factors that
can be used to generate a specific level of uncertainty
against which toxic effect concentrations/ambient
exposure ratios are compared. For instance, a limited
data base of a few acute toxicity tests on one species
could require a level of uncertainty of 1,000 (derived
from three uncertainty factors): 10X for species sen-
sitivity, 10Xfor variability in effluent toxicity, and 10X
for the acute-chronic ratio (ACR). (The ACR is a con-
version factor, not an uncertainty factor. For simplic-
ity, it is used in this context as an uncertainty factor.)
To illustrate the use of uncertainty factors, suppose a
Table 3-1. Uncertainty Factors
Uncertainty
Source
Effluent
Variability
Uncertainty
Factor
10
Rationale
Range in effluent toxicity
not observed to exceed one
order of magnitude
Data
Source
I2a, 3]
100 Heavy metals observed to
vary two orders of magnitude
in POTW effluents
[4]
Species
Sensitivity
Acute/ Chronic
Ratio
10 Acute or chronic tests
using one vertebrate
and one invertebrate
10 Value observed from
on-site whole effluent
toxicity testing
[2bJ
15, 6, 7, 8,
9, 10]
discharger exhibits potential toxicity in a screening test.
Further testing is required. The discharger conducts an
acute toxicity test on a fish and measures an LC50 of
40%. The discharger's IWC is calculated by dividing
the mean effluent flow, 3.0 cfs, by the 7Q10 of the river,
which is 10,000 cfs. A level of uncertainty of 1,000 is
calculated using the three applicable uncertainty fac-
tors; [10 (species sensitivity) x 10 (effluent variability)
x 10 (ACR)]. To decide whether or not further testing
is warranted, the regulatory agency compares toxicity
and exposure to the level of uncertainty:
LC50 (in percent)
IWC*
LC50 (40%)
3 cfs
> level of uncertainty
= 1,334 > 1,000
10,000 cfs + 3 cfs
x 100
In this case, the ratio is greater than the calculated level
of uncertainty (1,000). This discharger is given a low
priority for further assessment. However, if the LC50
was 4% for this discharger, the ratio would not be
greater than the level of uncertainty and further analy-
sis would be warranted:
LC50 (4%)
3 cfs
— = 133 < 1,000
10,000 cfs + 3 cfs
x 100
EPA recommends the concept of levels of uncertainty
be used in judging the adequacy of a site-specific data
for evaluating impact and calculating effluent require-
ments. These factors are perhaps more appropriate for
the whole-effluent approach to toxic impact assess-
ment. The species sensitivity and ACR uncertainty fac-
tors have been eliminated in the single chemical impact
assessment through the criteria development process.
Effluent variability can be considered the major uncer-
tainty factor for chemical-specific evaluations.
*The IWC can be calculated in two ways depending on the source
of the facility's water supply. Use IWC = Qw/Qr where the source
is the receiving water. Use IWC = Qw/Qw + Qr where the source
is not the receiving water (Qw = wastestream flow; Qr = receiv-
ing water flow.
15
-------
Recommendations for conducting tiered toxicity tests
are presented in Box 3-1.
Box 3-1
TESTING RECOMMENDATIONS
EPA recommends a two-tiered approach be used
in toxic impact assessment. The first tier,
SCREENING, is conducted to separate individual
dischargers into those with potential water qual-
ity standards violations requiring further analysis
and those where such violations are improbable.
The second tier, DEFINITIVE DATA GENERATION,
is conducted to generate sufficient data to quan-
tify violations of water quality standards so that
appropriate regulatory action can be taken to
eliminate that violation. This tier of testing can be
subdivided into progressively more detailed analy-
sis. Figure 3-1 shows a schematic representation
of the toxicity data generation process. Some
methods for toxicity testing are listed in Table 3-2.
Initial dilution
screen
If dilution =
10,000: 1 or more,
no testing
If pass criteria, ^_
no testing
Toxicity
screening
Definitive data
generation
|f NOEL
IWC
level of uncertainty,
no further tests needed
If NOEL <
IWC
level of uncertainty,
more tests needed
Exposure
Permit limit
derivation
Figure 3-1. Data generation: tiered testing.
Table 3-2. Methods Manuals for Toxicity Testing
1. General Listing of Available Toxicity Tests
U.S. Environmental Protection Agency. 1982. Environmental
Effects Test Guidelines. Office of Toxic Substances,
Washington, D.C. EPA 560/6-82-002.
U.S. Environmental Protection Agency. 1982. Pesticide Assess-
ment Guidelines. Office of Pesticide Programs. Washington,
DC. EPA 540/9-82-018 through 028.
2. Acute Toxicity Tests
Peltier, W. and C. I. Weber, 1985. Methods for Measuring the
Acute Toxicity of Effluents to Aquatic Organisms, Third
Edition. Office of Research and Development, Cincinnati,
OH EPA-600/4-85-013.
3. Chronic and Subchronic Toxicity Tests
Mount, D.I., and T.J. Norberg. 1984. A seven-day life cycle
cladoceran toxicity test. Env. Tox. Chem. 3(3) p. 425-434.
Horning, W., and C.I. Weber, ed. 1984. Methods for measuring
the chronic toxicity of effluents to aquatic organisms.
Environmental Monitoring and Support Laboratory,
Cincinnati, OH (in preparation). EPA-600/4-85-014.
Norberg, T.J., and D.I. Mount. 1985. A new subchronic
fathead minnow (Pimephales promelas) Toxicity Test. U.S.
Environmental Protection Agency, Environmental Research
Laboratory, Duluth, MN. (in Press).
Birge, W., and J. Black. 1981. In-situ acute/chronic
toxicological monitoring of industrial effluents for the NPDES
Biomonrtoring Program using fish and amphibian embryo-
larval stages as test organisms. U.S. Environmental
Protection Agency, Office of Water Enforcement and Permits,
Washington, D.C. OWEP-82-001.
4. Other Toxicity Tests
Bulich, A.A. 1982. Microtox System Operating Manual.
Beckman Publication No. 015-555879. Sectarian Instruments,
Inc. Carlsbad, CA.
Tier 1. Screening
The goal of screening is to identify, with the most cost-
effective method available, potential water quality
impact situations and to avoid collecting unnecessary
data in cases in which impact will probably be minimal.
After conducting a screening analysis, cases should fall
into two categories: those cases in which it is improb-
able that water quality is being adversely impacted, and
those cases in which water quality is more likely to be
impacted. In this first tier, false positives are more
desirable than false negatives. Therefore, the assump-
tions and decision triggers used in evaluating the data
should be conservative (i.e., stringent).
Determining which NPDES dischargers are candidates
for water quality- based toxics control is of obvious con-
cern. It is difficult to give recommendations as to which
dischargers out of the active 65,000 NPDES permittees
should conduct toxicity testing. This is a regulatory
decision which must be made on a site-specific basis
by State and regional regulatory authorities. Some fac-
tors pertinent in this decision are listed below.
• Dilution — Toxic impact is directly related to avail-
able dilution. The lower the available dilution, the
higher the potential for toxic effect. One recom-
16
-------
mendation is if concentrations of instream efflu-
ents reach 1% or higher during critical low flow
periods, those effluents should be given high pri-
ority for toxicity assessment. Assessment of the
amount of dilution available on a national basis at
annual low flow and at 7Q10 (Figure 3-2) shows
that a majority of NPDES dischargers discharge to
areas of low available dilution during certain critical
periods.
• Type of industry — Although this factor should be
carefully assessed, some generalizations can be
made. The primary industrial categories should be
of principal concern. Secondary industrial cate-
gories have less potential for toxicity than primary
industries. Within the primary industrial categories
no generalizations should be made. Toxicity prob-
lems associated with types of industries become
site specific.
• The type and volume of industrial input (for
8,000-
£
a>
fe 6,000-
o
5
o 4,000-
2,000-
Low Flow (7Q10)
15,863 13,159 Majors)
Majors
Minors
2,084
2,467
2,771
1,628
1,241
470
ISWAJJAVMi
8,000-
2
o>
J6 6,000-
Q
"° 4,000-
I
^2,000-
1-10 10-20 20-50 50-100 100- 1,000- >10,000
1,000 10,000
Dilution (stream/effluent)
Annual Mean Flow
7,908
5,006
3,450
1,771
2,477
3,849
2,065
1-10 10-20 20-50 50-100 100- 1,000- >10,000
1,000 10,000
Dilution (stream/effluent)
Figure 3-2. National distribution of NPDES dilution conditions at
7Q10 and at annual mean flow.
POTWs) — POTWs with large loadings from
indirect dischargers (particularly primary indus-
tries) are high priority candidates for toxicity
testing. However, absence of industrial input is no
guarantee of absence of toxicity problems.
• Existing data on toxic pollutants — Discharge
monitoring reports (DMRs) and application form
2C data provide some indication of the presence
of toxicants. Again, careful assessment is required.
The presence or absence of the 126 priority pol-
lutants is not an absolute indication of the pres-
ence or absence of toxicity. There are thousands
of toxicants that may be causing effluent toxicity,
not just 126. Also, low levels of several toxicants
combined can cause toxicity where individually
they would not.
• History of toxic impact and compliance problems—
Regulatory authorities are usually well aware of
particular problem areas. Such discharge sites
would be priority candidates for toxicity analysis.
Given these factors, priorities for toxicity analysis can
be set. Combinations of factors, such as low available
dilution, poor compliance record, and clustered
industrial and municipal discharges are obvious high
priorities.
Where cases of water quality standards violations are
already documented, it may be appropriate to skip the
screening tier and go to the next tier. This is perhaps
more applicable to situations where the chemical-
specific approach is to be used because regulatory
agencies usually have more chemical-specific data
available. Effluent toxicity data are mostly unavailable
for NPDES dischargers. For more complex or more tox-
icologically unknown discharge situations, data gener-
ation for whole-effluent toxicity screening will probably
be needed to characterize the site fully.
The decision triggers associated with the tiers are obvi-
ously directly related to how exposure is calculated.
The different options for exposure assessment are
detailed in Section 5 of this document. In general, the
method(s) used for estimating exposure and effluent
effect should be comparable in terms of cost and
required effort. Therefore, in the screening tier, as deci-
sions are made on simple, inexpensive toxicity tests,
simple exposure estimates such as assuming complete
dilution at design low flow (e.g., 7Q10) should be
employed. However, this assumption should be used
cautiously since discharges often create shore-hugging
plumes that mix slowly. As subsequent tiers of analy-
sis are used, more sophisticated exposure analysis
procedures, such as those described in Section 5,
should be used.
A recently developed source of whole-effluent toxicity
data can be useful in toxicity screening. The Complex
Effluent Toxicity Information System (CETIS) is a data
management system designed to catalogue the results
of whole-effluent toxicity testing of municipal and
17
-------
industrial wastewaters. CETIS is a data file within EPA's
In-house Software (IMS) system. IMS was developed to
augment the data management and analytical capabil-
ities of STORET. A user's manual is available [11a, 11b].
Data are obtained from EPA Regional and State NPDES
permitting programs and from published literature.
CETIS can be used to assess a discharge in a specific
industrial category for toxicity potential by comparing
it to a similar discharge for which toxicity data are
already available.
Whole-effluent Toxicity Screening
Several procedures can be used to screen for effluent
toxicity. One option is to run acute toxicity tests, cal-
culate the lowest LC50, divide it by available dilution,
and compare it to the appropriate level of uncertainty.
The ratio of LC50/IWC should be greater than this level
of uncertainty. If it is not, the second tier of testing is
required.
Another option for screening is to measure effluent tox-
icity using simple, fast screening tests conducted in the
Box 3-2
RECOMMENDATIONS FOR WHOLE-EFFLUENT TOXICITY SCREENING
• Individual Dischargers — Compare receiving water flow rate (in terms of whatever water quality-based
design low flow is specified by the State) to average effluent flow rate.
— If dilution exceeds 10,000 to 1, and there is a reasonably rapid mix of the effluent outside of the
rapid initial dilution area in the receiving water, then the effluent should be given a low priority for
any further attention.
— If dilution is less than 10,000 to 1, or mixing is not rapid and toxicity within a plume is of concern,
then toxicity screening tests should be performed.
— If dilution is between 1,000 to 1 and 10,000 to 1, or a poorly mixed effluent plume in a large receiv-
ing water (> 10,000 to 1 dilution) is of concern, conduct acute toxicity screens as follows:
1. Collect four to six effluent samples on one day (grab or short term composite), quarterly. Con-
duct screening tests (24-hour) in 100% effluent, using a daphnid and a fish, on each sample.
2. If 50% mortality or greater is observed in three samples, the potential for toxicity is assumed
and further testing is required.
3. If 50% mortality or greater is observed for two or fewer samples, the discharge should be given
a low priority for further analysis.
— If dilution is less than 1,000 to 1, conduct chronic toxicity screens (short term chronic tests are
recommended) as follows:
1. Collect four to six effluent samples (24-hour composite) on four to six successive days. Con-
duct static screening tests (seven-day) in 100% effluent, using a cladoceran and a fish, on each
sample.
2. If a 50% or greater effect is observed between controls and test organisms, the potential for
toxicity is assumed and further testing is required.
3. If less than 50% effect is observed, the discharge should be given a low priority for further
analysis.
Acute tests can be used in these dilution situations, but it should be noted that there will be cases
where no acute toxicity is measured but the effluent is chronically toxic.
—Where dilution is less than 100 to 1, the use of a toxicity-testing-based screening procedure is not
recommended. Screening has already been accomplished through dilution analysis. Even in
discharge situations where no toxicity is observed in screening tests, the narrow margin between
effect concentration and available dilution suggests more complete effluent toxicity characteriza-
tion is mandatory. If uncertainty factors are applied in a 100 to 1 discharge situation, dilution alone
would mandate further testing. Where very limited dilution is available, it is recommended that
toxicity-testing screening be skipped and the discharger be required to begin DEFINITIVE DATA
GENERATION procedures. An example of this situation is described in Section 8.
• Ambient Toxicity Analysis — Use ambient toxicity analysis to identify areas of instream toxicity
associated with specific dischargers. This analysis may be most useful when conducted by the regula-
tory agency, but dischargers may be required to conduct the tests in conjunction with effluent tests.
A systematic plan for identifying problem areas is recommended. This procedure is useful for multiple
source discharge situations. The analysis should be conducted concurrently with discharge-specific
screening and must be done at low flow conditions. A procedure is described in Appendix C.
18
-------
absence of serial dilutions at 100% effluent concentra-
tion. If a pre-specified toxicity level is observed, the
potential for toxic impact exists and further analysis is
conducted. Uncertainty factors are not used in this
option because LC50s are not generated. Also, dilution
factors (described in Box 3-2) serve the same function
as the uncertainty factors used with the first option
described.
One recently developed procedure, ambient toxicity
testing, is useful in screening stream reaches and dis-
crete watershed areas for existing instream toxic con-
ditions. The procedure, described in Appendix C, uses
short term chronic toxicity tests to measure the toxic-
ity of samples of receiving water taken above, at, and
below (or around) outfalls. The utility of this approach
in screening is that actual instream toxicity is meas-
ured, signifying the probability of toxic impact. No
extrapolation for exposure or ACR is needed. Further,
impact from multiple-source discharge situations,
which may not be apparent from individual outfall
screening data, is identified. Finally, the technique can
provide an assessment of the persistence of effluent
toxicity. The procedure must be conducted during an
appropriate, design low-flow period.
Chemical-specific Screening
There are two levels of analysis for chemical data. Both
are based on the margin between measured chemical
concentration and the State water quality standards or
the national water quality criteria. The first is simple
dilution analysis and comparison with State standards
or national criteria. Where high background concentra-
tions exist, the concentration in the receiving water
prior to discharge should be included in the calculation.
The second level of analysis is to use fate models to
estimate persistence. The EPA has recently supported
the development of a chemical-specific screening
approach that is applied to lakes, rivers, estuaries, and
coastal systems [12]. It contains methods for esti-
mating waste loadings from urban and rural point and
nonpoint sources and for predicting the levels of con-
taminants in waterbodies. Methods are included for
conventional pollutants (nutrients, dissolved oxygen,
biochemical oxygen demand, coliforms, temperature,
sediment transport, and salinity), metals (currently
being expanded), and the organic priority pollutants.
Considerable supporting information is provided in the
manual, particularly on rate constant information for
toxicants. The tools are intended to be used with a desk
top calculator or microcomputer.
EPA's recommendations for chemical-specific screen-
ing are presented in Box 3-3.
Tier 2. Definitive Data Generation
Whole-effluent Definitive Data Generation
Once screening has indicated the potential for toxic
impact, further data may be needed to determine the
Box 3-3
RECOMMENDATIONS FOR
CHEMICAL-SPECIFIC SCREENING
• Examine existing data to determine presence of
specific toxicants for which criteria, standards
or other toxicity data are available. Sources of
data include:
— Permit application forms, DMRs, and permit
files.
— Pretreatment industrial surveys.
— STORET for U.S. Geological Survey (USGS)
flow data and ambient monitoring data.
— Industrial effluent guidelines development
documents.
—The Treatability Manual [13].
• Use simple dilution assumptions to compare
effluent data and corresponding estimated
instream concentrations to the concentrations
specified in the national water quality criteria,
State water quality standards, or the literature
for the toxicant(s) of concern.
• Use simple fate models and screening proce-
dures [12] for situations where better estimates
of instream concentrations are needed.
• Follow the procedures for screening for human
health problems described in Section 4.
• In making screening decisions from limited
data, an uncertainty factor of 10 or 100 can be
used to account for a toxicant's potential con-
centration variability. In this way, effluent varia-
bility is addressed in the chemical-specific
screening process.
degree of impact caused by the discharge. In this tier,
whole-effluent toxicity testing is one of two procedures
that may be conducted to provide the data necessary
to measure the worst case ambient toxicity conditions
or to estimate those conditions through a limited data
extrapolation process. Identifying and quantifying the
worst case conditions require an analysis of toxic effect
and exposure assessment (detailed in Section 5) and
an analysis of the variability associated with these con-
ditions at the site.
EPA's recommendations for generating whole-effluent
definitive data are presented in Box 3-4.
Chemical-specific Definitive Data Generation
The recommended procedures for monitoring heavy
metals and organic chemicals are described in this part
of the Effluent Characterization Section. Analyses for
all toxicants are described in Standard Methods of
Water and Wastewater Analyses (ASTM, 15th edition,
1982). Detailed recommendations on the design of
appropriate sampling programs for rivers, lakes, and
estuaries are included in the Monitoring and Data Sup-
port Division's Technical Guidance Manual for Perform-
19
-------
Box 3-4
RECOMMENDATIONS FOR WHOLE-EFFLUENT DEFINITIVE DATA GENERATION
Requirements for toxicity analysis are site specific. Recommendations for testing in this section will
be presented in terms of eliminating levels of uncertainty. Analytical procedures will be presented together
with those uncertainty factors which can be eliminated if those procedures are used. Again, a simple
relationship can be applied to determine whether to require more data, whether to stop testing and begin
the process of establishing permit conditions, or whether to cease analysis because of a wide margin
between toxicity and the IWC: LC50 or NOEL (in %)
TT-— > level of uncertainty
where the level of uncertainty equals the uncertainty factors multiplied together. It should be stressed
that this equation is used to evaluate the need for additional analysis. The relationship is not the basis
for the development of permit limits or conditions. These procedures are discussed in Section 6.
• Initial or "Baseline" Jesting
— Tests — Conduct acute toxicity tests on two species. A fish and an invertebrate are recommended.
—Frequency — Conduct monthly on grab or composite samples, whichever is appropriate. Where
court orders, administrative orders, or legal deadlines require permits to be issued within a shorter
timeframe, schedule these tests to conform with the deadline.
— Uncertainty factors — 10X to 100X for effluent variability, 10X for species sensitivity variability,
and 10X for ACR where chronic toxicity is of concern, but no chronic data are available. Level of
uncertainty = 1,000 to 10,000.
• Eliminate Effluent Variability Factor
— Tests — Conduct acute toxicity tests on two species. A fish and an invertebrate are recommended.
—Frequency — Examine manufacturing processes, treatment plant design and retention times (actual,
not design), and variability of measured parameters to estimate the variability of toxicity. It may
be impossible in a practical sense to assess toxicity variability, but an estimate should be made ini-
tially. This factor is perhaps the most important source of uncertainty in the toxicity assessment
process.
For effluents exhibiting short term (12 to 48 hours) variability, measure the highest toxicity during
a short period of duration (one discharge cycle or a 24-hour period). Four to six samples (grab or
short term composite) taken during this period should be tested. Tests are conducted monthly for
at least one year. Another option is to require a continuously pumped, flow-through acute test on
each species. This will integrate the effects of a variable toxic concentration but will not provide
a quantification of effluent variability (see Appendix B).
For effluents exhibiting suspected long term variability, schedule testing frequency to conform to
expected changes (weekly, monthly, seasonal, process changes) in effluent composition, if known
(see Appendix B). For long term variability, a year-long monitoring program may be required to deter-
mine long term variability.
— Uncertainty Factors — 10X for species sensitivity and 10Xfor ACR where chronic toxicity is of con-
cern but no chronic data are available. Level of uncertainty = 100.
• Eliminate Species Sensitivity Factor
— Tests — Conduct acute toxicity tests on a total of three to five species. The species should be
representative of several different groups including fish, invertebrates, and plants.
—Frequency — Same as above.
— Uncertainty Factor — 10X where chronic toxicity is of concern but no chronic data are available.
Level of uncertainty = 10.
• Eliminate ACR Factor
— Tests — Conduct short term chronic toxicity tests on three species. Since the available test proce-
dures short enough to be practical in effluent characterization are limited, to eliminate the species
sensitivity factor where chronic testing is mandatory, the use of Ceriodaphnia, fathead minnow
growth test, or other short term chronic tests currently available is recommended.
—Frequency — Same as above.
— Uncertainty Factor — Level of uncertainty equals one if these data are generated.
• Other Considerations
— Use of Dye Studies — Dye studies are strongly recommended for effluent characterization. They
are relatively inexpensive and provide data needed to plan an assessment of an effluent's impact
on the receiving water. A dye study should be included in any Tier 2 analysis unless mixing is known
to be rapid and complete. Procedures for conducting a dye study in an effluent characterization are
described in the Lima, Ohio, report [5].
20
-------
ing Wasteload Allocations [13-15]. In general, plug-flow
sampling is recommended for rivers, and vertical pro-
file monitoring is recommended for lakes. A combina-
tion of slack water and intensive surveys is recom
mended for lakes, rivers, reservoirs, and estuaries.
It is important for each toxicant, whether it is a heavy
metal or an organic chemical, that the analytical equip-
ment achieve the best resolution and lowest detection
limit possible. As a minimum, the detection limit should
be at least as low as the concentration of the water
quality standard or criterion. EPA generally recom-
mends high-resolution, capillary-column gas chromato-
graphs (GO for organic chemical analysis and graphite-
furnace atomic adsorption (AA) for metals.
If heavy metals are being monitored for fate and trans-
port modeling, bed sediment should be analyzed for the
following parameters: solids concentration, particle
size distribution, total toxicant concentration, dissolved
toxicant concentration, and percent water in sediment.
If organic chemicals are being investigated, an addi-
tional analyisis must be conducted that measures the
weight fraction of organic carbon in particles less than
or equal to 50 micrometers (/mi) (fine) and in particles
greater than 50 /un (coarse). This organic carbon meas-
urement should be determined using a wet oxidation
method if high carbonate soils are being analyzed, and
a gravimetric analysis if low carbonate soils are pres-
ent. These analyses are described in soil chemistry
textbooks.
If heavy metals are being monitored for fate and trans-
port modeling, the water column should be analyzed
for the following parameters: total toxicant concentra-
tion, dissolved toxicant concentration, total suspended
solids (TSS), total organic carbon (TOC) and dissolved
organic carbon (DOC), pH, temperature, hardness and
total alkalinity, dissolved oxygen (DO), ammonia, chlo-
rine residual, and specific conductivity. If organic chem-
icals are being monitored and modeled, additional
analyses should be conducted: weight fraction of
organic content in suspended particles less than or
equal to 50 ^m and in suspended particles greater than
50 nm and secchi disk depth. The TOC and DOC data
indicate humic acid content of waters and the poten-
tial for metal complexing. The pH, temperature, hard-
ness, and total alkalinity data are needed to calculate
the appropriate water quality criteria for metals and
organic chemicals. The DO, ammonia, and chlorine
residual measurements can be used to interpret
observed toxicity and biotic status. Conductivity meas-
urements confirm model representation of transport,
and secchi disk depths indicate the potential for pho-
tolysis of organics.
Monitoring of toxic substances requires special qual-
ity control procedures beyond those necessary for
monitoring conventional parameters because toxicants
generally occur in trace concentrations and are fre-
quently unstable. The most up-to-date sampling and
handling procedures recommended by EPA for a num-
ber of toxic and conventional parameters are detailed
in a recent publication [17].
Samples that will be filtered for particulate and dis-
solved fractions should be delivered to the laboratory
for filtration within one or two hours maximum for
metals and filtered and preserved on site. For organic
chemicals that may photolyze and biodegrade readily,
samples should be exposed to the atmosphere as lit-
tle as possible to avoid the loss of volatile compounds.
Sample bottles must be clean and made of materials
that will not contaminate the samples, either plastic or
glass, depending on the analyses to be performed.
Replicate samples should be taken and analyzed to
assess the variability of measurements caused by sam-
pling techniques and site heterogeneity. A blank and
one sample spiked with each of the toxicants should
be analyzed with each group of samples.
The documentation of all field and laboratory proce-
dures is needed to ensure the defensibility of toxics
monitoring data. A sample observation sheet should be
filled out for each station providing observations on sur-
face conditions. For each sampling location, a sample
log book should be kept with a record of the in situ
results and the numbers and times of water column
measurements. All samples should be properly labeled
and numbered with preprinted forms and labels. Addi-
tionally, a log book should be kept in the laboratory in
order to record each sample as it arrives in the lab and
to document the analytical results.
Once these data have been generated (the responsibil-
ity of the permittee), the discharge is analyzed for water
quality problems using the procedures discussed in
Section 5.
EPA's recommendations for generating chemical-
specific definitive data are presented in Box 3-5.
Use of Toxicity Testing in Multiple-source
Discharge Situations
No formal definition exists to determine what a
multiple-source discharge situation is. One definition
could be that it exists where impact zones overlap.
Where more than one discharge is contributing to toxic
impact, additional data may be needed. Testing that
provides the regulatory authority with the information
necessary to separate the relative impact of each
source can be conducted. In multiple-source discharge
situations, additivity, antagonism, and persistence of
toxicity may become more important. Therefore, more
detailed testing designed to measure these factors
would be warranted.
Two options for assessment can be identified in these
cases. The regulatory agency may decide to handle
each source separately. The procedures described
21
-------
Box 3-5
RECOMMENDATIONS FOR CHEMICAL-SPECIFIC
DEFINITIVE DATA GENERATION
The choice of chemical-specific data generation
procedures is based on the type of modeling anal-
ysis found to be most appropriate for the specific
discharge situation. Steady state analysis is more
stringent since it assesses the worst case
exposure situation (where a proper, toxicology-
based, critical low flow is used) and requires com-
parable treatment. Dynamic or probabilistic
modeling will generally result in more cost-
effective treatment requirements because these
procedures more accurately assess actual
exposure conditions. A complete discussion of
the choice of modeling techniques is found in
Section 5.
• Limits to be Based on Steady State WLAs —
Conduct critical condition sampling as appro-
priate for rivers, lakes, and estuaries. GC anal-
ysis (for organics) and AA analysis (for metals)
for the specific toxicant(s) of concern should be
conducted at some designated critical period.
• Limits to be Based on Dynamic or Probabilistic
WLAs — Conduct comprehensive sampling of
effluent. Multiple grab or composite effluent
samples should be taken over a critical season
or the entire year to define effluent variability.
GC analysis (for organics) and AA analysis (for
metals) for the specific toxicant(s) of concern
should be conducted using the recommended
frequency.
previously would be used. Another option is to treat
each discharge as an interactive component of the
whole system. If this option is selected the regulatory
agency could follow the testing procedures presented
in Box 3-6. Again, testing is the resposibility of the per-
mittees. It is assumed that screening has been con-
ducted, and the potential for toxic impact has already
been shown.
Use of Effluent Characterization Data and
Decision Criteria
Where levels of uncertainty for the last testing tier com-
pleted are not exceeded, the discharger should be given
a low priority for regulatory actions designed to con-
trol toxicity. Where the level of uncertainty is exceeded
by the toxic level measured, the data generated should
be used as the basis for toxics control. The process of
exposure assessment and permit issuance is described
in Sections 5 and 6.
Bioaccumulation
Bioaccumulation is the accumulation of chemicals in
the body of an animal to concentrations greater than
the ambient concentration. Bioaccumulation occurs
both through uptake across the gill membranes (bio-
concentration) and through ingestion of contaminated
food (biomagnification). Some chemicals can accumu-
late to hazardous levels in fish when the concentration
in the water is below a toxic level. Therefore, effluents
that have an acceptable level of toxicity may contain
hazardous concentrations of bioaccumulative pol-
lutants.
The potential impacts of bioaccumulative pollutants
include: 1) accumulation of toxic pollutants in aquatic
biota resulting in effects on survival or reproduction,
2) economic impacts resulting from fish tainting, and
3) exposure of humans to hazardous chemicals
through drinking water or consumption of contami-
nated fish or shellfish. Some chemicals bioaccumulate
but have no apparent adverse effects. However, any
chemical that has high potential for persistence and
bioaccumulation should be a matter of concern until it
can be demonstrated that there are no adverse environ-
mental and human health effects resulting from dis-
charge.
Chemicals that bioaccumulate include metals, organic
compounds, and organometallic compounds. Metals
and organometallic compounds are adequately
addressed through the EPA ambient criteria for human
health and aquatic life. Similarly, application of the EPA
ambient criteria for organic compounds will adequately
control those compounds because the criteria are
based on bioaccumulation where appropriate. The
problem in addressing effluent discharges involves
organic compounds that are not among the priority pol-
lutants for which there are criteria.
The first step in addressing bioaccumulative pollutants
in effluents is to determine whether or not such pollu-
tants are present. The second step determines if such
pollutants are hazardous. The final step calculates an
acceptable discharge rate. Since bioaccumulation is
perhaps the most important human exposure route,
bioaccumulation assessment is discussed in Section
4, Human Health.
22
-------
Box 3-6
RECOMMENDED MULTIPLE-SOURCE TOXICITY TESTING PROCEDURES
Tests
— For effluent-dominated (a suggested definition is where the effluent(s) make up 1% or greater of
the flow) receiving waters (assuming total flow input from all sources including other effluent flows
is measured), conduct chronic toxicity tests following the testing procedures previously described.
— For stream-dominated (where the effluent(s) make up less than 1% of the flow) receiving waters
(assuming total flow input from all sources including other effluent flows is measured), conduct
acute toxicity tests following the testing procedures previously described.
—An additional data requirement in multiple-source discharge situations is the assessment of rela-
tive and absolute toxicity of each source so that appropriate permit conditions can be set for
individual dischargers. The following procedure is suggested.
1. Conduct one set of toxicity tests on the effluents using a control dilution water, either from an
upstream, uncontaminated receiving water station of known high quality with a similar chemi-
cal makeup or using reconstituted water. This gives an absolute toxicity measurement of the
effluent.
2. Run a parallel set of toxicity tests on the effluent using a dilution water taken directly upstream
from the point of discharge. This dilution water may be contaminated with upstream effluents
or other toxicant sources. The purpose of this test is to establish toxicity measurements of the
effluent as it is mixed at its point of discharge. This is a relative effluent toxicity measurement.
The relative testing procedure could result in a change in the standard concentration-effect curve
generated by the testing.
The dilution water for the relative toxicity test may cause significant mortality (or growth or
reproductive effect) at the lower effluent concentrations (including the 100% diluent control con-
centration) if the diluent from the receiving water is toxic. Of course, it will be toxic if an upstream
source is toxic. This mortality does not invalidate the test. Instead, analysis of toxicity trends
resulting from the relative toxicity tests can be used to assess the effluent's toxicity in relation
to other sources and ambient receiving water conditions. However, there must always be a control
with no toxicity in the dilution water for quality assurance and determination of absolute toxic-
ity of the effluent.
3. Conduct ambient toxicity tests to: 1) determine whether or not the effluent has a measurable
toxicity after mixing, 2) measure instream persistence of toxicity from all sources contributing
to instream toxicity, and 3) determine combined instream toxicity resulting from the mixing of
multiple, point and non-point sources of toxicity. See Appendix C for a discussion of ambient
toxicity testing procedures.
The ambient testing can be conducted by the regulatory agency or can be required of each dis-
charger who would then be responsible for the upstream and downstream areas around their
outfall. It should be conducted during a low flow period.
Frequency
—Conduct tests at appropriate frequencies as described in the previous section. All testing should
be conducted simultaneously by each discharger, if possible. At a minimum the tests should be con-
ducted concurrently starting within a short time period (one to two days). Repeated ambient tox-
icity analyses will be desirable when variable effluents are involved. Effluent toxicity data showing
variability can be used to assess what frequency will be most applicable. The level of repetition for
variability analysis should be similar to that used in effluent variability analysis.
Other Considerations
— Dye studies of effluent dispersion for rivers, lakes, reservoirs, and estuaries are strongly recom-
mended. This allows analysis of effluent concentration at the selected sampling stations above and
below the discharge points.
—The procedures suggested in this multiple-source section are based on actual multiple-source site
investigations conducted under the Complex Effluent Toxicity Testing Program. Site reports from
that study can be used to obtain further description of the toxicity testing procedures used to analyze
multiple-source toxic impact [5,6].
23
-------
References
1. Bergman, H., R. Kimerle, and A. W. Maki (editors).
1985. Environmental Hazard Assessment of
Effluents. Pergamon Press, Inc., Elmsford, New
York.
2a. Kimerle, R., W. Adams, and D. Grothe. 1985.
Tiered Assessment of Effluents. In Environmen-
tal Hazard Assessment of Effluents. H. Berg-
man, R. Kimerle, and A. Maki (eds.).
2b. Kimerle, R., A. Werner, and W. Adams. 1983.
Aquatic Hazard Evaluation, Principles Applied to
the Development of Water Quality Criteria. In
Aquatic Toxicology and Hazard Assessment (7th
Symposium). R. Cardwell and R. Purdy, (eds.) (in
press).
3. Macek, K. 1985. Perspectives on the Application
of the Hazard Evaluation Process. In Envi-
ronmental Hazard Assessment of Effluents.
H. Bergman, R. Kimerle, and A. Maki (eds.).
4. DiToro, D. 1985. Exposure Assessment for Com-
plex Effluents: Principles and Possibilities. In
Environmental Hazard Assessment of Effluents.
H. Bergman, R. Kimerle, and A. Maki (eds.).
5. Mount, D., N. Thomas, M. Barbour, T. Norberg,
T. Roush, and W. Brandes. 1984. Effluent and
Ambient Toxicity Testing and Instream Commu-
nity Response on the Ottawa River, Lima, Ohio.
Permits Division, Washington, D. C., Office of
Research and Development, Duluth, MN,
EPA-600/2-84-080, August, 1984.
6. Mount, D. I., and T. J. Norberg-King (editors).
1985. Validity of Effluent and Ambient Toxicity
Tests for Predicting Biological Impact, Scippo
Creek, Circleville, Ohio. U.S. Environmental Pro-
tection Agency, EPA/600/3-85/044, June,
1985.
7. Mount, D. I., et al. (editors). (In press). Validity of
Effluent and Ambient Toxicity Tests for predict-
ing Biological Impact, Five Mile Creek, Birming-
ham, Alabama. U.S. Environmental Protection
Agency, EPA 600/in preparation.
8. Mount, D. I., et al. (editors). (In press). Validity of
Effluent and Ambient Toxicity Tests for Predict-
ing Biological Impact, Back River, Baltimore Har-
bor, Maryland. U.S. Environmental Protection
Agency. EPA 600/in preparation.
9. Mount. D. I., et al. (editors). (In press). Validity of
Effluent and Ambient Toxicity Tests for Predict-
ing Biological Impact, Naugatuck River, Water-
bury, Connecticut. U.S. Environmental Protec
tion Agency. EPA 600/in preparation.
10. Mount, D. I., et al. (editors). (In press). Validity of
Effluent and Ambient Toxicity Tests for Predict-
ing Biological Impact, Skeleton Creek, Enid,
Oklahoma. U.S. Environmental Protection
Agency. EPA 600/in preparation.
11a. Crane, J. L, A. Pilli, and R. C. Russo. 1984. CETIS:
Complex Effluents Toxicity Information System.
Data Encoding Guidelines and Procedures.
Office of Research and Development, Duluth,
MN, EPA-600/8-84-029. November, 1984.
11b. Crane, J. L, A. Pilli, and R. C. Russo. 1984. CETIS:
Complex Effluent Toxicity Information System.
CETIS Retrieval System User's Manual. Office
of Research and Development, Duluth, MN,
EPA-600/8-84-030, November, 1984.
12. Mills, W., et al. 1982. Water Quality Assessment:
A Screening Procedure Toxic and Conventional
Pollutants. Parts 1 and 2. Office of Research and
Development, Athens, GA EPA 600/6-82-004
A,B. September, 1982.
13. U.S. Environmental Protection Agency. 1983. The
Treatability Manual. Vol. I-V. U.S. EPA Office of
Research and Development. EPA 600/2-82-
001e (revised January 24, 1983). GPO Stock No.
055-000-00237-1.
14. U.S. Environmental Protection Agency. 1983. Tech-
nical Guidance Manual for Performing
Wasteload Allocations, Book II. Streams and
Rivers. U.S. EPA Office of Water Regulations and
Standards, Washington, D.C.
15. U.S. Environmental Protection Agency. 1984. Tech-
nical Guidance Manual for Performing
Wasteload Allocations, Book III. Estuaries. U.S.
EPA Office of Water Regulations and Standards,
Washington, D.C.
16. U.S. Environmental Protection Agency. 1984. Tech-
nical Guidance Manual for Performing
Wasteload Allocations, Book IV. Lakes, Reser-
voirs, and Impoundments. U.S. EPA Office of
Water Regulations and Standards, Washington,
D.C.
17. U.S. Environmental Protection Agency. 1982. Test
Methods — Technical Additions to Methods for
Chemical Analysis of Water and Wastes. Office
of Research and Development, Cincinnati, OH
EPA 600/4-82-055, December, 1982.
24
-------
4. Human Health Hazard Assessment
Overview
The assessment of human health hazard, as with the
assessment of aquatic impact, depends on two factors:
1) the toxicological properties of the pollutants dis-
charged and 2) the degree of exposure of the target
populations to the pollutants. However, health hazards
are considerably more difficult to assess than aquatic
impacts. The reasons are that latent effects can occur
many years after exposure, toxicological data are
always indirect because of the inability to test humans,
and routes of exposure are difficult to project, particu-
larly since there is almost always a treatment system
between released effluent and human exposure. A vari-
ety of factors other than those mentioned above can
also influence toxicity.
Assessing effluent-related human exposure is a rela-
tively new area for regulatory agencies. Most of the
effort has been concentrated on drinking water, the
industrial workplace, and other routes of exposure such
as food. Most of these problems have been addressed
using pollutant-specific chemical analysis, toxicologi-
cal information, and exposure analysis to assess and
regulate exposure. This approach tends to break down
for effluents because of the diversity of chemicals pres-
ent and the lack of analytical methods, toxicological
information, and means to assess interactions. The use
of biological tests to test effluents for health effects is
an attractive alternative; however, the degree to which
such tests reflect potential human health hazards is not
known. Although biological testing using short term
assays cannot yet be used to quantify health hazards,
they can be used for screening to identify potential
hazards.
This section will provide background information on
health effects, a review of chemical vs. biological
approaches, recommendations for screening and pri-
ority setting, and recommendations for assessment
and control.
Background
Various toxicological effects in the human population
are thought to be caused by environmental exposure
to chemical substances. Toxic effects will be classified
into two broad categories: 1) genotoxic effects such as
carcinogenicity and heritable mutation and 2) systemic
or target organ effects that impair the functioning of
specific organ systems.
Genotoxicity is the ability of a substance to damage an
organism's genetic material (DNA). Damage to the
DNA of reproductive cells can impair reproductive abil-
ity or can produce a change in the DNA structure that
could be passed on to offspring as a heritable mutation.
A chemical must interact with germ cells and perma-
nently change the linear structure of the DNA and pass
that change to offspring before it can be classified as
a human mutagen.
Damage to the DNA of non-reproductive cells can
result in uncontrolled cell proliferation leading to can-
cer. Many classes of carcinogens are electrophilic or
metabolized in vivo to form electrophilic reactants that
may ultimately interact with cellular DNA. This is
thought to be the initiating and critical step for geno-
toxic carcinogens. Other subsequent changes are still
necessary before a malignancy appears. The fact that
many carcinogens are also mutagens has led to the idea
that mutation tests may be predictive of carcinogenic
properties. This generalization on the action of carcino-
gens, however, has not been demonstrated to apply to
all chemical carcinogens.
Systemic toxicity is the ability of a substance to affect
the function of one or more organ systems. Although
a comprehensive discussion of these effects is beyond
the scope of this manual, the following target organ
toxicities have been observed from oral exposure to
chemical substances (the target organ is indicated in
parentheses): hepatotoxicity (liver), renal toxicity (kid-
ney), cardiovascular toxicity (heart and blood vessels),
neurotoxicity (brain and peripheral nerves), reproduc-
tive toxicity (testes or ovaries) and hematopoietic
effects (bone marrow). A subchronic or chronic whole
animal testing protocol is typically employed to detect
target organ toxicities among the major organ systems.
Toxicological results from mammalian studies have
been generally accepted for purposes of directly
estimating potential hazards to man.
Risk assessment must include an estimate of exposure.
Ingestion of drinking water is the most obvious route
of exposure to waterborne pollutants since a broad seg-
ment of the population could be affected. There are
other less obvious but potentially significant routes of
exposure. Fish and seafood contaminated through the
bioaccumulation of pollutants by aquatic organisms
may be a significant source of exposure. Absorption
through the skin that may occur during bathing or
swimming, and inhalation exposure that may occur
25
-------
during breathing of humidified air in the home are also
pathways for toxic substances.
Bioaccumulation
Bioaccumulation is a potentially significant route of
exposure to human populations. Pollutants that are
health hazards may also cause severe impacts on
aquatic organisms and other wildlife through bio-
accumulation.
Of the two processes that constitute bioaccumulation
(bioconcentration and biomagnification) bioconcentra-
tion or uptake through the gills is the easier process to
study. Therefore, most evaluation techniques are based
on bioconcentration with the inference that, by limit-
ing bioconcentration, biomagnification is also con-
trolled. The measure of a chemical's tendency to
bioconcentrate is termed the bioconcentration factor
or BCF and is defined as:
BCF =
accumulated concentration in tissue
exposure concentration in water
Bioconcentration has been extensively studied and the
following conclusions have been drawn: 1) the BCF for
a compound is constant over a wide range of exposure
concentrations, 2) compounds with low BCFs reach
steady state residue concentrations quickly, whereas
compounds with high BCFs may never reach steady
state, and 3) the BCF is correlated very well with the
n-octanol/water partition coefficient. This correlation
allows the n-octanol/water coefficient to be used to
estimate the BCF of a compound (see Figure 4-1). The
partition coefficient also provides a way to separate
potentially bioaccumulative pollutants from other pol-
lutants.
Status of Chemical vs. Biological
Approaches
Most health-related toxicological information has come
from the testing of biological organisms. Almost all of
this work, however, has been done on single chemicals
and not on mixtures. This is because of the experimen-
tal simplicity and the need to assess risks of single
chemicals under the Toxic Substances Control Act
(TSCA), Federal Insecticide, Fungicide, and Rodenticide
Act (FIFRA), and Food and Drug Administration (FDA)
statutes. Results from toxicological experiments on a
single chemical are combined into a dose response
curve for that chemical describing its toxicological
potency. This is then combined with an exposure model
to estimate risk.
A body of toxicological knowledge has been built to the
point where an untested chemical's toxicological prop-
erties can be estimated by comparing its structure to
the structure of tested chemicals. This field of science
is called structure activity relationship (SAR) and is
rapidly developing. There are computer programs that
can estimate toxicological properties. However, the use
of SAR requires considerable expertise and is not
recommended as a casual exercise. It has also come
under some criticism and is not fully accepted, espe-
cially as a regulatory basis for toxics control.
Effluents present a difficult problem, in contrast to sin-
gle chemicals, because they may be composed of a
mixture of chemicals. A second problem is that the con-
stituents of the effluent may interact to change the
effects of a chemical compared to its effects in "pure"
form. These problems severely limit the approach of
identifying the individual chemicals in the effluent and
conducting a risk assessment for each chemical
independently. A logical alternative is to test an effluent
directly using biological tests. This approach also has
severe limitations. First, very little research has been
conducted on complex mixtures (of which effluents are
a good example) to determine how various tests
respond. Second, the higher animal testing that is
needed to quantify the dose-response relationship (the
"potency" of the effluent) would be extremely costly.
Third, effluent testing presents several practical prob-
lems such as the need to handle the continual change
in composition typical for most effluents, the need to
concentrate samples to get a dose-response curve, and
the need to deal with interferences from cytotoxic com-
ponents of the effluent.
Some of the biological tests used to assess the toxicol-
ogy of pure chemicals can be used practically to test
effluent samples. These are the short term tests used
in the first tiers of the single chemical evaluation proc-
ess. In that process, the results of these tests are used
to decide whether more definitive (and much more
expensive) testing is needed. Therefore, they constitute
screening tests. These short term tests are based on
cellular level response and do not measure the actual
response that higher organisms, such as mice or
humans, would exhibit to the toxicant. However, they
do indicate whether the test substances are biologically
active and provide some measure of that activity. Each
test system measures a different type of response so
that a battery of several tests is usually considered nec-
essary. Research is currently underway within EPA to
evaluate the use of various biological tests for effluent
testing.
The chemical-specific approach should be used to
quantify risks accurately and to determine treatment
requirements in the assessment of effluent health haz-
ards. It is important to note that the priority pollutant
list is not sufficient for chemical-specific analysis
because it is so limited in scope. There are many more
hazardous toxicants discharged than are listed. Iden-
tifying pollutants requires techniques such as process
evaluation and extensive analysis of GC/MS data. It can
be extremely expensive, however, to identify important
chemicals or classes of chemicals and determine their
toxicological properties. Therefore, a priority setting
process is needed. For setting priorities, biological tests
26
-------
U-
O
m
O) 1
Log BCF = 0.85 log P - 0.70 • Fathead minnow
R = 0 947 n Rainbow trout
N = 59 •A Bluegill
0 12345678
Log P (n-octanol/water)
Figure 4-1. Relationship between the bioconcentration factor of
54 organic chemicals in fish to the log P of the chemical.
along with a chemical review of expected effluent com-
ponents can provide the most cost-effective approach.
Priority Setting
The evaluation of a discharge for potential hazard con-
sists of two components — estimating exposure and
estimating the toxic effects of the effluent. The follow-
ing example of a priority setting approach is intended
to provide a starting point for more detailed State-
specific priority ranking systems.
This example approach consists of evaluating each
component (exposure and effects) and assigning a
value of high, medium, or low depending on the mag-
nitude of potential risk. This provides a two-way matrix
for ranking sources. The matrix could then be used to
rank sources in order to assign priority for further inves-
tigation. Alternatively, decision criteria could be de-
veloped in advance so that, as soon as a source is
screened, an appropriate action could be taken.
Exposure
Three major routes should be assessed when evaluat-
ing potential human exposure to a toxicant in water.
These are ingestion by drinking water, ingestion of con-
taminated aquatic organisms, and intake through con-
tact recreation. Exposure due to drinking water requires
knowledge of the concentration of pollutant and an
estimate of the quantity of water drunk. This latter
value is generally assumed to be two liters per adult or
one liter for a child. If the concentration is to be based
on dilution calculations rather than direct measure-
ment, it should be done using a low flow estimate such
as the 30-day average low flow with a five-year return
period.
Consumption of aquatic organisms can be estimated
using the national average consumption rate of
6.5 grams (g)/day. If the local population is known to
consume more or less fish than this, the exposure
assessment would be more accurate if consumption
was based on local fish sales or harvest.
Finally, the location and extent of use of recreation
areas where exposure might take place should be
documented. Types of intake through this route might
include dermal absorption, accidental swallowing, and
cuts. Generally, only crude estimates can be made of
this route.
Effects
Two types of information should be collected in order
to evaluate the effects that pollutants contained in the
discharge may have. First, the discharger should pro-
vide an estimate of all chemicals that may be present
in the wastewater. This can be accomplished by analyz-
ing the manufacturing processes and listing all chemi-
cals used, produced, or created as byproducts. Where
the effluent is complex, an identification of classes of
compounds thought to be present would suffice for this
screening analysis. Some attempt to estimate the con-
centration or volume of each chemical should be made.
This information can then be reviewed against the tox-
icology data bases to identify which chemicals are
potentially hazardous. When only a class of com-
pounds is identified as present, the toxicology informa-
tion available for any members of the class could be
used to evaluate that component of the effluent.
The discharger should also, in addition to conducting
a chemical review, be required to conduct short term
biological tests. Biological test results are an important
supplement to a chemical review because the vast
majority of chemicals lack toxicology information and
because of the inability to identify many of the chemi-
cals in complex wastewaters. A variety of toxicity tests
exist and are accepted for assessing potential human
health hazards [1,2], but most of these have not been
used for effluent toxicity impact assessment because
they are either impractical (too expensive or time con-
suming), or not sufficiently well developed to provide
reliable or meaningful data. Recently, a variety of short-
term tests that provide relatively rapid results at afford-
able costs have become sufficiently well developed to
be considered for use in screening health hazards
associated with complex effluents [3,4].
A battery of tests is needed in order to test for the major
types of potential impact. Many of the genotoxic
effects can be tested for using bacterial or cell culture
techniques. Systemic or target organ effects, however,
can only be tested using whole animal studies. The best
means to assess systemic effects is by using sub-
chronic toxicity procedures designed to determine the
effects that may occur with repeated exposure over a
part of the average life span of an experimental animal.
However, such studies are expensive ($100,000 and
27
-------
over) and probably beyond the cost constraints for a
screening level analysis. These methods are discussed
later in the section on risk assessment. There are no
inexpensive short term tests that can assess the poten-
tial for systemic toxicity effects.
Several short term assays have been developed to
assess genotoxic effects. The two major types of
effects are point mutations and clastogenic activity
(chromosomal breakage). To assess point mutation
effects, the Ames test is recommended. The Ames test
can be followed by a mammalian gene mutation test
using cells in culture if confirmation is desired. To
assess clastogenic effects either the mammalian sis-
ter chromatid exchange test or a mammalian cell chro-
mosomal aberrations test should be conducted. Both
of these tests typically use Chinese hamster ovary cell
cultures and involve cytologic examination after
exposure to determine if chromosomal effects are evi-
dent. The Organization of Economic Cooperation
Development (OECD) test methodology is recom-
mended [5]. EPA's Office of Toxic Substances and
Office of Pesticides Programs also have published test
methods (see References [17] and [18], Section 1) that
are consistent with the OECD tests.
Some effluent samples may not need to be concen-
trated to measure a response, but the majority of sam-
ples should be concentrated. When samples are
concentrated, the response should be calculated in
terms of the preconcentrated sample (e.g., Ames
results would be reported as revertants per milliliter (ml)
of original sample). Caution should be taken when col-
lecting waste effluent, concentrating complex mix-
tures, and extracting the chemicals for toxicological
testing. Ion exchange and liquid/liquid extraction tech-
niques are most commonly used and are recom-
mended. If an effluent is extremely variable, screening
assays should be repeated periodically to ensure that
potentialy hazardous discharges are detected.
Test results should be evaluated to decide whether a
high, medium, or low toxic effect potential exists. Con-
siderable scientific judgment is required. For example,
Ames results are expressed as the number of revertants
per ml original sample. A response twice the solvent
blank (control) is considered a positive. A scale could
be constructed where less than twice the control is
negative, up to 100 revertants/ml is a "low," between
100 and 1000 revertants/ml is a "medium," and greater
than 1000 revertants/ml is a "high" response.
Evaluation
The State or Region should review the results of a
screening program and establish actions for each level
of potential risk indicated. For example, a discharger
with either a high exposure risk or a high effects risk
might automatically be required to conduct a detailed
assessment or institute controls. A medium risk in both
exposure and effects might require further review of the
data and a case-specific decision about whether to
require additional assessment. A medium and a low risk
might indicate the need for limited testing to ensure
that the low is really indicative of the risk. Low risk in
both exposure and effects should receive low priority
for further assessment.
Assessment and Control
When screening has indicated a high potential for
health hazard, further assessment is required. A
chemical-specific approach is recommended to evalu-
ate the discharge constituents. The first half of this
process involves characterizing the composition of the
wastewater. It should be recognized from the outset
that it is usually impossible to identify all the com-
pounds contained in a wastewater. Typically, only a
small fraction of the total organic carbon (TOC) can be
accounted for as specific chemicals. Therefore, a sub-
stantial effort should be put into identifying constit-
uents through means other than chemical analysis.
The best way to accomplish this is through a detailed
process evaluation.
A process evaluation is a study in which components
in the wastewater are determined from an analysis of
feedstocks, manufacturing processes, products, by-
products, and pollution control in place. The result is a
list of compounds or classes of compounds with a high
probability of being present in the wastewater. Chem-
ical analysis should also be conducted for not only the
priority pollutants but also nonpriority pollutant peaks.
Particular attention should be paid to bioaccumulative
pollutants. There are three approaches to identifying
non-priority bioaccumulative pollutants in effluents.
Aquatic biota can be collected upstream and down-
stream of an effluent outfall and analyzed for specific
compounds or any compound present in downstream
biota that is not present in upstream biota. This
approach is expensive. There is also the problem of
locating sources in multiple-source situations. How-
ever, this approach is the only way to assess bio-
accumulation.
Another approach is to rely on the list of compounds
used or manufactured that may be present in the efflu-
ent. For each compound, the n-octanol/water partition
coefficient can be looked up or estimated from chemi-
cal SARs. EPA recommends that any compound for
which the logarithm of the partition coefficient (log P)
is greater than 3.5 be flagged for further evaluation and
possible control [6]. Computing the log P of industrial
chemicals from structure has been the objective of a
six-year project at Pomona College. A computer pro-
gram that accurately calculates log P is available to all
Regions and States. Contact Dr. Oilman Veith in EPA's
Duluth Laboratory, 6201 Congdon Blvd., Duluth, MN
55804, (218) 727-6692 or FTS 783-9550.
28
-------
Third, if the effluent is too complex to list all com-
pounds, the effluent can be analyzed using high pres-
sure liquid chromatography (HPLC). The partition
coefficient of a compound is directly related to its reten-
tion time using HPLC separation systems [7]. Since
water is the carrier solvent, effluent samples may be
directly injected into the HPLC system, and those com-
pounds corresponding to log P greater than 3.5 can be
detected or collected and identified [8].
To determine if bioaccumulative compounds are pres-
ent, the HPLC can be connected to an infrared detec-
tor to measure peaks. To identify compounds, the
peaks can be collected and subsequently analyzed
using GC/MS methods.
Once pollutants or classes of chemicals present in the
wastewater have been identified, the next step is to
evaluate their toxicological properties. Sources of infor-
mation that may be useful are listed below.
• Water quality criteria — Developed by the Office
of Research and Development (ORD) and available
through EPA Office of Water Regulations and Stan-
dards, (202) 245-3030. Criteria documents are
available for 65 compounds or classes of com-
pounds which review toxicology and health effects
and, if sufficient data exist, present ambient
criteria to protect human health.
• The Carcinogen Assessment Group (CAG) List —
The CAG of EPA ORD has published a draft list of
chemicals that are known to be potent carcino-
gens. A revised list with potency factor and haz-
ard ranking is currently under development.
Contact CAG (RD-689), Washington, D.C.,
(202) 382-7341.
• Acceptable daily intakes (ADIs) for non-genotoxic
pollutants — The Environmental Criteria and
Assessment Office of EPA ORD is developing two
publications. One is a methodology and ranking of
44 chemicals based on chronic toxicity. The other
is a summary of ADIs for oral exposure. They are
currently in draft. Contact the Office listed above
at Mail Stop MD-52, Research Triangle Park, North
Carolina 27711, (919) 541-4173.
• Health Advisories — These are issued by both the
EPA Office of Drinking Water and the National
Academy of Science. The EPA Office of Drinking
Water also establishes National Interim Primary
and Secondary Drinking Water Standards (see 40
CFR 141). Contact Dr. Khris Khana, Office of Drink-
ing Water, Washington, D.C., (202) 382-7588.
• Chemical Information System (CIS) — This exten-
sive data system is a collection of several data
bases. The user is able to search all data bases
using the structure, substructure, and full or partial
name of over 250,000 substances. Formerly
supported by ERA, it is now operated by Chemi-
cal Information Services, Inc., Dr. Al Fein, 7215
York Road, Baltimore, Maryland 21212,
(800) 247-8737 (Hotline).
• National Library of Medicine (NLM) — The NLM
maintains two data bases of toxicology informa-
tion: Toxicology Information Online (TOXLINE) and
Registry of Toxic Effects of Chemical Substances
(RTECS). These data bases are available through
the CIS system or can be used separately by con-
tacting Medlars Customer Service, 5600 Rockville
Pike, Bethesda, Maryland 20209, (301) 496-6193
or 638-8480.
• Dialog — This data system is designed primarily to
locate and retrieve bibliographic information and
abstracts of articles. Contact Dialog Customer
Service, 3460 Hillview Avenue, Palo Alto, Califor-
nia 94304.
• QSAR — This is a computer program that will pre-
dict parameters useful for exposure assessment
such as BCF, biodegradation and hydrolysis rates,
etc. and will screen chemical structures for fea-
tures that may contribute to the induction of can-
cer. Contact Lee Faulkner, Director, Center for Data
Systems and Analysis, Montana State University,
Bozeman, Montana 59717, (406) 994-4481.
• Michigan's Critical Materials Register — Chemi-
cals are numerically scored as to their hazard, and
those posing high hazard are included in the reg-
ister. Contact MDNR Office of Toxic Material Con-
trol, PO Box 30028, Lansing, Michigan 48909,
(517) 374-9640.
• Office of Pesticides Programs estimates of ADIs
for various pesticides.
• WHO/IPCS Environmental Health Criteria
Documents.
• WHO/IPCS Monographs on Pesticides.
• FDA ADIs.
Toxicology information should be used to narrow the
field of pollutants to be investigated. Chemicals and
classes of chemicals on this shortened list are then
evaluated in terms of the potential exposure to target
populations. Exposure assessment is the "determina-
tion or estimation ... of the magnitude, frequency, dur-
ation, and route of exposure" (Proposed Guidelines for
Exposure Assessment; 49 FR 46304, November 23,
1984). Two principal routes of exposure should be
investigated for each chemical or class of chemicals —
drinking water and bioaccumulation. Each of these
routes is discussed below.
The means for evaluating the drinking water route of
exposure depends on how the toxicological information
is expressed. If there is an acceptable drinking water
concentration for the chemical, the object is then to
estimate the concentration at drinking water intakes.
Two factors should be considered: dilution and degra-
dation. Concentration after degradation can be esti-
mated based on half-life and time of travel. The analysis
of dilution is not critical when the variability associated
with the toxicological information and the wastewater
concentration estimate is considered. Using the mean
annual river flow should be appropriate for most pollu-
29
-------
tants. If a pollutant is a fast-acting toxicant, the 30-day
average low flow might be more appropriate.
If the available toxicological information is expressed
in terms of an acceptable dose (milligrams per kilogram
[mg/kg] body weight), the exposure must be calculated
in terms of dose. The values normally used by EPA for
estimation of waterborne exposure are the daily con-
sumption of 2 liters of water and 6.5 g of fish, or other
food derived from the affected area, by a 70 kg adult
for the lifetime of the population. The value for food
consumption can be adjusted for differences in diet if
the pollutant is localized. A second scenario is used by
the Office of Drinking Water in deriving Health Effects
Advisories (formerly SNARLS). In order to protect the
most sensitive group, exposure is determined using a
10 kg child drinking 1 liter of water per day.
Assessing the bioaccumulation route of exposure
involves estimating fish body burden and human con-
sumption. BCF can be calculated from the partition
coefficient or may be available in the literature. As
stated above, EPA typically uses a daily consumption
of 6.5 g of fish by a 70 kg adult for 60 to 70 years.
These values are used for national risk assessments
and may be altered if local information is available.
Impacts to wildlife should also be considered in evalu-
ating the discharge of toxicants.
If the exposure exceeds levels that the toxicology infor-
mation indicates as no effect levels, regulatory controls
are needed. Discharge requirements are simply back-
calculated from the exposure or dose level that is
deemed acceptable using the same assumptions as
were used in the assessment. How the acceptable
exposure or dose was selected should be described in
the permit fact sheet.
References
1. U. S. Environmental Protection Agency. 1982.
Workshop on Short-term Bioassays for Estimat-
ing Carcinogenic Risk. J. Am. Coll. Tox. 1,
pp. 1-186.
2. Woodhead, A. D., and M. D. Waters (editors).
1983. Short-term Tests for Environmentally In-
duced Chronic Health Effects. EPA-600/8-83-
002.
3. Sexton, N. G. 1979. Biological Screening of Com-
plex Samples from Industrial/Energy Processes.
EPA-600/8-79-021.
4. McGeorge, L. S. , J. B. Louis, T. B. Atherholt, and
G. J. McGarrity. 1983. Mutagenicity Analyses of
Industrial Effluents: Background and Results to
Date. Office of Science and Research, Depart-
ment of Environmental Protection, State of New
Jersey, Trenton, NJ
5. Organization of Economic Cooperation and
Development. 1984. Guidelines for Testing
Chemicals. Section 4—Health Effects. Director
of Information, OECD, 2, rue Andre-Pascal,
75775 Paris CEDEX 16, France.
6. U. S. Environmental Protection Agency. 1983.
Testing for Environmental Effects under the
Toxic Substances Control Act. Office of Toxic
Substances (TS-796), Washington, D.C.
7. Veith, G., N. Austin, and R. Morris. 1979. A Rapid
Method for Estimating Log P for Organic Chem-
icals. Water Res. 13, pp. 43-47.
8. U. S. Environmental Protection Agency. 1984.
Draft Test Guideline CG-1410 Partition Coeffi-
cient (n-Octonol/Water). Estimation by Liquid
Chromatography. Office of Toxic Substances
(TS-796), Washington, D.C. 20460
9. U. S. Environmental Protection Agency. 1982. Pes-
ticide Assessment Guidelines. Office of Pesti-
cide Programs, Washington, D.C. EPA 540/
9-82-through 028.
30
-------
5. Exposure and Wasteload Allocation
Introduction
At this point in the toxics control process, a water qual-
ity problem area has been identified. Screening ana-
lyses may have been done to assess the extent of
toxicity, or the decision may have been made to pro-
ceed directly to a wasteload allocation (WLA). Criteria
that will protect the receiving water have been deter-
mined from the following sources (see Section 2 and
Appendix D):
• For specific toxicants, approved State standards
(numerical or narrative) have been used. Site-
specific criteria may have been derived with an
EPA-approved methodology.
• For effluent toxicity, approved State standards
have been used. Various species may have been
tested for sensitivity to the toxicants, and criteria
have been derived using an acute-chronic ratio
(ACR) and species sensitivity factor as appropriate.
• For both specific toxicants and effluent toxicity,
appropriate durations and frequencies for criteria
compliance have been selected.
The next step is to assess the exposure of the resident
aquatic community to toxic conditions. Exposure
assessments involve predictions of how much of a
waterbody is subject to concentrations exceeding
water quality criteria, for how long, and how frequently.
The spatial and temporal extent of aquatic life exposure
to toxicants will vary depending on variations in the
assimilative capacity of the receiving water and varia-
tions in effluent composition and quantity. WLA models
are used to predict exposure and to calculate the
effluent quality required to meet the criteria and pro-
tect the beneficial uses of the receiving water. This sec-
tion describes the types of WLA modeling that can be
performed. Data requirements for each of these model-
ing techniques are also described so that the definitive
data generation procedures presented in Section 3 can
be designed to support the specific type of WLA model-
ing a regulator chooses to perform.
Part 1 of this section discusses EPA's mixing zone
policy and explains how this policy affects the allow-
able toxic load that can be discharged from a treatment
plant. At present most WLA studies do not include mix-
ing zone analyses since they concentrate on conven-
tional pollutants that have their greatest impact
downstream of the point of complete mixing. Mixing
zone modeling has generally been performed only for
discharges to large rivers, lakes, or estuaries where
water quality problems are confined to the mixing area.
Now that WLA studies will be conducted for toxicants,
mixing analyses may be needed in more situations
since many of these pollutants exert their maximum
toxicity close to the point of discharge. Part 1 describes
the tracer studies and mixing zone models that can be
used to: 1) ensure that the discharge conforms to the
State's allowable mixing zone dimensions; 2) prevent
the mixing area from extending into critical resource
areas; and 3) provide boundary conditions for the com-
pletely mixed WLA models discussed in Part 2 of this
section.
Part 2 of this section describes the various types of
models for completely mixed conditions that can pre-
dict the effect of wastewater discharges on the dura-
tion and frequency of toxicant concentrations in a
receiving water. For steady state modeling of toxic dis-
charges, Part 2 recommends the use of toxicologically
based design flows for rivers and critical conditions for
lakes and estuaries. As a replacement for steady state
models, Part 2 also recommends the use of continuous
simulation or probabilistic models to predict more
accurately the duration and frequency with which
criteria may be exceeded. These non-steady state
models are appropriate for river, lake, and estuary ana-
lyses where adequate data are available.
Part 1 — The Mixing Zone and Toxic
Wasteload Allocations
Overview
When wastewater is discharged into a waterbody, its
transport may be divided into two stages with distinc-
tive mixing characteristics. Mixing and dilution in the
first stage is determined by the initial momentum and
buoyancy of the discharge. The second stage covers a
more extensive area in which the effect of initial
momentum and buoyancy is overridden, and the waste
is mixed primarily by ambient turbulence. In large rivers
or estuaries, this second stage mixing area may extend
for miles before uniformly mixed conditions are
attained. In some instances, such as larger lakes or
coastal bays, completely mixed conditions are never
reached in the waterbody.
As shown in Table 5-1, most States specify spatial
dimensions for regulatory mixing zones that do not
31
-------
Table 5-1. State-by-State Mixing Zone
State
Alabama
Alaska
Arizona
Arkansas
California
Colorado
Connecticut
Delaware
District of Columbia
Rorida
Georgia
Hawaii
Idaho
Illinois
Indiana
Iowa
Kansas
Kentucky
Louisiana
Maine
Maryland
Massachusetts
Michigan
Minnesota
Mississippi
Missouri
Montana
Nebraska
Nevada
New Hampshire
New Jersey
New Mexico
New York
North Carolina
North Dakota
Ohio
Oklahoma
Oregon
Pennsylvania
Rhode Island
South Carolina
South Dakota
Tennessee
Texas
Utah
Vermont
Virginia
Washington
West Virginia
Wisconsin
Wyomino
Guam
Puerto Rico
Virgin Islands
Water Body
0
river, streams
lakes
streams
lakes
large streams
small streams
0
0
streams
streams
lakes
estuary
streams, rivers
lakes, estuaries
0
streams
estuaries
0
all
streams
streams
streams
streams
streams
streams
streams
0
0
streams
Lake Michigan
streams
0
streams
0
0
streams
streams
streams
streams
streams
0
streams
receiving watercourse
mouth of receiving
streams
0
0
streams
0
streams
0
streams
0
streams
0
0
warm water fish
streams
cold water fish streams
lakes
streams
o
streams
streams
IMZ
FM2
streams
Dimensions
Dimensions
0
<1/3CS
< 10% SA
« 1 /2 CS, < 500 meters length
< 10% SA
<1/4CS
< 2 / 3 CS
0
0
< 1/4 CS
< 1/3 CS
< 10% SA
< 10% CS
< 800 meters
< 10% total length
< 125,600 m2 (600- radius)
< 10% SA
0
0
0
0
< 600 ft radius
<1 1 A PQ
1 /H l-O
< 1/4 CS
«1/4CS
«1/4CS
< 1/3 CS
< 1 /4 CS
< 1/4 CS
0
0
< 1/4 CS
< 1000 ft radius
< 1/4 CS
0
< 1 /4 CS
0
o
0
<1/4CS
< 1/4 CS (thermal)
< 1/4 CS
< 1/2 CS (thermal)
0
< 1/4 CS
41/3CS
< 1/5 CS
< 1/4 CS
o
0
< 1 /4 CS (thermal)
0
< 3/4 CS or 100 yards of
stream width
0
< 1/4 CS
0
< 1/4 CS
0
0
<33% CS
<20%CS
< 300' any direction
<1/4CS
Q
< 1 /4 CS, < 5 stream width
length
< 1/4 CS
<400ft
<4000ft
< 1/4 CS
conform to the dimensions of the physically based mix-
ing areas described above. State regulations dealing
with streams and rivers generally limit mixing zone
widths or cross-sectional areas and allow lengths to be
determined on a case-by-case basis. In the case of
lakes, estuaries, and coastal waters, dimensions are
sometimes specified for the surface area that can be
affected by the discharge. The surface area limitation
usually includes the underlying water column and ben-
thic area. State specification of mixing zone dimen-
sions means that the WLA cannot be based on the
assumption that the wastewater is completely mixed
in the receiving water. State standards require that
criteria be met at the edge of the regulatory mixing zone
1) to provide a continuous zone of passage that meets
water quality criteria for free-swimming and drifting
organisms and 2) to prevent impairment of critical
resource areas.
Even when State regulations include detailed geomet-
ric specifications for mixing zones, a smaller mixing
zone or no zone at all can be required in site-specific
cases. Bioaccumulative pollutants in particular may
exert unexpected impacts if a mixing zone is allowed.
A regulator should consider prohibiting a mixing zone
in situations where an effluent is known to attract fish.
In such cases, provision of a continuous zone of pas-
sage around the mixing area will not serve the purpose
of protecting aquatic life. A review of the technical liter-
ature on avoidance/attraction behavior revealed that
the majority of toxicants elicited an avoidance response
or a neutral response [1]. The 53 toxicants in this review
included a variety of heavy metals, organochlorine
compounds, organophosphates, phenols, and resin
acids. Six individual pollutants (copper, cadmium, mer-
cury, total residual chlorine, chloroform, and NTA) were
reported as eliciting attractive responses in one or more
of the tested species. It should be noted that fish may
not be attracted to effluent containing these attractive
pollutants if other chemicals that cause avoidance
behavior are present. The reverse is also true; an attrac-
tive pollutant or preferred temperature may overcome
the avoidance reaction to specific chemicals in an
effluent.
The purpose of this part of the Exposure Assessment
Section is: 1) to describe EPA's mixing zone policy and
2) to summarize the monitoring and modeling tech-
niques that are available for mixing analyses. In general,
the modeling techniques are applicable to routine sit-
uations. Depending on the complexity of the hydro-
dynamics of the discharge and the receiving water,
more detailed studies and/or the advice of experts may
be required. The policy and techniques apply to both
individual toxicants and effluent toxicity.
EPA Mixing Zone Policy
CS = cross-sectional area SA = surface area 0 = not listed
Source: Update of information in U.S. EPA Draft Technical Guidance Manual for
the Regulations Promulgated Pursuant to Section 301 (g) of Clean Water Act of 1977.
The EPA mixing zone policy is described in the 1983
Water Quality Standards Handbook [2]. The 1983
policy describes the mixing zone as an "allocated
32
-------
impact zone" where numeric water quality criteria can
be exceeded as long as acutely toxic conditions are
prevented. "Acutely toxic" conditions are defined as
those conditions lethal to aquatic organisms passing
through the mixing zone. Lethality is a function of the
magnitude of pollutant concentrations and the duration
of organism exposure to those concentrations.
EPA's water quality criteria provide guidance on the
magnitude and duration of pollutant concentrations
causing lethality. The criterion maximum concentration
(CMC) for whole-effluent toxicity prevents lethality or
acute effects in the tested species. The CMC for
individual toxicants prevents acute effects in all but a
small percentage of the tested species. Exposure to
concentrations in excess of the CMC must be limited
to minutes to prevent lethality or acute effects. Figure
5-1 illustrates how estimates of organism exposure
time can be made by comparing swimming or drifting
rates to the dimensions of concentration gradients
within the regulatory mixing zone. Lethality can be
prevented in the mixing zone in one of two ways. The
first method is to prohibit lethal concentrations (values
in excess of the CMC) in the pipe itself. The second
approach is to use high rate diffusers and to ensure that
the CMC is met within a short distance from the out-
fall. Details on how to use this second method are
provided in the next subsection.
Proper design of a WLA study for a particular water-
body requires estimation of the distance from the out-
fall to the point where the effluent mixes completely
with the receiving water. This determination can be
based on the results of a tracer study or the simple
desktop calculations described in the following subsec-
tions. If the mixing area is insignificant in size, as will
occur with effluent-dominated streams, completely
mixed conditions can be assumed, and the fate and
transport models described in Part 2 can be used for
the WLA.
At completely mixed conditions, two WLAs should be
performed — one to meet the CMC and one to meet
the criterion continuous concentration (CCC) at the
required duration and frequency. Appendix D describes
appropriate durations and frequencies for the CMC and
CCC. As described in Part 2 of this section, both steady
state and dynamic modeling can be performed to make
these allocations. The more stringent of the two WLAs
will determine the final permit limit.
If completely mixed conditions do not occur within a
short distance of the outfall, the WLA study should rely
on mixing zone monitoring and modeling. Just as in the
case of completely mixed models, mixing zone ana-
lyses can be performed using both steady state and
dynamic techniques. State requirements regarding the
Average concentration
= 30 mg/l
ET2 = 17 min.
Average concentration
= 15 mg/l
ET2 = 52 min.
Source: National Academy of Science. Water Quality Criteria 1972.
Figure 5-1. Hypothetical predicted concentrations of toxicant in an effluent plume and times of passage of migrating fish.
33
-------
mixing zone will determine how water quality criteria
are used in the WLA.
If a State allows a mixing zone, the WLA must ensure
that, under chronic design conditions, concentrations
at the mixing zone boundary meet the CCC. The CCC
for whole-effluent toxicity is defined as a No Observed
Effect Level (NOEL) that will prevent lethality and
growth and reproduction impairment in the tested spe-
cies. EPA's CCC for individual toxicants prevents unac-
ceptable effects on the survival, growth, or
reproduction of all but a small percentage of the tested
species. If a State does not allow a mixing zone, con-
centrations in the pipe itself cannot exceed the CCC.
The following subsections describe how lethal condi-
tions can be prevented in the mixing zone of each type
of receiving water. General guidelines are provided on
how to monitor mixing zone concentrations and how
to design appropriate outlet structures. Some desktop
and computer models for predicting mixing zone
dimensions will also be summarized.
Design Criteria to Prevent Lethality in the
Mixing Zone
In order to prevent lethal conditions in the regulatory
mixing zone, the State can prohibit lethal concentra-
tions in the pipe itself or require high rate diffusers and
CMC compliance within a short distance of the outfall.
If the high rate diff user alternative is selected, the out-
fall discharge velocity must exceed three meters per
second (ten feet per second). This will provide turbu-
lent initial mixing and will minimize organism exposure
time. In order to limit exposure time further, the outfall
design must ensure that the most restrictive of the fol-
lowing conditions are met:
• The CMC must be met within 10% of the distance
from the edge of the outfall structure to the edge
of the regulatory mixing zone in any spatial
direction.
• The CMC must be met within a distance of 50
times the discharge length scale in any spatial
direction. The discharge length scale is defined as
the square-root of the cross-sectional area of any
discharge outlet. In the case of a multiport diffuser,
this requirement must be met for each port using
the appropriate discharge length scale of that port.
This restriction will ensure a dilution factor of at
least 10 within this distance under all possible cir-
cumstances, including situations of severe bottom
interaction, surface interaction, or lateral merging.
• The CMC must be met within a distance of 5 times
the local water depth in any horizontal direction
from any discharge outlet. The local water depth
is defined as the natural water depth (existing prior
to the installation of the discharge outlet) prevail-
ing under mixing zone design conditions (e.g., low
flow for rivers). This restriction will prevent locat-
ing the discharge in very shallow environments or
very close to shore, which would result in signifi-
cant surface and bottom concentrations.
General Recommendations for Tracer
Studies
Analysis of the mixing zone should be performed under
critical conditions of minimal dilution. Some of those
design conditions are summarized in the following sub-
sections dealing with specific waterbodies. It should
be noted, however, that occasionally certain off-design
conditions may cause reduced mixing and water qual-
ity problems. Off-design conditions are described in the
subsections below and should also be addressed in
analyses.
Once the critical design condition has been selected for
a waterbody, tracer studies or predictive models can be
used to provide data on the dimensions and dilution of
the wastewater plume during this critical period. The
concentrations in the discharge jet or plume should be
superimposed over a map of the various resource zones
of the waterbody. The map will illustrate whether the
State's required zone of passage is provided and
whether the plume avoids critical resource areas. The
WLA must be calculated to provide the appropriate
zone of passage and to prevent detrimental impacts on
spawning grounds, nurseries, water supply intakes,
bathing areas, and other important resource areas.
A number of references that describe how to conduct
tracer studies for mixing analyses are available [3-10].
For WLA studies in which a discharge is already in oper-
ation, tracer studies can be used to determine concen-
tration isopleths in the mixing zone. This type of study
inherently includes any discharge-induced (or first
stage) mixing. However, if the outfall is not yet in oper-
ation, it is impossible to determine discharge-induced
mixing by tracer studies. Tracer studies can be used in
these situations to determine characteristics of the
ambient (or second stage) mixing. For ambient mixing
studies, the tracer release can be either instantaneous
or continuous. Instantaneous releases are used fre-
quently to measure longitudinal dispersion, but can also
be used to determine lateral mixing in rivers [11] and
lateral and vertical mixing in estuaries, bays, reservoirs,
and lakes. For waterbodies with significant flow veloc-
ities, continuous releases of tracer are normally used
to determine lateral and vertical mixing coefficients.
Continuous releases can also be used to determine
three-dimensional concentration isopleths for steady
state conditions. The tracer study must be made at crit-
ical design conditions in order to use the results directly
for WLAs. If a tracer study for ambient mixing is con-
ducted at near-to-design conditions, the observed data
can be used to determine dimensionless mixing coeffi-
cients. These coefficients can then be extrapolated to
critical conditions using hydraulic parameters [11]. A
tracer study at near-to-design conditions can also be
used to determine the computer model required to pre-
34
-------
diet critical condition mixing and to provide the coeffi-
cients needed for that WLA model.
General Recommendations for Outfall
Design
Surface discharges at the shoreline of a waterbody are
not recommended for toxic discharges. They usually
have an impact along the shoreline when there is a sig-
nificant crossflow, and they yield high surface concen-
trations.
Submerged discharges offer more flexibility in meeting
the design goals for toxic discharges. They may be in
the form of a single pipe outlet or of multiport diffusers,
giving rise to one or several submerged discharge jets.
Submerged multiport discharges do, however, have cer-
tain limitations in their use. They are not feasible in very
shallow waterbodies or where periodic dredging or con-
siderable scour and deposition occurs.
Specific design objectives for submerged discharges
should be to avoid direct surface impingement and bot-
tom attachment of the submerged jet or jets. Surface
and bottom impacts should be evaluated at critical
design conditions (low flow or high stratification) and
at off-design conditions (higher flow or lower stratifi-
cation) in order to ensure the best placement and
design of the diffuser. Multiport diffusers provide more
dilution than single outlets, but the alignment of the
diffuser with the receiving water flow direction
influences how much dilution will be provided. If the
outlet structure is directed parallel to the direction of
flow, dilution under high ambient velocities (off-design
conditions) may be worse than under low velocities
(critical design conditions). Some of the complexities
of multiport diffusers have been summarized by Jirka
[12], Roberts [13], and Holley and Jirka [11].
In rivers the preferred arrangement for a submerged dis-
charge is to direct the outlet into the flow direction or
vertically upward. In order to deal with the reversing
currents of estuaries and coastal bays, the preferred
arrangements for offshore discharges are parallel
diffuser alignment (tee diffuser) or perpendicular
diffuser alignment (staged diffuser) [12]. In lakes and
reservoirs, the preferred arrangement for a negatively
buoyant discharge is to direct the diffuser vertically
upward. A positively buoyant, vertically directed jet
could penetrate stratification, so the preference for this
discharge is to orient the diffuser at a slight angle above
the horizontal.
River and Run-of-river Reservoir Analyses
Rivers and run-of-river reservoirs are waterbodies that
have a persistent through-flow in the downstream
direction and do not exhibit significant natural density
stratification. Interim recommendations on toxicolog-
ically based design flows for mixing zone and com-
pletely mixed steady state modeling of rivers are
described in Appendix D of this document. Run-of-river
reservoirs with residence times less than 20 days at
critical conditions should also be analyzed using these
design flows [14]. Regulated rivers may have a mini-
mum flow in excess of these toxicological flows. In
such cases, the minimum flow should be used in WLA
modeling.
When a waste is discharged into a river, any discharge-
induced mixing (i.e., mixing associated with the
momentum and/or buoyancy of the effluent) occurs
first. The second stage of mixing is controlled by
ambient turbulence. The ambient mixing first accom-
plishes mixing over the depth, if this was not provided
by the discharge-induced mixing, and then ambient
mixing accomplishes mixing over the width. A final
stage occurs after the waste is fully mixed throughout
the cross-section. For nearly steady state conditions in
rivers, the concentration distribution in this final stage
is influenced primarily by the flow velocity and any
biochemical and physico-chemical processes.
Models for Discharge-induced Mixing
The first stage of mixing is controlled by jet momentum
and buoyancy. This stage generally encompasses most
of the regulatory mixing zone. Especially in shallow
environments, it is important to ascertain whether near
field instabilities occur. These instabilities, associated
with surface and bottom interaction and localized recir-
culation cells extending over the entire water depth,
can cause build-up of effluent concentrations. Criteria
for these instabilities and specialized predictive models
have been developed [11,12].
In the absence of near field instabilities, horizontal or
nearly horizontal discharges will create a clearly defined
jet in the water column that will initially occupy only
a small fraction of the available water depth. The fol-
lowing equations and models are designed for such sta-
ble near field conditions.
A minimum estimate of the dilution available in a mix-
ing zone can be made using this equation:
S = 0.31
D
Where S = flux-averaged dilution
s = distance from outlet to edge
of the mixing zone
D = diameter of outlet
The equation provides a minimum estimate of mixing
zone dilution because it is based on the assumptions
that ambient velocity is zero and that the discharge is
neutrally buoyant. The equation also assumes that the
jet is strong enough to reach the edge of the mixing
zone. The same equation can be used to estimate the
minimum dilution within the mixing zone before the
CMC must be met to prevent lethal conditions.
Mixing graphs that include the effect of discharge
buoyancy and ambient velocity can be found in Holley
35
-------
and Jirka [11], Fischer et al. [15], and Wright [16]. They
are useful and accurate for design to meet criteria at
the edge of the regulatory mixing zone.
More detailed design data for the mixing zone can be
obtained from the use of computer models based on
integral jet techniques. The models recommended here
are available through the EPA. Other models can be
used if well documented and verified with existing data.
It is important to note that most models represent an
idealization of actual field conditions and must be used
with caution to ensure that the underlying model
assumptions hold for the site-specific situation being
modeled. In general these buoyant jet models require
the following input data: discharge depth, effluent flow
rates, density of effluent, density gradients in receiv-
ing water, ambient current speed and direction, and
outfall characteristics (port size, spacing, and orienta-
tion). Model output includes the dimensions of the
plume at each integration step, time of travel to points
along the plume centerline, and the average dilution at
each point. The following models are available for use
in riverine analyses.
MERGE is an ocean plume model that can be used for
riverine analyses by setting ambient density gradients
to zero. The model analyzes positively buoyant, co-
flowing multiple-port discharges [17]. Current speed is
allowed to vary with depth. The multiple diffuser ports
must be equally spaced and in line but may be oriented
at any common elevation angle.
LINE is an ocean plume model that can be used for river-
ine analyses by setting ambient density gradients to
zero. The model accounts for adjacent plume interfer-
ence from a line source of positively buoyant, multiple-
port discharges [18]. Unlike MERGE, diffuser alignment
can be specified for LINE. Current speed is allowed to
vary with depth.
OUTPLM is an ocean plume model that can be used for
riverine analyses by setting ambient density gradients
to zero. The model analyzes a positively buoyant, co-
flowing single-port discharge into a waterbody with a
uniform ambient current [19, 20]. The model does not
include the dynamics of surface impingement or the
development of a bifurcated plume after impingement.
A model that includes these features has been devel-
oped but is not available from EPA [21].
Mixing zone analyses for run-of-river reservoirs can use
all the models listed under discharge-induced mixing
for reservoirs, except for PLUME, which requires an
ambient velocity of zero.
Models for Ambient Mixing
The models for discharge-induced mixing can be used
to predict concentrations in the regulatory mixing zone
when there is strong jet mixing. Beyond this point, the
second stage of mixing in rivers is controlled by ambient
turbulence. Pollutant concentrations at the end of this
second zone must be predicted to provide boundary
conditions for the completely mixed fate and transport
models described in Part 2 of this section. Concentra-
tion distributions within the ambient mixing zone may
also be needed to estimate concentrations at important
resource areas or at downstream dischargers.
If there is no discharge-induced mixing associated with
buoyant jet action of the discharge, then mixing over
the depth of the river must be accomplished by ambient
mixing. For a neutrally buoyant, soluble effluent dis-
charged with low velocity at the surface or at the bed
of a stream, the flow distance required to achieve com-
plete vertical mixing is on the order of 50 to 100 times
the depth of water in that portion of the channel where
the effluent is discharged [22]. For a discharge that is
either lighter (positively buoyant) or heavier (negatively
buoyant) than the ambient water but still has no excess
momentum, the flow distance for mixing over the depth
will be greater. In the normal case with a high-velocity
jet designed to prevent lethality in the mixing zone, mix-
ing over the depth will be accomplished primarily by jet
action, and the distance required for this vertical mix-
ing will be much shorter.
In general, ambient mixing must also accomplish mix-
ing over the width of a stream to bring the effluent to
the completely mixed condition. For situations where
the width of the zone that is mixed by the discharge-
induced mixing is much smaller than the width of the
river, the flow distance (Xm) required to achieve the
completely mixed condition may be estimated from an
equation of the form [15,11]:
= mW2u
Uy
where W=width of the river, u=flow velocity for the
critical design flow, Dy = lateral dispersion coefficient
as discussed in the next paragraph, and m = a parame-
ter whose value depends on the degree of uniformity
used to define "complete mixing" and on the trans-
verse location of the outfall in the stream. If completely
mixed conditions are defined as a 5% variation in con-
centration across the stream width, the value of m
would be 0.3 to 0.4 for a discharge near the center of
river flow (not the center of river width) and 0.4 to 0.5
for a discharge near the edge of the river. If, because
of other uncertainties in the problem, a 25% variation
across the width can be accepted as being completely
mixed, then the corresponding ranges of values for m
for the two locations would be 0.1 to 0.2 and 0.2 to 0.3.
For a very small stream, Xm may be only a few hundred
feet; for medium and large streams, Xm is normally
several miles to several tens of miles. For diff users that
initially spread the discharge across a significant part
of the river width or for cases where the discharge-
induced mixing causes mixing across a significant part
of the river width, the values of m and Xm can be
smaller than the ones indicated here. For distances
greater than Xm, the completely mixed models dis-
cussed in Part 2 of this section may be used. For shorter
36
-------
distances, the maximum concentrations in the river can
be much greater than those predicted by completely
mixed models and should be estimated using the fol-
lowing equations.
The lateral dispersion coefficient for most rivers can be
calculated with the following equation [15, 11]:
Dy = 0.6 du* ± 50%
where d = water depth at critical design flow and
u*=shear velocity. The coefficient (0.6) can vary
between 0.3 and 1.0 depending on the type and degree
of irregularity of the channel cross sections. Values
approaching and exceeding 1.0 are normally associated
with significant channel meandering [22]. The follow-
ing equation for shear velocity should be used [15]:
u* = (gdS)14
where g=acceleration due to gravity, S = slope of the
channel, and d = water depth.
Once an estimate of the distance to complete mixing
is made with the above equations, the following equa-
tion can be used to predict concentrations upstream of
this distance:
CpCLW
Cx =
where Cx = maximum pollutant concentration at dis-
tance x from the outlet, Ce=effluent concentration,
Qe= effluent flow, Qs = critical design stream flow,
X=distance from outlet, Dy = lateral dispersion coeffi-
cient, W=stream width, and u=flow velocity for the
critical design flow.
It should be noted that this estimate of Cx is a worst
case prediction since the equations assume no signifi-
cant discharge-induced mixing and a neutrally buoyant
effluent. A more accurate way to predict concentra-
tions within this second stage of mixing is to use the
methods of Yotsukura and Sayre [22]. In order to use
this approach, however, the value of Dy and pollutant
concentrations after discharge-induced mixing must be
known from tracer studies and/or from the detailed
modeling discussed in the previous subsection.
The PSY model can also be used to predict ambient
mixing in shallow, freshwater streams where water
depth is small in proportion to the width. PSY is a
steady state two-dimensional plume model that
predicts dilution of a surface discharge in a shallow
receiving water where the plume attaches to both bot-
tom and near shore [23]. Uniform vertical mixing is
assumed to occur at the point of discharge.
Lake and Reservoir Analyses
This receiving water category encompasses lakes and
reservoirs with residence times in excess of 20 days at
critical conditions [14]. In the case of long and narrow
reservoirs, areas above the plunge point can be ana-
lyzed as rivers and areas below as reservoirs [14]. Since
effluent density relative to the ambient water can vary
over seasons, no one season or stratification condition
can be selected as the most critical dilution situation
for all cases. In general, four seasons should be ana-
lyzed to determine the most critical period(s) for mix-
ing zone analyses. All seasonal analyses should
assume an ambient velocity of zero unless persistent
currents have been documented. Special attention
should be given to rising water level periods since pol-
lutants can move back into coves and accumulate
under these conditions. Location of discharges in coves
and dead-end embayments should be prevented when-
ever possible.
Models for Discharge-induced Mixing
The same discharge-induced mixing equations and
models available for riverine analyses are also appropri-
ate to predict concentrations in the regulatory mixing
zone of lakes and reservoirs. The only difference is that
ambient velocity should always be set equal to zero
unless persistent currents occur in a site-specific case.
In addition, the following models are appropriate for
lake and reservoir analyses:
PLUME analyzes a positively buoyant, single-port dis-
charge into an arbitrarily stratified water body [24, 25].
No current is allowed in PLUME so stagnant conditions
must prevail. Both surfacing and non-surfacing plumes
can be traced with the model.
PDS is a steady state three-dimensional plume model
that simulates the mixing of a surface discharge in a
deep receiving water [26,27]. The plume is assumed
to make no contact with the bottom or shoreline.
Ambient current must be less than one-half the dis-
charge velocity.
PDSM is a steady state three-dimensional plume model
that predicts the dilution of a surface plume in a deep
receiving water [27]. PDSM assumes that the plume
is not attached to the bottom but that high ambient cur-
rents cause the plume to attach to the near shore.
DKHDEN is an ocean plume model that can be used for
riverine analyses by setting ambient density gradients
to zero. The model is designed for submerged, posi-
tively buoyant discharges at sufficient depth that the
plume never surfaces [28]. The effects of a single or
merging multiple-port diffuser can be simulated, but
the outlets must be directed parallel to the flow direc-
tion (co-flowing diffuser design). Current speed is al-
lowed to vary with depth.
Models for Ambient Mixing
Discharge-induced mixing is especially crucial for lakes
and reservoirs as ambient mixing processes play a
lesser role. For WLA modeling purposes, the discharge-
induced mixing concentrations can be used as bound-
ary conditions for two-dimensional models that are
appropriately configured for this input. These models
are referenced in Part 2 of this section.
37
-------
Estuary and Coastal Bay Analyses
This receiving water category encompasses estuaries,
which are defined as having a main channel reversing
flow, and coastal bays, which are defined as having sig-
nificant two-dimensional flow in the horizontal direc-
tions. For both waterbodies, the critical design
conditions recommended here are based on astronom-
ical, not meteorological, tides.
In estuaries without stratification, the critical dilution
condition includes a combination of low water slack at
spring tide for the estuary and design low-flow for river-
ine inflow. In estuaries with stratification, a site-specific
analysis of a period of minimum stratification and a
period of maximum stratification, both at low-water
slack, must be made to evaluate which one results in
the lowest dilution. In general, minimum stratification
is associated with low river inflows and large tidal
ranges (spring tide), whereas maximum stratification
is associated with high river inflows and low tidal
ranges (neap tide).
After evaluating either stratified or unstratified estu-
aries at critical design conditions, an off-design condi-
tion should be checked. The off-design condition
recommended for both cases is the period of maximum
velocity during a tidal cycle. This off-design condition
results in greater dilution than the design condition, but
it causes the longest extension of the plume. Extension
of the plume into critical resource areas may cause
more water quality problems than the high concentra-
tion, low dilution situation.
Recommendations of a critical design for coastal bays
are the same as for stratified estuaries. The period of
maximum stratification must be compared with the
period of minimum stratification in order to select the
worst case. The off-design condition of maximum tidal
velocity should also be evaluated to predict the extent
of the plume.
Models for Discharge-induced Mixing
Most of the discharge-induced mixing equations and
models recommended for riverine and reservoir ana-
lyses are also appropriate to predict concentrations in
the regulatory mixing zone of estuaries and coastal
bays.
A minimum estimate of the dilution available in a regula-
tory mixing zone can be made with the same equation
used previously:
S = 0.31 f
Mixing graphs that include the effect of discharge
buoyancy, ambient velocity, and stratification can be
found in Fischer et al. [15] and Wright [16].
Models for Ambient Mixing
For estuaries that are completely mixed with regard to
salinity, the equations presented for rivers can be used
to estimate concentrations between the outlet and the
point of complete mixing. These equations will be
applicable to only a very few unstratified estuaries,
however, since the time required to mix across the estu-
ary must be significantly less than: 1) the time required
for the effluent to pass out of the unstratified part of
the estuary, 2) the time required for the effluent to pass
into a segment of greatly changed cross-section, or 3)
the time required for the substance to decay.
mW2u
Dy = 0.6 du* ± 50%
u* = 0.10ut
CeQeW
Qs(7rDyX/ur
where all the variables are defined as for rivers except
that u in the equation for Xm and Cx should be the
velocity associated with the freshwater flow into the
estuary, whereas ut in the equation for u* should be
based on an average total velocity (including the tidal
component).
For WLA modeling purposes, two-dimensional models
can be configured to use discharge-induced mixing
concentrations as input. These models are referenced
in Part 2 of this section.
Ocean Analyses
Critical design conditions and modeling and monitor-
ing techniques for oceans are described in the 301(h)
Technical Support Document and the 301(h) publica-
tion entitled Initial Mixing Characteristics of Municipal
Ocean Discharges [29,17]. This information will not be
repeated here.
Part 2 — Wasteload Allocation Modeling
Overview
At the present time, most States and EPA Regions use
steady state models that assume the wastewater is
completely mixed with the receiving water in order to
calculate WLAs for pollutants. As explained in Part 1
of this section, the assumption of completely mixed
conditions may be adequate in most situations involv-
ing conventional pollutants since they have the greatest
impact downstream of the discharger. Now that WLAs
will be conducted for toxicants, mixing analyses may
be needed in more situations because many of these
pollutants exert maximum toxicity close to the point of
discharge. The completely mixed models described
here in Part 2 are appropriate for any effluent-
dominated receiving waters or situations in which mix-
ing is relatively complete within a short distance. In
multiple-discharge situations, these models will also be
needed to link the concentrations predicted at the edge
of allowable mixing zones to downstream impacts.
38
-------
The purpose of this part of the Toxics Control Support
Document is to improve existing toxicant modeling
procedures by: 1) recommending for steady state
modeling of rivers the use of critical flows that are
based on toxicological concerns and 2) recommending
dynamic modeling techniques for rivers, lakes, and
estuaries that are better suited to toxic exposure
assessments than the steady state approach. Section
6 will describe how to improve the procedure for deter-
mining permit limits from the WLA results.
Steady State Modeling for Rivers, Lakes,
and Estuaries
A steady state model requires single, constant inputs
for effluent flow, effluent concentration, background
receiving water concentration, receiving water flow,
and meteorologic conditions such as temperature. The
frequency and duration of ambient concentrations
predicted with a steady state model must be assumed
to equal the frequency and duration of the critical
receiving water conditions used in the model. The varia-
bility in effluent flows and concentrations also affects
receiving water concentrations, but these effects can-
not be predicted with constant input. Steady state
models can be improved for toxic WLAs by: 1) using
design flows that will ensure criteria compliance at the
appropriate duration and frequency and 2) calculating
both acute and chronic WLAs.
Appendix D of this document describes the toxicolog-
ical basis for selecting receiving water flows for steady
state modeling. The duration of the design flow is
based on the maximum exposure time that will prevent
acute or chronic effects. The duration of flow is as-
sumed to apply to the duration of the allowable effluent
concentration or load. For example, if the flow used is
a seven-day average value, the allowable load is con-
sidered to be a seven-day average. The return period or
frequency of the flow is based on the number of years
required for biological population recovery after criteria
have been exceeded.
As an interim measure. Appendix D recommends
specific flows for steady state modeling of rivers. These
flows are suggested for use only until a detailed gui-
dance manual is developed on how to select toxicolog-
ical flows using site-specific data. At the present time,
there are no recommended toxicological flows for
steady state modeling of lakes, reservoirs, or estuaries.
The critical conditions recommended for these water-
bodies in Part 1 are worst case situations that are not
derived from toxicological duration and frequency data.
These worst case conditions should be used until fur-
ther guidance is provided.
Another improvement in steady state toxics modeling
can be realized by performing two WLAs — one for
maximum criteria and one for continuous criteria. EPA
is encouraging the States to adopt two-number water
quality criteria that specify a CMC and a CCC. The CMC
and CCC have recommended durations and frequen-
cies associated with them (See Appendix D). Steady
state WLA models should be used to calculate the
allowable effluent load that will meet the maximum
criterion at the acute design flow and the allowable load
that will meet the continuous criterion at the chronic
design flow. Calculation of these values will ensure that
the treatment system is designed to comply with both
criteria. Calculation of two WLAs will also provide
needed data on effluent variability. This information will
aid in the statistical derivation of permit limits that is
discussed in Section 6.
Dynamic Modeling Techniques for Rivers,
Lakes, and Estuaries
As explained in Appendix D, the extent of biological
impairment from toxic discharges depends on the dura-
tion of criteria violations as well as the number of times
these violations occur. The duration and frequency of
violations depend, in turn, on the daily variation in
receiving water and effluent flow combined with daily
variation in effluent toxicity. Although the steady state
modeling approach can only consider the impact of
receiving water variability on criteria violations,
dynamic modeling techniques can predict the effects
of both receiving water and effluent variability. These
more thorough methods calculate an entire probabil-
ity distribution for receiving water concentrations
rather than a single worst case based on critical con-
ditions. The prediction of complete probability distribu-
tions allows the risks inherent in alternative treatment
strategies to be quantified. The use of probability dis-
tributions in place of worst case conditions has been
accepted practice for years in water resource engineer-
ing where it was found to produce more cost-effective
design of bridge openings, channel capacities, flood
plain zoning, and water supply systems. The same cost
effectiveness can be realized for pollution controls if
probability analyses are used. For example, comparison
of receiving water concentration probability distri-
butions can be used to determine which treatment sys-
tem causes criteria to be exceeded less frequently —
a less expensive system producing a higher mean tox-
icity concentration but more stable effluent versus a
more expensive facility producing a lower average con-
centration but higher-variability effluent.
The dynamic modeling techniques have an additional
advantage over steady state modeling in that they
determine the entire effluent concentration distribution
required to produce the desired frequency of criteria
compliance. Maximum and monthly average permit
limits can then be obtained directly from this distribu-
tion. Existing practice has been to use a steady state
model to calculate only a chronic WLA. This generates
a single allowable effluent value and no information
about effluent variability. If the steady state model is
used to calculate both acute and chronic wasteloads,
39
-------
limited information on variability will be provided, but
the entire effluent distribution will not be predicted. As
described in Section 6, steady state WLA values are
more difficult to use in permits and enforcement.
The remainder of this subsection describes the
dynamic methods, which include continuous simula-
tion models, the lognormal probabilistic dilution model,
and Monte Carlo simulation using steady state or con-
tinuous simulation models. Recommendations on
when to apply these methods are also presented.
Continuous Simulation Models
As shown in Figure 5-2, a continuous simulation model
uses each day's effluent flow (Qe) and concentration
data (Ce) with each day's receiving water flow (Qs)
and background concentration (Cs) to calculate down-
stream receiving water concentrations [30]. The model
predicts these concentrations in chronological order
with the same time sequence as the input variables
(Cb vs. time). The daily receiving water concentrations
can then be ranked from the lowest to the highest with-
out regard to time sequence. A probability plot can be
constructed from these ranked values, and the occur-
rence frequency of any one-day concentration of
interest can be obtained (Cb vs. frequency). Running
average concentrations for four days, or for any other
averaging period, can also be computed from the daily
concentrations, ranked in order of magnitude, and
presented as a probability plot (see Figure 5-3).
The probability plot generated by the continuous simu-
lation model using existing effluent data will indicate
whether criteria are predicted to be exceeded more fre-
quently than desired. Appendix D discusses how to
select the appropriate occurrence frequency (or return
period) based on the biological recovery period required
for a specific waterbody. If recurrence intervals of 10
or 20 years are desired, at least 30 years of flow data
should be available in order to provide a long enough
record to estimate the probability of such rare events.
The same data requirements are also true for the log-
normal probabilistic and Monte Carlo methods.
If the existing effluent is predicted to cause too fre-
quent instream violations of water quality criteria, addi-
tional treatment is necessary to protect the receiving
water. Two possible approaches can be taken to deter-
mine the allowable effluent distribution. One approach
is based on the simplifying assumption that treatment
will only change the magnitude of effluent concentra-
tions. No changes are assumed to occur in effluent
flows or in the relative variability of effluent concentra-
tions. With these assumptions, no additional model
runs are needed to determine the allowable distribution
for effluent concentrations. The new required effluent
concentration distribution is assumed to be the same
as the existing distribution except that it is reduced in
magnitude by whichever is greater — the percentage
necessary for the one-day average concentrations to
meet the CM C or the four-day average concentrations
CL
Standard
Time
- Standard
Frequency
Figure 5-2. Continuous simulation modeling schematic.
50
-40 —
O)
i30 —
4 days
*c
•510
P
o
•rrrrr. 30days
i i i i i i i i i
» 99.5
Percent of Time Concentration Is Less Than or Equal
0.5 1 2
i ( l
— -
100
*1°A9
o 42.
I Ml
7
Recurrence Interval (years)
Figure 5-3. Concentration-frequency curves.
20
40
-------
to meet the CCC at the desired recurrence interval. Sec-
tion 6 includes the details on how permit limits are der-
ived from the mean and coefficient of variation of
effluent concentrations determined from this analysis.
Information concerning effluent concentration means
and variability can be obtained from databases on exist-
ing treatment plants and from Effluent Guideline
documents.
It may not be valid to assume that effluent concentra-
tions after treatment will have the same coefficient of
variation as concentrations before treatment. (The
coefficient of variation is the standard deviation of a
parameter divided by its mean value.) Studies have
documented that advanced secondary treatment
increases the coefficient of variation of BOD and TSS
concentrations compared to secondary treatment.
Investigations must be conducted to determine how
treatment processes for heavy metals, organic chemi-
cals, and effluent toxicity will change the variability of
these constituents. To account for a change in varia-
bility, an alternative approach must be used to deter-
mine the allowable effluent distribution. Iterative model
runs will have to be performed using different concen-
tration means with the effluent "future treatment" var-
iance until a mean is found that meets the criteria at
the desired recurrence intervals. These iterative model
runs will require stochastic generation of effluent input
data since daily effluent concentrations will not be
available for the hypothetical treatment schemes. The
required "future treatment" mean and coefficient of
variation of effluent concentration can then be used as
described in Section 6 to set permit limits.
EPA's Monitoring and Data Support Division (MDSD)
has developed an interactive program that will auto-
matically create daily effluent flow and concentration
data for input to continuous simulation models. The
program is included in the DYNTOX portion of the
ANNIE program available from the EPA Modeling Cen-
ter at the Athens, Georgia, Environmental Research
Laboratory (ERL) [31]. If the observed data base is fairly
complete but missing a few points, a linear interpola-
tion scheme is used to fill in the gaps. If data are scarce,
a lag-one Markov method is used to generate daily data
stochastically. The lag-one Markov method uses the
mean, standard deviation, and daily correlation coeffi-
cient of the observed data to create random sequences
of data all having the same statistical properties. The
interactive program is written in FORTRAN and is avail-
able for use on mainframe or IBM PC-compatible com-
puters.
Continuous simulation models have the following
advantages:
• They can predict the frequency and duration of
toxicant concentrations in a receiving water.
• They include fate processes for specific toxicants.
• They include transport processes for rivers, lakes,
and estuaries.
• They are capable of analyzing single or multiple
pollutant sources.
• They incorporate the cross-correlation and inter-
action of time-varying pH, flow, temperature, pol-
lutant discharges, and other parameters.
• They incorporate the effect that the serial corre-
lation of daily flows and other parameters has on
the persistence of criteria violations.
• They are capable of synthesizing long term stream-
flow records for ungauged rivers using precipita-
tion and evapotranspiration data.
• Their long simulation times prevent the initial con-
ditions used in the model from affecting the cali-
bration of fate and transport processes.
Continuous simulation models have the following dis-
advantages:
• They require more input data and calibration/verifi-
cation data than a steady state model.
• They require timeseries input data.
Lognormal Probabilistic Model
Without resorting to the continuous simulation method
of computing time variable receiving water concentra-
tions, this probabilistic method uses the lognormal
probability distributions of input variables to calculate
probability distributions of output variables [32]. As a
result, the method requires only the relevant statisti-
cal parameters of the input variables (medians and
coefficients of variation [CVs]) rather than the actual
timeseries data needed for continuous simulation. If
one-day average receiving water concentrations must
be predicted for comparison with the CMC, lognormal
probability distributions of daily input data are needed.
If four-day average concentrations must be predicted,
the lognormal probability distributions of four-day aver-
age input data are required. Because this probabilistic
model as yet cannot incorporate fate and transport
processes, it can only be used to predict the concen-
tration of a substance after complete mixing in a river
and before decay or transformation significantly alters
the concentration.
Two methods are available to compute the probability
that downstream toxicants (or effluent toxicity) will
exceed criteria — an approximate method of moments
and numerical integration. The computations for both
procedures can be programmed for desktop personal
computers. EPA's MDSD has developed an interactive
program that will perform the numerical integration cal-
culations for the lognormal probabilistic dilution model.
The program is included in the DYNTOX portion of the
ANNIE program available from the EPA Modeling Cen-
ter at the Athens, Georgia, ERL [31]. The interactive pro-
gram is written in FORTRAN and is available for use on
mainframe or IBM PC-compatible computers.
The method of moments will be discussed here. It is
based on the following equation:
CT = Ce 0 + Cs (1-0)
41
-------
where CT = downstream concentration
Ce = effluent concentration
Cs = upstream concentration
0 = effluent flow fraction (Qe/(Qe + Qs))
Estimates of the mean and variance of effluent concen-
tration, effluent flow, and upstream concentration can
be made by regressing the natural log of each of these
variables against Z, the quantiles of a standard normal
random variable. A three-parameter lognormal method
can be used for streamflows since the two-parameter
distribution may produce a poor fit for the low-flow tail.
This problem is due to the fact that some streams have
a base flow that does not fluctuate appreciably.
Once the logarithmic means and standard deviations
of effluent flow and upstream flow are computed from
the regression least squares fit, the distribution of the
dilution factor (0) can be computed. Assuming that 0,
1-0, Ce, and Cs are lognormally distributed, the
products Ce0 and Cs(1-0) will also be lognormal.
Even though it is not true that the sum of two lognor-
mal variables is lognormal, this assumption must be
made to proceed with the method. The method of
moments also requires that effluent and upstream con-
centrations be uncorrelated with flow since no con-
venient moment approximations are available to
account for cross correlation. The numerical integra-
tion method does include a correlation coefficient and
this is the method used in the DYNTOX interactive pro-
gram. The arithmetic moments of Ce, Cs, and 0 can
be back calculated from the logarithmic values and
used to compute the arithmetic moments of the down-
stream concentration (Cj) as follows:
U(CT) = U (Ce) U (0) + U (C») [1-U(0)1
a2 (CT) = °2 (0) [u(Ce)l - u (Cs)]2
+ aMCe) [ff2 (0) + u
-------
The approaches for calculating the allowable effluent
concentration distribution using Monte Carlo simula-
tion are the same as those described for the continu-
ous simulation model. The assumption can be made
that treatment will only change the magnitude of con-
centrations, not their CV. The new required effluent
concentration distribution is assumed to be the same
as the existing distribution except that it is reduced in
magnitude by whichever is greater — the percentage
necessary for the one-day average concentrations to
meet the CMC or for the four-day average concentra-
tions to meet the CCC at the desired recurrence inter-
vals. If it is known that the coefficient of variation of
effluent concentrations will change after additional
treatment, iterative model runs will have to be per-
formed using different concentration means with the
appropriate "future treatment" variance until a mean
is found that meets the acute and chronic criteria at the
desired recurrence intervals. Whichever approach is
used in the WLA, the procedures described in Section
6 can be used to set permit limits. Information concern-
ing effluent concentration means and variability can be
obtained from databases on existing treatment plants
and from Effluent Guidelines documents.
EPA's MDSD has developed an interactive program that
will randomly select the sets of input data required for
Monte Carlo simulations. The program is included in the
DYNTOX portion of the ANNIE program available from
the EPA Modeling Center at the Athens, Georgia, ERL
[31]. The interactive program is written in FORTRAN
and is available for use on mainframe or IBM PC-
compatible computers.
Monte Carlo simulation has the following advantages:
• It can predict the frequency and duration of toxi-
cant concentrations in a receiving water.
• It can be used with steady state or continuous
simulation models that include fate processes for
specific toxicants.
• It can be used with steady state or continuous
simulation models that include transport
processes for river, lake, and estuary predictions.
• It can be used with steady state or continuous
simulation models that are designed for single and
multiple pollutant source analyses.
• It can estimate the uncertainty in model
predictions.
• It does not require timeseries data.
• It does not require model input data to follow a
specific statistical function.
• It can incorporate the cross-correlation and inter-
action of time-varying pH, flow, temperature, pol-
lutant discharges, and other parameters if the
analysis is developed separately for each season
and the results combined.
Monte Carlo simulation has the following disadvan-
tage: it requires more input data and calibration/verifi-
cation data than a steady state model.
Calculating the Return Period
Two common methods exist to calculate the return
period for a given concentration from probabilistic
modeling. Herein, for convenience, they are termed:
1) the percentile method and 2) the extrema method.
The percentile method uses a listing of all individual
daily concentrations and ranks them. The return period
for a concentration is then calculated based on the per-
centile occurrence. In the extrema method, only annual
extreme values are used in the ranking. The return
periods calculated from these two methods are equally
valid statistical representations. When using the per-
centile method, results express an average return
period and include multiple occurrences within any
year. On the other hand, the extrema method describes
the return period for an annual extreme and includes
only the extreme of multiple occurrences within any
year. The DYNTOX program uses the percentile method
to calculate return periods.
General Recommendations for Model
Selection
The reliability of the predictions from any of the model-
ing techniques depends on the adequacy of the data
used in the analyses. The minimum data required for
model input includes receiving water flow/volume,
effluent flow, effluent concentrations, and background
concentrations. In many locations, streamflow data
should be sufficient for both steady state and dynamic
river models. At least 30 years of flow data are required
if violations of the CMC and CCC must be evaluated at
rare recurrence intervals of once in 10 or 20 years.
Measurements of effluent toxicity or individual toxi-
cants can be much more limited. Table 5-2 indicates the
effect of sample size on the confidence with which a
modeler can assume that the largest and smallest
measurements bracket 75%, 90%, 95%, or 99% of
the toxic concentrations discharged from a pollutant
source. According to Table 5-2, which is based on order
statistics, 30 effluent toxicity measurements are
needed to provide 100% confidence that the range of
the sampling data covers 75% of the actual toxicity.
Twelve samples result in 84% confidence that 75% of
the actual toxicity is represented by the data. This table
can be used as a guideline to help determine how many
measurements should be made for a reasonable esti-
mate of effluent variability.
If only a few toxicant or effluent toxicity measurements
are available, Level 1 or 2 steady state assessments
should be used (See Table 5-3). Modeling should also
be limited to steady state procedures if a daily receiv-
ing water flow record is not available. If State regula-
tions do not require a specific design flow for river
WLAs, toxicologically based design flows can be used
to calculate the allowable effluent load. Appendix D
describes how to select these design flows. Fate and
transport models or dilution calculations can be used
for individual toxicants. At the present time, only dilu-
43
-------
Table 5-2. Confidence Associated With a Tolerance Limit State-
ment
n
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
25
30
40
50
60
70
80
90
100
P = .75
.16
.26
.37
.47
.56
.63
.70
.76
.80
.84
.87
.90
.92
.94
.95
.96
.97
.98
.99
1.00
—
_
_
—
_
—
—
P = .90
.03
.05
.08
.11
.15
.19
.23
.26
.30
.34
.38
.42
.45
.49
.52
.55
.58
.61
.73
.82
.92
.97
.99
.99
1.00-
—
—
P = .95
.01
.01
.02
.03
.04
.06
.07
.09
.10
.12
.14
.15
.17
.19
.21
.23
.25
.26
.36
.45
.60
.72
.81
.87
.91
.94
.96
P = .99
.00
.00
.00
.00
.00
.00
.00
.00
.01
.01
.01
.01
.01
.01
.01
.01
.02
.02
.03
.04
.06
.09
.12
.16
.19
.23
.26
Confidence y with which we may assert that 100 P percent of the popula-
tion lies between the largest and smallest of a random sample of n from
that population (continuous distribution assumed).
Adapted with permission from Annals of Mathematical Statistics, Vol.
29, No. 2., June 1958, pp. 599-601, from article entitled "Tables for
Obtaining Non-Parametric Tolerance Limits" by Paul N. Somerville.
Source: Natrella, M.G. Experimental Statistics. National Bureau of
Standards Handbook 91, U.S. Department of Commerce, August
1963.
Table 5-3. Exposure Assessments
Option Type of Model
Level 1 Steady state
Level 2 Steady state
Level 3 Continuous
simulation
Monte Carlo or
Lognormal
probabilistic
Comments
7Q10 or other State
required low flow
Toxicologically based
design flow
Batch process or
seasonal effluent
time series data
No batch process or
seasonal trends in
effluent data
Effluent data only
defined by statistical
parameters
tion calculations or first order decay equations are
recommended for effluent toxicity analyses. Section 1
discusses the conservative/additive assumption for
toxicity.
If adequate receiving water flow and effluent concen-
tration data are available to estimate their frequency
distributions, one of the non-steady state modeling
procedures should be used to develop more cost effec-
tive treatment requirements. If the effluent data exhibit
significant seasonal differences or batch process
trends, the Level 3 continuous simulation approach
may be the easiest dynamic modeling method to use.
The best results will, of course, be obtained if daily
effluent flows and concentrations are available for
model input for an entire year. The lag-one Markov tech-
nique can be used to generate daily effluent data for the
entire simulation period as long as adequate measure-
ments for the site specific plant (or a similar one) are
available to estimate a day-to-day correlation coeffi-
cient and to determine when seasonal or batch proc-
ess changes in effluent quality occur. The DYNTOX
program described in the continuous simulation sub-
section can be used for this Markov generation of data.
If adequate receiving water flow and effluent concen-
tration data are available and if effluent data exhibit no
seasonal or batch process trends, the Level 3 lognor-
mal and Monte Carlo methods may be easier and
require less computer time than the continuous simu-
lation approach. The DYNTOX program described in the
sections on these modeling techniques can be used to
perform the analyses.
Effluent Toxicity Modeling
In order to apply the steady state, continuous simula-
tion, or probabilistic methods to effluent toxicity model-
ing, the percent effluent measurements should be
converted to toxic units (TUs). A TU is the reciprocal
of the dilution that produces the test endpoint. An
acute toxic unit (TUa) is the reciprocal of the effluent
dilution that causes fifty percent of the organisms to
die by the end of the acute exposure period. A chronic
toxic unit (TUC) is the reciprocal of the dilution that
causes no unacceptable effect on the organisms by the
end of the chronic exposure period. The WLA must
ensure that the CMC and the CCC are met in the receiv-
ing water at the desired duration and frequency. The
CMC is recommended as one-third the LC50 value.
Thus the CMC for toxicity measured with acute tests
would be equal to the following:
CMC = 0.3 TUa
The CCC for toxicity measured with chronic tests is
recommended as the following:
CCC = 1.0 TUC
The first step in the WLA process is to calculate the
allowable acute effluent toxicity that meets the CMC
in the receiving water at the duration and frequency dis-
44
-------
cussed in Appendix D. If effluent toxicity measure-
ments for model input are based on acute toxicity tests
using fewer than three species, it is recommended the
CMC be divided by a factor of 10 to account for the
uncertainty in sensitivity (CMC = 0.03 TUa).
The next step in the WLA process is to calculate the
allowable chronic effluent toxicity that meets the CCC
in the receiving water at the duration and frequency dis-
cussed in Appendix D. It is recommended that the CCC
be divided by the species sensitivity factor of 10 if
measurements are based on tests of fewer than three
species (CCC = 0.1 TUC). In order to compare the
allowable acute toxicity value to the allowable chronic
toxicity value, the numbers must be converted to the
same unit using the following relationship: 1 TUa =
(ACR)(TUC), where the ACR is determined from tests
on the effluent. It is important that the ACR used for
WLA purposes be based on actual data and not as-
sumed to be 10 or 20 as in the screening procedure
(Section 3). The value of this ratio will influence
whether the acute or chronic WLA is more stringent
and is used to calculate the permit limit using Section
6 methods.
The following subsections recommend models for tox-
icity and individual toxicants for each type of receiv-
ing water — rivers, lakes, and estuaries. Detailed
guidelines on the use of fate and transport models for
individual toxicants are included in the WLA guidance
available from EPA's MDSD [39-41].
Specific Model Recommendations — Rivers
and Run-of-river Reservoirs
EPA's WLA guidance for modeling individual toxicants
in streams and rivers recommends the following steady
state models [39]: Simplified Lake/Stream Analysis
(SLSA), Michigan River Model (MICHRIV), Chemical
Transport and Analyses Program (CTAP), Exposure
Analysis Modeling System (EXAMS), and Metals
Exposure Analysis Modeling System (MEXAMS). All
these models except MEXAMS can simulate both
organic chemical and heavy metal fate and transport
in rivers. MEXAMS is designed to predict the ionic
speciation of metals only. All of these models are suit-
able for analysis of multiple reach, multiple pollutant
source discharges, except for SLSA, which simulates
a single reach and a single pollutant source. The DYN-
TOX program discussed in the dynamic modeling sub-
sections can be used to develop input data to run these
steady state fate and transport models with Monte
Carlo techniques to predict instream concentration-
frequency curves.
EPA's WLA guidance recommends the following con-
tinuous simulation models for individual toxicants in
streams and rivers: Estuary and Stream Quality Model
(WASTOX). Chemical Transport and Fate Model (TOX-
IWASP), Channel Transport Model (CHNTRISI), Finite
Element Transport Model (FETRA), Sediment Con-
taminant Transport Model (SERATRA), Transient One-
dimensional Degradation and Migration Model
(TODAM), and Hydrologic Simulation Program —
Fortran (HSPF). All of these models are designed for
multiple reach, multiple pollutant source analyses of
both organic chemicals and heavy metals. The DYIM-
TOX program discussed in the dynamic modeling sub-
sections can be used to develop the input data for these
continuous simulation fate and transport models to pre-
dict instream concentration-frequency curves. In addi-
tion the lognormal probabilistic dilution model, included
in DYNTOX can be used for single discharge dilution
analyses to predict concentration-frequency curves.
Several references that list reaction rates for the vari-
ous fate processes affecting organic chemical and
heavy metals are available [42-44]. At the present time,
the fate of effluent toxicity in a receiving water is not
understood. Even if a decay rate for toxicity can be
measured on a given day in a site-specific situation,
there is no way as yet to know how this rate is affected
by temperature, pH, or other environmental conditions.
There is also no way to know how this rate may change
when new treatment is installed. Instream measure-
ments of toxicity should be made at least once per sea-
son in order to identify any time-varying trends in
site-specific fate processes. These monitored decay
rates can then be used in steady state or continuous
simulation fate and transport models to predict receiv-
ing water toxicity, assuming that the rates will not
change with future treatment.
If instream toxicity measurements are not available,
effluent toxicity cannot really be modeled. ERA ERL-
Duluth has begun a three-year study of the persistence
of toxicity that should eventually improve the model-
ing of toxicity. As an interim measure, it is recom-
mended that effluent toxicity modeling be limited to
dilution calculations and assumptions that toxicity is
additive and conservative. Toxicity is expected to be
additive even when the toxicity of one effluent affects
invertebrates and the toxicity of a downstream dis-
charge affects fish or algae. For rivers and run-of-river
reservoirs with a detention time less than 20 days, the
following dilution equation should be used assuming
completely mixed conditions:
C =
CSQS
CeQe
CL
where C = downstream concentration (TUC or TUa)
Cs = upstream concentration (TUC or TUa)
Qs = upstream flow (cfs)
Ce = effluent concentration (TUC or TUa)
Qe = effluent flow (cfs)
For multiple dischargers, this equation must be applied
sequentially to find the concentration as a function of
distance downstream. The equation can be used for a
steady state analysis if Qs is set equal to the design
flow, Qe is set equal to the design plant flow, and Ce is
45
-------
calculated to meet the CMC and CCC. This equation
can also be used with the continuous simulation, log-
normal probabilistic, or Monte Carlo methods. For these
dynamic analyses, a series of Ce, Qe, Cs, and Qs values
is used.
If instream toxicity measurements are available and a
first order decay rate for toxicity can be estimated, the
following equation should be used:
C = C0e
K/—\
in
where C = downstream concentration (TUC or TUa)
C0 = concentration after the point source
discharge has mixed completely with
the river (TUC or TUa)
x = distance downstream of complete mix
point
u = velocity of river
K = measured decay rate
The DYNTOX programs, described in the subsections
on dynamic modeling techniques, can be used to per-
form non-steady state modeling of effluent toxicity in
rivers. The dilution calculation or first order decay equa-
tion can be used for continuous simulation and Monte
Carlo analyses; the lognormal probabilistic method can
be used for dilution calculations only.
Specific Model Recommendations — Lakes
and Reservoirs
EPA's WLA guidance for modeling individual toxicants
in lakes and reservoirs recommends the following
steady state models [40]: SLSA, CTAP, EXAMS, and
MEXAMS. As mentioned in the above section on river
modeling, all of these models except MEXAMS can
simulate the fate and transport of both organic chemi-
cals and heavy metals. Only SLSA is limited to single
reach and single pollutant source analyses, the other
models handle multiple reaches and sources. The DYN-
TOX program, discussed in the dynamic modeling sub-
sections, can be used to develop input data to run these
steady state fate and transport models with Monte
Carlo techniques to predict concentration-frequency
curves.
The following continuous simulation models for
individual toxicants in lakes and reservoirs are also
recommended: TOXIWASP, CHNTRN, WASTOX, SER-
ATRA, HSPF, and Toxic Organic Substance Transport
and Bioaccumulation Model (TOXIC). All these models
are designed for multiple segment, multiple pollutant
source analysis of both organic chemicals and heavy
metals. The DYNTOX program, discussed in the
dynamic modeling subsections, can be used to develop
the input data for these continuous simulation fate and
transport models to predict concentration-frequency
curves.
The same equations used for toxicity analyses in rivers
can also be used in a steady state, continuous simula-
tion, or probabilistic analysis of long, narrow, shallow
impoundments with high inflow velocities. Wider,
deeper lakes require more complicated analyses since
prolonged detention times (> 20 days) and stratifica-
tion exert a significant impact on water quality. The
prolonged detention times make it essential that receiv-
ing water measurements of toxicity be available to esti-
mate decay factors. These measurements should be
made at least once per season in order to identify any
time-varying trends in toxicity fate processes. Steady
state or continuous simulation fate and transport
models for lakes can then be run with monitored decay
rates for toxicity. Toxicity is expected to be additive
even when the toxicity of one effluent affects inver-
tebrates and the toxicity of another affects fish or algae.
A simple steady state analysis can be performed using
the following equations [45]:
Tw = V/Q
C = C,n/(1+TwK)
where Tw = mean hydraulic residence time
V = lake volume at critical conditions
Q = mean total inflow rate at critical
conditions
C = steady state lake concentration
(TUC or TUa)
Cin = steady state inflow concentration
(TUC or TUa)
K = first order decay rate
If effluent is discharged to a stratified lake and mixes
only with the hypolimnion or epilimnion, the volume of
that layer only should be used to calculate mean
hydraulic residence time (Tw). The mean total inflow
rate (Q) and the inflow concentration (Cin) should be
calculated as the sum of all sources to the lake, includ-
ing point source, nonpoint source, and tributary inputs.
Specific Model Recommendations —
Estuaries and Bays
EPA's WLA guidance for modeling individual toxicants
in estuaries recommends the following steady state
models [41]: EXAMS and MEXAMS. EXAMS is
designed for heavy metal and organic chemical ana-
lyses; MEXAMS predicts the fate of ionic species of
heavy metals. Both models handle multiple segments
and multiple pollutant sources. The DYNTOX program,
discussed in the dynamic modeling subsections, can
be used to develop input data to run these steady state
fate and transport models with Monte Carlo techniques
to predict concentration-frequency curves.
EPA's WLA guidance recommends the following con-
tinuous simulation models for individual toxicants in
estuaries: WASTOX, TOXIWASP, and FETRA. These
models are designed for multiple segment, multiple pol-
lutant source analyses of both organic chemicals and
heavy metals. The DYNTOX program, discussed in the
dynamic modeling subsections, can be used to develop
46
-------
input data for these continuous simulation fate and
transport models to predict concentration-frequency
curves.
Dilution calculations for effluent toxicity discharges to
an estuary are complicated by the oscillatory motion
of the tides and possible stratification of the estuary.
The prolonged detention times make it essential that
field measurements of toxicity be available to estimate
decay factors. These measurements should be made
at least once per season in order to identify any time-
varying trends in toxicity rate processes. Steady state
or continuous simulation fate and transport models for
estuaries can then be run with monitored decay rates
for toxicity. Toxicity is expected to be additive even
when the toxicity of one effluent affects invertebrates
and the toxicity of another affects fish or algae. A sim-
ple steady state analysis can be performed using the
following equations for a non-conservative pollutant
entering from the river at the head of an estuary [45]:
B, =
C, = C,-_,
M-1
.-kt
B,
where r, = exchange ratio for segment i as defined
by modified tidal prism method
t = flushing time
f, = fraction of freshwater in segment i
C| - non-conservative pollutant concentration
in segment i (TUC or TUa)
k = decay rate of pollutant
The following equations should be used for a non-
conservative pollutant entering along the side of an
estuary:
' fi in)
c, = C0 n
Q = c0 n
i = l...n
where C,
C0
f n 1 — 0 — r
'
(n)
n =
for segments
downstream
of outfall
for segments
upstream of
outfall
non-conservative pollutant mean con-
centration in segment i (TUC or TUa)
conservative constituent mean con-
centration in segment of discharge
exchange ratio for segment i as
defined by the modified tidal prism
method
number of segments away from out-
fall (n = 1 for segments adjacent to
outfall, n = 2 for segments next to
these, etc.)
fraction of freshwater in segment i
f0 = fraction of freshwater in segment of
discharge
S, = salinity in segment i
S0 = salinity in segment of discharge
k = decay rate
t = flushing time
The details on how to calculate exchange ratios and
flushing times for estuaries are included in Part 2 of
EPA's Water Quality Assessment Manual [45]. This
manual also describes how to perform these calcula-
tions for stratified estuaries using a two-dimensional
box model analysis.
Allocation Procedures For Waste Loads
WLAs for water quality-based permits must be per-
formed in accordance with EPA regulations [46,47].
Point sources, non-point sources and natural back-
ground sources all contribute to the total load of a pol-
lutant in a waterbody. The allowable total maximum
daily load (TMDL) is defined as the total such that any
additional loading would produce a violation of water
quality standards. Allocation of the TMDL must be
made considering technical, socioeconomic, institu-
tional, and political constraints. States have used vari-
ous allocation schemes, and some require that a
particular method be used. Table 5-4 shows 19 poten-
tial allocation strategies [48]. In all cases, the antidegra-
dation provisions and the other requirements of the
water quality standards regulations must be met [47].
The most commonly used allocation methods can be
summarized as: 1) equal percent removal, 2) equal
effluent concentrations, and 3) a hybrid method. The
equal percent removal approach has two versions —
the overall removal efficiencies of each pollutant source
are required to be equal or the incremental removal effi-
ciencies must be equal. The equal effluent concentra-
tion approach can also be applied in two versions —
equal final concentrations or equal incremental concen-
tration reductions. This method is similar to the equal
percent removal method if influent concentrations at
all sources are approximately the same. If one plant has
substantially higher influent levels, however, requiring
equal effluent concentrations will result in higher over-
all treatment levels than the equal percent removal
approach.
The third commonly used method of allocating waste
loads is a hybrid method in which the criteria for waste
reduction may not be the same for each point source.
One facility may be allowed to operate unchanged,
whereas another may be required to provide the entire
load reduction. More generally, a proportionality rule
may be assigned that requires the percent removal to
be proportional to the input loading to the treatment
plant. In such situations, larger sources would be
required to achieve higher overall removals.
Several different allocation schemes should be ana-
lyzed before a final strategy is selected. In selecting the
47
-------
Table 5-4. Waste Load Allocation Methods
1. Equal percent removal (equal percent treatment)
2. Equal effluent concentrations
3. Equal total mass discharge per day
4. Equal mass discharge per capita per day
5. Equal reduction of raw load (pounds per day)
6. Equal ambient mean annual quality (mg/l)
7. Equal cost per pound of pollutant removed
8. Equal treatment cost per unit of production
9. Equal mass discharged per unit of raw material used
10. Equal mass discharged per unit of production
1 la. Percent removal proportional to raw load per day
11b. Larger facilities to achieve higher removal rates
12. Percent removal proportional to community effective income
13a. Effluent charges (dollars per pound, etc.)
13b. Effluent charge above some load limit
14. Seasonal limits based on cost-effectiveness analysis
15. Minimum total treatment cost
16. BAT (industry) plus some level for municipal inputs
17. Divide assimilative capacity to require an "equal effort among
all dischargers"
18a. Municipal: treatment level proportional to plant size
18b. Industrial: equal percent between BPT and BAT, i.e..
Allowable = BPT — (BPT - BAT)
100
19. Industrial discharges given different treatment levels for dif-
ferent stream flows and seasons. For example, a plant might
not be allowed to discharge when stream flow is below a cer-
tain value. Above that value, but below another value, the
plant would be required to use a higher level of treatment
than BPT. Finally, when stream flow is above an upper value,
the plant would be required to treat to a level comparable to
BPT.
Source: Chadderton eta/(19).
final allocation method, the relationship between
uncertainty of prediction versus cost should be exam-
ined. Uncertainty can vary between allocation
schemes, as discussed by Chadderton et al. [48].
References
1. Versar, Inc. 1984. Draft Assessment of Interna-
tional Mixing Zone Policies, Avoidance/Attrac-
tion Characteristics, and Available Prediction
Techniques. Prepared for EPA Office of Pesti-
cides and Toxic Substances, Washington, D.C.
2. U.S. Environmental Protection Agency. 1983.
Water Quality Standards Handbook. Washing-
ton, D.C.
3. Hetling, L. J., and R. L. O'Connell. 1966. A Study
of Tidal Dispersion in the Potomac River. Water
Resources Research, Vol. 2.
4. Baily, T. E. Fluorescent-Tracer Studies of an Estu-
ary. 1966. Journal of the Water Pollution Control
Federation, Vol. 38.
5. Yotsukura, N., H. B. Fischer, and W. W. Sayre.
1970. Measurement of Mixing Characteristics
of the Missouri River Between Sioux City, Iowa,
and Plattsmouth, Nebraska. U.S. Geological Sur-
vey Water-Supply Paper 1899-G.
6. Fischer, H. B. 1968. Methods for Predicting Disper-
sion Coefficients in Natural Streams, with Appli-
cations to Lower Reaches of the Green and
Duwamish Rivers, Washington. U.S. Geological
Survey Professional Paper 582-A.
7. Kisiel, C. C. 1964. Dye Studies in the Ohio River.
Proceedings of Symposium on Diffusion in
Oceans and Fresh Waters, Columbia University.
8. Thruston, A. D. 1970. A Fluorometric Method for
the Determination of Lignin Sulfonates in Natu-
ral Water. Journal of the Water Pollution Control
Federation, Vol. 42.
9. Wilson, J. F. 1968. Fluorometric Procedures for
Dye Tracing. Techniques of Water Resources
.Investigations of the U.S. Geological Survey,
Book 3, Chapter A12.
10. Johnson, M. C. 1984. Fluorometric Techniques for
Tracing Reservoir Inflows. Instruction Report
E-84-1, U.S. Army Engineer Waterways Experi-
ment Station, Vicksburg, Mississippi.
11. Holley, E. R., and G. H. Jirka. 1985. Mixing in Rivers.
U.S. Army Engineer Waterways Experiment Sta-
tion, Vicksburg, Mississippi.
12. Jirka, G. H. 1982. Multiport Diffusers for Heat Dis-
posal — A Summary. Journal of the Hydraulics
Division, Proc. ASCE, Vol. 108.
13. Roberts, P. J. W. 1979. Line Plume and Ocean Out-
fall Dispersion. Journal of the Hydraulics Divi-
sion, Proc. ASCE, Vol. 105.
14. Ford, D. E., and L. S. Johnson. 1984. An Assess-
ment of Reservoir Mixing Processes. Technical
Report E-84, U.S. Army Engineer Waterways
Experiment Station, Vicksburg, Mississippi.
15. Fischer, H. B., E. J. List, R. C. Y. Koh, J. Imberger,
and N. H. Brooks. 1979. Mixing in Inland and
Coastal Waters. Academic Press.
16. Wright, S. J. 1977. Mean Behavior of Buoyant Jets
in a Crossflow. Journal of the Hydraulics Divi-
sion, Proc. ASCE, Vol. 103.
17. U.S. Environmental Protection Agency. 1984. Draft
Initial Mixing Characteristics of Municipal Ocean
Discharges. Corvallis, Oregon.
18. Roberts, P. J. B. 1979. A Mathematical Model of
Initial Dilution for Deepwater Ocean Outfalls.
Proceedings of ASCE Specialty Conference on
Conservation and Utilization of Water and
Energy Resources.
19. Winiarski, L. D., and W. E. Frick. 1976. Cooling
Tower Plume Model. EPA 600/3-76-100, Cor-
vallis, Oregon.
20. Winiarski, L. D., and W. E. Frick. 1978. Methods of
Improving Plume Models. Presentation at Cool-
48
-------
ing Tower Environ., University of Maryland, Col-
lege Park, Maryland.
21. Jirka, G., and H. L M. Fong. 1981. Vortex Dynamics
and Bifurcation of Buoyant Jets in Crossflow.
Journal of the Engineering Mechanics Division,
Proc. ASCE, Vol. 107.
22. Yotsukura, N., and W. W. Sayre. 1976. Transverse
Mixing in Natural Channels. Water Resources
Research, Vol. 12.
23. Pailey, P. P., and W. W. Sayre. 1978. Model for
Shore-Attached Thermal Plumes in Rivers. Jour-
nal of the Hydraulics Division, Proc. ASCE, Vol.
104.
24. Baumgartner, D. J., and D. S. Trent. 1970. Ocean
Outfall Design: Part I, Literature Review and The-
oretical Development. NTIS No. PB-203 749.
25. Baumgartner, D. J., D. S. Trent, and K. V. Byram.
1971. User's Guide and Documentation for Out-
fall Plume Model. NTIS No. PB 204-577/BA.
26. Prych, E. A. 1972. A Warm Water Effluent Ana-
lyzed as a Buoyant Surface Jet. Svergis
Meterologiska Och Hydrologiska Institut, Serie
Hydrologi, Stockholm, Sweden.
27. Shirazi, M. A., and L. R. Davis. 1974. Workbook of
Thermal Plume Prediction, Vol. 2: Surface Dis-
charge. EPA R2/72-005b.
28. Teeter, A. M., and D. J. Baumgartner. 1979. Predic-
tions of Initial Dilution for Municipal Ocean Dis-
charges. CERL Publ. No. 043, Corvallis, Oregon.
29. U.S. Environmental Protection Agency. 1982. Re
vised Section 301(h) Technical Support Docu-
ment. EPA 430/9-82-011, Washington, D.C.
30. Donigian, A., and W. Waggy. 1984. Simulation —
A Tool for Water Resource Management. Water
Resources Bulletin, Vol. 10 (2).
31. LimnoTech, Inc. 1985. Users Manual for DYNTOX
Portion of ANNIE. U.S. EPA, Washington, D.C.
32. DiToro, D. 1984. Probability Model of Stream Qual-
ity Due to Runoff. Journal of Environmental
Engineering, Proc. ASCE, Vol 110 (3).
33. Tiwari, J. L., and J. E. Hobbie. 1976. Random
Differential Equations as Models of Ecosystems:
Monte Carlo Simulation Approach. Math Bio-
science, Vol. 28.
34. Malone, R. R, D. S. Bowles, W. J. Grenney, and
M. P. Windham. 1979. Stochastic Analysis of
Water Quality. UWRL/Q-79/01, Utah State
University.
35. Hornberger, G. M., and R. C. Spear. 1980. Eutro-
phication in Peel Inlet-l. The Problem-Defining
Behavior and a Mathematical Model for the
Phosphorus Scenario. Water Research, Vol. 14.
36. Ford, D. E., K. W. Thornton, A. S. Lessem, and
J. L. Norton. 1981. A Water Quality Manage-
ment Model for Reservoirs. ASCE Proc. of Sym-
posium on Surface Water Impoundments.
37. Scavia, D., W. F. Powers, R. P. Canale, and J. L.
Moody. 1981. Comparison of First-order Error
Analysis and Monte Carlo Simulations in Time-
dependent Lake Eutrophication Models. Water
Resources Research, Vol. 17.
38. Thornton, K. W, D. E. Ford, and J. L. Norton. 1982.
Use of Monte Carlo Water Quality Simulations
to Evaluate Reservoir Operational and Manage-
ment Alternatives. Presented at Third Interna-
tional Conference on State-of-the-art Ecological
Modeling, Colorado State University.
39. Delos, C., W. Richardson, J. DePinto, R. Ambrose,
P. Rodgers, K. Rygwelski, J. St. John, W.
Shaughnessy, T. Faha, and W. Christie. 1984.
Technical Guidance Manual for Performing
Wasteload Allocations, Book II Streams and
Rivers, Chapter 3: Toxic Substances. EPA
440/4-84-022.
40. HydroQual, Inc. 1985. Draft Technical Guidance
Manual for Performing Wasteload Allocations,
Book IV Lakes, Reservoirs, and Impoundments,
Chapter 3 Toxic Substances.
41. Southerland, E., R. Wagner, and J. Metcalfe. 1984.
Draft Technical Guidance Manual for Performing
Wasteload Allocations, Book III Estuaries.
42. Callahan, M., M. Slimak, N. Gabel, I. May,
C. Fowler, J. Freed, P. Jennings, R. Durfee, F.
Whitmore, B. Maestri, W. Mabey, B. Holt, and C.
Gould. 1979. Water-related Fate of 129 Priority
Pollutants. Volumes I and II. EPA 440/4-79-029.
43. Mabey, W., J. Smith, R. Podoll, H. Johnson, T. Mill,
T. Chou, J. Gates, I. Partridge, and D. Vanden-
berg. 1981. Aquatic Fate Process Data for
Organic Priority Pollutants. EPA 440/4-81-014.
44. Lyman, W. 1982. Handbook of Chemical Property
Estimation Methods. McGraw-Hill, New York.
45. Mills, W. B., J. D. Dean, D. B. Porcella, S. A. Gherini,
R. J. M. Hudson, W. E. Frick, G. L. Rupp, and G.
L. Bowie. 1982. Water Quality Assessment: A
Screening Procedure for Toxic and Conventional
Pollutants. EPA 600/6-82-004b.
46. U.S. Environmental Protection Agency. 1985.
40 CFR Parts 35 and 130, Final Rule-Water
Quality Planning and Management. Federal Reg-
ister, Vol. 50, January 11.
47. U.S. Environmental Protection Agency. 1983.
40 CFR Part 131, Revised Water Quality Stan-
dards Regulation. Federal Register, Vol. 48,
November 8.
48. Chadderton, R. A., A. C. Miller, and A. J. McDon-
nell. 1981. Analyses of Waste Load Allocation
Procedures. Water Resources Bulletin, Vol. 17.
49
-------
6. Permit Requirements
Overview
At this point in the water quality control process, a
wasteload allocation (WLA) has been developed for a
discharger if violations of the State standards are
predicted or occuring. The WLA is based on ambient
criteria and the exposure of the resident aquatic com-
munity to toxic conditions. Where human health is of
concern, the WLA is based on safe levels and the route
of exposure. The next step is to derive permit limits to
enforce the WLA. In certain cases permit limits will be
different than the WLA values.
The WLA provides a measure of effluent quality that
is necessary to protect water quality in the receiving
water. It is important to consider how the WLA
addresses variability in effluent quality. For example, a
WLA for nutrients or bioaccumulative pollutants could
be expressed as the required average effluent quality
because the total loading of these pollutants is of con-
cern. On the other hand, a WLA for toxic pollutants
should be expressed as a maximum value for the
effluent because the concentration of these pollutants
is of more concern than the total loading. It is impor-
tant to recognize that the duration and frequency with
which the required effluent quality level is exceeded are
important aspects of a WLA.
Permit limits are designed to require a particular level
of effluent quality. Effluent quality is variable, and limits
are set at a level so that if the plant maintains the
desired level of performance, the probability of exceed-
ing the limits is very low (i.e., the probability is < 0.05).
If limits are set too high, a discharger not meeting the
desired level of performance would not exceed the
limits when typical monitoring frequencies are em-
ployed. If limits are set too low, a discharger meeting
the desired level of performance would often exceed
the limits. In either case, determination of compliance
and enforcement would be compromised.
There is a significant risk of incorrectly enforcing a WLA
if effluent variability and the probability basis for both
the WLA and the permit limits are not considered. For
example, a steady state WLA may specify an effluent
value with the assumption that it is a value never to be
exceeded. The same value used as the daily maximum
permit limit could allow the WLA value to be exceeded
perhaps an unacceptable amount of time without ob-
serving permit violations using typical monitoring re-
quirements. Even more confusion could result in
translating a longer duration WLA requirement (e.g., a
four-day average) into daily maximum and average daily
permit limits. Much of this section, therefore, concerns
ensuring that permit limits are derived to implement a
WLA requirement correctly.
This section describes the basic principles of permit
limit derivation and provides recommendations for
deriving limits from various types of WLA output. This
section also addresses other types of permit require-
ments and expression of limits.
Basic Principles of Effluent Variability
Effluent quality varies over time. If BOD data for a typi-
cal sewage treatment plant are plotted against time, the
day-to-day concentration variations can be seen (see
Figure 6-1a). Some of this behavior can be described by
constructing a frequency-concentration plot of the
same data (see Figure 6-1b). From this plot, we see that
190.0
OJ
1 135.0
£ 90.0
§
a. Concentration — Time Plot (days)
o
o
O
m
45.0
0.0
0.10
§ 0.05
-------
most of the time, BOD concentrations are near the aver-
age value. The amount of data at low concentrations
is limited by the maximum treatment efficiency. There
is much more data available at higher concentrations,
resulting from periods when treatment efficiency
drops.
From the shape of this curve, a standard statistical dis-
tribution can be fitted. The choices for statistical dis-
tributions include normal (bell shaped), lognormal
(skewed), or other variations on the lognormal distribu-
tion. From the vast bulk of data that has been exam-
ined in this manner, it is reasonable to make a first
assumption that treated effluent data follow a log-
normal distribution.
Consequently, any treatment system can be described
by using the mean concentration of the parameter of
interest (i.e., the long term average [LTA]) and the vari-
ance (or coefficient of variation [CV]) and by assum-
ing a particular statistical distribution (usually
lognormal). The advantage of describing effluent
characteristic behavior statistically is that the entire
distribution of values can be projected from limited
data, and limits can be set at a known probability of oc-
currence. Figure 6-2 illustrates a hypothetical distribu-
tion of daily data.
Permit limits are set at the upper bounds of accepta-
ble performance and are values not to be exceeded. Re-
quirements are usually expressed using two types of
permit limits. The daily maximum permit limit is the
maximum allowable value for any single observation.
The average daily or "monthly" permit limit is the max-
imum allowable value for the average of all observations
obtained during one month (average daily limits for
weekly periods are also used for publicly owned treat-
ment works [POTWs]). If permit limits are set too high
relative to the LTA, a discharger not complying with the
desired level of performance will not exceed the limits.
If permit limits are set too low, a discharger that is com-
plying with the desired level of performance may fre-
quently exceed the permit limits.
In setting the average daily (i.e., monthly) limit, the num-
ber of samples used to compute the average is a fac-
tor. As the number of observations (n) increases, the
probability distribution of the n-day average values be-
comes narrower around the LTA. Therefore, the 95th or
99th percentile value for an n-day average will ap-
proach the LTA as n increases.
The frequency of monitoring should determine the val-
ue of n for calculating the average daily (monthly) lim-
it. However, if limits are derived using a value for n that
is too low, the limit will be high and a discharger could
disguise a poorly operating plant by increasing sam-
pling during periods of good operation to lower the
reported average. Therefore, some intermediate basis
may be needed for average daily limits regardless of ac-
tual monitoring frequency. EPA recommends basing n
on the actual monitoring frequency where it is expect-
Long-Term Average
95th
99th
Figure 6-2. Hypothetical daily effluent parameter probability
distribution.
ed that monitoring frequency will not exceed the re-
quired frequency. For limits that are inexpensive to
monitor (such as metals) n should not be less than 10.
EPA has used n = 10 as the basis for average daily re-
quirements in many effluent limitations guidelines.
Permit Limit Derivation
Limits for any parameter can be derived for any plant
from the LTA and CV for the parameter. In many cases,
the LTA and CV values can be estimated. The same
procedures are used for technology-based and water
quality-based requirements. The actual pollution con-
trol requirements are dictated by the LTA and CV. Table
6-1 presents the general procedures for deriving daily
maximum limits and average daily limits. Technical
statistical details of these procedures may be found in
Appendix E and References 1 and 2.
The procedures in Table 6-1 for calculating the average
daily (monthly) limits are based on an assumption that
monitoring observations are independent of (uncor-
related with) each other. This is a reasonable assump-
tion for most industrial treatment plants. Some
treatment plants (particularly large biological systems)
exhibit serial correlation of monitoring observations. In
such cases it would be more accurate to use an esti-
mate of the variance of the n-day averages that incor-
porates this correlation. Such an estimate of the CV can
be generated from historical data where treatment will
not change significantly. For example, to derive an es-
timate for n = 8, construct a data set of eight-day aver-
ages that approximates the monitoring that will be used
for the monthly limit (e.g., every Tuesday and Friday).
Then derive the CV for those values. Calculation of the
average daily permit limit in this case would then be
identical to the procedures used to calculate the daily
maximum in Table 6-1 except that the CV for the n-day
averages would be used instead of the CV for daily
values.
Ensuring Consistency with the WLA
As previously stated, the WLA for a discharger will al-
ready have been completed. The permit writer should
be especially careful to ensure that the limits designed
to enforce the WLA are consistent with the assump-
52
-------
Table 6-1. Statistical derivation of permit limits.
Definition of Statistics
!,X2 ..... Xk daily pollutant measurements
E(X) = long-term average = (X., + X2 + . . . + Xk)/k
k _
V(X) = variance = 2 (X; -X)2/(k-1)
CV = coefficient of variation = >/V(X)/E(X)2
---------- Derivation of Limits
Given values for E(X) and CV proceed as follows:
1. Estimate o2as o2 = ln(CV2 + 1)
2. Estimate n as \a = In E(X) - .5o2
3. Calculate daily limits as eM + Zo
a. For 95th percentile limit Z = 1.645, therefore, daily limit 95 = eM + 1.645o
b. For 99th percentile limit Z = 2.326, therefore, daily limit 99 = eA* + 2.326o
4. Calculate o2 = ln[1 + «e°2-1)/n>]
where n = assumed observations per month
5. Calculate ^ = \JL + (o2- on2)/2
6. Calculate average daily (monthly) limit as e (Mn + Zon)
a. For 95th percentile limit Z = 1 .645, therefore, average daily limitgg = eMn + 1 .645on
b. For 99th percentile limit Z = 2.326, therefore, average daily limitge = eMn + 2.326on
Example
Assume E(X) = 4 ug/l and CV = 1.0 and interested in 10-day average (n = 10)
1.o2= ln(1.02+ 1) = .6931
2. n = In4-.5(.6931) = 1.0397
3a. Daily Iimitg5= exp[1.0397 + 1.646^.69311 = 11.1 ug/l
3b. Daily limitgg = exp[1.0397 + 2.326 V -6931 1 = 19.6 ug/l
4. o2,0= 1n [1 + He.6931 -1 )/10H = .0953
5. n10= 1.0397 + [(.6931 - .0953)/2] = 1.3386
6a. Average daily limitgg = exp[1.3386 + 1.645 V.0953] = 6.3 ug/l
6b. Average daily limitgg = exp[1.3386 + 2.326 V -0953] = 7.8 ug/l
----------- Derivation of Average Daily Limit Based on 30 Samples ----------
Given values for E(X) and CV proceed as follows:
For 30-day average limitgg calculate E(X) [1 + ' — (CV)]
For 30-day average limitgg calculate E(X) [1 + -^=E (CV)]
Example
Assume E(X) = 4 ug/l and CV = 1.0
30-day average Iimit95= 4 [1 + ^ (1.0)1 = 5.2 ug/l
30-day average Iimit99 = 4 [1 + (1.0)1 = 5.6 ug/l
V30
tions and requirements associated with the WLA. • Daily maximum and average daily (monthly) limits
Potential problem areas are described below. must be developed so that they are consistent with
. . each other and reflect the required level of per-
• A WLA requirement may be based on a particular formance
probability that the WLA limit will be exceeded. . A WLA va|ue specified as an average (e.g., a seven-
This probability may not be consistent with the consecutive-day average) must be translated into
probability used to set limits. daily maximum and average daily limits.
• A WLA that assumes the discharge is steady state
(i.e., not changing over time) requires an assump- There are several ways that a WLA may be developed
tion regarding how the effluent may vary. and expressed. Table 6-2 presents the major types of
53
-------
WLA output and options for deriving limits. Each of the
WLA output types is addressed below.
WLA Output Type 1 -
Required Effluent Performance
The least ambiguous way that a WLA can be specified
is as the required effluent performance in terms of the
LTA and CV of the daily values. When a WLA is ex-
pressed as such, there is no confusion about assump-
tions used and the translation to permit limits. A permit
engineer can easily design permit limitations to achieve
the WLA objectives.
The types of exposure analyses that yield a WLA in
terms of required performance are the continuous
simulation, Monte Carlo, and lognormal probabilistic
analyses. Although these analyses are slightly more
complex than a steady state analysis, they are preferred
because they specifically address effluent variability.
Interactive computer programs that are available to
facilitate use of these WLA models are described in
Section 5.
The permit limit derivation procedures described above
are used to derive daily maximum and average daily
(monthly) limits from the required effluent LTA and CV.
As described above, EPA generally uses 10-day aver-
ages to derive the daily average limit (regardless of ac-
tual monitoring frequency). Depending on the
monitoring that is expected to be done, other values of
n may be used in deriving the daily average limits on a
case-by-case basis.
WLA Output Type 2 -
Single Value from a Steady State Analysis
The vast majority of WLAs are reported as a single val-
ue for effluent quality. An example of such a require-
ment is "copper concentration must not exceed 0.75
milligrams per liter (mg/l)." Steady state analyses as-
sume that the effluent is constant and, therefore, that
the WLA value will never be exceeded. This presents
a real problem in deriving permit limits because permit
limits must reflect variability. A second problem is that
the effluent concentrations may be occasionally great-
er than the permit limits without registering limit viol-
ations because of the frequency of the monitoring
used.
The proper enforcement of this type of WLA depends
on the parameter limited. For nutrients, BOD, and bioac-
cumulative pollutants, the WLA value should general-
ly be used as the average daily permit limit. However,
the impact associated with toxic pollutants is much
more time dependent as reflected in the four-day aver-
age duration for the CCC (see Section 2). Therefore, the
WLA value for toxic pollutants should be used as the
daily maximum permit limit.
The disadvantage of this approach is that the average
daily limits must be derived without any information
about the variability of the effluent parameter. Without
other information, most permit engineers divide the
daily maximum limit by 1.5 or 2.0 to derive an average
daily limit (depending on the expected range of varia-
bility).
The second approach available when a WLA is derived
as a single value using a steady state assumption is to
use the methodology described in Reference 3. This
methodology is used to determine whether it is ap-
propriate to enforce the WLA result on a basis less strin-
gent than the daily limit. The methodology involves
redoing the WLA using the iognormal probability anal-
ysis to determine the frequency with which the CMC
(the one-day criterion) would be exceeded under three
separate scenarios for permit limits.
WLA Output Type 3 —
Steady State Values with a Specified
Duration or with a Specified
Permit Limit Probability Basis
WLAs may begin to be reported as four- or seven-day
average values that then would require translation to
permit limits. The "toxicologically based steady state"
WLA approach provides one-day and four-to-seven day
average requirements based on acute and chronic pro-
tection. Also, if a WLA value did have a specified prob-
ability of being exceeded, permit limits could then be
derived to enforce the requirement consistent with the
probabilities commonly used to set limits. The follow-
ing methodology was developed to accomplish these
objectives. The methodology consists of the following
general steps:
• Step 1 — An effluent performance level (LTA and
CV) that will meet the WLA requirement is back-
calculated. Where two requirements are specified
based on different duration periods, two perfor-
mance levels are back-calculated.
• Step 2 — Permit limits are then derived directly
from whichever performance level is more re-
strictive.
The benefits of this methodology are: 1) limits may be
adjusted relative to the fixed value WLA to account for
a different permit limit probability basis, 2) the daily
maximum and average daily limits can be made consist-
ent to maximize enforceability, and 3) WLA require-
ments specified as an average can be translated to
standard permit limits.
To adjust limits to correct for probability basis requires
that the probability basis for the WLA value must be
specified. Steady state WLAs usually mean the value
may never be exceeded. If "never" was interpreted to
mean no more than one day per year, for example, the
probability would be 0.003 (1 day/365 days). Making
an assumption about the probability that the WLA may
be exceeded allows this methodology to be used to cor-
rect for inconsistent probability bases.
54
-------
Table 6-2. Wasteload allocation (WLA) outputs and derivation of
limits.
WLA Output Type Derivation of Limits
1. Required effluent Derive maximum and average limits
performance directly
2. Single value from a steady Incorporate value as daily maximum
state analysis limit
Use reference [3] to assess
variability and the degree to which
the value can be incorporated as
an average daily limit
3. Steady state values with Derive required long-term average/
specified duration or variability estimates for each
specified exceedence requirement; select limiting require-
probability ment; derive limits from long-term
average/variability
To use the methodology only to translate a WLA speci-
fied as an average to permit limits (and not correct for
probability), merely use the same probability for the
WLA as that used for deriving limits (e.g., 0.05). The
methodology is described below in five steps.
Step 1 — Estimate Effluent Variability
The first step is to estimate the daily CV the parame-
ter will have after the plant complies with the require-
ments. If variability is mostly related to production,
current data may be used. If future variability will be
substantially different, the CV must be estimated. Ta-
ble 6-3 provides data on effluent parameter CVs. (The
variability estimates for sewage treatment reflect
secondary treatment not advanced secondary treat-
ment.) Discharges of toxic pollutants are generally more
variable than discharges of conventional pollutants. It
is important to use an estimate of the CV that is as low
as can be reasonably achieved. In this procedure, the
LTA is back-calculated using the CV and the WLA re-
quirement. Table 6-4 illustrates how the CV affects the
relationship between the LTA and various measures of
extreme performance.
Table 6-3. Typical coefficients of variation
Coefficient
Number of Variation
Treatment Process of Plants BOD TSS
Trickling filter rock 64 .40 .50
Trickling filter plastic 17 .50 .65
Conv act. sludge 66 .65 .85
Contact stabilization 57 .60 .70
Extended aeration 28 .70 .65
Rotating bio. contactors 27 .60 .70
Oxidation ditch 28 .60 .70
Stabilization pond 37 .50 .65
Pollutant Coefficient of Variation
Cr .99
Cu .60
Fe .57
Ni .81
Zn .84
Tss .66
Coefficient of Variation
Plant Number BOD (n) TSS (n)
12015 1.01 46 .85 195
12072 .97 392 .63 395
12026 .95 44 .49 53
12036 .74 366 1.12 364
12097 1.08 222 1.21 249
12098 1.37 24 1.52 25
12117 .70 39 .81 51
12160 .92 34 1.11 32
12161 .55 249 .99 355
12186 .71 54 .50 54
12187 .21 12 .26 12
12136 1.02 110 1.16 111
12248 .58 50 .55 52
12257 .64 56 .92 56
12294 .93 56 1.25 50
12307 1.55 39 1.34 38
1 From Table C-2, Reference [1].
2 From Table 3, page 14 of 10-18-83 memorandum from H. Kahn to E. Hall titled,
"Revisions to Data and Analysis of the Combined Metals Data Base."
3 From preliminary descriptive statistics generated on pharmaceutical data by SRI
International, 11-12-82.
is not necessarily the same as the probability basis for
Step 2 — Derive Performance for One-day
WLA Requirement
The second step is to determine an effluent distribu-
tion that will meet the one-day WLA requirement (here
assumed to be acute protection). Toxicity requirements
must be converted to toxic units (TUs) (100/toxicity in
% effluent) in order to use these equations. The WLA
will have been conducted assuming some probability
of occurrence of the WLA value. Where the WLA does
not specify an assumed effluent occurrence frequency,
use a required occurrence probability of 0.01 or 0.05
(i.e., the 99th or 95th percentile). As explained above,
the assumed probability that the WLA will be exceeded
permit limit calculation. One purpose of this method-
ology is to derive different values based on different
probability bases for the WLA and permit limits.
Using the acute protection effluent value (WLA), the
required probability of occurrence, and the CV, derive
the log mean using:
H = In (acute WLA) - Zu
where: acute WLA = the effluent acute protection
value, Z = the critical value statistic (1.645 for 0.05
probability and 2.326 for 0.01 probability), and
a = ^ln(CV2 + 1).
The LTA is derived from n and a using:
LTA = e"+ 5ff2
55
-------
Table 6-4. Relationship between coefficient of variation and
various permit limits for the same long-term average.
Permit Limits (ppb)
Daily
Maximum
Daily Avg. Daily Avg.
(10-day) (30
-------
When toxicity limits are used, additional description of
the limit is required. The limit should be stated in Part I
as "effluent toxicity" in the parameter column with
"maximum TUs," "minimum LC50," or "minimum
NOEL" in parenthesis underneath. The numerical
values should be placed in the appropriate concentra-
tion column followed by TU or a percent sign. A foot-
note should direct the reader to Part III, Additional
Conditions. If the monitoring requirements cannot be
fitted into the columns in Part I, a reference such as
"see Part III" should be used.
The further description in Part III should accomplish the
following:
• Explain how the limit is expressed (e.g., the limita-
tion is the minimum LC50 expressed as percent
effluent, or the limitation is the maximum acute
toxic units [TUa]).
• Specify the test organism and the test methods.
• Describe any special reporting or followup require-
ments (e.g., requirements to conduct a toxicity
reduction evaluation).
The language in Part III should be modified as needed
to suit the situation. The following language should be
used only as an example:
• The effluent toxicity limitation contained in Part I
is the allowable chronic toxicity to the most sen-
sitive of three test species. It is expressed as the
maximum NOEL in percent effluent. The required
test species and the procedures to follow are
described in Methods for Measuring the Chronic
Toxicity of Effluents to Aquatic Organisms,
EPA-600/4-85-014, August, 1985.
• The permittee shall conduct monitoring of effluent
toxicity once per month. One 24-hour composite
sample shall be collected and tested within 24
hours of collection. Results shall be reported as the
NOEL. Any test that does not meet quality control
requirements as described in the above referenced
methods shall be repeated using a freshly collect-
ed sample as soon as practicable.
• If the permittee does not comply with the effluent
toxicity limitation in Part I, the permittee shall sub-
mit, if requested by the Director, within 45 days,
a plan and schedule for conducting a toxicity
reduction evaluation. The toxicity reduction evalu-
ation, when completed, shall determine how the
permittee can achieve the effluent toxicity limita-
tion including an implementation schedule. After
review of the plan by EPA, the permittee shall con-
duct the evaluation within the specified time
frames. Upon completion of the toxicity reduction
evaluation, this permit may be modified, or alter-
natively revoked and reissued, in order to incor-
porate appropriate permit conditions and
compliance schedules.
Toxicity Reduction Evaluation and
Indicator Limits
One mechanism that can be used in bringing a dis-
charger into compliance with a difficult water quality-
based requirement is a toxicity reduction evaluation
(TRE). A TRE is a study conducted to determine what
control options are effective for complying with either
toxicity or chemical concentration requirements. In
most cases the plant manager should be responsible
for conducting a TRE. A TRE can be done prior to per-
mit issuance, during the permit term in response to a
monitoring trigger, during the permit term in response
to limits being exceeded, or in response to an adminis-
trative order.
The purpose of a TRE is threefold: 1) to isolate causa-
tive pollutants or manufacturing processes that pro-
duce the chemicals of interest; 2) to identify control
options and determine the effectiveness of each op-
tion; and 3) to identify a compliance monitoring indi-
cator and demonstrate its effectiveness. Of these three
purposes, only the second is essential to developing a
control plan for the facility. Toxicity may be used as a
control parameter without identifying causitive chem
icals.
The first step of a TRE is to try to determine what chem-
icals are causing toxicity or, for specific chemicals,
what manufacturing processes are producing the pol-
lutants. This can be accomplished by an analysis of
feedstocks, products, by-products, and treatment de-
sign and operation. Chemical analysis, in-plant toxici-
ty testing, and wastewater chemical fractionization
followed by toxicity or chemical analysis of the frac-
tions may also be useful to isolate causes. A fractiona-
tion procedure to isolate toxic constituents and
fractions has been developed [4]. The principal advan-
tage to isolating causative factors is that cost-effective
process modifications may be available.
The second and essential step is to determine how ef-
fective alternative control options are in reducing the
discharge of toxic pollutants. If specific chemicals are
involved, the effects of process modifications or treat-
ment changes can be estimated using chemical and
treatability data, or determined through bench-scale
studies. When effluent toxicity is involved, bench-scale
studies should be used to determine the reduction in
wastewater toxicity associated with different control
options. The draft 301 (g) guidance contains informa-
tion on treatability studies [5]. EPA is also developing
industry-specific studies of toxicity control [6-10].
The third step is to explore the use of an indicator lim-
it. An indicator limit is a limit on some pollutant or pa-
rameter that will ensure compliance with a separate
chemical or toxicity requirement. The burden of proof
for establishing indicator limits rests with the discharg-
er. The benefits of using an indicator for compliance
monitoring are lesser cost for the permittee and great-
57
-------
er frequency of monitoring and consequently better en-
forcement for the regulatory agency.
The utility of an indicator limit must be demonstrated
by the permittee. Such a demonstration may be based
on: 1) similarities between causative pollutants and a
convenient monitoring parameter (e.g., oil and grease
limits to ensure control of polynuclear aromatic hydro-
carbons) or 2) the effectiveness of a particular treat-
ment process and a control parameter for that process
(e.g. limits on total organic carbon [TOC] to ensure prop-
er performance of an activated carbon process). Where
an indicator limit is used, periodic monitoring of toxicity
to test the validity of the indicator limit should be
required.
Other Permit Requirements
Operational Controls
The most effective control of toxic pollutant discharges
may result from operational requirements. For example,
a TRE or other sources may reveal that spills during ma-
terial handling are causing discharges of toxic chemi-
cals. Requirements for spill reduction, containment,
and treatment may eliminate the discharge. Such re-
quirements may be instituted as Best Management
Practices (BMP) requirements under sections 402(a)
and 304(e) of the Clean Water Act (see 40 CFR Part
125 Subpart K). Additional technical information is con-
tained in a publication entitled NPDES Best Manage-
ment Practices Guidance Document [11].
Anti-degradation and Anti-backsliding
The procedures in this document for developing water
quality-based permit limits for toxics at a source will
normally result in new or more stringent water quality-
based limits than those contained in a previously issued
permit. In a limited number of cases, however, it is con-
ceivable that less stringent water quality-based limits
could result. In these cases, permit limits must conform
to existing Federal regulations governing both anti-
degradation (existing instream water uses shall be
maintained and protected) and anti-backsliding (issu-
ance of permit limits that are less stringent than those
contained in the existing permit is prohibited). The per-
tinent regulations are 40 CFR Part 131 (see especially
§131.12) and 40 CFR Part 122 (see especially §122.44[l]
and 122.62). Anti-backsliding is frequently the subject
of debate in State legislatures and in Congress. Permit
writers should keep apprised of recent statutory or
regulatory developments in this area.
Data Collection
Section 308 of the Clean Water Act and correspond-
ing State statutes authorize the imposition of monitor-
ing and data collection requirements on any owner or
operator of a point source discharge. The only limita-
tion on this authority is a test of reasonableness. There
must be a reasonable need for the information, and the
schedule and costs of the requirements must be
reasonable. Regulatory agencies have been granted
broad discretion in defining what is reasonable.
Requirements to conduct biological assessments, tox-
icity reduction evaluations, and in-plant monitoring are
all authorized. In a recent challenge to certain aspects
of 308 authority, the court firmly upheld regulatory
agencies' authority to require in-plant testing of waste-
streams and manufacturing processes (see Mobil Oil
Corporation vs. EPA, 716F.2d 1187, 7th Cir. 1983).
Requirements to collect data may be implemented
through a direct request (commonly referred to as a
"308 letter"), permit reporting requirements, or an
administrative order. If sufficient data can be collected
prior to permit issuance, a 308 letter is the best means
to obtain the data. An example 308 letter is presented
as Figure 6-3.
Permit requirements to collect data should be used
when longer term data (e.g., for several seasons) are
needed and there are insufficient data to develop and
include water quality-based limitations in the newly is-
sued permit. The permit should include a statement
that the permit may be modified or revoked and reis-
sued if the data indicate violation of State water quali-
ty standards. Several agencies have experienced
problems in negotiating study plans based on a gener-
alized permit requirement. If the permit requirement is
non-specific (such as requiring the development and
execution of a plan of study), minimum requirements
should be included.
58
-------
References
1. U.S. Environmental Protection Agency. 1979.
Development Document for Existing Source
Pretreatment Standards for the Electroplating
Point Source Category, Appendix XII. EPA
440/1-79/003.
2. U.S. Environmental Protection Agency. 1982.
A Statistical Analysis of the Combined Metals
Industries Effluent Data, Appendix I. Report sub-
mitted by JRB Associates under EPA Contract
68-01-6347. Office of Water Regulations and
Standards (WH-586), Washington, D.C.
3. U.S. Environmental Protection Agency. 1984.
Technical Guidance Manual for Performing
Wasteload Allocations, Book VII, Permit Averag-
ing Periods. Office of Water Regulations and
Standards (WH-553), Washington, D.C.
4. Walsh, G. E., and R.L. Games. 1983. Determina-
tion of Bioavailability of Chemical Fractions of
Liquid Wastes Using Freshwater and Saltwater
Algae and Crustaceans. Env. Sci. Tech., Vol. 17,
pp. 180-182.
5. U.S. Environmental Protection Agency. Draft Tech-
nical Guidance Manual for Regulations Promul-
gated Pursuant to Section 301 (g) of the Clean
Water Act. Appendix C —Bench Scale Treat-
ability. Available from Permits Division (EN-336),
Washington, D.C.
6. U.S. Environmental Protection Agency. 1983.
Toxicity of Leather Tanning and Finishing
Wastewaters. Office of Water Enforcement and
Permits (EN-336), Washington, D.C.
7. U.S. Environmental Protection Agency. 1983.
Toxicity Reduction Manual for the Textiles
Manufacturing Industry. Office of Water En-
forcement and Permits (EN-336), Washington,
D.C.
8. U.S. Environmental Protection Agency. (In press).
Toxicity of Iron and Steel Wastewaters. IERL, Re-
search Triangle Park, NC.
9. U.S. Environmental Protection Agency. (In press).
Toxicity of Organic Chemical Industry
Wastewaters. IERL, Cincinnati, OH.
10. U.S. Environmental Protection Agency. (In press).
Toxicity Reduction of Pulp and Paper Mill
Wastewater. IERL, Cincinnati, OH.
11. U.S. Environmental Protection Agency. 1983.
NPDES Best Management Practices Guidance
Document. Office of Water Enforcement and
Permits (EN-336), Washington, D.C.
59
-------
Figure 6-3.
Example 308 letter.
UNITED STATES ENVIRONMENTAL PROTECTION AGENCY
WASHINGTON. D.C. 20460
CERTIFIED MAIL NO.
RETURN RECEIPT REQUESTED
Ms. Ann Powell
Plant Manager
Chemico Corporation
Anytown, USA
Dear Ms. Powell:
RE: NPDES No. XX000123
Chemico Anytown
The U. S. Environmental Protection Agency (EPA) is preparing
to reissue the NPDES permit for the Chemico Anytown facility.
Preliminary analyses conducted by EPA and water quality monitoring
conducted by the [State Water Control Authority] indicate that the
discharges from the Chemico Anytown facility may be violating
State water quality standards. In order for EPA to fulfill its
responsibilities under the Clean Water Act, 33 U.S.C. 1251, et seq. ,
additional information regarding the nature of the discharges from
the Chemico facility is required.
Therefore, you are hereby required (1) to perform the sampling
and analysis programs described below, (2) to maintain for possible
use by EPA all records regarding plant operations during the sampling
period, and (3) to submit reports resulting from the sampling and
analysis programs within the time limits specified. These require-
ments are imposed pursuant to the authority provided in Section
308 of the Clean Water Act, 33 U.S.C. Section 1318. Failure to
comply with this request may result in enforcement proceedings
under Section 309 of the Clean Water act, 33 U.S.C. Section 1319,
which could result in the judicial imposition of civil or criminal
penalties. Please refer to the attachment to this letter for a
summary of our authority and your rights to confidential treatment
of certain information.
Effluent Toxicity Characterization Study -
In order to characterize the magnitude and variability of the
toxicity of the discharge from Outfall 001, Chemico shall determine
the wastewater acute toxicity every two weeks for a period of four
months commencing within 30 days of receipt of this letter. Waste-
water samples composited over 24 hours shall be tested using at a
minimum the following test organisms: pimephales promelas (fathead
minnow) from one hour to 30 days old and Daphnia spp. (water flea)
from zero to 24 hours old. Results shall be expressed as the 48-hr
LC5 0 with the 95 percent confidence interval. Dilution water used
in the analysis may be either receiving water collected immediately
upstream of the discharge or another source of clean water such as
the water used for culturing organisms. However, the analysis of
at least one sample shall be replicated for each test organism
using the other source of dilution water.
Toxicity analyses shall be conducted in accordance with
Methods for Measuring the Acute Toxicity of Effluents to^ Aquatic
Organisms EPA-600/4-78-012, Revised July 1978.Each sample shall
also be analyzed for the parameters currently limited in the
NPDES permit for outfall 001 and the following parameters:
chlorine, cyanide, total dissolved solids, chloride, and ammonia.
Associated production levels, ancillary activities, and the
performance of the treatment systems shall be recorded for each
wastewater sampling.
The results of the toxicity analysis together with the
required chemical analyses and plant operations information shall
be provided for the first two samples within 45 days of initiation
of: the sampling program and for subsequent samples every 30 days
thereafter.
Bioaccumulative Pollutant Characterization Study -
In order to identify potentially bioaccumulative pollutants
discharged from Outfall 001, Chemico shall analyze effluent
samples as described below. Wastewater samples for analysis
shall be composited over 24 hours once per month for a period of
four months commencing within 30 days of receipt of this letter.
High performance liquid chromatography techniques shall be
used to separate the sample constituants. Methods shall conform
to EPA draft method CG-1410 partition Coefficient (n-octanol/
water) Estimation by Liquid Chromatography Tesj: Guideline.
USEPA.Office of Toxic Substances. Washington DC (copy attached).
All chemicals corresponding to log Kow greater than 4 shall be
collected and analyzed using gas chromatography/mass spectroscopy.
The identification and quantification of pollutants shall conform
to EPA proposed method 625 (44 FR 69540). Alternate methods may
be substituted with prior approval by EPA. In addition to
measuring the priority pollutants, a reasonable effort shall be
made to identify and quantify no fewer than the 20 largest non-
priority pollutant peaks on the total ion plot (reconstructed
chromatogram) except that peaks less than 10 times the peak to
peak background noise need not be identified.
60
-------
Figure 6-3 (cont.)
Identification of the non-priority pollutant peaks shall be
attempted by reference to the EPA/NIH computerized library of
mass spectra, with visual confirmation by an experienced analyst;
quantification may be an order of magnitude estimate based upon
the response of an internal standard. If more than 20 peaks
exceed 10 times the background noise, those not identified shall
be reported by relative retention time and magnitude based upon
the response of an internal standard.
'ith the total ion
i— ~_—J-*td within
The results of the analysis together with the tot,
Lots and description of the analysis shall be provide-
> days of each sample collection.
If vou have anv Questions olease contact Kermit R
Pi
45
If you have any questions please contact Kermit Riter at
213-456-7890. Thank you for your cooperation.
Sincerely yours.
Director,
Water Division
Attachment
cc: [state contacts]
AUTHORITY AND CONFIDENTIALITY PROVISIONS
This request for information is made under authority provided by Section 306
of the Clean water Act, 33 U.S.C. 1318. Section 308 provides that: "whenever
required to carry out the objective of this Act, ...the Administrator shall
require the owner or operator of any point source to (i) establish and
maintain such records, ( ii ) make such reports, (in) install, use, and
maintain such monitoring equipment and methods (including where appropriate,
biological monitoring methods), ( iv) sample such effluents. .. and (v) provide
such other information as he may reasonably require; and the Administrator or
his authorized representative, upon presentation of his credentials, shall
have a right of entry to... any premises in which an effluent source is located
or in which any records. ..are located, and may at reasonable times have access
to and copy any records. . .and sample any effluents....1
Please be advised that the submission of false statements may subject you to
federal prosecution under 18 U.S.C. 1001 and that this or any other failure to
comply with the requirements of Section 308 as requested by U.S. EPA may
result in enforcement action under the authority of Section 309 of the Clean
Hater Act, which provides for specified civil and/or criminal penalties.
Confidentiality
U.S. EPA regulations concerning confidentiality and treatment of business
information are contained in 40 CPR Part 2, subpart B. Information may not be
withheld from the Administrator or his authorized representative because you
view it as confidential. However, when requested to do so, the Administrator
is required to consider information to be confidential and to treat it
accordingly, if disclosure would divulge methods or processes entitled to
protection as trade secrets (33 U.S.C. 1318(b) and 18 U.S.C. 1905), except
that effluent data (as defined in 40 CPR 2.302
-------
7. Compliance Monitoring
Overview
Section 308 of the Clean Water Act and equivalent
State statutes authorize the imposition of monitoring
requirements "whenever required to carry out the ob-
jectives of this Act" and require the permittee to grant
access to treatment facilities and records for inspec-
tions by regulatory authorities and their designated
agents. Whenever the permit limits the value of a dis-
charge parameter, the permit will also require self-
monitoring of that parameter. The permit may also in-
clude monitoring of parameters that are not limited.
Regulatory authorities conduct inspections of permit-
ted facilities to verify compliance status.
The permittee is responsible for conducting monitor-
ing, maintaining records, and submitting reports. In ad-
dition, the permittee is required to collect samples that
are representative of their discharges and follow speci-
fied analytical methods. Failure to comply with any of
these requirements is a violation of the Clean Water
Act.
Self-monitoring Requirements
The permit must prescribe the self-monitoring proce-
dures, frequency of analysis, and the locations and
procedures for sample collection. Permit conditions re-
quire that analytical methods referenced in
40 CFR 136 be used for analysis. When methods are
not included in Part 136, the permit must specify the
methodology to be followed. Most self-monitoring data
are reported on a Discharge Monitoring Report (DMR).
Permittees are required to maintain all records associat-
ed with monitoring for a period of three years.
The frequency of monitoring is the most important fac-
tor in determining compliance with the permit limita-
tions. As explained in the previous section, permit limits
are set quite high relative to the required LTA perfor-
mance. Therefore, sampling and analysis must be suffi-
ciently frequent to detect violations of permit
limitations at a non-complying facility. On the other
hand, monitoring can be costly — particularly for tox-
ic materials. The amount of monitoring that should be
required is a function of how much variability is ex-
pected for the pollutant parameter, the compliance his-
tory of the facility, and the potential for water quality
impacts due to non-compliance.
As a general guideline, if the value of a parameter is
limited by the permit, the monitoring frequency should
be at least once per month. For most toxic pollutants
and effluent toxicity, monitoring once per week or more
frequently may be needed because water quality im-
pacts may result from non-compliance. Examples of
monitoring frequencies are discussed in Section 8.
The total cost of toxic pollutant monitoring could be
reduced by requiring a high initial monitoring frequency
that can be decreased if the permittee consistently
meets the limit. The overall purpose of such require-
ments is to establish a compliance history with a higher
monitoring frequency during an initial period and to re-
duce the monitoring frequency for facilities that rou-
tinely comply with permit conditions. The requirements
should be pre-specified in the permit and should take
into consideration the basis of the permit limitations.
For example, if weekly monitoring shows the permittee
complies with the limitation, monitoring frequency may
be reduced to once per month. Another approach is to
concentrate monitoring frequency during critical peri-
ods to protect water quality.
Quality assurance (QA) is an important element of self-
monitoring. If the referenced testing methodology in-
cludes QA procedures, the permit should require com-
pliance with the procedures. If the methodology
contains inadequate QA procedures, the permit should
reference and require other QA procedures or set forth
QA procedures in the permit. When QA standards are
not met, the analysis should be repeated. EPA and
many States also maintain laboratory QA programs.
These programs can include laboratory inspections,
"blind" sample analysis, and laboratory certification.
Inspections
Inspections are conducted to verify permittee compli-
ance status determined from self-monitoring data.
Compliance inspections may include reviewing records,
inspecting treatment facilities, evaluating laboratory fa-
cilities and performance, and collecting samples for
analysis. EPA has defined several types of inspections
based on the tasks that are included [1]. The inspec-
tions that focus on toxics control, Compliance Sam-
pling Inspections for toxics, and Compliance
Biomonitoring Inspections, can also provide useful in-
formation for water quality assessment. Therefore, the
regulatory authority should consider coordinating these
inspections with water quality assessments and per-
63
-------
mit reissuance activities. Procedures for inspecting tox-
icity testing facilities and measuring effluent toxicity
are detailed in the Compliance Biomonitoring Inspec-
tion Manual [2].
Regulatory authorities should also consider requiring
the use of ambient toxicity testing, described in Appen-
dix C, or other ambient testing as a monitoring tool to
determine if the receiving water is being protected, i.e.,
the designated use is maintained or restored.
Enforcement
Regardless of the statistical basis used to develop a per-
mit limit, exceeding the limit is a violation. The permit-
tee must advise the regulatory agency of any instance
in which a permit limit is exceeded in a written non-
compliance report within five days after the violation
is detected. This report must include a description of
the non-compliance and its cause, the period of non-
compliance, and if the non-compliance has not been
corrected, the anticipated time it is expected to con-
tinue, and steps taken to reduce, eliminate, and prevent
recurrence of the non-compliance. New regulations for
reporting instances of non-compliance, including those
related to water quality-based controls, have been pro-
mulgated (50 FR 34648, Monday, August 26, 1985).
If the permittee suspects analytical problems may have
contributed to the violation, the permittee may perform
additional analyses. The results of any additional ana-
lyses must also be reported. The regulatory agency
should review the non-compliance report, request addi-
tional information from the permittee if it is needed, and
review other relevant compliance data when an
instance of non-compliance is detected. The regulatory
agency decision to initiate enforcement actions, con-
duct inspections, establish a compliance schedule, or
continue review of the permittee's self-monitoring
should be made consistent with the principles of EPA's
Enforcement Response Guide [3].
In some cases where the regulatory agency believes
that a permittee's discharge may be causing an adverse
water quality impact, a 308 letter can be used to ob-
tain additional information and monitoring from the per-
mittee. If such information shows that a permittee has
failed to meet permit conditions or effluent limits but
the type of remedy needed is unclear, the regulatory
agency may establish a compliance schedule for fur-
ther testing, evaluation of a remedy, and implementa-
tion of additional treatment. This compliance schedule
could be implemented by an administrative order or
consent decree. For example, a compliance schedule
may require that the permittee conduct biological and
chemical analysis several times during the year before
a permit is reissued. The compliance schedule should
identify triggers or thresholds that would require fur-
ther monitoring or corrective actions, define or refer-
ence the types of analysis required, and establish
criteria that would indicate whether further controls
and limits were needed. The schedule should also in-
clude milestones for reporting and providing addition-
al toxic controls if they are needed.
References
1. U.S. Environmental Protection Agency. 1984.
NPDES Compliance Inspection Manual. OWEP.
June NTIS # PB85-115897.
2. U.S. Environmental Protection Agency. 1979.
Interim NPDES Compliance Biomonitoring
Inspection Manual. MCD-62. Office of Water
Enforcement (EN-338), Washington, D.C.
3. U.S. Environmental Protection Agency. 1985.
Enforcement Management System. Office of
Water Enforcement and Permits, Washington,
D.C.
64
-------
8. Case Example
In this section, an example of a water quality-based dis-
charge situation is described to illustrate the use of the
procedures detailed in the previous sections. All of the
recommendations and options are not incorporated in
the example. Although the site is fictional, much of the
data have been taken from previously conducted
studies.
River Mile 13.8 Bibury STP
Treatment: Activated sludge, Phos-
phorus removal
Discharge Flow: 2.3 MGD average
(1 MGD industrial input)
Dilution at 7Q10 (assuming com-
plete mixing): 14%
Site Description
The Linville River is a small to medium sized river flow-
ing through an area of mixed land use. The river is 60
miles long and its banks are typified by small towns
separated by wooded areas. The 7Q10 of the river
above the first discharger is 10.7 million gallons per day
(MGD) (16.6 cfs). From its monitoring programs, the
State had indications that the water quality of the mid-
dle section was somewhat degraded. The dischargers
located in this section (starting from the farthest
upstream) are Cibesi Corporation, a metal finishing
plant and foundry, Nicklaus By-Products Company, a
coal tar extractor producing a variety of organic sol-
vents, and the town of Bibury sewage treatment plant
(STP). All other sources in the area discharge into the
STP. NPDES permits for both industrial sources (which
the State permit writers had previously issued with
Best Professional Judgment [BPJ] Best Available Tech-
nology [BAT] permit limits) had expired, and the STP
permit was coming up for renewal.
The location and a description of each discharger is
provided below:
River Mile 22.1
River Mile 21.4
Cibesi Corporation
Manufacturing (SIC 3362 + 3325
Foundry; 3471 Electroplating)
Treatment: equalization; neutraliza-
tion; settling.
Discharge Flow: 0.5 MGD process
wastes
Dilution at 7Q10 (assuming com-
plete mixing): 4.5%
Nicklaus By-Products
Coal Tar Extractor (SIC 2865)
Treatment: aerated lagoon and
chlorination
Discharge Flow: 3.0 MGD process
and sanitary wastes
Dilution at 7Q10 (assuming com-
plete mixing): 21%
Criteria Determination (Section 2)
Linville River is classified for the propagation of aquatic
life and is used for a variety of recreational activities.
The State reviewed available data to identify pollutants
that may be causing water quality impacts. The follow-
ing pollutants were identified: cyanide (CN), phenols,
copper (Cu), cadmium (Cd), nickel (Ni), chromium
(Cr+6), and zinc (Zn). Assorted organic priority pollu-
tants, including an array of chlorinated phenolic com-
pounds, had been measured at Nicklaus By-Products
on several occasions. However, presence and concen-
trations were extremely variable.
The State had conducted a trial evaluation of site-
specific criteria modification and had adopted the EPA
criteria for copper and cyanide as State standards. No
other toxicants were added to the State's numerical
water quality standards so EPA criteria were also used
for the other toxicants in this study.
Criteria Concentrations*
(at 100 mg/l hardness)
Pollutant
CN
Cu
Cr+6
Cd
Ni
Zn
Phenols
CMC (pg/l)
22.0
16.0
11.0
4.5
1,800.0
320.0
CCC (/tg/l)
4.2
11.0
7.2
4.5
96.0
47.0
* These numbers are based on previous EPA criteria values.
EPA criteria are available only for specific phenolic com-
pounds, so the State did not have a recommended
value for the total phenols measured in this study.
The State applied the numerical criteria described in
Section 2. Thus, each discharger must meet the criteria
of 0.3 acute toxic units (TUa) within the mixing zone
65
-------
and 1.0 chronic toxic units (TUC) at the edge of the
mixing zone.
The State also examined the question of duration and
frequency criteria for the site. State water quality stan-
dards are one-number chronic values. State regulations
specify that the standards apply only at flows equal to
or greater than the 7Q10. EPA recommended that two
wasteload allocations (WLAs) be performed; one for
the acute criterion and one for the chronic criterion.
Since one-day duration is more appropriate for acute
criteria than seven days, the State decided to use a
1Q10 for the acute WLA and the state-specified 7Q10
for the chronic WLA.
Effluent Characterization (Section 3)
Chemical Data Available (in micrograms/liter [/tg/l])
Cibesi Corporation Nicklaus By-Products
Pollutant
CN total
Cu
Cr+6
Cd
Ni
Zn
Phenol
(4AAP)
DMR*
530.0
1130.0
290.0
210.0
1130.0
620.0
—
Form 2C
2300.0
2130.0
215.0
660.0
3130.0
2560.0
—
DMR*
210.0
—
—
—
—
1 200.0
Form 2C
420.0
—
_
—
—
6120.0
* Two-year average of reported monthly averages from the DMRs.
Chemical screening using simple dilution calculations
showed that the pollutants present in two of the three
effluents could be exceeding water quality criteria
values at various times after mixing. Chemical data for
the Bibury STP showed only low levels of metals. Due
to the complex nature of the effluents involved and the
lack of adequate effluent toxicity data, the regulatory
authority concluded that effluent toxicity analysis was
necessary. For Cibesi, a mixture of both cyanide and
five metals at potentially toxic levels suggested a
whole-effluent toxicity assessment. An analysis of
instream concentrations of copper and cyanide also
seemed appropriate since the State had numerical
standards for these toxicants and since they were pres-
ent at very high levels relative to their effect levels. For
Nicklaus By-Products, the presence of high levels of
phenols (for which no chemical-specific characteriza-
tion is feasible) suggests whole-effluent toxicity analy-
sis is the most efficient way to assess toxic impact. For
the STP, low dilution capacity also suggested toxicity
testing even though available chemical data had not
indicated any specific problems.
Initial Permit Limits
The State calculated initial toxicity-based permit limits
for Cibesi and Nicklaus By-Products using the equation:
allowable effluent toxicity < criterion x dilution factor
For Cibesi, the initial limits were 4.8 TUa and 19.3
TUC. For Nicklaus, the initial limits were 0.4 TUa and
1.5 TUC. Derivation of these limits is explained in the
following subsection entitled Effluent Toxicity
Wasteload Allocation. Both companies objected to the
establishment of toxicity-based permit limits, arguing
that they were meeting their BAT permit limits, they
were being given water quality limits for some specific
chemicals, and no toxic impact caused by their dis-
charges had been proved. The State responded that
the nature of their discharges, particularly those of
Nicklaus, made toxicity assessment mandatory due to
the low dilution available, the number of toxicants
known to be present in their discharges, and the legal
requirement to enforce the "no toxics in toxic
amounts" criterion in the State standards.
After reviewing the situation, each company agreed to
conduct toxicity testing. Specific testing procedures
were determined as follows.
Tier 1 — Toxicity Screening (Section 3)
The regulatory authority determined on the basis of
dilution alone that a strong potential exists for water
quality standards violation (toxic impact) for each dis-
charge. Accordingly, the State's normal toxicity
screening procedures were bypassed, and definitive
toxicity data procedures were implemented for each
permittee.
Tier 2 — Definitive Data Generation —
Toxicity (Section 3)
Requirement
The three permittees were given the same definitive
data generation requirements via §308 letters. They
were required to conduct effluent toxicity tests follow-
ing a tiered toxicity testing procedure as described in
the U.S. EPA Technical Support Document (Office of
Water, September, 1985). Dilution available at low-flow
periods on the Linville River falls within the effluent-
dominated category of receiving waters. Therefore,
chronic toxicity data will be required.
Uncertainty factors to be applied:
• For species sensitivity
— 10X, where two test species are used.
— 1X, where three test species are used.
• For effluent variability:
— 100X, where testing is limited to one test per
test species quarterly.
— 10X, where testing is limited to two tests per
test species monthly.
— 1X, where testing conducted on four samples
per test species monthly.
The regulatory authority will determine further data
generation requirements when the uncertainty factors
66
-------
analysis is made on the basis of the following compar-
ison at each test tier:
NOEL
IWC
> level of uncertainty
where IWC is the instream waste concentration and
NOEL is the No Observed Effect Level. If, after these
tests, the regulatory authority determines no more data
are warranted due to a large difference between effect
level and level of uncertainty, testing can be terminated.
Where no such difference exists, further testing
requirements will be determined.
Special Considerations
Several special considerations have been provided by
the State to reduce testing costs, where appropriate:
• If three species are tested, only the most sensitive
species need be tested after the first two months
of testing identifies that species.
• If the difference between each monthly sample's
NOEL is less than 50% after the first two months
of testing, the sampling frequency can be reduced
by half.
• Where greater than 100% effect levels are
observed for all tests after the first two months of
testing, data generation procedures can be
terminated.
Definitive Data Generation Requirements
for Each Example Permittee
Cibesi Corporation
TESTING
Based on dilution calculations, the company has an
IWC of 4.5%. If they conducted chronic toxicity tests
and consistently measured NOELs of greater than
100% (no observable toxicity) their best NOEL/IWC
ratio would be:
> 100%
0.5 MGD
= 22
10.7 MGD + 0.5 MGD
x 100
They would be able to pass successfully a testing tier
using an uncertainty factor of 10 (22 is greater than 10)
but not a testing tier with an uncertainty factor of 100
or greater. Bearing in mind the testing costs and the
special considerations described above, Cibesi
negotiated a testing requirement (Tier 2 testing) of four
tests (on consecutive-day samples) per species con-
ducted over a two-month period using Ceriodaphnia
and the fathead minnow. The company and the State
agreed that this testing procedure was the most effi-
cient since the results of the first two months of test-
ing would dictate whether or not additional data
generation would be required.
RESULTS
January, 1983
Ceriodaphnia NOEL
Fathead minnow NOEL
Sample
10%
70%
25%
>100%
1%
50%
5.5%
>100%
These first months' results surprised the company and
pointed to the probable need for toxicity reduction
procedures. At the lowest toxicity level (1%) no uncer-
tainty factor comparison could be passed (NOEL/IWC
is less than one). The regulatory authority determined
that a complete toxicity assessment was required to
establish permit limitations. A testing procedure was
established to provide a more complete effluent charac-
terization: conduct four tests (samples collected on
consecutive days) on three species (must include
Ceriodaphnia) monthly (or every two months) for a total
of at least six months. The special considerations listed
above will apply. Accordingly, the following data were
generated:
Sample
February, 1983 1 234
Ceriodaphnia NOEL 10% 25% 5.5% 1%
Fathead minnow NOEL 70% >100% 50% >100%
Selenastrum NOEL >100% >100% >100% >100%
April, 1983 1 2 3 4
Ceriodaphnia NOEL 5.5% 1% 5.5% 1%
Fathead minnow NOEL 50% 25% NA* 70%
Selenastrum NOEL >100% >100% >100% >100%
June, 1983 1 2 3 4
Ceriodaphnia NOEL
Fathead minnow NOEL
Selenastrum NOEL
5.5% NA* 1% 1'
dropped as a requirement;
not sensitive
dropped as a requirement;
not sensitive
1 234
Ceriodaphnia only:
August, 1983 1% 1% 5.5% 10%
September, 1983 5.5% NA 10% 8%
* Not available (test failed).
These data were sent to the regulatory authority by the
Cibesi Corporation where they were used to establish
a WLA.
Nicklaus By-Products
TESTING
The company had an estimated IWC of 21.0%. If they
conducted chronic toxicity tests and consistently
measured NOELs of greater than 100% (no observable
toxicity) their best NOEL/IWC ratio would be:
67
-------
> 100%
3.0 MGD
= 4.7
10.7 MGD + 0.5 MGD + 3.0 MGD
x 100
They would not be able to pass sucessf ully any tiered
testing that applied an uncertainty factor (since 4.7 is
less than 10, the minimum uncertainty factor). This
would require the company to implement a testing pro-
gram that would generate a complete database. No
tiered assessment utilizing some uncertainty factor or
combination of factors would be allowed for this dis-
charge situation. Therefore, the company proposed the
following testing requirement: conduct chronic toxic-
ity tests on three species (Ceriodaphnia, Pimephales,
and Selenastrum) using four samples taken on con-
secutive days per month for a total of at least six
months. The special considerations described above
should apply. The company requested that it be allowed
to conduct an analysis of persistence and instream
degradation if toxicity were measured. The regulatory
agency agreed and revised the requirement to say that
if the first month's testing showed toxicity, the com-
pany would be allowed to analyze the persistence of the
toxicity measured. Accordingly, the following data was
generated:
Sample
January, 1983 1 2 3 4
Ceriodaphnia NOEL 50% >100% 10% 70%
Fathead minnow NOEL 5.5% 10% <1% 1%
Selenastrum NOEL 10% 50% 20% 50%
The extreme toxicity of the effluent, particularly to fat-
head minnow, pointed to the need for toxicity reduc-
tion procedures. However, the company argued that
phenolic compounds were probably causing the toxic-
ity. These compounds, according to Nicklaus By-
Products, were not persistent after discharge and lost
their toxic effect after mixing. The company cited the
lack of fish-kills and obvious stream degradation as evi-
dence that toxicity was non-persistent.
The regulatory authority cited two procedures for per-
sistence analysis the company could use: shelf aging
and ambient toxicity testing. Further, the company was
given the option of conducting a biosurvey to assess
instream degradation of the biota in the river. The com-
pany first conducted tests following a sample aging
procedure in which an effluent sample was held for 24,
48, and 96 hours on the shelf prior to toxicity testing
with the following results:
Pimephales promelas seven-day growth test:
24-hour-old sample NOEL = 1.0%
48-hour-old sample NOEL = 5.5%
96-hour old sample NOEL = 8.0%
These results were inconclusive. Next, the company
conducted ambient toxicity testing following proce-
dures detailed in EPA's Technical Support Document
(Appendix C). A dye study of effluent dispersion is
required at a design low-flow period to determine mix-
ing characteristics.
The dye study showed the effluent rapidly mixed
after discharge. The effluent reached equilibrium
(IWC = 30%) approximately 500 feet below the
discharge point.
Next, six ambient stations were established along the
river:
Station
Location
1
2
3
4
5
6
RM
RM
RM
RM
RM
RM
RM
RM
31
22
21
21
21
21
20
16
.2
.1
.4
.4
.4
.5
.4
.4
Upstream (headwaters) control
(Cibesi Corporation discharge)
Directly upstream of discharge
(Nicklaus By-Products discharge)
Directly downstream of discharge
At point of complete mixing
One mile downstream
Five miles downstream
Two test species were used with the following results:
Species Stream Station
1
Ceriodaphnia
(# young/female)
(% survival)
24.1 10.4
100 60
5.5
20
8.0
40
9.1
60
16.0
90
Fathead minnow
(mean weight in mg) 0.47 0.40
(% survival) 90 90
- - - 0.31
0 0 0 30
Ambient toxicity results showed a pattern of instream
toxicity that would be expected if effluent toxicity data
were compared to effluent concentrations instream. At
Station 1, both species showed standard growth and
reproduction. At Station 2, below Cibesi, no effect was
seen for fathead minnows, but the Ceriodaphnia were
somewhat affected (as expected given instream con-
centrations of their effluent and the sensitivity of the
organism to the Cibesi discharge). Samples from sta-
tion 3, 4, and 5 below Nicklaus, showed lethality to the
minnows. Mortality was observed at the point of dis-
charge, at complete mixing, and at one mile down-
stream, respectively. Complete recovery for fathead
minnows was not observed even at Station 6, five miles
downstream. Ceriodaphnia did show recovery at this
station. Toxicity for Nicklaus By-Products was thus con-
sidered persistent by the regulatory authority at the
low-flow period observed.
The State determined that Nicklaus By-Products was
impacting the Linville River. The State then required the
company to generate additional effluent toxicity data
(following the procedures previously described) so that
appropriate limits could be calculated.
68
-------
RESULTS
Sample
February, 1983 1 234
Ceriodaphnia NOEL —- excessive control mortality —-
Fathead minnow NOEL 1% 10% 5.5% 1%
Selenastrum NOEL 50% >100% >100% >100%
March, 1983
Ceriodaphnia NOEL
Fathead minnow NOEL
Selenastrum NOEL
April, 1983
Fathead minnow NOEL
Ceriodaphnia NOEL
Selenastrum NOEL
1
100% >100% >100% >100%
50% 70% 30% 10%
>100% >100% >100% >100%
1234
10% 5.5% 2.5% 5.5%
dropped as a requirement;
not sensitive
dropped as a requirement;
not sensitive
1234
5.5%
10%
10%
5.5%
1%
3%
1%
1%
Fathead minnow only:
May, 1983
June, 1983
These data were presented to the regulatory authority.
The State will use the data to develop a WLA.
Bibury STP
TESTING
Based on dilution calculations, Bibury STP has an IWC
of 14%. If the STP conducted chronic toxicity tests and
consistently measured NOELs of 100%, their best
NOEL/IWC ratio would be:
> 100%
2.3 MGD
= 7.2
(10.7 + 0.5 + 3.0 + 2.3) MGD
100
They would not be able to pass successfully a testing
tier using any uncertainty factor (since 7.2 is less than
10, the minimum uncertainty factor). This would
require the STP to implement a testing program that
would generate a complete database. No tiered assess-
ment utilizing some uncertainty factor or combination
of factors will be allowed for this discharge situation.
Therefore, the STP agreed to conduct the following
testing: run chronic toxicity tests on three species
(Ceriodaphnia, Pimephales, and Selenastrum) using
four samples taken on consecutive days over a six-
month period. The special considerations previously
described will apply. The STP began testing with the
following results:
Sample
February, 1983 1 2 3 4
Ceriodaphnia NOEL >100% >100% >100% >100%
Fathead minnow NOEL >100% >100% >100% >100%
Selenastrum NOEL >100% >100% >100% >100%
1
April, 1983
Ceriodaphnia NOEL >100% >100% >100% >100%
Fathead minnow NOEL >100% >100% >100% >100%
Selenastrum NOEL >100% >100% >100% >100%
Since no chronic toxicity was observed during the first
two months of testing, the regulatory authority deter-
mined that the Bibury STP effluent was not toxic, and
further testing was terminated. No further water
quality-based limitation derivation procedures were
conducted.
Tier 2. — Definitive Data Generation —
Chemical-specific (Section 3)
Cibesi Corporation and Nicklaus By-Products were
required to conduct two plug-flow sampling programs
for cyanide during low flows approximately equal to
the WLA design flows of 1Q10 and 7Q10. In addition,
Cibesi was required to include copper in their plug-flow
sampling program. Cyanide was selected for Tier 2
monitoring because the State has numerical water
quality standards for this pollutant and it is discharged
at potentially acutely toxic levels by both Cibesi and
Nicklaus. Monitoring was required for copper because
the State has numerical water quality standards for this
pollutant. DMR and 2C data indicated that this metal
is discharged from Cibesi at concentrations signifi-
cantly above acutely toxic levels. Monitoring and
modeling of phenols is not required since many forms
of this compound may be present and each one would
require identification for comparison to EPA criteria.
Whole-effluent toxicity analyses should result in the
control of phenols as well as the other toxic com-
pounds without requiring detailed pollutant-specific
analyses of each chemical.
Data from one sampling period are needed for WLA
model calibration. A second sampling period is then
required for model verification. Flow conditions should
be similar for both sampling events so that calibrated
settling and resuspension rates for copper will be trans-
ferable to the verification period. Pollutant loadings
from Cibesi and Nicklaus must vary during the two
periods so that verification is a good test of the
calibrated rate coefficients.
Requirement
The firms were required to conduct two plug-flow
sampling analyses by following a dye slug along the
Linville River. Plug-flow sampling involves the collec-
tion of water quality samples at each station and each
effluent as the leading edge and peak of the aye slug
is passing that point.
69
-------
Results
Seven instream stations were established along the
river:
Station Location
1 RM 22.2 Directly upstream of Cibesi
RM 22. 1 (Cibesi Corporation discharge)
2 RM 22.0 Downstream of Cibesi at point of
complete mixing
3 RM
21 .65 0.35 miles downstream of Cibesi
4 RM
21.45 0.7 miles downstream of Cibesi
(directly upstream of Nicklaus)
RM 21 .4 (Nicklaus By-Products discharge)
5 RM 21.3 Downstream of Nicklaus at point of
complete mixing
6 RM17.6 3.8 miles downstream of Nicklaus
7 RM 13.8 7.6 miles downstream of Nicklaus
(directly upstream of Bibury STP)
Tables 8-1 and 8-2 show the results of the two sampling
programs.
Exposure Assessment (Section 5)
Effluent Toxicity Wasteload Allocation
The dye study showed that mixing is rapid and com-
plete. Therefore, a mixing zone analysis is not required
to assess exposure, and the State used a complete mix
WLA approach. A steady state WLA for effluent tox-
icity was performed for Cibesi Corp. and Nicklaus By-
Products. As explained in Section 3, the analysis is
based on the assumption that toxicity is conservative
and additive. The ambient toxicity data collected by
Nicklaus By-Products indicate that this assumption is
reasonable.
The input data required for the WLA include: the
chronic criteria, effluent flow, ambient toxicity
upstream from Cibesi, and the critical design flow of
the Linville River. The instream chronic criterion is one
toxic unit, measured as chronic toxicity to Ceriodaph-
nia (the most sensitive species to the Cibesi effluent)
or fathead minnows (the most sensitive species to the
Nicklaus By-Products effluent). Criteria for duration and
recurrence are as described above. Effluent flow is
given as 0.5 MGD for Cibesi and 3.0 MGD for Nicklaus
By-Products. Toxicity upstream from Cibesi is assumed
to be zero since there are no significant dischargers
above this point and ambient toxicity data collected by
Nicklaus By-Products showed good water quality (no
measureable toxicity) at the control station above the
discharge points.
The toxicity effects of both effluents are assumed to
be comparable so that wasteloads can be allocated
between them even though two different test organ-
isms turned out to be most sensitive. The rationale is
that both species merely represent the more sensitive
Table 8-1. Plug Flow Sampling /1
Stream Station Data
Constituent 1 234567
Total CN (ug/1) 0 26.8 25.7 24.6 66.4 41.9 26.3
Dissolved CN (ug/1) 0 26.8 25.7 24.6 66.4 41.9 26.3
Total Cu (ug/1) 0 57.2 56.1 55.1 42.7 34.6 29.2
Paniculate Cu (ug/1) 0 31.9 30.8 29.8 25.3 17.2 11.8
Dissolved Cu (ug/1) 0 25.3 25.3 25.3 17.4 17.4 17.4
TSS (mg/1) 20 21 20.3 19.6 24.2 16.5 11.3
Temperature (°C) 20 20 20 20 20 20 20
Hardness (mg/1 CaCOs) 100 100 100 100 100 100 100
Flow(cfs) 15 15.8 15.8 15.8 20.4 20.4 20.4
Velocity (ft/sec) 0.5 0.5 0.5 0.5 0.5 0.5 0.5
Effluent Data
Cibesi Nicklaus
Total CN (ug/1) 530 210
Total Cu (ug/1) 1130 -
TSS (mg/1) 40 40
Flow (cfs) 0.8 4.6
Table 8-2. Plug Flow Sampling 12
Stream Station Data
Constituent 1234567
Total CN (ug/1) 0 124.3 119.1 114.1 186.6 117.8 74.4
Dissolved CN (ug/1) 0 124.3 119.1 114.1 186.6 117.8 74.4
Total Cu (ug/1) 0 115.1 112.9 110.9 84.6 68.8 58.0
Paniculate Cu (ug/1) 0 58.0 55.8 53.8 49.8 34.0 23.2
Dissolved Cu (ug/1) 0 57.1 57.1 57.1 34.8 34.8 34.8
TSS (mg/1) 15 16.9 16.3 15.7 23.8 16.3 11.1
PH 7777777
Temperature (°C) 20202020202020
Hardness (mg/1 CaCOs) 100 100 100 100 100 100 100
Flow (cfs) 14 14.8 14.8 14.8 19.4 19.4 19.4
Velocity (ft/sec) 0.5 0.5 0.5 0.5 0.5 0.5 0.5
Effluent Data
Cibesi Nicklaus
Total CN (ug/1) 2300 420
Total Cu (ug/1) 2130 —
TSS (mg/1) 50 50
Flow (cfs) 0.8 4.6
end of the range of potential sensitivities. The meas-
ured toxicity from each effluent will "overlap" for the
many species in the receiving water. The toxic effects
measured by Ceriodaphnia and minnows can be
assumed comparable for the purpose of developing a
toxicity WLA (see Section 1).
Since the State water quality standards specify that
they apply only at or above the 7Q10, this flow is
selected as the chronic design flow for Linville River.
A 1Q10 is selected as the acute design flow since the
70
-------
one-day period is more appropriate for acute event
duration (see Appendix D).
A U.S. Geological Survey (USGS) flow gauging station
is located on the Linville River directly above the Cibesi
outfall. The low-flow statistics program in STORET cal-
culates the 1Q10 at this station as 8.9 MGD or 13.7 cfs
and the 7Q10 as 10.7 MGD or 16.6 cfs.
This multiple-discharge WLA is based on flow-
weighting of the assimilative capacity. There are many
other ways of doing a multiple-source WLA (e.g.,
allocating the total allowable load based on flow
proportion, cost parity, etc.). Cibesi will be allowed to
use 86% of the assimilative capacity of the river since
Cibesi flow is only 14% of the total point source flow
to the river. Consequently, the instream toxicity below
Cibesi must equal 0.86 times the criterion for both
acute and chronic effects. Nicklaus By-Products will be
required to provide the remainder of the treatment
needed to maintain the criteria for toxic effects.
The WLAs for acute protection were developed using
an acute-chronic ratio (ACR) of 10, which was deter-
mined from the chronic toxicity testing data by look-
ing at 48 versus the 168 hour (seven-day) NOEL. The
effluent toxicity required for Cibesi to meet the maxi-
mum acute criterion is calculated as follows:
Chronic Criterion = 1.0 TUC
(CSQS) + (CeQe)
Maximum Acute
Criterion
Qe + Qs
where Cs = upstream TUa
Qs = acute critical flow (1Q10)
Ce = required effluent TUa
Qe = effluent flow
Maximum Acute Criterion = 0.3 TUa
(0) (8.9) + (Ce) (0.5)
0.5 + 8.9
< (0.86) (0.3)
Ce < 4.85 TUa
or using the ACR,
Ce < 48.5 TUC
The effluent toxicity required for Nicklaus By-Products
to meet the maximum acute criterion is calculated as
follows:
(0.258) (9.4) + (Ce) (3.0)
Ce < 0.43 TUa
Ce < 4.3 TUC
(0'3)
The effluent toxicity required for Cibesi to meet the
chronic requirement is calculated as follows:
(CSQS) + (C.QL)
Qe + Qs
< (0.86) Chronic Criterion
where Cs = upstream TUC
Qs = chronic critical flow (7Q10)
Ce = required effluent TUC
Qe = effluent flow
(0) (10.7) + (Ce) (0.5)
0.5 + 10.7
Ce < 19.3 TUC
1(0.86)
The effluent toxicity required for Nicklaus By-Products
to meet the chronic requirement is calculated as
follows:
(0.86) (11.2) + (Ce) (3.0)
11.2 + 3.0
Ce ^ 1.5 TUC
< 1
Pollutant-specific WLA
At Stations 2 and 5, the WLA model used the complete
mix mass balance equations presented above. At the
other stations the equation for cyanide consisted of the
following:
CNX = CM,, e
-kx/u
where CN
CN
= cyanide concentration at
distance x (^g/l)
U = upstream cyanide
concentration (^g/l)
k = first order decay rate (I/day)
x = downstream distance (ft)
u = stream velocity (ft/sec)
The equations for copper at the other stations con-
sisted of the following:
TSSX = TSSU e
Cut = Cud (1
Cup = Cut - Cuc
-W/HU/U)
KPTSSX
where TSSX = total suspended solids concen-
tration at distance x (mg/l)
TSSU = upstream total suspended
solids concentration (mg/l)
Ws = settling velocity of TSS
(ft/day)
H = depth of water (ft)
x = downstream distance (ft)
u = stream velocity (ft/sec)
Cu, = total copper concentration at
distance x (^g/l)
Cud = dissolved copper concentra-
tion at distance x (/tg/l)
Cup = particulate copper concentra-
tion at distance x (/tg/l)
Kp = copper partition coefficient
(I/kg)
The following rate coefficients were calibrated and veri-
fied with the data from the two sampling events:
K = I/day for cyanide
Ws = 0.82 ft/day
Kp = 60,000 I/kg
71
-------
The calibrated constants were within the range
reported in the technical literature. Resuspension was
not observed during low flow so modeling of TSS and
Cu did not include this term. The WLA model described
above was used with the calibrated rate coefficients
and the 1Q10 flow to calculate the allowable acute
loads for copper and cyanide. The model was also used
with the 7Q10 flow to calculate the allowable chronic
loads. The same flow-weighting rule used for the
multiple-discharge WLA of generic toxicity was also
used for cyanide. Tables 8-3 and 8-4 show the predicted
instream results of acute and chronic WLAs for copper
and cyanide.
The following assumptions were made for the WLA:
• Upstream of Cibesi TSS = 15 mg/l.
• Upstream of Cibesi CN and Cu equal zero.
• Stream velocity at all stations is 0.5 ft/sec (as indi-
cated by monitoring data).
• Depth of flow at all stations is one foot (as indi-
cated by the monitoring data).
• New treatment will result in Cibesi and Nicklaus
effluent TSS = 30 mg/l.
The 1985 EPA national criteria for copper specify the
"active" form of the metal. Acid-soluble copper is dis-
tinguished from other forms by the analytical technique
used to measure it. Acid-soluble copper is always less
than total copper and greater than dissolved copper.
The WLA done for this situation used the conservative
assumption that total copper must be equal to the
acute and chronic criteria at the appropriate critical
flows. The 1985 cyanide criteria specify "free cyanide."
Since all the cyanide in this situation is dissolved, all the
chemical is considered to be in the free form.
In order to meet the 1985 instream criteria for copper
and cyanide with flows of 0.5 MGD at Cibesi and
3.0 MGD at Nicklaus, the following loads are allowable:
Table 8-3. Acute WLA Results
Cyanide WLA: Cibesi
Acute one-day CN load
(Ib/day) 1.59
Acute one-day CN
concentration (^g/l) 370
Chronic seven-day CN load
(Ib/day) 0.37
Chronic seven-day CN
concentration (/tg/l) 85.0
Copper WLA:
Acute one-day Cu load
(Ib/day) 1.25
Acute one-day Cu
concentration (jig/l) 290
Chronic seven-day Cu load
(Ib/day) 1.03
Chronic seven-day Cu
concentration (^g/l) 239
Nicklaus
0.74
30.0
0.15
6.0
Stream Station
Constituent
Total CN (ug/1)
Dissolved CN (ug/1)
Total Cu (ug/1)
Paniculate Cu (ug/1)
Dissolved Cu (ug/1)
TSS (mg/1)
1
0
0
0
0
0
15
2
20.4
20.4
16.0
7.8
8.2
15.8
3
19.5
19.5
15.7
7.5
8.2
15.3
4
18.7
18.7
15.5
7.3
8.2
14.8
5
21.4
21.4
11.8
6.2
5.6
18.5
6
13.5
13.5
9.9
4.3
5.6
12.7
7
8.5
8.5
8.5
2.9
5.6
8.7
Table 8-4. Chronic WLA Results
Stream Station
Constituent
Total CN (ug/1)
Dissolved CN (ug/1)
Total Cu (ug/1)
Paniculate Cu (ug/1)
Dissolved Cu (ug/1)
TSS (mg/1)
1
0
0
0
0
0
15
2
3.9
3.9
11.0
5.3
5.7
15.7
3
3.7
3.7
10.9
5.2
5.7
15.2
4
3.5
3.5
10.7
5.0
5.7
14.7
5
4.0
4.0
8.5
4.4
4.1
17.9
6
2.5
2.5
7.1
3.0
4.1
12.2
7
1.6
1.6
6.1
2.0
4.1
8.3
Permit Limit Derivation (Section 6)
The method for deriving permit limitations is a result
of the way that the WLA for the discharger is
expressed. In this example, the WLAs for both dis-
chargers were developed using the acute and chronic
steady state approach. This approach yields require-
ments expressed as a one-day value and a seven-day
average value for effluent quality. To derive limits, the
State permit writer had to determine which require-
ment was more limiting and to derive daily maximum
and average daily limits to enforce that requirement
properly. To illustrate the calculations, the derivation of
limits for the copper WLA for Cibesi is described in
detail below. The outcome for the other WLAs is sum-
marized in Tables 8-5 and 8-6 and described in general.
The WLA requirements for copper were specified as
follows: the one-day concentration should be less than
290 /*g/l, and the seven-day average concentration
should be less than 239 j*g/l. The State determined that
the occurrence probability for the WLA values should
be 0.01. They estimated that the CV for copper after
compliance would be 0.60 by looking at the develop-
ment documents for metal finishing and electroplating.
The log variance, a, was calculated as follows:
a2 = In (CV2 + 1)
= In (0.602 + 1)
= 0.3075
a = Vb.3075 = 0.5545
The z value corresponding to a probability of 0.01 is
2.326.
The required long term average (LTA) needed to meet
the one-day requirement was calculated as follows:
72
-------
WLA = 290 /tg/l
H = In (WLA) - Z a
= In 290 - 2.326 (0.5545)
= 4.3801
LTA = e"+0-5a2
_ 4.3801+0.5(0.3075)
— 6
= 93 /tg/l
The LTA needed to meet the seven-day requirement
was calculated as follows:
WLA = 239 /tg/l
Table 8-5. Cibesi Corporation
p7 = In (WLA) - Z Vln [1 + ((ea2-1)/7)]
= 4.9556
H = M - 0.5 a2 + 0.5 In [l + Ke"'-!)/?)!
= 4.8269
LTA = e"*0-5"1
= 145 /tg/l
The LTA needed to meet the one-day WLA value is 93
/tg/l. The LTA needed to meet the seven-day WLA value
is 145 /tg/l. The one-day or acute WLA requirement is
more limiting for this plant. Therefore, permit limits
were derived based on a required LTA of 93
To calculate permit limits, the permit writer needed to
determine the probability basis for the permit limits (not
to be confused with the probability basis for the WLA
values) and the assumed frequency of observation for
the monthly limits. For Cibesi, it was decided to require
daily monitoring because of the potential water qual-
ity impact and because copper analyses are inexpen-
sive. The compliance monitoring was considered
sufficient to allow a permit limit probability basis of
0.01 instead of 0.05. The assumed number of observa-
tions for the monthly limit was n = 30. Permit limits
were derived using Figure 6-3 as follows:
Daily maximum:
H = In (LTA) - 0.5 o-2
= In 93 - 0.5(0.3075)
= 4.3801
limit = eli+2a
_ 4.3801 + 2.326(0.5545)
= 289 |tg/l
Average daily (monthly):
limit = LTA (1 +
= 93 (1 +
= 117/tg/l
(CV))
(0.60))
A summary of the copper requirements can be seen in
Table 8-5. In this case it was the one-day WLA require-
ment that was more limiting for the plant even though
the WLA value was higher than the seven-day require-
Wasteload Allocation
Parameter Duration Value
Required
L.T.A.
Permit Limits
Daily
Basis Monthly Maximum
Toxicity
1-day 48.5 tuc 10.3 tuc 95th
7-day 19.3 tuc 8.0 tuc n = 10
12.5 tu. 21.8tu.
CN
1-day 370 ug/l 69 ug/l 99th
7-day 75 ug/l 36 ug/l n = 30
52 ug/l 71 ug/l
Cu
1-day 290 ug/l 93 ug/l 99th
7-day 239 ug/l 145 ug/l n = 30
117 ug/l 289 ug/l
Table 8-6. Nicklaus By-Products
Wasteload Allocation
Permit Limits
Parameter Duration Value
Required
L.T.A.
Daily
Basis Monthly Maximum
Toxicity
CN
1-day
7-day
1-day
7-day
4.3 tuc 0.9 tuc 95th
1.2tuc
1.5tuc 0.7 tuc -
30 ug/l
6 ug/l
6 ug/l
1.9tuc
12 ug/l
ment. The daily maximum permit limit is the same as
the WLA value only because the one-day WLA was
limiting and because the probability basis for the WLA
and the limits are the same.
Table 8-5 also describes Cibesi's requirements for tox-
icity and cyanide. The toxicity CV was estimated from
the toxicity data collected during impact assessment.
The value for CV was 0.96 and was assumed not to
change. The CV for cyanide was obtained from the
electroplating development documents and was esti-
mated as 1.10. For both these parameters, the seven-
day WLA requirement was limiting. The toxicity limits
illustrate the greatest change needed to properly
enforce the WLA. The factors that affect the limits are
as follows:
• The WLA occurrence probability was set at 0.01,
but the permit limit basis was 0.05 because
monitoring would be infrequent (once a week).
• The seven-day WLA was the limiting requirement.
If daily effluent values were allowed to go to the
one-day WLA value, non-compliance with the
seven-day WLA requirement is projected.
• The chronic WLA value is a seven-day average,
whereas the monthly permit limit was derived as
a ten-day average.
The requirements for Nicklaus By-Products (Table 8-6)
were significantly more stringent as a result of the
73
-------
reverse flow-weighting approach used in the WLA. The
WLA for toxicity was as follows: seven-day average
less than 1.5 TUs and one-day value less than 4.3 TUs
measured as chronic toxicity to fathead minnows. The
2 5 datum points generated during the evaluation of the
discharge had a mean of 45.5 TUs and a CV of 0.93.
As with Cibesi, the variance was assumed constant,
and the probability that the WLA would be exceeded
was set at 0.01. The LTA required to meet the seven-
day WLA requirement was more limiting than the LTA
needed to meet the one-day WLA.
The daily maximum limit was calculated as 1.9 TUs.
The average daily limit turned out to be below the
detection limit for a toxicity test. (The detection limit
of toxicity test is 1.0 TUs since samples cannot be
diluted any less than 100% sample.) Because the per-
mittee could not meet an average limit set at the detec-
tion limit and still take advantage of the short term
variability allowed under the daily maximum, the State
set the average limit at 1.2 TUs. The value was derived
considering the weekly monitoring requirement and
assuming that three measurements would be below the
detection limit (< 1.0 TUs) and one could be as high as
1.9 TUs. It was felt these limits best approximated the
required effluent performance.
For cyanide, the one-day WLA value was 30 /*g/l and
the seven-day value was 6 /tg/l. Since the State had no
basis to estimate the CV, and the seven-day value was
so close to the detection limit for cyanide (5 /*g/l), the
State did not use a statistical analysis. The treatment
that the company proposed consisted of recycling;
segregation of high strength wastestreams; polymer
settling; alkaline chlorination; recombination of waste-
streams; extended aeration-activated sludge with
powered activated carbon; and clarification. It was felt
that no cyanide would be discharged. The State set
cyanide limits of 12 IJLQ/\ daily maximum and 6 /*g/l aver-
age monthly with daily monitoring for the first year.
74
-------
Appendices
A. Policy for the Development of Water Quality-based Permit Limitations for
Toxic Pollutants
B. Sampling
C. Ambient Toxicity Testing and Data Analysis
D. Duration and Frequency
E. Lognormal Distribution and Permit Limit Derivations
-------
Appendix A.
Policy for the Development
of Water Quality-based Permit Limitations
for Toxic Pollutants
A-1
-------
9016Federal Register / Vol. 49, No. 48 / Friday, March 9. 1984 / Notices
[OW-FRL-2533-1J
Development of Water Quality-Based
Permit Limitations for Toxic Pollutants;,
National Policy
AGENCY: Environmental Protection
Agency fEPA).
ACTION: Notice.
SUMMARY: EPA has issued a national
policy statement entitled "Policy for the
Development of Water Quality-Based
Permit Limitations for Toxic Pollutants."
This policy addresses the technical
approach for assessing and controlling
the discharge of toxic substances to the
Nation's waters through the National
Pollutant Discharge Elimination System
(NPDES) permit program.
FOR FURTHER INFORMATION CONTACT:
Bruce Newton or Rick Brandes, Permits
Division (EN-336), Office of Water
Enforcement and Permits, U.S.
Environmental Protection Agency,
Washington, D.C. 20460. 426-7010.
A-2
-------
Federal Register / Vol. 49, No. 48 / Friday, March 9, 1984 / Notices
9617
SUPPLEMENTARY INFORMATION: As the
water pollution control effort in the
United States progresses and the
"traditional" pollutants (oxygen
demanding and eutrophying materials)
become sufficiently treated to protect
water quality, attention is shifting
towards pollutants that impact water
quality through toxic effects. Compared
with the traditional pollutants,
regulation of toxic pollutants is
considerably more difficult. The
difficulties include (1) the great number
of toxic chemicals that may potentially
be discharged to receiving waters and
the difficulties in their analysis; (2) the
changes in the toxic effects of a
chemical ^resulting from reactions with
the matrix of constituents in which it
exists; and (3) the inability to predict the
effects of exposure to combinations of
chemicals.
To overcome some of these problems,
EPA and the States have begun to use
aquatic toxicity tests and various human
health assessment techniques to
complement chemical analyses of
effluents and receiving water samples.
Because these techniques or their
application to effluent testing are new,
EPA and the States have been cautious
in their use. Based on EPA's evaluation
of these techniques and the experiences
of several States, EPA is now
recommeding the use of biological
techniques as a complement to
chemical-specific analyses to assess
effluent discharges and express permit
limitations. EPA has issued these
recommendations through a statement
of policy and is developing a technical
guidance document to help implement
the policy.
The complete test of the national
policy statement follows:
Policy for the Development of Water
Quality-Based Permit Limitations for
Toxic Pollutants
Statement of policy
To control pollutants beyond Best
Available Technology Economically
Achievable (BAT), secondary treatment,
and other Clean Water Act technology-
based requirements in order to meet
water quality standards, the
Environmental Protection Agency (EPA)
will use an integrated strategy
consisting of both biological and
chemical methods to address toxic and
nonconventional pollutants from
industrial and municipal sources. Where
State standards contain numerical
criteria for toxic pollutants, National
Pollutant Discharge Elimination System
(NPDES) permits will contain limits as
necessary to assure compliance with
these.standards. In addition to enforcing
specific numerical criteria, EPA and the
States will use biological techniques and
available data on chemical effects to
assess toxicity impacts and human
health hazards based on the general
standard of "no toxic materials in toxic
amounts."
EPA, in its oversight role, will work
with States to ensure that these
techniques are used wherever
appropriate. Under section 308 and
section 402 of the Clean Water Act (the
Act), EPA or the State may require
NPDES permit applicants to provide
chemical, toxicity, and instream
biological data necessary to assure
compliance with standards. Data
requirements may be determined on a
case-by-case basis in consultation with
the State and the discharger.'
Where violations of water quality
standards are identified or projected,
the State will be expected to develop
water quality-based effluent limits for
inclusion in any issued permit. Where
necessary, EPA will develop these limits
in consultation with the State. Where
there is a significant likelihood of toxic
effects to biota in the receiving water,
EPA and the States may impose permit
limits on effluent toxicity and may
require an NPDES permittee to conduct
a toxicity reduction evaluation. Where
toxic effects are present but there is a
significant likelihood that compliance
with technology-based requirements will
sufficiently mitigate the effects, EPA and
the States may require chemical and
toxicity testing after installation of
treatment and may reopen the permit to
incorporate additional limitations if
needed to meet water quality standards.
(Toxicity data, which are considered
"new information" in accordance with
40 CFR 122.62(a)(2), could constitute
cause for permit modification where
necessary.)
To carry out this policy, EPA Regional
Administrators will assure that each
Region has the capability to conduct
water quality assessments using both
biological and chemical methods and
provide technical assistance to the
States.
Background
The Clean Water Act establishes two
principal bases for effluent limitations.
First, existing dischargers are required
to meet technology-based effluent
limitations that reflect the best controls
available considering economic impacts.
New source dischargers must meet the
best demonstrated technology-based
controls. Second, where necessary,
additional requirements are imposed to
assure attainment and maintenance of
water quality standards established by
the States and approved by EPA. In
A-3
establishing or reviewing NPDES permit
limits, EPA must ensure that the limits
will result in the attainment of water
quality standards and protect
designated water uses, including an
adequate margin of safety.
For toxic and nonconventional
pollutants it may be difficult in some
situations to determine attainment or
nonattainment of water quality
standards and set appropriate limits
because of complex chemical
interactions which affect the fate and
ultimate impact of toxic substances in
the receiving water. In many cases, all
potentially toxic pollutants cannot be
identified by chemical methods. In such
situations, it is more feasible to examine
the whole effluent toxicity and instream
impacts using biological methods rather
than attempt to identify all toxic
pollutants, determine the effects of each
pollutant individually, and then attempt
to assess their collective effect.
The scientific basis for using
biological techniques has advanced
significantly in recent years. There is
now a general consensus that an
evaluation of effluent toxicity, when
adequately related to instream
conditions, can provide a valid
indication of receiving system impacts.
This information can be useful in
developing regulatory requirements to
protect aquatic life, especially when
data from toxicity testing are analyzed
in conjunction with chemical and
ecological data. Generic human health
effects methods, such as the Ames
mutagenicity test, and structure-activity
relationship techniques are showing
promise and should be used to identify
potential hazards. However, pollutant-
specific techniques are the best way to
evaluate and control human health
hazards at this time.
Biological testing of effluents is an
important aspect of the water quality-
based approach for controlling toxic
pollutants. Effluent toxicity data in
conjunction with other data can be used
to establish control priorities, assess
compliance with State water quality
standards, and set permit limitations to
achieve those standards. All States have
water quality standards which include
narrative statements prohibiting the
discharge of toxic materials in toxic
amounts. A few State standards have
criteria more specific than narrative
criteria (for example, numerical criteria
for specific toxic pollutants or a toxicity
criterion to achieve designated uses). In
States where numerical criteria are not
specified, a judgment by the regulatory
authority is required to set quantitative
water quality-based limits on chemicals
and effluent toxicity to assure
-------
9018
Federal Register / Vol. 49, No. 48 / Friday, March 9, 1984 / Notices
compliance with water quality
standards.
Note.—Section 308 of the Act and
corresponding State statutes authorize KPA
and the States to require of the owner/
operator any information reasonably required
to determine permit limits and to determine
compliance with standards or permit limits,
Biological methods are specifically
mentioned. Toxicity permit limits are
authorized under Section 301 and 402 uml
supported by Section 101.
Application
This policy applies to EPA and the
States. The policy addresses the use of
chemical and biological methods for
assuring that effluent discharges are
regulated in accordance with Federal
and State requirements. This policy was
prepared, in part, in response to
concerns raised by litigants to the
Consolidated Permit Regulations (see FR
52079, November 18,1982). Use of these
methods for developing water quality
standards and trend monitoring are
discussed elsewhere (see 48 FR 51400,
November 8,1983 and Basic Water
Monitoring Program EPA-440/9-76-025).
This policy is part of EPA's water
quality-based control program and does
not supersede other regulations, policy.
and guidance regarding use
attainability, site-specific criterid
modification, wasteload allocation, and
water quality management.
Implementation
State Role
The control of toxic substances to
protect water quality must be done in
the context of the Federal-State
partnership. EPA will work
cooperatively with the States in
identifying potential water quality
standards violations, assembling
relevant data, developing appropriate
testing requirements, determining
whether standards are being violated,
and defining appropriate permit limits.
Note.—Under sections 303 and 401 of the
Act, States are given primary responsibility
for developing water quality standards and
limits to meet those standards. EPA's role is
to review the State standards and limits and
develop revised or additional standards or
limits as needed to meet the requirements of
the Act.
Integration of Approaches
The type of testing that is most
appropriate for assessing water quality
impacts depends on the type of effluent
and discharge situation. EPA
recommends that an integrated
approach, including both biological and
chemical techniques, be used to assess
and control water quality. The principal
advantages of chemical-specific
techniques are that (1) chemical
analyses are usually less expensive than
biological measurements in simple
cases; (2) treatment systems are more
easily designed to meet chemical
requirements than toxicity requirements;
and (3) human health hazards and
bioaccumulative pollutants can best be
addressed at this time by chemical-
specific analysis. The principal
advantages of biological techniques are
that (1) the effects of complex
discharges of many known and
unknown constituents can be measured
only by biological analyses; (2)
bioavailability of pollutants after
discharge is best measured by toxicity
lusting; and (3) pollutants for which
there are inadequate chemical analytical
methods or criteria can be addressed.
Pollutant-specific chemical analysis
techniques should be used where
discharges contain few, well-quantified
pollutants and the interactions and
effects of the pollutants are known. In
addition, pollutant-specific techniques
should be used where health hazards
are a concern or bioaccumulation is
suspected. Biological techniques should
be used where effluents are complex or
where the combined effects of multiple
discharges are of concern. EPA
recognizes that in many cases both
types of analysis must be used.
Testing Requirements
Requirements for dischargers to
collect information to assess attainment
or nonattainment of State water quality
standards will be imposed only in
selected cases where the potential for
nonattainment of water quality
standards exists. Where water quality
problems are suspected but there is a
strong indication that complying with
BCT/BAT will sufficiently mitigate the
impacts, EPA recommends that
applicable permits include testing
requirements effective after BCT/BAT
compliance and reopener clauses
allowing reevaluation of the discharge.
The chemical, physical, and biological
testing to be conducted by individual
dischargers should be determined on a
case-by-case basis. In making this
determination, many factors must be
considered, including the degree of
impact, the complexity and variability of
the discharge, the water body type and
hydrology, the potential for human
health impact, the amount of existing
data, the level of certainty desired in the
water quality assessment, other sources
of pollutants, and the ecology of the
receiving water. The specific data
needed to measure the effect that a
discharger has on the receiving water
will vary according to these and other
factors.
An assessment of water quality
should, to the extent practicable, include
other point and nonpoint sources of
pollutants if the sources may be
contributing to the impacts. Special
attention should be focused on Publicly
Owned Treatment Works (POTW's)
with a significant contribution of
industrial waste-water. Recent studies
have indicated that such POTW's are
often significant sources of toxic
materials. When developing monitoring
requirements, interpreting data, and
determining limitations, permit
engineers should work closely with
water quality staff at both the State and
Federal levels.
A discharger may be required to
provide data upon request under section
308 of the Act, or such a requirement
may be included in its NPDES permit.
The development of a final assessment
may require several iterations of data
collection. Where potential problems are
identified, EPA or the State may require
monitoring to determine whether more
information is needed concerning water,
quality effects.
Use of Data
Chemical, physical, and biological
data will be used to determine whether,
after compliance with BCT/BAT
requirements, there will be violations of
State water quality standards resulting
from the discharge(s). The narrative
prohibition of toxic materials in toxic
amounts contained in all State
standards is the basis for this
determination taking into account the
designated use for the receiving water.
For example, discharges to waters
classified for propagation of cold water
fish should be evaluated in relation to
acute and chronic effects on cold water
organisms, potential spawning areas,
and effluent dispersion.
Setting Permit Limitations
Where violations of water quality
standards exist or are projected, the
State and EPA will determine pollution
control requirements that will attain the
receiving water designated use. Where
effluent toxicity is an appropriate
control parameter, permit limits on
effluent toxicity should be developed. In
such cases, EPA may also fequire a
permittee to conduct a toxicity reduction
evaluation. A toxicity reduction
evaluation is an investigation conducted
within a plant or municipal system to
isolate the sources of effluent toxicity,
specific causative pollutants if possible,
and determine the effectiveness of
pollution control options in reducing the
effluent toxicity. If specific chemicals
are identified as the cause of the water
A-4
-------
Federal Register / Vol. 49. No. 48 / Friday, March 9. 1984 / Notices 9019
quality standards violation, these
individual pollutants should be limited.
If a toxicity reduction evaluation
demonstrates that limiting an indicator
parameter will ensure attainment of the
water quality-based effluent toxicity
requirement, limits on the indicator
parameter should be considered in lieu
of limits on effluent toxicity. Such
indicator limits are not limits on
causative pollutants but limits
demonstrated to result in a specific
toxicity reduction.
Monitoring
Where pollution control requirements
are expressed in terms of a chemical or
lexicological parameter, compliance
monitoring must include monitoring for
that parameter. If an indicator
parameter is used based on the results
of a toxicity reduction evaluation,
periodic toxicity testing may be required
to confirm the adequacy of the indicator.
Where biological data were used to
develop a water quality assessment or
where the potential for water quality
standards violations exist, biological
monitoring (including instream
monitoring) may be required to ensure
continuing compliance with water
quality standards.
EPA believes that the intelligent
application of an integrated strategy
using both biological and chemical
techniques for water quality assessment
will facilitate the development of
appropriate controls and the attainment
of water quality standards. EPA looks
forward to working with the States in a
spirit of cooperation to further refine
these techniques.
Policy signed February 3,1984 by Jack E.
Ravan, Assistant administrator for Water.
Dated: February 16,19H4.
jack E. Ravan,
Assistant Administrator for Water.
|FR Doc. 84-6445 Filltd 3-fl-84, 8.-J5 .jm|
BILLING CODE 66SO-50-M
A-5
-------
Appendix B.
Sampling
B-1
-------
Appendix B.
Sampling
Sampling
The objective of an effluent or instream sampling pro-
gram is to obtain a sample (or samples) from which a
representative measure of the parameter of interest can
be obtained. Unfortunately, many of the industrial and
municipal NPDES sampling protocols presently in use
are carryovers from schemes used for calculating load-
ings of nutrients and oxygen-demanding substances,
or were developed to evaluate treatment plant opera-
tional efficiency. Sampling for individual toxicants and
particularly for effluent toxicity can require more
specific (and thus different) sampling procedures.
Wastewater variability is an important consideration in
selecting the method and frequency of sampling for
both chemical analysis and toxicity testing. Industrial
waste characteristics have been shown to vary in fre-
quency, intensity, and duration (1). As noted by Bender
(2), the sources of effluent variability include both ran-
dom and systematic components that influence both
daily and annual characteristics of waste discharges.
Although toxic pollutant loading may be of primary con-
cern in assessing human health impact or bioaccumu-
lation, loading may be of lesser importance in toxicity
assessment than frequency, intensity, and duration of
peak toxic discharge. Sampling must be tailored to
measure the type of toxicity of importance for that dis-
charge: either long term (chronic) impact, which is a
more constant effect, or short term (acute) impact,
which is more variable and subject to peaks of intensity.
There are several chemical parameters for which con-
tinuous analysis is possible. These include pH, temper-
ature, dissolved oxygen, and other parameters involving
instantaneous measurement. All other types of meas-
urement involve some time period over which the anal-
ysis is conducted. Toxicity tests require an exposure
period. Chemical tests require sample preparation and
analysis. There is no continuous analysis method for
toxicity.
It should be noted that although it is difficult to design
a representative sampling program for toxicity analy-
sis, the problems are of no greater magnitude than simi-
lar problems associated with obtaining a representative
sample for conventional pollutants.
Sampling Methods
Continuous Flow Samples
For toxicity testing, the test organisms may be exposed
to serial dilutions of a sample continuously pumped
from the effluent pipe or ditch. In the case of effluents,
if optimum accuracy is desired, then the ratio of
effluent flow/test chamber volume can be scaled to
simulate the time-varying concentration at the mixing
zone boundary.
Although flow-through methods can provide a realis-
tic simulation of time-varying exposure, they are rela-
tively expensive and are usually conducted on site.
Therefore, flow-through methods may only be practi-
cal where the goals of the analysis of impact require
this type of testing or where treatment costs are suffi-
ciently high that this type of analysis can be required.
A flow-through exposure method is not a continuous
analysis because only one result or data point is
obtained at the end of the test. However, the con-
tinuous exposure does provide some measure of time-
varying exposure effects.
Discrete Samples
Grab or flow composited sampling provide a discrete
sample for chemical analysis or toxicity testing. Static
or renewal toxicity tests using discrete samples result
in exposure of test organisms to a constant effluent
composition over the period of the tests, or for the
period between renewals.
If discrete samples are collected during peaks of
effluent toxicity, then constant concentration exposure
static tests provide a measure of maximum effect.
Depending on the duration of a peak and the composit-
ing period, composited samples may not be useful for
examining toxicity peaks because the compositing
process tends to dilute the peaks. Composited samples
are usually appropriate for chronic tests where peak
toxicity of short duration is of less concern. The aver-
aging effect of compositing may be misleading when
testing for acute toxicity.
B-2
-------
Grab samples must be collected at sufficiently frequent
intervals to provide a high probability of sampling daily
peaks. Fortunately, static toxicity tests are relatively
inexpensive and can be done on shipped samples; thus,
it may be cost effective to conduct individual tests on
a series of grab samples collected over a 24-hour
period.
Sampling Frequency
Non-random effluent variability, resulting from batch
processing, variable loadings, etc., is often known or
can be determined. Therefore, the first step in design-
ing a sampling program for chemical analysis or toxicity
testing is to select the annual sampling frequency
based on available site-specific operational information.
This is important in selecting sampling periods for both
continuous flow and discrete sampling methods.
If discrete sampling methods (grabs or composites) are
used, then random variations between and within days
for each sampling period must be considered. It is
important to recognize the tradeoff between the long
term (between days) frequency and short term (within
days) frequency of sample collection and analysis for
toxics. At present, the permit requirements for sam-
pling and analyzing chemical parameters are site
specific and generally involve a single grab or 24-hour
composite sample collected at daily, weekly, or monthly
intervals. Unfortunately, a sampling scheme involving
a single daily grab or a 24-hour composite sample can
conceal the presence of those daily extreme values that
may be of importance. To optimize sampling cost and
effectiveness, it may be desirable to reduce long term
frequency so that daily frequency can be increased.
For example, a weekly grab or composite involves 52
analyses per year. It may be more efficient to reduce
the annual frequency to monthly or bimonthly, but col-
lect and analyze four or eight grabs daily. Either scheme
(12x4 or 6x8) would involve 48 analyses per year vs.
52 for the weekly single sample approach. Assuming
that daily toxic events of environmentally significant
intensity and duration would not be masked by short
Box B-1
RECOMMENDATIONS
The initial sampling design step should involve stratification of sampling periods to account for non-
random sources of variation (e.g., batch processing). The second step includes selection of the frequency
and the method of sampling to be conducted within each sampling period. Depending on site-specific
considerations, several options are available.
• Flow-through Methods — Ideally, for both acute and chronic effluent toxicity tests, the exposure
of biota should simulate the time-varying concentration at a predetermined point in the receiving
water. For regulatory purposes, the critical point is often the edge of the mixing zone where the
waste should exhibit neither acute nor chronic toxicity. Therefore, if warranted by site-specific
factors, it is recommended that test biota be exposed to a continuously collected flowthrough
sample of serially diluted effluent. If no systematic annual variations (e.g., batch processing) are
known or suspected, flow-through testing can be conducted at a minimum of quarterly intervals
for at least one year.
• Grab Sample Methods — Grab samples are recommended for chemical analyses and for acute and
chronic toxicity tests where site conditions (such as wastewaters that are known to have relatively
constant composition) do not require use of continuous flow methods. Grab samples of effluent
or receiving water may be used for static or renewal acute toxicity tests, which may be conducted
on site or at a remote lab. The design of a toxics grab-sampling program must take into account
the trade-off between long term and short term sampling intensity. Where there is no ponding of
wastes or retention time is insufficient for thorough mixing, it is important to collect or analyze a
sufficient number of samples to provide a measure of daily spikes. Therefore, to minimize analyti-
cal costs where daily fluctuations are known or suspected, the annual sampling frequency should
be reduced in favor of more intensive daily sampling. It is recommended that on an annual cycle,
grab sampling and analysis include a minimum of four to six daily grabs collected monthly. An option
could include the use of short term (four-hour) composites rather than grabs. If site-specific data
are available to indicate that treatment system retention time is adequate to minimize daily varia-
tions, then the daily replicates may be omitted in favor of more frequent annual sampling (e.g., weekly
or semimonthly rather than monthly). If, to minimize costs in screening tests, only single samples
are collected at infrequent intervals (e.g., quarterly) an uncertainty factor for variability should be
used in the toxicity evaluation (see Section 3).
• Composite Sample Methods — If static or renewal methods are used for evaluation of toxicity, it
is recommended that 24-hour, continuous-flow composite samples be collected. Considerations
of annual frequency are the same as those for grab samples.
B-3
-------
term composites, it might be more efficient to collect
eight samples each composited over a three-hour
interval.
If costs or other constraints prohibit satisfactory daily
and annual replication of sampling, then a level of un-
certainty must be introduced into the calculations used
to evaluate waste toxicity (see Section 3, Table 3-1).
Box B-1 presents EPA's recommendations on sampling
methods.
References
1. Nemetz, P.N., and H.D. Dreschler. 1978. The role
of effluent monitoring in environmental control.
J. Water, Air, and Pollution. 10:477-497.
2. Bender, E.S. 1984. Sources of variations in effluent
toxicity tests. In: Environmental Hazard Assess-
ment of Effluents. (H. Bergman, R. Kimerle, and
A.W. Maki, eds.) Proceedings of the Fifth Pells-
ton Environmental Workshop. Cody, WY,
August, 1982.
B-4
-------
Appendix C.
Ambient Toxicity Testing and Data Analysis
C-1
-------
Appendix C.
Ambient Toxicity Testing and Data Analysis
Ambient Toxicity Analysis
Ambient toxicity testing procedures are useful where
measurement of toxicity levels after discharge is impor-
tant in the assessment of toxic effluent impact. This
is particularly true where impact is caused by the pres-
ence of multiple point sources. The purpose of this test-
ing is to provide an analysis of toxicity levels instream
from whatever sources of toxicity are affecting the
receiving water.
Procedures
The basic ambient toxicity testing procedure is to
expose test organisms to receiving water samples
taken from selected sampling stations above, at, and
below the discharge point(s). Since effluent concentra-
tions after discharge are often relatively low, chronic
toxicity tests should be conducted so that the tests are
sensitive enough for the purpose.
The methods available for chronic testing of sufficiently
short duration are limited. Two organisms for which
short term chronic toxicity tests are available are
Pimephales promelas and Ceriodaphnia sp.
The following procedures are used:
• Select instream sampling stations based on the
mixing characteristics involved in the specific dis-
charge situation.
• Collect a daily grab sample or a daily composite
sample of receiving water from each station.
• Use a renewal testing method to expose test
organisms to the daily samples collected at each
station. Use an appropriate number of replicates
(10 for Ceriodaphnia) for each sampling station. No
dilution series is required where screening is the
primary goal.
• Testing must be conducted at a low flow period.
However, it is not necessary to conduct the tests
at the critical low flow period. Testing is best when
relatively stable flow occurs during the test period.
• Record the results of the testing in the format
shown in Table C-1. The survival of the test organ-
isms and the effect on their growth or reproduc-
tion are used as endpoints. Figure C-1 plots the
results in graphic form so that the pattern of
ambient toxicity can be observed.
Selecting Sampling Stations
The selection of sampling stations is determined by the
characteristics of the site. When determining stations,
the following factors should be considered:
• Mixing and flow — The mixing characteristics of
the discharge site are useful to determine the
placement of sampling stations. Knowledge of
concentration isopleths allows the regulatory
authority to place stations at locations instream
Table C-1. Young Production and Percent Survival of Ceriodaphnia in Ambient Toxicity Tests at Ottawa River, Lima, Ohio
Station
1
2
3
3A
3B
4
4A
5
6
7
8
8A
9
Station Description
Above Lima
Above STP
Below STP
Midway between STP and refinery
Above refinery
Above chemical plant
Below chemical plant
Shawnee Bridge
Route 117
Allentown
Rimer
"Boonie" Station
Kalida
River
Mile
46.0
37.7
37.4
37.3
37.1
36.9
36.3
35.4
32.6
28.8
16.0
8.0
1.0
Young
Female
15.5
14.1
0
0
0.4
7.5
11.1
5.7
12.6
16.8
17.4
25.0
25.6
SD
8.0
2.1
—
_
—
3.6
4.6
4.0
3.8
6.1
9.5
3.3
5.5
Final
Survival
90
0
0
0
0
10
30
0
10
100
80
100
100
Daily Survival
1
100
100
100
100
90
100
100
90
100
100
100
100
100
2
100
100
100
100
90
100
100
90
100
100
90
100
100
3
100
100
10
10
40
100
100
90
100
100
90
100
100
4
90
100
0
0
0
100
100
90
100
100
90
100
100
5
90
90
0
0
0
100
100
90
100
100
90
100
100
6
90
10
0
0
0
50
40
60
100
100
80
100
100
7
90
0
0
0
0
10
30
0
10
100
80
100
100
C-2
-------
15-
5-
80
65
48 32
River Kilometers
16
Figure C-1. Ceriodaphnia young production in water from various
stream stations on the Ottawa River, Lima, Ohio.
that correspond to concentrations measured in the
dilution series in the effluent tests. For example,
where effluent testing shows the effluent NOEL is
10%, an instream station should be placed where
dilution is estimated to create a 10% instream
waste concentration. In this way, the size of a toxic
plume can be measured. Sampling stations should
be placed where the effluents exist at relatively
constant and relatively specific concentrations.
Test at specific low flow conditions, if possible.
Presence of tributaries or other sources of dilution
will influence positions and numbers of stations.
Where smaller tributaries have several point
sources on them, treat the tributary as a point
source. Obvious non-point source areas should
also be used to set stations.
• Existing biological data — Where biosurvey data
are available, sampling station location should be
influenced by the more obvious trends in impact.
In particular, control stations and recovery stations
can be determined by biosurvey data.
• Single point sources — Single point source situa-
tions should be bracketed with an above station,
an immediate mixing station, several intermediate
stations corresponding to different instream con-
centrations, and a recovery station. Of course, a
control station should be established.
• Presence of other point sources — Multiple point
source situations require the placement of more
stations between discharge points. Each source
should be bracketed by sampling stations.
There are four areas or zones that can be recognized
when establishing the sampling stations for ambient
toxicity testing:
• Zone 1 — An upstream zone before the effluent
enters.
• Zone 2 — A zone of mixing.
• Zone 3 — A zone after mixing and before additional
dilution water enters.
• Zone 4 — A zone where additional dilution occurs
either from effluents or tributaries.
All possible combinations of occurrences are not prac-
tical to discuss but must be sorted out for each site.
Some generalizations are important to mention:
• Any upstream sources of contaminants, such as
other discharges, will confound the individual
effects of a downstream discharge. For example,
Zone 3 of the downstream discharge may occur in
Zone 4 of an upstream discharge. This does not
invalidate the measurement of ambient toxicity. It
only makes it difficult to attribute amounts of
response to each individual discharge. Response
to the instream mixture is what is measured.
• Careful location of sampling stations in Zone 3 is
critical. Zone 3 is the only place where toxicity
decay rates of any one discharger can be measured
and then only if there are no upstream discharges,
or if there are, only if that upstream effluent is
stable in that reach.
• In Zone 4, not only is degradation of the effluent
toxicity occurring, but there is dilution of it by other
effluents and tributaries. Depending on the site cir-
cumstances, one may not be able to learn anything
about the ambient toxicity characteristics of the
effluent of concern in this zone.
• To emphasize, what can be measured in each zone
depends on the above considerations. In the more
complex situation, only an estimate of ambient
toxicity at each station can be obtained. No infor-
mation about one effluent's toxicity decay rate will
be available where several toxic effluents mix. In
the most simple situation of one discharge and no
dilution downstream for a long distance, Zone 3
will be large enough to get a good measure of tox-
icity decay.
Analysis of Ambient Toxicity Measurement
• When used in screening, the ambient toxicity data
can identify areas in receiving waters where ambi-
ent toxicity exists instream. Attributing such
impact to specific point sources (particularly
where several sources discharge) may require
effluent toxicity testing.
• Except when used for screening purposes,
ambient toxicity measurements must be inter-
preted in conjunction with effluent toxicity test
data if conclusions are to be drawn concerning
changes in toxic effect after discharge. The same
species must be used in both the ambient and the
effluent toxicity tests.
• When analyzing the data, the performance of the
animals at each station downstream is compared
to that of the animals exposed to receiving water
without the effluent of concern in it but contain-
ing all other upstream additions. The result is an
integration of effects from all contaminants and
components and represents not only the toxicity
of the effluent of concern but also the interactions
of it with other effluents.
C-3
-------
• Where the downstream stations show toxic effect
at the concentrations measured as toxic in the
effluent toxicity tests, effluent toxicity can be con-
sidered to be occurring instream, after discharge.
• Where the toxic effect decreases from station to
station downstream in the absence of further dilu-
tion, the effluent toxicity is degrading. If the decay
rate is rapid (e.g., no toxicity at the closest instream
station to the discharge point), the effluent has a
non-persistent toxicity. Where the decay rate is
more gradual, toxicity is more persistent. The rate
of decay of toxicity together with mixing data
allows the regulatory authority to approximate a
receiving water toxicity impact area. That impact
area can then be compared to the appropriate state
water quality standards when establishing control
requirements.
• In some cases, ambient toxicity may increase in
relation to effluent toxicity measurements. Either
upstream sources of toxicity exist or some factor
in the receiving water is reacting with the effluent
to increase its toxicity. Again, the pattern and mag-
nitude of change in toxicity should be analyzed.
Differences in toxicity levels between stations will
reveal what is happening to the effluent as it is
mixed instream and interacts with the constituents
of the receiving water.
• Trend analysis in the raw test data is of importance
in the interpretation of ambient toxicity data. As
used in this context, trend analysis means observ-
ing toxic effect as it occurs in the test itself and
relating it to what is occurring instream (plug flow,
intermittent discharge, peak toxicity of effluents).
Using time-of-travel data or receiving water flow
rates and patterns, observe effects on the test
organisms from day to day. There may be a pattern
of mortality which can be linked to discharge
events. For example, in C-1 the data indicate late
mortality at downstream stations on days 6 and 7.
Flow rates for the river in this example correlated
this mortality to the downstream movement of a
toxic slug illegally discharged upstream.
C-4
-------
Appendix D.
Duration and Frequency
D-1
-------
Appendix D.
Duration and Frequency
The following discussion describes the EPA's recom-
mendations for duration and frequency as they apply
to the assessment and control of toxic pollutants. It
should be noted that the recommendations presented
are based directly on national water quality criteria.
They are subject to change if and when EPA determines
any changes, improvements, or updates are needed.
Two important factors must be accounted for when
assessing exposure conditions and setting water
quality-based permit limits for toxicants in NPDES
permits: duration of adverse exposure and frequency
with which criteria are exceeded. Specifications for
these factors are provided in the national water qual-
ity criteria documents. They can be defined as follows:
• Duration of exposure considers the amount of time
organisms will be exposed to toxicants. It is ex-
pressed as that period of time over which the in-
stream concentration is averaged for comparison
with criteria concentrations. The duration of ex-
posure is determined scientifically using primarily
laboratory data.
• Frequency is defined as how often exposures that
exceed the criteria can occur during a given period
of time (e.g., once every ten years) without unac-
ceptably affecting the community. Ideally, a dyn-
amic model of the exposed population would be
used to select the frequency. Currently this model-
ing is beyond our capabilities, so the choice of the
frequency represents the best judgement of how
often a population can be reduced without affect-
ing its viability.
Duration
Several factors must be considered in order to properly
control duration. Toxic effect is a function of both mag-
nitude (concentration) and duration (time). Toxicity
tests measure concentrations that cause adverse
impacts over the testing period. In addition to limiting
the concentration of toxicants, the regulatory agency
must also specify and limit the length of time the
instream biota are exposed to those concentrations. For
example, a very brief exposure to a relatively high con-
centration may be less harmful than a prolonged
exposure to a lower concentration. Therefore, two dura-
tion specifications need to be set: a maximum duration
for protection against acute effects, and a maximum
duration for protection against chronic effects.
For acute effects, the duration must be as short as prac-
ticable because acute toxicity can occur quickly. An
instantaneous value as a maximum limit would be most
protective. Determination of an instantaneous maxi-
mum would require impractical and resource-intensive
continuous monitoring. Therefore, acute criteria are
expressed as a criterion maximum concentration
(CMC) occuring in a one-hour averaging period. The
hourly average is based on a daily measurement in most
cases.
It is more complicated to set the tolerable duration of
exposure for chronic effects. Since chronic toxicity
occurs over an extended time period, chronic effect
concentrations can be exceeded for a relatively short
period of time with no adverse effect so long as the
CMC is not exceeded. Chronic criteria are expressed as
a criterion continuous concentration (CCC). The objec-
tive is to maintain the concentration of a toxicant (or
toxicity) at or below the CCC over a protracted period
of time, but still allow some excursion above the CCC
that will not cause toxic effect. The question is, how
long can concentrations exceed the CCC without some
adverse impact occurring?
Exposure concentrations are usually continuously fluc-
tuating. Figure D-l graphically represents how concen-
trations could vary over time. Accurate measurement
of the fluctuations over time would require continuous
monitoring. As stated above, most measurements are
-CMC'
i
<-> CCC*
T— Unacceptable Exposure
\ (average above the CMC)
r Acceptable Exposure
\ (average below the CCC)
Unacceptable \
Area of
Chronic Effect
Exposure (average x
above the CCC)
1 \ 1 1 1 1
4 8 12 16 20 24 28
Duration (in days)
"CMC = criterion maximum concentration
"CCC = criterion continuous concentration
Figure D-1. Idealized representation of duration of acceptable and
unacceptable toxic concentrations.
D-2
-------
taken on a daily basis and consist of either instantane-
ous measurements or some form of composite meas-
urement. Water quality analysts also typically examine
environmental parameters as an average over some
time period. Because concentrations can be above the
CCC without causing adverse effects, there is con-
siderable temptation to specify the CCC in terms of
average exposure. However, if the period during which
exposure is averaged is long, periods of concentrations
above the CCC can produce unacceptable toxic effects
without the average concentration exceeding the CCC.
CCCs are usually based on test concentrations that are
constant. Therefore, they do not provide data for
effects of exposures fluctuating about an average, even
though this is the type of exposure that occurs in re-
ceiving waters.
The following examples illustrate what happens when
organisms are exposed to toxic effluents at specific
concentrations known to produce acute and chronic
toxic effects. These four examples are taken from the
Complex Effluent Toxicity Testing Program (a multi-year
EPA research effort examining the toxicity of industrial
and municipal effluents). The examples show situa-
tions where substantial toxicity to the test species
Ceriodaphnia affinis results from exposure at concen-
trations below the CMC and above the CCC for periods
of time from five to seven days.
Lima STP
LC50 = 100%
Dilution required to meet the CMC
(LC50x0.3) = 30%
NOEL = 5.5%
At a 30% effluent concentration, 40% of the Cerio-
daphnia died during days 6 and 7. At a 10% concentra-
tion, there were no deaths after four days. On days 5
and 7, however, young production was 24% and 30%
less than the production resulting from exposure to a
3% effluent concentration. These data show that ex-
posure concentrations above the CCC (NOEL) but be-
low the CMC (0.3 xLC50) can definitely be harmful.
Petroleum Refinery
LC50 = >100%
Dilution required to meet the CMC
(LC50x0.3) = >30%
NOEL = 5.5%
There were no Ceriodaphnia fatalities up through seven
days' exposure at a 30% effluent concentration, but
there was zero young production during that period. At
a 10% concentration, there were no mortalities. How-
ever, at 5 and 7 days, young production was 43% and
40% less than the young production resulting from
exposure to a 3% effluent concentration.
Birmingham Coke Plant
LC50 = >100%
Dilution required for the CMC
(LC50x0.3) = >30%
NOEL = 5.5%
At a 30% effluent concentration, there were no Cerio-
daphnia deaths, but zero young production for all days.
At a 10% effluent concentration, there were no mortal-
ities. On days 4, 5, 6, and 7, young production was
reduced 57%, 94%, 98%, and 86%, respectively, as
compared to young production resulting from exposure
to a 3% effluent concentration.
Naugatuck STP
LC50 = >100%
Dilution required for the CMC
(LC50x0.3) >30%
NOEL = 17%
At a 30% effluent concentration, 10% of the Cerio-
daphnia died after the 4th day, and young production
was reduced on days 5 and 7 by 24% and 62% as com-
pared to production resulting from exposure to a 10%
effluent concentration.
No similar data for fathead minnows (also tested) can
be given because the only measurements made on
days 5, 6, and 7 are for lethality and the CMC (0.3 times
the LC50) is supposed to eliminate lethality. Based on
data for effluents currently available, the CMC as now
set seems to prevent lethality. In order to gather simi-
lar data for fatheads, exposures would have to be ter-
minated on days 5 and 6 to get the weights. This has
not been done.
These data combined with other observations taken
during chronic toxicity testing show that four days is
the maximum period during which concentrations may
be continuously above the CCC (it must be remem-
bered that the CMC should not be exceeded for more
than one hour). After four days, impact begins to occur.
Since the longest duration allowable above the CCC
must be limited to four days or less, in the majority of
discharge situations the maximum period in which the
average exposure should not exceed the CCC is 4 days.
Exposure to substantial fluctuations about a mean con-
centration causes more adverse effect than continuous
exposure to the mean concentration. The averaging
periods recommended to restrict duration of adverse
exposure have been made short enough to restrict
allowable fluctuation in the instream concentrations.
The CCC stream flow averaging period used for steady
state wasteload allocation (WLA) modeling may be as
long as 30 days in situations involving publicly owned
treatment works (POTWs) designed to remove ammo-
nia where limited variability of effluent pollutant con-
D-3
-------
centration and resultant concentrations in receiving
waters can be demonstrated. In cases where low vari-
ability can be demonstrated, longer averaging periods
for the ammonia CCC (e.g., 30-day averaging periods)
would be acceptable because the magnitude and dura-
tion of concentrations above the CCC would be suffi-
ciently limited.
Frequency
Developing a WLA involves assessing the probability
that an event (criteria being exceeded) will occur. To be
completely certain that an event will never happen
would require unnecessarily stringent pollution con-
trols. Ecosystems can tolerate occasional impacts
provided they are not too severe or too frequent. The
question is how often criteria can be exceeded with-
out adversely affecting the community. Exceeding
criteria here refers to either a one-hour average exceed-
ing the CMC or a four-day average exceeding the CCC
(Figure D-l).
The guidance for how frequently criteria should be
allowed to be exceeded is developed along two lines.
These are based on the severity of the expected bio-
logical impact and are discussed below.
Considerations for Setting Frequency for
Minor Stresses
Criteria may be exceeded either as a result of predict-
able impacts from regulated pollutants or because of
spills (which are not regulated discharges under permit
limits but can have drastic effects on the community).
In the Complex Effluent Toxicity Testing Program, eight
field studies evaluating the use of toxicity tests to pre-
dict biological impact were conducted, and ambient
toxicity measurements were taken over a seven-day
period. During three of these studies, spills of pollutants
resulting in acute toxicity were documented. This
seemingly high frequency of spills suggests that their
effects may be as important as impacts caused by nor-
mal variations in effluent composition, concentration,
and dilution flow.
Predictable impacts, the occurrence of low streamf low
and high effluent concentration or toxicity, can be
reasonably well modeled and incorporated into the per-
mit. These impacts should most often be minor. Spe-
cial care should be exercised when there are many
outfalls in a small reach of receiving water, however,
because if one discharge exceeds the criteria due to low
flow, criteria violations are very likely to occur at other
outfalls at the same time. The several "minor" events
may thus add up to a "major" one.
Communities that are not under stress can withstand
an event better than a community that is under stress.
Recovery time from small events in unstressed systems
has been found to be a few months to a year (Table D-1).
Logically then, the allowable frequency can be greater
for unstressed communities.
In recovery studies where there was no toxicant
residual present, most fish species could repopulate in
as few as three weeks if conditions were advanta-
geous. The recovery time depended on the season of
the year. Recovery is most rapid in late summer to fall.
When determining the frequency that criteria may be
exceeded, the extent of the area affected should also
be considered. A reach of river having an affected area
of a few hundred feet will probably recover in a few
weeks, whereas a reach of several miles may take
years. Recovery of lakes or isolated systems is believed
to be very slow.
The considerations given to frequency should include
the number of events that are expected in an individual
year, the number of event-free years, and the number
of years between events. The temporal distribution of
events is dependent upon stream flow, effluent varia-
bility, and stream chemistry variability. Therefore, it is
difficult to predict when and exactly how many events
will occur. Most biological communities would not be
sufficiently affected if on the average there was one
event every three years. This means that there should
be two years on the average between events. If multi-
ple or very long events occur in one year, if uncontrolled
spills occur, or if there are large areas of impact in rela-
tion to the size of the waterbody, there should be a
longer period until the next event. It is possible in very
select situations that a frequency of once every one or
two years on the average might be acceptable. For such
cases, it would have to be demonstrated that the
affected area was small and the potential for biologi-
cal recovery was high.
Considerations for Setting Frequency
for Areas under Major Stresses and
Sensitive Communities
The frequency with which criteria can be allowed to be
exceeded will depend on the structure and function of
the community and the spatial relationships to other
non-affected areas. The impact on two different com-
munities, given the same toxic effects, will depend on
the availability of other unexposed receiving waters
from which animals can repopulate an area affected by
an exposure. If the sensitive individuals killed are long-
lived and produce few young, the impact on the com-
munity will be greater than if short-lived organisms with
higher reproduction potential are affected.
It is now believed that some ecosystems do not have
the ability to recover in one human life span.
Ecosystems can be classified as to their ability to
recover after a severe stress over a large area as follows:
D-4
-------
Ecosystem Type
Oceans
Lakes
Rivers
Estuaries
Ability to Recover
Low
Slight
Moderate
High
Cairns et al. [1] considered the following factors to be
important in evaluating ecosystem recovery:
• Existence of nearby epicenters.
• Transportation or mobility of dissemules (eggs,
spores).
• Condition of habitat following pollution stress.
• Presence of residual toxicants.
• Chemical and physical environment quality.
• Age and stability of ecosystem.
Characteristics of ecosystems that can be restored in
structure and function are:
• Indigenous species accustomed to highly variable
environmental conditions.
• System has high structural and functional
redundancy.
• Low stream order, flow dependable, good turbulent
diffusive action, high flushing.
• High water buffering.
• Not close to ecological Transition Zone.
Based on the above reference [1], ecosystem recovery
of the Clinch and Roanoke Rivers was evaluated and ap-
proximations of species recovery were developed.
Years After
Severe Major Stress
1st
2nd
3rd
Species
Reestablished
40-60%
60-80%
80-95%
In the case of the Clinch River, the large clams had not
recovered 13 years after the initial stress. It is probable
in this system as well as in other reported cases that
residual toxicity may have delayed the recovery.
Another important consideration is that a population
that is recovering from a severe stress is similar in
nature to a young system that is more vulnerable and
cannot recover as fast, i.e. repeated events cause ever
increasing impacts. The value placed on resources,
especially as to "important species," will also be a
factor in setting the frequency of allowed events.
The time required for various types of ecosystems to
recover from a stressful event should be a primary con-
sideration in determining the allowed frequency.
Although there is not enough known to draw definitive
conclusions relative to system type, some relative com-
parisons can be made with existing information. Much
of this information is drawn from studies conducted
subsequent to major oil spills and therefore reflects a
bias toward near-shore saltwater systems affected by
complex mixtures. Some work has been done on the
recovery of freshwater systems from toxic spills. Table
D-2 summarizes the relative susceptibility of various
Table D-1. Biological Recovery Studies
Site
Clinch River
Va.-Tenn.
Smith Branch
Illinois
Six sites on 4
streams in Louisiana
Latort Spring Run
Pennsylvania
Pine River
Pennsylvania
Codorus Creek
Pennsylvania
Cold Brook
New Jersey
Straight River
Minnesota
Toxicant
Caustic water;
Fly ash pond
Drought
Electro-fishing
Pesticide
Oil spill
Pesticide
Oil
Zinc
Degree ot
Damage
Severe
Severe
Decimation
Severe
Severe
Severe
Moderate
Slight
Community Effect
Benthos 100%,
fish 100%
Fish
Fish
Benthos 98%
fish
Benthos
Benthos 80%,
fish 90%
Benthos, fish (trout)
Benthos, fish,
zooplankton
Recovery Time
After 2 years, not
fully recovered
At 3 weeks, 21 of
29 species recovered;
Red Ear, 2 years later
4 of 6 sites in one
year; other sites
repopulated by other
species
1 year at one
of the sites; two
years for complete
recovery
In 4 years 50% of
species recovered;
residual oil
Good recovery in 2
months; continuing
good recovery at 6
months
2 years for recovery
of y-o-y trout, 1 year
for benthos
At 6 weeks,
recovering
Reference
Cairns, et al. 1971
Larimore, Childers
and Heckraffe. 1959
Berra, Gunning. 1970
Schott. 1981
Schott. 1982
Schott. 1983
Frey. 1973
Frey. 1974
Frey. 1975
Schott. 1980
Collier. 1981
Carlson. 1984
D-5
-------
aquatic systems to initial impact and estimates time
required for recovery following major stress.
Based on the above recovery data, an ecosystem capa-
ble of reestablishment would be back to its normal
function in three to four years after a major stress.
Based on logic, one would not want an ecosystem in
a constant state of recovery and increased vulnerabil-
ity from multiple stresses. An ecosystem severely
stressed every few years will not reach its full potential.
Where the communities are under stress from
1) uncontrolled spills, 2) large areas of impact relative
to the size of the water body (e.g., because of multiple
dischargers), or 3) exposure that will be extended for
longer time periods, the frequency should be reduced
to once every four or more years on the average.
Recommendations for Duration
and Frequency
The recommendations listed below are lexicologically
based specifications to be used as inputs to WLA
models. Depending on the modeling approach, differ-
ent inputs may have to be used to ensure full attain-
ment of these biological criteria. It is possible that a
variety of WLA design flows could result in the attain-
ment of the criteria. The following are the toxicologi-
cally based recommendations for ambient concentra
tions.
• The one-hour average concentration should not
exceed the CMC more than once every three*
years on the average.
• The four-day average concentration should not
exceed the CCC more than once every three* years
on the average.
The limitations of duration and frequency apply to both
single chemical criteria and to effluent toxicity. The
terms used are:
• For single chemicals the one-hour average concen-
tration is equal to one-half the Final Acute Value
and for general toxicity it is one-third the LC50
using the most sensitive of at least three test
species.
• For single chemicals the four-day average concen-
tration is equal to the lowest of the Final Chronic
* If the biological community is under stress because of spills,
multiple discharges, etc., or has a low recovery potential, or if a local
species is very important, the frequency should be decreased.
Table D-2. Relative Susceptibility of Various Aquatic Systems to Initial Impact
Ecosystem
Type
Rivers
Headwaters
Middle Reach
Slow
Lakes
Estuaries
Marine
Beaches
Rock Shore
Tidal Flat
Marshes
Open Water
Time Between Major Stresses (in years)
3
50-70% of species
recovered
50-75% of species
recovered
50-75% of species
recovered
Most species would
not be recovered
Principally clams
and molluscs are
recovered
Beaches are in state
of final repopulation
Colony communities
not recovered
Principally bivalves
not recovered
Annual plants and
short life span
organisms recovered
Very small area
repopulated
5
Recovery less than
95% of species
State of constant
recovery less than
95% of species
Recovery less than
95% of species
Biological integrity
not maintained
Clam and mollusc
populations still
reduced
Repopulated and
probably recovered
Colony communities
generally recovered
Bivalves still
reduced
Long-lived plants not
reestablished; most
organisms recovered
Long life span
organisms in recovery
10
Recovered
Recovered
Recovered
State of constant
recovery
Recovered
Recovered
Recovered
Recovered
Rnal stages of
recovery
Most species present
20
Recovered
Recovered
Recovered
Rnal state of
recovery
Recovered
Recovered
Recovered
Recovered
Recovered for very
large systems
Recovered except for
very large systems
100
Recovered
Recovered
Recovered
Recovered
Recovered
Recovered
Recovered
Recovered
Recovery depends
upon size of area
affected
Recovery depends
upon size of area
affected
*For those capable of recovery
Recovered is defined as greater than 95 percent of the original species being present and those species capable of being present would be present.
D-6
-------
Value, the Final Plant Value, and the Final Residue
Value unless data show that a lower value should
be used (National Guidelines), and for general tox-
icity it is the NOEL using the most sensitive of at
least three test species.
The use of national or site-specific criteria in calculat-
ing a WLA for waste treatment facilities requires selec-
tion of either a steady state or a dynamic model. The
concentrations, averaging periods, and frequencies
with which the CCC and CMC criteria are allowed to
be exceeded can be used directly in a dynamic model.
Such models are preferable in that they give a more
accurate picture of the stress on the environment by
allowing prediction of the number and extent of events.
However, limited data or other factors may make the
use of this type of model impractical. In these cases,
the regulatory authority may need to rely on a steady
state model.
The use of steady state models requires selection of a
design flow such that the biological criterion is not
exceeded more often than allowed by the specified fre-
quency. The allowed frequencies and the durations of
the averaging periods specified in the criterion should
not be used directly to calculate steady state design
flows using extreme value analysis. For example, if a
criterion specifies that the four-day average concentra-
tion should not exceed a particular value more than
once every three years on the average, this should not
be interpreted as implying that 4Q3 is appropriate to
use as the design flow. The criteria durations and fre-
quencies are biological specifications for protection of
aquatic organisms, and their use directly as a design
flow using extreme value analysis would result in differ-
ent frequencies for different receiving waters. Many of
these would not be protective. Properly selected design
flows, in terms of log Pearson extreme value calcula-
tion, may be appropriate for some receiving waters.
Other methods for determination of flows are under
consideration by the Agency.
Detailed guidance on the calculation of design flows to
achieve biologically based water quality criteria will be
provided by EPA. Until this guidance is developed, EPA
recommends the use of 1Q5 and 7Q5 as CMC and CCC
design flows for unstressed systems, and 1Q10 and
7Q10 for stressed systems. A system may be con-
sidered to be under stress if there are 1) uncontrolled
spills, 2) large areas of impact relative to the size of the
waterbody (e.g., because of multiple dischargers), or
3) exposure that will be extended for a longer time
period. The longer return frequency should also be used
if there is a locally important aquatic species that is
known to be particularly sensitive to the type of impact
expected (e.g., trout exposed to low dissolved oxygen
[DO]). These design flows are considered to be conser-
vative for the vast majority of cases. Design flow
regimes other than these have been used under certain
circumstances and in qualified site-specific situations.
They may continue to be used with proper justification.
States should, until publication of the design flow guid-
ance document, contact their EPA Regional Office for
the latest information.
References
1. Cairns, J., Jr., J. S. Grossman, K. L Dickson, and
E. E. Herricks. 1971. The Recovery of Damaged
Streams. ASB Bull. Vol 18(3), pp. 79-106.
2. Larimore, R. W., W. F. Childers, and C. Heckrathe.
1959. Destruction and Re-establishment of
Stream Fish and Invertebrates Affected by
Drought. Tran. Amer. Fish. Soc. (88),
pp. 261-285.
3. Berra, T. M., and G. E. Gunning. 1970. Repopula-
tion of Experimentally Decimated Sections of
Streams by Longear Sunf ish, Lepomis megalotis
megalotis (Rafinesque). Tran. Amer. Fish. Soc.
(4), pp. 776-781.
4. Schott, R. J. 1981. Aquatic Biology Investigation
Letort Spring Run.
5. Schott, R. J. 1982. Aquatic Biology Investigation
Letort Spring Run.
6. Schott, R. J. 1983. Aquatic Biology Investigation
Letort Spring Run.
7. Frey, R. F. 1973. Aquatic Biology Investigation Pine
Run. Memo Report, 30 October 1973.
8. Frey, R. F. 1974. Aquatic Biology Investigation Pine
Run. Memo Report, March 1974.
9. Frey, R. F. 1975. Aquatic Biology Investigation Pine
Run. Memo Report, September 1974.
10. Frey, R. F. 1977. Aquatic Biology Investigation Pine
Run. Memo Report, June 1977.
11. Schott, R. J. 1980. Aquatic Biological Investiga-
tion. Ageway, Inc./Herbicide Spill/Fish Kill
Codories Creek. Memo Report, September
1980.
12. Collier, C. R. 1981. Biological Survey of Cold Brook,
New Jersey. Bety Converse, Murdoch, Inc.,
Plymouth, PA.
13. Carlson, A. R., and T. H. Roush. 1985. Site-specific
Water Quality Studies of Straight River, Minne-
sota: Complex Effluent Toxicity, Zinc Toxicity,
and Biological Survey Relationships. Environ.
Res. Lab-Duluth Project Report.
D-7
-------
Appendix E
Lognormal Distribution and Permit Limit Derivations
E-1
-------
Appendix E.
Lognormal Distribution and Permit Limit Derivations
Introduction
This appendix is intended to provide supporting infor-
mation for use of the lognormal distribution in techni-
cal analysis related to toxics control in general and in
permit limit calculations in particular. Descriptive
details of this approach are augmented with empirical
evidence indicating that the lognormal distribution is
the appropriate distribution for permit limit derivations.
The concept and calculations are first presented from
the standpoint of describing existing effluent quality,
and then this framework is applied to derive water
quality-based permits. The discussion addresses sam-
pling frequency as it relates to the monthly average
daily limit and the treatment of correlated data.
Example Data
To provide some practical insight into the concepts and
procedures that follow, we first review three example
data sets. These data are actual measured concentra-
tions (in mg/l) of metals in the treated effluent of a foun-
dry and are displayed in Figure E-1. Observe that for
each of the metals — lead, copper and zinc — the vast
majority of the concentrations tend to be relatively low,
but a small, although substantial, number are quite
large by comparison. These patterns are typical meas-
urements of pollutants in wastewater, whether they be
concentrations, mass loadings, or generic toxicity units
(TUs). First, we will examine whether the normal dis-
tribution adequately describes these data and, then,
whether the lognormal distribution does any better.
Normal Distribution
Most, if not all, of us have encountered the normal prob-
ability distribution in a number of applications. Excel-
lent introductions and reviews of this distribution are
found in numerous statistical, engineering, and scien-
tific texts, as for example in Reference [1]. Only a terse
review will be given here.
Recall that the normal distribution has a probability
density function given by
1
f(y) =
exp [(y-/
f or - oc < x < oc. The symbols ^ and a are the popula-
tion mean and standard deviation. They, along with
2.00-
; 1.50-
i.oo
0.60
a. Lead
0.00
2.00-
1.50-
1.00-
0.50
0.00-1,.
0
2.00-
1.50-
1.00-
i 0.50
20
40
Sample Number
60
80
b. Copper
20 40
Sample Number
60
c. Zinc
0.00-i,.
20 40
Sample Number
60
Figure E-1. Effluent concentrations.
other parameters, can be estimated from a set of
independent observations Y1f Y2, ... , Yk from a nor-
mally distributed population according to the formulas
in Table E-1.
The quantity zp in Table E-1 is the standardized Z-score
for the pth percentile given by a table for the normal
distribution [1]. That is, zp is such that the probability
of a standardized, normally distributed random variable
being less than zp is p or
E-2
-------
Prob [Y < n + <7Zp] = p
where Y is a normally distributed random variable with
mean n and variance a2. The coefficient of variation
(CV), the ratio of the standard deviation to the mean,
is a dimensionless measure of the relative variability of
a distribution. The CV will be used frequently in these
procedures, especially in reference to the lognormal
distribution.
Table E-1, Population Parameters and Sample Estimates for a
Normal Distribution
POPULATION PARAMETER
MEAN = u Y = ;
VARIANCE = o2
STANDARD DEVIATION = o s
COEFFICIENT OF VARIATION s/ Y
= O/fi
PERCENTILE = n + zpo
SAMPLE ESTIMATOR
Y,/k
Probability and Log-probability Plots
Next, we will review how one can determine whether
a normal distribution models a population based on a
limited set of measurements. Consider an independent
sample of size k, labeled X1r X3, ..., Xk. Let U1( ..., Uk
be the sample of Xs in ascending order in which (J-\ is
the smallest X and Uk is the largest. Now for each U,,
find z, from the normal table such that
P[Z
-------
ity plots for the copper and zinc data exhibit similiar
patterns. Do these probability plots indicate that the
normal distribution is an adequate model for these
effluent data? The answer is given by consideration of
an alternative model: the lognormal distribution.
To construct a log-probability plot, set
Y, = In(Xj) or log(X,)
for i = 1, 2, ... , k, and then prepare a probability plot
for the Yj. Here, natural logarithms or logarithms to the
base e, denoted by ln(x) = loge(x), are used.
Logarithms to the base 10 can also be used.
o-
-1-
a. Lead
• • I ' '
0.0
Z Score
1.5
3.0
b. Copper
0
i-i
; -2
-31
-4J
-3.0
0.0
Z Score
•I -T
1.5 3.0
c. Zinc
o. j
H
-2-
-3-
-4-1
-3.0
-1.5
' I '
0.0
'I '
1.5
3.0
Z Score
Rgure E-3. Lognormal probability plots.
A log-probability plot for the lead data is presented in
Figure E-3. Without question, these lead data follow a
straight line on the log-probability plot much more
closely than the same data plotted on the probability
plot in Figure E-2. The straight line in the log-probability
plot is the best-fit model based on a lognormal distri-
bution. The log-probability plots for the copper and zinc
data sets exhibit the same pattern.
This empirical analysis demonstrates that these data
sets follow a lognormal distribution. This finding is not
peculiar to these data in particular. Experience has
shown that the same patterns emerge in analyses of
effluent concentrations and mass loadings. The degree
of fit for the lognormal distribution does vary somewhat
from application to application, but not enough to alter
the conclusion that effluent pollutant discharges are
generally lognormally distributed.
Lognormal Distribution
A random variable X behaves according to the log-
normal distribution if the random variable Y = ln(X) is
normally distributed. Effluent data are lognormally dis-
tributed. Therefore, the analysis procedures for analyz-
ing lognormal data are similar to those for the normal
distribution once the logarithmic transformation has
been applied. The analysis procedure is:
• Transform the observation to the log-space.
• Perform analyses and obtain results.
• Transform the results for the log-space back to the
observation-space.
The logarithmic transformation introduces specific
relationships between the parameters in the normal and
lognormal distributions. For instance, when X is lognor-
mally distributed, then Y = ln(X) is normally dis-
tributed. If Y has mean ^ and variance a2, then the mean
of X is given by
E(X) = e" + °2/2.
Other relationships between the lognormal distribution
of X in the observation-space and the normal distribu-
tion of Y = ln(X) in the log-space are given in Table E-2.
Table E-2. Normal and Lognormal Distributions
Y-NORMAL X-LOGNORMAL
(log-space) (observation space)
VARIABLE
POPULATION PARAMETER
MEAN
MEDIAN
VARIANCE
STANDARD DEVIATION
COEFFICIENT OF VARIATION
Y = In (X) X = exp(Y) =
ef
ol\n
02,^2 _ 1(
2/2(eo2
- 1)1/2
E-4
-------
-cr=0.1
1.50
1.00-
0.50-
O.OOH
Rgure E-4. Examples of lognormal densities.
The shape of the lognormal probability density func-
tion is determined by the parameter a, which is the stan-
dard deviation in the log-space. For a relatively small
value of a, say a = 0.1, the lognormal density appears,
at least to the eye, symmetric. For relatively large values
of a> 1, it is highly skewed to the right. In the kinds of
applications of interest here, the distributions are
almost always highly skewed, and the values of a are
generally in the range of 0.2 oa Daily
•-*•* Ten-Day
20
T
40
Sample Number
"I"
60
80
1=1
= -1.30
Figure E-5. Daily and 10-day average: Lead data.
and
s2 = £ (y,-y)2/73
1=1
= (1.03)2.
By transforming these parameter estimates back to the
observation-space, the estimates of the mean and stan
dard deviation of the concentration distribution art
found to be 0.46 mg/l and 0.63 mg/l, respectively. The
estimated coefficient of variation, therefore, is 1.37.
From Table E-1, the estimate of an upper percentile of
the lognormal distribution is given by
exp(y + zp sy) mg/l.
Therefore,
95th limit (z = 1.645) = 1.48 mg/l
and
99th limit (z = 2.326) = 2.99 mg/l.
Monthly Permit Limit
The monthly permit limit is the 95th or 99th percen-
tile of the distribution of the arithmetic mean of the
daily effluent values. The distribution is based on the
assumed number of samples that will be taken during
each month. This assumed number is generally recom-
mended to be 10 samples per month.
We again consider the lead concentration data that
were used for the daily maximum limit example in the
previous section. To illustrate the effects of averaging
observations, the data are displayed again in Figure E-5
together with averages of successive subsets of size
10 (i.e., in that figure, samples 1 through 10 were aver-
aged; samples 11 through 20 were averaged, and so
on). The average values are connected by a straight line.
Observe that these averages exhibit much less varia-
tion than the individual data. To obtain the monthly
limit, we calculate the percentile value of these
averages.
Again, we assume the daily observations are indepen-
dent and are drawn from a lognormal distribution. If we
assumed that n samples are taken each month, then
E-5
-------
the percentile for the distribution of monthly averages
is computed according to the table below. The formulas
in the table are based on the assumption that the sum
of lognormally distributed random variables is dis-
tributed according to a lognormal distribution. Theoret-
ically this assumption is not true, but the formulas,
nonetheless, provide a very accurate approximation as
demonstrated in Reference [3] and elsewhere.
The formulas for computing monthly averages based
on an assumed n samples per month are as follows:
Table E-3. Formulas for Computing Monthly Averages Based on
An Assumed k Samples Per Month.
pth percentile = exp(y"n + zp sn)
where s2n = ln(1 + le"2 -1)/n)
yn = y + (s2-s2n)/2
and y"and s are as in previous section
To illustrate the monthly limit calculations, consider the
lead data set used in the previous section to calculate
the daily permit limit. Recall for those data
k = 74.
y -1.30.
s = 1.03.
Using n = 10, from the equations above we have
s10 = 0.42 and
Vio = 0.86.
The 95th percentile limit (z95 = 1.645) is therefore
exp(y10 + (0.42X1.645)) = 0.85 mg/l,
which is considerably less than the daily 95th percen-
tile limit for the same data set of 1.48 mg/l.
Comment on Independence Assumption
In this and in the previous section, we have assumed
that all the observed pollutant levels are independent
or uncorrelated. In the case of the monthly limit deri-
vation, this assumption can be quite critical. If the
effluent levels are correlated, the actual monthly 95th
percentile can be substantially higher than that derived
from the analysis based on the independence assump-
tion. There is essentially no effect, however, on the cal-
culated daily limit validity. Whenever the data show a
positive correlation, that correlation should be factored
into the limitation; correlated or dependent data are
addressed in the subsection entitled Correlated Data.
Effect of Assumed Number of Samples
Per Month
The monthly limit procedures presented in this section
allow one to derive a monthly limit for an assumed num-
ber of samples taken during the month, denoted by n.
One can calculate a 95th percentile for n = 1, for n = 2,
and for n = 30. How do the results behave as a func-
tion of n? For each of three pollutants, monthly limita-
tions for each value of an assumed number of samples
are presented in Figure E-6. The general pattern is that
the calculated limitations are relatively insensitive to n
for n > 8. This analysis was used in part to support the
general recommendation that n should not be less than
10 for pollutants that are inexpensive to monitor.
Effect of Actual Number of Samples Taken
During the Month
Whatever value of n is selected to determine the
monthly limitation, the discharger may not take exactly
n samples during every month. For example, we derived
95th percentiles based on the assumption of 10 sam-
ples per month. What is the effect of the permittee tak-
ing a different number of samples, say 12, during the
month? This effect can be assessed by calculating the
theoretical probability that the monthly average (95th
percentile in this example) based on n = 10. This prob-
ability can be calculated for any number of samples
taken. These calculations are displayed for this exam-
ple in Figure E-7; the actual number of samples is plot-
ted on the X-axis, and the percentile is plotted on the
Y-axis. Observe that the calculated probability is rela-
tively insensitive unless the actual number of samples
is substantially less than 10.
Consistency of Limits
It is important to emphasize that these procedures pro-
vide consistent daily and monthly limits. They are con-
sistent in that they both incorporate the same level of
effluent quality and they are equally stringent. Enforce-
ability is enhanced because both limitations are equally
stringent. Limitations that dictate different levels of per-
formance are avoided. Accordingly, the permittee can
be expected to meet both the daily and monthly limi-
tations together.
Derivation of Water Quality-based Limits
Technology-based effluent quality limits (BAT, Best
Conventional Technology [BCT], secondary) are derived
from treatment system performance. Water quality-
based limits are derived from the required treatment
system performance necessary to comply with the
wasteload allocation (WLA). The mathematical expres-
sions for water quality limits are the same as those for
technology-based effluent quality limits; the major
difference is that the parameters in those expressions
are derived from the WLA. The procedure is illustrated
by an example.
We assume that for the existing treatment system,
which may or may not satisfy the WLA requirements,
the level of performance is known. Returing to the lead
data example, we have from the section on Daily Permit
Limits:
E-6
-------
Zinc
0 10 ' 20
Assumed Number of Samples
Rgure E-6. Limitations as a function of assumed number of
samples.
30
Actual Number of Samples Taken
1.0-j
3 b. Copper - 96%
0.9
I0.8
0.7-
10
• I'
11 • •
20
30
Actual Number of Samples Taken
O D D O O D
10
Actual Number of Samples Taken
Figure E-7. Probability associated with monthly averages.
LTA = 0.46 mg/l
and
CV = 1.37,
where LTA is the long term average. We assume further,
for this example, that new requirements can be satis-
fied by a fixed percentage reduction in effluent load-
ings, say by addition of a filter. This kind of modification
will have the impact of reducing the LTA but will not
change the CV. (Other kinds of modifications, such as
the addition of a polishing pond would reduce the CV,
which should be incorporated into the analysis.)
Suppose the acute WLA requirment is a one-day (com-
posite or grab sample) concentration requirement given
as
Acute WLA = 0.50 mg/l of lead
and the requirement is to be satisfied 99 % of the time.
We must, therefore, satisfy
exp(/ia + z99
-------
LTAC = 0.032 mg/l
and
CV = 1.37.
The corresponding 95th percentile limits are
Daily limit = 0.104 mg/l
and
Monthly limit = 0.051 mg/l,
assuming 10 samples per month.
These limits are internally consistent and ensure that
both the acute and chronic WLA will be satisfied.
Correlated Data
The analysis in this Appendix to this point has been
predicated on the assumption that measured effluent
pollutant loadings are independent or uncorrelated. If
data are strongly correlated, which is frequently the
case with POTWs, the monthly permits can be in sub-
stantial error if the correlation is not taken into account
in the analysis. In this section we address the incorpo-
ration of dependence in the calculations.
A major factor that determines whether effluent levels
are highly correlated is the retention time of the
wastewater treatment system. If the retention time is
large relative to the time between effluent samples,
then those samples will be correlated in most cases.
In municipal systems, for example, the retention time
is frequently a matter of days, and sampling is often
conducted on a daily basis. The effluent levels, conse-
quently, are substantially correlated. However, in many
industrial systems, for instance a physical/chemical
treatment system for electroplating wastewaters, the
treatment system retention time is relatively short: four
to eight hours. Daily effluent levels from these kinds of
systems are generally uncorrelated or statistically
independent. These general patterns are the same
irrespective of the kind of pollutant in question.
To illustrate the importance of accounting for correla-
tion in these analyses, we consider an example. In Sec-
tion C of Reference [4], an aggregate coefficient of
variation for public treatment works with activated
sludge is given as 0.85 for daily TSS concentrations.
The average TSS for the same plants is 14.3 mg/l. The
standard deviation for daily TSS is, consequently,
(14.31(0.85) = 12.15 mg/l.
If the daily TSS concentrations are independent, then
the standard deviation of a 30-day average (30 sam-
ples) would be
12.15 -s- ^30 = 2.22 mg/l.
However, for these same plants, the coefficient of var-
iation for a 30-day average is reported to be 0.45, yield-
ing a standard deviation of
(14.3)(0.45) = 6.43 mg/l
for the 30-day TSS average. This value is almost three
times larger than that value based on the assumption
of independence, which indicates that the daily TSS
levels are strongly correlated. Because the actual stan-
dard deviation, which incorporates the correlation, is
so much larger than that based on the independence
assumption, it must be reflected in the analysis. Other-
wise, the results would be in substantial error.
The theoretical framework for including correlation is
as follows. Recall that if the random variables Xi, X2,
... , Xn are independent with variance a2 then
Var(X) = trj/n
where X is the average of X,, X2, ... , Xn. If the X, are
correlated then
_ n-1
Var(X) = ff'[1/n + 2/n2 £ (n-j)Cjl
i=i
where e, is the correlation coefficient between X, and
Xl+). When the e, are positive they have the effect of
increasing the variance of X. The QJ, in principle, range
from - 1 to + 1, but in this application the coefficients
are almost always positive. The e, are defined by
[X, -
If the correlation coefficient e, = 1, then X, = Xl+j, but
if e, = 0, then X, and X,+, are independent or uncor-
related. As ej increases from 0 to 1, then Xl+j tends
towards X|. In the extreme case of e, = 1, there is no
reduction in variance due to averaging.
Having an estimate of Var(X), one can compute a
monthly average limit using the formulas to compute
the monthly limit given previously. For example, con-
sider a monthly average limit based on an assumed 30
samples per month. We first would compute from the
numerical illustration
daily variance of X,
ne ~ actual variance of 30-day average
= (12.15)2/(6.43)2
= 3.57
and use the value ne in the lognormal formulas rather
than n = 30. Although there are actually 30 samples
included in this 30-day average, there are only an
equivalent of ne - 3.57 independent samples because
of the strong correlation. Using the value ne instead of
n incorporates the correlation into the calculations
involving an average. Note, if the Xi are uncorrelated
then ne = n.
This general approach allows one to incorporate depen-
dence into distributional calculations for an n-sample
average by first computing
__ daily variance _
e ~ actual n-sample average variance
where these variances are of the actual pollutant levels
in the observation-space, not the log-space.
E-8
-------
Estimating ne from Correlated Data
To compute the actual n-sample variance when daily
plant data are available, say X,, X2, ... , Xk, one can
compute the cross-correlations and utilize the formula
for the variance given previously in this subsection. If
for instance n = 10, and if we assume a sample is taken
every third day, the cross-correlations Q3, QG, ... , Q30
would be required.
Alternatively, one can very much reduce the computa-
tion and data requirements by introducing a simple
model that states that the effluent concentration on
day i is linearly related to the previous day's concentra-
tion plus a random variation. This model is expressed as
xi = e xi-i + Cj-i,
where e is an independent error term. This model (first-
order autoregressive model — see Reference [6]) has
been applied to estimate effective average sample sizes
for effluent data [5,7]. Using this model one can esti-
mate ne directly by
K
n(1-e2) - 2e(1-cn) '
where Q, the lag-1 correlation, is estimated by
k-1
(x, -x)(xl+1 -x)
This formulation is for the n samples taken on succes-
sive days during the month. It, however, can be modi-
fied. For example, assume that n samples are taken
during the month and a sample is taken every third day.
Then the adjusted formula is
dO)2(1-e3)2
10(1 -
- e30)
which is obtained by substituting e3 forg because for
the given model the correlation between X, and Xi+j is
e1- Suppose, for instance, Q = 0.80 is an estimate of
the lag-1 correlation derived from daily data. Then
n = (100X0.238)
e 7.378 - 1.022
= 3.74
The 10-sample average has an effective sample size of
3.74.
In the subsection entitled Monthly Permit Limit we
calculated a monthly 95th percentile lead limit of
0.85 mg/l based on an assumed 10 samples a month.
That calculation was based on the data being indepen-
dent. If, however, the data are correlated and the effec-
tive sample size of ne = 3.74, we have
and
83.74 = 0.639,
V3.74 = -0.974,
and the 95th percentile monthly limit becomes
exp(-0.974 + (1.645X0.639)) = 1.08 mg/l.
The effect of this correlation is to increase the monthly
limit from 0.85 mg/l to 1.08 mg/l. If daily monitoring
were assumed (i.e., n = 30), the relative increase due
to the correlation would have been substantially larger.
Conversely, if a low sampling frequency were assumed,
say n = 4, the difference between assuming independ-
ence and incorporating correlation would have been
slight.
References
1. Benjamin, J. R., and C. A. Cornell. 1970. Probabil-
ity, Statistics and Decision for Civil Engineers.
McGraw-Hill, Inc.
2. Aitchison, J., and J. A. C. Brown. 1963. The Log-
normal Distribution. Cambridge at the University
Press.
3. Schleher, D.C. 1977. Generalized Gram-Charlier
Series with Application to the Sum of Log-
Normal Variates. EEE Tranactions on Information
Theory.
4. U.S. Environmental Protection Agency. Technical
Guidance Manual for Performing Waste Load
Allocations, Book VII, Permit Averaging Periods.
Monitoring and Data Support Division, Contract
68-03-3131-WA9, Washington, D.C.
5. Rossman, L. A. 1984. Memorandum: Variance
Reduction for n-Day Averages of Autocorrelated
Data. U.S. EPA, Cincinnati, Ohio, to H. Biswas,
U.S. EPA, Monitoring and Data Support Division,
Washington, D.C., June 5, 1984.
6. Box, G. E. R, and G. M. Jenkins. 1976. Time Series
Analysis: Forecasting and Control. Revised Edi-
tion, Holden-Day.
7. U.S. Environmental Protection Agency. 1983.
Development Document for Effluent Limitations
Guidelines and Standards for the Organic Chem-
icals and Plastics and Synthetic Fibers Industry:
Proposed. EPA 440/1-83/009-b, February, 1983,
pp. B-20-22.
E-9
1}. U . S . GOVERNMENT PRINTING OFFICE. I 98 7 - V 8 1 - ! 9 7 / 5 2 9 5 0
-------
U.S. Environmental Protection Agency
Region 5, Library (PL-12J)
77 West Jackson Boulevard, 12th Floor
Chicago, IL 60604-3590
------- |