vvEPA
United States
Environmental Protection
Agency
Office Of Water
(4204)
EPA 833-B-96-001
.April 1996
Interim Guidance For
Performance-Based Reduction
Of NPDES Permit Monitoring
Frequencies
-------
-------
UNITED STATES ENVIRONMENTAL PROTECTION AGENCY
WASHINGTON, D.C. 20460
APR I 9 1996
MEMORANDUM . '
SUBJECT: Interim Guidance for Performance - Based Reductions of
NPDES Permit Monitoring Frequencies
FROM: Robert Perciasepe, Assistant
Office of Water
A. Herman, Assistant Aominist
J&fficG of Enforcement and Complianc
TO: Regional Administrators
Regional Water Division Directors
Regional Counsels
' •* i • - * ' ,
We are pleased to transmit for your use, this interim
guidance for reducing reporting and monitoring under the NPDES
permit program.
This interim guidance helps to fulfill one of the main
directions in the President's Regulatory Reinvention Initiative
for EPA — reducing unnecessary reporting while at the same time
maintaining a high level of environmental protection for the
Nation.
NPDES authorities can grant relief to regulated facilities
that have a record of good compliance and pollutant discharges at
levels below permit requirements. This relief provides
incentives for voluntary reductions of pollutant discharges
through such means as reuse and recycling.
This interim guidance is the culmination of extensive work
among our offices, several Regions and States and consultation
with outside stakeholders representing industry, the
environmental community, and municipalities. These stakeholders
generally favor the approach, which has benefitted considerably
from their input.
Rrcyctod/ftoeyctafeto
MMMI wtth Soy/CM» M •> M*«
contain* tt
-------
-------
-.. :• - - 2 --.,'•
We encourage you to begin now to implement this interim
guidance through the regular NPDES permit issuance process where
EPA has permit authority, and to work with your NPDES States to
adopt this policy as soon as possible. EPA Region VI will soon
begin two pilot projects in Oklahoma and Louisiana to assess the
strengths and issues associated with the guidance. Based on the
results of these pilot projects, we will make revisions to this
interim guidance as necessary.
We look forward to working with you on this important
endeavor.
cc: Robbi Savage, ASIWPCA
Robbie Roberts, ECOS
-------
-------
PERFORMANCE-BASED REDUCTION OF MONITORING FREQUENCIES
INTERIM GUIDANCE FOR PERFORMANCE-BASED REDUCTION
OF NPDES PERMIT MONITORING FREQUENCIES
Introduction
The President's Regulatory Reinvention Initiative for the Environmental Protection Agency
(EPA) established an interim goal of reducing reporting and monitoring by at least 25%.
This goal is also embodied in the Office of Water's Agenda for the Future, which sets forth
. program priorities for the coming years for EPA and States.
Based on these directions, EPA's Offices of Water and Enforcement & Compliance
Assurance developed this Interim Guidance to reduce regulatory burdens associated with
reporting and monitoring based on a demonstration of excellent historical performance by
facilities subject to NPDES permit requirements. Under this,guidance, facilities can
demonstrate this historical performance through both compliance and enforcement history and
a demonstrated ability to consistently reduce pollutants in their discharge below the levels
necessary to meet existing permit requirements. Facilities will also be expected to maintain
these performance levels to continue to receive the reductions. Reducing burdens in this
manner will also provide incentives for voluntary reductions of pollutant discharges through
such means as reuse and recycling.
The approach for determining the degree of burden reduction available to individual facilities
is statistically sound and will not reduce the ability of EPA and States to determine non-
compliance with permit requirements. .
This guidance should also prove useful in setting monitoring frequencies for industrial users
of POTWs. EPA has not studied whether the variability of industrial users' effluent is
similar to that for NPDES permittees. Pretreatment control authorities may choose to apply
.this policy to their industrial users with effluent similar to that discussed in this guidance.
Future reductions to NPDES that can be integrated into this burden reduction initiative
include ongoing ambient monitoring efforts within the Office of Water.
Summary of Decision-Making Process
The guidance applies to both major and minor individual NPDES permits for direct
.discharges and will be implemented through the existing NPDES permitting cycle for
facilities.
INTERIM GUIDANCE APRIL 1996 -1-
-------
PERFORMANCE-BASED REDUCTION OK MONITORING FRKQUENCIKS
The following steps are to be taken when determining if a particular facility is eligible tor
reductions, and, if so, the amount of these reductions. These steps are also described in
more detail in the next section of the guidance.
1) Facility1 Enforcement History
Each facility's enforcement history is analyzed to assess eligibility for reductions under the
guidance. Criminal convictions under any environmental statute and NPDES civil judicial
and administrative enforcement actions are criteria considered in determining eligibility.
2) Parameter-by-Parameter Compliance History
For each eligible facility the compliance history for each .parameter controlled in its existing
permit is examined for Significant Noncompliance violations and/or effluent violations for
critical parameters. These critical parameters are determined at the discretion- of the
permitting authority and could include pollutants which pose a higher risk to human or
environmental health. ,The results of this examination determine which parameters are
eligible for monitoring reductions.
3) Parameter-by-parameter Performance History
The permitting authority then calculates, for each eligible parameter, the two-year composite-
average at each outfall. The composite average is compared with the permit limit, and the
information in .Table 1, which is based on the existing monitoring frequency, to determine
the potential monitoring frequency reduction.
4) Continued Eligibility for Reductions
EPA and States would continue to monitor each parameter for significant noncompliance and
any effluent violations of critical parameters, failure to submit DMRs, and any new
enforcement actions. If violations based on these do occur, the permitting authority may
require increased monitoring in accordance with a Section 308 or 309 order (or State
equivalent).
5) Future Reductions for Ambient Monitoring
Based on the facility's agreement to participate in an ambient monitoring program, along
with other stakeholders in a watershed, additional reductions could be provided, at the
discretion of the permitting authority.
The term "facility" as used in this document refers to the regulated entity.
INTKRIM GUIDANCE APRIL 1996 -2--
-------
PERFORMANCE-BASED REDUCTION OF MONITORING FREQUENCIES
Timing of Decisions
f
Monitoring reductions should be considered during permit reissuance. Reductions based on
facility performance may also be considered if the permit is reopened to accommodate other
issues. The permitting authority may, at their option, modify the permit solely to reduce
monitoring requirements if sufficient resources are available. Monitoring requirements are
not considered effluent limitations under section 402(o) of the Clean Water Act, and
therefore anti-backsliding prohibitions would not be triggered by reductions in monitoring
frequencies.
Permit monitoring requirements may, at the permit issuing authority's option, contain
conditions for decreases in monitoring if specified performance conditions are met and/or
require increased monitoring if performance levels drop. Although such conditions have
sometimes been used in NPDES permits in the past, these conditions cannot now be tracked
in the Permits Compliance System .(PCS) data base system, if the permitting authority has
sufficient resources to manually track changed reporting frequencies, such provisions could '.
be included in the permit when the monitoring frequencies are adjusted based on changed
performance. Increased monitoring requirements if performance levels are not maintained
will be incorporated through enforcement orders under Sections 308 or 309 of the Clean
Water Act (or State equivalent).
Entry Criteria for Participation
1) Facility Enforcement History
Criminal 'Actions (all environmental statutes) '
Facilities which have been criminally convicted under any Federal or State
environmental statute .of falsifying monitoring data or committing violations which
presented an imminent and substantial endangerment to public health or welfare will
not receive any reductions at any time in the future2.
• Facilities convicted of any other criminal violation under any Federal or State
environmental statute will not receive any reductions for five years.
Whenever the permit writer, on a case-by-case basis, determines that there has been a wholesale
change in ownership and management, that facility may become eligible for consideration under this
guidance as a new permittee.
INTERIM GUIDANCE APRIL 1996 -3-
-------
PERFORMANCE-BASED REDUCTION OF MONITORING FREQUENCIES
. Reductions will be available for those facilities where an individual employed by the
permittee, but not the permittee itself, was convicted of a criminal violation under any
Federal or State environmental statute, provided the permittee discovered and self- •
disclosed' the violation, and took prompt action to correct the root cause in order to
prevent future criminal violations.
Civil Judicial Actions (Clean Water Act/NPDES related) .
• Facilities ate eligible for consideration of reductions 1 year after completion of
injunctive relief and payment of penalty.
Administrative Actions (Clean Water Act/NPDES related)
• Facilities are eligible for consideration after the permittee has complied with
Administrative Penalty Order (APO) or Administrative Order (AO) (including State
equivalent) requirements, and payment of any assessed penalty.1, A permittee that is
issued an AO, in conjunction with reissuance of its permit, to extend a compliance
schedule, may be eligible if the permittee is in compliance with the interim milestones
and schedule in the AO.
For example, in order to comply with a newly promulgated effluent guideline, an
industrial sector may be required to install a new technology. Some facilities may not
be able to attain the new technology immediately so an AO is issued at the time the
facility's permit is reissued. The AO sets a compliance schedule to allow the
permittee additional time to install the technology needed to meet the new effluent
guideline limitation.
2) Parameter-by-Parameter Compliance
The permitting authority will examine each of the following entry criteria:
Significant Noncompliance for Parameters under Consideration
• A facility may not have had any Significant Noncompliance (SNC) violations for the
parameters for which monitoring/reporting reductions are being considered during the
last two years and,
Any Effluent 'Violations of Selected Parameters
• A facility may not have had any effluent violations of selected (critical) parameters
during the last year. The "selected parameters" can be permit-specific and would be
determined at the discretion of the permitting authority. These parameters could
INTERIM GUIDANCE APRIL 1996 -4-'
-------
PERFORMANCE-BASED REDUCTIPN OF MONITORING FREQUENCIES
include pollutants which pose heightened risks to human or environmental health,
.such as highly toxic or bioaccumulative compounds, • - .'
3) Parameter-by-Parameter Performance History
• At a minimum, the two most recent years of monthly average effluent data
representative of current operating conditions for the parameter at the particular
outfall wjill be used to calculate the long term average discharge rate for use in Table
i.. - -, ' ' ••;•'-•••.':.'
• The baseline monitoring, frequencies in Table 1 of this guidance will normally be
considered the.level of monitoring in the existing effective NPDES permit. It is
important to recognize that permittees that receive monitoring frequency reductions in
accordance with Table 1 or Table 2 are still expected to take all appropriate measures
to control both the average level of pollutants of concern in their discharge (mean) as
well as the variability of such parameters in the discharge (variance)', regardless of
any reductions in monitoring frequencies granted from the baseline levels. Reliance
on monitoring the discharge at a reduced frequency as the sole means of tracking and
controlling the discharge could increase the risk of violations.
Table!
Ratio of Long Tertn Effluent Average
to Monthly Average Limit
Baseline .
Monitoring 75-66% 65-50% .49-25% , <25%'
7/wk 5/wk 4/wk 3/wk 1/wk .
6/wk 4/wk 3/wk 2/wk 1/wk
5/wk 4/wk 3/wk 2/wk 1/wk
4/wk 3/wk 2/wk 1/wk 1/wk
3/wk 3/wk 2/wk 1/wk 1/wk
2/wk 2/wk 1/wk 2/mo ' 1/mo
1/wk 1/wk , 1/wk 2/mo l/2mos
2/month 2/mo 2/mo 2/mo I/quarter
I/month 1/mo 1/mo ,\ I/quarter l/6mos
Note: See above eligibility requirements.
INTERIM GUIDANCE APRIL 1996 -5-'
-------
PERFORMANCE-BASED REDUCTION.OK MONITORING FREQUENCIES
New permittees should go through one permit cycle (5 years) before being eligible for
consideration for reduced monitoring.
Facilities would not normally be considered for reductions in monitoring frequencies
below once per quarter, except in unusual circumstances of reliable performance at
the requisite levels and outstanding compliance/enforcement histories.
Facilities which satisfy the entry criteria but are not experiencing discharges of 75%
or less of their permitted levels of water quality-based parameters may still be eligible
for reductions in monitoring/reporting frequencies at the discretion of the permitting
authority. To control an increased risk of undetected violations, monitoring should
only be reduced for such parameters if the applicant can demonstrate a very low ,
variation in the concentrations being discharged.
Parameters that show a long-term (2 year) average discharge between die permitted
concentration and 76% of a water quality-based permit limit should demonstrate a
coefficient of variation (ratio of standard deviation to average) of 20,% or less. An
additional safeguard should stipulate that parameters which showed any exceedance of
the monthly average limitation during the two year averaging period would not be
subject to monitoring reductions. It should be noted that discharges with a long-term
average at or near the permit limit have a probability of reporting a violation 50% of
the time, regardless of low coefficient of variation or sample size. Reductions may
be made as shown in Table 2 below:
Table 2
Ratio of Long Term Effluent Average to Monthly Average Limit
100-76%
Baseline .Reduced
Monitoring Monitoring
7/wk 6/wk
6/wk 5/wk
5/wk . 4/wk
4/wk 4/wk
3/wk , " 3/wk
2/wk ' 2/wk
1/wk 1/wk
2/month 2/month
1/month 1/month
INTERIM GUIDANCE APRIL 1996 -6-'
-------
PERFORMANCE-BASED REDUCTION OK MONITORING FREQUENCIES
4) Residency Criteria for Continued Participation
• Permittees are expected to maintain the performance levels that were used as the basis
for granting monitoring reductions. To remain eligible for these reductions, the
permittee may not have any SNC violations for effluent limitations of the parameters
for which reductions have been granted or failure to submit DMRs, or may not be
subject to a new formal enforcement action. For facilities that do not maintain
performance levels, the permitting authority may require increased monitoring in
accordance with a Section 308 or 309 Order (or State equivalent).
Special Considerations {
Discontinuous data: Monitoring should not be reduced using the methodology described
above if effluent data have not been continuously reported over the period of time being •
considered. Effluent averages from interrupted or discontinuous data :sets may not be
representative of long-term performance. Monitoring frequencies for discharges that are
intermittent or short-term, such as seasonal discharges and highly variable, batch processes,
should not be assessed or reduced using the methods described in this guidance and would
need to be considered on a case-by-case basis.
Independent/Dependent Control Parameters: The procedures for reductions described in
this guidance are intended for effluent parameters which are normally independently
controlled by the permittee., That is, for each parameter limited in the permit there should be
significantly different control mechanisms/factors-either in the permittee's treatment,
pretreatment, or process operations. In situations where there are several parameters, each
of which could be used to measure the performance of a given system, it will generally be
appropriate to primarily, monitor only the best indicator parameter. For example, if a
biological treatment system can be evaluated by either BOD, CBOD, COD, or TOC
measurements; it would be normally appropriate to require monitoring of only one of these
oxygen demanding parameters. -
The permitting authority should, therefore, examine the parameters being monitored from
each facility during the permit issuance process to establish which parameters are
independently controlled and/or which can be used to determine the proper operation of a
facility. Monitoring of other parameters can be either eliminated or reduced to a minimum
frequency. , '
Monitoring Frequency "Floor": Current federal NPDFiS regulations do not establish a-
monitoring frequency."floor" but do'establish a reporting frequency floor of once/per year.
The monitoring, frequency from which reductions could be made in this guidance is
considered to be the level of the monitoring in the existing effective NPDES permit. It is
important to recognize that the guidance given in Table 1 does not advocate any reductions in
INTERIM GUIDANCE APRIL 1996 -7-
-------
PERFORMANCE-BASED REDUCTION OF MONITORING FREQUENCES
statistical confidence in the ability of a permitting authority to determine whether or not a
permit limit is being violated at reduced monitoring frequencies. The guidance also does not
advocate any reductions for parameters that are currently monitored only once/quarter:
The permitting authority may, however, consider other factors specific to the State or
facility. For example, a State policy may establish the baseline. If a facility has already
been given monitoring reductions due to superior performance, the baseline may be a
previous permit. As a point of reference, Federal regulations do not stipulate minimum
monitoring frequencies but do require that reporting cannot be less than once per year.
Future national guidance may also be used to establish a baseline for monitoring.
Exceptions: The permitting authority may elect to maintain higher monitoring levels in •
individual situations where there may be a particular interest in human health, endangered
species, or a sensitive aquatic environment. An example would be where a permitting
authority has assessed water quality problems in a watershed and determined which point and
nonpoint sources are particularly critical from the standpoint of protection of aquatic
resources (e.g., endangered species) and .human health (e.g., drinking water source), The
permitting authority may well decide not to reduce monitoring of critical point sources in
these instances, while continuing to monitor -the overall situation.
Applicability to Minor Facilities: Minor facilities are fully eligible for reductions under
this guidance, even though they are not automatically tracked for SNC in the Permits
Compliance System Database. (Avoidance of SNC is one of the minimum criteria that
should be met for participation in this program.) However, permitting authorities may apply
the SNC criteria on a case-by-case basis to minor facilities in order to allow them to
participate in this program based on permit-specific effluent compliance.
Implementation of Guidance: Where EPA is the permitting authority, it would apply this
guidance upon permit reissuance, and consider at that time, whether reductions in monitoring
and reporting frequencies were appropriate based upon the compliance/enforcement arid
performance history of the facility. EPA does not possess adequate resources to routinely
reopen, modify, and reissue currently effective permits to revise monitoring frequencies.
However, individual permitting authorities may elect to reopen and modify permits to reduce
monitoring frequencies consistent with this guidance if resources permit.
Limits below Levels of Detection: This guidance does not recommend reductions in
monitoring frequencies in cases where stringent water-quality based limits (WQBELs) are
below levels of quantitation (the level at which a constituent present in a wastewater sample
can be reliably detected and quantified). Permittees with these types of limits will normally
be deemed to be in compliance when monitored levels are below the level of quantitation;
however, by definition, it is not scientifically, possible (until analytical methods improve) to
certify that the WQBELs are actually being achieved. Thus, EPA feels it would be ...
inappropriate to develop national guidance establishing reductions from established
INTKKIM GUIDANCE APRIL 1996 -8-
-------
PERFORMANCE-BASED REDUCTION OF MONITORING FRKQUKNCIES
monitoring frequencies for these types of limits. However, individual permitting authorities
may still use their discretion in considering reductions on a case-by-case basis.
Use of Daily Maximum Values: .This guidance does not provide a specific methodology for
considering daily maximum permit values when considering monitoring/reporting reductions.
However, EPA is in the process of implementing a revised definition of SNC that accounts
for daily maximum violations. The new definition will be included in the entry criteria of
this proposal. In the interim, permitting authorities should consider such situations on a
case-by-case basis. There may be concerns over instances where, for example, there are
acutely toxic conditions in a receiving water due to violations of daily maximum permit
limitations. In such cases, the permitting authority may elect to maintain higher monitoring
levels. In addition, it is important to recognize that dischargers who frequently violate daily
maximum permit Limitations will likely be unable to achieve high levels of performance in
monthly average limits and effectively would not be eligible to participate in this program on
that basis. In addition, such facilities may also trigger one of the various
compliance/enforcement-based entry criteria. . '• '
Applicability of this program to indirect users of POTWs: Many elements of the national
Pretreatment program parallel the NPDES permit program. In general, therefore, the same
overall logic embodied in this guidance may be extended to industrial users of POTWs (Ills),
where appropriate. However, EPA has not investigated whether monitoring data of industrial
users of POTWs (lUs) can be characterized with similar coefficients of variation. (Tables 3,
4, and 5 were generated for facilities with coefficients of variation of 20%, 60%, and 80%,
respectively.)
Where monitoring frequencies are already near the minimum required by regulation (e.g.,
twice per year for significant industrial users), the reductions in this guidance would not
apply. EPA has begun a dialogue among State and EPA Regional Pretreatment Coordinators
to more fully discuss possible pilot projects and statistical analyses.
Incentives for Ambient Monitoring: This interim guidance focuses primarily on criteria for
reducing reporting and monitoring used for determining compliance with NPDES permit
requirements. It is our intention to reduce burdens associated with these activities where
good compliance and permitting performance can be demonstrated and maintained. Another
important policy direction for EPA and State water programs is the need to focus our
resources more effectively on the problems facing individual places. This Community Based
Environmental Protection (CBEP) strategy is embodied through our watershed protection
approach. One of the most important aspects of a successful watershed protection approach
is to get the best possible monitoring information on the conditions, causes and sources of
impairment, and relative impact of these sources on the overall health of a watershed and the
effectiveness of our control actions in a watershed. The approach described below for
obtaining ambient monitoring information from point sources will also help provide important
linkages among other important activities such as more comprehensive of our waters under
INTERIM GUIDANCE APRIL 1996 -9-
-------
PERFORMANCE-BASED REDACTION OK MONITORING FREQUENCIES
Section 305(b), effluent trading in watersheds, and improved Total Maximum Daily Load
(TMDL) analyses.
This information needs to be gathered and used, where available, from a variety of sources,
including municipal and industrial point source dischargers: These point sources could
provide a great deal of valuable ambient monitpring information that could be very helpful in
making better watershed-based decisions. While certain information may be unique to an
individual watershed, there needs to be a core group of environmental indicators, such as
attainment of designated uses in State water quality standards and fish consumption
advisories, that each watershed will need to measure. NPDES dischargers could often
provide valuable information to help measure these core indicators of the overall health of
the watershed. ' " .
Therefore, in order to encourage NPDES dischargers to Voluntarily provide this information
or collect additional ambient monitoring information, permitting authorities may consider .
granting additional reductions in compliance reporting and monitoring/over artd above the
reductions granted based on good performance if permittees agree to collect or provide
additional ambient monitoring information. Prior to granting these additional reductions,
permitting authorities should reach agreements with the dischargers on how this information
will be provided or collected and how it will be used to give all key stakeholders a better
picture of the overall health of the affected watershed. The amount of additional reduction
will be at the discretion of the permitting authority who should work collaboratively with
State and watershed agencies who design and implement monitoring programs to support
environmentally based decisions. This closer integration of ambient and compliance
monitoring may also be included in EPA/State agreements to support the National
Environmental Performance Partnership System (NEPPS).
Finally, any additional reductions provided should be done so in a manner consistent with the
framework and other criteria described in this guidance.
Future Actions
The burden reductions recommended under this guidance will be available immediately.
Over the next 12-18 months, EPA will also conduct detailed pilot studies in two States,
Louisiana, and Oklahoma, to closely monitor implementation of the guidance. Based on
information from these pilot studies and other information, EPA will consider modifications
to this interim guidance as appropriate.
INTERIM GUIDANCE APRIL 1996 -10--
-------
PERFORMANCE-BASED REDUCTION OF MONITORING FREQUENCIES
Supporting Statistical Study
Effect of Sample Size on Probability of Violation
EPA has done a statistical analysis on the effect of sampling frequency on compliance
assessment. The basic premise underlying a performance-based reduction approach is that
maintaining a low average discharge relative to the permit limit results in a low probability
of the occurrence of a violation for a wide range of sampling frequencies.
The probability of the occurrence of a violation of a monthly average permit limit was
calculated. Tables 3, 4 and 5 display the percentage of time that a monthly average permit
violation will be reported given sample size and a long-term average to permit ratio. This
probability is dependent on the true long-term average of the .discharge, the'piermit limit, and
the monthly sampling frequency. The variables of long-term average and permit limit are
both reflected in the tables by expressing these as a ratio. Tables 3, 4, and 5 assume a .
normal distribution of monthly averages and show the effect of altering the assumed
coefficient of variation, using 20%, 60%, and 80%, respectively.
Obviously, the best estimate of the true monthly average discharge is obtained by daily '_
sampling. One can assess the true violation rate of a discharge by looking at the probability
calculated assuming sampling was done daily (30 times per month). In order to maintain
compliance with a permit limit, the long term average level of the discharge must be
controlled at a level less than the permit limit. Reducing the sample size, while increasing
the probability that a violation will be reported, does not change the underlying probability of
reporting a violation associated with a baseline estimate of the monthly average, calculated
with 30 samples. With a constant performance, the probabilities of reporting a permit
violation increase as the sample size is reduced from daily sampling because the variance of
the average is inversely .proportional to the sample size.
Looking at the true violation rate of a facility sampling daily and operating at 75% of their
permit limit, these tables show that the probability oTa violation in a given .month is 1 % or
less. If the long-term average discharge is 65% of the permit limit, the true percentage of
violation is less than 1%.' As sample size decreases for a given discharge/limit ratio, the
expected percentage of time that the average of the samples collected during the month will
exceed the permit limit increases. For example, Table 5 demonstrates that at a ratio of 65%,
the expected violation rate is effectively zero. If a subsample of 8 samples per month is
taken instead of 30, the facility has a 3% chance of reporting a violation. If only one sample
per month is taken, the chances of reporting a violation increase to 25%. The facility
performance (true monthly average discharge) has not changed, thus "missed." monthly
average violations are not an issue. The probabilities calculated for very low sampling
ATTACHMENT TO INTERIM GUIDANCE A-L
-------
PERFORMANCE-BASED REDUCTION OF MONITORING FREQUENCIES
frequencies reflects the risk assumed, by the discharge operator that monthly average
violations will be reported when in fact the process average is under permit limit. If facility
performance degrades during the permit term and sampling has been reduced, it can be seen
that the facility will have probability of reporting violations at a higher rate, even if the long-
term average is still below the permit limit. An example will illustrate this point. Table 5
shows that if a facility was'judged to be at 75% of their permit limit and reduced sampling
from 16 to 12 times per month, the probability of violation would change from
approximately 5% to 7%. If the long-term average performance degraded to 90% of the
permit limit, the 12 monthly samples would yield expected monthly average permit violations
32% of the time instead of 29.% of the time if 16 samples were collected.
Table 5 shows probabilities calculated using a more conservative assumption of 80%
coefficient of variation. The results show that facilities with a long term average of less than
or equal to 75% have essentially no chance of violating a monthly average limit, hence
facilities with this performance would be good candidates for performance-based monitoring
reductions. The reductions in Table 1 were designed to maintain approximately the same
level "of reported violations as that experienced-with their current, (baseline) sampling.
ATTACHMENT TO INTERIM GUIDANCE
A-2.
-------
PERFORMANCE-BASED REDUCTION OF MONITORING FRKQ'UKNCIJO
Table 3
Probability of Reporting Monthly Average Permit
Violations at 20% Effluent Variability
(CV = 0.20; Normal Distribution)
• Monthly Sample Size
'LT A/Permit
100%
•95%
90%
85%
80%
75%
70%
65%
60%
; 55%
• 50%
40%
30%
20%
30
50%
7%
0%
0%
0%
0%
0%
0%
0%
0%
0%
0%
0%
0%
28
50%
8%
0%
0%
0%
0%
0%
0% •
0%
0%
0%
0%
0%
0%
24
50%
10%
0%
0%
0%
0%
0%
0%
0%
0%
0%
^0%
0%
0%
20
50%
12%
1%
0%
0%
0% ,
0%
0%
0%
0%
0%
0%
o%.
0%
16
50%
15%
1%
0%
0%
0%
0%
0%
0%
0%
0%
0%
0%
0%
12
50%
18%
3%
0%
0%
0%
0%
0%
0%
: 0%
0%
0%
0%
0%
8
50%
23%
6%
1%
0%
0%
0%
0%
0%
0%
0%
0%
0%
'0%
4
50%
30%
13%
4:%
1%-
0%
0%
0%
0%
0%
0%
0%
0%
0%
2
50%
35%
22%
1-1 fc
4% ;
1%
0%
0%
0%
'0%
0%
0%
0%-
0%
1
50%
40%
29%
19%
11%
5%
2%
0%
0%
0%
0% •'-
0%
0%
0%
1 Ratio of calculated average of at least 2 years of effluent data to monthly average permit limit.
ATTACHMENT TO INTERIM GUIDANCE
A-3
-------
PERFORMANCE-BASED REDUCTION OF MONITORING FRKQUENCIKS
Table 4
Probability of Reporting Monthly Average Permit
Violations at 60% Effluent Variability
(CV = 0.60; Normal Distribution)
Monthly Sample Size
'LTA/Permit
100%
95%
90%
85%
80%
75%
70%
65%
60%
55%
50%
40%
30%
20%
30
50%
32%
16%
5%
1%
0%
0%
0%
0%
0%
0%
0%
0%
0%
28
50%
32%
16%
6%
1%
0%
0%
0%
0%
0%
0%
0%
0%
0%
24
50%
33%
18%
7%
2%
0%'
0%
0%
0%
0%
0%
0%
0%
0%
20
50%
35%
20%
9%,
3%-
1%
0%
0%
0%
0%
0%
0%
0%
0%
16
50%
36%
23%
12%
5%'
1%
0%
0%
0%
0%
0%
0%
0%
0%
12
50%
38%
26%
15% •
7%
3%
1%
0%
0%
0%
0%
0%
0%
0%
8
50%
40%
30%
20%
12%
6%
2%
1%
0%
0%
0%
0%
0%
.0%
4
50%
43%
36%
.28%
20%
13%
8%
4%
1%
0%
0%
0%
0%
0%
2
50%
45%
40%
34%
28%
22%
16%
10%
6%
3%
1%
0%
0%
0%
1
50%
47%
43%
38%
34% .
29%
24%
18%
13%
9%
5%
1%
0%
0%
Ratio of calculated average of at least 2 years of effluent data to monthly average permit limit.
ATTACHMENT TO INTERIM GUIDANCE A-4
-------
PERFORMANCE-BASED REDUCTION OF MO\ITORI\G FRK
-------
PERFORMANCE-BASED REDUCTION OF MONITORING FREQUENCIES
Detailed Protocol for Calculating Probability of Reporting Permit
Violations
Calculation of probabilities for Tables 3-5
Probability distributions may be used to model effluent data and -assess the probability of
permit violations. The models provide a logical and consistent methodological framework
for using observed performance data to assess permit limitations in an objective manner.
The goal of the limitations is to establish performance levels that enforce good treatment and
ensure that water quality objectives are met. In deriving limitations, sufficient allowance for
variation in treatment performance is provided such that a well-operated treatment system
should be capable of compliance with the limitations at all times. In using probability
models as the basis for limits, it is necessary to select a percentile value such that, within the
context of the model, any meaningful limit will have a non-zero probability of being
exceeded. . •
1 • * •*,
The results shown in the tables here are. derived from probability distribution functions that
may be used to model effluent data. That is, the processes are assumed to operate over time
in a manner that is consistent with past performance. No intervention to change the process
or exert more or less control over the discharge is assumed.
Calculation of the probability that a reported permit violation will occur depends upon: the
number of individual samples taken during the month, the long-term discharge level, the
variance of the discharge concentrations, the probability distribution of the individual samples
during the'month, and the permit limit.. There are two probability distributions commonly
used to model effluent data: the lognormal distribution and the normal distribution. The
lognormal distribution usually provides a good fit to data sets'comprised of individual
effluent measurements because such data typically have two critical lognormal characteristics:
they are positive valued and positively skewed. Positive skewness means that the data are
characterized by a tendency for a preponderance of measurements in the lower range of
possible values with relatively fewer measurements stretched out over a wider range of
possible upper values. The lognormal also has the property that the logarithms (natural or
base 10) of the data are normally distributed. The normal distribution has the well-known
"bell shape" and is mathematically straightforward so that working with the logarithms of
effluent data is relatively uncomplicated.
The asymptotic distribution of sample averages is normally distributed. That is, the average
of a sample of individual measurements will have a distribution that is approximately
normally distributed regardless of the distribution of the individual measurements. The
quality of the approximation depends on several factors including the number of individual
measurements being averaged and the form of the underlying distribution. Although
individual effluent measurements are rarely normally distributed, it is reasonable in many
situations to approximate the distribution of the averages of effluent measurements with a
normal distribution and thus the normal approximation is used in many cases as a model for
monthly average effluent limitations. The results in Tables 3-5 are based on the assumption
ATTACHMENT TO INTERIM GUIDANCE A-6
-------
PERFORMANCE-BASED REDUCTION OF MONITORING FKKQUE.VCJKS
of a normal distribution for the averages of effluent measurements. Extensive discussion on
the statistical modelling of effluent data and methodology for setting effluent limitations are
contained in EPA's 1991 Technical Support Document for Water Quality-based Toxics'.'
Control (TSD).
The results of calculating probability of a reported violation of a monthly average permit
limit are shown, in Tables 3 through 5 under different conditions. The purpose of these
tables is to provide some insight into the effects of changing monitoring requirements. The
probability of exceeding the monthly limit when the long-term average of the discharge is at
the desired value can be thought of as the Type I error rate (alpha-level) of the monitoring
program. When the long-term average exceeds the desired limit, the probability of
exceeding the monthly limit is now the monitoring program's ability to detect violation
increases if the long-term average increases over the desired level. It should be understood
that if permit limits are held constant and performance measures such as long term average
discharge and variability of treatment do not change, then reducing the number of monitoring
measurements used to calculate the monthly average causes the probability of a violation to
increase for all values of the long term average less than the monthly average permit limit.
This has a two-fold effect: 1) the chances of reporting a violation even when the lorig term
average is less than the desired level (the Type I error rate) go up 2) the sensitivity (ability to
detect violations) of the program increases. The Tables also show that if the average
discharge level is held well below the monthly average limit, the chances of a violation are
small. The thee tables reflect three different levels of variation in the underlying daily data
as measured by the coefficient of variation. The coefficient of variation (CV) is the ratio of
the standard deviation of the distribution to the mean and is often expressed as a percentage.
The CV is a convenient measure for summarizing the relative variability in a data set. The
results in Tables 3,4, and 5 use CVs of 20%, 60% and 80% respectively. A coefficient of
variation of 60% was used in the TSD to describe a typical level, of variation for lognormally
distributed effluent data.. CVs of 80% and 20% were used to show the effects of .higher and
lower levels of variability.
The probability distribution of the average of N daily measurements taken during a month,
MN, is given by the following normal probability density function:
2*a
N
where \L is the mean or long term average, and a is the
standard deviation of the daily discharges. If ^ is the maximum monthly average allowed by
the permit, then the probability that the monthly average exceeds the permit maximum is .
given by P(MN>/x,). Using simple algebra this probability can be rewritten as:
ATTACHMENT TO INTERIM GUIDANCE A-7
-------
PERFORMANCE-BASED REDUCTION OF MONITORING FREQUENCIES.
where *(.) is the standard
= i-d>( -ti^- ), normal cumulative probability
function. (the Microsoft®Excel
built-in function NORMDIST).
Since
where C is the coefficient of variation, then the probability of a monthly average exceeding
the maximum allowable can be calculated using C, N, and the ratio of the long-term average
to the maximum allowable monthly average using NORMDIST. This is how the values in
Tables 3, 4, and 5 were calculated.
Alternate approaches to probability calculations:
The probabilities in Tables 3-5 were calculated with the assumption thai the distribution of
the sample means is normal. Individual sample values are generally best fit to a lognormal
distribution. As discussed in the TSD, the mean of small samples from a lognormal
distribution is in most cases approximately lognormal. Probabilities can be calculated
assuming a lognormal distribution by two different methods, a Monte Carlo technique and
the Microsoft Excel built-in function LOGNORMDIST. The resulting probabilities will be
very close to those in the normal distribution table for the sample sizes and discharge levels
under consideration for monitoring reductions, although the probabilities calculated from
these two distributions may not be comparable for all sample sizes and all discharge levels.
The statistical evaluations used in this analysis are intended for use only to illustrate the
effect and benefits of this strategy, alternative statistical techniques and approaches may be
utilized in other situations. . .
ATTACHMENT TO INTERIM GUIDANCE A-8
-------
PERFORMANCE-BASED REDUCTION OF MONITORING FREQUENCIES
NPDES Burden Reduction Analysis
The analysis to estimate the NPDES burden reduction used the SAS Language and data from
the Permit Compliance System (PCS) database. The procedure, assumptions, and results are
summarized below:
• The universe for this study was all major facilities with measurement data in PCS
(6,477) for the two-year evaluation period of'1/93 to 12/94. This evaluation period
was chosen in order to have as large a universe as possible since the Commonwealth
of Virginia and the State of California have not entered measurement data into PCS
for 1995. -.:.-'
The facility entry criteria for enforcement history were approximated by eliminating
permittees for consideration that have effluent violations for either an active formal
judicial action or an active formal administrative order (AO) for 1995.
The parameter entry criterion, evaluated per outfall, was the elimination of
parameters for consideration that have had any Significant Non-Gompliance (SNC)
violations during the two-year evaluation period.
For each parameter eligibleJor burden reduction, the long-term average (LTA) for
the two-year period was calculated and compared to the monthly average limit.
The amount of burden reduction was calculated to be the ratio of the difference
between the monthly average limit and the LTA divided by the monthly average limit.
This approximates the reduction presented in Table 1 of the guidance for LTA to
monthly average limit ratios up to 75%.,
• No reduction for parameters not meeting the 75% ratio threshold.
Table 6
Municipal
Non-municipal
Total
Burden Reduction
- 27%
24%
26%
ATTACHMENT TO INTERIM GUIDANCE
A-9
-------
------- |