United States
Environmental Protection
Agency
Enforcement And
Compliance Assurance
(2224A)
EPA 530-R-94-019
October 1994
Waste Analysis Guidance for
Facilities That Burn
Hazardous Wastes
Draft
Notice
This document is a preliminary draft. It has not been formally
released by EPA and should not at this stage be construed to
represent Agency policy. It is being circulated for comment on its
technical accuracy and policy implications.
-------
WASTE ANALYSIS GUIDANCE FOE
FACILITIES THAT BURN HAZARDOUS WASTES
-------
DISCLAIMER
The policies set out in this document are not final Agency action,
but are intended solely as guidance. They are not intended, nor
can they be relied upon, to create any rights enforceable by any
party in litigation with the United States. EPA officials may
decide to follow the guidance provided in this manual, or to act at
variance with the guidance, based on an analysis of specific site
circumstances. The Agency also reserves the right to .change this
guidance at any time without public notice.
-------
CONTENTS
Section - - .. ' : ' Page
1.0 INTRODUCTION ................. 1
1.1 BACKGROUND 1
1,2 PURPOSE ........ 1
1.3 REGULATORY REQUIREMENTS ....... ; 2
/': ' .;.-' .
2.0 WASTE ANALYSIS PLANS . . . . . . ... . . ..: . . . . .... 4
1 2.1 SAMPLING METHODS FOR EACH FEED STREAM .5
2.2 METHODS OF PREPARATION AND ANALYSIS OF SAMPLES . . 5
2.3 SAMPLING AND ANALYSIS STRATEGY 6
2.4 SAMPLING LOCATION . . i. ...... 6
2.5 QUALITY ASSURANCE AND QUALITY CONTROL . . . . . . .7
I ' ,
3.0 SAMPLING AND ANALYSIS STRATEGIES . . . ..;._.. . .... 7
3.1 SAMPLING AND ANALYSIS BY BATCH ........... 8
3.2 QUALIFICATION OF A FEED STREAM ...%...... 10
3.3 STATISTICAL ANALYSIS ....... .i . 13
3.3.1 Upper Tolerance Limits ......' 16
3.3.2 Statistical Approach: Compliance
Issues .................. 18
4.0 INCOMPLETE DATA, OUTLIERS, AND DETECTION LIMITS . . . .' 21
4.1 OUTLYING DATA POINTS . . . . . . . . i. . . ,. . . .21
4.2 INCOMPLETE ANALYTICAL DATA ; 22
4.3 DETECTION LIMITS ..........:....... 22
5.0 MANAGEMENT OF RESIDUES :. 23
5.1 'DETERMINATION OF THE BEVILL EXEMPTION 24
5.2 FREQUENCY OF SAMPLING AND ANALYSIS . ;. 25
5.3 CONSIDERATIONS . . ........... 25
6.0 DOCUMENTATION TO DEMONSTRATE COMPLIANCE ........ 27
Appendix " '
A Table of K Factors for Calculation of Tolerance Limits
-------
1.0 INTRODUCTION
1.1 BACKGROUND
On January 23, 1981, the United States Environmental Protection
Agency (EPA), pursuant to requirements of the Resource Conservation
and Recovery Act of 1976 (RCRA), promulgated regulations governing
the combustion of hazardous waste (HW) in incinerators (HWI). On
February 21, 1991, EPA promulgated regulations governing the
burning of hazardous wastes in boilers and industrial furnaces
(BIF). The regulations are intended to protect human health and
the environment from exposure to emissions from the combustion of
hazardous wastes. Regulations governing such activities are
codified at 40 Code of Federal Regulations (CFR) Section (§) 264
Subpart O and,§265 Subpart O for permitted and interim status HWIs,
respectively, and at 40 CFR §266 Subpart H for BIFs.
!
1.2 PURPOSE
The,purpose of this document is to provide guidance to facilities
and EPA Regional and state personnel regarding .appropriate
approaches to sampling and analyzing feed streams to ensure
compliance with EPA requirements for waste analysis for hazardous
waste combustion devices. This document describes three
alternatives for demonstrating compliance. The alternatives
provide a uniform approach to documenting compliance with limits on
constituent feed rates established during compliance testing or a
trial burn. The three alternatives are batch analysis,
qualification of a feed stream, and statistical analysis.
The concepts presented in the following sections meet the intent of
the regulations and can be implemented in a way that is consistent
with daily operations of the facility. Guidance also is provided
in this document for analysis of residues generated from the
combustion of hazardous wastes. This document does not cover
specific methods of sampling and analysis for units regulated under
RCRA and does _not preclude EPA or state personnel from taking
enforcement actions related to waste analysis.
- DRAFTS Do Not Cite Or Quote -
-------
1.3 . REGULATORY REQUIREMENTS1
.To ensure proper combustion of hazardous wastes, the HWI and BIF
regulations and individual operating permits establish limits on
operating parameters for combustion units. The limits ensure that
the maximum levels of emissions of the constituents of concern from
the combustion units are low enough to ensure acceptable levels of
constituents in ambient air, as specified in EPA regulations or the
facility's permit (levels protective of human health and the
environment). To demonstrate this, air dispersion modeling and/or
emissions testing is used to establish operating limits for the
facility that ensure that a facility's emissions do not exceed the
regulatory levels. Emissions tests are conducted during the trial
burn for HWIs and BIFs attempting to obtain operating permits or
during the compliance test for BIFs operating under interim status.
Some examples of operating parameters contained in the BIF
regulation and some HWI permits that generally are established
through air dispersion modeling and/or emissions testing are:
« Continuous monitoring and recording of the flow rates and
composition of hazardous waste, other fuels, and feed
stocks for industrial furnaces to yield the feed rates of
10 metals (mercury, lead, cadmium, chromium, barium,
beryllium, arsenic, thallium, silver, and antimony),
chlorine and chlorides, and ash
» Maximum and minimum temperature limits for the burning
zone
Maximum production rate (for example, steam)
Continuous emissions monitoring for oxygen, carbon
monoxide, and hydrocarbons
Appropriate operating parameters for air pollution
control equipment
Operating permits and regulations require that combustion
facilities maintain, monitor, and record established operating
parameters while burning HW to document compliance (for example, 40
CFR §266.102(e)(10), §266,103(k), §264/265.73). Of the various
operating parameters, feed rate limits for metals, chlorine and
chlorides, and ash are key elements for which facilities must
maintain records to demonstrate compliance. For example, the BIF
This guidance document discusses requirements that arc generally included in tbe regulations and/or individual permits for combustion facilities.
Because tbese requirements may vary by type of status of * combustion unit, it is necessary to consult tbe regulations and/or permit for requirements specific
to a particular uak,
- DRSJTs Do Not Cite or Quote -
-------
.; x . ' ' .--.--. 3' . ;t, _'-...;
regulations , specify that the feed rate limits for metals, total
chlorine and chloride, and asti are to be "established and monitored
by knowing the concentration of the substance in each feed stream
and the flow rate of the feed stream" (for example, 4.0 CFR
§266.103(c)(4)(iv)(D)). The flow rate must be monitored under the
continuous monitoring requirements specified in a permit or the
regulations (for example, 40 CFR §266.103(c)(4)(iv)(D) of the BIF
regulations or 40 CFR §264.347 of the incinerator regulations), In
other words, the feed rate for each metal, chlorine and chloride,
and ash in the total feed streams must be established and monitored
continuously (for example, 40 CFR §266.103(c.) (3)) . The term
"total feed streams" includes anything that is fed to the unit (for
example, liquid and solid hazardous wastes, raw materials, fuels,
nonhazarvdous wastes, and off-gas streams from production processes,
see 56FR7176). ,
In addition, compliance with all the other limits on operating
parameters (for example, operating limits on air pollution control
devices and temperature) may not be adequate to establish
compliance with emission limits, if the feed streams fed into the
combustion unit are not characterized and monitored properly. It
is easy to conclude, then, that analysis of constituents of concern
in the feed streams is the starting point in demonstrating
compliance with many requirements governing combustion of hazardous
waste. However, the regulations do not require that specific
methods be used in sampling and monitoring the concentrations of
each substance. Various interpretations therefore have been put
forth of what constitutes compliance with requirements for waste
analysis at such facilities. «
Since waste analysis is the basis for knowing the concentrations of
constituents and demonstrating compliance with requirements
governing feed rates, a waste analysis plan describing the specific
procedures that will be followed to obtain accurate, representative
results is necessary to support the analysis* EPA's waste analysis
regulations at 40 CFR §264.13, 265.13 state that before a Waste is
treated, stored or disposed, the facility must obtain a detailed
analysis of the waste, which, at a minimum, "must contain all the
information which must be known to treat store or dispose of the
waste" in compliance with relevant standards. In addition, the
regulations governing permitted incinerators and BIFs, set forth
under 40 CFR §264.341 and §266.102, require detailed analysis for
concentrations of constituents as necessary to. ensure that the
waste feed is "within the physical and chemical composition limits
specified" in the permit. Because of the uncertainty associated
with most production processes, the Agency has found that process
knowledge alone does not generally give the type of precise
information necessary to establish and monitor feed rate limits.
Therefore, any facility choosing to rely on process knowledge alone
will be intensely scrutinized and runs the risk that the Agency's
- DRAFTS Do Hot Cite Or Quote -
-------
own sampling will demonstrate that the facility's waste analysis
method did not produce the information required to demonstrate
knowledge of the constituent concentrations in the feed streams.
Further, the requirements for waste analysis at 40 CFR
§264.13(a) (3) and (b)(4) and at §265.13(a)(3) and (b)(4) state that
analysis must be repeated at a frequency sufficient to ensure that
it is accurate and up-to-date. The following sections of this
document provide facilities guidance on demonstrating compliance
with the waste analysis requirements for monitoring feed rates
through: ..."
« Development of a waste analysis plan
\ . *
Selection of the appropriate frequency for sampling and
analysis
Quality assurance and quality control of data from
analysis
Documentation that demonstrates compliance
Included is a discussion of the requirements for analysis of
residues generated from combustion of hazardous wastes. Because
some of those residues have the potential to be hazardous to human
health and the environment, proper sampling and analysis is also
necessary to make this determination.
2.0 WASTE ANALYSIS PLUMS
Waste analysis is the backbone of the RCRA program and the
hazardous waste combustion regulations discussed above. Therefore,
every facility that treats, stores, or disposes of hazardous wastes
is required to develop a waste analysis plan (WAP) (40 CFR §264.13
and §265.13). Elements of the WAP that are particularly applicable
to combustion facilities are discussed here. General contents and
development of a WAP will not be covered; for such general
information the reader should refer to the EPA guidance Waste
Analysis At Facilities That Generate, Treat, and Dispose of
Hazardous Waste, (OSW1R [Office of Solid Waste and Emergency
Response] # 9938.4-03, April 1994). The document is available from
the National Technical Information Services (NTIS), publication #
PB94-963-603.
In this document, the term waste analysis plan refers to a written
document, prepared by each regulated facility, that defines the
sampling and analysis protocols and frequency through which the
facility determines the concentration of regulated constituents in
each feed stream at all times. The WAP is not limited to hazardous
waste feed streams, but includes all feed streams, such as
- DRAFT: Do Not Cite Or Quote -
-------
nonhazardous wastes, fossil fuels, and raw materials when they (the
nonhazardous feed streams) are cofired with hazardous waste. The
waste analysis plan should be amended .with the appropriate
information when new units are added, processes change, new
regulations are promulgated, or permit modifications are issued
that affect analysis of feed streams before treatment, storage, or
disposal of those feed streams.
/ . , . t
Some items that may be contained in the WAP are: a description of
treatment activities conducted at the.facility; identification and
classification of HW generated, treated or managed at the facility
and of their quantities; and descriptions of HW units and operating
procedures (for example, use of safety equipment); and other
pertinent information. Some specific items that treatment
facilities such as BIFs and HWIs must include in the WAP, per 40
CFR §264.13, §265.13, and §268.7 are discussed below:
2.1 SAMPLING METHODS FOR EACH FEED STREAM
Sampling methods may be included in the WAP eithesr by reference to
sampling methods described in 40 CFR §261 Appendix I (for example,
specific methods set forth in EPA publication SW-846 or specific
American Society for Testing and Materials (ASTM) methods) or by
specifying an equivalent standard sampling procedure for the
selected analytical method. The WAP must describe measures used to
ensure that the analytical sample(s) is representative of the
entire feed stream (40 CFR §264.13 and §265.13). Representative
samples may be grab samples or composite samples. In general,
compositing of samples should be used only to account for spatial
variations within a single sample lot (for example, a rail car load
of coal or'a truck load of limestone). Compositing should not be
used to reflect the concentrations of constituents in a group of
waste containers that originate at any one of several sources. If
a facility's regulatory limits on feed rates are specified on a
time-average basis (that is, hourly rolling average), compositing
also may be used to account for temporal variations in the sample
lot. In such cases, the compositing period should not exceed the
regulatory averaging period for that sample lot (that is,
compositing of several sample lots being burned at different times
is not appropriate). - If the facility is subject to an
instantaneous constituent feed rate limit, temporal compositing
should not be used. Test methods in SW-846 provide'more detailed
information on sampling methods.
2.2 METHODS OF PREPARATION AND ANALYSIS OF SAMPLES
Methods of preparation and analysis of samples must be specified in
the WAP for each regulated constituent in each feed stream (40 CFR
§264.13 and §265.13). This requirement can be met either by
reference to standard methods of preparation and analysis of
- DRAFTS Do Not Cite Or Quote -'
-------
samples (specific methods in EPA publication SW-846 or specific
ASTM methods) or by specifying a 'step-by-step procedure for
preparation and analysis of samples. An SW-846 method must be used
when required by regulation. If an SW-846 method is not specified
in the regulations, it is recommended that SW-846 methods be used
whenever the methods are both available and appropriate for the
sample matrix; however, other equivalent methods generally may be
used. In addition, any laboratory that is to conduct the analysis
and meet requirements for quality assurance and quality control
(QA/QC) procedures (testing methods, laboratory procedures for
handling of the sample, and others) should be specified in the WAP.
2.3 SAMPLING AND ANALYSI13 STRATEGY
An acceptable strategy is one that, in combination with the data
from continuous monitoring of the feed rate, provides reasonable
assurance that all constituent feeds are within allowable limits
before they are fed and that the limits on feed rates will not be
exceeded at any time while waste is being burned. After-the-fact
knowledge of feed rates of constituents is not an acceptable way to
determine compliance with the regulations. The strategy should
outline the frequency at which the feed streams will be sampled and
analyzed. Supporting documentation should be kept on record to
justify the selection of the frequency of sampling and analysis,
2.4 SAMPLING LOCATION
The location from which a sample is to be collected is important in
determining the appropriate sampling method and assessing the
ability to obtain a representative sample. The location also may
influence the results of the analysis, thereby affecting the feed
rates, as well as the choice of an appropriate frequency of
sampling and analysis. Examples of appropriate sampling locations
include:
For an on-site, continuous process that generates one
waste stream, a sample may be obtained from the pipeline
that feeds hazardous waste to the combustion unit.
However, such sampling should be implemented only at
facilities at which it is known, through a statistical
profile, that none of the concentrations of constituents
in the feed streams is above the maximum allowable
limits. . ,
For a batch process, such as a tank filled with hazardous
wastes, a representative sample of the entire batch in
the tank should be obtained and analyzed before the
contents of the tank are fed. Potential for
stratification of the wastes should be considered during
the sampling procedure. Continuous mixing or
- DRAFT: Do Not cite Or Quote -
-------
reeirculating of the contents of a tank reduces the
significance of the degree of heterogeneity of the waste.
For a lot of containerized wastes from the same waste
stream, a representative sample may be obtained by
compositing samples from the containers. ASTM Method
D140-70 may be used to estimate the number of containers
within the lot to be sampled. Each sample should be
considered acceptable only if the particular, waste sample
closely resembles all other samples (for example, in
color). The composite or representative sample should be
analyzed before the wastes are fed to the combustion
unit. ; ;
Whatever sampling location is selected, the location should be
identified clearly and its selection justified in the WAP.
2.5 QUALITY ASSURANCE AND QUALITY CONTROL
The facility's QA/QC procedures for sampling and analysis should be
stated in the WAP. Sources of information on developing a QA/QC
procedure include: 1) Chapter One of SW-846, "Quality Control"; 2)
Guidance on Setting Permit Conditions and Reporting Trial Burn
Results (EPA/625/6-89/019); and 3) Handbook - QA/QC Procedures for
Hazardous Waste Incineration (EPA/625/6-89/023).
All the factors discussed above can influence the quality of the
analytical results. Therefore, they should be addressed in a site-
specific WAP as part of a facility's demonstration that the waste
streams will be sampled and analyzed in a manner that complies with
requirements for monitoring of feed rates of constituents. In
their WAPs, facilities also should set forth procedures for
evaluating analytical data with respect to outliers, completeness,
and detection limits, as discussed in greater detail in Section IV
of this document. ;
3.0 SAMPLING AND ANALYSIS STRATEGIES
This section presents options for sampling and analysis programs to
ensure compliance with either the permit or the regulatory
requirements discussed above.
The BIF rule and some permits for hazardous waste incinerators
require combustion facilities to continuously monitor the feed
rates of selected metals, chlorine and chlorides,, ash, and wastes
(40 CFR §266.103 (c) (iv) (D). and §264 and 265.347). As discussed
above, to satisfy this requirement, the feed rate of each feed
stream must be monitored continuously and the facility operator
must "know" the concentration of each regulated constituent in each
, - DRAFT: Do Not Cite Or Quote -
-------
- 8
feed stream. The "requirements for a continuous monitor are
provided in 40 CFR 266.103(c)(4)(iv)(B)(i). A logical and coherent
sampling and analysis program for regulated constituents-, and
continuous monitoring of feed rates of feed streams, are
fundamental aspects of a compliance strategy that ensures that
limits on maximum feed rates of regulated constituents are not
exceeded. Knowledge of the concentrations of regulated
constituents ' in each feed stream should be based on -an ongoing
sampling and analysis program. Fundamentally, the "knowing", of
concentrations of regulated constituents allows the calculation of
feed rates for those constituents for any point in time at which
hazardous waste is being burned. That calculation then can be
compared with regulatory limits.
When a sampling and analysis program is established, several
factors should be considered, including: variability of the feed
stream, sampling location, and proximity of levels of regulated
constituents to established limits. The following discussion
describes three generally acceptable approaches to sampling and
analysis. Other strategies may also be acceptable and will be
considered on a case^-by-case basis.
'
3.1 SAMPLING AND ANALYSIS BY BATCH
Sampling and analysis by batch, is a strategy most appropriate for
facilities that have multiple feed streams in which concentrations
of regulated constituents vary greatly and for facilities that
receive wastes from off-site. Multiple storage and feed tank
systems may be necessary to properly execute this compliance
strategy. The batch methodology reguires that once a
representative sample has been taken from a tank and analyzed, no
other material can toe added to the tank. Results of laboratory
analysis must be known before wastes are burned; therefore,
laboratory turnaround time may be, a consideration. "The measured
concentrations of the regulated constituents establish a maximum
feed rate that is at or below the regulatory limit. For tanks that
do not have agitation systems, stratification of the contents is
possible. Therefore, care should be taken to ensure that a
representative sample is obtained. The objective of a batch
strategy is to enable a facility to calculate a maximum feed stream
rate based on measured concentrations of regulated constituents.
The facility also can calculate the actual feed rate of regulated
constituents at actual feed rates of feed streams and actual
concentrations of constituents in any given instance. Batch
sampling and analysis is a relatively simple and straightforward
methodology for ensuring compliance. Examples that illustrate
generally acceptable and unacceptable ways of complying with this
strategy are given below:
- DRAFT: Do Not cite Or Quote -
-------
» A simple example of a case in which sampling and analysis by
batch is generally appropriate is a facility that receives
hazardous waste from many off-site sources and blends the
wastes on site. Such a facility may conduct some preliminary
analysis on each waste stream before it is accepted and
discharged to the storage tank system. Wastes are accumulated
in one of three continuously agitated" mixing and storage
tanks. When a tank is full, a representative sample of the
waste in the tank is obtained and analyzed. After the
facility receives the results of the analysis, the waste is
fed as a batch to the combust ion, unit. .Once characterized, no
other material (for example, hazardous wastes, used oil, or
fuels) is added to the tank being fed, and incoming wastes are
accumulated in the remaining two tanks. Calculations of feed
rates are based on the results of analysis of that batch.
A facility generates several waste streams from relatively
consistent production.processes. One or all of the streams
may be piped to a storage tank at any given time and in
quantities determined by production. A sample to determine
the concentrations of metals, ash, and chlorine and chlorides
is taken once from the storage tank for preparation of the
certification of compliance, and again six months later in
preparation for emissions testing to revise the certification
of compliance. The. two samples show variations in
concentrations of constituents? as shown however,
concentrations in both samples are below limits on feed rates.
In the second example, the facility performed analytical
determinations but did not consider how the results would be used
to document compliance. For example^ to calculate a feed rate at
a given point in time, which of the two results (if either) should
be used to determine compliance? How can the facility prove that
the two samples include the variations in concentrations of
constituents, considering that the various process streams exist in
different ratios in the burn tank at any given time? The facility
should consider several options that are more reliable compliance
protocols. One option may be a batch feed operation, in which the
three streams would be collected in the tank and an analytical
determination made after preparation of the batch and before
feeding. Under this option, a given set of analytical results
correlates directly with the period when a particular batch is fed.
Drawbacks are associated with the approach: frequent analysis is
necessary (every tank) and installation of several new tanks may be
appropriate. . ,
- DRAFT: Do Not Cite Or Quote -.-
-------
10 .
3.2 , QUALIFICATION OF A FEED STREAM
As' long as a facility can ensure that the feed rates for regulated
constituents are at or below regulatory limits, it is not necessary
to know the exact or actual constituent feed rate. This
alternative is a variation of the batch sampling and analysis
strategy and may be appropriate for facilities that have complex
feed management systems and those that have a continuous demand for
steam or production rate requirements (this does not imply a
constant feed rate of hazardous waste, since such wastes often are
cofired with other fuels). The qualification strategy is similar
to a batch strategy in that all feed streams are sampled and
analyzed for all constituents identified in the permit or
regulations at some point in the feed stream management system
before they are fed into the combustion device. After-the-fact
knowledge of constituent concentrations or feed rates of
constituents is not acceptable. The qualification strategy also
can be an appropriate approach for facilities that generate
multiple waste streams in various quantities and at various times.
This strategy can be implemented in two ways. There are two
variables for the calculation of constituent feed rates: The
concentration of the regulated constituent in the feed stream and
the feed rate of the stream. If one variable is fixed, the other
variable can be adjusted to ensure that the regulatory limit is not
exceeded. This approach is illustrated below as:
FR = (C) (Q)
where: ' .
FR = The regulatory fee,d rate limit of . a
constituent for the feed stream (unit
weight/time)
C = The concentration of the constituent in the
feed stream
Q = The feed rate of the feed stream
The feed rate limit, FR, has a maximum, value that cannot be
exceeded. (It should be noted that the sum of all Qs must be at or
below the allowable hazardous waste feed rate for the Qs that
represent hazardous waste feed streams '(40 CFR §266.103(c)(l)(i)).
It follows that both C, the concentration of the constituent, and
Q, the feed rate of the feed stream, can vary, as long as their
product does not exceed FR. Two options using this approach are:
- DRAFT: Do Not cite Or Quote -
-------
Qualification based on predetermined feed rate of the
total feed system
This option can be used by a facility that requires a
relatively constant feed rate, Q, to meet production
needs or demand for steam. The fixed feed rate also
fixes the maximum acceptable concentration, C, that can
be present without exceeding the maximum regulatory limit
FR. In other words, the concentration, C, can vary below
its maximum limit. Thus, when using this compliance
strategy, a facility would analyze each batch of waste or
feed material for regulated constituents before
acceptance of the waste into the feed management systems.
If each regulated constituent is determined to be at, or
below its maximum allowable concentration (determined by
the fixed feed rate Q so that thei product of the
concentration, C, and Q does not exceed FR, the
regulatory limit) , that batch is; qualified for
combustion. For determination of compliance, the "known"
concentration of a regulated constituent is the maximum
concentration at which the material meiy be ''qualified."
The qualified material then may be blended without
restriction with other qualified feed streams without
further analysis, since FR is at or below limits for the
fixed Q. , . . ' .
* Qualification based on predetermined concentrations
This approach sets a maximum limit on C (that is, fixes
this variable), the concentration of a regulated
constituent, and allows Q to vary below a maximum value
determined in such a way that the product of C and Q does
not exceed FR, To use .this compliance strategy, a
facility performs the required waste analysis on incoming
batches of waste before the batches are mixed. Rather
than doing another analysis of,the blended wastes, the
waste stream is considered to have the. concentration of
* regulated constituents found in the batch having the
highest concentrations. The facility then calculates a
maximum feed rate, Q. Compliance with the regulatory
limits on feed rates will be ensured as long as Q remains
at or below its maximum value.
Implementation of the qualification strategy can vary widely
depending on the complexity of a' facility's feed; stream and waste
management system (for example, presence of an interconnection and
isolation system for storage tanks) . It is alisd possible for a
facility to reestablish a predetermined maximum concentration, C,
that could be lower than a previously established level and thus
allow an increased feed rate, Q. In any event, as is true of the
- DRAFTS Do Not Cite Or Quote -
-------
12
original limits, a change in limits on concentrations should be
well documented, justified, and specifically associated with a
particular time period during which the waste is burned.
Application of the compliance strategy based on use of
predetermined concentrations " is illustrated by the following
example:
A facility that receives liquid hazardous wastes from
off-site has a series of storage, blend, and burn tanks,
as shown below:
Combustion
Unit(s)
Level 3
Level 2
Level 1
Burn tanks
Blend tanks
Storage tanks, containers
and transport vehicles
All incoming shipments are received, sampled, and analyzed at Level
1 and determined to be below the maximum concentration limits, C,
which are based on a fixed feed stream feed rate limit, Q. Thus,
constituents past level 1 in the system will have concentrations
less than the limits, because the operator does not allow transfers
into the system of waste having concentrations above the maximum
concentration limits. This sampling and analysis strategy also can
be applied effectively at Level 2. When monitoring concentrations
at Level 2, for example, it is not necessary that the actual
concentrations of the individual loads delivered to the Level 1
tanks be below the maximum limits, C, since all concentrations that
pass through levels 2 and 3 to be fed to the combustion unit must
be at or below maximum concentrations C. However, the facility may
not be able to apply this strategy at Level 3 since after-the-fact
knowledge of constituent . concentrations or feed rates of
constituents is not acceptable. Weighted averages should not be
used to determine levels of concentrations. Compositing of samples
from different levels and tanks is not acceptable. If a
concentration, C, anywhere in the system before the level being
monitored exceeds a predetermined maximum concentration limit, the
- DRAFT: Do Not Cite Or Quote -
-------
' . . . ' .13 ' . ','.".-'..-'....
contents of the tank can be reblended with, other fuels until the
concentrations are lowered (and resampled and analyzed before
transfer to the next level) , or a new maximum C can be established
and applied in future calculations of feed rates. Under certain
circumstances, blending of wastes (for the batch or qualification
strategies) may require a permit.
3.3 STATISTICM,
A statistical approach can characterize concentrations, of constituents in
fossil fuels, raw materials, or wastes generated on-site. It is appropriate
for "consistent" feed streams (for example, hazardous waste generated by a
specific on-site production process, coal produced from a specific mine or
seam, or limestone ore produced from a specific quarry), for which there is
reasonable expectation that each constituent will be normally distributed
about a mean. It should not be used at facilities that receive wastes from
off site. The approach demands that the operator of the facility have
sufficient knowledge of the source of the feed material to be aware of any
change that is likely to affect the sample distribution. When such a change
is known to have occurred, the facility operator should not rely on. this
approach until a statistical profile of the "new" , feed stream has been
developed. Through statistical analysis, the owner or operator ultimately
will develop a program that specifies a frequency at which sampling and
analysis are to be conducted to ensure, with an .appropriate degree of
confidence, that feed rates are not exceeded. It 'also' should be understood
that, with the use of a statistical approach, there is a finite probability
that a : facility can be found to be out of compliance based on sampling and
analysis. If such a circumstance occurs, use of a statistical sampling and
analysis strategy is not a shield against enforcement action and the adequacy
of the analysis may be considered in penalty calculations.
Because this approach should be used to characterize waste streams only as-
generated. It should not be used after the waste has been blended with any
other waste, fuel, or raw material. It is therefore generally not
appropriate in any case in which the hazardous waste is generated at a
facility that is not under the same ownership and control as the facility
that burns the waste. (This approach may be appropriate however, in cases in
which raw materials and fossil fuels are produced by entities other than the
facility that burns the waste, provided that there is a contractual
requirement that the burner be notified of changes that could have
significant effect on concentrations of constituents in those feeds.)
When using any statistical approach, facilities should be guided by the
following principles:
The statistical analysis should be based,. on 'actual analytical
results. As discussed above, Process knowledge alone is not
generally sufficient.
- DRAFT: Do Not Cite Or Quote -
-------
14
The operator of the facility should demonstrate at least a 95
percent probability and confidence that the maximum concentration
of any sample will not exceed an allowable limit.
A continuing sampling and analysis program should be established to
demonstrate that the statistical distribution does not change over
time.
Of the several approaches to conducting a statistical analysis, the use of an
upper tolerance limit is discussed here. This is the same method described
in 40 CFR §266 Appendix IX Section 7.2 of the BIF regulations for Statistical
Methodology for Bevill Residue Determinations. There also is a useful
discussion of the application and calculation of upper tolerance limits in
Statistical Intervals: A Guide for Practitioners , by Gerald J. Hahn and
William Q. Meeker (ISBN 0-471-88769-2) . For reasons discussed below, this
approach is a recommended approach to waste analysis at combustion
facilities.
A general overview of statistical analysis is an appropriate starting point
in understanding the approaches to be discussed. The underlying concept of
statistical analysis is the development of a mathematical model for the
expectation (or prediction) of a random variable within a given population.
Such a model, commonly known as a probability distribution model, gives the
probability that a random variable, x, lies between two values. Development
of the model is simple and is illustrated below:
For a sample of random values for a given data set, one could find the
average value for the sample,' which is called the arithmetic mean, X. This
is expressed as:
n
where:
X = Numerical value of sample point n
n = Number of samples
X = Mean of X
- DRAFT: Do Not cite Or Quote -
-------
' - 15 ." . .. .
Next, the distribution of the sample values.about the mean is desired. The
most common distribution is called a "normal" or "Gaussian" distribution.
Graphically, this distribution is represented by a bell-shaped curve, as
shown below: .
o
-4 -3 -2 - -I
Value of X
Solid curve:
Dotted curve:
Normal distribution, with mean of
deviation of 1. .
Normal distribution, with mean -of
deviation of 1.5. " . !
0 and standard
0 and standard
As this curve shows, lower and upper values of .the data set can be calculated
with known probabilities. The shape of the curve depends upon the scatter or
dispersion of the values about the mean and is often referred to as a "two-
sided" distribution. Evaluating the dispersion or scatter about the mean can
be done by calculating the standard deviation. The standard deviation, s,
can be calculated as follows:
s =
(x-x)
Un-x)2
The objective is to describe a population represented by the samples, for
which any given sample can be found to be between a set of upper and lower
"limits. From the samples, a confidence interval for the unknown population
mean can be constructed. This interval consists of two values, the upper and
.lower limit. Given 'certain assumptions about the population, the chance that
these values straddle the unknown mean is a certain percentage.
- DRAFT: Do Not Cite Or Quote -
-------
16
3.3.1 Upper Tolerance Limits
One approach that satisfies the criteria set forth above is based on upper
tolerance limits. This approach is outlined in the paragraphs below. For a
more detailed description, see Meeker and Hahn (1991).
If a "variable is normally distributed and the sample mean, standard
deviation, and number of samples are known, it is possible to estimate the
probability that a fixed percentage of the sample population will not exceed
a certain value. That value is known as an upper tolerance limit (UTL) . Fox-
purposes of this guidance, the minimum UTIj that should be used in lieu of
continuous analysis of waste is the value of the one-sided upper 95 percent
tolerance bound that exceeds at least 95 percent of. the sample population.
In other words, we can say with 95 percent .confidence that 95 percent of all
individual, samples will not exceed the UTL. Therefore, if a facility
generates a good initial database to establish the UTL for the constituents
of concern, and subsequent, sampling and analysis shows that the
concentrations are below the UTL, the waste can be considered the type of
waste for which the UTL was calculated. The UTL values then may be
considered the "known" concentration for each constituent in that feed
stream. .
Although concentrations of constituents in any single sample are likely to be
well below the UTL, feed rates always should be calculated as if each
constituent were present at its UTL. The UTL is adjusted continually to
reflect new information from analysis. The UTL for each constituent is
calculated as follows:
Step 1: Using all valid analyses of the subject feed stream, calculate the
mean concentration (X) and the standard deviation (s) for the
samples.-
Step 2: Using the equation below, calculate the upper tolerance limit, UTL
(o.95;o.9s) so that there is at least 95 percent confidence that at
least 95 percent of all samples will not exceed the UTL. Values
for K are obtained from a table for calculating one-sided tolerance
bounds for a normal distribution (see Appendix A) .
= X + (K (lKrip)) (s)
where:
1-a = The desired level of confidence that at least 100 (p) percent
of the individual samples will be below the UTL.
p = The decimal fraction of samples that will be predicted to fall
below the UTL.
n = The number of samples.
- DRAFTS Do Not Cite Or Quote -
-------
Table 1 in Appendix A lists values of K for l-a=0.95, with p=0.95.
Statistical references may be consulted for other values of i-a. Linear
interpolation may be applied to obtain values of n that are not tabulated.
This guidance recommends that, if a , UTL is to be used to demonstrate
compliance, i-a must be >0.95 and p must be >0.95. A more conservative (that
is, higher) UTL may be used to decrease the necessary frequency of sampling
and analysis, as described in the following step.
Step 3. Determine the appropriate frequency of sampling and analysis
, according to the following equation.
number of samples = (a^ic) (₯)
year
where: '' _ " | "-./'
acalc = One minus the level of confidence used to calculate the
UTL; at a 95 percent confidence level, a^,. = (1-0.95) =
0.05
Y = days per year on which waste is generated
N
For facilities that meet the minimum requirements of this methodology
(estimating concentrations based on "«,,,. 0,95), the feed stream should be
sampled and analyzed on at least 5 percent of the days on which it is
generated. If the facility chooses to use a more conservative UTL, where a^
>0.95, the burden of sampling and analysis will be reduced.
In qualitative terms, as the statistical confidence that an allowable feed
limit on, constituents will not be exceeded increases, the frequency with
which sampling and analysis are necessary decreases. However, at a minimum,
each feed stream should be analyzed at least once per year. Also, sampling
dates should be spaced evenly throughout the year.
Most statistical tests assume that the data come from a normal distribution.
The normal distribution is the assumed underlying model for such procedures
as calculation of tolerance intervals. if the data are not distributed
normally, false conclusions can result if the data follow a more skewed
distribution like lognormal. Therefore, checking the data for normality is
an important step in statistical calculations. EPA has available a useful
discussion of evaluating data for normality in a document titled Statistical
Analysis of Ground-water Monitoring Data at RCRA Facilities, Addendum To
Interim Final.Guidance - Draft (EPA/500/R-93/003, July 1992). The document
is available for sale through the RCRA docket at (202) 260-9327. Copies cost
$0.15 per page. r
- DRAFTS Do Not cite Or Quote -
-------
18
3.3.2 Statistical Approach: Compliance Issues
No statistical approach can guarantee true continuous compliance with short-
term constituent feed rate limits. There is always a finite possibility that
concentrations of constituents in any given sample will exceed the UTL. This
fact is accepted in statistical characterization. If the sampling and
analysis indicates that the UTL has been exceeded, then the following is
recommended:
O continue to calculate constituent feed rates using the UTL
o immediately following receipt of an analysis that exceeds the UTL for
any constituent, the facility should begin daily sampling and analysis
of that feed stream. Daily analyses should continue until all regulated
constituents are below their UTL for three consecutive days.
i
O if the feed stream exceeds the UTL for the same constituent 2 or more
times while conducting the daily sampling, the facility should
immediately cease using the statistical .approach for that feed stream
until a new feed profile is developed (using data obtained after the
initial UTL exceedance).
It should be noted that, at facilities that have more than one waste stream/
the maximum concentration of different regulated constituents can occur in
different waste streams; thus, UTLs should be calculated for the different
waste streams; and the UTLs are then composited for all waste streams.
The following example illustrates the calculations for the UTL and for
determining sampling frequency.
A facility generates one waste stream on site from a relatively
constant production process.. The stream has been analyzed several
times for metals, ash, and chlorine and chlorides. The analyses
revealed some .variations in concentrations of constituents. The
level of chromium (Cr) is near the allowable feed rate limit, but
the levels of all other constituents are well below the limits
The facility would like to use the results in its WAP to
demonstrate that the values are below the. concentrations of
constituents used in calculating feed rate calculations. The
facility also would like to specify more frequent analysis for
chromium than for the other constituents.
Because not all the constituents are well below the allowable
limits, it is appropriate in a case such as this to specify
different frequencies of analysis for different constituents. This
UTL of the tolerance interval can be compared with the feed rate
limit for each constituent. The WAP should specify that the upper
limit of the tolerance interval is s to . be used in determining
compliance, and, with future analyses at some reduced frequency, to
verify that concentrations remain below the UTL.
- DRAFT: Do Not Cite Or Quote -
-------
. -19 ,. . . ..
Data from analysis for chromium used for statistical calculations and
calculations of feed rate are: ,
!
- ' Ten samples were analyzed; n 10.
- The mean of the data (average) is calculated; x = 2.39 mg/L.
- The standard deviation is calculated; s = 1.53. . .
The facility established a maximum acceptable concentration of
Cr of 8.47 mg/L for calculations of feed rate.
* _ \
Using the procedures described earlier, UTL is calculated as follows:
\ , .
UTL = X + Ks .
: - \
where: . ' ',.'. . ' - '
X = Mean of the samples (Cr concentration - mg/L)
s = Standard deviation of samples
k = 2.911 f or n =10 (sample size)
\ " ;- "
UTL = 2.39 + (2.911) (1.53)
UTL 6.84 mg/L - .
The UTL of 6,84 mg/L then is compared with the maximum acceptable
concentration of Cr of 8.47 mg/L. For a fixed feed stream feed rate at or
near its maximum, use of the UTL provides a safety margin that ensures that
the regulatory limit for Cr is not exceeded,.
The facility also should determine the' frequency of sampling and analysis for
Cr as part of its sampling and analysis program. That program should be
described in detail in the waste analysis plan. Assuming that the facility
generates waste for 365 days per year, the frequency should be determined as
follows:
Number of samples/year = [ (a^) ] (₯)
where: . . ' - " '. y , "
[(
-------
20
Thus, the facility should sample -the feed stream a minimum of 19 times per
year (rounding up provides an extra degree of certainty). The minimum
frequency of sampling should be once per year.
Site-specific factors can influence the choice of a statistical approach to
compliance and its associated sampling and analysis strategies. Even if a
facility follows the procedures outlined in its WAP, problems related to the
results of analysis may arise. The issues that arise most often are
incomplete data, outliers, and detection limits. These issues are discussed
further in a later section of this document. ,
Stiaunaarv* ' ' .
Below are listed factors to be considered when selecting a sampling and
analysis strategy for each methodology: sampling and analysis by batch,
qualification of the feed stream, or the statistical approach. Such factors
include, but, are not limited to:
Sampling and Analysis by batch
- Appropriate for feed streams generated both on and off-site
- Appropriate for multiple waste streams produced from on-site
processes - - '- -
- Simplicity
- Ease of documentation of compliance
- Waste management system (burn tanks, blend tanks, and sample
location)
- Economic factors related to laboratory analysis
Qualification of the Feed Stream
- Appropriate for multiple feed streams generated on site
- Flexibility with regard to feed rates or constituent
concentrations
- Possible complexity of documentation of compliance
- More complex methodology to establish and execute than a batch
system , -
- More appropriate for situations in which a constant production
rate to generate steam is necessary
Statistical Approach
- Appropriate for "as-generated" waste streams originating oh-
site .. : '.
- Fossil fuels . . t ''.'
- Raw materials ' .
- Requirements for maintenance of the database
- Possible requirements for periodic reestablishment of
statistical baselines for feed streams
Do Not Cite or Quote -
-------
.-,-"' -" ' 21 . . : _ '- .--..
- Measurable and finite probability that a facility might be out
of compliance
Minimum frequency of sampling of once per year
4.0 INCOMPLETE DATA, OUTLIERS, AND DETECTION LIMITS
Because it is important to have complete and accurate data on waste analysis,
it is appropriate to discuss the issue in further detail. The facility's
QA/QC procedures should be set forth in the WAP, The QA/QC procedure should
outline a protocol for dealing with incomplete data, outlying data points,
and detection limits. Such information may be requested during an inspection
and will play an, important role in determining compliance. Therefore, a
facility should maintain it.
4.1 OUTLYING DATA POINTS
In waste analysis data, an outlying data point is one that does not appear to
be within a reasonable or expected numerical range. Such an assumption most
likely will be based on historical data with which a comparison can be made.
When it is suspected that an outlier has occurred, the facility should
determine why it has occurred. The quality assurance procedures submitted
for the analytical test of the sample should include detailed and objective
rejection criteria for all outlying data points. Those criteria could
include procedures for documenting outliers and determining why outliers
occur and what corrective action should be taken to prevent such events from
occurring in the future. Several references are available for evaluating
outliers. For example, a facility may evaluate the validity of its data
using ASTM Method E 178-80, "Standard Practice for Dealing With Outlying
Observations." in applying that and other methods, the underlying
assumptions of the methodology should be kept in mind (for example, ASTM
Method E 178-80 states that "the criteria for outliers are based on an
assumed underlying normal (Gaussian) population or distribution"). Data that
are .suspected of being outliers, but in fact are the result of the character
of the feed stream, or data that cannot be explained otherwise as an error in
sampling or analysis, are not outliers and should bet included in the data
set. Outliers caused by an error in sampling should be corrected through
immediate resampling and reanalysis of the feed stream. Outliers caused by
errors in analysis often are corrected through reanalysis of the sample.. If
the holding time of a sample has expired, the facility should resample and
perform the analysis again. However, it is recommended that the facility
take two or three samples at the same time; if one sample exhibits ah
outlier, the remaining samples can be analyzed. If the facility is using
statistical .analysis and has an outlying data point above the calculated UTL,
it is suggested .the facility use the value of that data point in the
calculations of feed rate until resampling or reanalysis shows different
results. All procedures for identifying and discounting outliers should be
documented in advance in the WAP. . r
,- DRAFT! Do Not cite br Quote -
-------
22
4.2 INCOMPLETE ANALYTICAL DATA
Data from waste analysis are considered incomplete when results for one or
more regulated constituents are missing from the analytical report.
Incomplete analytical data is generally unacceptable in demonstrating
compliance. A facility using sampling and analysis by batch or qualification
of the feed stream should reanalyze. However, analyses conducted pursuant to
a WAP that specifies different frequencies of analysis for different
constituents would not be considered incomplete. Otherwise, for the
statistical program, the facility should use the reported data and reanalyze
to get results for all constituents. As discussed earlier, all quality
assurance should be conducted, documentation gathered, and corrective
measures taken to prevent recurrence of this problem in the future.
4.3 DETECTION LIMITS
The BIF regulations and some HWI permits specify the use of testing methods
set forth in SW-846 for some constituents (for example, 40 CFR §266.106(a)
for metals). Limitations associated with these methods, such as the
detection limits, can present problems in the effort to use statistical
analyses to determine sampling frequencies for some facilities that generate
wastes on-site. When using these methods, such facilities may find that
results of analysis are at or near the detection limit(s) for the
constituent(s). Consequently, it may be difficult to develop a statistical
distribution for the constituent because most of the distribution is below
the detection limit. Therefore, the facility may not be able to use a
statistical approach to determining an appropriate frequency of sampling arid
analysis. In such situations, it may be appropriate to specify that the
facility sample and analyze more frequently (for example, every batch).
Possible solutions to this problem include:
The facility can use/when appropriate, the SW-846 Method 6020 (see
58FR46052) for analysis. (This method was promulgated in January
13, 1995 in the second update of the third edition, second update
of SW-846.) This inductively coupled plasma-mass spectrometry
(ICP-MS) method is a multielement, simultaneous method for the
analysis of inorganic analytes. It is capable of testing for
metals at much lower levels than other SW-846 test methods [parts-
per-billion (ppb) instead of parts-per-million (ppm)]. Use of the
method where allowed by the regulations or permit conditions may
provide better analytical data upon which to base development of a
statistical distribution, because the results of analysis would be
less likely to fall below the lower detection limits.
The facility can, when appropriate, develop a mathematical model to
estimate the statistical distribution of .constituents below the
detection limit. For example, EPA has recommended, in the guidance
document on groundwater monitoring referred to above, the simple
substitution method, under which nondetected results are replaced
by one-half the detection limit. A mean and variance then can be
- DRAFT: Do Not cite Or Quote -
-------
: .. ' 23 . ' ; ,
calculated by assuming all measurements were observable with the
same precision* Another model that may be used, when appropriate,
is the maximum likelihood estimator (MLE). Cohen (1959, 1961)
developed the MLE for calculating the mean and variance of a
, distribution based on the mean and variance of the detected values,
the difference between the mean of the detected values and the
censoring point, and a factor that depends on the proportion of the
data,that are nondetected results. This approach has been found to
work best for small, normally distributed results. A discussion
this Cohen's method can be found in Statistical Methods for
Environmental Pollution Monitoring, J?.O, Gilbert, Van Nostrand,
1987. Also, detailed discussions of other approaches to handling
nondetects can be found in Statistical Analysis Of Groundwater
Monitoring Data At RCRA Facilities, Addendum To Interim Final
1 Guidance (OSWER # EPA/500/R-93/003, July 1992).
Facilities that are having difficulty implementing one of the
approaches described above should establish a frequency of sampling
and analysis for all constituents of at least once per year.s,
5.0 MANAGEMENT OF RESIDUES
Management of residues generated during combustion is an important element in
the operations of facilities that burn nonexempt hazardous wastes. Because
such.residues may be considered hazardous, the relevant concepts are similar
to those discussed in the previous sections of this guidance. For this
document, the term "residue" includes bottom ash generated in the combustion
unit and/or fly ash that is collected in an air pollution control device. As
is the case in the generation of any solid waste, a determination must be
made whether the residues are hazardous wastes. There are three regulatory
requirements governing residues generated from the combustion of hazardous
wastes: . ' . ;- ^
Listed Hazardous Wastes:, All residues derived from the combustion
of a listed hazardous wastes remain listed wastes (40 CFR
§261.3 (c) (2)) [until delisted] and are subject to the land disposal
restrictions (LDR) requirements codified in 40 CFR §268 in
disposing of such residues.
Characteristic Wastes: Residues derived from the combustion of
characteristic wastes remain hazardous unless they no longer
exhibit any characteristic of a hazardous waste. The facility
should sample and analyze the residue to determine whether it
exhibits any of the characteristics. Further, if the waste was one
classified in EPA waste code D002 or D012 through D043 at the point
' of generation, the residue must be analyzed for "underlying
hazardous constituents,11 as defined at 40 CFR §268.2, that can be
reasonably expected to have been present in the waste at the point
of generation (40 CFR §268.7(a)). When meinaging such residues,
- DRAFT: Do Not Cite or Quote --
-------
24 '
facilities also, must comply with LDR requirements set forth at 40
CFR §268.9, "Special Rules Regarding Wastes That Exhibit a
Characteristic." , .
Bevill Exemption: Section 3001(b)(3)(A) of RCRA exempts certain
types of residual materials (generally "high-volume, low-hazard"
materials) from regulation under RCRA Subtitle-C; this is commonly
referred to as the Bevill exemption. Examples of Bevill exempt
material include,residues generated primarily from the combustion
of coal or other fossil fuels and cement kiln dust. The BIF
regulations define which BIF residues are subject to this exclusion
from the definition of hazardous waste (40 CFR §266.112).
5.1 DETERMINATION OF THE BEVILL EXEMPTION
The promulgation of the BIF regulations specifically addressed the issue of
continued applicability of the Bevill exemption when devices burn hazardous
wastes (40 CFR §266.112). This regulation specifically states that "a
residue derived from the burning or processing of hazardous waste in a boiler
or industrial furnace is not excluded from the definition of a hazardous
waste under §261.4(b)(4),(7), or (8) unless the device -and the owner or
operator meet the requirement (described below)." The first requirement
states that the device must be a boiler that burns at least 50 percent coal
or an ore or mineral furnace or cement kiln that processes at least 50
percent by weight normal, nonhazardous materials. The second requirement
mandates testing to determine whether the residues have been affected
significantly by the hazardous waste, thus causing them to no longer be the
"high-volume, low-hazard" material that the Bevill exemption was intended to
cover. That determination is achieved through, either of two tests that show;
» Test One: The waste-derived residue does not contain toxic
constituents at concentrations significantly higher than is
exhibited by the residue generated when hazardous wastes are not
burned.
Test Two: The concentration of toxic constituents does not exceed
health-based limits identified in the regulation.
i
The regulation at 40 CFR §266.112 requires that the waste-derived residue be
sampled and analyzed "as often as necessary to determine whether the residue
generated during each 24-hour period" meets requirements to qualify under the
Bevill exclusion. However, no specified frequency is identified for making
such a determination. Therefore, the discussions that follow will focus
primarily on the issue of frequency of sampling and analysis as it affects
facilities that attempt to claim the Bevill exemption.
- DRAPTs Do Not Cite Or Quote -
-------
,. . 25 : - . " '
5.2 FREQUENCY OF SAMPLING AND ANALYSIS
The first step a facility may take in determining the frequency of sampling
and analysis is to develop a WAP that addresses the frequency of sampling and
analysis of the residues and management practices for disposal of the
residues or to include those subjects in its overall WAP for the facility.
Two options a facility may consider when selecting a frequency, which are
similar to the options described earlier, are sampling and analysis by batch
and sampling at a reduced frequency, with statistical analysis. Both
approaches may be appropriate for residues generated from the combustion of
characteristic wastes.
* Sampling and analysis by batch: Since the regulation governing the
Bevill exemption requires that sampling and analysis be
representative of residues generated during a 24-hour period, daily
sampling and analysis is acceptable for a batch frequency. The
results of such analysis then should be compared with the limits
established through either of the two tests described earlier.
« Statistical Analysis; If a facility chooses* to sample and analyze
less frequently than daily, the facility should be prepared to
provide a technical justification of the appropriateness of the
lesser frequency and an explanation of how the results of analysis
represent the 24-hour periods during which residues were not
sampled to determine eligibility for the Bevill exemption. Using
the.methods of statistical analysis described earlier, a facility
might be able to establish that a less frequent sampling is
adequate. The facility should consider that, when sampling at a
reduced frequency based on statistical analysis, there is some
chance that the facility will be out of compliance.
S.S CONSIDERATIONS
When determining the frequency of sampling and analysis, several site-
specific factors should be considered. Some of those factors are discussed
below.
Sampling: To obtain a representative sample for a 24-hour period,
40 CFR §266.ll2(bj (1) (ii) and (2) (iii) state that one or more
samples may be taken, provided that the sampling does not exceed
: the 24-hour period. If more than one sample is taken, the samples
may be composited or analyzed separately. The regulations do not
specify the number of samples that may be taken within the 24-hour
period. However, the facility should specify in the WAP a
frequency that will account for any spatial or temporal variations
in the residues. The location from which the sample is taken is
another factor that should be considered in obtaining a
representative sample. According to requirements set forth at 40
CFR §266.112, the residues must riot conta in toxic compounds at
levels above the limits established in either of the tests
- DRAFT: Do Not Cite Or Quote -
-------
26
discussed earlier that could reasonably be attributed to the
hazardous waste. Therefore, samples should be collected in a
manner that minimizes any environmental contamination that is not
attributable to the burning, or processing of hazardous waste.
Samples should be taken from a location as close to the residue
outlet(s) as practicable; that location should be identified in the
WAP. Any sampling' conducted at a location other than that
specified in the WAP may not be considered valid.
Management of Residues: Because of the potentially large volumes
of residues generated in any 24-hour period, it is possible that a
facility may have disposed of the residue after a sample of the
residue had been taken but before the results of analysis had been
received. The problem arises when results show that the residue is
a hazardous waste and the residue is disposed of in a non-hazardous
disposal area or unit. A similar problem becomes especially
significant for facilities conducting sampling at reduced
frequencies. As the preamble to the BIF regulations set forth in
the August 27, 1991 BIF Federal Register states that "if the waste-
derived residue is sampled and analyzed less often than on a daily
basis, and subsequent analysis determines that the residue fails
the test and is fully regulated hazardous waste, the Agency
considers all residue generated since the previous successful
analysis to be fully regulated hazardous waste absent documentation
otherwise." In addition, residue generated after the failed test
may also be considered a hazardous waste until the next passing
test. Therefore, in all of these,scenarios, the facility risks not
only the residue becoming subject to the RCRA regulations, but also
the disposal area or unit becoming subject to RCRA Subtitle C
requirements. Resampling of the residue in the disposal area would
not generally be acceptable, because the area would not normally be
the appropriate sampling location and the residue found there may
not be representative of > the residue generated over a 24-hour
period. To minimize the extent to which disposal areas are subject
to RCRA regulations, a facility may want to implement certain,
disposal management practices. For example, management practices
controlling disposal of residues into a quarry on-site may include:
- Transfer by a dedicated truck for disposal
- Disposal in specific segregated locations in the quarry
- Documentation of disposal practices and locations ' ,
« Other Factors: Other factors to consider when selecting a
frequency of sampling and analysis include:
- Feed rate of wastes
- Levels of volatility of metals in the waste . '
- Physical form of the waste (for example, solid rather
than liquid)
- Waste feed system
- DRAFT: Do Not Cite Or Quote -
-------
27 '..'...'"
- Levels and types of organic constituents in the waste
(for example, difficulty of destruction or formation of
by-products)
- Levels and types of metals regulated under RCRA, other
than those regulated by the BIF regulations (for example,
selenium)
- Changes in feed streams j '
- Changes in operating conditions or equipment
- Operating conditions when sampling compared with those
when not sampling
- Trends in partitioning of metals in fly as compared with
bottom ash
5 . '-" *
6.0 DOCUMENTATION TO DEMONSTRATE COMPLIANCE
Documentation of compliance consists of detailed and complete records of a
facility's activities that are regulated by either permit conditions or
regulatory requirements (e.g., 40 CFR 266.103(j)). Some conditions governed
by permits can be measured and recorded directly, while others may require
indirect measurement or calculation. For example,.real time and continuous
monitoring systems for such stack gases as CO, O2» and SO2 are in widespread
use. However, no monitors are available for effectives real-time measurement
of metals or total chlorine in the stack gases. Therefore, compliance with
those limits on emission rates are demonstrated indirectly by calculation of
the feed rates of the constituents of interest.
Essentially, compliance with a permit or the regulations is demonstrated by
showing that a facility burned wastes only under certain specified
conditions. Those conditions usually are stated as maximum conditions,
minimum conditions, or conditions over a. specified range. Documentation and
recording .of those conditions provide the basis for determinations of
compliance, not only by personnel of EPA or state agencies but also by the
facility itself. The operators of a facility should have a detailed
understanding of permit or conditions governed by permit or regulation that
determine compliance or noncompliance. Since no two facilities are identical
in permitted or regulatory operating conditions, no all-encompassing check
list is feasible. However, the items listed below form a basis for a logical
and coherent approach that will enable a facility to document combustion
conditions and other parameters required by permit or regulation. The list
is not all inclusive, and each facility should tailor it to that facility's
own needs.
Facility operators should consider the following: . ,
. Specific responsibility for compliance should be assigned. While
owners and operators ultimately are responsible for compliance,
they may not be involved in this issue daily. -. Therefore, a
designated individual, with the necessary number of backups, should
- DRAFT: Do Not Cite or Quote -
-------
28
be responsible for ensuring that activities necessary for
compliance are carried out and documented. (This step is the basis
of all subsequent actions).
A written compliance methodology is recommended. It should
identify the responsibilities of individuals associated with the
combustion of hazardous waste. The document should.be tailored to
the specifics facility, with emphasis on regular use by operators.
In addition, it should be, available to personnel of EPA or the
appropriate state agency. Decision trees and actions to be taken
in the event of failures or other situations should be a principal
focus of the document. Conditions to be satisfied before hazardous
wastes are burned should be set forth in the document.
A WAP that contains the items required by permit or regulation is
vital and mandatory (40 CFR §264/265.13). The plan, at a minimum,
should cpver sampling locations and frequency of sampling,
statistical methodology (if applicable), procedures for handling
outliers and nondetects, methods of preparation and analysis"of
samples, and QA/QC procedures and should identify the laboratories
to be used. The foregoing list is not all-inclusive, since it may
be necessary to meet other requirements. The result of application
of procedures in a properly constructed WAP is that .the facility
will know what it .burned and when. As an example, for any day on
which waste is. burned, corresponding data on waste analysis should
be available that can be linked directly with that day's activity.
The data should be available from whatever method of sampling and
analysis is used. A facility's lack of this information can lead
to determinations of noricomplianceo
Although not specifically required, computer hardware and software
systems can be useful in providing a record of the most crucial
data. The systems can be tailored to meet almost any permit
condition or requirement. They enable the inspection of operating
records with relative ease. Care should be taken to select a
system that meets the needs of the facility, can accommodate
upgrades of software, and facilitates compliance rather than merely
reporting on compliance status.
Recordkeeping that.documents compliance is recommended. Records
should include, but not be limited to: waste analysis data;
records of continuous monitoring of feed stream rates; statistical
data; data on permit or operating limits, such as temperatures;
parameters for air pollution control devices; and any significant
operating requirements or constraints. It is important to note
that recordkeeping to support the sampling and analysis strategy
used will provide data on operating conditions and limits that form
the basis for enforcement actions --for example, feed stream
qualification values or UTLs. Finally, if the facility chooses to
change from one compliance strategy to another (for example, from
- DRAFT: Do Not cite Or Quote -
-------
' ;'' 29 ' :
sampling and analysis by batch to statistical analysis), complete
and supportable records should be kept to document the change. It
also is prudent to keep records of compliance in a single location,
with a remote backup. This precaution applies not only to computer
data storage but also to paper records.
Again, these items should be merely a starting point in the documentation
effort and should be adjusted or expanded to meet the specific needs of the
facility.
The objective of maintaining these items is that the facility will be able to
demonstrate compliance to regulatory officials and the public. For example,-
if an inspector has a-, difficult time determining that a facility is in
compliance with feed rates for various constituents, the inspector's
underlying assumption may be that the facility is having the same difficulty.
Such an assumption could lead inspectors to believe that the facility is out
of compliance. Well-maintained documentation, can prevent potentially
unnecessary actions.
- DRAFT: Do Hot Cite Or Quote -
-------
Appendix A
Table of K Factors foe Calculation of Tolerance Limits
-------
TABLE 1. K factors for Calculation of Tolerance Limits for
95 Percent Confidence and 95 Percent Proportion
5«:*S*}SS£»aiBp.i'B«il3 iJBJB «:₯ %:W^J::-i.5;^g::g:::;::::^^^yv^::
2 . ' :' '
3 '
' - 4 -'
. . 5 ' . '
6 .
7
8
.' ' 9
10
11
12
13
14
15
1.6
17
; is
19
20
21
22
23
24
25
26
7
5
4
3
3
3
3
2
2
2
2
2
2
2
2
2
2
2
2
2
2
2
2
.260-
.656
. 144
,203
.708
.398
.187
.031
.911
.815
.736
.670
.614
.566
.524
.486
.458
.423
.396
.371
.349
.328
.303
,.292
From: Hahn, Gerald S. and William Q. Meeker.
Intervals: A Guide for Practitioners.,
(ISBN 0-471-88769-2)'.
1991. Statistical
Wiley Interscience.
- DRAFT! Do NotlCite Or
------- |