EPA/600/A-92/031
PAST: The Potential ARARs Selection Tool
Daniel Greathouse
Risk Reduction Engineering Laboratory
U.S. Environmental Protection Agency
Cincinnati, Ohio
Jay Clements
Applied Technology Division
Computer Sciences Corporation
Cincinnati, Ohio
ABSTRACT
The 1986 Superfund Amendments and Reauthorization Act specified
that any remedial actions at Superfund sites must comply with applic-
able or relevant and appropriate regulations (ARARs), including
federal, state and local environmental statutes. Identifying these legal
requirements for a particular hazardous waste site can be a complex
and time-consuming process.
The U.S. EPA Risk Reduction Engineering Research Laboratory
is developing a prototype knowledge-based system to aid in screening
regulations that determine cleanup requirements based on unique site
characteristics. The decision logic follows that presented in the U.S.
EPA documents such as CERCLA Compliance with Other Laws
Manuals (Parts I and H). Hence, it takes into consideration the chem-
icals at the site, the particular location features of the site and the
proposed remediation methods.
INTRODUCTION
SARA modified the 1985 NCP to require remedial actions to comply
with all federal and state environmental requirements. SARA defines
two classes of legal requirements that must be addressed prior to
remediation work at a Superfund site. "Applicable requirements are
those cleanup standards, standards of control, and other substantive
environmental protection requirements, criteria, or limitations promul-
gated under Federal or State law that specifically address a hazardous
substance, pollutant, contaminant, remedial action, location or other
circumstance at a CERCLA site. Relevant and appropriate require-
ments are those cleanup standards, standards of control, and other
substantive environmental protection requirements, criteria or limi-
tations promulgated under Federal or State law that, while not "applic-
able" to a hazardous substance, pollutant, contaminant, remedial
action, location, or other circumstance at a CERCLA site, address
problems or situations sufficiently similar to those encountered at the
CERCLA site that their use is well suited to the particular site."'
Collectively these environmental requirements are known as ARARs.
In 1988, 200 U.S. EPA hazardous waste decision-makers were sur-
veyed to identify needs and opportunities for development of auto-
mated decision support systems. An ARARs screening aid was among
the most frequently expressed needs. Several characteristics of the
ARARs selection problem may explain this finding. First, many of
the decision-makers have technical rather than legal backgrounds and
interests. Although quite competent in their respective scientific fields,
many of those charged with ARARs decision responsibility feel uncom-
fortable operating in the unfamiliar legal domain. Second, most
decision-makers have insufficient time to adequately review the large
body of federal and state regulations that must be examined in order
to conduct a thorough ARARs search. Finally, the criteria and rules
for determining whether a regulation is relevant and appropriate hi*
continued to evolve over time. Thus, an acceptable decision 1 yt*
ago may be judged unacceptable by today's standards. The aforen*#.
tioned characteristics describe a problem which is amenable to sob-
tion via an expert system.
Briefly, an expert system is computer software that is constructed
using special coding techniques which allow the program to reprodna
the knowledge, experience and decision logic used by experts in i
certain field of specialization. Expert systems technology can be ustrf
to encapsulate the experience of environmental law professionals la!
subsequently dispense this knowledge to the appropriate scientific com-
munity to assist with ARARs selection. The automated system az
search the large body of information under consideration much more
efficiently than even the most experienced environmental lawyer
Furthermore, by implementing an expert system solution, the process
of cataloging the most current regulations and policy enhancement!
can be centralized, thereby alleviating individual decision-makers of
this responsibility. It is anticipated that a system such as the one
described will be a valuable asset for use in the U.S. EPA Region^
Offices, state environmental departments and the offices of remedi*-
tion support contractors who currently perform many of the tasb
associated with environmental restoration.
Typically, ARARs are identified during the Feasibility Study (FS)
stage of a site investigation by project engineers associated with the
performing institution (consulting firm, Army Corps of Engineetv
Principal Responsible Party, etc.). The list of selected ARARs b
reviewed by the U.S. EPA Regional Project Managers (RPMs) *nd
then sent to the attorneys in the respective U.S. EPA Office of Region*!
Counsel for final review and concurrence. The Policy and Analys°
Staff (PAS) in the U.S. EPA Office of Solid Waste and Emergency
Response is responsible for establishing the criteria for identification
of ARARs under varying circumstances. Hence, the Agency attorney1
and the PAS are the experts relative to the selection of ARARs- Theft-
fore, a situation exists in which a number of people in widely scatter®
geographic areas and organizational entities need the expertise retail*"
by a limited number of recognized experts. This is a typical technologj
transfer scenario which fits nicely with the paradigm of expert system^
Expert systems provide a means of readily capturing this decision lop*-
in a form that can be easily distributed to others.
The U.S. EPA Risk Reduction Engineering Laboratory has bees
developing an expert system to facilitate selection of potential
for approximately 1 year. This paper focuses upon and details
development effort.
SYSTEM OBJECTIVES
Clearly, the global objectives of any automated decision SUPP^
system are that it provide accurate and reliable advice that is of s"
632 RISK ASSESSMENT

-------
r • nt utility to the targeted users to warrant the time and resources
faired to use it. More specifically, the system must provide advice
increases the efficiency of targeted decision-makers in addition
^proving the quality of their decisions. In terms of ARARs iden-
"fication, the objective is to provide an easy to use system that
"^lately selects the ARARs for commonly encountered, typical sites
Vj situations. This point is emphasized because it is not possible
entirely automate the ARARs selection process. The most complex
Rations will require specialized legal advice which cannot be
presented with today's limited knowledge based tools. Furthermore,
to these complex cases, the ARARs eventually followed at a site may
the result of negotiation or litigation. However, if an automated
Mproach can be employed to aid in the more routine scenarios, scarce
fcgal resources can be freed to address the more involved cases.
The selection process follows the general approach described in the
U.S. EPA published manuals CERCLA Compliance with Other Laws1
,ad CERCLA Compliance with Other Laws, Part II,1 Options are
puvjded by the system to permit selection of chemical, location and/or
jction specific regulations. It will provide the capability to assess the
[tgulatory implications of alternative remedial alternatives without
requiring the user to consult multiple sources such as the Code of
Federal Regulations, State regulations, CERCLA Compliance manuals,
supplemental Office of Solid Waste and Emergency Response
(OSWER) ARARs publications and NCP information. By having the
developers of the PAST System work closely with both the U.S. EPA
Policy and Analysis Staff and Agency attorneys, PAST will serve as
the mechanism for the implementation of the latest ARARs decisions.
As newly hired personnel conduct consultations with the system, it
will tend to function as a training tool for these less experienced
personnel. Finally, the system will serve to promote completeness and
uniformity in ARARs assessment among decision-makers in scattered
geographic and organizational entities.
SYSTEM DESCRIPTION
Several initial assumptions shaped the development of the PAST
system. It was assumed that the most common computing platform
available to potential users of the system is the IBM compatible per-
sonal computer. In addition to its wide availability throughout the U.S.
EPA, this machine is common in the offices of state environmental
protection organizations and Superfund remediation contractors,
However, due to the complexity and volume of information required
to solve the ARARs problem, a powerful computing environment was
required. A compromise was reached which allowed the developers
to target the PAST system to more advanced, 386 based personal com-
puters. Specific hardware requirements are documented within the
User's Guide for the system.
Another assumption was that the knowledge necessary for solution
of the ARARs problem could be best represented in a rule-based
format. Furthermore, the system would be originally coded as a proto-
type and then the iterative development methodology will be applied
to refine this prototype into a fully functional system. These refine-
ments will represent the knowledge gained through a number of
interviews with persons considered expert in the ARARs selection
Process. Finally, the system will be coded at the highest possible level,
ie., the system developers will concentrate upon representation of
knowledge and relationships required for solution of the ARARs
problem rather than upon development of the underlying software
required to process such knowledge and relationships. All of these
Actors mandate that a suitable knowledge-based development tool be
Procured and utilized for implementation of the system. After analysis
of the commercial market for such tools, KAPPA, a product of Intelli-
°°rp of Mountain View, California, was chosen for the initial prototype.
In addition to providing the rule-based reasoning capabilities re-
tired by the ARARs application, KAPPA implements the object-
oriented programming paradigm which has been shown to increase
Programmer productivity and reduce resources necessary for system
®aintenance. Also, KAPPA runs within the Microsoft Windows com-
puting environment. This emerging standard for personal computing
Provides operating system support for a graphical user interface, multi-
asking, and interprocess communications. The graphical user inter-
face simplifies access to and use of the system. It also enables the
developers to implement device independent graphics for system I/O
with a minimum of programming and maintenance effort. Multitask-
ing and interprocess communications are also used by the developers
to build a more sophisticated application than could be built under
DOS.
Although the system knowledge base is written in KAPPA, the user
interface is written in ToolBook by Asymetrix Corporation of Bellevue,
Washington. ToolBook provides for rapid and flexible design of
Microsoft Windows user interfaces and as such is more suitable for
development of a graphical user interlace than KAPPA. As its name
implies, ToolBook uses the metaphor of a book for application de-
velopment. Each screen within the application is called a page. Tool-
Book can quickly move from page to page within a book. All system
interaction with the user is handled by ToolBook. Input data collected
by ToolBook is passed to KAPPA for processing. Likewise, conclu-
sions generated by KAPPA are passed to ToolBook for display to the
user. Interprocess communication between KAPPA and ToolBook is
accomplished through the use of Windows Dynamic Data Exchange
(DDE). DDE allows these two distinct applications to share informa-
tion in a common format. The performance of DDE is such that in-
formation can be passed between the two applications almost as quickly
as information can be accessed in a single application.
System rules define the relationships between site characteristics
and potential ARARs. These rules embody the decision criteria used
by the system. For example, the following Action Specific rule:
MakeRule( excavation	rulel, [Anlnstance excavation]
Anlnstance:exists
AppendReg( Anlnstance, regulation, CFR 40 268 SubpartD )
defines the fact that whenever an excavation action takes place at a
Superfund site, 40 CFR 268 Subpart D is a potential ARAR at this
site. It is apparent from this example that the system must determine
Superfund site characteristics before potential ARARs can be com-
puted. In general, site characteristic information must be supplied to
the system by the user.
Within PAST, site characterization information is represented in
objects. Objects are created to parallel the real world tangible items
and concepts which play a role in the specification and
solution of the ARARs problem. Objects can be thought of as reposi-
tories of related data. Slots within the objects are used to hold the
individual data elements. For example, if excavation of contaminated
soils occurs at a site, the system would contain an EXCAVATION object
to represent this action. Possible slots within this object might be depth,
volume, soil type, etc. The values of these slots would be used by
the rules to make determinations concerning ARARs.
Creation and manipulation of objects is controlled via KAPPA Ap-
plication Language (KAL) code. The user interface initially requests
key pieces of site information from the user. For each site datum
gathered by the user interface, there is a corresponding slot within
a KAPPA object. These slot values are set within KAPPA via a message
sent by the user interface. In addition to setting the slot value, the
slot is asserted to KAPPA's inference engine. Once asserted, this datum
causes system rules that reference this slot in their premise to be
examined. Depending upon the format of the rule, additional infor-
mation may be required and requested from the user. If all the condi-
tions of the premise are TRUE, the rule is executed. At this point,
the actions specified in the conclusion of this rule are carried out.
Usually this entails either setting additional slot values or concluding
that a particular regulation is indeed an ARAR.
Figure I shows the most general levels of the KAPPA object hier-
archy. This object hierarchy can be thought of as a very lai^e tree
diagram. The more specific branches of this tree are not shown in
Figure 1 for the sake of readability.
The ARARs object has three children: Location Specific, Chemi-
cal Specific, and Action Specific. These objects store each of these
three types of regulations after they are identified by the rule base
as potential ARARs.
The Regulations object is the toot node of a large subtree which
contains one object for each regulation defined within the system. Each
RISK ASSESSMENT 633

-------
Root
ARARS 	
Regular Ions..
Sit© crwacte*"is
-<
Locatfon Soeclflc
Cnemrcal Specific
Action Specific
contamf rants ., .
)ocatI on.,.
rsfnedfal action...
Figure 1
Genera] Levels of the KAPPA Object Hierarchy
specific regulation object contains a brief description of the content
of the regulatory citation. In general, these objects are referenced in
the conclusions of the system rules.
The Site Characteristics object is the root node of the tree which
stores the Superfund site characterization data input by the user. Each
of the three classes of site characteristics (contaminants, location,
remedial action) is represented as a subtree of Site Characteristics.
In general, objects within these subtrees are referenced in the premise
of the system rules. As such, they are used as the basis for determining
potential ARARs.
The location subtree is the least complex of the three Site Charac-
teristics subtrees. The extent of this subtree is shown in Figure 2. Each
of the child objects of location has one or more slots which correspond
to user supplied information concerning locational characteristics of
the site.
contarn i nants
characterj j
meta is
radioactive
mun i t i ons
pest i c i des
a Iiphat ics
aromatics
d i ox i ns
phenoIs
pahs
organometa [ l iCs
PCBs
asbestos
cyanides
oils
ac ids
gases
water related
*1Idlife related
location 	
historic character 1st Ics
geographies
Figure 2
Location Subtree
The knowledge representation for a chemical specific analysis is
somewhat more complex. Site characterization information for the
chemical specific analysis is found in the contaminants subtree of Site
Characteristics. The extent of this subtree, as it exists when the system
is originally started, is shown in Figure 3.
The location subtree (Fipre 2) described above is a static hierarchy.
Although slot values are added as the user inputs information, the
number and relationship of the objects is constant. This is not true
of the contaminants hierarchy. As shown in Figure 3, the contaminants
subtree has several subclass objects which indicate categories of poten-
tial pollutants for Superfund sites. These categories correspond to the
ToolBook selection menu page named contaminants.
In some cases, when a category is selected from this contaminant
menu, a page showing the specific types of contaminants in this
category is shown. Each type of contaminant is listed next to a check
box so that zero or more of these contaminant types can be selected.
By selecting a specific contaminant, the user is defining the existence
of this pollutant at the site. Upon doing so, ToolBook sends a message
to KAPPA and KAPPA responds by creating a new class object using
the CreateContamClass function, litis newly created class is a child
of the contaminant category class and has the name of the specific
contaminant type chosen. A slot named exists is automatically inherited
from the contaminants object. This exists slot is Asserted, which will
cause rules which reference this slot in their premise to be examined
and possibly executed.
Site characterization information for the action specific analysis is
represented b>y the remedial action subtree of the Site Characteristics
object. The immediate descendents of remedial action are shown in
Figure 4. These immediate descendent objects serve to categorize the
specific technologies defined in the system. The full extent of the
remedial action subtree is too large to be diagrammed.
634 RISK ASSESSMENT
rerredi e» I oct ior» -
Figure 3
Contaminants Subtree
{ 6ur*foce«mtl I izetion ir»Hu treat"**!
.. deweter fng. ..
-	%torave ¦ . .
-	soll collection..,
s. oss col lection. .
» gee treatment..
. drurJdetyr is conte I rwnent .
. groundwater trettu treat**?** . • •
air water aeoaraiar
physical treatment, , .
¦chemical treatment...
• biological treatment...
thermo I destruction...
Figure 4
Remedial Action Subtree
Like the contaminants subtree, remedial action is a dynamic hier-
archy. As the user defines the remedial alternatives under considera-
tion for a site, instance objects are created to represent these cleanup
technologies. This operation is performed by the KAPPA Create-
Technology function. Instance objects ait given arbitrary names «
the format parentX, where parent is the name of the technology
of which the instance is a descendent and X is some positive
These instance objects can be uniquely identified by the value of we
stream name slot which is set to the user defined treatment stream
in which this technology is implemented. Unlike the location
chemical specific rales, the results of the action specific rules an-P~
into the regulation slot of the technology instance that triggered them-
The KAPPA CompileReport function takes the values from
nology instance regulation slots and formats the output in the P°~l
tial slot of Action Specific so that the triggering technology and stiew
name is listed prior to the regulatory citations.

-------
As previously mentioned, analysis of action specific ARARs is not
a stand alone domain. Often information on site contaminants is neces-
sary to determine all action specific requirements. Within the user
interface, each treatment stream can have zero or more contaminants
associated with it. The specific contaminants available for associa-
^on are taken from the list of all site contaminants defined by the user.
This is why, if in the preliminary data entry stage the user requests
an action specific analysis but no chemical specific analysis, menus
for definition of site contaminants still appear. Associating a con-
taminant with a treatment stream indicates to the system that this con-
taminant is present in the medium being acted upon by this treatment
stream. When this association is made, an instance object is created
in the KAPPA object hierarchy to represent this contaminant. This
operation is done through the CreateContamlnst function. Like tech-
nology instances, contaminant instances are given arbitrary names in
the format parentX where parent is the name of the contaminant class
created when this contaminant is defined at the site (on the ToolBook
contaminants page). X is some positive integer. The stream name slot
within the contaminant instance is used to uniquely identify the ob-
ject. If a requirement is dependent upon both a specific remediation
technology and a specific contaminant within the media, the rule must
check for the existence of the technology and the contaminant and
then verify that these exist within the same treatment stream. The
following rule is an example:
MakeRule( pcb	rulel, [aContam pcbs Anlnstance storage],
aContam:exists And Anlnstance:exists And
aContam: stream_jiame
H = Anlnstance: stream	name,
AppendReg( Anlnstance, regulation, CFR 40 761 65 )
AppendReg( Anlnstance, regulation, CFR 40 761 180 )
This rule embodies the fact that 40 CFR 761.65 and 40 CFR 761.180
only apply when storage is used to house PCB contaminated waste
ai a site.
After all such rule inferencing is complete, a function is invoked
which generates the system output report. Output citations are segregat-
ed into chemical, location and action specific categories. For each
citation, a short descriptive title is attached that allows the user to
ascertain the subject of this regulation. The report is displayed in a
scrollable window for easy viewing. Options which allow the user
to print or save the report are also available.
FEATURES (CURRENT AND PROPOSED)
The ARARs selection approach used in the PAST system is similar
to that described in the U.S. EPA documents entitled CERCLA Com-
pliance with Other Laws and CERCLA Compliance with Other Laws,
II. Searches can be performed for chemical, location and/or action
specific regulations. In response to the Chemical Specific option, the
system outputs regulatory citations that apply based upon the presence
of specific contaminants at the site. Often these are references to the
tobies of maximum contaminant limits that are contained in the Code
of Federal Regulations. Future additions to the system will also include
access to the appropriate state regulations. Selection of the Location
Specific option causes the system to inquire about the location charac-
teristics of the site. This user supplied location information is used
the system to select and output citations to the CFR that are poten-
cy applicable or relevant and appropriate. Examples of site charac-
fcnstics that trigger the Location Specific rules are location in a flood
Plain, wetlands or a historical preservation site. Selection of Action
Specific ARARs is a task requiring more complex reasoning on the
of the system and more input from the user. Treatment trains must
* specified for each contaminated medium at the site. Additionally,
Multiple treatment trains for the same medium type may be defined,
"is process allows the user to represent a case in which the site is
Se8regated into distinct operable units. Based upon the remediation
Jfchnologies and contaminants of concern within each treatment train,
r* system provides a list of regulatory citations that are potential Action
Specific ARARs.
The prototype system which was completed at the end of Septem-
r '^91, only contained information about federal regulations that
are applicable. The system is currently being expanded to include rules
for identification of relevant and appropriate federal regulations. Regu-
lations for a few selected states are also being included. Other features
to be added at a later date include an output table that provides a brief
description of the regulatory requirements for the potential ARARs.
It will also include an ability to access the full text for any of the listed
citations. Context sensitive Help and Explanation facilities will be in-
corporated to define technical terms used by the system and enable
the user to interactively examine the system's reasoning process and
rationale for selection of potential ARARs.
DISCUSSION
Compliance with ARARs is one of the key considerations in the
selection and design of a treatment program for Superfund sites. Hence,
it is very important that the ARARs selection process be as accurate
as possible. The major concern is that the system not miss any regu-
lations that may potentially be ARARs. This type of error could be
very expensive. It might be necessary to perform major modifications
to the selected treatment process, result in further contamination to
the environment and/or have significant human health consequences.
On the other hand, if too many regulations are selected, time and
resources will be required by knowledgeable persons to carefully screen
the list of selected regulations and/or determine those treatment com-
ponents that might be unnecessary for the selected design. Hence,
it is important that the system be as accurate as possible, erring on
the side of over selection of regulations.
One of the concerns with this system is the computer and resource
requirements to make it useful while still useable by the targeted user
community (project engineers in the Alternative Remedial Contract
Strategy [ARCS] firms and U.S. EPA regional project managers). To
expedite the design and development process and reduce the antici-
pated long-term maintenance effort, high level special purpose
microcomputer tools are being used (Toolbook for the user interface
and KAPPA for knowledge representation and inferencing). These tools
are integral with the Windows environment. Also, due to the over-
head associated with these tools and the size and complexity of the
ARARs selection process, considerable computer memory is needed
(approximately 4 megabytes of RAM). Another concern is the distri-
bution costs associated with the KAPPA expert system shell. Due to
these concerns, the issues associated with distribution of the end
products to targeted users still have not been resolved.
Acquiring accurate and reliable information has been one of the
difficulties encountered while developing this system. The final
authority relative to ARARs selection rests with the Agency attor-
neys. Also, the U.S. EPA Policy and Analysis Staff (PAS) formulate
guidance for ARARs selection. Since no comprehensive documents
exist about the ARARs selection process, it is necessary to rely on
the input of the attorneys and representatives of PAS. As might be
expected, the supply of knowledgeable persons from these groups is
limited, making it very difficult to get the input that is needed. As
is normally done, we had to become quite familiar with the regula-
tions ourselves and to rely on the input of persons who perform the
initial ARARs selection and review. This is a matter of concern since
it has been reported that at least 50% of the initial documents listing
the ARARs for sites are in error. Hence, the need for thorough review
of the rules and knowledge contained in the system.
Due to the constant evolution of rules/criteria for identifying ARARs
and changes in regulations themselves, it is anticipated that main-
tenance will require a constant commitment of resources. A large
proportion of the persons who have provided input to the system caution
that without continued maintenance, so that the system reflects current
regulations and ARARs selection processes, the system will have
limited utility.
CONCLUSIONS
Development of a system to select regulations that may be potential
ARARs is a complex and challenging effort. The potential economic
and environmental consequences of errors in the selection process
warrant, however, the costs of system development. Additionally, when
the cost of time currently spent on ARARs selection throughout the
RISK ASSESSMENT 635

-------
Agency is taken into consideration, it is very likely that use of the
PAST system can save the U.S. EPA many times the development and
maintenance costs,
REFERENCES
1. U.S. EPA, CERCLA Compliance with Other Laws, EPA'540/G-89/Q06. US.
EPA, Washington, DC, August 1988, p.xiii.
2. U.S. EPA, CERCLA Compliance with Other Laws: Part II. Clean a:
and Other Environmental Statutes and State Requ,rem'
EPA/540/G-89/009, U.S. EPA OSWER Directive 9234.1-02,
DC, August 1989.
«6 RISK ASSESSMENT

-------
TECHNICAL REPORT DATA
(Please read Instructions on the reverse before complet'
1. REPORT NO. 2.
EPA/600/A-92/031
3.
4. TITLE AND SUBTITLE
PAST - The Potential ARARS Selection Tool
5. REPORT DATE
6. PERFORMING ORGANIZATION CODE
7. AUTHOR(S)
Daniel Greathouse and Jay Clements
8. PERFORMING ORGANIZATION REPORT NO.
9. PERFORMING ORGANIZATION NAME AND ADDRESS
Risk Reduction Engineering Laboratory
26 W. Martin Luther King Drive
Cincinnati, Ohio 45268
10. PROGRAM ELEMENT NO.
11. CONTRACT/GRANT NO.
68-W0-0043
12. SPONSORING AGENCY NAME AND ADDRESS
Risk Reduction Engineering Laboratory—Cincinnati, OH
Office of Research and Development
U.S. Environmental Protection Agency
Cincinnati, OH 45268
13.TYPE.OF	REPORT AND PERIOD COVERED
published paper
14.	SPONSORING AGENCY CODE
EPA/600/14
15.supplementary notes Pr0J-ect officer = Daniel Greathouse, FTS/684-7869 COM/569-7869
HMCRI's 12th Annual National Conference Proceedings - Hazardous Materials Control/
Superfund '92. 12/3-5/91. Washington. D C . PP *637-636
16. ABSTRACT
The 1986 Superfund Amendments and Reauthorization Act (SARA) specified that any
remedial actions at Superfund sites must comply with applicable or relevant and
appropriate regulations (ARARS), including Federal, state, and local environmental
statutes. Identifying these legal requirements for a particular hazardous waste
site can be a complex and time consuming process.
The EPA Risk Reduction Engineering Research Laboratory is developing a prototype
knowledge-based system to aid in screening regulations that determine clean-up
requirements based on unique site characteristics. The decision logic follows that
presented in the EPA documents such as CERCLA Compliance with Other Laws Manuals
(Part I and II). Hence, it take into consideration the chemicals at the site, the
particular location features of the site, and the proposed remediations methods.
17. KEY WORDS AND DOCUMENT ANALYSIS
a. DESCRIPTORS
b.IDENTIFIERS/OPEN ENDED TERMS
c. COSATI Field/Croup
software, regu1 ations,
environment,
expert systems
hazardous waste

18. DISTRIBUTION STATEMENT
RELEASE TO PUBLIC
19. SECURITY CLASS (This Report)
IINm ASSTFTFD
21. NO. r\f PAGES
7
20. SECURITY CLASS (This page)
IIMfl ASSTFTFn
22. PRICE
EPA Form 2220-1 (R«v. 4-77) pbeviou

-------