Chemical Accident Safety: EPA's Responsibilities for Preparedness,
Response, and Prevention (Letter Report, 06/27/96, GAO/PEMD-96-3).

GAO reviewed the Environmental Protection Agency's (EPA) role in
implementing a national chemical accident safety policy, focusing on:
(1) trends in chemical accident occurrence and impact; (2) EPA chemical
accident preparedness activities; (3) the effectiveness of EPA response
to chemical accidents; and (4) EPA actions to help prevent chemical
accidents.

GAO found that: (1) although chemical accidents may be increasing, there
is no clear evidence on accident trends; (2) accident data is incomplete
and sometimes inaccurate; (3) EPA has vigorous programs for accident
preparedness and response, but legislation has caused EPA to shift its
primary focus to chemical accident prevention; (4) EPA assists in
developing industry, state, and local accident preparedness programs,
but those programs often fail to inform the public of potential accident
risks; (5) EPA is using several sources, including the Internet, to
disseminate accident data; (6) EPA is required to assess the nature and
seriousness of chemical accidents and take charge of response operations
if the responsible party or state and local officials cannot handle the
incident and coordinate other agencies' direct response efforts; (7)
local authorities rely heavily on EPA for accident assessments, since
they often do not have the training or resources to perform initial
assessments; (8) EPA has collaborated with industry and professional
associations to further understand safety management issues; (9) EPA has
issued regulations requiring facilities to develop and implement risk
management plans, including emergency response programs; (10) EPA, the
Occupational Safety and Health Administration, and the National
Transportation Safety Board have delineated their responsibilities to
investigate chemical accidents; and (11) although EPA has developed a
large repository of information on accident prevention issues, the
information is not being utilized at the community and industry level.

--------------------------- Indexing Terms -----------------------------

 REPORTNUM:  PEMD-96-3
     TITLE:  Chemical Accident Safety: EPA's Responsibilities for 
             Preparedness, Response, and Prevention
      DATE:  06/27/96
   SUBJECT:  Accident prevention
             Hazardous substances
             Information gathering operations
             Federal/state relations
             Public relations
             Information dissemination operations
             Interagency relations
             Data bases
             Emergency preparedness
             Environmental policies
IDENTIFIER:  Bhopal (India)
             EPA National Response Team
             EPA Accidental Release Information Program
             DOT Hazardous Materials Information System
             EPA Emergency Response Notification System
             EPA Chemical Safety Audit Program
             EPA Acute Hazardous Events Data Base
             Internet
             Ohio
             Pennsylvania
             California
             Connecticut
             Texas
             EPA National Contingency Plan
             
******************************************************************
** This file contains an ASCII representation of the text of a  **
** GAO report.  Delineations within the text indicating chapter **
** titles, headings, and bullets are preserved.  Major          **
** divisions and subdivisions of the text, such as Chapters,    **
** Sections, and Appendixes, are identified by double and       **
** single lines.  The numbers on the right end of these lines   **
** indicate the position of each of the subsections in the      **
** document outline.  These numbers do NOT correspond with the  **
** page numbers of the printed product.                         **
**                                                              **
** No attempt has been made to display graphic images, although **
** figure captions are reproduced.  Tables are included, but    **
** may not resemble those in the printed version.               **
**                                                              **
** Please see the PDF (Portable Document Format) file, when     **
** available, for a complete electronic file of the printed     **
** document's contents.                                         **
**                                                              **
** A printed copy of this report may be obtained from the GAO   **
** Document Distribution Center.  For further details, please   **
** send an e-mail message to:                                   **
**                                                              **
**                                            **
**                                                              **
** with the message 'info' in the body.                         **
******************************************************************


Cover
================================================================ COVER


Report to the Administrator of the Environmental Protection Agency

June 1996

CHEMICAL ACCIDENT SAFETY - EPA'S
RESPONSIBILITIES FOR PREPAREDNESS,
RESPONSE, AND PREVENTION

GAO/PEMD-96-3

Chemical Accident Safety

(973788)


Abbreviations
=============================================================== ABBREV

  AHE - Acute Hazardous Events
  ARIP - Accidental Release Information Program
  CAA - Clean Air Act Amendments of 1990
  CAMEO - Computer Aided Management of Emergency Operations
  CERCLA - Comprehensive Environmental Response, Compensation, and
     Liability Act of 1980
  CSA - Chemical Safety Audit
  CSHIB - Chemical Safety and Hazards Investigation Board
  DOT - Department of Transportation
  EPA - Environmental Protection Agency
  EPCRA - Emergency Planning and Community Right-To-Know Act of 1986
  ERNS - Emergency Response Notification System
  HMIS - Hazardous Materials Information System
  HMTUSA - Hazardous Materials Transportation Uniform Safety Act of
     1980
  LEPC - Local Emergency Planning Committee
  NTSB - National Transportation Safety Board
  OPA - Oil Pollution Act of 1990
  OSHA - Occupational Health and Safety Administration
  SARA - Superfund Amendments and Reauthorization Act
  SERC - State Emergency Response Commission

Letter
=============================================================== LETTER


B-261768

June 27, 1996

The Honorable Carol M.  Browner
Administrator, Environmental
 Protection Agency

Dear Ms.  Browner: 

Over 10 years ago, a catastrophic chemical accident occurred in
Bhopal, India, which claimed the lives of thousands of residents,
injured many others, and displaced many more from their homes and
businesses.  The magnitude of this event, coupled with an identified
potential for similar accidents in the United States, were catalysts
for strengthening national policy on chemical accident preparedness
and response.  Several specific laws were passed, giving the
Environmental Protection Agency a central role in the implementation
of chemical accident safety policy.  The National Response Team,
chaired by EPA, provides support for local and state officials who
respond to chemical accidents.  EPA also conducts accident
preparedness and prevention activities. 

EPA's role has been central in these areas.  However, efforts to
implement chemical accident safety policy have not been
comprehensively evaluated.  We examined the extent of EPA's role in
accident preparedness, response, and prevention.  Specifically, we
assessed (1) EPA's databases on accident occurrence and impact, (2)
EPA's chemical accident preparedness activities, (3) the
effectiveness of EPA's response to chemical accidents, and (4) the
steps EPA has taken to help prevent the occurrence of chemical
accidents. 


      BACKGROUND
---------------------------------------------------------- Letter :0.1

Federal policy on chemical accidents has chiefly been formulated over
the last 25 years and is administered under five major specific acts. 
They are the Comprehensive Environmental Response, Compensation, and
Liability Act of 1980, the Hazardous Materials Transportation Uniform
Safety Act of 1990, the Emergency Planning and Community
Right-To-Know Act of 1986, the Clean Air Act Amendments of 1990, and
the Clean Water Act as amended by the Oil Pollution Act of 1990. 
Table 1 summarizes this legislation and EPA's responsibilities for
chemical accident safety. 



                                Table 1
                
                  Environmental Legislation and EPA's
                        Chemical Accident Safety
                            Responsibilities

Legislation                               Responsibilities
----------------------------------------  ----------------------------
Comprehensive Environmental Response,     Clean up of hazardous waste
Compensation, and Liability Act of 1980   sites and land and inland
(CERCLA), P.L. 96-510                     waterway spills of hazardous
                                          substances


Emergency Planning and Community Right-   Facilitate state and local
To-Know Act of 1986 (EPCRA), P.L. 99-     accident contingency
499                                       planning, public
                                          participation, and access by
                                          individuals and communities
                                          to information regarding
                                          hazardous materials in their
                                          locales

Hazardous Materials Transportation        Participate in the oversight
Uniform Safety Act of 1990 (HMTUSA),      of grants awarded for the
P.L. 101-615                              training and education of
                                          hazardous materials
                                          employees

Clean Water Act as amended by Oil         Facilitate contingency
Pollution Act of 1990 (OPA),              planning for the petroleum
P.L 101-380                               industry including issuing
                                          regulations for tank vessel
                                          and facility response plans

Clean Air Amendments of 1990 (CAA), P.L.  Develop a list of at least
101-549                                   100 chemicals that may
                                          result in significant harm
                                          to human health or the
                                          environment

                                          Develop accident prevention
                                          programs
----------------------------------------------------------------------
EPA's chemical accident planning and prevention responsibilities are
conducted by the Chemical Emergency Preparedness and Prevention
Office.  The Office of Emergency and Remedial Response provides
resources for responding to accidents involving hazardous substances
and for those involving oil releases.  (Oil accident planning and
prevention responsibilities are also the responsibility of this
office, although these issues are not addressed in detail in this
report).  While several federal agencies have responsibilities for
chemical accident safety policy, EPA (along with the Department of
Transportation [DOT] and the Occupational Safety and Health
Administration [OSHA]) has the primary regulatory authority.  Other
EPA responsibilities include facilitating the efforts of local
emergency planners and assisting at accident sites when local
resources are insufficient.  EPA provides training to local emergency
responders and works with Local Emergency Planning Committees (LEPCs)
to share information with communities. 


   OBJECTIVES, SCOPE, AND
   METHODOLOGY
------------------------------------------------------------ Letter :1

In designing our evaluation, we found that EPA's chemical accident
safety activities include accident preparedness, response, and
prevention, and thus we examined the following evaluation questions: 

1.  What have been the recent trends in accident occurrence and
impact? 

2.  What are EPA's accident preparedness activities? 

3.  How effectively does EPA respond to chemical accidents? 

4.  What steps has EPA taken to prevent the occurrence of chemical
accidents? 

Many federal agencies play a role in chemical accident preparedness,
response, and prevention; however, our study is limited to EPA's
role.  It is also limited to accidents involving hazardous chemicals
(including petroleum products) produced by private industry. 
Radiological substances are not included.  A portion of the report is
based on case studies of six accidents that have occurred since
January 1993.  These cases constitute significant accidents with
major impact on human health or the environment; EPA coordinators
were on-site in each case.  The criteria for case selection included
the occurrence of at least 5 injuries, 1 death, 100 evacuees, or
$50,000 in property damage.  The cases were evenly split between
fixed-facility and transportation accidents and included urban and
rural communities.  We identified these cases through the assistance
of officials at the EPA regional offices. 

We used multiple sources of information to address our four
evaluation questions.  We synthesized information from EPA documents,
legislation, and National Response Team publications.  We also
utilized data from three national accident databases:  the Accidental
Release Information Program (ARIP), the Hazardous Materials
Information System (HMIS), and the Emergency Response Notification
System (ERNS).  In addition, we obtained data from accident case site
visits, including interviews with state, local, and industry
officials.  Finally, we reviewed EPA's Chemical Safety Audit (CSA)
Program database and interviewed pertinent federal officials. 

We conducted our evaluation between July 1994 and December 1995 in
accordance with generally accepted government auditing standards. 


   RESULTS IN BRIEF
------------------------------------------------------------ Letter :2

We found that EPA has vigorous programs for accident preparedness and
response.  Legislation enacted during the last several years,
however, has shifted the agency's primary focus to the prevention of
chemical accidents.  Through risk management planning, the agency has
conducted a number of efforts to further the prevention of accidents,
including collaboration with industry and professional associations. 

EPA has developed a large repository of information on accident
prevention.  A major barrier to the prevention of accidents, though,
remains the relative lack of involvement of community residents and
some sectors of industry.  To maximize the potential for accident
prevention, information that facilitates prevention must be made
available and used by industry and the local level. 


   PRINCIPAL FINDINGS
------------------------------------------------------------ Letter :3


      ACCIDENT OCCURRENCE AND
      IMPACT
---------------------------------------------------------- Letter :3.1

The available data on chemical accidents do not provide clear
evidence on accident trends.  While accidents do not appear to be
getting more severe, neither is there a downward trend pointing to
effective accident prevention activities.  The data do suggest that
the total number of chemical accidents has been increasing over time,
peaking at over 40,000 incidents occurring in 1994, but this may be
attributable to more aggressive reporting and increased economic
activity. 

For example, we found that as industrial production increases, the
number of fixed-facility accidents also increases.  Similarly, for
transportation accidents, even after accounting for the increasing
mileage being driven by trucks, there is an increase in the reported
number of chemical accidents.  Assuming continued economic growth,
then, we can assume an increase in the number of accidents, unless
specific prevention measures are implemented.  However, the impact of
fixed-facility and transportation accidents, which includes deaths,
injuries, and property damages, have changed little over time. 
(Appendix I provides a complete discussion of these results.)

Making definitive conclusions about how and why chemical accidents
occur is hindered by data quality problems.  Unverified data are
reported, causal factors are not always included, and underreporting
of accident data exists.  As shown in table 2, each of the databases
used to chronicle chemical releases has significant limitations. 

These limitations include unverified data, underreporting, and the
unavailability of data on the causes of accidents.  (Appendix I
provides a full discussion of these limitations.)



                                     Table 2
                     
                     Information Included and Limitations of
                            Chemical Release Databases

Database       Information included             Database limitations
-------------  -------------------------------  --------------------------------
Emergency      Initial reports of release data  Release data are reported when
Response       submitted to the National        events occur and before they are
Notification   Response Center, EPA regional    verified, hence, ERNS is prone
System (ERNS)  offices, and until 1989, U. S.   to inaccuracies about the number
               Coast Guard field offices.       of deaths, injuries, economic
                                                damage, and chemicals involved.

                                                An estimated 5 percent of
                                                reports are duplicated.

                                                Causal information not always
                                                included.

Accidental     Information on the causes of     Includes only accidents at fixed
Release        the most serious or potentially  facilities and is limited to
Information    serious releases and preventive  those releases that involve
Program        practices before and after a     injury and have off-site impact,
(ARIP)         release. Cases are identified    which are not representative of
               through ERNS, and data are       all release cases. The vast
               collected through a              majority of cases are excluded.
               questionnaire.
                                                Database has not always been
                                                updated on a timely basis.

Acute          Accident information from        Information, collected from
Hazardous      secondary sources, including     secondary sources, has not been
Events (AHE)   newspapers, state government     verified. (EPA cautions that
               office files, and EPA office     such sources are particularly
               files.                           prone to error.)

                                                Database was developed only over
                                                a 5-year time frame and is no
                                                longer active.

Hazardous      Chemical release data from       Limited to the information that
Materials      transportation incidents. The    a transportation carrier has
Information    data are used to monitor DOT's   knowledge of and reports.
System (HMIS)  hazardous materials
               transportation program.          In some cases, information on
                                                chemicals may be destroyed
                                                during the events surrounding
                                                the incident, and it is likely
                                                that the carrier does not know
                                                what material was being
                                                shipped.

                                                There is no assurance that DOT
                                                is receiving all incident
                                                reports as the reporting burden
                                                is placed upon carriers.
--------------------------------------------------------------------------------

      ACCIDENT PREPAREDNESS
---------------------------------------------------------- Letter :3.2

EPA administers programs to help local communities prepare for
chemical accidents, and one way the agency does this is by training
local and state responders.  EPA exceeds its requirements by offering
this training.  Most of those trained find it useful, but relatively
few local responders have taken the training, and state and local
agencies may not have the resources to provide it internally. 
However, national firefighter associations do provide a significant
amount of such training, thereby filling some of the gaps. 

Under the Emergency Planning and Community Right-To-Know Act of 1986,
EPA assists in developing accident preparedness programs for
industry, the states, and local communities.  The act requires that
state and local communities be involved in emergency planning and
response with EPA in a facilitative role.  Under the act, EPA was
given a lead role in making it possible for community residents to
obtain information about community chemical hazards.  The provisions
of the legislation rely on public awareness to encourage the
minimization of accident risks.  Specifically, Local Emergency
Planning Committees are responsible for informing community residents
about the potential accident risks, but a sizable number of them were
not actively disseminating this information.  However, EPA has made
accident data available to the public, industry, and local and state
governments through several sources, including the Internet. 
(Appendix II discusses this issue in more detail.)


      ACCIDENT RESPONSE
---------------------------------------------------------- Letter :3.3

EPA is required to assess the nature and seriousness of chemical
accidents.  If neither the party responsible for the accident nor the
locality or state can handle the incident, EPA is required to take
charge of response operations.  We examined a variety of chemical
accident cases in which EPA played a major role.  The fixed-facility
cases include chemical explosions at plants in Ohio and Pennsylvania,
while the transportation accidents involved truck and train cases in
Texas, California, and Connecticut.  (See appendix III for
descriptions of these cases.)

One of EPA's key functions is to assess the facts about a chemical
accident, such as the probable cause or causes and effects upon human
health and the environment.  Typically, these efforts are used to
facilitate accident response by local or state authorities who are
likely to have immediately responded to the accident.\1 In the
accident cases we examined, EPA officials performed these functions
and were commended for their efforts by local responders. 

As noted in the National Contingency Plan, EPA on-scene coordinators
are required to assess whether or not responsible parties or local
and state officials can handle incidents.\2 The length of response
time varied considerably (from fairly immediate to approximately 15
hours), but EPA officials coordinated with on-site responders by
telephone to ascertain whether or not they were needed.  In the case
with the longest response time, EPA coordinated with state officials
and discovered that an immediate response was unnecessary.  In one
transportation case, an EPA coordinator arrived at the accident site
3 hours after it occurred and met with the local fire department
incident commander to discuss options for managing the case.  In this
case, before the arrival of the EPA coordinator, the local responders
had immediately taken command of the incident by alerting area
residents and applying response procedures to contain the accident. 

The capabilities of local responders are very important, given that,
depending on the location of an incident, an EPA coordinator could
take several hours to reach the accident site.  While on-scene
coordinators are stationed at the 10 EPA regional offices and at
smaller facilities, they cannot always be on-site immediately.  As a
result, local responders must usually make initial assessments about
the need for evacuation, containment, and treatment technologies. 
Local authorities told us that they usually do not have the chemical
training or equipment to perform these tasks, and they rely highly on
EPA resources.  As one local responder noted, his background was in
firefighting, not chemicals. 

Under the National Contingency Plan, EPA is responsible for
coordinating response efforts among appropriate federal, state, and
local agencies and officials.  In one transportation case, EPA
participated in a meeting attended by local, state, and other federal
officials.  The group decided to let a fire burn out rather than
treat it with materials that would contaminate a nearby creek.  In
another case, EPA oversaw a cleanup function of hazardous chemicals,
but (as described by a local responder) did not micromanage the
event.  For another transportation case, EPA was described by a local
fire department official as acting in a "support role," while state
and local authorities conducted the direct response functions.  A
road that was likely to be closed for a week was reopened within 72
hours because of the coordination efforts of EPA with those of the
state and other federal officials. 

In all the cases, EPA and the technical assistance teams were heavily
involved in monitoring activities to ascertain the effects of the
accident upon the area's environment.  In accordance with the
contingency plan, this monitoring supported the local responders, who
generally had less expertise in this technical area.  While local
responders are usually able to control some of the more obvious
effects of an accident (such as fire), they are less likely to have
the expertise to mitigate such effects as groundwater pollution and
air emissions. 

In summary, EPA was effective in assessing the incidents we studied
and coordinating response efforts.  As noted by local responders, the
EPA officials were quite skilled at providing needed expertise and
coordinating response options, while not micromanaging any of the
cases. 

Under the National Contingency Plan, EPA is also responsible for
ensuring that local responders can handle the responsibility of
making area residents aware of accident risks and protecting natural
resources from harm.  (The local responders' inability to do so would
effectively shift the responsibility to EPA.) EPA was not involved in
evacuating residents and handling their concerns.  Local officials
directed evacuation efforts since EPA officials often did not arrive
on-site until several hours after the incident.  In the cases we
examined, local resources were quite sufficient for these efforts. 
However, in two cases, EPA officials did attend public meetings
following the incidents to advise citizens about accident
preparedness and prevention. 

For one fixed-facility case, the head of the EPA regional office that
had jurisdiction for the accident personally attended a public
meeting regarding response and cleanup concerns.  According to a
local responder in this case, the citizens were positively impressed
that an official of this rank attended the meeting.  In the cases
with natural resources at risk, EPA officials contacted the
appropriate trustees as required by the contingency plan.  For
example, in one transportation case where a creek was contaminated,
an area water quality control board, the local county water district,
and the department of fish and game were contacted. 

Given that community residents were at risk, media outlets were
interested in covering some of these cases.  According to one local
responder, media representatives were very impatient with him in
their quest to obtain information.  This responder relied upon the
public relations expertise of EPA.  This responder noted that, in
working with the media, EPA was able to garner respect. 

In summary, EPA was also able to ensure timely communication of
accident risks to the public.  The agency's ability to deal with the
public and media representatives was appreciated by local responders. 

As noted above, EPA is charged with ensuring that the entity
responsible for the accident provides cleanup and pays associated
costs.  For all cases in our study, we found that EPA was successful
in meeting this requirement.  In most cases, those responsible
engaged contractors who worked with government officials to clean up
the accidents. 

Two of the cases we reviewed occurred in rural areas, which are
unlikely to have the technical resources to manage the response to
chemical accidents.\3 Furthermore, Local Emergency Planning
Committees in rural areas tend to be inactive or quasi-active,
according to a recent study.\4 However, in our cases, we found a high
level of assistance given to such areas by neighboring community fire
departments, state officials, and ultimately, EPA staff members and
technical assistance teams.  While such assistance networks may not
be formalized (as demonstrated by the weak rural LEPC system), in the
cases we examined, the assistance was important in providing response
resources to these areas, facilitating risk communication, and
following other mandated response procedures. 


--------------------
\1 As discussed previously, EPA has the authority to take control of
accident response functions if local, state, or other authorities are
unable to handle an incident.  However, in our cases, this did not
occur. 

\2 The National Contingency Plan is a federal preparedness plan
developed over a number of years that attempts to deal
comprehensively with oil spills and hazardous substance accidents. 

\3 "Rural" is defined as an area with fewer than 2,500 residents. 

\4 See Nationwide LEPC Survey (Washington, D.C.:  George Washington
University Department of Public Administration, Oct.  1994). 


      ACCIDENT PREVENTION
---------------------------------------------------------- Letter :3.4

As mandated by the Clean Air Act Amendments of 1990, EPA has
developed a risk management planning program regulation that requires
facilities to identify and plan for the possibility of chemical
accidents.  In developing this program, EPA has conducted a number of
efforts to obtain information on accident prevention.  The agency has
worked with industry and professional associations to understand the
basics of process safety management issues, which show that accidents
are best prevented by comprehensive management systems.  The agency
has also conducted a number of outreach efforts.  However, the agency
has noted that the level of demand for accident prevention
information should be increased at the industry and community levels. 

The Clean Air Act Amendments of 1990, which added section 112(r),
require facilities handling regulated substances in quantities that
exceed specified thresholds to develop integrated plans for managing
accidents, which must be registered with appropriate parties and be
made available to the public.  Facilities must identify chemical
hazards, design and maintain safe operating procedures, and minimize
the consequences of releases when they occur.  EPA is required to
develop a list of at least 100 substances that cause, or may be
reasonably anticipated to cause, death, injury, or serious adverse
effects to human health or the environment when released in the air. 
In addition, the act requires EPA to develop "reasonable regulations
and appropriate guidance" to provide for the prevention and detection
of accidental releases and for responses to such releases.  These
regulations are required to include, as appropriate, provisions
concerning the use, operation, repair, and maintenance of equipment
to monitor, detect, inspect, and control releases, including training
of personnel in the use and maintenance of equipment or in the
conduct of periodic inspections.  The regulations, issued in May
1996, require covered facilities to prepare and implement risk
management plans that include a hazard assessment, a prevention
program, and an emergency response program.  Specifically, the hazard
assessment must analyze the off-site consequences of the worst
possible accident.  In addition, the risk management planning will
improve accident information reporting. 

EPA has conducted several main activities geared toward risk
management planning efforts.  These include conducting a review of
emergency systems, operating the Accidental Release Information
Program and the Chemical Safety Audit Program, initiating process
safety management activities, and conducting outreach efforts.  These
are described in greater detail in appendix IV. 

The Clean Air Act Amendments of 1990 required the establishment of
the Chemical Safety and Hazards Investigation Board to obtain
information about accidents that could be used to prevent future
occurrences.  Modeled after the National Transportation Safety Board
(NTSB), the CSHIB was to consist of five members, nominated by the
President and confirmed by the Senate.  The functions of the Board
were to investigate the causes of any fixed-facility accidental
releases resulting in a fatality, serious injury, or substantial
property damage and to publicize results of the investigations.\5

The CSHIB has never become operational.  The Senate has confirmed
three nominees; however, none has been sworn in.  The administration
proposed rescinding the Board's funding for fiscal year 1995, and its
proposed budgets for fiscal years 1996 and 1997 requested no funds
for the Board.  The administration stated it was including in its
budget requests additional funds for EPA and OSHA to enhance their
investigation and prevention efforts. 

In January 1995, EPA and OSHA developed a memorandum of understanding
outlining their accident investigation program.\6 To carry out their
responsibilities, these agencies are undertaking joint accident
investigations and, on the basis of their experience, are developing
an investigation protocol.  The memorandum of understanding will be
reexamined in light of their joint experience.  The agencies will
develop criteria for selecting accidents for both joint and
individual agency investigations.\7 Public reports will detail
findings on the causes of accidents and provide safety
recommendations. 

The agencies also agreed to continue staff training in accident
investigation and to train staff on the new investigation protocol. 
Training will be made available to other federal agencies, industry,
labor, and other interested parties.  In addition, EPA and OSHA will
establish an internal and external expert review process, including
"blue ribbon" panels of stakeholders for reports and recommendations. 

EPA can use information obtained in accident investigations in legal
proceedings brought against alleged industry violators, and the
public, in turn, could use this information in tort liability cases. 
However, if the CSHIB conducted investigations, the public would be
prohibited under the law from using information in this manner,
effectively weakening the "public pressure" provisions of Community
Right-To-Know legislation. 

In summary, to prevent accidents, industry must be able to obtain
quality information.  The replacement of the CSHIB with joint
EPA/OSHA oversight may meet this need if industry fully cooperates
and if the agencies undertake to fulfill the mandates of the Board. 


--------------------
\5 The Board was to enter into a memorandum of understanding with
NTSB to ensure coordination of functions and to limit duplication of
activities, allowing NTSB to investigate transportation-related
chemical releases. 

\6 See EPA/OSHA Joint Accident Investigation Program (Jan.  18,
1995). 

\7 OSHA currently investigates incidents in which a worker is hurt;
EPA investigates fixed-facility environmental accidents with impacts
outside of facilities.  The joint accident investigation protocol
currently being developed will more specifically delineate each
agency's investigatory responsibilities.  As noted, transportation
accident investigation is NTSB's responsibility. 


   CONCLUSIONS
------------------------------------------------------------ Letter :4

In this report, we examined several components of EPA's chemical
accident safety policy, including accident preparedness, response,
and prevention.  Arguably, the prevention of accidents is the
ultimate goal of agencies and other organizations concerned with
chemical accident safety policy.  Accident prevention activities, if
effective, reduce risks to communities and lower costs to
municipalities and private organizations.  As noted in a report
prepared for EPA by the Massachusetts Institute of Technology,
accident mitigation systems are expensive, while accident prevention
can be much less costly.\8

One of the major barriers to accident prevention is inadequate
information and, perhaps most importantly, lack of demand for that
information at the local, state, and industry level.  Information
from prior accidents is essential to take a "lessons learned"
approach in preventing future accidents.  Furthermore, this
information must be disseminated to community residents, state
governments, and industry.  However, as noted by EPA, the major
challenge is to stimulate interest in accident prevention at these
levels.  Oftentimes, community residents and industry officials do
not consider the importance of accident prevention until after an
accident occurs, which severely limits the extent of accident
prevention activities. 

As noted above, EPA directs several efforts to advance accident
prevention activities.  For the risk management planning process, the
agency has developed a large repository of information on accident
prevention issues, especially on safety management procedures.  EPA
has utilized ARIP and the CSA Program to support this effort.  By
working with industry, professional associations, and academia, the
agency has built a large amount of expertise in this area. 

However, a notable amount of this knowledge is not being utilized at
the community level.  While EPA has outreach efforts, agency
officials have noted that some community residents and industry
officials remain unconvinced about the possibility of the occurrence
of a chemical accident.\9 As a result, the agency could improve the
extent to which "lessons learned" are provided to the local level. 


--------------------
\8 See The Encouragement of Technological Change for Preventing
Chemical Accidents:  Moving Firms From Secondary Prevention and
Mitigation to Primary Prevention (Cambridge, Mass.:  MIT Center for
Technology, Policy, and Industrial Development, July 1993). 

\9 One example of outreach efforts is EPA, Managing Chemicals Safely: 
Putting It All Together (April 1992). 


   RECOMMENDATIONS
------------------------------------------------------------ Letter :5

Based on these findings, we recommend that the EPA Administrator (1)
initiate improvements in prevention activities by increasing the
extent to which the "lessons learned" from risk management planning
efforts are conveyed to industry, state and local government and the
public, and (2) work with industry, state and local government, and
the public to stress the importance of how this information can be
used to facilitate the prevention of accidents. 


   AGENCY COMMENTS
------------------------------------------------------------ Letter :6

EPA provided comments on a draft of this report.  Most of the
agency's concerns pertained to our findings on EPA accident
prevention activities.  EPA suggested that our discussion of accident
prevention did not provide sufficient information on recent agency
programs and activities.  We agreed with these comments and made
appropriate changes. 

EPA made several comments about the Accidental Release Information
Program.  We had noted that accident trend information cannot be
developed from the program, and the agency responded that this is not
the program's intent.  However, we believe that trend information is
a vital part of understanding the causes of accidents.  We had noted
the program's bias toward larger events, and EPA told us that agency
resources limit its coverage.  We concurred with this point and made
the appropriate change in the report.  Several other technical
changes were made throughout the report where appropriate. 

We are sending copies of our report to interested congressional
committees and to officials in EPA's Chemical Emergency Preparedness
and Prevention Office and the Office of Emergency and Remedial
Response.  We will also send copies to other interested parties, and
we will make copies available to others upon request.  If you have
any questions, please call me at (202) 512-3092.  Major contributors
to this report are listed in appendix V. 

Sincerely yours,

Kwai-Cheung Chan
Director for Program Evaluation
 in Physical Systems Areas


TRENDS AND IMPACT OF ACCIDENT
OCCURRENCE
=========================================================== Appendix I

Below, we examine the recent trends in accident occurrence and
impact.  Within the limitations of the available data on chemical
accidents, we profile the frequency of chemical accidents between
1987 and 1994 and their seriousness (as measured by deaths, injuries,
and property damage.)

ACCIDENT OCCURRENCE


      CHEMICAL INCIDENT RELEASES
------------------------------------------------------- Appendix I:0.1

Table I.1 and figures I.1 through I.3 display information on initial
notifications of chemical incident releases reported to the Emergency
Response Notification System.\1



                                    Table I.1
                     
                     Incidents Reported to Emergency Response
                              Notification System\a

Year         1987     1988     1989     1990     1991     1992     1993     1994
--------  -------  -------  -------  -------  -------  -------  -------  -------
Total      26,662   28,554   33,337   34,185   35,483   35,693   37,204   39,817
 reports
Reports        84      110      128       99      124      126      122      134
 involvi
 ng
 deaths
Total     1,147\b    455\b      201      148      162      163      159      185
 deaths\b
Deaths         43       16        6        4        5        5        4        5
 per
 thousan
 d total
 reports
Reports       520      569      751      688      732      690      772      921
 involvi
 ng
 injurie
 s
Total       2,673    1,656    2,360    1,963    1,933    1,743    2,025    2,730
 injuries
Injuries      100       58       71       57       54       49       54       69
 per
 thousan
 d total
 reports
--------------------------------------------------------------------------------
\a The search for this information, which is based on initial
notification data and may be subject to change, was performed on May
1, 1995, and reflects data as of that date. 

\b Figures for 1987 and 1988 include human and animal deaths. 

As shown in table I.1, the number of fixed-facility accident reports
has been rising steadily over the period 1987 to 1994, from 26,662 to
39,817.  However, according to EPA officials, more aggressive
accident reporting may have contributed to this increase.  In
addition, industrial activity has steadily increased over this time
frame; as a result, the level of accident frequency also rose. 

Figure I.1 presents the results of a model where a measure of
economic activity (an industrial production index developed by the
Federal Reserve Board) is used to test the effect upon accident
frequency.\2 The figure demonstrates that as industrial production
increases, the number of accidents also increases as more
opportunities exist for such events to occur.  Thus, without specific
interventions aimed at accident prevention, the number of accidents
would be expected to increase in the future, assuming continued
economic growth. 

   Figure I.1:  Total Incidents
   Reported to ERNS Compared to
   Industrial Production

   (See figure in printed
   edition.)

Source:  EPA and Federal Reserve Board. 

Reported incidents can vary greatly in their severity.  The number of
initial reports of accidents involving deaths has varied between 84
and 134.  These reports have not followed a clear trend from 1987 to
1994, although the number of reported deaths has remained in a fairly
narrow range since 1989 (prior to that year, ERNS data included
nonhuman deaths as well as human deaths).  The number of initial
reports involving injuries has ranged from 520 to 921, while the
number of total injuries has ranged from 1,656 to 2,730.  Figure I.2
chronicles the numbers of deaths and injuries reported to ERNS. 

   Figure I.2:  Deaths and
   Injuries Reported to ERNS\ a

   (See figure in printed
   edition.)

\a Calendar years 1987 and 1988 include human and nonhuman deaths. 

Source:  EPA. 

As shown in figure I.3, initial reports of total property damage
(standardized to 1994 dollars) have ranged from approximately $6
million to slightly over $15 million. 

   Figure I.3:  Total Property
   Damage Reported to ERNS

   (See figure in printed
   edition.)

Source:  EPA. 

To demonstrate the seriousness of these chemical accidents, we
developed the following two ratios for each year:  deaths per
thousand total reports and injuries per thousand total reports.  The
results of those calculations, also presented in table I.1, show that
while the number of reported cases has been increasing over time, the
number of deaths per thousand reports has remained fairly constant
since 1989.  The estimated number of injuries per thousand reports
reached a high point in 1987, but has remained fairly constant since
then. 

We also calculated the average cost of property damage (as estimated
in initial reports) per report involving such damage.  Figure I.4
provides the results.  The average cost has varied from slightly less
than $40,000 to well over $70,000, dropping significantly in 1992 and
remaining fairly steady in 1993 and 1994. 

   Figure I.4:  Average Cost of
   Property Damage per Damage
   Incident

   (See figure in printed
   edition.)

Source:  EPA. 

Therefore, while the number of reported chemical accidents has been
increasing, overall--with the exception of property damage
costs--their outcome, as measured by initial reports, has changed
little over time.  Furthermore, the apparent increase in accident
occurrence may partially be the result of more aggressive reporting
of these accidents and increasing levels of production. 

The above analyses focus upon actual accident impacts.  We also
analyzed the extent to which populations were at risk during chemical
accidents.  To accomplish this, we obtained data from ERNS on the
number of people evacuated (by the time of the initial report)
between 1991 and 1994.\3 In 1991, approximately 34,000 people were
evacuated.  This decreased to 19,000 in 1992, grew to 28,000 in 1993,
and fell back to about 17,000 in 1994.  While chemical accidents have
maintained relatively low actual impact, these evacuation figures
suggest that, in the judgment of those responsible for evacuations,
the potential for harm is fairly high. 


--------------------
\1 ERNS data are gathered as a part of initial accident notification. 

\2 We conducted a statistical analysis utilizing a regression model
that demonstrated that, on average, a 1-percent increase in economic
activity (as measured by an industrial production index) was
associated with a 2.24-percent increase in the number of accidental
releases. 

\3 This information was not collected before 1991. 


      TRANSPORTATION ACCIDENTS
------------------------------------------------------- Appendix I:0.2

Figures I.5 through I.12 summarize information about chemical
accidents from 1987 until 1994 for highway, railway, and "other"
(including air and water) transportation modes.  These data are
adapted from the Hazardous Materials Information System, maintained
by the Department of Transportation.  As shown in figure I.5, highway
accidents have been the most common and have risen fairly steadily
from about 5,000 to 14,000 incidents over this time frame. 

   Figure I.5:  Incidents by
   Transportation Mode

   (See figure in printed
   edition.)

Source:  DOT. 

As demonstrated in figure I.6, even after accounting for the
increasing mileage being driven by trucks (rates per billion truck
miles), there is an increase in the reported number of chemical
accidents. 

   Figure I.6:  Chemical Incident
   Rates--Highway\ a

   (See figure in printed
   edition.)

\a Data for 1994 are not available. 

Source:  DOT. 

Railway cases have held fairly constant (approximately 1,000 per
year), while other accidents have slightly risen in recent years to
approach the number of railway cases.  After accounting for mileage
for railway cases (figure I.7), a slight downturn in incident rates
is found. 

   Figure I.7:  Chemical Incident
   Rates--Railway (Freight)\a

   (See figure in printed
   edition.)

\a Data for 1994 are not available. 

Source:  DOT. 

Figure I.8 depicts deaths from chemical transportation accidents. 
These resulted only from highway incidents, peaking in 1988 at 19 and
ranging to as low as 8. 

   Figure I.8:  Deaths by
   Transportation Mode (Highway)
   and Incident Year\a

   (See figure in printed
   edition.)

\a There were no reported deaths for railway, air, or water. 

Source:  DOT. 

As shown in figure I.9, the injury toll for highway accidents
involving chemicals has risen markedly since 1988 to as high as 500,
although it declined in 1994 to approximately 400.  Injuries from
railway accidents have generally been increasing over time, while
injuries attributed to other modes have remained fairly constant
since 1989. 

   Figure I.9:  Injuries by
   Transportation Mode and
   Incident Year

   (See figure in printed
   edition.)

Source:  DOT. 

As shown in figure I.10, property damage estimates show no clear
trend for highway and railway chemical accidents, although other
modes have consistently caused little damage. 

   Figure I.10:  Property Damage
   by Transportation Mode and
   Incident Year

   (See figure in printed
   edition.)

Source:  DOT. 

To depict the seriousness of these transportation accidents, we
calculated ratios that demonstrate the number of deaths and injuries
per 1,000 reports.  As shown in figure I.11, the ratio of deaths to
total reports is fairly flat, while the ratio of injuries to total
reports does not follow a clear trend. 

   Figure I.11:  Ratio of Deaths
   and Injuries to Total
   Transportation Accident Reports

   (See figure in printed
   edition.)

Source:  DOT. 

We also calculated the average cost (in dollars) of property damage
per report involving such damage.  Figure I.12 shows that the average
property damage for transportation accidents has ranged from
approximately $1,800 to $5,000, peaking in 1987. 

   Figure I.12:  Average Cost of
   Property Damage per Incident
   Report\a

   (See figure in printed
   edition.)

\a Includes highway, railway, air, and water. 

Source:  DOT. 

DATABASE LIMITATIONS

ERNS data are compiled from initial reports of accident data
submitted to the National Response Center, EPA regional offices, and
until 1989, U.S.  Coast Guard field offices.\4 ARIP data are gathered
by sending survey questionnaires to facilities that have reported
serious chemical releases through ERNS.  This information is gathered
to develop accident prevention programs.  The AHE database contains
accident information from secondary sources, including newspapers,
state government office files, and EPA office files.  HMIS contains
chemical accident data from transportation incidents; the information
is used to monitor DOT's hazardous materials transportation program. 

EPA has identified several factors that seriously limit the extent of
chemical accident data reliability and validity.\5 The agency has
noted that there is significant evidence of data underreporting. 
Many accidents are not reported to federal authorities.  In some
cases, those involved in accidents have been unaware of reporting
requirements. 

There has been no independent verification of information provided
from industry; much of the data are based upon estimates that cannot
be readily validated.  Reports and surveys completed by facilities
may not disclose what could be self-incriminating information.  There
are inconsistencies among various industry officials who report
information:  questions are not uniformly interpreted among all
respondents and technical expertise varies.  Thus, information about
accidents is largely incomplete, often lacking details regarding such
basic technological issues as type of production process involved. 

Situational factors, such as management issues and policies, are not
routinely reported.  These could provide unique reasons as to why
accidents occur, and accident scenarios could then be developed. 
Having incomplete data makes it difficult to learn about accidents. 

Information is not systematically gathered on "near misses," which
eliminates a wide range of knowledge on accident potential. 
According to a recent study on accident prevention conducted by the
Massachusetts Institute of Technology, the data reporting systems
that exist are somewhat duplicative and also incompatible.\6 Terms
differ among databases, such that variables cannot be linked or
otherwise combined for analysis.  This problem is compounded by the
different time periods that the databases cover.  For example, ERNS
has been in existence since 1986, while AHE data were compiled for
only 5 years (1982-86).\7

In response to these problems, EPA published, in September 1995, a
users' guide to federal accident release databases.  This document
describes each database, including the data fields, reporting
criteria, and points of contact.  It also identifies common
information among the databases to allow users to cross-search for
information on a particular accident.  EPA has also made accident
databases accessible via the Internet. 

Trend analyses with these accident data are difficult to construct. 
While the existing databases give frequency counts of accidents, they
do not allow us to determine how these counts relate to changes in
industrial production.  For example, an increase in the number of
accidents could be a result of increased levels of production rather
than a marked decrease in factory safety.\8

The accident databases have specific inherent problems.  The ERNS
database is prone to inaccuracies.  Accident data are reported to
ERNS when events occur and before the information is verified.\9

As a result, at the time of reporting, it may be unclear how many
casualties resulted, the extent of economic damage, or the types of
chemicals involved.  Duplicate reports may also exist within ERNS,
estimated at 5 percent of the total cases. 

The ARIP database depicts a limited subset of accidents.\10

Cases meeting specific criteria are included, but the vast majority
of cases are excluded, and the cases included are not statistically
representative of all accident cases. 

At one point, ARIP was not updated on a timely basis.  Until July
1995, the most recent data that could be accessed from ARIP were from
1993.  As noted by an EPA official, contractual problems involving a
private company delayed the updating of the database, which generally
maintains a 6-month to 1-year lag time.  However, according to EPA,
that problem has now been resolved. 

The AHE database, compiled from secondary sources, has not been
verified.  EPA cautions that such sources are particularly prone to
error.  Furthermore, the AHE database covers only a 5-year time
frame, and it is no longer active. 

The HMIS database is limited to the information that a transportation
carrier has knowledge of and reports.  In some cases, information on
chemicals may be destroyed during the events surrounding the
accident, and according to a DOT official, a driver or train operator
may not know what materials are being shipped.  Furthermore, DOT has
no assurance that it is receiving all accident reports as the
reporting burden is placed upon carriers. 


--------------------
\4 Typically, the National Response Center is contacted by a
representative of the party responsible for the accident, local
response personnel, individuals that notice the consequences of a
release, or a witness to the accident. 

\5 See A Review of Federal Authorities for Hazardous Materials
Accident Safety, EPA (Dec.  1993). 

\6 See The Encouragement of Technological Change for Preventing
Chemical Accidents, MIT (1993). 

\7 See Environmental Protection:  EPA's Problems With Collection and
Management of Scientific Data and Its Effort to Address Them
(GAO/T-RCED-95-174; May 12, 1995) for a general overview of EPA's
problems with gathering and managing data on pollutants. 

\8 Companies generally produce a mix of product types.  Different
products utilize different types of chemicals, and the "product mix"
changes over time.  Assuming that chemical types are associated with
varied levels of accident risk, an increase in the use of certain
chemicals may result in a greater risk of an accident.  Currently
available accident data do not account for the extent of product mix. 

\9 EPA has noted that ERNS does have the capability to capture
follow-up information on releases and spills, but more resources
would be required to collect such information.  Currently, parties
responsible for incidents are not required to report updated data. 

\10 Generally, accidents are included if significant off-site impact
(death, injury, evacuations, shelter-in-place) or environmental
damage (wildlife kills, significant vegetation damage, soil
contamination, ground and surface water contamination) occur. 


CHEMICAL ACCIDENT PREPAREDNESS
========================================================== Appendix II

The national strategy on chemical accident safety policy depends
heavily on local communities to prepare for chemical accidents.  EPA
provides resources aimed at furthering the mission of this strategy. 
Generally stated, EPA depends upon Local Emergency Planning
Committees to prepare communities for chemical accidents and, in
turn, provides assistance to them.  EPA provides various tools,
publications, and technical assistance to LEPCs.  However, as noted
in a recent national survey of LEPC members, many of them are
unfamiliar with these resources.\1 For example, one-third of LEPC
members are unfamiliar with the software system that is used to map
accident hazards, and nearly 30 percent of the LEPC members are
unfamiliar with the most well-known EPA publications, such as NRT1
Planning Book, Green Book and Chemicals in the Community.\2 Other
publications, such as Managing Chemicals Safely, Opportunities and
Challenges, and Making It Work have even lower familiarity ratings. 

Local Emergency Planning Committees are required to maintain a number
of functions.  They must

  -- have a chairperson,

  -- have an emergency coordinator,

  -- have an information coordinator,

  -- have members representing local interest groups,

  -- hold formal public meetings,

  -- advertise meetings to the public,

  -- design an emergency response plan,

  -- have a plan incorporating at least nine key elements, and

  -- review the plan once a year. 

The 1994 survey of LEPCs found that these organizations, generally,
have filled the leadership positions, held meetings, maintained
committee membership, and correctly developed and maintained their
emergency plans.  However, the study found that public communications
mandates are not followed to a significant extent.  For example, less
than half of the LEPCs publish newspaper notices of the public
availability of their emergency plans, and only 70 percent of them
advertise meetings to the public.  The study also found that rural
and small town LEPCs tend to be quite inactive and could profit from
additional resources and guidance. 

A study from Tufts University reported that LEPC communication with
the public was passive and unidirectional; only modest attempts were
made to facilitate citizen understanding of title III data on
chemical hazards.\3 One impetus for the passage of EPCRA was the lack
of public knowledge about chemical accident safety issues, so this
finding suggests that such knowledge could still be limited. 

The partnership of EPA and local communities in supporting local
preparedness activities is not very strong or consistent.  To be
effective, local accident preparedness depends upon effective
communication between EPA, local and state agencies and departments,
and the public.  Federal law places a significant responsibility upon
LEPCs to prepare communities for chemical accidents, but the LEPCs
have not sufficiently communicated the risks associated with such
accidents to the public.  As a lead federal agency with
responsibilities in accident response, EPA has a large role in
ensuring that the federal-local relationship works properly.  To
examine how these linkages could be improved, EPA sought the advice
of experts in the field.  These experts have suggested that EPA
should bolster its efforts to assist the LEPCs in carrying out the
requirements of title III of the Superfund Amendments and
Reauthorization Act (SARA).\4 They suggested that EPA take an active
role in evaluating LEPC performance, in assisting LEPCs in
identifying possible funding sources, and in assessing the extent to
which LEPCs have utilized information.  Overall, the participants
stressed the growing importance of LEPCs in managing chemical
accident risk, especially as industry is required to develop Risk
Management Plans in accordance with the Clean Air Act Amendments of
1990. 

Despite these weaknesses, nearly 80 percent of LEPCs are functioning
and do execute many of their main responsibilities.  Of these
functioning LEPCs, a majority have filled their leadership positions
and committee appointments.  In addition, most hold regularly
scheduled meetings and have completed and submitted emergency
response plans to the appropriate State Emergency Response
Commission, which serves as a link between LEPCs and EPA.\5

Despite the relative weakness of the LEPCs in communicating the risks
of accidents to the public, EPA has made accident data from ERNS
available to the public, industry, and governmental agencies through
three different methods.  ERNS data can be obtained by contacting the
manager of the database, by calling an information line, or by
accessing the Internet. 


--------------------
\1 Nationwide LEPC Survey, GWU (Oct.  1994). 

\2 The software system is Computer Aided Management of Emergency
Operations (CAMEO). 

\3 Risk Communication and Community Right-To-Know:  A Four Community
Study of SARA Title III, Tufts University Center for Environmental
Management (Mar.  1991). 

\4 The Future of Local Emergency Planning Committees:  Report of a
September 13-14, 1993 Meeting.  Session sponsored by the Center for
Risk Management at Resources for the Future, Washington, D.C.  (The
30 participants represented federal, state, and local government,
labor unions, the press, LEPCs, and State Emergency Response
Commissions [SERCs].)

\5 These commissions have the authority and resources necessary to
implement federal law for chemical accident preparedness. 


ACCIDENT CASE STUDY DESCRIPTIONS
========================================================= Appendix III

Below, we present detailed information about the accident cases
examined for this report.  Cases 1 through 3 involve fixed
facilities, while cases 4 through 6 are transportation accidents. 

FIXED-FACILITY ACCIDENTS


      CASE 1
----------------------------------------------------- Appendix III:0.1

In summer 1993, a building exploded at a chemical plant in Elyria,
Ohio.  This plant produces organic peroxides used as catalysts in
plastic production.  These peroxides are highly flammable and very
reactive.  The explosion occurred as workers combined two chemicals,
which ruptured two tanks of sulfuric acid.  The accident was caused
by an overheated pump that went dry and ignited a nearby chemical
tank.  A fire followed the explosion and burned down the building and
an adjacent one.  Local and state officials were immediately
dispatched to the scene.  About 30 minutes after the first explosion,
a second one occurred, severely damaging another building in the same
plant. 

A chemical cloud was released after the explosions, and several
thousand area residents were evacuated from area neighborhoods.\1

The chemical cloud resulted from the release of two chemicals from
ruptured supply pipes.  Although no deaths resulted from the
incident, approximately 75 people were treated for minor respiratory
difficulties and acid vapor burns. 

The EPA on-scene coordinator and technical assistance team arrived on
the site within 5 hours of the first explosion.\2 At the time of
their arrival, the chemical cloud was drifting in a northern
direction, toward a housing development.  The EPA official and
technical assistance team monitored the air, and results suggested
that area residents were not in danger.  Within 10 hours of the first
explosion, the fire was extinguished and no contaminants were found
in the air.  Cleanup efforts began, and residents were allowed to
return to their homes. 


--------------------
\1 Estimates range from 4,000 to 7,000. 

\2 The teams are comprised of EPA contractors that assist in
emergency response functions, such as sampling and monitoring. 


      CASE 2
----------------------------------------------------- Appendix III:0.2

In spring 1994, an explosion occurred in a plant processing area at a
large petroleum plant in Bel Pre, Ohio, and three employees were
killed.  It caused two buildings and a tank farm to catch fire. 

Several EPA regions became involved in the response effort.  Within 2
hours of the incident, the EPA on-scene coordinator arranged for a
technical assistance team to fly over the area to assess the
situation.  The team discovered that several 300,000-gallon styrene
tanks were on fire, with flames shooting as high as 400 feet.  Foam
was used in an attempt to suppress the fires. 

Additional technical assistance teams were dispatched by the EPA
staff.  The smoke plume was observed approximately 25 miles north of
the accident site.  However, evacuation efforts were limited to the
immediate area, where 1,500 people were temporarily displaced. 

Within 8 hours of the incident, significant progress had been made in
extinguishing the fires.  Air monitoring continued until most of the
fires had been contained, while water sampling continued for several
more hours.  Samples were submitted for analysis. 


      CASE 3
----------------------------------------------------- Appendix III:0.3

In summer 1994, an explosion occurred at a chemical manufacturing
plant in Bristol Township, Pennsylvania.  Local responders arrived
within a few minutes and found heavy smoke and employees trapped
inside.  Within 30 minutes, the county hazardous materials team was
requested to respond, and the county emergency management coordinator
was en route to the scene.  Approximately 150 area residents were
evacuated. 

Within 2 hours, all victims were removed; these included three
employees of the plant who were injured.  Forty others were taken to
the hospital as a precaution.  Within 2 hours, EPA officials and
contractors had begun to monitor the air and water quality.  EPA was
assisted by the U.S.  Coast Guard and county officials.  The incident
did not escalate beyond the initial explosion and fire, although the
scene was secured by the local responders for 2 days. 

TRANSPORTATION ACCIDENTS


      CASE 4
----------------------------------------------------- Appendix III:0.4

In spring 1994, an 18-wheel tractor-trailer truck carrying a large
quantity of the chemical aldicard veered off the road, struck a
highway sign, and caught fire.  This accident, which occurred in a
suburb of Dallas, killed the driver of the vehicle.  No other
casualties were reported. 

An EPA regional office, in the vicinity of the accident, received
word of the accident through media reports.  Within 2 hours, an EPA
on-scene coordinator and technical assistance team were at the site
conducting monitoring activities.  These officials worked with state,
county, and local officials to contain the accident. 

Because of a toxic plume, city fire department officials evacuated at
least 5,000 residents downwind of the site.  Early the next day,
sampling results confirmed that no danger existed for the area, so
the evacuation order was lifted. 


      CASE 5
----------------------------------------------------- Appendix III:0.5

Early in 1994, a freight train derailed in Fremont, California. 
Because of poor lubrication, the railcars were unable to negotiate a
curve and the train derailed.  The derailed cars struck a large beam
and two containers of hazardous materials fell into an adjacent
creek, which serves as a source of drinking water for area residents. 
One of the containers caught fire and burned for several hours.  An
estimated 3 million gallons of water were contaminated.  Within 4
hours, an EPA on-scene coordinator arrived and met with the local
responders.  Several other government officials were involved in the
case, including state, county, and federal responders in addition to
EPA.  A railway representative was also on-site.  No injuries were
reported. 

Within 10 hours, the fire burned out.  The responders implemented a
plan to flush the contaminated water into the sewer.  Within 2 days,
the sampling of surface water revealed no detection of chemicals. 
Groundwater was sampled for a much longer period of time; by the end
of the third month, the concentration of chemicals was below the
maximum contaminant levels for drinking water. 


      CASE 6
----------------------------------------------------- Appendix III:0.6

In spring 1994, a dump truck collided with a gasoline tanker in
Montville, Connecticut.  The dump truck gouged a large hole in one of
the gasoline tanker's fuel compartments.  The accident caused an
explosion and fire, resulting in the death of the dump truck driver
and injuries to five people.  Two of the injured were hospitalized. 
Several homes and a small motel were evacuated.  The escaped gasoline
went into a brook and ignited, burning about half an acre of
wetlands.  Several telephone poles and utility lines were burned
down, and five other vehicles were severely damaged.  Several area
fire departments responded quickly, and the fire was extinguished
within 2-1/2 hours.  Later, state officials arrived to supervise
cleanup operations.  Early the next morning, an EPA on-scene
coordinator arrived along with a technical assistance team. 
Groundwater and air monitoring was initiated, and the tests indicated
no contamination.  However, the soil was contaminated; it was
excavated and replaced with clean soil within 2 days of the accident. 


EPA PREVENTION ACTIVITIES
========================================================== Appendix IV

This appendix provides an elaboration on EPA's accident prevention
activities.  These activities include a review of emergency systems,
ARIP and CSA Program, process safety management, and outreach efforts
conducted by the agency. 

EMERGENCY SYSTEMS REVIEW

Before the enactment of the Clean Air Act Amendments of 1990, EPA
conducted a review of emergency systems used to monitor, detect, and
prevent chemical accidents.\1 The review focused on 21 chemicals
chosen from the list of extremely hazardous substances developed
under SARA.\2 A panel of experts reviewed the preliminary findings
and made suggestions.  The findings and recommendations were
submitted to the states for comment and were also reviewed by
officials from industry, trade associations, other Federal agencies,
and environmental interest groups.  EPA found that prevention of
chemical accidents requires a comprehensive, integrated approach. 
Such an approach must consider the hazards of the chemicals in
question, the process' hazards, the capabilities of the site's
personnel, and the possible impact on the community. 

A comprehensive approach requires that management be committed to
installing, maintaining, and updating appropriate technologies and
providing personnel training.  In addition, the facility should be
involved with the local community, with industry, and with
professional groups in order to show its commitment to safety.  These
findings were consistent with those of industry and professional
organizations, including the Chemical Manufacturers Association,
American Petroleum Institute, and the Center for Chemical Process
Safety.\3 On the basis of this report, EPA recommended that industry
take the primary responsibility for preventing accidents and ensuring
the safety of its workers and the public health of the community.  In
addition, a recommendation was made to form a Chemical Accident
Prevention advisory committee to develop a strategy for implementing
the prevention of chemical accidents. 

The emergency systems review study (completed in 1988) recommended
that LEPCs increase public awareness of accident prevention issues. 
However, as discussed in the letter, studies of LEPCs completed as
recently as 1994 find that accident risk issues are not well
communicated to community residents.  Placing accident prevention
responsibilities upon industry does not complete the other important
link of community involvement.  Such an approach to accident
prevention is less likely to be effective if risks are not
communicated to the public. 

ARIP AND CSA PROGRAM


--------------------
\1 See Review of Emergency Systems, Report to Congress, Section
305(b) Title III Superfund Amendments and Reauthorization Act of
1986, EPA Office of Solid Waste and Emergency Response (June 1988). 

\2 In order to begin identifying facilities to be covered by the new
SARA requirements, EPA was directed to develop a list of extremely
hazardous substances.  Facilities using certain threshold quantities
of such substances are covered. 

\3 In addition to EPA's programs, private industry has established
various initiatives geared toward accident prevention.  The Chemical
Manufacturers' Association, a trade group representing the chemical
industry, utilizes their "Responsible CARE" program to promote safe
chemical manufacturing, transportation, and disposal.  The American
Petroleum Institute, through the "Strategies for Today's
Environmental Partnership" program, promotes the reduction of
accidental petroleum spills, fires, and explosions. 


      ACCIDENTAL RELEASE
      INFORMATION PROGRAM
------------------------------------------------------ Appendix IV:0.1

Both ARIP and the CSA Program are based on findings from the
emergency systems review.  ARIP focuses on accident prevention by
collecting data on the causes of accidents, and the CSA Program is
used to encourage accident prevention and identify problem areas that
may result in accidents.\4 ARIP is an information collection tool
begun by EPA in 1986 to obtain data on accidents so that past history
could be used to prevent such incidents in the future.  As noted by
EPA, this program is designed to serve many purposes, including

  -- Identifying problems (that is, facilities showing a persistent
     pattern of small releases that may foreshadow more severe future
     releases) and alerting facility management to the problem;

  -- Heightening corporate awareness and involvement in preventing
     accidental releases through a thought-provoking questionnaire;\5
     and

  -- Providing LEPCs and SERCs with important information useful both
     in preparing emergency response plans mandated by title III and
     in working with facilities to reduce hazards through
     prevention.\6

As EPA has recently noted, ARIP provides significant information in
the risk management planning process required by the Clean Air Act
Amendments of 1990.  To that end, ARIP serves to provide data in the
development of risk management plans for accident prevention. 

With ARIP, information on the causes of serious accidents and
preventive practices before and after an accidental release is
collected.  ARIP includes information on the most serious or
potentially serious releases.\7 Generally, accidents are included if
significant off-site impact (death, injury, evacuations) or
environmental damage (wildlife kills, significant vegetation damage,
soil contamination, ground and surface water contamination) occurs.\8
ARIP information is combined into a national database, and analyses
are disseminated to personnel involved in chemical accident
activities in industry, in EPA regional offices, and in the
community.  The ARIP database contains information on approximately
4,700 serious chemical incidents that have occurred since 1987.  The
overall response rate to the questionnaire is nearly 100 percent. 

After identifying an incident that meets the criteria, EPA sends a
questionnaire to the facility, requesting the following information: 
the chemical and amount released; the environmental media affected;
and the number of injuries, evacuations, and deaths.\9 In addition,
information is gathered on the duration of the release, the
conditions preceding the release, the existence of any hazard
assessment, and actions taken to prevent a recurrence. 

EPA distributes bulletins to state and local officials that present
ARIP data.  This information includes discussions of accident causes
and the steps taken to prevent them.  EPA regions use ARIP data as
one of the bases for selecting facilities for the Chemical Safety
Audit Program (see below).  Finally, analyses of ARIP have also been
used to set policy and prepare legislation.  For example, ARIP data
were used by EPA headquarters to prepare the list of facilities that
will be required to comply with the Risk Management Plans mandated
under the Clean Air Act Amendments of 1990, section 112(r). 

Since no other accident databases provide comprehensive information
on the causes of accidents, EPA relies heavily on ARIP to help meet
its prevention responsibilities.  EPA refers to the program as "the
best available database on the causes of, and means of preventing,
chemical accidents."\10

The Accidental Release Information Program exhibits three major
limitations in furthering the EPA accident prevention goals.  The
first limitation is the fairly narrow scope of the program.  The
database includes only a limited subset of accidents.  Reports are
biased toward larger, more severe, and more frequent releases.\11
While these requirements include "extremely hazardous" substances,
some problematic chemicals are still ignored, increasing the accident
vulnerability of industries with less common production processes. 
For example, accidents associated with flammable or petroleum
products are excluded.  As noted above, one of the purposes of ARIP
is to uncover national trends of accidentally released chemicals and
how they happen.  However, since ARIP exclusively focuses on serious
releases, any national trends in accidents that may be widespread but
that are not considered "serious" will not be documented.  EPA has
noted that ARIP is not designed to obtain national trends on accident
occurrence.  In addition, no other databases provide trend
information.  We believe that this greatly limits the extent to which
accident occurrence can be understood. 

The second limitation stems from the point, noted above, that a major
purpose of ARIP is to provide information for Local Emergency
Planning Committees to use to promote accident prevention by
industry.  However, LEPCs are quite weak in carrying out their public
communications responsibilities.  The Community Right-To- Know
provisions of SARA title III provide the mechanism to increase the
public's awareness of chemical hazards, thereby encouraging community
involvement in responding to and preventing chemical accidents by,
for example, lobbying industry to take the necessary preventive
steps.\12 However, three major studies have shown fairly weak
communications between LEPCs and the community.\13

With such weak linkages, the public is unlikely to be aware of the
important prevention data gathered by ARIP, especially given the
reliance that EPA has on the program to promote accident prevention. 
However, as EPA has noted, the risk management planning process
requires the use of accident history data that may be used to
encourage local action on accident hazards. 

Finally, on at least one occasion, data gathered for ARIP were not
updated on a timely basis.  Because of organizational difficulties,
the lag time in obtaining the data can be as great as or longer than
1 year.  Until July 1995, the most recent data for ARIP were from
1993, limiting the extent to which new information on accident
prevention can be quickly disseminated.  Recently, however, this
situation has been corrected, and the data are being updated
regularly.  Furthermore, EPA is providing accident data online
through the Internet. 


--------------------
\4 In addition, EPA maintains an oil pollution prevention team, which
is charged with preventing petroleum accidents in underground storage
tanks. 

\5 EPA sends ARIP questionnaires to facilities that report
significant or frequent releases to the National Response Center, if
these releases fit under the confines of the program. 

\6 See Why Accidents Occur:  Insights From the Accidental Release
Information Program, EPA Office of Solid Waste and Emergency Response
(July 1989). 

\7 ARIP is intended to include reported incidents of CERCLA hazardous
substances or EPCRA extremely hazardous chemicals.  Further details
on ARIP can be found in EPA's Federal Authorities for Hazardous
Materials Accident Safety (Dec.  1993). 

\8 See Accidental Release Information Program (ARIP) Fact Sheet, EPA
(Nov.  1993). 

\9 Facilities are initially identified from data provided to ERNS. 

\10 Why Accidents Occur (July 1989), p.  13. 

\11 EPA has noted that ARIP's bias toward more serious releases is
intentional.  Agency resources do not allow smaller events to be
included, therefore larger accidents are surveyed to maximize the
amount of information obtained from each facility. 

\12 As a result of the Community Right-To-Know provisions,
information on toxic chemicals is made available to the public. 
These provisions authorize public involvement in emergency response
planning.  They require that the public have access to data that
allows them to take appropriate actions. 

\13 See Communicating With the Public About Hazardous Materials:  An
Examination of Local Practice, EPA (Apr.  1990); Risk Communication
and Community Right-To-Know, Tufts University (Mar.  1991); and
Nationwide LEPC Survey, GWU (Oct.  1994). 


      CHEMICAL SAFETY AUDIT
      PROGRAM
------------------------------------------------------ Appendix IV:0.2

The Chemical Safety Audit Program was started by EPA in 1988 as a
means of identifying the causes of accidental releases of hazardous
chemicals and ways to prevent them.  The program seeks to obtain
information about industrial safety practices; promote safety
awareness among industry; share safety information with communities,
companies, and other interested groups; and develop a database with
the results of safety audits.\14

Under CERCLA, EPA has the authority to enter a facility and obtain
information.  Between the beginning of fiscal year 1989 (the
beginning of the CSA Program) and the close of fiscal year 1994, EPA
had undertaken 281 chemical safety audits and completed 270 audit
reports and was conducting as many audits as agency resources
allowed.\15 Audit teams have reviewed more than 175 hazardous
chemicals, including 159 CERCLA hazardous substances and 69 listed in
EPCRA as extremely hazardous substances. 

An important element of the CSA Program is the development of
personnel with the expertise to conduct safety audits.  EPA has
trained about 800 people in the analysis of process hazards, standard
operating procedures, prevention and mitigation systems, safety
audits, incident investigation, and interview techniques.  These
training sessions include EPA personnel as well as state and local
staff.\16 Audited facilities are selected in various ways; EPA does
not require regions to select audit sites in a prescribed manner.\17
Selection factors, to a certain extent, depend on the priorities of
the region.\18 Nationwide, about 50 percent of the facilities
targeted are identified through the ARIP database because of their
history of accidental releases.  Proximity to sensitive populations,
high population density, or an industry's concentration in an area
are also considered in the selection of audit sites.  EPA regional
offices may target certain processes or chemicals.  Furthermore, an
audit might be initiated at the request of a citizen or state or
local government or from other agency referrals. 

The audit entails a review of facility process characteristics,
emergency planning and preparedness activities, hazard evaluation,
release detection techniques, and several other areas.  In
particular, the facility's community emergency response planning
procedures for public notification of releases are studied.\19

Results of the audit are detailed in a report that is available to
other facilities, trade associations, community groups, and state and
local officials.  Weaknesses as well as strengths in preventive
practices (both operational and managerial) are discussed, along with
recommendations for improvement. 

Facilities are also chosen for audit under the CSA Program based on
public requests and local concern.  For example, citizens in a
community can work through their LEPC to request an EPA audit.  This
procedure is authorized under the Community Right-To-Know provisions
of SARA title III.  However, citizens are unlikely to realize the
risks associated with facilities in their communities, and LEPCs are
not very effective in promoting the availability of this information. 
As noted in a recent study, most functioning LEPCs receive few
inquiries from community residents.\20 During a 12-month period (June
1993-June 1994), more than 40 percent of the LEPCs received no
inquiries, while only a quarter received more than six.  This
apparent lack of public concern can result in fewer requests for
audits, which may limit EPA's effectiveness. 

EPA makes presentations to trade associations and other industry
groups, community groups, and state and local officials on the
lessons learned from the CSA Program.  However, an EPA official told
us that only about 20 such presentations are made each year, which
limits the exposure of this information. 

EPA follows up on some, but not all, of the safety audits.  According
to an EPA official, of the 10 EPA regions, five conduct follow-up
procedures to some extent, although only three of these conduct this
activity extensively.  As a result, only about 20 percent of cases
are followed up, and the EPA official estimates that 80 percent of
audited recommendations in these cases are implemented.  While some
follow-up does occur, EPA has no formal mechanism to ensure feedback
from facilities on their adoption of EPA recommendations.  Facility
compliance with recommendations is, for the most part, voluntary, so
there is no assurance that they will be implemented.\21

PROCESS SAFETY MANAGEMENT

Process safety management is used to identify the potential risks at
a facility and establish a systematic method for reducing those
risks.  EPA has incorporated this approach in the risk management
planning regulations issued under the Clean Air Act Amendments of
1990.  EPA believes that process safety management, over time, will
improve facility safety.  The agency envisions implementation of the
program as a philosophy that must be embraced by both management and
workers.\22 Although the chemical safety management program may vary
from facility to facility, all programs perform the following: 

  -- Take an inventory of hazardous materials at the site;

  -- Review the entire production process;

  -- Undertake studies to identify potential hazards, to assess the
     likelihood of accidents, to evaluate their potential
     consequences, and to address the serious problems first;

  -- Establish and follow a regular program of preventive
     maintenance;

  -- Develop standard operating procedures and training programs for
     employees;

  -- Manage changes in the operation so that accidents do not occur
     as a result of changes;

  -- Investigate and document accidents and near-accidents;

  -- Develop emergency response plans and coordinate them with local
     emergency planners; and

  -- Share information with the local community. 

Many of these components of chemical safety management already exist
in some facilities (such as training); the idea of chemical safety
management is to bring them together into a coordinated policy
strongly supported by top management. 

EPA provides funding to the American Institute of Chemical Engineers'
Center for Chemical Process Safety to conduct projects on process
safety management.  This center has representatives from 80 companies
worldwide. 

As stated previously, accident prevention activities undertaken by
industry become more effective through strong communication links
between industry, LEPCs, and community residents.  Industry's
commitment to process safety management and communication of risks
along with public participation in understanding and acting on the
risks are essential. 


--------------------
\14 See Chemical Safety Audit Program:  FY 1994 Status Report, EPA
(May 1995). 

\15 An EPA official stated that agency resources allow approximately
60 safety audits to be completed each year. 

\16 This program is designed to develop chemical safety audit
expertise both within and outside EPA, although only EPA staff are
authorized to enter facilities. 

\17 There are two important requirements for audit site selection: 
(1) A facility may be entered only if there has been a release
outside facility walls, or if there is "reason to believe" that there
is a threat of a release of a CERCLA hazardous substance; and (2) A
facility with pending or active legal actions against it must be
identified through the Office of the Regional Counsel and the SERC to
ensure that an audit would not interfere with such actions.  (See
Chemical Safety Audit Program [May 1995].)

\18 For example, EPA Region 6 (which includes Texas, Oklahoma,
Arkansas, New Mexico, and Louisiana) has concentrated on facilities
on the gulf coast because of their potential for incidents associated
with natural disasters such as hurricanes. 

\19 The Chemical Emergency Preparedness and Prevention Office has
provided regions with a Guidance Manual for EPA Chemical Safety Audit
Team Members (June 1993), which details the elements of the audit and
has a framework for the report. 

\20 Nationwide LEPC Survey, GWU (Oct.  1994). 

\21 However, if serious problems are found during the audit, EPA has
sufficient authority to remedy them. 

\22 See Managing Chemicals Safely, EPA (March 1992). 


      EPA OUTREACH EFFORTS
------------------------------------------------------ Appendix IV:0.3

As noted in a recent study commissioned by EPA, gaps in information
severely limit firms' ability to prevent accidents.\23 According to
this study, firms tend to lack information on the hazards inherent in
current production processes, the alternative technologies that would
lessen accident risk, and the costs of serious accidents.  However,
EPA has been conducting a number of outreach efforts designed to
better inform industry about the hazards essential to accident
prevention.  For example, the agency has produced documents for
industry that provide technical guidance for hazard assessment,
chemical profiles, and information- sharing requirements. 

In addition, the agency conducts special studies of chemicals and
writes advisories that are distributed to the LEPCs.  A recent
example of this practice is the Hydrogen Fluoride Study (on a widely
used and highly toxic and corrosive substance) reported to Congress
in 1993.  EPA found that accidental releases of this chemical can be
prevented through process safety management principles and
recommended that facilities handling this substance work closely with
the LEPCs to increase public awareness for adequate emergency
response.  This report also was distributed to all industries that
use this chemical (about 500 facilities). 

The agency also works with organizations such as the Center for
Chemical Process Safety, an association of 80 chemical companies
worldwide.  EPA provides the Center with funding to do special
projects such as a primer for LEPCs and a book on putting process
safety management in nontechnical language.  In addition, EPA works
closely with the American Institute of Chemical Engineers, which
provides technical resources to the agency.  Other EPA prevention
activities include funding university studies and working with small
businesses that have little knowledge of chemicals.  EPA advises them
on the safe management of hazardous substances. 


--------------------
\23 See The Encouragement of Technological Change for Preventing
Chemical Accidents, MIT (July 1993). 


MAJOR CONTRIBUTORS TO THIS REPORT
=========================================================== Appendix V

PROGRAM EVALUATION AND METHODOLOGY
DIVISION

Marcia G.  Crosse, Assistant Director
Lawrence S.  Solomon, Senior Evaluator
Nila Garces-Osorio, Social Science Analyst
Robert E.  White, Project Adviser
Venkareddy Chennareddy, Referencer

*** End of document. ***