&EPA
                United States
                Environmental Protection
                Agency
              Office of Policy
              Analysis (PM-220)
              Washington, D.C. 20460
EPV230-06-89-061
December 1989
Acute Hazardous Events
Data  Base (1989)
Final Report
  •*«_ • .
   *V-,  • ••
                                                    Recyf.o Paper

-------
                              EPA-230-06-89-061
  Acute Hazardous Events
      Data Base (1989)
           Final Report
            Prepared for:

       Office of Policy Analysis
Office of Policy, Planning and Evaluation
 U. S. Environmental Protection Agency
            Prepared by:

   Industrial Economics, Incorporated
     2067 Massachusetts Avenue
   Cambridge, Massachusetts 02140

     EPA Contract No. 68-W8-0038
           December 1989

-------
                ACUTE HAZARDOUS EVENTS DATA BASE

                        Table of Contents

FOREWORD	ii

ACKNOWLEDGEMENTS	iii

LIST OF EXHIBITS	iv

EXECUTIVE SUMMARY	I

1.  PROJECT BACKGROUND AND DATA BASE CHARACTERISTICS	1-1

    1.1  Proj ect Background	1-1
    1.2  Methodology	1-3

         1.2.1  Data Development	1-4
         1.2.2  Data Editing	1-6
         1.2.3  Development of Related Data Bases	1-6

    1.3  AHE/DB Characteristics	1-8
    1.4  Use of the AHE/DB	1-10

         1.4.1 Source of Information on Specific Events	1-10
         1.4.2 Perspective-Setting for Regulatory Actions....1-11
         1.4.3 Perspective-Setting for Policy Decisions	1-11
         1.4.4 Guidance for State or Local Actions	1-12

    1.5  Cautions for Interpretation and Suggestions for
         Additional Efforts	1-12

         1.5.1 Completeness of the Data Base	1-13
         1.5.2 Getting to the 'Truth — Cost/Quality
               Trade-offs	1-15
         1.5.3 The Problems of Causation, Consequences and
               Priorities	1-17
         1.5.4 Chemicals, Their Properties and Priorities....1-18

    1.6  Report Organization	1-20


2 .  CHARACTERISTICS OF THE MORE SEVERE EVENTS	2-1

    2.1  Events Resulting in Deaths or Injuries	2-1

         2.1.1 Extent of Deaths and Injuries	2-3
         2.1.2 Circumstances of Release	2-4
         2.1.3 Types of Facilities Involved	2-5
         2.1.4 Chemicals-Involved	2-8
         2.1.5 Relationship of Event Severity to Chemical
               Properties	2-10

-------
          2.1.6 Variation of Event Severity with Chemical
                Properties	2-14

     2.2  Evacuation Events	2-16

          2.2.1  Extent of Damage	2-16
          2.2.2  Circumstances of Release	2-17
          2.2.3  Types of Facilities Involved	2-18
          2.2.4  Chemicals Involved	2-20

     2.3  Most Severe Events Involving Atmospheric Releases
          of Acutely Toxic or Explosive Chemicals	2-22

          2.3.1  General Characteristics	2-22
          2.3.2  Significance of Event End Effects	2-24
          2.3.3  Large Scale Domestic Releases of Extremely
                 Toxic Chemicals	2-24
          2.3.4  Comparison with Major Industrial Disasters
                 of Similar Origin	2-26
3 .    CHARACTERISTICS OF ALL EVENTS	3-1

     3.1  Extent of Damage	3-1
     3.2  Circumstances of Release	3-2

          3.2.1  In-Plant Events	3-2
          3.2.2  In-Transit Events	3-4

     3.3  Types of Facilities Involved	3-5
     3.4  Chemicals Involved	3-6
     3.5  Geographic Patterns	3-7
APPENDICES

     A.  Data Coding and Format Sheets
     B.  Data Source Profiles
     C.  Chemicals in the ARE Data Base
     D.  Data Base Summary for All Events
     E.  Data Base Summary for Death or Injury Events
     F.  Data Base Summary for Evacuation Events
     G.  Data Base Summary for Events Reporting Spill/Vapor
         Release or Fire/Explosion as Worst End Effect

-------
                             NOTICE
     This  report has  been  reviewed  by  the U.S.  Environmental
Protection Agency, and approved for publication.  Approval does not
signify that the contents necessarily reflect the views and policy
of  the  Agency,  nor does  mention  of trade  names or  commercial
products constitute endorsement or recommendation for use.

-------
                             FOREWORD
     In 1985 EPA published  criteria  for selecting substances for
a  listing of  acutely hazardous  substances.    In addition,  EPA
published  the  list  resulting  from the  application  of  these
criteria.  The  data base that is the subject  of  this report was
originally developed as a part of the process of establishing this
listing criteria.   Information  from the data base  was  used in
drafting  the  published  criteria.   The  results of  this  initial
effort, which  involved the  compilation of information on  3,121
records,  are described in a December 1985 report, titled "Acute
Hazardous Events Data Base," EPA Report Number 560-5-85-029.

     Since this initial  effort,  we have refined and  supplemented
the  data  base.    The  data  base   now  contains  6,190  records
representing 10,933 events.   This represents a large  share of the
most  severe  fixed  facility  events,   and  a  large  number  of
transportation related events,  which  occurred in the United States
over the  five-year  period from 1982  to  1986.   This data base has
been used both in  the  analysis  of  EPA policy options,  and in
researching factors and  issues related  to  the  release of  acutely
hazardous   substances.      Current   plans    include   continued
supplementation of  the  event  records  contained  in   the  AHE/DB,
verification of information  on significant events, and the removal
of remaining duplicate event records.
                                 ii

-------
                        ACKNOWLEDGEMENTS
     This study was conducted under EPA Contract Number 68-W8-0038
by Industrial Economics,  Incorporated.   The principal members of
the project team were: Dr. James Cummings-Saxton, Ms. Amy J. Barad,
Mr. Marc  Benoff,  Ms.  Janet A. Gochman,  William C.  Morrison,  Dr.
Samuel J. Ratick,  Mr. Robert E. Unsworth, Mr. Mark R.  Veksler and
Mr. Anthony P. Wood.

     The  project  team received valuable and generous assistance
from individuals in several organizations in acquiring and making
use  of  the  data  that  went  into  this   study.    Among  these
organizations  are  EPA's  Office  of  Solid  Waste  and  Emergency
Response  and  the  Transportation Systems   Center  of  the  U.S.
Department of Transportation  (DOT), which  provided  access to the
National Response Center's data base; U.S.  DOT for  access to its
Hazardous Material Information System; EPA's Region 7, which made
its computerized  data  base  available;  the Centers  for  Disease
Control  of the  Department  of Health  and  Human  Services;  the
California Highway Patrol  and Office of Emergency Services; the New
Jersey Department of Environmental Protection; the New York State
Attorney  General's  Office;   the  Ohio  Environmental  Protection
Agency; and the Texas Air Control Board.

     Several EPA employees had  active  roles in the  project.  Mr.
Frederick W.  Talcott  has  been responsible  for the  project since
its  inception.    Mr.  Talcott  has  had  a  substantial  role  in
conceptualizing the data  base,  assembling and coding  records,
reviewing interim products, and directing research using the data
base.   Members of the Chemical  Emergency Preparedness Program of
EPA's  Office of  Solid  Waste  and  Emergency Response  provided
suggestions and support throughout the project.   EPA's Office of
Toxic Substances provided early funding, and individuals in that
office made valuable  contributions.   Members of  the EPA Chemical
Accident  Prevention Work  Group have helped us  shape the final
project.
                               111

-------
                         List of Exhibits


Exhibit        Description

1.2-1          Data Sources for the AHE Data Base

1.2-2          Distribution of Events by Source and Year

1.2-3          AHE/DB's Three Interconnecting Data Bases

2.1-1          Occurrence of Injuries - All Events and D/I Events

2.1-2          Occurrence of Deaths - All Events and D/I Events

2.1-3          In-Plant Events by  Location — All  Events  and D/I
               Events

2.1-4          In-Plant Events  by Cause - All Events and D/I Events

2.1-5          In-Plant Events by  Event Context  - All Events and
               D/I Events

2.1-6          Frequency of Reported  End  Effects -   All  Events,
               D/I Events and Evacuation Events

2.1-7          In-Plant vs. In-Transit Events - All Events and D/I
               Events

2.1-8          In-Transit  Events  by  Mode  - All Events  and D/I
               Events

2.1-9          In-Transit  Events by Cause  — All Events  and D/I
               Events

2.1-10         Distribution of  Events by Facility Type - All Events
               and D/I Events

2.1-11         Reported Number of Deaths by Facility Type

2.1-12         Reported Number of Injuries by Facility Type

2.1-13         Frequency of Events Reporting Deaths  by Facility
               Type

2.1-14         Frequency of Events Reporting Injuries by Facility
               Type

2.1-15         Frequency of Reported End Effects by Facility Type
               - for D/I Events
                                iv

-------
                  List of Exhibits  (continued)


Exhibit        Description
2.1-16         Distribution of Events by Chemical for D/I Events:
               10 Chemicals  Representing the  Largest Number  of
               Events

2.1-17         Substances Most Frequently Reported in D/I Events

2.1-18         Chemical Property Characterization Metrics

2.1-19         Frequency  Distribution  of  AHE/DB  Chemicals  by
               Chemical  Property Ratings  of  Primary  Substance
               Released

2.1-20         Frequency Distribution of AHE/DB Events by Chemical
               Property Ratings

2.1-21         Role  of  Fire/Explosion  vs.  Toxicity as  Apparent
               Cause of Deaths or Injuries

2.1-22         Event Severity by Toxicity Rating

2.1-23         Event Severity by Flammability Rating

2.1-24         Event Severity by Reactivity Rating

2.2-1          Occurrence of Injuries — All Events and Evacuation
               Events

2.2-2          Occurrence of  Deaths  — All Events  and  Evacuation
               Events

2.2-3          In-Plant vs.  In-Transit Events — All  Events and
               Evacuation Events

2.2-4          Frequency of Reported End Effects - All Events and
               Evacuation Events

2.2-5          In-Plant  Events  by   Location  -  All  Events  and
               Evacuation Events

2.2-6          In-Plant Events by Cause - All Events and Evacuation
               Events

2.2-7          In-Plant Events by Event Context  — All  Events and
               Evacuation Events

-------
                   List  of Exhibits  (continued)


Exhibit        Description

2.2-8          In-Plant Events by Location — Total Amount Released
               and Amount Released in Evacuation Events

2.2-9          In-Plant Events by Cause — Total Amount Released and
               Amount Released in Evacuation Events

2.2-10         In-Transit  Events  by  Mode   —   All   Events  and
               Evacuation Events

2.2-11         In-Transit  Events  by  Cause  —  All  Events  and
               Evacuation Events

2.2-12         In-Transit Events by Mode -  Total Amount Released
               and Amount Released in Evacuation Events

2.2-13         Distribution of Events by Facility Type - All Events
               and Evacuation Events

2.2-14         Frequency  of   Events   Reporting  Evacuations  by
               Facility Type

2.2-15         Percentage of Evacuees by Facility Type - Residents
               Versus Workers

2.2-16         Substances Most Frequently Reported in Evacuation
               Events

2.3-1          Air Releases of Toxic Chemicals  Leading to Two or
               More Deaths

2.3-2          Air Releases of Toxic Chemicals Leading to Forty or
               More Injuries

2.3-3          Air Releases of  Toxic  Chemicals  Leading to  Two
               Thousand or More Evacuees

2.3-4          Events with Highest Quantity-to-Toxicity Ratio

2.3-5          Ten Major Chemical-Related Industrial Disasters in
               the Twentieth Century

3.2-1          Distribution by Release Size for All Events

3.2-2          In-Plant Events by Location — Total Amount Released
               and Amount Released in D/I Events
                               VI

-------
                  List of Exhibits  (continued)

            •

Exhibits       Description

3.2-3          In-Plant Events by Cause - Total Amount Released and
               Amount Released in D/I Events

3.2-4          Distribution by Release Size for In-Plant Locations
               — Events Involving Process Vessels

3.2-5          Distribution by Release Size for In-Plant Locations
               — Events Involving Storage Vessels

3.2-6          Distribution by Release Size for In-Plant Locations
               — Events Involving Valves or Pipes

3.2-7          Distribution by Release Size for In-Plant Locations
               — Events Involving Disposal

3.2-8          Distribution by Release Size for In-Plant Locations
               - Events Involving Heating or Cooling

3.2-9          Distribution by Release Size for In-Plant Locations
               — Events Involving Vehicles Not in Transit

3.2-10         Distribution by Release Size for In-Plant Locations
               — Events Involving Equipment Failure as Cause

3.2-11         Distribution by Release Size for In-Plant Locations
               — Events Involving Operator Error as Cause

3.2-12         Distribution by Release Size for In-Plant Locations
               — Events Involving Upset Conditions as Cause

3.2-13         Distribution by Release Size for In-Plant Locations
               - Events Involving Fire or Explosion as Cause

3.2-14         Distribution by Release Size for In-Plant Locations
               — Events Involving Arson or Vandalism as Cause

3.2-15         Distribution by Release Size for In-Plant Locations
               - Events Involving Disposal as Cause

3.2-16         Distribution by Release Size for In-Plant Locations
               — Events Involving High Pressure or Temperature as
               Cause

3.2-17         Distribution by Release Size for In-Plant Locations
               — Events Involving Power Fluctuations as Cause
                               vii

-------
                   List  of Exhibits  (continued)


Exhibit        Description

3.2-18         Distribution by Release Size for In-Plant Locations
               - Events Involving Sparking or Lightning as Cause

3.2-19         Distribution by Release Size for In-Plant Locations
               — Events  Involving  Loading or Unloading  as Event
               Context

3.2-20         Distribution by Release Size for In-Plant Locations
               - Events Involving Steady State Conditions as Event
               Context

3.2-21         Distribution by Release Size for In-Plant Locations
               - Events Involving Start-Up as Event Context

3.2-22         Distribution by Release Size for In-Plant Locations
               - Events Involving Shut-Down as Event Context

3.2-23         Distribution by Release Size for In-Plant Locations
               - Events Involving Maintenance as Event Context

3.2-24         Distribution by Release Size for In-Transit Events
               Involving Trucks

3.2-25         Distribution by Release Size for In-Transit Events
               Involving Rail

3.2-26         Distribution by Release Size for In-Transit Events
               Involving Pipeline

3.2-27         Distribution by Release Size for In-Transit Events
               Involving Barge

3.2-28         Distribution by Release Size for In-Transit Events
               Involving Other Modes

3.2-29         In-Transit Events by Mode  - Total Amount Released
               and Amount Released in D/I Events

3.2-30         Distribution by Release Size for In-Transit Events
               Involving Leaks as Cause

3.2-31         Distribution by Release Size for In-Transit Events
               Involving Collisions as Cause

3.2-32         Distribution by Release Size for In-Transit Events
               Involving Derailments as Cause


                               viii

-------
                  "List of  Exhibits  (continued)

Exhibit        Description
3.2-33         Distribution by Release Size for In-Transit Events
               Involving Over-Turned Trucks as Cause
3.2-34         Occurrence  of Evacuations  — All  Events  and D/I
               Events
3.4-1          Distribution of Events  by  Chemical for All Events
               — 10 Chemicals Representing the Largest Number of
               Events
3.4-2          Substances Most Frequently Reported in All Events
3.5-1          Spatial Distribution of All Events
3.5-2          Spatial Distribution of Death Events
3.5-3          Spatial Distribution of Injury Events
3.5-4          Spatial Distribution of Evacuation Events
3.5-5          Spatial Distribution of AHE/DB Sources
.3.5-6          Spatial  Distribution  of  Value of  Shipments  of
               Manufactured Products (1985)
                                IX

-------
                        EXECUTIVE SUMMARY
PROJECT BACKGROUND

     The  Acute  Hazardous  Events  Data  Base  (AHE/DB)  has  been
assembled as one component of EPA's  ongoing  effort to assess and
respond to  the dangers posed to  the U.S.  public  and  workers by
sudden, accidental  releases of toxic chemicals.  The initial AHE/DB
contained over  three  thousand event records,  which  through data
sampling represented about seven thousand events.  An interim final
report issued  in December,  1985,  titled "Acute  Hazardous Events
Data Base"  (EPA 560-5-85-029), characterized the 1985 version of
the AHE/DB.

     Since  1985, the  AHE/DB  has  been expanded by  the Agency and
adapted for use in  support of Title  I  and  Title  III applications
under the Superfund Amendments and Reauthorization Act (SARA) of
1986.  The principal enhancements substantially expand the number
of  records  and augment the  AHE/DB  with  supportive  data  bases
containing information on the chemicals and companies involved in
the reported events.   These  enhancements enable  the AHE/DB to be
used more effectively to support the mandates of  the Accidental
Release Information Program (ARIP) , the activities being undertaken
in the realm of Community Right-to-Know, and other policy analyses
conducted by EPA.

     Because of EPA's  primary interest in preventing events similar
to that which occurred at Bhopal,  India in 1984,  several criteria
were  used  to  screen   event  data  for  inclusion  in the  AHE/DB.
Emphasis was placed on incidents that had acute hazard potential,
i.e., incidents resulting in  deaths,  injuries  or evacuation were
given highest  priority.   In  performing data  collection  for the
AHE/DB, events  that resulted in releases  of hazardous chemicals
were  given  priority,   while  those that  involved the  release of
petroleum-related  products  and  non-hazardous  substances  were
largely  excluded.   Priority also was given  to  incidents  that
involved  air  releases of  chemicals.   Events  that  involved the
release of large quantities of chemicals were given priority, and
events  that involved the  release  of less  than  one pound  of
materials were  excluded.
AHE/DB CHARACTERISTICS

     The AHE/DB is composed of four relational data bases.  These
data bases are maintained  on  IBM AT-compatible personal computer
equipment.   Two of  these  data  bases  (MAIN and  SAMPLE)  contain
information  transcribed  from existent  chemical  release  event
reports.   In addition, the study team developed two  data bases
(CHEM and  COMFAC)  that provide  information on  the  chemicals and

-------
facilities  involved  in  events  reported  in  the  AHE/DB.    The
information contained  in these latter two data  bases is derived
from a large number of additional sources that provide information
beyond that contained in the event reports.  These data bases are
further described below.

     The 1989 version of the AHE/DB contains 6,190 records, which
through sampling represent 10,933 separate events.  The data come
from 38 sources.   National  sources  include the National Response
Center  (7,705 events),  United  Press International and Associated
Press  (279  events),  and the U.S. Department of Transportation's
(DOT's) Hazardous Materials Information System (168 events) .  Other
sources include EPA's Region 7 (504 events), 26 daily newspapers
(327 events),  and  six  offices of  five state  governments  (1,820
events).  Event  records in the AHE/DB encompass the period from
before  1980  to 1987.   However,  the major data  gathering effort
focused on  the five-year period from  1982 to 1986.   The AHE/DB
contains an  average of 2,070  events per  year  for  that period,
ranging from 1,501 events for  1982  to  2,619  events for 1984.  Of
these events, 112 reported one or more fatality, and 968 reported
one or  more  persons injured.   Altogether, 288  deaths and 10,803
injuries are reported  in the AHE/DB.  Of  the  4,315 injuries for
which severity  could be determined, 1,472 required hospitalization.

     The Company/Facility Data Base  (COMFAC)  contains information
on  facilities   involved  in  AHE/DB  events.1    This  information
includes: facility name; address; a  unique hierarchical identifier
containing  the three-digit  zip code  for  the  facility and  an
indication of any parent/subsidiary relationships; and a four-digit
Standard Industrial Classification   (SIC)  code for that facility.
All  facilities  in  COMFAC  were  classified into  one  of  five
categories:  primary producers  (facilities that produce chemicals) ;
secondary producers (facilities that use chemicals to manufacture
another  product);   distributors;  end  users;  and  waste-related
activities.

     The Chemical  Data Base  (CHEM)  contains information  on the
chemicals involved  in the events reported in the  AHE/DB.   This
information includes:   chemical name;  Chemical  Abstracts Service
(CAS)  number; four-digit  SIC  of the originating industry  (i.e.,
the industry that produces the  chemical); a commodity or specialty
chemical designation;  and production characteristics.  In addition,
CHEM contains  chemical property information such  as:   toxicity
data,  including  "Immediately Dangerous to  Life and Health"  (IDLH)
values; physical characteristics, including vapor pressure, boiling
point,  melting  point,  and  air half-life  estimates;  and hazard
ranking information, including National Fire Protection Association
     1For in-transit events,  COMFAC contains  information  on the
company  reportedly  involved  in  the release  (e.g.,  XYZ Trucking
Company).

-------
reactivity and flammability rankings.  CHEM also identifies whether
or not  the  chemical is  listed under RCRA,  CERCLA or SARA (Title
III, section  302  and 313).
CHARACTERISTICS  OF THE MORE SEVERE EVENTS

     There exist a number of ways  to  analyze events.   One way is
to consider the  subset of the most  severe  events (i.e.,  those that
resulted   in   deaths   or   injuries),   by  reviewing   the  event
characteristics  data  presented  in  the  AHE/DB.    This  section
demonstrates  some of the attributes of the most severe events in
the AHE/DB.


Causes of Events at Fixed  Facilities

     Fire/explosion  was  the  attributed  cause  more   often  in
death/injury  (D/I)  events  at fixed facilities than  for  all events
at fixed facilities,  accounting for one-eighth  of the D/I events.
Equipment failure  was  the reported  cause of more than one-third of
all events, and  one-fifth of the D/I  events.   Operator error was
the reported  cause in approximately one-eighth of all  events and
D/I events.
                     Characteristics of Events by Event Cause
                           Fixed Facility Events

                   Attributed     Percent of Death/.  Percent of
                    Cause        Injury Events   All Events

                   Fire/Explosion       13X         4X

                   Equipment Failure     21X        36X

                   Operator Error       13X        14X
Facility Type

     Each event location is assigned a facility type in the AHE/DB.
The first three facility types are primary producers (any facility
in SICs 28 or 29,  i.e.,  Chemical and Allied Products, and Petroleum
Refining and Allied Products),  secondary producers  (including all
other manufacturing sectors, i.e.,  SICs 20  to  39,  except  28 and
29), and  end users  (including residential and commercial  users,
i.e., SICs  01  to  17, 48, 49  except  495,  and 52 to  99).   Although
events that occurred at primary producers  rank  second in terms of
the number  of  D/I events, they  account for the largest  share (39
percent) of events  in which deaths occur.   Events  that occurred at

-------
distributors (i.e., events at wholesale,  retail and transhipment
facilities),  which account for the  largest number  of D/I events,
account  for only eight  percent of the total number of fatalities
reported in the data base.  End users  and secondary producers rank
second and third in terms  of  the number of fatalities, accounting
for  17  percent  and nine percent  of the  deaths,  respectively.
Events at facilities involved in waste disposal activities account
for  two  percent of the  fatalities  reported  in  the data  base.
Twenty percent  of deaths occur at facilities for which insufficient
information is  available  to determine facility type.

     Although most  events involving  distributors  are in-transit
events  (69.8 percent),  some  of  them occur  at fixed facilities.
Similarly,  while a  large majority  of  events involving primary
producers (88.0 percent),  secondary producers  (94.0 percent), end
users  (86.0 percent), and waste-related facilities (81.8 percent)
occur at fixed  facilities, a small fraction are in-transit events.
                     Characteristics of Events- by Facility Type
                            Death/Injury Events

  Facility       Percent of Death/  Percent of Death  Percent of   Percent of   Percent of
   Type	Injury Events       Events       Deaths    Iniurv Events   Injuries:

  Primary Producers      26*           39X         43X       26X       31X

  Secondary Producers    13X           15X         9X       13X       18X

  Distributors         31X           12X         8X       32X       25X

  End Users          13X           18X         17X       13X       9X

  Waste: Disposal        1X           3X         2X       IX       IX

  Unknown            15X           13X         20X;       15X       16X

  * Total may not sum to 100 due to rounding.
    Primary producers also rank  first with regard to the number of
injuries—3,343  injuries  occurred  at  such   facilities,  which
represents  31 percent of the  total number of injuries reported in
the  data  base.    Distributors  rank  second  when  injuries   are
considered,  accounting  for 25  percent of  all  reported injuries,
while  secondary producers  account for 18 percent.   End users  and
waste  disposers account for nine percent  and less than one percent
of the injuries,  respectively.   Sixteen  percent of  the injuries
occurred  at facilities for  which  insufficient  information  is
available to determine  facility type.

-------
In-Transit Events

     In addition to the primary cause of the event and the facility
type at  which the event  takes place, the' situation in  which the
event  takes  place (i.e.,  fixed facility  or in-transit)  helps to
define the event.   For example,  in-transit incidents  represent a
slightly  larger percentage of  D/I events  than they  do  for all
events—37 percent versus  29  percent.   Among in-transit events,
truck  transport  represents  a  significantly  greater  share  (66
percent versus  49  percent)  and rail transport a  smaller share (28
percent  versus  43 percent)  for  events  resulting  in deaths  or
injuries, as  compared to  all in-transit events.
                        Characteristic* of Events by••;••'•
                   .  •'•'•.••:'.:•. .  •  ln-Tnn»it Bocte  .  '.-:;.. ....".

                              Percent of Death/ Percent of
                    Mode         Injury Events  AH Events

                    Truck Transport     66X        49*

                    Rait Transport     28X        43X
Reportable Quantity

     When the  quantities of substances released  in  D/I events are
compared to  their reportable quantities  (RQs),2  35  percent of the
D/I  events  for  which  released  quantities are  reported  involve
quantities less  than the RQ.   Note that the release  of chemicals
in quantities  less  than their RQs would not usually be reported to
the  National  Response  Center  or  other  jurisdiction.   Only  17
percent  of the  D/I events  involved chemical  release  quantities
greater than 100 times the RQs of the chemicals  involved.
     2RQs determine those  releases that must  be reported  under
section  304 of CERCLA.   That  is,  all releases of substances in
quantities  greater than their RQ would normally be reported to an
appropriate authority.

-------
                        Quantities Released in O/i Events

                           Qtlatiomnip to ffeporticl* Otantity
               1-100 X (O C23.3IQ
                                               rmn no c» an
                                     1-10 X OOC33 ST)
Events  Involving Evacuations

     Evacuations were reported to  have  occurred  in 829  of  the
10,933  events represented in the AHE/DB.   The 29 media sources  (26
newspapers,  2 wire services and 1 trade journal)  contributed 54
percent of the evacuation records.   It is important to note that
the NRC data  source, which is the primary source for 70 percent of
AHE/DB  records,  does not  identify whether or not evacuations were
carried out.   In the  405 events for  which the  number of evacuees
was identified,  the total number of persons  evacuated was 435,851.

     Injuries were associated  with  40 percent of  the evacuation
events  compared to only  nine percent of all  events.   Deaths were
associated with five percent of the evacuation events compared to
one percent of all events.   For some of the  evacuation events it
was possible  to distinguish between evacuations of workers and non-
workers.   Of  the 308,822  people evacuated in these events, 280,250
(91 percent)  were residents and 28,572 (nine percent) were workers.
                      Characteristics of Evacuation Events

                                       Percent of
                                    Evacuation Events

                    Involving Injuries         40X

                    Involving Deaths           5X

                    Involving Deaths or Injuries   40X

-------
     Events  at  facilities classified as  distributors and primary
producers account for the majority of evacuation events  (26 and  25
percent, respectively).  End users and secondary producers account
for  the next largest  shares  (14  and 13  percent,  respectively) .
Compared  to  their  frequency  of  involvement  in  all  events,
distributors  and end  users  have a  higher frequency and primary
producers a  lower frequency of  involvement  in evacuation events.
Secondary producers  and waste disposal facilities  account for  a
similar proportion of all events (11  and one percent, respectively)
and the subset of events involving evacuations (13 and one percent,
respectively).
                      Characteristics or Evacuation Events

                          Fact Hty Type (Pixad f*ci Hty 6v«nt«)
                OittriDutori (2V.0*3
                                          Pr.nary a*oauc«r> C23-QK)
                                            End LMTt C" 00
LARGE SCALE DOMESTIC  RELEASE OF TOXIC CHEMICALS

     Potential toxic  effects vary both with chemical toxicity  and
the quantities of the substances released.   In the AHE/DB,  2,875
event records contain information on  both quantities of material
released and acute toxicity.  For those events for which we  had  the
necessary  data,  we divided quantity  released by  the  IDLH  value
attributed to that chemical to  establish  an empirical measure of
the potential toxic  effects of  each release.  Using this measure,
25  AHE/DB  event  records  involved  releases whose  quantity-to-
toxicity ratio was at least one-half  of that associated with  the
release at  Bhopal, and  17  event records  involved releases  whose
ratio exceeded that at Bhopal.

-------
     For  the  25 events  whose  quantity-to-toxicity ratio  was at
least one-half of that associated with the release at Bhopal, only
one event reported deaths  (two  fatalities),  five events reported
injuries  (ranging  from  five  to 650 people  injured) ,  and seven
reported evacuations  (up to 20,000  people).   The average numbers
of  injuries  and evacuations were  somewhat  greater for these 25
high-damage-potential events than for the AHE/DB as a whole, and
the average number of deaths was approximately the same.  However,
given the vast quantity of materials released in the high-damage-
potential events, their severity remains surprisingly low.

     The difference between the potential for human harm and the
actual consequences of these events  appears to be due, in part, to
a  combination  of  the  following factors:    (1) the  chemical's
physical state at the  time of release was liquid  or solid, and much
of the material  remained liquid or  solid when  released, or other
physical/chemical   properties   prevented   widespread   airborne
exposure; (2) few people were close to  the point of release; (3)
weather conditions were favorable;  (4) preparedness  at the site of
release included procedures to mitigate the extent of release and
population exposure;  or (5) evacuation or other emergency response
may have  removed potentially  exposed victims.    In  addition, the
data  on  these  events  are  sometimes  incomplete  (e.g.,  the
concentration of some materials released, such  as sulfuric acid,
may have been less than 100 percent).  In most cases quantity data
represent "best estimates," which may overstate the actual quantity
released.
RELATIONSHIP OF EVENT SEVERITY TO CHEMICAL PROPERTIES

     Acute chemical  toxicity is represented  in CHEM by  an IDLH
value assigned to each chemical.  Toxicity values were obtained for
approximately  29  percent of the 651  chemicals in  the  CHEM Data
Base.  Releases of these chemicals  account  for 67 percent of the
events in the  AHE/DB.  The  "most toxic"  chemicals in CHEM  (those
chemicals in the top one-third of toxicity for  chemicals with known
IDLH  in  CHEM)  account for  more than  61 percent of  the  release
events in the  AHE/DB.  Thus,  it appears  that  severity of outcome
has some relationship to toxicity.

     Information on chemical toxicity and properties, combined with
event characteristics, was  used to determine whether casualties
resulting from release events were attributable to the toxicity of
the chemicals released.  Using this information, approximately 65
percent of the injuries and 13 percent of the deaths reported for
events in the AHE/DB were apparently caused by chemical toxicity.
In contrast approximately 35 percent of the injuries  and 87 percent
of the deaths were apparently caused by fires or explosions.
                                8

-------
CAVEATS AND INTERPRETIVE GUIDANCE

     The purpose  of  developing and maintaining the  AHE/DB is to
provide a historical  perspective on the types of accidental events
that release potentially hazardous substances in the United States,
the substances  involved,  the causative factors leading  to their
release, and the end  effects of these releases.  The AHE/DB was not
intended to provide a precise estimate of the nation-wide frequency
of events,  the quantities released,  or  their consequences.   In
fact,  information developed recently  indicates  that the AHE/DB may
significantly under-report less severe events (events that did not
result  in deaths,  or  which  resulted in very  few injuries  or
evacuations).

     In attempting to  interpret the  reported data,  some cautions
are in order.   First, neither the study team nor EPA have performed
an independent  review  of  the  circumstances or  characteristics of
the  reported  events,   nor  have  they verified  the  information
contained in most of  the event records in  the data base.  The data
base is best viewed as  an  accurate  compilation  and codification of
previously  reported  information  on  acute  hazardous  material
releases.  A number  of reviewers have cited  instances  where data
available through other sources  (including EPA and company records)
conflicts  with  information   contained  in the  AHE/DB.    Where
possible,  the  AHE/DB  has been  adjusted in these  instances.
However,  for  the vast  majority of  the  events (greater  than 96
percent) , the  information contained  in the data base  remains an
unverified transcription of existent records.

     Second,  the  term "injury"  is   used to  refer to  physical
manifestations ranging from temporary respiratory or eye irritation
treated  on-site  to  critical   injury   resulting   in   prolonged
hospitalization.   In  an  attempt  to  categorize  the severity of
injuries,  where possible  we coded  whether or  not the  injured
parties  were hospitalized.   While  this   parameter  serves  as  a
surrogate for injury  severity,  it should be noted that not all data
sources  used  in  developing  the  AHE/DB  provided information on
hospitalization.   Also,  it  is  important to  note  that  not all
hospitalizations are necessarily serious.

     Third,  direct  evidence  of causality rarely  exists  in the
contributing sources to the  data base.    Some of the  deaths or
injuries reported to  have occurred in association with a particular
event may  not, in  fact,  have  been  caused by the  release  of  a
chemical during the  event.  For example,  about 13  percent of all
events  with reported  casualties  list  the  cause  as  collision,
derailment or overturned  truck.   For these events  it is possible
that some of the  reported casualties may  be  the  direct result of
the accident, not the  release of a chemical  substance  during the
event.  Whenever possible, reports of deaths and injuries resulting

-------
from the physical impact of a collision were removed from the data
base.   However,  it is not  possible  to  identify how many  of the
deaths and injury reports remaining in the data base fall into this
category.

     Finally,  in some  cases  the source  of  data  being used  to
document an event did not provide exact numbers of persons injured
or evacuated, or quantity of substances released.  In these cases
standardized rules were  used  to quantify such terms as  "a few,"
"about 50," and  "a thousand barrels."  While  these  rules will on
average yield reasonable  estimates, in specific cases the estimated
values made may vary significantly from the true (albeit unknown)
value.

     Thus, the nature and limitations of the AHE/DB must be fully
understood and  carefully  observed in deriving any conclusions from,
or in addressing any corporate or public policy concerns using, the
information  contained in the  data  base.   We suggest a careful
reading of  the  full  AHE/DB report.  Readers should  apply this
historical  data  together  with  other  information  in  setting
priorities  or   taking actions  affecting  subsets  of  chemicals,
facilities or circumstances for the purpose of  accident prevention,
mitigation  and emergency  preparedness.    Additional  information
could include chemical,  physical and toxicological characteristics
of    released    substances;   national    production   and   use
characteristics; quantities of substances commonly present during
production,  distribution,   use  and  disposal; as well  as  local
circumstances.   Readers are encouraged to contact the EPA project
officer with any questions or concerns.
                                10

-------
1.   PROJECT BACKGROUND AND DATA BASE CHARACTERISTICS
1.1  PROJECT BACKGROUND

     The  Acute  Hazardous  Events  Data  Base  (AHE/DB)  has  been
compiled  as  part of  the U.S. Environmental  Protection Agency's
(EPA) review  of the dangers  posed  to the public  and industrial
workers by sudden, accidental releases of toxic chemicals.  Concern
regarding such  releases  was heightened by the accidental release
of a pesticide  intermediate  that occurred on  3  December 1984 in
Bhopal,  India,  which  killed  over 3,000  and  injured over 200,000
persons.  Shortly thereafter, the EPA formed  an Acute Hazards List
Workgroup to investigate the safety-related characteristics of U.S.
industry  with  regard  to  accidental  chemical  releases,  and  to
identify those  chemicals that  appear  to represent unusually high
risks.   In late May,  1985 the decision was made  to  assemble an
Acute Hazardous Events  Data  Base as  part of the  Acute Hazards
Analysis review process.

     An important factor in the initial AHE/DB development effort
was the need for a rapid  assemblage  of data  to support various
aspects of the analysis process.  A team of three contractors was
formed,  with Industrial  Economics,   Inc.  (lEc)   responsible  for
overall  project  direction.     The  other  two  contractors  were
Management  Technology  and Data  Systems,  Inc.   (MTDS)  and  PEI
Associates, Inc.  By mid-August, 1985, the AHE/DB contained 3,121
events,  which through  data  sampling  represented  6,928  release
incidents.  An interim final report titled,  "Acute Hazardous Events
Data Base,"  (EPA  560-5-85-029)  was  issued  in  December,  1985.
Copies of the data base were made available to the public at this
time.

     Because of EPA's primary interest in preventing events similar
to that which occurred at Bhopal, several criteria were used to
screen event data for  inclusion in the AHE/DB.  Emphasis was placed
on  incidents that  had acute  hazard  potential,  i.e.,  incidents
resulting in deaths, injuries or evacuations.   In addition, events
     1Paul Shrivastava,  Bhopal:  Anatomy  of  a Crisis (Cambridge,
Massachusetts:  Ballinger Publishing Company, 1987), p.  65.
                               1-1

-------
that  resulted  in  releases  of  hazardous  chemicals  were  given
priority,  as  were  incidents  which  involved  air  releases  of
chemicals or  large quantities  of chemicals.  Events  that solely
involved  the  release  of petroleum-related  products  and  non-
hazardous substances were not included.   Events that involved the
release of less than one pound of materials were excluded.2

     Since 1985, the  AHE/DB  has been expanded  to  include 10,933
event records, and adapted for  use  in support of Title I and Title
III   policy   analyses   under   the   Superfund  Amendments   and
Reauthorization Act  (SARA)  of 1986.   The  principal  adjustments
included augmentation of the  data base with  data on substances and
companies  involved in  the  reported events,  and  substantially
expanding the number of  event records.  These modifications enable
the AHE/DB to be used more effectively to support the mandates of
the  Accidental Release Information  Program  (ARIP),  in  policy
analysis issues concerning Community Right-to-Know,  and  in other
programs being planned or implemented by EPA.  Examples of further
modifications made to the AHE/DB include:

     o    Supplementing  event  records  in  the  data base  (where
          information was available) with additional information
          characterizing events, such as whether injuries resulting
          from  the chemical  release  required  hospitalization;
          distinguishing between workers and non-workers evacuated
          as a result of  a release event; and  adding data fields
          that further  define  the circumstances surrounding the
          release events.

     o    Reviewing the events contained within the  AHE/DB with
          regard to  the  revised  inclusion criteria.   That is,
          events solely  involving the release of petroleum products
          were excluded from the  data  base.  As a  result,  3,489
          events involving polychlorinated biphenyls (PCB's), and
          122 events  involving the release of  petroleum-related
          products (e.g., kerosene, fuel oil)  were transferred to
          separate data bases.   The purpose of this effort was to
          focus the  AHE/DB  on events  involving toxic  chemical
          releases with the  potential  for  inducing  acute health
          effects on an exposed population.
      In addition,  commonly occurring  incidences  not reported in
the AHE/DB  include:   reports  of  oily substances or foul odors;
incidences  not  related  to the release of a chemical  (such  as a
truck accident which does not  result  in a chemical  release); and
instances in which only non-hazardous materials were involved.
                               1-2

-------
     o    Verifying and supplementing the information contained in
          the AHE/DB  for  some of the most serious  events,  i.e.,
          those  events that  led to deaths,  a  large number  of
          injuries or  evacuations, or which  involved  the release
          of a large quantity of a highly toxic substance.

Additional examples of policy analyses  which  have been performed
using the AHE/DB include:

     o    Conducting  a number  of analytical  studies using  the
          expanded AHE/DB, including the development of descriptive
          statistics on evacuation  events,  the appraisal  of  the
          relationships between event characteristics and chemical
          properties,  and an assessment of the type of facilities
          at which release incidents have occurred.

     o    Developing a  structure  to  provide  information  to state
          and local emergency  response committees to help them meet
          their responsibilities  under  Title  III of SARA.   This
          includes the  development of  "Historical Performance at
          Comparable Facilities" reports.

     o    Analysis  of  the  scope  and   comprehensiveness  of  the
          information contained in the AHE/DB, including assessing
          the level of under-reporting  and bias  in  the reporting
          of events in the AHE/DB.

     Section 1.4 of this report provides additional information on
analyses which have been conducted using the data base.

     It is important to note that the data collection criteria used
in assembling the AHE/DB has purposely biased the data base toward
events  with  characteristics   considered  to  hold  the  greatest
potential for sudden/  large scale harm to the populace due to the
release of hazardous chemicals.  Thus, the data base should not be
viewed as  being  representative of the population of  all acutely
hazardous  release  events,  or  of the types of  outcomes resulting
from the release of hazardous materials.
1.2  METHODOLOGY

     The primary purpose of the AHE/DB is to provide an historical
perspective  on  the  characteristics  of  events  involving  the
accidental  release of  potentially hazardous  substances  in  the
United  States.  These  characteristics  include  the  substances
involved, causative factors leading to a substance release,  and the
end effects of a release (e.g., human health or property damage).
The AHE/DB  is not  intended to provide a  basis for  the precise
estimate of nation-wide frequencies of events, quantities released
                               1-3

-------
or  the consequences  of these  events.   Rather,  the AHE/DB  is
intended to provide  an assessment of the general  scope  of these
types of events.


1.2.1  Data Development

     Our  intention  in  developing  the  AHE/DB  was  to  convert
information describing  the  characteristics  of a large  number of
accidental releases  into a  form  convenient for statistical and
policy analysis.   In our initial data gathering effort  we found
that  there  was  no  one  data  source  which  was  sufficiently
comprehensive either  in its breadth or its depth  of  coverage to
supply the overall perspective on accidental releases in the U.S.
This was our approach:   (1)  access large  numbers of records from
several different  types of  data  sources;  (2)  screen out event
records that were not relevant to the AHE/DB  (such as events solely
involving petroleum products),  as  well as duplicate event records;
(3)  take   samples  of  certain  categories  of  records;3  and  (4)
transcribe  the   information,  in  a standardized  format,   into  a
computerized data base.

     More than  twenty separate  contributing sources were used in
developing the AHE/DB.  (Exhibit 1.2-1)  These sources differed in
the amount, quality and type of information provided.  Emphasis was
placed on  locating and using sources that held  the potential to
yield a large  number of usable records—such as  DOT'S  "National
Response Center  Data  Base",  New Jersey's  "Emergency Response/Air
Release Data  Base",  and New  York's  "Attorney General's Toxic
Accident Data Base"—or those data bases that provided information
on  the more  serious  events,  i.e., those  that  led to  deaths,
injuries or evacuations.

     Data coding involved organizing the information provided from
these different sources into the format specified in a single two-
sided coding sheet (an example of the coding sheet is provided in
Appendix  A).     The  coding process  organized  the  descriptive
information contained in the original sources into a formalized and
consistent set of categories.  This format enables the generation
of summary statistics and cross-tabulations as  presented in this
report.  Since the AHE/DB represents a standardized transcription
     3Events that did not involve deaths, injuries or evacuations,
but which were reported in  the NRC  data base, and which met all of
the AHE/DB inclusion criteria, were sampled at a frequency of one-
in-ten  (i.e., one event out of every ten events  in the NRC data
base was coded in the AHE/DB).  Thus, through sampling, the 6,190
records in  the  data base statistically  represent 10,933 events.
All of the analysis and exhibits in this report are based on this
weighted number of events.
                               1-4

-------
of information  from  a variety of sources, each  of which provide
differing event information, comment sections are provided in the
data  base to  retain information that  does  not fit  the  coding
scheme.   The  comment fields are intended to  prevent  the loss of
important  descriptive information  about  the  release incidents.
Comment information is often a verbatim transcription of the text
included  in the contributing  source.  In  other cases, the coders
utilized the comment field to report a summary of the information
contained in more extensive event records.  The 1985 coding format
was re-organized  and expanded during the second data collection
effort, based on  the experience gained  in  the  initial  coding
effort.   This revision included the  development of standardized
fields to contain information which would otherwise only be found
in the comment fields.

     The following percentages quantify the general nature of the
availability of data from the AHE/DB:

     o    87.5  percent   of  the   data   fields   containing
          information in a  text  format'(e.g., company name,
          city or state)  contain data.

     o    71.5  percent   of  the   data   fields   containing
          categorical  data (e.g.,  event  context,  quantity
          units or  end effects)  and boolean  fields  (e.g.,
          existence or absence of deaths,  property damage or
          impact on the environment) contain information.

     o    58.7 percent of  the  numeric fields (e.g., number of
          deaths, amount of property damage or total quantity
          of substances released) contain information.

In  addition,   an  average  of  one  printed  line of  descriptive
information is found  in the comment fields  for each record, with
a range of zero to 14 lines of descriptive information presented.

     Appendix A presents  the  instruction sheet  utilized by data
coders  in making consistent  categorical choices.    Coders were
trained   professionals  with  backgrounds   or   experience   in
engineering,  chemistry,   risk  assessment   or   other  relevant
disciplines.  A data coding program was developed for this effort
that provided error-checking to prevent inappropriate information
from being entered into any one data field.

     In order to maximize  the accessibility of the AHE/DB to EPA
policymakers, as well as  other interested parties,  the AHE/DB is
maintained in dBASE III format in a PC environment.4
     4dBASE III is a registered trademark of Ashton-Tate.


                               1-5

-------
1.2.2  Data Editing

     Data editing is an on-going process, including:  (1) assuring
that the coded records accurately reflect the relevant data source;
(2) combining information  from two or more  sources  for the same
event; and (3) utilizing the results of in-depth research of events
in the AHE/DB to upgrade information provided for the more serious
events.   However,  neither  the study  team  nor  EPA  has made  an
independent review  or verification  of  the  information  in most of
the records.  Thus,  we did not discover events or information about
events that were  not previously recorded elsewhere.   Neither did
we "second-guess" the  information  contained in the sources used.
Thus,  the  AHE/DB is best viewed  as an  accurate  compilation and
codification of previously reported information on acute hazardous
material releases.

     Since the 1988  Draft Final Report  was completed, data editing
efforts have  focused on (1)  verifying  the  accuracy of  coding for
events cited in  this report; (2) identifying chemical name synonyms
and standardizing chemical  names  used in the  data base;  and (3)
identifying  and removing duplicate event  records  for  the  same
event.  As  noted  above,  verification  of events involves assuring
that  the  coded records  in  the  AHE/DB accurately  reflect  the
information  source(s)  used.   No  attempt  has been  made  to  use
alternative sources  to verify this information.  The identification
and  standardization of chemical  name synonyms  is  intended  to
improve  the accuracy and  usefulness   of the data base,  through
application  of  the  most commonly  used names  for  all  substances
reported as released.

     In order to fully assess the occurrence of duplicates in the
data base, the study team constructed an  expert system designed to
identify duplicate  records  based  on a series  of  decision rules.
These  rules  considered  variables  such as the  date of  the event,
event  location,  chemicals  released and the  facility name.   This
system has  enabled  the  team to remove  over 114 duplicate records
thus  far.   While this  is  believed to represent the majority of
duplicate event records, undoubtedly some duplicates remain in the
data base, since many are difficult to identify due to missing or
incorrectly reported information.


1.2.3  Development of Related Data Bases

     Two  other  data bases have  been developed  to augment  the
AHE/DB.  These data  bases provide  information on the facilities at
which  events  have   occurred,  as   well  as information  on  the
properties of chemicals involved in events  reported in the AHE/DB.
Both of these relational data bases contain  information beyond that
which  is  provided in the  event records.   Working  from standard


                               1-6

-------
published  sources,   we created  brief  summary  dossiers  on  the
facilities and the chemicals appearing in the event records, to the
extent  they   are   specifically   identified  in  any  one  event
record.  Linking the AHE/DB to these other data bases facilitates
analysis  of  the relationships  between  event  characteristics,
including end effects,  with facility and chemical characteristics.
(Exhibit 1.2-3)

     The  first of  these  relational  data  bases is  the Company/
Facility data  base   (COMFAC),  which  contains  information  on the
facilities reported to have been involved in AHE/DB events.  This
information  includes:     the  facility   name,  address,   unique
hierarchical identifier containing the three digit zip code for the
facility and an indication of any parent/subsidiary relationships,
and the 4-digit Standard Industrial Classification (SIC) code for
the  facility.   Information  on the number  of employees  is  also
included for some of the  facilities in COMFAC.  COMFAC is linked
to the AHE/DB through the hierarchical facility identifier.

     Facilities  in  COMFAC were  organized  into  five categories:
primary producers (facilities  that produce chemicals); secondary
producers   (facilities  that   use  chemicals  to  produce  other
products); distributors (which  includes events at wholesale, retail
and  transshipment  facilities);   end  users;  and  waste-related
activities.   Note that while  most  events  at  primary producers,
secondary producers, waste-related  facilities and end users are
fixed  facility events, a  small  fraction are  in-transit  events.
Likewise,  though most events  involving distributors are in-transit
events, some involve fixed facilities.

     The second  relational data base  is the  Chemical  Data  Base
(CHEM), which contains  information on chemicals involved in AHE/DB
events. This information includes chemical name, chemical abstract
service (CAS)  number,  the 4-digit SIC code  of the industry that
produces the chemical,  an  indication  of  whether the  substance is
a  commodity   or  specialty  chemical,   as  well  as  production
characteristics.    In   addition,  CHEM  contains  information  on
chemical properties such  as:   toxicity,  including  "Immediately
Dangerous  to   Life  and Health"  (IDLH)  concentrations;  physical
characteristics, including vapor pressure, boiling point,  melting
point and air half-life estimates; and hazard ranking information,
including the National Fire Protection Association reactivity and
flammability rankings.   This  data base also  identifies whether the
chemical is included on any of the following lists:  SARA Title III
section 302 and  313,   CERCLA  Reportable Quantity,  or the  RCRA
Hazardous Waste list (Codes P or U).
                               1-7

-------
     For  both  COMFAC and  CHEM,  care  was taken  to  eliminate
redundancies  in  the identification of  facilities  and chemicals.
A unique facility name was agreed upon with  all other spellings or
name presentations  maintained for reference.   Similarly,  unique
chemical names and CAS numbers were established.
1.3  AHE/DB CHARACTERISTICS

     o    The  AHE/DB currently  includes  6,190 records,  which,
          through sampling, represent 10,933 events.

     o    The records included in the AHE/DB encompass the period
          from before 1980 to the present.  However,  the major data
          gathering  effort focused on the  five-year  period from
          1982 to 1986.   The AHE/DB contains an average of 2,072
          events  per year  for that  period,  ranging  from 1,501
          events for 1982 to 2,619  events for 1984.  (Exhibit 1.2-
          2)

     o    Of the  events  reported  in the AHE/DB,  1,019 involved
          deaths or  injuries.

     o    Of the 10,933 events  represented in the data base: 4,324
 -^       involved primary producers;  1,522 secondary producers;
          2,689 distributors;  1,239 end-users;  154 waste related
          activities; and 1,005 occurred where  the facility type
          was not specified in the source reporting the event.

     o    Although  most  events  involving  distributors are  in-
          transit  events  (69.8  percent),  some occur at  fixed
          facilities.  Similarly, while a large majority of events
          involving  primary producers  (88.0  percent),  secondary
          producers  (94.0 percent), end users (86.0 percent), and
          waste-related  facilities  (81.8  percent)  occur at fixed
          facilities, a small fraction are in-transit events.

     o    For  events involving deaths or  injuries:  267 involved
          primary   producers;   132   secondary  producers;   317
          distributors;   133  end-users;  and  13  waste  related
          activities.   Information  was not available on facility
          type for 157 D/I events.

     o    Of the  10,933 events represented  in  the AHE/DB, 8,444
          involved commodity chemicals and 834 involved specialty
          chemicals.  For the  remaining 1,655 events, either the
          released substance was not specified,  or if specified,
          could not be  classified as  a specialty  or commodity
          chemical.
                               1-8

-------
The  AHE/DB  contains  information  on  substances  and
quantities  released,   event  location,  description  of
causative factors and  description of  end  effects.   The
latter include:  deaths, injuries, evacuations of workers
and non-workers, as well  as  property  and  environmental
damage estimates.

Records  in  the  AHE/DB were  gathered  from  41  separate
sources, the most significant of which included:

     The  National  Response  Center  data  base  (U.S.
     Department  of   Transportation)   which   contains
     information on hazardous substances released to air,
     water and land throughout the U.S.   This data base
     includes information on  in-plant and transportation
     related  releases.   Through a stratified  random
     sampling scheme applied to the NRG data,  the 2,962
     records  reported  in the  AHE/DB  from  this source
     represent a total of 7,705 events.

     News Wire Services (UPI and AP).   The United Press
     International   and   Associated    Press   provide
     nationwide  coverage of  large events  and  events
     having  severe  consequences.     Events  involving
     evacuations,  injuries  and  deaths  commonly  are
     reported in this source.

     Marsh  and  McLennan  maintains  a  data  base  which
     characterizes incidents  included among the 100 worst
     industrial accidents  from 1956 to 1984  (rated by
     property damage).   One-tenth of these events  are
     transportation related.   The  remainder are  split
     evenly between chemicals  processing  and petroleum
     refining.

     EPA Region  7  maintains a data base of  incidents
     affecting all media  in  Iowa, Kansas, Missouri  and
     Nebraska.

     A variety of regional newspapers were  reviewed to
     gather  information,  mainly  on  large  events  and
     events having severe  consequences.   These sources
     provide regional  coverage with emphasis  on events
     in the relevant state.

     A number of state sponsored reporting systems were
     accessed for data.  These included:   the Texas Air
     Control Board data base, which covers air releases
     from manufacturing  facilities  in Texas;  the  Ohio
     EPA  data  base,   which  includes  information  on
     manufacturing and transportation  related  releases
                     1-9

-------
               of hazardous  substances  in  Ohio; the  California
               Highway Patrol and California  Office of Emergency
               Services data bases which emphasize transportation
               related events and releases to land in that state;
               and  the New  Jersey  Department   of  Environmental
               Protection and the New York State Attorney General' s
               Data Bases, which focus on releases with potentially
               significant human health  or  economic consequences
               in these states.

               The  U.S.  Department  of  Transportation  Hazardous
               Materials   Information   System,    which   tracks
               transportation   accidents   involving   hazardous
               materials.
1.4  USE OF THE AHE/DB

     The AHE/DB was  initially designed  for a limited purpose—to
set into historical  perspective the list  of Extremely Hazardous
Substances that EPA  was developing  in  1985.  It  performed that
function, in showing that this  list  was relevant to the kinds of
releases which occur, and the substances most often  released in the
U.S.  Of the more  serious events (i.e.,  those with reported deaths
or injuries), almost one-half involved the release of one or more
substances on  the original Extremely Hazardous  Substances  (EHS)
list,  while another  30  percent  of  the  events involved substances
with known fire or explosive hazards.

     The AHE/DB has been applied to numerous additional analyses.
Some of these are summarized below.
1.4.1  Source of Information on Specific Events

     Although the AHE/DB is not a  census  of  events,  but rather a
partial and informal sampling of event records, it does represent
an excellent source of data on a broad range of acutely hazardous
release events.   For this reason,  it has been used  to identify
specific types of events for further study. Some examples include:

     o    EPA's  Office  of Toxic  Substances used  the  AHE/DB  to
          examine ways  in which the Toxic Substances Control Act
          (TSCA) premanufacture review  process might be modified
          to focus more directly on the potential  for accidental
          releases of new chemicals.

     o    An  industry   trade  association,    in   examining  how
          evacuations are performed in  the U.S.,  used the AHE/DB
          to  identify  candidate  evacuation events   for  further
          study.   The  researchers  then conducted detailed case
                               1-10

-------
          studies on  about  50 of these  events,  using additional
          documentary   evidence,   including   interviews   with
          individuals involved in the events.

          A  firm  specializing  in  training films  that  review
          chemical release events and suggest appropriate responses
          has  used  the  data  base  to  identify  events  involving
          chemicals of concern.
1.4.2   Perspective-Setting for Reo^ilatory Actions

     The data  collection process used  in developing  the AHE/DB
created a sample of event records that do not fully represent the
population of  acutely hazardous release events.   However,  while
recognizing  the  biases  that  resulted  from  our  chosen  data
collection process,  it is reasonable to examine patterns within the
variables used to characterize  events  contained in  the AHE/DB.
This  report does so in  its  numerous  figures  and  tables—by
examining substances, causation, event types and end effects.  In
addition, EPA has used the information contained in the AHE/DB to
pursue   aspects   of   potential   regulatory  or   other   program
implementation actions.  For example:

     o    EPA's Office of Pesticide Programs examined the frequency
          of  chlorine  releases  from  swimming  pools  and  the
          consequences of these releases.  These  releases were then
          compared to other sources of chlorine releases in a re-
          examination  of the  use classification of swimming pool
          pesticidal substances.

     o    EPA's Office of Air and Radiation Programs examined the
          environmental loadings and reported  acute health effects
          of a set  of substances  in considering  the potential
          regulation   of   various   sources   of   air  releases.
          Specifically, the AHE/DB  has been  used  to  examine the
          extent to which accidental releases contribute to human
          health risks.
1.4.3   Perspective-Setting for Policy Decisions

     In addition to aiding  in  perspective setting for regulatory
actions, the  AHE/DB has been  applied to policy  analysis tasks.
Specifically:

     o    The AHE/DB was one element used  in clarifying the extent
          of accidental  releases  in the U.S.  and their potential
          for human health and  environmental damage.  In addition,
                               1-11

-------
          it  has  been  used  by   the   internal  EPA  workgroup
          implementing the  new  Community Right-to-Know  program
          under SARA Title III.

          The  data  contained  in the AHE/DB have  been used  by
          offices in EPA attempting to  examine  priorities  from a
          comparative  risk  perspective,  with  the  potential  for
          changing policy priorities within the Agency.

          The AHE/DB has been used in setting  the  initial threshold
          levels for  the  Accidental Release Information  Program
          (ARIP) as  a means of  focusing attention on  the small
          fraction of  reported events warranting  follow-up data
          collection.
1.4.4   Guidance for State or Local Actions

     EPA and lEc have done preliminary work in developing profiles
of the types and severity  of releases occurring at various classes
of facilities.   This work on "Historical  Behavior  at Comparable
Facilities"  is  preliminary  in  nature.    As  such,  it  must  be
carefully understood prior to application.  However,  it is possible
that by aggregating records into logical groupings of events, the
data base may help local officials and others to understand better
the range of accident types that they may  have  to plan for, even
though the historical record present in the AHE/DB  for a specific
area may be limited.  We have supplied information to EPA regional,
state or local officials  for this purpose.   In  addition,  we have
discussed  a more generalized  profile   format  with  individuals
involved  in  activities   of  the  Massachusetts  State  Emergency
Response Commission  (SERC), and  several  Local  Emergency Planning
Committees  (LEPCs) in Massachusetts.
1.5  CAUTIONS FOR INTERPRETATION AND
     SUGGESTIONS FOR ADDITIONAL EFFORTS

     Several aspects of the AHE/DB deserve particular notice.  It
is necessary that  the reader take account of  the  limitations of
the data base,  in  order to make the most appropriate  use of the
data while avoiding misinterpretations.   In  each of the sections
that follow, we  describe how the data base  was  compiled and how
this affects interpretation of the data.  In addition,  we outline
additional research tasks that are underway  to assess  better the
implications of these factors.
                               1-12

-------
1.5.1     Completeness of the Data Base

     Initially, the project team hoped that one, or at most a few
sources of  data  would be sufficient to  characterize  the extent,
scope,  and  attributes of  chemical accidents.   However,  it  was
determined that no single set of sources  was adequate to serve the
purposes  of  the  study,   either  in  identifying  events  or  in
documenting the circumstances surrounding  them.   For example, we
found that  events that were  reported  in  the  news media might not
be found in  EPA or state data sources,  while the majority of events
reported  in EPA and  state data  sources attracted  little media
attention.  Among those events  reported  in more than one source,
we found  that  the type of  technical  information provided varied.
That  is,  one  source might  track the  quantity of  material(s)
released, but not the number  of  deaths or injuries associated with
release events. The result of this original scoping effort was the
development of a multi-source study design.   This methodology is
similar   to  many  epidemiological studies,   in  which  private
physicians,  public and private hospitals, clinics, school records
and other sources are tapped to estimate the number of cases of a
particular disease.

     As we entered the second phase of the study (concluding with
this   report),  we   developed   strategies   to  increase   the
representativeness of the AHE/DB  for  those  types of accidental
releases which qualify for inclusion.  The goal was  to develop a
complete census of the more severe accidental releases, i.e., those
with any fatalities, with more than a few  reported injuries or with
more than a few dozen persons evacuated.   To accomplish this goal
we added more years of coverage from some  of the original national,
state and   regional  sources.   We  also extended  the  scope  of our
research  effort  by compiling information  from  a number of news
media sources  (each of which provided varying regional coverage),
and reviewed documentation of acutely hazardous release events that
were developed by researchers in  this  field.  As each additional
source  of  information  was  accessed,  we noted that  a  higher
proportion  of the  events  collected  duplicated  those  already
identified in the AHE/DB, while  a correspondingly lower proportion
of new events were added to the data base.  This state of declining
marginal  returns,  expressed  as  the  increasing  cost per  event
identified,  suggested that we had achieved a reasonably complete
tally of the more severe events.  In addition,  we  had increased the
total  number  of  events  in  the  data base  to  the  point  where
descriptive  statistics   developed  from  the   data   base  held
significance for planned analyses.

     However,  it remains reasonable to question how  complete the
AHE/DB  is in recording information on the most serious events.
After completing  the current version of the AHE/DB, we investigated
this question  through  intensive investigations  in the  states of
                               1-13

-------
Louisiana, Massachusetts, New York and South Carolina for the year
1986.  This effort was carried out in cooperation with researchers
from  the  Centers  for  Disease Control  in  Atlanta,  Georgia.   It
involved a review of information on acutely hazardous events that
resulted  in  deaths,  injuries or  evacuations  that were available
from local and regional news sources,  various state agencies, EPA
regional  offices,  local police  and  fire departments,  and local
officials  with knowledge  of severe  chemical  accidents.    This
additional data collection  effort identified a total  of  185 new
death,  injury or  evacuation events  for  the year  1986.    More
importantly, this additional data collection effort indicated that
the sources used to compile  the AHE/DB may not have been reporting
a significant fraction of those chemical accidents which resulted
in human deaths,  injuries or evacuations in that time period.  The
information  from  this effort is currently under review,  but our
initial conclusion is that progressively more intense searching of
event data  sources will yield additional events with  a moderate
numbers of casualties and evacuations.

     This last point  is  worthy of further  consideration.   Within
the current AHE/DB, the ninety-eighth percentile of severity (i.e.,
events in the top two percent of  severity) includes events with 79
or more persons injured,  15 or more persons  killed or 8,000 or more
persons  evacuated.     The   additional   data  collection  effort
identified  only  four  injury events  and no  death or  evacuation
events  that would have  been in  this top two percent range  of
severity.  The ninetieth percentile of  severity  (i.e., events in
the top 10 percent of severity)  entails 24  or more injuries, five
or  more  deaths  or  2,500  or  more  persons  evacuated.    The
supplemental  data  collection effort  identified only  five injury
events and no death or evacuation events of this magnitude.  Half
of  all  death  events,  61 percent of  all  injury  events,  and  59
percent of all evacuation events uncovered during  the additional
data collection effort are at or below the median level  of severity
for  events  in the AHE/DB  (two  injuries,  one fatality and 250
persons evacuated).  Thus, although more  intensive data collection
is likely to uncover additional events which resulted in a serious
outcome, the majority of the most serious events were reported in
sources contributing to the AHE/DB.

     The reader should therefore  be aware that  the AHE/DB does not
represent a complete census of events.  Moreover, any attempts made
to  achieve  a complete  census  of acutely  hazardous  substance
releases will experience increasing costs per reported event.  In
practical terms,  a census may be  infeasible for the time period we
have  addressed.   EPA  is pursuing this  topic further,  first  to
define the extent  of  under-reporting  among the data  sources used
in developing the  AHE/DB; second to assess the relative severity
of the  unreported  events as compared to those in the  data base;
and finally to assess  the need for additions and adjustments to the
data  base.  Note  that the  advent of  SARA Title  III, with its
                               1-14

-------
increased reporting requirements,  together with efforts by EPA and
others to increase compliance with existing reporting requirements,
and generally greater awareness of the existence and importance of
such  requirements   may  have  already  combined  to  increase  the
availability of more  complete  reports  on events  occurring today.
The  extent  to  which  this  structural  shift  in the  nature  of
reporting  has  already  occurred  is   a  topic   for  additional
investigation.


1.5.2    Getting to the Truth — Cost/Quality Trade-offs

     Because of  the unique  design  of this  study—acquiring  and
coding  large  numbers  of  reports  of  accidental  releases  of
substances  that were reported  in  sources  which  were  easily-
accessible to the study team—we were able to  compile a substantial
historical  record   (representing  over  10,000 events)  at a  very
modest cost (between one-half hour to one hour of professional time
per event, including study design, data acquisition, entry, editing
and reporting).  The  study team  is  familiar with other efforts,
albeit with different goals from this study that have invested ten
to 40  times  this effort,  and of  one investigation,  involving 50
case studies, which  involved about 100 times this effort per event.
Finally, several investigations of highly important or widely noted
accidents have had  many months  of professional effort  devoted to
them—in  some  cases more  than  one-thousand  times the  intensity
allocated for each of the records in the AHE/DB.

     Clearly it is  possible to  apply any given degree of effort to
the investigation of accident histories.  With increased effort per
event,   one   is    increasingly   likely   to  develop   a   full
characterization of these events.  But,  with limited resources, the
scope of coverage diminishes.  The reader should  recognize where,
based  on  the study design,  we chose to  be  on the continuum of
effort per event, and  should be aware of the inevitable limits that
this choice  places  on interpreting  the  results.   That  is,  this
study  was designed  to  provide  a  broad  coverage of  accidental
chemical  releases rather  than  the  complete  documentation  of  any
single event.

     The  reader  should   also  recognize   that  even   generous
expenditures per event still would leave sizable gaps in the data.
It would be fair to  say that twenty  times the effort produces less
than twenty  times  the information.   In  fact, the  circumstances
surrounding  many   of  the  remaining   events  are   so  complex,
contentious or remote in time that,  no matter how large the level
of effort dedicated  to their study,  substantial uncertainties will
likely always remain.
                               1-15

-------
     To  investigate,  understand,  and  illustrate  the  difficulty
involved  in  determining the "true" circumstances  surrounding an
event, given both time  and  expenditure constraints,  we performed
an in-depth analysis of  a relatively large and well-known incident
which occurred on 30 October 1987.  The  incident involved a release
of hydrofluoric acid and a  mixture of  hydrocarbons at a Marathon
Petroleum refinery in Texas which resulted in a number of injuries,
and the evacuation of individuals from the surrounding community.
Reports made within one  hour of  the  initial release,  both to the
NRC and  to  EPA's regional office, identified the  main substance
released, gave no quantity,  and reported no casualties.  A report
made to  a state  authority within one day of the event listed an
amount released that was five times lower than the amount that the
facility  later indicated  to be the best estimate.   On the day of
the event, a news media source reported a quantity released based
on an estimate of the capacity of the tank releasing the chemicals.
This  amount  was  five  times  greater  than the  quantity  later
indicated by  the  facility as  having been released.   A follow-up
report by government investigators  about  six  months after  the
accident, and  a report  by  the  facility  made  to  the  Accidental
Release Information Program  one year after the event, represent two
historical  references  which  contain  "final"  quantity-released
estimates.

     The  source  used for this  event by  the AHE/DB was  a  media
report, as the official records were not available at the time this
event was coded into the data base.  If the study team had access
to the NRC record and relied solely  on the  information contained
within it,  the AHE/DB  would  have indicated  that  there  were no
casualties or evacuations related to this event.  There would also
have been no estimate of  the quantity released.  If  we had used the
amount reported  to  a state authority  one day later,  the AHE/DB
would have under-estimated  the  quantity released by  a factor of
five.  The source we used to code this event apparently erred by
a similar factor, but in the opposite direction.   After reviewing
a draft of this report,  the  firm  owning the  facility contacted the
study team,  and disputed the volume  estimate, which it indicated
to be too high by a  factor of five. We  have  indicated, both in the
text and in  the exhibits where appropriate, where potential errors
of  a  similar  nature might have occurred.    To reiterate,  this
incident serves to illustrate the potential  problems of relying on
any one single report as the final word on any particular event.

     There are no practical  solutions to the problems described in
this section,  other than a very large expenditure of  effort to
identify  and  correct  such discrepancies.    However,  it is likely
that over-estimates and  under-estimates both of release quantities
and other aspects of these events may tend to offset one another,
on average.   It is  also  interesting  to note,  for this particular
event, a  reduction  in the quantity estimated as  released changed


                              1-16

-------
its importance  (ranked  in terms of quantity divided  by  IDLH,  as
presented on Exhibit 2.3-4) by only a very modest amount, given the
combined quantity and toxicity of the substances involved.

     Proposed  future research  includes an  in-depth  search  for
detailed information related to a  subset of  AHE/DB events.   This
effort will most likely  entail both a random sample of events from
the data base,  as well as  a  census of  all  "significant" events
(based on  some  measure  of number of deaths, injuries or evacua-
tions) .  The results of this effort will provide an indication of
the overall level of the  quality of the data both in the AHE/DB,
and for other reporting sources, such as the NRC.


1.5.3  The Problems of Causation.  Consequences and Priorities

     The circumstances surrounding most chemical accidents are at
least moderately complex,  and could take several pages of text to
describe.   Most  records  of incidental  release  events  that  are
readily available contain  less than one-half page of text and data.
Attributing causation is often a difficult, if not impossible task
for  many  events.   Releases   may involve  multiple  pieces  of
equipment,  several individuals and  several separate events.  Thus,
although causation  is  ascribed based  on  the sources   of  data
utilized by the  AHE/DB this factor should be interpreted carefully.

     However, since information  is  available  for a large number of
events  within  the  data  base,   some  conclusions can be  drawn
concerning causation.  One  instructive example  involves  the role
of chemical toxicity versus that of fire and explosions and other
circumstances   as   causative   factors  of   human   casualties.
Information is  presented  later  in  the  report that indicates that
fires or explosions  appear to be responsible for  the bulk of the
fatalities recorded in the AHE/DB,  including those few events that
reported 10 or more fatalities.  In contrast, the toxic properties
of the released substances seem  to  have caused the majority of the
injuries among the records.

     One reviewer of a draft  of  this report  asserted  that,  in
general, collisions were a more important cause of fatalities among
the  transportation  accidents   than are chemical  toxicological
properties.  However, detailed review of the data seems to indicate
a more complex  profile.   More than 46  percent of the in-transit
events involving injuries  or fatalities report "leaks"  as the cause
of the event; 34  percent  report either "collision",  "derailment"
or "over-turned truck"  as the  primary cause of the  event.   The
remaining events reported  the cause as either  "other" or "unknown."
Thus, independent of an  assessment  of the count of fatalities, the
major fraction of serious in-transit events seem to be related to
situations other than collisions or wrecks.   In addition,  the vast
majority of  the total  number of injured persons  were from those


                              1-17

-------
events which  resulted  in  a half-dozen or  more casualties each.
In  fact,  for some  events the  number of  injuries run  into  the
hundreds.  Thus, since it  seems  implausible that more than a few
persons would be injured  by the impact resulting  from a collision,
we  conclude  that properties of  the chemicals  involved  in these
events were most likely  responsible for the vast majority of the
injuries  from these types of  events.  Thus,  the  potential  for
exposure  and  risk  (including  death)   to  individuals  near  a
transportation released  event clearly existed  during many of the
events reported in the data base.

     In a similar vein, one reviewer has asserted that there is no
evidence  that links  chemical  releases  at  fixed  facilities  to
fatalities  in  the  community,   i.e.,  that  all  releases  from
facilities that resulted  in deaths  involved  only  workers.  Some of
the  sources  used  in  developing   the  AHE/DB   include  specific
information  or  other  indications  that the persons  reported  as
killed during an event were, indeed, workers. Most  sources are not
as direct about this aspect,  but we agree that little information
exists to indicate that any non-workers were killed by accidental
releases reported in the data base.   However,  it is important to
note that in those cases where it is possible to distinguish non-
workers  from workers,    non-worker evacuations  outnumber worker
evacuations  by  almost  ten to one.   We have not  made  a similar
assessment for events involving injuries,  but several events with
the  largest  counts  of  injuries involved  individuals  from  the
surrounding communities.  Thus, the potential for exposure and risk
(including death) to individuals  outside the plant  site appears to
have existed for many of the events reported in the data base.

     The above discussion indicates that "causation" and "outcome"
are complex  and often  difficult to define.  In addition,  it  is
misleading to focus solely on any one factor, to the exclusion of
the others.   The examples  given  above suggest various additional
analyses  that may further illuminate  the full  range  of factors
related to acutely hazardous substance releases.


1.5.4  Chemicals. Their Properties and Priorities

     With 927 unique substances  reported  as released in the data
base,  not all  of them  could be   equally  important  for  policy
planners and  researchers studying  incidental releases of acutely
hazardous  substances.3   In  this  report  we have  organized  the
chemicals and the events in which they were  released in a variety
of ways—by  number of  reported deaths,  injuries and evacuations;
by number of events resulting in deaths,  injuries or evacuations;
     5Detailed information is available in CHEM for 70.2 percent,
or 651 of these substances.
                               1-18

-------
by their association with relatively large quantities of substances
released; and by the type of activity or plant at which the release
occurred.   Displaying the data  in  these ways, we find  that the
chemicals ranking highest in terms of these metrics were common to
many release events, while only a handful  of events are associated
with chemicals  which  rank order 10  and below.  Since  only a few
chemicals represent a  large  percentage  of the  release  events, it
is possible to ascribe undue importance to a particular substance
or  circumstance,  and conversely  to  implicitly  downplay  other
equally or more important chemicals or situations.

     One aspect of this problem was raised by a  group that reviewed
a draft of this  document.  Polyvinyl  chloride (PVC) is a ubiquitous
plastic material, used in a multitude of solid  shapes.  When it is
involved in  a fire, it burns and subsequently contributes chemicals
containing  chlorine  to  the  resulting  smoke.   Several  events
involving PVC have been reported—four included  reported casualties
and nine entailed evacuations.  These facts raise the question as
to whether it makes sense  to  consider this widely-used material in
a manner similar to highly toxic gases or volatile liquids.

      The answer to this question would  depend  on an assessment of
the physical characteristics of PVC  or any other similar substance
in question.  Questions which might  be asked include: is it easily
flammable or does it release particularly hazardous substances when
reacting with oxidizers?   Is there  some circumstance surrounding
the burning  of this material that results in severe or modest human
exposure  potential?   Are  there  common  practices  during  the
production,   distribution,  storage or use of  this material  that
present some risk to human health or to the environment?

     Related  to this  issue  are broader questions regarding the
physical state of substances involved in accidental releases.  It
seems reasonable to assume that gases have a greater potential to
reach humans and to cause potential harm than do liquids or solids.
Among all events,  4.6 percent of instances involved substances that
are  solids  at  standard  temperature and pressure,  77.4  percent
involved liquids  and  27.7  percent involved gases  (see  Exhibit D-
1) .6 But, among the events with reported human casualties, solids
play a role  in 12.2 percent of events, liquids  in 73.6 percent and
gases in 31.8 percent (see Exhibit E-l).  Thus, all three physical
states play  significant roles in reported chemical accidents, with
solids  having a  relatively  more important  role  in, events  with
deaths or injuries than they do among the  less  serious events, and
liquids being much more  likely to be involved in serious events
than gases.   It is important to note that fires or explosions can
     6 Note that  these percentages sum  to more than  100,  since
events may  involve the release of  more  than one  substance  in a
variety of physical states.
                               1-19

-------
effectively convert liquids and solids into gases, and can result
in changes in the chemical  composition of the substances involved.
Additionally,  the processing  or use  of  materials at  elevated
temperatures or pressures,  or in the presence of other substances
may explain the ability of these substances to cause damage.

     Thus,  in general,  it  would appear  incorrect to  use  the
"typical" physical state of a substance as the only measure of its
human exposure potential.  Proposed research into this topic area
includes a  detailed  review of the accident histories  of several
substances  that  are  normally handled as  solids  or  liquids,  but
which have  been  involved  in release events  that resulted  in  a
significant number of deaths, injuries or evacuations.

     In conclusion, we suggest a careful  reading of the full AHE/DB
report.   Readers  should  apply this  historical  data  together with
other information in setting priorities or taking actions affecting
subsets of chemicals, facilities or circumstances for the purpose
of  accident prevention, mitigation  and emergency  preparedness.
Additional  information  could  include   chemical,  physical  and
toxicological  characteristics  of  released substances,  national
production  and  use   characteristics,  quantities of  substances
commonly present during  production, distribution,  use and disposal
as well as local circumstances.
1.6  REPORT ORGANIZATION

     The  remainder of  this  report is  divided  into two  major
sections and seven appendices.  The major sections are:

     o    Characteristics of the More Severe Events; and

     o    Characteristics of All Events.
     The  first  section  reviews  the  characteristics of  events
involving deaths, injuries or evacuations.  These characteristics
include the extent of property damage or human health  impacts; the
circumstances of  release  (such as  location,  cause,  process state
or transit mode); the types of facilities involved; the chemicals
released;  and  the  relationship of  event  severity  to  chemical
properties.
            *
     The second section provides a  similar overview of information
contained in the AHE/DB for all events, including those events not
involving  deaths,  injuries or evacuations.   Frequently reported
chemicals, quantities released, media affected, environmental and
property damage, and evacuations are discussed.
                               1-20

-------
     The seven appendices provide detailed background information
on the AHE/DB.    The Data Coding and  Format  Sheets (Appendix A)
identify the information included -(although few records include all
of this information).  Data Source Profiles (Appendix B) identify
the principal characteristics of the various sources of data used
in developing the AHE/DB.  Appendix  C provides  a listing of all of
the chemicals included  in  the data  base;  the  4-digit SIC code of
the industrial  sector that  produces the chemical;  the Chemical
Abstracts Service  (CAS)  identification number; and the number of
incidents,  injuries  and fatalities  that occurred  as a  result of
events at which the chemical was released.  Appendices D, E and F
provide summary outputs (referred to as "Encyclopedia of Data11) of
AHE/DB information for all  events, for  death or injury events, and
for evacuation events, respectively.  Appendix G provides a similar
summary for those  events that reported a  spill/vapor  release or
fire/explosion as the worst end-effect.
                               1-21

-------
2.   CHARACTERISTICS OF THE MORE SEVERE EVENTS
2.1  EVENTS RESULTING IN DEATHS OR INJURIES

     Each of the records coded  into  the  AHE/DB were examined for
event severity, specifically for indications of human casualties.
Altogether, injuries were  reported in 9.1 percent  of  the events
and fatalities in 1.0 percent.   In 8.2 percent of the records in
the data base,  insufficient information was available to determine
if  injuries had occurred.    For  5.9   percent  of the  records
insufficient information was  available   in  the  source  used  to
determine if any  fatalities had occurred as a  result  of the release
event.  (Exhibits 2.1-1  and 2.1-2)  Although some of these records
may in fact represent events that involved injuries or deaths, it
is not possible to estimate the degree of  under-reporting of deaths
or injuries without further research.

     Some  cautions  are  called  for in interpreting  the reported
casualty data:

     o    Reported Deaths or Injuries; Only a small percentage of
          the  information contained  in the event  reports used to
          construct the AHE/DB  have  been independently verified,
          either by EPA or  its contractors.   Thus,  one should
          interpret these  data  as "deaths or  injuries  that were
          reported by one or more of the contributing sources."

     o    Causation;    Direct  evidence  of  causation  is  rarely
          provided in the contributing sources to the AHE/DB.  Some
          of the deaths or  injuries  reported to have occurred in
          association with  a particular  event may  not,  in fact,
          have been caused  by  the   event.    For  example,  some
          casualties may well have been  caused by the triggering
          accident that released a substance into the environment
           (e.g.,  a  truck driver whose injuries were  caused by a
          collision,  and  not  the  succeeding  exposure  to  the
          materials released).  Whenever  possible  deaths that were
          explicitly the  result of  factors  other  than chemical
          toxicity,  such as truck collision were removed from the
          data base.  However, it is not possible to identify the
                               2-1

-------
          proportion of deaths and injuries that remain in the data
          base that may have resulted from factors unrelated to the
          release of a chemical substance.

     o    Range of Severity;  Although  some of the sources used in
          developing the AHE/DB indicated in detail  the type and
          severity of the reported injuries, most sources did not.
          To categorize the severity of injuries for those records
          where  information  was not   provided, we  coded,  where
          possible,  whether  or not   the  injured  parties  were
          hospitalized.  This variable then served as a surrogate
          for injury severity.  While  this  categorization scheme
          provides some information concerning injury severity, it
          should  be noted that not   all  data sources  provided
          information  on  hospitalizations,   and   that  not  all
          hospitalizations are necessarily serious in nature.  That
          is, even  in  the  case  of hospitalizations,  the severity
          of  an  individual  injury  could range  from  temporary
          respiratory  or  eye  irritation  to  a   more  critical
          condition leading to prolonged hospital  treatment.

     o    Numbers  of  Injuries;   Some of  the sources used  in
          compiling the  AHE/DB  provide exact counts  of injuries
          whereas others  (explicitly or  implicitly)  provide only
          approximations.  If  a source quantified the number of
          injuries  this number  was  recorded   in  the  database.
          However, it was sometimes necessary to translate phrases
          such as "a few"  or "about 50"   into  specific counts of
          injuries.    Thus,   the  accuracy  of  the  information
          contained in  individual records which make-up the data
          base varies.

     Although there  is  understandable  interest  in  making nation-
wide estimates of the total number of deaths and injuries resulting
from  accidental  releases  of acutely   hazardous substances,  this
study was  not designed with that purpose in mind.   Rather,  the
objective was to understand better the nature,  causes and effects
of  accidental releases.   Before  drawing conclusions  regarding
national  totals,  the   reader  should  be  aware of  the  caveats
presented above and consider the following factors:

     o    The 29  media sources  used  (26 daily  papers,  two wire-
          services  and one trade journal)  contributed  only  six
          percent of the total number of events.  But, these events
          represent 28 percent  of the total number of death/injury
          (D/I) events  in  the data base.   Although these sources
          together cover the period 1980 to 1986, no one source was
          exhaustively  searched  for  the entire period,  and most
          were  systematically  searched for  four  years  or less.
          (Exhibit 1.2-3 and Appendices D and E)
                               2-2

-------
     o    Although each  of the  26 daily  papers  had  a  separate
          regional focus,  some  also reported moderate to severe
          accidents from other regions of the country.  Thus, some
          events appeared  in  two or more of the  sources.   Given
          this, we believe that most of the events which received
          national media attention during the period 1980 to 1986
          are included in the data base.

     o    Some events involving human casualties appeared in only
          one  of  the sources searched.   This suggests  that one
          would find reports of other events of similar magnitude
          both by  increasing  the "depth" of the  data collection
          (i.e., by including  additional years of data  for sources
          already accessed), and  by the "breadth" of the search for
          events  (i.e.,  by examining   other  media sources  with
          varying regional focuses).

     o    Some types  of events may not be reported  in the data base
          due to the circumstances under which the events occurred,
          such as an event involving casualties that  was limited
          to a  work-place. ' Federal OSHA  and  state  occupational
          safety  and  health  records  have  not  been  examined  to
          determine the extent of  over-lap between those records
          and the ones in the  AHE/DB.  Thus, it is not possible to
          estimate the number of additional records that might be
          identified by accessing these types of sources.

     o    Many  of the sources  used in constructing  the AHE/DB
          reported on  events during  the  initial  stages of the
          incidents.   In  fact, in  some  cases  the  reporting party
          was  responsible for coordinating some  aspect of the
          emergency  response  effort.    Such reports  tend  to  be
          preliminary in nature,  are frequently incomplete in some
          way, and in some cases, contain inaccurate information.


2.1.1  Extent of Deaths and Injuries

     o    One  thousand  and nineteen events (9.3  percent of all
          events)  involved deaths  or injuries.  The  total number
          of  deaths  reported was  288  and the  total number  of
          reported injuries was 10,803.   (Exhibit D-2)1

     o    Of the 6,434  injuries for which severity is known,  1,472
          required hospitalization. (Exhibit D-2)
     Exhibits  using  an  alphabetic designation,  such  as  "D-2",
refer to exhibits presented in the appendices.
                               2-3

-------
     o    For those  events in  which the  number  of  injuries  is
          known, the average number of injuries was 11.2,  and the
          average number of injuries requiring hospitalization was
          5.0.  (Exhibit D-2)

     o    For those events  in which  deaths were reported  to have
          occurred, the average  number of deaths was 2.6. (Exhibit
          D-2)


2.1.2     Circumstances of Release

     o    Vapor releases were reported  as an end  effect  in 63.3
          percent of the D/I events,  spills in 56.5 percent, fire
          in 25.0 percent and explosion in 17.7 percent. (Exhibit
          2.1-6)
          In-Plant Events

          For the population  of  all AHE/DB events, and  for that
          subset  of events  that  involved deaths  or  injuries,
          storage and process  locations jointly accounted for one-
          half of the reported events.  Valves and pipes were the
          second most frequently  reported location for all events,
          but ranked third among D/I events.  (Exhibit 2.1-3)

          Insufficient  information was available to determine event
          location  for  twice as  large  a  percentage  of the D/I
          events  as  compared  to  all   events.     However,  the
          percentage  of  events  at  process vessels,   storage
          locations, vehicles not in-transit,  heating/cooling and
          other locations are nearly  the same for  all events and
          D/I events.  The percentage  of events involving releases
          from valves,  piping and fittings is ten percent greater
          for all  events as  compared to  D/I events.   (Exhibit
          2.1-3)

          Fire and explosion was reported as the attributed cause
          of events almost four times  as often for D/I events than
          for all events,  accounting for 5.2 percent of all events
          and 19.1 percent of D/I events.   Equipment  failure was
          the reported  cause of slightly less  than one-half of all
          events (46.9  percent),  but somewhat less than one-third
          of the D/I events (30.9 percent) .   Operator error was the
          reported cause in approximately 18 and 19 percent of all
          events and D/I events,  respectively.  (Exhibit 2.1-4)

          The event context was reported as "steady state" in more
          than one-half of all events,  and approximately 40 percent
          of D/I events.  Maintenance  operations and start-up were
                               2-4

-------
          both reported twice  as  frequently  as the event context
          for D/I events than  for all  events.   The percentage of
          events for which insufficient information was available
          to  determine  context   was   greater  for  D/I  events
          (approximately one-third) than  for all events reported
          in the data base (one-quarter).  (Exhibit 2.1-5)


          In-Transit Events

     o    In-transit incidents  played  a slightly  larger  role in
          the  subset  of  D/I  events  than  they  did  for  all
          events~37.4 percent versus 29.0 percent, respectively.
          (Exhibit 2.1-7)

     o    A larger share of events involving deaths and injuries,
          as compared to all events,  reported truck transport as
          the in-transit mode  (66.1 percent versus 48.6 percent).
          A smaller share of D/I events, as compared to all events,
          reported  rail  transport as  the in-transit mode (27.6
          percent versus 43.3 percent).  (Exhibit 2.1-8)

     o    Collision was reported as the in-transit cause of events
          twice  as  frequently  for the subset  of  D/I  events as
          compared to all  events.   Leaks accounted for the largest
          share of in-transit releases for both the subset of D/I
          events and for all  events (46 percent).  (Exhibit 2.1-9)


2.1.3  Types of Facilities Involved

     The industrial classification code of  the  facility  (for fixed
facility events), or the company  (for in-transit events) that was
reported to have been involved  in  the release event was identified
for 85  percent  of the 1,019  events that  reported  deaths and/or
injuries.  The study team  assigned a  4-digit Standard Industrial
Classification (SIC) code  to the  facility  or company reported to
have been  involved in  the release event, based  on information
contained in the Dun and Bradstreet Reference  Book.   Facilities
were then  further classified into five broad industrial groups:
Primary  Producers.  including  Chemical  and Allied  Products,  and
Petroleum Refining and Allied Products  (SICs  28  and 29) ; Secondary
Producers.  including all other manufacturing sectors (SICs 20 to
39,  except 28  and 29) ;  Distributors, those  firms  involved in
transporting, storing and distributing chemical products  (SICs 40
to 47,  50 and 51);  End Users, including residential and commercial
users (SICs 01 to 17, 48,  49 except 495, and 52 to 99); and Waste
Disposal (SIC 495).
                               2-5

-------
     Note that while most events  at  primary producers,  secondary
producers,  waste-related  facilities  and   end  users  are  fixed
facility events,  a small fraction are in-transit events.  Likewise,
though most events involving  distributors  are in-transit events,
some involve fixed facilities.

     o    Of all D/I events, the largest share (31.1 percent) was
          associated with distributors. Primary producers, one of
          the  areas  of  emphasis  in  most  emergency  response
          planning,  were associated with a slightly smaller share
          (26.2 percent).  Secondary producers and end users were
          associated with equivalent shares of D/I  events  (13.0
          percent and 13.1 percent, respectively).  Waste disposal
          facilities were associated with only 1.3 percent of the
          D/I events.  (Exhibit 2.1-10)

     o    Although primary producers ranked second among facility
          types in terms of the number of  D/I events reported in
          the data base,  events at these facilities accounted for
          the  largest  share  (43.4  percent)  of  the deaths  that
          occurred.   While  distributors were associated with the
          largest number  of events, these events accounted for only
          7.6 percent of the fatalities.  End users and secondary
          producers  were associated  with   17.4  percent and 8.7
          percent  of  the  fatalities,  respectively.     Events
          associated with waste disposal activities accounted for
          2.4 percent of the fatalities.  (Exhibit 2.1-11)

     o    When ordered by the  total number of injuries reported in
          association with events, events that occurred at primary
          producers rank  first among facility types—3,343 injuries
          occurred at such  facilities,  representing  31.0 percent
          of  the total.    Events that occurred at  distributors
          ranked  second   in  terms   of the  number  of  injuries
          reported,  accounting for 25.3 .percent  of all  injuries,
          while  secondary producers  accounted for  18.0 percent.
          End users  and waste  disposers were the reported facility
          type for 8.6 percent and 0.7 percent of the  injuries,
          respectively.  Eighteen  percent of all injuries occurred
          at  facilities  for  which  insufficient information was
          available to determine facility type.   (Exhibit 2.1-12)

     o    The  frequency  at which  fatalities occurred  among all
          events, by  facility type,  is shown in  Exhibit 2.1-13.
          Primary producers were  the  reported facility type for
          43.6 percent of the  9,928 events for which facility type
          could  be  determined, and  of those events 45  (or 1.0
          percent)  resulted in fatalities.   Events which occurred
          at secondary production facilities demonstrate comparable
          frequency  of  fatalities (1.1  percent  of  these  events
          reported fatalities) while  only 0.5 percent of events at
                               2-6

-------
          distribution  facilities  involved  fatalities.    While
          events that  occurred  at end users  and  waste disposers
          reported fatalities more frequently  (1.6 percent and 1.9
          percent, respectively),   it is important  to note that
          these higher frequencies may be the result of biases in
          the  reporting of  only  the more  severe  events  which
          occurred at end users and waste disposers.

     o    Similar patterns are demonstrated when the frequency of
          injury  is   considered,   except  that   in  this   case
          distributors were the most frequently reported facility
          type (11.8 percent of all events reported at distributors
          resulted in  injuries).   End  users and  waste disposal
          ranked second and third in terms of frequency  of reported
          injury   events   (10.3   percent    and   8.4   percent,
          respectively).    Events  which  occurred  at  primary
          producers and secondary producers were  less  likely  to
          involve  injuries   (6.0  percent   and   8.2  percent,
          respectively).   (Exhibit 2.1-14)

     o    Vapor  releases  were the  most  frequently  reported end
          effect associated with  D/I events for all facility types,
          except  distributors.     For events that  occurred  at
          distribution facilities,  spills were the most frequently
          reported end effect  (81.4 percent), and vapor releases
          the  second  most frequently  reported   (57.4  percent).
          Fires and explosions each  accounted for 20 to 37 percent
          of  the  D/I   events  for all   facility  types,  except
          distributors and waste disposers.   Fires were reported
          in only 13.3 percent of the events at distributors, and
          explosions in only 7.9 percent.  Fires were reported in
          one-half of  the events  at  waste disposal facilities.
          (Exhibit 2.1-15)


     It is  important  to note that  the  determination of facility
type is based  on the  identification of the two digit  SIC  of the
facility at which  the  release occurred.  Thus,  facility type is
only meant  as  a  general indication of the  industry in which the
facility is a part.  For example, while all producers of chlorine
are in SICs 28, many consumers of chlorine are also  in SICs 28 and
29.   That  is,  not all events  involving chlorine  at  facilities
reported  as  "primary  producers"  in  the  AHE/DB  occurred  at
facilities  which   produce  chlorine.     Thus,   facility   type
designations are only a general indicator of  the primary activities
at a facility.
                               2-7

-------
2.1.4     Chemicals Involved
          The 11 chemicals most frequently reported in association
          with D/I events were reported as released in 43 percent
          of such events (see Exhibit 2.1-16 for a graph of these
          substances by number of D/I events).*

               Chlorine was  reported as released in  the largest
               number of  D/I events  (121,  or  12 percent  of D/I
               events).   Five  deaths  and 1,054  injuries  were
               reported   in   association   with  these   events.
               (Exhibits 2.1-16 and 2.1-17)

               Ammonia (anhydrous),  hydrochloric acid and sulfuric
               acid were involved,  respectively, in 66, 64 and 62
               D/I events, together representing 19 percent of such
               events, at which  seven deaths and  1,524  injuries
               occurred.  (Exhibits 2.1-16 and 2.1-17)

               The  seven  chemicals  that  were  the  next  most
               frequently reported in conjunction with D/I events
               (ammonia, sodium hydroxide,  nitric acid,  toluene,
               styrene, phosphoric acid and toluene diisocyanate)
               accounted for 12 percent of  D/I events.  Forty-four
               deaths  and   1,291   injuries  were   reported  in
               association with these events. (Exhibits 2.1-16 and
               2.1-17)

          Chlorine  was  the   most   frequently  reported  chemical
          associated  with   D/I  events  at  primary  producers,
          secondary  producers,   end  users  and  waste  related.
          Chlorine was reported as released for 22 percent of the
          D/I events that occurred at primary producers; 15 percent
          of D/I events  at secondary producers; and four percent
          of D/I events at distributors.   (Exhibits E-7, E-9, E-ll
          and E-13)

          After chlorine, the next three most frequently reported
          chemicals  in association  with  D/I  events  at  primary
          producers were ammonia (anhydrous), hydrochloric acid and
          sulfuric acid.  Together  these chemicals  accounted for
          54 events,  or  20 percent  of the total number of events
          that occurred at primary production facilities.  (Exhibit
          E-7)
     2Since any given  event  may involve more  than one chemical,
there is  overlap in the  numbers of events, deaths  and injuries
reported in association with more  than one chemical.  Therefore,
all percents or  combined  totals of events,  deaths or injuries in
this section should be interpreted as upper bounds.


                                2-8

-------
     o    For events that occurred at secondary producers, the four
          chemicals  most  frequently  reported  as  released,  in
          addition to chlorine, were  ammonia, ammonia (anhydrous),
          hydrochloric acid  and  toluene diisocyanate.   Together
          these  substances  accounted  for  28 D/I  events,   or  21
          percent of the total number  of events that occurred at
          secondary production facilities.   (Exhibit E-9)

     o    For D/I events that occurred  at distribution facilities,
          the three chemicals most  frequently reported as released
          were  sulfuric  acid,   hydrochloric  acid  and  ammonia
          (anhydrous).  Together they accounted for 74 events, or
          23  percent of  the total  number  of  D/I events  that
          occurred at distribution facilities.  (Exhibit E-ll)

     o    For the 21 chemicals reported as released  in the largest
          number of D/I events, the average quantities of substance
          released per event are shown in Exhibit  2.1-17.   Among
          D/I events for the 11 most frequently reported chemicals,
          sulfuric acid releases  involved, on average, the largest
          quantities  (51,000 pounds),  followed by nitric  acid
          (17,000 pounds), and sodium  hydroxide and hydrochloric
          acid (each at 11,000 pounds).  These averages are largely
          biased by a few extremely large events. When the largest
          event of each substance is  excluded, the largest average
          quantities  involve hydrochloric  acid (9,000  pounds),
          anhydrous ammonia,  and sulfuric acid (6,000 pounds each) .

     o    When the quantities of  substances released in D/I events
          are compared to the reportable quantities  (RQs)  for these
          substances, 36.1 percent  of the D/I events for which the
          quantity  released  was  known,   involved  releases  of
          quantities of  substances less than their RQs.   It is
          interesting to note that the release of quantities less
          than  the  RQs would not  typically  be reported  to the
          National Response Center or other jurisdiction.  (Exhibit
          E-4)

     o    Only 14.3 percent of the  D/I  events involved the release
          of a  chemical substance  in quantities greater than 100
          times the RQ of the substance.  (Exhibit E-4)


     Again, it  is important  to understand the limitations of the
data base.   Many substances are  involved  in a large  number of
release incidences due to:   (1)  the high volume of the substances
used in commerce or industry, or   (2) their association with more
hazardous  substances.    For example,  sodium  hydroxide  occurs
frequently in release events reported in the data base.   However,
sodium hydroxide is not comparable in terms of toxicity with many
                               2-9

-------
of the more hazardous substances found in the data base.  The large
number of  events  involving sodium hydroxide  undoubtedly reflect
the  ubiquity of  this  substance  in  industry  and commerce,  and
possibly reflect its association (and subsequent co-release) with
more hazardous substances.
2.1.5  Relationship of Event Severity To Chemical Properties

     Information on five  physical  and toxicological properties—
acute toxicity, flammability, reactivity, volatility and processing
mode—is contained  in  CHEM,  for those  chemicals  associated with
events reported  in  the AHE/DB.   These data were  collected from
various sources, which are identified  in the  subsections below.
In addition, a determination  of the physical state of the released
chemicals was made based on information contained in the original
event reports  for most of the records  in  the AHE/DB.   Exhibit
2.1-18 summarizes  the type  of information used  and  the rating
scales employed  for  each property.    Note that  the  following
analyses are based on the  chemical  properties  of the  primary
substance reported for each event.


Acute Toxicity

     The acute  toxicity  of  many  of  the  chemicals reported  as
released in  the  AHE/DB are  characterized  in  CHEM   using values
compatible with "levels of concern" (LOG) as defined in EPA's CEPP
Technical Guidance Document.3  Where available,  National Institute
for Occupational Safety and Health (NIOSH)  measures designated by
the term "Immediately  Dangerous to Life and Health (IDLH)" were
used as a measure of acute toxicity.  Additional IDLH values were
drawn from the CEPP guidance document.

     For purposes of  this analysis IDLH values are grouped into
five LOG categories,  termed LRATEs,  which  are shown  in  Exhibit
2.1-18.   The higher LOG indices correspond to hazard  categories
previously established by organizations attempting to segregate
chemicals with regard  to the  hazards attending  their usage.  LRATE
4 corresponds to EPA's "Extremely Hazardous Substances" criteria
for categorizing  chemicals  for reporting  under  SARA  Title  III
section 302.  LRATE 3  refers to "Hazardous  Chemicals"  subject to
SARA  Title III  section  311 and  312  material  safety  data sheet
     ^.S.  EPA,  Chemical  Emergency  Preparedness Program,  Site
Specific  Technical Guidance  for  Hazards  Analysis;    Emergency
Planning for Extremely Hazardous Substances. Draft, August 1987.
                               2-10

-------
(MSDS) and chemical inventory requirements.   An LRATE value of 2
is equivalent to the SARA Title III section 313 "toxic" reporting
category.  LRATEs 1 and  0  designate  chemicals in the least toxic
categories.

     o    LRATEs were determined  for approximately 29 percent of
          the 651 chemicals in CHEM.   (Exhibit 2.1-19)

     o    Releases of chemicals for  which LRATEs were determined
          accounted for  66 percent of the events  in the AHE/DB.
          (Exhibit 2.1-20)

     o    Forty-one percent of the events in the AHE/DB involved
          chemicals  in   the  highest  toxicity  group  (LRATE  4) ,
          although only  10 percent of the chemicals reported as
          released in the AHE/DB fall into this toxicity category.
          A slightly  larger proportion of chemicals involved in
          D/I  events are   in  the  highest  toxicity  group  (19
          percent).  (Exhibits 2.1-19 and 2.1-20)

     o    Relatively  few  of  the  events  involving  deaths  or
          injuries,   or   of all events  reported  in  the  AHE/DB,
          involved chemicals  whose  toxicity  places them  in  the
          LRATE 3 category. This  is  generally consistent with the
          lower  frequency  of  LRATE 3 chemicals  in the CHEM data
          base.  (Exhibits  2.1-19 and 2.1-20)


Flammability

     Information  on  the flammability  of chemicals  was obtained
primarily from  the  National Fire  Protection  Association (NFPA),
which rates substances for  flammability (referred to in this report
as "FRATE")  on a scale  of 0  to  4,  with FRATE 4  being the most
hazardous level.4 In  cases where  NFPA ratings were not available,
flash points and boiling  points were used to assign  similar ratings
based on  a  scheme developed by the  National  Academy of Sciences
(NAS).   When  neither  NFPA  ratings nor physical property data were
available, FRATEs were  assigned according to NFPA criteria,  for
cases where the necessary  information was available.
     National Fire Protection Association  (NFPA), Fire Protection
Guide on Hazardous Materials. 1984.

     'National Academy of Sciences  (NAS), System  for Evaluation of
Bulk Water Transportation of  Industrial Chemicals:  A Report to the
Department of Transportation. U.S. Coast Guard.  1974.
                               2-11

-------
     o    Information   on   flammability    was    obtained   for
          approximately  65  percent  of  the  chemicals  in  CHEM.
          Releases of these chemicals represent 87 percent of all
          events in the AHE/DB.  (Exhibits 2.1-19 and 2.1-20)

     o    Forty  percent  of all  events  involved  non-flammable
          substances  (FRATE 0) ,  and  approximately  30  percent
          involved chemicals in the two highest flammability rating
          groups (FRATEs 3 and  4) .  The distribution of D/I events
          by flammability rating is similar to the distribution for
          all events, with the  exceptions that a larger percentage
          of these events report  FRATE  as "unknown," and a smaller
          percentage  report the  highest  flammability  category.
          (Exhibit 2.1-20)


Reactivity

     Reactivity was characterized using the NFPA definition, "the
susceptibility of materials  to  release  energy either by themselves
or in  combination  with water."  Observed  substance  responses to
fire  exposure,  shock  and pressure  are used to  determine such
susceptibility.   As with flammability,  NPFA assigns degrees of
reactivity  (RRATE)  on a scale of 0  to 4,  with 4  being  the most
reactive.

     o    Reactivity ratings (RRATEs) were determined for only 40
          percent  of the chemicals  in CHEM.  Very  few  of the
          chemicals  were  assigned  to  the  highest  reactivity
          category; only 4.0 percent were assigned RRATEs of 3 or
          4.  A similar pattern holds for D/I events,  although in
          this case RRATEs could be  identified  for 56 percent of
          the released substances. (Exhibit 2.1-19)

     o    A comparable pattern is  reflected when considering the
          number of events, with only 3.2 percent of all events and
          2.0   percent  of  D/I  events involving  chemicals with
          RRATEs 3 and 4.  (Exhibit 2.1-20)


Volatility

     Vapor pressure, which indicates the tendency of molecules to
escape from liquids or solids,  is used as a measure of volatility
in CHEM.  Ratings  of high, moderate and  low volatility (VRATES)
were assigned to the chemicals based on criteria established by
                               2-12

-------
previous  investigators  (identified  in  Exhibit  2.1-18).6   These
categories are compatible with the chemical selection criteria used
for the SARA Title III section 305(b) Review of Emergency Systems.

     o    Vapor pressures have been identified for only 33 percent
          of the chemicals in CHEM.   These  chemicals were reported
          as released in one-half of all AHE/DB events.  The rated
          chemicals are evenly divided into the low,  moderate and
          high volatility groups.   (Exhibits 2.1-19 and 2.1-20)

     o    A slightly greater percentage of the  events (for which
          chemical vapor pressure is known) involved chemicals in
          the  low  volatility  group.    A similar  pattern  was
          demonstrated  for events  involving deaths and injuries.
          (Exhibit 2.1-20).


Physical State

     A chemicals's physical state  at the  time of an  event can be
an  important  factor determining the severity of an  event.   The
AHE/DB contains information on physical state, specifically:

     o    Information regarding the physical state of the primary
          substance released was  available for  99  percent of the
          events reported in the AHE/DB.   (Exhibit D-l)  Note that
          many events involve the release of multiple substances,
          sometimes  with   different  physical  forms.     Thus,
          percentages sum to greater than 100.

     o    Releases of liquids predominate for both D/I events and
          all events, representing the chemical state  in 72 percent
          of D/I events and  74  percent of all events.  (Exhibits
          D-l and E-l)

     o    Gases  are the second  most commonly  reported  physical
          state,  reported in  32 percent  of D/I  events  and 27
          percent of all events.  (Exhibits D-l and E-l)

     o    Solids were reported  as released much less frequently,
          —only 12 percent of D/I events and four percent of all
          events reported the release of solids.   (Exhibits  D-l and
          E-l)
     Versar, Inc., Physical-Chemical Properties and Categorization
of RCRA Wastes According to Volatility,  prepared  for the U.S. EPA,
February, 1985; and U.S.  EPA Office of Toxic Substances, Section
305 (b)  Review of Emergency Systems — Chemical  Selection Criteria.
Draft, January, 1987.
                               2-13

-------
Processing Mode
     As a proxy for processing mode,  chemicals reported as released
in the  AHE/DB are  designated as either commodity  or  specialty
chemicals.  This proxy  is  based  on  the assumption that commodity
chemicals, which are produced  in large quantities,  are  likely to
be produced in  continuous processing modes,  whereas  specialty
chemicals,  which  are produced in small quantities  to  meet more
narrowly  defined specifications,  are  more  likely  to  be  batch
processed.

     o    Approximately 41 percent  of the chemicals in CHEM are
          commodity chemicals, 35 percent  are  specialties and 24
          percent could not be designated.  (Exhibit 2.1-19)

     o    Seventy-seven percent  of  all  events  involve  commodity
          chemicals, while only  seven  percent  involve  specialty
          chemicals.  Although the shares of D/I events involving
          commodity  and specialty chemicals are similar to the
          shares  for all events, the distribution  of  chemicals
          involved  in   D/I  events  is more  heavily weighted  to
          commodities.   (Exhibits 2.1-19 and 2.1- 20)


2.1.6  Variation of Event Severity With Chemical Properties

     Acute  hazardous events  can be   separated  into two groups
depending upon whether or not a fire or  an  explosion occurred
during the  event.   In  those  cases where a fire  or explosion was
reported as an end effect,  the severity of the event, in terms of
number  of  injuries  or  deaths,  appears  to  be  predominantly
determined  by  the  characteristics of  the  fire  or explosion. For
those events where a fire  or an explosion was not reported as an
end effect,  we assume that  an  important factor contributing to the
severity of the event was the toxicity of the chemicals involved.
Since multiple end effects are often reported for a single event,
individual  records  were  examined  in  order  to  consider  all
information when classifying  the  cause of the deaths or  injuries.


Variation With Toxicity

     o    Approximately 65 percent of the injuries and  13 percent
          of the deaths reported  in  association with events in the
          AHE/DB were  apparently caused by  toxicity.    (Exhibit
          2.1-21)
                               2-14

-------
     o    For the  subset of events  that resulted in  deaths and
          injuries which were thought  to  be caused  by chemical
          toxicity, more than two-thirds of the events, for which
          a chemical toxicity rating  could be determined, involved
          highly toxic chemicals (LRATE 4).  (Exhibit 2.1-20)

     o    Events involving chemicals with  high  LRATEs had a much
          wider range of severity  (in  terms  of  numbers of deaths
          and injuries)  than those with a low LRATE.   (Exhibit
          2.1-22)   A  chi-square  analysis of  the grouped  data
          suggests that  this phenomenon for  D/I events is due to
          the underlying distribution of the two variables in the
          data set.  That is, statistical analysis indicates that
          there may be a relationship between severity  (as measured
          by  the number  of  deaths  and injuries)  and level  of
          chemical toxicity.

     o    Chi-square analysis on the  toxicity ratings of chemicals
          involved in  all events  in the  AHE/DB,  for  which the
          primary end effect was a vapor release  or spill, suggests
          that a  relationship  exists between event  severity and
          chemical toxicity.  The analysis  indicates that severity
          is  related  to toxicity  in determining whether  or not
          deaths or  injuries  will  occur, but that the degree of
          severity, i.e., number of  deaths or  injuries, does not
          depend on toxicity.   That  is, if an  accidental release
          of  a  highly  toxic chemical  occurs,  the event  is  more
          likely to  cause deaths  and/or injuries.   However,  we
          cannot predict the number of deaths or injuries.


Variation of Events Severity
With Flammability and Reactivity

     o    Approximately 87 percent  of the deaths and 35 percent of
          the injuries reported in association with events in the
          AHE/DB  were  apparently  caused by  fire or  explosion.
          (Exhibit 2.1-21)

     o    A chi-square analysis  indicates that there is  no apparent
          relationship between the  severity of events,  as measured
          by  the  numbers  of  injuries    plus  deaths,  and  the
          flammability  or reactivity  ratings  of the chemicals
          involved. (Exhibits 2.1-23 and 2.1-24)
                               2-15

-------
2.2  EVACUATION EVENTS

     The expanded  coding format for  the present version  of the
AHE/DB allowed each record to  incorporate information on whether
or not an evacuation was performed,  how many people were evacuated,
and of those, how many were  residents and how many were workers.
Evacuations occurred in  association with 829  (or 7.6 percent)  of
the 10,933 events in the  AHE/DB.  Of the  remainder, 5,067 (or 46.4
percent)  of the events did not have associated evacuations, while
for 5,037 events (or 46.1 percent)  it was not reported whether an
evacuation took place.  The 29 media sources (26 newspapers, 2 wire
services and  one trade  journal)  contributed  54 percent  of the
evacuation records.

     Similar caveats to those discussed for death  and  injury events
apply  to this  analysis, particularly regarding  the  number  of
persons evacuated and the distribution among residents and workers.
2.2.1     Extent of Damage

     o    Over  800  events  (or  7.6 percent  of all  events)  were
          reported to involve evacuations. In  the  405 events for
          which the number  of evacuees  was  identified,  the total
          number of people  evacuated was  435,851.   (Exhibits D-l
          and D-2)

     o    Injuries  were  associated with 39.8  percent  of  the
          evacuation events, compared to  only  9.1  percent of all
          events.  (Exhibit  2.2-1)  Deaths were  associated with 4.8
          percent of the evacuation events compared to 1.0 percent
          of all events.  (Exhibit 2.2-2)

     o    For some  of the  evacuation events it was  possible to
          distinguish  between  the  evacuation  of  workers  and
          non-workers.  Of the 335,323  persons evacuated in these
          events, 308,183 (91.9  percent) were residents and 27,140
          (8.1 percent)  were workers.   (Exhibit D-2)

     o    For events  where  the  number  of persons evacuated was
          reported, the average  number  of evacuees per event was
          1,076.  For those  events in which a distinction was made
          between  residents and  workers, the  average  number of
          residents evacuated was 973  and the average  number of
          workers evacuated was 99.  (Exhibit D-2)
                               2-16

-------
2.2.2     Circumstances of Release

     o    In-transit  events played  a  slightly  larger  role  in
          evacuation events as compared to all events—34.4 percent
          of all  evacuation events  versus 29.0  percent  of  all
          events.   (Exhibit 2.2-3)

     o    Spills were reported  as an end  effect in 49.9 percent of
          evacuation events, vapor release  in  76.1  percent,  fire
          in 32.3  percent  and  explosion  in  15.1.   Fires  and
          explosions  were   reported   nearly  five  tines  more
          frequently in association with evacuation events than for
          all  in-plant  events.   Vapor  releases  also were  more
          frequently reported,  and spills less frequently, as  end
          effects in evacuation events as compared to all events.
          (Exhibit 2.2-4).


          In-PIant Events

     o    Storage locations were  reported to be  involved in 35.1
          percent of in-plant evacuation events,  compared to 28.1
          percent  for   all  events.     Vehicles  not-in-transit
          accounted for  5.7 percent of  evacuation events and 2.8
          percent of all events.  Process locations accounted for
          a smaller share of evacuation events (19.5 percent) when
          compared to all events (22.4 percent), as did valves and
          pipes (14.3  percent  for evacuation events versus 25.4
          percent for all events).  (Exhibit 2.2-5)

     o    Fires and  explosions were  reported  as the  attributed
          cause four  times more  frequently in  evacuation events
          (21.5 percent)  when compared to all in-plant events (5.2
          percent).    Sparking  and  lightning,   and  arson  and
          vandalism were also more frequently  reported as causes
          in evacuation events.  Equipment failure represented the
          event cause in 46.9 percent of  all events, but only 33.3
          percent of evacuation events.    (Exhibit 2.2-6)

     o    Steady state condition was the  in-plant event context in
          over one-half of all events (51.0 percent) and evacuation
          events (55.9 percent) .  Start-up conditions were reported
          as the in-plant event context three times more frequently
          for evacuation events (4.4 percent)  than for all events
          (1.8  percent),  while loading and unloading was reported
          as the in-plant event context one-half as frequently (9.6
          percent)  for evacuation events as for  all events (16.2
          percent). (Exhibit 2.2-7)
                               2-17

-------
          Events involving releases at disposal locations accounted
          for 83.0  percent  of the  total quantity  of substances
          released during in-plant evacuation events, as compared
          to 5.9  percent of the  substances released  during all
          in-plant events.  The large share of  material released
          at disposal locations during in-plant evacuation events
          was attributable to a single event which released a very
          large  quantity  of  substance  (16,500,000  pounds  of
          hazardous waste).   (Exhibit 2.2-8)

          Fire and/or explosion, which was the attributed cause for
          5.7 percent of  the total  quantity of  material released
          in all in-plant events, accounted for 84.0  percent of the
          material released during in-plant events which involved
          evacuations.    (Exhibit 2.2-9)   This  aberration results
          from the same,  large  single  release of 16,500,000 pounds
          of hazardous waste.
          In-Transit Events

     o    Truck  modes  were  responsible  for  50.2  percent  of
          evacuation events while 45.3 percent involved rail.  This
          represents approximately the same  percentages as their
          shares in all events.  (Exhibit 2.2-10)

     o    Derailment was the attributed  cause  in  24.9 percent of
          the evacuation events, which is approximately  four times
          more frequent than derailment's share of all events (6.5
          percent).  Collisions were the attributed cause in 21.4
          percent of the evacuation events, which is almost twice
          the share of all  events accounted for by collision (11.6
          percent).  (Exhibit 2.2-11)

     o    Rail-associated releases  accounted  for two-thirds of the
          quantity of substances released in in-transit evacuation
          events.  However, rail related events  represented only
          22.3 percent of  the  quantity of  substances released in
          all in-transit events.   (Exhibit  2.2-12).


2.2.3     Types of Facilities Involved

     o    Of the 829 events for which evacuations were reported,
          65.6 percent occurred at fixed  facilities.   (Exhibit F-l)

     o    We were  able  to identify  industrial  sectors  for  the
          facilities and/or  firms  reported  in association  with
          release events in 79  percent of the 829 events reporting
          evacuations.   (Exhibit 2.2-13)
                               2-18

-------
Events that  occurred at distribution  facilities (25.9
percent)  and primary production facilities  (25.0 percent)
accounted for the majority of evacuations.  Events that
occurred  at end  users  (14.5 percent)  and  secondary
producers (12.6 percent) accounted for the next largest
shares.    Events  occurring  at  firms  involved  with
waste-related activities accounted for only 1..1 percent
of the evacuation events.  (Exhibit 2.2-13)

Compared to their frequency of involvement  in all events,
distributors  and  end  users  have  a  slightly  higher
frequency and  primary producers a  significantly lower
frequency of involvement in evacuation events.  The rates
of involvement  for facilities classified  as  secondary
producers  and  waste  related  are  slightly  lower  for
evacuation events than for all events.   (Exhibit 2.2-13)

For events that occurred at facilities coded as primary
producers, 4.8  percent involved evacuations.  Similarly,
evacuations were reported for 6.8 percent of  events at
secondary  production facilities,  8.0  percent   of  the
events at distributors,  and 9.7 percent  of the events
reported at end-users.   (Exhibit 2.2-14)

On average, more people were  evacuated in distributor-
related events than  in  any  other sector.   Distributor-
related  events  accounted  for  25.8  percent  of  the
evacuation  events,   and 38.0  percent  of all  persons
reported as  evacuated.   In contrast,  end-user-related
evacuation events  tend  to  involve a  smaller  number of
evacuees, i.e.,  while  14.5 percent of the  evacuation
events involved end-users, these  events contributed only
7.3 percent of all reported evacuees.  (Appendix 0)

Of the 288 events  in  which evacuees can be identified as
workers or residents, residents,  on average,  accounted
for 90.7 percent of the persons evacuated while workers
accounted for 9.3 percent.  The proportion  of workers and
non-workers varied greatly by sector:

Primary Producers:     10.0 percent were workers
                       90.0 percent were non-workers

Secondary Producers:    34.5 percent were workers
                       65.5 percent were non-workers

Distributors:           2.9 percent were workers
                       97.1 percent were non-workers
                     2-19

-------
          End-Users:             21.2 percent were workers
                                 78.8 percent were non-workers

          Waste Related:         82.9 percent were workers
                                 17.1 percent were non-workers
2.2.4     Chemicals Involved
          The ten chemicals most frequently reported in association
          with evacuation events accounted for  releases at up to
          43 percent of such events.7  (Exhibit  2.2-16)

               Chlorine was involved in the most events  (89), that
               representing  10.7  percent  of  evacuation  events
               reported in the AHE/DB.  (Exhibit 2.2-16)

               Hydrochloric acid, ammonia  (anhydrous) and sulfuric
               acid  were  involved  in 76,  67  and 30  evacuation
               events, respectively.  Together, they accounted for
               21 percent of such events.   (Exhibit 2.2-16)

               The six chemicals next most frequently reported in
               conjunction with evacuation events accounted for 11
               percent of such events.  (Exhibit 2.2-16)
     o    Chlorine was  the most frequently reported  chemical in
          association with evacuation events at primary producers,
          secondary  producers,  end  users   and  waste  related
          facilities.   Chlorine was  reported as released  in 18
          percent of the  evacuation  events  at primary producers;
          12 percent of the events   at  secondary producers; four
          percent at distributors;  18 percent  at  end users; and 44
          percent at  waste disposal facilities.   (Exhibits F-7,
          F-9, F-ll, F-13 and F-15)

     o    After chlorine, the six substances next most frequently
          reported   as   released   at  primary   producers   were
          hydrochloric acid, ammonia (anhydrous), vinyl chloride,
          ammonia, sulfuric acid, and nitric acid. Together, these
      Since any given event  may involve more  than  one chemical,
there is an overlap in the numbers of events, deaths and injuries
reported in association with more  than  one chemical.   Therefore,
all percents or combined totals  of events,  deaths or injuries in
this section should be interpreted as upper bounds.


                               2-20

-------
chemicals account for 71 events, or 34 percent of the 207
events which occurred at primary production facilities.
(Exhibits F-6 and F-7)

For  events  that occurred  at secondary  producers,  the
three chemicals most frequently reported  as released, in
addition to chlorine, were ammonia (anhydrous), ammonia
and nitric acid.   Together,  these substances accounted
for  20  events, or  19  percent of  the total number of
events that occurred at  secondary producers.  (Exhibit
F-9)

For events that occurred  at distributors, the three most
frequently reported chemicals were  hydrochloric  acid,
ammonia  (anhydrous)  and  sulfuric acid.   Together they
accounted for 50 events or 23 percent of the total number
of events that occurred at  distributors.  (Exhibit F-ll)

The average quantities of substances released per event
for  the 19  chemicals that  were  released  during  the
largest number of evacuation  events are shown in Exhibit
2.2-16.   Among evacuation  events,  releases of  methyl
alcohol  represent,  on  average, the  largest quantities
(975,000 pounds of  methyl  alcohol  were released during
evacuation events),  followed  by  sulfuric acid (130,000
pounds) and phosphoric acid (100,000 pounds).  However,
these  average quantities  are  heavily influenced by a
small  number  of  very  large  releases.  Excluding  the
largest  single  release  for each  substance  changes  the
rank ordering.  The chemical  with the largest adjusted
average  release is  still methyl  alcohol  (75,000 pounds
in  one  release),  followed by  styrene  (15,000  pounds
average  for  five releases),  hydrochloric  acid (14,200
pounds average  for  27 releases),  sulfuric acid (11,800
pounds  average for 12  releases),  and  ammonia   (9,000
pounds average for eight releases).

When  quantities of materials  released  in  evacuation
events are compared to the reportable quantities (RQs)
for  these  substances,  25.4  percent of  the evacuation
events   for  which  release  quantities  were  reported
involved quantities less  than  the RQ.   That  is,  the
release  of such quantities of substances would  not be
reported to the National Response Center or most other
governmental jurisdictions.  (Exhibit  F-4)

Only 19.7 percent of the evacuation events involved the
release  of  substances in  quantities  greater  than  100
times the released chemical's RQ.  (Exhibit F-4)
                     2-21

-------
2.3  MOST SEVERE EVENTS INVOLVING ATMOSPHERIC RELEASES
     OF ACUTELY TOXIC OR EXPLOSIVE CHEMICALS

     As stated at the beginning of this chapter, some caveats are
necessary in interpreting this data. These are summarized below.

     First,  only  a   few  of   the   reported  events  have  been
independently verified.   Thus,  one should interpret  the data as
"deaths or  injuries  reported by one or more  of the contributing
sources."  Second, direct evidence  of  causality rarely exists in
the contributing sources to  the data base.   That is,  some of the
death or injuries reported to have occurred in association with a
particular event may not,  in  fact, have been caused by the release
of a chemical during an event.  For example, an event involving the
release of  a substance following  a motor vehicle  collision may
involve a death related to the impact of the collision.  Although
deaths explicitly resulting  from a collision were removed from the
data  base whenever  possible,  it   is  not possible to  identify
causality in all cases. Third,  the  term "injury" is used to refer
to physical manifestations ranging  from temporary respiratory or
eye  irritation treated  on-site to critical  injury  leading  to
prolonged hospital treatment.   Finally,  in some cases the source
of data used did not provide exact  numbers  of persons injured or
evacuated.  In these cases standard rules were  used to code such
terms as "a few" and "about 50" into specific numbers.


2.3.1     General Characteristics

     o    Air releases of toxic chemicals that led to two or more
          deaths are listed in Exhibit 2.3-1.   Events involving the
          release of a chemical vapor,  a fire or an explosion, or
          a substance with a  gaseous physical  state are classified
          here as  "air releases."   Thirty-six such  events were
          identified as having occurred during the five-year period
          from 1982 to 1986,  an average of seven events per year.
          These events were reported to have led to 127 deaths in
          total.  These air release events comprised 32.1 percent
          of all events with fatalities and  44.1 percent  of the
          total number of fatalities.

     o    The most serious single event, with respect to the number
          of fatalities was an explosion and fire associated with
          gunpowder and fireworks  that  led to  21 deaths in June of
          1985.
                               2-22

-------
Four-fifths  of  these  events  reporting  two  or  more
fatalities were limited to"two or three deaths, and only
two (five percent)  of the events resulted in ten or more
deaths.

Events  involving 40  or  more  injuries  are  listed  in
Exhibit 2.3-2.  Forty-two such events were identified as
having occurred during the five year period from 1982 to
1986, an average of eight events per year.  These events
resulted in 4,121 injuries.  These most severe air injury
events account for 4.3 percent of all injury events and
38.1 percent of all injuries.

The most serious single event, with  respect to the number
of injuries,  was a vapor release of aldicarb oxime in
August, 1985, which resulted in 430 injuries.

Three-quarters  of   the  most severe  events  involving
injuries had  90  or  fewer reported  injuries,  while only
four  events  (ten  percent)   resulted  in 200  or  more
injuries.

Events that  involved  2,000  or more evacuees are listed
in Exhibit  2.3-3. Thirty-six such  events were reported
during  the   five  year   period  from   1982  to  1986,
representing, on average, seven events per year.  These
events resulted in  a  total of 188,075 reported evacuees.
These most severe air evacuation events account for 4.3
percent of all evacuation events and 60.9 percent of all
evacuations.

The most  serious event,  with respect to the  number of
evacuees was  an  explosion and  fire associated with the
release  of 22,500  pounds  of  phosphorus,  causing  the
evacuation of 40,000 people  in July of 1986.

Two-thirds of the most severe evacuation events involved
4,000  or  fewer evacuees,  while  three  events  (eight
percent) led to 10,000 or more evacuees.

Transportation events played a significant  role in the
most severe evacuation events,  accounting for one-third
(12)  of   the  36   most  serious   evacuation  events.
Furthermore, transportation  events account for  13 percent
of the events with deaths and  38 percent of the events
involving  injuries.  (Exhibit E-19)    The  most severe
transportation  event  with   fatalities  involved  seven
deaths.
                     2-23

-------
     o    The  number of  most  severe events  involving  deaths,
          injuries and  evacuations,  shown  in  Exhibits  2.3-1  to
          2.3-3, appears to have remained relatively constant over
          the  1982 to 1986 period,  with  the exception  of 1985.
          This increase in 1985 may have been a result of greater
          public interest and more comprehensive reporting of such
          events in the immediate aftermath of the Bhopal tragedy
          rather  than an  actual  increase  in  the occurrence  of
          accidents.     The number of severe incidents  listed  in
          Exhibits 2.3-1 to 2.3-3 is too few and the time frame too
          short to provide  a reliable basis for  evaluating these
          types of trends.


2.3.2     Significance of Event End Effects

     o    The most significant factor in determining the severity
          of events involving fatalities appears to be whether or
          not a fire  or an explosion  occurred.  Most of the events
          in which two or more deaths occurred (listed in Exhibit
          2.3-1) involved a fire and/or an explosion  (86 percent) .

     o    Five  (or 14 percent) of the  events listed  in Exhibit
          2.3-1  occurred  in situations  where chemical  toxicity
          appeared to be  the primary factor in determining event
          severity, i.e., where a spill or vapor release occurred
          but  fire and  explosion did not.   In all  cases,  these
          toxicity-dominated events were restricted to two or fewer
          deaths.

     o    Toxicity appears to be a somewhat more important factor
          in determining the severity of events involving injuries.
          For  injury events,   toxicity  was  indicated  to  be  the
          primary factor in 25 (or 60 percent)  of the most severe
          events, and accounted for three of the four events which
          resulted in 200 or more injuries.

     o    Fires  and/or  explosions appear  to have  had  the  most
          significant   role  in  determining  the   severity   of
          evacuation events, accounting for 25 (or 70 percent)  of
          the  36  evacuation events  listed  in  Exhibit  2.3-3,  the
          three most severe events,  each of which led to reported
          evacuations of 10,000 or more persons.


2.3.3  Large Scale Domestic Releases of Extremely Toxic Chemicals

      The potential toxic effects  of  a chemical release event vary
with the  toxicity and  the volume of  substance  released.   Five
thousand two hundred  and  seventy  eight events  in the AHE/DB have
information  on both   quantities  and  acute  toxicity.    For those


                               2-24

-------
events, we divided  quantity released by the IDLH value attributed
to that chemical to establish a  relative measure  of the potential
toxic effects of each of these releases.   Exhibit 2.3-4 lists the
25 events in the AHE/DB having highest quantity-to-toxicity ratios.

     A  number of  caveats  are  necessary in  order to  interpret
properly  the  information  presented on  Exhibit  2.3-4.    First,
substances are commonly distributed in solution, the concentration
of which is dependent on the intended use of the product.  Second,
since the  IDLH  value presented  on Exhibit 2.3-4  is based  on the
toxicity of a concentrated form of each substance, the  ratio of
quantity released to toxicity for events involving the release of
a chemical  in solution will  be overstated.    In  addition,  these
results are highly dependent on the accuracy  of  the  information
gathered in developing the AHE/DB.  Errors which overestimate the
quantity of a substance  released  will  result  in an over-estimate
of the  ratio  of quantity  to toxicity.  Given  these caveats,  the
reader is cautioned not to over-interpret the significance of this
analysis.

     In addition, a number of substances appearing on Exhibit 2.3-
4 are not generally classified as  highly toxic  substances, and are
not  comparable,  in  terms  of toxicity,  with  many  of the  more
hazardous  substances found in  the  AHE/DB.   For  example,  sodium
hydroxide (caustic soda)  appears on this list four times, possibly
due  to  the ubiquity of  this substance in commerce and industry
(i.e., it is likely to be involved in more events simply due to its
common use), and to the large quantities of caustic soda commonly
stored  and shipped  (i.e.,  when  released,  the average  quantity
released  will  be   large).    However,  sodium   hydroxide  is  not
generally considered an extremely hazardous substance.

     The listing presented on Exhibit 2.3-4 is important  not so
much  for  identification  of particular  events,  or  substances
released during these events, as  for the  insights it  may provide
into those events  with potentially  severe end  effects.   Thus, it
can serve as a starting point in the identification of events for
which potentially catastrophic outcomes were avoided, due to event-
specific conditions, preparedness planning or  intervention on the
part of plant personnel or emergency response personnel.

     o    By  this   measure,   25  events  involved  releases  whose
          quantity-to-toxicity ratio was greater than one-half that
          generally associated with the release at Bhopal,  and 17
          events involved releases whose quantity-to-toxicity ratio
          exceeded that of Bhopal.

     o    Among the 25 events whose quantity to toxicity ratio was
          at least one-half of that associated with Bhopal (Exhibit
          2.3-4), only one event reported deaths (two fatalities).
          Five events reported injuries (ranging from five to 650
                               2-25

-------
          people injured),  and six  reported  evacuations  (up  to
          20,000 people).   This is a somewhat greater frequency of
          severe consequences than that observed among all events.
          The average numbers  of  injuries and evacuations   were
          somewhat  greater for  these 25  high-damage  potential
          events when compared  with  the  AHE/DB as a whole.   The
          average number  of deaths  was  approximately the  same.
          However,  given the vast amount  of materials released in
          the events in these high-damage potential events,  their
          severity remains surprisingly low.

     o    The difference between the potential for human harm and
          the actual consequences of these events appears  to  be
          due, in part,  to a combination  of  one or more  of the
          following factors:

               The chemical's physical state at the time of release
               was  liquid or  solid,  and  much  of  the  material
               remained liquid  or solid  when released,  or  other
               physical/chemical properties prevented  widespread
               airborne exposure.

               Few people were close to the point of release.

               Weather conditions were favorable.

               Preparedness  at  the  site  of  release  included
               containment  of   materials  released,  or   other
               equipment or procedures to  mitigate the  extent of
               release and population exposure.

               Evacuation  or  other  emergency  response  may  have
               removed potentially exposed victims.


2.3.4     Comparison with  Major Industrial Disasters of Similar
          Origin

     o    Ten  of this  century's  largest  industrial  disasters
          related to large scale atmospheric releases of toxic or
          explosive chemicals are listed  in Exhibit 2.3-5.  Bhopal
          is by far  the most severe,  accounting for more than 3,000
          deaths to date,  with  reports  of  more  than  200,000
          injuries,   40,000  of which  have  been  classified  as
          serious.
                              2-26

-------
The most serious event of this type to occur in the U.S.
is the  explosion of  a  fertilizer ship in  Texas City,
Texas in  1947.   The  total  number of  fatalities (552)
associated  with this ammonium  nitrate related  event
places it in a  category  of  only  three  other events (in
addition to Bhopal) where the death toll exceeded 500.

Toxicity represented  the primary human health  risk in
only two of  these industrial disasters.   One of these was
the 1976 release of TCDD in Seveso, Italy that caused 700
homes to be  permanently abandoned and hundreds of animals
to die,  but did not involve any human fatalities. Thus,
Bhopal is the only large scale chemical release event for
which substance toxicity can be attributed as the cause
of a large number of human deaths.

The  average number of  fatalities for  the ten  events
listed in Exhibit 2.3-5  is  nearly 600,  far  larger than
the  number  of  fatalities that occurred at any  of the
events  listed  in Exhibit  2.3-1   for the  1982 to 1986
period.    The largest number of fatalities among  the
latter is 21.

An important factor contributing to the total number of
deaths in Bhopal was the lack of warning and instructions
in proper response  given to the  populace prior  to and
during the period of release.  In Seveso, Italy hundreds
of  animals were  killed because they  could  not  be
evacuated effectively. The long term effect of the event
was the contamination  of the soil and the abandonment of
a large land area.

Thus, the data in Exhibit 2.3-5 indicate that events of
severe magnitude have  to date been associated either with
situations in which massive explosions have occurred or
in which toxic materials have been allowed to reach the
surrounding  populace   without  a timely  or  effective
response by emergency response officials.
                     2-27

-------
3.0  CHARACTERISTICS OF ALL EVENTS
3.1  EXTENT OF DAMAGE
          Injuries were  reported for 9.1  percent of  the  10,933
          events in the AHE/DB.  (Exhibit 2.2-1)

          Deaths were reported for only 1.0 percent of the events.
          (Exhibit 2.2-2)

          Evacuations were reported  for  only 7.6 percent  of the
          events; however,  for 46 percent of the events it was not
          stated whether evacuation was required.   (Exhibit D-l)1

          A total  of approximately one-half million  people were
          reported as evacuated  as  a result of chemical release
          events.      This   total   excludes   non-quantitative
          descriptions of  number  of people  evacuated,  such  as
          "several people" or "all employees."  (Exhibit D-2)

          In 288 events, accounting  for  308,822 evacuees,  it was
          possible  to  distinguish  between  the  evacuation  of
          residents and workers.  These evacuation events involved
          a total  of over 280,000 residents and  28,000  workers.
          (Exhibit D-2)

          When  evacuations  were  reported,  an  average  of  1,076
          people per event were  evacuated.  The number of people
          evacuated ranged from one to 40,000 people.   (Exhibit D-
          2)

          Property damage  was reported  for  13.5  percent  of all
          events.  In 43.2 percent of the  events  property damage
          was stated not to have occurred,  and in 43.3 percent of
          the events it was not stated whether such damage had
     1As   noted   elsewhere   in  this  report,  exhibits  using  an
alphabetic designation,  such as "D-2", refer to exhibits presented
in the appendices.
                               3-1

-------
          occurred.   Estimates  of property  damage are  usually
          expressed  in  very  general  terms,  ranging up-to  $100
          million dollars for one event.   (Exhibit D-l)

          Similarly, less than three percent of the 10,933 events
          were reported to have resulted in environmental damage,
          while one-third were  reported  not to have resulted in
          environmental damage.   For the majority of events (64.2
          percent), the  reporting  source does not  state whether
          environmental damage occurred.   (Exhibit D-l)
3.2  CIRCUMSTANCES OF RELEASE

     o    Of  the  10,933  events,  71  percent  occurred at  fixed
          locations;  the  remainder  occurred  during  transit.
          (Exhibit 2.2-3)

     o    Spills were reported as an end effect in 68.3 percent of
          the events; vapor release in 39.3  percent;  fire in 6.7
          percent; and explosion in 2.9  percent.   The severity of
          end effects was  estimated  to  be  (in decreasing order)
          explosion,  fire, vapor  release and  spill.    Note  that
          multiple end-effects may be reported, therefore, percents
          do not total to 100.  (Exhibit 2.2-4)

     o    Of  the  7,807 events  for which  quantity released  was
          reported,  those events involving the release  of more than
          100,000 pounds of material (which represent three percent
          of  the  events  in  the data base)  accounted for  88.0
          percent of the  total  quantity  of  reported  releases.
          (Exhibits 3.2-1 and D-6)


3.2.1  In-Plant Events

     o    Storage vessels, process vessels,  and  valves or piping
          accounted  for  approximately   equal  shares  of  event
          locations for the 7,760 in-plant  events  (22 percent to
          28 percent each).  (Exhibit 2.1-3)

     o    Events occurring at storage vessels  accounted for  75.2
          percent of the quantity of material released as a result
          of in-plant events.  However, when deaths or injuries are
          involved,  "vehicles not in-transit"  accounted for  50.7
          percent of  the quantity released,  while  waste disposal
          and pollution control  equipment accounted  for  another
          35.4 percent.  In each case, the disproportionate shares
          of quantities of materials released are attributable to
          a single event with a large release quantity.  (Exhibit
          3.2-2)
                               3-2

-------
Equipment failure was  identified as the cause  of 46.9
percent  of  all  events  that  occurred  in-plant,  but
accounted for only 32.0 percent of the quantity reported
as released in-plant.   In comparison, operator error was
the attributed  cause  for  18.2 percent of  the in-plant
events, while accounting for 46.0 percent of the quantity
released.    For  D/I   events,  sparking  and  lightning
accounted for 2.2 percent of the  events,  but  over 50
percent of the quantity of material released.  The large
share of material released attributable to sparking and
lightning was due to  a  single  large  quantity  release
event.   (Exhibits 2.1-4 and 3.2-3)

In-plant events  primarily  occurred under  steady state
conditions (51.0 percent) or during loading or unloading
operations (16.2 percent).   Operational context was not
reported for one-quarter of the incidents.   (Exhibit 2.1-
5)

Less than six percent  of the  in-plant events accounted
for over 92 percent of the quantity of material released.
Those events  that released more  than  100,000 pounds of
substances (3.2  percent of in-plant events)  accounted for
over  89 percent of  the  total  quantity  of  material
released.  (Exhibits 3.2-1 and 0-6)

Events that involved releases of over 100,000 pounds of
material accounted for the majority of material released
for   all  in-plant   locations   except   valves/pipes,
heating/cooling and "other" locations.  (Exhibits 3.2-4
through 3.2-9)

The  majority of events  that  listed event cause  as
equipment failure, upset conditions, fire/explosion, high
temperature  and  pressure,  power related,  sparking and
lightning, other or  unknown involved the release of less
than 1,000 pounds of substances.  The majority of events
that  listed  operator  error,  arson  and   vandalism  or
disposal related activities as event cause involved the
release  of   less   than  10,000  pounds   of  material.
(Exhibits 3.2-10 through 3.2-18)

Events that involved the release of over 100,000 pounds
of  material  accounted for  the majority  of  material
released  for  all  causes  except  disposal  and  power
related.  (Exhibits 3.2-10 through 3.2-18)
                     3-3

-------
     o    Over 80 percent of the material released in all in-plant
          event contexts (except the "other" category) was related
          to  events  that released  50,000  pounds  or  more  of
          material.  (Exhibits 3.2-19  through 3.2-23)


3.2.2  In-Transit Events

     o    In-transit  events  accounted  for 29.0  percent of  the
          events in the AHE/DB.  (Exhibit 2.1-7)

     o    Slightly  less  than  one-half  of  all in-transit  events
          (48.6  percent)  involved trucks;  another  43.3  percent
          involved rail cars.   (Exhibit 2.1-8)

     o    For events involving the release of hazardous materials
          during transportation,  46.3 percent were  the result of
          a leak, 11.6 percent were the  result of a collision, 6.9
          percent were the result  of  over-turned trucks  and 6.5
          percent were the result of derailments.  (Exhibit 2.1-9)

     o    Over one-half of the events  for  each transit mode type
          (except pipelines)  involved  releases of less than 1,000
          pounds of material.  The majority of events that occurred
          from pipelines involved the release of over 1,000 pounds
          of material.  (Exhibits 3.2-24 through 3.2-28)

     o    The quantity reported as released from events occurring
          at pipelines accounted for approximately one-half of the
          total quantity of material released in in-transit events.
          However, the bulk of this volume released was contributed
          by a few large pipeline spills; otherwise, pipelines were
          the  reported  transportation  mode  for  only  a  small
          fraction of release  events and quantities released.  The
          quantities reported  as released in rail and truck events
          accounted for 22.3 and 18.5 percent of the total quantity
          reported as released in in-transit events, respectively.
          (Exhibit 3.2-29)

     o    Events that involved releases of over 100,000 pounds of
          material  accounted   for  almost  all  of  the  quantity
          reported as  released during  pipeline and barge events.
          For rail events, incidents involving the release of over
          50,000 pounds of material accounted for over 85 percent
          of the material released.  (Exhibits 3.2-24 through 3.2-
          28)
                               3-4

-------
          Over 70 percent of the  in-transit  events that involved
          rail transport  involved  releases  of  less than  1,000
          pounds of material, over one-half released 100 pounds or
          less of material,  and 23.5 percent involved the release
          10 pounds or less of material.  (Exhibit 3.2-25)

          No event  that  involved "other" transit  modes (usually
          air transport) reported a release quantity of over 25,000
          pounds.   (Exhibit 3.2-28)

          For  all  in-transit  event  causes,  except  over-turned
          trucks, the  majority of  material  was released  during
          events that involved the release of over 100,000 pounds
          of material.  For  events in which an "over-turned truck"
          was identified as the cause of a release, approximately
          50 percent  of  the  total  volume of  the material  was
          released in volumes ranging  from 1,000 to 50,000 pounds.
          (Exhibits 3.2-30 through 3.2-33)
3.3  TYPES OF FACILITIES INVOLVED

     o    Facility type was identified  for about 91 percent of the
          events in the AHE/DB.  Assigned facility types range from
          primary producers to end-users, with 39.6 percent of the
          events occurring in  the  Chemicals  and Allied Products,
          and Petroleum Products sectors (referred  to in the AHE/DB
          as the "Primary Producers" sector).  Over 24 percent of
          the events  occurred in  the  distribution  sector,  13.9
          percent in the secondary producer  sector,  11.3 percent
          in end-user applications and 1.4 percent in waste-related
          activities.   In  9.2  percent  of the events insufficient
          information  was  provided  in the  reporting source  to
          identify facility type.  (Exhibit 2.1-10)

     o    Although most events  involving distributors  are  in-
          transit  events   (69.8  percent),  some of   these  events
          occurred at fixed facilities.  Similarly,  while a large
          majority of  events  involving  primary producers  (88.0
          percent), secondary producers (94.0 percent), end users
          (86.0  percent),   and  waste-related  facilities  (81.8
          percent) occur at fixed facilities, a small fraction are
          coded as in-transit events.

     o    Events  occurring   at  primary  production  facilities
          accounted for a greater percentage of reported deaths and
          injuries than  for  all  other industry sectors.   Events
          occurring at facilities  classified as  "distributors"
          accounted for the largest share of  evacuees.  (Exhibits
          2.1-11, 2.1-12)
                               3-5

-------
          While events occurring at secondary production facilities
          accounted  for  only  14  percent  of all  events,  they
          represent over 40 percent of the total reported quantity
          of material released  in the AHE/DB.   Distributors and
          end-users, on the other hand,  represent twice as large
          a percentage of  the number of  AHE/DB events as they do
          of total quantity released.   (Exhibits  D-l,  D-10, D-13,
          D-16)
3.4  CHEMICALS INVOLVED
          The ten chemicals most frequently reported in association
          with AHE/DB accidental release  events  were involved in
          43 percent of all reported events.  (Exhibits 3.4-1 and
          3.4-2)2

               Releases of sulfuric acid were reported for the most
               events:  1,045 or 9.6 percent of all of the events
               in  the  AHE/DB.    Five  deaths and  365  injuries
               occurred as a  result of these events.   (Exhibits
               3.4-1 and 3.4-2)

               Chlorine, ammonia (anhydrous)  and hydrochloric acid
               were involved, respectively,  in 750,  628  and 530
               events,  together representing 17  percent  of all
               events  in the  AHE/DB.   Seven deaths and 2,213
               injuries  occurred  as  a result  of these  events.
               (Exhibits 3.4-1 and 3.4-2)

               Six  other  chemicals   (sodium  hydroxide,   methyl
               chloride, phosphoric acid,  ethylene oxide,  toluene
               and vinyl chloride)  were involved in 16 percent of
               all events reported  in the AHE/DB.   Twenty deaths
               and  683  injuries occurred as a  result of these
               events.  (Exhibits 3.4-1 and 3.4-2)

          Sulfuric acid was the most frequently reported chemical
          associated  with events  at  all  facility types  except
          "waste  related."    Sulfuric  acid was  involved  in  11
          percent of the  events at  primary producers, 11 percent
          of the events at  secondary producers and  9 percent of
          the events at distributors.  (Exhibits D-9, D-12 and D-
          15)
     2Since any given  event  may involve the  release  of multiple
chemicals,  double counting  exists  when  summing  the  number  of
events,  deaths  or  injuries  reported  for  various  chemicals.
Therefore, all percentages or combined totals  of events, deaths or
injuries should be viewed as upper bound estimates.


                               3-6

-------
     o    After  sulfuric  acid,  the  next  four most  frequently
          reported chemicals at  primary  producers  were chlorine,
          ammonia (anhydrous),  methyl  chloride and ethylene oxide.
          These chemicals accounted for 1,195 events,  or 28 percent
          of the 4,324 events that occurred at primary producers.
          (Exhibits D-8 and D-9)

     o    For  events  that  occurred  at  secondary  production
          facilities, the five most frequently reported chemicals
          were  sulfuric  acid,  chlorine,  ammonia  (anhydrous),
          phosphoric acid  and  hydrochloric  acid.   Together these
          substances accounted  for 509 events, or 33 percent of the
          1,522  events  that   occurred   at  secondary  production
          facilities.  (Exhibit D-12)

     o    For events that occurred at distributors,  the four most
          frequently reported  chemicals,  in addition to sulfuric
          acid, were sodium hydroxide, hydrochloric acid, ammonia
          (anhydrous) and phosphoric acid. Together they accounted
          for 614 events, or 23 percent of the 2,689 events which
          occurred at distributors.   (Exhibit D-15)

     o    The average quantities of substances released per event
          are  shown  in  Exhibit  3.4-2  for  the  20  chemicals
          associated with the largest number of events.  Releases
          of sulfuric acid involved the largest quantities (88,000
          pounds) , followed by sodium hydroxide (83,000  pounds) and
          methyl alcohol  (52,000 pounds).

     o    When  release quantity  is  compared  to the  reportable
          quantities (RQs) for the chemicals  involved, 30.9 percent
          of the events  involved  the release  of substances  in
          quantities less than the RQ, i.e., these releases would
          not normally be reported to the National Response Center
          or most other jurisdictions.   (Exhibit D-5)

     o    Only 13.1 percent of the events involved the release of
          substances in  quantities  greater  than 100  times  their
          RQs.    (Exhibit D-5)
3.5  GEOGRAPHIC PATTERNS

     Geographic distributions  of all  AHE events,  death events,
injury events  and  evacuation events are  shown  in Exhibits 3.5-1
through 3.5-4.   It  is important to note three caveats regarding the
spatial patterns shown in these exhibits:
                               3-7

-------
          The existence of high numbers of incidents in some states
          may be a  result  of effective  enforcement^of  reporting
          requirements by authorities in those statesrj.ower numbers
          of incidents  in  other states  may  be a result  of less
          aggressive enforcement.

          The distribution of  events  by state also  reflects the
          geographic emphasis of the data collection process.  The
          data  collection  process  attempted to  minimize  such
          biases, but was  restricted  by the  need to  access more
          readily available data sources, such as those that had
          been computerized.  Exhibit 3.5-5 graphically depicts the
          coverage achieved in each state.  California, Texas, New
          York and New Jersey received the most scrutiny.

          We also  would expect the  geographic  distribution  of
          events  to  be  dependent   upon   the  distribution  of
          manufacturing activity across  states.  That  is, states
          with extensive manufacturing activity, and thus, chemical
          production and use,  are likely  to have higher numbers of
          incidents.  Exhibit 3.5-6 demonstrates  that California,
          Texas, New York,  Illinois and Michigan  all demonstrated
          total values of shipments for manufactured products that
          exceeded 120 billion dollars in 1985.
Exhibits 3.5-1 to 3.5-4 demonstrate the following:

     o    Texas was the  only state with more than  1,000 events.
          Ohio, New Jersey, New York, California and Louisiana also
          demonstrated a large number of incidents.  (Exhibit 3.5-
          1)

     o    New York, New Jersey and  Texas had the largest number of
          release events involving deaths.   (Exhibit 3.5-2)

     o    New  Jersey  and California had the  largest number  of
          incidents   involving   injuries,   followed   by  Texas,
          Louisiana, Illinois, Ohio and New York.  (Exhibit 3.5-3)

     o    The spatial  distribution of evacuation events was similar
          to that of injury events.  New Jersey leads in the number
          of events involving evacuations,  followed by California
          and Louisiana.  (Exhibit 3.5-4)
                               3-8

-------
For each of the spatial  distributions,  the majority of
states fall in  the categories representing  the lowest
frequency of events:  39 states had  250 or fewer total
events; 44  states  had four  or  fewer death  events;  41
states had 25 or fewer injury events and 43  states had
25 or fewer evacuation events.
                     3-9

-------
Report Exhibits

-------
                                Exhibit 1.2-1

          DATA SOURCES FOR THE AHE DATA BASE
National Data Sources
       National Response Center
       United Press International
       Dept. of Transportation Hazardous
       Material Information System
       New York Times
       PEI Associates
       Associated Press
       Marsh & McLennan
       Chemical Week
Regional Data Sources
       Region 7 EPA
       New Orleans Tlmes-Plcayunne
       Chicago Tribune
       Los Angeles Times
       Boston Globe
State Data Sources
Number of
 Records
   7,705
     236
     168

      90
      80
      43
      46
       2

   8.367
     504
     102
      62
      35
      12

     715
 Period
1982-87
1980-85
1986-87

1965-87
1981-86
1980-86
1956-84
1985-86
1979-85
1980-86
1980-86
1980-86
1985-87
       New Jersey Department of
         Environmental Protection
       New York Attorney General
       Texas Air Control Board
       Ohio EPA
       California:
         Office of Emergency Services
         Highway Patrol
       Additional Local Newspapers
     513
     433
     428
     252

     133
      61
      24
Other
   1.844

       4
1980-87
1982-85
1980-86
1981-85

1984-85
1983-85
1980-86
        TOTAL
  10,933  *
    National Response Center (NRC) data included 7,705 accidental
   releases of Comprehensive Environmental Response. Compensation
   and Liability Act (CERCLA) chemicals during 1982 through 1986.
   Of these, the 2,435 events designated as air releases or having
   reports of deaths, injuries or evacuations were all included in
   the AHE Data Base.  One-tenth of the remaining 5,270 events
   events were randomly selected for inclusion In the AHE Data Base.
   As a result, the 2.962 NRC-origin records In the AHE Data Base
   represent 7,705 events: and the 6,190 total records in the AHE
   Data Base represent 10,933 incidents.
Source:  Industrial Economics, Inc.

-------
                                                Exhibit 1.2-2

                       DISTRIBUTION OF EVENTS BY SOURCE AND YEAR
SOURCE
National Response Center
EPA Region 7
United Press International
DOTs Haz. Mat'l Inf. System
PEI Associates
Associated Press
Marsh & McLennan
Chemical Week
Newspapers:
   New Orleans Times Picayunne
   New York Times
   Chicago Tribune
   Los Angeles Times
   Boston Globe
   Additional Local Newspapers
States:
   New Jersey Dept. of
      Environmental Protection
   New York Attorney General
   Texas Air Control Board
   Ohio EPA
   California Office of
      Emergency Services
   California Highway Patrol

Other
TOTAL
Before
1980 or
Unknown
15
1
2
0
0
0
33
0
0
5
0
0
0
3
0
0
0
0
0
0
0

1980
0
1
3
0
0
19
4
0
26
9
4
4
0
0
66
0
19
0
0
0
0

1981
0
99
32
0
1
9
1
0
30
13
5
8
0
2
75
0
64
1
0
0
3

1982
1,065
141
38
0
3
3
3
0
10
12
4
7
0
0
80
2
63
70
0
0
0

1983
1.532
106
71
0
3
8
3
0
13
3
9
2
0
2
90
91
56
58
0
32
0

1984
1.860
106
58
0
1
3
2
0
6
1
13
1
0
2
57
149
122
100
111
26
1

1985
1.499
50
32
0
30
0
0
1
5
34
13
6
1
14
102
191
103
23
22
3
0


1986 1987
1.730
0
0
167
42
1
0
1
12
12
14
7
3
1
42
0
1
0
0
0
0
4
0
0
1
0
0
0
0
0
1
0
0
8
0
1
0
0
0
0
0
0

Total
7.705
504
236
168
80
43
46
2
102
90
62
35
12
24
513
433
428
252
133
61
4
59
155   343    1,501   2,079    2,619   2,129   2,033
15   10.933
Source: Industrial Economics, Inc.

-------
                          Exhibit 1.2-3
 AHE/DB's Three  Interconnecting Data Bases
             NATIONAL SOURCES OF EVENTS DATA
      National Response Center     New York Times
          DOTs HMIS        PEI Associates
        Marsh & McLennan       Chemical Week
                                 United Press International
                                    Associated Press
             REGIONAL SOURCES OF EVENTS DATA
         Region 7 EPA    New Orleans Times - Pfcayunne    Boston Globe
        Los Angeles Times        Chicago Times        Other Local Papers
               STATE SOURCES OF EVENTS DATA
   New Jersey DEQ
New York Attorney Genera.
                       Texas Air Control Board
                           Ohio EPA
                        EVENTS
                     DATA   BASE
      CHEMICALS
      DATA BASE
                                    COMPANY-
                                    FACILITIES
                                    DATA BASE
            t
                                         t
     EPA / NIOSH Sources
          RTECS
          NFPA
                                     Event Reports
                                   Dun and Bradstreet
Source: Industrial Economics, Inc.

-------
                            Exhibit 2.1-1
                 Occurrence of Injuries
                   DEATH/INJURY EVENTS
                          No Injuries  Unknown
                           1.9%     0.5%
                                          Injuries
                                           97.6%
                         Number of Events » 1,019
                         ALL EVENTS
Unknown
                                   Injuries
                                    9.1%
                 No Injuries
                   8Z7%
                        Number of Events = 10,933
Source: Industrial Economics, Inc.

-------
                            Exhibit 2.1-2

                 Occurrence of Deaths

                  DEATH/INJURY EVENTS
                           Unknown    Deaths
                            5.4%      n.2%
                 No Deaths
                   83.4%
                        Number of Events = 1,019
                         ALL EVENTS

                          Unknown Deaths
                           5.9%   1.0%
                                         No Deaths
                                           93.1%
                        Number of Events = 10,933
Source: Industrial Economics, Inc.

-------
                                 Exhibit 2.1-3

               In-Plant  Events  By  Location


                      DEATH/INJURY EVENTS
                         Unknown
                          17.6%
 Process Vessel
    22.1%
    Vehicle not In Transit  4.2%

       Heating/Cooling  1.7%

             Disposal  3.0%
                    Valve/Pipe
                      15.8%
   Storage Vessel
      29.9%
                             Number of Events » 638
                             ALL EVENTS
                              Unknown
                                9.1%
                    Other 3.5%

       Vehicle not In Transit  2.8%
         Heating/Cooling 1.9%

            Disposal  6.8%
Process Vessel
   22.4%
                  Valve/Pipe
                    25.4%
    Storage Vessel
      28.1%
                              Number of Events = 7,760
Source: Industrial Economics, Inc.

-------
                                   Exhibit 2.1-4
                  In-Plant  Events  By Cause
                       DEATH/INJURY  EVENTS
                 Other 2.0%
        Sparking/Lightning 2.2%
               Power  1.1%
     High Pressure/Temp  2.7%

       Arson/Vandalism  2.2%
                Fire/Explosion
                   19.1%
                           Unknown
                            15.1%
Equipment Failure
    30.9%
                   Upset Conditions  5.6%
                               Number of Events = 638
                                                 Operator Error
                                                    19.1%
                  Other 7.5%
       Sparking/Lightning  0.3%
               Power 1.7%
     High Pressure/Temp  2.0%
             Disposal  0.4%
       Arson/Vandalism  0.9%

         Fire/Explosion  5.2%
             Upset Conditions
                 4.8%
                               ALL EVENTS
                             Unknown
                              1Z2%
   Equipment Failure
       46.9%
                     Operator Error
                        18.2%
                               Number of Events = 7,760
Source: Industrial Economics, Inc.

-------
                               Exhibit 2.1-5

          In-Plant Events  By  Event  Context

                     DEATH/INJURY  EVENTS
             Unknown 32.0%
                                          Loading/Unloading
                                             16.5%
               Other 0.8%

            Maintenance 6.3%

                Shut Down 0.8%
                    Start-Up 19%

                            Number of Events = 638
Steady State
  39.8%
                            ALL EVENTS
                Unknown 25.0%
                                          Loading/Unloading
                                             16.2%
             Other 1.8%

         Maintenance  3.1%
           Shut Down 1.2%
              Start-Up 1.8%
                                             Steady State
                                               51.0%
                           Number of Events = 7,760
Source: Industrial Economics, Inc.

-------
(D
UJ
                             Exhibit 2.1-6
           Frequency of  Reported  End  Effects
             ALL EVENTS, DEATH/INJURY EVENTS
                   AND EVACUATION EVENTS
                 Vapor Release
Fire
Explosion
Other
                          End Effect
                             All Events
                             Death/Injury
                             Evacuation
   Note: Summation across end effects may exceed 100% as each event may exhibit more than one end effect
   Source: Industrial Economics, Inc.

-------
                           Exhibit 2.1-7

            In-Plant vs.  In-Transit Events


                  DEATH/INJURY  EVENTS
               In-Transit
                37.4% /:•:
                                           In-Plant
                                            62.6%
                        Number of Events = 1,019
                         ALL EVENTS
                 In-Transit
                 29.0%
                                          In-Plant
                                          71.0%
                         Number of Events = 10,933
Source: Industrial Economics, Inc.

-------
                            Exhibit 2.1-8
              In-Transit Events  By Mode
                  DEATH/INJURY EVENTS
                     Barge 2.6%
                           PipeVne
                            2.4% 1.3% other
                         Number of Events = 381
                         ALL EVENTS
                    Barge 4.5%
                           Pipeline
                            2.8% 0.7% Other
                                            Truck
                                            66.1%
                                             Truck
                                             48.6%
                        Number of Events = 3,173
Source: Industrial Economics, Inc.

-------
                              Exhibit 2.1-9
              In-Transit Events  By  Cause
                    DEATH/INJURY EVENTS
                              Unknown
                               4.5%
                Other  14.7%
       Truck Turnover  5.5%
         Derailment  6.0%
                 Collision 22.8%
                                                  Leak
                                                  46.5%
                           Number of Events - 381
                           ALL EVENTS
                             Unknown
                              5.6%
             Other 23.1%
        Truck Turnover 6.9%
               Derailment 6.5%
                     Collision 11.6%
                           Number of Events -3,173
                                                  Leak
                                                  46.3%
Source: Industrial Economics, Inc.

-------
                               Exhibit 2.1-10
       Distribution  of  Events  By  Facility  Type
                     DEATH/INJURY EVENTS
                         Unknown
                          15.4%
               Waste Related
                  1.3%
               End User
                13.1%
                                              Primary Producer
                                                 26.2%
                  Distributor 31.1%

                           Number of Events = 1,019
                                                  Secondary Producer
                                                      13.0%
                            ALL EVENTS
                             Unknown
                              9.2%
               Waste Related  1.4%
                  End User
                   11.3%
                 Distributor
                  24.6%
                                                   Primary Producer
                                                      39.6%
                                         13.9% Secondary Producer
                           Number of Events = 10,933
Source: Industrial Economics, Inc.

-------
                            Exhibit 2.1-11
  Reported  Number  of  Deaths By Facility  Type
              Unknown
              20.5%
Waste Related
   2.4%
     End User
      17.4%
                      Primary Producer
                         43.4%
                  Distributor
                    7.6%
Secondary Producer
     8.7%
                         Number of Deaths = 288
 Source: Industrial Economics, Inc.

-------
                              Exhibit 2.1-12
   Reported  Number of  Injuries By Facility Type
                 Unknown
                  16.5%
Waste Related 0.7%
    End User
      8.6%
           Distributor  x£
            25.3%
     Primary Producer
        31.0%
Secondary Producer
    18.0%
                        Number of Injuries = 10,803
 Source: Industrial Economics, Inc.

-------
                              Exhibit 2.1-13
      Frequency  of  Events  Reporting  Deaths
                       By Facility Type
     2.0%-
 2   1.5% -
 I
 in
     1.0%-
     0.5% -
     0.0%
            Number of Events with Typed Facilities = 9,928
            Total Number of Events = 10,933
           Primary
           Producer
Secondary
Producer
Distributor   End User
                              Facility Type
Waste
Related
All Types
Source: Industrial Economics, Inc.

-------
                              Exhibit 2.1-14
      Frequency  of  Events  Reporting Injuries
                       By  Facility  Type
    14.0%
    12.0% -
    10.0% -
     0.0%
            Number of Events with Typed Facilities = 9,928
            Total Number of Events = 10,933
           Primary
           Producer
Secondary
Producer
Distributor   End User
                              Facility Type
Waste
Related
All Types
Source: Industrial Economics, Inc.

-------
                           Exhibit 2.1-15
        Frequency  of Reported End  Effects
     By  Facility Type for Deattylnjury  Events
       Primary
       Producer
Secondary
Producer
Distributor
End User
Waste
Related
                       Facility Type
                                                       Spill
                                                       Vapor Release

                                                       Fire
                                                       Explosion

                                                       Other
Total
Source: Industrial Economics, Inc.

-------
                         Exhibit 2.1-16
        Distribution of  Events by Chemical
               for Death/lnjury Events

              10  CHEMICALS REPRESENTING
           THE LARGEST NUMBER OF EVENTS
       150-1
                         Substance Name
Note: Up to three substances may be reported as released during any one event in the AHE/DB. Thus,
double counting will result from summing the number of release events across substances.
Source: Industrial Economics, Inc.

-------
RANK
                                             Exhibit 2.1-17

      SUBSTANCES MOST FREQUENTLY REPORTED IN DEATH/INJURY EVENTS *
Notes
        SUBSTANCE NAME'
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
CHLORINE
AMMONIA (ANHYDROUS)
HYDROCHLORIC ACID
SULFURIC ACID
AMMONIA
SODIUM HYDROXIDE
NITRIC ACID
TOLUENE
STYRENE
PHOSPHORIC ACID
TOLUENE DIISOCYANATE
BENZENE
PHOSGENE
SULFUR DIOXIDE
HYDROFLUORIC ACID
ACRYLONITRILE
METHYLENE CHLORIDE
HYDROGEN SULFIDE
ACETIC ACID
METHYL ALCOHOL
METHYL ETHYL KETONE
121
66
64
62
26
26
18
16
14
12
12
11
10
10
9
9
9
9
8
8
a
»OF
DEATH
AND
INJURY
(0/1)
EVENTS
121
66
64
62
26
26
18
16
14
12
12
11
10
10
9
9
9
9
8
8
a


fOF
DEATH
EVENTS
4
1
1
4
• 2
1
4
7
1
0
2
2
2
0
2
1
1
2
0
3
4



#OF
DEATHS
5
1
1
5
6
1
21
13
0
0
3
16
2
0
3
1
2
3
0
9
6


#OF
INJURY
EVENTS
121
65
64
61
26
26
18
15
13
12
11
11 .
10
10
8
9
8
9
8
8
7



#OF
INJURIES
1,054
248
911
365
308
39
361
70
32
369
112
436
303
55
664
63
481
157
41
29
101
TOTAL
QUANTITY
RELEASED
FROMD/I
EVENTS
(U>«.)
47.028
414,346
296,650
1.690.977
57,633
159,094
149,266
27.563
52,601
2.166
16,146
7.800
6,800
1,277
270.562
900.000
4,116
38
19,536
104.250
90
AVERAGE
QUANTITY
RELEASED
FROM 0/1
EVENTS
(Lbr) "•
644
9.636
10,595
51,242
5,763
10,606
16.585
9.188
8,767
361
3.229
3.900
3.400
426
90,187
225.000
1,372
38
3,907
34,750
90

LARGEST
QUANTITY
RELEASED
(Lbs.)
10.000
160.000
45.375
1.500.000
48.750
120,000
135.000
22.500
45.000
1.875
15.000
7.500
5,000
1.125
270.000
400.000
2.053
38
17,000
55.500
90
       Includes all events In which the substance appears as the primary, secondary or tertiary substance.
       All substances are commodity chemicals.
       Average quantities are computed using only those events that reported quantities.
Source: Industrial Economics, Inc.

-------
                                               Exhibit 2.1-18
                       CHEMICAL PROPERTY CHARACTERIZATION METRICS
       Chemical Property        Basis for Determination
                                              Rating Scale
       Acute Toxicity
         (LRATE)
NIOSH IDLH (level of concern)        0	1	2	3	4
or CEPP Level of Concern          >500     50-500     5-50      0.5-5    0-0.5
                              LOW                                    HIGH
       Flammability
        (FRATE)
NFPA Rating
Flash Point
Boiling Point
Qualitative Information
  0...
LOW
                            ..4
                            HIGH
       Reactivity
       (RRATE)
NFPA Rating
Qualitative Information
  0...
LOW
.1	2.	3.,
	4
   HIGH
       Volatility
       (VRATE)
Vapor Pressure measured
in mm Hg
  0...
  <1
LOW
          .1.
        1-100
	2
 >100
  HIGH
       Physical State
Solid/Liquid/Gas (as
reported for each event)
       Processing Mode
        (COMMSPEC)
lEc's assessment of whether
substance is Commodity (high
production volume) or
Specialty (low production
volume)
Source: Industrial Economics, Inc.

-------
                                Exhibit 2.1-19

            FREQUENCY DISTRIBUTION OF AHE/DB CHEMICALS
                    BY CHEMICAL PROPERTY RATINGS
LRATE
FRATE
RRATE
VRATE
COMMSPEC
0
1
2
3
4
U
Total
0
1
2
3
4
U
Total
0
1
2
3
4
U
Total
0
1
2
U
Total
C
S
U
0
4
37
23
42
116
222
54
29
23
41
33
42
222
73
18
23
8
2
98
222
35
42
39
106
222
128
59
35
           Total    222
                 Death and Injury
                    Chemicals
                  Number  Percent
  0.0%
  1.8%
 16.7%
 10.4%
 18.9%
 52.3%

100.0%

 24.3%
 13.1%
 10.4%
 18.5%
 14.9%
 18.9%

100.0%

 32.9%
  8.1%
 10.4%
  3.6%
  0.9%
 44.1%

100.0%

 15.8%
 18.9%
 17.6%
 47.7%

100.0%

 57.7%
 26.6%
 15.8%

100.0%
LRATE
FRATE
RRATE
VRATE
Total

   0
   1
   2
   3
   4
   U

Total
COMMSPEC
Total

   0
   1
   2
   U

Total

   C
   S
   U
                                   All Events
                                   Chemicals
                                Number  Percent
0
1
2
3
4
U
0
14
66
45
66
460
0
2.2%
10.1%
6.9%
10.1%
70.7%
651

115
 82
 49
111
 65
229

651
           Total
651

 70
 77
 69
435

651

267
229
155

651
100.0%

 17.7%
 12.6%
  7.5%
 17.1%
 10.0%
 35.2%

100.0%
0
1
2
3
4
U
147
42
46
20
6
390
22.6%
6.5%
7.1%
3.1%
0.9%
59.9%
100.0%

 10.8%
 11.8%
 10.6%
 66.8%

100.0%

 41.0%
 35.2%
 23.8%

100.0%
Source: Industrial Economics, Inc.

-------
                                   Exhibit 2.1-20

                 FREQUENCY DISTRIBUTION OF AHE/DB EVENTS
     BY CHEMICAL PROPERTY RATINGS OF PRIMARY SUBSTANCE RELEASED
LRATE
FRATE
RRATE
VRATE
   0
   1
   2
   3
   4
   U

Total

   0
   1
   2
   3
   4
   U

Total
COMMSPEC
Total

   0
   1
   2
   U

Total

   C
   S
   U

Total
                       All Events
                   Number    Percent
    0
   96
 1820
  782
 4484
 3751

10933

 4408
 1154
  364
 1686
 1433
 1888

10933
10933

 2079
 1860
 1557
 5437

10933

 8367
  732
 1834

10933
  0.0%
  0.9%
 16.6%
  7.2%
 41.0%
 34.3%

100.0%

 40.3%
 10.6%
  3.3%
 15.4%
 13.1%
 17.3%

100.0%
0
1
2
3
4
U
4534
873
1622
281
61
3562
41.5%
8.0%
14.8%
2.6%
0.6%
32.6%
100.0%

 19.0%
 17.0%
 14.2%
 49.7%

100.0%

 76.5%
  6.7%
 16.8%

100.0%
LRATE
FRATE
                                  RRATE
VRATE
   0
   1
   2
   3
   4
   U

Total
                                                       Total
COMMSPEC
Total

   0
   1
   2
   U

Total

   C
   S
   U

Total
                                                      Death and Injury
                                                          Events
                                                     Number   Percent
   0
  11
 128
  51
 452
 377

1019
                   1019
1019

 150
 143
 111
 615

1019

 695
  80
 244

1019
  0.0%
  1.1%
 12.6%
  5.0%
 44.4%
 37.0%

100.0%
0
1
2
3
4
U
371
149
46
123
86
244
36.4%
14.6%
4.5%
12.1%
8.4%
23.9%
                   100.0%
0
1
2
3
4
U
433
71
121
16
4
374
42.5%
7.0%
11.9%
1.6%
0.4%
36.7%
100.0%

 14.70/0
 14.0%
 10.9%
 60.4%

100.0%

 68.2%
  7.9%
 23.9%

100.0%
Source: Industrial Economics, Inc.

-------
                        Exhibit 2.1-21
       Role of  Fire/Explosion  vs. Toxicity as
      Apparent Cause of  Deaths or  Injuries
                                                      Fire/
                                                      Explosion

                                                      Toxicity
              % of Injuries               % of Deaths

                      Injuries or Deaths
Source: Industrial Economics, Inc.

-------
  1000
                            Exhibit 2.1-22

            Event Severity  By Toxicity  Rating
          TOXICITY AS CAUSE OF DEATH  OR INJURY
   100 ^
'c*
   10-
»           i


•           *
           «
           •
           •
•           •
•           •
           *

           •
                     1          2          3


                            Toxicity Rating
   Note: See Exhibit 2.1-18 for explanation of rating metrics.
   Source: Industrial Economics, Inc.

-------
                           Exhibit 2.1-23

         Event Severity By  Flammability  Rating
      FIRE/EXPLOSION AS CAUSE OF DEATH OR INJURY
  1000
   100-
(0

.2
*!Z


"£*
CO

I

-------
                            Exhibit 2.1-24
          Event Severity By  Reactivity  Rating
      FIRE/EXPLOSION AS CAUSE OF DEATH OR INJURY
  1000
   100-
(0
O
+
CO
0>
Q
   10-
                          Reactivity Rating
   Note: See Exhibit 2.1-18 for explanation of rating metrics.
   Source: Industrial Economics, Inc.

-------
                            Exhibit 2.2-1
                 Occurrence of Injuries

                    EVACUATION  EVENTS
                            Unknown
                             6.0%
                 No Injuries
                   54.2%
                         Number of Events = 829
                         ALL EVENTS
                          Unknown
                           8.2%
Injuries
 9.1%
                 No Injuries
                   82.7%
                                             Injuries
                                              39.8%
                         Number of Events = 10,933
Source: Industrial Economics, Inc.

-------
                            Exhibit 2.2-2

                 Occurrence of Deaths

                    EVACUATION EVENTS
                           Unknown Deaths
                            6.6%   4.8%
                 No Deaths
                   88.5%
                         Number of Events = 829
                         ALL EVENTS

                          Unknown Deaths
                           5.9%   1.0%
                                         No Deaths
                                           93.1%
                        Number of Events = 10,933
Source: Industrial Economics, Inc.

-------
                           Exhibit 2.2-3

            In-Plant vs. In-Transit Events

                   EVACUATION  EVENTS
               In-Transit
               34.4% /:;:
                 In-Transit
                 29.0%
                                           In-Plant
                                           65.6%
                         Number of Events =* 829
                         ALL EVENTS
                                          In-Plant
                                          71.0%
                        Number of Events = 10,933
Source: Industrial Economics, Inc.

-------
I
LU
                             Exhibit 2.2-4
           Frequency  of Reported End  Effects
            ALL EVENTS AND EVACUATION EVENTS
            All Events
            Evacuation
                                    Explosion
Other
                          End Effect
   Note: Summation across end effects may exceed 100% as each event may exhibit more than one end effect
   Source: Industrial Economics, Inc.

-------
                                 Exhibit 2.2-5
                In-Plant Events  By  Location
                       EVACUATION  EVENTS
                           Unknown
                            14.2%
Process Vessel
   19.5%
                     Other
                     4.8%

         Vehicle not in Transit
              5.7%

        Heating/Cooling 1.7%

             Disposal 4.8%
                     Valve/Pipe Ni
                      14.3%
      Storage Vessel
         35.1%
                              Number of Events - 544
                              ALL  EVENTS
                               Unknown
                                9.1%
                     Other 3.5%
        Vehicle not in Transit  2.8%
         Heating/Cooling  1.9%

             Disposal 6.8%
  Process Vessel
     22.4%
                   Valve/Pipe
                    25.4%
     Storage Vessel
        28.1%
                              Number of Events = 7,760
Source: Industrial Economics, Inc.

-------
                                   Exhibit 2.2-6
                  In-Plant  Events  By  Cause
                         EVACUATION EVENTS
                              Unknown
                               11.0%
                    Other 3.1%
          Sparking/Lightning  1.5%
                Power  1.3%
      High Pressure/Temp 2.4%
             Disposal 0.2%
        Arson/Vandalism 3.3%
    Equipment Failure
        33.3%
                Fire/Explosion
                   21.5%
                   Upset Conditions  4.0%
Operator Error
   18.4%
                               Number of Events = 544
                               ALL  EVENTS
                             Unknown
                              12.2%
                  Other 7.5%
       Sparking/Lightning  0.3%
               Power 1.7%
     High Pressure/Temp  2.0%
             Disposal  0.4%
       Arson/Vandalism  0.9%

         Fire/Explosion  5.2%
             Upset Conditions
                 4.8%
        Equipment Failure
            46.9%
                     Operator Error
                        18.2%
                               Number of Events = 7,760
Source: Industrial Economics, Inc.

-------
                               Exhibit 2.2-7
          In-Plant  Events By  Event Context
                      EVACUATION  EVENTS
                                    Loading/Unloading
                                        9.6%
               Unknown 23.4%
             Other 0.9%

         Maintenance 5.2%

           Shut Down 0.7%

             Start-Up 4.4%
                                             Steady State
                                               55.9%
                           Number of Events » 544
                            ALL EVENTS
                Unknown 25.0%
                                         Loading/Unloading
                                             16.2%
             Other 1.8%

         Maintenance 3.1%
           Shut Down  1.2%
              Start-Up 1.8%
                                            Steady State
                                              51.0%
                           Number of Events = 7,760
Source: Industrial Economics, Inc.

-------
                             Exhibit 2.2-8
              In-Plant Events  By  Location
      AMOUNT RELEASED  IN EVACUATION  EVENTS
                          Other 0.8% Process Vessel 3.0M>

                                        Storage Vessel 9.9%

                                            Valve/Pipe
                                             3.3%
              Disposal 83.0%
                      Total Amount » 20.0 Million Pounds
            AMOUNT RELEASED IN ALL EVENTS

                              Unknown
               Vehicle not in Transit 7.1% 0<9% Process Vessel 5.2%
              Heating/Cooling  0.1%
               Disposal  5.9%

               Valve/Pipe
                 5.5%
                                          Storage Vessel
                                            75.2%
                       Total Amount = 346.2 Million Pounds
Source: Industrial Economics, Inc.

-------
                              Exhibit 2.2-9

                In-Plant Events By  Cause


      AMOUNT RELEASED IN  EVACUATION  EVENTS
                                Unknown

                        Other 1.1%  O-4** 7.3%  Equipment Failure

                                           4.4% Operator Error

                                              Z8%  Upset Conditions
                Fire/Explosion
                   84.0%
                       Total Amount =• 20.0 Million Pounds
            AMOUNT RELEASED IN  ALL EVENTS
                        Other 5.3%
                                Unknown
                                 1.8%
           Sparking/Lightning  6.8%

                Power  0.2%
       High Pressure/Temp 0.9%
        Arson/Vandalism 0.5% .

         Fire/Explosion 5.7%


      Upset Conditions 0.8%
Equipment Failure
   32.0%
                  Operator Error
                     46.0%
                       Total Amount = 346.2 Million Pounds
Source: Industrial Economics, Inc.

-------
                           Exhibit 2.2-10
              In-Transit  Events By  Mode
                    EVACUATION EVENTS
                            Pipeline
                             1.1%
                      Barge 3.2%   O-4** Other
                         Number of Events = 285
                         ALL EVENTS
                     Barge 4.5%
                            Plpeine
                            2.8% 0.7% Other
                                              Truck
                                              50.2%
                                              Truck
                                              48.6%
                        Number of Events = 3,173
Source: Industrial Economics, Inc.

-------
                             Exhibit 2.2-11
              In-Transit Events By Cause
                     EVACUATION  EVENTS
                              Unknown
                               3.9%
                     Other 8.1%
          Truck Turnover  6.7%
          Derailment 24.9%
                                         Collision  21.4%

                           Number of Events - 285
             Other  23.1%
        Truck Turnover 8.9%
                           ALL EVENTS
                             Unknown
                              5.8%
               Derailment 8.5%
                     Collision  11.6%
                          Number of Events -3,173
                                                  Leak
                                                 46.3%
Source: Industrial Economics, Inc.

-------
                        Exhibit 2.2-12
            In-Transit Events By Mode
     AMOUNT RELEASED IN EVACUATION EVENTS
                   Barge
                   16.5%
                   Total Amount a 9.3 MilDon Pounds
          AMOUNT RELEASED IN ALL EVENTS
             Pipeline
             48.3%
                                10.9% Barge

                   Total Amount = 53.4 Million Pounds
Source: Industrial Economics, Inc.

-------
                               Exhibit 2.2-13
       Distribution  of Events  By Facility  Type
                      EVACUATION EVENTS
                      Unknown
                      21.0%
            Waste Related
                1.1%
                End User
                 14.5%
                                               Primary Producer
                                                  25.0%
                  Distributor 25.9%

                            Number of Events = 829
                                                  Secondary Producer
                                                       12.6%
                            ALL EVENTS
                             Unknown
                              9.2%
               Waste Related 1.4%
                  End User
                   11.3%
                 Distributor
                  24.6%
                                                   Primary Producer
                                                      39.6%
                                         13.9% Secondary Producer
                           Number of Events = 10,933
Source: Industrial Economics, Inc.

-------
                              Exhibit 2.2-14
   Frequency  of  Events  Reporting  Evacuations
                       By Facility Type
 i
 §
 UJ
    12.0%
    10.0% -
     8.0%-
6.0%-
     4.0%-
     2.0%-
     0.0%
            Number of Events with Typed Facilities = 9,928
            Total Number of Events =• 10,933
           Primary
           Producer
              Secondary
              Producer
Distributor   End User
                             Facility Type
Waste
Related
All Types
Source: Industrial Economics, Inc.

-------
                            Exhibit 2.2-15
     Percentage of  Evacuees  by Facility Type
               Residents versus  Workers
                            Number of Events - 220
                            Total Evacuated - 259,030
  10% H
  o%
                                                         Residents
                                                         Workers
       Primary
       Producer
Secondary
Producer
Distributor
End User
Waste
Related
All Types
                        Facility Type
Source: Industrial Economics, Inc.

-------
                                        Exhibit 2.2-16

     SUBSTANCES MOST FREQUENTLY REPORTED IN EVACUATION EVENTS
BANK   SUBSTANCE NAME *
 1    CHLORINE
 2    HYDROCHLORIC ACID
 3    AMMONIA (ANHYDROUS)
 4    SULFURIC ACID
 5    AMMONIA
 6    NITRIC ACID
 7    VINYL CHLORIDE
 8    SODIUM HYDROXIDE
 9    HYDROGEN SULFIDE
10    STYRENE
11    TOLUENE
12    PHOSPHORIC ACID
13    TOLUENE DIISOCYANATE
14    PHOSGENE
15    METHYL ALCOHOL
16    ETHYLENE OXIDE
17    SULFUR DIOXIDE
18    POLYVINYL CHLORIDE
19    BENZENE

»OF
EVACUATION
EVENTS "
89
76
67
30
21
20
14
12
12
12
10
10
10
9
9
9
8
8
8

#OF
EVACUATIONS
(ALL SUBS)
38.037
47.243
12.126
14,055
5.158
17.124
17,686
5.070
3.022
17.250
2,950
18.760
4.958
1.109
3.275
6.225
630
18.250
5.050
TOTAL
QUANTITY
RELEASED
(Lb«)
52,777
1,508.389
415,086
1,641.870
183.612
152.776
150,696
40.723
1.669
157,238
46.875
300,563
1,058
100
1.950.000
225.026
0
0
19.500
AVERAGE
QUANTITY
RELEASED
(Lb«) —
1,199
53.871
11,860
126,298
20.401
19.097
37.674
8.145
334
26.206
15.625
100,188
529
50
975.000
56.257
0
0
9,750
LARGEST
QUANTITY
RELEASED
(Lbt)
20,000
1,125.000
160.000
1.500.000
112.500
135.000
150.000
30.000
1.000
82.500
30.000
300.000
938
90
1.875.000
202.500
0
0
12,000
Notes
     All substances are commodity chemicals.
     This Includes all events In which the substance appears as a primary, secondary, or tertiary substance.
     Average quantities are computed using only those events that reported quantities.
Source: Industrial Economics. Inc.

-------
                             Exhibit 2.3-1

AIR RELEASES OF TOXIC CHEMICALS LEADING TO TWO OR MORE DEATHS
                             (1982-1986)
DATE
04-Apr-«6
12-Jun-86
17-Jan-8e
OI-May-86
27-May-8e
OS-Jun-86
os-Jui-ae
25-Jun-85
20-May-85
20-Oec-85
18-Apr-85
26-Apr-85
29-May-8S
19-Jan-85
06-Feb-85
03-Mar-e5
04-Apr-85
18-Jul-85
05-Dac-85
: SUBSTANCES RELEASED
SUBSTANCE1 SUBSTANCE2
GUNPOWDER AND/OR FIREWORK
TOLUENE NITROETHANE
AMMONIUM DICHROMATE
ACET ALDEHYDE
POLYVINYL CHLORIDE TOLUENE DIISOCYANATE
BUTADIENE FORMALDEHYDE
TRICHLOROETHYLENE
GUNPOWDER AND/OR FIREWORK
GUNPOWDER AND/OR FIREWORK
XYLENE TOLUENE
TOLUENE
POTASSIUM SODIUM NITRATE
AMMONIA
SULFUR
NITROGLYCERINE
HYDROFLUORIC ACID
ACETYLENE
METHYL ALCOHOL VARNISH
HYDROGEN NAPHTHA
END EFFECTS
SUBSTANCES 1ST 2ND 3RD 4TH
EX
AMMONIA Fl
EX
Fl
METHYLENE CHLORIDE Fl
SP
SP
EX
Fl
METHYL ETHYL KETONE VR
EX
EX
EX
Fl
EX
SP
Fl
Fl
EX
Fl
EX
Fl
SP
EX
EX
VR
Fl
EX
Fl
Fl
UU
UU
EX
UU
VR
EX
EX
Fl
UU
SP
VR
EX
VR
Fl
UU
UU
UU
EX
UU
UU
UU
SP
UU
UU
UU
UU
VR
UU
UU
UU
VR
UU
VR
UU
UU
UU
UU
UU
UU
UU
VR
UU
UU
UU
UU
UU
NUMBER
OF
INJURIES
0
4
18
2
0
6
25
6
0
7
1
17
10
7
0
0
0
1
45
NUMBER
OF
DEATHS
8
3
2
2
2
2
2
21
e
3
3
3
3
2
2
2
2
2
2
NUMBER
OF
EVACUEES
0
0
0
700
0
2.000
0
0
0
0
0
0
0
0
0
0
0
100
0

-------
                                                                 Exhibit 2.3-1
                                                                 (Continued)
                        AIR RELEASES OF TOXIC CHEMICALS LEADING TO TWO OR MORE DEATHS
                                                                 (1882-1964)

DATE
11-May-84
16-Aug-84
10-Aug-«4
22-Aug-M
13-Oec-M
27-May-83
04-Dec-83
25-Jan~S3
28-Aug-83
ie-Nov-83
03-Oct-83
14-Jun-«3
21-Jul-«3
18-May-82
02-Jun-«2
13-Jan-82
14-Jan-82
<^8TANCE8BajEA8f33
SUB8TANCE1 SUBSTANCE2
PLASTIC POLYSTYRENE
ACETONE NITRIC AGIO
COAL TAR LIGHT OIL
SPECIALTY OILS
SULFURIC ACID
EXPLOSIVES
PETROLEUM PRODUCTS INSECTICIDE
EXPLOSIVES
SODIUM NITRATE
EXPLOSIVES
METHYL BROMIDE MALATHION
METHYL ISOBUTYL KETONE 2-NITROPROPANE
METHANE CHLORINE
NITROGEN
NITRIC ACID
HYDROGEN SULFIDE
UNKNOWN
'-:< 1
SUBSTANCES 1ST
URETHANE Fl
EX
EX
Fl
EX
EX
ANTIFREEZE Fl
EX
EX
Fl
EX
VR
EX
VR
EX
VR
EX
ENDEf
2ND
UU
Fl
Fl
UU
Fl
Fl
EX
Fl
Fl
EX
Fl
UU
Fl
UU
Fl
UU
Fl
•fECT!
3RD
UU
VR
UU
UU
UU
UU
VR
UU
UU
UU
VR
UU
SP
UU
VR
UU
UU
if:
4TH
UU
UU
UU
UU
UU
UU
UU
UU
UU
UU
UU
UU
VR
UU
SP
UU
UU
NUMBER
OF
INJURIES
0
1
21
1
0
1
4
0
10
20
6
0
0
3
04
15
11
NUMBER
OF
DEATHS
3
3
2
2
2
11
6
4
4
2
2
2
2
2
2
2
2
NUMBER
OF
EVACUEES
0
100
0
0
0
0
1.800
0
0
173
0
0
0
0
6.000
0
550
           FIVE YEAR TOTAL
                                                                                                          318
                                                                                                                    127
                                                                                                                                10.223
End Effect Codes: SP - spill; VR - vapor release: Fl - fire; EX - explosion; UU - unknown

Note:       Only a small percentage of the information contained in the event reports used to construct the AHE/DB have been independently verified.
           either by EPA or its contractors. Thus, one should interpret these data as 'deaths or injuries that were reported by one or more
           of the contributing sources.* In addition, direct causation is rarely provided in the contributing sources to the AHE/DB. See additional
           cautions for interpretation in Chapter 2.

Source: Industrial Economics, Inc.

-------
                              Exhibit 2.3-2
                               'l •'•

AIR RELEASES OF TOXIC CHEMICALS LEADING TO FORTY OR MORE INJURIES
                              (1082 -1086)
DATE
09-Jul-86
27-Mar-ee
2S-Apr-86
21-Fab-ee
22-Jul-8a
ie-Jul-86
30-May-se
11-Aug-8S
2t-Jan-8S
21-Mar-«5
26-Jun-8S
20-Apr-«5
05-Sep-85
30-Sap-8S
16-Jul-85
26-Jul-eS
13-Apr-85
22-Nov-85
04-Aug-a5
05-Dec-85
13-Mar-85
27-Aug-85
06-Ocl-84
10-May-84
15-Nov-84
08-Jan-84
22-Jan-84
17-Jul-fl4
25-OCI-84
SUaSTANCES RELEASED
SUBSTANCE1 SUBSTANCES
PHOSPHORUS
HYDROCHLORIC ACID
UNKNOWN
CHLORINE
ETHYLENE GLYCOL
SULFURIC ACID
UNKNOWN
ALDICARB OXIME METHYLENE CHLORIDE
DIMETHOATE
POLYVINYL CHLORIDE
PARATHION PARAQUAT
METHYL ETHYL KETONE PHOSGENE
UNKNOWN
CARBON MONOXIDE
POLYVINYL CHLORIDE HYDROCHLORIC AGIO
CHLORINE
BENZENE SULFURIC ACID
SILVER CYANIDE
EXPLOSIVES
HYDROGEN NAPHTHA
ACRYLIC RESIN
URANIUM OXIDE
MALATHION
HYDROCARBONS BENZENE
METHYL ISOCYANATE
HYDROGEN SULFIDE AMMONIA
PHOSPHORUS OXYCHLORID
AMMONIA
CHLORINE
SUBSTANCES 1ST
EX
SP
Fl
SP
SP
SP
VR
VR
VR
VR
GASOLINE VR
HYDROGEN CYANIDE Fl
VR
VR
Fl
VR
URANIUM EX
EX
Fl
EX
VR
SP
VR
LACQUER EX
VH
SP
SP
EX
SP
END EFFECTS
2ND 3RD 4TH
Fl
VR
VR
VR
VR
VR
UU
UU
SP
Fl
Fl
VR
UU
UU
VR
UU
Fl
VR
EX
Fl
SP
VR
UU
Fl
UU
VR
VH
VR
UU
VR
UU
UU
UU
UU
UU
UU
UU
UU
UU
EX
UU
UU
UU
UU
UU
VR
UU
UU
VR
UU
UU
UU
VR
UU
UU
UU
UU
UU
SP
UU
UU
UU
UU
UU
UU
UU
UU
UU
UU
UU
UU
UU
UU
UU
UU
UU
UU
UU
UU
UU
UU
UU
UU
UU
UU
UU
UU
NUMBER NUMBER
OF OF
INJURIES DEATHS
400
115
00
71
65
50
40
430
200
143
134
80
72
67
66
65
61
46
45
45
40
40
161
125
111
78
72
60
48
0
0
0
0
0
0
0
0
0
1
0
0
0
0
0
0
0
0
0
2
0
1
0
1
0
0
0
0
0
NUMBER
OF
EVACUEES
40,000
0
0
740
0
0
5.000
3.100
0
8.250
2.000
74
260
0
10.000
0
0
200
4.500
0
0
0
0
0
600
1.200
0
0
0

-------
                                                                  Exhibit 2.3-2
                                                                  (Continued)
                      AIR RELEASES OF TOXIC CHEMICALS LEADING TO FORTY OR MORE INJURIES
                                                                  (1902-1880)
DATE
SUBSTANCE1
8UBSTANCE2
SUBSTANCES
                                                                                      END EFFECTS
1ST  2ND  3RD  4TH
NUMBER  NUMBER
  OF       OF
INJURIES  DEATHS
 NUMBER
   OF
EVACUEES
30-Mar-84
SULFURIC ACID
                                                 SP   VR   UU   UU
                                                                                                             40
                                                                                                                                 200
IS-Jun-AS
10-Aug-«3
04-Apr-83
25-Jul-83
22-Dec-«2
13-Sep-«2
12-Nov-«2
31-Oct-«2
02-Jun-82
02-Jun-82
13-OCI-82
22-Oct-«2
PARATHION
SODIUM HYDROSULFIT
NITRIC ACID
UNKNOWN
METHYLACRYLATE
SODIUM SULFIDE
ETHYLENE OXIDE
CHLORINE
NITRIC ACID
PHOSGENE
VINYL BENZENE
TRICHLORO ETHYLENE
SP
SP
SP
EX
VH
Ft
VR
Fl
UU
VR
Fl
EX
UU
UU
EX
VR
118
70
40
40
0
0
0
0
1.600
107
6.000
300
                                  PARAQUAT
                                  FLUOROCARBON
                                  TOLUENE DIISOCYANATE
                                                    ORTHENE
SP
VR
SP
SP
EX
VR
SP
SP
VR
Fl
VR
VR
Fl
SP
VH
VR
UU
UU
UU
UU
VR
UU
UU
UU
UU
UU
UU
UU
SP
UU
UU
UU
355
100
77
70
04
00
61
40
0
0
0
1
2
1
0
0
400
2.000
1.000
0
6.000
0
350
300
         FIVE YEAR TOTAL
                                                                                                           4.121
                                                                                                                   02.141
End Effect Codes: SP - spill: VR - vapor release; Fl - fire; EX - explosion; UU - unknown
Note:     Only a small percentage of the Information contained in the event reports used to construct the AHE/OB have been Independently verified.
         either by the EPA or its contractors. Thus, one should interpret these data as 'deaths or injuries that were reported by one or more of
         the contributing sources.' In addition, direct causation is rarely provided in the contributing sources to the AHE/OB. See additional
         cautions for interpretation in Chapter 2.
Source: Industrial Economics. Inc.

-------
                                 Exhibit 2 £-3

AIR RELEASES OF TOXIC CHEMICALS LEADING TO TWO THOUSAND OR MORE EVACUEES
                                 (1982-1086)
EOATE
08-JUM6
06-Aug-86
30-May-se
08-Jun-86
16-Jul-45
21-Mar-85
22-Jun-aS
27-Nov-8S
04-Aug-85
09-Jun-86
24-M«y-86
13-NOV-85
11-Aug-S5
16-Mar-85
10-Aufl-45
22-Mar-85
26-Jun-«5
04-Sep-flS
14-May-«4
31-Dec-84
10-Apr-«4
01-Jun-84
12-May-83
04-Apr-83
26-Jan-83
05-Nov-83
leUBSTANCES RELEASED
SUBSTANCE1 SUBSTANCE2 SUBSTANCES
PHOSPHORUS
CHLORINE
UNKNOWN
BUTADIENE
POLYVINYL CHLORIDE
POLYVINYL CHLORIDE
METHYL BROMIDE
LINURON
EXPLOSIVES
VINYL CHLORIDE
HYDROCHLORIC ACID
BROMINE
ALDICARB OXIME
TOLUENE DIISOCYANATE
CHLORINE
OCTYL ALCOHOL
PARATHION
HYDROCHLORIC ACID
SULFURIC ACID
ETHYLENE OXIDE
METHYL ALCOHOL
HYDROCHLORIC ACID
PLASTIC
NITRIC ACID
SULFURIC ACID
AMMONIA (ANHYDROUS)

CALCIUM HYPOCHLORITE

FORMALDEHYDE
HYDROCHLORIC ACID

PHOSPHATE AMMONIUM NITRATE


CALCINED ALUMINA
HYDROGEN SULFURIC ACID

M ETHYLENE CHLORIDE



PARAQUAT GASOLINE

COPPER SULFATE AMMONIUM PERSULFATE

PLASTIC





1ST
EX
Fl
VR
SP
Fl
VR
Fl
Fl
Fl
EX
EX
SP
VR
EX
VR
SP
VR
SP
SP
SP
SP
SP
VR
SP
SP
Fl
END EFFECTS
2ND 3RD
Fl
VR
UU
EX
VR
Fl
VR
EX
EX
Fl
Fl
VR
UU
Fl
UU
Fl
Fl
VR
VR
VR
Fl
VR
Fl
VR
VR
UU
VR
UU
UU
Fl
UU
UU
UU
VR
UU
SP
SP
UU
UU
VR
UU
UU
EX
UU
UU
UU
UU
UU
UU
Fl
UU
UU
4TH
SP
UU
UU
VR
UU
UU
UU
UU
UU
UU
UU
UU
UU
UU
UU
UU
UU
UU
UU
UU
UU
UU
UU
EX
UU
UU
NUMBER
OF
INJURIES
400
34
40
6
se
143
12
13
45
0
21
1
430
26
3
0
134
0
21
0
0
4
0
40
10
1
NUMBER
OF
DEATHS
0
0
0
2
0
1
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
NUMBER
OF
EVACUEES
40.000
8.800
6.000
2.000
10.000
8.250
7.600
6.026
4.600
4.000
3.800
3.600
3.100
3.000
3.000
2.000
2.000
2.000
3.500
2,500
2.000
2.000
6.400
6.000
3.000
2.000

-------
                                                       Exhibit 2.3-3
                                                       (Continued)

     AIR RELEASES OF TOXIC CHEMICALS LEADING TO TWO THOUSAND OR MORE EVACUEES
                                                       (1082-1M6)
EDATE
ll-Oec-82
02-Jun-«2
06-May-82
26-Feb-«2
28-Sep-82
30-Nov-«2
U-Sep-82
10-Apr-82
21-Apr-«2
13-Sap-«2
SUBSTANCES RELEASED
SUBSTANCE! SUBSTANCE2
ACROLEIN
NITRIC ACID
FUMARIC ACID
PHOSPHORUS THICHLORI
VINYL CHLORIDE TETRAETHYL LEAD
STYRENE
CHLORINE HYDROGEN PEROXIDE
CHLORINE
CHLORINE
SODIUM SULFIDE PARAQUAT
SUBSTANCES




HYDROCHLORIC ACID

SODIUM HYDROXIDE


ORTHENE
1ST
EX
EX
EX
SP
Fl
EX
EX
VR
Fl
VR
END EFFECTS
2ND 3RD
Fl
Fl
R
VR
EX
Fl
Fl
UU
VR
Fl
VR
VR
VR
UU
SP
VR
UU
UU
UU
UU
4TH
UU
SP
UU
UU
VR
UU
UU
UU
UU
UU
NUMBER NUMBER
OF OF
INJURIES DEATHS
0
64
10
2
6
11
1
12
0
100
0
2
0
0
0
0
0
0
0
0
NUMBER
OF
EVACUEES
20.000
6.000
3.600
3.000
3.000
3.000
2.500
2.100
2.100
2.000
         FIVE YEAR TOTAL
                                                                                                   1.643
184.075
End Effect Codec SP - •pill: VR - vapor rotate; Fl - firs; EX - explosion; UU - unknown
Note:     Only a small percentage ol the information contained in the event reports used to construct the AHE/DB have been Independently
         verified, either by the EPA or its contractors. Thus, one should interpret these data as 'deaths or injuries that were reported by one
         or more of the contributing sources.' In addition, direct causation is rarely provided in the contributing sources to the AHE/DB. See
         additional cautions for interpretation in Chapter 2.
Source: Industrial Economics. Inc.

-------
                     Exhibit 2.3-4
EVENTS WITH HIGHEST QUANTITY-TO-TOXICITY RATIO
                                           Number
Plant or
Date
12/11/82
11/12/85
10/30/87
10/09/85
09/14/85
11/25/85
07/31/81
08/13/84
12/01/82
10/28/84
08/25/85
01/05/83
12/24/81
06/26/85
08/11/83
State Transit
LA P
FL
TX
MO
TX
N/A
KS
Ml
NE
CA
CA
AZ
NE
CA
NJ
P
P
T
T
P
T
P
T
P
T
P
P
P
P
Physical
State
GS
LQ
LQ
GS
LQ
LQ
LQ
LQ
LQ
LQ
LQ
LQ
LQ
LQ
LQ
GS
GS
GS
LQ
LQ
LQ
Substance(s)
Acrolein
Sulfuric Acid *
Hydrofluoric Acid * *
Hydrochloric Acid *
Sulfuric Acid *
Sodium Hydroxide *
Ammonia (Anhydrous)
Ammonia *
Hydrazine
EVENT TOTAL
Trichloroethane
Hydrogen Peroxide *
Sodium Hydroxide '
Sodium Hydroxide '
Nalco 1370
EVENT TOTAL
Sodium Hydroxide. *
Parathion
Paraquat
Gasoline
EVENT TOTAL
Hydrochloric Acid *
Phosphoric Acid *
Phosphorus Oxychloride
EVENT TOTAL
Quantity
(Ibs) |
480,000
6.000.000
270.000
1.125.000
1.500.000
1,500.000
1.575.000
1.500,000
N/A
1.500.000
23.250.000
375.000
1.125.000
938.000
N/A
938,000
750.000
3.750
3,750
4,500
12,000
N/A
N/A
51,600
51,600
IDLH
[gm/m3)
0.010
0.320
0.020
0.150
0.320
0.330
0.350
0.350
0.100
5.430
0.100
0.330
0.330
N/A
0.330
0.240
0.002
N/A
0.150
N/A
0.030
End Effects Number
Quant/IDLH 1st 2nd
48,000,000 EX Fl
18,750.000 SP UU
13.500.000 SP VR
7,500.000 SP VR
4.700.000 SP UU
4.500.000 SP UU
4.500,000 SP VR
4,300.000 SP UU
N/A
4,300.000
4.300.000 SP UU
3,800.000 SP UU
3,400,000 SP UU
2.800.000 SP UU
N/A
2,800,000
2,300.000 SP UU
16.000 VR Fl
1.900.000
N/A
1.900,000
N/A VR UU
N/A
1,700.000
1 .700,000
3rd 4th
VR UU
UU UU
UU UU
UU UU
UU UU
UU UU
UU UU
UU UU
UU UU
UU UU
UU UU
UU UU
UU UU
EX UU
UU UU
Injured
0
0
650
0
6
0
0
0
0
0
0
0
0
134
0
Hospit-

ajized FataMties
0 0
0
140
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
Number
Non-Worke
Evacuated Evacuees
20,000 20,000
0
3.000
0
300
0
0
0
0
0
0
0
0
2.000
80
0
3,000
0
0
0
0
0
i
0
0
0
0
0
2,000
0

-------
                                                        Exhibit 2.3-4 (continued)

                         EVENTS WITH HIGHEST QUANTITY-TO-TOXICITY RATIO
             Plant or Physical
  Date   State Transit  State   Substance(s)
                                                Number
Quantity   IDLH               End Effects   Number Hospit-           Number  Non-Worker
  (Ibs)    (gm/m3) Quant/IDLH 1st 2nd 3rd 4th Injured  alized Fatalities   Evacuated  Evacuees
05/24/76



01/22/84
11/03/86
06/08/86


04/10/85
02/13/85
11/02/84
04/16/84
05/13/85



LA



IL
ME
TX


TX
TX
Wl
IL
OH



P



P
P
T


P
P
P
P
P



LQ
LQ
LQ

LQ
LQ
LQ
LQ

GS
GS
LQ
LQ
GS
GS
GS

Ethylene Oxide
Propylene Oxide
Polyglycol Ether
EVENT TOTAL
Phosphorus Oxychloride
Potassium Cyanide
Butadiene
Formaldehyde
EVENT TOTAL
Sulfur Dioxide
Sulfur Dioxide
Nitric Acid *
Hydrochloric Acid '
Acrylonitrile
Hydrogen Cyanide
Acetonitrile
EVENT TOTAL
2,400.000
N/A
N/A
2,400.000
45,000
75.000
466,600
157,500
624,100
339.000
324.000
277,500
150.000
37,000
61.000
N/A
98.000
1.440
4.730
N/A

0.030
0.050
44.080
0.120

0.262
0.262
0.260
0.150
8.650
0.070
6.690

1.700.000 EX Fl
N/A
N/A
1 ,700.000
1.500.000 SP VR
1.500,000 SP UU
11.000 SP EX
1.300.000
1 .300.000
1.300.000 VR UU
1.200.000 VR UU
1.100.000 SP UU
1.000,000 SP UU
4.000 VR UU
900.000
N/A
900.000
UU UU



UU UU
UU UU
Fl VR


UU UU
UU UU
UU UU
UU UU
UU UU



0



72
0
5


0
0
0
0
0



0



12
0
1


0
0
0
0
0



0



0
0
2


0
0
0
0
0



0



0
0
2.000


0
0
0
0
0



0



0
0
0


0
0
0
0
0



End Effect Codes: SP = Spill; VR = Vapor .Release; Fl = Fire; EX = Explosion; UU = Unknown
N/A = Information was not reported in the sources used to characterize these events.

Note:    * These substances are commonly distributed in solutions of less than full concentration. Thus acute toxicity (as measured by IDLH) may be overstated.
        • * The firm that owns the facility at which this event took place submitted additional information indicating that the quantity reported as released in the
          AHE/DB is overstated. Thus the quantity-to-toxicity ratio is overstated, though this event would remain on this list even if the quantity released is modified.
Source: Industrial Economics. Inc.

-------
                                                         Exhibit 2.3-5
                            TEN MAJOR CHEMICAL-RELATED INDUSTRIAL DISASTERS IN THE TWENTIETH CENTURY
                                                                                          Number of    Number of    Number of
Year
1984
1984
1984
1978
1976
1974
1948
1947
194Z
1921
Location
Bhopal. India
Mexico City. Mexico
Sao Paulo. Brazil
Los Alfaques, Spain
Seveso, Italy
Flixborough, England
Luduigshafen, FRG
Texas City, USA
Tessenderloo. Belgiua
Oppau. Germany
Description of Accident
Methyl isocyanate leak at chemical plant
LPG explosion at processing plant
Gasoline explosion following pipeline rupture
Propylene explosion following leak fro* lorry
TCDD leak at chemical plant
Cyclohexane explosion at chemical plant
Dimethyl ether explosion in factory
Aomoniin nitrate explosion while loading ship
Ammonium nitrate explosion in chemical plant
Ammonium nitrate explosion in fertilizer factory
Deaths
>3,000
>550
508
216
0
28
24S
552
200
561
Injuries
200.000
4.248
NA
NA
NA
NA
3,800
300
1.000
1.900
Evacuees
200,000
350,000
3,000
NA
>958
3.000
NA
NA
NA
NA
Sources:  Kleindorfer and Kunreuther, 1987; Kletz, 1985;  Chemical  Engineering News.  1988.

Note:     Seveso and Flixborough are included in this list  because of their significance
          in affecting public opinion regarding the dangers associated with chemicals
          processing and transportation.

-------
                              Exhibit 3.2-1
              Distribution  By Release  Size
                          for All  Events
    100% -/
     90%-
           Number of Events = 7,807
           Amount Released - 399.6 Million Pounds
                                                                  Number
                                                                  Quantity
          1-10   11-100   101-1K  1K-10K  10K-25K 25K-50K 50K-100K  > 100K

                           Release Size (Ibs)
Source: Industrial Economics, Inc.

-------
                          Exhibit 3.5-6
       Distribution of Value of  Shipments of
           Manufactured Products  (1985)
                             Value of 1985 Shipments in $ Billions
                             • More than $179 Billion
                             E~3 $120 Billion to $179 Billion
                             E3 $60 Billion to $119 Billion
                             D Less than $60 Billion
Source: Industrial Economics, Inc.

-------
                             Exhibit 3.2-2
              In-Plant  Events By Location
     AMOUNT RELEASED IN DEATH/INJURY EVENTS
                              Unknown
                               0.3%  6.1% Process Vessel
        Vehicle not in Transit
            50.7%
                                         6.0% Storage Vessel

                                            1.3% Valve/Pipe
                                              Disposal
                                               35.4%
                      Total Amount » 46.6 Milton Pounds
            AMOUNT RELEASED  IN ALL EVENTS

                               Unknown
               Vehicle not in Transit 7.1%  a9%  5.2% Process Vessel
              Heating/Cooling 0.1%
               Disposal 5.9%

               Valve/Pipe
                 5.5%
                                          Storage Vessel
                                             75.2%
                       Total Amount = 346.2 Million Pounds
Source: Industrial Economics, Inc.

-------
                               Exhibit 3.2-3

                In-Plant Events  By Cause


      AMOUNT  RELEASED IN  DEATH/INJURY EVENTS
                           Unknown 0.2%  Equipment Failure 5.9%
                                          Operator Error 2.1%
                                            Upset Conditions 0.3%
     Sparking/Lightning  50.7%
  Fire/Explosion
    35.9%
                                        Mgh Pressure/Temperature  4.6%

                        Total Amount = 46.6 Million Pounds
             AMOUNT RELEASED IN ALL EVENTS
                         Other 5.3%
                                Unknown
                                  1.8%
            Sparking/Lightning  6.8%

                 Power 0.2%
       High Pressure/Temp  0.9%
         Arson/Vandalism  0.5%

         Fire/Explosion  5.7%

      Upset Conditions 0.8%
Equipment Failure
    32.0%
                   Operator Error
                     46.0%
                        Total Amount = 346.2 Million Pounds
Source: Industrial Economics, Inc.

-------
                          Exhibit 3.2-4
            Distribution  By Release  Size
                 for  In-Plant Locations
         EVENTS INVOLVING PROCESS VESSELS
    80%-

    70%-

    50% H
    40%-j
          Number of Events = 1,389
          Amount Released = 18.1 MilGon Pounds
H  Number
1  Quantity
         1-10   11-100   101-1K  1K-10K  10K-25K  25K-50K 50K-100K > 100K
                       Release Size (Ibs)
Source: Industrial Economics, Inc.

-------
                           Exhibit 3.2-5
             Distribution By  Release Size
                 for  In-Plant  Locations
         EVENTS INVOLVING STORAGE VESSELS
    100%-^

    90%-

    80%-

    70%-
 *""  50% H
 -5
Number of Events = 1,665
Amount Released = 260.1 Million Pounds
                                             a  Number
                                             H  Quantity
         1-10   11-100  101-1K  1K-10K  10K-25K 25K-50K 50K-100K  > 100K
                        Release Size (Ibs)
Source: Industrial Economics, Inc.

-------
    100% •
    90%-
                           Exhibit 3.2-6
             Distribution By  Release Size
                 for  In-Plant  Locations

          EVENTS INVOLVING VALVES OR  PIPES
          Number of Events = 1,746
          Amount Released = 19.0 Milfion Pounds
                                                          Number
                                                          Quantity
         1-10   11-100   101-1K  1K-10K  10K-25K 25K-50K 50K-100K  > 100K
                        Release Size (Ibs)
Source: Industrial Economics, Inc.

-------
                           Exhibit 3.2-7
             Distribution By Release  Size
                 for In-Plant Locations

               EVENTS INVOLVING DISPOSAL
    100%-/
          Number of Events = 370
          Amount Released = 20.5 Million Pounds
                                                           Number
                                                           Quantity
         1-10   11-100  101-1K   1K-10K  10K-25K  25K-50K 50K-100K > 100K
                        Release Size (Ibs)
Source: Industrial Economics, Inc.

-------
 ^
                          Exhibit 3.2-8
            Distribution  By Release  Size
                 for  In-Plant Locations
        EVENTS INVOLVING HEATING OR COOLING
    90%-
    80%-
    70%-
3  60%^

*~  50% H
    40%-
          Number of Events » 94
          Amount Released = 0.5 Million Pounds
                                                         Number
                                                         Quantity
         1-10   11-100   101-1K  1K-10K  10K-25K  25K-50K 50K-100K > 100K
                       Release Size (Ibs)
Source: Industrial Economics, Inc.

-------
 .8
                          Exhibit 3.2-9
             Distribution  By Release Size
                 for  In-Plant  Locations
     EVENTS INVOLVING VEHICLES NOT IN TRANSIT
    90%-
    80%-
    70%-
    50%-
          Number of Events = 177
          Amount Released = 24.1 Million Pounds
Number
Quantity
         1-10   11-100  101-1K  1K-10K  10K-25K 25K-50K 50K-100K  > 100K
                        Release Size (Ibs)
Source: Industrial Economics, Inc.

-------
                         Exhibit 3.2-10
            Distribution By  Release Size
                 for In-Plant  Locations
  EVENTS INVOLVING EQUIPMENT FAILURE AS CAUSE
    100% •
    90%-
         Number of Events = 3,026
         Amount Released = 110.9 Million Pounds
                                                        Number
                                                        Quantity
         1-10   11-100  101-1K  1K-10K  10K-25K 25K-50K 50K-100K > 100K
                       Release Size (Ibs)
Source: Industrial Economics, Inc.

-------
                         Exhibit 3.2-11
            Distribution By  Release Size
                 for In-Plant  Locations

   EVENTS INVOLVING OPERATOR ERROR AS CAUSE
    90%-
    80%-
    70%-
50%-
    40%-
    30%-
    20%-
    10%-
         Number of Events = 1,168
         Amount Released 3 159.1 MilPon Pounds

                                                         Number
                                                         Quantity
         1-10   11-100  101-1K  1K-10K  10K-25K 25K-50K 50K-100K  > 100K
                       Release Size (Ibs)
Source: Industrial Economics, Inc.

-------
                          Exhibit 3.2-12
            Distribution By  Release Size
                 for In-Plant Locations

   EVENTS INVOLVING  UPSET CONDITIONS AS CAUSE
    100% •/
    90%-
          Number of Events = 289
          Amount Released = 2.8 Million Pounds
                                                        Number
                                                        Quantity
         1-10   11-100  101-1K  1K-10K  10K-25K 25K-50K 50K-100K > 100K
                       Release Size (Ibs)
Source: Industrial Economics, Inc.

-------
                         Exhibit 3.2-13
            Distribution By  Release Size
                 for In-Plant  Locations

   EVENTS INVOLVING FIRE OR EXPLOSION AS CAUSE
          Number of Events = 104
          Amount Released = 19.5 Million Pounds
                                                        Number
                                                        Quantity
         1-10   11-100  101-1K 1K-10K  10K-25K 25K-50K 50K-100K > 100K
                       Release Size (Ibs)
Source: Industrial Economics, Inc.

-------
                         Exhibit 3.2-14
            Distribution  By Release Size
                for In-Plant Locations


 EVENTS INVOLVING ARSON OR VANDALISM  AS CAUSE
          Number of Events = 42
          Amount Released = 1.7 Million Pounds
                                                       Number
                                                       Quantity
         1-10   11-100  101-1K  1K-10K 10K-25K  25K-50K 50K-100K  > 100K

                       Release Size (Ibs)
Source: Industrial Economics, Inc.

-------
 4—•
                          Exhibit 3.2-15
             Distribution By  Release Size
                 for In-Plant  Locations

         EVENTS INVOLVING DISPOSAL AS CAUSE
    90%-
    80%-
    70%-
50%-
    40%-
    30%-
    20%-
    10%-
     0%'
          Number of Events = 17
          Amount Released = 0.1 Million Pounds
                                                          Number
                                                          Quantity
         1-10   11-100  101-1K  1K-10K  10K-25K 25K-50K 50K-100K  > 100K
                        Release Size (Ibs)
Source: Industrial Economics, Inc.

-------
                         Exhibit 3.2-16
            Distribution  By  Release Size
                for  In-Plant Locations

          EVENTS INVOLVING HIGH PRESSURE
              OR TEMPERATURE AS CAUSE
    100% •/
    90%-
    80%-
    70%-
          Number of Events = 133
          Amount Released = 3.1 Million Pounds
                                                       Number
                                                       Quantity
         1-10  11-100   101-1K  1K-10K  10K-25K 25K-50K 50K-100K > 100K
                      Release Size (Ibs)
Source: Industrial Economics, Inc.

-------
                         Exhibit 3.2-17
            Distribution  By Release Size
                for In-Plant Locations


 EVENTS INVOLVING POWER FLUCTUATIONS AS CAUSE
          Number of Events a 102
          Amount Released = 0.7 MilOon Pounds
                                                       Number
                                                       Quantity
         1-10   11-100  101-1K  1K-10K 10K-25K  25K-50K 50K-100K  > 100K

                       Release Size (Ibs)
Source: Industrial Economics, Inc.

-------
                          Exhibit 3.2-18
            Distribution  By Release  Size
                 for  In-Plant Locations
EVENTS INVOLVING SPARKING OR LIGHTNING AS CAUSE

ioo%-f


90%-


80%-


70%-
    40%-


    30%-


    20%-


    10%-
     0%-
          Number of Events = 6
          Amount Released = 23.6 Million Pounds
                                                         Number
                                                         Quantity
         1-10   11-100  101-1K  1K-10K  10K-25K  25K-50K 50K-100K  > 100K

                       Release Size (Ibs)
Source: Industrial Economics, Inc.

-------
 I
    90%-

    80%-

    70%-
    40%-(
    30%-
    20%-
    10%-
                         Exhibit 3.2-19
            Distribution  By Release Size
                for In-Plant Locations

             EVENTS INVOLVING LOADING OR
             UNLOADING AS EVENT CONTEXT
          Number of Events = 1,127
          Amount Released = 113.3 Million Pounds
Number
Quantity
         MO   11-100  101-1K  1K-10K  10K-25K 25K-50K 50K-100K  > 100K
                       Release Size (Ibs)
Source: Industrial Economics, Inc.

-------
 .3
                         Exhibit 3.2-20
            Distribution  By Release Size
                for In-Plant Locations

    EVENTS INVOLVING STEADY STATE CONDITIONS
                   AS EVENT CONTEXT
    90%-
    80%-
    70%-
    50%-
    40%-
         Number of Events = 3,045
         Amount Released = 206.2 MilKon Pounds
   Number
H  Quantity
         1-10   11-100  101-1K  1K-10K  10K-25K 25K-50K 50K-100K > 100K
                       Release Size (Ibs)
Source: Industrial Economics, Inc.

-------
                          Exhibit 3.2-21
            Distribution  By  Release Size
                 for In-Plant Locations

   EVENTS INVOLVING  START-UP AS EVENT CONTEXT
    100% -r
    90%-
    80%-
    70%-
          Number of Events = 101
          Amount Released = 6.1 Million Pounds
                                                         Number
                                                         Quantity
         1-10   11-100  101-1K  1K-10K  10K-2SK 25K-50K 50K-100K  > 100K
                       Release Size (Ibs)
Source: Industrial Economics, Inc.

-------
                         Exhibit 3.2-22
            Distribution  By Release Size
                for  In-Plant Locations
 EVENTS INVOLVING SHUT DOWN AS EVENT CONTEXT
          Number of Events = 76
          Amount Released = 1.1 Million Pounds
                                                       Number
                                                       Quantity
         MO   11-100  101-1K  1K-10K 10K-25K  25K-50K 50K-100K  > 100K
                       Release Size (Ibs)
Source: Industrial Economics, Inc.

-------
                         Exhibit 3.2-23
            Distribution By  Release Size
                 for In-Plant  Locations

EVENTS INVOLVING MAINTENANCE AS EVENT  CONTEXT
 3
100%-

90%-

80%-

70%-
    50% H
    40%-
    30%-
    20%-
    10%-
          Number of Events = 171
          Amount Released a 2.5 MilOon Pounds
                                                     Number
                                                     Quantity
         1-10   11-100  101-1K  1K-10K  10K-25K 25K-50K SOK-100K > 100K
                       Release Size (Ibs)
Source: Industrial Economics, Inc.

-------
                            Exhibit 3.2-24
    100% •


    90%-


    80%-


    70%-
    50%-
    40%-
    30%-
    20%-
             Distribution  By Release  Size
                   for  In-Transit  Events


                      INVOLVING TRUCKS
Number of Events = 1,170
Amount Released = 9.9 Million Pounds
                                                  Number
                                                  Quantity
          1-10   11-100   101-1K   1K-10K  10K-25K 25K-50K 50K-100K > 100K

                         Release Size (Ibs)
Source: Industrial Economics, Inc.

-------
                           Exhibit 3.2-25
             Distribution By Release Size
                   for  In-Transit  Events

                        INVOLVING  RAIL
           Number of Events = 523
           Amount Released = 11.9 Million Pounds
                                                             Number
                                                             Quantity
          1-10   11-100  101-1K  1K-10K 10K-25K 25K-50K 50K-100K > 100K
                         Release Size (Ibs)
Source: Industrial Economics, Inc.

-------
                            Exhibit 3.2-26
             Distribution  By Release  Size
                   for  In-Transit  Events
                     INVOLVING PIPELINE
    100% -1
    90%-
           Number of Events = 64
           Amount Released = 25.8 Million Pounds
                                                             Number
                                                             Quantity
          1-10   11-100   101-1K  1K-10K  10K-25K  25K-50K 50K-100K  > 100K
                         Release Size (Ibs)
Source: Industrial Economics, Inc.

-------
                           Exhibit 3.2-27
             Distribution  By  Release  Size
                  for In-Transit Events

                      INVOLVING BARGE
    100% -i
    90%-
           Number of Events = 106
           Amount Released = 5.8 Million Pounds
                                                            Number
                                                            Quantity
         1-10   11-100  101-1K  1K-10K  10K-25K 25K-50K 50K-100K > 100K
                         Release Size (Ibs)
Source: Industrial Economics, Inc.

-------
                           Exhibit 3.2-28
             Distribution By Release  Size
                  for  In-Transit Events


                 INVOLVING OTHER MODES
           Number of Events = 10
           Amount Released = 32,296 Pounds
                                                           Number
                                                           Quantity
         1-10   11-100  101-1K  1K-10K  10K-25K  25K-50K 50K-100K > 100K

                        Release Size (Ibs)
Source: Industrial Economics, Inc.

-------
                         Exhibit 3.2-29
             In-Transit Events By Mode
    AMOUNT RELEASED IN  DEATH/INJURY EVENTS
                       Barge
                       7.7% 0.1%
                                      Truck
                                      28.2%
                    Total Amount * 5.9 Million Pounds
          AMOUNT RELEASED  IN ALL EVENTS
                                    Truck
                                    18.5%
             RpeUne
             48.3%
                                 10.9% Barge
                    Total Amount = 53.4 Million Pounds
Source: Industrial Economics, Inc.

-------
                          Exhibit 3.2-30
             Distribution By  Release  Size
                  for  In-Transit Events
               INVOLVING LEAKS AS CAUSE
           Number of Events = 813
           Amount Released - 6.9 Million Pounds
                                                           Number
                                                           Quantity
         1-10   11-100
101-1K   1K-10K  10K-25K  25K-50K 50K-100K > 100K
    Release Size (Ibs)
Source: Industrial Economics, Inc.

-------
                           Exhibit 3.2-31
    90%-
    80%-
    70%-
 -   60% H
    50% H
    40% H
    30%-
    20%-
    10%-
             Distribution  By Release Size
                  for In-Transit  Events


            INVOLVING COLLISIONS AS CAUSE
          Number of Events = 250
          Amount Released = 10.6 Million Pounds
Number
Quantity
         1-10    11-100  101-1K  1K-10K  10K-25K 25K-50K 50K-100K  > 100K

                        Release Size (Ibs)
Source: Industrial Economics, Inc.

-------
                           Exhibit 3.2-32
             Distribution By  Release  Size
                  for  In-Transit Events

            INVOLVING  DERAILMENT AS CAUSE
    100% y
    90%-
    80%-
    70%-
           Number of Events = 62
           Amount Released = 4.8 Million Pounds
                                                          Number
                                                          Quantity
         1-10   11-100  101-1K  1K-10K  10K-25K 25K-50K 50K-100K > 100K
                        Release Size (Ibs)
Source: Industrial Economics, Inc.

-------
                          Exhibit 3.2-33
            Distribution  By Release  Size
                  for In-Transit Events
      INVOLVING OVER-TURNED TRUCKS AS CAUSE
 o
    10(3%-i
    90%-
    80% -
    70%-
    50% H
    40%-
    30%-
    20%-
          Number of Events = 181
          Amount Released = 1.6 Million Pounds
Number
Quantity
         1-10   11-100   101-1K  1K-10K  10K-25K  2SK-50K 50K-100K > 100K
                       Release Size (Ibs)
Source: Industrial Economics, Inc.

-------
                            Exhibit 3.2-34

              Occurrence  of Evacuations

                   DEATH/INJURY EVENTS
               Unknown
                35.5%
              Unknown
               46.1%
Evacuations
 32.8%
                                         No Evacuations
                                            31.7%
                         Number of Events = 1,019
                         ALL EVENTS
                                   Evacuations
                                    7.6%
                                              No Evacuations
                                                 46.4%
                        Number of Events = 10,933
Source: Industrial Economics, Inc.

-------
                           Exhibit 3.4-1
Distribution of Events  by Chemical  for All  Events
    i
    0)
    .a

    3
               10 CHEMICALS REPRESENTING
            THE LARGEST NUMBER OF EVENTS
       1200-1
       1000-
800-
                          Substance Name
 Note: Up to three substances may be reported as released during any one event in the AHE/DB. Thus,
 double counting will result from summing the number of release events across substances.
 Source: Industrial Economics, Inc.

-------
                                                               Exhibit 3.4-2

                               SUBSTANCES MOST FREQUENTLY REPORTED IN ALL EVENTS *
RANK   SUBSTANCE NAME"
 1    SULFURIC AGIO
 2    CHLORINE
 3    AMMONIA (ANHYDROUS)
 4    HYDROCHLORIC ACID
 5    SODIUM HYDROXIDE
 6    METHYL CHLORIDE
 7    PHOSPHORIC ACID
 6    ETHYLENE OXIDE
 9    TOLUENE
10    VINYL CHLORIDE
11    METHYL ALCOHOL
12    NITRIC ACID
13    TETRACHLOROETHYLENE
14    STYRENE
15    AMMONIA
16    HYDROGEN SULFIDE
17    ACETONE
18    METHYLENE CHLORIDE
19    BENZENE
20    METHYL ETHYL KETONE


WEIGHTED
TOTAL*
1.045
750
628
530
488
330
254
215
215
202
189
185
145
144
131
125
119
114
98
86



MAIN
205
690
518
250
88
280
44
215
75
172
49
75
35
54
111
115
29
44
48
36



SAMPLE
84
6
11
28
40
5
21
0
14
3
14
11
11
9
2
1
9
7
5
5

fOF
DEATH
EVENTS
4
4
1
1
1
0
0
0
7
2
3
4
0
1
2
2
1
1
2
4


fOF
DEATHS
5
5
1
1
1
0
0
0
13
6
9
21
0
0
6
3
3
2
16
6

tOF
INJURY
EVENTS
61
121
65
64
26
2
12
4
15
7
8
18
3
13
26
9
5
8
11
7


• OF
INJURIES
365
1.054
248
911
39
31
369
108
70
66
29
361
11
32
308
157
9
481
436
101

• OF
EVACUATION
EVENTS
30
89
67
76
12
5
10
9
10
14
9
20
2
12
21
12
3
6
8
4


• OF
EVACUEES
14.055
38.037
12.126
47.243
5.070
4.600
18.760
6.225
2.950
17.686
3.275
17.124
25
17.250
5.158
3.022
451
3.750
5.050
474
TOTAL
QUANTITY
RELEASED
(Lbi)
70,097.694
182.907
3.569.711
5.416,514
30.454.552
187.419
2,807.019
2.698.023
2,137.992
250.110
7,321,262
4,110.549
828.663
520.660
1.977.861
1,167.942
1,056.318
141.309
1.777.925
333.650
AVERAGE
QUANTITY
RELEASED
(Lbs)++
87.513
343
8.187
14.719
82.533
689
15.173
14.584
18.920
1.573
51.924
32,623
5.962
5.985
31,395
13.904
14.671
1.766
23.394
4.766
LARGEST
QUANTITY
RELEASED
(Lb.)
6.000.000
20.000
1.575.000
1.125.000
1.500.000
37.500
300.000
2,400.000
172,500
150,000
1,875.000
277.500
67.500
82.500
1,500,000
1,000.000
63.000
7.500
400.000
30.000
Notes:
      Includes all events in which the substance appears as the primary, secondary or tertiary substance.
      All chemicals are commodity chemicals.
      The weighted total equals MAIN plus ten times SAMPLE.
      Average quantities are computed using only those events that reported quantities.
Source: Industrial Economics, Inc.

-------
                               Exhibit 3.5-1
                  Distribution of All  Events
                                   Number of Events
                                   • More than  1000 Events
                                   m 501 to 1000 Events
                                   El 251 to 500 Events
                                   H 101 to 250 Events
                                   D 51 to 100 Events
                                   D 10 to 50 Events
                                   D Less than 10 Events
Source: Industrial Economics, Inc.

-------
                             Exhibit 3.5-2
              Distribution  of  Death  Events
                                 Number of Death Events
                                 • More than 10 Events
                                 B 5 to 9 Events
                                 ED 3 to 4 Events
                                 EJ 2 Events
                                 ED 1 Event
                                 D No Events
Source: Industrial Economics, Inc.

-------
                              Exhibit 3.5-3
               Distribution  of  Injury Events
                                   Number of Injury Events
                                  • More than 100 Events
                                  • 51 to 100 Events
                                  ED 26 to 50 Events
                                  H 11 to 25 Events
                                  Q 6 to 10 Events
                                  D 1 to 5 Events
                                  D No Events
Source: Industrial Economics, Inc.

-------
                             Exhibit 3.5-4
           Distribution of Evacuation  Events
                                 Number of Evacuation Events
                                 • More than 100 Events
                                 m 51 to 100 Events
                                 D 26 to 50 Events
                                 Q 11 to 25 Events
                                 E3 6 to 10 Events
                                 D 1 to 5 Events
                                 D No Events
Source: Industrial Economics, Inc.

-------
                            Exhibit 3.5-5
           Distribution of AHE/DB  Sources
                                Source Types
                                • National, Regional and State
                                E3 National and State Only
                                E3 National and Regional Only
                                D National Only
Source: Industrial Economics, Inc.

-------