Document Z4wyD4yeZva76M7EjdJ74RrEJ

The Society of the Plastics Industry, Inc. 1275 K Street. N.W., Suite 400 Washington. D.C. 20005 (202) 371-5200 flVreJrfk. To: Mcririeth Scheck Catherine Randazzo Betsy Shirley Fran Lichtenberg^ From: Margaret Rogers Date: Junel2, 1989 \ Subject: Acute hazardous events database -- source of chemicals listed in Senate Air Toxics Legislation Exploring on behalf of the Vinyl Institute the basis for the inclusion of polyvinyl chloride in the catastrophic release section of the Senate Air Toxics Bill, S.816,1 learned that the basis for the chemicals listed is a draft document referenced above prepared by an EPA contractor. The purpose of the project begun at EPA several years ago was to "prevent another Bhopal." In addition to polyvinyl chloride being listed, styrene, methylene chloride, and toluene diisocyanate are listed as well. There may be other chemicals of concern to the plastics industry. I just received this document June 9 and SPI staff is currently reviewing it. The EPA staff in charge of the project believes it is a flawed database and needs significant investigation and correction to be developed into a useful database. I look forward to discusing this document with you. cc: Lew Freeman, SPI Pat Toner, SPI VVV 000005279 an 1 "(1) a status report on standard-setting under sub 2 sections (d), (f), and (g); 3. "(2) information with respect to compliance with 4 such standards including the costs of compliance expe 5 rienced by sources in various categories and sub6 categories; 7 "(3) development and implementation of the 8 national urban air toxics program; 9 "(4) recommendations of the Chemical Safety and 10 Hazard Investigation Board with respect to the pre 11 vention and mitigation of sudden, accidental releases; 12 and 13 "(5) such recommendations for additional legisla 14 tion to achieve the purposes of this section and section 15 129 as the Administrator may deem appropriate. 16 "(r) Authorization.--There are authorized to be ap 17 propriated such sums as are necessary to carry out the provi 18 sions of this section.". 19 Sec. 3. The Clean Air Act is amended by adding the 20 following new section: 21 "prevention of sudden, accidental releases 22 "Sec. 129. (a) Purpose and General Duty.--It 23 shall be the objective of the regulations and programs author 24 ized under this section to prevent the sudden, accidental re 25 lease and to minimize the consequences of any such release of 26 any substance listed pursuant to subsection (b) or any other S 018 IS k, 52 1 extremely hazardous substance. The owners and operators of 2 facilities producing, processing, handling, or storing such sub3 stances have a general duty to identify hazards resulting from 4 such releases using appropriate hazard assessment tech5 niques, to design and maintain a safe facility taking such 6 steps as are necessary to prevent releases, and to minimize 7 the consequences of sudden, accidental releases which do 8 occur. Nothing in this subsection shall be interpreted, con9 strued, or applied to create any liability for compensation for 10 bodily injury or property damages to any person'which may 11 result from sudden, accidental releases of such substances. 12 "(b) List of Substances.--The Administrator shall, 13 not later than twelve months after the date of enactment of 14 this section, propose and, not later than twenty-four months 15 after such date, promulgate, a list of not less than fifty sub16 stances which may, as the result of sudden events, be re17 leased in concentrations that may reasonably be anticipated 18 to cause acute adverse health effects in humans. The list 19 shall be drawn from, but not be limited to, those substances 20 on the list established by section 302 of the Emergency Plan21 ning and Community Right-to-Know Act of 1986 and shall 22 include those substances with the greatest likelihood to cause 23 death, injury, property damage, or evacuations as the result 24 of sudden, accidental releases. The initial list shall include 25 sulfuric acid, chlorine, anhydrous aimmonm, hydrochloric s sis is 0000528i . 53 1 acid, sodium hydroxide, methyl chloride, phosphoric acid, 2 ethylene oxide, toluene, vinyl chloride, methyl alcohol, nitric 3 acid, styrene, tetrachloroethvlene, ammonia, hydrogen sul4 fide, acetone, methylene chloride, benzene, methyl ethyl 5 ketone, toluene diisocyanate, phosgene, polyvinyl chloride, 6 sulfur dioxide (10 per centum or more by volume), formalde7 hyde, butadiene, hydrofluoric acid, and acrylonitrile. The Ad8 ministrator shall from time to time (but not less often than 9 every five years) review and revise the list established by this 10 subsection adding substances which, as a result of sudden 11 events, may be released in concentrations that may reason12 ably be anticipated to cause acute adverse health effects in 13 humans. At the time any substance is placed on such list, the 14 Administrator shall establish a threshold quantity for such 15 substance, taking into account the toxicity (including short16 and long-term health effects), reactivity, volatility, dispersi17 bility, combustibility, or flammability of the substance. 18 "(c) Hazard Assessments.-- 19 "(1) Not later than twelve months after a sub20 stance is listed under subsection (b), the owner or oper21 ator of each facility at which such substance is present 22 in amounts greater than the threshold quantity shall 23 conduct and complete a hazard assessment for each 24 such substance present at the facility. The purpose of 25 such assessment shall be to anticipate the consequences s aie is vvv 000005292 DRAFT FINAL DRAFT 31 August 1988 ACUTE HAZARDOUS EVENTS DATA BASE (1968) Prepared for: Office of Policy Analysis Offics of Policy, Planning and Evaluation U.S. Environmental Protection Agency under EPA Contract Ho. 68-W8-0038 Prepared by: Industrial Economics, Incorporated 2067 Massachusetts Avenue Cambridge, Massachusetts 02140 31 August 1988 0OOOO52B3 VW 'I DRAFT 31 Augusc 1988 ACUTE HAZARDOUS EVENTS DATA BASE Table of Contents FOREWORD.............................................................................................................................................i ACKNOWLEDGEMENTS.......................................................................................................................ii EXECUTIVE SUMMARY.....................................................................................................................1 1. PROJECT BACKGROUND AND DATA BASECHARACTERISTICS........................1-1 1.1 Project Background....................................................................................l-l 1.2 Methodology......................................................................................................1-3 1.2.1 Data Development.................................. 1-4 1.2.2 Development of RelatedData Baaaa................................1-6 1.3 AHE Data Baaa Characteristics............................................................1-7 1.4 Representative Uses of the AKE Data Base.......................1-9 1.5 Report Organization......................................................................... ....l-il 2. CHARACTERISTICS OF THE MORE SEVERE EVENTS.........................................2-1 2.1 Death or Injury Events..............................................................................2-1 2.1.1 Extent of Damage...........................................................................2-3 2.1.2 Circumstances of Release..................... 2-4 2.1.3 Types of Facilities Involved..........................................2-5 2.1.4 chemicals Participation.......................... .....2-7 2.1.5 Relationship of Events Severity to Chemical Properties...........................................................................................2-9 2.2 Evacuation Events.................. >15 2.2.1 Extent of Damage...........................................................................2-15 2.2.2 Circumstances of Release......................................................2-16 2.2.3 Types of Facilities Involved............................................2-17 2.2.4 Chemicals Participation.........................................................2-19 2.3 Most Severe Events Involving Atmospheric Releases of Acutely Toxic or Explosive Chemicals...............................2-20 2.3.1 General Characteristics..............................................,....2-20 2*3*2 significance of Event End Effects............2-22 2*3*3 Large Scale Domestic Releases of Extremely Toxic Chemicals......................................................... 2-2 3 2*3*4 Comparison with Major Industrial Disasters of Similar Origin............................... 2-24 Wy 000005284 DRAFT 31 August 1988 V - i ft * * 3. CHARACTERISTICS OF AU. EVENTS.........................................................................3-1 3.1 Kctint of Danaqt...............................................................................................3-i 3.2 Circuastaneas of Ralaasa.........................................................................3-2 3.2.1 In-Plant Evanta.......................... 3-2 3.2.2 In-Transit Evanta......................................... 3-4 3.3 Typaa of Facilitiaa Involved...............................................................3-5 3.4 Chamicala participation.................................. 3-6 3.5 Gaographic Pattams......................................................................................3-7 APPENDICES A. Data Coding and Fornat Shaata B. Data Sourca Profilaa C. Chaaicals in tha AHE Data Basa * D. Data^Basa Sunaary for All Evanta E. Data Basa suanary for DaatB or Injury Evanta F. Data Basa suamary for Evacuation Evanta G. Data Basa Suamary for Racorda Raporting Spill/Vapor Ralaaaa or Fira/Exploaion as Worst Endtk-Effact t VVV 000005285 it ' DRAFT 31 Augusc 1988 EXECUTIVE SUMMARY PROJECT BACKGROUND The Acute Hazardous Events Data Base (AHE/DB) has been assembled as ona component of EPA's ongoing affort to assass and raspond to dangars posad to tha U.S. public and workers by suddan, accidantal ralaasas of toxic chemicals. Tha initial ahe/DB containad 3,121 avant records, which through data saapling raprasantad 6,928 events. An interim final report issued in December, 1985, titled "Acuta Hazardous Events Data Base" (EPA 560*5-85*029), presented tha analytic results obtained using tha 1985 version of tha AHE/DB. Since 1985, tha AHE/DB has baan expanded by Industrial Economics, Inc. and adapted for use in support of Title I and Title III applications under tha Superfund Amendments and Reauthorization Act (SARA) of 1986. Tha principal enhancements vara to substantially expand tha number of records and to augment tha AKE/DB with supportive data bases containing information on tha chemicals and companies involved in the reported events. These enhancements enable the AKE/DB to be more affectively used in supporting the mandates of the Accidantal Release Information Program (ARIP), the activities being undertaken in the realm of Community Right-to-Knov, and other applications being implemented by EPA. Because of EPA's primary interest in events similar in some vays to the accident in Bhopal, several criteria were used to screen event data for inclusion in the AHE/DB. Emphasis vas placed on incidents that had acute hazard potential, with incidents resulting in deaths, injuries, or evacuations given highest priority. Events that resulted in releases of hazardous chemicals were given priority, and those that released petroleum related and other types of substances vere not included. Priority vas also given to incidents involving air releases of chemicals. Events releasing large quantities of chemicals vere given priority and events with releases of less than one pound of materials were not included. AHE/DB CHARACTERISTICS The 1988 version of the AHE/DB contains 6,305 records, vhich through sampling represent 11,048 separate events. The data come from 41 sources. National sources include the National Response Center (7,755 events). United Press International and Associated Press (286 events), and DOT'S Hazardous Materials Information System (171 events). other sources included EPA's l VVV 00G0052B6 DRAFT 31 Auguat 1988 \t Region 7 (510 events), 28 daily newspapers (340 avanta), and six offleas of fiva stats governments (1,855 avants) . AHE/DB records encompass the period froa toafora 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 2,100 events par year for that period, ranging froa 1,515 events for 1982 to 2,638 events for 1984'. The Coapany/Facility (COMFAC) Data Base contains information on facilities involved in the AHE/DB avants. This information includes; facility naas, address, unique hierarchical identifier containing the three-digit zip code for the facility and an indication of any parent/subsidiary relationships, and four-digit standard Industrial Classification (SIC) for that facility. All facilities in COMFAC vers classified into one of five categories: -primary producers (facilities that produce chemicals); secondary producers (facilities^that use the chemicals to produce another product); distributors; end users; and waste-related activities. The Chemical (CHZM) Data Base contains information* on the chemicals involved in the events in the AHE/DB. This information includes; chemical name, Chemical Abstracts Service (CAS) number, four-digit SIC of the originating industry, 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" (IDLE) concentrations; physical characteristics, including vapor pressure, boiling point, melting point, and air half-life estimates; and hazard ranking information, including National Fire Protection Association reactivity and flammability rankings. CHEM also identifies whether or not that chemical is listed under RCRA, CERCLA, or SARA (Titles 302 and 313). CHARACTERISTICS OF THE MORE SEVERS EVENTS Altogether, 309 deaths are reported in the AHE/DB, plus 11,341 injuries. In the 6,567 injuries for which severity could be determined, 1,587 required hospitalization. Fire/explosion was the attributed cause almost four times as often in death/injury (D/I) events as for all events, accounting for onefifth of the D/I events. Equipment failure was the reported cause of slightly less than half of all events, and somewhat less than one-third of the D/I events. Operator error was the reported cause in approximately one-fifth of all events and D/I events. In-transit incidents play a slightly larger role in death/injury events than they do in overall events -- 37 percent versus 29 percent. Among in-transit events, truck transport represents a significantly increased share (66 percent versus 49 2 WV 000005287 i * . ft DRAFT 31 August 1988 percent) and rail transport a dacraaaad share (28 parcant varsus 44 percent) for avanta raaulting in deaths/injuries aa compared to all in-transit avanta. Although primary producara rank aacond in number of daath and injury avantsr thay account for tha largest ahara (39 parcant) of avanta in vhich daatha occur. Diatributora, who account for tha largest number of daath/injury avanta, account for only 7 parcant of tha fatalitiaa. End uaara and aacondary producara rank aacond and third in raaponaibility for fatalitiaaf accounting for 21 parcant and 9 parcant of tha daatha. raapactivaly. Waata diapoaal activitiaa account for 2 parcant of tha fatalitiaa. Ninataan parcant of daatha occur at facilitiaa of unknown typa. Primary producara again rank firat with ragard to number of injuriaa -- 3,377 injuries1 -occur at auch facilitiaa, rapraaanting 30 parcant of tha total. Diatributora riaa to tha aacond rank whan injuriaa ara conaidarad, accounting for 25 parcant, and aacondary producara account for 17 parcant. End uaara and waata disposers? account for 9 parcant and laaa than 1 parcant of tha injuriaa, raapactivaly. Eightaan parcant occur at facilitiaa of unknown typa. whan tha quantitiaa ralaaaad in daath/injury avanta ara comparad to tha raportabla quantitiaa (RQa) of tha participating chemicals, 37 parcant of tha D/X avanta in vhich quantitiaa ara raportad involve quantitiaa laaa than tha RQ, i.a., tha ralaaaa of such quantitiaa uaually would not ba conaidarad of sufficient concern to justify its report to tha National Response Canter or other jurisdiction. Only 13 parcant of tha D/I avanta involva chemical ralaaaa quantitiaa greater than 100 times tha RQs of tha involved chemicals. Evacuations occurred in 855 of tha total 11,048 avanta in tha AHE/DB. Half of tha remaining avanta did not have associated evacuations, and for an almost equal numbar of avanta it was not known whether an evacuation took place. The 29 madia aourcaa (26 newspapers, 2 wire serviceev and 1 trade journal) contributed 54 parcant of tha total numbar of evacuation records. In tha 419 avanta for uBirh tha numbar of evacuees was identified, tha total numbar of people evacuated was 464,677. Injuriaa vara associated with 40 parcant of tha evacuation avanta compared to only 9 pareant of all events. Daatha vsra associated with 5 parcant of tha evacuation avanta compared to l parcant of all events. For soma of tha evacuation avanta it vaa possible to distinguish between workers and non-workers evacuated. Of tha 335,323 people evacuated in those events, 308,183 (92 percent) vara residents and 27,140 (8 parcant) vara workers. 3 VVV 00Q005288 DRAFT 31 August 1988 I * Distributers and primary producers account for tha majority of evacuations. End ussrs and secondary producer* account for tha next largest shares. Compared to their frequency of involvement in all events, distributors and end users have a higher frequency and primary producers a lover frequency of involvement in evacuation events. The rates for secondary producers and waste related are the same for all events and evacuation events. Large Scale Domestic Release of Toxic Chemicals Potential toxic effects vary with the toxicity and the quantity of substance relearned. Among the AHE/DB events, 3,415 have information on both quantities and acute toxicity. For those events, we divided quantity released by the IDtH value attributed to that chemical to establish an empirical:-measure of the potential toxic effects of each release. Using this measure, 29 AHE/DB events involved releases whose quantity-toxicity ratio was at least half that associated with the refcsmM* mt-Bhopal, and 17 events involved releases whose ratio exceeded that at Bhopal. Among the 29 high-damage-potential events, only one event reported deaths (five fatalities), six events reported injuries (ranging from one to S50 people Injured), and five reported evacuations (up to 20,000 people). The average numbers of deaths, injuries, and evacuations were somewhat greater than among the AHE/DB subgroup of events reporting deaths, injuries or evacuations. However, given the vast quantity of materials released in the high-damage-potential events, the severity of these effects appears surprisingly low. The difference between the potential for human harm and the actual consequences in these events appears to be due in part to combinations of the following factors: (1) the chemical's physical state at the time of release was liquid and much of the material remained liquid, or other physical/chemical properties prevented wide spread 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 containment of materials released or other equipment or procedures to mitigate the extent of release and population exposure; ear (S) evacuation or other community 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)? and quantity data represent in most cases best estimates that may overstate the actual quantity released. 4 VVV 000005239 t -- * _ 41 DRAFT 31 Auguae 1988 Relationship of Event Severity to Chea,|^*t Properties Acute chemical toxicity vaa characterized in CHEH using values compatible with the "levels of concern (LOC)" defined in EPA's CEPP Technical Guidance Document. Relative toxicity values were obtained for approximately 62 percent of the 681 chemicals in the CHEM Data Base. These chemicals account for 82 percent of the events in the AHE/DB. chemicals in the highest toxicity group represent only 15 percent of the chemicals in CHEM, but account for over 44 percent of the total number of events in the AHE/DB. Approximately 75 percent of the injuries and 13 percent of the deaths reported for events in the AHE/DB have toxicity ascribed as the apparent cause. In contrast approximately 25 percent of the injuries and 87 percent of the deaths were apparently caused by fires or explosions. 5 VVV 000005290 t DRAFT 31 Auguse 1988 1. PROJECT BACKGROUND AND DATA BASE CHARACTERISTICS 1.1 PROJECT BACKGROUND Tii* Acute Hazardous Events Data Base (AHE/DB) has been assembled as part of EPA's review of the dangers posed to the U.s. 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 December 3, 1984 in Bhopal, India, killing over 3,000 and injuring over 200,000. Shortly thereafter, EPA formed an Acute Hazards List Workgroup to investigate the safety-related characteristics of the U.s. industry with regard to accidental chemical releases that could pose exceptional risks to human health, and to identify those chemicals that appear to represent unusually high risks. In late May of 1985 the decision was made to assemble an Acute Hazardous Events Data Base (AH2/DB) as part of the Acute Hazards Analysis review process. An important factor in initial AHE Data Base development was the need for it to be assembled quickly in order to support other aspects of the analysis process. A teas of three contractors was formed, with Industrial Economics, Inc. responsible for overall direction. The other two contractors were Management Technology and Data Systems, Inc. (MTDS) and PEI Associates, Inc. By midAugust, the AHE/DB had been expanded to 3,121 events, which through data sampling represented 6,928 actual incidents. An interim final report titled, "Acute Hazardous Events Data Base," (EPA 560--5--85--029) was issued in December, 1985, and copies of the data base itself were made available to the public at that time. Since-1985, the AHE/DB has been expanded and adapted for use in support of Title I and Title III applications under the Superfund Amendments and Reauthori2ation Act (SARA) of 1986. The principal adjustments were to augment the data base with 1-1 VVV 00000*5291 OBAFT 31 Augusc 1998 -11 additional data on tha chemicals and companies involved in the reported events, and to substantially axpand tha number of records. Thaaa adjustments enable tha AHE/DB to ba more affactivaly usad in supporting tha aandatas of tha Accidental Ralaaaa Information program (ARIP), tha activitias baing under taken in tha realm of Community Right-to-Know, and other applications baing implemented by EPA. Tha changes made to tha AHE/DB and associated activitias undertaken include: o Additional data from previously used and nev sources vara coded and added to the data base, more then doubling its size. o All data in the AHE/DB vas recoded to include additional information on events, such as discerning whether or not injured people required hospitalization, distinguishing between workers and non*workers evacuated during an event, and adding new categories describing the circumstances of release for events. o Two nev data bases related to the AHE/DB were developed to augment its analytical capabilities: one contains information on the chemicals involved, and the other describes facility characteristics. o Events in the original AHE/DB were reviewed to determine the continued relevance with respect to the revised screening criteria for the data base. As a result, 3,489 events involving polychlorinated biphenyls (PCB's), and 122 reports of events releasing only petroleum related products (e.g., kerosene, fuel oil) were moved to separate data bases. The purpose was to strengthen the AHE/DB1s focus on events involving toxic chemical releases with the potential for inducing acute health effects on the exposed population. This report excludes PCBs and petroleum products from the data summaries presented. o Some events in the original AHE/DB were edited to correct and augment the information provided. Emphasis vas placed upon "serious" events, those that led to deaths, injuries or evacuation, or involved large releases of highly toxic chemicals. o A number of analytical studies were undertaken with the expanded AHE/DB, including development of descriptive statistics on evacuation event:., appraisal of the relationships 1-2 Wv 00oosi9ti i DRAFT 31 Augusc 1988 between event assasraant of occurrad. characteristics and chemical propart ia, and the typo of facilities at which tho accidents o Work has begun on davoloping a atructura for providing information to tha stats and local rssponss committees with rsspoct to thsir raaponsibilitiaa undsr Titls 111 of SARA. This includss boginning ths dsvslopmsnt of "Historical Performance at comparable Facilities" reports. o Analysis of the scope and comprehensiveness of the information in the AHE/DB has continued. Because of EPA's primary interest in events similar in some ways to the accident in Bhopal, several criteria vers used to screen event data for inclusion in the AHE/DB. Emphasis was placed on incidents that had acute hazard potential, with incidents resulting in deaths, injuries, or evacuations given highest priority. Events that resulted in releases of hazardous chemicals were given priority over events that released petroleum related and other types of substances. Priority was also given to incidents involving air releases of chemicals. Events releasing large quantities of chemicals were given priority and events with releases of less than one pound of material were not included. Through imposition of these data collection criteria, the AHE/DB has been purposely biased toward events considered to have the greatest potential for sudden large scale harm to the populace due to the release of hazardous chemicals. Thus, the susoaary statistics obtained from the data base should not be viewed as being representative of the totality of the types of risks involved in the release of potentially hazardous materials. 1.2 METHODOLOGY The main purpose of the AHE/DB is to provide a historical perspective on the hinds of accidental events that . release potentially hazardous substances in the United States, the substances involved, the causative factors leading to their release, and their effects. The AHE/DB was not intended to provide a basis for the precise estimate of nation-wide 1-3 Wy 000005293 DRAFT 31 AuguftC 1988 _ > ft I frtquftnclts of events, quantities released or their consequences, but, rather to provide an assessment of tha general scopa of avanta 1.2.1 Data Davalopnant Our intantion warn to convart information from a large number of diffarant kinds of accidantal ralaasaa into a fora convaniant for suaaary and analysis. Wa found that no ona data sourca was sufficiently coaprahanaiva aithar in its braadth or its dapth to supply tha ovarall parspactiva on aeeldantal ralaasaa in tha U.S. Thus, va adoptad tha approach of: Cl) accessing larga numbers of racords from savaral diffarant kinds of data sources; (2) scraaning out irralavant or duplieata racords; (3) taking a saapla of certain categories of records; and (4) transcribing the information into a computerized data base. Davalopnant of tha currant version of tha AHE/DB evolved through a nuabar of stages. Tha process by which data is devel oped and maintained in tha data base, which enconpasses tha coding, entering and editing of tha data obtained fron diffarant sourcas, has ranained essentially tha sane throughout, however. Mora than twenty separata contributing sourcas have bean used to provide data to tha AHE/DB. (Exhibit 1.2-1) These sourcas differed in tha extant, quality, and type of information provided. Emphasis has bean placed on locating and using sourcas that potentially can yield a larga number of usable racords -- such as DOT'S "National Response Canter Data Base", New Jersey's "Emergency Response/Air Release Data Base", and New York's "Attorney Ganaral's Toxic Accident Data Base" -- or those that provide information on "serious" events, i.e., those that lad to deaths, injuries or evacuations. Tha racords included in tha AHE/DB encompass tha period from before 1980 to tha present. However, tha major data gathering effort focused on tha five-year period from 1982 to 1986. Tha AHE/DB contains an average of 2,100 events par year for that period, ranging from 1,515 events for 1982 to 2,638 events for 1984. (Exhibit 1.2-2) Data coding involves organizing the information provided from tha diffarant sourcas into tha format specified in a single two-sided coding sheet (Appendix A) . The coding proems 1-4 VVV 000005294 ' 'A DRAFT 31 Augusc 1988 organize# and characterizes tha descriptive information containad in tna original sources into a consistent set of categories that enable the generation of summary statistics and cross tabulations. Because the AHE/DB is comprised of data that are taken from many different contributing sources and restructured into a single consistent format, comment sections are used in the coding form to maintain information that does not fit the coding scheme, thereby preventing the loss of important unique accident information. The comment information is usually a verbatim tran scription of the text included in the record of the contributing source, and this is often quite terse. In other cases, the coders entered precis of more extensive records. As mentioned previously, the original 1985 coding format vas re-organized and expanded during the present study. The choice of nev categories that vere added and the re-coding of the existing data from the original AKE/D8 vas made possible by reviewing data contained in these comment fields. Appendix A also contains an instruction sheet developed to assist the data coders in making consistent categorical choices. Coders vere junior level to senior level professionals vith background training or experience in engineering, chemistry, and risk assessment, among other relevant disciplines. A data structure for the AHE/DB in Dbase III format vas developed in response to the EPA's specification that the AHE/DB be compatible vith most of EPA's installed PC capability. To expedite entering the coded data into the proper dBase III format, a data entry program vas developed vith screens that replicated the data coding forms and vith built-in error checking capability that prevented improper categories from being entered in each of the fields. Data editing is an on-going process vhose activities included assuring that the coded records accurately reflect the data source information as received, combining information from duplicate sources for the same event, and utilizing the results of in-depth reviews of events in the ABE/OB to upgrade signifi cant records* However, neither the study team nor EPA has made an independent review or verification of the information in moat of the regards. Thus, we did not discover events or information about events that was not previously recorded elsewhere. Neither 1-5 VVV 000005295 _^c DRAFT 31 August 1988 I did WS "fldeond-guassN the information containad in the sources usad. The AHE/DB can ba viavad as an accurata compilation and codification of previously reported information on acute hazardous material releases. 1.2.2 Development of Related Data Bases Two other data bases in dBase III format have been developed to augment the AHE/DB by providing information related to the facilities at which events have occurred and data on the properties of chemicals involved in the event*. Sinking the AHE/DB to th*e other data bases facilitates thevsnalysds of the relationships among event characteristics and facility and chemical characteristics. (Exhibit 1.2-3) Both of the related data bases contain some information beyond that provided in the event records of the contributing sources. working from other standard published sources, ve created brief summary dossiers on the facilities and the chemi cals showing up in the event records, to the extent to which they could be specifically identified. The company/facility (COKFAC) data base contains information on the facilities reported to be involved in the AHE/DB events. This information includes: the facility name, its address, unique heirarchical 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) for that facility. For some of the facilities in COMFAC information on the number of employees is also included. COKFAC is linked to the AHE/DB through the hierarchical Identifier. All facilities in COKFAC were organized into one of five categories: primary producers (facilities that produce chemicals)? secondary producers (facilities that use the chemicals to produce another product); distributors; end users; and waste-related activities. Ultimately, through the informa tion contained in COKFAC, other data sources such as the OTS Toxic Substances Inventory at the Dunn and Bradstreet ratings may be accessed to provide financial characteristics and other relevant feCllity information. 1-6 VVv 0005^96 r *t DRAFT 31 Augusc 1988 The Chamical (CHEM) Data Baft* contains information on tha chemicals involved in tha avants in tha AHE/DB. This information includas chamical idantification information such as: tha chami- cal name, its chamical abstract systam (CAS) number, tha 4-digit SIC of tha industry that producas it, whether it is a commodity or spacialty chamical, and production charactaristics. m addi tion, CHEM contains information on chamical propartias such as: toxicity information, including "Immediately Dangarous to Lifs and Health" (IDLH) concantrations; physical charactaristics, including vapor pressure, boiling point, malting point and air half-life estimates; and hazard ranking information, including National Firs Protection Association reactivity and flammability rankings. CHEM also identifies vhathar or not that chamical is included on any of tha following lists: SARA Section 302 list, CERCLA Reportable Quantity list, SARA Section 313 list, or RCRA Hazardous Waste list (Codas P or U). For both COMFAC and CHEM, care was taken to eliminate redun dancies in tha identification of facilities and chemicals. A unique facility name was agreed upon with all other spallings or name presentations maintained for reference. Similarly, unique chemical names and CAS numbers vara established. 1.3 AHE/DB CHARACTERISTICS o The AHE/DB currently includes 6,305 records, which, through sampling, represent 11,048 events. o Of those records, 1,045 events involved deaths or injuries. o Of the 11,048 events, 4,770 events involved a release that affected air, 2,114 affected water, and 5,157 affected the land. (For any event more than one medium could be affected). o Of the 11,048 events, 4,371 events related to primary producers, 1,538 to secondary producers, 2,711 to distributors, 1,239 to end-users, 158 to waste related activities, and 1,033 events occurred where the facility type was not specified. 1-7 VVV G00005297 DRAFT 31 August 1988 - 1 ft o For events involving deaths or injuries, 273 involved prinary producers, 136 secondary producers, 320 distributors, 137 end-users, and 14 waste related; facility type could not be determined for 165 D/I events. o Of the 11,048 events, 8,513 events involved commodity chemicals, 779 involved specialty chemicals, and in the remaining 1,756 events the chemical identify either was not specified or could not be classified as specialty or commodity. o 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, and property and environmental damage. o The types of accidents not contained in the ARE Data Base include: citizen complaints of oily substances, foul odors, and the like; incidents that are not chemical related; and incidents in which only non-hasardous materials are involved. o Records in the AHE/DB were gathered from 41 separate sources, the most significant of which included: The National Response Center (U.S. Department of Transportation): Contains nationwide information on hazardous substances released to air, water, and land. Includes in-plant and transport releases. Through a stratified random sampling scheme applied to the NRC data, the 3,013 records included in the AHE/DB represent a total of 7,755 NRC-recorded events. News Hire Service (DPI): Nationwide coverage of large events and events having severe consequences. Events involving evacuations, injuries, and deaths commonly reported in this media. Marsh and McLennan: U.S. incidents included among the 100 worst industrial accidents from 1956 to 1984 (rated by property damage). One-tenth are transportation related; the remainder are split evenly between chemicals processing and petroleum refining. EPA Region 7: Incidents affecting all media in Iowa, Kansas, Missouri and Nebraska 1-8 VVV 00000529a T- 'i DRAFT 31 August 1988 Newspapers: Hsvs reports, mainly of large events and avants having severe conaaquancaa. Rational coverage with emphasis on stats avanta. Stat: a Reporting systems: Taxaa Air Control Board data covar air ralaaaaa from manufacturing facilitiaa. Ohio EPA data includa manufacturing and tranaportation ralaaaaa. California data amphaaiza tranaportation and ralaaaaa to land. New Jarsay and New York data focua on ralaaaaa with potentially significant affacta. 1.4 REPRESENTATIVE USES OF AHE/DB Tha AHE/DB vaa initially daaignad for a limited purpose -to sat into hiatorical perspective tha liat of extremely hazardous substance* that EPA was developing in 1985. It performed that function, and showed that tha list waa relevant to tha kinds of ralaaaaa and tha substances often ralaaaad in tha U.S. Of tha more serious avants (i.e., those with reported deaths or injuries), almost half involved tha ralaaaa of one or more substances on tha original Extremely Hazardous Substances (EHS) liat, and another 30 percent of tha avanta involved substances with known fire or explosive hazards. Beyond this "reality checking" function, however, EPA and others have found various additional uses for tha AHE/DB. Source of Information on Specific Events Although tha AHE/DB is not a canaus of avanta, but rathar a partial and informal sampling of racorda of avanta, it is now broad enough in its coverage to be a good sourca of eventapacific information for further inquiries. For instance: o EPA'a Office of Toxic Substancas usad tha listing of events as a sampla frame for examining ways in which the TSCA premanufacture review process might be modified to more directly focus on the potential for accidental ralaaaaa. For chemicals of tha type under consideration, thay examined the AHE/DB records and other EPA data on the causae and consequences of accidental releases. 1-9 VVV 000005299 DRAFT 31 August 1988 - :* o A trad* association that is examining hov evacuations are performed in the U.S. ussd ths AHE/DB to hslp in their iden tification of events of interest. The researchers then conducted detailed case studies on about fifty events, using documentary evidence and interviews with individuals involved. Perspective-Setting for Specific Regulatory Actions The sampling approach for the AHE/DB is unusual, and it would be erroneous to ascribe scientific statistical significance to the aggregated data. Nevertheless, in the process of acquiring and coding first 3,121, and now 8,305, records it is valid to examine the relative importance of various measures shoving up within^ such an extensive collection of historical records. This report does so in its numerous Sigores and tables -- examining substances, causation, event types, and end effects. EPA has used this data to pursue some aspects of potential regulatory or other program implementation actions. For example: o The Office of Pesticide Programs has examined the frequency of chlorine releases from swimming pools and their consequences versus 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 has examined the environmental loadings and reported acute health effects of a set of substances in preparing for potential regulation of various sources of air releases -- do accidental releases contribute significantly to the health risfcs? o The AHB/QB has been used in setting the initial threshhold levels for the Accidental Release Information Program (ARIP) -- as a means to focus attention on a small fraction of reported events most deserving of follow-up data collection. Perspective-Setting for Policy Decisions Similar to the topic above, the extensive nature of the AHE/DB allows us to gain some overview of what situations seem relatively more or less important, and has contributed to setting the scene for policy discussions. 1-10 VVV 000005300 fi DRAFT 31 August 1988 o The AHB/DB was one element used in clarifying the extant of accidental ralaaaas in tha U.S. and thair potential for human and environmental damage, and hae been used by the internal CPA workgroup implementing the SARA Title III amendments dealing with the new community right-to-know program. o The data has been used by offices in EPA attempting to examine priorities from a comparative risk point of view, with the potential of modifying existing emphases within the Agency. guidance for State or Local Actions EPA and IEc have done preliminary work in developing profiles of the types and severity of releases possible from various classes of facilities. The work on "Historical Behavior at Comparable Facilities" is experimental at this point, and it must be carefully applied. However, it is possible that aggrega ting records by logical groupings of events may help local officials or others to grasp the range of accident types that they may have to plan for in their local areas, even though the historical record in that specific area may be quite thin. In some cases, we have supplied printed or computer records to EPA regional, state, or local officials. In addition, we have discussed a more generalized profile format with people involved in the Massachusetts state Emergency Response Commission (SERC) activities and with several Local Emergency Planning Committees (LEPCs > in Massachusetts. 1.5 REPORT ORGANIZATION The remainder of this report is divided into two major sections and ten appendices. The major sections are: o Characteristics of the More Severe Events, and o Characteristics of All Events. The section on the more severe events discusses events involving either deaths, injuries, or evacuations from ths perspective of extent of damage, circumstances of release (such as location, cause, process state, transit mode, etc.), types of facilities involved, chemicals participation, and the relation ship of events severity to chemical properties. 1-11 VVV 000005301 DRAFT 31 August 1988 -! The section on all events provides a similar ovarviav of information assembled for all events in the AHE Data Basa, including thoaa not involving deaths, injurias, or evacuations. Frequently reported chemicals, quantities released, media affected, environmental and property damage, and evacuations are discussed. The aection on the circumstances of release describes end effects, event location, caose, production configuration, intransit mode, and relationships between these factors and quantity released for all events. The seven appendices provide detailed background material regarding the information contained in the ABE Data Base. The Beta Coding and Format Sheets (Appendix A) Identify the informa tion included in a complete record, although few records include all of this data. Data Source Profiles (Appendix B) identify the principal characteristics of the source data files that provide the information assembled within the AHE Data Base. Appendix C provides a list of all chemicals included in the data base, including their common name ; 4-digit SZC code of* the producing sector; their Chemical Abstracts Service (CAS) identification number; and the number of incidents, injuries, fatalities with which they are associated. Appendices D, E, and F provide summary outputs of AHE Data Base information for all events, for death or injury events, and for evacuation events. Appendix G provides a similar summary for those events reporting a spill/vapor release or fire/explosion as the worst end-effect. 1-12 VVV 000005302 i DRAFT 31 August 1988 2. CHARACTERISTICS OF THE MORE SEVERE EVENTS 2.1 DEATH OR INJURY EVENTS Each of the original records coded into tha data base was examined for indications of human casualties. In most cases, the record unambiguously indicated whether or not the event led to deaths or injuries. Altogether, injuries were reported in 9.2 percent of the events, and fatalities in 1.1 percent of the events. Another 8.4 percent of the records indicated injuries as unknown, and 6.0 percent of the records listed deaths as unknown. (Exhibits 2.1*1 and 2.1*2) Although some other records in the data base may represent events that involved injuries or deaths, it was not possible to estimate the degree of under-reporting without follow-up on event reports. Some cautions are necessary in interpreting the reported casualty data: o Reported Deaths or Injuries: Only a fsw of the events in this study have been independently verified, either by ERA or its* contractors. Thus, one should interpret these data as "deaths or injuries reported by one or more of the contributing sources." o Causation: Direct evidence of causation rarely exists in the contributing sources of these records. 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 a 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 released materials.) Whenever possible, deaths explicitly resulting from a collision were removed from the data base. However, it is not possible to identify how many of the deaths and injuries remaining in the data base fall into this type of situation. 2-1 VVV 000005303 DRAFT 31 Auguc 1988 I o Rang* of Stvritv: Although soot records indicat* in detail the type and seventy ot the reported injuria*, most records do not. To categorize tha savarity of injuries for thaaa racorda, where possibla va codad whathar or not tha injurad parti** vara hospitalized. This then served as a surrogate for injury severity. While this categorization provides son* resolution of tha question of severity, it should be noted that not all data sources provided information on hospitalizations and not all hospitalizations are necessarily serious. Thus, even in tha case of hospitalizations an individual "injury" could range from temporary respiratory or eye irritation treated on-site to critical injury leading to prolonged hospital treatment. o Numbers: Some of the source record* provide counts of injuries that suggested precision*- vhareas others (explicitly or implicitly) vers approximate counts. If the source provided a number, it was recorded, but sometimes it was necessary to translate phrases such as "a fear* or "about 50" into specific numbers. Thus, the accuracy of the individual records composing the data base is highly variable. Although there is understandable interest in making nation wide estimates of deaths and injuries, the study was not designed with that purpose in mind. Rather, the objective was to better understand the nature, causes, and effects of accidental releases. Before drawing conclusions regarding national totals, the reader should carefully assess the cautionary notes above and consider the following factors: o Twenty-nine media sources (26 daily papers, two wireservices, and one trade journal) contributed only six percent of the total number of events, but those events represented 29 percent of the total death/injury events. Although these sources together cover the period I960 to 1966, 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 Appendix 0) o Although each of the 26 daily papers had a separate regional focus, some also reported moderate to severe accidents from other parts of the country. Some events appeared in two or three of these sources, as well as occasional other sources. Thus, it is believed that most of the serious events that received national media attention during the period have been included in the data base. 2-2 VVV 000005304 I DRAFT 31 August 1988 o On the other hand, soma events involving human casualties shoved up in only one of the sources searched. This suggests that one would find other events of similar magnitude both by searching additional years for the sources used in this study, and by examining other media sources with different regional specializations. o Some )cinds of events may not be reported to one of the contributing sources due to the circumstances under which the events occurred, such as an event involving casualties that are limited to the 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 AKE Data Base. Thus, it is not possible at this point to estimate how many additional records might be identified by accessing those sources. o Many of the reports were obtained from sources that obtained the data during the initial stages of the incident, and the recorders of the data also were responsible for coordinating aspects of the emergency response effort. Such reports tend to be preliminary in nature -- frequently being incomplete in some ways, and in some cases containing Inaccurate information. 2.1.1 Extent of Damage o events) is 309 D-2) One thousand forty-five events (9.5 percent of all involve deaths or injuries. The total number of deaths and the total number of injuries is 11,341. (Exhibit o In the 6,567 injuries for which the severity is known, 1,587 required hospitalization. (Exhibit D-2) o For events in which the number of injuries is known, the average number of injuries is 11.4, and the average number of injuries requiring hospitalization is 2.7. (Exhibit D-2) * Exhibits using an alphabetic designation, such as D-2, refer to exhibits in the Appendices. 2-3 WV 000005305 DRAFT 31 August 1988 o For events in which deaths ara reported to have occurred, the average number of deaths is 2.7. (Exhibit D-2) 2.1.2 Circumstances of Release o Vapor releases are reported as an end effect in 63.3 percent of the D/X events, spills in 56.0 percent, fire in 25.8 percent, and explosion in 17.8 percent. (Exhibit 2.1*6) In-Plant Events o For both all events and death/injury (D/X) events, storage and process locations jointly account for half of the reported events. Valves and pipes is the second most frequently involved location in all events, but ranks- third among D/I events. (Exhibit 2.1-3) o Reports of unknown location approximately double for D/X events; but the percentages for events at process vessels, storage locations, vehicles not in transit, heating/cooling, and other locations remain nearly the same for all events and D/X events. Reports of events in valves, piping and fittings decrease by ten percent when D/X events are compared to all events. (Exhibit 2.1-3) o Fire and explosion is reported as the attributed cause almost four times as often in D/X events as for all events, accounting tor 5.4 percent of all events and 20.4 percent of D/I events. Equipment failure is the reported cause of slightly less than half of all events (46.8 percent), but somewhat less than one-third of the D/X events (29.9 percent). Operator error is the reported cause in 18 to 19 percent of all events and D/I events. (Exhibit 2.1-4) o Steady state conditions is the event context in more than half of all events, and approximately 40 percent of D/X events. Maintenance operations and start-up both are reported twice as frequently as event contexts for D/X events as for all events. Unknown context increases from approximately one-quarter of all events to nearly one-third of D/X events. (Exhibit 2.1-5) 2-4 VVV 00000*5306 ' ,L DRAFT 31 Augusc 1988 In-Transit Evtnta o In-transit incidents play a slightly larger rola in dsath/injury events than they do in overall svants -- 37.0 parcant varsus 28.9 parcant. (Exhibit 2.1-7) o Truck transport raprasants a significantly increased shara (65.6 parcant varsus 48.6 parcant) and rail transport a dacraasad shara (28.2 parcant varsus 43.5 parcant) for avants resulting in deaths/injuries than for all in-transit events. (Exhibit 2.1-8) o Collision is reported as the in-transit causa twice as frequently for D/I events than for all avants. Leaks account for the largest shara (46 parcant) of in-tran#it releases for both D/I events and all avants. (Exhibit 2.1-9) 2.1.3 Types of Facilities Involved Industry sectors vara identified for the responsible parties in 84 percent of the 1,045 events reporting deaths and/or injuries by assigning 4-digit Standard Industrial Classification (SIC) codas. Sources for assigning SIC codas included the Dun & Bradstreet Reference Book and identifying information in the event records. Five industry groups vara classified: Primary Producers. including Chenical and Allied Products, and Petroleum Refining and Allied Products (SICs 28 and 29)? Secondary Producers, including all other nanufacturing 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). o Of all death/injury events, the largest share (30.6 percent) were associated with distributors. Primary producers, who receive principal emphasis in most emergency response planning, were responsible for a slightly smaller share, at 26.1 percent. Secondary producers and end users were responsible for equivalent shares of death/injury events at 13.0 percent and 13.1 percent, respectively. Waste disposal facilities accounted for only 1.3 percent of the D/I events. (Exhibit 2.1-10) 2-5 VVV 000005307 DRAFT 31 August 1988 o Although prinary producers ranX second in the number of death/injury events, events at these facilities account for the largest share (41.4 percent) of the deaths that occur. Distributors, vho account for the largest number of events, account for only 7.1 percent of the fatalities. End users and secondary producers ranX second and third in responsibility for fatalities, at 20.7 percent and 9.4 percent shares, respectively. Waste disposal activities account for 2.3 percent of the fatalities. (Exhibit 2.1-11) o Primary producers again ranX first with regard to number of injuries -- 3,377 injuries occur at such facilities, representing 29.8 percent of the total. Distributors rise to the second ranX when injuries are considered, accounting for 24.4 percent, and secondary producers account for 17.0 percent. End users and vaste disposers account for 94 percent and 0.7 percent of the injuries, respectively. Eighteen percent of all injuries occur at facilities of unXnown type. (Exhibit 2.1-12)* o The frequency at which fatalities occur among all events at each facility type is shown in Exhibit 2.2-13. Primary producers account for 43.6 percent of the 10,015 events for which facility type could be determined, and of those events 46 (or 1.1 percent) result in fatalities. Secondary producers incur a comparable frequency of fatalities (1.2 percent) among their events, while distributors incur only an 0.5 percent frequency of fatalities. End users and waste disposers incur somewhat higher frequencies, at 1.7 percent and 1.9 percent, respectively. However, the number of events in these two categories is small, and the higher frequencies may represent the reporting of only the more severe events for end users and waste disposers. o Similar results are obtained when injury frequency is considered, except that in this case distributors are ranXed first, with 11.8 percent of all events reported at distributors resulting in injuries. End users and waste disposal ranX second and third, at 10.6 percent and 9.0 percent respectively. Primary producers and secondary producers report injuries less frequently when relemmes occur, experiencing 6.1 percent and 8.3 percent injury frmqnmncies, respectively. (Exhibit 2.1-14) o Vapor releases are the most frequently reported end effect associated with D/Z events for all facility types, except distributors. In the case of distributors, spills are the most frequently reported end effect (at 81.6 percent), with vapor 2-6 VVV 000005303 i DRAFT 31 August 19S8 releases Mcond at 57.5 parcant of tha avanta. Flraa and xploaiona aach account for 20 to 37 parcant of tha D/I avants for all facility types, axcapt distributors and vaata diapoaars. Firaa ara raportad in only 14.1 parcant of tha avants at distributors, and explosions in 8.4 parcant, firaa ara raportad in half tha avants at vaata diapoaars. (Exhibit 2.1-15) 2.1.4 Chemicals Participation o Tha alavan chemicals most fraquantly raportad in aasociation with daath/injury avanta account for up to 45 parcant of such avants (saa Exhibit 2.1-16 for a graph of tan of thasa subatancaa by shara of daath or Injury avanta).* - chlorina is involvad in tha largast numbar of avants (125), rsprasanting 12 parcant of daath/injury avanta; fiva daaths and 1,086 injuriaa occur in thasa avanta. (Exhibits 2.1- 16 and 2.1-17) - Hydrochloric acid, ammonia (anhydrous), and sulfuric acid ara involvad, raspactivaly, in 68, 67 and 63 daath/injury avants, togathar rsprasanting 20 parcant of such avants and accounting for 6 daaths and 1,598 injurias. (Exhibits 2.1-16 and 2.1-17) - Tha savan chaaicals naxt Boat fraquantly raportad in conjunction with daath/injury avants (sodium hydroxida, ammonia, nitric acid, toluana, styrana, toluana diisocyanata, and phosphoric acid) account for 13 parcant of such avants; 42 daaths and 1,293 injurias occur in thasa avants. (Exhibits 2.1-16 and 2.1-17) o Chlorina is tha most fraquantly raportad chamical associatad with daath/injury avants at primary producars, secondary producars, and usars, and vasts ralatad; and ranks fourth at distributors. Chlorina is involvad in 22 parcant of Because any givan avant may involve mors than ona chamical, there cam be overlap in tha numbers of avants, daaths, and injurias vhen discussing tha participation of more than ona chamical. Therefore, all percents or combined totals of avants, daaths or injurias in this section should be interpreted as upper bounds. 2-7 VVV 000005309 DRAFT 31 Augusc 1988 I the D/Z events at primary producer*; 15 percent of D/I wants sacondary producars; and 5 parcant of D/I avants distributors. (Exhibits E-7, E-9, E-ll, and E-13) at at o Aftar chlorine, tha naxt thraa most fraquantly raportad chamicals involvad in D/I avants at primary producars ara ammonia (anhydrous), hydrochloric acid, and sulfuric acid; togathar thay account for 55 avants (20 parcant of tha total) at primary producars. (Exhibit E-7) o At sacondary producars, tha four most fraquantly raportad chamicals, in addition to chiorins, ara ammonia, ammonia (anhydrous), hydrochloric acid, and toluana diisocyanata; togathar thay account for 28 D/I avants (21 parcant of tha total) at sacondary producars. (Exhibit B-9) o At distributors, tha thraa most fraquantly raportad chamicals associated with D/I avants ara sulfuric acid, hydrochloric acid, and ammonia (anhydrous). Togathar thay account for 77 avants (24 parcant of tha total) at distributors. (Exhibit E-ll) o Tha avaraga quantitias ralaasad par event ara shown in Exhibit 2.1-17 for tha nineteen chemicals representing tha largest number of D/I avants. Among death/injury avants for tha top eleven most fraquantly raportad chamicals, sulfuric acid releases involve on avaraga tha largest quantitias (51,000 pounds), followed by nitric acid (17,000 pounds), and hydrochloric acid and sodium hydroxide (each at 11,000). o Whan tha quantitias ralaasad in death/injury avants ara compared to the reportable quantitias (RQs) of the participating chamicals, 37.0 parcant of tha D/I avants in which quantitias ara known involve quantitias lass than tha RQ, i.a., tha release of such quantitias usually would not be considered of sufficient concern to justify its report to tha National Response Canter or other jurisdiction. (Exhibit E-4) o Only 13.1 parcant of tha D/I avants involve chemical release quantities greater than 100 times tha RQs of tha involvad chemicals. (Exhibit E-4) 2-8 VVV 000005310 r- , i DRAFT 31 Augusc 1968 2.1.5 Relationship of Event Severity Tr> Prooertiaa Information on fiva physical and toxicological propsrtias -- acuta toxicity, flammability, reactivity, volatility, and procassing mode -- was antarad into CHEK for the chemicals associatad with avants in tha AHE/DB. Tha data vara collactsd froa various sourcas, which ara idantifiad in tha sutosactions balov. In addition, information on tha physical stata of tha participating chaaicals was provided by tha AHE/DB records for most avants. Exhibit 2.1-18 suaaarizas tha typo of information usad and tha hazard rating scales employed tor each property. Tha following analyses ara based on tha chemical properties of only tha primary substance involved in each event. Acuta Toxicity Acute toxicity of chemicals was characterized in CHEM using values compatible with tha "levels of concern (LOC)" defined in EPA's CEPP Technical Guidance Document. 1/ Where available. National Institute for Occupational Safety and Health (HXOSH) measures designated by the tarn "Immediately Dangerous to Life and Health (IDIH)" ware used to indicate the concentration (in g/a3) at which each chemical poses a significant risk if it is accidantaly released. Where IDLH values had not been developed by NIOSH, proxy values were estimated on the basis of animal lethality values using methods designated in the CEPP Technical Guidance Document. The IDIH values ware grouped into five LOC categories, termed XRATEs, as shown in Exhibit 2.1-18. The higher LOC categories correspond to hazard categories previously established by organizations attempting to segregate chemicals with regard to the hazards attending their usage. IRATEs 3 and 4 correspond to EPA's "highly toxic" criteria for categorizing chemicals for reporting under SABA Title III Sections 311 and 312. LRATS 2 is equivalent to SARA Title Ill's*' "toxic" reporting category. LRATEs 1 and 0 designate the lowest LOC categories. 1/ U.S. EPA Chemical Emergency Preparedness Program, site Specific Technical Guidance for Hazards Analysis: Emergency Planning for Extremely Hazardous Substances, Draft August 1987. 2-9 VW 000005311 DRAFT 31 August 1988 11 o LPATE* vara determined for approximately 62 percent of the 681 chaaicals in CHEM, as shovn in Exhibit 2.1-19 by ths distribution of all avant chemicals for tha fiva LRATE lavals. o Tha chemicals for which LRATEs vara determined account for 82 parcant of tha avants in tha AHK/DB, as shovn by tha distribution of all avants for tha fiva IRATE lavals in Exhibit 2.1-20. o Forty-six parcant of tha total number of avants in tha ABE/OB involvad chaaicals in tha highest toxicity group (LRATE 4), although only IE parcant of tha nuabar of chaaicals fall into tha IRATE 4 catagory. A slightly higher nuabar of chaaicals (25 parcant) involvad in death/injury avants vara in tha highast toxicity group. (Exhibits 2.1-19 and 2.1-20) o Ralativaly few of either death/injury or all avants involvad chaaicals in tha LRATS 3 category. This is generally consistent vith tha lever frequency of IRATE 3 rhea1cals in tha CHEH data base. (Exhibits 2.1-19 and 2.1-20) Flammability Information on tha flaaaability of chaaicals vas obtained priaarily from tha National Fire Protection Association (NFPA), which has rated substances for flaaaability (FRATE) on a scale of 0 to 4, with FRATE 4 being tha most hazardous level. 2/ In cases where NFPA ratings vara not available, flash points and boiling points vara used to assign similar ratings based on a National Academy of Sciences (NAS) schema. 3/ When neither NFPA ratings nor physical property data vara available, PRATEs vara assigned according to tha NFPA criteria based on available qualitative information. 2/ National Fire Protection Association (NFPA), Fire Protection Guide on Bmsardous Materials, 1984. 3/ National Academy of Sciences (NAS), System for Evaluation of Bulk Water Transportation of industrial Chaaicals; A ^wport to tha Department of Transportation, U.s. Coast Guard, 1974. 2-10 VVV 000005312 fi DRAFT 31 August 1988 o Information on flammability was obtained for approximately 65 percent of the chemicals in CHEM. These chemicals represent 84 percent of all events in the AKE/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 tvo highest flammability rating groups (FRATEs 3 and 4). The distribution of death/injury events by flammability rating is similar to the distribution for all events, with the exceptions of a larger percent in the unknown category and a smaller percent in the high flammability category. (Exhibit 2.120) Reactivity Reactivity was characterized using the WFFA definition -- "the susceptibility of materials to release energy either by themselves or in combination with water". Observed responses to fire exposure, shock, and pressure are used to determine such susceptibility. As with flammability, NPFA assigns dsgrees of reactivity (RRATE) on a scale of o to 4. o Reactivity ratings (RRATEs) were determined for only 40 percent of the chemicals in CHEM. Very fev of the chemicals rated for reactivity are highly reactive -- only 3.8 percent are rated at RRATEs 3 and 4. A similar pattern holds for death/injury events, although in this case RRATEs could be identified for 56 percent of the chemicals. (Exhibit 2.1-19) o A comparable pattern is reflected in the events, with only 3.3 percent of all events and 1.9 death/injury events involving chemicals with RRATEs (Exhibit 2.1-20) number of percent of 3 and 4. Volatility Vaporpressure, which indicates the escaping tendency of molecules 'ffeam liquids or solids, is used as the measure of volatility in CHEM. Ratings of high, moderate, and low volatility (VRATES) were assigned to the chemicals using criteria (identified in Exhibit 2.1-18) that had been established by 2-11 VVV 000005313 DRAFT 31 August 1988 i previous investigator*. 4/ 5/ These catagorias ara compatible with tha chemical salactlon criteria usad for tba SARA Title III Section 305(b) Review of Emergency Systems. o Vapor pressures have bean identified for only 32 percent of tha chemicals in CHEM thus far. These chemicals participate in 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 The distribution of events indicates that slightly more of the events (for which chemical vapor pressure is known) involve chemicals in the low volatility group. A similar pattern is seen for the death/injury subset of events. 20) . (Exhibit 2.1- Physical State The chemicals's physical state at the time of the event is used in this part of the analysis. Therefore, different events with which a chemical is associated may involve different physical states. o Information regarding the physical state of the primary substance is available for 99 percent of the events in the AHE/DB. (Exhibit D-l) o Liquids predominate in both death/injury events and all events, being designated as the chemical state in 72 percent of D/I events and 73 percent of all events. (Exhibits D-l and E-l) o Gases are the second most common physical state occurring in 32 percent of death/injury events and 27 percent of all events. (Exhibits D-l and E-l) A/ Versar, Inc., Physical-Chemical Properties and Categoriza tion of BOA Hastes According to Volatility, prepared for u.s. Environmental Protection Agency, February 1985. 5/ U.S. ERA Office of Toxic Substances, Section 305(b) Review of Emergency Systems -- Chemical Selection Criteria, Draft January 1987. 2-12 r VVV 000005314 fI DRAFT 31 August 1988 Processing Mod* Ai a proxy for processing mode, the chemicals vara dasignatad as either commodity or apacialty. This proxy la basad on the assumption that commodity chemicals, which ara producad in larga quantities, ara likely to ba producad by continuous processing sodas; whereas specialty chemicals, producad in small quantities to meat more narrowly defined specifications, ara more likaly to ba batch processed. o Approximately 42 percent of the chamicals in CHEM are commodity chamicals, 35 percent ara specialties, and 24 percent could not ba designated. (Exhibit 2.1-19) o seventy-seven percent of all events involve commodity chemicals, and only seven percent involve specialty chemicals. Although the shares of death/injury events involving commodity and specialty chemicals ara similar to the shares for all events, the distribution of chemicals involved in death/injury events is more, heavily weighted by commodities (Exhibits 2.1-19 and 2.120) Variation of Event Severity With Chemical Properties Acute hazardous' events can be separated into two groups depending upon whether a fire or explosion did or did not occur. 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, as discussed in Section 2.1. In those events where a fire or explosion was not an end effect, it is assumed that an important factor contributing to the severity of the event is the toxicity of the chemicals involved. Since multiple end effects often are reported for a single event, individual records were examined in order to consider all information when classifying the cause of the deaths or injuries. 2-13 vvv OOOOG^IS DRAFT 31 Augusc 1988 I I Variation With Toxicity o Approximately 75 percent of tha injuries and 13 parcant of tha deaths reported for events in tha AHE/DB vara apparently caused by toxicity. (Exhibit 2.1-31) o In tha subset of events involving deaths and injuries thought to be caused by chemical toxicity, more than tvo-third* of the events with rated chemicals involved highly toxic (URATE 4) chemicals. (Exhibit 2.1-22) o Events involving chemicals with a high URATE have a much wider range of severity (in terms of numbers of deaths plus injuries) than those with a low IRATE. (Exhibit 2.1-22) A chi- sguara analysis of the grouped data suggests that this phenomenon for death and injury events is due to the underlying distribution of the tvo variables in the data set rather than any dependence upon one another. o Chi-square analysis on the toxicity ratings ef chemicals involved in all events in the AHE/OB for which the- primary end effect was a vapor release or spill suggests that a relationship exists betveen event severity and chemical toxicity. The analysis Indicates that severity is related to toxicity in terms of 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 mors likely to cause deaths and/or injuries, but ve can not predict how many. Variation of Events Severity with Flammability and Reactivity o Approximately 87 percent of the deaths and 25 percent of the injuries reported for events in the AHE/OB were apparently caused by firs or explosion. (Exhibit 2.1-21) o There is no apparent relationship betveen 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) 2-14 VVV 000005316 tr DRAFT 31 August 198* 2.2 EVACUATION EVENTS The expanded coding format for tha praaant varsion of the AHE/OB allowed each racord to incorporata information on vhathar or not an evacuation vaa required, hov many people vara evacuated, and of those, hov many vara raeidenta and hov many ware workers. Evacuations occurred in 855 of the total 11,046 events (7.7 percent) in the AHE/DB. Of the remainder, 5,094 (46.1 percent) of the events did not have associated evacuations, and for an almost equal 5,089 events (46.1 percent) it vas not known whether an evacuation took place. The 29 media sources (26 newspapers, 2 wire services, and one trade journal) contributed 54 percent of the total number of evacuation records. similar caveats to those discussed for death and injury events apply to this analysis, particularly regarding the number of people evacuated and the distribution among residents and workers. 2.2.1 Extent of Damage o 855 events (7.7 percent of all events) were reported to involve evacuations. In the 419 events for which the number of evacuees vas identified, the total number of people evacuated vas 464,677. (Exhibits 0-1 and D-2) o Injuries vers associated with 40.1 percent of the evacuation events compared to only 9.2 percent of all events. (Exhibit 2.2-1) Deaths were associated with 4.7 percent of the evacuation events compared to 1.1 percent of all events (Exhibit 2.2-2). o For some of the evacuation events it vas possible to distinguish betveen vorkers and non-vorkers evacuated. Of the 335,323 people evacuated in those events, 308,183 (91.9 percent) were residents and 27,140 (8.1 percent) were vorkers. (Exhibit D-2) o mhen the number of people evacuated vas reported, the average number of evacuees per event vas 1,109 For those events in which a distinction vas made betveen residents and vorkers, the average number of residents evacuated was 1,034, and the average number of vorkers was 91. (Exhibit D-2) 2-15 VVV 000005317 DRAFT 31 August 1988 iL 2.2.2 Circumstances of Release o In-transit avants play a slightly largar rola in avacuation avants -- 34.6 parcant tor evacuation avants varsus 28.9 parcant for all avants. (Exhibit 2.2-3). o Spills ara raportad as an and affact in 50.1 parcant of avacuation avants, vapor ralaasa in 76.1 parcant, fira in 32.5 parcant, and axploaion in 15.1. Piras and explosions ara raportad nearly five tines more frequently for avacuation avants than for all in-plant events. Vapor ralaasas ara also acre frequently raportad, and spills lass frequently, as and affects in avacuation avants than in all avants. (Exhibit 2.2-4). In-Plant Events o Storage locations ara involved in 34.9 parcant of inplant avacuation avants, cosparad to 28.0 parcant involvaaant in all events. Vehicles-not-in-transit account for 5.5 parcant of avacuation avants and 2.8 parcant of all events. Process locations account for a smaller share of avacuation avants (19.7 parcant) whan compared to all avants (22.4 percent), as do valves and pipes (14.3 parcant tor avacuation avants varsus 25.3 parcant for all avants). (Exhibit 2.2-5). o Piras and explosions ara raportad as the attributed causa four times more frequently in avacuation avants (22.2 parcant) .as compared to all in-plant avants <5.4 parcant). Sparking and lightning, and arson and vandalism ara also more frequently raportad as causes in avacuation avants. Equipment failure represents the event causa in 46.8 parcant of all avants, but only 32.4 parcant of avacuation avants. (Exhibit 2.2-6). o Steady state condition is the in-plant event context in over half of all avants (51.0 parcant) and avacuation avants (56.0 percent). Start-up conditions ara raportad as the in-plant event context three times more frequently for avacuation avants (4.5 percent) than for all avants (1.8 parcant), while loading and unloading is raportad half as frequently (9.1 parcant) in avacuation events as in all avants (16 parcant). (Exhibit 2.2-7) 2-16 VW 000005318 r- 't DRAFT 31 August 1988 o Disposal locations ara responsible for 83.6 percent of the amounts released in in-plant evacuation events, compared to 6.0 percent for all in-plant events. The large share of disposal locations in in-plant evacuation events is attributable to a single large quantity release. (Exhibit 2.2-8). o Fire and/or explosion, which is the attributed cause for 5.8 percent of the amount released in all in-plant events, accounts for 84.6 percent of the amount released in in-plant evacuation events. (Exhibit 2.2-9). In-Transit Events o Truck and rail nodes are equally involved in evacuation events -- 49.3 percent of evacuation events Involve truck nodes and 46.3 percent involve rail. This is approximately the same percentages as their shares for all events. (Exhibit 2.2-10). o Derailment is the attributed cause in 24.7 percent of the evacuation events, which is approximately four times sore frequent than derailment's share of all events (6.5 percent). Collisions are the attributed cause in 21.3 percent of the evacuation events, occurring alsost twice as frequently as colli sions's share of all events (11.7 percent). (Exhibit 2.2-11). o Rail-associated releases account for two-thirds of the quantity released in in-transit evacuation events. However, rail represents only 22.1 percent of the quantity released in all in-transit events. (Exhibit 2.2-12). 2.2.3 Types of Facilities Involved o Of the 855 events reporting evacuations, 65.4 percent occur at fixed facilities. (Exhibit F-l) o Industry sectors were identified for the responsible parties in 78 percent of the 855 events reporting evacuations. (Exhibit 2*2-13) 2-17 VVV 000005319 DRAJT 31 Auguc 1988 - I o Distributors (25.8 psrcsnt) and primary producsrs (24.8 psrcsnt) account for ths majority of svacuations. End ussrs (14.2 psrcsnt) and sscondary producsrs (12.4 psrcsnt) account for ths naxt largest shares. Waste-related activities account for only 1.2 percent of the evacuation events. (Exhibit 2.2-13) o compared to their frequency of involvement in all events, distributors and end users have a slightly higher frequency and primary producers a significantly lover frequency of involvement in evacuation events. The rates for secondary producers and vasts related are slightly lover for evacuation events than for all events. (Exhibit 2.2-13) o At primary producers, 4.9 percent of the events reported evacuations, .8 percent at secondary producers, 8.2 percent at distributors, and 9.8 percent of the events at endusers. (Exhibit 2.2-14). o on average, more people are evacuated in a distributorrelated event than for any other sector. Distributors account for 25.8 percent of the evacuation events, resulting in 38.8 percent of all people reported evacuated. In contrast, end-userrelated evacuation events tend to involve a smaller number of evacuees, with the 14.2 percent end-user share of evacuation events contributing to 7.1 percent of all reported evacuees. (Appendix D) o Of the 298 events in vhich the types of evacuees can be identified as vorJcer or resident, residents on average account for 91.9 percent of the number of people evacuated and workers account for 8.1 percent. In the end-user and secondary chemical producer sectors, a greater than average share of workers are evacuated, 20.0 percent and 37.2 percent respectively, while distributors report almost no workers evacuated (1.0 percent). (Exhibit 2.2-15). 2-18 VVV 000005320 fL DRAFT 31 August 1988 2.2.4 Qiemteals Participation o The ten chemicals most frequently reported in associa tion with evacuation events account for up to 42 percent of such events (Exhibit 2.2-16) - chlorine is involved in the most (93) events, rep resenting 10.9 percent of evacuation events. (Exhibit 2.2-16) - Hydrochloric acid, ammonia (anhydrous), and sulfuric acid are involved in 79, 68, and 31 evacuation events, respectively. Together, they account for 21 percent of such events. (Exhibit 2.2-16) - The six chemicals next most frequently reported in conjunction with evacuation events account for 11 percent of such events. (Exhibit 2.2-16) o Chlorine is the most frequently reported chemical associated with evacuation events at primary producers, secondary producers, end users, and waste related; and ranks fourth at distributors. Chlorine accounts for 18 percent of the evacuation events at primary producers; 11 percent of the events at secondary producers? 4 percent at distributors, 17 percent at end users, and 40 percent at waste disposal. (Exhibits F-7, F-9, F11, F-13 and F-15) o After chlorine, the next five most frequently reported chemicals at primary producers are hydrochloric acid, ammonia (anhydrous), sulfuric acid, nitric acid and ammonia; together they account for 66 events, 31 percent of the 212 events occurring at primary producers. (Exhibits F-6 and F-7) Because any given event may Involve more than one chemical, there can be overlap in the numbers of events, deaths, and injuries when discussing the participation of more than one chemclal. Therefore, all percents or combined totals of events, deaths or injuries in this section should be interpreted as upper bounds 2-19 000005321 DRAFT 31 August 1988 e At secondary producers, the three most frequently reported cheaicals, in addition to chlorine, are ammonia (anhydrous) , anonia, and nitric acid; together they account for 21 events (20 percent of the total) at secondary producers. (Exhibit 7-9) o At distributors, ths three most frequently reported cheaicals are hydrochloric acid, ammonia (anhydrous), and sulfuric acid. Together they account for 53 events (24 percent of the total) at distributors. (Exhibit 7-11) o The average quantities released per event are shown in Exhibit 2.2-16 for the 21 cheaicals representing the largest number of evacuation events. Among evacuation events, methyl alcohol releases on average involve the largest quantities (1,000,000 pounds), followed by butadiene (170,000 pounds), and sulfuric acid (130,000 pounds). o When the quantities released in evacuation events are compared to the reportable quantities (RQs) of the participating cheaicals, 26.7 percent of the evacuation events in which quantities are reported involve quantities less than the RQ, i.e., the release of such quantities usually would not be considered of sufficient concern to justify its report to the National Response Center or other jurisdiction. (Exhibit 7-4) o Only 19.5 percent of the evacuation events involve chemical release qualities greater than 100 times the RQs of the involved cheaicals. (Exhibit 7-4) 2.3 . MOST SEVERS EVENTS INVOLVING ATMOSPHERIC RELEASES 07 ACUTELY TOXIC OR EXPLOSIVE CHEMICALS 2.3.1 General Characteristics o Air releases of toxic cheaicals that led to two or more deaths are listed in Exhibit 2.3-1. Thirty-seven such events were identified as having occurred during the five-year period froa 1982 t*j!986, or an average of seven events per year. These events in.q$al were reported to have led to 135 deaths. o Ifee most serious single event with respect to mortality was an explosion and fire associated with gun powder and fire works that led to 21 deaths in June 1965. 2-20 VVV 000005322 DRAFT 31 Augusc 1988 o Three quarters of tha moat savara avants rsporting fataiitiaa vara limited to two or three deaths, and only tvo (fiva parcant) of tha avants lad to ton or more deaths. o Events involving forty or more injurias ara listad in Exhibit 2.3-2. Forty-six such avants vara idantifiad as having occurrad during tha fiva yaar pariod from 1982 to 1986, or an avaraga of nina avants par yaar. Thasa avants lad to 4,411 injurias. o Tha most sarious singla avant with raspact to injuries reported was a vapor release of aldicarb oxime in August 1985 to which 430 injuries vara attributed. o Nearly thraa quarters of tha most severe evants involving injuries had 90 or favor reported injurias, vhile only four avants (nina percent) lad to 200 or aora injuries. o Evants involving 2,000 or aora evacuees ara listad in Exhibit 2.3-3. Thirty-seven such avants vara reported during the fiva yaar pariod froa 1982 to 1986, repraaanting an avaraga of seven par year. These avants lad to a reported 190,575 total evacuees o Tha aoat sarious avant vith raspact to number of evacuees vas an explosion and fire associated vith tha release of 22,500 pounds of phosphorus, reportedly causing tha evacuation of 40,000 people in July 1986. o Kora than tvo-thirds of tha aost severe evacuation evants involved. 4,000 or fever evacuees, vhile thraa evants (eight percent) lad to 10,000 or aora evacuees. o Transportation avants played a large role in tha most severe evacuation events, accounting for nearly half (17) of tha 37 evacuation avants. In contrast, transportation evants represent only 29 percent of tha avants in tha entire AHE/DB. Furthermore, transportation avants did not represent a large proportion of tha death and injury avants, accounting for 13 percent of tha avants vith deaths and 37 percent of tha events involving injurias. (Exhibit E-19) Tha most severe transportation avant vith fatalities involved four deaths. 2-21 vvv 00000^323 DRAFT 31 August 1988 I e The number of the most severe events involving death, injury, and avaeuation events, shown in Exhibits 2.3-1 -to 2.3-3, appaars to hsva raaainad relatively constant ovar ths 1982 to 1986 pariod, axcapt that oora avants vara raportad in 1985. Tha apparant incraasa in 1985 say hava baan a rasult of graatar public intarsst and more comprehensive raporting of such avants in tha immediate aftermath of tha Bhopal tragady rathar than an actual incraasa in tha occurranca of accidants. Tha number of savara incidants listad in Exhibits 2.3-1 to 2.3-3 is too fav and tha time frame too short to provida a raliabla basis for avaluating thasa typas of trands. 2.3.2 Significance of Evant End Effacts o Tha most significant factor in tha determination of tha savarity of avants involving fatalitias appaars to ba. vhathar a firs or axplosion occurred. Most (86 parcant) of tha avants in which tvo or nora deaths occurred (listad >ia Exhibit 2.3-1) involved firs and/or explosion. o Five (14 percent) of tha avants listad in Exhibit 2.3-1, occurred in situations vhara chemical toxicity appeared to ba tha primary factor in determining evant severity, i.e., vhara spill or vapor release occurred but fire and axplosion did not. In all cases, thasa toxicity-dominated avants vara restricted to tvo or fever deaths. o on the other hand, toxicity appaars to ba a somewhat more important contributing factor affecting tha severity of avants involving injuries. For injury events, toxicity was indicated to ba tha primary factor in 26 (57 parcant) of tha most severe events, and to account for three of the four avants leading to 200 or more injuries. o Fire and/or axplosion appaars to play tha most significant role in determining the severity of evacuation events, accounting for 26 (70 percent) of the 37 evacuation events listed in Exhibit 2.3-3 and all of the most severe events, each of which lad to reported evacuations of 10,000 or more people. 2-22 vvv 00000532^ DRAFT 31 August 1988 2.3.3 Seal# Domestic Releases of gxtrwptlv Toxic Chemicals Potential toxic effects vary with the toxicity and the volume of substance released. Three thousand four hundred fifteen events in the AHE/DB have information on both quantities and acute toxicity. For those events, we divided quantity released by the IDLH value attributed to that chemical to establish an approximate relative measure of the potential toxic effects of each release. Exhibit 2.3-4 lists the 29 events in the AHE/DB having largest quantity/IDIii values. o By this measure, 29 events involved releases whose quantity-toxicity ratio was greater than half that associated with the release at Bhopal, and 1? events involved releases whose ratio exceeded that at Bhopal. o Among the 29 high-damage potential events in Exhibit 2.3-4, one event reported deaths (five fatalities), five events reported injuries (ranging from one to 650 people injured), and five reported evacuations (up to 20,000 people) -- a somewhat greater frequency of severe consequences than that observed among all events. The average numbers of deaths, injuries, and evacua tions also were somewhat greater than the AHE/DB subgroup of events reporting deaths, injuries or evacuations. However, given the vast amount of materials released in the events listed in Exhibit 2.3-4, the frequency and severity of these effects is surprisingly lover than might be expected. o The difference between the potential for human harm and the actual consequences in these events appears to be due in part to combinations of the following factors: - The chemical's physical state at the time of release was liquid and much of the material remained liquid, 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 material* released, or other equipment or procedures to mitigate the extent of release and population exposure. 2-23 vvw 00OOO53?-5 DRAFT 31 August 1988 - Evacuation or other community response may have removed potentially exposed victims. o 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 that may overstate the actual quantity released. 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, including 40,000 classified as serious. o The most serious event of this type to occur in the u.s. is the explosion of a fertilizer ship in lexas City in 1947. The death total of 552 associated with this aaonium. nitrate related event places it vith only three other events listed in Exhibit 2.3-5 (in addition to Bhopal) where the death toll exceeded 500. o Toxicity acted as the primary factor in only two of the events. 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 event related primarily to toxicity factors in which a large number of people have died. o The average number of fatalities for ths tsn evsnts listed in Exhibit 2.3-5 is nsarly 600, far largsr than any of ths U.S. svsnts from ths AHE/DB listsd in Exhibit 2.3-1 for ths 1982 to 1986 period. The largest numbsr of fatalitiss among ths lattsr svsnts is 21. o Am important factor contributing to ths total numbsr of dsaths in Bhopal was the lack of warning and instructions in propsr rseponse givsn to ths populacs ovsr ths period of release. In Seveso, Italy hundreds of animals vsre killed because they 2-24 VVV 000005326 DRAFT 31 Augusc 1988 could not b avacuatad affactivaly. Tha long taro affact of tha avant vaa tha contamination of tha soil and tha abandonaant of a larga land araa. o Thus, tha data in Exhibit 2.3-5 indicata that avants of this nagnituda hava to data baan associatad aithar with situations in which massiva axplosions hava occurrad or in which toxic matarials hava baan allowad to raach tha surrounding populaca without a tiaaly or affactiva rasponsa by tha public officials. 2-25 000005327 DRAFT 3! Augusc 1988 3.0 CHARACTERISTICS OP ALL EVENTS 3.1 EXTENT OP DAMAGE o injuries ara reported in 9.2 percent of tha 11,0-48 vnti in tha AHE/DB. (Exhibit 2.2-1) o Daaths are reported in only 1.1 pareant of tha avanta. (Exhibit 22--2) o Evacuations ara raportsd in only 7.7 pareant of tha avants; hovavar, in 46 pareant of tha avants it vas not statad whether evacuation vas required. (Exhibit D-l) o A total of approximately one-half million people have bean evacuated in all avants which recorded tha number of evacuations. This total excludes non-specific descriptions of number of people evacuated, such as "several people" or "all employees." (Exhibit D-2) o In 29S events, accounting for 335,323 evacuees, it vas possible to distinguish batveen tha evacuation of residents and workers. These evacuation events involved a total of over 308,000 residents and 27,000 workers. (Exhibit D-2) o When evacuations are reported, an average of 1,109 people per event ara evacuated. Tha number of people evacuated ranges from 10 people to 20,000 people* (Exhibit D-2) o Property damage vas reportad in 13.6 percent of all events* In 43.1 percent of the events property damage is stated not to occur, and in 43.4 percent of the events it is not stated whether such damage occurred. Estimates of property damage are usually sxprsssed in very general terms; they range up to $100 million for one event. (Exhibit D-l) 3-1 vvv DRAFT 31 August 1988 . J 1. o Similarly, lass than three percent of tha n,c*8 events involva environmental damage, while one-third are reported not to lnvolva environment damage. In tha majority of avanta (64*2 parcant), it was not atatad whathar environmental damage occurred. (Exhibit D-l) 3.2 CIRCUMSTANCES OF RELEASE o Of tha XI,048 avanta, 71 parcant occur at a fixad loca tion; tha remainder occur during transit. (Exhibit 2.2-3) o Spilla ara raportad as an and affact in 68.0 parcant of tha avanta; vapor ralaasa in 39*6 parcant; fira in 6.9 parcant; and axploaion in 3.0 parcant. Tha severity of and affacts vaa estimated to ba (in dacraaaing ordar) axploaion, fira,v vapor ralaasa, spill, and unknown. Nota that multiple and-effects may ba raportad; tharafora, percents do not total to 100 parcant. (Exhibit 2.2-4) o Of tha 7,756 avants for which quantity' released was raportad, tha 3.0 parcant of avants ralaasing mors than 100,000 pounds accountad for 88.1 parcant of tha quantity ralaasad. (Exhibits 3.2-1 and D-6). 3.2.1 In-Plant Evants o account evants. Storaga vassals, process vassals, and valves or piping for approximately equal aharaa of tha 7,850 in-plant (Exhibit 2.2-5) o Storaga vassals account for 76.0 parcant of tha quantity released in in-plant avants. However, whan deaths or injuries ara involved, "vehicles not in transit" accounts for 50.4 parcant of tha quantity ralaasad, with waste disposal and pollution control equipment accounting for another 35.2 parcant. These large shares of tha D/I avants ara attributable to single large events in each case. (Exhibit 3.2-2) o Equipment failure is identified as the cause of 46.8 parcant of all events occurring in-plant, but accounts for only 32.2 parcant of tha quantity raportad released in-plant, in comparison, operator error is tha attributed cause in 18.2 3-2 VVV 000005329 DRAFT 31 August 1988 percent of the in-plant events, accounting for 45.7 percent of the quantity released. For D/I events, sparking and lightning account for 2.1 percent of the events, but over 50 percent of the quantity released. The large share attributable to sparking and lightning is due to a single large event. (Exhibits 2.1-4 and 3.2- 3) o In-plant events primarily occur under steady state conditions (51.0 percent) or during loading or unloading operations (16.0 percent). Operational configuration is not reported for one-quarter of the incidents. (Exhibit 2.2-7) o Less than six percent of the in-plant events account for over 92 percent of the amount released. Those events greater than 100,000 pounds in release size (3.2 percent of in-plant events) account for over 89 percent of the quantity released. (Exhibits 3.2-1 and D-6) o Events involving releases over 100,000 pounds account for the majority of material released for all in-plant locations except valves/pipes, heating/cooling, other, and unknown locations. (Exhibits 3.2-4 through 3.2-9) o The majority of events listing equipment failure, upset conditions, fire/explosion, high temperature and pressure, power related, sparking and lightning, other, and unknown causes release less than 1,000 pounds. The majority of events listing operator error, arson and vandalism, and disposal related activities as causes release less than 10,000 pounds. (Exhibits 3.2- 10 through 3.2-18) o Events involving releases over 100,000 pounds account for the majority of material released for all causes except disposal related, power related, and unknown causes. (Exhibits 3.2-10 through 3.2-18) o Over 80 percent of the material released in all in- plant event contexts, except other and unknown categories, are from events releasing 50,000 or more pounds. (Exhibits 3.2-19 through 3.2-23) 3-3 WV 000005330 DRAFT 31 August 1988 ' ;l 3.2.2 In-Translt Evsntt o In-transit events account for 28.9 avanta. (Exhibit 2.2-3) percent of tha o slightly lass than half of all in-transit avsnts (48.6 parcant) involva trucks* anothar 43.5 parcant involva rail cars. (Exhibit 2.2-10) o Of tha ralaasas of hazardous materials during transportation, 46.2 parcant ara tha rasult of a leak, 11.7 parcant ara tha rasult of a collision, 6.9 parcant ara from ovartumad trucks, and 6.5 parcant ara from derailments. (Exhibit 2.2-11) o ovar half of tha avsnts for aach transit mods, .. axcapt pipalinas, involva ralaasas of lass than 1*000 pounds. Tha majority of avants occurring in pipalinas ralaasa ovar 1,000 pounds. (Exhibits 3.2-24 through 3.2-28) o Tha quantity raportad ralaasad in avsnts occurring in pipalinas accounts for approxiaataly half of tha total quantity ralaasad in in-transit avants. However, tha bulk of this was contributad by a few larga pipalina spills; otharvisa, pipalinas ara rasponsibls for only a small fraction of ralaasa avants and quantities. Tha quantities raportad ralaasad in rail and truck avants respectively account for 22.1 and 18.8 parcant of tha total quantity raportad ralaasad in in-transit avants. (Exhibit 3.2-29) o Events involving ralaasas ovar 100,000 pounds account for almost all of tha quantity raportad ralaasad in pipalina and barge avants. In rail avants, incidents involving ovar 50,000 pounds account for ovar 85 parcant of tha quantity raportad released; while for trucks, ralaasas of ovar 50,000 pounds account for more than 40 parcant of tha amount of material ralaasad. (Exhibits 3.2-24 through 3.2-28) o Ovar 70 parcant of tha in-transit avants involving rail ralaasa las# than 1,000 pounds, over half ralaasa 100 pounds or lass, and 23.6 parcant ralaasa 10 pounds or lsss. (Exhibit 3.225) 3-4 VVV 000005331 t f* DRAFT 31 August 1933 o Ho event involving "other" transit modes (usually alrplanss) has a rsportsd quantity of ovsr 25,000 pounds. (Exhibit 3.2-28) o For all in-transit svsnt causss sxcspt ovsr-tumsd trucks, ths majority of aatarial is from svsnts rslsasing ovsr 100,000 pounds. For ovsr-tumsd trucks as ths causs of rslsass, approximately 50 psrcsnt of ths matsrial is rslsassd in svsnts with rslsass sizes ranging from 1,000 to 50,000 pounds. (Exhibits 3.2-30 through 3.2-33) 3.3 TYPES OF FACILITIES INVOLVED o industry ssctors wars idsntifisd for ths rssponsibls partiss in 90.6 psrcsnt of ths 11,048 svsnts in ths AHE/DB* Ths rsportsd rssponsibls partiss rangs from primary producsrs to endussrs, with 39.6 psrcsnt of ths svsnts occurring in thm-Chemicals and Allied Products and Petroleum Products (Primary) sectors. Ovsr twenty-four psrcsnt of the events occur in ths trans portation/distribution sector, 13.9 psrcsnt in ths secondary producer sector, 11.2 psrcsnt in end-user applications, 1.4 psrcsnt in waste-related activities, and 9.4 psrcsnt unknown. (Exhibit 2.2-13) o Primary producsrs account for a greater percentage of rsportsd dsaths and injuries than all other industry ssctors. Distributors account for ths largest share of evacuees. (Exhibits D--8, D--11, D-14, D-17, and 0-20) o .While secondary producsrs account for only 14 psrcsnt of all svsnts, they rslsass ovsr 40 psrcsnt of ths total rsportsd quantity in ths AHE/DB. Distributors and end-users, on ths other hand, have twice as large a share of svsnts as they do of quantity rslsassd. 3-5 VVV 0QGQG5332 DRAFT 31 August 1988 : ft. 3.4 CHEMICALS PARTICIPATION e The tan chemicals most frequently reported in associa tion with AHE/DB accidental release events are involved in up to 43 percent of such events (Exhibits 3.4-1 and 3.4-2) - Sulfuric acid is involved in the most (1,058) events, representing 9.6 percent of all events; five deaths and 380 injuries occur in these events. (Exhibits 3.4-1 and 3.4-2) - Chlorine, ammonia (anhydrous), and hydrochloric acid are involved, respectively, in 765, 644, and 536 events, together rspeesenting 18 percent of such events and accounting for seven deaths and 2,304 injuries. (Exhibits 3.4-1 and 3.4-2) - The other six chemicals (sodium hydroxide, methyl chloride, phosphoric acid, ethylene oxide, toluene, and vinyl chloride), most frequently reported in conjunction with AH2/DB events are involved in 16 percent of such events; 20 deaths and 685 injuries occur in these events. (Exhibits 3.4-1 and 3.4-2) o Sulfuric acid is the most frequently reported chemical associated with events at all facility types except waste related. Sulfuric acid is involved in 11 percent of the events at primary producers; ll percent of the events at secondary producers; and 9 percent of the events at distributors. (Exhibits D--9, 0-12, and D-15) o After sulfuric acid, the next four most frequently reported chemicals at primary producers are chlorine, ammonia (anhydrous), methyl chloride, and ethylene oxide; together they account for 1,219 events, 28 percent of the 4,371 events occurring at primary producers. (Exhibits D-8 and D-9) a Because any given event may involve more than one chemical, there can be overlap in the numbers of events, deaths, and injuries when discussing the participation of more than one chemical. Therefore, all percents or combined totals of events, deaths, or injuries in this section should be interpreted as upper bounds. 3-6 VVV 000005333 DRAFT 31 August 1988 o At secondary producers, the four moat fraquantly raportad chemicals, in addition to sulfuric acid ara chlorina, ammonia (anhydrous), hydrochloric acid, and phosphoric acid; together thay account for 341 avants, 22 parcant of tha 1,538 avanta occurring at secondary producers. (Exhibit D-12) o At distributors, tha four most frequently raportad chemicals, in addition to sulfuric acid, ara sodium hydroxide, hydrochloric acid, ammonia (anhydrous), and phosphoric acid. Together thay account for 623 avants, 23 parcant of tha 2,711 avants occurring at distributors. (Exhibit 0-15) o Tha average quantities ralaasad par event ara shovn in Exhibit 3.4-2 for tha tvanty chemicals participating in tha largest average number of events. Methyl alcohol releases involve tha largest quantities (100,000 pounds), folloved by sulfuric acid (86,000 pounds), and sodium hydroxide (81,000 pounds) o Whan tha quantities released in events ara compared to the reportable quantities (RQs) of the participating chemicals, 31.3 percent of the events in which quantities are reported involve quantities less than the RQ, i.e., the release of such quantities usually would not be considered of sufficient concern to justify its report to the National Response Center or other jurisdiction. (Exhibit D-5) o Only 12.6 percent of the events involve chemical release quantities greater than 100 times the RQs of the involved chemicals. (Exhibit 0-5) 3.5 GEOGRAPHIC PATTERNS The geographic distribution by state of all events, death events, injury events, and evacuation events are shovn schemati cally in Exhibits 3.5-1 through 3.5-4. It is important to note three caveats with regard to the patterns shovn in these exhibits: o This existence of high numbers of incidents in some states may be a result of effective enforcement by those states of the requirement that such incidents be reported; lower numbers of incidents in other states may be a result of less aggressive enforcement. 3-7 Wv 0000053 34 DRAFT 31 August 1988 - I ft o The distribution of svsnts by stats also reflects ths geographic emphasis of the data collection process, which 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 comprehensive scrutiny. o We would also expect the geographic distribution of events to be dependent upon the level of manufacturing activity in the state. That is, states with extensive manufacturing activity, hence chemical activity, are likely to have higher numbers of incidents. Exhibit 3.5*6 shows that California, Texas, New York* Illinois, and Michigan all had manufacturing values of shipments exceeding 120 billion dollars in 19SS. Exhibits 3.5-1 to 3.5-4 show the following: o Texas is the only state with more than 930 events. Ohio, New Jersey, California, and Louisiana have the four next largest numbers of incidents. (Exhibit 3.5-1) o New York and New Jersey have the largest number of events involving deaths, and Texas has the third largest number of death events. (Exhibit 3.5-2} o New Jersey has the largest number of incidents invol ving injuries. California has the second largest, followed by Texas, Louisiana, Illinois, Ohio, and New York. (Exhibit 3.5-3) o The trend for evacuation events is similar to that for injury events. New Jersey leads in number of events involving evacuations, followed by California. (Exhibit 3.5-4) o For each of the spatial distributions, the majority of states fall in the category representing the lowest frequency of events: 42 states have fever than 310 total events, 44 states have fewer than five death events, 43 statss have fewer than 40 injury events, and 43 states have fewer than 33 evacuat4rn events. 3-8 VVV 000005335 r i A< DRAFT 31 August 198* exhibit 2.2-16 DISTANCES MOST FREQUENTLY REPORTED IN EVACUATIOM EVENTS SANK 1 2 3 4 5 6 7 8 9 10 11 12 13 14 IS ----^16 ' 17 18 19 20 21 SUBSTANCE NAME CHLORINE HYDROCHLORIC ACID AJSOtIA (ANHYDROUS) SULFURIC ACID NITRIC ACID AASCNIA VINYL CHLORIDE HYOROGEN SULFIDE SCO!UN HYDROXIDE STYRENE TOLUENE 01 ISOCYANATE ETNYLENE OXIDE PHOSGENE TOLUENE PHOSPHORIC ACID POLYVINYL CHLORIDE BENZENE SULFUR DIOXIDE FORMALDEHYDE BUTADIENE METHYL ALCOHOL S OF EYAC EVENTS * ALL SUBS TOTAL QUANTITY RELEASED (Lba.) AVEIAQS QUANTITY RELEASED (Lba.) ** LARGEST QUANTITY RELEASED (Lba.) 93 38,126 53,000 1,200 20,000 79 49,693 1,500,000 52,000 1,125,000 6S 12,201 450,000 12,000 160,000 31 13,055 1,600,000 130,000 1,300,000 21 17,142 150,000 17,000 135,000 21 3,156 180,000 20,000 14 17,666 150,000 30,000 112,500 150,000 12 3,022 1,700 330 1,000 12 5,070 41,000 8,100 30,000 12 17,250 160,000 26,000 11 4,959 16,000 5,400 82,500 15,000 11 9,725 350,000 80,000 202,500 ID 1,139 100 SO 90 10 2,450 47,000 16,000 30,000 10 18,760 300,000 100,000 . 300,000 9 18,250 00 9 4,800 20,000 9,800 0 12,000 S 630 00 0 8 2.414 340,000 120,000 466,390 8 3.962 470,000 170,000 93,750 8 3,278 2,000,000 1,000,000 1,875,000 * All subatancaa art eoHRodity cfcaaicala. ** Thi* inelixSa* ail avsnts in iMticft tba nistmu ippaira as a primary, aacsndary, or tartiary siatanca. Avarass qjant1t1as art ecaputad uain* only thoaa avsnta that raportad quantitiaa. VVV 000005336 DRAFT 31 August 1988 WV 000005337 Exhibit 2.3-1 AIR RELEASES Of IONIC CHEMICALS LEADING 10 1U0 OR MORE PLAINS <1902 1906) BATE .......................... SUBSTANCES RELEASED.......................................END EFFECTS - SUBSTANCE I SUBSTANCE 2 SUBSTANCE 3 1ST 2ND 3A0 ATM NUMBER NUMBER NUMBER Of Of OF INJURIES DEATHS EVACUEES 0 04-Apr-06 01 May 66 I3-Jtav06 27 -May06 17 Jan-06 06 Jul 06 00 Jun 06 GUNPOWDER ACEIALDCHTDE UNKNOUN POLYVINYL CHLORIDE TOLUENE OI1SOCVANAIE AfMCMIlM DICHOfMATE TRICNIOBOEIRVLENE BUTADIENE ? ' fORMALDEHTDE NEINTLENE CHLORIDE EX FI UU UU SP EX FI VR EX FI UU UU FI IX VR UU EX FI VR UU SP Vt UU UU SP EN FI VR 000 0 4 200 4 30 020 10 2 0 2S 2 0 5 2 2,000 2S-Jun-03 20 Nay OS GUNPOLAER/IIKUORKS GUNPOWDER F EX fl UU UU FI EX UU UU OS Dec OS' NVOROGEN 1 NAPHTHA EX FI VR UU 26 Apr 03 POTASS l(M '||l| SODIUM NITRATE EX UU UU UU 29 Nay 03 AMMONIA 20 Oec 09^ XYLENE TM TOLUENE EX UU UU UU METHYL ETHYL FI EX UU UU 10 Apr 0? TOLUENE ] 06 Feb 03 NIIROCLVCERINE EX FI UU UU EX UU UU UU 10 Jul 03 MEIHYL ALCOHOL VARNISH FT EX UU UU 19 Jan 03 SULFUR FI EX SP UU 04 Apr 03 ACE1YLENE Jt, FI EX UU UU 03 Mar 03 NTDROFLUORlC MfLID SP n UU UU I1' ' >1 0 u. 1 t 5 21 0 090 43 3 0 to 3 0 10 3 0 7 S0 1 30 020 1 2 100 720 020 020 ft H -s: < o a o o o g-1 *s>> > DATE tUOSTAMCE 1 Exhibit 2.3-2 AIK RELEASES Of TOXIC CHEMICALS LEADING 10 FORTY 00 NODE INJURIES (1902 1906) SUNSIANCCS RflEASEO SUOSIAMCE 2 SUOSIAMCE 5 END EffCCIS 1SI 2ND MO AIN UMBER HLMSEO NUMBER OF OF OF INJUIES DCAINS EVACUEES OSJulOO 27MarA6 26-Ai-M 2SApr06 21 fab06 22 Ju(B6 16JulS6 )OHiyU PNOOPNIMm WMOCMLOtIC ACIO ACRYLONITRILE UNKMOUN CHLORINE ethylene glycol SULFURIC ACIO UMCNGUI EX fl VI so 400 0 40,000 SP VI Ul Ul IIS 0 0 IX Fl VI Ul 111 0 500 fl M uu Ul 90 0 0 SO M uu uu 71 0 740 SO VR Ul uu 65 0 so VR uu uu SO 0 0 M UU uu Ul 40 0 5,000 11 Aug-05 ALD1CAM OXIME NEIHHENE CHLORIDE VI uu U1 Ui 450 0 3,100 21 Jan 05 OIMETNCMTE V* SP uu UU 200 0 0 26- Jin BS PA0A1MI0N PARAQUAT CASOLIHE VI Fl EX uu 154 0 2,000 21 Mar-05 POLYVINYL CHLORIDE VI Fl UU UU 90 1 0,250 27-Jun OS 20-Apr-OS NAtATNION MEIMVl ETHYL KEIQME PARAQUAT PHOSGENE MEIMVL OAQHIOE HYDROGEN CYANIDE VI Fl uu uu Fl VI uu uu 90 0 1,000 00 0 74 os sp as UNKNOUM M UU Ul uu 72 0 260 >16JulB5 POLYVINYL CHLORIDE NVMOCNLOIIC ACID Fl VI uu uu S6 0 10,000 26-Jul as CMIOHNE VI uu uu uu SS 0 0 15 Apr OS BENZENE SULFURIC ACID LXUNIUN ER fl Ml uu SI 0 0 19-Apr-OS POLVCNLOIIMAIEO PUSUC METHYL ETHYL KETONE FT VI SP uu so 0 so DIPHENYLS 22 Nov BS 04 AugOS os o*c as 50-Sep-aS 27 Aufl as U Mar OS SILVER CVANIOE EXPLOSIVES HTOSOCEN CANBON MONOXIDE UIANIUN OXIDE ACIVLIC RESIN NAPN1NA EX VR uu uu Fl EX uu uu EX Fl Ml Ul Ml UU UU UU SP w Ul uu VI SP UU Ul 46 0 200 45 0 4.SOO 45 s 0 42 0 a 40 1 0 40 0 0 ' DRAFT 31 Auguoc 1988 i '*1 DRAFT 31 A u g u s t 1988 W v 000005339 MIC Exhibit 2.3-1 (continued) AIR RELEASES Of IQIIC CHEMICALS LCADINO IQ |UO Of MORE REAMS <1962 - 1906) SUSSTANCE 1 SUSSTANCES RELEASED.......................................END EFFECTS -* SUSSTANCE 2 SUSSTANCE S 1ST 2ND 3RD 4M NUN0CR NLMSER NUNKI OE Of Of INJURIES DEATHS EVACUEES 20 0ac04 IS-At^AL 1lNeyA4 16At-04 22-AurSL D-DmM SOLVENT ACETONE RUSTICS COAL 1AR SPECIALTY OILS SULFURIC ACIO NITRIC ACID POLYSTYRENE LISNT OIL URETHANE 27-Nay A3 D4DecA3 2S-Jan AS 2A-Aug-A3 16 Nov-AS 03 Oct A3 21 Jul-AS HlrnAS EXPLOSIVES PETROLELM PtOOUCIS EXPLOSIVES SODIUM NITRATE EXPLOSIVES METHYL MOM IDE ME THAME MEINVL ISOAUTYL KETONE INSECTICIDE MALATMION CHLORINE 2-NIIROPROPANE ANIIFREE2E VR FI EX UU CK FI VR Ui FI UU Ul UU ER FI Ul UU FI Ul UU UU EX FI LAI UU i IX FI liU UU FI EX vty.uu EX FI UF ' UU EX FI IN l uu FI EX (JjlWkJU EX FI JKu EX FI kTuu VR UU 10 * UU 10-Nay 02 02-Jrn A2 13 Jan-02 14 Jan A2 NITROGEN NI1RIC ACIO HYDROGEN SULFIDE LMKNOIM FIVE YEAR TOTAL U P' t. V UU UU, .UU IN ft vbI^kp urlu VO UU EX Ft uu^Iaj .i 73 0 1 3 100 03 21 2 12S 02 0 1 11 0 4 s 1.600 04 0 10 4 0 26 2 173 S20 920 020 320 M 2 S.DOO IS 2 0 11 2 S50 324 13S 9.01! End Effect Codes: SP - spill; VR - vapor release; FI - lira; EX - explosion; UU - uiknMi ' DKAFT 31 Auguac 1968 VVV ^ 0 0 0 0 5 3 4 0 DATE SURSTANCE 1 Exhibit 2.3 - 2 (continued) AIM RELEASES Of TOXIC CHEMICALS LEADING TO TOfciY 08 MORE INJURIES (1902 ' 1906) SURSTANCES RELEASE!)................................ SURSTANCE 2 fURSIANCE ) - - EDO EFFECTS 1ST 2ND 3R0 AIN NUMM8 NLMBER NUNRER Of Of Of INJURIES DEATNS EVACUEES 22-Rc-02 1JS*>i2 12-Nov-02 Il-OcC-tt 02-AA-02 O2-JW-02 TBWVUCRHAU SORILM NVRRQSULFIDE ETMTLENE ONIPE CNL08INC NITRIC ACID PMORCENE n-oct-02 22-Oct-02 VINTL REN2ENE IRICMLOtOElNTLEHE FIVE TEAR IOIAL PARAQUAT flilOROCARROW TOLUENE 01ISOCYANATE ORTNENE SP VR uu UU VR fl VR UU SP VR UU UU SP VR UU UU EX fl VR SP VR SP Ul uu SP VR UU uu SP VR UU uu 3SS 100 T7 n 64 60 SI 40 4,411 0 400 0 1.S00 0 1.000 1a 2 S.000 10 0 550 0 300 12 93.224 End EffocC Codn: SP ' ipiU; Mi - vapor toImko; FI - lira; EX apio* ion; Ul unknown il RELEASESA lt Of TOXIC CHEMICALS LEADING 10 K B IT OR MORE INJURIES 0 9 8 2 - 1966) II DRAFT 31 Augusc 1988 H 3 58 3i iSS :5315S 31 is* 1*1 aa 333 33333 3 3 f# 3 3 3 3 3=S 5 3a 3S353? 5a32aa s e 33 33 33 3= :5 n 2a a* M9la* 2 u * 1 8 222 222222 22 22 5 i k I I 11 5 a 4j Ii 2 SO^ N^ mP OO W ^ ^ MOl S2 VVV GQG00534-L DRATT 31 August 1988 VVV 0 0 0 0 0 5 3 4 2 Exhibit 2.3-3 All RELEASES or (Otic CKNICALS LEADING TO IU0 TM0USAN0 OB MORE EVACUEES (1982 - 1986) BATE NUMER NUMBER NUMBER ....................................... SUBSTANCES IUEASCD .......................................................... END EFFECTS -*- OF Of OF UMIAK* 1 SUBSTANCE 2 SUBSTANCE I 1ST 2B0 5ND 4TN INJURIES MAINS EVACUEES OS-Jut St OSAia-M SR-May-tt OO-JuvRB PHOSPHORUS CHLORINE iMwnuM BUTADIENE FORMA1 BCHVOE IX FI VR SP 400 0 40.000 FI VR UU>'a,UU VR UU uw Juu 14 0 6.000 40 0 5.000 SP EX FI lt l 5 2 2,000 >16JulB5 POLYVINYL OR.OHM NYBROCNLORIC ACID uk UU FI VI S6 0 10.000 *21-Nar-BS POL WINTL CNLOBIM VR FI UU UU 90 1 0,250 22* Jui-BS 27-Nov-BS ME INTI BROMIDE LIHURON PBOSPNATE AMUNILM NITRATE FI VI 3p- FI EX 12 0 7.500 IS 0 5.025 04'AuB'tS EXPLOSIVES FI EX UF- UU 45 R 4.500 09-Jwi-BS 2S-NM-K VINYL CNLOBIM NYBROCNLORIC ACID CALCINED ALUMINA NYDROFLORIC ACIB SULFURIC ACIO EX FI J" EX FI 0 0 4,000 21 0 S.SOO 10'An K ETHYLENE ONIDE BUTYL ACRYLATE CK FI yu R 0 5,500 11*AuB'8S ALBICAM MINE NEIRVLENE CHLORIDE VR UU W UU 4SO 0 5.100 16-Mm* RS TOLUENE 01 ISOCYANATE IX FI VR UU 25 0 5.000 10-AuB'S) CHLORINE VR UU UU UU S 0 5,000 04-Sp-OS NYBROCNLORIC ACID SP VR UU UU 0 0 2,000 26-Jui-8$ RAAAINION PARAQUAT GASOLINE VR FI EX UU ISA 0 2,000 29-Mm"8S OCTYL ALCOHOL SP FI III uu 0 0 2.000 H-llirM 29-Dac*ft4 31-Oac-S4 01-JuvM IO Aim* M SULFURIC ACIO VTffffflUtl ETHYLENE MIOE NYBROCNLORIC ACID MEINTL ALCOMOL COPPER SULFATE PLASTIC V AMMONIUM PERSULFATE . SP VR IR| uu SP VR UU UU SP VB uu uu SP VI UU.j|U SP FI urlKu 21 0 3.500 R D 2,500 C 0 2,500 4 0 2,000 0 0 2,000 L V f