Loading...
HomeMy WebLinkAboutMINUTES - 09161997 - C54 TOI BOARD OF SUPERVISORS CONTRA COSTA COUNTY .Qo FROM: William Walker, MD '�. � Health Services Director Os�'4 C0927' DATE: September 4, 1997 SUBJECT: Report on Hazardous Materials Incidents, Notification, and Complaints SPECIFIC REQUEST(S)OR RECOMMENDATION(S)&BACKGROUND AND JUSTIFICATION Recommendation: ACCEPT a report by the Health Services Department on Hazardous Materials Incidents, Notifications, and Complaints, 1989 through 1996. Background: In August, 1996, the Health Services Department prepared a Report on Hazardous Materials Data: Incidents, Notifications, and Complaints in Contra Costa County, 1989-1996, to assist in policy development concerning industrial safety. The Department has revised and updated that report to include data for all of 1996. The updated report also includes statistical analyses of the data with regard to trends in incidents and notifications over time. Fiscal Impact: None. CONTINUED ON ATTACHMENT: yes SIGNATURE -L=��') L RECOMMENDATION OF COUNTY ADMINISTRATOR RECOMMENDATION OF BOARD COMMITTEE APPROVE OTHER SIGNATURE(S): .1L ACTION OF BOARD ON "d �M�M•� �l� /997 APPROVED AS RECOMMENDED OTHER VOTE OF SUPERVISORS I HEREBY CERTIFY THAT THIS IS A TRUE UNANIMOUS (ABSENT ) AND CORRECT COPY OF AN ACTION TAKEN AND ENTERED ON THE MINUTES OF THE BOARD AYES: NOES: OF SUPERVISORS ON THE DATE SHOWN. ABSENT: ABSTAIN: Contact Person: Elinor Blake 370-5022(5) p ATTESTED CC: Health Services Director PHIL BATCHEL R,CLERK OF THE B ARD OF Hazardous Materials Department(via HSD) SUPERVISORS AND COUNTY ADMINISTRATOR BY: 1 DEPUTY r ,str, REPORT ON HAZARDOUS MATERIALS DATA INCIDENTS, NOTIFICATIONS AND COMPLAINTS IN CONTRA COSTA COUNTY 1989 - 1996 Sue Loyd Dena Hutchin Laura Brown Contra Costa County Health Services Department August 5, 1997 EXECUTIVE SUMMARY Contra Costa County is home to a large industrial base that includes five petrochemical refineries, a half dozen chemical plants, two power plants and numerous smaller facilities. Incidents involving hazardous materials have prompted the Board of Supervisors to initiate policy discussions concerning the need for improved safety. This report was prepared to assist in that discussion. An analysis was conducted of the Health Services Department data base of hazardous materials incidents,notifications, and complaints from January 1, 1989 through December 31, 1996. With regard to environmental reporting, the County has one of the most advanced notification policies in the nation,resulting in the reporting of releases that are often well below regulatory thresholds. Incidents were divided into two levels to differentiate the severity of the incident. Level I incidents are those that posed an imminent threat to the public health, the environment, or property, and required immediate action by the Health Services Department. Level II incidents did not pose an imminent threat to public health,the environment, or property. A notification is a report, usually made by a company, concerning a potential release, a release that did not go off- site, or a release that did not injure anyone. More detailed definitions are in the body of the report. Of some 7500 reports reviewed, about 2201 concerned industrial facilities. Thirty-seven were classified as Level I incidents, 102 as Level II incidents, 2008 as notifications, and 54 as complaints. Analysis of the data showed no apparent trend in the number of incidents; they seem to be neither increasing nor decreasing over time. Notifications increased, as we would hope-- the County has encouraged industry to notify even on events (and potential events) for which no other agencies require reporting. Our Notification Policy and industry's voluntary compliance have proven extremely beneficial. We attempted a comparison of our data with data from other areas of the county with a similar industrial profile. Unfortunately, such a comparison could not be made due to differences in the reporting structures and other issues concerning the state and federal data bases we reviewed.. Review of the Level I incidents showed no single initiating cause. The Health Services Department has requested information from the facilities about the initiating, contributing, and root causes of the Level I incidents. From the initial responses, it appears that most facilities have not conducted root causes analyses on past incidents. As of this writing, a number of facilities are beginning to conduct such analyses for Level I incidents. When root cause analysis becomes industry practice, valuable information will be available to improve safety. -i- INTRODUCTION Outside of Los Angeles County, Contra Costa County is home to the largest number of major industrial facilities in California. It is located on the northeastern edge of the San Francisco Bay and is ideally situated for heavy industry. The Delta's deepwater channel provides an important commerce link between the West Coast and Pacific Rim countries. A large industrial base is located along the shoreline of the Sacramento River, Carquinez Straits, San Pablo Bay, and San Francisco Bay. Contra Costa County is home to five refineries, a half dozen chemical plants, two power plants and numerous smaller facilities. Recent hazardous materials industrial incidents prompted the Board of Supervisors to initiate a policy discussion about the need for and nature of a new or revised County policy that would improve safety. This report was prepared to document the data which will assist in that discussion. The report is a review and analysis of the County Health Services Department database of incidents,notifications, and complaints concerning hazardous materials received by the County between January 1989 and December 1996. The industrial community in Contra Costa County must adhere to federal and state as well as local hazardous materials Iaws, regulations, and policy. California statutes are some of the most restrictive and safety-oriented in the nation. Considering the amount of hazardous materials stored and used in the county, safety requirements and environmental standards are imperative for a secure workplace and community. The following paragraphs briefly summarize the five federal, state, and local laws and policies governing hazardous materials handling and storage that are implemented by County agencies, primarily the Health Services Department. Numerous other regulations are implemented by other agencies,notably the Bay Area Air Quality Management District,the Regional Water Quality Control Board,the Department of Toxic Substances Control, Cal/OSHA,the State Fire Marshall (which regulates pipelines),the State Lands Commission, the US Coast Guard, and others. Hazardous Materials Emergency Planning and Community Right to Know Program The California hazardous materials emergency planning and community right-to-know law(also known as the Business Plan Program or the Hazardous Materials Release, Response and Inventory Program)covers some 1200 businesses in Contra Costa County, including the oil refineries and chemical plants. The purpose of the regulation is to protect the health and safety Page 1 of the public and of emergency responders, as well as to ensure the community's right to know about chemical use in the County. This regulation requires the training of all employees who may come into contact with hazardous materials. Training consists of hazardous materials storage and handling procedures, spill prevention,emergency response procedures, and clean-up procedures. All businesses storing more than specified quantities of hazardous materials must report an inventory of chemicals and submit an emergency response plan to the Health Services Department. Risk Management and Prevention Plan (RMPP)Program The goal of the California RMPP is to prevent the accidental release of acutely hazardous materials and to prepare for public protection in the event of a release. The RMPP evaluates _ potential-off-site consequences of the worst credible accident scenario for emergency response planning. Contra Costa County is one of the leading jurisdictions in implementing the RMPP. As of December 1996, fifty-one RMPPs are on file with the County. Federal Risk Management Plan (RMP) Program The Federal RMP evaluates flammable hazardous materials as well as some acutely hazardous materials. Under this program,the worst-case (rather than worst credible case) scenario for off- site consequences must be assessed. The Health Services Department anticipates responsibility for implementation of this program beginning in 1998; an integrated federal and state program will replace the current RMPP. Process Safety Management (PSM) The Federal Occupational Health and Safety Administration (OSHA) adopted PSM standards in 1992. Whereas the RMPP and RMP regulations are concerned with off-site impact to human health and the environment,the PSM standards consider employee safety. This federal regulation, and its more stringent state Cal/OSHA counterpart, establish procedures for process safety management that protect employees by preventing or minimizing the consequences of chemical accidents involving highly hazardous chemicals. A key provision of PSM is process hazard analysis - a careful review of what could go wrong and the safeguards that must be Page 2 . f implemented to prevent releases of hazardous chemicals. Employers must identify those processes that pose the greatest risks and evaluate those first. County Notification Policy In addition to the state and federal regulations governing hazardous materials, Contra Costa County maintains a notification policy adopted by the Board of Supervisors in 1992. The policy requires a facility to promptly notify the Health Services Department of a hazardous materials release or threatened release that poses a present or potential hazard to human health and the environment. The importance of this notification policy cannot be overestimated. The policy results in industry reporting very small spills, in addition to the required reporting of releases above federally-specified reportable quantities. This policy is critical to the effectiveness of the Community Warning System, which is one of the most advanced alert and notification systems in the nation. COUNTY DATA ON INCIDENTS, NOTIFICATIONS, AND COMPLAINTS The County's hazardous materials "incident files"are comprised of three distinct types of reports: Incident -An incident is an event involving the release or threatened release of hazardous materials occurring within the geographical limits of Contra Costa County, which may pose an imminent threat to the public health, the environment, or property, and would require immediate action. For the purpose of this report, incidents have been subdivided into Level I and Level II events. These levels will be described later in the report. Notification -Under the County's Notification Policy, a notification is a report (usually made by a company)that a release may occur, or that a release has occurred but did not go off-site, injure anyone, and was contained, or an activity will occur such as flaring or on-site emergency drills that could cause concern to citizens observing it. This notification would not require a response other than documentation and internal notification of the County's Incident Response Team. Page 3 CQmplaial- A complaint is information received by the County,usually from a private party, that involves past activities or activities which would be completed by the time an inspector could arrive on-scene. These activities may violate the hazardous materials handling, storing, processing and removal laws. Based on the information received the activity does not present an imminent threat to the public health,the environment, or property. Methodology Health Services Department staff reviewed the Contra Costa County Hazardous Materials Incident Report Database for January 1, 1989 through December 31, 1996. Over 7,500 database entries were examined. These entries were screened for selection of reports by the following industrial facilities: Chevron,Dow, DuPont, General Chemical (Bay Point and Richmond), Pacific, Rh6ne-Poulenc, Shell,Tosco, and Unocal. The two Pacific Gas and Electric Company power plants (Antioch and Pittsburg) and KemWater, formerly Imperial West Chemical (Antioch and Pittsburg) were also included because they were included in a report presented to the Board of Supervisors' Ad Hoc Committee on Industrial Safety by the Contra Costa Building Trades (July 29, 1996). The total number of reports for the selected facilities was 2201. Of this total, 2008 were notifications, 139 were incidents, and 54 were complaints. The data presentation below evaluates the incidents and notifications in more detail for the time period January 1989 through December 1996. The remaining 5300 database entries consisted of complaints, referrals, and clean-up requests about non-facility events such as illegal discharge of waste onto public property, drug labs, and neighborhood complaints. Incidents,Notifications, and Complaints Concerning Hazardous Materials Figure I shows the number of incidents,notifications and complaints for the selected facilities from January 1989 through December 1996. It includes additional events for the selected facilities not recorded in the database. These additional events are noted in Attachment 1 with an Page 4 a a �o N O R = � "L O� ao, � L n o. 00 C u � A co E C n� U C4 u N a L C t o a� R N wr R CD c� L O. V iz z •� N N S z � u - cc 0 a vi C = II E 1� Y i o F N i g� � I E c a a B P � u en jagmnN e V h e Incidents As mentioned previously, an incident is an event involving the release or threatened release of hazardous materials occurring within the geographical limits of Contra Costa County, which may pose an imminent threat to the public health,the environment,or property, and would require immediate action. For the purpose of this study we reviewed each file for content. As a result, incidents have been further divided into two levels to differentiate the severity of the incidents. Level I-These are incidents/events involving the release of hazardous materials occurring within the geographical limits of Contra Costa County which may have posed an imminent threat to the public health. the environment, or 1ranerty, and required immediate action by Health Services. These events may consist of explosions, large fires,releases of toxic or hazardous substances creating off-site impacts,or occurrences that resulted in serious worker injuries. Level II-These are incidents/events involving the release of hazardous materials occurring within the geographical limits of Contra Costa County,which did not nose an imminent threat to the public health.the environment, or property. These events may include small fires, releases of toxic or hazardous substances that may cause an odor impact but did not pose an imminent threat to the public health and safety, and occurrences that resulted in non-volatile spills to the environment. _Figure 2 presents the distribution over time of incidents reported in the County from January 1989 through December 1996. It includes the 139 incidents for the selected facilities and 3 additional events that occurred at facilities other than the selected facilities. These additional incidents are as follows: the 7/29/92 incident at Texaco Pipeline; the 2/1/96 incident at Air Products; the 8/22/92 incident at EIectro Forming. 'Two 1994 incidents responded to by staff and reported by an industrial facility are not included in Level 11 text or figures as those incidents did not involve hazardous materials. One involved algae in a rainwater puddle;the other,pollen on cars. Page 6 a a N NON � � L 'C = u o as A C� A cc C v o i I� I I i I i i to I i I r O vi O v+ O N N jagmnN Causes of Incidents Attachment 1 offers a closer look at Level I incidents. For each incident,the date, initiating cause, category of incident,and chemicals) involved are presented, as well as additional descriptive details. A good deal of information regarding root causes and contributing causes of incidents is not available for review because most facilities do not yet conduct root cause analysis. From Attachment 1,no single cause or trend can be identified as responsible for the Level I incidents that have occurred in our County since 1989. Notifications Contra Costa County received a total of 2008 notifications from the selected industries for the reporting period January 1989 through December 1996. A notification as defined in the County's notification policy is: "a report usually made by a company that a potential release may occur; or a release has occurred but did not go off-site, injure anyone, and was contained; or an activity such as flaring or on-site emergency drills that could cause concern to citizens observing it. This would not require a response other than documentation and Incident Response Team notification." Documentation of notifications began in 1990. However, companies were not required to notify until the policy was adopted in the fall of 1991. A review of the incident files for 1989 showed that thirty-four would have been notifications under the Notification Policy. As Figure 3 shows, most of the notifications listed in the database were releases. Please note that 1989 notifications are not included, as they were not documented. The types of events reported in Notifications are described below. Page 8 • r •p2�us`tut �, sauntut 00 vuei`d •a saMod �D .o uMou�tu(l u'1 N � on iopo .o r t � „ 1 U WIG �► �`, �{ �� aWPi b3IS �c to E 7 � N � r d OO � O D 0 O O Releases The most common type of notification is a release. For the purposes of this report the term "releases"includes small spills, leaks, and potential releases. Most of these releases occur on the industries' property,have no off-site impact,and are mitigated by trained crews. These releases result in negligible risk to the general public. The quantity/volume of most of these releases are below federal guidelines for reportable quantities but are reported to the County under the Notification Policy. Flaring Flaring is the second most common notification and includes normal flaring and the potential for greater than normal flaring. Flaring is a controlled and regulated safety mechanism designed to manage excess gas production by the complete combustion of the flared gasses to form primarily carbon dioxide and water. The Bay Area Air Quality Management District has established strict limitations on the contaminant levels in the gasses, such as sulfur and particulates. Industries maintain continuous stack analyzers to monitor compliance and are required to report exceedence to the Air District, Equipment Notification with regard to equipment issues is the third most common type of notification. It includes changes in equipment condition (e.g., a piece of equipment or unit being shut down or started up, equipment failing to operate, new equipment or unit being put into service), demolition of equipment, and equipment malfunctions with no associated consequences. Drills The fourth most common type of notification is the report of drills, including site drills, fire training, and site alarm testing. Page 10 Fire Fire notifications include small process and non-process fires such as insulation fires, electrical fires, grass fires, control room fires, and cafeteria fires. The major facilities maintain their own fire-fighting personnel for responding to these situations. Other "Other" includes notifications of steam leaks, cooling tower plumes,potable water leaks, and other items generally not associated with chemical releases. Odor Notifications of odor were reported to the County between January 1989 and July 1996. These notifications were related to on-site odors with no identified associated source. Unknown This category includes notifications for which the report was incomplete and did not state the nature of the notification. Power Notifications with regard to power outages with no associated release were reported. Power outages may affect the process units or process equipment. If the power outages resulted in an associated release, the notification was counted as a release and not as a power outage. Alarms The County also receives notifications for alarm activation not associated with a release or upset, such as false alarms. Injury Notifications of injury have been subdivided into those associated with worker exposure to hazardous materials releases ("Injury/Rel."on Figure 3) and those not associated with worker Page 11 exposure to hazardous materials ("Injury"). Please note that the County Notification Policy does not require notification of worker injuries to Health Services. Derailments A review of notifications regarding derailments at the facilities selected for review, reveals that the derailments involved rail cars that derailed or jumped the tracks, and were not associated with chemical releases. Summary of Notifications Figure 4 shows the subject of notifications by percentages for the selected industries. Of all notifications from January 1989 through December 1996,49% involved on-site releases. Reports of flaring comprised 16%of all the notifications. Reports of equipment issues and drills respectively, were 11%and 7%of the total. The remaining 17%involved fires, odors on-site, and other. In Figure 4, `other' includes alarm, derailment,worker injuries associated with exposure to hazardous materials, worker injuries not associated with exposure to hazardous materials, power, and unknown. Figure 4 Notifications by Percentage Selected Facilities Releases Drill 49% 7% Odor 4/o 0 Other 7% Flare Fire Equipment o 16/0 6% 11% Figure 2-Oth&Includes: Alarm, Derailment, Injuries/Rel., Injuries, Power, and Unknown Page 12 Complaints The 54 complaints indicated on Figure 1 are specific complaints relating to facility activities. None of the complaints related to Level I incidents; but some may relate to Level II incidents or notifications. STATISTICAL ANALYSIS OF INCIDENT AND NOTIFICATION DATA Two conventional statistical tests were employed to evaluate the data: a simple linear regression model (and correlation coefficient), and a goodness-of-fit test. A simple regression model considers the straight-line relationship between two variables, as depicted by the regression line. The correlation coefficient(r) is a measure of how well the linear regression fits the actual data. Figure 5 illustrates the strength of the linear relationship(r) between two variables. A goodness-of-fit test, in turn,compares a certain pattern with the observed data. (See attachments 2 and 3 for assumptions and analysis). Regression analyses were performed to evaluate the relationship between the number of incidents /notifications (y), and the chronology (x) inclusive of 1989 to 1996. The regression lines and correlation coefficients (r) for the number of Countywide incidents spanning the years under study (Figures 6, 7, and 8) are similar to those at selected facilities (Figures 9, 10, and 11). Countywide Level I incidents indicate a very weak positive correlation, whereas total incidents and Level II incidents show a weak negative correlation. In contrast,notifications display a strong positive correlation (Figure 12). The goodness-of-fit test, comparing the annual average number of total and Level I incidents between 1989 and 1996 with the actual number of incidents, suggests that the differences in observed and expected frequencies are small enough to be explained by chance. This is not true for the number of Level II incidents; the observed values do not"fit"the expected values, and the differences may not be entirely accounted for by random sampling error. Page 13 Figure 5 Correlation Coefficient (r) Perfect Positive Linear Correlation rel X Perfect Negative Unear Correlation jAq t +* *a »n �•.- s�' UN " X No Linear Correlation r-0 wr- i ♦ hV X Page 14 O. O. h a a O� ON Y�I ~ � u � Q a 40 u u u ea 0 i I i i cc � T O+ O V1 O snO v1 O ry N N jagwnu .o a� r d r M O� Q^ r �O � � � � � � W A � �• a �, 1 t� d � � � p w `�, t- d ,,. f. CA d � � �� .♦ 7. � i � i � �� .► w � rte•, � �i � � � ��1 ��� rn � � �� � � � 1 � �� �� � � � � fl 1�' � V � ' ��`�`' �1 � 1 �, �'` �, t, ,, cpi` t � ti ..- 1 � � ,� � � � t � � ,, ;� ' ` ;1 �`` `' }` ' t 1 '� � t � � ', `�, 1 ,1 1 �� �l � � ��� �� ���� � � ��,t �� �l p� � �. i r '`:` '�l � � � t'' 1 �., `ti .o �` oo r- 1a4��� ,✓ � 0 r r .1� r r rr d \ rn G o d 1 t \ \ ` 1 ` r a i r r N o� o� a �o — a� o, L u � ; •"" LRt o-o R � � I h N "p h T a� — I i i 00 o - M - jagwnx .a rr a v, er rn o, r•r L as as v C� M A � c t i i r � t •� i s y .i 1 I ' i y � i iaquinN �c o, c, h v o� o, as v as A O^ ON 0cCN^ L j 0" 4)� � ami on a 1 N L.y CA +.r w+ I tJ co � I � I I O I i NN -- jagounNI .a i C V` . I' Q. Q� e^' I N f' oa � R ,�,, a � � i '�' % �� y as °� W �'� a> �� � °' c? � .� pr +'' ` ` ` �1 W' � 1 is � � �� o � �` � � ,1 a� r ,� 0 '�, � o o a N O t*i �aAw°� �-- o DISCUSSION .Contra Costa County has one of the most advanced policies with regard to environmental reporting (see Attachment 4 for a discussion of other incident databases reviewed for comparison purposes). The County's Notification Policy requests reporting of releases that are often well below regulatory thresholds. About 2201 of the 7500 reports reviewed concerned industrial facilities. Of those 2201 reports, 40 were classified as Level I, and may have posed an imminent threat to the public health,the environment, or property; 102 were classified as Level II incidents that include events such as small fires or releases that may have caused smoke or bad odors but did not pose an imminent threat to the public health or environment. Of the 2008 notifications from selected facilities for January 1989 through December 1996, 49% involved releases that did not go off-site and caused no worker injury. Reports of flaring comprised 16%of all the notifications. Reports of equipment issues and drills respectively,were 11% and 7% of the total. Conventional statistical analyses of the data suggests that the relationship between the number of incidents and the years over which they occurred shows no apparent trend, i.e.,the total number of incidents appear to be neither increasing nor decreasing over time. The County's database includes information regarding the type of incident that occurred but rarely the initiating, contributing, or root causes of an incident. The Health Services Department has requested information from the facilities about initiating, contributing, and root causes of the Level I incidents. From the initial responses, it appears that most facilities have not conducted root cause analyses on past incidents. As of the date of this report, a number of Contra Costa sites are beginning to conduct root cause analyses for Level I type incidents. This is an encouraging sign. Once root cause analysis begins to be an industry practice valuable information will be created as County review of the 40 Level I incidents demonstrates that no single initiating cause is responsible. The County's Notification Policy, which resulted in the bulk of information used to prepare this report, requires a call to Health Services to report events for which no other agencies require reporting. When the policy was developed in 1991,the industrial community expressed concern that the number of notifications would be used inappropriately to suggest a level of risk to the Page 22 community that was not present. In addition,they were concerned that the policy would have a negative impact on their ability to conduct business. Nevertheless, industries have voluntarily complied, and we urge that the notification data not be used in a way that discourages continued voluntary compliance. The policy has proven extremely beneficial to county residents and to people who work and go to school here, as it provides our department with an early"heads-up" for events that might be of public health or community concern. We believe that this report provides an accurate portrayal of incidents,notifications, and complaints reported in Contra Costa County from 1989 to 1996. Page 23 Attachment 1 Do a > -COO 0 W = 0 0 Ch Op ia9 00 C: >� . co bo =0 at 0 go Eto 5 Q. Q6 s 9 . 2, 4v. S , .e t g I " =Q - -5 z zed: 40.3 boo IT c c 0 P. T 2 to -S E ta 2 E ot cr to :rd E aoc6 E rr ej as a to o > m bcas t7 not U E Q, fu 41 < C6. Go V00 00 00 % %n 00 Z cli v1on Cz ;:� to u o G >• s 3 r u G A N A � G 3 o L' a t' •%s ,�y .E ° � E .N o G o G � C :: Op V v o0 u .p �., nU � � r .� .N u C ,,.Q• +Vn ..• ✓ O d C Gi � � p � � `� Ci td P � G •O � r yc u p E O 'd p ✓ A p �j y „ E G '? ` u G w u ao 0Ro 63 c "ed u .o. ` u u O Qy G y G W O C 3 E •• �✓ A ' u "O C ari E p E C� N .� u v- 34 d v 3ipN p V ti'►' C u w 0 i O ✓A O A �: .u+ u 0 g . N �O pis v e � 2 E �o. o" fcu o05 a .p .- � � 0 s � y 3 ' o s s C D N ra ✓ Q U E s A u Via Co A A C ✓� u �, G E E �r, E •3 v y 9 O O O V 'D ��„ oh G r U A A { v v o N E � u u u o v tY W G C G E C+ � 'G '> �. r u S px•,• 'G`p ✓ amu^' . W . N d A ✓ Q,. d u y G• .0 " '� N O u ✓ J::IM u 4 i a •7 '�" d o d v u G �i+ �' u J v u op G• '� $ U G o 7 v N ♦ u 30 d 3 N `" � v +^ ♦ fT u •u � � s u N r � N � m 0 o � DO « t E .° 3 u b c a v y o uQ � 401� •►. ✓�, � a U ti D EVS " � O ilN U ... x ,� �„� N � 7�p � .b a^ (/N NNV a0 3 oo u u cam- a c a-o o m d vu .. G ° c co H G it c c o con o o sa x t tit,3 JD c o 7 S � to t U tai > C w m v c W E a G41i" IE w � U 0 pp A 4 u Cd C3. 10- OD 4= C u V LOD ti O C � u N ted U P P- .i u o y H e a u o o U G• o N u OD Mv, V L• C V V 'N .0 ed ro •t « . W G v. V v C 4 NCJ C p :� C C 7 C C .a O C i ..Tr vN, ° V r y.. •V CO 'a U a, 'te V ro V V r- 0 00 p, oGpp 1 > w w C .? v > ,Ci u =6 G G t) n to c0 d: v, o H c v� tic dE inM t. ed V to u to > f* M eel, o • O�G N MCT Oo N C> N as G •u✓.. N u chi OF w u Mml . O G V ^y u ..- � es t� .a b �G '^' w �s •� � ,. N E � ca u � o V ✓ a o s E 3 �- L' L' V es .� "= o ,o F� o o ° s ,A ° E u •d u u �,, o o N s u .' ,o s p �► E c v o0 .V+ v Z-0-00 3 O G °m O "t0 p V p ° G o V G ✓a4 uG G 61wap td VD " x ✓ G '�- uw 5 00 oEE " TyE L' u cr c a o- m es $ G r VS p. P w ✓o o 0. a ��'u a .4 on , �i p ° on to vto a a-o E u G c T � ?p O V oD p �'Q r'• O •,�,, a 0 4 x '� E •p •� r y o ✓ u u � u � til w I'° E G u .� ol G tT c ✓ U '' W oo E w o ° Nuo A tD p v U u � r V w ° .G a C p OD �' ✓ uu d m .p "N N G V O. ..- w c U •u `b' ou N � c o. y.a '°°o'o w 5 c C y o G y bR w O V d r u y ' i °✓ ✓ p $ 00 o • o N t c ol Lx- C4 < O N � N � A a ■ s © « % •_ . a$ 4 10 « _ # � 6 a } eeee % — c OJk � � S � t % R o � � « / � \ t © i « § 6 � g � 0 T-) e t � � � a . � t � % % k t � # % 5 t k ' co $ t o « � k t ƒ � > « « � . t CPO � p Attachment 2 Goodness-of-fit Tests Ho: The proportion of incidents are evenly distributed over the years of study (1989-1996). HI: The proportion of incidents are not evenly distributed over the years of study(1989-1996). Assumptions for the goodness-of-fit test: 1. The experiment consists of n identical repetitions/trials, 2. Each repetition consists of K possible outcomes (K>2), 3. The trials are independent, 4. The probability of the various outcomes remains constant for each trial. Where n= number of incidents/trials K = outcomes (1989 - 1996). Page A2-1 Total Incidents Countywide January 1989 - December 1996 Goodness-of-fit test Observed Expected (O-E)^2/E User Input 19 17.75 0.088028169 Deg_of Freedom 7 (No.of outcomes- 1) 20 17.75 0.285211268 Alpha 0.05 (Significance level) 17 17.75 0.031690141 Test Stat 9.774648 (From computations on the left) 18 17.75 0.003521127 Computed Values 14 17.75 0.792253521 Critical Value 14.067 28 17.75 5.919014085 P-value 0.2017077 12 17.75 1.862676056 14 17.75 0.792253521 142 9.774647887 Test Ha: The proportion of total incidents countywide are evenly distributed over the years of study. HI: The proportion of total incidents countywide are not evenly distributed over the years of study. Result Do not Reject Ho Not statistically significant at the 5%significance level. . . Page A2-2 Level I Incidents Countywide January 1989 - December 1996 Goodness-of-fit test Observed Expected (0.E)^2/E User Input 6 5 0.2 Deg_of Freedom 7 (No.of outcomes- 1) 1 5 3.2 Alpha 0.05 (Significance level) 5 5 0 Test—Stat 8 (From computations on the left) 9 5 3.2 Computed Values 4 5 0.2 Critical Value 14.067127 6 5 0.2 P-value 0.3325939 3 5 0.8 6 5 0.2 40 8 Test Ho: The proportion of Level I incidents countywide are evenly distributed over the years of study. Hl: The proportion of Level I incidents countywide are evenly distributed over the years of study. Result Do not reject Ho Not statistically significant at the 5%significance level. Page A2-3 Level II Incidents Countywide January 1989 - December 1996 Goodness-of-fit test Observed Expected (O-E)^2/E User Input 13 12.75 0.004902 Deg_of Freedom 7 (No.of outcomes- 1) 19 12.75 3.0637255 Alpha 0.05 (Significance level) 12 12.75 0.0441176 Test—Stat 14.39216 (From computations on the left) 9 12.75 1.1029412 Computed Values 10 12.75 0.5931373 Critical Value 14.067127 22 12.75 6.7107843 P-value 0.0446302 9 12.75 1.1029412 8 12.75 1.7696078 102 14392157 Test Ho: The proportion of Level I1 incidents countywide are evenly distributed over the years of study. H,: The proportion of Level II incidents countywide are not evenly distributed over the years of study. Result Reject Ho At the 5%significance level the results are statistically significant. Page A2-4 Total Incidents Selected Facilities January 1989 -December 1996 Goodness-of-fit test Observed Expected (O-E)^2/E User Input 19 17.375 0.1519784 Deg_of Freedom 7 (No.of outcomes- 1) 20 17.375 0.3965827 Alpha 0.05 (Significance level) 17 17.375 0.0080935 Test—Stat 10.58273 (From computations on the left) 16 17.375 0.1088129 Computed Values 14 17.375 0.6555755 Critical Value 14.067127 28 17.375 6.4973022 P-value 0.1578841 12 17.375 1.6627698 13 17.375 1.1016187 139 10.582734 Test Ha:The proportion of Level I incidents for selected facilities are evenly distributed over the years of study. H,: The proportion of Level I incidents for selected facilities are not evenly distributed over the years of study. Result Do not reject Ho Not statistically significant at the 5%significance level. Page A2-5 Level I Incidents Selected Facilities January 1989 - December 1996 Goodness-of-fit test Observed Expected (O-E)^2/E User Input 6 4.625 0.408783784 Deg_of Freedom 7 (No.of outcomes- 1) 1 4.625 2.841216216 Alpha 0.05 (Significance level) 5 4.625 0.030405405 Test—Stat 5.594595 (From computations on the left) 7 4.625 1.219594595 Computed Values 4 4.625 0.084459459 Critical Value 14.067127 6 4.625 0.408783784 P-value 0.5877998 3 4.625 0.570945946 5 4.625 0.030405405 37 5.594594595 Test Ho: The proportion of Level I incidents for selected facilities is the same every year. HI: The proportion of Level I incidents for selected facilities is not the same every year. Result Do not reject Ho Not statistically significant at the 5%significance level. Page A2-6 Level II Incidents Selected Facilities January 1989 - December 1996 Goodness-of-fit test Observed Expected (O-E)^2/E User Input 13 12.75 0.004901961 Deg_of Freedom 7 (No.of outcomes- 1) 19 12.75 3.06372549 Alpha 0.05 (Significance level) 12 12.75 0.044117647 Test_Stat 14.39216 (From computations on the left) 9 12.75 1.102941176 Computed Values _ 10 12.75 0.593137255 Critical Value 14.067127 22 12.75 6.710784314 P-value 0.0446302 9 12.75 1.102941176 8 12.75 1.769607843 102 14.39215686 Test Ho: The proportion of Level II incidents for selected facilities are evenly distributed over the years of study. H 1: The proportion of Level II incidents for selected facilities are not evenly distributed over the years of study. Result Reject Ho Statistically significant at the 5%significance level. Page A2-7 Total Notifications Selected Facilities ' January 1989 - December 1996 Goodness-of-fit test Observed Expected (O-E)^2/E User Input 34 251 187.6055777 Deg_of Freedom 7 (No.of outcomes- 1) 39 251 179.059761 Alpha 0.05 (Significance level) 91 251 101.9920319 Test—Stat 894.8048 (From computations on the left) 282 251 3.828685259 Computed Values 258 251 0.195219124 Critical Value 14.067 389 251 75.87250996 P-value 6.36E-189 486 251 220.0199203 429 251 126.2310757 2008 894.8047809 Test Ho: The proportion of notifications for selected facilities are evenly distributed over the years of study. H,: The proportion of notifications for selected facilities are not evenly distributed over the years of study. Result Reject Ho At the 5%significance level the results are statistically significant. Page A2-8 Attachment 3 Total Incidents Countywide January 1989 -December 1996 Figure 6 -Linear Regression 1989 19 1990 20 1991 17 1992 18 1993 14 1994 28 1995 12 1996 14 Total Incidents-Level I and Level II Countywide January 1989-December 1996 sn u x Is E 0 Z_. l0 S 0 IM9 19911 IMI IMJ 1197 119, IMS 19% Ynr Equation of a Linear Regression: Y=a+bx Equation for this Figure: Y=19.67+(-.5476)(x) Coefficient of Determination: r2=.07 Correlation Coefficient: r=-.27 Page AM Level I Incidents Countywide January 1989 - December 1996 Figure 7 Linear Regression 1989 6 1990 1 1991 5 1992 9 1993 4 1994 6 1995 3 1996 6 Level I Incidents Countywide January 1989-December 1996 w 9 E3 a Z 1 0 1989 19911 1991 1991 1"1 199 1995 1996 Ynr Equation of a Linear Regression: Y=a+bx Equation of this Figure: Y=4.67+(.0952)(x) Coefficient of Determination: r2=.01 Correlation Coefficient: r=.10 Page A3-2 Level II Incidents Countywide January 1989 - December 1996 Figure 8 Linear Regression 1989 13 1990 19 1991 12 1992 9 1993 10 1994 22 1995 9 1996 8 Level H Incidents Countywide January 1989 -December 1996 u m IS ` I y E 0 z ,o 3 0 1919 1990 1991 1993 1993 1991 1995 19% Ynr Equation of Linear Regression: Y=a+bx Equation of this Figure: Y=15+(-.6429)(x) Coefficient of Determination: r2=.09 Correlation Coefficient: r=-.31 Page A3-3 Total Incidents Selected Facilities " January 1989 - December 1996 Figure 9 Linear Regression 1989 19 1990 20 1991 17 1992 16 1993 14 1994 28 1995 12 1996 13 Total Incidents-Level I and Level 11 Selected Facilities January 1989-December 19% >b u 20 IS a Z 10 5 0 1929 1990 1991 1992 1991 1991 1995 19% Veer Equationof a Linear Regression : Y=a+bx Equation for this Figure: Y=19.5+(-.61)(x) Coefficient of Determination: r'=.08 Correlation Coefficient: r=-.29 Page A3-4 Level I Incidents Selected Facilities January 1989 - December 1996 Figure 10 Linear Regression 1989 6 1990 1 1991 5 1.992 7 1993 4 1994 6 1995 3 1996 5 Level I Incidents Selected Facilities January 1989-December 1996 s 0 z 1989 1990 1991 1991 1997 19% 1995 19% V=r Equation of a Linear Regression: Y=a+bx Equation of this Figure: Y=4.75+(.036)(x) Coefficient of Determination: r==.002 Correlation Coefficient: r=.05 Page A3-5 Level II Incidents Selected Facilities January 1989 - December 1996 Figure 11 Linear Regression 1989 13 1990 19 1991 12 1992 9 1993 10 1994 22 1995 9 1996 8 Level II Incidents Selected Facilities January 1989-December 1996 v m _ u E O 2 10 S 0 1919 1991- 1991 1991 1997 1994 1195 19% Yew Equation of a Linear Regression: Y=a+bx Equation of this Figure: Y=15+(-.643)(x) Coefficient of Determination: r2=.09 Correlation Coefficient: r=-.31 --Page A3-6 Notifications Selected Facilities January 1989 - December 1996 Figure 12 Linear Regression 1989 34 1990 39 1991 91 1992 282 1993 258 1994 389 1995 486 1996 429 Notifications,Selected Facilities January 1989-December 1990 E a Y aoo IOD g 0 Iw IRO IRI IRl IM IRI IR7 19% Year Equation of a Linear Regression: Y=a+bx Equation for this Figure: Y=6.417+69.881(x) Coefficient of Determination: r2=.91 Correlation Coefficient: r=.96 Page A3-7 Attachment 4 For comparison purposes,we attempted to compare data from our County with incident data collected in various State and National databases. We endeavored to compare not only the data compiled in this report but also data compiled in the databases we reviewed to other areas in the United States. Los Angeles, California,Texas City,Texas, and Baton Rouge, Louisiana were selected for comparison purposes because of the relative density of industry as compared to the density found in Contra Costa County. Unfortunately, we were not able to compare the data reviewed due to issues with the various databases; comparisons for our County with other jurisdictions is very difficult due to the fact that the reporting structure is not the same nation- wide. Below is a summary of the databases reviewed for comparison purposes. California Hazardous Materials Incident Response(CHMIRS) computer database system administered by the California Office of Emergency Services. The program collects and analyzes statistical data from state and local governmental agencies regarding accidental. hazardous materials releases. Releases include fixed facility and transportation incidents.- Reports ncidents:Reports are gathered from all responding agencies which tends to create duplication in the database for a single incident. Additionally, governmental agencies may not fill out a CHMIRS report for all such incidents. Emergency Response Notification System (ERNS) computer database system administered by the Environmental Protection Agency. The database contains information on reports of oil and hazardous substance releases that have occurred throughout the United States and have been reported to the National Response Center,the EPA Regional Offices, or the United States Coast Guard. In California, the CHMIRS database is also reported in the ERNS data. Reports in the ERNS database range from traffic accidents involving a drum of hazardous waste, spills of antifreeze in driveways, and releases from industrial sites. We have informally discussed our concerns regarding the ERNS database with the EPA (Mr. Dana Stalcup, USEPA in Virginia). Accidental Release Incident Prevention (ARIP)computer database system administered by the Environmental Protection Agency. The database is a sub-set of the ERNS database, but for fixed facility incidents. Follow-up questionnaires about the facility and circumstances of the incident are sent to the facility for completion. Data obtained from the questionnaires is then included in the ARIP. Page A4-] a