Document 1Gnbvq9bbg7L91ga9mJZkgqq

<I r I * 4 i. %f +\ I ANALYSIS OF LOSSES IN THE CHEMICAL INDUSTRY AS IT APPLIES TO THE POLYVINYL CHLORIDE INDUSTRY * C. W, Conaway, J. A. Davenport (Speaker), G. P. Norstrom Industrial Risk Insurers SUMMARY: This paper details an analysis of losses to Industrial Risk Insurers in the Chemical Industry for the period 1978 through the first ten months of 1980. This analysis shows that the type of plant that is the most susceptible to loss is the type which predominates the PVC Industry. Industrial Risk Insurers is an Association of 45 leading insurance companies. IRI has a substantial stake in the Chemical Industry in general and the PVC Industry in particular. Our liability in force is approximately $500 billion Approximately $63 billion cr 14% of this liability is in the Chemical Indus try. According to an article by Cameron, et al^, the annual capacity of PVC producers in the United States is about 7.2 billion pounds. IRI now insures 3.7 billion pounds of this capacity, or 51%. IRI recently did an analysis of the losses in the chemical and allied classes for the years of 1978, 1979, and the first nine months of 1980. During this period, there were 967 occurrences which produced a loss to Industrial Risk * Insurers of $116,336,882. It is important to note that during this period there were no catastrophic losses -- that is, a loss greater than $20 million. PERIL ANALYSIS In every analysis that we have ever run of losses in the chemical industry, explosions have been shown to be the most expensive type. In the analysis * A Paper delivered to the Vinyl Chloride Safety Association Meeting at the Holiday Inn, St. George, Bermuda, October 23, 1980. * VAB.0001035293 r : . P-i'w tlHfr I- !-' - 1" -tpii-V.' . r t' V / * of the 1978-1980 losses, explosions represented almost 602 of the dollars H from only one quarter of the occurrences. (See Table 1) # OCCUPANCY ANALYSIS The next thing we looked at in the loss analysis is the occupancy class where the losses occurred. (See Table 2) The extra heavy hazard chemical class is that involving organic peroxide and explosive manufacture, nitra tions, Claisen and Grignard reactions, and other very vigorous and thermally unstable chemical reactions. The petrochemical occupancies include processes utilizing petroleum hydrocarbons or similar feed stocks. Olefin facilities predominate this occupancy class. The heavy hazard class is one of the lar gest classes and includes most polymerization processes like PVC plants, as well as solvent extraction, sulfonation, low pressure hydrogenation, and other processes involving flammable and combustible materials. The light hazard class involves primarily inorganic operations. The remaining classes are self-explanatory. The table shows that the bulk of the dollar loss is in the petrochemical and heavy hazard class. The figures show that there is a 422 chance that a loss will occur in these two classes and 75% of the dollar loss was in these two classes. One item of note is the excellent loss record of the pharmaceutical class. This experience should be tempered with the understanding that firms in this class usually purchase their insurance with a lower deductible than that of firms with predominantly petrochemical or heavy hazard chemical opera- H tions. Therefore, the higher frequency is to be expected. -V 1 r - r ft w.n - ;i tf wiB.-ff V?!- -n4* < r. ' tfT^- r v r - ' '*' h VAB.0001035294 -1.'.* Ml t ........................................^ t -3- SIZE OF LOSS ANALYSIS Table 3 shows an analysis of the size of loss. Here it is noted that 50.7% of all occurrences reported to the IRI resulted in no dollar loss while a mere 1.3% of all occurrences result in loss greater than $2.5 million. Clearly, major losses in the chemical industry can be characterized as having relatively low frequency but enormous dollar loss associated with them. It has been suggested that this data is support for Murphy's Law. That, is* if you examine the trend, ever-decreas ing frequency produces an ever-greater loss. This would logically lead us to the point of view that the loss which will probably never happen will always be large. LARGE LOSS ANALYSIS Next, we examined a typical six-quarter segment of this three-year period. (See Table 4.) This was an in-depth analysis of larger losses; that is, occurrences which have produced dollar loss greater than $100,000 combined * i property damage and business interruption. In this analysis we attempted to isolate the critical factors which are of importance to safety and pro perty conservation engineers. We specifically examined explosion and fire losses from the point of view of the type, cause, location, and key contri buting factors which are associated with these more severe occurrences. Notice that in this case the explosion occurrence tracks our other analyses. LARGE EXPLOSION LOSS ANALYSIS We first examined the type of loss for explosion losses greater than $100,000. (See Table 5.) It is noted here that the explosion losses involving chemical reaction developed 44.2% of frequency and 85.9% of total VAB.0001035295 . I --j" H-I r a +*!.* --n-+w 4- . i- .- - j. 4 - ".I i+ ' -r . w - T A -4- dollar loss. Fired equipment explosions produce another 23.5% of frequency and all "other" explosions comprise the other 32.3% of occurrence. However, the dollar loss associated with these occurrences produce only 14.1% of dollar loss. For those of you who are familiar with the IRI's vapor cloud theories, be assured that if a major vapor cloud explosion were to have occurred during this period, the data would be slanted towards this "other" category. The work that we have done on vapor cloud explosions has indicated that none of the PVC plant releases with subsequent explosion could be classed as an unconfined, free-air vapor cloud explosion. Vie feel that the amount " of material released has not been sufficient to produce an unconfined cxplosion outside the reactor building. As reactors get bigger, this probability is increasing. Examining the chemical reaction loss cause analysis (see Table 6), it is noted that accidental and uncontrolled chemical reactions are the causes in 73.3% of all cases. Recalling that chemical reactions are responsible for 44.2% of all major explosions, it is apparent that about one-third of all major explosions in the chemical industry will be due to either accidental or uncontrolled chemical reaction. "Chemical reaction-accidental" is the term for an incident caused by the accidental contact of materials. "Chemi- cal reaction-uncontrolled" is the intentional combination of materials which -4 become uncontrolled as the result of agitation or coolant loss, a contami nant, instrument failure, etc. We now examine where in the plant this type of incident, an explosion caus ing damage greater than $100,000, would be more likely to occur. (See Table 7). We see that the odds are approximately two to one, or that 66.7% of all losses will occur in an enclosed process building as opposed to an open .. . I, . . r-4^ p ^^ i I. y :+ -i-l- 1. IT .. .1- > VAB.0001035296 T- 11- process structure or other areas, such as a tank farm, laboratory building, etc. In this enclosed building, the type of process which will probably be involved will be that of a batch chemical reaction as contrasted to a continuous chemi cal reaction system or other type of unit (Table 8). The last area we will examine as respects major explosion losses will be that of contributing factors (See Table 9). Needless to say, in any given loss, one or more contributing factors may be involved in determining the magnitude of loss. Often, the si2e of the loss is proportional to the number of contri buting factors. Accordingly, we will only identify the four major contributing factors which we have found to be involved in large explosion losses during this period. The first and most major factor is the rupture of a vessel or equipment, or the mechanical failure of piping, valves, agitators, etc. These failures would be at pressures below the design pressures. Three other ele ments which find their way into 20% of all losses during this period are the w human element or carelessness, inadequate venting resulting in vessel over pressure, and, finally, the improper location of equipment or congestion. LARGE FIRE LOSS ANALYSIS Even though explosions caused the greatest loss, it is still important to examine the fire losses. Table 10 is a breakdown of the causes of fires in the large loss category. The principal cause is that of a flammable liquid or gas release or overflow. The next major cause is the overheating of hydro carbon materials, followed by the failure of a pipe or fitting, static elec- * tricity or electric sparks, electrical or mechanical breakdown. Oust as with VAB.0001035297 * -6- the explosion losses, we must keep the process materials in the pipes or vessels. As in the case of the explosion peril, we have attempted to place the location in the plant where such incidents will occur (Table 11). We find that, in contrast to the explosion incident, the frequency of loss in an enclosed process building versus an outdoor process structure is approximately equal in the case of major fire losses. This would lead us to the conclusion that leaks occur in process equipment outside of buildings as often as inside. This leak inside of a building is, how ever, much more likely to cause an explosion. Examining the major contributing factors (Table 12), we see that the lack of automatic sprinkler protection is the key factor which has been involved in almost 40% of these fire losses. The basis of the founding of the old Factory Insurance Association as well as the Factory Mutuals was the insuring of risks which were adequately protected by automatic sprinklers. Still, today inadequate sprinkler protection is a major con tributing factor in our fire losses. CONCLUSIONS > To summarize the conclusions from this analysis: 1) Taken together, explosion and fire losses provide the greatest frequency and severity of incident in the chemical industry.2 2) The explosion loss is of much more severe character than that of the fire loss. h A VAB.0001035298 f js * i f! i r- t ' i i y : 4 -p: t -v-; :: !: it iT 'f--| I- r -t 'f- .. k. <i I 4 h 4 -7- 3) Major explosion losses will be most likely caused by either accidental or uncontrolled chemical reaction. 4) Major explosion losses will occur in enclosed process buildings involving batch chemical reactions. 5) The magnitude of loss will be contributed to by the rupture of vessels or equipment. 6) In the case of major fire losses, the cause will be due to the release of either a flammable liquid or gas and will occur v/ith equal frequency in enclosed buildings or outdoor process struc tures. 7) The principal contributing factor to major fire loss will be the lack of automatic sprinklers or water spray. Industrial Risk Insurers believes that to reduce losses in the Chemical Industry a vigorous loss prevention program that has the full backing of top management is vital. IRI has published a booklet (2^whiich out lines our "Guiding Principles for Protection of High Hazard Chemical and Petrochemical Plants." A LITERATURE CITED 1) Cameron, J. 0., A. J. tundeen and J. II. McCulley - "Trends in Sus pension PVC Manufacture," Hydrocarbon Processing, March 1980.2 2) Industrial Risk Insurers, Loss Prevention and Protection for Chemical and Petrochemical Plants, Harcroni, Connecticut J I1-.' <it = ` rMI-*'! ' l"h 'I- 4'.............................'* -fr " S ` VAB.0001035299 | |. i- r- q- . . ..................... -1 J'f TABLE 1 LOSSES IN THE CHEMICAL INDUSTRY IRI CHEMICAL AND ALLIED CLASSES 1978 - 1980 PERIL ANALYSIS PERIL Explosion Fire Windstorm All Other ' % OCCURRENCES 23.6% 42.4Z 14.3% 19. 7% 1007: % $ LOSS 58.27. 26.3% 8.8% 6.77o 100% TABLE 2 LOSSES IN THE CHEMICAL INDUSTRY - XEX CHEMICAL AND ALLIED CLASSES 1978 - 1980 OCCUPANCY ANALYSIS OCCUPANCY CLASS 63 - Extra Heavy Hazard 64 - Petrochemicals 65 - Heavy Hazard 66 I - Light Hazard 67 - Paint, Dyestuffs, Inks 68 - Soaps and Veg. Oil 69 - Pharmaceutical & Fine Chem. - { OCCURRENCES 4.8% 9.3% 32.6% 2 2.8% 10.5% 6.07. . 14.0% 1007. % $ LOSS 4.6% 27.4% 47.7% 11.67. 4.3% 1.0% 3.47. 100% VAB.0001035300 -9- 4 TABLH 3 LOSSES IN THE CHEMICAL INDUSTRY - IRI CHEMICAL AND ALLIED CLASSES 1978 - 1980 SIZE OF LOSS ANALYSIS SIZE 4 No Claim $ 0-$ 9,999 $ 10,000 - $ 24,999 $ 25,000 ~ $ 49,999 $ 50,000 - $ 99,999 $ 100,000 $ 249,999 $ 250,000 - $ 499,999 $ 500,000 - $ 999,999 $1 ,000,000 $2 ,499,999 $2 ,500,000 & Over % OCCURRENCES 50.7% 14.6% 9.4% 6.1% 6.1% + 4.0% 4.4% - 2.0% > 13.27, 1.5% 1.37.-- % $ LOSS 0 i.4% 1,.3% 1.,7% - 3..4% 4, 6% 12,.9% 11..7% > 93.27. 17,.0% 47 .0% _ TABLE 4 LOSSES IN THE CHEMICAL INDUSTRY IRI CHEMICAL AND ALLIED CLASSES LARGE LOSS*ANALYSIS TYPICAL 6-QUARTER PERIOD TYPE OF I.,OSS ' % OCCURRENCE Explosion 51.4% Fire 33.3% Windstorm 1.5% All Other 13.8% 100% 4 *Loss > $100,000 net of deductible 7. $ LOSS 73.27. 20.1% .2% 6.5% 1007. A * :j . 1m' r. H O*: ' i' '.pi j: *- VAB.0001035301 ..... h -10- TABLE 5 LOSSES IN THE CHEMICAL INDUSTRY IRI CHEMICAL AND ALLIED CLASSES EXPLOSION LOSSES > $100,000 TYPE ANALYSIS TYPE OF LOSS Chemical React!or: Boiler Explosioix Furnace Explosion Explosion - Other r1 f* nVj 44. 27. 17.77. 5.87, O'* . J0*/7o 7. $ LOSS 85.97. 5.37. .87. 8.07. 1007. A 4 f'ALE 6 LOSSES 1M Till; CHEMICAL INDUSTRY IRI CHEMICAL AND ALLIED CLASSES MAJOR CHEMICAL REACTION LOSSES $100,000 CAUSE ANALYSIS CAUSE Chemical Reaction (Aceidratal) Chemical Reaction (Uncontrolled) Decomposition of Unstable Materials Other 7, NO. OF LOSSES 33.3%v >73.37. 40.07. x 13.37. 13.47. 1007. . VAB.0001035302 -11- w TABLE 7 *- LOSSES IN THE CHEMICAL INDUSTRY P IP.I CHEMICAL AND ALLIED CLASSES CHEMICAL REACTION LOSSES > $100,000 LOCATION ANALYSIS LOCATION i^i > Enclosed Process Buildin Open Process Structure Other % OF NO. OF LOSSES a. 66. Tl 20.07. 13.27. 100% * t ** . TABLE 8 LOSSES IN THE CHEMICAL INDUSTRY IRT. CHEMICAL AND ALLIED CLASSES CHEMICAL REACTION LOSSES $100,000 OCCUPANCY ANALYSIS OCCUPANCY mi ill Chemical Reaction Process - Batch Chemical Reaction Process - Continuous Recovery Unit Evaporation Unit Other 7. OF NO. OF LOSSES I laP^wiPi I M 11 60% 13.6% 6.6% 6.6% 13.2% 1007. VAB.0001035303 7k .1* V * r ! f ifi H1Clq. \\ %- A, 1 " - - f-1' - 4 TABLE 9 LOSSES IN THE CHEMICAL INDUSTRY IRI CHEMICAL AND ALLIED CLASSES CHEMICAL REACTION LOSSES> $100,000 MAJOR CONTRIBUTING FACTOR ANALYSIS CONTRIBUTING' FACTOR HIP IP^l I >111 I -- Rupture of Vessel or Equipment or Mechanical Failure Human Element - Carelessness Inadequate Venting Improper Location of Equipment or Congestion % OF NO. OF LOSSES 40.0/o 20.07. 20.07, 20.0% TABLE 10 LOSSES IN THE CHEMICAL INDUSTRY IRI CHEMICAL AND ALLIED CLASSES FIRE LOSSES $100,000 CAUSE ANALYSIS CAUSE % OF NO. OF LOSSES Flammable Liquids or Gas Release, Overflow 28.5% Overheating 24.3% Failure of Pipe or Fitting 14.3% Static Electricity or Electric Spark . 9.5% Electrical or Mechanical Breakdown 9.4% All Other 14.0% 100% -13- TABLE 11 + LOSSES IN THE CHEMICAL INDUSTRY IRI CHEMICAL AND ALLIED CLASSES FIRE LOSSES > $100,000 LOCATION ANALYSIS LOCATION Enclosed Process Building Outdoor Prccess Structure Other % OF NO, OF LOSSES 38.5% 33.3% 28.2% 100% A * TABLE 12 LOSSES IN THE CHEMICAL INDUSTRY IRI CHEMICAL AND ALLIED CLASSES ' FIRE LOSSES > $100,000 MAJOR CONTRIBUTING FACTOR ANALYSIS CONTRIBUTING FACTOR Sprinklers Needed Human Element Vapor Laden Atmosphere Excessive Residue or Deposits % OF NO. OF LOSSES 38.1% 19.0% 14.3% 9.5% t.... .....................' J ... ... 'T*' : r,*1 un**! K ft SV. '.bqQJfc W*1T!' fW"*:-'- * VAB.0001035305