Managing Business Risks from Major Chemical

Transcription

1 Managing Business Risks from Major Chemica Process Accidents Mariana Bardy 1, Dr Luiz Fernando Oiveira 2, and Dr Nic Cavanagh 3 1 Head of Section, Risk Management Soutions Savador, DNV Energy Soutions South America 2 Head of Department, DNV Energy Soutions South America 3 Head of Safeti Product Management, DNV Software, London, UK Traditiona risk management has focused on effects on peope, particuary when using Quantitative Risk Assessment (QRA). Nowadays a much broader approach is being adopted as companies begin to consider potentia accidenta osses in their decision making process. This paper describes a study where business risks from major accident hazards were assessed for a arge chemica process company. Resuts are presented in the form of F-Cost curves, simiar to traditiona F-N curves, representing cumuative frequencies of events versus their cost. These are evauated for various types of oss incuding property damage, business interruption, inventory oss, environmenta oss, cean-up costs, ega costs, fataity and injury costs, amongst others. Adjusting insurance vaues according to risk is one of the benefits of such an anaysis. Evauating these risk criteria heps organisations to negotiate with their insurers to agree maximum insured oss vaues that wi fit their strategy and satisfy their risk acceptance criteria. Aso, vaues for civi iabiity insurance can be reviewed, based on the hazards that can resut in fataities and injuries, as we as property osses, for third parties. This paper presents the resuts of this study which invoved the evauation of the company s maximum insured osses. The key driver was a board eve initiative to impement a more objective approach to quantifying this metric which serves as the main basis for setting the eve of insurance premiums. The resuts of the anaysis are presented as F-Cost curves and we describe the risk-based decision making process and criteria adopted by the company to set the maximum oss vaue to be insured for their pant. In concusion, this paper presents resuts from a rea ife exampe of the appication of the quantitative evauation of business risks, iustrating that a financia or business risk approach can hep management in day-to-day decisions when considering possibe effects of major chemica process accidents. Introduction and Background Over the ast 30 years, the management of risks associated with the operation of major accident hazard faciities has been high on the corporate agenda. This has been driven by a number of major accidents incuding Fixborough (1974), Bhopa (1984), Piper-Apha (1988) and, more recenty, Enschede (2000), Tououse (2001), Fuxys (2004) and Texas City (2005). A these, and many more, have resuted in significant fataities and injuries. The most recent three exampes have resuted in a tota of more than 60 fataities and 3000 injuries. This has further driven egisation such as the Seveso directives in Europe and the

2 EPA Risk Management Pan reguations in the US (Fewtre and Hirst, 1998). Legisation has generay been focused on reducing the risk of fataities and injuries, and righty so. However, in this same period there have been many high profie accidents which have resuted in few, or even zero, fataities and injuries, but enormous cost to business, both of the operator and more broady in the vicinity of the pant concerned. Companies have suffered significant financia osses and entire countries have seen major disruption from singe incidents invoving reativey sma direct asset oss and sometimes no fataities. The reease of dioxin at Seveso, Itay, on 9th Juy 1976 resuted in no direct fataities. However, this incident required the evacuation and decontamination of a wide area north of Mian. Athough no immediate fataities were reported, kiogram quantities of the substance, which can be etha even in microgram doses, were widey dispersed. This resuted in an immediate contamination of some ten square mies of and and vegetation. More than 600 peope had to be evacuated from their homes and as many as 2000 were treated for dioxin poisoning. This was a key driver in changing the reguation of major hazard faciities across Europe through the so-caed Seveso directives, subsequenty brought to the statute books in a member states of the European Community. From 3rd February 1999, the obigations under the Seveso II Directive have been mandatory for industry as we as the pubic authorities of the Member States responsibe for the impementation and enforcement of the Directive. So, athough causing no immediate fataities, the Seveso incident had an enormous cost both to the operator and the environment as a whoe. In 1998 at Esso s Longford iquified petroeum gas processing pant in Austraia there was a massive exposion, kiing two workers and injuring eight. Athough in comparison with some of the events described above, the number of fataities and injuries were reativey sma, gas suppies to the state of Victoria were severey affected for severa months after the incident. Most of the state s gas suppy was cut for amost two weeks with severe disruption for a further 2 months and a tota estimated cost to industry of $1.3 biion. The Exxon Vadez oi spi on 24th March 1989 resuted in no fataities but in addition to the direct costs to Exxon, fines of around $150 miion doars were imposed aong with a $900 miion civi settement. The ingering oi spi aso had affects on the environment which are difficut to put a vaue on even now. In recent years focus has moved from improved safety and compiance with egisation to a regime where companies need to ook at improvements which can be shown to deiver benefits directy to their bottom ine. In genera, good safety means good business, as exempified above, and the business risk concept is a way of demonstrating this to senior management. The techniques for QRA and Consequence anaysis can be extended to assess the financia consequences of accidents and associated financia risk exposure. In today s competitive business environment key drivers are improved financia performance, maximised up-time, reduced insurance costs or reduced risk of interruption to business resuting from an accident. The typica questions to be answered are: If I have an incident, what wi it cost? What is the maximum oss I can incur as the resut of an accident? 2

3 How can I minimise the ikeihood of an incident resuting in oss of production? What risks am I exposed to from a financia standpoint? How can I perform Cost Benefit Anaysis on my operationa business risks? With greater goba competition and much more chaenging margins, there is now ess cash avaiabe for activities that do not contribute directy to the bottom ine, or are perceived as such. The Q of QRA need not ony be fataities, but effects on the environment, downtime or doars. By quantifying impacts on peope, operations or assets, anaysts are better abe to estimate the ikey costs of an incident or incidents in terms of down-time, asset damage, persona injury and oss of ife, brand damage, environmenta cean-up, itigation and compensation, and so on. Buiding on the methodoogies and modes used over many years for QRA and Hazard Anaysis and buit into many standard software toos (Worthington and Witox, 2002 and Cavanagh, 2001), we have extended the cassica QRA methodoogy to cacuate financia risks and consequences in addition to the more traditiona fataity risks and effects on ife. This paper goes on to describe the methodoogy we have used in extending the cassica QRA methodoogy to take account of other risk measures such as business interruption, environmenta impact, oss of production and so on. The Financia Risk Concept QRA techniques have been used over many years to assist in the management of the safe operation of process pants. The focus has usuay been on compiance with safety egisation and the approach taken dependant on whether a quantitative risk based approach is egisated (such as the Purpe Book in the Netherands) or a consequence based approach is used (such as RMP in the USA). Either way, both can be extended to assess the financia risks associated with a pant or the cost of a singe event occurring (financia consequence). The cassica QRA methodoogy provides a technique for quantifying the risks associated with the activities invoved in the production and processing of chemicas and petrochemicas. In order to quantify risks it is necessary to first identify a possibe risk situations, quantify them in terms of event consequence and ikeihood and compare them with acceptance criteria. The main questions to be answered by a QRA are what can go wrong, what are the potentia effects if it does go wrong, how often is it ikey to go wrong and is it important if it does. Typica outputs of a QRA study are ocation specific risk contours, estimates of individua risk and the F-N curve for representation of societa risk. Individua risk can be defined as the frequency at which an individua may be expected to sustain a eve of harm from the reaisation of specified hazards and is usuay taken to be the risk of death expressed as a risk per year. Societa Risk is defined as the reationship between the frequency and the number of peope suffering a given eve of harm from the reaisation of specified hazards. It is normay taken to refer to the risk of death expressed as a risk per year and dispayed as F-N curves. These describe the cumuative frequency (F) of a event 3

4 outcomes eading to N or more fataities and this representation of societa risk highights the potentia for accidents invoving a arge number of fataities. Risk to ife is ony one of the risks inherent in the operation of a process pant which may be reaised by the occurrence of an accident. Others incude risk to the environment, risk to assets and equipment and risk to financia performance. Furthermore, a these risks can have a cost associated with them which can be cacuated and integrated in the same way as fataity risk provided appropriate cost parameters are avaiabe for each cost category. The workfow and data required to extend the concepts of QRA to Financia Risk Anaysis (FRA) are summarised in Figure 1. Typica contributors to overa financia osses resuting from an accident may incude: Impact on peope in terms of fataities and injuries Property damage incuding capita costs to repair or repace damaged equipment and damage to other assets Business interruption incuding ost production from origina faiures Cost of ost inventory, again from sources and other damaged equipment Environmenta damage, incuding cean-up costs, fines, impact on anima and pant ife Pus many other outcomes with financia impact incuding ega costs, fines, oss of reputation, brand damage, compensation, reduction of share prices and so on Typica output from a financia risk anaysis, ooking purey at consequences of a singe accident may be tota cost of a singe faiure case, tota cost per outcome or per cost category, or both, and cost ranking per scenario. This is of use in assessing areas of a pant where a singe accident may resut in unacceptabe high risk of oss. Extending this to risk, measures such as Estimated Annua Average Loss and Estimated Maximum Loss may be cacuated. Aso, F-Cost curves (anaogous with the F-N curves for societa risk from a traditiona QRA) can be generated, aong with other metrics (Cavanagh and Linn, 2006). The Estimated Annua Average Loss (EAAL) heps to identify the factors contributing to the highest risks in financia terms. The rate of financia oss for a given event outcome is the product of the financia oss for that outcome and its frequency, and the tota rate of financia oss is the sum over a outcomes. Each event in an FRA mode has a ikeihood (typicay occurences per year) and consequence (typicay in USD, GBP, Euro or other currency). The product of these is the EAAL for the event in USD/year, for exampe. By aggregating the EAAL for each event, the EAAL for the entire faciity can be quantified. So the EAAL for a pant represents the average annua oss rate expected across the compete ifetime of the faciity. With respect to Estimated Maximum Loss (EML), this provides an estimate of the maximum oss that coud be sustained at a faciity due to a major accident. By assessing the financia consequences of a possibe outcomes for a possibe reease scenarios and integrating these over a possibe reease directions, an accurate and defendabe estimate of the maximum oss which coud be sustained at the faciity can be derived. This is aso the position where the F-Cost curve crosses the cost axis on the F-Cost graph so from Figure 3, for exampe, we can see that the EML is cose to $1biion. 4

5 Figure 1. Cassica risk anaysis methodoogy extended to incude financia risk cacuations 5

6 Other approaches have defined EML as a worst-case or financia consequence measure rather than an absoute maximum possibe oss with a given frequency. Marsh s SLAM mode, for exampe, defines EML as The argest oss that coud resut from a singe incident in the pant and uses the CAM exposion mode for assessment of overpressure whist SwissRe s ExToo focusses on damage to property ony using the TNT mode, determiningfinancia oss due to an individua exposion and extrapoating this to estimate Maximum Possibe Loss (MPL), again without consideration of its ikeihood of occurrence. Chippinda and Butts (2004) adopted a simiar approach to the current authors, using Phast as a consequence engine and performing the cost cacuations and risk summations through a number of spreadsheet modes. The advantage of the Safeti Financia mode used in this study is that asset, equipment, source, popuation and ignition information is entered onto a map directy through the existing Safeti GIS and grouping and combination functionaity enabing mutipe combinations to be anaysed easiy. For the mode under consideration here, the key eements are:- A geographica mode of the faciity and surroundings incuding popuation, ignition and asset and equipment data sets A compete set of major accident hazard scenario faiure cases for the faciity A set of representative weather conditions and their directiona probabiities (wind rose) Estimation of the ikeihood of each faiure case Modeing of the range of potentia consequences for each faiure case Assessment of the impact of each faiure case on the pant, surrounding assets and popuation Cacuation and assessment of the financia risks associated with these impacts reported in terms of F-Cost curves, tota oss rates, Estimated Annua Average Losses (EAAL) and Estimated Maximum Losses (EML). (See for exampe Evans and Thakora, 2004) Typica uses of this kind of financia risk anaysis incude Aiding the decision making process with risk reduction recommendations supported by cost benefit anaysis techniques Reducing exposure to financia risk by assessing the reative benefits of different risk mitigation strategies Comparison of financia risk exposure for a range of process conditions Financia risk trends with time Direct assessment of financia risks from major process pant hazards Demonstrating a strong cuture of corporate socia responsibiity (CSR) and adherence to the principes of tripe bottom ine (TBL) reporting which expicity considers an organisation's economic, environmenta and socia performance (Ekington, 1994) Better understanding of appropriate eves of insurance in terms of both maximum insured osses and deductibe eves The next section goes on to describe in more detai the methodoogy we have adopted and its impementation in the Safeti Financia mode. 6

7 Methodoogy and Mode Financia Risk Cacuations The tota financia risk is the summation of the tota risk due to impacts on peope (fataities and injuries), impacts on equipment and other assets in terms of damage and repacement cost, cost of business interruption, cost of environmenta impact and the cost of other outcomes such as ega costs, fines, brand damage, etc. The methodoogy used in this mode considers financia risk in terms of the foowing asset types: Popuation; Origina Source Equipment; Other Specific Equipment; Other Assets (buidings, non specific pant, infrastructure, etc.); User Defined costs. Cost contributors within each of these asset groups may incude equipment damage, ost inventory, business interruption and environmenta ceanup. For popuation zone anaysis, the cost of each reease is cacuated summing up the cost of a fataity and the cost due to injuries. If the anaysis invoves specific equipment that constitutes a source of a reease of hazardous materia or that can be affected by the accident, the foowing different costs coud be considered: Equipment repair: the cost of repairing the damage or repacing the equipment; Equipment business interruption cost: the financia oss incurred when production is hated because of the damage; Equipment ost inventory cost: the vaue of any inventory that had to be scrapped as a resut of the damage. Equipment environmenta: the cost of ceaning up any environmenta effects associated with the damage; Or any other costs associated with the damage. The same cost categories indicated for equipment can be appied to any asset zone (such as an industria area, process area, adminstrative buiding, residencia area) that can be affected by a hazardous event. The tota cost reated to any accidenta reease is cacuated considering a the above individua costs and any aditiona cost that can be defined by the anayst, as environmenta fees and ega costs. For cases with reativey ow costs, tota cost, COST T, is assumed to be given by a simpe summation. For cases with arger costs, the vaue wi be given by the power reation: COST Tota = max (COST T, 0.5 COST T 1.05 ) This is aready commony used in Risk Based Inspection (RBI) cacuations (Topais and Cavanagh, 2001, Risk Based Inspection, 2000). The power reation accounts in a 7

8 genera way for the additiona costs such as itigation, fines, etc., which increase with the size of the incident. This is commony referred to as escaation. Damage Leve and Vunerabiity Factor Concept The simpest method for assessing the impact of each consequence on a receptor is to define a threshod vaue for each, above which the receptor suffers 100% damage and beow which it suffers no damage. This concept wi be famiiar to Safeti users, where typicay 2 threshod vaues are used for each outcome type. For exposions, jet-fires and poo fires an upper and ower damage eve are defined, with a vunerabiity factor between zero and 1 associated with each eve. For fash fires a fraction of LFL is defined beow which no damage occurs and above which maximum damage occurs, as defined by the appropriate vunerabiity factor. For toxic reeases, a concentration eve is provided beow which no damage occurs and above which maximum damage occurs, again based on the reevant vunerabiity factor. Toxic damage is intended primariy for use with assets where damage wi resut in some kind of poution which has an environmenta cean up cost associated with it. A simiar approach is taken for the effects on popuation and this is aready we documented within the Safeti mode (Worthington and Witox, 2002). From the effect zone size and ocation determined using the consequence modes avaiabe in Phast, the damage eves described above, the associated vunerabiity factor and the overap between assets and effect zones, the eve of damage can be cacuated, and converted to a tota cost for each cost category. In Figure 2, the bod inner and outer footprints represent typica upper and ower threshod damage eves respectivey for a singe reease scenario and weather state. Aso shown are typica popuation and asset zones and a singe equipment item. Any receptor Figure 2. Definition sketch for asset damage due to upper and ower damage eves 8

9 outside the ower threshod vaue indicated by the bod outer ine wi be undamaged. Any receptor between the inner and outer threshod boundaries wi be damaged to the degree indicated by the appropriate vunerabiity factor for the ower threshod. Any receptor within the inner threshod boundary wi be damaged to the degree indicated by the appropriate vunerabiity factor shown for the upper threshod. So, for exampe, in Figure 2, if upper and ower vunerabiity factors are set to 1.0 and 0.5 respectivey, then the section of asset area A outside the outer footprint wi be unaffected, 50% of asset area A between the inner and outer footprints wi be affected and 100% of asset area A within the inner footprint wi be affected. Simiary, for the singe equipment item iustrated, this wi be 100% affected, since it is in the inner zone, as wi the popuation which is aso within the inner zone. Source Costs Sources or faiure cases resut in hazardous reeases which in turn resut in damage to other assets. However, each reease wi have a cost associated with it whether it impacts other assets or not. In order to cacuate the source costs, the foowing information is required in addition to the reease source term information. Each source is defined as a particuar equipment type, which wi have its own unique outage time and repair cost, based on the information in Tabe 1. The mode aso aows user defined equipment types to be added with their own outage time and repair costs. Tabe 1. Outage time and repair cost ook-up tabe for range of equipment types Equipment type Outage time (day) Repair/repace cost ($) Large Pipes 7 50,000 Medium Pipes 4 20,000 Sma Pipes 2 5,000 Compressors ,000 Exchangers 5 50,000 Vesses 7 40,000 Fiters 1 10,000 Reactors 14 80,000 Tanks 7 80,000 Pumps 0 5,000 Heater 5 60,000 Coumn 21 10,000 Other/Genera 7 20,000 Mobie Buidings 5 25,000 Brick Buidings ,000 Asset Zones 5 1/m 2 None 0 0 9

10 In addition the foowing information is required to cacuate costs in each of the cost categories described earier: Cost of ost Production per day Cost of Environmenta cean-up per kg Other Environmenta costs for other effects per kg Vaue of ost inventory The Inventory mass and spied mass, which may not be the same, are aso required, athough the spied mass can be cacuated as part of the discharge cacuation performed in the source term modeing if desired. Popuation zone costs Popuation information can be entered through the GIS in the same way as in the Safeti mode, as can ignition source information. However, for financia risk cacuations based on popuation the foowing additiona parameters are required: Cost of a Soitary Fataity Number of Injuries Per Fataity Cost of a Singe Fataity Among Many Fataities Cost of One Injury Equipment costs An equipment item in terms of financia risk is essentiay the same as a source, merey acting as a point receptor to a hazard zone generated by a source. From that point of view an origina source and an item of equipment require the same financia data. Typica exampes of specific items of equipment which you may want to enter into the mode as equipment rather than sources or assets are individua arge capita vaue items of pant which do not represent a hazard themseves, or storage faciities for non hazardous materia which have a high capita vaue. Asset set costs An asset is a generic receptor type, which wi resut in a cost being incurred if it is impacted by a hazard zone with a arge enough damage eve. An asset zone is treated in the same way as an equipment item in terms of financia risk, but with no contribution from environmenta costs. Assets can be defined at a point, or over an area. Typica exampes of assets may be on-site buidings ike contro rooms and offices, off site buidings and infrastructure incuding, for exampe, houses, commercia buidings, factories, warehouses, etc. Areas of a pant which do not contribute directy to risk as possibe sources of hazardous reeases may be defined by a series of assets, to account for background pant vaue. 10

11 Financia Risk Parameters Much of the data described above is avaiabe as a set of defaut parameters within the mode which may be modified on a case by case basis for each receptor and cost category type. Data avaiabe as defauts incudes: Damage eves and vunerabiity factors for each outcome type, cost category and receptor type Cost of ost production per day, environmenta cean-up per kg, environmenta costs for other effects per kg and vaue of ost inventory Outage time and repair cost for each equipment type with abiity to add new types Financia information reating to fataities and injuries incuding cost of a soitary fataity, cost of an injury and number of injuries per fataity and cost of singe fataity amongst many in order to dea with the concept of risk aversion (see for exampe Wiggins 1984) when deaing with mutipe fataity events. In addition, cacuations and resuts may ony be required for a sub-set of the compete set of cost categories avaiabe. For exampe, in certain situations it is undesirabe to pace a vaue on a ife and therefore you may wish to excude fataity and injury costs from the cacuations. The compete set of cost categories is origina source, fataity, injury, repair and repacement, ost inventory, business interruption and user defined. Any of these can be excuded at any eve, either from defauts, so that they are not incuded in cacuations and never appear in the resuts, at the study eve so that they are not cacuated for a particuar study and don t appear in the resuts for that study, or at the resuts eve so that they are cacuated but not dispayed individuay in the resuts. The study described beow was used to assess the financia risks associated with the business interruption cost category ony. Practica Exampe of Managing Business Risks using FRA This rea ife study was performed by DNV Energy in order to support its cient in assessing the adequacy of the maximum insurance vaues for their pant. The cient wanted to review the maximum vaue for their insurance based on a quantitative risk assessment of accidents that coud happen on their site. A typica QRA study was deveoped based on the cassica approach supported by Safeti and the workfow iustrated in Figure 1. Subsequenty equipment data was added to the mode using the GIS capabiities of Safeti Financia and appropriate repair costs were defined for each asset zone within the pant. Accident scenarios from eeven different systems on two of the cient s Oefin pants were anaysed. Frequencies were estimated for a range of typica faiure case scenarios categorised within each of the systems. These were combined with the financia oss consequences cacuated using Safeti Financia. The combined resut from the frequencies per year and the financia consequences in US doars based ony on equipment damage costs were presented as F-Cost curves, thus representing cumuative frequency with which osses exceeding a particuar doar vaue woud be expected to occur. 11

12 The major phases of the anaysis were: Phase 1: Identification of reease events, based on materia, reease conditions, inventory, contro systems, ocation and confinement, that coud represent significant osses to the site. Phase 2: Likeihood and consequence cacuation for each seected event. Phase 3: Financia oss cacuation for each scenario and resuts assessment. The processes on each pant which were incuded in the anaysis were: Pant Oefins A: Propyene refrigeration, Propyene separation, Ethyene refrigeration, Ethyene separation, Demethanizer and Deetanizer. Pant Oefins B: Propyene refrigeration, Propyene separation, Ethyene separation, Deetanizer and Demetanizer, A the cacuations were performed using the Safeti Financia software, deveoped by DNV Software. Safeti Financia cacuates the frequency of occurrence of each representative reease scenario. This is based on individua equipment reease frequencies factored by the number of pieces of equipment within the segment or area on the pant for which that scenario type coud occur. This is achieved by counting equipment items for which simiar reeases may occur and factoring the individua reease frequencies based on the equipment count. These frequencies are then factored by probabiities associated with each possibe weather condition (in terms of wind direction and veocity the wind rose), ignition source probabiities for immediate and deayed ignition, as we as event trees providing the probabiities with which each possibe outcome type is expected to occur (e.g. probabiity of ignition, fash fire, jet fire, VCE, etc.). This methodoogy is described in detai in the Safeti user manuas (Worthington and Witox, 2002). The financia consequences are estimated using the Phast modeing integrated in Safeti Financia. This takes account of the reease conditions such as materia, temperature, pressure, inventory and reease hoe size. These source-terms are then used to cacuate the associated hazard zones from which the cost of repair of the equipment on the instaation damaged by the reease is cacuated. The cost of repair within each area, based on the damage for each reease scenario, are then cacuated considering the hazard zone, as foows: F Repair,h,o = v Repair,h C Repair a h,0 where v Repair,h is the repair vunerabiity for hazard zone h that is defined for each outcome type, as shown in Tabe 2 beow, C Repair is the cost of repairing or repacing the entire area and a h,o is the fraction of the area that is covered by hazard zone h for outcome o. The resut from the anaysis are used to create the F-Cost curves shown in Figure 3, from which we can see the estimated repair costs we can expect to incur with frequencies between 0.2 and 10-6 (or 1 in a miion) occurrences per year. The bue curve shows expected osses considering a possibe consequences from the eeven events anaysed, considering day weather conditions and the pink curve is reated to night weather conditions. 12

13 Tabe 2. Repair vunerabiity per outcome (based on API RP 581 (2000)) Outcome Threshod Maximum Poo Fire 12,5 kw/m 2 50% damage 37,5 kw/m 2 100% damage Jet Fire Fireba Fash Fire 100% LFL 100% damage Exposion 0,3 bar 50% damage 0,5 bar 100% damage Based on the above, the cient decided to use an accumuated frequency of 10-5 / year as acceptabe for their maximum oss, which impies a tota insured oss of USD 1 biion. The main contributors from each system to the maximum expected osses are shown in Tabe 3 beow. From this we can see that the 3 argest contributors to expected maximum oss are the Propyene Refrigeration in Pant A, Propyene Separation in Pant A and the Propyene Separation unit within Pant B. It shoud aso be noted that a scenarios with a frequency of occurrence ess than 10-9 /year were negected and that a Figure 3 F-Cost curve 13

14 Tabe 3. Maximum oss by system System Event number Maximum cost (USD) Pant Oefins A Propyene Separation Propyene Refrigeration Ethyene Separation Demethaniser Ethyene Refrigeration Deethaniser Pant Oefins B Propyene Separation Deethaniser Propyene Refrigeration Ethyene Separation Demethaniser scenarios with osses above USD 1 biion had frequencies of occurrence of ess than 10-8 /year. Driing down further we can use the faciities of Safeti Financia to identify the scenarios contributing most to the overa financia risk in terms of Expected Maximum Loss (EML). This enabes us to isoate the key faiure cases we shoud focus on shoud we wish to recommend mitigating measures for reducing overa financia risk exposure. In this study the foowing were found to be the argest contributors. EML above USD 1 biion Propyene Refrigeration Pant Oefins A Propyene Separation Pant Oefins A Propyene Separation Pant Oefins B EML above USD 900 miions Ethyene Separation Pant Oefins A We have aso ooked at individua F-Cost curves for each system in order to assess more accuratey which system are contributing most to the financia risk exposure at different frequencies of occurrence and these are iustrated in Figure 4. From here we can see that the main contributors at frequencies of 10-6, 10-5 and 10-4 per year, in order of highest contributor are: 10-6 /year: Systems 6, 10, 2, 1 and /year: Systems 10, 2, 6, 1 and /year: Systems 10, 2, 9, 3 and 11 14

15 Figure 4. F-Cost curve per system Future Potentia The resuts presented above represent the actua conditions of the anaysed systems. In order to reduce the estimated osses as we as the expected osses for each frequency, mitigation measures can be evauated and the resuts reviewed. This mitigation shoud focus on inventory isoation to minimize consequences and reduction of frequency of reease. A cost benefit anaysis can be performed, comparing the reduction on the insurance premium per year with the cost for the impementation of the mitigation proposed. This way a risk management approach based on the continuous reduction of the estimated osses can be used. Aso, as an extension to the study presented in this paper, other specific insurance products such as third party insurance can be anaysed using the same approach taking account of property damage as we as business interruption and fataity/injury costs. Concusions We have extended the cassica approach to QRA to enabe the cacuation of broader financia or business risks and buit this extended methodoogy into the Safeti QRA mode. 15

16 We have then used this mode to assess the financia risk exposure of an oefin production pant focusing on financia risk from accident scenarios resuting in damage or destruction of equipment and other assets. The Safeti Financia FRA mode has been used to make recommendations on appropriate eves of insurance. We have aso used the mode to identify the main contributors to these broader business risks in order to propose measures to reduce our cient s exposure to possibe financia osses. References Cavanagh, N.J., Cacuating Risks, Hydrocarbon Engineering, Voume 6, Number 6, Paadian Pubications, London, June Cavanagh, N.J. and Linn, J., Apri 2006, Process Business Risk A methodoogy for assessing and mitigating the financia impact of process pant accidents, AIChE Goba Safety Symposium, Annua Conference of Centre for Chemica Process Safety, Orando, Forida, pp Chippinda, L. and Butts, D., June 2004, Managing the Financia Risks of Major Accidents, Annua Conference of Centre for Chemica Process Safety, Emergency Panning: Preparedness, Prevention and Response, Orando, Forida, pp Ekington, J., Towards the sustainabe corporation: Win-win-win business strategies for sustainabe deveopment., Caifornia Management Review 36, no. 2: , Evans, J. and Thakora, G., Tota Loss Prevention Deveoping Identification and Assessment Methods for Business Risks, 11th Internationa Loss Prevention Symposium, Prague, Fewtre, P. and Hirst, I.L., A review of high cost chemica/petrochemica accidents since Fixborough 1974, IChemE Loss Prevention Buetin, No. 140, Apri Risk Based Inspection, Base Resource Document, API Pubication 581, Second Edition, October Topais, P. and Cavanagh, N.J., Optimising Inspection Regimes. Hydrocarbon Engineering, Voume 6, No.10. October Wiggins, J. H., Risk Anaysis in Pubic Poicy, Proceedings of Victoria Division, Engineers Austraia, Risk Engineering Symposium 1984: Engineering to avoid Business Interruption, Worthington, D.R.E. and Witox, H., 2002, SAFETI Risk Modeing Documentation Impact of Toxic and Fammabe Effects, DNV Software Risk Management Soutions, October