Safety Use Maximum-Credible Accident Scenarios for Realistic and Reliable Risk Assessment Faisal I. Khan, Memorial Univ. of Newfoundland Posing various possible incidents — rather than just the worst-case one — illuminates those that are really important and are most likely. Such knowledge can enhance safety and planning for emergencies. T here have been many methodologies proposed for the risk assessment in the chemical process industries (CPI). Among them, the most notable ones are quantitative risk analysis, probabilistic safety analysis, worst-case methodology for risk assessment and optimal risk analysis. A critical review of these methodologies is available (1). The key points of these methods are: Quantitative risk analysis (QRA) — This method is comprised of four steps: hazard identification, frequency estimation, consequence analysis and measure of risk. The first step answers the question: What can go wrong? This is the most important step because hazards that are not identified will not be quantified, leading to an underestimated risk (2, 3). The techniques used for hazard identification include hazard indices, hazard and operability (HAZOP) studies, failure mode and effect analysis (FMEA), what-if analysis and checklists. After the hazards are identified, the scope of a QRA is defined. The second step asks: How likely is the occurrence of each accident? Answering this question means quantifying of the probability of each accident scenario. The third step aims to quantify the negative impacts of the envisaged accident scenario. The consequences 56 www.cepmagazine.org November 2001 CEP are normally measured in terms of the number of fatalities, although they can also be determined by the number of injuries or value of the property lost. The last step of a QRA is to calculate the actual risk. Probabilistic safety analysis (PSA) — Different techniques can be combined to carry out PSA (4, 5, 6). PSA provides a framework for a systematic analysis of hazards and quantification of the corresponding risks. It also establishes a basis for supporting safety-related decision-making. The methodology and the procedures followed for the PSA of a typical chemical installation handling a hazardous substance can be outlined in seven major steps: (1) hazard identification; (2) accident-sequence modeling; (3) data acquisition and parameter estimation; (4) accident-sequence quantification; (5) hazardous substance-release-categories assessment; (6) consequence assessment; and (7) integration of results. Worst-case methodology for risk assessment — An excellent work on hazard and risk screening based on worst-case methodology is presented by Hirst and Carter (7). The method does not discuss use of worstcase scenarios in detailed risk assessment, but rather outlines the development of ready-made and easy-touse risk indices, such as the risk integral, scaled risk integral and approximate risk integral, for early planning Modularization of Complete Plant into Manageable Units and decision-making. It is recommended that Hazard Identification if the values of computed risk indices are • HIRA Technique above acceptance criteria, then a detailed Aid to Develop risk assessment be made. Although this Accident Scenario methodology computes these indices based on the worst-case of an accident in a unit, it Qualitative Hazard Assessment makes the computation more realistic by • optHAZOP procedure Aid to Develop • TOPHAZOP tool considering many parameters, such as condiAccident Scenario tional plume probability, the population distribution factor, and weather and directional probabilities. This method does not actually Quantitative Hazard Assessment/Consequence Assessment assess the hazard/risk in a unit, since the • MOSEC for fire and explosions worst-case scenario remains almost constant • HAZDIG for toxic release and dispersion • DOMIFFECT for cascading effects for different units; it just depends upon the type/quantity of the chemical in the unit (8). Optimal risk analysis (ORA) — This is a fairly new means for risk analysis (1, 9). ORA involves four steps: (1) hazard identifiProbabilistic Hazard Assessment • PROFAT tool cation and screening; (2) hazard assessment (both qualitative and probabilistic); (3) quantification of hazards or consequence analysis; and (4) risk estimation (Figure 1). This pro■ Figure 1. cedure is named ORA as it is swifter, less exRisk Estimation Steps and methods used pensive to implement, less time-consuming to conduct an optimal risk and more-precise than alternative analyses analysis (ORA). (1, 10). Central to all of these methods is predicting accident scenarios; a set of scenarStop ios is developed and subsequently analyzed for detailed consequences. Significant advancement has been made in developing newer means for hazard and risk assessment, There is a wide variation in the risk-assessment studies consequence modeling, and user-friendly computer-aided conducted by different groups. Foremost in this is the diftools. However, while foreseeing worst-case scenarios is ference in the envisaged accident scenarios. For the same common, little attention is paid in envisioning credible sceunit, one group foresees a small leak from the joints and narios. Moreover, the scenarios developed by safety engiestimates risk based on that, while another group imagines neers often describe just one type of release, without giving an explosive release of all vessel contents. Moreover, it has much detail about its mode and further escalation. This point also been observed that only one or two scenarios are deis cited in the magnum opus on loss prevention by Lees (11): veloped for a possible accident in a unit, and these may not “For a potential release, it is necessary not reflect the true important possible incidents. This is due to only to identify a source but also to decide on the the absence of any homogeneous system for accident-scenature of the release which could occur … The nario envisaging and credibility assessment. identification process should not stop at the point A credible accident is one within the realm of possibiliwhere a release occurs, but in principle should be ty and is likely to be severe enough to cause significant continued to embrace the consequences of the redamage. However, what constitutes reasonable probability lease and the failures and other events, which may and significance of credibility are mostly qualitative, based allow these consequences to escalate. In practice, on the subjective judgment of the analyst. A small error or this aspect is usually treated as part of the hazard a bit of ignorance in these subjective judgments (in particuassessment. Whilst this is a reasonable approach, lar, the development of accident scenarios) can yield errothere is a danger that, unless the identification of neous results or make the study meaningless. The more rethe escalation modes is treated with a thoroughalistic the accident scenario, the more accurate the foreness matching that applied to the identification of casting will be for this type of accident, its consequences the release modes, features which permit escalaand associated risks. This would help in developing more tion and against which measure might be taken appropriate and effective strategies for crisis prevention will be missed. The identification of the modes of and management (12). escalation appears to be a rather neglected topic.” Thus, it is necessary to have some systematic procedure CEP November 2001 www.cepmagazine.org 57 Safety Accidental Release of Chemical Operational Data Temperature Pressure Capacity of the unit Chemical characteristics, etc. Release under high pressure Normal continuous release Normal instantaneous release Non-operational Parameters Chemical properties Toxicity Atmospheric conditions Quantity released Site characteristics Safe No Yes No Is the chemical flammable/reactive? Is the chemical toxic? Yes Yes Fire Fireball Flash fire Pool fire Jet fire Explosion CVCE BLEVE VCE Vented explosion Dispersion of Toxic Chemical Toxic load ■ Figure 2. Non-operational Parameters Chemical characteristics Ignition source Quantity of chemical released Atmospheric conditions Degree of confinement Site characteristics The logistics of generating an accident scenario. Accident Scenario: Sequence of events or guidelines to envisage all probable accident scenarios, and, further, to decide which are the most credible. Also, a scenario should describe the complete situation (right from release mode to the subsequent events). Accident scenario This is a description of an expected situation. It may contain a single event or a combination of them. In most of the past risk-assessment reports, accident scenarios have been proposed as single events, which is not a valid way of imagining an incident. Past case studies show that an accident occurs as a sequence or combination of events. Creating a scenario does not mean that it will occur, only that there is a reasonable probability that it could. A scenario is neither a specific situation nor a specific event, but a description of a typical situation that covers a set of possible events or situations (Figure 2). It is the basis of the risk study; it tells us what may happen so that we can devise ways and means of preventing or minimizing the possibility. An accident scenario forms a focal point of a heuristic process. It enables use of the wisdom of hindsight and state-of-the-art knowledge to evaluate its impact in forecasting accident situations. The scenario is a reference 58 www.cepmagazine.org November 2001 CEP point, as well as a link between the past, present and future. Such scenarios are generated based on the properties of chemicals handled by industry, physical conditions under which reactions occur or reactants/products are stored, as well as geometries/material strengths of vessels and conduits, in-built valves and safety arrangements, etc. External factors, such as site characteristics (topography, presence of trees, ponds, rivers in the vicinity, proximity to other industries or neighborhoods, etc.) and meteorological conditions, need also be considered. Worst-case accident scenario The Emergency Planning and Community Right to Know Act (EPCRA) was passed in 1986, in the wake of the tragic chemical accident in Bhopal, India. EPCRA established some systems to cope with chemical emergencies including Local Emergency Planning Committees (LEPCs). LEPCs must prepare comprehensive emergency plans outlining local chemical hazards and emergency response procedures. For these plans, the U.S. Environmental Protection Agency (EPA) recommends that LEPCs either prepare themselves or request facilities to prepare emergency plans based on worst-case scenarios (8, 13). Worst-case accident scenarios indicate the geographic area (the vulnerable zone) affected by the worst-possible accident at a facility in which people would be at risk of life and health. These scenarios typically consider the nearinstantaneous release of the entire amount of a chemical stored at a facility and assume the failure of mitigation and safety systems. EPA and other regulatory agencies have been canvassing the worst-case scenarios for emergency planning; these scenarios are also critical for pollution prevention. The Clean Air Act Amendments (CAAA) of 1990 require companies to prepare risk management programs, including worst-case accident scenarios, and make them available to the public. Many companies are questioning the significance of the worst-case accident scenario concept and opposing the national and public data systems that include the risk management plan (RMP) on the Internet (8). Although, the author agrees with CAAA and Right to Know Act, the worst-case accident scenario may not be the best approach. This author feels that RMPs and other emergency plans should be based on the maximum-credible accident scenarios. Maximum credibility should be set by regulatory agencies, depending upon the vulnerability of a site. This would overcome the following major limitations of the worst-case-accident-scenario approach: • In the worst-case method, emergency or risk-management planning is done mostly based on the release and dispersion of toxics as they cover maximum distances and may cause a large number of fatalities by a short exposure, as in Bhopal. However, it is not always true that a release of toxic chemical would be the most disastrous one, particularly if the domino effect were considered (12, 14, 15). The 1984 incident in Mexico and one in Vishakhapatnam, India (1997) are clear evidence of this (15, 16). • In using the worst-case scenario, only one-parameter damage potential is considered, and the probability of its occurrence is generally ignored. However, the probability of an accident is also equally important. Past accident analysis reveals that frequent, but less-damage-causing accidents can create large financial losses and, often, can escalate to catastrophic proportions due to negligence. A detailed description of the maximum-credible-accident-scenario approach is now presented. Maximum-credible accident scenarios In using maximum-credible accident scenarios (MCAS), the central criterion is what constitutes a credible accident. A credible accident is defined as: an accident that is within the realm of possibility (i.e., probability higher than 1 × 10–6/yr) and has a propensity to cause significant damage (at least one fatality). This concept (11, 17, 18, 19) comprises both parameters — probable damage caused by an accident and its probability of occurrence. There may be types of accidents that may occur frequently, but would cause very little damage. And there may be others that may cause great damage, but would have a very low probability of occurrence. Both would be considered accidents. A credible accident scenario should contain two sets of information: a description of the situation and its probability of occurrence. The description must not reduce the freedom of finding solutions and must not restrict the means available for solution. A good accident scenario should describe the most prime cause of an event. An example: Define a leak rate instead of an explosion pressure, because here, one could go further and describe the cause of the leak as well. There may be number of accidents that occur quite frequently, but due to proper control measures or lesser quantities of chemicals released, they are controlled effectively. A few examples are a leak from a gasket, pump or valve, release of a chemical from a vent or relief valve, and fire in a pump due to overheating. These accidents generally are controlled before they escalate by using control systems and monitoring devices — used because such piping and equipment are known to sometimes fail or malfunction, leading to problems. On the other hand, there are less problematic areas/units that are generally ignored or not given due attention. This is because few or even no accidents have been reported. In such situations, even a small leak may lead to a disastrous accident. Past accident analysis reveals that most of the catastrophic accidents occurred in ignorance (the accident was not foreseen) and either in areas marked yellow (not highly hazardous) or where the control arrangements were inadequate (control measures based on less credible scenarios). The disaster at Vishakhapatnam proves that most of the risk and hazard studies are lacking in envisaging the credible accident scenarios, therefore, the control measures or emergency plans are not so effective. Methodology for MCAS The first step of the MCAS methodology develops all plausible accident scenarios in the unit (Figure 3). In the second step, damage radii are calculated for each scenario. This can be done using quantitative hazard indices. There are two options available — one is to use Dow’s Fire and Explosion Index (for fire and explosion scenarios) and the Mond Toxic Index (for toxic and corrosive releases and dispersions) (9, 11, 20). The other is to employ the indices proposed by Khan and Abbasi (1, 21); a fire and explosion damage index (FEDI), and toxic damage index (TDI) for flammable and toxic chemicals, respectively. In the next step, the probability of each accident scenario is estimated. This can be done using either industryspecific data (failure rates of various components used in a process unit) or the data available in the literature (frequencies of occurrence of the same event under similar conditions). The later process of probability estimation is easy, but a little crude, and has been subject to criticism. However, this author feels that this procedure is adequate, as the objective is to get a rough estimate of probability. If the accident is found to be credible, more accurate estimation of its probability would be done in the subsequent step of de- CEP November 2001 www.cepmagazine.org 59 Safety Take One Unit Develop All Plausible Accident Scenarios Consider One Accident Scenario Flammable Toxic Is the Chemical Flammable or Toxic? Both Flammable and Toxic Calculate Factor A Calculate Factor B Calculate Factor AA Calculate Factor C Calculate Factor BB Calculate Credibility Factor L1 Calculate Credibility Factor L2 of formulating the accident scenario. According to the characteristics of the chemical, accident scenarios can be divided into three main groups: fire and explosion; release and dispersion of a toxic or corrosive fluid; and both fire and explosion, and toxic events. In real life, many times the third type of accident occurs. Once damage radii and probabilities are known for each damaging event, three factors — A, B and C — are computed using site-specific information, such as population density, asset density of the site, etc., to define credibility. The estimating procedures follow for each of the three incident groups. Scenarios involving fires and explosions Financial loss: Factor A accounts for the damage to property or assets and may be estimated for each scenario using: Ai = (AR)i × (PR)i × (AD)i /UFL (1) Calculate Total Credibility Factor L A = minimum (1, ΣAi) Classify Credibility of the Scenario Is it Credible? No Yes List the Scenario Are all Units Over? Yes Short-list the Most Credible Accident Scenarios ■ Figure 3. Steps in the most-credible-accident scenarios (MCAS) method. tailed risk assessment. The use of more reliable methods of probability estimation (e.g., fault-tree analysis) not only requires large sets of data, but also large amounts of computational time. These costs are not justifiable at this stage 60 www.cepmagazine.org where i is the number of events (from i = 1 to n), i.e., fire, explosion. UFL is the unacceptable financial loss. For example, a loss of $1 million/100 yr may be just tolerable to an organization, so a loss higher than this unacceptable. We suggest a value of $10,000/yr for UFL. Fatalities: Similar to the factor for financial loss, the fatality factor, B, is estimated for each accident scenario: PD1 = PD1 × PDF1 No November 2001 CEP (2) (3) Bi = (AR)i × (PR)i × (PD1)i/UFR (4) B = minimum (1, ΣBi) (5) UFR, the unacceptable fatality rate, has a suggested value = 10–2. PDF1, the population distribution factor, reflects the heterogeneity of the population distribution. If the population is uniformly distributed in the region of study (~2,000-m radius), the factor is assigned a value of 1; if the population is localized and away from the point of accident, the lowest value of 0.2 is assigned. Values for this parameter has been adapted from Ref. 9. Importance Factor 1 0.8 0.6 Nomenclature 0.4 0.2 0 0 1 5 10 50 Distance of Vulnerable Ecosystem from the Accident Site, km ■ Figure 4. Importance factor, IM, is a measure of the spread of potential damage. A AD AR B BB C CC IM L L1 L2 PD1 = = = = = = = = = = = = PD2 = Ecosystem damage: Factor C signifies ecosystem damage, which can be estimated as: Ci = (AR)i × (PR)i × (IM)i/UDA (6) C = minimum (1, ΣCi) (7) UDA, the unacceptable damage area, has a suggested value of 1,000 m2/yr. IM, the importance factor, is 1 if the damage radius is higher than the distance between accident and location of a sensitive ecosystem, i.e., lake, forest, bird sanctuary, etc. IM is quantified using Figure 4, developed with the help of Ref. 22. In this reference, the authors use a parameter ecosystem damage-penalty for the quantification of an accident hazard index, AHI. This parameter was quantified based on a comprehensive Delphi (22). (Delphi is a technique to quantify subjective parameters through an opinion survey of a team of experts. Delphi was used to quantify some parameters of the AHI.) These three factors are combined together to yield L1 by: L1 = [1 – (1 – A)(1 – B)(1 – C)] (8) Scenarios involving toxic release and dispersion Unlike for fire and explosion, two factors are estimated here, BB and CC, for fatality and ecosystem damage, respectively. These are computed using the following equations: PD2 = PD2 × PDF2 PDF1 = PDF2= PR = UDA = UFL = UFR = WPF = factor for damage to property or assets asset density in the vicinity of the event, up to ~500-m radius), $/m2 area inside the damage radius, m2 factor for fatalities factor for toxic release and dispersion for fatalities factor for ecosystem damage factor for toxic release and dispersion for ecosystem damage importance factor, from Figure 4, dimensionless total credibility factor, dimensionless credibility factor for fire and explosion hazard, dimensionless credibility factor toxic hazard, dimensionless population density in the vicinity of the event (fire and explosion) up to ~2,000-m radius, persons/m2 population density in the vicinity of the event (toxic release) up to ~2,000-m radius, persons/m2 population distribution factor for fire and explosion, dimensionless population distribution factor for toxic release and dispersion, dimensionless probability of occurrence of an event, /yr unacceptable damage area, m2/yr unacceptable financial loss, $/yr unacceptable fatality rate, persons/yr weather probability factor, dimensionless study (an area of ~2,000-m radius). The method of quantification is same as for PDF1. WPF represents the likelihood of the weather condition used in the dispersion estimation. Generally, the maximum possible damage area is estimated considering a slightly stable or stable condition. However, this condition may not prevail at all times, so 0.0 Uncertainty Zone 0.2 ■ Figure 5. Credibility Zone Classification of credibility. (9) 0.5 BBi = (AR)i × (PR) × (PD2) × (WPF)/UFR (10) BB = minimum (1, maximum of BBi) (11) Maximum Credibility Zone where i denotes the particular chemical released. In case of an accident scenario involving more than one chemical, BB is computed for each and the highest value is used. PDF2 defines the population distribution factor, which reflects the heterogeneity of the population distribution in the region of 1.0 CEP November 2001 www.cepmagazine.org 61 Safety Table 1. Data setting for the ammonia study. Parameters Value Chemical involved Delineation of MCAS The credibility ranking is shown Figure 500 m.t. 5. Region 0–0.2 signifies the zone of uncerLiquefied tainty, which means that the envisaged sceStorage narios do not pose much threat either due to 15˚C very low probability of occurrence or damage potential. This zone signifies the tolera6.5 atm ble risk zone. Region 0.2–0.5 signifies the 0.40 credible scenarios, meaning they are likely 250 persons/km2 to occur and may cause enough damage. $300/m2 Region 0.5 onwards signifies MCAS, which means the developed scenarios are 0.3 highly vulnerable to cause catastrophes. 0.3 Once all of the credible and MCAS 1.0 have been identified, they are further studied to decide the most credible ones as per the analyst criterion. The quantitative value of this credibility criterion is defined considering: the objective of study, available time and resources, and the operational constraints. Therefore, the significance of the term “most credible” varies widely, depending upon the analyst or team of analysts conducting the study. This would further short-list important accident scenarios. The short-listed scenarios may be further processed for damage potential estimation, risk estimation, and finally, to develop hazard mitigation/minimization or disaster management strategies. The proposed approach is now used to study the storage of liquefied ammonia. Ammonia Quantity of the chemical involved Phase of the chemical Unit operation Operating temperature, T Operating pressure Degree of conjunction at the site Site population density (within region of 2,000-m radius) Asset density (within region of 500-m radius) Population distribution factor Weather probability factor Importance factor this factor estimates the probability of this atmospheric condition occurrence. This is quantified using the statistical weather data of the local area. For example, if a slightly stable condition exists 30% of the time during a year, then the WPF is considered to be 0.30. Similarly, CC is computed as: CCi = (AR)i × (PR) × (IM)/UDA (12) CC =minimum (1, maximum of CCi) (13) Finally, these two factors are combined to give a credibility factor L2 for toxic release and dispersion: L2 = [1– (1 – BB)(1 – CC)] (14) Scenarios involving combination of fire, explosion and toxic release To estimate the credibility of accident scenarios involving both type of events, L1 and L2 are combined as follows: L = (L12 + L22)1/2 (15) Case study: Ammonia A vessel stores 500 metric tons of liquefied ammonia at of 15°C and 6.5 atm. The vessel is connected with one input line, one outflow line, a pressure-relief valve and other conventional safety devices. The vessel is in one corner of a fertilizer plant where the population density is 250 persons/km2, and asset density around the unit is $300/m2 (Table 1). There is bird sanctuary about 1,000 m away from the site. A total of five different accident scenarios are envisaged in the unit: Table 2. Credibility factors for the scenarios in the ammonia study. Accident Scenario Damage Radius, m Frequency of Occurrence, (/yr) Fire and Explosion A C L1 Credibility BB CC L2 L Scenario 1 2,500 5.0E–05 1.00 0.98 1.00 1.00 Scenario 2 1,100 4.0E–04 1.00 1.00 1.00 1.00 Scenario 3 250* 1,270 7.0E–5 0.79 0.35 0.86 350† 1,200 1.0E–06 0.01 0.00 0.01 950 8.0E–05 0.51 0.22 0.61 Scenario 4 Scenario 5 * Damage radius for BLEVE (boiling liquid/expanding vapor explosion). † Damage radius for CVCE (confined vapor cloud explosion). 62 B Toxic Load www.cepmagazine.org November 2001 CEP 0.41 0.00 0.10 0.00 0.01 0.00 0.47 0.00 0.98 0.01 0.61 Table 3. Result of consequence analysis for Scenarios 1, 2, 3 and 5 for ammonia study. Parameters Scenarios and Their Likely Impacts Scenario 1: Continuous Release Explosion: Total energy released, kJ Peak overpressure developed, kPa Variation of overpressure in air, kPa/s Shock wave velocity, m/s Duration of shock wave, ms Scenario 2: Continuous Release near Ground Scenario 3: BLEVE with Dispersion BLEVE 1.11E+07 1,556.7 575.7 865.5 14 Scenario 5: Evaporation with Dispersion NA NA NA NA 745.6 6.5E+06 NA NA NA 0.17 0.30 0.02 NA NA NA NA NA Toxic release and dispersion: Instantaneous (I)/continuous (C) Puff/plume characteristics: Concentration at center of puff/plume, kg/m3 Concentration at the edge of puff/plume, kg/m3 Dia. of puff/plume at end of lethal zone, m Radius of the lethal zone (based on LD50), m C Plume 3.14E–02 3.14E–03 210 2,500 C Plume 4.15E–03 4.15E–04 170 1,100 I Puff 3.78E–04 3.78E–05 375 1,270 C Puff 1.15E–04 1.15E–05 115 950 Domino checking: Location of the unit from primary event, m 60 m 60 m 60 m 60 m NA NA NA NA NA NA 3.15E+05 310 1.0 NA NA NA 3.15E+05 145.4 NA Missile Characteristics: Initial velocity of fragment, m/s Kinetic energy of fragment, kJ Penetration ability at 50 m: Concrete structure, m Brick structure, m Steel structure, m Fire: Radius of fireball, m Duration of fireball, s Energy released by fire ball, kJ Radius of pool fire, m Burning area, m2 Burning rate, kg/s Radiation heat flux, kJ/m2 Domino effect due to heat load: Total heat received, kJ Heat intensity, kJ/m2 Probability of domino effect due to fire Domino effect due to overpressure: Explosion energy, kJ Peak overpressure, kPa Probability of domino effect due to overpressure Domino effect due to missile: Explosion energy, kJ Missile velocity, m/s Probability of domino effect due to missile after meeting target NA 0.95 NA = Not applicable. Scenario 1: High pressure in the vessel causes the pressure-relief valve (at the top of the vessel) to open, which leads to a continuous release of ammonia to the atmosphere until 80% of the chemical is released. Scenario 2: Due to improper maintenance or other problems, a leak develops in the vessel’s input or output pipeline. The leaking area is believed to be 40% of the pipeline’s cross-sectional area. This scenario is modeled as continuous release of liquid ammonia near ground level causing subsequent evaporation and dispersion. Scenario 3: High pressure develops in the vessel either due to overfilling or to a runaway reaction. The instantaneous release of high pressure causes the vessel to fail as a boiling-liquid, expanding-vapor explosion (BLEVE), and the released chemical disperses into the atmosphere. Scenario 4: Excessively high pressure develops in the vessel beyond the design capacity of the pressure relief valve. This causes vessel to burst as a confined vapor cloud CEP November 2001 www.cepmagazine.org 63 Safety explosion (CVCE). The instantaneously released chemical disperses into the atmosphere. Scenario 5: Ammonia is released from the joints, causing a pool of liquid to form. The released chemical subsequently evaporates into the atmosphere and disperses. Discussion These five scenarios were assessed for the credibility estimations and results are presented in Table 2. It is evident from the table that Scenarios 1, 2, 3 and 5 come in the range of maximum credible range. Scenario 4 falls into the uncertainty range, and although it poses a considerably high damage potential, credibility is nullified due to its very low probability. However, since Scenario 2 has a high frequency of occurrence, but does not pose much threat, still, it does fall in maximum credibility range. The other scenarios have adequate damage potentials and probabilities of occurrence. Among MCAS, Scenario 1 engulfs largest damage area. The short-listed scenarios were processed for detailed consequences and the summary of the results is presented in Table 3. As per the study, Scenario 1 poses a severe threat due to a toxic load over an area of more than a 2,500-m radius. The damage potentials of Scenarios 2 and 5 are limited to areas of ~1,000-m radius. Note that Scenario 3 is not only vulnerable for toxic load (over an area of ~1,270-m ra- Literature Cited 1. Khan, F. I., and S. A. Abbasi, “Techniques for Risk Analysis of Chemical Process Industries,” J. of Loss Prevention in Proc. Ind., 11 (2), p. 91 (1998). 2. Van Sciver, G. R., “Quantitative Risk Analysis in the Chemical Process Industries,” Reliability Eng. & System Safety, 29, p. 55 (1990). 3. “Guidelines for Chemical Process Quantitative Risk Analysis,” Center for Chemical Process Safety, AIChE, New York, p. 125 (1989). 4. Popazoglou, I. A., et al., “Probabilistic Safety Analysis in Chemical Installations,” J. of Loss Prevention Proc. Ind., 5 (3), p. 181 (1992). 5. Kafka, P., “Probabilistic Safety Assessment: Quantitative Process to Balance Design, Manufacturing and Operation for Safety of Plant Structures and Systems,” Principal Division Lecture, Trans. SmiRT, 11, p. 23 (1991). 6. Kafka, P., “Important Issues Using PSA Technology for Design of New Systems and Plants,” GRS mbH, Garchirg, Germany (1993). 7. Hirst, I. L., and D. A. Carter, “A ‘Worst Case’ Methodology for Risk Assessment of Major Accident Installations,” Proc. Safety Progress, 19 (2), p. 78 (2000). 8. Laplante, A., “Too Close to Home: A Report on Chemical Accident Risks in the United States,” U.S. Public Interest Research Group, Washington, DC (1998). 9. Khan, F. I., and S. A. Abbasi, “Risk Assessment in the Chemical Process Industries: Advanced Techniques,” Discovery Publishing House, New Delhi, India, p. 393 (1998). 10. Khan, F. I., and S. A. Abbasi, “Risk Analysis of a Typical Chemical Industry Using ORA,” J. of Loss Prevention Proc. Ind., 14 (1), p. 59 (2001). 11. Lees, F. P., “Loss Prevention in the CPI,” Butterworths, London, pp. 26–28 (1996). 64 www.cepmagazine.org November 2001 CEP dius), but also for the shock wave developed due to the BLEVE. The damage-causing shock wave would be operative over an area of ~250-m radius. It was concluded that current safety measures were not adequate and needed review in order to reduce the risk/hazard potential to a tolerable level. It was advised that safety-related decision-making (planning for disaster management or emergency planning) should be done, conCEP sidering Scenarios 1 and 3. FAISAL I. KHAN is currently a visiting research professor at the Memorial Univ. of Newfoundland (MUN) (Faculty of Engineering & Applied Science, MUN, St John’s, NF, A1B 3X5, Canada; Phone: (709) 737-7652 or 8963; Fax: (709) 737-4042; E-mail: fkhan@engr.mun.ca). Before moving to MUN, he had served about two years at the Birla Institute of Science and Technology (BITS) in Pilani, India. He also headed the Computer Aided Environmental Management unit at the Centre for Pollution Control and Energy Technology for four years. Khan has over 70 research publications, along with four books to his credit. He is the coauthor of “Risk Assessment in the Chemical Process Industries: Advanced Techniques.” He is a cowinner of the S. K. Mitra Award of 1998. Khan holds BS and ME degrees from Aligarh Muslim Univ. and the Univ. of Roorkee in chemical engineering and computer-aided process plant design, and he has secured II and I rank, respectively, with first-class honors. He has a PhD from Pondicherry Univ. in environmental systems engineering with special reference to risk assessment and environmental impact assessment. 12. Khan, F. I., and S. A. Abbasi, “Studies on the Probabilities and Likely Impacts of Chains of Accidents (Domino Effect) in the Fertilizer Industry,” Proc. Safety Progress, 19 (1), p. 45 (Spring 2000). 13. Skelton, B., “Process Safety Analysis: An Introduction,” Gulf Publishing, Houston, p. 201 (1997). 14. Khan, F. I., and S. A. Abbasi, “Models for Domino Effect Analysis in the Chemical Process Industries,” Proc. Safety Progress, 17 (1), p. 121 (1998). 15. Khan, F. I., and S. A. Abbasi, “Major Accidents in the Process Industries and Analysis of Their Causes and Consequences,” J. of Loss Prevention Proc. Industries, 12, p. 361 (1999). 16. Khan, F. I., and S. A. Abbasi, “The Worst Chemical Industry Accident of the 1990s — What Happened and What Might Have Been: A Quantitative Study,” Proc. Safety Progress, 18 (1), p. 135 (1999). 17. Hagon, D. O., “Use of Frequency-Consequence Curves to Examine the Conclusion of Published Risk Analysis and to Define Broad Criteria for Major Hazard Installations,” Chem. Eng. Res. Dev., 62, p. 381 (1984). 18. “The Tolerability of Risk Formation from Nuclear Power Stations,” Health and Safety Executive, HM Stationary Office, London (1998). 19. Ale, B. J. M., “Risk Analysis and Risk Policy in the Netherlands and EEC,” J. Loss Prevention Proc. Ind., 4 (1), p. 58 (1991). 20. Scheffler, N. E., “Improved Fire and Explosion Index Hazard Classification,” Proc. Safety Progress, 13 (4), p. 214 (1994). 21. Khan, F. I., and S. A. Abbasi, “Hazard Identification and Ranking (HIRA): A Multi-Attribute Technique for Hazard Identification,” Proc. Safety Progress, 17 (3), p. 16 (1998). 22. Khan, F. I., and S. A. Abbasi, “Accident Hazard Index: A Multi-Attribute Scheme for Process Industry Hazard Rating,” Trans. IChemE (Environmental Protection and Safety), 75B, p. 217 (1997).