Microsoft Word - Efficient Applications of Risk Analysis in the Chemical Industry and Emergency Response Efficient Applications of Risk Analysis in the Chemical Industry and Emergency Response Roberto Bubbico Dipartimento di Ingegneria Chimica – Università di Roma “La Sapienza” Via Eudossiana 18, 00184, Roma (Italy) Abstract Despite it is now used in many technical and industrial areas, Risk Analysis is sometimes still considered by many plant managers as a methodology too complex and too time- consuming to be of practical use. Conversely, it will be shown that Risk Analysis can of- ten provide very useful information on the safety of selected industrial activities. Also, the introduction of proper simplifying assumptions can even wider the range of applica- tion with great benefit of all stakeholders. Keywords: Risk Analysis, Safety, Chemical industry, Hazardous materials, Emergency response 1. Introduction The introduction of the first techniques for the calculation of the risk posed by a given technical activity, dates back to the end of the 1970s in the area of the nuclear and aeronautics industries. Some of them are nowadays quite well known among practitioners, such as the Probabilistic Risk Assessment (PRA)1,2, the Prelimi- nary Hazard Analysis (PHA), and so on3,4. However, in the recent years, these techniques, or some modifications of their original versions, have been adopted to a large extent, also in many new areas, such as finance, medicine, biology, etc., different from the native ones5-7. This is because of the high flexibility of the methods and to the capability of a very detailed and powerful analysis of the system under study, whether an industrial installation, an informatics network, a fi- nancial system and so on. Among the many methodologies availa- ble, the quantitative approaches are able to provide more detailed and useful in- Journal of Risk Analysis and Crisis Response, Vol. 1, No. 2 (November 2011), 92-101 Published by Atlantis Press Copyright: the authors 92 formation but, on the other hand, they are usually much more time-consuming and demanding in terms of computational ef- fort. Furthermore, due to the uncertainties in the input data or to the lack of adequate information, in some cases the accuracy and reliability of the results are a matter of discussion. Despite these difficulties, Risk Analysis has gained a good reputation among prac- titioners, especially in its quantitative version (Quantitative Risk Analysis, QRA)8,9. With specific reference to the area of the chemical and process indus- tries, the main problems involved in its use will be here critically analysed and a number of useful examples of application presented. 2. Risk Analysis Risk Analysis is a powerful and compre- hensive methodology that can be adopted to assess of the level of the risk associat- ed with a given activity (industrial, eco- nomical, etc.)5,6. It is a stepwise proce- dure and more than one single component technique are available for performing each of the tasks within the whole proce- dure. The need for a stepwise procedure is giv- en by the many aspects which must be considered to estimate the value of the risk. These aspects depend on the area of application, the risk possibly being linked to a financial loss/income, to the safety of an industrial activity and so on. In the specific case of the chemical industry, different variations can be found in the definition of risk in the literature (see for example the paper by Kaplan and Gar- rick9, for a more thorough analysis), but it is generally agreed that risk is a function of a combination of the impact of a num- ber of selected hazardous events, and of their probability/frequency of occurrence. Of course, based on this definition, the value of this function strongly depends on the identified harmful events (accident scenarios). As a consequence, in order to estimate the risk, the following four tradi- tional main steps composing the whole analysis are required:  Hazard identification  Consequence calculation  Frequency estimation  Risk evaluation As mentioned above, in order to accom- plish each of these steps, different tech- niques can be adopted, thus providing the methodology with a high level of flexibil- ity. In fact, on one hand, the methodology can be applied to any stage of the lifetime of an industrial installation: from the ini- tial preliminary design stage, to engineer- ing phase, up to the actual operation of the plan, possibly taking into account all structural and procedural changes (man- agement of changes). On the other hand, the methodology can be applied to a vari- able level of detail for the same stage of the plant: all the specific pieces of equipment can be studied, along with the associated accidents and their frequency of occurrence and magnitude of conse- quence; or larger subsystems of the plant can be, at least preliminarily, considered as a black box, to allow a more general and quicker analysis to be carried out. Of course, once the critical subsystems have been identified, they can be analysed to a higher level of detail. Another very important aspect is that the methodology can be either qualitative or Published by Atlantis Press Copyright: the authors 93 quantitative. From this point of view, it worth noting that different conclusions can be drawn by these two approaches. A typical example: it is generally assumed that an accident deriving from one single initiating event, is more critical than one occurring from the combination of two or more initiating events. This sounds rea- sonable. However, when quantitatively analysed, the frequency of occurrence of the first accident can result even much smaller than that of the second one, thus revealing a misleading conclusion of the qualitative approach. Therefore, when planning a risk analysis study, it is im- portant to clearly state since the begin- ning the scope and aim of the analysis. Besides this aspect, it must be said that some of the techniques and models used for performing these tasks can be rather complex and often require skilled and ex- perienced personnel. And even in the case of relatively simple models, a minimum level of uncertainty and approximation is implicit. As examples of such complex models, in the area of consequence calculation, the mathematical models adopted for the as- sessment of the fraction of aerosol gener- ated after a liquid release, or those used for the calculation of the dispersion of a heavy gas, can be mentioned. In the case of the dispersion of a heavy gas in a com- plex environment, such as a urban area or a congested industrial site, the application of these models becomes even more dif- ficult and usually require a trained ana- lyst, familiar with the fluid-dynamics and thermodynamics. Toxicological information are available only for a limited number of substances, and therefore, even when dispersion can be assessed rather accurately, effect mod- els cannot be used without a high level of uncertainty. Other difficulties arise when there is a lack of historical data required as input parameters for some models. For exam- ple, the values of the frequency of occur- rence of some events (release of a materi- al from a containment system, rupture or failure of a piece of equipment, the igni- tion of a flammable mixture, and so on) are hardly found in the literature, or, when available, they are referred to sys- tems/conditions other than those under investigation. In these cases the needed parameters have to be estimated by means of specific techniques (Fault Tree Analysis, Event Tree Analysis, etc.), which represent another important source of uncertainty. In fact, besides the quality of the data used to quantitatively solve these models, even the definition of the model itself is prone to uncertainties and errors. Based on the above considerations, it ap- pears that the inherent structure of Risk Analysis, on one hand gives rise to a very flexible and powerful methodology for the assessment of the safety of an indus- trial activity, but, on the other hand, is a source of uncertainty and errors. Also, the higher the accuracy of the results re- quired, the longer the time and the calcu- lation burden for performing the analysis. Just to represent more clearly the uncer- tainties we are dealing with in this area, it can be useful to mention a famous study carried out in the past at a European lev- el10,11: a number of Risk Analysis teams were provided with the same system to analyse, with specific indications on the techniques to use, and so on. Yet, the Published by Atlantis Press Copyright: the authors 94 possible assumption of hypotheses, the selection of the incidents to study and of the input data (frequencies, and so on) were on their own judgment. At the end of the study, the results obtained by the different teams ranged over several order of magnitudes, and the influence of the many uncertainties was not enough to ex- plain this variability. Only when much more restricting instructions were given to the teams, a much more acceptable variability was obtained. Based on these considerations, a strong debate sometimes arises among practi- tioners, especially process engineers and plant personnel, about the actual reliabil- ity and effectiveness of the methodology. Different arguments are also debated in the literature12,13. Despite these difficulties and sources of uncertainties, Risk Analysis still repre- sents a useful and sometimes unique methodology for the assessment of the safety of a given industrial activity, espe- cially when a comparative use is to be carried out: e.g. when different layouts of the plants or alternative production pro- cesses have to be evaluated. This is now generally recognized and in fact the use of this technique is nowadays also rec- ommended by different regulations in force in Europe14 as well as in other Countries all over the world. 3. Examples of application In the following paragraphs, a number of examples of application of the methodol- ogy to various industrial activities will be reported. However, besides the “conven- tional” use of Risk Analysis, it must be noted that its component techniques are widely used in a number of other applica- tions: consequence evaluation15-17 , deci- sion making18-20, process design21 and others. 3.1. Transportation Risk Analysis Any process industry managing hazard- ous substances as raw materials or final products to be sent to the market, requires the transportation of these materials to/from the production site. This specific task has to be considered as an integral part of the general production activity. As a consequence, in the framework of the assessment of the risk connected with the whole production activity, the analysis of the risk associated with the transportation phase has to be taken into consideration along with that connected with the other production units. This has been recog- nized recently, and a specific version of the original Quantitative Risk Analysis technique has been set up, usually re- ferred to as Transportation Risk Analysis (TRA)22. This is an extension of the tradi- tional quantitative risk analysis technique, with the distinctiveness that the risk source is not in a stationary location, but is moving along a given route in a con- tinuously changing environment. There- fore, in order to apply the technique to transportation activities, the knowledge of a large amount of information is required, partly strictly connected with the territory characteristics, and partly relative to other parameters. Just to mention a few exam- ples: the local distribution of population along the route, the site-specific accident rates, the local weather conditions, the accidents evolution in relation with the local specific characteristics, and so on. The need of this considerable amount of information often prevent the application Published by Atlantis Press Copyright: the authors 95 of this very useful methodology to practi- cal cases, with heavy detriment of the safety for the involved working person- nel, for the exposed population or for the environment. In order to overcome these complexities, different approaches have been proposed in the literature, and a number of methods of implementation of the technique have been devised (see [15] for some refer- ences). In most of the cases, the proposed varia- tions are simplified versions of the more rigorous approach23,24. For example, a first solution can be to distinguish the pa- rameters which are really route- dependent from those which are not23. In the first group can be considered the re- lease scenarios, the probability of a release scenario after a given accident and the probability of an outcome case following a given release scenario. Other parameters, such as the accident rate, the population distribution and the weather conditions, have to be considered as route-dependent factors. However, some simplifying assumptions are possi- ble also in this case, at least as a first ap- proximation: the accident rate can be considered as a function of the type of road (highway, state-road, etc.); typical values of population densities can be as- sociated to a reduced number of different types of built-up areas ranging from ur- ban to rural, and so on. All these assump- tions allow to reduce the amount of in- formation to be collected. Of course, each simplification will in- volve advantages and disadvantages. In summary, the pros consist in the availa- bility of a less demanding procedure, re- quiring a reduced amount of input data, and in a more “user-friendly” approach, allowing the technique to be applied even by less experienced practitioners. On the other hand, the use of too many or too “heavy” simplifications can lead to a less than acceptable accuracy of the results; this can finally result in the adoption of erroneously biased decisions during risk management. Even if the accuracy of the results cannot be expected to be very high, in some cas- es, such as a preliminary analysis, it can provide very useful information and al- low an efficient use of time. In Figure 1 a comparison between a simplified and a rigorous approach is shown. Fig. 1. Comparison of F-N curves obtained by a simplified and a rigorous approach [10] In fact, a similar simplified approach has been proposed in the literature25 to identi- fy critical areas along a specific route and thus provide effective strategies, such as a different distribution of the dangerous goods traffic, alternative routes and/or limitation to the transportation itself. Al- so, the adoption of techniques based on the graph theory allowed to improve the planning of the emergency phase. A more efficient way to manage the TRA calculations involves the use of Geo- graphic Information Systems (GIS). By 1E-09 1E-08 1E-07 1E-06 1E-05 1E-04 1E-03 1E-02 1 10 100 1000 10000 N F ( # /y r) rail road Published by Atlantis Press Copyright: the authors 96 this methodology, a tool capable of per- forming in a simple and relatively quick way all the steps of a TRA can be de- vised. The consequences of an accident are in- fluenced by many parameters, and some of them depend on the specific location (e.g. accident rate, on-route and off-route population, weather conditions and so on). GIS databases are able to provide some of these data already directly linked to the location along the route network. Different commercial GIS databases are already available for almost any Country in the world, providing at least some basic information (road and rail network, land use, and so on). Nonetheless, the values of other im- portant parameters (accident rates, popu- lation and meteorology) have to be col- lected from different sources and then manually introduced in the GIS system. Depending on the available data and their initial format, these phases (data acquisi- tion and manipulation) may be quite long (e.g. see [26, 27]). Besides the GIS sys- tem, a database containing information on the probabilities of occurrence of all the possible accident outcomes, and on the magnitude of the consequences of such dangerous events under the selected refer- ence weather conditions, has to be pre- pared. Of course, these data are available only after the chemical products of interest and the weather conditions to be associated with the considered accident have been de- fined. The larger the number of chemicals and/or the number of weather conditions adopted, the “heavier” the database and the more accurate the calculation results. By such a GIS system, the calculation proce- dure is very simple: once the product and the schedule of the transportation (number of trips per season) have been defined, and once the route of interest has been selected, all the related data are automatically up- loaded by the software and used for the calculations. In Figure 2, the selection of a specific route between given origin and destination is shown. Fig. 2: Selection of the route of interest on the GIS map [26] The local information about many differ- ent parameters (population, road charac- teristics, etc.) can be automatically dis- played on the same map by simply click- ing on the proper GIS tool buttons (Fig- ure 3). Fig. 3: Site-specific information on the GIS map [26] Road Seg- ment ID Probab. Wind Di- rection Weather Conditions Popula- tion D Published by Atlantis Press Copyright: the authors 97 Both the individual risk as a function of the distance from the route, for each route segment, and the societal risk, in terms of the cumulative F-N curve of the route, can be finally displayed (see Figures 4 and 5). Sometimes, the procedure can be also speeded up by using one of the GIS selec- tion tools, such as the fastest or shortest route selection option. Fig. 4: Individual risk as a function of the dis- tance from the road. Fig. 5: Societal risk: F-N curve for the whole route selected. If different routes can be chosen for the same transportation activity, the construc- tion of the risk curves for each of them would allow to identify the one which is characterized by the lower level of risk. At the same time, the knowledge of the individual risk profiles along the selected route make it possible to locate the most hazardous spots: this would allow, for example, to slightly modify the route, achieving a lower risk value with minor changes and, correspondingly, with neg- ligible cost increases. It is thus apparent that such a tool is able to permit a fast and relatively accurate investigation of the alternative routes or transportation modalities for a given ac- tivity, and to provide a comprehensive risk management tool. Different examples of application can be found in the literature. Here, reference will be made to a previous study28 where the overall risk associated with the trans- portation activities of hazardous materials in a whole region of Italy has been re- duced by simply modifying the distribu- tion of the transported chemicals in terms of transport modalities and schedule. The data concerning the transportation of dangerous goods in Sicily (Italy) were obtained from different sources (Civil Defence Office of the Prefettura of Mes- sina, Federchimica, Italian railways com- pany, harbour offices, etc.). Also, the routes travelled by the products were identified with a reasonable degree of ac- curacy, thus allowing to get the specific accident frequencies and other character- istics of the travelled routes. The analysis of the data allowed to identify 51 road transportation cases and 4 rail ones, and the use of a specifically devised GIS tool led to the construction of the following overall F-N curve (Figure 6): Published by Atlantis Press Copyright: the authors 98 Fig. 6: Societal risk for the initial transporta- tion activity. By re-arranging and optimizing the dis- tribution of the transport among road, rail and intermodal and, consequently, modi- fying the amounts of materials transport- ed and the routes travelled, a new risk curve is obtained. The advantages of the new organization of the transport, is demonstrated by the risk reduction shown in Figure 7, where a comparison between the two curves (before and after re- distribution) is reported. Fig. 7: Societal risk for the optimized trans- portation activity. 3.2. Emergency support tool A GIS application quite similar to that set up for the analysis of the risk associated with the transportation of hazardous ma- terials, can represent a useful support tool in the case of an emergency. In fact, in the case of an accident, if its location is identified on the relevant GIS map, the im- pact areas of all the possible outcome cases associated to the corresponding accidental scenario (dependent on the substance, the type of release, the environmental condi- tions, etc.) can be easily displayed on the map itself. If specific information, such as the actual wind direction at the time of the accident, are available, a rather accurate prediction of the territory at risk (of fire in the case of a flammable release, of toxic exposure in the case of a toxic cloud, etc.) can be identified and very useful infor- mation can correspondingly be obtained: e.g. the urban areas to be evacuated by the population, the roads and/or railways to be prohibited or, conversely, those to be used by the emergency teams for reaching the location of the accident; similarly, if the location of centres of interest (fire brigade and police stations, civil protection offices, hospitals, etc.) are reported on the map, it might be possible to know what centres need to be activated following the emer- gency. In Figure 7 the possible impact are- as following the release of ammonia from a road tank are represented. The examples reported here are only a few of the many possibilities offered by such a kind of tool. 1E-06 1E-04 1E-02 1E+00 1 10 100 1000 10000 N F ( # /y r) Rail Total Road 1E-06 1E-04 1E-02 1E+00 1 10 100 1000 10000 N F ( # /y r Before mitigationAfter mitigation Published by Atlantis Press Copyright: the authors 99 Fig. 8: Impact areas following a release of ammonia from a road tank. 4. Conclusions A quick presentation of some of the most common applications of the Risk Analy- sis technique in the industrial sector and in the civil protection and emergency planning area has been reported in the present paper. It should be apparent that, despite the complexity of the procedure and the con- siderable effort in terms of calculation and data collection, Risk Analysis still represents in many cases the only method able to provide useful information on the safety and quality of a selected industrial activity. It represents a quantitative ap- proach and, despite the absolute values obtained cannot be considered as “the” value of the investigated risk, especially when applied in a comparative analysis, the results provided can be trusted with a sufficient level of confidence. Furthermore, when a number of appropri- ate simplifying assumptions are intro- duced, the application can be definitely easier and thus the range of applications and of possible users becomes quite larg- er with great benefit for all stakeholders: the responsible of the activity under in- vestigation and the exposed population. References 1. NUREG, U.S. Nuclear Regulatory Com- mission, “PRA procedures guide: a guide to the performance of probabilistic risk as- sessment for nuclear power plants”, NU- REG CR-2003, Washington D.C., 1984. 2. NUREG, U.S. Nuclear Regulatory Com- mission, “Probabilistic safety analysis procedures guide”, NUREG CR-2815, Washington D.C., 1985. 3. N. Hyatt, “Guidelines for Process Hazards Analysis, Hazards Identification & Risk Analysis”, Dyadem Press, Richmond Hill, Ontario, Canada, 2003. 4. CCPS, Center for Chemical Process Safe- ty, “Guidelines for hazard evaluation pro- cedures”, 2nd edition, AIChE, New York, 1992. 5. G. Koller, “Risk Assessment and decision making in business and industry- A prac- tical guide”, Chapman and Hall/ CRC, Boca Raton, FL, 2005. 6. T. Aven, O. Renn, “The role of Quantita- tive Risk Assessment for characterizing risk and uncertainty and delineating ap- propriate risk management options, with special emphasis on terrorism risk”, Risk Analysis, Vol 29/4, pp. 587-600, 2009. 7. H. Li, G.E. Apostolakis, J. Gifun, W. Van Schalkwyk, S. Leite, D. Barber, “Ranking the risks from multiple hazards in a small community”, Risk Analysis, Vol 29/3, pp. 438-456, 2009. 8. CCPS, Center for Chemical Process Safe- ty, “Guidelines for Chemical Process Quantitative Risk Analysis”, 2nd edition, AIChE, New York, 2000. 9. S. Kaplan, B.G. Garrick, “On the quantita- tive definition of risk”, ”, Risk Analysis, Vol 1/1, pp. 11-27, 1981 10. Amendola, “Uncertainties in systems reli- ability modeling: insight gained through European benchmark exercise”, Nuclear Engineering Design, 93, 215-225. Am- sterdam, The Netherlands, Elsevier Sci- ence, 1986 Imp a ct a re a s tox ic clo ud fla sh fire uvce jet fire po ol fire 2 0 2 4 Kilom e ters EW S N Published by Atlantis Press Copyright: the authors 100 11. Amendola, S. Contini, I. Ziomas, “Uncer- tainties in chemical risk assessment: re- sults of a European benchmark exercise”, J. of Hazardous Materials, 29, 347-363, 1992 12. Z. Ciechanowicz, “Risk analysis: re- quirements, conflicts and problems”, Computers and security, 16/3, 223-232, 1997 13. M. Elisabeth Paté-Cornell, “Uncertainties in risk analysis: six levels of treatment”, Reliability engineering and systems safe- ty, 54, 95-111, 1996 14. European Commission Council Directive, 96/82/EC, On the control of major acci- dent hazards involving dangerous sub- stances, 9 December 1996 15. J. Tixier, G. Dusserre, S. Rault-Doumax, J. Ollivier, C. Bourely, “OSIRIS: software for the consequence evaluation of trans- portation of dangerous goods accidents”, Environmental Modeling & Software, Vol. 17/7, 627-637, 2002 16. R. Bubbico, M. Marchini, “Assessment of an explosive LPG release accident: A case study.” J. of Hazardous Materials, 155, pp 558-565, 2008 17. P.A. Davies, F.P. Lees, “The assessment of major hazards: the road transport envi- ronment for conveyance of hazardous ma- terials in Great Britain”, J. of Haz. Mat., 32, pp. 41-79, 1992 18. I.A. Papazoglou, G.S. Bonanos, Z.S. Nivolianitou, N.J. Duijm, B. Rasmussen, “Supporting decision makers in land use planning around chemical sites. Case study: expansion of an oil refinery”, J. of Hazardous Materials, 71, pp 343-373, 2000 19. K.H. Harrington, S.E. Rose, “Wisely make investment decisions via risk map- ping”, Chemical Engineering Progress, 95/12, 63-66, 1999 20. Fabiano, F. Currò, E. Palazzi , R. Pastori- no, “A framework for risk assessment and decision-making strategies in dangerous good transportation”, J. of Loss Preven- tion in the Proc Ind., 93, pp. 1-15, 2002 21. F.I. Khan, S.A. Abbasi, “Inherently safer design based on rapid risk analysis”, J. of Loss Prevention in the Proc Ind., 11, pp. 361-372, 1998 22. CCPS, Center for Chemical Process Safe- ty, “Guidelines for Chemical Transporta- tion Risk Analysis”, New York: AIChE, 1995. 23. R. Bubbico, S. Di Cave, and B. Mazza- rotta, “Risk analysis for road and rail transport of hazardous materials: a simpli- fied approach”, J. of Loss Prevention in the Proc Ind., Vol 17/6, pp. 477-482, 2004. 24. C.B. Goh, C.B. Ching, R. Tan, “Risk analysis for the road transportation of haz- ardous chemicals in Singapore – a meth- odology”, J. of Loss Prevention in the Proc Ind., 8, pp. 35-39, 1995 25. Fabiano, F. Currò, A.P. Reverberi, R. Pas- torino, “Dangerous good transportation by road: from risk analysis to emergency planning”, J. of Loss Prevention in the Proc Ind., 18, pp. 403-413, 2005 26. R. Bubbico, S. Di Cave, B. Mazzarotta, “Risk analysis for road and rail transport of hazardous materials: a GIS approach”, J. of Loss Prevention in the Proc Ind., Vol 17/6, pp. 483-488, 2004. 27. R. Bubbico, S. Di Cave, B. Mazzarotta, “La gestione del rischio nel trasporto di merci pericolose via terra”, Convegno Scientifico Nazionale “Sicurezza nei sis- temi complessi”, III edizione, Bari, 19-21 ottobre 2005, Sessione III/2, (in Italian), 2005. 28. R. Bubbico, G. Maschio, B. Mazzarotta, M.F. Milazzo, E. Parisi, “Risk manage- ment of road and rail transport of hazard- ous materials in Sicily”, J. of Loss Preven- tion in the Proc Ind., Vol 19/1, pp. 32-38, 2006. 29. G. Purdy, “Risk analysis of the transporta- tion of dangerous goods by road and rail”, J of Haz. Mat., 33, pp. 229-259, 1993 Published by Atlantis Press Copyright: the authors 101