Modern Risk Analysis based on PRA - Essay Example

?Topic: Modern Risk Analysis based on PRA XXXXXXXXXX XXXXXXXX XXXXXXX XXXXXX XXXXXXXXXX Introduction Probabilistic Risk Analysis (PRA) is a procedural use of probability distributions in the determination of levels of uncertainty associated with estimates of risk. Variables that define a risk equation may be one or more in number, and the variables are defined as a probability distribution in a probabilistic risk assessment. The risk equation does not define the variables as single numbers in the probability distribution. This risk assessment leads to an output of ranging risks prior to the experiences of the receptors. Proper probabilistic risk assessments require adequate description of the input parameters. For this to happen, it requires that distributional data be available and further be adequate in describing the already identified input parameters. PRA purely employs probability and probability distributions in the characteristic analysis. Probability denotes the chances of occurrence of an uncertain phenomenon. The uncertainty constitutes occurrence of risk. Through PRA, risk assessments can be carried out and the levels of risk therein identified. Use of probability in the assessments of risk can be used in the understanding, quantifying and management of risk. Such a process can further be analysed in relation to the limitations of quantifying risk using probability. Reasons why probability is chosen as opposed to other measures of uncertainty Probability quantifies the description of levels of risk, characterized by the aspect of uncertainty or variability associated with risk estimates. Risk therefore becomes comprehensively characterized by using probability, which would not be the case when point estimate measure of risk is used (UKOOA, 2006, pg. 134). This feature therefore makes probability a better measure of risk than the point estimate method. Quantitative analysis of risk allows for diverse treatment of uncertainty variables used in the determination of underlying risks prior to the probability of occurrence of the identified risks. Risk management requires the use of flexible tools of analysis as well as availability of vast information in regard to what is being accounted for. In this regard, probability becomes the best measure of uncertainty. Uncertainty constitutes risks, and it is the risks that risk managers ought to address. Probability allows for flexibility in the analysis and treatment of variables in the probability equation (U.S. Environmental Protection Agency, 2005, pg. 78-79). Following this, a large volume of information can be derived, thereby allowing risk managers to make choices among alternatives. Other measures of uncertainty do not provide for flexibility and variant information, making risk managers fixed to the choices they make or to the scope that they can manage risks using such measures as a baseline. Managers need to assess and evaluate high-end risks, and the best gateway to succeeding in that is using probability in measuring levels of risk. Limitations of quantitative approach to risk The quantitative approach to risk is time consuming. It is procedural and treatment and analysis of variables require adequate time. Step by step consideration of variables is necessary in order to ensure that each and every aspect that constitutes risks is identified and accounted for (Stern and Fineberg, 1996, pg. 157). This process is characterized by huge requirement of resources. Adequate resources need to be pulled into place. Finances are required at every step of the quantification process. Gathering data and information is expensive. The management should be aware of these requirements before such an approach to risk is adopted. In order to come to a critical and fundamental understanding of the risks through the quantification approach all required information should first be in place. Data collection, cleaning, editing, analysis and reporting require that enough resources be allocated for the purposes. Knowledge and skills are also great component of quantitative approach to risk (Costa-Gomes, Crawford and Broseta, 2001, pg. 1198). A number of professionals are required in order to make this approach operational. It draws together assessors, reviewers and risk managers. These personnel need to be knowledgeable and skilled enough in order to make this approach fully operational. The results of the model adopted also depend on the expertise of the team in place to deal with risks using the quantitative approach. This aspect is considered complex and comprehensive as compared to other approaches, for example the point estimate approach. Over and above this fact, many input variables used in the quantitative approach lack frequency distributions that are exhaustive in documentation. Management of risks in the context of decision theory The use of probability in the assessments of risk is not meant to produce probability numbers and leave the process at that. Risk analysis in the context of probability application is tailored towards supporting and improving decision making (Aven, 2003, pg. 105). Risk and uncertainty employs prediction ad forecasting of future happenings in the present day. Decisions about the future necessitate making them today in order to maintain functionality of an entity over time. Risk management and decision making have to be integrated in order to minimize risks and maximize the expected future benefits. In order to achieve this, the decisions made need to be good, rational and should fully account for the problem that characterizes the decision being made. Probabilistic risk assessments provide an understanding of underlying risks based on the analysis of those risks. Once risks have been identified, they are used to influence the decisions made by risk managers in regard to the welfare and performance of the firm. Risk management is prioritized once the risks have been identified. The firm and the firm at large must work its way out the risks. Risk management takes a number of alternatives. These are: reducing the risks, eliminating the risk, transferring the risks or ignoring the risks (Camerer, 2005, pg. 130). Each alternative affects the decision making process independently, thereby determining whether the decision made is appropriate or note. For example, the applicability of each identified risk management option on cost and benefit analysis vary. On the same note, making a Bayesian decision analysis in the context of risk management requires drawing of rational inferences from the risk analysis process (Aven, 2003, pg. 106). Decision making based on the risk management criteria relies on an optimization model prior to the decision being made. Optimization requires that choices be made among alternatives. The alternatives are directly or indirectly related to the optimization problem, such that the alternative that best suits the optimization model is selected. This may be in a maximization or minimization problem, based on the interest of the optimization model. Decision making is also characteristic of formal risks and analysis; the reason why risk management is integral to the process. Managerial judgments can therefore be based on risk analysis and management, allowing for review processes that step by step result into a rational decision. The figure below is a summary of an integration of risk management and decision making, where the decision made is rational based on probability, risk analysis and the opinion of the managerial team. Figure 1: Risk Management and Decision Making Source: Aven, A (2003, pg. 108) Role of Probability in Understanding and Quantification of Risks For one to understand and quantify risks there is need to follow the Ten Step Procedure for Conducting PRA (Camerer, 2005, pg. 133). This will be discussed below. Role of Probability in Understanding Risks Methodology Definition This is the first step in formulating a PRA. It involves reviewing analytical options which may include using a computer program, consulting experts among others. After this, the requirements and necessities for the assessment should be evaluated. Merits and demerits of each analytical option are now selected depending on the advantage and disadvantage of each option (NORSOK, 2001, pg. 98). Familiarization and Information Assembly It is vital for one to become familiar to the process or system under consideration. Among those to be investigated include the safety measures, standard procedures and physical layout. This will help in identification of what could go wrong and how this can be corrected. There is need to review any information about failure that have occurred in the past. All details gathered in this step should be well documented for future reference and review. Identification of Initiating Events Internal initiating events should be formed in this third step. These are abnormal events in the system that might lead to a risky exposure if not properly remedied. This can be done by brainstorming or by use of Failure Mode and Effects Analysis (FMEA) (Kunreuther & Heal, 2002, pg. 220 – 226). The vital role here is to discover all possible internal threats and hazards. Initializing events are then group according to the similarity of characteristics. For example, all events that require the same prevention measure or contribute to a similar risk are grouped together. This helps in the next steps and also in enhancing concentration on significant category rather than redundant event. Sequence or Scenario Development This step involves doing an enquiry on the consequences of internal initiating events. This is mainly done by employing event tree analysis. An event tree is a figure that describes all the possible results of an initiating event. An event tree is developed by starting with the initiating events and then identifying all its consequences. These consequences are then treated as new events and other consequences are identified. This goes on and on. This process leads to growth of the trees (Moyer, McGuigan and Kretlow, 2009, pg. 225). An example of a decision tree can be formed by making an initializing event to be accidental fire. This has its consequences which are activation of the sprinkler system and an emergency call to a fire department. The sprinklers then become the next event. There are two possible consequences which are succeeding or failing. In the next adjacent branch, there is succeeding or fail for the fire fighters emergency call. After the fire fighters, there are no further consequences and this mark the end of the tree. Figure 1: The Event Tree (Example event tree courtesy of Relex Software System Analysis [http://www.event-tree.com/].) Event trees contain vital points. This may include points where a critical safeguard may or may not function for which there is little information. Using the example above, there is concern to establish how likely the sprinkler system is to work. To get this information, an internal system analysis should be performed. Fault tree analysis is a useful at this point in time. It is similar to an event tree since it starts with an event. However, unlike the event tree, it traces the causes rather than following the consequences. Figure 2: Fault Tree Internal Events External to the Facility Here, any event that spreads beyond the organization or facility boundary but originates within the system is considered. These are events such as internal fires or energy discharges. These events are rare but when they occur, they should be dealt with as internal events in the third to fifth step. External Events In this step, the last category of events is put into consideration. These are the ones originating from outside the system and may include factors such as storms, political violence among others... They are analyses as internal events in the third to fifth step. Dependent Failure Considerations As long as a system is independent, redundant systems are installed as a risk mitigation procedure. This is actualized by creation of multiple success paths. Elements that are susceptible to common-cause failures should be explicitly integrated into fault and event tree if this is possible. Failure Data Analysis Failure data for all the events in the fault and event tree should be obtained. This may be from past experience or expert’s opinion. This data is important for outages due to test, repair, and maintenance (Kunreuther & Heal, 2002, pg. 235 – 237). Quantification: Role of Probability Quantification of Risk. Data acquired from step 9 is used fault and event tree sequences. This are reduced to get probability and severity information. Using the example above, assumed the data obtained has the following probabilities and severity: Probability (sprinkler system success): 0.8 Probability (fire department success | sprinkler success): 0.95 Probability (fire department success | sprinkler failure): 0.5 Cost (fire handled OK): – $1,000 Cost (partial damage): – $500,000 Cost (system destroyed): – $3,500,000 Figure 3: Event Tree with Quantification From figure three the probability of an OK outcome will be (0.9 X 0.95) = 85.5%; the probability of a Partial Damage outcome) will be (0.9 X 0.05) + (0.1 X 0.7) = 11.5%; and the probability of system destruction will be (0.1 X 0.3) = 3%. After combining the probability and severity information, the overall average risk of accidental fire damage can be determined as: (0.855 X $1,000) + (0.115 X $500,000) + (0.03 X–$3,500,000) = –$163,355. Communication of Risk Presenters of risk information from a PRA should determine the critical information and level of detail to be presented to stakeholders. Steps for communication involve (1)determine the audience, (2) determine the needs of the audience, (3) formulate a communication plan, (4) assure clarity of presentation, (5) present information, (6) post-meeting review of presentation and community feedback, and finally (7) update information where necessary for future assessments and presentations (Kunreuther & Heal, 2002, pg. 245 – 249). Conclusion PRA is heavily employed in modern risk assessment despite its limitations. This is because of its strength in empirical verifiability. The various steps of should be followed in the understanding, quantification and management of risk using probabilities. Communication of result of a PRA should be clear. A risk manager should be conscious of how this information could be received. A review of presentation and community feedback is necessary for future assessments and presentations. References Aven, A, (2003), Foundations of Risk Analysis: A Knowledge and Decision-Oriented Perspective, Terje Aven: John Wiley & Sons, Ltd. Camerer, C. (2005), Three cheers – psychological, theoretical, empirical – for loss aversion, J. Marketing Res. 42, 129 – 133. Costa-Gomes, M., Crawford , V.P. , and Broseta , B. (2001), Cognition and behavior in normal-form games: an experimental study, Econometrica 69 , 1193 – 1235. Finley, B. and D. Paustenbach. (1994), The benefits of probabilistic exposure assessment: Three case studies involving contaminated air, water, and soil, Risk Analysis 14(1): 53-73. Grinband, J., Bayer, H., Figner, B. et al. (2007), Neural correlates of risk and return in risky decision making, Working Paper, Center for the Decision Sciences (CDS), Columbia University. Holt , C.A. and Laury , S.K. (2001), Risk aversion and incentive Effects, Am. Econ. Rev. 92, 1644 – 1655. Kunreuther, H. and Heal, G. (2002), Interdependent Security: Journal of Risk and Uncertainty. 26:201-49. Moyer, R., McGuigan, J. and Kretlow, W. (2009), Contemporary Financial Management, 6th ed, New York: West Publishing. NORSOK, (2001), Risk and emergency preparedness analysis, Z-013, Oslo: Norwegian Technology Standards Institution. Rugen, P. and Callahan, B. (1996), An Overview of Monte Carlo: A fifty year perspective. Human Health and Ecological Risk Assessment 2(4): 671-680. Singpurwalla, N.D. (2002), Some cracks in the empire of chance, International Statistical Review, 70: 53–79. Stern, P.C. and Fineberg, H.V. (1996), Understanding Risk, Washington DC: National Academy Press. U.S. Environmental Protection Agency. (2005), Guidance on Risk Characterization for Risk Managers and Risk Assessors, Washington, DC: Risk Assessment Council. UKOOA (2006), A framework for risk related decision support. Industry guidelines, London: Offshore Operators Association. Read More