Risk analysis is a key feature of modern decision making, for both government and industry. While not yet mandated by government regulations, quantitative risk analysis is an increasingly preferred method of hazard evaluation based on numerical estimation of incident frequency and consequences. Internationally, there appears to be a gradual trend towards increasing use of quantitative risk assessment methods in the chemical process industry – fully for the relatively small number of instances where they can be applied, and partially in semi-quantitative approaches. MKOPSC has been long involved in the research in the area of quantitative risk assessment, as well as provided consultancy to the industry. For example, the Center conducted several comprehensive quantitative subsystem hazards analysis study covering process, Chemical Agent Transfer System, electrical power system, fire protection system, Heating, Ventilation, and Air Conditioning system, Ton container decontamination system, product offsite transportation, and their related support systems for Newport Chemical Demilitarization Facility, which is designed to destruct VX stockpile. All the undesired events can be reduced to an acceptable level with suggested changes of process design change, operating specifications, and maintenance procedures. The safety impacts of all risk reduction suggestions have been quantified and discussed as well.
Research activities in this area are highlighted in the following sections.
Risk analysis and management is becoming increasingly important to the process industry to meet safety criteria and regulations. As a powerful and systematic tool, fault tree analysis (FTA) has been adapted to the particular need of chemical process quantitative risk analysis and found great applications.
The central idea of this research is to capture the cause-and-effect logic around each item of equipment directly into mini fault trees (Wang, 2004). The results generated from the program have been compared to published results and verified to be correct. This work proposes a computer-aided methodology for fault tree synthesis (Wang et al, 2002). Many special features are designed to prevent incidents, which are crucial in the synthesis of fault trees. The prototype program developed in this research was to illustrate and test the methodology against case studies. The final product of this research is to provide a computer package that standardize the procedure of quantitative risk analysis and help decision makers to decide more formally and more cost-effectively. Last but not the least, though this work is originally developed for the application in the chemical process industry, it has the extendibility to other fields such as electrical systems, the nuclear industry, and the aerospace industry.
More than 3.1 billion tons of hazardous materials (HazMat) are shipped in the United States annually. According to the Department of Transportation (DOT) statistics, 156,442 Hazmat transportation incidents occurred during 1995 to 2004, resulting in a total of 221 deaths and 3,143 injuries. To limit these incidents, optimization of HazMat transport involves comparisons of alternatives in the domain of risk, as well as an analysis of the tradeoffs between cost and risk. The numerical methodology developed to assess the transportation risk was hard to employ directly by decision or policy makers.
The procedure takes into account the effects of hazardous materials type, environmental, truck configuration, and road conditions to both the accident frequency and consequence. A numerical procedure, which allows the coupling of time effectiveness and mathematical accuracy, will be developed for the individual risk evaluation, and therefore provides criteria for the route selection of hazardous materials transportation. User-friendly software on transportation risk analysis and the route selection can be developed based on this research. With sufficient data, the incident frequency of different road could be measured given the data of affecting parameters, and then the general models could be built to assess the incident frequency for any kind of road. The transportation risk analysis results can be used in a decision-making system. The results from the risk assessment, cost evaluation, and routing methodology would be inputs to the system. Then the decision would be made about the emergency response, evacuation procedures, resource management, public protection and so on.
Quantitative risk assessment requires failure and event data. The ideal situation in quantitative risk assessment is to have sufficient in-house data, however, due to various restrictions, laboratory data or data from generic data sources are often used in reliability studies. Reliability data can often deviate by a factor of three or four, and a factor of ten is not unusual, as suggested by Kletz (1999). In particular, there is few reliability or failure rate data that has been collected for the equipment in the LNG industry. Usually failure rate data from nuclear, oil and gas industry like OREDA handbook are borrowed for reliability studies and risk assessment of the LNG facility. This brings uncertainty which is hard to predict and makes the analysis results unreliable and easy to attack.
An LNG reliability database has been suggested by several LNG industry companies. A survey to update GRI’s historical failure rate database (last data in 1980s) is being considered. Our idea is to gather available data from other industry such as oil and gas industry and cryogenic industry and adjust the data through Bayseian analysis to predict reasonable failure rate data for the equipment used in the LNG facilities. Unit operations fault tree models are being considered for the new plant designs, particularly the offshore facilities. An industry consortium is being considered for this activity. Upon successful execution of this project, it may be expanded to the chemical process industry as well.
In the chemical process industry it is easier to calculate the loss expenses after an accident has already occurred. However, calculating the monetary losses that the industry could incur as a result of an accident which has not yet occurred is not so easy. It is of our interest to find a systematic approach to quantify future risks including the societal losses in monetary terms.So we propose the path of unifying effective risk identification, quantification and mitigation for a process using QRA for any process such as the Chlorine process, along with the financial and economic consequences resulting from a suitable VaR model. The risks to be used as key factors by VaR model will arrive from the QRA methodology. VaR is a thorough risk management measure that represents the worst possible loss of a portfolio over a specified time horizon in monetary terms. Utilisation of VaR is fitting not only because it bridges the gap between “engineers and scientists who calculate process risk and, the business leaders and policy makers who evaluate, manage, or regulate risk in a broader context,” but also because VaR gives the value of total cost-benefit analysis (TBA) of entire portfolio via single probability distribution function (p.d.f) for value (Fang, Mannan and Ford, 2003). Another advantage of using VaR is its generality and the possibility of comparing portfolio values between different time horizons once a VaR measurement is acquired. The portfolios are being studied by using standard statistical probability distribution functions to acquire a VaR measurement.
We believe that only estimating the likelihood of occurrence of any event is insufficient and that any risk involved should also be effectively communicated to the industry personnel, the policy makers and the public in order to properly mitigate the risks involved in a process. Therefore, the results of QRA and VaR of a portfolio will be utilised to communicate societal risk using effective risk communication means such as FN curves, FAR, etc.
Quantitative risk assessment (QRA) in chemical industry aims to quantify risk as a function of occurrence probabilities and consequences of major accident scenarios. However, this process itself includes uncertainties. System component failure rates used in form of point values from laboratory data or generic data have generally been accepted in the past by industry to estimate the system failure occurrence probability. This practice may mislead the QRA evaluation and subsequent decision-making. Because the results of a QRA provide information to prevent losses in major accident hazards and aid in many decisions on risk management, it is important to increase accuracy of the results. Therefore, analysis on uncertainties associated with a QRA is crucial to evaluate the QRA, how close the evaluation is from reality, and how the risk is reliably identified to make good decisions that affect chemical process safety design.
Using Monte Carlo simulation, the distribution of system failure probability is presented to characterize the uncertainty of the result. Bayesian theory is used to update failure rates of the equipment, combining generic reliability information in available databases and plant specific equipment testing data to enhance information about component reliability. In a FTA for the level control of distillation unit, it was found that the uncertainty of basic event probabilities has a significant impact on the top event probability distribution; and the top event probability prediction uncertainty profile shows that the risk estimation was improved by reducing uncertainty though Bayesian updating on the basic event probabilities.
Entire distributions of top event probabilities replaces point values in a risk matrix to guide decisions employing all of the available information rather than only using input mean values as in the conventional approach. The resulting uncertainty guides where more information or reduction of uncertainties is needed the most to avoid overlap with intolerable risk levels.
Quantitative risk assessment has been accepted as a mature methodology for analyzing accident scenarios and quantifying the risk in nuclear power plants, chemical process plants, structure industries, and space systems. The chemical process system is a dynamic and complicated system. Various time-dependent effects such as season changes, aging of process equipment, physical processes, stochastic processes, and operator response time are involved in a chemical process system. With the accumulated experience of quantitative risk assessment and the progressive awareness of dynamical characteristics of reliability and safety, the conventional approaches show limitations for addressing dynamic issues in quantitative risk assessment for chemical industry, for example, the system dynamic response on time, process variables, and human behavior. Therefore, a dynamic tool is needed to assess dynamic risk in the chemical process industries.
In this study, the methodology is designed for continuous operational risk assessment. Process variable evolution follows physical/engineering laws, and this evolution is also governed by the performance of the components within the system under assessment. Discrete event simulation is applied to study the stochastic process behavior of a specific component. Then the process variable evolution directed along discrete event paths is simulated to obtain the real time probability of process variable to exceed safety boundaries. This approach will be used to study the level control of a high pressure Oil/Gas separator in an offshore plant.
In addition to generating a better understanding of real-time risk associated with a dynamic system in chemical process industries, this research will provide a platform for continuous operational risk assessment using a standard computational space storage and time consumption. This approach also aids in monitoring the most critical components within a system even though their importance or criticality varies with time. This platform is implemented as an ongoing model to guide implementation and continual updating of safety program components such as risk-based and cost-effective monitoring, testing, maintenance, reliability assessment, component replacement timing, shutdown times, and timing of other operational decisions including selection of minimal reliability criteria during maintenance shutdowns.