Operational risk assessment of chemical industries by exploiting accident databases

Accident databases (NRC, RMP, and others) contain records of incidents (e.g., releases and spills) that have occurred in the USA chemical plants during recent years. For various chemical industries, [Kleindorfer, P. R., Belke, J. C., Elliott, M. R., Lee, K., Lowe, R. A., & Feldman, H. I. (2003). Accident epidemiology and the US chemical industry: Accident history and worst-case data from RMP*Info. Risk Analysis, 23(5), 865–881.] summarize the accident frequencies and severities in the RMP*Info database. Also, [Anand, S., Keren, N., Tretter, M. J., Wang, Y., O’Connor, T. M., & Mannan, M. S. (2006). Harnessing data mining to explore incident databases. Journal of Hazardous Material, 130, 33–41.] use data mining to analyze the NRC database for Harris County, Texas.Classical statistical approaches are ineffective for low frequency, high consequence events because of their rarity. Given this information limitation, this paper uses Bayesian theory to forecast incident frequencies, their relevant causes, equipment involved, and their consequences, in specific chemical plants. Systematic analyses of the databases also help to avoid future accidents, thereby reducing the risk.More specifically, this paper presents dynamic analyses of incidents in the NRC database. The NRC database is exploited to model the rate of occurrence of incidents in various chemical and petrochemical companies using Bayesian theory. Probability density distributions are formulated for their causes (e.g., equipment failures, operator errors, etc.), and associated equipment items utilized within a particular industry. Bayesian techniques provide posterior estimates of the cause and equipment-failure probabilities. Cross-validation techniques are used for checking the modeling, validation, and prediction accuracies. Differences in the plant- and chemical-specific predictions with the overall predictions are demonstrated. Furthermore, extreme value theory is used for consequence modeling of rare events by formulating distributions for events over a threshold value. Finally, the fast-Fourier transform is used to estimate the capital at risk within an industry utilizing the frequency and loss-severity distributions.     

A novel tool for Bayesian reliability analysis using AHP as a framework for prior elicitation

Quantitative risk assessment (QRA) is a powerful and popular technique to support risk-based decisions. Unfortunately, QRAs are often hampered by significant uncertainty in the frequency of failure estimation for physical assets. This uncertainty is largely due to lack of quality failure data in published sources. The failure data may be limited, incompatible and/or outdated. Consequently, there is a need for robust methods and tools that can incorporate all available information to facilitate reliability analysis of critical assets such as pipelines, pressure vessels, rotating equipment, etc. This paper presents a novel practical approach that can be used to help overcome data scarcity issues in reliability analysis. A Bayesian framework is implemented to cohesively integrate objective data with expert opinion with the aim toward deriving time to failure distributions for physical assets. The Analytic Hierarchy Process is utilized to aggregate time to failure estimates from multiple experts to minimize biases and address inconsistencies in their estimates. These estimates are summarized in the form of informative priors that are implemented in a Bayesian update procedure for the Weibull distribution. The flexibility of the proposed methodology allows for efficiently dealing with data limitations. Application of the proposed approach is illustrated using a case study.     

On risk-based maintenance: A comprehensive review of three approaches to track the impact of consequence modelling for predicting maintenance actions

Since gas plants are progressively increasing near urban areas, a comprehensive tool to plan maintenance and reduce the risk arising from their operations is required. To this end, a comparison of three Risk-Based Maintenance methodologies able to point out maintenance priorities for the most critical components, is presented in this paper. Moreover, while the literature is mostly focused on probabilistic analysis, a particular attention is directed towards consequence analysis throughout this study. The first developed technique is characterized by a Hierarchical Bayesian Network to perform the occurrence analysis and a Failure Modes, Effects and Criticality Analysis to assess the magnitude of the adverse outcomes. The second approach is a Quantitative Risk Analysis carried out via a software named Safeti. Finally, another software called Synergi Plant is adopted for the third methodology, which provides a Risk-Based Inspection plan, through a semiquantitative risk analysis. The proposed study can assist asset manager in adopting the most appropriate methodology to their context, while highlighting priority components. To demonstrate the applicability of the approaches and compare their rankings, a Natural Gas Regulating and Measuring Station is considered as case study. The results showed that the most suited method strongly depends on the available data.     

Uncertainty reduction for improved mishap probability prediction: Application to level control of distillation unit

At all levels, the understanding of uncertainty associated with risk of major chemical industrial hazards should be enhanced. In this study, a quantitative risk assessment (QRA) was performed for a knockout drum in the distillation unit of a refinery process and then probabilistic uncertainty analysis was applied for this QRA. A fault tree was developed to analyze the probability distribution of flammable liquid released from the overfilling of a knockout drum. Bayesian theory was used to update failure rates of the equipment so that generic information from databases and plant equipment real life data are combined to gain all available knowledge on component reliability. Using Monte Carlo simulation, the distribution of top event probability was obtained to characterize the uncertainty of the result. It was found that the uncertainty of basic event probabilities has a significant impact on the top event probability distribution. The top event probability prediction uncertainty profile showed that the risk estimation is improved by reducing uncertainty through Bayesian updating on the basic event probability distributions. The whole distribution of top event probability replaces point value in a risk matrix to guide decisions employing all of the available information rather than only point mean values as in the conventional approach. The resulting uncertainty guides where more information or uncertainty reduction is needed to avoid overlap with intolerable risk levels.     

Uncertainty analysis in the nuclear industry: Analytical unavailability modelling incorporating ageing of safety components

There is a growing interest in analyzing the possibility for current nuclear power plants operation extension. In that sense, life management programs, considering safety components ageing, are being developed and employed. On the other side, the large uncertainties of the ageing parameters as well as the uncertainties associated with most of the reliability data collections are widely acknowledged.This paper deals with uncertainty analysis associated with specific ageing rates database. The analysis is conducted using an analytical unavailability model applied for a selected safety system in a nuclear power plant. The most important problem is the immense uncertainty associated to the component ageing data sets as well as the lack of the very data in general, which would correspond to the more detailed modelling of ageing.New probability distributions, encompassing the ageing rates available in the considered data set, are suggested. The obtained results indicate the extent to which the uncertainty of the considered ageing data set, given the inherently assigned probability distribution, influences the performed unavailability calculations. Additionally, comparative analysis regarding the insights gained out of the application of the suggested probability distributions is conducted.     

Fault detection and diagnostics of a three-phase separator

A high demand of oil products on daily basis requires oil processing plants to work with maximum efficiency. Oil, water and gas separation in a three-phase separator is one of the first operations that are performed after crude oil is extracted from an oil well. Failure of the components of the separator introduces the potential hazard of flammable materials being released into the environment. This can escalate to a fire or explosion. Such failures can also cause downtime for the oil processing plant since the separation process is essential to oil production. Fault detection and diagnostics techniques used in the oil and gas industry are typically threshold based alarm techniques. Observing the sensor readings solely allows only a late detection of faults on the separator which is a big deficiency of such a technique, since it causes the oil and gas processing plants to shut down.A fault detection and diagnostics methodology for three-phase separators based on Bayesian Belief Networks (BBN) is presented in this paper. The BBN models the propagation of oil, water and gas through the different sections of the separator and the interactions between component failure modes and process variables, such as level or flow monitored by sensors installed on the separator. The paper will report on the results of the study, when the BBNs are used to detect single and multiple failures, using sensor readings from a simulation model. The results indicated that the fault detection and diagnostics model was able to detect inconsistencies in sensor readings and link them to corresponding failure modes when single or multiple failures were present in the separator.     

Optimization of test interval for ageing equipment: A multi-objective genetic algorithm approach

Reducing the unavailability of safety systems at nuclear power plants, by utilizing the probabilistic safety assessment (PSA) methodology, is one of the prime goals in the nuclear industry. In that sense, optimization of test and maintenance activities, which are defined within the technical specifications, represents quite popular and interesting domain. Obtaining optimal test and maintenance schedule is of great significance for improving system availability and performance as well as plant availability in general.On the other side, equipment aging has gradually become a major concern in the nuclear industry since the number of safety systems components, that are approaching their wear-out stage, is rising fast. Nuclear power plants life management programs, considering safety components aging, are being developed and employed. The immense uncertainty associated to the available component aging rates databases poses significant difficulties in the process of incorporation and quantification of the aging effect within the PSA and, subsequently, in the decision-making process.In this paper, an approach for optimization of surveillance test interval of standby equipment with highly uncertain aging parameters, based on genetic algorithm technique and PSA, is presented. A standard standby safety system in nuclear power plant is selected as a case study. A Monte Carlo simulation-based approach is used to assess uncertainty propagation on system level. Optimal test interval is derived on the basis of minimal system unavailability and minimal impact of components aging parameters uncertainty. The results obtained in this application indicate the fact that risk-informed surveillance requirements differ from existing ones in technical specifications as well as show the importance of considering aging data uncertainties in component aging modeling.     

Application of Bayesian informative priors to enhance the transferability of safety performance functions

IntroductionSafety performance functions (SPFs) are essential tools for highway agencies to predict crashes, identify hotspots and assess safety countermeasures. In the Highway Safety Manual (HSM), a variety of SPFs are provided for different types of roadway facilities, crash types and severity levels. Agencies, lacking the necessary resources to develop own localized SPFs, may opt to apply the HSM’s SPFs for their jurisdictions. Yet, municipalities that want to develop and maintain their regional SPFs might encounter the issue of the small sample bias. Bayesian inference is being conducted to address this issue by combining the current data with prior information to achieve reliable results. It follows that the essence of Bayesian statistics is the application of informative priors, obtained from other SPFs or experts’ experiences.MethodIn this study, we investigate the applicability of informative priors for Bayesian negative binomial SPFs for rural divided multilane highway segments in Florida and California. An SPF with non-informative priors is developed for each state and its parameters’ distributions are assigned to the other state’s SPF as informative priors. The performances of SPFs are evaluated by applying each state’s SPFs to the other state. The analysis is conducted for both total (KABCO) and severe (KAB) crashes.Results, conclusions and practical applicationsAs per the results, applying one state’s SPF with informative priors, which are the other state’s SPF independent variable estimates, to the latter state’s conditions yields better goodness of fit (GOF) values than applying the former state’s SPF with non-informative priors to the conditions of the latter state. This is for both total and severe crash SPFs. Hence, for localities where it is not preferred to develop own localized SPFs and adopt SPFs from elsewhere to cut down on resources, application of informative priors is shown to facilitate the process.     

基于贝叶斯网络的高速动车组运营故障分析

为研究影响高速动车组正常运行的各故障因素间的因果关系,分析其耦合强度,将故障因素分为人、机、环3类,从系统角度又将机器因素分为5个子系统。人、机、环3类共识别27个故障因素;使用K2算法生成贝叶斯网络结构,引入扩展因果效应算法确定节点优先次序作为K2算法的先验知识,采用EM算法学习贝叶斯网络参数,构建基于贝叶斯网络的高速动车组运营故障分析模型;以209个CRH详细故障报告为例,对故障因素的故障发生概率进行排序并分析因素间的影响强度和灵敏度。结果表明:牵引供电系统故障发生概率较高;车门系统故障、牵引变流器故障易由内部零件故障引起,外界异物击打对受电弓影响较大;人、环因素更易引起多故障耦合;环境因素对牵引供电系统表现出较高的灵敏度。贝叶斯网络在分析高速动车组运营系统故障问题上具有可行性,分析结果有助于提升运营单位的管控能力。     

Time-related distribution of occupational accidents

Time-related distributions of the frequency of occupational traumatic accidents (3,309) were analyzed in two plants. Data on the distribution of work-related accidents (676,389) throughout Ontario between 1980–1984 were also compiled and compared with the data gathered in the two plants. Time-related distributions of accident frequency were similar for both plants and for the province of Ontario, with accidents peaking around noon. There was a direct relationship between the number of accidents and the level of the activity of the plants. At night, however, between 2 a.m. and 4 a.m., despite a great decrease of activity level, there was a slight increase in accident frequency.     

An integrated method for safety pre-warning of complex system

In large-scale and complex industrial systems, unplanned outages and hazardous accidents cause huge economic losses, environmental contamination, and human injuries, due to component degradation, exogenous changes, and operational mistakes. In order to ensure safety and increase operational performance and reliability of complex system, this study proposes an integrated method for safety pre-warning to analyze the current safety state of each component and the whole system indicating hidden hazards and potential consequence, and furthermore predict future degradation trends in the long term.The work presented here describes the rationale and implementation of the integrated method incorporating HAZOP study, degradation process modeling, dynamic Bayesian network construction, condition monitoring, safety assessment and prognosis steps, taking advantage of the priori knowledge of the interactions and dependencies among components and the environment, the relationships between hazard causes and effects, and the use of historical failure data and online real-time data from condition monitoring.The application of the integrated safety pre-warning approach described here to the specific example of the gas turbine compressor system demonstrates how each phase of the presented method contributes to completion of the safety pre-warning system development in a systematic way.     

基于Bayesian网络的常减压装置设备故障概率分析

为分析影响常减压蒸馏装置平稳运行的设备失效模式及故障部件,基于1 151条设备故障数据,采用Bayesian网络分析方法,分别对离心泵、压缩机、电动机构建基于Bayesian网络的设备故障概率分析模型,分析故障部件、失效模式、故障后果之间的定量概率关系。研究结果表明:离心泵、压缩机、电动机停运的关键致因部件分别为轴承箱密封故障、活塞环故障、轴承故障,同时得到导致设备停运的故障部件敏感度排序。研究结果有助于提高设备故障风险防范及检维修工作效率,同时可为备件优化方案提供思路。     

An uncertainty and sensitivity analysis of dynamic operational risk assessment model: A case study

In Dynamic Operational Risk Assessment (DORA) models, component repair time is an important parameter to characterize component state and the subsequent system-state trajectory. Specific distributions are fit to the industrial component repair time to be used as the input of Monte Carlo simulation of system-state trajectory. The objective of this study is to propose and apply statistical techniques to characterize the uncertainty and sensitivity on the distribution model selection and the associated parameters determination, in order to study how the DORA output that is the probability of operation out-of-control, can be apportioned by the distribution model selection. In this study, eight distribution fittings for each component are performed. Chi-square test, Kolmogorov–Smirnov test, and Anderson-Darling test are proposed to measure the goodness-of-fit to rank the distribution models for characterizing the component repair time distribution. Sensitivity analysis results show that the selection of distribution model among exponential distribution, gamma distribution, lognormal distribution and Weibull distribution to fit the industrial data has no significant impact on DORA results in the case study.     

基于贝叶斯网络的人因可靠性评价

提出一种贝叶斯网络的人因可靠性评价(HRABN)方法,其中的每个因子对应于贝叶斯网络中的节点,该方法可对人因可靠性作定量分析和定性分析。在定性分析上,节点的因果关系(HRA中的因子关系)及需要改进的薄弱节点都直观地显示在层次图中;在定量分析方面,对节点因子后验概率的推断通过HRA中的先验信息(包含仿真数据、现场操作及专家知识等)和最新信息得到。如果人因可靠性贝叶斯网络中的每个节点的先验概率分布和后验概率分布都已知,模型的可信性就可通过贝叶斯因子进行定量验证。贝叶斯网络扩展性好,当有新的节点因子需要考虑时,只需要补充对应的节点;笔者的方法也能很好地应用在不同行业的HRA。     

Dynamic failure assessment of an ammonia storage unit: A case study

Chemical Process Industries usually contain a diverse inventory of hazardous chemicals and complex systems required to perform process operations such as storage, separation, reaction, compression etc. The complex interactions between the equipment make them vulnerable to catastrophic accidents. Risk and failure assessment provide engineers with an intuitive tool for decision making in the operation of such plants. Abnormal events and near-miss situations occur regularly during the operation of a system. Accident Sequence Precursors (ASP) can be used to demonstrate the real-time operating condition of a plant. Dynamic Failure Assessment (DFA) methodology is based on Bayesian statistical methods incorporates ASP data to revise the generic failure probabilities of the systems during its operational lifetime.In this paper, DFA methodology is applied on an ammonia storage unit in a specialized chemical industry. Ammonia is stored in cold storage tanks as liquefied gas at atmospheric pressure. These tanks are susceptible to failures due to various abnormal conditions arising due process failures.Tank failures due to three such abnormal conditions are considered. Variation of the failure probability of the safety systems is demonstrated. The authors use ASP data collected from plant specific sources and safety expert judgement. The failure probabilities of some safety systems concerned show considerable deviation from the generic values. The method helps to locate the components which have undergone more degradation over the period and hence must be paid attention to. In addition, a Bayesian predictive model has been used to predict the number of abnormal events in the next time interval. The user-friendly and intuitive nature of the tool makes it appropriate for application in safety assessment reports in process industries.     

石化装置火灾爆炸贝叶斯网络与防护层集成分析

复杂的石油化工装置在运转过程中存在诸多不确定因素,易发生火灾、爆炸等重大事故,给安全生产带来极大威胁。考虑到传统的系统安全分析方法在风险评估中存在一定局限性,引入贝叶斯网络与防护层集成分析模型。应用GeNIe软件将系统故障树转成贝叶斯网络,根据贝叶斯双向推理进行故障预测和诊断,快速识别系统薄弱环节并确定为风险贝叶斯故障节点,结合防护层分析提出相应的独立防护层,确定剩余风险水平。实例应用表明,所构建的贝叶斯网络与防护层集成分析模型对复杂系统进行风险评估是可行的,较传统的事件树、故障树分析方法更加科学、合理。     

Dynamic Bayesian networks for reliability evaluation of subsea wellhead connector during service life based on Monte Carlo method

The subsea wellhead connector is a critical connection component between subsea Christmas tree and subsea wellhead for preventing the leakage of oil and gas in the subsea production system. Excited by cyclical loadings due to environmental forces and the other support forces, the subsea wellhead connector is prone to the failure, which could lead to the loss of subsea tree or wellhead integrity and even catastrophic accidents. With the Monte Carlo simulation method, this paper presents a reliability analysis approach based on dynamic Bayesian Networks, aiming to assess the failure probability of the subsea wellhead connector during service life. Take the driving ring component of the subsea wellhead connector as an example to demonstrate the reasonability of the proposed model. The generation data is processed by the transform between the numerical value and the state variable. Based on the stress-strength interference theory, the structure reliability of the driving ring with 96.26% is achieved by the proposed model with the consideration the aging of the material strength and the most influential factors are figured out. Meanwhile, the corresponding control measures are proposed effectively reduce the failure risk of the subsea wellhead connector during service life.     

基于贝叶斯网络的地震次生燃气管道泄漏事件链构建

地震次生燃气管道泄漏事件是导致地震灾害影响扩大的主要原因之一,为了解建 筑物内地震次生燃气管道泄漏事件的发展过程,参考国内外相关的文献和统计资料,确 定建筑物内地震次生燃气管道泄漏事件贝叶斯网络的节点变量和取值范围。根据节点变 量及其逻辑关联性构建贝叶斯网络结构图,通过对国内外研究数据的统计并结合专家经 验估算确定各节点变量的条件概率。利用贝叶斯网络工具计算建筑物属性和环境变量在 不同状态取值下建筑物遭受破坏、燃气泄漏、燃气扩散引发二次灾害等关键节点变量的 后验概率。通过实例分析得出,建筑结构和地震烈度是建筑物遭受破坏的主要影响因素 ,风速、大气稳定度对燃气扩散引发二次灾害有显著影响。     

Bayesian networks make LOPA more effective,QRA more transparent and flexible,and thus safety more definable!

Quantitative risk analysis is in principle an ideal method to map one’s risks, but it has limitations due to the complexity of models, scarcity of data, remaining uncertainties, and above all because effort, cost, and time requirements are heavy. Also, software is not cheap, the calculations are not quite transparent, and the flexibility to look at various scenarios and at preventive and protective options is limited. So, the method is considered as a last resort for determination of risks. Simpler methods such as LOPA that focus on a particular scenario and assessment of protection for a defined initiating event are more popular. LOPA may however not cover the whole range of credible scenarios, and calamitous surprises may emerge.In the past few decades, Artificial Intelligence university groups, such as the Decision Systems Laboratory of the University of Pittsburgh, have developed Bayesian approaches to support decision making in situations where one has to weigh gains and costs versus risks. This paper will describe details of such an approach and will provide some examples of both discrete random variables, such as the probability values in a LOPA, and continuous distributions, which can better reflect the uncertainty in data.     

Dynamic risk assessment for underground gas storage facilities based on Bayesian network

Loss of the underground gas storage process can have significant effects, and risk analysis is critical for maintaining the integrity of the underground gas storage process and reducing potential accidents. This paper focuses on the dynamic risk assessment method for the underground gas storage process. First, the underground gas storage process data is combined to create a database, and the fault tree of the underground gas storage facility is built by identifying the risk factors of the underground gas storage facility and mapping them into a Bayesian network. To eliminate the subjectivity in the process of determining the failure probability level of basic events, fuzzy numbers are introduced to determine the prior probability of the Bayesian network. Then, causal and diagnostic reasoning is performed on the Bayesian network to determine the failure level of the underground gas storage facilities. Based on the rate of change of prior and posterior probabilities, sensitivity and impact analysis are combined to determine the significant risk factors and possible failure paths. In addition, the time factor is introduced to build a dynamic Bayesian network to perform dynamic assessment and analysis of underground gas storage facilities. Finally, the dynamic risk assessment method is applied to underground gas storage facilities in depleted oil and gas reservoirs. A dynamic risk evaluation model for underground gas storage facilities is built to simulate and validate the dynamic risk evaluation method based on the Bayesian network. The results show that the proposed method has practical value for improving underground gas storage process safety.