Ioannis A. Papazoglou

Probabilistic safety analysis in chemical installations

Abstract A set of procedures and corresponding methodologies for probabilistic safety assessment (PSA) in chemical installations is presented. State-of-the-art methodology for PSA in nuclear power plants is tested for suitability and applicability to chemical installations and applied on a refrigerated ammonia storage facility. The procedural steps comprise hazard identification, accident sequence modelling, data acquisition and parameter estimation, accident sequence quantification, hazardous substance release categories assessment, consequence assessment and integration of results. The results of the application indicate that the methodology applied is adequate from the procedural and completeness point of view and that no major aspect of PSA of a chemical installation is overlooked. A need for integrated computerized tools has, however, been identified.

Mathematical foundations of event trees

Abstract A mathematical foundation from first principles of event trees is presented. The main objective of this formulation is to offer a formal basis for developing automated computer assisted construction techniques for event trees. The mathematical theory of event trees is based on the correspondence between the paths of the tree and the elements of the outcome space of a joint event. The concept of a basic cylinder set is introduced to describe joint event outcomes conditional on specific outcomes of basic events or unconditional on the outcome of basic events. The concept of outcome space partition is used to describe the minimum amount of information intended to be preserved by the event tree representation. These concepts form the basis for an algorithm for systematic search for and generation of the most compact (reduced) form of an event tree consistent with the minimum amount of information the tree should preserve. This mathematical foundation allows for the development of techniques for automated generation of event trees corresponding to joint events which are formally described through other types of graphical models. Such a technique has been developed for complex systems described by functional blocks and it is reported elsewhere. On the quantification issue of event trees, a formal definition of a probability space corresponding to the event tree outcomes is provided. Finally, a short discussion is offered on the relationship of the presented mathematical theory with the more general use of event trees in reliability analysis of dynamic systems.

Modeling emergency evacuation for major hazard industrial sites

A model providing the temporal and spatial distribution of the population under evacuation around a major hazard facility is developed. A discrete state stochastic Markov process simulates the movement of the evacuees. The area around the hazardous facility is divided into nodes connected among themselves with links representing the road system of the area. Transition from node-to-node is simulated as a random process where the probability of transition depends on the dynamically changed states of the destination and origin nodes and on the link between them. Solution of the Markov process provides the expected distribution of the evacuees in the nodes of the area as a function of time. A Monte Carlo solution of the model provides in addition a sample of actual trajectories of the evacuees. This information coupled with an accident analysis which provides the spatial and temporal distribution of the extreme phenomenon following an accident, determines a sample of the actual doses received by the evacuees. Both the average dose and the actual distribution of doses are then used as measures in evaluating alternative emergency response strategies. It is shown that in some cases the estimation of the health consequences by the average dose might be either too conservative or too non-conservative relative to the one corresponding to the distribution of the received dose and hence not a suitable measure to evaluate alternative evacuation strategies.

Comprehensive risk assessment for rail transportation of dangerous goods: a validated platform for decision support

Abstract Currently, the most advanced and well documented risk assessments for the transportation of dangerous goods by railway take into account: (i) statistics-based loss of containment frequencies, (ii) specification of potential consequences for a given release situations using event tree methodology as an organisational tool and (iii) consequence calculation models to determine a risk figure known as CCDF (Complementary Cumulative Distribution Function). Such procedures for the risk assessment (including for example decision-making on preventive measures) may offer only a limited insight into the causes and sequences leading to an accident and do not allow for any kind of predictive analysis. The present work introduces an enhanced solution, and a related software platform, which attempts to integrate loss of containment causes and consequences with systems infrastructure and its environment. The solution features: (i) the use of a detailed Master Logical Diagram, including fault/event tree analysis to determine a loss of containment frequency based on different initiating events, scenarios and specific basic data, (ii) the characterization of a resulting source term following a release situation, and (iii) the calculation of various potential impacts on the neighbouring site. Results are wrapped into a CCDF format for each selected traffic segment. The risk-related results are integrated on a software platform, structured as a decision support system using intelligent maps and a variety of GIS (Geographical Information System) data processing procedures. The introduction of the hot spot approach, allows us to focus on the most risk-relevant areas and to use information on various railway infrastructure elements (e.g. points, tunnels), are the basis of the new models employed. The software is applicable to any railway transportation system, comprising its technical infrastructure, rolling stock, human actions, regulation and management procedures. It provides the determination of the annual societal risk due to potential accident scenarios, while also revealing information on the potential causes of an accident taking into account spatial parameters. The approach and software have been validated by a case study done for a particular traffic segment of the Swiss Federal Railway company.

Functional block diagrams and automated construction of event trees

Abstract Functional block diagrams are introduced as graphical representations of the function of a system. A functional block provides the output of a system as the outcome of a joint event defined by the inputs to the system and its various states. Functional blocks corresponding to different subsystems are combined together to form a functional block diagram representing the functional characteristics of the combined system. Conversely, a complex system represented by a single functional block is decomposed to constituent components with a corresponding functional block diagram. It is demonstrated that a functional block diagram is an alternative representation of an event-tree corresponding to the outcome space of the underlying joint event. The concept of output space partition is introduced and an algorithm is developed for generating the most compact form of the event-tree consistent with a given partition. This algorithm forms the basis of a technique for the automated construction of an event-tree starting from a developed functional block diagram. This technique is amenable to computerization. Finally, as a demonstration of the technique, the event-trees of a boiling water reactor (BWR) are developed.

On the management of severe chemical accidents DECARA: A computer code for consequence analysis in chemical installations. Case study: Ammonia plant

Abstract A computer package is presented for the integrated risk assessment of accidental releases of hazardous substances. DECARA provides an integrated risk analysis including source-term strength evaluation, estimation of the hazardous cloud dispersion and quantification of health impacts. Multiple accidents, each with a certain probability of occurrence can be handled. Dispersion of heavier as well as lighter-than-air gases released instantaneously or continuously, can be simulated. Time-varying release rates are possible. Uncertainty analysis can be performed including both parameter and modelling uncertainty. The code calculates the unconditional fatality probability at any point around the site of release. Isorisk curves or maximum individual risk versus distance from the source can be generated. The computer package is portable and available for running in personal computers.

A dynamic screening algorithm for multiple objective simulated annealing optimization

Abstract This work proposes new multiple objective optimization (MOO) technology, using a Monte Carlo-based algorithm stemmed from simulated annealing (SA). Since the expected result in MOO tasks is usually a set of Pareto-optimal solutions, the optimization problem states assumed here are themselves sets of solutions. The stochastic search follows a series of reversible state transitions at constant probability, to enjoy convergence properties of stationary Markov processes. The proposed technology is tested against the optimal design of a process system involving equipment placed in a serial/parallel arrangement, with three optimization objectives: the system cost, reliability and weight.

An auditing methodology for safety management of the Greek process industry

Abstract Process Risk Management Audit (PRIMA), a methodology aiming at assessing the impact on safety of the management system in a plant is outlined. The methodology has been applied in two plants in Greece within the framework of a cross-national application and results concerning both the methodology and the Greek plants are reported and discussed.

Risk-informed assessment of the technical specifications of PWR RPS instrumentation

A special Markov reliability model is used to simulate the effects of frequency of testing and allowable bypass time of the analog channels and logic trains on the performance of a pressurized water reactor, reactor protection system (RPS). The effects of these technical specifications (TSs) on the unavailability of the RPS and on the frequencies of spurious scrams, core damage, anticipated transients without scram, and large early releases are assessed. The use of these results to support decision making concerning changing the TSs in light of the relevant policy statement by the U.S. Nuclear Regulatory Commission on risk-informed decision making is demonstrated. The analysis includes point-value calculations sensitivity and uncertainty analysis.

Dynamic failure assessment of incidents reported in the Greek Petrochemical Industry

Abstract This paper presents a Bayesian statistical analysis on real incident data collected from the Greek Petrochemical Industry for a period of 6 years (1997–2003). The analysis provides an assessment of the database to support predictions of dynamically updated incident occurrence frequencies. Results are reported for two different categories of incidents, namely the industrial and the occupational ones.