Ulrich Hauptmanns

A procedure for analyzing the flight of missiles from explosions of cylindrical vessels

Abstract The fragmentation of storage tanks or other equipment of the process industry, for example caused by boiling liquid expanding explosions (BLEVEs), and the consequent missile generation is a problem in industrial safety whose importance is underlined by accidents such as that of Mexico City. It is inscribed in what is called the “Domino Effect”. A method is presented to calculate the trajectories of missiles from horizontally mounted cylindrical vessels using analytical solutions of the equations of motion. The principal input parameters of these equations, e.g. drag and initial fragment energy, are stochastic or uncertain. Hence, they are represented by statistical distributions, whose parameters are determined as far as possible from experimental results or findings from past accidents. If no such evidence is available, reasonable assumptions, e.g. constant probability density function, are made. The Monte Carlo method is used to propagate the effect of the stochastic and uncertain input parameters through the calculation. Numerical results thus obtained are compared with evidence from accidents.

The human error rate assessment and optimizing system HEROS — a new procedure for evaluating and optimizing the man–machine interface in PSA

Abstract A new procedure allowing the probabilistic evaluation and optimization of the man–machine system is presented. This procedure and the resulting expert system HEROS, which is an acronym for Human Error Rate Assessment and Optimizing System, is based on the fuzzy set theory. Most of the well-known procedures employed for the probabilistic evaluation of human factors involve the use of vague linguistic statements on performance shaping factors to select and to modify basic human error probabilities from the associated databases. This implies a large portion of subjectivity. Vague statements are expressed here in terms of fuzzy numbers or intervals which allow mathematical operations to be performed on them. A model of the man–machine system is the basis of the procedure. A fuzzy rule-based expert system was derived from ergonomic and psychological studies. Hence, it does not rely on a database, whose transferability to situations different from its origin is questionable. In this way, subjective elements are eliminated to a large extent. HEROS facilitates the importance analysis for the evaluation of human factors, which is necessary for optimizing the man–machine system. HEROS is applied to the analysis of a simple diagnosis of task of the operating personnel in a nuclear power plant.

The multi-class binomial failure rate model

The impact of common cause failure (CCF) on PSA results for NPPs is in sharp contrast with the limited quality which can be achieved in their assessment. This is due to the dearth of observations and cannot be remedied in the short run. Therefore, the methods employed for calculating failure rates should be devised such as to make the best use of the few available observations on CCF. The Multi-Class Binomial Failure Rate (MCBFR) Model presented here achieves this by assigning observed failures to different classes according to their technical characteristics and applying the BFR formalism to each of these. The results are determined by a superposition of BFR type expressions for each class, each of them with its own coupling factor. The model hence integrates additional technical information on CCFs and leads to a better quality of results whenever this information is available. This is demonstrated by evaluating CCFs observed for combined impulse pilot valves in German NPPs.

Uncertainty and the calculation of safety-related parameters for chemical reactions

Data uncertainties are present in any engineering calculation. Their impact may be particularly important if they concern safety-related parameters of process plants. Procedures are presented to deal with such uncertainties. Their propagation through calculations by means of a Monte-Carlo approach is shown. The adiabatic temperature rise and the time for achieving a specified fraction of conversion are evaluated for an example, the esterification of acetic anhydride, in order to show the influence of uncertainties and to demonstrate that the conventional procedure of using point values may produce a misleading basis for design decisions.

Analytical propagation of uncertainties through fault trees

Abstract A method is presented which enables one to propagate uncertainties described by uniform probability density functions through fault trees. The approach is analytical. It is based on calculating the expected value and the variance of the top event probability. These two parameters are then equated with the corresponding ones of a beta-distribution. An example calculation comparing the analytically calculated beta-pdf (probability density function) with the top event pdf obtained using the Monte-Carlo method shows excellent agreement at a much lower expense of computing time.

Integration of Stochastic Effects and Data Uncertainties into the Design of Process Equipment

Stochastic effects and data uncertainties are present in any engineering calculation. Their impact may be particularly important if they concern the design of process equipment. A calculation model for the dynamic behavior of a heat exchanger and procedures to deal with the related uncertainties are presented. Their propagation through the calculation by means of a Monte Carlo approach is shown. The temperature at the heat exchanger outlet and the step response of a sudden variation in the heat exchanger inlet temperature are simulated and evaluated by way of example. It is demonstrated that the inclusion of stochastic effects and uncertainties provides a more reliable basis for design decisions and hence reduces the probability of errors.

A decision-making framework for protecting process plants from flooding based on fault tree analysis

The protection of process plants from external events is mandatory in the Seveso Directive. Among these events figures the possibility of inundation of a plant, which may cause a hazard by disabling technical components and obviating operator interventions. A methodological framework for dealing with hazards from potential flooding events is presented. It combines an extension of the fault tree method with generic properties of flooding events in rivers and of dikes, which should be adapted to site-specific characteristics in a concrete case. Thus, a rational basis for deciding whether upgrading is required or not and which of the components should be upgraded is provided. Both the deterministic and the probabilistic approaches are compared. Preference is given to the probabilistic one. The conclusions drawn naturally depend on the scope and detail of the model calculations and the decision criterion adopted. The latter has to be supplied from outside the analysis, e.g. by the analyst himself, the plant operator or the competent authority. It turns out that decision-making is only viable if the boundary conditions for both the procedure of analysis and the decision criterion are clear.

Different sets of reliability data and success criteria in a probabilistic safety assessment for a plant producing nitroglycol.

The lack of plant-specific reliability data for probabilistic safety assessments usually makes it necessary to use generic reliability data. Justifiably different assessments of plant behaviour (success criteria) lead to different models of plant systems. Both affect the numerical results of a probabilistic safety assessment. It is shown how these results change, if different sets of reliability data and different choices of success criteria for the safety system are employed. Differences in results may influence decisions taken on their basis and become especially important if compliance with a safety goal has to be proved, e.g. a safety integrity level. For the purpose of demonstration an accident sequence from a probabilistic safety assessment of a plant producing nitroglycol is used. The analysis relies on plant-specific reliability data so that it provides a good yardstick for comparing it with results obtained using generic data. The superiority of plant-specific data, which should of course be acquired, cannot be doubted. Nevertheless, plant safety can be improved even if generic data are used. However, the assignment to a safety integrity level may be affected by differences in both data and success criteria.

Probabilistic Safety Analysis of a Plant for the Production of Nitroglycol including Start-up and Shut-down

Probabilistic safety analyses for process plants generally only address accidents occurring during production operation (cf.[1], [2]). This is in contrast with the numerous start-ups and shut-downs required in most chemical plants. For this reason an investigation is presented which covers all the three operational phases mentioned. It was performed for a plant producing nitroglycol, a very powerful explosive. The presentation is limited to the production part, although other plant areas as, for example, storage were investigated as well [3].

Feasibility study for a passive trip system to prevent a runaway reaction in a batch reactor

Reactors for carrying out exothermic reactions are amongst others equipped with trip systems. Normally these are active systems requiring a number of components such as sensors, pumps or valves to function for a successful trip. They may, for example, use the injection of a reaction inhibitor or the dumping of the reactor contents into a knock-out tank. The availability of such systems, i.e. their probability of functioning on demand, largely depends on their degree of redundancy. However, the possibility of common cause failures places a limit on increasing their availability by raising their degree of redundancy. Nevertheless, a trip system may reach a high availability if instead of stepping up its redundancy a passive system is used. The design of such a passive trip system for batch reactors is described and its feasibility is demonstrated by experimental investigations of three different types of reactions.