Jl Jan Rouvroye

Comparing safety analysis techniques

Abstract In process industry Safety Instrumented Systems (SIS) and Emergency Shutdown Systems (ESD) are very important for the management/reduction of risk. In new standards (e.g. Ref. [1] ) on functional safety of electrical/electronic/programmable electronic safety-related systems a quantification of the achieved safety is often required. These new standards do not prescribe how to calculate the achieved safety. Only guidelines and recommendations are given. The problem with this approach is that all kinds of different analysis techniques will be used and in industry the results of the analysis will be compared. These different analysis techniques all use different methodologies and assumptions, which implies that the results may not be comparable. In this paper an approach for comparing different analysis techniques and the qualitative and quantitative results from this comparison are described. The author suggests that, because of the differences in the analysis techniques, one analysis technique is to be preferred. The Enhanced Markov Analysis technique, described in this paper, could be used for this purpose because it covers most aspects relevant for quantification of safety.

Managing product reliability in business processes 'under pressure'

Product reliability is often seen as a product attribute. Models with different degree of sophistication analyze and predict the reliability of a product as a function of the internal structure (such as components and their relation). The practical relevance of these models, in relation with the (business) processes in which the related products are actually used, is not often addressed. Different types of reliability issues, however, can be relevant for products in different industrial contexts. This paper will present a classification model to describe different business processes, based on the degree of product innovation. It will also propose a taxonomy that can be used to classify different types of reliability problems. As this paper will demonstrate, only certain combinations of reliability problems are relevant for certain business processes. It will also show that, given certain technology trends, some combinations will become more relevant in the future. The final part of this paper will demonstrate that especially for these combinations many of the existing reliability analysis and prediction methods can be considered inadequate.

New quantitative safety standards : different techniques, different results?

Abstract Safety Instrumented Systems (SIS) are used in the process industry to perform safety functions. Many factors can influence the safety of a SIS like system layout, diagnostics, testing and repair. In standards like the German DIN no quantitative analysis is demanded (DIN V 19250 Grundlegende Sicherheitsbetrachtungen fur MSR-Schutzeinrichtungen, Berlin, 1994; DIN/VDE 0801 Grundsatze fur Rechner in Systemen mit Sicherheitsaufgaben, Berlin, 1990). The analysis according to these standards is based on expert opinion and qualitative analysis techniques. New standards like the IEC 61508 (IEC 61508 Functional safety of electrical/electronic/programmable electronic safety-related systems, IEC, Geneve, 1997) and the ISA-S84.01 (ISA-S84.01.1996 Application of Safety Instrumented Systems for the Process Industries, Instrument Society of America, Research Triangle Park, 1996) require quantitative risk analysis but do not prescribe how to perform the analysis. Earlier publications of the authors (Rouvroye et al., Uncertainty in safety, new techniques for the assessment and optimisation of safety in process industry, D W. Pyatt (ed), SERA-Vol. 4, Safety engineering and risk analysis, ASME, New York 1995; Rouvroye et al., A comparison study of qualitative and quantitative analysis techniques for the assessment of safety in industry, P.C. Cacciabue, I.A. Papazoglou (eds), Proceedings PSAM III conference, Crete, Greece, June 1996) have shown that different analysis techniques cover different aspects of system behaviour. This paper shows by means of a case study, that different (quantitative) analysis techniques may lead to different results. The consequence is that the application of the standards to practical systems will not always lead to unambiguous results. The authors therefore propose a technique to overcome this major disadvantage.

A comparison study of qualitative and quantitative analysis techniques for the assessment of safety in industry

The primary function of safety systems is avoiding personal injuries or death and environmental pollution. Major problem is that, at this moment, there is no standardised method for the assessment for safety in the process industry. Many companies and institutes use qualitative techniques for safety analysis while other companies and institutes use quantitative techniques. The authors of this paper will compare different quantitative and qualitative techniques and will show that different analysis techniques show widely different results. It will also be shown that the qualitative techniques are highly dependent on human judgement while the quantitative techniques suffer from highly uncertain underlying data. Therefore in the second part of the paper a new quantitative technique will be presented that can be used for the assessment of safety, availability and triprate in the process industry. This technique can take into account effects of uncertain data and identify parameters that influence safety and/or availability dominantly. Finally the paper presents an example of the new technique applied to a practical system used in Dutch petrochemical industry.

Developing MESA: An accelerated reliability test

This paper describes the on-going research on an accelerated reliability test strategy called MESA (Multiple Environment Stress Analysis) intended to find in a fast and efficient manner (potential) reliability problems during the design phase of high volume consumer products. This test has shown promising results on its capability to provoke failures, especially on failures formerly registered as NFF (no-fault-found). The strategy applies combinations of stressors in a statistically efficient test scheme. To validate results, MESA must be able to reproduce failures with similar stressor configurations and test scheme. First pilot tests have been completed. The results show that it is possible to induce realistic, previously unknown, field failures in an efficient manner during the design stage. Considering that this test was the first pilot on the development of MESA, many opportunities are still left to be further explored.

Risk reduction through partial stroke testing

The process industry uses a variety of trip valves that provide emergency shutdown functions and, thus, play a very significant role in the safety of process plants. The reliability of such valves must be commensurate with the criticality of the process they safeguard and must comply with good engineering practices, industry standards, and regulatory requirements. Trip valves are tested periodically with a full closure test, usually known as the full-stroke test to ensure the operational reliability of the trip valve system. This type of test may present practical difficulties related to process disruption and associated cost and safety issues. Therefore, a partial stroke test of safety trip-valves has been proposed, which reduces the implications of the full-stroke test and provides an indication of the reliability/availability of the trip-valve system response on actual process demands.

Improving state-of-the-art power plant availability using Bayesian Networks

The Integrated Gasification Combined Cycle technology in combination with coal as a natural resource is claimed to be one of the leading power generation alternatives for the near future. One of the main challenges facing this technology today is plant availability. Due to the specific characteristics of this technology, current practices cannot sufficiently address plant availability. A different approach is needed. This paper proposes the use of Bayesian Networks as an extension of existing methods. Ultimately, the approach described in this paper is directed towards scenario-informed decision making to improve plant availability. This includes the prioritization of potential scenarios of unavailability and allocation of resources to prevent or mitigate the most serious scenarios, as well as operator training and identification of indicators of potential unavailability. A case study that demonstrates the application of the initial steps of the proposed approach has been conducted within a Dutch IGCC plant. A Bayesian Network was constructed for the syngas treatment unit of the plant, in which identified unavailability scenarios were modeled. Results from the case study indicate that our approach can help a company to identify, quantify, and prioritize scenarios, which can act as an input to improve availability management.