Improvement of Safety Instrumented System (SIS) Performance through Detection Diversity

Abstract

Abstract Detection coverage plays an essential role in defining the capacity of a fire and gas system to perform properly when required (ISA-dTR84.00.07 2008). The concept seems clear: As the probability of a sensor detecting a hazard increases, so does the likelihood that fire and gas systems (FGS) are able to mitigate more severe consequences. Several reports document the importance of detection coverage. Statistical studies of hydrocarbon gas releases by the Health and Safety Executive (HSE) calculate the detection rate of gas detectors at 60% (HSE 2000) and 76% (HSE 2003a). Another report by the same agency on gas turbine incidents between 1991 and 2004 puts gas detection coverage at about 78% (HSE 2008). The collected historical record suggests that, even if an FGS design complies with a high safety integrity level (SIL) target, a reduction in risk is not achieved unless detector coverage is high. An independent calculation of risk reduction under different scenarios of detection coverage supports the empirical findings (Choo 2008). An important, but often neglected element of detection coverage is the contribution of applying two or more sensing technologies toward risk reduction. Detection diversity improves the odds that a hazard is detected early on, independent of the number of detectors installed, their reliability, and geographic coverage. Diversity is often effective, because detection methods share few common failures. By drawing on the advantages of each, several different types of detectors can be more effective at mitigating the spread of an accident sequence than any one type alone. In this report we examine how fire and gas detector coverage can be improved by means of detection diversity. The study applies Markov models to illustrate the potential risk reduction as a function of detection diversity. Monte Carlo simulations are also used to estimate the contribution of detection diversity on attaining a specific SIL. The results confirm that field devices have a substantial effect on SIS performance. The paper concludes with recommendations on the application of several detection technologies to protect offshore facilities. Note: This paper is not about whether or not FGS can be considered part of a SIS. The authors understand the arguments for and against this.