Proceedings of the Eighth International Conference on Probabilistic Safety Assessment & Management (PSAM)
304 Evaluation of the Analytical Bottom-Up SIL Proof by Statistical Top-Down Methods (PSAM-0242)
Download citation file:
- Ris (Zotero)
- Reference Manager
According to the IEC 61511 a Safety Integrity Level (SIL) has to be assigned to all new safety instrumented loops in chemical and pharmaceutical industrial plants. An important SIL criterion is represented by the probability of failure on demand (PFD) defined as a probability interval for each different SIL. The PFD can be calculated based on the individual loop structure and the failure rates of all the loop components. Commercial software tools are available to assist the calculation. But, as a rule, these bottom-up methods need many assumptions and suffer from uncertain data, especially for inline components like sensors and actuators. The consequence is the data sensitivity of the PFD outcome yielding a sensitivity bandwidth for the PFD rather than a crisp PFD value. This bandwidth has been calculated for different technical examples by the use of commercial software tools.
On the other hand statistical data are available from more than 20,000 safety loops collected every year by 37 companies, which are European NAMUR members. These data are based on the behavior of the complete safety loops in use. Together with the known proof intervals the data have been divided into four groups (temperature, pressure, level, quality). They are analyzed yielding PFD confidence intervals as the essential information of the data pool. By internal comparison among the four groups significant differences become obvious. The main result indeed is delivered by confronting the former results of the bottom-up approach (PFD sensitivity bandwidths) with the outcome of the top-down approach (PFD confidence intervals). Again, unexpected differences are shown which are thoroughly elaborated in the sense of a bottom-up method evaluation.