Research On Safety Integrity Level Of Safety Instrumented System Based On Bayesian Network

With the rapid development of modern industrial production, people’s material life has increasingly been enriched and the quality of life has been greatly improved. At the same time, due to the large amount of industrial production equipment put into use, to control the potential risks of catastrophic accident hazard has become the problem that can not be ignored. It has become increasingly difficult for people to tolerate the various risks that endanger their lives, property and the environment. In order to improve industrial safety, safety instrumented systems (SIS) are usually used to play its safety features before the happening of security incidents. By this way, in order to avoid accidents or to reduce the harm caused by the accident, the safety instrumented system has been more and more widely used in modern process industry. It is the critical facilities to protect the process of industrial safety. Thus, safety and reliability of the safety instrumented system its own become the assurance that make it timely perform its safety function when the accident occurred foundation. However, due to its structure, hardware and software, the environment and other causes of safety instrumented systems, verifying the safety instrumented system whether meets the expected requirements has become an important issue for functional safety studies at home and abroad. To research the verification method about the safety integrity level of safety instrumented system is precisely to make safety instrumented system play its proper safety features accurately and timely and to ensure production safety. Therefore, the research of safety integrity level authentication methods has become a hot topic in the field of functional safety.Currently, there are three main methods of safety integrity level verification of safety instrumented system:fault trees, reliability block diagram and Markov models. Fault tree model which is simple and intuitive with a clear diagram can show the series and parallel relationship of the syste, while unlike the difficulties of Markov model for large complex systems modeling. It can make both qualitative and quantitative analysis. But the fault tree is based on the assumptions that the state of events is two-state resistance and the logic of fault states is deterministic. Because of many events of safety instrumented systems with polymorphism, fault tree is difficult to accurately describe the events of polymorphism. Secondly, the amount of calculation using fault tree analysis to modele large systems is very large. The amount of the logic gates can be used to characterize the size of the fault tree. The amount of calculation using fault tree analysis usually will be increases exponentially with its size.For fault tree analysis can not describe the lack of polymorphism, a approach based on bayesian network to verify the safety integrity level of safety instrumented system is proposed. The basic idea is that transform the fault tree model of the system in accordance with the correspondence into a bayesian network, and get the conditional probability table of the bayesian network nodes obtained from the fault tree logic gate. Then use the joint probability distribution to calculate the safety integrity level of the entire system, achieving the purpose of verification. Such verification method overcomes the shortcomings of the fault tree analysis method that can not solve the polymorphism problem. And it reduces the computational complexity of a large complex safety instrumented systems when validating its safety integrity level.In this article, the bayesian networks of different redundant structure (lool, loo2, loo2D) were analyzed, and on this basis, the safety integrity level calculation method of the three redundant structural were shown. It is indicated that the proposed method not only validated suitable for direct hardware system consisting of components, can also be used for different redundancy structures system to calculate safety integrity levels under different failure modes.In this paper, with the example of recycle hydrogen gas pressure furnace safety instrumented system, use the method presented in this paper which based on bayesian network to compute and analyze the safety integrity level of safety instrumented systems. Because of the introduction of bayesian network, it becomes simpler and accurater to calculate the security integrity level. It is also able to improve the system sensors part and actuators part by redundant configuration according to its verification results. Thereby their overall safety instrumented systems safety integrity level improved. This process and conclusion indicates that the method is applicable to compute the safety integrity level of safety instrumented systems, and the calculation has the characteristics of simple.