Currently,the calculation of the average probability of dangerous failure on demand(PFDavg)for safety instrumented systems(SIS)was based on the method of complete proof testing(as-suming 100％coverage of proof testing).However,in practical application,this completed proof testing only exists in ideal situations,especially when faced with complex systems or maintenance strategies,and the calculation errors was difficult to accept.This article is reported that the failure partitioning under in-complete proof testing,and based on the new failure classification,uses Markov state transition to con-struct a probability model.Based on this model,PFDavg is calculated to reduce the error in probability calculation.
万方数据