| description abstract | This paper addresses dangerous failures of stainless steel (SS) trim springoperated pressure relief valves (SOPRV) due to a particular failure mode (SStoSS adhesion), which is not currently being included in SOPRV failure rates. As a result, current methods for estimating or predicting failure rates for SS trim SOPRV significantly underestimate these failure rates and, consequently, overestimate the safety provided by the SOPRV as measured by its average probability of failure on demand (PFDavg) or its corresponding safety integrity level (SIL). The paper also illustrates the critical importance of root cause analysis (RCA) of dangerous SOPRV failures in understanding the impacts of various failure modes. Over 1300 proof test results for both new and used SS trim SOPRV from the Savannah River Site (SRS) were identified. RCA was used on the failed valves to classify those failed due to SStoSS adhesions. Statistical analysis of the data convincingly demonstrates adhesions, previously assumed to be only an instorage failure phenomenon, are also an inservice failure mode, which needs to be included in SOPRV failure rates. The paper discusses the factors which potentially influence the adhesion failure mode, and suggests a possible approach to including this mode in failure rate predictions. An example illustrates how current failure rate models overestimate SS trim SOPRV safety by 1 or 2 orders of magnitude. | |
