I got an email saying that the exida's www.SILSafeData.com minimum failure rates were way too high.  The email went on to say that his REAL field failure data showed a result of 150 FITS for a remote actuated valve assembly.  The lowest SILSafeData limits for clean service, Class IV leakage, close to trip assembly of a solenoid valve, pneumatic actuator and ball valve are 480 FITS and other selections give numbers in the 2000 FITS range.

exida’s SILSafeData tables are based on data from three field failure studies and over 350 billion unit operating hours of manufacturers field return data.  One set of field failure data that we consider to be of top quality is the OREDA data set from SINTEF in Norway. In a detailed study of the OREDA databook [Reference 1] we found remote actuated valve assemblies to be in the range of 2000 FITS and up.  So even taking the optimistic (lowest) number of 2000, this 150 FIT number seems to have no credibility.

How can this be? exida has performed a review of the field failure data collection procedures for several companies, especially those operating plants that use the exida SILStat data collection software. In those reviews we have found lots of issues that could influence the results [Reference 2].  These include:

1. Discarding real failures classified as “Systematic” [Reference 3]
2. A data collection process that misses many real failures
3. Data is collected for control valves that move frequently as opposed isolation valves used for safety
4. The failure data does not include drift/calibration failures when they can be calibrated out
5. A data collection process that only counts manufacturing defects
6. And many more

We must remember that any SIF failure that prevents the SIF from responding to a safety demand could result in an accident.  ALL failures need to be counted in any field failure analysis.  OREDA does this.  

In this case we do not know the reasons for the optimistic (and potentially dangerous) results presented as an argument to reduce the lower limits on SILSafeData.  We could dig into this data collection process to answer the question and exida does offer this valuable service.  Without that study, I trust OREDA and the resulting SILSafeData limits. 

References

1. exida White Paper, “Explaining the Differences in Mechanical Failure Rates: exida FMEDA Predictions & OREDA Estimations”, www.exida.com
2. exida White Paper, “Field Failure Rates - The Good, The Bad, The Ugly”, www.exida.com
3. exida White Paper, “Random versus Systematic Failures – Issues and Solutions”, www.exida.com