Going back in time to the turn of the century (remember 2000?), the amount and quality of failure data threatened to doom the whole probabilistic analysis approach.  I recall a letter in IEEE Spectrum magazine from Patrick D. O’Conner, a well known reliability engineering author, who expressed the opinion that IEC 61508 was a unless burden that would force widespread waste of engineering time. 

We at exida recognized that probabilistic analysis had great potential to allow companies to improve safety while actually lowering cost however more realistic models and failure data were needed.  We have studied hundreds of field failure data sets and incident reports for more than a decade.  The first thing to become clear was that there was a need for better models.  We also recognized that all proof tests are not created equal.  Proof test coverage was an essential variable in probabilistic failure analysis, and is now a commonly accepted fact.  Currently, our research is focused on site specific failure rates.  Stay tuned to the exida website to see updates on this research.

Today, more realistic data has become available.  The exida Safety Equipment Reliability Handbook has become the standard reference worldwide.  Although some criticize the failure rates as being too low, others criticize the failure rates as being too high, therefore these differences must be investigated.  Although exida has documented over sixty billion unit operating hours of failure data, we continue to gather data and are always looking for improvement. That is how one sets the pace.