I had an engineer from a European Certification Body ask me why exida would bother paying any attention to these failure rate numbers, especially on mechanical devices? He said that he does not know anyone who actually uses them! To answer why exida pays attention to failure rates, we must ask what are we trying to accomplish by using probabilistic analysis to evaluate risk reduction provided by any automatic protection function. Why does IEC 61508 define the need for probability of dangerous failures? 

Some say that probabilistic analysis is done to check the requirements box. Special interests must have pushed to get probabilistic analysis put into the functional safety standards. Some say that probabilistic analysis is done to show how good an automatic protective safety function might perform under perfect maintenance? Are we trying to figure out how much of the problem is caused by product manufacturers? Pardon me, exactly where can we find perfect maintenance? This sounds like a “blame the product manufacturer” game?

Or are we attempting to realistically evaluate risk reduction achieved by our designs while operating in various conditions?  Can we use this to guide the safety improvements identified by the numbers? Can we use the numbers in combination with the cost of a false trip or the cost of an accident event to justify improvements?  Most operators prefer the realistic evaluation!  Top tier companies used the numbers to guide and justify improvements. 

A realistic evaluation must look at all the variables [1]. A realistic evaluation must use failure rates without extreme data filtering where real failures are excluded under the name of “systematic.“ [2] When realistic information including operations and maintenance capability [3] is entered into the calculations, the resulting numbers can show what needs to be improved and can show the results of those improvements when updated failure data is analyzed at a later time.  This is valuable. 

 It is sad that some just want to check the box.

References:

• [1] The Key Variables Needed for PFDavg Calculation
• [2] Random versus Systematic Failures – Issues and Solutions
• [3] Assessing Safety Culture via the Site Safety Index