@Khalid Kalbiyev The certificate indicates that the reliability assessment for this PLC was made on the machinery FS standard. So you hardly find PFDavg values there. So you have to find other solution for this issue. So, instead you have some PFH numbers which you can treat as λ and use them in your own calculations of PFDavg. Just take number of DIs and DOs that are needed to manage all input and output signals. Then add CPU. And also don't forget about other components which are part of the safety PLC (like power supply, communication modules, etc.). They are components in series in a reliability block diagram - just add them and calculate final probability of failure within your mission time with some assumed proof test interval and proof test coverage (may be a generic data from safety logic controller).

I appreciate your work maintaining this forum. I have many ideas for topics, but not enough time to ask them before my vacation :)

Hello everyone. Quite long time without answer so maybe I can trigger this topic once again. I really wonder your opinion about above high demand statement for the SIFs with breakers as final element. I have many doubts about this approach in the context of real possible demand for SIFs which protect rotating machines in process industry. My experience says that these critical ones usually work for long time without any stoppage, so should be treated as low demand. And one more: good quality breakers have a high operational reliability but usually don't have IEC61508 compliance documentation.

@Harvey Dearden Same approach in my many projects. It's difficult to find a more common sense solution today. Confirming product quality based on the manufacturer's quality system and a declaration of compliance with IEC electrotechnical standards relevant to the implementation. However I look forward others point of view, it's always inspiring to find some new ideas.

I want to return to the topic of legacy systems, because I've just read your additional article on LI about this topic :) We're talking about systems that typically have dozens or even hundreds of SIF functions and even more devices. The problem certainly arises when we look at the devices installed before IEC61511. Let's assume they've all been operating for decades without any problems but for some reasons we want to do sth with their compliance. Doing assessment and relying 100% on generic reliability data can be a trap. Let's look at current equipment reliability data declared by manufacturers. Their expected lifetimes show that devices in legacy systems have typically exceeded these values ​​long time ago. Without evidence of testing, servicing, and refreshes, there's no way to formally confirm that these systems have the reliability we're aiming for. I think the approach you're presenting is based on common sense. There's probably no better method than successive analysis of individual system components and categorizing their importance based on risk assessment. And, whenever possible, upgrade to newer solutions based on real needs with cost-effective way.

Mr Bachir, I don't know French but I tried to translate it. You are right about the need for a structured and well-defined way of collecting data. But what is your proven method for verifying that the devices to which you assign 'prior-use' data are not actually at the end of their useful lifetime, and that they won’t cause an increased probability of a major failure tomorrow?

I agree with Anth. A banefit of your proposal is very minimal. Look at the failure rate of safety PLC in comparison to other elements - it's a very small part of overall system PFDavg. So you should aim at improving other parts rather the best one.