Let me begin with an apology. In what follows I might sound harsh; please appreciate that I feel strongly about some things, and the words could probably have been better chosen. I mean no offence and I respect your views. In this case I happen to disagree with your statements, but do not claim to be right. It is just another view.
Maintenance would not exist if plants did not fail.
Economics dictate the level of gold plating. To build a plant that would never fail is not an economically viable proposition. Having established that, let us distinguish between items that can be run to failure because their failure does not matter and those whose failure costs us an arm and a leg (sometimes literally!). We have to find out which ones these are, whether at the design stage or in the operating stage. Some kind of analysis is required, be it a FMECA or RCM or some other method.
The second issue is this: the designer builds in a certain level of reliability. In practice we achieve a much lower reliability (30-60% in terms of run lengths), because we dont operate or maintain it correctly. Getting this right can double the run lengths we achieve today, getting plant availability up by 3-5%. How much money can that make? Let us think about that before we quibble about RCM costs.
In nearly every RCM study I have been involved in, we found 3-5% new failure modes that were 'hidden' and we did not realize were there. Just think of what that means. Before the study we were not aware of these sleeping tigers. A very small number of them may have resulted in a Piper, Bhopal,a BP Texas etc., but we need to know which ones they are.
We have a primary duty to run our plant safely. Hidden failures are always a problem for technical integrity, but if we dont know they exist, what then?
If 'streamlined' or turbo or jet RCMs can do the same job cheaper, let us go for it, BUT not if they can miss any sleeping tigers. Oh yes, we do need to know ALL the credible failure modes, not pick and choose the juicy ones only. We only need to miss one really important hidden failure to suffer badly.
Because this work is done prior to certification, RCM was developed as a process developed for the design phase of an assets life cycle.
I think you are rewriting history. You know very well what happened in the 1950s and 1960s to Civil Aviation in the US. Specifically what happened to Continental Airlines after they doubled the frequency of engine overhauls and even the FAA did not know what else to do. Nowlan & Heap and United were not working on a new process to build better planes. They were trying to salvage their existing fleet from going down in flames.
For that they needed to study failures SYSTEMATICALLY. Out of that study RCM was born. Its sister MSG-1 was used for building a reliable plane. The tool was developed to stop planes falling out of the sky; it was adapted to build better planes.
Whether in the design or operating phase, I would argue you need to know
- why you need something i.e. its function
- how it can fail, i.e., functional failure
- how it manifests itself i.e failure mode
- what happens when it fails i.e. consequences
- whether this matters
- what we can do about it
- what if we cant do something about it
I am not a 'classical' RCM fan. Straightjackets of any kind are not my cup of tea. However to argue that shortcuts are fine even if they do not address these issues is, for me, not acceptable, on cost or resource availability grounds. If they do address these issues, I am all for it.
If one is standing in the docks when the company is sued for neglect, it may not be easy to say we could not afford the cost or spare the people to do a proper job. Worse if one has to tell a grieving spouse.
Let me repeat the apology. I am only writing this spurred on by Steven's analysis of your post, which other readers may have similarly analysed.