H Svanels,
At the level you are looking at your equipment (that is at plant level) you will almost certainly see a bath tub curve unless you do something to change that. That something is called the right maintenance.
How relevant do you think your maintenance activities are?
The failure curve we aim for at plant level is one that heads South no matter how old the equipment is. We like to achieve that through a combination of the right maintenance and modification where needed.
Regards
Steve

Welcome svannels, enjoy the game. If you see the hot dog salesguy call him for me. I want one with a Pepsi.

This message has been edited. Last edited by: Eugene,

Hi folks,
I had the occasion to refer to the SAE Standard JA1011 and found the following statement. Thought it might be useful in the context of the thread.
5.6.1 The failure management selection process shall take account of the fact that the conditional probability of some failure modes will increase with age (or exposure to stress), that the conditional probability of others will not change with age, and the conditional probability of yet others will decrease with age.

Dear Steve,

Many thanks for the quote. You reinforced the point I made earlier.

quote:

5.6.1 The failure management selection process shall take account of the fact that the conditional probability of some failure modes will increase with age (or exposure to stress), that the conditional probability of others will not change with age, and the conditional probability of yet others will decrease with age.

However, as you point out, the conditional probability tends to level out and flatten, as complexity rises, a point that N&H pointed out in 1968. I would argue that at systems level or higher, the curve is an 'E' or 'F'. So what is the point? Plants dont grow old, bits of it do, especially at failure mode level. Repair or replace those bits on time, and voila, the Plant as a whole can last a long, long time. End of life is more a question of obsolescence, not hours, cycles, miles or number of operations. There are Refineries built in the 1930's or 40's which are running efficiently and economically today. They started life with a design life of 25 years!

Thanks once again Vee.
I think we are in agreement - I just thought I would assist with some deeper perspectives /opinions of the Nowlan and Heap work and in particular the failure patterns used inappropriately at times (not all the time - just some times). The SAE standard recognises that failure modes can have three conditional probability curves (failure patterns). Data is often an aggregation of these as data is never perfect.

The original question posted by RCM2 was this:

<<As u know there are 6 failure patterns which the equipment or part will fail in one of those patterns Moreover , the first 3 patterns consider as age related and the others are random failure.>>

The statement that there are 6 failure patterns can be argued. First of all, I have seen double humped patterns which shows there can be more than six, and secondly, the answer depends on whether you are looking at data retrospectively or doing predictive analysis such as RCM in which case we should be working at failure mechanism level and there are only three as suggested in the SAE Standard.
I think that a lot of people make the wrong statements due to not knowing the context of the data and the context of the problem they are solving. This can lead to poor analysis.
Regards,
Steve

Just as an adjunct to the debate on failure patterns, it might be interesting to note that the bath tub curve is alive and well (as of 2002) and business cases are still being predicated on it.

Try this link:
http://www.nao.org.uk/publications/nao_reports/01-02/0102757.pdf

Having spent 15 years in the UK MoD trying to get the RCM message across (and with some success in the Royal Navy) what I thought were pockets of resistance turned out to be whole battalions .....

Malcolm Regler

Do we really need to plot these failure patterns in order to make our PM program realistic?

Absolutely not Josh. Most maintenance should be condition based which has very little (if anything at all) to do with the failure patterns. For the relatively small number of occasions you are looking at prescribing hard time replacement or refurbishment decisions (you are going to replace or refurbish items at prescribed intervals regardless of their condition) then the idea is to change something before its safe or economic life expires. The failure pattern in useful on these occasions. In industrial applications however, there are often too many variables and too few data points to make accurate assesment. You will find people who seem to be able to construct some kind of magic with three data points and make some predictions. All I can say about this is that those people must be a heck of a lot smarter than me.