Gday Steve,

I tend to agree with you. I have heard this view a couple of times, particularly from some of the military quarters.

The thinking, which in itself is hard to fault, is that you need to take into account the fact that the inspection task may not capture the indications of a potential failure every time.

The use of human senses for condition monitoroing would be a good example of where this could be observed I suppose.

Technological methods are more likely to capture the warining signs of failure in my experience.

Steve,

I do not know if you mean the idea of inspection intervals being fixed at 50% of the P-F interval, or in some circumstances at 33% of the P-F is an old-wife's tale when you say

quote:

I think these rules came from one of those old wives that wrote the old wives tales... or as someone once so delicately put it at one of our workshops.... " around here - someone writes something on the dunny wall and before you know it it is policy".

In which case, please meet an old wife. I have explained the basis of the 50% and 33% rules on pages 190, 191 of my book. So I guess I qualify for that title. Right now I am away from base, so I dont have access to my copy of Nolan & Heap. I believe they have also explained this principle.

Nice to get your feed back Vee... I can not find my Nowlan and Heap either... I think I must have lent it to somebody and forgotton to get it back. Anyway - I am looking for a good answer on this one so hopefully you can provide it. My memory is far from perfect but I don't recall this being in Nowlan and Heap. I would be happy if it is, so I at least know the source... By the way - can you send me a free autographed copy of your book?

Steve,

Just to clarify. (Because I have become confused here...)

We agree that the frequency of the routne task in a PdM task should be less than the P-F Interval don't we?

We also seem to be in agreement that around 50% is probably good enough for most if not all failures right?

So what we are trying to get to is the thin issue of whether or not 33% and other lesser, or greater, periods are justifiable. is this right?

quote:

We also seem to be in agreement that around 50% is probably good enough for most if not all failures right?

Daryl... I don't buy this one. I would recommend the basis of the interval is to allow enough time after detection to rectify the fault without causing significant disruption to the operation. This is completely independent of the PF interval so by factoring the PF to come up with an interval takes the logic out of the process and applies a rule that does not make sense.

quote:

So what we are trying to get to is the thin issue of whether or not 33% and other lesser, or greater, periods are justifiable. is this right?

I am suggesting this whole factor thing is not the best way 50%/30% or what ever.
Estimated the order of magnitude (days, weeks 1month, 2 months etc) of the PF (be conservative if you think it is hard to pinpoint), then figure out how much time the planners will need to get a corrective in place and then make selection of the interval required to inspect and minimise the production impact. No point selecting half the PF interval as the inspection interval if the lead time for the parts is 70% PF.
Regards
Steve

Darryl,

Good to hear from you. Email me sometime when you have time at ricky.smith@ivara.com

To continue the discussion on the PF interval, What most experts say is that you take half of the PF interval but the problem is time based failures only account for 20-30% of failures based on all the studies I have seen. I was a DS Maintenance Company Commander in the US Army Reserve recently maintaining all types of equipment for over 100 coalition units in Kuwait and Iraq. The time based PMs do not work well and I proved it on this deployment with sand storms, extreme heat, severe floods, etc. We found equipment failures were mainly random in nature. This why the US Military uses the 1/3 rule in order to be on the "safe side". The "safe side" is based on risk and consequences. With risk and consequences come cost in labor and material. If you go back to what Darryl was saying about RCM defining 4 different types of routine maintenance which are PdM, PS, PR, and DM. This issue is very complex. This is why I recommend all maintenance professionals have some level of RCM Training in order to understand failures and how we can predict or prevent them. If anyone would like a graphic showing these 4 types of routine maintenance send me an email and I would be glad to send it to you.

How to reliably estimate the order of magnitude of the PF curve eg for a turbine, compressor & pump? Any guidelines?

How to get spare parts ontime for repairs? Some tips that I know when setting up maintenance strategy:

1) Build up equipment BOM with proper material master & inventory. For vital eqpt like a stand-alone compressor, keep capitalised spares and for essential eqpt eg pumps, keep min-max stock.

2) Set up a price agreement with a spare part supplier so he can deliver immediately upon request.

3) Also set up a service agreement with a vendor for overhaul if not done inhouse, to reduce lead time to start of work. Items 2 & 3 can be combined into a package contract for optimizing contracting activities.

quote:

How to reliably estimate the order of magnitude of the PF curve eg for a turbine, compressor & pump? Any guidelines?

Josh - the guidline is you need to ask the right people the right questions. In industrial equipment, you are unlikely to find the answers in a computer or a book.
Regards
Steve

quote:

This why the US Military uses the 1/3 rule in order to be on the "safe side". The "safe side" is based on risk and consequences. With risk and consequences come cost in labor and material.

In industry budget contraints play a higher role.

Ricky,
You are right on however, from experience most do not have adequate equipment history nor do they think they have time or take the time to calculate the P-F interval.

If true MTBF's and machine history is missing I start out using 3db or 0.707 of the P-F interval as a starting point then use my data to fine tune. It has worked well so far.

Reliability programs require accurate data without it what we are trying to achieve will fall short. I beleive this is why most programs fall short. Every once in a while we find a company that really understands and will approach world class performance.

I've seen it many times. I do believe the more we can automate and build into CMMS, work order systems and other means the data needed, the more successful the program will be.

Regards,
Spencer Hatfield

quote:

If true MTBF's and machine history is missing I start out using 3db or 0.707 of the P-F interval as a starting point then use my data to fine tune. It has worked well so far.

Steve - how do you use your data to fine tune?

quote:

Reliability programs require accurate data without it what we are trying to achieve will fall short..

Also - we are never going to have accurate failure data because the whole idea of maintenance is to eliminate the cause of data - which is failure. This depends on what data you are collecting I guess - and I am talking about failure data here.

Hi Daryl, tudo bem obrigado. I am sorry for being so late in answering your post from the 17th.

Let me add some ashes to this fire…

I know that many practitioners adopt a constant time interval between every two successive inspections which is made equal to some fraction of the PF interval. I even read Steve’s paper on this: TURNER, Steve, Understanding the differences between Cost Minimisation Algorithms and the RCM concepts developed by Nowlan and Heap, s.ed., 1978, but I don’t agree with his approach.

First reason: It doesn’t make sense in my mind to adopt a constant time interval when you are expecting a degradation failure mode to reveal. It should be a progressive decreasing interval instead, as it becomes more and more likely that the failure occurs as time progresses.

Second reason: If the consequences are merely operational (safety and environment are no concern), why not to accept a certain level of risk of not detecting in time the failure in progress (some point ahead of point P) and suffer the consequences? If it is only a matter of money, the consequences might not be too expensive as compared to inspections costs over time. It might be worth the risk. Thus, purely economic grounds wise; there may be a calendar of inspections that might return the minimum cost per unit time. Don’t you agree?

An example: Consider the case of a mechanical device that fails predominantly due to a wear out mode described by a Weibull distribution with shape parameter of 2 and a scale parameter of 8,000 hours. The accumulated life time is 3,000 hours and presented no sign of the proximity of a failure in the course of the last inspection which took place at 2,600 hours (400 hours ago). Time remaining before the next overhaul is 10,000 hours. The minimum acceptable reliability is 0.9, or in other words, the chance of a failure in progress not to be detected between any two successive inspections and a failure does occur is 0.1. According to my calculations the calendar should be:

3,672; 4,498; 5,193; 5,806; 6,361; 6,870; 7,345; 7,790; 8,212; 8,612; 8,995; 9,363; 9,716

In practice, all these figures would be rounded up to the nearest tenth or hundred, of course. As you can see, the time intervals become shorter and shorter.

I might describe an example comprising the PF interval in a next post…

I would like very much to hear your comments on this issue.

Regards,

Rui

This message has been edited. Last edited by: <Rui Assis>,

May we know who are the right people to ask, operators, consultants, OEM? And what are the right questions? Any examples, to illustrate your point better?

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

The idea of progressively decreasing inspection interval seems realistic. We would increase the frequency of inspection after detecting the onset of failure to know the rate of deterioration.

Rui,
Are you expecting Daryl to respond to your post or me?

quote:

Originally posted by Josh:
The idea of progressively decreasing inspection interval seems realistic. We would increase the frequency of inspection after detecting the onset of failure to know the rate of deterioration.

Absolutely but only after the onset of failure is there.

Rui, I agree with Josh and Steve on the basis of a potential failure being identified. However your arguement (correct me if I'm wrong) is that we should be reducing this period anyway which means we'll get to the point - if the equipment doesn't fail when we have calculated - we a re checking it every second or less????

Steve, I hope you can answer "Who to ask and what to question?" about estimating the PF curve. I didn't address specifically to you to allow others to chip in in this public forum.

If I ask operators, sometimes they tend not cooperative because they are scared of bad consequences from what they say. This really happens during RCA exercise?

What sort of questions are right to ask to get the right info and to encourage people to speak up. Is it "How long does it take for this machine to fail?" If I ask this question, won't some of the operators refer me to CMMS.

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

quote:

Steve, I hope you can answer "Who to ask and what to question?" about estimating the PF curve. I didn't address specifically to you to allow others to chip in in this public forum.

Josh – If you are looking for a condition based maintenance program the “who” to ask are the people who know the equipment / failure characteristics best. They can be the operators in a packaging line or it could be the university PHD who developed your structures model for a massive mast on a crane or dragline.
There are several ways to get the right inspection interval. The least successful is “What it the PF interval?”. The more successful approaches are to do some training in task selection concepts so that the people you are asking understand the reason for these questions. Then you should be talking about obtaining zero tolerance to unexpected failure – that means that if you are setting condition based maintenance, then the task needs to be robust and the interval should be set such that the failure will never occur without someone knowing. The question in this case is then … “How often should you inspect this equipment for that failure mode such that that failure mode would never occur without you knowing?” You may need to tease the answer out by saying – what about if you inspected every four weeks – would you miss some failures – oh ok – what about two weeks?... I hope you get the gist of this approach.
Other things such as leaks which have a variable PF – suggest to the group that the leak could be in existence for the whole interval of inspection if the leaks starts shortly after the last inspection – the critical question then is this “How often would you think you would tolerate a leak in that area without anyone knowing?”.
So there Josh – you have a start. Your question is a bit like asking a soccer player how to kick the ball…. There are many ways to score and experience and skill will determine how successful you will be at the end of the game. I guess that is why there is a market for good training and good coaching.

Guys,

I am absolutely flat out right now, but I will try to get back to these points over the next day or so. Great thread...!

Rui:

quote:

It doesn’t make sense in my mind to adopt a constant time interval when you are expecting a degradation failure mode to reveal

I am not talking about a degradation failure mode. I am talking about adopting a constant time interval to detect the warning signs of failure.

If we are talking about a degradation failure mode, such as wear of impellers on pumps, or wear of rail-track then I would propbably adopt a different approach. (Such as change out or refurbishment based on the calculable "life" due to wear, or via weibull analysis etc)

Steve,

quote:

then figure out how much time the planners will need to get a corrective in place and then make selection of the interval required to inspect and minimise the production impact. No point selecting half the PF interval as the inspection interval if the lead time for the parts is 70% PF.

As you say Steve, I don't buy this at all. Reliability is driven by the requirements of the plant, production process, or safety / environmental requirements. Not by the lead time of inventory items.

If you are going to have a timeframe of (say) two weeks, and the lead time of the item is three weeks, then you need to have one (at least, depending on population) on hand.

I will get back to this thread soon, apologies for the short reply.

(G'day Ricky.. I will write shortly, hope all is well where you are these days)