Maintenance Task Selection - Part 4
Summarised by : Sandy Dunn
Webmaster, Plant Maintenance Resource Center
This is Part 4 (the last part) of a summary from the plantmaint Maintenance discussion forum which discusses alternative approaches to Maintenance task selection, including RCM, PM Optimisation (PMO), RCMCost, and others - it also touches on Total Productive Maintenance (TPM)
Go back to Part 3 of this discussion
From Dana Netherton
From: Walt Sanford
Last thread from me. There is a difference between being rigorous
and being overly resource intensive. There are ways to apply the
RCM principles, or "rigor" as it has been referred to, without
compromising the integrity of the intent. It has to do with the
organization and automation of the approach, the order and
combining of steps, and the flexibility to adjust the amount of
time spent on an asset once the importance has been determined,
using the "correct" level of rigor.
Walt, IMHO, you make a great deal of sense, here. I don't know
whether you and I would be 100% in agreement in every detail of
the follow through, but I think you've put your finger on something
very important. As you go on to say ...
Here's the rub. Some people have discounted RCM as a solution
because they only have been exposed to one, unpalatable rendition,
which they found had more unsatisfied than satisfied
practitioners, and these are often folks who need it the most.
This has spawned alot of "truly non-RCM" offers to spring up as
alternatives. We have applied "RCM" to 1000's of systems and
can't remember an instance where the application did not conform to
the criteria outlined in SAE JA1011, Nowlan and Heap, DOD
applications, and original MSG-3 guidance (as many clients have
When my subcommittee drafted SAE JA1011, we did so with the
conscious intent of making something that could support different
processes that might all be "RCM processes". In my view, the
various "RCM processes" all offer a great benefit. Where they
differ, they differ chiefly in whether their cost is greater than, or less
than, their benefit.
It is certainly possibly to do RCM wastefully -- to do it in such a
way that their benefit is obtained at great cost (and in such a way
that little tangible benefit is actually realized on the shop floor). As
I have said from time to time, "You *can* do RCM 'dumb'. You can
also do RCM 'smart.' The trick is to know how to do RCM 'smart',
and how to avoid doing it 'dumb.'"
Here's an example of what I'm talking about. Last summer, I
attended a conference workshop at which a consultant said that
his RCM reviews (using his rigorous approach, in groups) typically
ran for 6 or 7 weeks of all-day meetings.
My jaw dropped. This would be 30 to 35 8-hour meetings -- 240 to
280 group-hours. I am accustomed to reviews that run for 5 to 15
meetings -- *half*-day meetings: 20 to 60 group-hours. While a
newly-trained group of people might take a little longer than usual, I
would be very seriously concerned if they were not finished after 20
half-day meetings. In my professional opinion, if they take that
long using the process I use, they would be doing something very
(BTW, needing fewer grou-hours to complete a review permits me
to run meetings on alternate days, rather than every day, while still
getting results reasonably quickly. That, in turn, can make group
members available for their normal work for anywhere from 70% to
80% of the work week.)
Of course, few folks do a review of an entire industrial facility. Most
folks subdivide things in some way. If you pick off a big bite, it can
take you a long time to chew it. Did this consultant routinely pick
off bigger bites than I do?
In the course of his workshop, he offered the results of two reviews
that he called "typical". The scope of his reviews looked almost
exactly the same as the scope of my reviews. So I don't think that
OK, maybe I didn't understand his scope exactly right. But even if
I'm wrong by 100% -- even if the scope I usually use is *half* the
scope he usually uses -- that doesn't account for his group-labor
being anywhere from 4 to 14 times larger than mine!
IMHO, he was doing RCM "dumb", and making it enormously
costly. His client was willing to use this rigorous approach *when*
and *if* he saw an enormous benefit to it ... but in most cases, his
client preferred to use a less rigorous process that the consultant
also offered in his "Fuller Brush kit" of approaches. No surprise
there -- given these alternatives, who wouldn't?
And this, I think, is how the "adjectival-RCM" processes arose.
Based on various accounts I've seen by the people who advocate
them, I think that people tried to use RCM -- they really did -- but
with the best of intentions they were unwittingly doing RCM
"dumb". And then, "having tried RCM", they gave up on rigorous
RCM and tried to "streamline" it or "turbo-ize" it or otherwise speed
it up -- honestly believing that they had retained RCM's essential
features, but not knowing how to do RCM in a way that was both
rigorous (JA1011-compliant) *and* "smart".
How is it that people might do RCM "dumb"? There are a number
of issues involved.
(And in this discussion BTW I'm deliberately going to mix
references to a bunch of different approaches I've seen. I am not
trying to single anybody out. I'm also deliberately not naming
names. In this discussion, IMHO, *who* is doing things is less
important than *what* is being done. I'll only say that I am
describing real approaches, ones that I've either used in the past
myself, or that I have seen other people advocate either orally or in
writing. Finally, I will cheerfully acknowledge up-front that I'm going
to be presenting my own opinions on things, opinions that formed
the practices I follow today. You can bet that other folks will have
other opinions and will follow other practices. *shrug* No surprises
there, either, right?)
One way to do RCM "smart", IMHO, is to have a process that is
presented clearly enough to be followed easily; this prevents
confusion and mistakes during the analysis. No surprises here --
but it's harder than it sounds.
Another is to have training that equips all participants with the
knowledge they need -- an investment that obviously reaps
dividends later. Again, no surprises here, and again it's harder than
Another is to organize the people who do RCM in the most efficient
and productive manner. Here you will find considerable differences
among the consultants who would advise you.
Some will bring a consultant in, who will interview your people and
produce a report. Personally, I don't favor that -- I think it's better to
for you (the client) to bring your people in and let them talk things
out face-to-face. Maintenance is a big elephant -- IMHO, let all the
blind men thrash it out among themselves (under the guidance of
an expert in the RCM process). Not only will they resolve the
disagreements better and faster, but you will have an honest shot
at getting them all to agree on the way the disagreements got
resolved. (And we all know how hard it can be to get two prima
donnas to agree on something, esp when they never meet face-to-
Again, some consultants will use groups, but will provide an
"expert" whose expertise consists solely of having observed a
number of meetings beforehand. Again, personally, I think that the
"expert" needs to be an *expert* -- someone with dedicated
specialist training in RCM. Once again, this investment reaps
dividends later (IMHO). But, again, it's harder than it sounds.
Some think that using software speeds things up. Personally, I
think that groups get bored when they sit in a room watching one
person push buttons; so relying on software to *execute* the
process can end up wiping out the advantage gained by working in
groups. (Software can and often should record the group's
information and decisions, but IMHO this is something that should
usually be done by the expert after the meeting, when the group
members have gone back to their "real" lives.)
And so forth. This is by no means an exhaustive list, but it gives
you a sense of the issues. Lots of issues, lots of aspects -- and
lots of opinions.
If you're wondering which opinion to follow, well, shop the way you
would shop for any other service. Get quotes from a range of firms
(timetable as well as price), get references, check the references,
and decide for yourself which firms has shown you that they can
meet your business needs (in terms of elapsed time and employee
labor diverted to the analysis, as well in terms of the direct cost of
consultant fees and expenses).
There has been a lot of claim laid to "RCM". There was a time
where we were compelled to call it something else because of public
perception (PMO, SRCM, etc.) Education on what it is from a
historical and wide perspective is recommended.
Yes, as we've seen, RCM done "dumb" gave it such a black eye in
some spheres that some consultants decided to back off from the
name itself. Of course, they were also backing off from the RCM
process, so in a way they were only being honest about what they
The sad thing is that in justifying their decision to turn away from
RCM, they made broad statements about how RCM "is" expensive
or resource-intensive. In other words, rather than retaining the
technical steps of the process (what's in SAE JA1011) and looking
for ways to deliver those steps more efficiently, they threw up their
hands and gave up on those steps altogether. *sigh*
Some also gave up on the notion of tackling the issue in an
organized fashion. "Break it down into its elements, and pick and
choose the elements you want." Good approach for an expert,
perhaps, but how many working people are expert enough to do
this sort of scramble-and-sort? If they ask for expert assistance,
how can they assure themselves that their expert is doing anything
more than throwing darts at a dartboard?
What we hope to do, with SAE JA1011, is to reclaim the name
"RCM" so that it *can* be used consistently by these working
people -- by the real people in real industrial plants whose jobs
(and, sometimes, whose survival) depend on getting that evaluation
The key attributes, and an adherence to them, are what is
important, not a name, or a book, or how much you have to spend.
Dana is correct that there is something in a name, but
unfortunately "RCM" has had too many renditions and
counter-renditions. Understanding the principles will allow you to
subjectively evaluate the plethera of alternatives and find what is
correct for you, call it what you wish..
*Why* are names important? Names are important because they
help ordinary people talk about important things in an organized
fashion without having to be a professor at some university.
Once we know what things we are talking about, we are then much
better-equipped to decide which things we want to use.
Which is what I see people on this list struggling with, day after
From John Moubray
I am due to present the article in question as a paper ("The Case Against
Streamlined RCM") at the ICOMS conference in Melbourne in June. It is also
due to be published in other magazines and presented at one or two other
conferences elsewhere later in the year. As a result, I believe that I have
an obligation to the conference organisers and magazine publishers not to
make it freely available on websites (my own and others) until after the
relevant events/publication dates. It was however published in "Maintenance
Technology" magazine in the USA in January, and I am sure you and anyone
else who is interested could get a copy from them. In doing so, I would be
most grateful if respondents could respect the copyrights in the article
until other conference organisers/magazine publishers have had the
opportunity to derive the benefit from providing forums for us all to air
our views on topics like this. (In other words, by all means please get hold
of a copy for your own use, but please refrain from disseminating it further
until (say) the end of October this year.)
From Dana Netherton
Steve et al.,
As a further followup, MT has placed John's article on its website,
at this URL:
So, as far as MT is concerned, they are (presumably) happy if
people to view it there.
Links to my own pair of MT articles about SAE JA1011 (June and
July/Aug 1999) on MT's web site may be found at www.mt-online.com/articles/06-99mm.cfm, and www.mt-online.com/articles/08-99mm.cfm
Notwithstanding any of this, John's warnings about copyright stand
WRT quoting the article too extensively in e-mails or other
correspondance. Provide the URL, but don't provide more text than
is needed in the course of "fair comment". (That's the U.S. rule-of-
thumb, at least. I haven't a clue regarding the rules around the
British Commonwealth, of course.)
From Shannon Hood
Each discussion I have heard of criticaility analysis seems to talk around
production loss. In this instance its not surprising it seems fruitless.
We need to remember we're supporting a business, not just production so that
means a criticality reveiw should consider (amongst other things)
When these things are considered, there's some surprising criticality
- Liklihood of production loss
- Result of production loss
- Safety issues (fire equipment)
- Environmental impacts
- Effects on product qualkity
- Costs to repair
- Recovery rate
- Machine complexity
- Inherent relability.
I don't want to analyse us to death, but if we should consider these things.
The reality is that RCM costs money and so does bad maintenace. So there's
a happy medium. If we suggest RCM should be done on everything then we are
inherently suggesting that the cost of doing the analysis on all machines is
ALWAYS less than the benefits we attain. Well, I have seen one site spend
$1800 in consulting fees and four trade mandays defining the optimum
strategy for an $800 pump. So for the cost of the analysis I could have
added two pumps to the store, paid for 32hours of trade checks (thats over 2
hours per month) and still have had change left over.
My suggestion is that the cost benmefit for RCM is a lot more likely for
your really critical machines and a lot less likely for your non-critical
machines - this is not rocket science. So start at the top and work your
way down til you think (or have done some really complicated analysis to
prove) its not worth it anymore.
Is criticality pre-analysis analysis? well yes. But so is a visit to a GP
before a specialist. An alternative approach would be to feel unwell and
visit every possible medical specialist and have every imaginable test, scan
and boipsy in the pursuit of avoiding pre-analysis analysis. Thorough, but
probably not sensible!?
From Mike Fitch
Sandy and list,
This is so you will know how the RCM
discussion is affecting at least 1 non consultant. As far as names, I have
to come down on Dana's side. I know very little about true RCM, PMO & such
(although I'm learning rapidly through this list), but I do know that when
someone refers to something, anything, I need to know exactly what there
referring to. In the U.S. we play a game called football. It's the best game
in the world (now I've gone & started another controversy), but if you go to
South America & play football, your just chasing a little kickball around
with no rhyme or reason that's intelligible to me. I could go on & on with
examples of misleading names. If the ISO has done anything to clear the
muddy water in the TLA flood, bully for them & goody for me. Now I can
examine RCM, it's successes, it's failures ( not some competing methodology
wrongly so named) & compare it with PMO (by the way, what exactly is PMO &
if 10 different people tell me 10 different things who should I believe?) &
it's successes & failures. From the comparisons maybe (& I do mean maybe)
our corporation can decide where to invest our time, money, & effort.
What I'm saying is forgetaboutit when it comes to debating
the names, accept the standard, name your competing TLA as you wish so the
difference is distinguishable, & let guys like me make as intelligent as
possible decisions on the possibilities!!!
Mike L. Fitch
From Tobias Kuners
New at the scene [all-the-way across in Europe], just some quick thoughts on
I cannot fully agree with Dana, and also want to add that there is more to
view then only production loss.
RCM does give answers; it helps understanding what equipment needs
preventive maintenance, and what does not (=RbM; Risk based Maintenance).
All based on independent data instead of opinions and 'feelings' of
mechanics & staff. Of course, you yourself have to set the limits, but once
that is done, the system fills itself. We developed a matrix in which we
could classify equipment based on the # of hours downtime to fix that part.
Knowing the nature of our business, we had set ourselves some goals. We also
identified a risk of failure. Both parameters 'automatically' bring an
equipment part somewhere in the grid of the RCM matrix. With Management we
sat together and agreed acceptable and unacceptable squares in the matrix
and with that data we designed our Preventive and Predictive maintenance
routines such that the equipment parts were pushed into less severe squares
of the matrix. In order to move forward we agreed to review several dozen
parts per week for a couple of months and at the same time used downtime
pareto's to tackled the most severe causes immediately when the occurred.
The latter process also gave us goodwill in the company; people recognised
that we were changing things for the better.
Besides production loss, and depending on the process and industry you're
in, you may want to consider spills to the environment, process safety,
quality related losses, all of which relate to equipment downtime to
repair/fix as input for the RCM matrix. To cover multiple criticality we
agreed that more then one criticality (i.e. safety and process) adds one
square to the risk of failure. More then one would be excessive with respect
to the Preventive Maintenance that would be required to bring that piece
back into an acceptable square in the matrix.
From Kim Matulich
Maintenance management is always evolving and through this evolution we see
many systems come and go. I challenge anyone to state there maintenance
management system is best as we all know this is just opinion. It is up to
all of us to make our own intelligent decisions on what suits our needs.
Mike if you are considering 'buying' a maintenance strategy first ask
yourselves a few questions
1. How much money can we spend on it?
2. How long have we got to get it working?
3. Is it going to make a defence?
4. Can we do it ourselves with our own people?
I'm not trying to down play the importance of any intelligent maintenance
strategy techniques and I am well aware that plant reliability, safety,
environment, etc all play a major part in the process and I'm not saying
don't use expert systems - just remember successes and failures don't
normally come from the systems, they come from the people and more
importantly the attitudes of the people. Good luck with your selection.
From Dana Netherton
Oh yes, when you use the RCM process you *do* get answers --
but RCM itself does not hand you answers on a silver platter. (It
does not say, "this asset is critical", without your involvement. In
fact, RCM itself doesn't use the word "critical". It simply gets on
with the task of finding out what you need to do with your assets.)
You have to apply its questions to your own information in order to
come up with those very useful answers. :-)
As for production, well, Richard Ellis's question of course was
about production -- but you're absolutely right, there *is* much
more to view than only lost production.
From Michael Doolan
Here's a question I've been dying to ask for quite a while now ......... essentially its addressed to the upper echelon of our maintenance guru's .....
here goes ..................
The aviation industry Globally seems from all media I've seen over the years, boasts the most regulated maintenance of any industry whether its mechanical , electrical or electronic. It seems to work extremely effectively considering the number of planes and hours spent flying etc. . ( of course there are a few that for unknown reasons fall out of the sky )
The aviation industry has its act together why cant the rest of our respective industries do like wise??
Why is general (or any) industry not regulated in a similar fashion?? i.e.: a Standard Global RCM. (to be customized slightly to each relevant industry)
All work we do in its simplistic form IS essentially the same.
For mechanical it is either rotating ,reciprocating or static or some variation of these forms.
Is it just because the CMMS programmers and companies refuse to come to a standard agreement so that the profit margins remain high or is it deeper, in that the training institutions globally (universities) can't/won't come to an agreement for some inane philosophical refusal to agree on anything apart from their own teaching curriculum.
I know there exists .... RCM, RCM2 and SAE JA1011, (not to mention the thousands of home grown versions around) Which, is the globally accepted minimum standard .............?
Don't get me wrong this post is not meant to inflame the Consultants or Programmers or Uni boffins on this mailing list, but rather to try to get a definitive response to the subject. I expect there will be quite a few differing opinions.
From Brandon Chennaux
As a former aircraft maintainer, (specializing in jet engines) the answer
is simple. Failure of a aircraft system in flight usually has catastrophic
effects, hence the high maintenance standards. We looked intently at all
performance indications and testing engines repeatedly to ensure optimum
performance. The cost otherwise would be too high.
For the rest of industry to take this approach, would be too costly, and the
benefits unrealized. If you break down in a factory, rarely do all workers
suffer anything other than an outage. Therefore managing only the critical
items, identified as critical due to cost, or importance in the plant or
process is the "cost effective" way to do business and manage assets.
From Terrence O'Hanlon
I understand the difference in consequences in aircraft vs. industrial plant
failures however is it impossible to standardize approaches without going
overboard. In other words, an optimized approach that balances safety, cost
avoidance and reliability that is appropriate for all (most) industrial
What are the barriers?
From John Moubray
The principles of true RCM can be applied to any organised system of human endeavour that you care to name, not only to physical assets of any kind in any industry. The question is what are the principles of true RCM?
These principle have been summarised in SAE JA1011. This standard does not attempt to spell out a specific RCM process - it merely lists the features that any RCM process must incorporate in order to be called "true" RCM. Two versions of RCM that comply with this standard are RCM2 and the Royal Navy's NES45. Navair's NAVAIR 00-25-403 guide is also being modified to comply with the standard (bearing in mind that these same Navair people played a major part in writing JA 1011).
Note that the section of Navair that uses 403 is applying it to aircraft used by the US Naval Air Command. This means that the principles embodied in JA1011 can indeed be applied to aircraft.
With regard to commercial aviation (scheduled airlines): the methodology that they use to develop prior-to-service maintenance programs is called MSG3 (for Maintenance Steering Group version 3). This document was first published under the auspices of the US Air Transport Association (ATA) in 1980, shortly after the Nowlan and Heap report. It was revised twice, in 1988 and 1993. In 1999, ATA convened a working group (the MSG WG) to carry out a substantial further revision intended to accommodate issues that have arisen since 1993. This working group includes representatives of the major airlines, aircraft manufacturers and aviation regulatory bodies (the FAA, Transport Canada and the European JAA). I am also an active member of this working group.
The revisions to MSG3 are due to be published in two or three stages, the first of which is due to be published in March this year (to be known as MSG3.2001).
Note that although MSG3 was originally based closely on Nowlan and Heap's report, it contains a number of crucial differences from RCM as defined in JA1011. (Anyone who wishes to know what these differences are is at liberty to get hold of a copy of MSG3 Revision 2 and JA1011, and work it out for themselves. It may also be wise to wait for MSG3.2001 and then MSG3.2002 before coming to any final conclusions.) However, MSG3 does not claim to be "RCM". In fact, the terms Reliability-centred Maintenance and RCM do not appear anywhere in the MSG3 document. One of the issues currently under discussion within the MSG WG is whether MSG3 needs to brought fully into line with JA 1011, and if so, what further modifications would need to be made. (The two documents are beginning to converge, and I suspect that they will converge fully in the fullness of time, but it may take a few years.)
Without going into a great deal more detail, Michael, the short responses to your comments are as follows:
- there is a global minimum RCM standard: it is SAE JA1011
- JA 1011 allows for differences in the detailed interpretation of RCM, as long as whatever is being called "RCM" incorporates certain basic principles.
It would be nice if everyone in the world of physical asset management sang from the same hymn sheet when it came to the basic principles of physical asset management strategy formulation. In my opinion, there is no reason why we shouldn't. However, bear in mind that none of the processes that are being discussed on this website even existed 50 years ago, so we are trying to move from a state of complete anarchy to one of (complete?) harmony. I believe that humanity being what it is, it is probably going to take another fifty years before we reach this state. This is way beyond the end of the working lives of all of us presently engaged in this correspondence, so all that each one of us can do is continue, in good faith, to try to understand and to propagate what we genuinely believe to be "best practice", and in so doing make the world a better and safer place for all who live on it.
From Raul Pereira da Costa
Thank you for the very good, good and not so good messages cross this web.
It has been very open (for some too open) discussion.
But please keep in mind that one size does not fit all.
Not only when you choose a pair of boots or when you try to choose a
strategy, a tool, a consultant, one software package, a book, ...
There is more than one valid approach to keep up our production facilities
in production with good (may be the lowest...) possible cost.
And try convince others that someone has the "silver bullet" for all
situations it's so unrealistic, as in the cowboy's movies in my young days.
Raul Pereira da Costa
From Walt Sanford
I agree whole heartedly with John on this one. Understanding what must be present to constitute RCM is what is important. It will allow you to properly evaluate what is out there, if RCM is what you want and need. However, processes available which do not meet the "RCM" criteria also may add value, as they may be focused on a different set of objectives, they just should not be called "RCM". Just be cognizant of what you want versus what you will get (and what you should not compromise). You will find out what works most efficiently for you, within your set of criteria.
From Brandon Chennaux
Terrance, IMHO, to standardize an approach would work only in a standardized
environment. The focus I am sure is to get the most out of what ever process
the widget is performing, however not all factories fit the same profile, or
run the same process to get produce the end product. Therefore lack of
flexibility which a strict "standardized approach" promotes can be a
barrier. I don't take the position as an expert, I am trying to learn from
some. My maintenance background taught me that nothing was set in stone, and
"everything" changes, from hardware to software to processes.
Standardization is good as long as it is flexible enough to change.. I know
From Jose Duran
I full agree with you.
I almost have stopped my participation (now reading when it is possible)
because of many of the comments are taken like personal stuff and people
start to be aggressive or defensive. The level of the discussion are many
times so low (offending or treating with lack of respect people and/or
organizations) because of the trend to "convince others" that we have the
magic spell to solve all the problems (call RCM, RCM2, RCA, RBMX3-5, a book,
an article, etc) and we forget that our companies are due to produce and to
implement all the well know tools is not the real mission.
From Scott Chapman
as a newish engineer who has only been playing maintenance engineer for a
couple of years i do not have years of experience to fall back on but i'll
offer my thoughts for what they're worth.
as i see it, it is good that all other industries are not regulated. if we
are regulated and all have to do things exactly the same, where is the
competitive advantage for my company? if i can save my factory several
thousand dollars by using a different method or idea, then our products
become cheaper than our competitors (or more likely we keep the price the
same and take the extra profit) and this means the boss likes me and i get
to come to work the next day to find more ways for him to get his bonus at
the end of the year.
From: Walt Sanford
I like the way this guy thinks. There is always a better way. Regulation
provide limits to the consequence of change, not micro-manage your business.
By the way, a standard is not a regulation, until someone tells you you have
comply. It is a means to define an approach.
From Ron Doucet
Keep in mind that with evolution, progress and the survival of the fittest also
What constitutes progress in ones person's eyes, with all of the best
intentions, could actually be detrimental and result in an unsafe process.
Although a textbook does an RCM process make, now we have a standard that allows
for evolution yet not deterioration.
You car example is great, but we have firmly defined what a car is and then we
let progress take over. As we evolved from the car to the jet engine we
actually gave then different names.
Now the criteria for RCM have been defines (SAE JA1011) now let progress
From Ron Doucet
Of course there is a difference between a "maintenance strategy" and a
"maintenance strategy formulation tool" such as RCM.
There are many viable maintenance strategies, and one size, by the nature of
there being an unlimited amount of operating contexts etc, does not fit all.
On the other hand, as for a maintenance strategy formulation tool, RCM II is
operating context independent. That is it will help you define the most
appropriate maintenance for your asset in your operating context. Therefore as a
maintenance strategy formulation tool, RCM II, being one size, can fit all.
Of course by this I mean that there are no assets nor processes for which RCM II
can not be applied successfully. I do not mean that it should be used to
develop a maintenance program for everything regardless of its importance or
consequence of failure.
From Kim Matulich
I agree RCMII is a good strategy formulation tool; however, I also believe
TDBU is a good tool and what I have seen of PMO I believe it also a good
tool. Mike was indicating that, as an organisation, they were considering
investing in a tool - let them as an independent organisation look at as
many as possible without being influenced by just one type of process.
These other tools will also assist in identifying the most appropriate
activity - actually most human brains (with the right input) can make that
From Scott Chapman
as i mentioned in an earlier post i am a newish engineer and i have been
reading the posts on this list for several months. now at the risk of
opening up a can of worms, can sombody please explain (as breifly as
posible) the philosophy behind the maintenance strategies such as RCM, TPM,
PMO etc.etc. i sincerely do not want this to turn into a who is better than
who debate as that serves no one. and to those of you who have implemented a
TLA strategy in your factory, mine etc, how did you do it? was it a gradual
process over time or did you go in boots and all. i am curious and i await
regards to all
From Ron Doucet
Scott, I hope this, as you say does not bring out all the consultants " guns a
blazin" trying to sell tools.
Here goes, short and sweet: (well it was meant to be)
RCM, reliability centred maintenance is a process used to determine what must be
done to ensure that any physical asset continues to do what its users want them
to do in their present operating context.. In other words RCM will determine
everything that is required including mtce tasks, operator tasks, training
requirements, SOP's, suggested and mandatory redesigns. THis is applicable to
new or old assets in any operating context. If you have been on this site for a
while you will know that I am referring to true RCM and specifically my
experience is with RCM II. We did a pilot, the concept was sound, then went in
head first and have literally saved millions. (could have saved more but the
PMO, stands for preventive maintenance optimisation, although my comments will
generate comments from Steve and Sandy, I am only commenting on the classical
approach to PM optimisation and not specifically the product called PMO 2000 or
any other product out there.
PMO in its simplest form optimizes your current PM tasks. IT is not a zeor based
approach. It takes your current PM tasks and runs them through some sort of
decision diagram that can validate whether the task you are doing is technically
feasible(can be done) and well written. The output of a PMO is clearer tasks,
tasks being more specific as to the failure mode being addressed and
corresponding remedial actions as opposed to the classical "Check Motor".
Another output of the PMO process should be the elimination of task that are
proved not technically feasible or are unsafe.
You have to be careful with PMO because the baseline, the old PM, is an
unvalidated baseline and was probably only developed, with best intentions,
through gut feel.
A PMO process, in my opinion (here goes) should not generate new mtce tasks if
it uses the old ones as the basis. It should only improve on them. We have a
formal PM optimisation process here that we use, it is very quick, approximately
2 minutes per task reviewed, but as I did not find what I was looking for
outside (as Dana said, he wanted to be able to sleep at night) I had to develop
it myself. We chose not to mix up PMO and RCM. One reviews tasks, the other
determines the right tasks and never the twain shall meet.
The benefits you will get from PMO are cleaner safer tasks and usually less of
them. In the end it does not mean you will have the right tasks. The benefits
that we have realised are useful, we have decreased tasks (people were probably
not doing them anyway) and have eliminated tasks that could have killed
people(this was a big bonus) and all of our task are much more precise and
cleaner and we hope to get reliability improvements out of this. After the
pilot, we have decided to review all of our PM and at the same time perform RCM
on our assets giving us the most trouble. ROI wise it is not comparable to RCM
but well worth doing. As you can not RCM the world in a day it is a good thing
to do in parallel with RCM. One warning though, make sure your process is
defensible as the base line you are using, the old PM, is probably not.
TPM, I read the book by Seichi Nakajima, TMP involves unconditional cleaning and
promotes operator involvement. In my observations there are a few things flawed
with TPM but I wont give you the whole list.
First, the cleaning is unconditional, you should only clean if there is a
Second, all failures conform to the bathtub curve. This is about 40 to 50 years
out of date. (read RCM II by John Moubray to understand the 6 failure curves)
Third, it is about operator involvement with no actual process for determining
the required tasks. It was invented in a place with no trades demarcation and is
not that easily applicable in places with trade demarcations.
TPM is pushed by the JIPM (Japanese Institute of Preventive Maintenance) who I
believe said "If you use TPM you will need RCM, if you use RCM you will not need
This is because RCM is all about people involvement both cross trades and with
operations and with a rigourous process backing it up, determines the right
tasks for everyone. Do what you have to do, have the right person do it and make
the decision while all working together.
Hope this helps
From Mike Atkin
I think you owe Ron an ice cream
From Larry Johnson
As the creator of the PM Optimization process I think that I am best
qualified to answer your question in this matter. I also included an
explanation of TPM and RBM at the end of this section.
A little background first:
I initially used PM Optimization in 1992 to help a nuclear power plant that
had suffered a series of unplanned shutdowns respond to the Nuclear
Regulatory Commissions' (NRC) concern that their PM program lacked any
technical basis. The power plant had, over many years, continuously added
PMs to their program in response to one-time failures (knee-jerk tasks),
Root Cause Failure Analysis, and industry technical notices. They knew that
most of the activities they were doing were effective; they just couldn't
say which ones.
I knew that RCM would be overkill for their needs - they didn't need to
create a new PM program from scratch, which is what RCM does, they just
needed to optimize the one they had. Drawing from my past experience with
RCM (3 years with Electrical Power Research Institute and other earlier
projects), I devised a process that employed many of the RCM methods to
validate and optimize their PM activities. It took under a year to complete
the project, the NRC accepted the results, and PM Optimization was born.
There seems to be a lot of misinformation being spread about PM
Optimization - mainly by consulting firms who stand to gain from such
action. I'll address a few of the hobgoblins right now.
1. PM Optimization is dangerous and leaves you vulnerable to lawsuits.
Nonsense. The longest running 'pure' PM Optimization program is now in its
eight year at a US nuclear power plant. Since its inception there has been:
a.) No PM preventable failures, b.) A 12% reduction in maintenance staffing,
c.) A 35% reduction in PM activities with no increase in corrective
maintenance, d.) Recognition by the Institute of Nuclear Power Operations
during their last seven evaluations as a strength, and e.) Recognition by
the NRC during two separate readiness inspections as a strength.
If you included all facilities actively using PM Optimization there would be
well over a 100 years of operating time between them. Not a single facility
has had a failure attributed to, or because of, the PM Optimization program.
2. PM Optimization doesn't allow you to add new PMs.
Yes, of course it does. In fact, we always end up adding new PMs to the
program as we go through analysis. PM Optimization also adds new equipment
to the PM program when an omission is noted.
3. PM Optimization misses potential failures that only RCM can find.
False - and true. PM Optimization effectively addresses equipment failure
within the existing PM Program but it obviously does nothing for the
equipment outside it. This seems to be the biggest misconception - that RCM
and PM Optimization are mutually exclusive. Nothing is further from the
truth. In fact, we often start with PM Optimization, establish an optimized
PM program, and the roll over to RCM for the remaining equipment in the
facility. Doing so allows us to show a quick return on investment while
management is still focused on us - which makes long term funding and
cooperation more forthcoming than with RCM alone.
Who can use it?
1. Any company that has implemented RCM. PM Optimization is ideally suited
for use in the Living Program after the technical recommendations from RCM
are implemented. RCM is not suited, nor was it ever intended, to be
anything else that a once-through evaluation to initially establish a PM
program. RCM's methodology, especially classical RCM, does not have the
provision to optimize task effectiveness after implementation. Because RCM
is a zero based approach, which means you start from scratch, you have to
"throw the baby out with the bathwater" each time you conduct a new cycle of
2. Any reliability program manager looking for a rapid return on investment.
In the long run RCM will show a return on investment, but there are very few
short-term gains. Let's face it; a company is not going to institute RCM
unless it improves the bottom line. This may be in the form of improved
public opinion, addressed environmental concerns, better worker safety, and
of course production. The longer it takes to show a return, the more likely
the program will not receive the funding it needs to survive.
PM Optimization is an effective strategy for reliability managers to produce
a quick return while securing long-term commitment from management.
3. Any facility with a mature PM program that seeks to establish a
documented basis and eliminate unnecessary PMs.
PM Optimization process:
PM Optimization employs many of the same analysis techniques as RCM.
However, PM Optimization is a more streamlined approach.
1. It starts with the equipment within the existing PM program.
2. All PM activities for the equipment are decomposed into their
constituent tasks (e.g. a Overhaul activity consists of visual inspections,
alignment checks, replace parts, etc.).
3. A Failure Modes and Effects Analysis (FMEA) evaluation is conducted
(Side note: FMEA has been around for over thirty years - long before RCM) as
a.) Group identical PM tasks performed on similar equipment. Similar
equipment has the same equipment type, operating environment, duty cycle,
b.) Identify equipment function(s) in respect to its contribution to system
and plant functions.
c.) Identify the failure mode(s) that the PM task is intended to address.
(e.g. a lubrication task might address 'fails to run' for a motor).
d.) With respect to the equipment function and the failure mode, determine
the failure consequences.
If any failure consequence is undesirable then the PM task is addressing a
Critical failure. Critical failures undergo PM task selection similar to
RCM; the exception is that you are scrutinizing PM task effectiveness, not
trying to construct a PM program from scratch and then comparing it to the
existing PM program.
Non-critical failures undergo a 'sanity check' before dropping the PM tasks
from the program. The sanity check assures that there are not reasons,
other than criticality, why the task should remain in the program. This may
include regulatory commitments, difficulty in obtaining parts, adverse
working conditions, etc.
That's the process in a nutshell. If you want more details then go to
http://www.fractalsoln.com/downloads.html to download the PM Optimization
Total Productive Maintenance
Total productive maintenance (TPM) is a maintenance optimization process
that promotes preventive maintenance set in the framework of equipment
ownership by multi-discipline teams consisting of operators, craft, and
engineers. TPM crosses organizional boundaries and the attempts to break
down the rigorous compartmentalization of work within a plant. The work
team concept determines which equipment gets what type of maintenance, at
what level, and using what technologies. The decision process used is less
formal and uses the model that best fits the specific application and work
group. Similar to the RCM process, this process also emphasizes predictive
maintenance. Some sites are using this concept to some extent now in what
are called "Keep It Running" (KIR), "Fix It Now" (FIN), or "Work It Now"
(WIN) teams. This is a relatively new initiative, and there has been some
limited success with this team concept, but the overall impact on the
maintenance process is still not known.
The reliability-based maintenance (RBM) process is a hybrid of the RCM and
TPM processes. RBM begins with a benchmarking phase in which an assessment
is made of the current maintenance practices, organization, personnel
attitudes, technologies used, work flow and practices, costs, and
performance measures. Using the information from this assessment, an action
plan for transition from the present to the future is developed. At this
stage, RCM is used to determine which equipment will get what level and what
type of maintenance and what technologies will be used in the maintenance.
Like RCM, RBM also emphasizes predictive maintenance instead of or as a
supplement to periodic maintenance. In a parallel effort, the organization,
work division responsibilities between departments, and workflow are
evaluated to determine if there is a need to recast the departmental
responsibilities and work flow process. From this two-pronged review, a set
of recommendations for maintenance, organizational changes, and reassigned
responsibilities are developed.
From Ron Doucet
Well Scott I guess you got exactly what you did not want. Specifically :
"i sincerely do not want this to turn into a who is better than who debate as
that serves no one."
The temptation was too much.I will refrain from commenting for now.
From Ross Kennedy
Thanks for the e-mail.
It was certainly an interesting reply from Ron Doucet. It is a pity he has
only read Nakajima's 1982 book about 1st Generation TPM which was poorly
translated by someone who had trouble understanding the TPM concepts
because they were so new to them at the time. This is why the book is
difficult to understand.
Maybe someone should suggest he read TPM in the Process Industry by Suzuki
which is about 3rd Generation TPM and explains why TPM as a precursor to
RCM is such a wise move for any industry where operators have an impact on
the performance of plant & equipment. Alternatively he could refer to my
paper entitled: "Examining the Processes of RCM and TPM: What do they
ultimately achieve and are the two approaches compatible?( www.plant-maintenance.com/downloads/RCMvsTPM.doc) " which I
presented at an IIR maintenance conference back in 1999. I have attached a
copy in case you want to share it with your colleagues.
From Martin Theobald
Put the lid back on that can, quick!!!!
From Scott Chapman
and to all who helped, here is your virtual icecream fresh from my factory.
enjoy it before it melts
thanks for your help
From Terrence O'Hanlon
Thanks for the detailed response regarding the PM Optimization process!
A couple of questions if you please:
1) You state that if "you include all facilities actively using PM
Optimization there would be
well over a 100 years of operating time between them." Is there a reason
PMO is not more widespread?
2) I thought that RBM was a trademark belonging to a hardware/software
vendor and was exclusively offered by them. Is there a public version of
On another note, I read Ron's post which arrived as I was typing this and I
urge Ron to stay plugged in. His very active posting of the last few days
have been very educational and this list needs more NAC's (not a consultant)
From Larry Johnson
You are welcomed Terrence.
As for your questions:
1) I can only vouch for facilities that my company has dealt with directly.
I do know however that Duke Engineering, SIRF (previously) and other
consulting firms are using PM Optimization on the same scale as RCM, but I
don't have direct contact with their clients. Besides, PM Optimization is
just one method of many that we use. Right now the hot button is Risk-based
2) I was unaware that RBM is trademarked - do you know by whom? It's
almost impossible the trademark a commonly used terms such as RBM, PMO, RCM,
From Terrence O'Hanlon
I appreciate your reply about the number of facilities using the PM
Optimization process. I have seen a great deal of reference to it lately.
RBM and Reliability Based Maintenance trademark information can found at
Computational Systems Inc. (CSI) web site here:
It is surprising when seemingly generic terms are granted this type of IP
From Larry Johnson
How about that. I learn something new every day. I did confirm
"Reliability-Based Maintenance" registration via the US Patent and Trademark
Office and it's legitimate.
My apologies to CSI for stepping on their turf.
From Steve Turner
Unfortunately, there is no standard for PMO hence you can't be wrong nor
right. However, when we (underline) do PMO we do not just review the
current PM's. In step 2, failure mode analysis, we do three things:
1. list the failure modes that are currently the subject of PM
2. Add to that list the failure modes that have happened but are not on
the original list, and
3. Review the technical documentation (mostly P&IDs) for failures that do
not appear on either and may have hazardous or serious commercial
consequences. These are mostly to do with protective systems.
Hope this helps to put PMO in the correct frame of reference.
From Ron Doucet
Steve, I understand what you are saying and the failure mode criteria you are
describing seems based on the guidelines given by John's RCM II book of which I
know you are very familiar with.
As you know I was just trying to be general in my explanation and did not want
to list the differences between the multitudes of all the PMO processes out
there. In my quest to be general in my comments I probably should have said the
"general approach to PMO" or the "main goal of PMO is to"....as opposed to the
Sorry for the misunderstanding, you are right, there is no classical approach to
base this on and thanks for saying what your PMO process does without attempting
to sell it.
From Mick Drew
Ron I think you have don a credible job of summarising some of the old world thinking re Maintenance Improvement methodologies, I have added a section to your summary of RCM,PMO, TPM called New Generation tools to give a more up to date view of the latest tools, which thanks to the advent of user friendly software proagrams is available to reliability practitioners, maintenance engineers, supervisors, technicians and operators.
Why new generation? because they use computer simulation features to take the guesswork and subjectiveness out of the maintenance optimisation process. Please note I am a highly paid consultant - but with a practising maintenance background and my business is to teach others the thinking techniques to use and apply these methodologies for themselves. And whilst those on this network thought they were safe from Aussie practitioners, ARMS Reliability Engineers have agreements in place to market the Isograph RAMS software range in the US and a network of instructors to provide skills based training in Reliability methods!
Director Arms Reliability Engineers
The New Generation of Maintenance Improvement Tools
I have included a summary below of the methodology we use. It is much faster to build a model of existing plant and use this to address the improvement effort. Time is not spent going through the resource hungry logic tree analytical process, instead models are preprepared through historical data analysis or in the case of new projects through reliability prediction techniques,and site resources are called in to validate the models and make decisions on new maintenance strategies or changes to existing strategies. We have found this to be the most rapid process and least costly and leaves you with quantified models for further improvement. For example, we have been engaged by an international company to develop a model of a new 1.4 mtpy refinery so the maintenance regime can be optimised against production thruput IN 6 WEEKS!. You just can't do that with any other process and particularly not RCM or PMO.
Also our interface to CMMS Systems is live and kicking!
To determine the optimum maintenance policy for equipment you must compare the cost of unplanned failure with the cost to prevent unplanned failure. The cost of unplanned failure must include all costs to the business not just equipment downtime costs.
In order to arrive at the optimal maintenance policy several maintenance strategies need to be considered and modeled. There are several predictive and preventative maintenance techniques that can be employed and compared to Redesign and RTF options.
Since business environments are dynamic,it is not a once off process, maintenance policies should be continually reviewed to ensure they reflect changes in operational, safety and environmental objectives.
Most organizations have the information they need to optimise their maintenance and continually review their decisions but they lack the tools to structure and process the information.
In most cases organizations have, or are about to invest in Computerised Maintenance Management Systems, with a belief that maintenance will improve in efficiency and effctiveness based on the information they can store and capture in the system. In reality little is done to ensure that the information that is entered into the system is configured in a way that generates easy to use data. Typically little more is done with the information than reporting on work types and cost. The result being that time is spent on entering information that is rarely read, and the CMMS is used generally by engineers and supervisors to generate budget predictions based previous years experiences.
Correctly configured the CMMS, along with the maintenance model, can become the focal points of the continuous maintenance improvement cycle. Through a maintenance modeling process, maintenance policies can be optimised and the resulting strategies downloaded into a CMMS system. This ensures that the optimal predictive, preventative maintenance routines are being followed or RTF where it is the most cost effective option.
Typically remarks are entered against a work order on completion of the job/task. The CMMS should be configured such that the information entered can easily be used to review policies. There are two steps to ensure the data entered is relevant:
Ensure that the user interface for data input is configured to extract the required information, and where applicable failure reporting codes match relevant codes in the maintenance model.
Train the workforce to understand the continuous improvement cycle and how important the knowledge they have and gain is. Relevant skills are root cause analysis and reliability principles.
The information collected by the CMMS should then be used in the maintenance model to continually refine equipment failure models, including frequencies and costs. The maintenance model can then challenge current maintenance strategies to enure that are optimal.
Ideally the process should be automated so that once initial maintenance modelling has been completed and the CMMS is configured correctly, as work orders are completed and information entered into the CMMS this data automatically revises relevane modelling parameters and checks the current maintenance strategy.
The continuous improvement loop uses and builds on CMMS use;
Initial Maintenance Model
The maintenance model provides the basis for the continuous improvement loop using the CMMS as its information source and review. Initial modelling involves developing failure impacts and costs of failure to the business
The model is derived using reliability engineering principles and reflects dominant failure modes and major risk failure modes.
Data for the model can be generated from work order history, generic industry databases, and site specific knowledge and experience.
Once optimised to meet operational, safety and environmental objectives, outputs from the model are;
- Failure predictions
- MTTR predictions
- Maintenance cost predictions
- Costs of failure
- Risk exposure
- Equipment Availabilities
- Maintenance Strategies/Intervals
- Prioritised list of design out items
- Labour predictions
- Spare Part predictions
The CMMS should be customised/configured such that the data entry is easy, relevant and can be linked to the maintenance model parameters.
Work Force Training
To facilitate the continuous improvement loop maintainers and operators need to understand cause and effect relationships and ideally be trained in root cause analysis. An understanding of the process of failure is also important if the work crews are to contribute to the refinement of maintenance plans.
An electronic link between the CMMS, other systems and the maintenance model allows dynamic analysis of maintenance and operational data that can contribute to revised maintenance plans. Data that could contribute to the policy review is;
The dynamic review will report periodically on
- Feedback from PM's, Inspections and Equipment Condition Monitoring
- Breakdown and Corrective Work Orders
- Equipment 'events'
- Product throughput
If changes to policies or intervals are excepted, relevant data in the CMMS will be automatically updated.
- Equipment condition
- Predicted failures
- Revised intervals
- Revised policies
The model will be able to continually report on;
For details see www.reliability.com.au
- Failure predictions
- MTTR predictions
- Maintenance cost predictions
- Costs of failure
- Risk exposure
- Equipment Availabilities
- Maintenance Strategies/Intervals
- Prioritised list of design out items
- Labour predictions
- Spare Part predictions
From Walt Sanford
Also be aware that "PMO" is a term used
not only by Larry's company but by
others as well. Ex: The "PMO" used by
EPRI in the fossil power arena is not as
described here, but is essentially an
RCM approach. PMO is not a trademarked
or licensed acronym. Much of what you
see and hear about "PMO" is likely
several different versions. Another
example of confusion in the marketplace
demonstrating the need for self
education. Don't buy a name, buy the
underlying process that fits.
From Larry Johnson
Walt's double right.
There are a lot of consultants out there that can talk the talk, but when
asked, can't show proof that their 'version' of RCM/PMO has ever been
I believe that ERIN Engineering, who also took the initials and renamed
their own reliability process "Plant Maintenance Optimization", prepared the
report he is referring to. Another EPRI report prepared by Duke Power in
1996 also identifies PMO/PM Optimization. In Duke's case they accurately
characterized the PM Optimization process - because Tom Weir, who managed
the project, was part of my initial project team when PM Optimization was
The one thing to come out of all this is my reversal of the need to
'standardize' the various methodologies. Although I am still suspicious how
SAE reached their standard, I now agree that there is a need for one.
From Ron Doucet
On the topic of data to determine maintenance managment policies may I
the following as food for thought.
As the food I present.........:
THE RESNIKOV CONUNDRUM
MANY BELIEVE IT IS NOT POSSIBLE TO DEVELOP VIABLE MAINTENANCE PROGRAMS
WITHOUT EXTENSIVE DATA ABOUT FAILURES....
BUT IF WE ARE COLLECTING LOTS OF DATA ABOUT FAILURES IT MUST BE BECAUSE
ARE NOT PREVENTING THEM...
SO LARGE QUANTITIES OF DATA MUST BE THE EVIDENCE OF THE FAILURE OF OUR
PREVENTIVE MAINTENANCE PROGRAMS (ESPECIALLY IF THE FAILURES HAVE
SO SUCCESSFUL PREVENTIVE MAINTENANCE MUST BE ABOUT PREVENTING THE
OF THE INFORMATION THAT SOME PEOPLE THINK WE NEED IN ORDER TO DECIDE WHAT
PREVENTIVE MAINTENANCE WE OUGHT TO BE DOING!
As a side note, where I work we have extensive data about unimportant
failures, we use this data to identify the most repetitive problem to
and to measure the results of our efforts at improving reliability.
failure data is an excellent tool for continuous improvement. It can help
identify your biggest alligator (or croc for the Aussies
We on the other hand we do not have a lot of data on failure that have
economic, safety or environmental consequences and nor do we want lots of
on these failure. Preferable we would want no data about serious failures.
Just food for thought
From Graham Oliver
Ron Doucet's message about failure data which he labeled as The Resnikov
Conundrum (hello?) seems, at first glance, to have something going for
it. In a word -- How can you collect failure data if you're doing a
good job of preventing failures.
But there may be two things wrong with it. One is that if you have no
failures (or almost none) you may be over-maintaining, i.e., replacing
parts too often. So you'd be getting no failures but spending tons of
money replacing parts that had loads of useful remaining life. An
optimal replacement policy should give you "some" failures".
The other thing is that the analysis of replacement intervals should
also take into account "suspended" observations. These are historical
records of parts being taken out of service for reasons other than
failure. They even include parts that are still working at the time an
analysis is done.
All of this is consistent with the data required when doing Weibull
Ron, and others are invited to look at RelCode on our website at
PS -- We don't like failures any more than you do. But the alternative
From Mick Drew
The goal is not to generate failure data, it is to understand the potential
for failure and predict or forecast future failures in order to decide the
most appropriate means to either prevent, predict, eliminate of RTF. The New
generation of modeling tools that simulate performance do not rely on
failure data at all. But a failure if or when it is found becomes another
excellent opportunity to improve the failure parameters, as are Suspensions.
From Dana Netherton
Graham Oliver's message about accepting failures in order to get
the data needed to optimize one's replacement polciy seems, at
first glance, to have something going for it. But there may be two
things wrong with it.
One is that the more important the failure, the fewer failures you
can afford to risk. The more people who might be killed by the
failure, the fewer failures you can afford to risk.
Ultimately (and this is the core of Resnikoff's Conundrum), the
failures you care most about are the failures you cannot afford to
have failure data about. Because the only way you can get that
failure data is to have those failures -- and those failures kill people.
The very most important failures kill lots of people.
Correct me if I'm wrong, but I am under the impression that no
responsible person would run that kind of experiment, and kill
those people, simply in order to gather failure data that might
"optimize" a maintenance program.
The other thing is that the analysis of replacement intervals
should also take into account "suspended" observations. These are
historical records of parts being taken out of service for reasons
other than failure. They even include parts that are still working
at the time an analysis is done.
All of this is consistent with the data required when doing
The other thing is that fewer parts have well-defined wearout ages
than people realize, especially parts in modern complex systems
with electronics, hydraulics, pneumatics, and so forth.
Weibull analysis only works when examining an item with a single
failure cause (what RCM calls a "failure mode"). When examining
an item with more than one failure cause, the Weibull curve shifts
toward random failure (beta=1) -- because the item's failures occur
in effect at random (from different causes, each time).
(In practice, the curve often shifts toward infant mortality, because
the intrusive work done to replace something often introduces new
problems that pop up soon after the item re-enters service.)
If you're replacing simple piece parts in your asset, you may be
dealing with items that have only one failure cause.
But if you're replacing assemblies -- say, electronic circuit cards,
or ball bearing assemblies, or air compressors -- then you're
dealing with items that are susceptible to many failure causes.
And then you can no longer speak in terms of wearout, or in terms
of "useful remaining life".
Which means that (in such cases) age-based replacement as a
policy (what RCM calls "scheduled discard") is not technically
feasible -- it doesn't respond to the technical characteristics of the
way that such assets fail -- and therefore cannot be "optimized", no
matter how much failure data you might have.
This news came out of the Reliability Programs that United Airlines
ran in the 1960s and early 1970s. Nowlan & Heap publicized it in
their US government report, published in 1978. *shrug* It's been
around long enough, I should think.
But perhaps it's still news in some circles.
-- Dana Netherton
From Peter Ball
Graham Versus Dana!
Both have an excellent *point of view* which could be worth watching.
Given that Nowlan and Heap RCM and The Resnikoff Conundrum - from
Mathematical Aspects of Reliability-centered Maintenance are both Seminal
Texts with N&H being 'out of print' we should emerge very much wiser.
From Jose Duran
There is an European project called MACRO. It has been dealing with poor data (or lack of data) handling in order to make several decisions:
The approach has been showing good results around the world
- Spares and materials holding
- When to maintenance
- When to inspect
- When to perform a shutdown
- Life Cycle costs
- Project evaluation
You can take a look of those tools in the next web pages:
Lack of data is something that several times produce analysis paralysis, but we need to be capable to make decisions with poor data. It is almost impossible in most cases to have real and hard failure data. Nobody will allow us to run to failure population of turbines in order to "know" failure rates.
CMMS in most of cases are filled with a lot of rubbish, and you can not rely on them, another good reason for not doing that is that almost never the failure report is linked with specific failure modes neither a time reference for specific failure modes.
Discussion ends.....(for the time being :-))
Copyright 1996-2009, The Plant Maintenance Resource Center . All Rights Reserved.
Revised: Thursday, 08-Oct-2015 11:53:49 AEDT