Do somebody have experience with this? I am thinking to apply RCM to the forklift maintenance, to improve the reliability of the equipment. I will appreciate any comments and suggestion.
Thanks in advance

Generally oil analyses (coolant analysis, etc.) is used on large lift trucks. Large reservoirs mean large expense and oil changes are necessary only when necessary and contaminates must be removed when necessary as these are not throw-away vechiles. ON a fleet this can easily become cost effective.

I have extensive RCM experience and truly believe in its value. However, I have also seen RCM applied far too early, resulting in a superficial analysis with too many unknowns. It adds value in either case, but certainly is a lot of effort in the latter.

Before applying RCM, ensure you have done the following:
1. Implemented basic maintenance programs such as cleaning, inspecting, lubricating, adjusting, and tightening.

2. Implemented any programs from vendors' recommendations.

3. Implemented maintenance inspections compliant with any workplace safety laws. An example is NDT of the forks (required in some states and provinces)

Keep very good records of your findings and you will get good data for your RCM analysis.

If you already achieved 1-3, then pick the problem subsystems. You could analyze the entire forklift, but first focus on the subsystems causing the most downtime or quantity of unexpected failures or excessive costs. Like all things there is a point of diminishing returns so put your effort where the immediate results are obtained.

Hullo Canpeng,

Since RCM is a process to determine what maintenance must be done, i.e., its starts with the premise that there is no maintenance program in place, your comment quoted below intrigues me.

quote:

I have also seen RCM applied far too early, resulting in a superficial analysis with too many unknowns. It adds value in either case, but certainly is a lot of effort in the latter.

While I agree with you that in the real world, it is better to get the basics right before moving into RCM application, it is perfectly possible and often desirable to apply RCM at the design stage. I have certainly done so, and it is very effective. After all the first serious application of RCM was on the Jumbo Jet, while it was still on the drawing board.

V.Narayan.

If we do rcm at design stage, how dependent are the results on the rcm doer's experience and knowledge?

Dear Josh,

To do RCM well, at any stage of a Project or operating Plant, the analysts must

1. Know the RCM process properly
2. Know the equipment and its degradation mechanisms and failure modes
3. Have default ailure rates for similar equipment if historical failure data is unavailable
4. Understand the operating context
5. Know the consequences of failure

These requirements apply whether RCM is done at the design or operating stage.

V.Narayan.

Where can we get items 2 & 3? From OREDA report?

Josh,

Item (2) from your experience, from Cvendors and/or Consultants.

For Item (3), there are a number of published sources, e.g., IEEE, OREDA, ESREDA and others. But these generic data may not be useful in your operating context. Talking to operators and maintainers often yeild good results.

V.Narayan.

For item 2, are there no generic degradation mechanisms & failure modes published like for item 3? I know they have to be tailored to site-specific operating context and taken with a pinch of salt.

Dear Josh,

The generic data bases provide failures per million hours for specific failure modes. They dont stste how the item degrades, though you may be able to infer this from the failure mode description. Vendors may provide both degradation info and failure rates.

The operating context is the really important factor, which affects the degradation mechanism AND the failure rate. Generic failure rates have a range covering perhaps two to four decades, and a mean value. Such a wide range means that most users select the mean value, which may not be applicable to their scenario. Your own estimate is often better than such 'foreign' data.

I think that both of us are digressing significantly from the the original question about Fork Lift Trucks, so we should stop here and perhaps start a new thread.

V.Narayan.

Vee, My comment regarding RCM application at far too early a time is directly related to your points #2 and #3 of your October 21 posting,
"2. Know the equipment and its degradation mechanisms and failure modes
3. Have default ailure rates for similar equipment if historical failure data is unavailable"

Without any basic maintenance program or work management system in place, a typical analysis team will produce a maintenance program with their best guesses. Not necessarily a bad thing, but one cannot expect an optimized program. The analysis team I refer to is not a team of design engineers, but a more typical one of operators, trades, and possibly vendor or contractor repair specialists.

You will agree that the best outcome from a typical RCM analysis depends on a solid knowledge base among the analysis team, and they are significantly aided by good equipment history if their knowledge levels are not as deep. That being said, it could be worthwhile to use RCM to define the basic program that will collect useful data for future analyses and optimization. I have taken that approach too.

If the forklift is from an established manufacturer, it should have a maintenance program suitable as a starting point if its application and modifications are not unique. That is why I would start there if nothing exists.

If you start a new thread, I would be interested in discussing how to prevent maintenance professionals from misinterpreting database information. A good example is with respect to mean failure rates for components with random conditional probabilities of failure.

Canpeng,

I would be very interested in discussing that theme if you start that thread.

I would like to take issue with a few things that get continually repeated here. Even thirty years out from the publication of the original report, and almost ten years out from the publication of the RCM Standard, there are still some fundamental misunderstandings I would suggest.

1 - RCM needs data to be initiated

This is both incorrect and I would suggest that it is an unethical form of developing maintenance regimes going into the 21st century.

Most historical analyses of maintenance data to create maintenance regimes fail dramatically. They do so for a range of reasons. (All of which I have spoken to previously)

1. The integrity of the data is not great in most systems around the world. I have now seen the CMMS implementations in a vast number of companies and countries and while the data integrity is improving it is still far from what I would call good.

Apart from data capture and system configuration issues there are a number of other practical reasons for this.

When a craftsman / tradesman is performing a service, anbd thgey find something wrong. What do they do?

If we are honest we will realise that they fix it there and then. They dont bother to raise a work order, get it scheduled or all the rest of it. They do what the do best, get things back to their original resistance to failure.

Similar happens when operations techs find a problem and they have the maintainer nearby.

So there are glaring holes even in the best managed CMMS systems. This is changing, in some cases maintainers do go and record the info etc. But lets not kid ourselves, times have changed. Resources are tight, and all of us are now doing 5 times the work we were doing ten years ago. So time is often the limiting factor here.

2. Many of the failure modes that we need to understand and analyse may not yet have occurred! This is a critical point to understand. If we go along thinking that everything we could possibly need to manage has already occurred then we are deluding ourselves and aiming to deliver an analysis that is at best limited.

This flies in the face of "understand degradation triggers etc" that is often quoted. How can you understand the degradation mechanisms of items that have never failed in the time yuou have had the items?

Furthermore many items, up to 89% from the original report, have random or not age related failure patterns. So the entire degratation mechanisms argument is representative of the time that we are trying to move away from, not towards! (Continually amazes me this one!)

3. When we have serious incidents with either operational, safety or environmental consequences, then they only occur once. Someone will do something to ensure that they don't occur again. Therefore we simply do not have the level of data availablke to us to do a comprehensive data analysis.

4. The often quoted and misinterpretted Resnikov conundrum. WHich states that analysis of failure data requires a large amount of failure information. Failures which could have operational, environmental, safety of economic only consequences for the company and surrounding public. Not a very ethical way to go about improving performance.

As a second point here there is often discussion around when a company is ready to perform RCM. Even to the point of having coupled together a range of indicators to detect poor performance. Itr amazes me that "experts" in the area still do not understand the ethical implications of the Resnikov conundrum.

Waiting for poor performance to occur is an unethical practice and is one that supports the outdated paradigm regarding the need for large amounts of data to initiate a reliability program.

RCM, or any reliability initiative for that matter, does not need data to get started. Many initiatives that are undertaken are done so with the use of the knowledge and engineering judgement of those involved.

For example, the new airbus plane was subjected to 2000 hours of testing only. Most decisions taken here, as with the range of others, were taken using the inherent capacity of the profesionals to judge the needs of assets, their potential failures and the ways in which we can treat these.

Don't get me wrong, if the data is available and has been produced in a reliable and consistent manner (an argument against national databases in some cases) then by all means lets use them. But if not then we need to press ahead regardless. Not wring our hands and wait for things to go wrong.

Regarding when to start RCM I disagree strongly with almost everything that has been written previously on this. I dont think that RCM is "another initiative to be implemented", I think it is the fundamental initiative without which much of the remainder of the Asset Management area is meaningless.

My 2 c worth,

Daryl,

In your note, you mention a few points on which I wish to comment. These are embedded into your note at the appropriate places so they can be read in context.

quote:

Canpeng,
2. Many of the failure modes that we need to understand and analyse may not yet have occurred! This is a critical point to understand. If we go along thinking that everything we could possibly need to manage has already occurred then we are deluding ourselves and aiming to deliver an analysis that is at best limited.

Comment: An Functional FMEA identifies failure modes whether they have occured in the past or not. So the question of deluding ourselves does not arise. When the 747 was built and its maintenance plan was created, no failures had occured, as it was still on the drawing board.

> How can you understand the degradation mechanisms of items that have never failed in the time yuou have had the items?

Comment: Just because a Heat Exchanger has not yet failed, or even built, does not mean we dont understand how it can be fouled or its tubes leak. This applies to all equipment that are not completely 'novel' in design or operation. When Bell created FTAs in the 1960's, their objective was to cover 'novel' items as well..

> Furthermore many items, up to 89% from the original report, have random or not age related failure patterns. So the entire degratation mechanisms argument is representative of the time that we are trying to move away from, not towards! (Continually amazes me this one!)

Comment: As you know very well, 89% was a statistical value of non-age related failures. A statistical distribution of human mortality during the constant conditional probability phase does not mean that the individuals involved in the sample did not have a failure mode, e.g., heart attack, cancer or a road accident. The randomness relates to timing of the event leading to their death. The cause may have been genetic disposition, lifestyle etc, in the first two cases, or say poor hearing/vision in the last case. Such precursors have a clear degradation path. Barring the bird strike variety of failures, in general all failure exhibits degradation. I would request you to provide examples of such failures that do not exhibit degradation mechanisms, not just statements, however firmly stated.

> RCM, or any reliability initiative for that matter, does not need data to get started. Many initiatives that are undertaken are done so with the use of the knowledge and engineering judgement of those involved.

Comment: At last I found something I could agree with you about. Age-exploration is always available to tune the analysis.

In my earlier note, I advised Josh to talk to maintainers and operators to collect data, not rely solely on published sources. But doing RCM without an understanding of degradation and ball-park or better failure rates is not very productive. And doing RCM on everything does not appeal to me; it will waste resources and money without commensurate improvements. In one of your earlier posts you talked about schemes to make Consultants rich. This is surely one of those.

Regards,

V.Narayan.

Vee,

Good to hear from you again. We do have different points of view, and I enjoy the debate to be honest.

Will try to answer a few of your points:

Firstly my comment that you have paraphrased was the criticality regimes are a scam to separate clients and their money. This remains true in the majority of cases apart from a certain few. For the vast majority of RCM sequencing tasks an approach focussing on simple prioritization initiatives would be adequate.

Your comment:

quote:

An Functional FMEA identifies failure modes whether they have occured in the past or not. So the question of deluding ourselves does not arise. When the 747 was built and its maintenance plan was created, no failures had occured, as it was still on the drawing board.

... has confused me. Surely we are in violent agreement here? The base of the discussion was to dispel the myth around the fact that data is a requirement to get RCM going. If so then what is the argument?

Your comment:

quote:

Just because a Heat Exchanger has not yet failed, or even built, does not mean we dont understand how it can be fouled or its tubes leak.

Is correct, but this is not the point I was making. The focus on the degradation argument brings into focus a mindset that argues all failure modes have an age or a life, this is patently untrue as I will go into briefly.

It was also to argue against the paradigm that state we need to already understand the potential failure modes on the items before we can analyse them. (As per your original posting)

This is not true, we will come to the fact during the analysis and drive them ouit rather than waste time and resources trying to work it all out before even attempting the RCM analysis.

quote:

As you know very well, 89% was a statistical value of non-age related failures..... I would request you to provide examples of such failures that do not exhibit degradation mechanisms, not just statements, however firmly stated.

In general most not all failures do have degradation triggers and most but not all do exhibit warning signs that they are failing. Not necesarily due to age which was the original argument but due to a range of other factors.

Vee, when we speak about maintenance within RCM we are not just talking about the grease the bearing, take the reading type of actions. We are talking about all activities that are required to maintain the operation of the asset or asset system within its current operating context.
This could well mean, and often does, operational procedures and activities, asset planning actions, as well as the standard maitnenance activities.

So, a number of failure modes that do not necesarily exhibit degradation are as follows:

• Within waste water treatment a vast number of failure modes will be driven by what comes down the pipes everyday. These may trigger the degradation or warning signs that you speak about, or they may cause the total failure immediately. THis is not the one of "bird strike" issue that you continually try to diminish in importance, but they are part of their operating environment every single hour!
  
  
• Human error in managing processes incorrectly (either deliberately or not)
  
  
• Human error from poor procedures
  
  
• Poor installation, design, materials for construction, construction iteslf, rebuild etc.. the vast number of iutems contributing to the infant mortality areas. Again around 64% as a statistical representation in the NH report. (reducing in later reports showing that the original report was very much a creature of its time)
  
  
• Many if not all electronics failures do not exhibit degradation signs or warning signs. Failure of SCADA cards or PLC cards (Some failure modes) is a strong case for this.
  

Age exploration is not always available to verify anything. Even this sometimes is out of the reach of many companies on a purely economic basis.

Lastly, and I am not going to do a Vee here but am going to continue to participate regardles of whether you agfree or not, I never said that RCM should be applied to everything neither do I support that view.

What I did say, and believe strongly, is that RCM is not just another initiativge that shoudl be implemented, it is the fundamental essence of asset management without which many of the rest of the initiatives are merely wasting the companies time and money. More than happy to go into this in substantial detail from my experience if you would like to.

Canpeng

You made mention that the manufacturer will have a maintenance program set up for equipment, based upon their history. This is true, and often written into the owners/users manual.

Unfortunately, as is found in most cases, if you were to apply all of the recommendations in the manufacturer's user manual, you would have a very expensive program if you have any number of that equipment.

For example, in a class I teach on motor management, following a day of classical motor maintenance (greasing, aligning, testing, etc.), I give the class a homework assignment to develop a maintenance program around a motor-only using the manufacturer's manual. These are often cited as the minimum requirements to maintain the warranty on the motor. The class will normally show up the following morning a little frustrated and upset having realized that there is no way they can apply that much maintenance to a plant full of motors.

The class continues with training in Condition-Based Maintenance, with training in such TOOLS as RCM, Backfit RCM, etc. They are then given homework on developing a program for a complete new pumping system, including distribution, VFD's, motors, pumps, etc. They usually come back with a reasonable program.

The lesson: Do not rely upon the manufacturer to develop the program. It will be excessive and may not apply to the application at hand.

Instead, you can utilize their history in development of the program, then follow later with a continuous improvement step, such as Backfit RCM, in order to evaluate the program's effectiveness and adjust, as necessary.

Howard

Howard,

I spent some time earlier detailing the failures of data focussed, or historical approaches.

How would you overcome these? (To make it easy lets say that you have a CMMS with good data integrity, which is not common)

Howard,
I agree with your posting regarding owner's manuals for motors and other similar equipment. Applying vendor's recommendations to large quantities of equipment can be sub-optimal, excessively expensive, and completely infeasible.

Getting back to the start of the thread, the question was regarding forklifts. MDeleo does not describe his specific goals or the current problems of the forklift, so we are guessing. Anyway, if his company's maintenance actions are solely reactive at this point with minimal preventive maintenance, I would expect immediate benefits by implementing the maintenance program in the vendor's manual. Simply put, start with the basics immediately, and use RCM on problematic subsystems and for general optimization. A reasonable "something" is better than nothing in the short term.

I have seen far too many companies where the maintenance manuals were shelved from day one. If the binders were pulled down and opened, we would have the greatest dust storm in history. I have come across many instances where failures, sometimes catastrophic costing millions of dollars, could have been avoided if the vendor's maintenance program was applied. I am not referencing major rebuilds with excessive part replacements, but basic lubrication, fitting, alignment, inspection for leaks, and so on.

Obviously if the owner feels the program is inappropriate, then they should modify it accordingly. My point is get the basics in place asap. Use RCM and other tools to build a comprehensive program if nothing exists, or validate / optimize an existing one, and focus on key systems and subsystems with the highest risks.

Readers should not misinterpret the statement "The lesson: Do not rely upon the manufacturer to develop the program. It will be excessive and may not apply to the application at hand." I agree to an extent, but two points must be made:

1 - Some manufacturers have very good basic maintenance programs. Most do not. It is a worthwhile effort to inspect the recommendations and consider the applicability, regardless of whether or not a tool like RCM will be applied. After all, a good practitioner using any tool for designing a maintenance program would consider the vendor's recommendations, troubleshooting tips, and so on found in the manual.

2 - A very small number of vendors are offering performance assurance contracts and extended warranties for their equipment, and always with the condition that their recommended maintenance programs are followed. I have seen some of those programs and they are well-designed, and usually customized, for the operating context. One may argue that it is just a money maker for the vendor. In the end, obviously, the purchaser should consider whether or not the program adds any value.

Canpeng:

Agreed.

Howard

Daryl,
Thank you for your detailed post of 23rd. I will address some of your observations, but in a different order from the one you have used. For convenience, I will include your points before adding my comments.
Daryl: The focus on the degradation argument brings into focus a mindset that argues all failure modes have an age or a life, this is patently untrue as I will go into briefly.
and,
Daryl: In general most not all failures do have degradation triggers and most but not all do exhibit warning signs that they are failing. Not necesarily due to age which was the original argument but due to a range of other factors.
Daryl: Many if not all electronics failures do not exhibit degradation signs or warning signs. Failure of SCADA cards or PLC cards (Some failure modes) is a strong case for this.
Comment: Degradation is not always dependant on age or always have to do with the P-F curve. It is simply a process (physical, chemical or biological) that explains why an item is no longer able to do its work properly. Brittle fracture is as much a degradation mechanism as a strainer fouling or catalyst decay; one has a metallurgical cause, the others have a physical/chemical cause. The degradation can take place in years, days or nano-seconds, but there is a process that makes the item lose its design intent. SCADA or PLC cards also degrade, e.g. by moisture or dust. Just because we know how it degrades does not always make us able to prevent the degradation or even ‘catch’ it in time. Certainly I have never correlated degradation with age as a default scenario, so I dont have that 'mind-set' about which you seem concerned.
Daryl: The base of the discussion was to dispel the myth around the fact that data is a requirement to get RCM going. If so then what is the argument? It was also to argue against the paradigm that state we need to already understand the potential failure modes on the items before we can analyse them. (As per your original posting) . This is not true, we will come to the fact during the analysis and drive them out rather than waste time and resources trying to work it all out before even attempting the RCM analysis.

Comment: I told Josh to
1. Know the RCM process properly
2. Know the equipment and its degradation mechanisms and failure modes
3. Have default ailure rates for similar equipment if historical failure data is unavailable
4. Understand the operating context
5. Know the consequences of failure
Let me clarify, if one is needed, that I had not asked to get all this before an RCM, only that he has access to all this i.e., while doing an RCM. In other words if these 5 points were ‘not possible’ don’t even bother to embark on an RCM. At no point had I said he has to get it before doing an RCM. I am sure you are not implying we don’t need data to do an RCM
Daryl: Vee, when we speak about maintenance within RCM we are not just talking about the grease the bearing, take the reading type of actions. We are talking about all activities that are required to maintain the operation of the asset or asset system within its current operating context.
This could well mean, and often does, operational procedures and activities, asset planning actions, as well as the standard maitnenance activities.
Comment: Give me some credit Daryl, you know me well enough so you don’t need to spell out what is maintenance or RCM.
Daryl: So, a number of failure modes that do not necesarily exhibit degradation are as follows:
• Human error in managing processes incorrectly (either deliberately or not)
• Human error from poor procedures
• Poor installation, design, materials for construction, construction iteslf, rebuild etc.. the vast number of iutems contributing to the infant mortality areas. Again around 64% as a statistical representation in the NH report. (reducing in later reports showing that the original report was very much a creature of its time)
Firstly my comment that you have paraphrased was the criticality regimes are a scam to separate clients and their money. This remains true in the majority of cases apart from a certain few. For the vast majority of RCM sequencing tasks an approach focussing on simple prioritization initiatives would be adequate.

Comment: Are you saying these are failure modes to tackle in the RCM process? If we find failures due to these causes, e.g. while doing an RCM or through an RCA, we can design it out, by training, procedure updates etc. I think we start an RCM with the premise that the item is properly designed, built, installed, operated and serviced in line with the vendor’s procedures. We don’t need an RCM process to sort this out, we need a Management Process! I would argue we don’t need to plan for ‘degradation’ related to any of these. That is not to say we will not find it during an RCM study, and the decision chart provides a solution for them. I reiterate therefore that for most of the failure modes we are looking at, we should be able to identify their degradation mechanism. Without that, the question ‘ what task shall we do’ will remain unanswered.

Daryl: Within waste water treatment a vast number of failure modes will be driven by what comes down the pipes everyday. These may trigger the degradation or warning signs that you speak about, or they may cause the total failure immediately. THis is not the one of "bird strike" issue that you continually try to diminish in importance, but they are part of their operating environment every single hour!
Comment: Let us consider some of them:
1. Pieces of wood , metal, plastic sheets, leaves, silt or other solids
2. Dissolved chemical or biological contaminants (including oxygen scavengers) that are not traceable to a source, or occur sporadically
3. Non-biodegradable bits of material, especially if they come sporadically
Item 1 has a clear degradation mode (except for items like silt), so RCM can find a solution. It may be uneconomic to pursue items 2,3 in terms of sources, but BOD, COD upstream of the Plant inlet will help design out some failure modes. But some of these are in the ‘bird-strike’ category. The statistical data on the ‘bird-strike’ frequency and probability density distribution will help design out some of the total failure scenario that you seem to suggest is inevitable.
Regards,

V.Narayan

Vee,

Okay we are getting somewhere on this one. Good to see. You have quoted me out of context a little bit here however.

My comment:

quote:

Surely we are in violent agreement here? The base of the discussion was to dispel the myth around the fact that data is a requirement to get RCM going. If so then what is the argument?

Was in response to your statement:

quote:

An Functional FMEA identifies failure modes whether they have occured in the past or not. So the question of deluding ourselves does not arise. When the 747 was built and its maintenance plan was created, no failures had occured, as it was still on the drawing board.

Thus the confusion as to why you thought this was a valid point to bring into the discussion.

YOur comment:

quote:

Let me clarify, if one is needed, that I had not asked to get all this before an RCM, only that he has access to all this i.e., while doing an RCM. In other words if these 5 points were ‘not possible’ don’t even bother to embark on an RCM. At no point had I said he has to get it before doing an RCM. I am sure you are not implying we don’t need data to do an RCM

Yes I am implying exactly that. Mainly due to the fact, as I have stated elsewhere, that the dat ais generally not available for what we need to analyse. There is a need to be pragmatic and to apply the method with what we have. I disagree with the five points, which you have simultaneously stated that you should have before the analysis and that you shouldnt have before the analysis.

Surely you of all people are familiar with Resnikovs conundrum. Are you implying that waiting until we have the failure data, therefore the failures which could have severe consequences, prior to trying to develop management regimes for asset systems?

These items can be uncovered during the analysis, particularly regarding failure causes and failure modes themselves. (There are many ways and techniques to get to this information Vee)

Your comment:

quote:

Give me some credit Daryl, you know me well enough so you don’t need to spell out what is maintenance or RCM.

Fair statement, I didnt mean to be condescending in any way. However, your later comments show what I was trying to address.

quote:

Are you saying these are failure modes to tackle in the RCM process? If we find failures due to these causes, e.g. while doing an RCM or through an RCA, we can design it out, by training, procedure updates etc. I think we start an RCM with the premise that the item is properly designed, built, installed, operated and serviced in line with the vendor’s procedures.

This is simply not true Vee, the vast majority of failures are due to either poor operations and other human error induced factors. A large number of additional failures are due to operations exceeding, oftentime deliberately, the inherent capacity of the physical asset. Either knowingly due to a range of factors, or unknowingly due to the effects of inter-related systems etc. (One of the common after effects of Root cause analysis in my experience)

SO in one hit you are trying to take the discussion back to "its all about asset failures due to degradation". Which is simply no thte case. As stated, maintenance is about the activities that we do to sustain and support the funcions (say, operations) of the equipment. Not just the on-condition, Preventive stuff but hte whole range.

Yes, these items can be sorted through one of changes, re-designs and other issues. But they first need to be identified through the RCM analysis and then acted upon through the decision diagram that is used. (Diagram 17 of the SAE JA1012 is a good one)

YOur comment:

quote:

We don’t need an RCM process to sort this out, we need a Management Process! I would argue we don’t need to plan for ‘degradation’ related to any of these. That is not to say we will not find it during an RCM study, and the decision chart provides a solution for them. I reiterate therefore that for most of the failure modes we are looking at, we should be able to identify their degradation mechanism. Without that, the question ‘ what task shall we do’ will remai