Rui,

To answer your question of how I performed my analysis:

1) I calculated the probability of a gasket failure prior to a PM, using a Weibull spreadsheet. Like the posts above have noted I also calculated ~20% for all three pumps.
2) Using this information I then calculated the MTTF – 5 cycles.
3) Then I used your spreadsheet to enter all three pumps each using two gaskets and the associated MTTF for each pump which resulted in my answer.

All in all it took me about 20 to 30 minutes of my time. I could probably do this in half the time if I became more practiced.

As Phillip pointed out there is a high failure rate with the pumps. From my perspective I would concern myself if this was one of my higher cost bad actors of all my maintenance strategies and I felt there were significant consequences at risk of occurring. As in all cases each individual will have different expectations of when to act and what is acceptable at their site.

As a side note: where I work I often have people ask me about improving the reliability of a particular "bad actor" – one that is an issue for the person that is asking. I find the culture is over looking the more substantial one off failures that no one saw coming (no historical failures), but cost us 20% of our maintenance budget and 3% of our throughput. In the next year or four something unexpected will fail again costing 10 to 30% of our maintenance budget and up to 5% of our throughput. There certainly are a handful of bad actors that cost us money all the time but they are not even close to this magnitude of total cost. I can only speak of where I work. It is my goal to shift our culture to not accept these unforeseen issues and begin applying a proactive maintenance strategy. I have been at this site for 6 months and so far we have prevented an issue that could have been 5-10% of our maintenance budget and 0.5% of our throughput. As for our Bad Actors we are good at preventing the larger consequences and we are good at repairing them. In time I hope we can address them too.

Jaz,

I am not sure that I can give you any input on identifying your unforeseen issues except to say have you investigated the PMO approach put forward by OMCS?

The reason I am writing is the term 'bad actor' being used in an engineering environment.

I am curious about the origins of the expression - what does it mean, where is it from and so forth. Can you shed any light on this?

Cheers

Bad actors are frequent events which are chronic in nature but once off events like fires or explosions are infrequent events which can be acute when they occur.

Which one costs more? I heard bad actors cost more like pump failures and the cost of acute events can be amortized.

P/s better start a new thread.

Phill,

Good questions - I imagine there are many different idioms used by different locations throughout the world. Josh does sum it up nicely. It is a term I have learned from the oil and gas industry in North America.

To go further with my thoughts and the point you made in a previous post:
If failures are expected at every 15 weeks I would label pump A as a bad actor. And considering I have taken the time to calculate the spare parts I would consider the consequence of the pump's failure to be "high" and from a personal perspective I would also like to think the parts are of a relatively higher cost (as I would let my accuracy drop off with diminishing returns on my invested time). So therefore I would probably analyze the situation further...perhaps a redesign is pragmatic.


Overall I liked the scenario from Rui as it is a good test for those business situations at my site that call for this level of analysis.

Phill,

To answer the question about my familiarity with PMO, yes, I did attend the seminar at the 2006 RCM conference in Las Vegas. This does not make me an expert – it only provided me with a flavour. The talk was interesting, but I was more taken by Neil Bloom's discussion on RCM and the unobvious failures real RCM can prevent. I thought the perspective opened the door for me in understanding "why RCM?".

For all,

I find many companies have the opportunity to generate savings in maintenance budgets, but when the unexpected occurs the potential savings had they been prevented would greatly outweigh the savings in streamlining the existing maintenance strategies (if I understand PMO correctly??? – that is a big "IF"). At the same time performing any kind of improvement is better than standing still.

I have noted that companies sometimes shy away from clearly linking these catastrophic events to maintenance usually stating there are a number of contributing factors (and as Josh mentioned some of these issues are amortized – further clouding the opportunity). I have found if enough questions are asked there is usually some equipment failure imbedded inside these catastrophic events. I am not suggesting that equipment failures alone are responsible for these catastrophic failures, but they play a role. Some people have used a "Swiss Cheese" analogy - enough faults occur in unison and eventually the barriers are overcome causing a larger catastrophe.

To continue with the analogy my goal is to reduce the frequency of the "Swiss Cheese" holes. Perhaps equipment failures can even be ascribed a larger role in failures – when equipment faults occur they may close the window of opportunities available to operators; perhaps even pushing operators into unknown territory which eventually leads to human error and a catastrophic event. Of course this is all conjecture, but for now it is an opportunity to follow up on.

Josh you are right - This should be in a new thread...but what should it be called – Jaz's philosophy on reliability? The ravings of the mad reliability coordinator?

I am glad to see so many answers and visits to this topic...

Phil,

You are absolutely right with regard to the numbers interpretation. It is just an example and so the failure rate should not be in question. For some reason, the periodicity of PM was chosen in such a way that the probability of gaskets to be replaced under a CM task is around 0.2. Many times people apply the rule of thumb: "90-10". That is to say that if no other criterion prevails, then choose 10% chance to CM or, which is the same, 90% to PM.

By the way, I considered gaskets as units and not as sets.

When you do know costs of CM and PM (materials, labour, opportunity and any penalties, if they apply), then you have to use maths once more to find out the proportion of CM versus PM that returns the minimal cost or the highest operational availability. This can be another case that I would like to put here in this forum.

Suppose that, the replacement of gaskets of pump A costs $600 and takes 30 hours during a CM task and $250 and 5 hours, respectively, during a PM task. If you select cost effectiveness as your decision criterion, what time interval (in whole days) should it be? And what if you shift your preference towards the criterion operational availability?


Svanels,

In my opinion, it shouldn't be left to the store's people the burden and the responsibility of deciding which the reorder points (is this what you mean by minimum stock?) are to be adopted. I think it should rather be the responsibility of the maintenance manager, if he is really supposed to be responsible for costs and availability of the facility.

If a vendor wants to sell more than the exact quantity that you need at a certain point in time and offers a lower unit price, once more, you have to perform some calculus because, it may happen that it is the most cost effective decision. Of course that the picture drastically changes if the equipment, where the parts belong, is to be discontinued in a short period, though the situation is also manageable in what calculus is concerned.

Jaz,

I don't think I quite understood the steps followed by you Jaz. According to your own words, if you calculated the MTTFs of the pumps, or the mean failure rate, you got demand as following:

Pump A: 0.00125 failures (or gaskets) per hour or 0.00125 x 16 x 2 = 0.04003 units/day
Pump B: 0.00094 failures (or gaskets) per hour or 0.00094 x 14 x 2 = 0.02627 units/day
Pump C: 0.00070 failures (or gaskets) per hour or 0.00070 x 13 x 2 = 0.01829 units/day

Total = 0.08459 units/day

Demand during LT = 1.5 x 30 x 0.08459 = 3.8068 units

Hence: Reorder point = 8 (for a service level of 98.39% according to the Poisson distribution) and not 4 (!?).

Behaving this way, one is putting too many units in stock (8 instead of 4) as one is erroneously considering parts needed for CM and PM altogether.

Tell me where I misinterpreted your reasoning.

You sound very motivated and you are certainly very knowledgeable and determined. You know what you actually have in hands and know where to go (based on numbers, what is surprisingly rare to witness). I am sure that you are doing a fantastic job in what reliability and efficiency are concerned.

Josh,

Hi. Because I am so often involved with such cases, I developed software (my son did it, more precisely...) in VBA (you might find interesting to take a look at the attached 4 screens of the application that deals specifically with the issue treated in this topic: Stock 0 shows the ID, Stock 1 and 2 ask for data and Stock 3 provides answers, despite it is in portuguese). This specific package was "baptized" as INES (a common Portuguese lady name).

So, whenever I have a case, I look at my paraphernalia of applications and choose the one that fits the case the best, and here I go. It may take just a few minutes to enter the data or hours when data are not available and I have to look through documents and inquire operators or engineers.

I just can't wait to have your answers to the problem described above.

Regards,

Rui

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

Gaskets_in_stock.zip (101 KB, 11 downloads)

Rui,

I will attempt to clarify my answer. I used the attached spreadsheet for determining my reorder point. I made a couple miscalculations in my original MTTF

Again to state my answer in a different manner I calculated the MTTF for each pump as 1/(probability of failure prior to PM) Multiplied by the PM timeframe. The probability of failure prior to a PM was calculated using a 2-parameter Weibull calculation. In all cases the pumps had a probability of failure of approximately 0.2. I think I accidentally entered the scale parameter on my first attempt as the PM timeframe.

So I performed the calculation again and found 4 gaskets equate to 99.14% (3 Gaskets would be 96.33%). Originally I had posted that 3 gaskets were required, but I would add in a fourth to keep the gaskets paired in stock.

Perhaps I am using the incorrect spreadsheet. I have not yet downloaded your answer Rui so that I may first describe my solution. After reporting this response I will download your answer.

And to further correct myself on the conversion of a 2500hr MTTF for Pump A to weeks - it is 22 Weeks if the pump is only run 16 hours a day (not 15 as I had erroneously noted).

**************
Edit: I had entered the wrong lead time and wrong MTTF - so I had to remove my old answer...
**************

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

Jaz's_Calculation.gif (13 KB, 14 downloads)

Rui,

Thanks for the kind words .

As for my work on the problem of the gasket reorder point - I am hopeful that my method and answer meets your approval, I see I have used a different method to answer the original question then yourself.

Additionally when trying to post my answer it took me 3 reviews to re-check my numbers on the spreadsheet (I hope my manager isn't reading this).

Hi Jaz,

I am sorry but it is my turn to recognize that I have also made a miscalculation in my yesterday's post. I corrected it already. I attach an image of the "Spares" Excel file where you can see that if you consider numbers the way I do in this file, you would overstock because you are considering all the gaskets that you are going to need over a period of one year, without separating the ones to be replaced without notice (the proportion to be kept in stock just-in-case) from the ones to be replaced during planned actions of PM (just-in-time).

In short, the "Spares" Excel file doesn't take into account PM´s actions. So it must be used solely for the case of random failure modes. When you are dealing with a wear out failure mode, then, you have to perform the calculations the way I did in the "Gaskets of 3 pumps" Excel file posted on the 5th of the current month.

As you can see, if you consider CM and PM actions apart, your reorder point is 4 units. If you don't, your reorder point increases (doubles coincidentally in this particular case) and becomes 8 units. The average stock increases too and the holding costs follows the trend.

Regards,

Rui

Spare_gaskets.JPG (29 KB, 9 downloads)

Rui,

Good Point. I was under the assumption the PM components would be ordered separately. I.e. the planner would order the gaskets and other parts as needed for the PM (have PM kits), but this could increase the administrative costs.

If I hadn't already mentioned this is what we did when I worked in mobile equipment. We kept the stock for the "random" events and ordered kits for the regular PMs. This method certainly kept the planners busy, but we really hadn't worked on streamlining the process.

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

quote:

In my opinion, it shouldn't be left to the store's people the burden and the responsibility of deciding which the reorder points

They already took it !!! But I can advise in terms of lead time, by knowing the item.

Stores is beanies territory (at least the thinkers who can't distinguish a fuse from bearing)

With repsect to who should decide reorder points there is one thing that I know for sure.

Whomever is given that responsibilty must also be acountable for BOTH the operational impact and the financial impact. This, afterall is what all the calculations are ultimately about.

In my opinion it is the misalignment of this accountability that results in excessively overstocked inventory.

Unfortunatley very few companies actually ensure this alignment. This means that either the maintenance guys decide what to hold (and how much) resulting in excess stock. Or the stores/finance guys decide in which case they hold the wrong stock and too little of it - negatively impacting performance (and morale).

In both cases the investment in inventory is largely a waste of the companies funds.

Cheers,