At the moment I am setting up preventive maintenance plans, conform RCM methodology. I use a FMEA matrix to determine the consequences of possible failures and dependent on the results, maintenance rules are defined. So far so good.
In our plant we have a lot of redundant equipment, because we can not take our plant out of operation. If for instance a redundant pump fails, the consequence is no production loss, so the advice from the FMEA will be: no preventive maintenance. But through the pump failure, the operation will become less reliable than before, because there is no back up. How to deal with redundancy and FMEA. is there someone who has experience with RCM strategies on redundant equipment?

Erik

Approach on, how does PdM put money in your pocket. Avoidance of catastrophic failure. Provide additional insight to underlying faults to enable increased MTBF. Least expence on actual maintenance performed.

Needless to say, PdM does fit somewhere in your facililty and expansions of those technologies gives greater cost justifibility.

Motor ($4,000), Pump ($6,000) coupling ($500). MTBF or L10 = 5 years or something common sense reasoning. A timely rebuild could mean bearing change only and may avoid catastrophic failure. I've caught motors where the O/B brg was bad only - so, knock off the end bell and 1 hr later back in service with a new bearing.

Monthly, quarterly, OA tracking of 1 pt/brg can make it cost justifable. I do many 1 pt/brg w/FFT and TWF but track the OA level.

Naturally you don't want $3,000 worth of analysis on a $1,000 motor.

quote:

If for instance a redundant pump fails, the consequence is no production loss, so the advice from the FMEA will be: no preventive maintenance.

What is the logic of FMEA?
* PM for the first, no PM for the second because it is redundant.
* No PM for both, each of them are redundant to the other one.

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

Hi Erik,
When you have redundancy, the maintenance analysis applied comes under the analysis of hidden failures. You can work out the risk level using a simple formula if there is one level of redundancy. The formula and an example are shown on the attached document. Note that MTBF is only a constant if the failures are random.
The formula deals with failures that are unexpected - in that you are doing nothing to prevent or predict them. The formula comes up with a failure finding interval based on these parameters. The idea is you set the frequency of failure finding or testing to match the risk levels you are willing to accept... but it certainly does not end there. You have other options. Some of these are listed below:
• Perform predictive and preventive maintenance on the pumps to reduce the rate of unexpected failure.
• Build in additional redundancy (not just pumps but valves and other things that may fail)
• Make some hidden failures evident - ensure your circuits are fail safe.
• Make all failure evident - run the standby all the time (expensive but is done on some standby generators)
You may be able to think of others.
Hope this helps
Steve
PS - The attached file is a word print file of some description.
Please email me at steve@omcsinternational.com if you cant read it.

FFI.mdi (37 Kb, 42 downloads) Failure Finding Interval Calculation

Steve what type of file it is? which program to use?

Erik,

quote:

If for instance a redundant pump fails, the consequence is no production loss, so the advice from the FMEA will be: no preventive maintenance. But through the pump failure, the operation will become less reliable than before, because there is no back up. How to deal with redundancy and FMEA

The basic process is one of managing risk effectively. there are two parameters of concern; consequence AND probability. In the FMEA, we evaluate the consequences, but we do need to look at probability as well.
In the case of redundancy being available, with a duty/standby operation,
A) For the duty eqpt., as you say rightly, there is no immediate consequence. hence a run-to-failure strategy is acceptable. However, this assumes attention to the failed item in a reasonable time frame. If the repair time is very long, say in many weeks or months, the Plant is exposed to the risk of the standby item also failing. That is the drop in reliability you mentioned. There are two ways of handling this.
1. Reduce repair time of duty item to as low as is practical, without incurring penalties such as overtime.
2. Even though run-to-failure is an acceptable strategy for duty item, do some minimum low cost maintenance such as condition monitoring, keeping item clean, lubricated properly and aligned.

B) For the standby item, as Steve points out, the standby item has a hidden failure viz. fail-to-start. With a test regime of starting the standby periodically and loading it fully for, upto 24 hours, that risk can be minimized.

Here it is in pdf.
The file was Microsoft Office Document Image Writer - obviously not a good choice. This computer I am working on does not have adobe writer.

Failure_Finding_Tasks.pdf (66 Kb, 55 downloads) Failure Finding Task Forumlae and example

Regarding the scenario in the Failure Finding Task document, please explain how the failure associated with pump B is evident. If Pump B is running and all of sudden stop running, pump C automatically takes over. There is no loss of operation. It seems to me that the operator may not be aware of the failure of Pump B because of the built in automatic feature.

Here is an actual scenario. A battery charging room has two redundant exhaust fans that remove the explosive fumes. One fan is always running and it is designed that the other fan will come on when the duty pump fails. The exhaust fans are belt driven. The exhaust fans are located in the ceiling. There is no preventive maintenance because of redundancy. While doing a walkaround inspection, I noticed a noisy belt. I put a request to investigate the noisy belt. The mechanic found out that the standby fan belt was running with a belt ready to go. The belt of the duty exhaust fan had already failed. With the actual scenario, I consider the failure of the duty fan belt hidden failure.

Hidden failures requires PMs aimed to test the condition and repair or replace as required (fault finding PMs).

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

For duty / stand-by arrangements with automatic switch over, I recommend that you also install some type of indicator or alarm when the stand-by equipment comes on-line. Something to alert the operator that the duty equipment has failed.

Shelley

quote:

There is no preventive maintenance because of redundancy

You mean the fans are running forever without somebody ever looking at it or getting in the room?

I never would open that door, if you have gas inside there, open the door could let oxygen in or something dangerous out.

I don't call this redundancy, I call it stupid!!

I would add to the run to failure definition

This does not mean that we forget to look at a running equipment until we hear kaboom!!

PM (preventive maintenance) is also from time to time put some drop of oils and dust it of.

With the run to failure strategy

• we hold minimal parts/consumables in stock
  
• looks sometimes at the equipment (3 months, 6 months or every year) to verify that it is still working or not have been stolen
  
• if it breaks down, we can either buy a new or have enough time to repair it
  

If you look at the wiper blades of your car, what maintenance strategy is common sense, yes run to failure, but I wouldn't apply the same method for the brakes, or the engine.
By the way in most parts of the world there is a mandatory annual inspection of your car by some authority.

Erik,

If you have the setup similar to my experience, I agree with Shelly, make the system functional failure evident, otherwise, PM is required like Eugene stated.

Rod,
If the B pump stops while running,
- can the operator see whether it is running if he goes there?, or
- is there any indication on screen or panel whether it is running?

When deciding if a failure is hidden, TIME is not an issue. If at any time after the event, the operator CAN know if the item has failed or not, THEN it is an evident failure.
In the case of a hidden failure, an operator CANNOT normally know by his senses or by instruments, e.g. he cannot know whether a pressure relief valve or gas detector is working or not, UNLESS he places a dem,and on it, i.e., a second event or failure is required to reveal the original failure.

Shelley,

quote:

For duty / stand-by arrangements with automatic switch over, I recommend that you also install some type of indicator or alarm

I hope you realise you are introducing a new hidden failure, that of the alarm or indicator itself failing to perform.

svannels: "or not have been stolen"

Hmmm! I think I need to update the drop down list of failures codes at SAP.

quote:

Originally posted by Vee:
I hope you realise you are introducing a new hidden failure, that of the alarm or indicator itself failing to perform.

We know... but it may prove to be more economical to add an alarm plus the PM or Calibration inspection(s) that the alarm system will require, that to face the costs of the main system failing.

The only equipment that is untouchable is the clock..
Because everybody is watching it

I'm amazed at the similarity of plants across the world / industry.