Rod,

quote:

The exhaust fans are belt driven. The exhaust fans are located in the ceiling. There is no preventive maintenance because of redundancy.

1. Belt drives can be hazardous as loose belts sometimes catch fire.
2. Belts become slack after some time and need periodic tightening.

So no PM seems a strange decision. Redundancy allows us to reduce maintenance work if ONLY production consequences are involved. If safety consequences are there, this decision needs a revisit.

in an explosive environment, perhaps the belt drive is not an ideal solution.

Darth,

quote:

prove to be more economical

I think you miss my point. It is not a question of economics. If we put an alarm, we DEPEND on it to wake us up when things go wrong. But the alarm itself may fail to light up or sound, AND as it is a hidden failure, we may never know its condition UNLESS we maintain IT by testing!

Vee, I think I understand your point, I recognize that this additional element bring another hidden failure to deal with; an additional fault finding PM to process, but at the end:
* which setting is better?
* which will be more economical?
* more reliable?
* easier to plan maintenance for?

*** equipment A, redundant equipment B, and alarm system to signal that A failed and B is running;
vs.
*** equipment A, redundant equipment B, no alarm system.

Do early notice of equipment A failure justify handling an additional component in the system?
Do additional alarm decreases overall reliability?

Darth,

quote:

Do early notice of equipment A failure justify handling an additional component in the system?
Do additional alarm decreases overall reliability?

Good questions, and the answer is NO. But before we add new christmas lights, let us examine what else we can do.
1. A toothed belt drive is intrinsically better than a V belt drive; it will not slip and cause a
fire. It also decreases bearing loads.
2. When there is a potential safety hazard, a belt tension checking PM does not seem onerous. At the same time, bearings can be greased, motor cooling fins cleaned, foundation bolts checked for tightness and a general look-see of the area. We will thus know when belts need replacement next.
These are relatively low cost actions, with potentially high returns (we can't burn the place down!)

My concern is this; in offshore subsea installations, people think that once you put equipment on the sea bed, they dont need attention, because 'you dont or cant see them, so they must be OK'. Exhaust fans on ceiling are similar, or pipelines in trenches partly filled with water or slime.

Every alarm we add needs a failure-finding routine in the CMMS. Every alarm we add has the potential to annunciate when there is no reason to. i.e., spuriously. Operators may fail to do even simple checks just because there is an alarm. In a Plant where there are too many spurious alarms (and by the way this is very common, I have discussed it on pages 97 and 101 in my book).

Surely the correct action is do the basic work before attempting re-engineering every problem; following RCM logic, we would only add alarms once these avenues are exhausted.

I am not against alarms; but they should only be added after careful thought, as every solution is itself a potential problem.

Thanks Vee. If I follow you, you are telling then:
* equipment A, redundant equipment B, no alarm system, plus basic PM work

is the first option to try.

Darth,
I am saying there is no universal recipe. In Rod's example, both fans are not easily accessible. There appears to be no indication by e.g., ammeter, alarm light, panel indicator or other means for the operator to know whether
Fan A is running, or
Fan B is running, or
Both fans are dead
This case is not the same as two pumps which we can access easily. In Rod's example, ANY failure is hidden, unless the fumes themselves alert us to a problem. My concern is that a principle that a redundant item can be run-to-failure because the standby will cut in automatically does not mean we dont examine secondary effects of failure modes, especially when safety is involved. A fan belt failure may stop the fan, and since the standby cuts in there is no loss of 'production' But a V-belt can slip before failing and catch fire. This is a safety consequence that is unacceptable. Even if this did not happen, if we have no way of knowing which fan is running, or if any fan is running at all, then we have an explosive vapour build up. That is far more dangerous.
So your solution of putting an indicator light is both economic and practical, because now you at least know which fan is running. But this is surely not enough. Because V-belts become slack and can slip, one MUST tighten them, so we cant avoid a PM. If the fans were exhausting a non-explosive mixture, these concerns would not be there. I hope I make the distinction clear.

Having redundant systems should provide greater opportunity for preventive/predictive maintenance. To allow equipment to run until failure in my opinion is irresponsible. Having redundant systems eleminates the excuse of not having opportunity to properly care for operating equipment. By taking advantage of the redundancies you could save a considerable resources in a short period of time. If your systems are not life safety I believe you should have to manually recover from a failure. Automatic recovery can be dangerous at best. All failures should be immediately known and appropriate action taken for recovery and repair.

Too many redundant systems create also another problem. No parts in store, nobody worries because there is always back-up, (but they forgot to report the inoperative back-up months ago)

Very good point Steve.
I have seen some plants that have been built with redundancy or bypasses almost everywhere. Can you believe this one plant could not get much above 80% of nameplate capacity....even with all the bypass and redundancy. Why - no one thought any of the machinery was critical. Something goes wrong - divert product somewhere, run a little slower for a while and the list of actions continues. After a few workshops we were beginning to wonder when we were going to start working on the critical equipment - We always assumed we were working on the critical stuff but the first two equipment items weren't considered critical. When I probed further I was told there wasn't any real critical equipment.
Well the whole attitude was wrong. No one was really accountable and no one measured anything substantial. When all this changed and the company started to count losses and people were held accountable things turned around. So it is easy to see how building failure tolerant plant can be a waste of money and effort and actually drive the wrong behaviour...
Also - the Japanese succeeded in manufacturing for many reasons but one of them was their approach to lean manufacturing and just in time. They set out to eliminate waste such as buffer stock which hid all the problems. A very good concept indeed.
Steve

It sounds like the case of having too much alarms being designed and installed that an alarm rationalisation exercise is necessary.

Josh... if you are responding to my post, these were not alarms - these were physical equipment items that were put in as stand by. In most applications if one pump was required then two were installed, if two were required three were installed.
Steve

Yes, there are different cases but both have a similarity ie overdesigned. In your case cited above, wasn't there any over-alarmed situation inaddition to over-redundancy & bypasses?

I don't think that there was an "over alarmed" situation. They had an attitude problem, nobody was accountable.
They could have 10-millions alarms, and still nobody would come into gear.

There were cases of over instrumentation... get this... We have a kiln that operates between x and y. There are eight temperature probes feeding information to the control centre about the kiln temperature. Inst calibrate these to 0.1% every three months. Operator says he goes to the outfeed conveyor and by hand feels the product to determine if it is ok... too dry turns the kiln down - too wet turns the kiln up. Infeed is variable... no set parameters.
Operator does not care if the kiln indication is within 5% yet instros calibrate to 0.1%

Beam me up Scotty
Doh?

I read an article by "Vee" that was extremely interesting that stated by staggering the duty-standby equipment to a disproportionate amount 7-weeks 1-week the return could be as much as 10-12% rise in MTBF and a like reduction in maintenance cost. Most redundant systems I have seen in plants typically split the time evenly. After reading the article I agree with him. Although this is not exactly what you requested it is a strategy that may prove beneficial.

A visual check at the time of switch over would give you six opportunities per year to pamper the units.

Hi,

quote:

I read an article by "Vee" that was extremely interesting that stated by staggering the duty-standby equipment to a disproportionate amount 7-weeks 1-week the return could be as much as 10-12% rise in MTBF and a like reduction in maintenance cost.

In the first instance, I suggest a pure duty standby operation where it is possible; i.e. say 8 weeks on and 1 day off. The interval may be as low as 4 weeks in some cases and 12 weeks in other cases. This depends on the fail-to-start failure-rate of the standby equipment on the one hand, and the required system availability on the other.
Under certain circumstances, this policy is difficult to implement. In THAT case, consider unequal operation, e.g., 7 weeks on, one week off.
With alternate running every two weeks, the duty and standby pumps together will have 26 starts, or 13 starts per pump.
With duty-standby at 7 weeks on 1 week off, we will have 13 starts for the two of them, or 6.5 starts per pump.
With duty-standby at ~8 weeks on, 1 day off, we will also have 13 starts for the two of them , on average, or 6.5 starts per pum pa.
Thus with a duty-standby regime, we have 6.5 starts p.a. vs 13 starts p.a. for the two weeks on two weeks off regime. THIS reduces seal failure rates by about 50% in theory, less so in practice. In the oil and gas industry, using OREDA figures,seal failure rate account for about 22% of pump failure rates. With fewer starts, we would expect seal failures to drop correspondingly. In theory, we could expect pump failure rates to drop from 100% of original value to ~89% of this value. This appears to be borne out in practice.

Erik: Using the FMEA approach and an RCM philosophy doesn't mean you should allow the analysis to result in an "operation (will become) less reliable than before, because there is no back up." Even though, as you state, the plant has lots of built-in redundancy, the very fact that the plant needs to run 24/7/365 should be the core foundation of the reliability efforts. There is never wasted effort to exploit the FMEA process to yield proactive maintenance tasks to enhance the reliability of individual equipment. Emphasis should be placed on widespread and deep applications of predictive maintenance techniques. As well, assuming continuous operations are critical as you state, there is much to be gained from RCM analyses of your systems in terms of redundancies and identifying possible weak reliability areas. Is the operation balancing the wear on all equipment within a family, or is the approach run-to-failure (RTF) then switching to an online back up? Gaining reliability at component level with thorough FMEAs and implemented tasks bolters the reliability of the plant as a whole; when combined with the global-view RCM you can then strengthen your systems overall. It may well be that you can then reduce the preventive maintenance task frequencies, saving these costs, once you have applied vibration, thermography, and ultrasonics maintenance programs on the assets as applicable. Good luck.

skm,
You advise

quote:

Is the operation balancing the wear on all equipment within a family,

Are you suggesting 50:50 running?