MTBF

Its all to do with statistics...

There's lies, damn lies and then statistics...

John.

MTTF (Mean time to failure.)

is possibly more relevant unless it is a repairable item.

It also depends on the standard deviation... 3000hrs +/- 2000 wouldn't really mean anything!

You need a full set of figures to work out anything meaninful.

Usually there is the 'bathtub' shaped curve for failures..eg. Early failures, a period of stability and reliability...then the wearout failures. That's why extended warranties are a rip off...if it survives the first year it will brobably be fine for the next 3-5.

Del

And, logically MTBF is mean time before failure (which is the same as MTTF) and mean is the statistical average of the values for a population or sample.

I don't believe it is...I may be wrong....not sure I can be arsed to research it tho'.

Thanks, Cat.

I realized my error immediately after replying. One of the interesting stories about MTBF relates to the early days of vacuum tube computers. There were so many tube failures at startup (after powering down) that the decision was made to keep them powered up continuously. Over weekends, holidays, etc. the computers were kept busy working on problems like rationalizing an irrational number. That significantly increased the tube MTBF.

Check this ....just as I thought.

Del

It's a pretty useful number for something like stocking replacement parts where you use a lot of them, for example, lamp bulbs. I would not use it for a nuclear power plant.

if it is just statistics then whats the need to calculate it?

Mohammed:

When a system is being procured for service, the using organization must know how to handle maintenance. For example, for a piece of aircraft equipment to be used in a given number of airplanes and flown for some given number of hours per month, how many failures can you expect (calculated down to the piece-part level)? From that (and a few other parameters, such as mean time to repair) you can determine a number of things, such as: Maintenance shop size and staff level, test equipment needs, spare parts inventory, test rigs needed and, even, spare aircraft requirements. And more.

DickL

thank you so much! sir, for a very good explanation. For an electronic product consisting of some thousand components(resistors, caps,diodes transistors,etc), it takes atleast 200 man hrs.,whatever software we use. So is it actually so useful for spenting that many hrs. But anyhow with your explanation if i can conclude that for repairable items how it is useful.

But i think for non-repairable products it just gives reliability information, and just a marketting feature.

But i think for non-repairable products it just gives reliability information.

Read my posts and link. ....MTTF is what you need for non-repairable items.

Del

of course the terminology for non repairable items is MTTF, but the its significance is what i need. For repairable items already we had a discussion but the same disc. doesnt apply for nonrepairable products......

please give me the clarification.

Mohammed:

One consideration is that, in the selection phase of a program, a procuring agency might very well use MTTF in choosing a system supplier. If everything else was equal, it would make sense to buy the system with the lowest failure rate, wouldn't it?

Beyond that, once the system was selected, you can answer your own question by putting yourself in the position of the procuring agency. Ask yourself the same kinds of questions that we discussed for MTBF.

If you are responsible for support of a fleet of 1000 airplanes with components that are subject to failure, what do you have to know to do your job? Failures per flight hour, time and skills required to isolate the failures, time and skills required to replace the failed LRU (Line Replaceable Unit), special equipment needs, number of spare LRUs, AOG (Aircraft On Ground) time per failure incident.

If you wish to become a serious supplier of such equipment, you should be thinking in these terms yourself as a guide to effective product design.

Best regards.

DickL

mtbf is used to prevent a major machine downtime cause by, example, parts wearout failure.

Simplest answer I can give, the longer the mtbf the less you fix or replace that item.

If you have a hard drive with 5k hours MTBF and another drive with 50K MTBF @ 2X $ of the first, what fits your needs better?

There may be times when a low MTBF is more cost effective than a high one ( Dat tapes are one example. a few bucks for alow MTBF tape or a few hundred for a high MTBF tape; but you only keep the tape for 5 or 6 uses ).

Generally MTBF is predicted using MIL- STD 217, can we validate the MTBF figure so obtained using some Hardware tests?

Mohammed:

It is not practical to "test" MTBF predictions. The scope and cost of a statistically valid test would be much too great. User groups do, however, keep records of failures over thousands, or millions, of operating hours and make a running calculation of MTBF. This makes it possible to evaluate supplier calculations and, also, to watch for anomalies which could point to correctable problems.

Even the failures which occur during environmental screening of new product would not provide accurate MTBF information. Those are "Infant Mortality" failures which occur at a higher rate than would be seen in mature product (for which MTBF calculations are intended).

Best regards.

DickL

Thank you so much for the reply.

HALT and HASS data is used for predicting MTBF. There r even softwares which does that. Can we compare the MTBF obtained by using ALT data and by using Mil-Hdbk-217. Is this really used in industry?

"Can we compare the MTBF obtained by using ALT data and by using Mil-Hdbk-217. Is this really used in industry?"

Mohammed:

Any experience-based data source should provide valid failure-rate data, but don't get complacent.

My experience was in Project Engineering for military and airline flight equipment. MIL-HDBK-217 was used extensively and successfully by Reliability Engineering specialists within Engineering as a tool for predicting the statistical random failure rate of mature equipment. That is NOT wearout rate as described in some of these messages. You can't predict the failure time of a particular disk drive, etc, within "a few %"

Standard deviations could be quite large, but, fleetwide, the analysis allowed for effective maintenance planning. Customers needed this information and required ongoing coordination throughout the life of the equipment. It was important to me because, frequently, our customers turned to us for maintenance equipment, training and spares.

These calculations would have no value for an individual who is buying one piece of equipment, but are critical for fleet operators.

Best regards:

DickL

The best use of MTBF is for depreciating the device. If you have a device that is on 24/7 divide the total MTBF by 24 and that is the expected lifetime, so you should be ready to replace the device around that time. Remember MTBF is an average of manufacturers tests and could be longer or shorter by a few %.

If you are applying MTBF to a full system like a Computer then check the various components MTBF's and budget accordingly. Components like motherboards and video cards are usually obsolete before they pass their MTBF. Parts like hard drives and dvd's can reach their MTBF before they are too obsolete and can be replaced easily to extend the life of the system.