Some problems related to MTBF calculations

Disclaimer: don't take me as much of an authority. I've worked with folks who have to do MTBF calculations and I poked my nose in their business a few times, but I have no formal training. I'll try to keep to things that I know to be true, but I may use odd terms and I can almost guarantee you that my approach will be unorthodox.

MTBF is more or less the reciprocal of the "real" parameter it work, which is the probability density per unit time that a failure will occur. Failures are assumed to happen due to a Poisson process (or set thereof) (so if you look up Poisson process you should find some useful reading).

I was going to say something brilliant about the Poisson process, but I find that I can't remember the details of how to calculate anything. Basically you start by making the assertion that for some short time interval T that the probability of a failure happening in that interval is T / MTBF -- this approximation holds for T

1. So for T = 1 hour and MTBF = 10000 hours, the probability of a single failure in that hour is about 1/10000, the probability of two failures in that hour is about 1e-8, etc. For a 10000-hour chunk of time, you make this into a binomial distribution with P_1 = 1/10000, so P_0 has to be

So: this is vague and incomplete, but I hope it helps.

thanks tim

i needed someone who 'd translate this into common probability language. probably i can understand what u wanted to say.... :)

I assume you've tried reading the set books for your course? Paying attention during lectures helps too.

Also try learning to write proper sentences (perhaps English is not your native language, in which case grammar mistakes will be forgiven), learn about some basic punctuation and capital letters, and drop the childish SMS abbreviations.

That's a nice trick question to fool students who don't understand basic probability...

We have sympathies for you.

Well, Tim Wescott probably made his best. While most of us are probably engineers and not mathematicians I have to add a few comments...

I don't thing you have enough information here. Anyone doing maintenance? MTTR?

I don't think you can calculate this. Probability of failure does not have a uniform distribution. It's more likely when the lamps are new, after that initial time probability if failure is very low for a long time and then increases again as the filement gets consumed.

can we solve these if we assume MTBF=1/failure rate and leave out all those practical problems.. [cause at the end of the day its just another assignment]

yes

w..

Clearly they're textbook problems; from the way they were worded I think you can assume that MTTR = 0, and MTBF is constant. (Presumably those come in a later chapter :-).

Most of the MTBF calculations I've seen are for the bottom of the bathtub; if you're selling something on it's MTBF then you're assumed to have done any necessary burn-in to proof the system against infant mortality, and if there are any wear-out mechanisms then you're supposed to state the MTTF of the wear components, or sell the system with a maintenance schedule that keeps the wear-out components replaced so you stay on the bottom of the bathtub.

yeah i assumed MTTR =0

otherwise there was no way of doin those problems

Actually, that's not correct either. Go back and read the questions again.

Also, could you provide a link to the course's syllabus, the on-line notes, or your instructor's e-mail? It would be handy to have an idea how much you've already covered in the course.

There is no MTTR given for the case where the repair guy is standing there and the equipment is broken. You have to assume some number, the implicit assumption is that Jensen Tools is now selling magic wands and the repair guy's name is Hermione Granger.

But yes, you can calculate the part of the MTTR that comes from Hermione's availability (presumably she hasn't gotten another Time Twister).

For sufficient values of "guy." ;-)

As I read it, though (with some assumptions, since the OP hasn't come back with a link to the syllabus) the questions can all be answered as either time to first failure or with no repair performed, as appropriate, with the equipment created de novo at the start of the question. I'd *guess* that's what they're looking for, at least.

basically i ve already solved the problems, though the answers are compromised theoretically, that was what the instructor needed.. anyway thanks 4 all the helps, guys.. Thiswas my first USENET thread, and the response i got here was simply amazing..

ciao..

@ Tim, thats exactly what we had to assume if these were practical problems, and as they r not, why disturb Hermione? :)

Kindly avoid using abbreviations. They only confuse the newsgroup.

Please do not top-post. Your answer belongs after (or intermixed with) the quoted material to which you reply, after snipping all irrelevant material. See the following links:

(taming google) (newusers)

Why don't the companies that manufacture bathroom fixtures design them so that the top and bottom are equally reliable? :-)