Author Topic: MIL-HDBK-217F, MTBF and reliability: Rant  (Read 855 times)

0 Members and 1 Guest are viewing this topic.

Offline John C

  • Newbie
  • Posts: 4
MIL-HDBK-217F, MTBF and reliability: Rant
« on: December 15, 2017, 07:04:12 pm »
MIL HDBK 217F.  Today I took money to employ this obsolete, unalloyed, 100% pure bullshit.
It is what they want and they have zero interest in mending their ways.

The underlying principle of assigning a constant hazard rate to individual components and then summing it all up has no foundation in science.  It hasn't become outdated - it was always utter nonsense.  In fact, I know of only one degradation mechanism with an actual constant hazard rate; ionising radiation induced SEE (single event effects). Unless you are working on space, nuclear or medical systems, this aint your problem.

The standard excuses I hear when some cardigan wearing old duffer ensconced in the greasy cubicle he has occupied for the past 25 years  insists that MIL HDBK 217F 1995(!) adds value:
    "We have always done it like that".
    "Safety department need a number".
    "The customer wants it".
    "We know it is conservative, but we apply a correction factor".

Believe it or not, I just came across a two orders of magnitude fudge factor. Seriously. They called this reliability engineering . :palm:

Well, I am a reliability engineer and sometimes this cretinous crap makes me wish I wasn't.

Stuff fails for one or more of three reasons: Change (degradation), variation (insufficiently mitigated by robustness)  and difference (design).  NOT because, perfectly randomly, "its time has come".

But there is hope, usually found under the banner Physics of Failure.  Research institutions like CALCE and some medium to smaller companies outside aerospace and military actually take reliability seriously. 

And please lend your support to Fred Schenkelberg and his campaign to eliminate the widespread misunderstanding of MTBF.   
http://nomtbf.com/

Sorry, just needed to vent.
 
The following users thanked this post: rsjsouza, 2N3055

Offline Pitrsek

  • Regular Contributor
  • *
  • Posts: 133
  • Country: cz
Re: MIL-HDBK-217F, MTBF and reliability: Rant
« Reply #1 on: December 16, 2017, 08:39:02 am »
Thank you for the web link, interesting!
Do you have a recommendation  for a well rounded resource about reliability/robustness of design? Mode of failures, distribution in time, de-rating, best practices etc. I'm no reliability engineer, I'd just like the stuff that I do to be more reliable/less overengineered  :).
I've read "The Design Analysis Handbook : A Practical Guide to Design Validation" - this was a rather interesting read and I'd like to read some more about the topic. The goal is use less of rules of thumb and "we have always done it like that" and more cold hard data and engineering.
Thank you. 
 

Offline John C

  • Newbie
  • Posts: 4
Re: MIL-HDBK-217F, MTBF and reliability: Rant
« Reply #2 on: December 16, 2017, 10:38:22 am »
Hi Pitrsek,
Recommending information resources is always tricky, because best practice in reliability engineering relies so much on the specific context.   That said, Practical reliability Engineering by Patrick D.T. O'Connor is essential reading for anyone concerned with the subject. Buy this book, it will serve you well.   ISBN 978-0-470-97981

In some cases, a reliability engineer is the "chief problem manager and solver" on a product development.   "the Toyota Way" an "Understanding A3 Thinking" are worthwhile reads.  Too big a topic to cover here really.

If engaged in quantitative ALT (Accelerated Life Testing) of components, the resources provided by Reliasoft at [url http://www.weibull.com/[/url] are useful. (And some of their software isn't half bad either).

If dealing with repairable systems (i.e., systems affected by degradation mechanisms during their service life) and are employing Reliability Centred Maintenance to deliver a level of availability,   Reliability Centered Maintenance   by John Moubray is the place to start. (Goes on a bit, but still the best reference on the subject I know of).  And if you really get into that, actually modelling maintenance scenarios using Monte Carlo simulation can be very illuminating (see Reliasoft Blocksim).

Whatever the context, for me a good DFMEA (Design Failure Modes and Effects Analysis) is the foundation.  Not some "part by part" standard-based nonsense that you will find by Googling the acronym, but an actual database of functions, cross referenced to design solutions, their failure modes and identified  knowledge to justify why they won't happen, of absence thereof... Every simulation, analysis, test or experiment can be referenced back to a knowledge gap in the FMEA - or else it is waste.
No sources for that I'm afraid, you'd just have to hire me  ;).
 
The following users thanked this post: Pitrsek

Offline Pitrsek

  • Regular Contributor
  • *
  • Posts: 133
  • Country: cz
Re: MIL-HDBK-217F, MTBF and reliability: Rant
« Reply #3 on: December 16, 2017, 08:15:14 pm »
Thank you for the recommendations!
 


Share me

Digg  Facebook  SlashDot  Delicious  Technorati  Twitter  Google  Yahoo
Smf