Continue to Site

Eng-Tips is the largest engineering community on the Internet

Intelligent Work Forums for Engineering Professionals

  • Congratulations SDETERS on being selected by the Eng-Tips community for having the most helpful posts in the forums last week. Way to Go!

Overwhelmed about testing please help!!

Status
Not open for further replies.

Viar

Automotive
Nov 12, 2007
4
I need some advice from some of the reliability professionals in the room. I work for a small company in the capacity of research and development and now have been asked to develop a reliability program. My company manufactures 12V and 24V air compressors primarily for automotive, OEM and industrial applications. We have been working on a military contract for about a year or so and going through all the mil spec testing (which is a learning curve for all of us). I have a test lab (PLC controlled)on site that will allow me to run our product and log data to narrow in on and pin point failures. Frankly I am not sure where to begin. I was asked to do this to help improve our current products and help validate new upcoming designs. Anyone know of a computer program or book to assist a non-statistician in accoplishing building the foundation to a reliability program. as an FYI I am not an engineer by trade I just happen to have a very firm knowldege and understaning of most things automotive,so here I am and, needless to say,my head is swimming due to the amount of information out there that I have been looking at to help guide me through this so.....HELP! Thank you in advance to anyone who can assist me in the slightest way

Sean
 
Replies continue below

Recommended for you

First and foremost, create a list of "things" that you want to measure / keep in control, i.e. output pressure.

Next, determine your limits for each "thing", i.e. output pressure +/- 2.5 PSI.

Then, figure out how to measure each thing. Use a pressure gauge that has 0.1 PSI resolution.

Finally, verify your measurement systems are repeatable and reproducible.

Hopefully, that will point you in the right direction.
 
Get access to an exisitng application for the product, even if it's not your product.
Then get it instrumented and collect data. Stuff like:
- temperature
- pressures
- voltage & current waveforms
- on/off cycling times
- time 'on' with respect to the primary 'mission' (eg if its part of an engine installation, how many hours will the compressor be on compared to engine running hours)
- vibration inputs from the compressors mounting points

Once you have this sort of info, you can start to think about replicating the real service loading of the part with concurrent, temperature, prssure loading and vibration.

It's not easy.

Alternatively, if you're in the USA, talk to Southwest Reasearch, San Antonio. If you're in the UK talk to MIRA, Nuneaton.
Neither are cheap but both are practised at this sort of thing.
You could also sift the papers of the SAE and the UK Institute of Mechanical Engineers for service loading and reliability studies.

Once you have devised a test and achieved some results, which should include systems that have been run to failure, you'll need some help from a reliable and practical statistician. Alternately, software from someone like Fulton Findings (California, I think) to conduct your reliability analysis.



Bill
 
>> You need to find out what your CONTRACT says, including whatever change orders there are. Are the testing being asked part of the contract?

>> You need to understand what type of contract this is; is it a fixed price contract, or is it a cost-plus development contract. This will affect your flexibility in addressing what your contract calls for.

>> You need to find out what your company bid. Likewise, this determines what.

>> You need to find out how your products fail. Hopefully, your company keeps records? Are there others that have failed, but were unreported because the customer switched to a more reliable product? Are your competitors' products more reliable?

>> You need to look at the technical requirements of the contract. What environment is the product specified for?, what vibration, shock, temperature, voltage, etc.?

Bear in mind that a reliability enhancement program is not cheap, particularly if you're only using in-house product, since these will not be eligible as shippable product and therefore generates no income.

>> You need to have a REALLY GOOD failure analysis capability, either inhouse or outside. You need to find out precisely what fails on each and every product you can lay your hands on. Are there any common themes in the failures? That's what you're primarily looking for, since common failures are the most likely to be analyzable for weak spots. Single failures might simply not be statistically significant.

There are a number of military and commercial experts in failures, physics of failures, and highly accelerated life testing (HALT).

TTFN

FAQ731-376
 
Thank you very much for all of your responses. To (hopefully) try and clear some of this up - We already have an outside lab assisting with the mil project the in house lab will be for fielding future issues that arise but it's primary function will be for life testing some of our most popular units. At the current time The units are set up to run based on our publicized duty cycles and I can collect pressure, temurature, amperage, voltage and run time (both total and cycling within specific PSI range). As I have gathered I will probably also need to find a way to monitor vibration.And I apologize for all of the inquirires but I am bound and determined to make this an invaluable part of this company, it's just been tough trying to wade through all of the available information.The number of units being tested at this point is very random, being 2 units per part number as manpower and time permit. Once again thanks to all who have responded.
 
You didn't say whether these units are run continuously or not. Startup and shutdown performance might also be both clues and factors to overall reliability. For example, one could monitor time it takes to reach pressure from a discharged and unpowered state. Over time, this should take longer as the compressor efficiency ought to decrease over time.

Additionally, automotive engines suffer their biggest reliability hits from startups; there might be similar effects here.

TTFN

FAQ731-376
 
The unit is filling a 5 gal tank starting from 0-145 PSI and then cycling from 110-145. And the current unit is a 100% duty cycle so it will be continuous - but - it will only be run for approx 8 hours a day, 5 days a week -
 
OK, so at the beginning of each day, the compressor is started up, and the tank is filled. When the pressure drops to 110, the compressor is turned on and cranks the pressure back to 145. At the end of the day, the compressor is turned off. So, a worst case scenario is one where the tank is depleted the instant the pressure reaches 145, and the compressor only rests during the time it takes the pressure to drop to 110.

So that would be the basic test cycle, run the compressor to 145, rest a few minutes, dump the pressure down to 110 and crank the compressor again. Intersperse all of that with a complete pressure discharge and a slightly longer rest.

Therefore, the time it takes to pressure up from 110 to 145 would be useful information to know. Other than vibration and current draw, this might be the only other indication of wearout or pending failure. Likewise, the intermittent complete perssure cycle and pump time from 0 psi would also tell you something about the overall performance of the pump. The rate of pressure drop when the pump is resting might also tell you something about seal integrity.

TTFN

FAQ731-376
 
I'm sorry the compressor runs continuously and the pressure drops via a dump valve that is actuated by a pressure sensor in the tank - so there is no rest time
 
Hi,

Msquality here. Try Mil-Handbook-217. Reliability predictions. Should be on software. Minitab.

Msquality
 
Status
Not open for further replies.

Part and Inventory Search

Sponsor