Question on testing...

[cuco33]

New here, and to reliability and life testing... Great website!

My company is currently using life testing methods to test some sealing issues we have with a product. The device uses ultrasonic energy to nebulize a medium. Currently the life test stations are not used to quantify how reliable the device is by any means but to test how good the tested seal held up within the desired lifespan. We are looking to use these life test stations and reliability methods in the future once the product is finalized but have been stumped on which model/relationship I should be looking at. The device can not be accelerated by means of increased temperature, humidity, or power. Time is the only factor we have to play with. With that said, we have 2 separate accelerated life stations. The first station runs the unit constantly without stopping, while the second cycles the unit as per the spec (time during use, time to cool down system, repeat). We have multiple inspections that the unit must pass after life testing (simplest form is only inspecting under 3 parameters).
First question, is there a way to quantify a relationship of the constant life test versus the cycled life test?
Second, how would I quantify how reliable a unit is by having 3 inspection parameters but only 1 accelerated variable (time)?

Many thanks!

Nuno Sousa


PS: Currently looking at taking a course within the reliability/accerated testing fields.

[DrDave]

Dear Nuno,

Accelerated testing assumes that you have increased a stress, and that in turn accelerates the failure rate or scales the time dimension for failures, depending on the model you select.

If you run an item more frequently than it would experience in usual use, that is called 'duty cycle' acceleration, and life is scaled proportional to the usual fraction of usage. (I.e., run 4 hours a day usually and run 24/7, you can accelerate by a factor of 6). That is not strictly accelerated testing (see paragraph above).

From my reading of your post, duty cycle acceleration is the only approach I see for you, unless there is another possible stress that induces failures more quickly. Those are usually found by thinking about the failure mechanism for the product. (For a sealer might it be UV exposure? ) It is worthwhile to ponder what stresses might accelerate failure. Remember that increasing stress usually doesn''t accelerate ALL the possible failure modes, so one must be very aware of the physics of failure for the product.

Accelerated testing is tricky. ReliaSoft has two good courses on the analysis for it, and I strongly recommend them. They have as a prerequisite the basic life data analysis course, which I also endorse. They are well worth the time and money. Disclaimer: I have taught these courses so I may be biased.

Of course, you have to bring your own knowledge of the product and its physics of failure to the course: they can't be expected to know your product better than you do.

[David]

Hi Nuno,

Given the scenario that you described, you could also use ALTA to determine the life-stress relationship between the constant usage and cycling. This can be done using indicator variables. An example involving this can be found at http://www.reliasoft.com/alta6/a6ex5/index.htm. In your case you would not be able to extrapolate to other stress conditions, but you would be able to determine which scenario is having a greater impact on the life of the product and by how much. Additional information on ALTA can also be found at http://www.reliasoft.com/alta6/index.htm.

I hope this helps.