Equivalency testing good practice?

I'm with you here...
We have evaluated alternative motor/gearbox assy's for our pumps. I usually end up running them at a high duty cycle...but not too high say 15 secs on every minute... the run-time soon mounts up.
The hard part is accurately logging and recording run times and results due to being distracted by other work .
Suddenly its two months later and you can't find the paperwork and my meticulous record keeping gives a definitive answer as to the suitability.

Del

Well, well..

It must depend upon components' "mechanical" and any other technical or non-technical specifications specified clearly at the first place for the original component.

Question! Is it possible to just comapre all specifactions and will it be possible to determine if a mechanical or electrical part(s) is (are) equivalent theretically and practically◊?

That's part of the problem. There were really no specs for the original part within the system. The part under test is a simple pressure switch that is adjustable from 30 to 150 psi. It is set to 50 psi at our factory and used to monitor the air pressure to the equipment. If it falls below 50 psi, you get a low pressure alarm. When the original unit was spec'd out, the only concern was that it would open when dropping below 50 psi. There was no testing actually done. The only thing they originally did was read a spec. Our quality guys want a detailed report stating the equivalency of the new proposed part versus the original. The requestor of the test has also asked for a life cycle test of 1 million cycles at 60 cycle/min. I feel we should test and validate the new component within the scope of it's operation in the equipment, not some unrealistic cycle test. That's why I want to test a couple of unit in the actual equipment taht can be set to run 24hrs a day, this would give us real world data, and also do a cycle test side by side with the original to a calculated number based on our equipment history. Thoughts?

Yaba-daba,

Have the concerns changed as to application of the part?

Considerable testing is usually done prior to a specification being assigned ya know. Maybe the time intensive procedures are done already?

That's why I want to test a couple of unit in the actual equipment taht can be set to run 24hrs a day, this would give us real world data, and also do a cycle test side by side with the original to a calculated number based on our equipment history. Thoughts?

Perfect, sensible and realistic.
Maybe run a dozen of each side by side (a one off is no good statistically, 30 would be a great number, but rather a lot).

Maybe try and get the set up to cycle the pressure at an accelerated rate to speed up the testing, but don't go too mad. Accelerate a test too much and it soon becomes non representative.

PS If your quality guys want it ...I hope they are doing some of this work. Now't worse that noncontributory quality guys who have no appreciation of what they are actually dealing with.

Del

Yea, I've read that 30 is a good number, but that requires a rather large test fixture and space is a premium in my lab. A sample size of 12 is much better. I usually test components in 10's. I definately do not want to test only one because a failure might just be an anomoly and not representative of all. If I test 10 though, this would produce a common failure mode, if there is going to be one. The fastest that I can accelerate the test is 1 cycle every 3 seconds: 1 sec/pressurize, 1 sec/exhaust, 1 sec delay of timing board. I also need to make sure the air controlling solenoid and counters that I am using will have adequate time to function properly. A 1 second cycle may be too quick, but I havn't tested them yet, but I think 1 sec is probably good. Any faster than that and it may be a problem. And lastly, no, the quality gys are not helping any bit and that's the way I'de rather it be. i do not want them in my lab playing around with my stuff. They will be billed for time and parts however

30 is statistically valid number for Student t test for difference between means between two conditions being tested.

If you can create more 'classes" then a Taguchi L-4 matrix will allow all samples tested to be control for other samples and thus get more info for less testing

normal conditions, normal conditions high temp; Accelerated conditions; accelerated conditions high temp.

just a thought.

milo

To answer this really requires more info on the material, and the application.

If it is metallic, Fatigue and Fracture testing todetermine forces withstood; determine crack growth behavior characteristics, and fracture mechanics- elastic plastic, Linear elastic all come to mind.

Strain or age hardening in the application, as well as Mechanical property comparison to prior part's material are tests that I would think make sense- is the part of a similar order of magnitude in Yield strength (level at which it deforms) or ductility (level of stretch, if you will) be fore it fils.

If shockloads are present, impact testing would be indicated as well.

The plastics folks probably have a selection of their favorite tests for characterization of performance.

Environmental testing for exposure to acid rain, pollutants, UV, or other factors might be indicated. If painted instructions are key part of this, and the lettering fades, you'd want to know that.

On a more practical note, I remember seeing a lawn tractor hooked up with a cable to a center tether with the steering adjusted to pull outwards. They putt 200 pounds of weight on the trailer hitch, and had speed bumps on the track and would gas it up and run it til 600 hours passed.

600 hours was their estimate of the actual life.

One caution, when you do testing, make sure your test reflects what you want to learn.I remember one particularly stupid test scheme for heattreated axles shafts wher they replaced to shear keys with hardened steel and the fiber gears to steel and then whined about the "shafts failing" when the yput 500 pounds on the hitch and popped the clutch from a full scream...

Duh. What did they think was going to happen?

These idiots ended up spending close to $20,000 on outside meatllurgical consultants to tell them that they 'Overloaded' the material, and it was not material fault, but attributable to overriding the designers intentions for the key/keyway and gear.

milo "don't know what to tell you if its made out of rubber glass or stone"

milo