This week, I didn't get a chance to come up with a good CAD related topic to blog about. My schedule just stole my mind share.
But what I did get, was a solid exposure to how the real world works, thanks to my aircraft powerplant classes at Mount San Antonio College.
In the course, I had to perform some "non-destructive testing".
Of course I've seen how the computers do it. I've heard how computer software simulation can "accurately predict how many cycles a component can take before it reaches it's fatigue life".
I've seen heads move up and down knowingly, about how the computer can help make more accurate predictions.
But I also read a real world "Air Worthiness Directive". This document is issued by the FAA when there's an incident that warrants a notification that could affect other operators.
In this case, the directive resulted from an incident, where an aircraft had a propeller "separate from the aircraft" due to crack propagating in the crankshaft!
I don't think that the pilot of that aircraft was thinking "the simulations didn't predict a failure at this time" when he was watching the propeller leave without him!
The result of this? Someone has to perform a real world test.
In class, we performed a "Zyglo" test on non-ferrous parts. In this test, a dye suspended in a penetrating oil is used. It penetrates cracks, and when illuminated with a black light, it glows, revealing the cracks.
Below is a video describing Zyglo.
We tested for pistons, and found cracks in the skirts of three of the four pistons, probably where they had been dropped.
|An example a piston in Zyglo. It's a blurry picture, but there's a crack near the bottom|
|Example of parts undergoing Zyglo test. Courtesy "Safari Helicopter Construction"|
For our ferrous components we conducted a "Magnaflux" test. In this test, a ferrous component is magnetized and an oil with fine iron filings is sprayed on the part. In this case, we had better luck. no cracks in the parts we tested.
|Connecting rods getting magnetized for a Magnaflux test|
|Image of a crack revealed in Magneflux. Courtesy J&M Machine Co|
But there's a lesson here. How much can a simulation predict? Can it predict that a part might get dropped and receive invisible damage? Can it know how many parts might be improperly manufactured? Can it predict that it's own inputs didn't reflect the forces the component would exist in the real world?
No. It can't. That doesn't mean simulation isn't a valuable tool, it's an incredibly valuable tool. But in the end. It's just that. One tool among many.
And it shouldn't be used to the exclusion of other tools.
That's the lesson I learned!