I quite agree with where Frank J Bohanan’s article goes in relation to real-world testing. I have found that simulation and modeling works great for generalizing some behavior of failure and/or part wear, and it can narrow down the field of failure greatly to a certain possible set of outcomes. When and if looking for specific failures, actual physical testing proves this, hands down.
Actual real-world testing also provides physical part failures that can be reverse analyzed. That in itself provides a huge amount of data that can be directly applied to the intent for part design, not only for the part that is under test, but also to the environment that the part is utilized in.
As a general example, let’s consider a basic rolling chassis component: an axle spindle. A rolling chassis is not going to roll very far with one falling off. So, a spindle needs to be designed such that it supports the payload of the rolling chassis. Is that payload always static or is it variable? Will the loading be “soft” or “spikey”? Will the spindle see progressive bearing wear or sometimes aggressive bearing wear (depending upon grease application/condition)? What if this spindle sees a pothole or a curb? What if this spindle sees a pothole or a curb at either desert temperatures or ‘cover every piece of bare skin’ winter temperatures?
Now throw in side-loading/cornering variables. Will this be a constant load due to perfect traction? Or are there going to be “vibratory” loads due to running across loose gravel from time to time?
Let’s say the spindle does fail in some way. Did the spindle physically break, wear down in the bearing journals, or bend out of operational specifications? Did the spindle break where it was anticipated? Is the fracture surface indicative of excessive load/wear? Or does the fracture surface indicate a part material issue was at hand? (Now there is a can of worms )
These are a lot of variables to explore. Modeling has made tremendous advances in helping to find multiple end-games to these variables. Combining those end-game results to tailor fit specific customer needs can help with reducing the potential for application liability and warranty claims. Using real-world testing can facilitate the combining of those results and help make for a more robust product design.
Joseph A Van Wanzeele, application engineer, Metalsa
November 19, 2015