Testing, testing, and more testing?

[emerjer]

I was curious as to what kind of tests Teams do validate and justify their design?

Also, is there a testing schedule to go along with the tests or are they more so spontaneous?

What's everyone's thoughts on computer analysis vs. physical testing.

[emerjer]

I was curious as to what kind of tests Teams do validate and justify their design?

Also, is there a testing schedule to go along with the tests or are they more so spontaneous?

What's everyone's thoughts on computer analysis vs. physical testing.

[Jersey Tom]

At this level, do heaps of physical testing. Too many unknowns.. both with regard to mass and inertia distributions, as well as real-life physical behavior, etc.

Depends what you're trying to validate. Most of the time you're tryin to get the car to go faster. Test a baseline, and test with Change X. See if its statistically significant.

For most teams, us included in 07, our testing schedule mostly was just trying to get base and transient fuel maps, our dry sump system working, getting the suspension balanced, and fixing anything that broke or was 'dicey'.

[Steve Yao]

Basically, test everything.

Computer analysis is not a complete replacement for physical testing. CAE helps you make the best use of your testing, but physical testing has to be done to validate the analysis.

Never assume your simulated results are correct unless you have physical test data to corroborate it. Especially in FEA, its easy to input the wrong loads, constraints, mesh size, element type, etc. Just because you get some nice plots and pretty pictures out of an application does not make it correct.

That said, we don't really test the majority of our parts before it goes on the car, but we do our best to have a car running ASAP and put as much time on course as possible before competition.

[Wesley]

I'd like to reiterate what Steve said. Computer analysis lets you do two things - make decisions for which changes to implement and test on an existing design, or establish a starting point for a new design - from which you will test and refine your model to make it match the real world.

It seems the most forgotten thing is that even the very best computer simulation is an approximation - and at that, one based on the number of correct or incorrect inputs it has. A well researched computer simulation will give you a good direction to go - but you have to do your actual development in the real world, with the design as a guideline.

Testing is absolutely critical - frame stiffness, suspension compliance, brake, shock, and engine dyno tests, flowbench tests, impact and stress tests, the list is almost endless.
Every byte of data that you gather about a system in the real world is information you can use to correlate and refine your digital model.

We've learned that you can't run things by the numbers without a bit of testing. Drive pins shearing, stress cracks forming, etc, because to make a calculation manageable, you always leave things out of the equation - and those should be things testing has shown to be insignificant.

[Kurt Bilinski]

And don't forget the human factor. That is, if CAE says the shocks must have a certain valving, yet your ace driver goes faster with different settings, well, there you go.

[B Hise]

<BLOCKQUOTE class="ip-ubbcode-quote"><div class="ip-ubbcode-quote-title">quote:</div><div class="ip-ubbcode-quote-content">hat is, if CAE says the shocks must have a certain valving, </div></BLOCKQUOTE>

are people using PVP curves to simulate the shock?