Design assumptions

[Pennyman]

Lately a few of our powertrain guys and I (suspension guy) were talking about dynamic loading conditions and how to approximate dynamic loadings as static loads with "shock factors" included on such components as bearings, sprockets, pushrods, wheels etc. Anything that would see a significant dynamic load.

So we tried to find some numerical methods to calculate ways to approximate things like maximum vertical inputs from road imperfections, or worst case scenarios with dropping the clutch with the axles held fixed, crazy stuff like that.

We looked at the impulse-momentum theorem, energy conversion, and other methods and came to the realization that there must be a lot of teams out there that are simply assuming a lot of these "max-g" situations. I remember reading 2g lateral, 2g longitudinal, 3g vertical somewhere on the forums...

One of our professors who works for a prominent defense company was explaining that the way they do it in the field is to analyze vibration data (from accelerometers) and then plot the Gaussian distribution (basically a histogram of likely g-loadings) to then design from. They also plot spectral density vs. frequency to make sure they stay away from the natural frequencies of the structure they've built.

I'm sure there's ways to go crazy with this stuff, and I'm not sure if I WANT to see where this rabbit hole goes just yet....but it makes me wonder about how in-depth one could get when designing a vehicle system. I remember hearing from Claude how TUG was looking for someone to focus on FFT for their suspension team, and I was blown away when I heard that.

It also makes me wonder how many assumptions teams are making when it comes to loading conditions on their cars...

[Pennyman]

Lately a few of our powertrain guys and I (suspension guy) were talking about dynamic loading conditions and how to approximate dynamic loadings as static loads with "shock factors" included on such components as bearings, sprockets, pushrods, wheels etc. Anything that would see a significant dynamic load.

So we tried to find some numerical methods to calculate ways to approximate things like maximum vertical inputs from road imperfections, or worst case scenarios with dropping the clutch with the axles held fixed, crazy stuff like that.

We looked at the impulse-momentum theorem, energy conversion, and other methods and came to the realization that there must be a lot of teams out there that are simply assuming a lot of these "max-g" situations. I remember reading 2g lateral, 2g longitudinal, 3g vertical somewhere on the forums...

One of our professors who works for a prominent defense company was explaining that the way they do it in the field is to analyze vibration data (from accelerometers) and then plot the Gaussian distribution (basically a histogram of likely g-loadings) to then design from. They also plot spectral density vs. frequency to make sure they stay away from the natural frequencies of the structure they've built.

I'm sure there's ways to go crazy with this stuff, and I'm not sure if I WANT to see where this rabbit hole goes just yet....but it makes me wonder about how in-depth one could get when designing a vehicle system. I remember hearing from Claude how TUG was looking for someone to focus on FFT for their suspension team, and I was blown away when I heard that.

It also makes me wonder how many assumptions teams are making when it comes to loading conditions on their cars...

[exFSAE]

You don't need to get that crazy.

For one, just using a 2-axis accelerometer to measure combined G-loading is generally just fine for getting your parts strong enough. Alternatively, you could design for the "two-wheeling" scenario.

Plus, on the wheel assembly (wheel, spindle, etc) the stiffness design criterion will sometimes mean you have to significantly overbuild the part from a pure stress perspective.

Regarding frequency analysis for suspensions... there's reasons to do it, it's fairly straightforward, but I wouldn't use it for stress design on these cars. More for vehicle tuning.

[Adambomb]

Yes, it is very easy to get completely overboard. We generally figure out "ideal" loading, then use a factor of safety of 2 to make up for shock loading. In practice that has worked well as long as you don't hit any curbs or design parts with huge stress concentrations. Although for suspension we're stiffness constrained more often than not.

Actually this whole thread brings up a good point that has taken me some time to learn, that when you are strapped for time and cash it is often better to do simple testing than to think about complex testing. We've spent a LOT of time playing around with our high-end cRio data aquisition system, but often there is another solution that could yield just as much useful data from, say, a $20 digital fish scale from Harbor Freight, or a good old zip tie on the shock rod. Begging to use university testing equipment has also proven helpful, especially when you consider there's often a tech that knows how to use it.

Not to say that all out on track data aquisition testing isn't worth it, just that it often pays to consider simpler solutions to save some time, allowing you to do more physical testing. Or maybe I'm just sick of setting up the car for DAQ testing. Now to convince myself to get the car all instrumented again...