
Re: MIL901D Shock simulation
Jared Conway Feb 12, 2013 4:53 PM (in response to Bryan Williams)Hi Bryan, what version of SolidWorks Simulation do you have access to? For this analysis you'll need SolidWorks Simulation Premium to run a linear dynamic  time history analysis.
There is a tutorial for a basketball in the software that should show you the process. The big difference is that you'll need to generate your curve in something like excel of the half sine pulse that is 13ms in duration. After the rest of the pulse if you want to watch how it is damped will be set to 0.
Don't forget to check:
1. mass participation
2. time step
3. mode shape capturing
4. mesh convergence

Re: MIL901D Shock simulation
Bryan Williams Feb 12, 2013 5:05 PM (in response to Jared Conway)Thank you for responding. I have generate the modal time history study with the sine pulse curve generated in excel and the study runs but takes a very long time and I have a difficult time getting my mass participation to get anywhere near appropriate. I have included up to 90 frequencies now taking upwards of 4 hours to run the simulation and only getting 5% participation in the shock axis. I have contacted my VAR to see if they can see anything that might be jumping out as an incorrect parameter or something of the sort. One thing that I find interesting is that when I watch the resource useage, it is using around 9% of available resources to solve the simulation...why only 9%?


Re: MIL901D Shock simulation
Mike Pogue Feb 12, 2013 5:47 PM (in response to Bryan Williams)Bryan,
The way we did this before we had sophisticated simulation software was to examine the equivalent static shock. It's pretty easy to compute, though I don't have it in front of me. It is the response as a function of input frequency/natural frequency.
The trick is, it never goes above 2 for an undamped system. So one shortcut is to double the acceleration and then apply the appropriate safety factors and treat the shock as a static load.

Re: MIL901D Shock simulation
Bill McEachern Feb 13, 2013 8:32 AM (in response to Mike Pogue)The whole point in doing a dynamic analysis is to find out what the dynamic amplifications are. These occur due to the frequency content of the input signal inducing resonances with the structure (typically the biggest uncertainty) and the inertially induced components (those with no resonnance but are due to the overall energy into the system. Guessing/estimating some static equivalent will work if say the minimum natural frequecy is sufficiently higher than the highest frequency content with significant energy input. If you do an FFT on the input signal you get a plot of frequency content vs. energy. Then you need to get an estimate of the inertial equivalent to come up withthe static load. Like a zero height impact is known to be double the static load.

Re: MIL901D Shock simulation
Mike Pogue Feb 13, 2013 10:58 AM (in response to Bill McEachern)I agreee with everything your said. But the amplification will never go above 2, with the caveat that the structure is simple enough to be treated as 1 DOF around the frequency of the impulse (f). Even if not, it will not likely go above 2.
If there are multiple modes between 1/2f and 2f, you can compute the transfer for each mode and then multiply them together. The formula is in the shock and vibration handbook, but I don't have it here. It's easiest to use the undamped version, because structural damping is hard to figure out, and I'm pretty sure the result is sensitive to your choice of damping. All you need for this technique is an estimate of the resonant frequencies. You can qualitatively look at the modes to see which are coupled spacially, or you can conservatively assume they are all coupled.
This is using the same equations that drive the FEA simulation, but solving analytically using simplifying assumptions. It is faster to compute the modes of a structure than to compute the frequency response. And the speed of the calculations lets you iterate the design.
I agree that you can get a more sophisticated result using Simulation, if you know what you are doing. But these techniques are used in aerospace with success. If I could avoid a lengthy analysis using a few conservative assumptions, I would. I've attached the ESS mathcad sheet.



Re: MIL901D Shock simulation
Jared Conway Feb 12, 2013 6:24 PM (in response to Bryan Williams)Hi bryan, sounds like it might be pretty stiff in the direction of the shock. It might be impossible to get the recommended amount of mass participation. It is only a guideline. In the end, you may need to accept 5% and then get a physical test done to see if you're close.
Regarding resources, not all processes are multi threaded. So depending on how many cores you have and whether you have hyperthreading on/off and what part of the process you're in, you might be limited by programming.
Mike, if you do dig that information up, I'd be curious to see it. There is a method to do this but I wouldn't consider it a replacement for dynamic analysis. However I have seen people suggest 23x the dynamic load for static as equivalent.

Re: MIL901D Shock simulation
Christian Mackin Sep 27, 2013 9:17 PM (in response to Bryan Williams)Did you ever get this to work? I have the same task. I need to simulate a 901D test on a computer chassis mounted in a rack. I am looking for the SRS plot for 901D to load into SWSimulation. Is Excel the only way to get the plot? Can you post the Excel file you used to make the 901D plot?

Re: MIL901D Shock simulation
Jared Conway Sep 28, 2013 12:49 AM (in response to Christian Mackin)Are you looking for the curve? He mentioned in the first pulse that it was a half sine with a certain pulse width. You'll need to generate the curve in excel to plug into sim. The only precreated input that exists in sim is a full sine.
