## General Mechanical

#### Behavior of a structure experiencing a shock

• Rohit2215
Subscriber

Hello Everyone..!!!

I am simulating a structure with a shock of 5g half sine shock wave for a given time like 13 milliseconds.

Is it correct to first analyze the structure in Static Structural Analysis for 5gs of acceleration before performing any transient analysis and is it correct to compare the results of static as well as transient analysis.

In static structural the maximum value of stress is 2465 Mpa and in Transient I am getting different values as I don't know the correct methodology of performing the transient analysis. I have attached some snapshots of what I have done. Please guide me. Thanks to everyone in advance...

• Dave Looman
Ansys Employee
Your methodology is good. For efficiency it would be better to start the acceleration pulse at time = 0, but otherwise everything makes sense. You do a static analysis to determine the non-dynamic response. This provides a good comparison with the dynamic response as you suggest. A modal analysis reveals the natural frequency of the structure. In your case it is relatively low compared to the duration of the excitation so a dynamic response less than the static is expected and the peak response occurs long after the peak acceleration. This is the classic response of a low frequency system to a shock loading.
• Rohit2215
Subscriber
Thanks a lot for your response.
• Rohit2215
Subscriber

• Rohit2215
Subscriber
I have some basic queries
1) What should be the boundary conditions for a structure that is mounted on shock mounts. Is it better to simulate the shock mounts by using the springs in all the three directions with defined stiffness and damping rather than assuming the fixed supports, for dynamic analysis.
2) Is it always desirable to fulfill the CFL condition to calculate the minimum time step or number of sub steps.
• Erik Kostson
Ansys Employee
Hi
2) We use an implicit method in Transient Structural in Ansys workbench mechanical for full transient analysis (not mode superposition like you have) which is unconditionally stable,allowing the use of larger increment time steps. CFL is more related to explicit dynamics for instance in the Explicit Dynamics system and LS-Dyna (explicit solutions).

So for mode superposition transient like you do and for the full implicit (linear no nonlinearities) method in Transient Structural in Ansys workbench mechanical that you use we only need to have a time step that is fine enough to capture the highest frequency (smallest period since T=1/f) of interest. Say in your case from the modes the highest frequency is 30 Hz, then the time step should be one tenth of that smallest period. So the period is (1/30Hz) ~ 0.033s, and thus the time step is equal to 0.003 s . Also I would recommend our course in time integration and dynamics.
As for the boundary, I suppose the fixed supports might give the most conservative (so highest stress) at the base of the structure, but springs might be also possible if the connection stiffness at the base is know which is not always the case unless one e.g., measures it perhaps. So it is up to the user which approach to take, we can not advice exactly which one to use, but that is as we said up to the user and their experience in doing these calculations.
All the best

Erik
• Rohit2215
Subscriber
Thanks a lot Erik for your support. I will surely go through the suggested courses.
• Erik Kostson
Ansys Employee
No worries happy to help .

In your case except of the time step that should be one tenth of that smallest period of interest. So the smallest period is(1/30Hz) ~ 0.033s, and thus the time step is equal to 0.003 s we also need to have a time step that captures your shock, and again we should have a a time step that is say one tenth of the shock period.
In your case the pulse period seems to be about 0.08 s, so 0.003 s is more than enough to capture that schock.
Finally you seem to run the full transient (not mode superposition), so we use the implicit scheme. Have in mind though that if you have nonlinearities since the we need the NewtonRap. iterations for everytime tep, and we might then need a smaller time step than 0.003 to get convergence (so perhaps have substeps maximum say to 10 - or define the minimum time step say as a start one tenth of the main time step of 0.003 s, so for the minimum 0.0003s.
All the best

Erik

• Rohit2215
Subscriber
Ok Got it.
Thanks again Erik.
To add to the discussion the above panel is designed to be mounted in a locomotive. Now a days we came across analysis of products related to defense as well. In which customer specifications provide us the value of shock in terms of Gs and its curve behavior along with the time. Like 57 Gs for 35 mili second full sine wave. Those structures are supported from the bottom as well as from top through the commercially available shock mounts. As per the results provided by our analysis team, our structures demands lot of support members to mitigate the generated stresses and to be within the yielding zone.
However I talked to other manufacturers also but their structure are pretty light weight as compared to ours for the same shock amount and period. I don't understand how but I want to know the correct methodology like selection of time step (solved as above), effect of shock mounts, weather base excitation to be considered or apply acceleration on all bodies, like such other factors which I may not know or our team may be missing to consider for the analysis and we are getting some incorrect results.
I request if you have any solve example of such a problem, please share as that will be a very great help for me.