Response spectrum analysis 3

[sachin46]

Dear all,
I am interested in carrying out transient analysis of structure where I have load on model as a "acceleration time history" from shock standard.
I have a question that in such case I have to do analysis as a normal transient analysis or any response spectrum analysis is needed for same? is any dynamics is involved in such problem?. Actually I am a mechanical engineer and not familiar with response spectrum analysis .
I am performing analysis in ANSYS 15.0 mechanical.
I will appreciate if anybody helps me in this regards
Thanking you in advance.

Regards

 

[gravityandinertia]

The type of dynamic analysis chosen usually depends on the input you want to put in, and the output you want to get out. I'm confused because you mention an "acceleration time history" which is typically used for a time history analysis or is converted to a PSD curve for a random vibration analysis. The input for a response spectrum analysis isn't a time history data.

To clear up the confusion, could you post the data you have?

 

[ThomasH]

Hi
Like gravityandinertia said, what data do you have to start with.

You mention a "shock standard", what shock standard do you refere to? Does the standard have any requirements for how you do the analysis?

Response spectrum analysis is common for earthquake engineering. There are methods to simplify the actual calculations but the basis is usually a response spectrum. And the response spectrum is usually created from time-histories based on real earthquakes, they are related. I would say that time-history is for time domain analysis while response spectra is for frequency domain analysis.

Thomas

 

[sachin46]

Dear gravityandinertia & ThomasH,
Thank you very much for your valuable reply.
Actually I am referring one of Naval standard ,where they have given the input as a "velocity pulse" based on weight of equipment.
Considering this I have two methods of doing this analysis. In first, I can convert this velocity into pressure and can perform time domain analysis.
In second method I can convert this into acceleration and used it as a acceleration time history.
For first method I can carry out full transient analysis but for second method I have a doubt how to carry out the analysis.
I will appreciate if you help me for same.
once again thanks for help

Regards,
SACHIN46

 

[gravityandinertia]

If I'm undersstanding correctly, if you are expecting stresses below yield and small displacements, you can perform a modal time history analysis.

If you are going to have large displacements, high stresses, you'll need to perform a non-linear dynamic analyis.

 

[rickfischer51]

For the second method the input is a curve showing your accelerations as a function of frequency. I'm not sure what a velocity pulse is , but if you have velocity vs time date, that could be converted to velocity vs frequency and that could be used as well. I've always had the spectrum supplied to me so I've never had to derive it from raw data, but I believe it comes from a PSD. Also, google DDAM. This method is typically used for shipboard equipment.

Rick Fischer
Principal Engineer
Argonne National Laboratory

 

[ThomasH]

Hi
When I work with responce spectra the data is ususlly based on a structural code. The spectra is then acceleration as function of frequency.
I can also create a response spectra from a time-history, acceleration as function of time. It is made through a transient analysis with Nastran.
Maybe it can be done based on PSD and random theory but I have never done that.

In my experience, if your system is linear, you can use either response spectra or time histories. I have tried (compared) them a few times and the results were comparable. In some cases nearly identical.
If the system is nonlinear you must use a time history. I have seen methods the use response spectra for nonlinear systems but they were approximations. So I have never actually used them for "real" applications.

Thomas

 

[GPapazafeiropoulos]

If the velocity pulse is provided as a function of time and your model is linear, then it would be more economical to perform response spectrum analysis. In this analysis:
(1) You convert the velocity pulse from the time domain to the frequency domain using standard procedures which produce the linear elastic response spectrum (you can find several on the web, and also one implemented in Matlab here:

http://www.mathworks.com/matlabcent...ponse-spectra/content/RESPONSE_SPECTRA/LERS.m

(2) You perform eigenvalue analysis in your structure to extract its most important eigenmodes and corresponding eigenvalues which are independent of the applied loading (either pressure or velocity pulse)
(3) You use the eigenvalues and the response spectrum produced in the two aforementioned steps to find the maximum response of your model for each of the eigenmodes considered in step 2. Then you combine these by a suitable combination method (CQC, SRSS, etc.).

If you have nonlinear model, then you have to resort to dynamic response history analysis (time history analysis) in which the model is analyzed step by step in time. This procedure is much more computationally demanding then response spectrum analysis.

HTH

George Papazafeiropoulos