Continue to Site

Eng-Tips is the largest engineering community on the Internet

Intelligent Work Forums for Engineering Professionals

  • Congratulations waross on being selected by the Eng-Tips community for having the most helpful posts in the forums last week. Way to Go!

Different result obtained from spectrum and transient analysis

Status
Not open for further replies.

mniki2007

Computer
Jul 3, 2007
12
hi dears:

I have a structure and do two seismic analyses with it.
1-spectrum analysis
2-transient analysis

But the thing that is very strange is different result obtained from them.
Stress, nodal displacement, and ......are different.
I expect that at least nodal displacement at top of structure to be the same,
But there are many differences between them, for example:

1-spectrum analysis: maximum x-component of displacement of top node of structure is: 35 cm
2-transient analysis: maximum x-component of displacement of top node of structure is: 3 cm

My seismic data in both analyses are the same and it is El Centro earthquake accelerogram.
But I integrate acceleration to obtain displacement versus time in transient analysis and do Fourier transform to obtain acceleration versus frequency for spectrum analysis.

In transient analysis I set displacement loads at base nodes in x direction and fix y and z directions.
Please someone help me why it happened?
Is it correct and common or I made mistake in some part of my analysis?
Thanks


Regards
 
Replies continue below

Recommended for you

hi
please some one help me to solve this problem.
it is urgent to me.

thanks
 
hi
please some one help me to solve this problem.
it is urgent to me.

thanks
 
hi
please some one help me to solve this problem.
it is urgent to me.

thanks
 
Just a suggestion, maybe also post it in the Mechanical Acoustics/Vibration engineering forum, you'll might get some answers from the vib side.

Tobalcane
"If you avoid failure, you also avoid success."
 
Now that I'm thinking about it, from a vib point of view, a spectrum analysis is different than a transient analysis. Spectrum analysis you will see how the model will react to a band of frequency (0 to 500hz). For transient, you will look at a specific frequency (may be 250 hz), but at different amplitudes (Gs). So, it would make sense that your analysis shows that the larger displacments are in the spectrum analysis because you have "struck" the natural frequency of the model. For the transient, it looks like you are within 3 octavs of the natural frequency.

Tobalcane
"If you avoid failure, you also avoid success."
 
Hi,
the question that arises to me is the following:
in transient analysis you excite only the X-direction; are you sure that, in the spectral analysis, the excitation is also confined to this single direction? Crossed-terms in resonance could otherwise give you a very strong X-response caused, for example, by a Z-excitation.

Other thing: during integration of the time-variant acceleration, did you provide correct values for the integration constants? If you don't, you alter the content of harmonics.

Last thing: did your transient analysis have a sufficient duration to capture all the significant response of the system? Would it be possible that you notice an increasing trend in the X-amplitude, so that you can argue that the max displacement has not been reached yet? Or, vice-versa, is the timestep sufficiently fine in order to capture the highest harmonics of interest? Say you have a peak rise from 1 to 35 mm in 10 ms followed by a decrease to 2 mm in another 10 ms: if you have a delta-time of 20 ms it is very unlikely that you will capture the real peak (except if your timestep is exactly "in-sync" with it, but it's a bit randomic!).

Regards
 
hi and thanks cbrn:

yes,i set Global Y-axis as spectrum direction,but my structure is symmetric, its cross section is circular then i think that there is no diffrent in x or y direction,but in spectrum analysis when i set global y -axis as specrum direction i check y displacement.do u think i have to change spectrum direction to x like transient analysis?

Question 2: my acceleration time data start like this t=0 s -->a=0 ,t=0.02s -->a= 0.003640 i set c1=0 and c2 =0,is it correct?

Question 3: my third frequency is 1.7721 then i use time step size az time=1/20*f =0.28215112and it is Approximately 0.02.is it correct?

thanks
regards
 
Hi,
1- no, I don't think so; if your structure is truely symmetric it should be fine as you're doing.
2- I suppose so; with c1=0 and c2=0 you are saying "initial displacement and velocity are null at start time". Being also a=0, your condition is "at rest".
3- if the max freq you are interested in is a bit less than 2 Hz, one cycle lasts for 0.5 s and so, adequately describing it with at least 10 points/cycle, leading to a timepoint interval of 0.05 s. If you use 0.02 s you should be OK...
The last point to verify is "does your transient analysis last sufficiently"? Do you notice an increasing tendency in the displacement response?

Regards
 
Status
Not open for further replies.

Part and Inventory Search

Sponsor