Seismosoft | Earthquake Engineering Solutions

Hi Dear all,

I am performing a dynamic Time-history analysis with SeismoStruct. Before I define the applied loads as an acceleration on the different (ground) nodes on the structure, I am using SeismoMatch to modify the original time history curve (match it to target spectrum, EC8). Everything works very nice so far, but if I now apply the modified Time history curve on the structure to get the acceration for the different storeys and want to get the response spectra, the peak is always far away from the original period T of the structure (T=0,45 s).

I have already used a few modified Time-history curves and the peak of the different response spectra always occurs too late. Here in my case, the maximum spectra occurs at T=0,7s.
Does anyone have an idea, why the peak of the response spectra for different storeys is so late and not in the range of 0,45s?

Thank you very much!

I'm not sure I fully understand the question. Let me see if I can interpret it correctly.

(1) You have a multistory model and apply an accelerogram which has a response spectrum with a peak PSA at T = 0.45 seconds.
(2) You record the acceleration history of each floor as a function of time.
(3) You generate the floor response spectrum from the recorded floor acceleration history.
(4) The peak PSA of the floor spectrum is at 0.7 seconds, rather than at the 0.45 seconds expected.

If I have interpreted incorrectly, perhaps you could elaborate. If my interpretation is correct, perhaps you could elaborate on the type of analysis being performed (linear or nonlinear).

Hi @huffte,
First of all thank you for your quick reply.

(1) 3 storey frame with nonlinear behaviour and lumped masses (=1 for Mx, My and Mz) on the edges on every floor
(example of Tutorial Nr. 4 of the SeismoStruct 2021 -PDF)
and Im not sure if the accelerogram has a peak at T=0,45s?
Anyway I want to apply several accelerograms on the structure.

(2), (3) and (4) Exactly!

I thought maybe the Eigenvalue Analysis contains an error.

Given that you are doing nonlinear analysis, remember that the instantaneous period of the structure is in a state of flux. Yes, perhaps the initial period is 0.45 seconds. Once inelasticity is introduced into the structure, the effective period lengthens. So, I am not convinced that your solution is in error. But You will have to make that decision yourself.

I would suggest trying a linear analysis of the same structure. In such a case, the expectation that the peak of the floor spectra would coincide with the first natural period of the structure is certainly more realistic.

Such an analysis would give you more confidence in the model I think. Best of luck.