Seismosoft | Earthquake Engineering Solutions

Hi,
I programmed displacement based fiber frame element at OpenSees and get the same (same numerical results-same numbers after digits) results from OpenSees. I also checked it with Seismostruct. Results were numerically different (and a little bit different). If the theory of OpenSees frame element and Seismostruct element are same you may (not sure!) have an error.
(Thanks for also the Online Course on Seismic Assessment & Retrofitting of Existing RC Structures, it was very valuable).
Regards
Ahmet Alper Parker

We are not aware of the details of the displacement based fiber element of Opensees and we cannot confirm that the theory is the same with our infrmDB element. However, we are pretty confident that SeismoStruct's displacement based fiber element is correct. It has been around for almost 20 years and it has been used by thousands of users.

Note that there are several factors that can differentiate the results between the two programs:
1) The location of integration sections along the length of the member
2) The number and location of monitoring points in each integration section. As far as I know, in Opensees the discretization is with rectangular areas, in SeismoStruct the areas are discretized with triangulation and the monitoring points are triangular.
3) The convergence criteria in inelastic analysis
4) The material models employed by each program.

Seismosoft Support

I strongly recommend you to have a look at the Verification Report that comes with the SeismoStruct installation (it is acessible from the corresponding toolbar button). You will notice that SeismoStruct has done pretty well in numerous blind prediction contests in the past, but you will also see that the examples with the infrmDB element give good results against the experimental values. Having said that and modesty aside, we strongly believe that SeismoStruct is numerically much stabler than Opensees, especially in the highly inelastic range. Moreover, we believe that our con_ma concrete material model is superior to similar concrete models by other competitive programs.

Your comment that the Seismostruct infrmDB element may have an error was probably lightly said.
Anyway, thank you for your comments on the Online Course on Seismic Assessment & Retrofitting of Existing RC Structures.

Seismosoft Support

I am trying to control my model.
While doing, in pushover analysis, the current seismosoft gives this message.
No masses have been defined in any of the three translational directions. SeismoStruct cannot run the eigenvalue analysis
It tries to do eigen analysis. I started the model in eigen. Then I finished the model in pushover. It does not solve the pushover analysis.
Could this be the source of problem? In settings, I tried to do eigen mode number 0, however, it asked for at least 1 eigen.
Thank you very much
Regards

The eigenvalue analysis is required for the calculation of the target displacement of the pushover analysis.
If you uncheck the corresponding checkbox in the Target Displacement module of the pre-processor and choose not to calculate the target dispacement, there should be no problem
Seismosoft Support

Is there a way I can send you files?

I did not say anything heavily or lightly. Anyway, I made sap2000 and Opensees models, they were same for linear elastic and displacement based beam element. Displacement based beam element used no fiber section. So element seems ok for OpenSees. Fiber results are only comparable for OpenSees and Seismosoft, so results are not so much agreeing. In a first step of 10 step static nonlinear analysis, the first step of Seismosoft seems similar to 10 percent of static analysis (Sap2000 and OpenSees), however, results of OpenSees and Seismosoft are not so similar. Maybe yours is ok, OpenSees is not, maybe formulation. Not know. Also, I used fiber sections, not quad, so I was able to compare OpenSees and Seismosoft. Anyway, I do not know where difference is coming.

Hi aaparker,
Opensees is also an accurate package, and to the best of our knowledge the results are very close to SeismoStruct in most of the cases.
Instead of trying to find problems in the accuracy in the two programs, you should try to identify the differences in your two model in Opensees and SeismoStruct. I would recommend to start searching for the differences in the material models employed (including the material properties, especially the strength and the stiffness, but also the hysteretic rules), and in the type of discretisation of the sections in monitoring points.

Seismosoft Support

I had programmed OpenSees's beam (fiber) by myself and get the same results. Do you have a theory manual? Explaining every detail in formulations?

If it is trade secret, do not share it