Comparing seismoestruc 2018 vs 2021

04-Unexpected behaviour/errors
crimaval
Posts: 13
Joined: 14 May 2020, 23:30

Comparing seismoestruc 2018 vs 2021

Post by crimaval »

I have to reproduce again a non-linear time history analysis that I did in 2018 with a concrete frame using linear viscous damper but now at 2021 software version. The problem is that I can reproduce the same structure with the new results been 1,25 times the results obtained in the 2018 version on average. In one simulation for example a got a top peak deformation of 0,11m in the 2018 version vs 0,16 m that I obtained at 2021 version. Moreover in the 2021 version the convergence is more difficult to achieve for GFRP RC frames than in 2018 version.
I want to know if it is normal to have this type of differences between the results obtained in both versions.
User avatar
seismosoft
Posts: 981
Joined: 06 Jul 2007, 04:55

Re: Comparing seismoestruc 2018 vs 2021

Post by seismosoft »

This is probably because the dashpot damping element until v2018 had one node, whereas in more recent versions it has two nodes (which is more standard in FE packages). It seems that there should be a problem with the modelling of dashpots.

Seismosoft Support
crimaval
Posts: 13
Joined: 14 May 2020, 23:30

Re: Comparing seismoestruc 2018 vs 2021

Post by crimaval »

The problem is that without a damper (the frame with columns and beams alone) the difference is 25%, however when I apply dampers the difference is 40% on average. To eliminate errors in the model, I import the model from 2018 version to 2021 version. How can be explained the difference in the models without dampers.
crimaval
Posts: 13
Joined: 14 May 2020, 23:30

Re: Comparing seismoestruc 2018 vs 2021

Post by crimaval »

I guess you have a problem with the model for the GFRP Bars, because when I make the simulation with steel reinforcement I do not have convergence problems. However, I use GFRP bar as reinforcement of concrete frames and the convergence problem arise, even when I make sure that the deformation are small enough to do not produce failure of any material.
User avatar
seismosoft
Posts: 981
Joined: 06 Jul 2007, 04:55

Re: Comparing seismoestruc 2018 vs 2021

Post by seismosoft »

What do you mean GFRP bars? You mean rebars modelled with the frp_tl material models? If there is a difference between v2018 and v2021, can you please send the v2018 files to support@seismosoft.com?
Seismosoft Support
crimaval
Posts: 13
Joined: 14 May 2020, 23:30

Re: Comparing seismoestruc 2018 vs 2021

Post by crimaval »

I sent it. Let me know when you replay.
User avatar
seismosoft
Posts: 981
Joined: 06 Jul 2007, 04:55

Re: Comparing seismoestruc 2018 vs 2021

Post by seismosoft »

Hi crimaval,
I ran the file with both v2018 and v2021 Release-3 and got exactly the same results (max. difference of 2e-4 in the displacement of node 50).
Please make sure that you did not change anything (.e.g program settings) in the file after you opened it with v2021.
Seismosoft Support
crimaval
Posts: 13
Joined: 14 May 2020, 23:30

Re: Comparing seismoestruc 2018 vs 2021

Post by crimaval »

Hi, thank you for the support. I check it and you are right, I had a different setting. But now I have another problem. In seismoestruct 2018, I made a simulation of five floors - 3 bays RC frame with a linear viscous damper using the dashpot element. In that version of the software, the handbook said that the dashpot element calculates the force taking into account the velocity with respect to the ground. This was a problem because the viscous damper velocity must be calculated relative to the end of the damper. To solve this problem I assumed that the structure vibrates in the first mode and made a correction in the damping coefficient of the dashpot element. In my case, the damper coefficient was 480 tonnes/sec. I assigned this magnitude to the dashpot on the first floor. however, for the second, third, fourth.. etc I divide the damping coefficient by the factor of 2, 3, 4... respectively.

Now I need to repeat the simulation in the 2021 version of the software using the link element. I tried some simulation and the results are quite different, I obtain 25-30% higher top floor deformation using the link element with the same damping coefficient of 480 tonnes/sec. Did you have any idea how can I model the dashpot element of the 2018 software version using a link element in the 2021 version of the software? I can only work with link element because I also need to obtain results with a nonlinear viscous damper, for comparison propose.
User avatar
seismosoft
Posts: 981
Joined: 06 Jul 2007, 04:55

Re: Comparing seismoestruc 2018 vs 2021

Post by seismosoft »

Hi crimaval,

I am not sure if I know of any efficient way to simulate viscous damping with link elements. I do not know whether if this is possible altogether.

Seismosoft Support
crimaval
Posts: 13
Joined: 14 May 2020, 23:30

Re: Comparing seismoestruc 2018 vs 2021

Post by crimaval »

Hi Seismosoft

In your 2021 seismoestruct, you have the options link element where we can select the curve model for each degree of freedom.
I chose the viscous damper curve that is available in the software option for the x and z direction with the respective damper's coefficient, and use infinite stiffness for the other degree of freedom. Moreover, link element is the normal way software model the viscous damper or not?.

I need to know the difference between the modelling of the dampers in 2018 vs the 2021 version of the software in other to reproduce the result in the 2021 version.
Post Reply

Return to “04-Unexpected behaviour/errors”