Since ver. 7 is going to be released soon, I wanted to address the following cases lead in "Range Check Error" for further improvement of this great package:
1- Apparently SStruct has a limit of 100,000 of data points for input acceleration (Load Curve). Trying to simulate one of the PEER columns with consecutive ground motions, the 100,000 limit appeared followed by range check error.
2- If trying to create a dynamic load curve by pasting from excel, if relatively large number of points (less than 100,000) range check error is likely to happen.
3- Changing the node color in 3D plot options gives range check error.
4- Trying to paste large number of Analysis Stress-Strain outputs leads to List index out of bounds.
I run simulations on multiple computers and experienced the above mentioned errors.
Many Thanks.
Common range check errors
-
seismosoft
- Posts: 1271
- Joined: 06 Jul 2007, 04:55
Re: Common range check errors
Many thanks Mostafa,
We tried both points 1 and 2 with SeismoStruct v7, we got the error mesage but without the range-check error, hence the bug will appear fixed in the new version. Do you feel that we should increase the 100,000 data points limit?
Regarding points 3 and 4, can it be that a previous range change error caused the problem. And what do you mean with 'large number of Analysis Stress-Strain outputs'?
SeismoSoft Support
We tried both points 1 and 2 with SeismoStruct v7, we got the error mesage but without the range-check error, hence the bug will appear fixed in the new version. Do you feel that we should increase the 100,000 data points limit?
Regarding points 3 and 4, can it be that a previous range change error caused the problem. And what do you mean with 'large number of Analysis Stress-Strain outputs'?
SeismoSoft Support
-
fakharifar.mostafa
- Posts: 124
- Joined: 06 Jan 2010, 05:34
Re: Common range check errors
Thanks for your consideration.
1- Regarding the 100,000 data points it seems adequate.
However, it appeared to me that there is one great feature missing in SStruct. If you want to simulate multi-component earthquakes (like bidirectional earthquake) loading applied to a single column, according to the "Time-History Stages" definition of time steps, you need to use either identical ground motions or adjust multiple ground motions duration in time intervals to have equal time steps (may miss important part of accelerogram)
If you also want to include different angle of earthquake attack, you need to change the node coordinates of the whole structure with regard to the reference (0,0,0) coordinate to include angle of inclination. Still feasible but not really convenient.
The problem number 4 already addressed in "List index out of bounds (#) error" surprisingly is working now (no clue). Although it gives the error but it runs the analysis.
1- Regarding the 100,000 data points it seems adequate.
However, it appeared to me that there is one great feature missing in SStruct. If you want to simulate multi-component earthquakes (like bidirectional earthquake) loading applied to a single column, according to the "Time-History Stages" definition of time steps, you need to use either identical ground motions or adjust multiple ground motions duration in time intervals to have equal time steps (may miss important part of accelerogram)
If you also want to include different angle of earthquake attack, you need to change the node coordinates of the whole structure with regard to the reference (0,0,0) coordinate to include angle of inclination. Still feasible but not really convenient.
The problem number 4 already addressed in "List index out of bounds (#) error" surprisingly is working now (no clue). Although it gives the error but it runs the analysis.
Re: Common range check errors
Just curious about why identical ground motions would need to be used fakharifar.mostafa? I believe that bi-directional loading could easily be accomplished given that typical ground motion recordings come in two orthogonal accelerograms with the same time step and same duration.
With regard to angle of ground motion, I believe it would be easier to rotate the orthogonal records themselves than to rotate the model. The rotation of components could be done in a spreadsheet which permits the user to input an angle and - voila - the rotated components pasted into an new text file and read into SeismoStruct. Granted, it still requires a few steps, but generally easier than rotating the model it seems.
Kind regards.
With regard to angle of ground motion, I believe it would be easier to rotate the orthogonal records themselves than to rotate the model. The rotation of components could be done in a spreadsheet which permits the user to input an angle and - voila - the rotated components pasted into an new text file and read into SeismoStruct. Granted, it still requires a few steps, but generally easier than rotating the model it seems.
Kind regards.
Tim Huff
-
fakharifar.mostafa
- Posts: 124
- Joined: 06 Jan 2010, 05:34
Re: Common range check errors
There is a bug for the "ccs: Composite Circular section" reinforcement.
For the ccs cross section if you try defining the reinforcement properties from the "Additional Reinforcing Bars" dialog box for the ccs cross section, analysis won't run and will receive an error stating "Reinforced ID out of bounds (C:\Users\Santon\...., line 616). Will have to close SStruct and run again.
However, if "Edit Reinforcement Pattern" is used to place the reinforcement by SSTruct, analysis will run fine with no problem. It seems at the moment for the ccs cross section customized reinforcement can't be defined, unless you add ONE reinforcement. Problem does work if you define only One reinforcement in the "Additional Reinforcing Bars".
I just brought up these matters for further improvement in ver 7.0.
For the ccs cross section if you try defining the reinforcement properties from the "Additional Reinforcing Bars" dialog box for the ccs cross section, analysis won't run and will receive an error stating "Reinforced ID out of bounds (C:\Users\Santon\...., line 616). Will have to close SStruct and run again.
However, if "Edit Reinforcement Pattern" is used to place the reinforcement by SSTruct, analysis will run fine with no problem. It seems at the moment for the ccs cross section customized reinforcement can't be defined, unless you add ONE reinforcement. Problem does work if you define only One reinforcement in the "Additional Reinforcing Bars".
I just brought up these matters for further improvement in ver 7.0.
-
fakharifar.mostafa
- Posts: 124
- Joined: 06 Jan 2010, 05:34
Re: Common range check errors
Thanks for the input huffte.
I should've been more specific for the first comment. What I meant was not considering the two orthogonal components of ONE IDENTICAL earthquake. You're totally right in that regard, if the orthogonal EQ components of one record (or even 3D components) are considered. I was asking for different EQ of different sources to be applied in different directions with time intervals between them.
The second suggestion seems to be great, saving lot of headaches [;)]. I wish I had asked this sooner, instead of recreating models many many times. THANK YOU [:)]
I should've been more specific for the first comment. What I meant was not considering the two orthogonal components of ONE IDENTICAL earthquake. You're totally right in that regard, if the orthogonal EQ components of one record (or even 3D components) are considered. I was asking for different EQ of different sources to be applied in different directions with time intervals between them.
The second suggestion seems to be great, saving lot of headaches [;)]. I wish I had asked this sooner, instead of recreating models many many times. THANK YOU [:)]
-
seismosoft
- Posts: 1271
- Joined: 06 Jul 2007, 04:55
Re: Common range check errors
Thank you for your input Mostafa,
We tested ccs section with v7 and everything is correct. We also increased the 100,000 time-steps limit (to 360,000).
SeismoSoft Support
We tested ccs section with v7 and everything is correct. We also increased the 100,000 time-steps limit (to 360,000).
SeismoSoft Support
-
fakharifar.mostafa
- Posts: 124
- Joined: 06 Jan 2010, 05:34
Re: Common range check errors
Simulating repaired frame buildings pushed over to highly nonlinear range, I get range check error in the degrading region of the frame's behavior.
https://www.dropbox.com/s/cljcvxnxtxrc7 ... s.jpg?dl=0
While I have developed the model to degrade, and increased the time steps up to 1000, however, instead of receiving a divergent error message, I get range check error. I know my frame is collapsed at that displacement level, just wanted to inform that there might be some bugs (MAYBE?!). Thanks.
https://www.dropbox.com/s/cljcvxnxtxrc7 ... s.jpg?dl=0
While I have developed the model to degrade, and increased the time steps up to 1000, however, instead of receiving a divergent error message, I get range check error. I know my frame is collapsed at that displacement level, just wanted to inform that there might be some bugs (MAYBE?!). Thanks.
-
seismosoft
- Posts: 1271
- Joined: 06 Jul 2007, 04:55
Re: Common range check errors
Hi Mostafa,
Due to the nature of the non-linear programs the solutions always take place with the presence of the out-of-balance forces, until convergence is reached. In the case of divergence the norm of the out-of-balance forces may reach values up to 1e300! (we have seen this), which causes numerical problems to the computer. Hence SeismoStruct employs the maximum tolerance criterion with a default value of 1e20, which is used to check this norm.
Can it be that you are using a very high maximum tolerance value?
What is the convergence creteria values that you are using? Is it just displacement-based?
Seismosoft Support
Due to the nature of the non-linear programs the solutions always take place with the presence of the out-of-balance forces, until convergence is reached. In the case of divergence the norm of the out-of-balance forces may reach values up to 1e300! (we have seen this), which causes numerical problems to the computer. Hence SeismoStruct employs the maximum tolerance criterion with a default value of 1e20, which is used to check this norm.
Can it be that you are using a very high maximum tolerance value?
What is the convergence creteria values that you are using? Is it just displacement-based?
Seismosoft Support
-
fakharifar.mostafa
- Posts: 124
- Joined: 06 Jan 2010, 05:34
Re: Common range check errors
As you correctly guessed, I set the Convergence criterion using Displacement/Rotation. However, the tolerance is indeed not really high, in order of 0.0001 to 0.1.
By the way, as I said, I know my already damaged frame is collapsed and SStruct simulates the right behavior (no comment on the degradation), I simply reported as thought range check error should not happen, unless there is a bug. Thanks for further improving this versatile nonlinear package.
By the way, as I said, I know my already damaged frame is collapsed and SStruct simulates the right behavior (no comment on the degradation), I simply reported as thought range check error should not happen, unless there is a bug. Thanks for further improving this versatile nonlinear package.