Dear all
i tried most of the solutions said in the forum.i decreased the no. of IS to 3,fibers=50,but still have the fbd-Tol flag during the analysis.
i guess too many rigid links might be the cause.(attached pls find the file)
Any help, really appreciated.
BR
Behnam
http://www.fileconvoy.com/dfl.php?id=g8 ... f114c831bb
Convergence problem in IDA
Re: Convergence problem in IDA
Hi Behnam. Could it be that the convergence problem is a legitimate failure of the structure? Everything seems to run fine for the first IDA analysis. I stopped during the second IDA but it seemed to be running fine as well.
Tim Huff
Re: Convergence problem in IDA
Hi Tim. you have always been really helpful.TNX
you are right .it is ok until the the 9th analysis(3.5g) but what should i do with the results? the analysis is not complete,how can extract the outputs to check them with another structural system .
Also if you don't mind i have other questions relating to my chevron bracings(attached are the files)?
1.i would like to know max displacement when the first yielding happens?how can obtain that?there are 8 braces ?you might say performance criteria?but how can get the max dis at the time of 1st yielding?
2.Also there is a big difference between my Maximum displacement in real(5,8,...meters!!!!) and artificial(0.38,.5meters) records in a certain eg: 1.73g IDA analysis.since i had checked the records via Seismo-match which matched with 7%div to my design spectrum.which brings big inconsistency to the inter-story drift ratios which are based on three limit states.(1.73g,1g,0.5g)
http://www.fileconvoy.com/dfl.php?id=gf ... ab608847c5
Please Help
you are right .it is ok until the the 9th analysis(3.5g) but what should i do with the results? the analysis is not complete,how can extract the outputs to check them with another structural system .
Also if you don't mind i have other questions relating to my chevron bracings(attached are the files)?
1.i would like to know max displacement when the first yielding happens?how can obtain that?there are 8 braces ?you might say performance criteria?but how can get the max dis at the time of 1st yielding?
2.Also there is a big difference between my Maximum displacement in real(5,8,...meters!!!!) and artificial(0.38,.5meters) records in a certain eg: 1.73g IDA analysis.since i had checked the records via Seismo-match which matched with 7%div to my design spectrum.which brings big inconsistency to the inter-story drift ratios which are based on three limit states.(1.73g,1g,0.5g)
http://www.fileconvoy.com/dfl.php?id=gf ... ab608847c5
Please Help
Re: Convergence problem in IDA
1. It seems highly unusual to apply the same curve as an acceleration at the base and as forces at the elevated nodes. Was this really your intention?
2. The file called Artif2 actually has the same loading as Artif1.
3. There is a scale factor of 1 on the applied accelerations. If your accelerograms are in units of 'g', then you should specify scale factors of 9.81 (since your units are set to meters) on the accelerations to apply the full accelerogram.
4. I loaded your accelerograms into SeismoSpect to observe integrated velocities and displacements and response spectra. I encourage you to do the same and observe the differences to convince yourself of the adequacy, or lack thereof, of the accelerograms.
2. The file called Artif2 actually has the same loading as Artif1.
3. There is a scale factor of 1 on the applied accelerations. If your accelerograms are in units of 'g', then you should specify scale factors of 9.81 (since your units are set to meters) on the accelerations to apply the full accelerogram.
4. I loaded your accelerograms into SeismoSpect to observe integrated velocities and displacements and response spectra. I encourage you to do the same and observe the differences to convince yourself of the adequacy, or lack thereof, of the accelerograms.
Tim Huff
Re: Convergence problem in IDA
5. Your real accelerograms have enormous drift at the end of the records and enormous spectral velocities at long periods. This may be part of the problem since, as the structure suffers damage, the effective periods increase (the structure softens). So, while you may have a reasonable match to spectral shape at periods up to about 2 seconds, the various spectra diverge greatly beyond this period. You may wish to re-try your matching but using a longer maximum period for the matching.
6. In your performance criteria, did you intend to specify negative strains for the criteria entitled 'Buckling'? You have entered positive strains as your criteria for both yielding and buckling. Buckling criteria should have negative strain targets.
7. You yielding perfromance criteria specify yield strains of 0.00676 for some of the braces, and 0.01014 for others. If I look at your material properties and divide Fy by E, I get a yield strain of 0.00169, about 1/4 of the lowest value you specified in performance criteria. Is there some reason you set the performance criteria yielding to 4 times the true value?
8. I did a eigenvalue analysis. 70% of the mass participates in the first mode, which has a period of 1.44 seconds. As the structure softens, the structure is likely operating at periods well beyond the 2 second value at which the accelerogram spectra begin to diverge.
6. In your performance criteria, did you intend to specify negative strains for the criteria entitled 'Buckling'? You have entered positive strains as your criteria for both yielding and buckling. Buckling criteria should have negative strain targets.
7. You yielding perfromance criteria specify yield strains of 0.00676 for some of the braces, and 0.01014 for others. If I look at your material properties and divide Fy by E, I get a yield strain of 0.00169, about 1/4 of the lowest value you specified in performance criteria. Is there some reason you set the performance criteria yielding to 4 times the true value?
8. I did a eigenvalue analysis. 70% of the mass participates in the first mode, which has a period of 1.44 seconds. As the structure softens, the structure is likely operating at periods well beyond the 2 second value at which the accelerogram spectra begin to diverge.
Last edited by huffte on 30 Dec 2015, 00:32, edited 1 time in total.
Tim Huff
Re: Convergence problem in IDA
I encourage you to use similar rationales to those outlined here in refining your model. You will learn a lot in the process of doing these things for yourself as you get more comfortable with the software.
Best of luck, Behnam.
Best of luck, Behnam.
Tim Huff
Re: Convergence problem in IDA
Dear Tim.Thanks
i used Seismo-Artif to create 7 records according to my design Spectrum which is EC8(Soil A) and my ag=.25g(PGA=1) .when i check them via SeismoSpect the acc matches but not the velocity & Dis.
Still very large Drift values.Also eg: in file 1 convergance flag in analysis 4!!!!
(About the 4x performance criteria ,they're based on different Limit states.)
Please help me on the records.
any other reviews ,highly appreciated
http://www.fileconvoy.com/dfl.php?id=g4 ... f56ee43dc3
i used Seismo-Artif to create 7 records according to my design Spectrum which is EC8(Soil A) and my ag=.25g(PGA=1) .when i check them via SeismoSpect the acc matches but not the velocity & Dis.
Still very large Drift values.Also eg: in file 1 convergance flag in analysis 4!!!!
(About the 4x performance criteria ,they're based on different Limit states.)
Please help me on the records.
any other reviews ,highly appreciated
http://www.fileconvoy.com/dfl.php?id=g4 ... f56ee43dc3
Re: Convergence problem in IDA
Your artificial records seem to be a reasonable match to the target spectrum, Behnam. Of course, you and your advisor will have to decide if you want to use these artificial records or a set of real records. Your accelerograms have a PGA of about 1.0 g. Your first IDA has a factor of 0.25. Your 11th and final IDA has a factor of 4.5. Is the intent really to apply records having a PGA of about 4.5 g for the final IDA? Maybe so, but that is very high. The only other suggestion I might make is in the damping. You might consider increasing the first mode from 0.7 to 1.5 or so, given that your first mode in your structure is about 1.44.
Tim Huff
Re: Convergence problem in IDA
Dear Tim
indeed i'd like to use real records since they are less demanding.but i can't find the right records which have max less than 10% mismatch according to EC8.
Please help me on this manner?
"www.isesd.hi.is" or "Peer" are the sources,what criteria should search in these websites?
in the seismomatch what should i do in order to match the records to my design spectra since my buildings(knee-moment frame & chevron bracing 4 to 12 story building) periods ranges from .38 to 2.4 seconds?should i change the scale factor ?how much?(would you please introduce me 7 records which will match my design spectra?)
PS:my IDA factors ranges from .25 to 9 in order to get dense data?what should i do about the convergence issue relating to structural collapse?
BR
indeed i'd like to use real records since they are less demanding.but i can't find the right records which have max less than 10% mismatch according to EC8.
Please help me on this manner?
"www.isesd.hi.is" or "Peer" are the sources,what criteria should search in these websites?
in the seismomatch what should i do in order to match the records to my design spectra since my buildings(knee-moment frame & chevron bracing 4 to 12 story building) periods ranges from .38 to 2.4 seconds?should i change the scale factor ?how much?(would you please introduce me 7 records which will match my design spectra?)
PS:my IDA factors ranges from .25 to 9 in order to get dense data?what should i do about the convergence issue relating to structural collapse?
BR
Re: Convergence problem in IDA
I would recommend PEER. Define a target spectrum at their website and the GM selection online tool will let you specify a period range of, say, 0.3-6 seconds. The online tool will pick the top 7 matches to your spectral shape over this range.
You could then take these records from PEER, scale them (PEER will give the recommended scale factors), and use SeismoMatch to get an even tighter for to the target spectrum.
I am not surprised that you have convergence criteria when you try to apply a 4.5g accelerogram. If your 'design' load is 0.25g, why not perform IDA with scale factors of, say, 0.10, 0.20, 0.25, 0.30, 0.35, 0.40, 0.50? You may have reason, but I still don't understand why you put such a severe loading with the high scale factors in IDA.
Besr of Luck, Behnam.
You could then take these records from PEER, scale them (PEER will give the recommended scale factors), and use SeismoMatch to get an even tighter for to the target spectrum.
I am not surprised that you have convergence criteria when you try to apply a 4.5g accelerogram. If your 'design' load is 0.25g, why not perform IDA with scale factors of, say, 0.10, 0.20, 0.25, 0.30, 0.35, 0.40, 0.50? You may have reason, but I still don't understand why you put such a severe loading with the high scale factors in IDA.
Besr of Luck, Behnam.
Tim Huff