Dear Tim
I used peer records as you kindly said.
1.could you please check the records?the average misfit is about 4% but the max misfit 21%.(i used the same SF suggested by Peer in seismo-match)
do they satisfy the EC8-1 ?
2.Also regarding the IDA scale factors i would like to have results in 3 different Limit states.(Pls take a quick look at pdf)what would be my Scale factors ?(0.125,.25,.43 ???)
REALLY APPRECIATED
BR
Behnam
http://www.fileconvoy.com/dfl.php?id=g9 ... 76ec3b5c42
Convergence problem in IDA
Re: Convergence problem in IDA
Behnam,
1. One good idea is to use records with scale factors in the range (0.5-4.00), but it is not completely necessary. Match to spectral shape is the single most important factor. Others include magnitude, soil type at recording station, distance from fault, and scale factor. I toyed with some records having magnitude closer to your 6.5 and with smaller scale factors and sent you the scaled and matched versions via your e-mail.
2. It is normal to have variations in spectral ordinates among the records. The average spectrum is the proper basis for comparison to the target more-so than the individual records.
3. I believe your scale factors should be 0.12, 0.24, and 0.4128. I may have missed something, but the paper gives a design PGA of 0.24, not 0.25?
4. You may wish to set up performance criteria based on chord rotation since drift criteria are used in the paper.
Your getting close, Behnam.
1. One good idea is to use records with scale factors in the range (0.5-4.00), but it is not completely necessary. Match to spectral shape is the single most important factor. Others include magnitude, soil type at recording station, distance from fault, and scale factor. I toyed with some records having magnitude closer to your 6.5 and with smaller scale factors and sent you the scaled and matched versions via your e-mail.
2. It is normal to have variations in spectral ordinates among the records. The average spectrum is the proper basis for comparison to the target more-so than the individual records.
3. I believe your scale factors should be 0.12, 0.24, and 0.4128. I may have missed something, but the paper gives a design PGA of 0.24, not 0.25?
4. You may wish to set up performance criteria based on chord rotation since drift criteria are used in the paper.
Your getting close, Behnam.
Tim Huff
Re: Convergence problem in IDA
ِDear Tim
thanks a lot for all your help.
I'm comparing two Structural Systems.1.moment knee frame(in which chord rotation of beams and buckling of knee braces are the key parameters 2.chevron concentrically braced frame(buckling(compression) and tension are key parameters)for performance criteria
i still can't understand A/Ad concept in the paper.if my project peak ground acceleration is ag=0.25g,PGA=1,what would be my project scale factors for the 3 limit states according to that paper(A/Ad=.5,.75,1.75)?also any suggestion for other scale factors in case not to have convergence flags?
These scale factors for the three Limit States are really important to me.because all the results depend on correct scale factors definition.
BR
Behnam
thanks a lot for all your help.
I'm comparing two Structural Systems.1.moment knee frame(in which chord rotation of beams and buckling of knee braces are the key parameters 2.chevron concentrically braced frame(buckling(compression) and tension are key parameters)for performance criteria
i still can't understand A/Ad concept in the paper.if my project peak ground acceleration is ag=0.25g,PGA=1,what would be my project scale factors for the 3 limit states according to that paper(A/Ad=.5,.75,1.75)?also any suggestion for other scale factors in case not to have convergence flags?
These scale factors for the three Limit States are really important to me.because all the results depend on correct scale factors definition.
BR
Behnam
Re: Convergence problem in IDA
The way I interpret the paper is this:
The design PGA is 0.24g. A/Ad is the performance level PGA divided by the design PGA. So at PGA = 0.5*0.24=0.12, the DL limit state proposes to limit interstory drift to 0.75%. At the design PGA = 1.00*0.24=0.24g, the SD limit state proposes to limit residual drift to 0.4%. And at PGA = 1.72*0.24 = 0.4138, the NC limit state proposes to limit drift to 3%. Later on the IDA goes from 0.2 to 1.2 in steps of 0.2 and from 1.2 to 12. Frankly, the paper is somewhat unclear as to why this was done.
Your ground motions were scaled and matched to a response spectrum having a PGA =1.0g. So to assess the DL limit state, the factor would be 0.12. And for SD and NC limit states, scale factors = 0.24 and 0.4138 respectively. However, it is certainly reasonable to include a wide range of scale factors in IDA as long as these three values are included somewhere in the range.
The design PGA is 0.24g. A/Ad is the performance level PGA divided by the design PGA. So at PGA = 0.5*0.24=0.12, the DL limit state proposes to limit interstory drift to 0.75%. At the design PGA = 1.00*0.24=0.24g, the SD limit state proposes to limit residual drift to 0.4%. And at PGA = 1.72*0.24 = 0.4138, the NC limit state proposes to limit drift to 3%. Later on the IDA goes from 0.2 to 1.2 in steps of 0.2 and from 1.2 to 12. Frankly, the paper is somewhat unclear as to why this was done.
Your ground motions were scaled and matched to a response spectrum having a PGA =1.0g. So to assess the DL limit state, the factor would be 0.12. And for SD and NC limit states, scale factors = 0.24 and 0.4138 respectively. However, it is certainly reasonable to include a wide range of scale factors in IDA as long as these three values are included somewhere in the range.
Tim Huff
Re: Convergence problem in IDA
Dear Tim
Tnx.
i seem to have convergence flag at the 2nd analysis(.24g).at this time 17.11 the accelerogram is not demanding at all.Why does it happen?
Record 1:
Time= 17.11002, dt= (Iter: 2 => fbd_Tol)
Also in record 2(0.24g) :
Time= 29.53016, dt= (Iter: 62 => Prd_Ite)
Time= 29.53020, dt= 0.00004 (Iter: 10 => Converg)
Time= 29.53020, dt= (Iter: 4 => fbd_Tol)
Do you have any suggestions?
there are a few points.any reviews?
1.i can not tweak much the Dis/Rot criteria .(not more .0001 & 1e-4)since it is a sensitive issue in dynamic analysis.
2.can i remove some part of long records?(40seconds)e.g:only the first 20s.)in order to save time
3. i thought of reducing the section fiber to 50 ,IS=3 .in a paper it was said 50 fibers would not be a problem however the seismo manual says 100.as simplification.
3.i unchecked stress recovery since we dont need much accuracy in elements internal forces.
4.changed Rayleigh Damping Mode 1=1.33 sec and Mode 2=.49 according to my eigenvalues.2% damping.
5.can not decrease the time step since the record time are long enough.
KINDLY YOURS
Behnam
http://www.fileconvoy.com/dfl.php?id=g3 ... b28fe861e7
Tnx.
i seem to have convergence flag at the 2nd analysis(.24g).at this time 17.11 the accelerogram is not demanding at all.Why does it happen?
Record 1:
Time= 17.11002, dt= (Iter: 2 => fbd_Tol)
Also in record 2(0.24g) :
Time= 29.53016, dt= (Iter: 62 => Prd_Ite)
Time= 29.53020, dt= 0.00004 (Iter: 10 => Converg)
Time= 29.53020, dt= (Iter: 4 => fbd_Tol)
Do you have any suggestions?
there are a few points.any reviews?
1.i can not tweak much the Dis/Rot criteria .(not more .0001 & 1e-4)since it is a sensitive issue in dynamic analysis.
2.can i remove some part of long records?(40seconds)e.g:only the first 20s.)in order to save time
3. i thought of reducing the section fiber to 50 ,IS=3 .in a paper it was said 50 fibers would not be a problem however the seismo manual says 100.as simplification.
3.i unchecked stress recovery since we dont need much accuracy in elements internal forces.
4.changed Rayleigh Damping Mode 1=1.33 sec and Mode 2=.49 according to my eigenvalues.2% damping.
5.can not decrease the time step since the record time are long enough.
KINDLY YOURS
Behnam
http://www.fileconvoy.com/dfl.php?id=g3 ... b28fe861e7
-
seismosoft
- Posts: 1316
- Joined: 06 Jul 2007, 04:55
Re: Convergence problem in IDA
Hi Behnam,
You can also check the following points:
(i) some of the structural nodes are not supported against the out of plane deformation (you can see this in 1-2 of the higher modes)
(ii) Through the equilibrium stages, you can employed adaptable time-steps, i.e. smaller steps when the earquake loading is high and larger when the analysis is not very demanding
(iii) we would try the analyses with more monitoring points per members (say increase 50 to 100) and more integration sections (from 3 ro 4).
(iv) try deactivating geometric nonlinearities or selecting the 'run with elastic properties' checkbox, in order to spot the exact reason for instability
Seismosoft Support
You can also check the following points:
(i) some of the structural nodes are not supported against the out of plane deformation (you can see this in 1-2 of the higher modes)
(ii) Through the equilibrium stages, you can employed adaptable time-steps, i.e. smaller steps when the earquake loading is high and larger when the analysis is not very demanding
(iii) we would try the analyses with more monitoring points per members (say increase 50 to 100) and more integration sections (from 3 ro 4).
(iv) try deactivating geometric nonlinearities or selecting the 'run with elastic properties' checkbox, in order to spot the exact reason for instability
Seismosoft Support
Re: Convergence problem in IDA
Dear SeismoSoft Support team
thank you.
regarding the (iii) i would like to inform that based on paper(page 17/28)"Modelling aspects of the seismic response of steel concentric braced frames,M. D’Aniello,Steel and Composite Structures, Vol. 15, No. 5 (2013) 539-566" there isn't much difference on results whether using 25 or 100 fibers.the main advantage is time saving.
could you please explain (iv),how can i spot the instability?
BR
Behnam
thank you.
regarding the (iii) i would like to inform that based on paper(page 17/28)"Modelling aspects of the seismic response of steel concentric braced frames,M. D’Aniello,Steel and Composite Structures, Vol. 15, No. 5 (2013) 539-566" there isn't much difference on results whether using 25 or 100 fibers.the main advantage is time saving.
could you please explain (iv),how can i spot the instability?
BR
Behnam