dear huffle

i sent the model and its explanation(pdf file). i can find your email address as Tim.Huff@tn.gov. is it ok? could you please check it if arrive?

i sent it by this email address: amir.nayebi86@gmail.com

best regards

happy new year

dear huffle

i sent the model and its explanation(pdf file). i can find your email address as Tim.Huff@tn.gov. is it ok? could you please check it if arrive?

i sent it by this email address: amir.nayebi86@gmail.com

best regards

happy new year

i sent the model and its explanation(pdf file). i can find your email address as Tim.Huff@tn.gov. is it ok? could you please check it if arrive?

i sent it by this email address: amir.nayebi86@gmail.com

best regards

happy new year

- amir_nayebe65@yahoo.com offline
**Posts:**9**Joined:**29 Nov 2013, 13:20

First, I will make a couple of comments which are not related to the solution problem, but which I believe are worth mentioning to you to help improve your model.

1. Your time step in the analysis is 0.01 seconds, while the input accelerogram has a time step of 0.005 seconds. It is good practice to use a time step no smaller than the time step of the input accelerogram and no smaller than one-tenth of the smallest significant natural period. You may want to change the number of steps on the loading stage from 11500 to 23000 in order to make the analysis time step equal to 0.005 seconds to match the input.

2. Your input accelerogram has a large terminal displacement drift at the end of the record. You can identify this using SeismoSignal and inspecting the integrated velocity and displacement. You can also use SeismoSignal to baseline correction to the accelerogram and remove the drift.

As for the solution problem, I was able to run the model by simply changing from the Skyline Solver to the Frontal Solver in the Analysis and removing the damping. You had initial-stiffness-proportional Raleigh damping specified and I changed it to none. Actually, the model will still run with your specified damping. I would encourage you to get into the literature on damping in inelastic time history analysis. There is a good bit of research to suggest that tangent-stiffness-proportional rather than initial-stiffness-proportional damping is more appropriate. I did want to check that it will run with no damping and it does.

I suspect that the large stiffness values of the links in the linear DOFs combined with large penalty factors are the source of the solution instability, though I am not certain. See the SeismoStruct Help section on Project Settings -> Constraints.

Regardless, perhaps a wise course of action would be to run your simpler, 2-storey model using the Frontal Solver and see if the results are close to those obtained for the simpler model using the Skyline Solver. While the Skyline Solver is the default - see Project Settings -> Analysis in the SeismoStruct Help system, this appears to be a case where the Frontal Solver might be useful.

I admit that it does seem strange that the 42nd link seems to be the point at which the solution instability occurs. I did not verify this but take your word for it.

Let me know if the change to the Frontal Solver does the trick for you.

Best of luck, amir.

1. Your time step in the analysis is 0.01 seconds, while the input accelerogram has a time step of 0.005 seconds. It is good practice to use a time step no smaller than the time step of the input accelerogram and no smaller than one-tenth of the smallest significant natural period. You may want to change the number of steps on the loading stage from 11500 to 23000 in order to make the analysis time step equal to 0.005 seconds to match the input.

2. Your input accelerogram has a large terminal displacement drift at the end of the record. You can identify this using SeismoSignal and inspecting the integrated velocity and displacement. You can also use SeismoSignal to baseline correction to the accelerogram and remove the drift.

As for the solution problem, I was able to run the model by simply changing from the Skyline Solver to the Frontal Solver in the Analysis and removing the damping. You had initial-stiffness-proportional Raleigh damping specified and I changed it to none. Actually, the model will still run with your specified damping. I would encourage you to get into the literature on damping in inelastic time history analysis. There is a good bit of research to suggest that tangent-stiffness-proportional rather than initial-stiffness-proportional damping is more appropriate. I did want to check that it will run with no damping and it does.

I suspect that the large stiffness values of the links in the linear DOFs combined with large penalty factors are the source of the solution instability, though I am not certain. See the SeismoStruct Help section on Project Settings -> Constraints.

Regardless, perhaps a wise course of action would be to run your simpler, 2-storey model using the Frontal Solver and see if the results are close to those obtained for the simpler model using the Skyline Solver. While the Skyline Solver is the default - see Project Settings -> Analysis in the SeismoStruct Help system, this appears to be a case where the Frontal Solver might be useful.

I admit that it does seem strange that the 42nd link seems to be the point at which the solution instability occurs. I did not verify this but take your word for it.

Let me know if the change to the Frontal Solver does the trick for you.

Best of luck, amir.

Tim Huff

- huffte offline
**Posts:**744**Joined:**22 Jul 2011, 10:19**Location:**Cookeville, Tennessee, USA

Dear huffle

at first tanke you for helpful comments and Thank you for taking the time .i read your comments and use them in my model. i run a 2D model(2story+3bay) with both solver (skyline and frontal).compare the absolute acceleration of the top floor. they were close to each other. their maximum diffrence was just 0.0005 and its connivance. i have 2 more comments with question:

as u saw my model, i used takeda hystersis model for all linkes element. when i want to define takeda model in M2(moment rotaion response curve for M2 degree of freedom) it needs 5 parameters(My and K and #945;,#946;0,#946;1). i used FEMA356 equations for calculating the yielding rotation of members(chapter 5 of FEMA356-Equation 5-1 and 5-2 for beams and columns).after that calculated yielding moment by this equation My=S.Fy and then K=MY/yielding rotaion.so now we have K and My. is it right to use in takeda parameters ?(K=initial stiffness & yielding rotation=#952;y)

Note: at first I modeld a 2D(3story+3bay) frame in X-Z direction(without linkes.means model has just beams and columns).then I use the eigenvalue analysis the first significant period of structure was 0.80 sec.after that I use linkes(takeda hysteresis model with initial stiffness(K) and yielding moment(My) as discuss them at last paragraph) in my models. Then run eigenvalue analysis again on new model(with linkes) and get the first period of structure about 1 sec. it mean that the period of structure increase when I use linkes(takeda hysteresis model). does sttifness of frame decrease?why??!!!!!. I think the initial stiffness of takeda hysteresis model should be more.but i do not know how to calculate it?

BEST REGARDS

at first tanke you for helpful comments and Thank you for taking the time .i read your comments and use them in my model. i run a 2D model(2story+3bay) with both solver (skyline and frontal).compare the absolute acceleration of the top floor. they were close to each other. their maximum diffrence was just 0.0005 and its connivance. i have 2 more comments with question:

as u saw my model, i used takeda hystersis model for all linkes element. when i want to define takeda model in M2(moment rotaion response curve for M2 degree of freedom) it needs 5 parameters(My and K and #945;,#946;0,#946;1). i used FEMA356 equations for calculating the yielding rotation of members(chapter 5 of FEMA356-Equation 5-1 and 5-2 for beams and columns).after that calculated yielding moment by this equation My=S.Fy and then K=MY/yielding rotaion.so now we have K and My. is it right to use in takeda parameters ?(K=initial stiffness & yielding rotation=#952;y)

Note: at first I modeld a 2D(3story+3bay) frame in X-Z direction(without linkes.means model has just beams and columns).then I use the eigenvalue analysis the first significant period of structure was 0.80 sec.after that I use linkes(takeda hysteresis model with initial stiffness(K) and yielding moment(My) as discuss them at last paragraph) in my models. Then run eigenvalue analysis again on new model(with linkes) and get the first period of structure about 1 sec. it mean that the period of structure increase when I use linkes(takeda hysteresis model). does sttifness of frame decrease?why??!!!!!. I think the initial stiffness of takeda hysteresis model should be more.but i do not know how to calculate it?

BEST REGARDS

- amir_nayebe65@yahoo.com offline
**Posts:**9**Joined:**29 Nov 2013, 13:20

The stiffness of the frame does, in fact, decrease upon the insertion of the links. Even for the linear degrees-of-freedom, the links mean that members which previously had rigid ends, now have "flexible" ends. I place the term "flexible" in quotes because the value of the stiffness specified in the link definition may be quite large compared to frame member stiffness values. But the point is, anything specified is a decrease in stiffness from the non-link model, which would be equivalent to the model with links and stiffness definitions of infinity.

Regarding your use of FEMA356, I am afraid that it must be up to you to determine the appropriate model to use for the connection behavior you wish to mimic in the structure using link properties. It is impossible for me to judge the suitability of your choice in this situation.

Best of luck amir.

Regarding your use of FEMA356, I am afraid that it must be up to you to determine the appropriate model to use for the connection behavior you wish to mimic in the structure using link properties. It is impossible for me to judge the suitability of your choice in this situation.

Best of luck amir.

Tim Huff

- huffte offline
**Posts:**744**Joined:**22 Jul 2011, 10:19**Location:**Cookeville, Tennessee, USA

hello

i need a stiffness degradation in just one of the story of a structure. i model a 2d frame(3story+3 bay). if i want to use the linke element just in one of the stories(forexample story 2 in bay 1).can i use linke element just in that story or i should model the linkes in all the stories?when we use the likne elements, is it force(rule) to use linkes in all the elements?or we can define some members in forexample infrmFB and use linkes elements in just some of the frame members?

tnx

i need a stiffness degradation in just one of the story of a structure. i model a 2d frame(3story+3 bay). if i want to use the linke element just in one of the stories(forexample story 2 in bay 1).can i use linke element just in that story or i should model the linkes in all the stories?when we use the likne elements, is it force(rule) to use linkes in all the elements?or we can define some members in forexample infrmFB and use linkes elements in just some of the frame members?

tnx

- amir_nayebe65@yahoo.com offline
**Posts:**9**Joined:**29 Nov 2013, 13:20

You may certainly place the links only where you want them. Of course, it is up to you to decide where it is appropriate to use the links to model the desired inelastic behavior of the real structure.

Best of luck with the interesting work.

Best of luck with the interesting work.

Tim Huff

- huffte offline
**Posts:**744**Joined:**22 Jul 2011, 10:19**Location:**Cookeville, Tennessee, USA

Dears,

In order to define the damping for gap elements. As this element has not mass we have to use stiffness-proportional damping method.we used C/K as the stiffness parameter; C is damping constant for the dash-pot and k is a linear axial stiffness of the colliding slabs. I have a question, regarding the stiffness proportional damping in the gap model,is it logical when I change initial stiffness with tangent stiffness my model provides different results????? (as my axial stiffness is linear and I used gap hook model)

Thanks,

In order to define the damping for gap elements. As this element has not mass we have to use stiffness-proportional damping method.we used C/K as the stiffness parameter; C is damping constant for the dash-pot and k is a linear axial stiffness of the colliding slabs. I have a question, regarding the stiffness proportional damping in the gap model,is it logical when I change initial stiffness with tangent stiffness my model provides different results????? (as my axial stiffness is linear and I used gap hook model)

Thanks,

- bavafa612006 offline
**Posts:**12**Joined:**04 Dec 2013, 11:25

Users browsing this forum: No registered users and 2 guests