PDelta Effect

 Posts: 2
 Joined: 25 Jan 2018, 12:12
PDelta Effect
Goodmorning everyone
I have a problem with a planar model of steel MRF .
In particular, in addition to the normal MRF structure, I introduced the leaning columns where I applied: the concentrated masses, the incremental forces (for a PushOver analysis) and the concentrated vertical forces (to take into account the gravity loads for the Pdelta effects). The model workswell with the one exception for Pdelta effects. In fact, there is no degradation structural capacity curve (Base shear vs top displacement), moreover also changing the level of the concentrated force, the results seem to be insensitive.
Do you have any idea what could be the modeling error? For an easier understanding of the problem, attached you will find the model
Thank you
Roberto
P.S. I have already set the geometric nonlinearity.
Dropbox Link:
https://www.dropbox.com/s/aougvq599c2a8 ... r.spf?dl=0
I have a problem with a planar model of steel MRF .
In particular, in addition to the normal MRF structure, I introduced the leaning columns where I applied: the concentrated masses, the incremental forces (for a PushOver analysis) and the concentrated vertical forces (to take into account the gravity loads for the Pdelta effects). The model workswell with the one exception for Pdelta effects. In fact, there is no degradation structural capacity curve (Base shear vs top displacement), moreover also changing the level of the concentrated force, the results seem to be insensitive.
Do you have any idea what could be the modeling error? For an easier understanding of the problem, attached you will find the model
Thank you
Roberto
P.S. I have already set the geometric nonlinearity.
Dropbox Link:
https://www.dropbox.com/s/aougvq599c2a8 ... r.spf?dl=0
Re: PDelta Effect
Roberto, the first thing I note in your model is that there are no end moment releases in the beams. It appears to me that the far right frame is intended to be a moment frame with all other bays intended to be standard shear connections with no moment at the end of the beams.
This is just my assumption since you mention leaning columns.
If my assumption is correct, then you have modeled a much stiffer structure than intended by making all bays moment resisting frames instead of just the rightmost bay. You could release the end moment for all beams not part of the moment frame. Given that this would be 5 of the 6 bays, this would be a dramatic decrease in overall stiffness and would render PDelta effects much more significant.
This could be a starting place. But as I say, I may be misinterpreting the intention of the model. I hope it helps.
This is just my assumption since you mention leaning columns.
If my assumption is correct, then you have modeled a much stiffer structure than intended by making all bays moment resisting frames instead of just the rightmost bay. You could release the end moment for all beams not part of the moment frame. Given that this would be 5 of the 6 bays, this would be a dramatic decrease in overall stiffness and would render PDelta effects much more significant.
This could be a starting place. But as I say, I may be misinterpreting the intention of the model. I hope it helps.
Tim Huff

 Posts: 2
 Joined: 25 Jan 2018, 12:12
Re: PDelta Effect
Dear Tim
First of all thank you for your quick reply, unfortunately I didn't explain my problem well.
The MRF structure consists of 5 bays to which the leaning columns are connected (on the right). The leaning elements are modeled as truss elements without weight, in order to not influence the structural behavior. On the nodes of the leaning columns (connected to the MRF by the diaphragm constraints) I applied the concentrated masses of each floor, the horizontal forces and the vertical forces .
When I run the model, the leaning column works well without influencing the structure and at the same time: the concentrated masses ensure a perfect dynamic behavior, the horizontal forces are perfectly transferred to the model, but the concentrated loads have no influence on the response which makes no sense to me.
This is my problem: why the concentrated vertical loads introduced to model the PDelta effect don't work? Are there any another ways to model the PDelta effect in SeismoStr.?
The only way that gives reasonable results is modelling the elements of the leaning elements with a infrmFB and assign to the elements extremity a “link element” with a very low (0.1) stiffness coefficient in M2 and M3 directions.
Attached you will find the Seismostruct model and few slides to better explain the main model features.
Thank you
Best regards
Roberto
Links:
https://www.dropbox.com/s/aougvq599c2a8 ... r.spf?dl=0
https://www.dropbox.com/s/qd1u5sub8ighw ... .pptx?dl=0
First of all thank you for your quick reply, unfortunately I didn't explain my problem well.
The MRF structure consists of 5 bays to which the leaning columns are connected (on the right). The leaning elements are modeled as truss elements without weight, in order to not influence the structural behavior. On the nodes of the leaning columns (connected to the MRF by the diaphragm constraints) I applied the concentrated masses of each floor, the horizontal forces and the vertical forces .
When I run the model, the leaning column works well without influencing the structure and at the same time: the concentrated masses ensure a perfect dynamic behavior, the horizontal forces are perfectly transferred to the model, but the concentrated loads have no influence on the response which makes no sense to me.
This is my problem: why the concentrated vertical loads introduced to model the PDelta effect don't work? Are there any another ways to model the PDelta effect in SeismoStr.?
The only way that gives reasonable results is modelling the elements of the leaning elements with a infrmFB and assign to the elements extremity a “link element” with a very low (0.1) stiffness coefficient in M2 and M3 directions.
Attached you will find the Seismostruct model and few slides to better explain the main model features.
Thank you
Best regards
Roberto
Links:
https://www.dropbox.com/s/aougvq599c2a8 ... r.spf?dl=0
https://www.dropbox.com/s/qd1u5sub8ighw ... .pptx?dl=0
Re: PDelta Effect
Ok Roberto. That's a bit more clear. Now my second inclination is to wonder about the rigid links. Would you get better results if you eliminated the rigid links and used rigid offset lengths when you define the elements? It seems that might be worth looking into.
Tim Huff
Re: PDelta Effect
Hi Seismosoft team,
I'm aware that Seismostruct can account for global pdelta effects (geometric nonlinearity) but do we need to include the leaning pinned columns with the concentrated loads in our model in order for Seismostruct to account for geometric nonlinearity ?
Thank you.
I'm aware that Seismostruct can account for global pdelta effects (geometric nonlinearity) but do we need to include the leaning pinned columns with the concentrated loads in our model in order for Seismostruct to account for geometric nonlinearity ?
Thank you.
 seismosoft
 Posts: 594
 Joined: 06 Jul 2007, 04:55
Re: PDelta Effect
All SeismoStruct frame elements have by default formulations to account for second order effects (provided of course that such option is selected in the program settings). Note however that second order effects become important in the structural response with large deformations.
Seismosoft Support
Seismosoft Support
Re: PDelta Effect
hi please help
Is it necessary to consider leaning columns (with pinned connection) for pdelta effect?
why somebody use this method?
Many scholars only consider geometric nonlinear effects.
Is it necessary to consider leaning columns (with pinned connection) for pdelta effect?
why somebody use this method?
Many scholars only consider geometric nonlinear effects.
 seismosoft
 Posts: 594
 Joined: 06 Jul 2007, 04:55
Re: PDelta Effect
As mentioned in the previous post, the second order effects are accounted for by automatically, if the corresponding option is selected in the program settings.
Seismosoft Support
Seismosoft Support