In a 3D RC frame I want to model a horizontal diaphragm by using the command rigidDiaphragm. All elements (columns and beams) are modeled with FBE and fiber sections so, in order to avoid experiencing unwanted-unrealistic axial forces (+moments) in FB Elements, I am going to place “artificial” truss elements of high stiffness between the existing ending nodes of the already defined beams.
I also want to define a master node, that will constrain the -in plane- DOF of all nodes of the storey and will be placed in the C.M of the storey (so the total mass of the storey will be concentrated on it) which does not belong to any of the existing elements (so it will be an “isolated” node). Should I restrain the rest DOF (out of plane) of the master node in order not to face any numerical problems?
Has anyone tried to use this kind of assumptions for modeling diaphragmatic action?
Master node-Diaphragm
Moderators: silvia, selimgunay, Moderators
Re: Master node-Diaphragm
yes you need o fix the out out of plane dof .. you also have to consider the rotation of the plane and it's effects on the translation and rotations at the other nodes in the plane. this will not be picked up using your truss solution.
you may be forced to use disp based elements or model the diaphragm with plane stress or plate elements.
you may be forced to use disp based elements or model the diaphragm with plane stress or plate elements.
Re: Master node-Diaphragm
Thank you for your answer.
BTW, as concerns DBE please clarify what i wrote in another post: http://opensees.berkeley.edu/community/ ... =2&t=60480
BTW, as concerns DBE please clarify what i wrote in another post: http://opensees.berkeley.edu/community/ ... =2&t=60480
Re: Master node-Diaphragm
fmk wrote:
> yes you need o fix the out out of plane dof .. you also have to consider
> the rotation of the plane and it's effects on the translation and rotations
> at the other nodes in the plane. this will not be picked up using your
> truss solution.
> you may be forced to use disp based elements or model the diaphragm with
> plane stress or plate elements.
Just as a notice: Why do you think that the plane's rotation and it's effects will not be picked up by using the truss solution? The behaviour of the diaphragm as a disk will be modeled by the constrain trough the rigidDiaphragm command and not by the truss elements. The later ones (i.e the truss elements) are intended to be used only for carrying the axial forces (that will be caused in FB Elements, combined with fiber sections, by the rigidDiaphragm constrain) instead of being carried by the real beams (FBE elements).
> yes you need o fix the out out of plane dof .. you also have to consider
> the rotation of the plane and it's effects on the translation and rotations
> at the other nodes in the plane. this will not be picked up using your
> truss solution.
> you may be forced to use disp based elements or model the diaphragm with
> plane stress or plate elements.
Just as a notice: Why do you think that the plane's rotation and it's effects will not be picked up by using the truss solution? The behaviour of the diaphragm as a disk will be modeled by the constrain trough the rigidDiaphragm command and not by the truss elements. The later ones (i.e the truss elements) are intended to be used only for carrying the axial forces (that will be caused in FB Elements, combined with fiber sections, by the rigidDiaphragm constrain) instead of being carried by the real beams (FBE elements).
Re: Master node-Diaphragm
A good solution is to manually use elements in order to model the diaphragm instead of the command. If let's say you have a rectangular slab, you could use two beam elements that connect the opposite corners, that are very rigid in the axial manner and for bending perpendicular to the slab that you are trying to model (as the rotations of the nodes should be the same, too). The other coefficients should be set to 0. In addition, you could place elements with the same characteristics parallel to the real elements and use ZeroLength elements at the edges of the real ones which set free the dofs that are constrained with the diaphragm members (the diaphragm elements should connect the main nodes directly).
Also, I think that the problem with the axial forces in inelastic members exists for both FBE and DBE, so you have to release the corresponding dofs whichever element you use.
Also, I think that the problem with the axial forces in inelastic members exists for both FBE and DBE, so you have to release the corresponding dofs whichever element you use.
Re: Master node-Diaphragm
Dear Lost,
Thank you for your remarks.
I am aware of the way you suggest about modeling the diaphragmatic action through rigid elements in the axial manner between opposite nodes but there are some obstacles with that: if there is not just one single slab (with a simple rectangular shape) but there are many connected individual slabs and the total shape of plan of the storey is too complex then there are many combinations of crossing types of that artificial rigid elements and it is not obvious which is the proper on in order to model the diaphragmatic action. If the full diaphragmatic action isn't obvious (i.e it is a relenting diaphragm) then i would choose the way you say indeed (using the proper values of axial stiffness for that artificial members).
That what you say about using ZeroLength elements at the edges of beams sounds interesting.
As concerns the axial forces in FBE and DBE i agree with you, as you can also see in the other related post "Diaphragm rigid" from what i am saying in there and fmk did clarified it by his last reply to that post.
The "idea" of placing artificial truss elements between the existing ending nodes of the already defined inelastic elements (beams) was taken from an other post (that was talking about 2Dframes and tried to "translate" it to 3D frames) http://opensees.berkeley.edu/community/ ... agm#p56394
Thank you for your remarks.
I am aware of the way you suggest about modeling the diaphragmatic action through rigid elements in the axial manner between opposite nodes but there are some obstacles with that: if there is not just one single slab (with a simple rectangular shape) but there are many connected individual slabs and the total shape of plan of the storey is too complex then there are many combinations of crossing types of that artificial rigid elements and it is not obvious which is the proper on in order to model the diaphragmatic action. If the full diaphragmatic action isn't obvious (i.e it is a relenting diaphragm) then i would choose the way you say indeed (using the proper values of axial stiffness for that artificial members).
That what you say about using ZeroLength elements at the edges of beams sounds interesting.
As concerns the axial forces in FBE and DBE i agree with you, as you can also see in the other related post "Diaphragm rigid" from what i am saying in there and fmk did clarified it by his last reply to that post.
The "idea" of placing artificial truss elements between the existing ending nodes of the already defined inelastic elements (beams) was taken from an other post (that was talking about 2Dframes and tried to "translate" it to 3D frames) http://opensees.berkeley.edu/community/ ... agm#p56394
Re: Master node-Diaphragm
Although the floor plan may be complex, it could be divided in rectangular or similar shapes, and the members could be placed in each one of them. I think it works and maybe it represents the reality better in those cases, too (for example when the floor plan of a real structure contains holes for stairs or the elevator). I would trust such a method more that a command that you don't know what assumptions it makes, especially for a complex floor plan.
Re: Master node-Diaphragm
So we agree that the method you say is more proper to use in case of not full diaphragmatic action... (in such case the stiffness of the artificial members should also be proper too and not just of "very high" axial stiffness).
My intention of the way i am trying to suggest is when the full diaphragmatic action is (almost) realistic and although the floor plan is "complex" too (e.g when there are beams that are supported indirectly). See for example figure 3 in the attached paper ftp://ftp.ecn.purdue.edu/spujol/Andres/ ... 1-0159.PDF
My intention of the way i am trying to suggest is when the full diaphragmatic action is (almost) realistic and although the floor plan is "complex" too (e.g when there are beams that are supported indirectly). See for example figure 3 in the attached paper ftp://ftp.ecn.purdue.edu/spujol/Andres/ ... 1-0159.PDF
Re: Master node-Diaphragm
That's an interesting paper! I thought that rigidDiaphragm was just a compact way of using many multiple point constraints with a master node, so there wouldn't be such a difference in the way it was utilized (lagrange multipliers or penalty method) and equivalent beams that force those constraints.
So I suggest that someone who knows better how the rigidDiaphragm command works and what assumptions it makes to shed light on this problem...
So I suggest that someone who knows better how the rigidDiaphragm command works and what assumptions it makes to shed light on this problem...
Re: Master node-Diaphragm
The differences of using different kind of constrain commands is another issue that can be discussed maybe later or in an other post.
My request is if the suggested way of modeling a (full) diaphragmatic action could be used and if not why.
I think that fmc's opininon will be usefull.
(The idea of placing artificial truss elements between the existing ending nodes of the already defined inelastic elements -beams- was taken from an other post, that was talking about 2Dframes and i tried to "translate" it to 3D frames, viewtopic.php?f=2&t=33760&p=56394&hilit=truss+elements+diaphragm#p56394)
My request is if the suggested way of modeling a (full) diaphragmatic action could be used and if not why.
I think that fmc's opininon will be usefull.
(The idea of placing artificial truss elements between the existing ending nodes of the already defined inelastic elements -beams- was taken from an other post, that was talking about 2Dframes and i tried to "translate" it to 3D frames, viewtopic.php?f=2&t=33760&p=56394&hilit=truss+elements+diaphragm#p56394)
Re: Master node-Diaphragm
the use of truss with a large stifness is just tthe same as using Penalty method to enforce the constraints. The advantage of the Penalty method for the 3d problem is that it constrains the nodes for the diaphragm action you are looking for (which you will not get with just truss elements on top of beam elements)
Re: Master node-Diaphragm
Thank you again for your reply fmk.
My intention of using these aditional truss elements is not to model any kind of constraint (i.e diaphragmatic action). The ONLY reason of using them is to avoid obtaining the "unwanted" axial forces (as well as the unwanted adittional moments too) that will be caused on nonlinear elements because the later ones are defined with RC fiber sections. I thought that maybe by placing some artificial truss elements only for that reason (that also would not cause any other kind of problem because the full diaphragmatic action is already defined through the rigid Diaphragm command) would "absorb" those adittional and unwanted axial forces from the nonlinear elements.
I suppose that if Zeroelements are defined with an axial release at the edges of beams (i.e inelastic elements) will "leave" the adittional truss elements to carry those axial forces don't they?
My intention of using these aditional truss elements is not to model any kind of constraint (i.e diaphragmatic action). The ONLY reason of using them is to avoid obtaining the "unwanted" axial forces (as well as the unwanted adittional moments too) that will be caused on nonlinear elements because the later ones are defined with RC fiber sections. I thought that maybe by placing some artificial truss elements only for that reason (that also would not cause any other kind of problem because the full diaphragmatic action is already defined through the rigid Diaphragm command) would "absorb" those adittional and unwanted axial forces from the nonlinear elements.
I suppose that if Zeroelements are defined with an axial release at the edges of beams (i.e inelastic elements) will "leave" the adittional truss elements to carry those axial forces don't they?