Hi there, I am new to modeling in Risa 3d. I have made this model of a small braced frame building. Typically I would model each lateral line in 2D and asses the deflection/strength that way. Our office has recently added RISA 3d so I thought I would model the whole structure. However, I am having difficulty with instabilities issues (common issue I know). I have tried playing around with the boundary conditions and member end releases with no luck yet. I have not yet applied any lateral loads yet either, only gravity. I have attached the file. Any tips or advice would be appreciated. Thanks!
Tek-Tips is the largest IT community on the Internet today!
Members share and learn making Tek-Tips Forums the best source of peer-reviewed technical information on the Internet!
-
Congratulations cowski on being selected by the Eng-Tips community for having the most helpful posts in the forums last week. Way to Go!
New to risa3d instability
- Thread starter EngRon
- Start date
JoshPlumSE
Structural
Rcagar -
It is generally considered poor form to double post the same question in multiple forums. I suggest you delete the thread in either this RISA-3D forum. Or the one in Structural Engineering General Discussion. I will respond further to whichever one still exists.... Once I have more time.
It is generally considered poor form to double post the same question in multiple forums. I suggest you delete the thread in either this RISA-3D forum. Or the one in Structural Engineering General Discussion. I will respond further to whichever one still exists.... Once I have more time.
- Thread starter
- #3
Thanks Josh, I did not realize there was a specific Risa thread, as suggested by another user in my other post. I have deleted the other thread and included your previous response.........
1) The beam column gravity system between column lines b-1 and c-1 is unstable in the X direction. It has zero lateral stability and is not connected to any diaphragm or bracing that could restrain it.
2) Look at member M1, neither of its two end nodes (N1 and N2) are directly restrained in the X direction. not really restrained in the X direction by M3 (at node N4). Put that just means M1 is supported at one location like a teeter totter at a child's playground.
That's all I have time for right now. But, I'll address some of the other issues later.
1) The beam column gravity system between column lines b-1 and c-1 is unstable in the X direction. It has zero lateral stability and is not connected to any diaphragm or bracing that could restrain it.
2) Look at member M1, neither of its two end nodes (N1 and N2) are directly restrained in the X direction. not really restrained in the X direction by M3 (at node N4). Put that just means M1 is supported at one location like a teeter totter at a child's playground.
That's all I have time for right now. But, I'll address some of the other issues later.
- Thread starter
- #5
Thanks for the responses jayrod12 and JoshPlumSE,
A 1.5" deep steel roof deck would be the diaphragm. I did not model it or connect the beam column gravity system to a diaphragm yet as I was trying to get the model working for gravity first. Josh, to your 1) how would I connect the beam column lines on b1 and c1 to the diaphragm? I believe I added a 0.91mm thick plate (in addition to the 1way member area load).
Thanks again.
A 1.5" deep steel roof deck would be the diaphragm. I did not model it or connect the beam column gravity system to a diaphragm yet as I was trying to get the model working for gravity first. Josh, to your 1) how would I connect the beam column lines on b1 and c1 to the diaphragm? I believe I added a 0.91mm thick plate (in addition to the 1way member area load).
Thanks again.
If the nodes at the tops of all of the columns are at the same elevation, you just specify one of those nodes to have a diaphragm in the correct plane. All nodes in that plane would be locked such that they would move together.
In Risa3d, without pairing it with RisaFloor, creates a rigid diaphragm only that essentially attaches all the nodes together via rigid links. You need to keep this in mind when using the diaphragm function and if you have a flexible diaphragm, that it won't properly attribute the lateral loading around.
You wouldn't be able to get the gravity cases to work if your columns aren't appropriately restrained at the top. Again, haven't opened your model yet, but when I have a flexible diaphragm with discreet bracing, often I need to provide lateral restraints at the top of the columns that prevents lateral movement. And then I apply the appropriate lateral loads to the members that are intended to receive them.
For example, a line of columns has one braced bay in the x-direction, I would then provide Z translation restraints to the tops of the columns that would be normally be provided by the flexible diaphragm. Then for each of the beams that is acting as a collector I'd provide a uniformly distributed axial load so that they are properly analyzed as beam columns. Also ensuring that the total expected load going into the bracing is correct.
If you are having tension only bracing, I find you'll have better luck if you specify the members as Euler Buckling, allowing them to take a minor amount of compression without throwing the entire solution out. Otherwise you may have convergence issues as the model will throw out any solution where a tension only member takes any compression load, regardless of how minor.
I should add three more things now that I've opened the model:
1) I never have all members going to a node as pinned. There needs to be at least one member there that has a fixed end, otherwise the program will throw instability warnings. If only a single member has a fixed end, the stability errors go away and when you look at the analysis it doesn't actually end up taking any additional load because all the other members at the node are pinned.
2) I find many instability warnings are a cause of the columns being able to spin on their axis when modelled how you have it. So I also often model all columns as fixed end releases (and all beams pinned unless they are intended on being fixed) and I also will model the boundary condition at the bottom of the column from a true pin, to a translation and y-axis rotation fixed, but leave the x and z axis rotation free.
3) I often will create a load combination of just dead load, with no P-delta analysis. Run this load case and look at the deflected shape. Usually any modelling errors will be abundantly clear just from that deflected shape.
So, My steps to try and correct your model would be as follows.
1) change all column end releases to fixed-fixed
2) change all pins at the bottom of the columns to include y-axis rotation as a reaction. You can look at the end to see if it's restraining any torsion and address that situation as required.
3) provide lateral restraints for the tops of the columns that don't currently have a brace in the line. So along grids b, c, d, e and f I would provide Z translation reaction and at nodes N2 and N14 I would provide X translation reaction.
3.1) In lieu of the reactions at the top of the columns being changed, you could put a diaphragm in place, but that would act as a rigid diaphragm. You would need to ensure your diaphragm performance matches that assumption.
I'd bet once you did those things, your model would be stable and run without even a stability warning (different than a stability error)
In Risa3d, without pairing it with RisaFloor, creates a rigid diaphragm only that essentially attaches all the nodes together via rigid links. You need to keep this in mind when using the diaphragm function and if you have a flexible diaphragm, that it won't properly attribute the lateral loading around.
You wouldn't be able to get the gravity cases to work if your columns aren't appropriately restrained at the top. Again, haven't opened your model yet, but when I have a flexible diaphragm with discreet bracing, often I need to provide lateral restraints at the top of the columns that prevents lateral movement. And then I apply the appropriate lateral loads to the members that are intended to receive them.
For example, a line of columns has one braced bay in the x-direction, I would then provide Z translation restraints to the tops of the columns that would be normally be provided by the flexible diaphragm. Then for each of the beams that is acting as a collector I'd provide a uniformly distributed axial load so that they are properly analyzed as beam columns. Also ensuring that the total expected load going into the bracing is correct.
If you are having tension only bracing, I find you'll have better luck if you specify the members as Euler Buckling, allowing them to take a minor amount of compression without throwing the entire solution out. Otherwise you may have convergence issues as the model will throw out any solution where a tension only member takes any compression load, regardless of how minor.
I should add three more things now that I've opened the model:
1) I never have all members going to a node as pinned. There needs to be at least one member there that has a fixed end, otherwise the program will throw instability warnings. If only a single member has a fixed end, the stability errors go away and when you look at the analysis it doesn't actually end up taking any additional load because all the other members at the node are pinned.
2) I find many instability warnings are a cause of the columns being able to spin on their axis when modelled how you have it. So I also often model all columns as fixed end releases (and all beams pinned unless they are intended on being fixed) and I also will model the boundary condition at the bottom of the column from a true pin, to a translation and y-axis rotation fixed, but leave the x and z axis rotation free.
3) I often will create a load combination of just dead load, with no P-delta analysis. Run this load case and look at the deflected shape. Usually any modelling errors will be abundantly clear just from that deflected shape.
So, My steps to try and correct your model would be as follows.
1) change all column end releases to fixed-fixed
2) change all pins at the bottom of the columns to include y-axis rotation as a reaction. You can look at the end to see if it's restraining any torsion and address that situation as required.
3) provide lateral restraints for the tops of the columns that don't currently have a brace in the line. So along grids b, c, d, e and f I would provide Z translation reaction and at nodes N2 and N14 I would provide X translation reaction.
3.1) In lieu of the reactions at the top of the columns being changed, you could put a diaphragm in place, but that would act as a rigid diaphragm. You would need to ensure your diaphragm performance matches that assumption.
I'd bet once you did those things, your model would be stable and run without even a stability warning (different than a stability error)
JoshPlumSE
Structural
rcgagar -
The only problem with specifying a "rigid diaphragm" is that it is so easy to forget it's there and it will cause the beams at that level to have zero axial force (because the diaphragm is so much stiffer in that plan than the beams).
I would suggest just throwing in a few X braces to halfway represent the diaphragm where you connect some of those gravity only columns back to the lateral resisting frames.... Just to get lateral stability out of the way.
You have a number of other issues as well related to rotation at some joints. Those are much more minor and can be fixed per the suggestions from Jayrod12.
Just to elaborate on item number 2: In a 3D analysis a column that is supported by a "pinned" boundary condition at the bottom and is connected to beams with pinned end is technically unstable for torsion. You and I (as reasonable engineers) know that there isn't any way for torsion to get into the column. But, the program isn't quite smart enough to know that all it sees is a zero in the main diagonal of the stiffness matrix and it wants to find a way to correct that instability. That's why we often add an additional rotational restraint about the global vertical axis (Y in your case).
Note: I corrected a (very) minor issue in Jayrod's response when I quoted him. I changed his reference from "y-axis" to "Y-axis". In RISA, they use the lower case x, y, and z to refer to the local axes of members and plates and such. Then they use the upper case X, Y, Z to refer to the GLOBAL axes of the model. In all my years working there, I never asked why Bruce / Roger did that. It seems to me that it would have made more sense to use X, Y, Z for the global axes then something totally different like 1, 2, 3, or i, j, k for the local axes.
None of us work for RISA anymore, but I still talk to Roger, so maybe I'll ask him one of these days.
The only problem with specifying a "rigid diaphragm" is that it is so easy to forget it's there and it will cause the beams at that level to have zero axial force (because the diaphragm is so much stiffer in that plan than the beams).
I would suggest just throwing in a few X braces to halfway represent the diaphragm where you connect some of those gravity only columns back to the lateral resisting frames.... Just to get lateral stability out of the way.
You have a number of other issues as well related to rotation at some joints. Those are much more minor and can be fixed per the suggestions from Jayrod12.
Just to elaborate on item number 2: In a 3D analysis a column that is supported by a "pinned" boundary condition at the bottom and is connected to beams with pinned end is technically unstable for torsion. You and I (as reasonable engineers) know that there isn't any way for torsion to get into the column. But, the program isn't quite smart enough to know that all it sees is a zero in the main diagonal of the stiffness matrix and it wants to find a way to correct that instability. That's why we often add an additional rotational restraint about the global vertical axis (Y in your case).
Note: I corrected a (very) minor issue in Jayrod's response when I quoted him. I changed his reference from "y-axis" to "Y-axis". In RISA, they use the lower case x, y, and z to refer to the local axes of members and plates and such. Then they use the upper case X, Y, Z to refer to the GLOBAL axes of the model. In all my years working there, I never asked why Bruce / Roger did that. It seems to me that it would have made more sense to use X, Y, Z for the global axes then something totally different like 1, 2, 3, or i, j, k for the local axes.
None of us work for RISA anymore, but I still talk to Roger, so maybe I'll ask him one of these days.
- Thread starter
- #8
Thanks for the quick and clear responses. The column end releases to fixed-fixed was fairly obvious and I should have done that originally. That being said, I tried both methods(Z or X reactions a the col tops (for ones not in braced lines), as well as just adding a rigid diaphragm) as suggested with success. Thanks again.
Nice catch Josh on the y versus Y. Need to proof read better.
Did you apply both the translation restraints and the diaphragm concurrently? If so, your bracing will never see any load. You can do one, or the other, if you still want it to analyze your bracing for you. Only you can determine which is most appropriate for your situation.
I'm glad it worked otherwise.
Make sure you heed Josh's warning about the rigid diaphragm and your beams seeing axial load. If you have a rigid diaphragm turned on in the model, your collectors and drag struts will not attract any load and will require you to check them manually.
Did you apply both the translation restraints and the diaphragm concurrently? If so, your bracing will never see any load. You can do one, or the other, if you still want it to analyze your bracing for you. Only you can determine which is most appropriate for your situation.
I'm glad it worked otherwise.
Make sure you heed Josh's warning about the rigid diaphragm and your beams seeing axial load. If you have a rigid diaphragm turned on in the model, your collectors and drag struts will not attract any load and will require you to check them manually.
- Status
Similar threads