Hello,
I have performed a response spectrum analysis in an FE program. It's a cantilever structure, but the geometry is a quite complex. It is a solid model with a mixture of solid/shell/beam element types. OK, so maybe this problem deserves to go in the FE forum, but I think its a generic issue and I would more like advice on the method.
For the response spectrum analysis, the modes are combined using square root sum of the squares. I am happy with the results, stresses, deflections, and reaction loads at the boundary conditions (nodal constraints).
Now here is the problem, there is a solid part of the model that is currently bonded together, at which I would like to evaluate forces/moments, so that a mechanical connection can be subsequently designed in detail by hand calculations. Just to clarify, the bonded connection is working fine in the linear modal analysis. I wanted to calculate the loads across the section, but since the loads are combined by square root sum of the squares, the direction (+/-) is lost. For the first-order modes, across the solid section is a typical bending distribution, positive (tensile) force at one side to negative (compressive) force at the other. When this is squared, everything becomes positive. Then, when combined (squared, summed, square rooted) with the other modes in the spectral analysis the loads are drastically over-calculated. After the combination, the force distribution is quite complex.
So my question is, what is the best way to get forces at a (solid) connection in a response spectrum analysis?
One thought I had would be to make the connection through a single node, but I would prefer not to do this since the area of the section is quite significant.
I guess one solution would be to use a time history analysis, but this would be a more costly option.
Another idea I had, would be to extract the accelerations from the RSA and re-apply them in a static step. I'm not sure if anyone has tried this before, or whether I would still in effect end up with the same problem since the combination is still SRSS.
Maybe another combination method could be used, I am aware of the Newmark combination method (e.g. 1.0 : 0.4 : 0.4 for AX : AY : AZ).
Anyway, thanks for reading. Any suggestions/experience are welcome.
I have performed a response spectrum analysis in an FE program. It's a cantilever structure, but the geometry is a quite complex. It is a solid model with a mixture of solid/shell/beam element types. OK, so maybe this problem deserves to go in the FE forum, but I think its a generic issue and I would more like advice on the method.
For the response spectrum analysis, the modes are combined using square root sum of the squares. I am happy with the results, stresses, deflections, and reaction loads at the boundary conditions (nodal constraints).
Now here is the problem, there is a solid part of the model that is currently bonded together, at which I would like to evaluate forces/moments, so that a mechanical connection can be subsequently designed in detail by hand calculations. Just to clarify, the bonded connection is working fine in the linear modal analysis. I wanted to calculate the loads across the section, but since the loads are combined by square root sum of the squares, the direction (+/-) is lost. For the first-order modes, across the solid section is a typical bending distribution, positive (tensile) force at one side to negative (compressive) force at the other. When this is squared, everything becomes positive. Then, when combined (squared, summed, square rooted) with the other modes in the spectral analysis the loads are drastically over-calculated. After the combination, the force distribution is quite complex.
So my question is, what is the best way to get forces at a (solid) connection in a response spectrum analysis?
One thought I had would be to make the connection through a single node, but I would prefer not to do this since the area of the section is quite significant.
I guess one solution would be to use a time history analysis, but this would be a more costly option.
Another idea I had, would be to extract the accelerations from the RSA and re-apply them in a static step. I'm not sure if anyone has tried this before, or whether I would still in effect end up with the same problem since the combination is still SRSS.
Maybe another combination method could be used, I am aware of the Newmark combination method (e.g. 1.0 : 0.4 : 0.4 for AX : AY : AZ).
Anyway, thanks for reading. Any suggestions/experience are welcome.
