RISA 3D Modeling - Shear wall over moment frames

[CCox]

In RISA, I am modeling a wood shear wall sitting on top of a moment frame line below. I am pretty new to using the wood shear wall panel elements, so I am seeing how accurate they are. I created one model with out the wall above and two with the wall above. The model without the wall has the wind and seismic applied directly to the frame lines. The models with the wall has the load applied at the correct diaphragm level. The only difference between the two models with wall above is that one has one shearwall element and the other has 3 separate shear walls. All three models have the same vertical distributed loads applied to the beams. For boundary conditions at the wall panels, I did not choose fixed, pin, or roller. Also, I did not assign a diaphragm at the beam centerline. I attached a picture of the three models.

In looking at the results, forces and moments from purely lateral loads are very similar for all three models. However, the vertical loading is another story. Beam moments and shears are different between the three models. They are higher for the model without the wall. Is the shear wall contributing to the stiffness matrix and carrying some of the load that the beams should be carrying?

 

[jayrod12]

In short, yes. We have run into this issue with masonry walls sitting on steel beams. The stiffness of the wall allows it to arch between columns so the steel beam picks up less load than we had originally expected.

The one way we worked around it was to adjust the stiffness of the wall until the beam was picking up the expected amount of load, but this would require you to have two models, one for lateral with no adjustment to stiffness, and one for gravity.

 

[CCox]

Thanks for the quick response. That makes sense to me, without spending too much more time on it. The best option for me here is to model them separately and superimpose the loads from the shear wall to the frame below.

For simple wall lines where the foundations go to the ground, the shear wall panel seems fairly accurate. Even for multistory perforated or shear force transfer type walls.

I was actually trying to model the entire building that has multiple roof elevations and combination shear wall/moment frames throughout. I was getting some really screwy results because of this problem. I have always been wary of FEA programs (including RISA) where you can create an entire 3D model, get it to converge and yet have very inaccurate results.

If JoshPlum sees this post, maybe he can chime in on this issue. Developing a modeling solution for combination elements like I have within RISA would be helpful. There could be an option to keep the wall stiffness separate from the global stiffness matrix, but separately be solved for determination of shear wall adequacy?

 

[JoshPlumSE]

Yes, Jayrod describes exactly what I expect is happening.

Personally, I'm of two minds about this:
1) Realistically, the wall is probably a lot stiffer than the steel beam. Therefore, any vertical load added after the wall is in place will be supported more by the wall than the beam. Meaning that the FEM results are telling you something useful that may be more accurate in some ways.

2) The way we traditionally design wood walls you prefer to see the vertical load come directly down. Or, at least come down via the studs. Therefore, the FEM analysis isn't what we would design by hand. Also, the FEM results don't account for concepts like staged construction. How, the steel beam is in place before the wood wall is framed or sheathed. Meaning that we have to realize that our traditional calculation methods are still useful maybe even preferred.

Regardless, we should always check our computer analysis work to make sure the results we're getting reasonably agree with what we would have traditionally done. What decision we make based on that check is up to the engineering judgment of every engineer.

Now, if this had been modeled in RISAFloor instead of RISA-3D, you would have seen different behavior. The gravity load from the wall would have come down directly into the beams and they would have been designed essentially the traditional way inside of RISAFloor. However, when the model jumps over to RISA-3D for lateral load analysis, you would still see this "FEM arching" effect with the vertical load. Now, if this leads to larger shear force in some panels (especially when you have openings in the walls) this might be telling you that some segments of the wall may be more susceptible to shear failure than you would have realized without the FEM analysis.

 

[CCox]

Thanks both of you for your feedback.

I will have to look at getting RISAFloor. If our office does more Revit/BIM projects, I could justify the time spent creating a 3D building model in RISAFloor.

Josh, I just saw on other posts you are no longer with RISA. Thanks for your help over the years. It has really been helpful having a technical go to guy on this forum for RISA. Undoubtedly, it was a big loyalty factor that kept many engineers using it. Hopefully RISA understands the benefit of someone providing technical assistance in this forum. I have seen comments from someone named Andrew that seems to be employed by RISA? Is he the RISA tech help guy now?

 

[CCox]

I see my question has already been answered on another post. Andrew is the "dude"