Confusing Results for Wall Panels in RISA

[andriver]

Hello ladies and gents,

I decided to post this in the structural section because it has more traffic than the RISA forum...

I am having an issue with some results RISA is giving me for a wall panel analysis I am doing. I will be submitting this question directly to RISA as well and will update this post when feedback is provided. In the meantime I wanted to get y'alls opinion on the matter.

I am analyzing an underground sump for a heavy equipment surcharge load. I have modeled a wall panel in RISA (24.5' tall and 53' Long). I applied a lateral earth pressure load and an equipment surcharge load to the ENTIRE wall panel. These loads are applied as surface loads and are out-of-plane loads to the wall. I run the model and get an applied shear and applied moments (in-plane and out-of-plane). I agree with the moment and shear capacity given by RISA.

Here's where things get funky

I recreate the model but instead divide the model into three equal length wall panels (24.5' high and 17.67' Long). I still apply the lateral earth pressure to each wall panel, however I only apply the equipment surcharge load to the center panel to replicate real-world conditions. So, I am applying 1/3 of the surcharge load that was applied in my first model and the same lateral earth pressure. I would expect to see results showing a lower out-of-plane moment but I get the opposite. My out-of-plane moment has gone up by some 30 kip*ft! I cannot reconcile these results with my logic. I have started from scratch and still end up with similar results...

Do you guys have any opinions, or am I missing something? I have attached a pdf showing both scenarios with the results.

 

[Teguci]

I'm not a RISA man but I am wondering if you are getting any deflection at node 8 for scenario 2. If you are then I'm thinking that adding a pinned boundary condition at node 8 (and the one next to it) should get you the expected results.

 

[BadgerPE]

Agreed with Teguci. In scenario 2 you have panels that are allowed to move and therefore, have more moment instead of fixed-fixed in Scenario 1.

 

[andriver]

Hmm I see what you are saying, but shouldn't my middle wall panel be rigidly attached to the other two adjacent wall panels?

If I add a fixity at those joints, this would not be accurate because the wall should be deflecting slightly at the center, and assigning a fixity there would prevent this deflection.

If this were a beam connection at a joint, I would be able to control moment/translation releases. Is it possible to do this with wall panels? Essentially how do I make the middle wall panel fixed to the adjacent wall panels?

 

[jayrod12]

I'm sure josh plumb will be here shortly but I believe if you mesh your panel into multiple panels you will get the continuity at the different panels. You'll want to have a few nodes along the vertical interface that tie each wall "section" together. I believe the wall panels act like plate elements in that they really only span node to node If your panel only has 4 nodes, it spans between these nodes.

 

[Teguci]

First, solve the presumed modeling problem. Add the pinned connections and verify that the results are expected. Then, to model the 2-way bending, remove the interior pin conditions and subdivide your wall panels into 5 or so elements and see how it behaves. Subdivide into 10 and compare to the 5. If you are approaching the same results add the percentage difference to the 10 panel results and design to that.

 

[andriver]

Ok your suggestions make sense, I am not sure why I expected wall panels to act differently than plate elements.

 

[JoshPlumSE]

I responded to the message you sent to the RISA support account already. But, in the interest of letting other folks know what is happening, I will respond (though more briefly) here as well.

There are two things to review here. 1) Analysis results and 2) Design results.

The analysis results can be seen by looking at the plate force contours and such. I believe these results are going to show you what you expect to see.

The design results, however, are based on assumptions that aren't valid for this model. Essentially, we're averaging the out of plane forces for the entire "region" of the wall. And, you only have one region for the wall. If you split the wall into smaller regions, you would probably get results that are more in-line with the multiple wall panel model.

Though the design results would still not be complete. In this case, the two-way flexural behavior of this wall is something that RISA doesn't handle. RISA only handles one way bending for out of plane walls.... at least as far as presenting code checks and such. Analysis wise, it is handled fine. But, you would have to figure out how to convert that analysis into design forces and code checks.

 

[andriver]

Josh,

Thank you for the quick response on here and via email, kudos to RISA!

I had believed that the regions were auto-meshing so that adding multiple regions didn't add any accuracy.

My goal in using this model is to compute the demand; shear and out-of-plane moment on the wall panel, which then I compare to my hand calculated capacities. Your final comment that I would need to figure out how to convert the analysis into design forces is somewhat confusing. If I created regions 1' wide, would I not be able to use the maximum calculated moment/shear as my design forces?

 

[JoshPlumSE]

The panel does get automeshed, so the regions don't add much accuracy to the ANALYSIS results. But, it's the design results where they become important.

Walls are two dimensional elements, so their forces vary with height and along their width. As opposed to 1 dimensional columns whose results only vary with height.

So, if the wall has 10 kip/ft shear at the center of the wall and 1 kip/ft of shear at the edge of the wall, what is the design force you should use for shear? This is where regions come into play. The design code checks in RISA are done on the regions themselves. So, if the force varies along the width of the wall, you want to make sure the regions are sized in a way that captures the maximum design values over a reasonable representative width.

Though, again, this is only so effective in your case. We'll capture the vertical steel for cantilever type wall bending very well. That's probably good for a center region of some width. But, the 2 way bending towards the sides won't get considered in the RISA code checks or suggested steel.