FEA for Wood Shear Walls 2

[E720]

Does anyone have experience running FEA simulations on wood shear walls. I live in a high seismic zone with incredible mountain views ..... which leads to a lot of steel moment frames in residential construction (I recently talked to an architect who basically told me it would be a crime against humanity to cover up those views .... I just needed 8 more inches of wall!). Anyway I figure one way my firm can get ahead of the rest is by limiting the number of residential steel moment frames we design. I don't think that our competitors use FEA so I figure that is one way to get ahead. The analysis may take longer than just using segmented shear walls, but is probably comparable to the amount of time it takes to design a steel moment frame, and it would save home-builders a ton of money, probably tens of thousands. I have some experience programming FEA and IGA (isogeometric analysis) plate and shell elements so I think I could write my own program in c++ with a front end in excel and visualize the results with LaTex maybe. But before I go chase that project here are some questions for the wise elders of the eng-tips community:
[ol 1]
[li]In your humble opinions do you think that by using an FEA simulation I could forego some residential steel moment frames that I otherwise couldn't by using segmented, perforated, or FTAO analysis techniques? [/li]
[li]Does the code allow analysis by FEA? This seems like a silly question but I looked through SPDWS and I couldn't find anything that said I could use another analysis technique that satisfied engineering principles.[/li]
[li]Do you think that I could model a wood shear wall with regular Reisnner-Mindlin plate elements? I feel like that would work better than Kirchhoff-Love plate elements because I could capture shear deformation, which may be significant in a shear wall. [/li]
[li]Any other tips or guidance on the subject. [/li]
[/ol]

 

[WARose]

1. In your humble opinions do you think that by using an FEA simulation I could forego some residential steel moment frames that I otherwise couldn't by using segmented, perforated, or FTAO analysis techniques?

I doubt it. You'd wind up with a lot of stress concentrations that you wouldn't know what to do with. Not to mention other issues I will get to in a moment.

2. Does the code allow analysis by FEA? This seems like a silly question but I looked through SPDWS and I couldn't find anything that said I could use another analysis technique that satisfied engineering principles.

I'm not sure I've ever seen anything in any wood design text or reference on it. There is probably a paper or two out there.....but that doesn't translate into good practice.

3. Do you think that I could model a wood shear wall with regular Reisnner-Mindlin plate elements? I feel like that would work better than Kirchhoff-Love plate elements because I could capture shear deformation, which may be significant in a shear wall.

STAAD uses Mindlin plate theory (at least the last time I checked). That's kind of a hybrid of in-plane and out of plane elements. It gives me good results in most cases.

However for wood shear walls.....there will be BIG issues. For one thing, I doubt a hand calculation for shear wall deflection would ever match the FEA output. Why? There are so many factors influencing the actual results that FEA simply won't catch. (Nail slip to name one.) You'd likely get much more from a hand calc than FEA.

I'd stick with the proven methods.

 

[NorthCivil]

If you are using FTAO, you are ahead of %90 of the pack. Beyond that, you will likely be violating some sort of clause in the wood standard.

my experience with this is that even if you switch to steel moment frames, by the time you dress it out, there are barely any savings in inches to be had vs. if the architect would have allowed 18" for a simpson WSW panel.

Until physics change, there will never be a substitute for depth when it comes to resisting moments generated.

if there is only one wall you need open, you might have some success analyzing the forces with a flexible shearwall-stiff diaghram approach using cantilevered diaghrams.

 

[JoshPlumSE]

I have a different take on this than many. I was heavily involved in RISA's development of a wood shear wall feature for RISA-3D. My thoughts on the subject:
1) You're not going to get much of an advantage for segmented or perforated as it's really just doing the same basic things we all do by hand. Maybe it give you a slightly better force distribution than just looking at wall length. I will say that RISA may make the calculation of perforated walls a little easier.

2) My belief is that RISA's use of FEA for FTAO walls is much more rational and robust than what most people are doing by hand. I'm not sure that this inherently results in needing fewer steel frames though.

3) The more complex the structure the harder it is to use RISA's wood shear walls. I love it when I'm looking at a 2D model of a single force line. It does a great job of sharing load between steel frames and wood walls. But, there is some massaging of the wood wall stiffness that you should probably do (accounting for hold down deformations and such). But, once you get into complex 3D structures, I find it to be "challenging".

4) You do NOT want the program to "optimize" these wood walls for you. Especially if you have a rigid diaphragm. Instead pick a few sheathing an nailing patterns you want to you and manually assign them to the individual walls.

 

[E720]

Thanks for all the tips. It seems like the consensus among those who have responded is that running an FEA simulation on wood shear walls would probably not be able to get rid of a steel moment frame. If that is true that would beg the question - Why would you ever run an FEA simulation on wood shear walls. I know there are a few programs that do this (like RISA). JoshPlum, you mentioned that it more accurately calculates the forces around openings in the case you are doing FTAO. I definitely believe that would be the case. What other benefits would come with using FEA to do shear walls? Why else would someone take the time to do this?

I guess that I still believe that there are some instances where "proven methods" would not allow wood shear walls to work but doing an FEA simulation would reveal that everything was just fine. Maybe I will have to make up some obscure example where that is the case . And I think that with an appropriate mesh and maybe a little higher order basis functions, or even smoother basis functions I could take care of most stress concentrations around the corners of openings (I guess easier said than done).

WARose - I've been lurking on this forum for a long time and really respect your opinion. Would you ever use a program like Risa to do shear walls? Why or why not?

NorthCivil - That is a really good suggestion about the cantilevered diaphragms. I haven't ever done that but I should look into it. It seems like it would solve a lot of my problems.

JoshPlum - What are other benefits of using RISA to do shear walls? Does it take the maximum shear stress in the panel or does it do an average? How does it nail slip in the deflection like WARose was asking about?

 

[WARose]

WARose - I've been lurking on this forum for a long time and really respect your opinion. Would you ever use a program like Risa to do shear walls? Why or why not?

First off: thank you. Secondly, yes, RISA would be fine for such a issue. (I'm just not a personal fan of RISA in general and for other applications.)

 

[JoshPlumSE]

I think there are times where FEA of the shear wall (particularly using RISA) is the ONLY way I could analyze an FTAO wall. The more complex the geometry, the more openings of different sizes between piers of different sizes the trickier it gets to do by hand or even by spreadsheet. You just have to make so many questionable assumptions that I feel much better with RISA's method.

There are plenty of caveats, of course. One thing to mention is that if you've got an opening that goes all the way to the floor you (in my opinion) must put a hold down on either side. Even if it's just a nominal one. RISA doesn't do that for some reason. It's more geared towards window openings.

 

[E720]

Does RISA take into account the top chord, or the stiffness of the studs at all?

 

[JoshPlumSE]

RISA uses Ortho-tropic plate elements to model the wood shear wall. The axial stiffness (i.e. vertical) is based on the studs and stud spacing. The shear stiffness is based on the sheathing and nailing pattern.

I'm not sure if it accounts for the top plate or sill plates at all in the analysis. Probably not. That would be a horizontal axial stiffness of the plate element. My guess is that it's based on the sheathing as well. Not sure why that matters much though. I usually apply my shear load as a distributed shear load at the top of the wall, so the wall doesn't really deform at the top plate.

One thing that can mess it up is when you have a concentrated vertical load on a wall, where you'd put a post in the wall. I believe the RISA help file has some modeling tips related to this. But, in my opinion, that's another case where the more complicated your situation the more challenging it is to use the feature. It can still be done. You've just got to put more effort into it. More effort into modeling it. More effort into interpreting your results to make sure they're right, they make sense, and how to use them for design.

 

[Brad805]

FEA for wood panel design is done at the research level using ANSYS or ABAQUS, but without that type of software do not see much advantage in this level of detail. The accuracy of the model will boil down to your assumptions regarding the panel stiffness. With ABAQUS or ANSYS you can model the studs, sheathing and define slip planes between the sheathing and the studs. Without doing so, I do not see this being any better than hand calcs. Downside of ABAQUS is a license will cost $30k+ and the analysis time is not practical for daily use.

 

[JoshPlumSE]

I feel like I should point out that the purpose of the RISA analysis of wood shear walls was to mimic what engineers are doing my hand. Shear resisted entirely by sheathing, vertical resisted entirely by studs. Moment resisted by force couple from chords / hold downs, et cetera. That's why it gives you the same results for segmented or perforated walls.

The design of FTAO walls is very similar to what was being done by hand. It just doesn't require simplifying assumptions (like a point of inflection at mid-height of each pier) that we would do by hand to solve the problem. It's certainly not trying to do the more complex type of non-linear analysis that Brad805 mentions. It's just trying to use the methods engineers were already using, but in a more generalized way.

 

[E720]

Thanks for all the input. I think that I will most likely not pursue this. The only thing I am getting caught up on is it seems like most people responding to this post are saying "Sure I would use ABAQUS or RISA but I wouldn't trust an engineer to write a program for himself/herself". I have trouble with that attitude because it assumes that there are these wise all-knowing engineering over-lords at big commercial places that know something we "regular engineers" don't. From my experience this is rarely the case and I think that this is a dangerous attitude to have in general (not just in the engineering sphere). I have written beam, truss, frame, plate, shell, and solid FEA programs before and I think that I could write a functional orthotropic plate program but it sounds like it wouldnt really be worth all the work. I really appreciate everyone's input.

 

[XR250]

I think if you are considering FEA for wood construction, you should be looking for a simpler solution.

 

[KootK]

Thanks for all the input. I think that I will most likely not pursue this.

That would absolutely be my recommendation as I see no low hanging fruit here as others have mentioned.

The only thing I am getting caught up on is it seems like most people responding to this post are saying "Sure I would use ABAQUS or RISA but I wouldn't trust an engineer to write a program for himself/herself".

As far as I can see, not a single person has said this. I will though. With all of the nail slip business, potential for panel shear buckling, and the panels acting like individual, rigid bodies rotating about their own centroids, I wouldn't trust your FEM work unless:

1) You did a crap ton of hand calc verifications to validate the programming and;
2) You correlated your results to some full scale testing.

Unless you're mad and/or independently wealthy like Voltaire, I just don't see a lone individual having the personal resources to get there.

Does the code allow analysis by FEA? This seems like a silly question but I looked through SPDWS and I couldn't find anything that said I could use another analysis technique that satisfied engineering principles.

Check out section D.5 of this which seems to endorse FEM as well as some other voodoo: Link. I wouldn't take that as code acceptance per se but it's something.

There's a misconception out in the wild that very complex problems warrant very complex solutions. In my experience, it's usually the opposite. Most highly complex problems require greatly simplified solutions. I would consder wood shear wall behavior to be in this category.


I like to debate structural engineering theory -- a lot. If I challenge you on something, know that I'm doing so because I respect your opinion enough to either change it or adopt it.