can long load bearing concrete shear wall be designed not to take lateral forces?

[4dmodeller]

Dear all,
I was wondering if you can help shed some light in my dilemma about Etabs as a lateral stability analysis tool.
Previously under my first company, I used to do a full etabs model and extract individual wall pier forces for design in a manual spreadsheet. This was all clear and is similar to how Etabs online tutorial have suggested.

At my new company, They have a design methodology that is radically different. They asked me to only draw lateral resisting shear wall (neglecting columns and other walls that may also long but are transferring and does not reach ground/ or precast wall that has minimal horizontal joint dowel connections - and thus assume not to be rigid enough to take lateral load in ultimate condition)+assign slab as membrane element with mesh option of rigid diaphragm so that it does not transfer any vertical load down the wall.
In effect, they want to extract lateral forces out of Etabs only and combine it with manual hand calc load rundown for design. Group of wall are grouped as single pier and designed as a box etc.

I was wondering if there's anything wrong in the second method?
Things that concern me about this approach is that
1. I found that despite no load transfer from slab (due to mesh option described earlier) I was still getting significant bending moment in the pier from self weight which doesn't seem right. I feel that the geometry of the slab that's connected to the wall are what causes the bending moment. But why would there be much if there's no load transfer between them?
2. can long load bearing concrete shear wall be designed not to take lateral forces? I feel that minimum connection to slab alone would cause the wall to take lateral load. perhaps it will fail and the load redistribute to other walls that are designed to take load? is that an acceptable way of design?

Sorry for the long message. Thanks all for your opinions

 

[msquared48]

I can think of a couple of details where that would be possible. So, yes.

Mike McCann, PE, SE (WA)

 

[KootK]

2. can long load bearing concrete shear wall be designed not to take lateral forces? I feel that minimum connection to slab alone would cause the wall to take lateral load.

I agree with your assessment. Certainly it's possible to create connection details that wouldn't transfer lateral shear. Those details are not at all common, however, and they would often be costly.

perhaps it will fail and the load redistribute to other walls that are designed to take load? is that an acceptable way of design?

Not in my opinion unless:

1) the walls that are expected to fail are made to fail in a ductile fashion that does not compromise their ability to support gravity loads etc and;

2) the walls' effect on load distribution prior failure is accounted for in the analysis and design of other LFRS elements (torsional irregularity etc).


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.

 

[chekre]

I used to do the following :
Let say i have 5 shear walls.
One of the wall is failing in shear. I used to assign modifiers for the GA compound for the failing wall in order to take a shear less or equal to its Vn and the remaining shear will be resisted by the other walls which are safe versus shear.
However, Kootk comment is very interesting regarding the torsional irregularity...

 

[4dmodeller]

I tend to agree with Chekre.
All walls must fail in shear before the building will fail. Hence, you can technically make some walls weak (by making them precast with minimal connection). These wall will barely take any load and other stiffened walls e.g. core can take majority of the load.
This approach would mean that no additional costly shear detailing would be required.

do you agree?

torsional irregularity is an interesting problem though. However, you are designing the lateral system for the ultimate condition. Load distribution prior to that is not of concern. do you think there is a major problem with this approach?

 

[chekre]

Yes i agree.
Moreover, if one of the resisting walls is failing, increasing its thickness or its length will not solve the problem since its Vn will increase due to the revised dimensions but since the Wall is stiffer, he will attract more load (axial, shear, bending) and your problem will persist maybe.

Regarding the torsional irregularity thing, you should recheck the whole system based on the final cracked model (where all your elements are safe versus the different forces) and determine in which category your building is.

 

[wannabeSE]

I have a couple thoughts assuming the precast wall connections fail in shear
1. The strength of the failing elements should not be included in the lateral analysis (unless a detailed analysis is performed to determine if any residual capacity at the design displacements can be justified - I am not sure but ASCE 41 might have some information)
2. What are the consequences if the precast connections fail. Will the wall fall and kill someone or prevent people from safely evacuating the structure. If the precast is designed to carrying gravity loads, will a collapse occur if the connections or walls fails?

 

[4dmodeller]

my understanding is if the connection that fails is the top/ bottom dowel connection, the panel itself doesn't fail. Hence, the panel should still be ok to take gravity load.

any thoughts on that?

The strength of the failing elements should not be included in the lateral analysis (unless a detailed analysis is performed to determine if any residual capacity at the design displacements can be justified - I am not sure but ASCE 41 might have some information)

this is why i've asked whether it is fair to create a model without these weak wall, and assign mesh as rigid diaphragm (does not transfer any gravity load). Thus, creating a lateral force only model. It is a much quicker model to create.

This approach doesn't allow me to calculate bending and shear due to eccentricity of gravity force properly though. does anyone know how to fix this issue? any thoughts?

 

[chekre]

U can model your weak walls using a frame section not the pier definition.
Release your frame at the top so they act as members that can support gravity loads only and create a rigid link through your wall length.
For example if ur wall is 8m long with a 25 cm as thickness, create a frame having the above properties and create a rigid link which u will draw it in the initial 8m length of shwar wall.
Dont know if it is clear

 

[4dmodeller]

Chekre,

what is the point of rigid link? so that wall act as one? it does not really matter if you are going to draw a series of frame with section property of a wall (perhaps 2m wide - so space frame out at 2 m c/c)...however,not too sure what you mean? please clarify.

 

[KootK]

torsional irregularity is an interesting problem though. However, you are designing the lateral system for the ultimate condition. Load distribution prior to that is not of concern. do you think there is a major problem with this approach?

Regarding the torsional irregularity thing, you should recheck the whole system based on the final cracked model (where all your elements are safe versus the different forces) and determine in which category your building is.

I'm not sure that I agree with these statements. We often employ 3D non-linear modelling when serious torsional irregularities exist. One of the reasons for that is simply because, when serious torsional irregularities are present, we have a hard time understanding the dynamic behavior of our structures. Consider that, up until the discounted walls "disappear" analytically, the following may be true:

1) Your building may be much stiffer than you assumed in your analysis.

2) Your building may attract more seismic load in response to a smaller event than you assumed in your analysis.

3) The first fundamental mode of vibration may actually be a torsional mode as opposed to a lateral, shear building sway mode (the assumption underpinning ELF procedures).

4) Conceivably, some members that attract little to no loading in your ELF analysis may see significant load prior to the disappearance of the discounted wall.

In essence, your lateral design may be a complete farce right up until the building undergoes enough seismic motion that that your breakaway components fail and, hopefully, the building starts to resemble the structure that you modeled in your design. Is that a deal breaker? Perhaps not. Does it have you wading nipple deep in murky seismic territory? You bet it does.

Some additional things to consider:

1) Because of point #4 above, all members should be capacity designed for at least the load that they may see up until the discounted wall disappears. And that raise the question of just what is the reasonable expected shear capacity of a wall? Expected over strength flexural capacity for walls is set out in many codes. I don't know that I've ever seen such a capacity defined for walls yielding in shear however.

2) A long shear wall is often a squat shear wall. And squat shear wall would need to a) yield it's vertical reinforcement and b) overcome any applied gravity loads in order to top out with regard to shear capacity. Defining the point in the load history where that would reliably occur is no trivial matter. This would be a good reason to keep the yielding to the panel connections rather than within the panels themselves.

3) If you're going to attempt to construct breakaway connection in your precast wall panels, perhaps it would be cleaner and easier to expend that same effort/expense in simply trying to create connections that allow lateral movement per your original post.

4) Wind. Your breakaway connections would still need to be at least strong enough to keep them from "failing" under wind loads I would think. And the rest of the structure would need to be designed for wind load actions assuming that the unintended shear wall under consideration was in tact and contributing to stiffness.

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.

 

[chekre]

Kootk so if i understand u arent a fan of reducing the shear Vu on a wall and redistribute it to other members ?

 

[KootK]

Kootk so if i understand u arent a fan of reducing the shear Vu on a wall and redistribute it to other members ?

I'm skeptical to be sure. It seems to me that you're asking your wall to form a "plastic shear hinge" such that the shear load / lateral deformation curve of the wall would be effectively bi-linear, as we would assume with shear walls that yield in flexure. Generally, in seismic design, one of our goals is to preclude concrete shear failures.

To be sure, however, I feel that your method is a substantial improvement over disregarding the problem wall altogether.


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.

 

[4dmodeller]

KootK

1) Your building may be much stiffer than you assumed in your analysis.

2) Your building may attract more seismic load in response to a smaller event than you assumed in your analysis.

3) The first fundamental mode of vibration may actually be a torsional mode as opposed to a lateral, shear building sway mode (the assumption underpinning ELF procedures).

4) Conceivably, some members that attract little to no loading in your ELF analysis may see significant load prior to the disappearance of the discounted wall.

No.1 - i agree
No. 2 - i agree, stiffer building mean lower natural period and higher base shear
No. 3 - this is the point i am trying to prevent. I know that you dont want torsion as your first mode (i was told by a senior engineer)..however, i do not know why? can you guide me to something that can explain it? or if you can it would be great. By making some wall disappear, i can make the building first mode a lateral shear sway.
No. 4 - I agree.
I wanted to note that I do not at all agree with this methodology but i've been asked by senior staffs/ directors in my new office to do so. It may be argued that by deleting some wall, you are making your "lateral resisting element" stiffer (and thus a conservative design for those elements)..however, if some of the "disappearing wall" does terminate into a transfer ..you are in effect underdesigning that transfer structure...i do not dispute that.

you are right..i dont know if i have heard of ductile shear yielding...but i don't agree that you can shed shear load to other lateral element...if one of them reach yield point...that is like a bolt connection..first row of bolt takes more than others but we still assume equal distribution.

2) A long shear wall is often a squat shear wall. And squat shear wall would need to a) yield it's vertical reinforcement and b) overcome any applied gravity loads in order to top out with regard to shear capacity. Defining the point in the load history where that would reliably occur is no trivial matter. This would be a good reason to keep the yielding to the panel connections rather than within the panels themselves.

this is why this method is applied to precast wall where vertical dowel connection can be nominal..and what fails is the connection...however, you are right..the point that this happens is trivial in nature.
If you break long panel into pieces and don't stitch the precast panel together however, the stiffness of that wall won't be as big as if they are all stitched to work monolithically. this may help make the argument more valid.

3) If you're going to attempt to construct breakaway connection in your precast wall panels, perhaps it would be cleaner and easier to expend that same effort/expense in simply trying to create connections that allow lateral movement per your original post.

what are you referring to here? which original post?

4) Wind. Your breakaway connections would still need to be at least strong enough to keep them from "failing" under wind loads I would think. And the rest of the structure would need to be designed for wind load actions assuming that the unintended shear wall under consideration was in tact and contributing to stiffness.

are you talking about global wind force? i believe that is the same as an EQ stability design..the disregarded wall can fail and go to the "lateral stability element". Again, i want to stress that i do not agree wholeheartedly with this method (disregarding wall and all). but i want to somehow find a way to justify that it is a reasonable way..in order for me to be able to do what i've been asked to do without feeling like the whole thing is all wrong..

 

[chekre]

Ok Kootk. Thank you.
However, what do u think is the best method for solving the shear problem since increasing the thickness will lead to a higher vn but at the same time, a higher load attraction.
Increasing the length of the wall (or adding another wall) also will be crucial to architect and he will not like it.
Thanks

 

[KootK]

I know that you dont want torsion as your first mode (i was told by a senior engineer)..however, i do not know why? can you guide me to something that can explain it

If your first fundamental mode of vibration is torsional, then the ASCE7 Equivalent Lateral Force procedure is no longer valid. You would need to execute a 3D response spectrum / time history analysis in order to capture the structural response adequately. Obviously, that can be done. For most projects of modest scale, however, the project budget doesn't leave room for such a detailed investigation.

.that is like a bolt connection..first row of bolt takes more than others but we still assume equal distribution.

The problem with that analogy is that bolt failures have been shown, through testing, to be ductile. To my knowledge, concrete shear failures have not.

this is why this method is applied to precast wall where vertical dowel connection can be nominal..and what fails is the connection...however, you are right..the point that this happens is trivial in nature. If you break long panel into pieces and don't stitch the precast panel together however, the stiffness of that wall won't be as big as if they are all stitched to work monolithically. this may help make the argument more valid.

I'm pretty skeptical about this. The panel stiffness will be more or less that of one large composite panel right up until you fail those stitch connections. And most common stitch connections that I'm familiar with would have:

a) pretty high ultimate capacities and, more importantly;
b) ultimate capacities that would be very difficult to predict with any accuracy.

If you post a detail showing your proposed stitch connection, I'd be happy to review it and provide further comment.

what are you referring to here? which original post?

Your very first post in this thread and the subject implied by the title of this thread. Namely, a wall connection detailed to prevent a concrete wall from attracting shear forces.

are you talking about global wind force?

Yes.

i believe that is the same as an EQ stability design..the disregarded wall can fail and go to the "lateral stability element".

So, every time that there's a strong gust of wind, the owner is going to have to go back and repair all of those failed stitch connections? You might want to run that by the owner.

but i want to somehow find a way to justify that it is a reasonable way..in order for me to be able to do what i've been asked to do without feeling like the whole thing is all wrong..

I feel your pain. Try not to let it get you down. In my experience, most low rise buildings are murky pits of uncertainty when it comes to lateral deign. In a 100 story building, the lateral forces are high but at least you have a good idea where they go. In a four story building, you're stuck interacting will all kinds of elements that you'd rather not have in the mix complicating things.

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.

 

[KootK]

Thank you. However, what do u think is the best method for solving the shear problem since increasing the thickness will lead to a higher vn but at the same time, a higher load attraction.

You're most welcome chekre. It's a tricky problem to be sure. Some possibilities would include.

1) Add, lengthen, or thicken walls elsewhere to create a more balanced layout and draw shear away from the problem wall.

2) Divide up the problem wall into several, narrower walls using vertical control joints. This may reduce the stiffness of the walls and, therefore, the amount of shear they will attract. You may also be able to arrange things such that the walls yield through the formation of a flexural base hinge rather than through shear yielding.

3) If it makes sense to do so (cladding panels etc), connect the wall to the diaphragm in a manner that facilitates lateral slip.

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.

 

[4dmodeller]

KootK,

I'm pretty skeptical about this. The panel stiffness will be more or less that of one large composite panel right up until you fail those stitch connections. And most common stitch connections that I'm familiar with would have:

that's why you dont stitch the panel together..i.e. a vertical control joint..then they are just separate smaller panel

please find attached stitch connection used..ut is rated for 200 kN..i want to ask you if you would consider eccentricity of shear to the center of shear stud group too? (my approach is to take the eccentricity from shear group centroid to edge of panel + gap between panel)..some say it is too conservative..and they can just work in shear without any eccentricity..this would dramatically increase the capacity of studs...some also say the eccentricity should be shear group centroid to edge of first panel + gap+ shear centroid to edge of second panel... i think the last one is too conservative...
what is your opinion on this?

yes, i prefer high rise where they load path is straightforward and builder is more willing to follow your advice.

 

[KootK]

what is your opinion on this?

I agree with your method. Eccentricity = 1/2 distance between stud group centroids.

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.

 

[4dmodeller]

KootK,

But if the connection is in the corner with the return wall, eccentricity can increase because you dont get double curvature in the stitch plate anymore (it will be a cantilever from one wall receiving v* from the return panel)