Masonry Walls and Shear Distibution Around Openings

[wannabeSE]

I typicaly analyze CMU shear wall by calculating the relative stiffness of the elements. Recently, I modeled a wall with RAM Elements masonry wall module and was surprised by the FEM results. The piers between window attracted more lateral load than I expected. I then performed a calculation by hand and the results were as I would expect. Next, I analyzed the same wall with ETABS an the results were consistent with RAM Elements FEM. The forces in the piers between windows is 70%± higher than hand calculations. This leaves me with a couple questions:

1. I am wondering if the wall actually behaves more like the FEM elastic analysis or the traditional relative stiffness method. Has there been any testing of masonry shear walls with openings to validate FEM modeling or relative stiffness method?

2. I am wondering when hand calcs should be used or whether FEM elastic analysis or more advanvced method (plastic analysis with equivalent frames or fibers) should always be used.

3. Are there any studies or articles related to this subject that I can review. There is one article that recommends FEM for walls with "large openings or walls with configurations that require the wall assemblage be analyzed as a frame" in Masonry Chronicles,

http://cmacn.org/PDF/Masonry_Chronicles_Spring_2009.pdf.

I'll be disappointd if single story CMU buildings should noo longer be analyzed with hand calcs and need a computer.

 

[Agent666]

I can definitely see torsional 3D effects increasing the load over and above simple tributary width hand calcs, similarly things like wall continuity above & below windows also affect the relative stiffness that is hard to account for in hand methods, which in turn affects the distribution of load between walls. Really depends on your assumptions for both the hand and computer analysis (such as how stiffness reductions and the type of analysis (modal or equivalent static), etc) are considered.

Basically hand methods should be limited to reqular structures that the behaviour is known to be modelled accurately enough by hand, anything irregular should be modelled in 3D or as otherwise defined in the codes you are working to (there should be rules around when a more robust analysis is required).

Regular/irregular in this context means as defined as per your code as it relates to buildings, even seemingly regular structures can fail one or more of the checks around vertical or horizontal mass distribution, stiffness discontinuities between storeys or the torsional sensitivity of a structure.

I often see the issue with people inappropriately using hand methods, and as you've shown when you analyse the system using a more appropriate rigorous analysis method the results can be quite different than what your expectations were based on a simplified hand check.

 

[wannabeSE]

I am working on a building with a flexible diaphragm. The discrepancy is within individual wall. My models and hand calc is 2D. The hand method I am using is well documented in older textbooks.

 

[Agent666]

Without knowing the exact arrangement you are looking at its hard to say if the hand calculation captures all the behaviours that a computer analysis might be considering. Can you maybe print an output of what you are looking at and how the hand analysis is breaking it down to equivalent elements that are easily assessed by hand. Hand analyses often include some underlying assumptions to simplify things for example.

Remember that these old hand methods most likely predate modern computer analysis, and I have no doubt many are in fact inappropriate in one way or another for certain types of structures, so like anything they may have been deemed appropriate once upon a time for all types of structures. These days with computer analysis at our fingertips we can obviously analyse things to the point where we are also making simplifications that can skew the answers, so what is really the right answer! Refining the mesh size in ETABs would be a good example of this!

For example refine your mesh size in the shell elements you are using in ETABS and the distribution of design actions will continue to change at each iteration, so at what point do you stop and deem the answer to be correct/right? If you keep dividing the mesh the answer will continue to converge on a given solution, however by this time occurs there is potential for the the mesh size might become impractically small for running any practical analysis as it will probably take until Christmas to complete or crash the program...