Deep Beams - Strut & Tie Vs Finite Element Model

[Drapes]

For the design of deep beams, the general consensus is to use the strut and tie method. As an alternative though, if a finite element model is set up using say SAP2000, would it be acceptable to simply extract the direct stresses from the analysis, and compare it to the max allowable stresses dictated by S&T, in order to verify the design?

This would involve first calculating the max allowable stresses at the struts and nodes depending on whether you have a CCC, CCT or CTT node, and then ensuring the S11 (horizontal), S22 (vertical) or SVM (von mises) stresses resulting from the finite element analysis are within those limits.

The benefit I see in doing this is to bypass the tedious process of manually resolving and setting up the often complex nodal geometries, especially when it involves non-hydrostatic nodes.

 

[Celt83]

you could use the FEM model to verify and help optimize your Strut and Tie "truss". To verify stresses you'd need to successively run and remove zero/near zero stress elements until you have a model that matches the strut and tie layout, I believe one or two of the deep beam programs out there in the wild does automates this process.

Open Source Structural Applications:

https://github.com/buddyd16/Structural-Engineering

 

[Drapes]

Thank you Celt83,

So why cant we simply use the direct stresses given from a pure FEM model, and compare them to the max allowable stresses dileneated by S&T?

Conventionally, you would set up your S&T model, including the geometry of the nodes at each strut/tie interface, and then calculate the stresses based on this geometry. These stresses would then be compared to the max allowable stresses, to determine if the design is valid.

What I'm curious to know is, instead of dealing with all the geometry, if you could use the results direct from the FEM model to determine your stresses. Obviously from a capacity point of view, you would still need to consider the general assumptions of the S&T to determine the max allowable stresses, for example the strut angles, and whether the nodes are CCC, CCT or CTT node etc, but from a demand point of view you could bypass all the geometry particulars and obtain the results directly from the FEM model.

 

[Celt83]

My hesitation with using the FEM results directly would be that they are likely based on a linear elastic analysis using the full gross cross sectional properties for the stresses. This would pretty much always yield lower stresses than reduced strut and tie model since there is more area to spread the forces over.

Open Source Structural Applications:

https://github.com/buddyd16/Structural-Engineering

 

[IDS]

Have a look at:

http://vectoranalysisgroup.com/software.html

I have recently downloaded this program, but not used it seriously yet.

It has detailed manuals and background information, and is designed exactly for this sort of analysis, i.e. non-linear finite element analysis in situations where standard beam bending assumptions do not apply.

Doug Jenkins
Interactive Design Services

http://newtonexcelbach.wordpress.com/