RISA-3D and the PCA design coefficients 1

[mann]

I am trying to compare results from the finite analysis program RISA-3D to the plate coefficients in "Rectangular Concrete Tanks" published by PCA. All of the results from the RISA run seems to be lower than those listed in the PCA charts. Does anyone have any suggestions about how to model the plate? i.e. how thick, mesh size etc.?
Or has anyone else tried this exercise?

 

[JAE]

I have found that generally, finite element analyses tend to give lower results than other empirical methods. I am not familiar with the PCA document you refer to but many handbooks on plate-type solutions tend to utilize approximate methods that give conservative results. Also, you must be very careful in modeling with FE as they can be tricky. Try using different mesh sizes (RISA has a quick way of sub-dividing elements down to smaller elements). Also, watch your boundary condition assumptions as this can also affect your results.

The FE results also assume a non-cracked section if you simply enter a constant thickness throughout your model. Does the PCA document indicate any assumptions as to cracking and loss of stiffness?

 

[mann]

Actually, the coefficients were developed using SAP90. "Results are provided from finite element analyses of two-dimensional plates subject to out-of-plane loads using SAP90. Convergence analysis was made to ensure the quality of the results." It goes on to say that the slab was assumed to act as a thin plate.

This is why I am concerned that I am not getting results closer to the coefficients listed in the PCA charts.

 

[Qshake]

I'm not familiar with the charts either but I gather that the PCA chart coefficients were developed using SAP 90. How much difference are we talking about...you are comparing two different numerical methods (FEA)which use slightly different algorithms. I would not expect the results be be exact but would expect the results to be somewhat close. Perhaps you should check out the element dimensions and refine your mesh.