Is STAAD Plate Up and Mesh Up Equivalent?

[petertor]

I have modeled an open top tank as square plates of concrete material plates filled with water and ran the stress contours:

http://i42.tinypic.com/2sb7uwz.jpg

Then I took that model and meshed it up into a much larger model with smaller quadrilateral plates (finite elements):

http://i40.tinypic.com/11s2jwj.jpg

Why are the Maximum Top (Principal Major Stress) contours different for the same model? I mean, radically different.
I've checked kN/m2 loading, and the meshup of the original correctly applied the pressures to the smaller sub-mesh.

I'm used to doing square concrete tanks using PCA factors for moment varying across the face from corner to middle and from bottom to top, and would have expected the corner stresses to be higher than the middle showing in plate up.

As soon as you can't trust your software, you're old school, and Bentley FAQ is about as helpful as MSoft.

 

[JoshPlumSE]

I'm not a STAAD guy (in fact I'm actually with the RISA support group), but your pictures do show a couple of basic modeling problems:

1) The original mesh size (4 plates wide x 4 plates tall) would seem to be a bit coarse. I always tell people to model at least 3 plates between points of inflection. In your case, the absolute minimum mesh size that I would think you could get away with would be 6x6. Though I would prefer to use something closer to a 12x12.

2) The boundary condtions remained the same in both models. The coarse meshed model is supported at EVERY one of the bottom joints. But, the fine meshed model is only supported once every 10 joints or so. Therefore, it doesn't have continuous support at the base.... This would result in a different force distribution.