The shell displacement is different w/ or w/o beams

A 10m x 10m flat shell, the thickness is 0.05m, the E modulus is 2.1e7 MPa, the shear modulus is 80770 MPa. A uniform load of 1kPa is applied download on the shell.

There are two models

  1. The 4 corners of the shell are simply supported.
  2. 4 beams are under the shell edges, the ends of the beam are simply supported., while 4 very stiff springs connect the support and the corner of the shell.

You would expect that the displacement of the shell in model 1 and model 2 should be very close.

The image below show the displacement contour of model 1(left) and model 2 (right) in Sap2000. The max disp in model 1 is 106.8 mm, while that in model 2 is 107.1 mm, they are very close.

The image below show the displacement contour of model 1(left) and model 2 (right) in Sofistik. The max disp in model 1 is 105 mm, which is close to that in Sap2000, while that in model 2 is only 55.6 mm.

As you can see, the 4 beams in Model 2 should have no effect on the result.
Then why the results in Sofistik are different?

The Teddy code is uploaded, please download it and change Sofistik1.dat to, then unzip it.
Sofistik1.dat (1.9 KB)

Please run Model 1.dat or Model 2.dat to get result.

Good morning netsonic_Sofistik.

I think the difference is that your springs doesn’t only constraint vertical movement but also rotation of the shell. The springs have CQ (Transversal stiffness) so the beams are restraining the rotation of the corners at the shells. If you remove the CQ constant from the springs you will get similar results (105mm).

Kind regards,

Sofimesh attached:
SofimshC - Model 2.dat (1.7 KB)

Hi rbastidam

The shell is a symetrical structure, the boundaries and loading are symetrical too.
The rotation about global Z axis at 4 corners should be zero.
Therefore, there is no rotation restraining.

rbastidam is correct:
You have managed to model an extra rotational restraint (due to springs and lever arms of said springs)

It’s evident if:

  • You check the principal moments for both systems in wingraf
  • Compare deformations at the corners in the animator (with amplified and equal amplitude, e.g. 50)