Different results between Reduced Results vs Core walls

[Anonymous]

Hi everyone !.

I have a simple panel 2mx3mx0.13m

I apply a force in the global x direction F = 1000 kg. I made a 5cm mesh for greater precision.

The results are different in terms of the moments, for the bottom part of the panel the results are close but for the Top part the difference is large, Reduced Results (M = 615kgf-m), Core walls (M = -25 kgf-m )

Results Core walls

Results Core wall

Reduced Results for panels (table)

I also made model in Etabs

I don't understand what is happening. 

Doing a quick manual calculation, the moment at the bottom would be 3000kgf-m.

And the moment at the top of the panel should be 0 kgf-m, however I don't understand why Robot gives a result of +615 kgf-m

[Krzysztof_Wasik]

Hi @Anonymous 

Robot display those diagrams between Finite Elements centers. So results are not presented for top and bottom edge. Mesh refinement near top and bottom edge will reduce difference.

Value 615 is strange. I cannot explain it without model verification.

[Anonymous]

Hi @Krzysztof_Wasik 

Thank you so much for taking the time to respond.

I attach the file.

[Krzysztof_Wasik]

Hi @Anonymous 

In my opinion it is integration accuracy for in plane forces when big deformation occurs locally. It is related to finite element type used in Robot

When I have defined equivalent horizontal linear force along top edge (instead point load in corner)  results looks as follows (no big local deformation effect).

Displacement

results

When I have defined rigid link along top edge of panel, leaving point force in corner, results are better as well.

 Results will be better when beam will be defined along panel top edge

[Anonymous]

Hi @Krzysztof_Wasik .

I was doing tests with cuts and with different dimensions of beams, also increased the force FX = 1000T I got very curious results.

For a 10cmx10cm beam, I have a shear = 1000.64T , Moment = +73.31 T-m

V10cmx10cm

For a beam 30cmx30cm

According to the results, without a beam a very large error is generated, but increasing the section of the beam gives a better result, but nevertheless there is still a lot of error. Is it possible that for a later version they improve this problem? The truth is that I am concerned about this problem, as the results are not reliable for the designs.

[Krzysztof_Wasik · Accepted solution]

Hi @Anonymous 

As I have mentioned before issue is caused finite element type used in Robot and its accuracy in in-plane force integration/calculation when big in plane deformation occurs. In my opinion improvement needs using different (or improved) type of finite element in Robot.

It is test example (quite specific). In my opinion, in real models, where such big deformations does not occur locally, inaccuracies will be smaller.