Learning Forum

[C Deho]

I am currently writing code to apply various nonlinearities by referring to theory references. I am using the Newton-Raphson method with the residuals of internal and external forces, but it seems that the internal forces are not being computed accurately.

For each element, I calculate the stress as shown in the figure below, then compute the nodal internal forces by summing the stresses at overlapping points, and finally obtain the total nodal internal force vector. Given that I am applying a force in the z-direction to only one node, all other nodes should ideally have values close to zero in that direction. However, this is not the case.

Internal forces are calculated by multiplying the element stresses by the transpose of the B matrix, and since both stresses and strains appear to be computed correctly, the issue does not seem to lie with this matrix.

It appears that the problem may originate from incorrect stress calculations. I am currently averaging the strains obtained at the Gauss points and then applying the D matrix to get the element stresses. While some nodes show internal forces that cancel out to zero, there is no discernible trend.

Could you suggest a method for calculating Element stresses? For reference, I am first calculating internal forces in a linear problem before applying nonlinear analysis (since the residuals should be zero in linear analysis).

Thank you.

[deepak.deepak]

Hi,

Please refer to, Geometry nonlinear FEM iteration process ( How to compute internal force?) (ansys.com), which is already answered.

Thanks

Deepak