
From:  logari81 
Subject:  [Getfemusers] Plasticity brick 
Date:  Fri, 21 Sep 2012 16:03:03 +0200 
Useragent:  Mozilla/5.0 (X11; Linux x86_64; rv:15.0) Gecko/20120827 Thunderbird/15.0 
Hi all,I have recently worked with the elastoplasticity brick and I have one relevant comment.
As a scalar measure of stress one can use either the euclidean norm of deviatoric sigma or the Von Mises stress. The difference is an sqrt(3/2) factor:
$\sigma_{vm} = \sqrt{3/2} \sigma^{dev}$In getfem, \sigma^{dev} is used through the whole brick implementation except one point. Only in the stresses calculated by compute_elastoplasticity_Von_Mises_or_Tresca the result is multiplied by sqrt(3/2).
The stresses calculated by "compute_plastic_part" and the input parameter "datathreshold" do not include the sqrt(3/2) factor. So the user of the brick has to divide the tensile yield strength \sigma_y by sqrt(3/2) before assigning it to "datathreshold" and he also has to multiply the results of "compute_plastic_part" with the same factor in order to receive stresses comparable to the Von Mises definition.
Is this behavior intentional? Wouldn't it be more consistent to either remove the factor from "compute_elastoplasticity_Von_Mises_or_Tresca" or add it to "compute_plastic_part" and "datathreshold"?
Best regards Kostas
[Prev in Thread]  Current Thread  [Next in Thread] 