tapan
2007-07-19, 11:50
Hi,
As per the abaqus user's manual, for isoparametric 3D elements with full integration, a technique called "selective reduced integration" is used wherein a modified deformation gradient F_bar is used which is related to the actual deformation gradient F as,
F_bar = F*(J_bar/J)^1/3
where J_bar and J are the jacobian of F_bar and F respectively. Thus when we use a umat for a C3D8 element the DFGRD1 and DFGRD0 that are passed in are F_bar and not the actual F. Anyone have any experience on how different is F_bar from F (I think it should be pretty similar unless you have high strain gradients within an element). Should an effort be made to recover the actual F from F_bar before using it in the calculations?
Thanks,
Tapan
As per the abaqus user's manual, for isoparametric 3D elements with full integration, a technique called "selective reduced integration" is used wherein a modified deformation gradient F_bar is used which is related to the actual deformation gradient F as,
F_bar = F*(J_bar/J)^1/3
where J_bar and J are the jacobian of F_bar and F respectively. Thus when we use a umat for a C3D8 element the DFGRD1 and DFGRD0 that are passed in are F_bar and not the actual F. Anyone have any experience on how different is F_bar from F (I think it should be pretty similar unless you have high strain gradients within an element). Should an effort be made to recover the actual F from F_bar before using it in the calculations?
Thanks,
Tapan