- 积分
- 0
- 注册时间
- 2008-9-7
- 仿真币
-
- 最后登录
- 1970-1-1
|
发表于 2011-6-7 02:06:59
|
显示全部楼层
来自 美国
本帖最后由 stone08 于 2011-6-7 02:11 编辑
The following is what I got from Abaqus support. Basically, it is what #33 described about but in a more detailed way. Just FYI:
During the analysis, Abaqus calculates all the element based field output variables (stresses, strains) at integration points where as the nodal variables such as displacements, temperatures are calculated at the nodes directly. During the post processing of the results, to calculate (transform) the integration point tensor quantities at nodes, Abaqus extrapolates the values and uses a conditional averaging method by default, which works as follows:
1.The stress tensor components are extrapolated from the integration points to the nodes.
2.Compute the scalar (invariant) from the extrapolated stress tensor components at element nodes.
3.Average the scalar if the threshold criteria (which by default is 75%) is met Compute the invariant by averaging the nodal scalar quantities.
For example, if a particular node is shared by 4 elements and that node is having contributions from all the four nodes, then from each element the stress components are extrapolated to the nodes and then scalar quantities are computed. After that all the scalars are averaged (depending on the threshold criteria) to get final invariant say Von Mises value. Note that with this approach, all tensor components averaging happens in a consistent basis. Abaqus calculates the threshold criterion, which is to be chosen by the user (default is 75%). If the relative nodal variation for each node included in the plot is less than your averaging threshold, the contributing values from elements are averaged at that node. If the relative nodal variation exceeds user setting (default is 75% Avg.), the values are not averaged. A value of 0% would cause no averaging and a value of 100% would give you smooth contours (with a value of 100%, one might not able be able to identify if the mesh refinement is fine enough to capture the stress concentrations, hence a default value of 75% is chosen which would give us a better insight). At some location in your model, the relative nodal variation exceeds the default 75% Avg and hence the stress values are not averaged in such cases. |
|