jfonken
Bioengineer
- Oct 30, 2020
- 6
Hi all,
As a subproject in my PhD project, I'm comparing Ansys and Abaqus (geometrically) nonlinear solution methods. I used a simple cylinder to represent an artery and modelled it with a simple Neo-Hookean material model. Large deformations (NLGEOM) are turned on (hence the non-linearity). I used 20-nodes hexahedral elements with reduced integration and a mixed UP-formulation.
When I compared the nodal results, I saw that the absolute differences in displacement were very small (10^-10 m). However, the absolute difference in stress goes up to 6*10^4 Pa (figure 'nodalResults',
The figure 'VonMises' ( shows the Von Mises stress distribution for the Ansys and Abaqus simulations. Which shows that the distribution is similar, but the values differ.
To find the cause of these differences in stress output, I looked at the stress at the integration points. The same (Gauss) interpolation points are used in Ansys and Abaqus, so the stresses at the integration points should resemble each other. Luckily, this is the case. The differences are small (up to 5 Pa for the normal stresses and up to 1 Pa for the shear stresses). See figure 'IntegrationPointResults' (
These results indicate that the differences in nodal stresses are caused by differences in the extrapolation and/or averaging method. I therefore want to know how the Ansys and Abaqus extrapolation and averaging methods exactly work and how they differ. However, I'm unable to find a clear answer to this question. For Ansys, I've found a table (figure AnsysDataVariation, explaining the assumed data variation used to extrapolate the stress values to the nodes, but I couldn't find any information about the coefficients.
As a subproject in my PhD project, I'm comparing Ansys and Abaqus (geometrically) nonlinear solution methods. I used a simple cylinder to represent an artery and modelled it with a simple Neo-Hookean material model. Large deformations (NLGEOM) are turned on (hence the non-linearity). I used 20-nodes hexahedral elements with reduced integration and a mixed UP-formulation.
When I compared the nodal results, I saw that the absolute differences in displacement were very small (10^-10 m). However, the absolute difference in stress goes up to 6*10^4 Pa (figure 'nodalResults',
The figure 'VonMises' ( shows the Von Mises stress distribution for the Ansys and Abaqus simulations. Which shows that the distribution is similar, but the values differ.
To find the cause of these differences in stress output, I looked at the stress at the integration points. The same (Gauss) interpolation points are used in Ansys and Abaqus, so the stresses at the integration points should resemble each other. Luckily, this is the case. The differences are small (up to 5 Pa for the normal stresses and up to 1 Pa for the shear stresses). See figure 'IntegrationPointResults' (
These results indicate that the differences in nodal stresses are caused by differences in the extrapolation and/or averaging method. I therefore want to know how the Ansys and Abaqus extrapolation and averaging methods exactly work and how they differ. However, I'm unable to find a clear answer to this question. For Ansys, I've found a table (figure AnsysDataVariation, explaining the assumed data variation used to extrapolate the stress values to the nodes, but I couldn't find any information about the coefficients.