www.kxcad.net Home > CAE Index > ANSYS Index > ANSYS AUTODYN

Hex Integration
The unstructured, 8 noded, constant strain, hexahedral element is available in two forms.
Finite volume formulation with exact volume integraion (Wilkins 1974) (Default)
Finite element formulation using approximate Gaussian quadrature (Hallquist 1976)
The finite volume formulation is the default option and recommended for most applications involving large deformation or warped meshes. This formulation is the same as that used in the Structured (IJK) Lagrange solver.
The finite element formulation is included as a solver option and provides efficiency improvements over the above. However, the accuracy of the element is reduced for warped elements.

Example performance comparison for the three Hex solvers now available in AUTODYN
Hex Hourglass Control
Hourglass control for the Unstructured Hex solvers is provided in two forms
AD standard
Flanagan-Belytschko
The "AD Standard" option is the default option and works well for most applications. This is the most efficient option in terms of memory and speed. A default viscous damping coefficient of 0.1 is recommended.
The "Flanagan-Belytschko" form of hourglass control is available as an option since the AD standard form may not perform well under large rigid body rotations. The Flanagan-Belytschko form of hourglass control is invariant under rotation hence overcomes this problem. Both viscous and stiffness based control is available. The default is stiffness based control with a coefficient of 0.1.
Tet Pressure Interaction
The Unstructured Tet element is available in two forms
Standard constant pressure (SCP) Tet
Average nodal pressure (ANP) Tet, (Burton 1996)
The SCP tetrahedral element is a basic, constant strain element and can be used with all the standard AUTODYN material models including erosion. Explosive burn logic is also available. The element is intended as a "filler" element in meshes dominated by hexahedral elements. The element is known to exhibit locking behavior under both bending and constant volumetric straining (i.e. plastic flow). If possible the element should therefore not be used in such cases.
The ANP tetrahedral element is an extension of the advanced tetrahedral element (Burton 1996) and can be used as a majority element in the mesh. The ANP tetrahedral overcomes problems of volumetric locking, which occur with the SCP tetrahedral element. however, the ANP tetrahedral element is still susceptible to shear locking in bending dominated problems. You should therefore be careful to verify their results in such cases.
For meshes containing a majority of tetrahedral elements, the ANP option is recommended.
Material Modelling Options Available with ANP-Tet Elements
|
Category |
Model |
|
Equation of State |
Linear |
|
|
Polynomial |
|
|
Shock |
|
Strength Model |
Elastic |
|
|
Viscoelastic |
|
|
Von Mises |
|
|
Johnson Cook |
|
|
Piecewise JC |
|
|
Zerilli Armstrong |
|
|
Steinberg Guinan |
|
|
Drucker-Prager |
|
|
MO Granular |
|
Failure Model |
Hydro (Pmin) |
|
|
Plastic Strain |
|
|
Principal Stress |
|
|
Principal Strain |
|
|
Principal Stress/Strain |
Parts containing ANP tetrahedral elements should only be filled with a single material. Multiple materials can be represented in a single body by creating multiple parts and joining them together.
Note that a hexahedral mesh will generally provide more efficient results than a tetrahedral mesh hence we only recommend the use of predominantly tetrahedral mesh models for convenience of mesh generation.

Comparison of SCP and ANP Tet Elements for an impact (Taylor test) involving large amounts of plastic deformation. The ANP-Tet fives results comparable with experiments.
1) Burton A.J., "Explicit, Large Strain, Dynamic Finite Element Analysis with Applications to Human Body Impact Problems", PhD Thesis, University of Wales, December 1996.
2) Bonet J, Burton A.J. "A simple averaged nodal pressure tetrahedral element for incompressible and nearly incompressible dynamic explicit applications". Communications in Numerical Methods in Engineering 1998; 14, 437-449.