I'm trying to undertake a 2D patch test for shear (Txy), but I cannot figure out how to apply a shear stress in ANSYS. I am presuming it has to do with applying a pressure, but it will only apply a positive or negative pressure perpendicular on the face.
Strictly speaking a traction can be any form of load applied to boundary normal, or tangential. Also note that all forms of loading like pressure are converted to nodal forces in the software. Rigid-body motions should be constrained by point (nodal) constraints rather than constraints along edges which will constrain all nodes along the edge and lead to an over-constrained model with associated higher stresses. If you are trying to check that the model can accurately represent statically admissible stress fields then you should apply the corresponding consistent nodal forces rather than distributing the net shear force uniformly across the section. Consistent nodal forces are calculated by integrating the shape functions with the boundary tractions along a model edge - see any good finite element text book.
Yes - 'pressure' is applied 'normal' to the surface; 'traction' is applied 'tangential' to the surface.
It sounds like you are doing convergence tests on patches of elements to see if they converge to the true value? The distribution of shear you require is probably quadratic? I have done this in the past and the best approach is to write yourself a small programme to evaluate the consistent nodal forces corresponding to the stress at the boundary. With a bit of ingenuity this programme can be quite general and include lower- and higher-order elements and can output the nodal forces in a format suitable for reading directly into ANSYS.
there are two ways either apply shear displacement mode or force mode. the easiest is to apply shear forces. As a quick and dirty method you can apply the stress as nodal forces along the edge, by selecting all edge nodes and dividing the shear force by their number count [use *get command]. it is of vital importance to preserve equilibrium of forces while doing that, i.e. no net moment should result from the application of the shear forces on the 4 edges. also equally important is to properly constrain the model. I would go for fixing two intersecting edges one edge against motion in its perpendicular direction and let it slide in its parallel direction, the other edge the other way around.
Strictly speaking a traction can be any form of load applied to boundary normal, or tangential. Also note that all forms of loading like pressure are converted to nodal forces in the software. Rigid-body motions should be constrained by point (nodal) constraints rather than constraints along edges which will constrain all nodes along the edge and lead to an over-constrained model with associated higher stresses. If you are trying to check that the model can accurately represent statically admissible stress fields then you should apply the corresponding consistent nodal forces rather than distributing the net shear force uniformly across the section. Consistent nodal forces are calculated by integrating the shape functions with the boundary tractions along a model edge - see any good finite element text book.
A simple way to apply a pure shear is to apply a case that sigma_x= - sigma_y=sigma, and to apply this condition in case your model is a cube or square to may apply positive pressure on vertical edges and equal negative pressure on horizontal edges.
- Badges
- Science topic