Common Boundary Condition Feature Settings
Boundary condition features control the mesh displacement or velocity at the boundaries of deforming domains. These features have a number of common settings, controlling where and how the constraint is enforced.
Excluded Entities
When a boundary condition is enforced using Pointwise constraints or Weak constraints, it may come into conflict with boundary conditions at adjacent boundaries. When this happens, you can use the Excluded Edges or Excluded Points sections (collapsed by default) to exclude affected edges or points from one of the boundary conditions. This ensure that the other, remaining, condition prevails.
If the boundary condition is enforced using Pointwise constraints, then no constraint is added at the constraint points that belong to the excluded entities. When using Weak constraints, the Lagrange multiplier variable will be constrained on the excluded entities, which effectively removes the constraint. Since the Nitsche method is implemented as a distributed contribution on the boundary, it cannot be disabled locally at edges or points. The Excluded Edges or Excluded Points sections are therefore not available when this method is used.
Constraint Settings
All moving mesh boundary conditions are of Dirichlet type, but there are a number of different ways these can be enforced:
Pointwise constraints:
The mesh displacement is prescribed using a pointwise constraint at Lagrange points of the same order as the geometry representation. The default Constraint method is Elemental, which means that constraints are applied for each Lagrange point in each element separately; if the constraint is discontinuous between elements, there will be multiple separate constraints at shared points. When using the alternative Nodal constraints the constraint expression is first averaged over all elements sharing a constraint point, before applying a single constraint.
Pointwise constraints are stable and always respected as long as they are self-consistent, but the solvers do not allow constraining time derivatives. This option is therefore not available for velocity conditions.
Nitsche method:
The mesh displacement or velocity is enforced using an incomplete Nitsche formulation, which can be seen as a modified penalty method. An equation contribution that penalizes deviation from the specified constraint is added everywhere on the boundary. The Nitsche method therefore does not satisfy the constraints exactly. You can to some extent control the accuracy by tuning the Stabilization factor. A higher value (default is 1) increases the penalty and therefore accuracy, but at the same time makes the system of equations increasingly ill-conditioned. The main advantage of the Nitsche method is that it does not interact with conditions at adjacent boundaries.
Weak constraints:
the mesh displacement or velocity is prescribed using weak constraint equations and Lagrange multiplier variables. Where adjacent constraints meet at edges or points, it is important that the number of free Lagrange multiplier variables agree with the number of degrees of freedom to be constrained.
If a weak constraint meets at an edge or point with a pointwise constraint or another weak constraint using a different Lagrange multiplier, then one of the Lagrange multipliers will often become underdetermined. The Lagrange multiplier is then not necessary for enforcing the given boundary conditions and can take any value without violating any equations. This, in turn, leads to solvers reporting a singular system. The solution to this problem is often to use the Excluded Edges or Excluded Points selections to remove one of the constraints from the shared edges or points.
By default, the Constraint type is set to Automatic, which generally is interpreted as pointwise constraints for deformation conditions and weak constraints for velocity conditions.