# Slip boundary conditions

Hi everyone,

is there “automatic” way to enable free-slip boundary conditions on one boundary and Dirichlet on another using the VectorH1 space?

Alternatives would of course to be a separate H1 space for each component each with different Dirichlet flags like here, using Nitsche to enforce the Dirichlet on only one component or to manually set the freedofs array like here.

Best wishes,
Henry

There is a “manual” way, which is pretty good.
See the post

Here is a snipplet that works for me:

``````V = VectorH1(mesh, order = 3)
Q = H1(mesh,order=2)
X = FESpace([V,Q])
freedofs = X.FreeDofs()
fesx,fesy = V.components
for el in fesx.Elements(BND):
if (el.mat=="dir"):
for dof in el.dofs:
freedofs.Clear(dof)
freedofs.Clear(fesx.ndof+dof) # y-component
if (el.mat=="neu"):
for dof in el.dofs:
freedofs.Clear(dof)
if (el.mat == "wall" or el.mat =="inlet" or el.mat =="outlet"):
for dof in el.dofs:
freedofs.Clear(fesx.ndof+dof) # y-component                     ``````

Hi Guosheng,

thank for your help! I was was hoping to avoid a loop over the elements, as I have to update the freedofs array in every time-step, but your way is probably still the best.

Best wishes, Henry

What is your criteria for selecting fixed dofs in each time step?

Hi Christopher,

within the bulk, the dofs are selected with element markers obtained via a level-set function (which changes in each time-step). The mesh boundary dofs will stay the same in each time-step.

Best wishes, Henry

Hm I guess then your best option is to create a map of element number to dofs and in each timestep clear the freedofs and then use the map to set the marked dofs. This is most likely faster than iterating over all elements.
Best
Christopher

Thank you for your advice! I think I can do it even more simply: Before the time loop, I create list of the dofs which need to be set on the slip boundary, and in each time step I loop just over this list and set the respective dofs.

Best wishes, Henry

As I have not seen this in the online documentation yet, for reference to anyone who might also need this: I have seen that the VectorH1 space now takes the arguments dirichletx, dirichlety and dirichletz to set separate Dirichlet BCs on the components of the space.