Feature/111 inflow outflow bcs #112
Conversation
Codecov ReportAttention: Patch coverage is
📢 Thoughts on this report? Let us know! |
This reduces the amount of duplicate code for default gpu backend
There was a problem hiding this comment.
It would make sense to add some simple tests for each of the new BCs in 2D and 3D to satisfy codecov.
This would likely be under a new PR, but with the variety of BCs available it could be useful to add a dedicated "Boundary Conditions" page under "Mesh Generation" in the docs.
| !! | ||
| !! \begin{equation} | ||
| !! \overleftrightarrow{f} = \begin{pmatrix} | ||
| !! \rho u \hat{x} + \rho v \hat{y} + \rho w \hat{z} \\ |
There was a problem hiding this comment.
Consistent spacing
| !! \rho u \hat{x} + \rho v \hat{y} + \rho w \hat{z} \\ | |
| !! \rho u \hat{x} + \rho v \hat{y} + \rho w \hat{z} \\ |
| integer :: ndiagnostics | ||
|
|
||
| ! Model parameters | ||
| real(prec) :: p0 = 10.0_prec**(5) ! Reference pressure for potential temperature () |
There was a problem hiding this comment.
Do you want Pascal as the unit here?
| real(prec) :: p0 = 10.0_prec**(5) ! Reference pressure for potential temperature () | |
| real(prec) :: p0 = 10.0_prec**(5) ! Reference pressure for potential temperature (Pa) |
| nhat = this%geometry%nhat%boundary(i,j,iEl,1,1:2) | ||
|
|
||
| this%solution%extBoundary(i,j,iEl,1:this%nvar) = & | ||
| this%hbc2d_noslip(this%solution%boundary(i,j,iEl,1:this%nvar),nhat) |
There was a problem hiding this comment.
Capitalize for consistency
| this%hbc2d_noslip(this%solution%boundary(i,j,iEl,1:this%nvar),nhat) | |
| this%hbc2d_NoSlip(this%solution%boundary(i,j,iEl,1:this%nvar),nhat) |
| nhat = this%geometry%nhat%boundary(i,j,iEl,1,1:2) | ||
|
|
||
| this%solutiongradient%extBoundary(i,j,iEl,1:this%nvar,1:2) = & | ||
| this%pbc2d_noslip(this%solution%boundary(i,j,iEl,1:this%nvar),nhat) |
There was a problem hiding this comment.
| this%pbc2d_noslip(this%solution%boundary(i,j,iEl,1:this%nvar),nhat) | |
| this%pbc2d_NoSlip(this%solution%boundary(i,j,iEl,1:this%nvar),nhat) |
While I agree with this, I've been thinking on this issue of the ever growing number of potential boundary conditions. What I'd propose is a new module with a class defined for mapping boundary conditions to procedure pointers. This class would be something like type SELF_BoundaryCondition
procedure(SELF_BCFunction),pointer :: bcfunction
integer :: bcflag_id
character(len=SELF_BCNAME_LENGTH) :: bcflag_nameWe could set such a class up as a linked list which could allow for dynamic insertion of boundary conditions on the fly.. Having methods like do concurrent(j=1:4,iel=1:this%mesh%nElem)
bcid = this%mesh%sideInfo(5,j,iEl) ! Boundary Condition ID
e2 = this%mesh%sideInfo(3,j,iEl) ! Neighboring Element ID
if(e2 == 0) then
! Get a function pointer for the boundary condition id
bfunc = this%boundaryconditions%GetBCFunc_for_id( bcid )
do i = 1,this%solution%interp%N+1 ! Loop over quadrature points
x = this%geometry%x%boundary(i,j,iEl,1,1:2)
this%solution%extBoundary(i,j,iEl,1:this%nvar) = bcfunc(x,this%t)
endif
enddo |
This PR adds inflow and outflow boundary condition integer parameters and overridable type bound procedures in the template model classes for 2D and 3D models.
Resolves #111