Discrete inner-product on non-uniform mesh

[MarcoArtiano]

Hi guys, I'm trying to write the same post-process that appears in BC-TGV.f90, but for cases where the mesh is non uniform, for example, x and z are uniform and y is stretched.

       do k=1,xsize3
          do j=1,xsize2
             do i=1,xsize1
                temp1=temp1+zpfive*((ux1(i,j,k))**2+(uy1(i,j,k))**2+(uz1(i,j,k))**2)
             enddo
          enddo
       enddo
       call MPI_ALLREDUCE(temp1,eek,1,real_type,MPI_SUM,MPI_COMM_WORLD,code)
       eek=eek/(nxc*nyc*nzc)

I thought to modify that as follows

temp1=zero
    do k=1,xsize3
       do j=1,xsize2
         if (j.eq.1) then
               dyv = (yp(2)-yp(1))*half
         elseif (j.eq.xsize2) then
            dyv = (yp(xsize2)-yp(xsize2-1))*half
         else
            dyv = (yp(j)+yp(j-1))*half
         endif
           do i=1,xsize1
              temp1=temp1+dyv*zpfive*((ux1(i,j,k))**2+(uy1(i,j,k))**2+(uz1(i,j,k))**2)
          enddo
       enddo
    enddo
    call MPI_ALLREDUCE(temp1,Ec,1,real_type,MPI_SUM,MPI_COMM_WORLD,code)
  Ec = Ec/(nxc*nzc)

But dunno if this is correct or not, or should I go for like computing the mean value in the cell center and multiplying for the volume of the cell, so I don't have to take half volume for the points collocated at the edge of the domain.

Thanks so much in advance

[mathrack]

Hello,

You can use something based on the estimation of the averaged velocity available in the channel flow subroutines : https://github.com/xcompact3d/Incompact3d/blob/master/src/BC-Channel-flow.f90#L200

Best regards,
Mathrack

[MarcoArtiano]

Oh thank you so much, I missed that !