A small question about speed correction

[yzl429]

Dear xcompact developers,
In the subroutine pre_correc(ux,uy,uz,ep) within navier.f90, when the boundary condition nclx1==2 is met, the pressure gradient dpdyx1(j,k) is used to correct the velocity: uy(1,j,k) = bxy1(j,k) + dpdyx1(j,k). However, in the earlier subroutine gradp(px1, py1, pz1, pp3), when the boundary condition nclx1==2 is met, dpdyx1(j,k) is calculated as py1(1,j,k)/gdt(itr), normalizing the pressure gradient. My question is whether py1(1,j,k) itself already incorporates a time component. Therefore, in uy(1,j,k) = bxyn(j,k) + dpdyxn(j,k), dpdyxn(j,k) appears to be equivalent to py1(1,j,k).

[mathrack]

Hello,

The pressure pp3 is computed using the divergence of the predicted velocity. The resulting pressure gradient thus includes the time step of the sub-iteration. See https://github.com/xcompact3d/Incompact3d/blob/master/src/navier.f90#L206

For a discussion on prediction / correction time schemes, see https://fischerp.cs.illinois.edu/cs598/assignments/guermond_minev_shen_CMAME_2006.pdf

Best regards,
Mathrack