Dear Neks, I found some suspicious codes in the subroutine x_average of navier5.f, which are ua(j,k,ey,ez) = ua(j,k,ey,ez)+dx2*wzm1(i)*u(i,j,k,e) w1(j,k,ey,ez) = w1(j,k,ey,ez)+dx2*wzm1(i) ! redundant but clear I think the wzm1 should be wxm1? Thanks, Peng