[gemini]

Hi everyone,

I want to calculate the gradient of a UDS by using Green Gauss theorem using cell based approximation for face values of UDS. Below you can see the corresponding c code. However i created a simple test case which contains a square domain (in which UDS solved using dirichlet type BCs) and test whether my gradient function and C_UDSI_G(c, t, uds_no) gives the same values. The result is that my function gives totally different and unrealistic gradient values while C_UDSI_G gives correct values,

So anybody can help me to correct the code below:

I used both uniform and nonuniform structured grids.

thanks so much

real evaluate_gradient(cell_t c, Thread *t, int i_uds, real *x_vec, real *y_vec, real *z_vec)
{
    int n;
    face_t f;
    Thread *tf;
    Thread *t0, *t1;
    cell_t c0, c1;
    real A[ND_ND];
    real grad_phi[ND_ND];
    ND_SET(grad_phi[0],grad_phi[1],grad_phi[2],0.0,0.0,0.0);
    
    c_face_loop(c, t, n)
    {
        f  = C_FACE(c,t,n);
        tf = C_FACE_THREAD(c,t,n);
        c0 = F_C0(f,tf);
        t0 = THREAD_T0(tf);
        if (!BOUNDARY_FACE_THREAD_P(tf))
        {
            c1=F_C1(f,tf);
            t1=THREAD_T1(tf);
        }
        else
        {
            c1=-1;
            t1=NULL;
        }
        F_AREA(A,f,tf);
        if (!BOUNDARY_FACE_THREAD_P(tf))
            NV_V_VS(grad_phi,=,grad_phi,+,A,*,(0.5*(C_UDSI(c0,t0,i_uds)+C_UDSI(c1,t1,i_uds))));
        else
            NV_V_VS(grad_phi,=,grad_phi,+,A,*,C_UDSI(c0,t0,i_uds));
    }
    *x_vec = grad_phi[0]/C_VOLUME(c,t);
    *y_vec = grad_phi[1]/C_VOLUME(c,t);
    *z_vec = grad_phi[2]/C_VOLUME(c,t);

    return 0.0;
}