[say2017]

Hello,

I am trying to simulate a heat pipe and have managed to compile the code with no errors. However, as soon I load in one of the source terms, I get an error that says "Divergence detected in AMG solver: x-momentum". I have tried changing the URFs and reducing the time step but I still get this error. I have included the code for the mass source, which is the only source term I am currently loading, below. Any help will be appreciated! Thank you.

DEFINE_SOURCE(mass_source,c,t,dS,i)
{
face_t f;
Thread *tf;
int n;
int nn;
real S = 0;

/* loop over all faces of cell */
c_face_loop(c,t,n){
/* find face and thread */
f = C_FACE(c,t,n);
tf = C_FACE_THREAD(c,t,n);

for(nn=0;nn<N_WICK_VAPOR_INTERFACE; ++nn){
if ((THREAD_ID(tf)==t_wall_wick_id[nn])||(THREAD_ID(tf)==t_wall_vapor_id[nn])){
if (THREAD_ID(tf)==t_wall_wick_id[nn]){
S += F_UDMI(f,tf,MVAR)/C_VOLUME(c,t);
}
else{
S -= F_UDMI(f,tf,MVAR)/C_VOLUME(c,t);
}
}
}
}
return S;
}

[ahmed425]

I'm not expert in UDFs but I always see that the arguments of any source function ends with a variable called eqn and at the end before the return statements there is a command like: dS[eqn]=0;

this might be one reason

the other thing that I can think of is you can initialize your model and solve for a number of iterations without the source then you can add the source and see what happens.

good luck

[Micael]

For verification, return zero value and put "S" in a UDM instead so you can visualise it and assess it is right.

Otherwise, ahmed425's idea to do few iterations before including the UDF is worth trying. Sometime source term at very first iteration can cause problem.

[say2017]

Thank you ahmed and micael. I have tried running it without the UDF at first and implementing it later but it still didn't work. However, I did finally manage to solve my problem. If anyone else has the same issue, try using the coupled solver especially if you have a transient simulation.