[cfdnewb123]

I have no idea why but when using this code, FLuent keep crashing while the command just show "999999: mpt_accept: error: accept failed: No error".
I have narrow down the issue to C_UDSI(c,t,...) as it works when I remove C_UDSI(c,t,...) and C_UDMI(c,t,...).

I also try initializing before using this however it still doesnt work.

Here is my code,

#include "udf.h"
#include "math.h"

const real C_1=-0.2;
const real C_2=0.4;
const real C_5=0.0;
const real CE_1=1.44;
const real CE_2=1.92;
const real SIG_K=1.0;
const real SIG_E=1.3;
const real C_MU_ke=0.09;
const real kappa=0.42;
const real E_log=9.793;

enum { UU, VV, UV};

DEFINE_ADJUST(rsm_adjust,domain)
{
Thread *t;
cell_t c;
real T;
real mu_t, c_mu;
real S11, S12, S21, S22;
real W11, W12, W21, W22;
real S, W, Sij, Wij;
real C_3, C_4, C_6, C_7;
real P11, P22, P33, Pkk;
real tau_w, u_tauw, u_star, y_star, Gk, u_mag;

thread_loop_c(t,domain)
if (FLUID_THREAD_P(t)) {
begin_c_loop(c,t)
{
S11 = 0.5*( C_DUDX(c,t)+C_DUDX(c,t) );
S12 = 0.5*( C_DUDY(c,t)+C_DVDX(c,t) );
S21 = 0.5*( C_DVDX(c,t)+C_DUDY(c,t) );
S22 = 0.5*( C_DVDY(c,t)+C_DVDY(c,t) );

W11 = 0.5*( C_DUDX(c,t)-C_DUDX(c,t) );
W12 = 0.5*( C_DUDY(c,t)-C_DVDX(c,t) );
W21 = 0.5*( C_DVDX(c,t)-C_DUDY(c,t) );
W22 = 0.5*( C_DVDY(c,t)-C_DVDY(c,t) );

T=C_K(c,t)/C_D(c,t);

S=T*sqrt( 2*(S11*S11+S12*S12+S21*S21+S22*S22) );
W=T*sqrt( 2*(W11*W11+W12*W12+W21*W21+W22*W22) );

Sij=S11*S11+S12*S12+S21*S21+S22*S22;
Wij=W11*W11+W12*W12+W21*W21+W22*W22;

c_mu=0.055;
C_3=2.0-exp(-SQR(S-W));
C_4=-32.0*SQR(c_mu);
C_6=-16.0*SQR(c_mu);
C_7=16.0*SQR(c_mu);

mu_t=c_mu*T*C_K(c,t);

C_UDSI(c,t,UU)= C_1*mu_t*T*(S11*S11+S12*S12-1./3.*Sij) + 2.*C_2*mu_t*T*(W12*S12) + C_3*mu_t*T*(W12*W12-1./3.*Wij) - 2.*C_4*mu_t*T*T *(W12*(S11*S12+S12*S22)) + C_6*mu_t*T*T*(S11*Sij) + C_7*mu_t*T*T*(S11*Wij);

C_UDSI(c,t,VV)= C_1*mu_t*T* (S12*S12+S22*S22-1./3.*Sij) + 2.*C_2*mu_t*T*(-W12*S12) + C_3*mu_t*T*(W12*W12-1./3.*Wij) + 2.*C_4*mu_t*T*T * (W12*(S11*S12+S12*S22)) + C_6*mu_t*T*T* (S22*Sij) + C_7*mu_t*T*T* (S22*Wij);

C_UDSI(c,t,UV)= C_1*mu_t*T* (S11*S12+S12*S22) + C_2*mu_t*T*(W12*(S22-S11)) + C_3*mu_t*T*(0) + C_4*mu_t*T*T*(W12*(S11*S11-S22*S22)) + C_6*mu_t*T*T* (S12*Sij) + C_7*mu_t*T*T* (S12*Wij);

C_UDMI(c,t,0)= -2.*mu_t*S11 + 2./3.*C_K(c,t) + C_UDSI(c,t,UU);
C_UDMI(c,t,1)= -2.*mu_t*S22 + 2./3.*C_K(c,t) + C_UDSI(c,t,VV);
C_UDMI(c,t,2)= -2.*mu_t*S12 + C_UDSI(c,t,UV);
}
end_c_loop(c,t)
}
}

[vinerm]

The purpose of the UDS is not what you are using it for. You only need UDMs. UDS are required if you want conservation equations to be solved. And for UDM and UDS, there are two pre-requisites.

1. Assign the required number of UDM (and UDS, if required)
2. Initialize or patch all UDMs (and UDS) throughout the domain

[cfdnewb123]

Thanks for the response but may I ask how do I assign the number of UDM?

Additionally, is just using UDM sufficient for my case as isn't it just storing memory?
Honestly, I have no idea what I am doing on udf but what I am trying to do is modify the Reynolds stress to be used by the solver. I based the above code from a sample code online I found online though I am not sure whether is it correct or not.

[vinerm]

Fluent solves for primitive fields, such as, velocities and pressure, not for stresses until you use RSM. Only when stress field is being solved, you can modify it, otherwise their modification will not affect the flow. As far as the UDF is concerned, it is doing nothing but saving results of some expressions in UDMs.

To enable UDMs, use User-Defined > Memory.

[cfdnewb123]

Thanks for the clarification. Does this mean that Fluent is unable to solve for nonlinear k-epsilon model where the Reynolds stress is modified?

Additionally, may I ask what cause C_UDSI() to be unusable here? Did I implement it wrongly?

[vinerm]

User can modify the turbulence model, however, the UDF is not doing that. All that the UDF is doing is calculating some functions and storing those in UDMs (or in UDS). But these are not being used to modify the flow in any manner. Hence, these are just passive expressions that can be used for post-processing but not for affecting the flow. If you want to affect the flow, then you have to modify parameters in the flow equations, such as, eddy viscosity.

[cfdnewb123]

Thanks I managed to get UDM to work well based on your suggestion. However, UDS still doesn't seems to work and you are right that I do not really need it here. But it may be good to know as I may need it next time. I actually try using the sample code giving by Fluent too but it encountered an error related to "signal".
https://www.afs.enea.it/project/nept...df/node103.htm

[cfdnewb123]

Oh wait never mind, I resolved this thanks so much!

[vinerm]

Just like UDM, you need to enable scalar under User-Defined > Scalar