[Mohammad74]

Hi,

First I must say I'm pretty new to Ansys Fluent UDFs. I'm trying to solve a simple problem with a cubic domain and a custom velocity profile at inlet boundary. I've read the Ansys users' guide on UDFs and I cant find any problems with my code, however when I begin the simulation, Fluent gives me this error:

"Divergence detected in amg solver: x-momentum"

Sorry if my question is a bit obvious but what am I missing? Can someone please help me?

BTW my Fluent version is 17.0 and I'm using MVS Express 2012. Also here is my code:

 
#include "udf.h"

#define A_f	22			/* Frontal built area */
#define A_p	22			/* Planar built area */
#define A_tot 100			/* Total land area */
#define H 10				/* Mean building height-Refrence height */
#define U_H	6			/* Wind velocity at height H */
#define A 4.4			/* Building arrangement constant */
#define Cd 1.2			/* Drag coefficient for an staggered array of cubes */
#define Beta 0.55		/* Building drag correction factor */
#define k_von 0.4		/* von-Karman constant */

DEFINE_PROFILE(x_velocity,t,i)
{
	double lambda_f, lambda_p, a, d, zz, z0, U_s, x[ND_ND], z;	/* Variable declarations */
	face_t f;
	lambda_f = A_f/A_tot;				/* Frontal area density */
	lambda_p = A_p/A_tot;				/* Planar area density */
	a = 9.6*lambda_f;					/* Building morphology constant */
	d = H*(1+(pow(A,-lambda_p))*(lambda_p-1));		/* Displacement length */
	zz = 0.5*Beta*Cd*pow(k_von,-2)*(1-(d/H))*lambda_f;
	z0 = H*(1-(d/H))*exp(-(pow(zz,-0.5)));			/* Surface roughness */
	U_s = (U_H*k_von)/(log((H-d)/z0));				/* Friction velocity */
	begin_f_loop(f,t)
	{
		F_CENTROID(x,f,t);
		z = x[2];
		
		if (z <= H)
			F_PROFILE(f,t,i) = U_H*exp(a*((z/H)-1));
		else
			F_PROFILE(f,t,i) = (U_s/k_von)*log((z-d)/z0);

	}
	end_f_loop(f,t)
}

[pakk]

Initialize, don't solve, plot velocity on your BC.

Is it what you expect?

My guess is that you made a mistake in one of the equations, which gives you to high velocities.

[Mohammad74]

Hello again and thank you for your response,

I did as u said and plotted velocity profile on inlet BC after standard initialization (hybrid initialization gives me a similar error). Unfortunately it does not plot anything...
So in order to check my equations I used Matlab to plot them and they work as they should be. I attached an image of the Matlab plot here too.
I also noticed that in initial values fields, values for k, omega and x velocity are set to " 1.#QNAN " !!
Any other ideas?

vplot.jpg

[pakk]

Originally Posted by Mohammad74

I also noticed that in initial values fields, values for k, omega and x velocity are set to " 1.#QNAN " !!
Any other ideas?

[

Fix that problem, and set them to number values!

[Mohammad74]

I already did that before posting the reply and it did not help.

[pakk]

If you replace your UDF boundary with a simple boundary (velocity 10 m/s for example), do you get the same error? If so, it has nothing to do with the UDF.

And maybe your fluid properties are wrong?

[Mohammad74]

Hi, I've tried that with a constant 6 m/s velocity at inlet and it worked fine.

Also other things that I've tried but still got the same error:

• I used another domain and mesh quality.
• Changed the working fluid.
• Used another turbulence model.
• I even tried to use a sample code for velocity profile given in Fluent's UDF guide but even that gave the same error!

[AlexanderZ]

check this code

#include "udf.h"
DEFINE_PROFILE(vel_profile,t,i)
{
real x[ND_ND]; /* this will hold the position vector */
real y;
face_t f;
begin_f_loop(f,t)
{
F_CENTROID(x,f,t);
y = x[1];
F_PROFILE(f,t,i) = y*6;
}
end_f_loop(f,t)
}

[Mohammad74]

Originally Posted by AlexanderZ

check this code

Thank you, this worked fine and I was able to run the simulation. But I'm still trying to figure out what is wrong with my UDF...

[AlexanderZ]

your code seems to be ok,
the only thing I would change is double type to real here:

	real lambda_f, lambda_p, a, d, zz, z0, U_s, x[ND_ND], z;	/* Variable declarations */

compile code

[Mohammad74]

Originally Posted by AlexanderZ

your code seems to be ok,
the only thing I would change is double type to real here:

nope, its still not working. gives the same error.

[pakk]

You have some tabs in your #define's, which might be a problem (I don't know if that is allowed or not).

And you can write some values to show if there is an intermediate problem, like this:

#include "udf.h"

#define A_f 22			/* Frontal built area */
#define A_p 22			/* Planar built area */
#define A_tot 100			/* Total land area */
#define H 10				/* Mean building height-Refrence height */
#define U_H 6			/* Wind velocity at height H */
#define A 4.4			/* Building arrangement constant */
#define Cd 1.2			/* Drag coefficient for an staggered array of cubes */
#define Beta 0.55		/* Building drag correction factor */
#define k_von 0.4		/* von-Karman constant */

DEFINE_PROFILE(x_velocity,t,i)
{
	double lambda_f, lambda_p, a, d, zz, z0, U_s, x[ND_ND], z;	/* Variable declarations */
	face_t f;
	lambda_f = A_f/A_tot;				/* Frontal area density */
	lambda_p = A_p/A_tot;				/* Planar area density */
	a = 9.6*lambda_f;					/* Building morphology constant */
	d = H*(1+(pow(A,-lambda_p))*(lambda_p-1));		/* Displacement length */
	zz = 0.5*Beta*Cd*pow(k_von,-2)*(1-(d/H))*lambda_f;
	z0 = H*(1-(d/H))*exp(-(pow(zz,-0.5)));			/* Surface roughness */
	U_s = (U_H*k_von)/(log((H-d)/z0));				/* Friction velocity */
	begin_f_loop(f,t)
	{
		F_CENTROID(x,f,t);
		z = x[2];
		
		if (z <= H)
			F_PROFILE(f,t,i) = U_H*exp(a*((z/H)-1));
		else
			F_PROFILE(f,t,i) = (U_s/k_von)*log((z-d)/z0);

	}
	end_f_loop(f,t)

Message("z=%f ",z);
Message("U_s=%f ",U_s);
Message("z0=%f ",z0);
Message("zz=%f ",zz);
Message("d=%f ",d);
}

This will just show the values of the last face that was visited, but it might just be enough to help understand what is going on.

[Mohammad74]

Hi,

So I've finally managed to solve this error. Apparently "#define" that I used to define some constants was not working properly so constant values were set to zero in equations. I used the method below to define my constants and its working now:

const float A_f=22, A_p=22, A_tot=100, H=10 ,U_H=6, A=4.4, Cd=1.2, Beta=0.55, k_von=0.4;

I'm still confused about why "#define" is not working though.

[pakk]

Originally Posted by Mohammad74

Hi,

I'm still confused about why "#define" is not working though.

Thank you for completely ignoring my message directly above yours.

[Mohammad74]

Originally Posted by pakk

Thank you for completely ignoring my message directly above yours.

No I did not ignore it. Those are not tabs but spaces and that is how the syntax of #define works.

And about using the "Message" command, I did try it but it was not working since my simulation could not even start to run that command.
Anyhow, thank you so much for taking the time to answer my question.

[pakk]

The code that you posted here has tabs. If you don't believe me, check for yourself, after A_f, A_p and U_H.

But anyway, good that it is solved.