[shivakumar]

After compiling udf in parallel mode, the error appeared in this manner. In fact, I changed the double to single precision as mentioned in another thread, still getting the same error. How to fix this?

Thanks in advance

================================================== ============================

Node 0: Process 9872: Received signal SIGSEGV.

================================================== ============================

================================================== ============================

Node 1: Process 6252: Received signal SIGSEGV.

================================================== ============================

================================================== ============================

Node 2: Process 3308: Received signal SIGSEGV.

================================================== ============================

================================================== ============================

Node 3: Process 2408: Received signal SIGSEGV.

================================================== ============================

================================================== ============================

Node 4: Process 6792: Received signal SIGSEGV.

================================================== ============================

================================================== ============================

Node 5: Process 9236: Received signal SIGSEGV.

================================================== ============================

================================================== ============================

Node 6: Process 3832: Received signal SIGSEGV.

================================================== ============================

================================================== ============================

Node 7: Process 9096: Received signal SIGSEGV.

================================================== ============================
MPI Application rank 0 exited before MPI_Finalize() with status 2
The fl process could not be started.

[`e`]

Here's an answer about what the SIGSEGV errors represent. If you still can't make sense of your error, perhaps post your UDF code so that we can read it.

[shivakumar]

Thanks for the link. The error was due to memory allocation. In fact, after allocating the memory also the same error is displaying.
I am working on supersonic flow over a cylinder with sharp edged spike placed at front. from the below code, I am trying to visualize the pattern of shock wave and viscous effects in front of the cylinder. Here is the code. I am attaching the equation as well and the equation in attached file is for 1D flow, my case is 2D. Can you help me?

#include "udf.h"
#include "mem.h"
#include "sg.h"
#include "math.h"

#define M_infi 1.2
#define a 340.26

DEFINE_ADJUST(sensors,d)
{
	#if PARALLEL
	
	Thread *t;
	cell_t c;
	
	real f_shock, f_viscous;
	real first_term, second_term;
	
	real f_spurious;
	
	f_spurious=0;
	
	d=Get_Domain(1);
	
	thread_loop_c(t,d)
	{
		begin_c_loop(c,t)
		{
			first_term = ( - a / NV_MAG(C_P_G(c,t))) * ( (C_R(c,t) * C_U_G(c,t)[0]) + (C_U(c,t) * C_R_G(c,t)[0]) + (C_R(c,t) * C_V_G(c,t)[1]) + (C_V(c,t) * C_R_G(c,t)[1]) );
			
			second_term =  ( ( (C_U(c,t) * C_P_G(c,t)[0]) * (C_V(c,t) * C_P_G(c,t)[1]) ) * M_infi ) / 
							
						( ( sqrt( (C_U(c,t) * C_U(c,t)) + (C_V(c,t) * C_V(c,t)) ) ) * (NV_MAG(C_P_G(c,t))) ); 
			
			f_shock = first_term + second_term;
			
			f_viscous = (C_MU_L(c,t)+C_MU_T(c,t)) / C_MU_L(c,t);
			
			if(f_shock>=1)
			{
				C_UDMI(c,t,0)=f_shock;
			}
			
			if(f_viscous>1.1)
			{
				C_UDMI(c,t,1)=f_viscous;
			}
			
			if((f_shock<1) && (f_viscous<=1.1))
			{
				C_UDMI(c,t,2)=f_spurious;
			}
		}
		end_c_loop(c,t)
	}
	#endif
}

Thanks in advance

[`e`]

You've used three UDM spaces with the IDs 0, 1 and 2. Have you enabled three UDM slots?

[shivakumar]

Originally Posted by `e`

You've used three UDM spaces with the IDs 0, 1 and 2. Have you enabled three UDM slots?

Yup. I did.

[shivakumar]

and still the error is same.

[pakk]

You use the pressure gradient, which might be removed from memory by Fluent.

Did you set "solve/set/expert", as mentioned in the Fluent manual section "3.2.3.7. Gradient (G) and Reconstruction Gradient (RG) Vector Macros"?

[shivakumar]

I did that one also but still the error is same.

solve/set/expert
use alternate formulation for wall temperatures? [yes] yes
Keep temporary solver memory from being freed? [yes] yes
Allow selection of all applicable discretization schemes? [yes] yes
Explicit underrelaxation value: [0.75] 0.75

>
iter continuity x-velocity y-velocity energy k omega Cl-1-cylinder Cd-1-cylinder time/iter
Library "C:\xxx\ansys15\libudf\win64\2ddp_node\libudf. dll" opened
================================================== ============================

Node 0: Process 5284: Received signal SIGSEGV.

================================================== ============================

================================================== ============================

Node 1: Process 6200: Received signal SIGSEGV.

================================================== ============================

================================================== ============================

Node 2: Process 6284: Received signal SIGSEGV.

================================================== ============================

================================================== ============================

Node 3: Process 8388: Received signal SIGSEGV.

================================================== ============================

================================================== ============================

Node 4: Process 5924: Received signal SIGSEGV.

================================================== ============================

================================================== ============================

Node 5: Process 7144: Received signal SIGSEGV.

================================================== ============================

================================================== ============================

Node 6: Process 2692: Received signal SIGSEGV.

================================================== ============================

================================================== ============================

Node 7: Process 7544: Received signal SIGSEGV.

================================================== ============================
MPI Application rank 0 exited before MPI_Finalize() with status 2
The fl process could not be started.

[pakk]

Then I guess you need to really debug it in the hard way.
Add messages like "Code ran to here", to find out when your code gives errors.

[shivakumar]

Thank you

Ok.. May be.... I resolved that equation properly. Let me check it once again. I have a doubt in the above code, To find out the du/dx, this macro is correct right? C_U_G(c,t)... I used the same macro earlier but it had yielded compilation error "real[2] .....". After that I added C_U_G(c,t)[0] and did not yield any error. I feel like this may be the problem.

[pakk]

Originally Posted by shivakumar

Thank you

Ok.. May be.... I resolved that equation properly. Let me check it once again. I have a doubt in the above code, To find out the du/dx, this macro is correct right? C_U_G(c,t)... I used the same macro earlier but it had yielded compilation error "real[2] .....". After that I added C_U_G(c,t)[0] and did not yield any error. I feel like this may be the problem.

I am sorry, but I don't understand what you mean here. But you seem to be confused on C_U_G(c,t), so I can say something about that.

C_U_G(c,t) is the gradient of the x-component of the velocity.This is a vector, so it has three components. The three components are dux/dx=C_U_G(c,t)[0], dux/dy=C_U_G(c,t)[1] and dux/dz=C_U_G(c,t)[2].

I don't know if that is the same as your "du/dx", because I don't know exactly what that is. I am not stupid, so of course I can see it is some derivative of some velocity, but the notation is exact enough to see if this is a scalar, a tensor or a vector. (Does it have 1 component, 3 or 9?)

[shivakumar]

I have just debugged the code and there is a problem in the first_term (code) statement.

In the first term of RHS of eqn, there is term

(du/dx * rho) + (d(rho)/dx * u)

The resultant term is the vector in this case because du/dx and d(rho)/dx are velocity gradient vector and density gradient vector respectively.

Now can you explain this?

[pakk]

It seems to me that you have already explained that it is a vector. I don't understand what it is you would like me to explain.

[shivakumar]

Thank you for the help. I deduced the whole equation into simple format and now fluent is working fine..

shiv