[majid_kamyab]

the case is a 2 phase vapor-liquid modeled by Eulerian framework and k-e
and there are only 2 species. there is a mass transfer between the phases
I have written a UDF for mass transfer but an error occurs when I try to run calculation in parallel mode and when I run in series access violation error occurs
The code is written below


#include "udf.h"
DEFINE_MASS_TRANSFER(liq_gas_source,c,t,from_index ,from_species_index, to_index, to_species_index)
{
real m_lg,c_meth_l,c_meth_v,k_l,dal,usg,kappa,m,c_sat;
dal=1.46*0.000000001;
usg=0.294188;
kappa=0.000484011/748;
m=1.204;
k_l=2*sqrt((dal/3.1415)*sqrt(usg*9.8/kappa));
m_lg = k_l*(C_YI(c,from_index,from_species_index)-m*C_YI(c,to_index,to_species_index));
return (m_lg);
}


the code can be interpreted but during the first calculation the following error occurs:




Updating solution at time level N... done.
iter continuity u-mixture-li u-mixture-ga v-mixture-li v-mixture-ga w-mixture-li w-mixture-ga k-mixture-li eps-mixture- ch3oh<l>-mix n-propanol-g vf-mixture-g time/iter
!417331 solution is converged
417331 7.7100e-04 2.3939e-04 6.4021e-05 5.8865e-04 1.0116e-04 2.8676e-04 8.3530e-05 1.4561e-04 2.6601e-04 0.0000e+00 0.0000e+00 2.7556e-05 0:01:40 100

999999 (..\src\mpsystem.c@1173): mpt_read: failed: errno = 10054

999999: mpt_read: error: read failed trying to read 4 bytes: Invalid argument
MPI Application rank 0 exited before MPI_Finalize() with status -1073741819
The Parallel FLUENT process could not be started.

[pakk]

When using parallel, compiling is usually much safer than interpreting. Try that.

[majid_kamyab]

Originally Posted by pakk

When using parallel, compiling is usually much safer than interpreting. Try that.

so there is no problem with the code?

[pakk]

Originally Posted by majid_kamyab

so there is no problem with the code?

I don't know. Your primary problem is that you try interpret for parallel calculations, where compile is more appropriate.

I could spend the next hour figuring out if your code would give errors if you would compile it, but I am not going to do that, because it is much faster for you if you just compile it and see what the result is.

[ghost82]

Did you read this?
http://www.cfd-online.com/Forums/ans...g-mpi-why.html

It seems your problem is not related to the code.

[lehoanganh07]

Your code have problem both series and parallel simulation?
I think the problem is C_YI(cell,thread,i) but your code C_YI(cell, index,i)
So I think it may cause error

[majid_kamyab]

Yes I have already read that
my RAM is 8 Gb and my mesh is about 200,000 Cells
so I dont think the problem is with the RAM

[pakk]

Is there a reason why you don't take 2 minutes to compile it and tell us if that solves the problem?

[majid_kamyab]

Originally Posted by pakk

Is there a reason why you don't take 2 minutes to compile it and tell us if that solves the problem?

Dear pakk the following error occurs when I try to load the library

the UDF library you are trying to load (libudf) is not compiled for parallel use in current platform (win64)

as I mentioned earlier I couldnt interpret it even in serial mode
I am a new user and I was trying to learn how to compile
and thats why I didnt answer quickly
thanks in advance

[majid_kamyab]

Originally Posted by lehoanganh07

Your code have problem both series and parallel simulation?
I think the problem is C_YI(cell,thread,i) but your code C_YI(cell, index,i)
So I think it may cause error

I think that the problem is with the C_YI too.
I have tried
Thread *gas=THREAD_SUB_THREAD(t,to_index);
Thread *gas=THREAD_SUB_THREAD(t,to_index);
but there are another parse errors in these lines
thanks in advance

[pakk]

Originally Posted by majid_kamyab

Dear pakk the following error occurs when I try to load the library

the UDF library you are trying to load (libudf) is not compiled for parallel use in current platform (win64)

as I mentioned earlier I couldnt interpret it even in serial mode
I am a new user and I was trying to learn how to compile
and thats why I didnt answer quickly
thanks in advance

Ok.
But my question was not what happens if you try to load the library, but what happens if you try to compile your file. Apparantly, that gave an error. Look at that error to see what is wrong with your code.

[pakk]

Example of how I would debug this:

I tried to compile your code, and saw the following error:

..\..\src\test.c(10) : error C2223: left of '->storage' must point to struct/union
..\..\src\test.c(10) : error C2223: left of '->storage' must point to struct/union

This means that something is wrong in (or before) line 10.
Line 10 is:

m_lg = k_l*(C_YI(c,from_index,from_species_index)-m*C_YI(c,to_index,to_species_index));

So I look at the definition of C_YI (Fluent UDF manual 3.2.3.6.):

C_YI(c,t,i)

cell_t c, Thread *t, int i

Translated to your code, c should be a cell_t, from_index should be a Thread*, and from_species_index should be an int.
But, from the definition of DEFINE_MASS_TRANSFER (2.4.6):
C is indeed cell_t, from_index is an int, and from_species_index is also an int.

So, what is wrong? The second input variable of C_YI.

Looking at variable types alone, the following looks more reasonable:

m_lg = k_l*(C_YI(c,t,from_species_index)-m*C_YI(c,t,to_species_index));

[majid_kamyab]

Originally Posted by pakk

Example of how I would debug this:

I tried to compile your code, and saw the following error:


This means that something is wrong in (or before) line 10.
Line 10 is:

m_lg = k_l*(C_YI(c,from_index,from_species_index)-m*C_YI(c,to_index,to_species_index));

So I look at the definition of C_YI (Fluent UDF manual 3.2.3.6.):


Translated to your code, c should be a cell_t, from_index should be a Thread*, and from_species_index should be an int.
But, from the definition of DEFINE_MASS_TRANSFER (2.4.6):
C is indeed cell_t, from_index is an int, and from_species_index is also an int.

So, what is wrong? The second input variable of C_YI.

Looking at variable types alone, the following looks more reasonable:

m_lg = k_l*(C_YI(c,t,from_species_index)-m*C_YI(c,t,to_species_index));

In this code the mass fraction of the gas phase and the liquid phase must be used but in the
C_YI(c,t,from_species_index)
the mass fraction of the cell is used isnt that right?
I have used THREAD_SUB_THREAD for identification of the gas phase and also the liquid phase is it correct?(I have mentioned in the last reply

[pakk]

The correct syntax depends on what you want your code to do.
Your equation is:

m_lg = k_l*(A-m*B);

Here, A and B are mass fractions. Which mass fractions do you want them to be?

In "C_YI(c,t,i)", c and t indicate which cell you want to look at, and i indicates which mass fraction you want to use. I am fairly sure about the c and the t, but to know which i you should use (from_index,from_species_index, to_index or to_species_index), you should look at the problem you want to solve. All four are acceptable from a programming point of view, they just give solutions to different physical problems.

[majid_kamyab]

Originally Posted by pakk

The correct syntax depends on what you want your code to do.
Your equation is:

m_lg = k_l*(A-m*B);

Here, A and B are mass fractions. Which mass fractions do you want them to be?

In "C_YI(c,t,i)", c and t indicate which cell you want to look at, and i indicates which mass fraction you want to use. I am fairly sure about the c and the t, but to know which i you should use (from_index,from_species_index, to_index or to_species_index), you should look at the problem you want to solve. All four are acceptable from a programming point of view, they just give solutions to different physical problems.

let me be more specific
there are two species ( methanol and propanol) and I want to calculate mass fraction of methanol IN THE LIQUID PHASE and the mass fraction of methanol IN THE VAPOR PHASE
so I dont want the mass fraction of the liquid phase or methanol in the cell.
as I think t is a mixture thread (super thread) and I think I am making mistake using THREAD_SUB_THREAD

[pakk]

I am not experienced enough in species' udfs to tell you what to use. I can only tell you that from a programming point of view, you select the species phase with the third input, and that should be an int.

[majid_kamyab]

Ive got my answer
The macro C_YI(cell, thread, I) should be used in the form of following.

thread_loop_c(thread, domain) { if( THREAD_ID(t) == 8 ) { begin_c_loop(cell, thread) {

C_YI(cell, thread, I) ; } } }

The value "8" specified above is the "thread id" that will be got in Define-Boundary Conditions . And , "I" of C_YI(cell, thread, I) shows species which you want to get the mass fraction.

[majid_kamyab]

Originally Posted by pakk

I am not experienced enough in species' udfs to tell you what to use. I can only tell you that from a programming point of view, you select the species phase with the third input, and that should be an int.

thanks a lot for the time you have spent on this Thread
Ive got my answer from another Thread
I posted it here too if you are interested