entropy generation with UDF
1 Attachment(s)
Dear all
I want to compute entropy generation in tube with constant heat flux. I wrote a udf to compute gradient temperature: # include "udf.h" # include "math.h" DEFINE_ADJUST(adjust_gradient, domain) { Thread *t; cell_t c; face_t f; domain = Get_Domain(1); /* Fill UDS with the variable. */ thread_loop_c (t,domain) { begin_c_loop (c,t) { C_UDSI(c,t,0) = C_T(c,t); } end_c_loop (c,t) } thread_loop_f (t,domain) { if (THREAD_STORAGE(t,SV_UDS_I(0))!=NULL) begin_f_loop (f,t) { F_UDSI(f,t,0) = F_T(f,t); } end_f_loop (f,t) } } DEFINE_ON_DEMAND(store_gradient) { Domain *domain; cell_t c; Thread *t; domain=Get_Domain(1); /* Fill the UDM with magnitude of gradient. */ thread_loop_c (t,domain) { begin_c_loop (c,t) { C_UDMI(c,t,0) =(pow(NV_MAG(C_UDSI_G(c,t,0)),2)); } end_c_loop (c,t) } } //////////////////////////////// after the iteration in custom field function, i defined: dissip=(2 * ((daxialvelocitydx) ^ 2)) + ((daxialvelocitydy) ^ 2) S(generation)=((thermalconductivitylam * udm0) / temperature ^ 2) + ((viscositylam * dissip) / temperature) with this way i compute entropy generation in fluent.I can compile my code and get answer without any problem but my result is wrong(not validated with paper). I think my code for compute gradient temperature is incomplete and i need to compute gradient temperature in face too, but i dont know how can i do that? please help with this problem thanks 
I think you can easyly skip all the troubles by directly using the temperature gradient provided by fluent?
However, to be able to access gradients you need to tell fluent to store them (check the udf manual section 3.2.3  solve/set/expert). A suggestion: instead of Code:
C_UDMI(c,t,0) =(pow(NV_MAG(C_UDSI_G(c,t,0)),2)); Code:
C_UDMI(c,t,0) =NV_MAG2(C_UDSI_G(c,t,0)); 
Dear dragos
How can i directly use gradient temperature provided with fluent? there was no gradient temperature(dT/dx) in custom field function panel to use!! I did that(solve/set/expert) but my result was not changed. I tried this code to compute gradient face temperature: # include "udf.h" # include "math.h" DEFINE_ADJUST(adjust_gradient, domain) { Thread *t; cell_t c; face_t f; domain = Get_Domain(1); /* Fill UDS with the variable. */ thread_loop_c (t,domain) { begin_c_loop (c,t) { C_UDSI(c,t,0) = C_T(c,t); } end_c_loop (c,t) } thread_loop_f (t,domain) { if (THREAD_STORAGE(t,SV_UDS_I(0))!=NULL) begin_f_loop (f,t) { F_UDSI(f,t,0) = F_T(f,t); } end_f_loop (f,t) } } DEFINE_ON_DEMAND(store_gradient) { Domain *domain; cell_t c; Thread *t; face_t f; domain=Get_Domain(1); /* Fill the UDM with magnitude of gradient. */ thread_loop_c (t,domain) { begin_c_loop (c,t) { C_UDMI(c,t,0) =(pow(NV_MAG(C_UDSI_G(c,t,0)),2)); } end_c_loop (c,t) } thread_loop_f(t, domain) { begin_f_loop(f,t) { F_UDMI(f,t,0) =(pow(NV_MAG(F_UDSI_G(f,t,0)),2)); } end_f_loop(f, t) } } ///////////////////////// but i got this error: ..\..\src\test.c(48) : error C2109: subscript requires array or pointer type ..\..\src\test.c(48) : error C2109: subscript requires array or pointer type ..\..\src\test.c(48) : error C2109: subscript requires array or pointer type ..\..\src\test.c(48) : error C2109: subscript requires array or pointer type /////////////////////////////////////////////// please help me i don't know how to solve my problem. i am reletively new to UDF. thanks 
It seems either I missunderstood again your problem, or you mixed up concepts.
To clarify: if you need temperature gradient only as a field function, then you don't need udf's. To get the temperature gradient available in field function you have to tell fluent to store it after each iteration (presuming that you have enabled the energy equation first). If you had red the manual page I suggested, you would have found, that storing the gradients is straight forward: Code:
/solve/set/expert 
Dear dragos
you solved my problem, thank for your help. As you said there is no need to write udf, i just used "/solve/set/expert". After that i changed criteria of energy from 1e6 to 1e12 and my solution was validated. thanks again 
Hi,
I am solving a 2D problem. I want to write the value of temperature gradient on a surface (I need the values). Can anyone tell me how to do it......I think a udf is to be written but I dont understand which macro to use. I have the end point coordinates of the surface(line) 
you don't need to write udf for calculating the gradient. you should just keed the data by typing a command in the console. I can't remember the command but you can find it in the UDF PDF file.

I want the gradient value on a wall i.e, surface.....will it be possible

I think it is possible. And the command to keep the memory is :
/solve/set/expert ... Keep temporary solver memory from being freed? [no] yes good luck 
hi,
could you please write the custom field function for entropy generation 
Dear Rahul,
Custom field function can be written based on the formula that I have attached in the first post. For adding gradient temperature you need to type the following statement in the cosole: /solve/set/expert...Keep temporary solver memory from being freed? [no] yes...Good luck 
Regarding Entropy generation
I tried to calculate the entropy generation using the command: solve/set/expert
Now, I got the reconstructed dTdx, dTdy and dTdz. Can these values directly be used in my custom Field function to compute entropy. Can any one help me out mail me: ashok_iit@hotmail.com 
how i enter attached equation in custom filed function?
Quote:
Hi mohammad how i enter equation that you attached in custom filed function? please help me about this thnx 
entropy generation custom field function
Dear all
I entered the entropy generation equation in custom field function.but my result is wrong(not validated with paper). I examined my results and found that Fluent calculate the temperature gradient, but it can not properly calculate the square of the temperature gradient!!! please help with this problem thanks 
There are two options:
 Fluent has a bug and calculates the square of the gradient of temperature wrong.  You did something wrong when you tried to let Fluent calculate the square of the gradient of the temperature. You seem to think that the second option is very unlikely. I think the first option is very unlikely. 
Dear pakk
of course, i think the first option is very unlikely. But I really do not know what's the problem. For a specific position within the computational domain, when i account the entropy based on the components separately, there is difference between the answer with custom field printing result. Interestingly, this difference can be seen when i want to calculate the square of the temperature gradient ! 
So you agree that it is possible that you did something wrong when you tried to calculate the square of the temperature gradient, but you don't know what it is you could have done wrong.
And what now? I think it is impossible to get help on which of your steps could be wrong, if you don't say which steps you took. But that is just my opinion. 
All times are GMT 4. The time now is 00:24. 