
[Sponsors] 
August 7, 2015, 15:19 
Different values for identical UDS and UDM

#1 
Member
Farzin
Join Date: Jul 2014
Posts: 42
Rep Power: 10 
Hello everyone,
I wrote a UDF that calculates a quantity for a User Defined Scalar, then in another c_loop values of UDS are stored in a User Defined Memory. After calculations, contours of the UDM and UDS differ from each other significantly. It seems that UDM contour is the correct one. But why they are different? Thanks for any suggestion, Farzin 

August 11, 2015, 06:32 

#2 
Senior Member
Bruno Machado
Join Date: May 2014
Posts: 271
Rep Power: 11 
Could you post the part of the code you are talking about?
Them we can see if there is any mistake and help you. 

August 11, 2015, 08:59 

#3  
Member
Farzin
Join Date: Jul 2014
Posts: 42
Rep Power: 10 
Quote:
Here's the related part of the UDF: Code:
DEFINE_ADJUST(adjust_vorticity_calc,domain) { int n; Thread *t,t0,t1; cell_t c,c0,c1; face_t f; //Calculating Vorticity for Cell Centers thread_loop_c(t,domain) { begin_c_loop(c,t) { C_UDSI(c,t,Vorticity_I) = VORI(c,t); C_UDSI(c,t,Vorticity_J) = VORJ(c,t); C_UDSI(c,t,Vorticity_K) = VORK(c,t); } end_c_loop(c,t); } //Calculating Vorticity Derivatives for (n=Vorticity_I; n<Vorticity_I+3; n++) { uds_derivatives(domain,n); } thread_loop_c(t,domain) { begin_c_loop(c,t) { C_UDMI(c,t,Vorticity_I) = C_UDSI(c,t,Vorticity_I); C_UDMI(c,t,Vorticity_J) = C_UDSI(c,t,Vorticity_J); C_UDMI(c,t,Vorticity_K) = C_UDSI(c,t,Vorticity_K); ND_D(C_UDSI(c,t,G_Vorticity_Ix),C_UDSI(c,t,G_Vorticity_Iy),C_UDSI(c,t,G_Vorticity_Iz),=,\ C_UDSI_G(c,t,Vorticity_I)[0],C_UDSI_G(c,t,Vorticity_I)[1],C_UDSI_G(c,t,Vorticity_I)[2]); ND_D(C_UDSI(c,t,G_Vorticity_Jx),C_UDSI(c,t,G_Vorticity_Jy),C_UDSI(c,t,G_Vorticity_Jz),=,\ C_UDSI_G(c,t,Vorticity_J)[0],C_UDSI_G(c,t,Vorticity_J)[1],C_UDSI_G(c,t,Vorticity_J)[2]); ND_D(C_UDSI(c,t,G_Vorticity_Kx),C_UDSI(c,t,G_Vorticity_Ky),C_UDSI(c,t,G_Vorticity_Kz),=,\ C_UDSI_G(c,t,Vorticity_K)[0],C_UDSI_G(c,t,Vorticity_K)[1],C_UDSI_G(c,t,Vorticity_K)[2]); } end_c_loop(c,t); } } void uds_derivatives(Domain *domain, int n) { MD_Alloc_Storage_Vars(domain, SV_UDSI_RG(n), SV_UDSI_G(n), SV_NULL); Scalar_Reconstruction(domain, SV_UDS_I(n), 1, SV_UDSI_RG(n), NULL); Scalar_Derivatives(domain, SV_UDS_I(n), 1, SV_UDSI_G(n), SV_UDSI_RG(n), NULL); return; } 

August 11, 2015, 09:52 

#4 
Senior Member
Bruno Machado
Join Date: May 2014
Posts: 271
Rep Power: 11 
Are you plotting in the Fluent software or in the CFDPost?
If fluent, plot the results of the CELL and try to compare both. Should be same. If you are plotting the NODE values, it can be different since it is a average. 

August 11, 2015, 10:21 

#5 
Member
Farzin
Join Date: Jul 2014
Posts: 42
Rep Power: 10 
I'm plotting computed cell values in FLUENT, not node values.
This order of difference cannot be made by averaging; UDSUDM.jpg 

August 11, 2015, 10:39 

#6 
Senior Member
Bruno Machado
Join Date: May 2014
Posts: 271
Rep Power: 11 
Make a test, plot the values of the UDS and UDM in the GUI.
Add this in your code, after saying UDS=UDM and run one iteration. if((c%2000) == 0) Message("UDM Vor_I %f UDS Vor_I%f\n UDM Vor_J %f UDS Vor_J%f\n UDM Vor_K %f UDS Vor_K%f\n" C_UDMI(c,t,Vorticity_I), C_UDSI(c,t,Vorticity_I), C_UDMI(c,t,Vorticity_J), C_UDSI(c,t,Vorticity_J), C_UDMI(c,t,Vorticity_K), C_UDSI(c,t,Vorticity_K)); 

August 11, 2015, 10:55 

#7 
Senior Member
Bruno Machado
Join Date: May 2014
Posts: 271
Rep Power: 11 
What I would also suggest is to add this part in an EXECUTE_AT_THE_END macro instead of putting it in the ADJUST macro
thread_loop_c(t,domain) { begin_c_loop(c,t) { C_UDMI(c,t,Vorticity_I) = C_UDSI(c,t,Vorticity_I); C_UDMI(c,t,Vorticity_J) = C_UDSI(c,t,Vorticity_J); C_UDMI(c,t,Vorticity_K) = C_UDSI(c,t,Vorticity_K); ND_D(C_UDSI(c,t,G_Vorticity_Ix),C_UDSI(c,t,G_Vorti city_Iy),C_UDSI(c,t,G_Vorticity_Iz),=,\ C_UDSI_G(c,t,Vorticity_I)[0],C_UDSI_G(c,t,Vorticity_I)[1],C_UDSI_G(c,t,Vorticity_I)[2]); ND_D(C_UDSI(c,t,G_Vorticity_Jx),C_UDSI(c,t,G_Vorti city_Jy),C_UDSI(c,t,G_Vorticity_Jz),=,\ C_UDSI_G(c,t,Vorticity_J)[0],C_UDSI_G(c,t,Vorticity_J)[1],C_UDSI_G(c,t,Vorticity_J)[2]); ND_D(C_UDSI(c,t,G_Vorticity_Kx),C_UDSI(c,t,G_Vorti city_Ky),C_UDSI(c,t,G_Vorticity_Kz),=,\ C_UDSI_G(c,t,Vorticity_K)[0],C_UDSI_G(c,t,Vorticity_K)[1],C_UDSI_G(c,t,Vorticity_K)[2]); } end_c_loop(c,t); } 

August 11, 2015, 11:34 

#8 
Member
Farzin
Join Date: Jul 2014
Posts: 42
Rep Power: 10 
Thank you Bruno, for your time and effort.
Printed values of the UDMs and UDSs are all equal. Plus, plot of vorticity values calculated by FLUENT is very similar to UDM (which should be the same as UDS values). So the problem should be due to the way they're plotted. My concern was vorticity (UDS) and its gradient which are used in another UDF, are not correct. 

August 11, 2015, 11:50 

#9 
Senior Member
Bruno Machado
Join Date: May 2014
Posts: 271
Rep Power: 11 
Glad I could help you.
Just to make sure the value you are getting is the correct one and the difference is in the plotting, you could compute the other UDF using the value of the UDM and from the UDS. You should get the same value. 

August 11, 2015, 12:04 

#10 
Member
Farzin
Join Date: Jul 2014
Posts: 42
Rep Power: 10 
I just replaced the whole ADJUST function by AT_THE_END, and now the vorticity UDS and UDM plots are both exactly the same as the one calculated by FLUENT.
Still don't know why but this solved the problem. Thank you Bruno for the suggestion. 

August 11, 2015, 12:33 

#11 
Senior Member
Bruno Machado
Join Date: May 2014
Posts: 271
Rep Power: 11 
I think it is because fluent solve all the equations you imputed before solving the governing equations, so your current value of some variables actually were the value of previous iteration. But glad you solve the problem.


January 26, 2016, 12:07 

#12 
Senior Member
Przemek
Join Date: Jun 2011
Posts: 245
Rep Power: 14 
Hi Bruno,
I have such a code: Code:
/******************************************************************** *********************************************************************/ #include "udf.h" DEFINE_EXECUTE_AT_END(myFunc) { Domain *d; Thread *t; /* Integrate dissipation. */ real sum_diss=0.; cell_t c; d = Get_Domain(1); /* mixture domain if multiphase */ thread_loop_c(t,d) { if (FLUID_THREAD_P(t)) { begin_c_loop(c,t) C_UDMI(c,t,0) = C_T(c,t); end_c_loop(c,t) } } } Static temperature and User memory 0 differ (exactly only in one location = at boundary). Can you tell what I am doing wrong?
__________________
best regards pblasiak 

January 26, 2016, 13:17 

#13  
Member
Farzin
Join Date: Jul 2014
Posts: 42
Rep Power: 10 
Quote:
It's because you didn't assign UDM value for the boundary and you're trying to get it. Actually, the value you get is an interpolated value calculated by fluent. If you want to set values of a boundary to the UDM, you should also add face UDM. Try this one: Code:
#include "udf.h" DEFINE_EXECUTE_AT_END(myFunc) { Domain *d; /* Integrate dissipation. */ real sum_diss=0.; cell_t c; Thread *t; d = Get_Domain(1); /* mixture domain if multiphase */ thread_loop_c(t,d) { if (FLUID_THREAD_P(t)) { begin_c_loop(c,t) C_UDMI(c,t,0) = C_T(c,t); end_c_loop(c,t) } } thread_loop_f(t,d) { if (FLUID_THREAD_P(t)) { begin_f_loop(f,t) F_UDMI(f,t,0) = F_T(f,t); end_f_loop(f,t); } } } 

January 26, 2016, 13:32 

#14 
Senior Member
Przemek
Join Date: Jun 2011
Posts: 245
Rep Power: 14 
Hi Bruno,
Thank you for your fast replay. I used your code with adding (face_t f) and t = LookupThread(d,11) but it gives me "received a fatal signal (Segmentation fault)". Code:
#include "udf.h" DEFINE_EXECUTE_AT_END(myFunc) { Domain *d; /* Integrate dissipation. */ real sum_diss=0.; cell_t c; face_t f; Thread *t; d = Get_Domain(1); /* mixture domain if multiphase */ thread_loop_c(t,d) { if (FLUID_THREAD_P(t)) { begin_c_loop(c,t) C_UDMI(c,t,0) = C_T(c,t); end_c_loop(c,t) } } t = Lookup_Thread(d,11); thread_loop_f(t,d) { if (FLUID_THREAD_P(t)) { begin_f_loop(f,t) F_UDMI(f,tf,0) = F_T(f,t); end_f_loop(f,t); } } } 11 is the ID of the boundary where I want to set value. Do you know why error occur?
__________________
best regards pblasiak 

January 26, 2016, 14:12 

#15 
Senior Member
Przemek
Join Date: Jun 2011
Posts: 245
Rep Power: 14 
I changed the code like that but it does not solve the problem
Code:
#include "udf.h" DEFINE_EXECUTE_AT_END(myFunc) { Domain *d; /* Integrate dissipation. */ real sum_diss=0.; cell_t c; face_t f; Thread *t; const int ID = 11; d = Get_Domain(1); /* mixture domain if multiphase */ thread_loop_c(t,d) { if (FLUID_THREAD_P(t)) { begin_c_loop(c,t) C_UDMI(c,t,0) = C_T(c,t); end_c_loop(c,t) } } t = Lookup_Thread(d,ID); thread_loop_f(t,d) { //if (FLUID_THREAD_P(t)) if (NNULLP(THREAD_STORAGE(t,SV_UDM_I))) { begin_f_loop(f,t) F_UDMI(f,t,0) = F_T(f,t); end_f_loop(f,t); } else { Message("Skipping FACE thread no. %d..\n", THREAD_ID(t)); } } } Can anybody help?
__________________
best regards pblasiak 

January 26, 2016, 14:15 

#16 
Member
Farzin
Join Date: Jul 2014
Posts: 42
Rep Power: 10 
Hi, I'm Farzin!
I do not get why you added the thread pointer to the boundary; Because the loop in below line provides the thread pointer to all face threads, including that particular boundary. Remove the line you added and try. If you want to keep it, you should remove the face thread loop. 

January 26, 2016, 14:25 

#17 
Senior Member
Przemek
Join Date: Jun 2011
Posts: 245
Rep Power: 14 
Ok I deleted it but I got received a fatal signal (Segmentation fault).
Do you know what is wrong?
__________________
best regards pblasiak 

January 26, 2016, 14:34 

#18  
Member
Farzin
Join Date: Jul 2014
Posts: 42
Rep Power: 10 
Quote:
It shouldn't have any problem. 

January 26, 2016, 14:49 

#19 
Senior Member
Bruno Machado
Join Date: May 2014
Posts: 271
Rep Power: 11 
I believe the values are different because the macro C_T gives the total temperature, so it might be different from the static temperature.
Have a look at this thread http://www.cfdonline.com/Forums/flu...mperature.html 

January 26, 2016, 16:53 

#20 
Senior Member
Przemek
Join Date: Jun 2011
Posts: 245
Rep Power: 14 
Hi Bruno and Farzin,
@Farzin: I compiled this code: Code:
#include "udf.h" DEFINE_EXECUTE_AT_END(myFunc) { Domain *d; /* Integrate dissipation. */ real sum_diss=0.; cell_t c; face_t f; Thread *t; const int ID = 11; d = Get_Domain(1); /* mixture domain if multiphase */ thread_loop_c(t,d) { if (FLUID_THREAD_P(t)) { begin_c_loop(c,t) C_UDMI(c,t,0) = C_T(c,t); end_c_loop(c,t) } } //t = Lookup_Thread(d,ID); thread_loop_f(t,d) { //if (FLUID_THREAD_P(t)) //if (NNULLP(THREAD_STORAGE(t,SV_UDM_I))) begin_f_loop(f,t) F_UDMI(f,t,0) = F_T(f,t); end_f_loop(f,t); } } @Bruno: I do not agree with you because I compared data for Static temperature and C_T that was assigned to UDM. Each value was the same but boundary value. So I agree with Farzin that I have to set face value also with F_T, however I do not know how??? Here mvee wrote that he solved this problem but he unfortunately didn't provide the code: http://www.cfdonline.com/Forums/flu...gudmuds.html
__________________
best regards pblasiak 

Tags 
udf, udm, uds 
Thread Tools  Search this Thread 
Display Modes  


Similar Threads  
Thread  Thread Starter  Forum  Replies  Last Post 
Adjusting UDM to UDS  mvee  Fluent UDF and Scheme Programming  16  June 26, 2018 05:30 
Segmentation violation  louiza  FLUENT  16  June 27, 2017 16:41 
UDS Unsteady with UDM  shashank312  FLUENT  3  May 15, 2013 16:35 
UDS stored to UDM do not show the same values  swati_mohanty  FLUENT  0  December 3, 2010 04:46 
usage of UDS & UDM  Tong  FLUENT  2  February 15, 2006 11:32 