Tutorial: Calculation of Flow Uniformity

Introduction
The purpose of this tutorial is to provide guidelines and recommendations for setting up and solving a ow uniformity problem with the help of a user-dened function (UDF). This tutorial demonstrates how to do the following: Modify the UDF for a specic case. Use the UDF to calculate the ow uniformity indexes for steady-state ow.

Prerequisites
This tutorial is written with the assumption that you have completed Tutorial 1 from ANSYS FLUENT 12.0 Tutorial Guide, and that you are familiar with the ANSYS FLUENT navigation pane and menu structure. Some steps in the setup and solution procedure will not be shown explicitly. For more details about UDFs, see ANSYS FLUENT 12.0 UDF Manual.

Problem Description
The problem considers a 3D ow passage. A schematic of the problem is shown in Figure 1. There are two ow inlets and an outlet. The postprocessing planes (post-01, post-02, post03, and post-04) have already been created with the mesh.

c ANSYS, Inc. September 25, 2009

Calculation of Flow Uniformity

Figure 1: Schematic of the Problem

Setup and Solution Preparation

1. Copy the les (flow-uniformity.msh.gz, cat-general.c, and input.txt) to the working folder. 2. Use FLUENT Launcher to start the 3D version of ANSYS FLUENT. For more information about FLUENT Launcher see Section 1.1.2, Starting ANSYS FLUENT Using FLUENT Launcher in ANSYS FLUENT 12.0 Users Guide. 3. Enable Double-Precision in the Options list. 4. Click the UDF Compiler tab and ensure that the Setup Compilation Environment for UDF is enabled. The path to the .bat le which is required to compile the UDF will be displayed as soon as you enable Setup Compilation Environment for UDF. If the UDF Compiler tab does not appear in the FLUENT Launcher dialog box by default, click the Show More >> button to view the additional settings. The Display Options are enabled by default. Therefore, after you read in the mesh, it will be displayed in the embedded graphics window.

c ANSYS, Inc. September 25, 2009

Calculation of Flow Uniformity

Step 1: Mesh 1. Read the mesh le (flow-uniformity.msh.gz). File Read Mesh... As the mesh le is read, ANSYS FLUENT will report the progress in the console. Step 2: General Settings 1. Retain the default solver settings. General 2. Check the mesh (see Figure 2). General Check

Figure 2: Mesh Display ANSYS FLUENT will perform various checks on the mesh and will report the progress in the console. Make sure the minimum volume reported is a positive number. 3. Scale the mesh to centimeters. General Scale... (a) Select cm from Mesh Was Created In drop-down list. (b) Click Scale and close Scale Mesh dialog box.

c ANSYS, Inc. September 25, 2009

Calculation of Flow Uniformity

Step 3: Models Select the k-epsilon turbulence model. Models Viscous Edit...

You will use the default uid properties of air for this problem. Hence, you need not make any changes to the material properties. Step 4: Boundary Conditions 1. Set the boundary conditions for vel-in-large. Boundary Conditions vel-in-large Edit...

(a) Select Magnitude, Normal to Boundary from the Velocity Specication Method drop-down list. (b) Enter 0.5 m/s for Velocity Magnitude. (c) Select Intensity and Hydraulic Diameter from the Specication Method drop-down list. (d) Enter 5% for Turbulent Intensity. (e) Enter 0.06 m for Hydraulic Diameter. (f) Click OK to close the Velocity Inlet dialog box. 2. Set the boundary conditions for velocity-in-small. Boundary Conditions vel-in-small Edit...

(a) Select Magnitude, Normal to Boundary from the Velocity Specication Method drop-down list. (b) Enter 2 m/s for Velocity Magnitude. (c) Select Intensity and Hydraulic Diameter from the Specication Method drop-down list. (d) Enter 5% for Turbulent Intensity. (e) Enter 0.02 m for Hydraulic Diameter. (f) Click OK to close the Velocity Inlet dialog box. 3. Set the boundary conditions for press-out. Boundary Conditions press-out Edit...

(a) Select Normal to Boundary from the Backow Direction Specication Method dropdown list. (b) Select Intensity and Hydraulic Diameter from the Specication Method drop-down list. (c) Enter 5% for Backow Turbulent Intensity. (d) Enter 0.06 m for Backow Hydraulic Diameter. (e) Click OK to close the Pressure Outlet dialog box. 4

c ANSYS, Inc. September 25, 2009

Calculation of Flow Uniformity

Step 5: Operating Conditions Boundary Conditions Operating Conditions... 1. Retain the default operating conditions. Step 6: Solution 1. Retain the default solution control parameters. 2. Initialize the ow eld from vel-in-large. Solution Initialization 3. Start the calculation for 100 iterations. Run Calculation The solution will converge in approximately 60 iterations. 4. Save the case and data les (flow-1.cas/dat.gz). Step 7: Postprocessing 1. Display lled contours of velocity magnitude. Graphics and Animations Contours Set Up...

(a) Enable Filled in the Options list. (b) Select Velocity... and Velocity Magnitude from Contours of drop-down list. (c) Select post-01, post-02, post-03, and post-04 from the Surfaces list. (d) Enable Draw Mesh in the Options group box. The Mesh Display dialog box will open. i. Disable Edges and enable Faces in the Options group box. ii. Deslect all the surfaces from the Surfaces list and click the Outline. iii. Click Display and close Mesh Display dialog box. (e) Click Display and close Contours dialog box.

c ANSYS, Inc. September 25, 2009

Calculation of Flow Uniformity

2. Manipulate the display using the Scene Description dialog box. Graphics and Animations Scene... (a) Select the surfaces as shown in the Scene Description dialog box.

(b) Click Display... in the Geometry Attributes group box to open the Display Properties dialog box.

i. Enable Lighting in the Visibility group box. ii. Disable Edges, Lines, and Nodes in the Visibility group box. iii. Enable Outer Faces in the Visibility group box. iv. Set the sliders for Red, Green, and Blue to 255 in the Colors group box. v. Set the slider for Transparency to 80. vi. Click Apply and close the Display Properties dialog box (see Figure 3). (c) Close Scene Description dialog box.

c ANSYS, Inc. September 25, 2009

Calculation of Flow Uniformity

Figure 3: Contours of Velocity Magnitude

3. Display velocity vectors (see Figure 4).

Figure 4: Velocity Vectors Colored by Velocity Magnitude

c ANSYS, Inc. September 25, 2009

Calculation of Flow Uniformity

4. Calculate the ow uniformity. (a) Set up the input le (input.txt). The input le (input.txt) has to be processed before executing the UDF. For the details about the input le, see Appendix 1: Details of the Input File. (b) Set the Number of User-Dened Memory Locations to 4. Dene User-Dened Memory... (c) Save the case le (flow.cas.gz). (d) Compile the UDF. Dene User-Dened Functions Compiled...

i. Click Add... and select the source le, cat-general.c. For details about the UDF le (cat-general.c), refer to Appendix 2: Contents of UDF. ii. Click Build to build the library. A Warning dialog box opens, asking you to ensure that the UDF source les are in the same folder that contains the case and data les. Click OK. iii. Click Load to load the newly created UDF library. (e) Execute the UDF. Dene User-Dened Execute on Demand... i. Select Uniform steady::libudf from the Execute On Demand drop-down list. ii. Click Execute. (f) The output on the screen and in the le are shown in Appendix 3: Output of UDF. The UDF reads the input le (input.txt) and generates the output le (output.txt) automatically in each simulation. The parameters of the output are explained in Appendix 4: General Structure of UDF. If you want to preserve the old results, rename the output.txt le. Otherwise the data for the next run will be appended to the old le.

c ANSYS, Inc. September 25, 2009

Calculation of Flow Uniformity

Appendix 1: Details of the Input File

The rst line is the number of postprocessing faces. You can dene more than one face zones for ow uniformity calculation. Each line after the rst line has the parameters of a face-zone. They are dened in the following order: POST BOUNDARY ZONE ID POST CELL ZONE ID 1 POST CELL ZONE ID 2 L XX L YX L ZX L XY L YY L ZY Further detailed description of the parameters is given as follows: 1. POST BOUNDARY ZONE ID is the ID of the face zone. 2. POST CELL ZONE ID 1 and POST CELL ZONE ID 2 are the IDs of two adjacent cell zones of the postprocessing face zone. They are the same if the face zone is inside a cell zone. 3. L XX, L YX, L ZX are the direction cosines of the x-axis of the local coordinate. The UDF uses a local coordinate for each face zone to compute the ow uniformity. The local x-axis will be the major axis, and local y-axis will be the minor axis. You should dene the direction of longer length as x-axis, and the other direction in the same plane as y-axis. 4. L XY, L YY, L ZY are the direction cosines of the y-axis of the local coordinate. The UDF uses a local coordinate for each face zone to compute the ow uniformity. The local x-axis will be the major axis, and local y-axis direction will be the minor axis. You should dene the direction of longer length as x-axis, and the other direction in the same plane as y-axis. The denition of the direction cosines may aect the eccentricity because it reects the local ow distribution on the plane. The gamma does not depend on the direction cosines.

c ANSYS, Inc. September 25, 2009

Calculation of Flow Uniformity

Appendix 2: Contents of UDF

The contents of the UDF le are as follows:
/**************************************************************/ /* */ /* User-Defined Functions for calculation of flow uniformity */ /* */ /**************************************************************/ #include "udf.h" int ncount; int npost; /*You need to modify these three lines*/ int POST_BOUNDARY_ZONE_ID[50]; int POST_CELL_ZONE_ID_1[50]; int POST_CELL_ZONE_ID_2[50]; float L_XX[50]; float L_YX[50]; float L_ZX[50]; float L_XY[50]; float L_YY[50]; float L_ZY[50]; float L_XZ[50]; float L_YZ[50]; float L_ZZ[50];

static void read_input() { int i=0; FILE *fpin; if((fpin=fopen("input.txt","r"))==NULL) { printf("Input file does not exist!"); } fscanf(fpin, "%d",&npost); for(i=0;i<npost;i++) { fscanf(fpin,"%d %d %d %f %f %f %f %f %f", &POST_BOUNDARY_ZONE_ID[i], &POST_CELL_ZONE_ID_1[i], &POST_CELL_ZONE_ID_2[i], &L_XX[i], &L_YX[i], &L_ZX[i], &L_XY[i], &L_YY[i], &L_ZY[i]); L_XZ[i]=L_YX[i]*L_ZY[i]-L_ZX[i]*L_YY[i]; L_YZ[i]=L_ZX[i]*L_XY[i]-L_XX[i]*L_ZY[i]; L_ZZ[i]=L_XX[i]*L_YY[i]-L_XY[i]*L_YX[i]; } fclose(fpin); } static void init_udm() { Thread *tc; cell_t c; Domain *domain; domain = Get_Domain(1); thread_loop_c(tc,domain)

10

c ANSYS, Inc. September 25, 2009

Calculation of Flow Uniformity

{ begin_c_loop(c,tc) { C_UDMI(c,tc,0)=0; C_UDMI(c,tc,1)=0; C_UDMI(c,tc,2)=0; C_UDMI(c,tc,3)=0; } end_c_loop(c,tc) } ncount=0; }

static void cat_post(int flag) { float xf[ND_ND], x_vmax_global[ND_ND], x_vmax, y_vmax; float x_face_local, y_face_local; float xn[ND_ND],x_node_local,y_node_local; float x_min, x_max, x_mid, y_min, y_max, y_mid; float L_major, L_minor; float eccent_x, eccent_y, eccent; float vmag, vmax, vavg, vratio, vel_space; float A_face_vec[ND_ND], A_face_mag, A_tot; float a_35, v_35, unif_35, a_65, v_65, unif_65, gamma; int k; face_t f, f_vmax; cell_t c0, c1; Thread *tf, *tc0, *tc1; FILE *fp; Domain *domain; domain = Get_Domain(1);

#define BIGNUM 1e20 if((fp=fopen("output.txt", "r"))==NULL) { fp=fopen("output.txt","w+"); fprintf(fp, "post-face eccent gamma fclose(fp); } fp=fopen("output.txt", "a"); /* find local y,z min/max extents based on node values */

unif_35

unif_65

vratio

vmax \n");

x_min=BIGNUM; y_min=BIGNUM; x_max=-BIGNUM; y_max=-BIGNUM; tf=Lookup_Thread(domain, POST_BOUNDARY_ZONE_ID[flag]);

c ANSYS, Inc. September 25, 2009

11

Calculation of Flow Uniformity

begin_f_loop(f,tf) { f_node_loop(f,tf,k) { xn[0]=NODE_X(F_NODE(f,tf,k)); xn[1]=NODE_Y(F_NODE(f,tf,k)); xn[2]=NODE_Z(F_NODE(f,tf,k)); x_node_local = NVD_DOT(xn,L_XX[flag],L_YX[flag],L_ZX[flag]); y_node_local = NVD_DOT(xn,L_XY[flag],L_YY[flag],L_ZY[flag]); x_min=MIN(x_node_local,x_min); x_max=MAX(x_node_local,x_max); y_min=MIN(y_node_local,y_min); y_max=MAX(y_node_local,y_max); } } end_f_loop(f,tf) x_mid=0.5*(x_min + x_max); y_mid=0.5*(y_min + y_max); L_major = x_max - x_min; L_minor = y_max - y_min;

Message("================================================================\n"); Message("post-%d\n",flag); Message("================================================================\n"); Message("Geometry information:\n"); Message("(x_min, x_mid, x_max) = (%f, %f, %f) [m]\n", x_min, x_mid, x_max); Message("(y_min, y_mid, y_max) = (%f, %f, %f) [m]\n", y_min, y_mid, y_max); Message("(L_major, L_minor) = (%f, %f) [m]\n\n", L_major, L_minor); /* velocity index (eccentricity) computation */ vmax=0.0; vavg=0.0; A_tot=0.0; tf=Lookup_Thread(domain, POST_BOUNDARY_ZONE_ID[flag]); tc0=Lookup_Thread(domain, POST_CELL_ZONE_ID_1[flag]); tc1=Lookup_Thread(domain, POST_CELL_ZONE_ID_2[flag]); begin_f_loop(f,tf) { F_AREA(A_face_vec,f,tf); A_face_mag=NV_MAG(A_face_vec); A_tot+=A_face_mag; c0=F_C0(f,tf); c1=F_C1(f,tf); vmag=0.5*(C_UDMI(c0,tc0,3)+C_UDMI(c1,tc1,3)); vavg+=vmag*A_face_mag; if(vmag > vmax) { vmax=vmag; f_vmax=f; } } end_f_loop(f,tf) vavg/=A_tot; vratio=vmax/vavg; F_CENTROID(x_vmax_global,f_vmax,tf); x_vmax= NVD_DOT(x_vmax_global,L_XX[flag],L_YX[flag],L_ZX[flag]); y_vmax= NVD_DOT(x_vmax_global,L_XY[flag],L_YY[flag],L_ZY[flag]); eccent_x=2.0*(x_vmax-x_mid)/L_major; eccent_y=2.0*(y_vmax-y_mid)/L_minor; eccent=sqrt(eccent_x*eccent_x+eccent_y*eccent_y);

12

c ANSYS, Inc. September 25, 2009

Calculation of Flow Uniformity

Message("Eccentricity information:\n"); Message("(x_vmax, y_vmax) = (%f, %f) [m]\n", x_vmax, y_vmax); Message("eccentricity = (%f, %f) -> %f \n\n",eccent_x, eccent_y, eccent); Message("vmax = %f m/s occurs at face %d\n", vmax, f_vmax); Message("vratio = %f\n", vratio); /* uniformity, gamma calculations */ v_35=0.35*vmax; v_65=0.65*vmax; a_35=0.0; a_65=0.0; gamma=0.0; begin_f_loop(f,tf) { F_AREA(A_face_vec,f,tf); A_face_mag=NV_MAG(A_face_vec); c0=F_C0(f,tf); c1=F_C1(f,tf); vmag=0.5*(C_UDMI(c0,tc0,3)+C_UDMI(c1,tc1,3)); if(vmag >= v_35) a_35+=A_face_mag; if(vmag >= v_65) a_65+=A_face_mag; gamma+=ABS(vmag-vavg)*A_face_mag; } end_f_loop(f,tf) gamma=1.0-gamma/(2.*vavg*A_tot); unif_35=a_35/A_tot*100; unif_65=a_65/A_tot*100; Message("uniformity index = %f/%f\n",unif_65,unif_35); Message("gamma = %f\n",gamma);

fprintf(fp, "post%d %9.3f %9.3f unif_35, unif_65, vratio, vmax); fclose(fp); } static void set_udm(int flag)

%9.3f

%9.3f

%9.3f

%9.3f \n", flag, eccent, gamma,

{ /* NOTE: THIS ROUTINE REQUIRES ALLOCATION OF */ /* 4 USER DEFINED MEMORY LOCATIONS IN FLUENT */ #define #define #define #define X_PRIME 0 Y_PRIME 1 Z_PRIME 2 V_MAG 3

float xc[ND_ND]; Thread *tc1,*tc2; cell_t c; Domain *domain; domain = Get_Domain(1);

tc1=Lookup_Thread(domain, POST_CELL_ZONE_ID_1[flag]); tc2=Lookup_Thread(domain, POST_CELL_ZONE_ID_2[flag]);

c ANSYS, Inc. September 25, 2009

13

Calculation of Flow Uniformity

begin_c_loop(c,tc1) { C_CENTROID(xc,c,tc1); C_UDMI(c,tc1,X_PRIME)=NVD_DOT(xc,L_XX[flag],L_YX[flag],L_ZX[flag]); C_UDMI(c,tc1,Y_PRIME)=NVD_DOT(xc,L_XY[flag],L_YY[flag],L_ZY[flag]); C_UDMI(c,tc1,Z_PRIME)=NVD_DOT(xc,L_XZ[flag],L_YZ[flag],L_ZZ[flag]); C_UDMI(c,tc1,V_MAG)=sqrt(C_VMAG2(c,tc1)); } end_c_loop(c,tc1) if(POST_CELL_ZONE_ID_1[flag] != POST_CELL_ZONE_ID_2[flag]) { begin_c_loop(c,tc2) { C_CENTROID(xc,c,tc2); C_UDMI(c,tc2,X_PRIME)=NVD_DOT(xc,L_XX[flag],L_YX[flag],L_ZX[flag]); C_UDMI(c,tc2,Y_PRIME)=NVD_DOT(xc,L_XY[flag],L_YY[flag],L_ZY[flag]); C_UDMI(c,tc2,Z_PRIME)=NVD_DOT(xc,L_XZ[flag],L_YZ[flag],L_ZZ[flag]); C_UDMI(c,tc2,V_MAG)=sqrt(C_VMAG2(c,tc2)); } end_c_loop(c,tc2) } }

DEFINE_ON_DEMAND(Uniform_steady) { int i=0; read_input(); init_udm(); for(i=0; i<npost;i++) { set_udm(i); cat_post(i); } }

Notes: The postprocessing face must be a real face zone and should be created in preprocessing. Only Uniform-steady can be used as a Execute On Demand function. You can use the EXECUTE ON DEMAND macro for calculations which do not participate in the solution iterations, such as postprocessing.

14

c ANSYS, Inc. September 25, 2009

Calculation of Flow Uniformity

Appendix 3: Output of UDF

Output on the screen: ================================================================ post-0 ================================================================ Geometry information: (x _min, x _mid, x _max) = (-0.030000, 0.000000, 0.030000) [m] (y _min, y _mid, y _max) = (-0.030000, 0.000000, 0.030000) [m] (L _major, L _minor) = (0.060000, 0.060000) [m] Eccentricity information: (x _vmax, y _vmax) = (0.002929, 0.001147) [m] eccentricity = (0.097642, 0.038245) -> 0.104865 vmax = 1.576765 m/s occurs at face 306 vratio = 2.062665 uniformity index = 23.614954/66.099402 gamma = 0.807598 ================================================================ post-1 ================================================================ Geometry information: (x _min, x _mid, x _max) = (-0.030000, 0.000000, 0.030000) [m] (y _min, y _mid, y _max) = (-0.030000, 0.000000, 0.030000) [m] (L _major, L _minor) = (0.060000, 0.060000) [m] Eccentricity information: (x _vmax, y _vmax) = (0.004870, -0.003402) [m] eccentricity = (0.162350, -0.113398) -> 0.198032 vmax = 1.495621 m/s occurs at face 278 vratio = 1.961932 uniformity index = 27.457386/72.582035 gamma = 0.813344 ================================================================ post-2 ================================================================ Geometry information: (x _min, x _mid, x _max) = (-0.044035, -0.029980, -0.015925) [m] (y _min, y _mid, y _max) = (-0.010000, -0.000000, 0.009999) [m] (L _major, L _minor) = (0.028110, 0.019999) [m]

c ANSYS, Inc. September 25, 2009

15

Calculation of Flow Uniformity

Eccentricity information: (x _vmax, y _vmax) = (-0.018510, -0.001128) [m] eccentricity = (0.816052, -0.112719) -> 0.823800 vmax = 2.378441 m/s occurs at face 53 vratio = 1.149897 uniformity index = 99.176368/100.000000 gamma = 0.956343 ================================================================ post-3 ================================================================ Geometry information: (x _min, x _mid, x _max) = (-0.030000, 0.000000, 0.030000) [m] (y _min, y _mid, y _max) = (-0.030000, 0.000000, 0.030000) [m] (L _major, L _minor) = (0.060000, 0.060000) [m] Eccentricity information: (x _vmax, y _vmax) = (0.000981, 0.005763) [m] eccentricity = (0.032705, 0.192102) -> 0.194866 vmax = 1.677601 m/s occurs at face 334 vratio = 2.189714 uniformity index = 21.332038/55.492872 gamma = 0.802111

Output in a le:

post-face post0 post1 post2 post3

eccent 0.105 0.198 0.819 0.195

gamma 0.807 0.812 0.956 0.802

unif _35 65.308 72.158 100.000 55.005

unif _65 23.475 27.038 98.426 21.096

vratio 2.071 1.969 1.146 2.197

vmax 1.582 1.501 2.371 1.682

16

c ANSYS, Inc. September 25, 2009

Calculation of Flow Uniformity

Appendix 4: General Structure of UDF

The UDF can be used for uniform steady-state analysis. It is used as a compiled UDF. There are three subroutines, init udm, set udm, and cat post in the UDF. You need to set four userdened memory locations, in which udm-0, udm-1, and udm-2 store the coordinates, and udm-3 stores velocity. The functions of the subroutines are as follows: init udm: It is used to initialize the user-dened memory. set udm: It computes the local coordinate in the selected cell zones based on the directional cosines provided by users. It stores the local coordinates in udm-0, udm-1, and udm-2, and velocity magnitude of the selected cell zones in udm-3. Cat post: Compute ow parameter for the postprocessing faces. Calculation of Flow Parameters:

Figure 5: Postprocessing Face

c ANSYS, Inc. September 25, 2009

17

Calculation of Flow Uniformity

1. Calculation of geometric parameters of the postprocessing face. (a) Length of major and minor axes: L major = x max x min L minor = y max y min where, L major = length of major axis L minor = length of minor axis x max = maximum x coordinate y max = maximum y coordinate x min = minimum x coordinate y min = minimum y coordinate (b) Center of the zone: x mid = x max x min 2 y max y min y mid = 2 (3) (4) (1) (2)

2. Calculation of velocity index (eccentricity). Compute the location (x and y coordinates) of the maximum velocity and the distance between this location and the center of the face. Find the maximum velocity and velocity ratio. Eccentricity () is dened as: x= y= 2 (x vmax x mid) L major 2 (y vmax y mid) L minor x2 + y2 (5) (6) (7)

Velocity ratio ( ratio) can be dened as the ratio of the maximum velocity to the average velocity. max ratio = (8) avg 3. Calculation of ow uniformity. Compute the uniformity index (area based), gamma, and vel space. The uniformity index is based on these reference velocities: 65 (65% of the maximum velocity) and 35 (35% of the maximum velocity).

18

c ANSYS, Inc. September 25, 2009

Calculation of Flow Uniformity

The uniformity is dened as: 65 = 35 = where, A 65 = the area of which velocity is higher than 65 A 35 = the area of which velocity is higher than 35 Gamma is dened as, =1 4. Output parameters. (a) On the screen: The parameters of the output on the screen can be explained as follows: (x max, y max) = Location of the maximum velocity eccentricity = ( x, y ) vmax = maximum velocity vratio = ratio uniformity index = 65/ 35 gamma = (b) In the output le: The parameters in the output le can be explained as follows: time = ow time eccent = gamma = unif 35 = 35 unif 65 = 65 vmax = maximum velocity (| i avg|A i) 2 avgA total (11) A 65 100A total A 35 100A total (9) (10)

Summary
In this tutorial, a ow uniformity problem was set up and solved for a steady state ow with the help of UDF.

c ANSYS, Inc. September 25, 2009

19