Tutorial: Turbulent Flow in a Compact Heat Exchanger

Introduction
The purpose of this tutorial is to evaluate the pressure drop and t h e heat transfer characteristics for liquid ammonia flowing through the specified interrupted fin heat exchanger at a given mass flow rate. At the end of the session, you can compare the computed values with the experimental data.

Prerequisites
This tutorial assumes that you have successfully completed the introductory training for the ANSYS FLUENT solver.

Problem Description
The compact heat exchanger for this problem consists of the staggered fins that enhance the heat transfer relative to a continuous- fin heat exchanger by promoting t h e turbulent mixing in the wake region behind each fin and taking advantage of the relatively high heat transfer coefficients as the boundary layers continuously re-form on the interrupted fins. A sketch of the geometry, with dimensions in mm, is shown in Figure 1. The geometry contains symmetry boundary conditions at the top and bottom planes.

Figure 1: Heat Exchanger Geometry

Assume that pressurized liquid ammonia at 240 K is to be heated in the heat exchanger and that the fin walls are at a constant temperature of 350 K. Considering the viscosity of the liquid ammonia, 0.000152 kg/m-s, and the characteristic length scale of 5.84mm for the half-width between the pins, t he Reynolds number is around 9000 and thus the flow is turbulent. Periodic boundaries will be used at the beginning and ending planes of the domain and a periodic mass flow rate of 1.385 kg/s will be used.

The performance of this heat exchanger will be expressed in terms of a drag coefficient, and a Stanton number, St. CD is based on the drag force density of the fluid , and the mean velocity, U :

CD ,
the

D,

the fin surface area

A,

1 2 The drag coefficient, , represents both the viscous and the pressure drag components. The Stanton number depends upon the average heat transfer coefficient and the thermal capacitance of the flow:

Preparation
1. Develop a strategy for modeling this turbulent flow. The strategy will include decisions on which turbulence model to use, what type of near-wall treatment to use, what type of mesh to use (quad, tri, or hybrid), and how to distribute the nodes. Be sure to estimate the friction velocity, u , so that you will be able to specify the proper node spacing in the direction normal to the fin walls.

Setup and Solution

Step 1: Mesh
1. Read the mesh file, htx.msh., into a 2D double precision version of ANSYS FLUENT. Note: There are periodic boundaries in the existing mesh; however, in some cases the source mesh might not have the periodic boundaries assigned. Should you ever have a situation like that then you can make the corresponding boundaries as periodic by using some TUI commands:

The references to zone names and ID numbers above may differ from yours, since your boundary zones might be named and numbered differently. 2. Check the mesh. Mesh Check 3. Scale the mesh from microns to meters. Mesh Scale One needs to use the scaling factors of 1.e-6 in both x & y directions. Please press the Scale button, check the Domain Extents field in the same panel and once satisfied with the obtained values you can close the Scale Mesh panel 4. Display the Mesh. Display Mesh The computational domain and mesh will become extremely small and invisible after downscaling by a factor of 1.e+6. Please press Fit to Window button to see the mesh.

Figure 2: Graphics Display of the Mesh

Step 2: Models
1. Retain the default solver settings. 2. Enable the Energy Equation 3. Choose the k-epsilon (2- equation) turbulence model under

Define Models Viscous a. Select RNG k-epsilon Model from the list of three available k-epsilon models. b. Choose the Enhanced Wall Treatment as a Near-Wall Treatment. c. Retain the default values for the other parameters.

Step 3: Materials
Materials 1. Copy ammonia-liquid (nh3<l>) from the database.

Step 4: Operating Conditions

Define Operating Conditions 1. Retain the default operating conditions.

Step 5: Boundary and Periodic Conditions

1. Specify the Cell Zone Conditions. Cell Zone Conditions Select ammonia-liquid in the Material Name field for the fluid cell zone and retain the default values for the other parameters. 2. Specify the Boundary Conditions. Boundary Conditions a. Set the Temperature for wall-bottom and wall-top zone to 350 K. b. Press the Edit button for the periodic zone and ensure that a translational periodic type has been selected. 3. Specify the Periodic Conditions Boundary Conditions Press Periodic Conditions tab. a. Under the Type, select Specify Mass Flow b. Set the Mass Flow Rate and Upstream Bulk Temperature to 1.385 kg/s and 240K, respectively.

Step 6: Solution
1. Initialize the flow field. Solution Initialization Initialize... You can assign the initial values as follows: Initialized Quantity Streamwise Velocity, U Turbulent Kinetic Energy, k Turbulent Dissipation Rate, Temperature, T Value Evaluation Method From mass flow rate: 1.385/610 5.84 10 0.39 m/s Assuming the inlet turbulence intensity of 10%: 0.1 0.39 1.52 10 m2/s2 Assuming turbulent viscosity ratio of 100 for Re=9000:

0.39 m/s

1.610-3 m2/s2 8.310-3 m2/s3 240 K

100

8.3 10 m2/s3

Constant temperature of oncoming flow

Note: The above evaluations are not strictly necessary, but the more realistic initialization can speed up the convergence rates or even prevent the divergence at earlier stages of the calculation.

2. Enable the plotting of residuals during the calculation. Solution Monitors Residuals Press the Edit button or double click in the Residuals Print, Plot field and the Residual Monitors panel should open. Please make sure that plotting and printing options for the residuals are switched on. In addition, it is recommended to reduce the convergence criteria values for the solved equations, especially, if the double precision solver is used. In some cases the solution might still be changing even if the residuals dropped by three orders of magnitude. Either the individual values for the equations can be lowered or the Convergence Criterion option should be changed from absolute to none. 3. Enable the monitoring of the surface Stanton number on the longest wall (wall-top). Solution Monitors Surface Monitors... Press the Create button and the Surface Monitor panel will open. Specify a report of Surface Stanton Number for wall-top as shown in the Figure below.

Enable monitoring of the drag coefficient on the longest wall (wall-top). Solution Monitors Drag... Press the Edit button and the Drag Monitor panel will open.

Specify a report of the Drag Coefficient for wall-top in the x direction.

Set the appropriate reference values for use in reporting the drag coefficient and surface Stanton number. 4. Problem Setup Reference Values Fill in the values as shown in the Figure below.

Note: The reference Area should be equal to the area of wall-top, which can be calculated using the Surface Integrals panel. Cut and Paste option can be used but there is no need to carry over all 8 digits after the decimal point and 6 digits after the first meaningful digit.
5. Set the solution parameters. Solution Solution Methods Set the Discretisation for the Momentum, Turbulence Kinetic Energy, Turbulence Dissipation Rate and Energy equations to the Second Order Upwind scheme.

Note: In this example the use of the second-order discretisation schemes can be utilized from the beginning of the calculation. However, there are a lot of cases where a user might need to start from the first order. Once the first-order solution is obtained then one can use the higher discretisation in order to get a more accurate solution. Note: It is often the case that the use of the higher order discretisation schemes for the mean equations (i.e. momentum, energy & species) gives a clear advantage over the first order schemes. However, the use of these more accurate schemes for the turbulent quantities might

or might not be so advantageous, but is likely to hinder the convergence rates. Therefore, in such cases it is advisable to ramp up the discretisation for the mean equations first and only then do the same for the turbulence equations.
6. Set the solution controls. Solution Solution Controls Deselect the Energy in the list of Equations after pressing the Equations button in the Solutions Control panel. You will begin the calculation by solving only the flow and turbulence equations. Once this solution has converged, you will turn off the flow and the turbulence equations and then solve only the energy equation. Finally, you will solve all equations together.

7. Calculate the flow and turbulence solution until you are happy with the convergence. Run Calculation Calculate

Figure 3:Scaled Residuals

Check the near-wall mesh resolution by visualizing the near-wall y+ value. The best way to visualize that value is by plotting the contour plots of Turbulence and Wall Yplus. One needs to switch off the display of Node Values so that no interpolation of the calculated cell values is done. This way, the actual calculated near-wall cell y+ values can be displayed. In this tutorial, the maximum near-wall y+ value is around one and no mesh refinement is required; however, in a general situation one can adapt the mesh to achieve a required level of refinement by using the Adapt Yplus/Ystar a. If you use the wall functions as a near-wall treatment, then the near-wall y+ value should be ideally between 50 and 500. b. If you use the two-layer zonal model, then the y+ value at the adjacent cell should be in the order of 1. 8. Turn off the flow and turbulence equations, and enable the energy equation. 9. Calculate the energy solution until convergence and then increase the under-relaxation factor for the energy equation from 0.95 to 0.98 and iterate again. As a final step, please increase the same under-relaxation factor to 1 and achieve the final solution. The plot of the energy residuals is shown in the Figure 4 below. The Stanton number monitor is shown in Figure 6.

Note: Please bear in mind that unlike the rest of the under-relaxation factors, the solution for the energy equation requires for its under-relaxation factor to be as close to unity as possible. Values lower than unity might result in a different solution for the energy equation.

Figure 4: Convergence History for the Energy Equiation

Figure 5: Drag Convergence History

Figure 4: Convergence History of Surface Stanton Number on wall-top

10. Save the case and data files. 11. Enable solution of all equations, solve until convergence, and save the final case and data files. Note: This step is not really needed in flows like this where the properties can be assumed constant and the influence of buoyancy is negligible. Otherwise, it can be very helpful to approach the solution in such way.

Step 7: Postprocessing
1. Enable reflection of the display across the symmetry planes. Display Views Select the plane, symmetry-bottom and symmetry-top under Mirror Planes, click Apply, and close the panel. The Display will be updated, as shown in Figure 7

2. Display filled contours of static pressure (Figure7)

Figure 5: Contours of Static Pressure, Pa

The pressure contours show a region of high pressure near the stagnation point at the front of the fin as well as the regions of relatively low static pressure where the flow accelerates around the curved front portion of the fin. 3. Display filled contours of velocity magnitude (Figure 8). The velocity contours show the stagnation and flow acceleration regions at the front of the fin, the growth of a thin boundary layer along the length of the fin, and a wake region immediately downstream of the fin. The wake region promotes turbulent mixing and enhances overall heat transfer.

Note: For this flow rate, the wake region settles out before the next fin is encountered and ensures that a new boundary layer with a relatively high heat transfer coefficient will grow on subsequent fins.
4. Display velocity vectors (Figure 9). 5. Display filled contours of temperature in the range between 240K and 290K (Figure 10).

Figure 6: Contours of Velocity Magnitude, m/s.

Figure 7: Velocity Vectors at the Front Showing the Stagnation Point Region

Figure 8: Contours of Temperature, K

The plots for the turbulent quantities are shown in Figures 11-13. The turbulent kinetic energy plot in Figure 11 and the turbulent viscosity ratio in Figure 13 show the regions where the most of the mixing takes place ( i.e in the wakes) as well as the regions where the turbulence is produced by shear i.e. the frontal impingement regions and along the surfaces. Figure 12 highlights the regions of higher dissipation rate. These normally coincide with the regions where the turbulence is produced. All of these figures can shed some light how accurate our assessment was for the initial conditions earlier in this tutorial.

Figure 11: Contours of the Turbulent Kinetic Energy, m2/s2.

Figure 12: Contours of Turbulent Dissipation Rate, m2/s3.

Figure 13: Contours of Turbulent Viscosity Ratio

Summary
For liquid ammonia at a moderately low bulk Reynolds number over the heat exchanger surface t he experimental data in terms of pressure drop performance is cD = 0.019 and in terms of heat transfer performance is St = 3.87 103 . Now that you have successfully completed the tutorial, you can review this work by checking the following: 1. 2. 3. 4. 5. 6. Which type of mesh did you use? How many cells were required? How accurate was the calculated drag coefficient? How accurate was the calculated Stanton number? What would you have done differently? Please have a look at the minimum calculated temperature. Is it realistic? If not, please try to identify the region where this is the case and try to assess whether it is the turbulence closure itself or the near-wall treatment or the higher order numerical discretisation that might be responsible for the error.