Mihir Meetarbhan

0922031

CFD ASSIGNMENT

NUMERICAL SIMULATION OF FLUID FLOW AND HEAT TRANSFER

MIHIR MEETARBHAN

0922031

08/12/11

Mihir Meetarbhan

0922031

PROBLEM UNDER INVESTIGATION

The problem being investigated is the steady flow of air, which is viscous and incompressible (i.e., constant density), through a duct which has a heated plate midway through its lower part. The duct also has an elbow which is at 35 just after the hot flat plate. The study is being carried out in order to investigate the formation of boundary layers at the wall of the duct and the heat transfer taking place between the hot plate and the working fluid. When air enters the duct, it comes into contact with the walls of the duct. As the walls are not smooth, there is friction between the air flow and the walls. Hence this causes the particles of air close to the wall to slow down. As this occurs, the faster layer of air above collides with the slower particles and form a boundary layer. This is a slow moving layer of air where the velocity is zero close to the wall. The following are the operating conditions of the fluid flow in the duct: U (m/s) = Air velocity T (C) = Air temperature Tw (C) = Flat plate wall temperature L1 (m) = Length of duct bend (lower wall) L2 (m) = Length of flat plate L3 (m) = Length of duct before bend L4 (m) = Length of duct after bend H1 (m) = Duct height H2 (m) = Distance between the flat plate and lower duct wall () = Bend sudtended angle The problem was investigated using the Reynolds-averaged Navier-Stokes (RANS) approach. The RANS equations are used to describe turbulent flows. These equations can be used with approximations based on knowledge of the properties of flow turbulence to give approximate time-averaged solutions. The assumption (known as the Reynolds decomposition) behind the RANS equations is that the time-dependent turbulent (chaotic) velocity fluctuations can be separated from the mean flow velocity. This transform then introduces a set of unknowns called the Reynolds stresses, which are functions of the velocity fluctuations, and which require a turbulence model (e.g., the two-equation k-epsilon model) to produce a closed system of solvable equations. The reduced computational requirements for the RANS equations, while still significant, are orders of magnitude less than that required for the original Navier-Stokes equations. Another advantage of using the RANS equations for steady fluid flow simulation is that the mean flow velocity is calculated as a direct result without the need to average the instantaneous velocity over a series of time steps. The CFD model, as shown in figure , was created using the software GAMBIT. The model was divided into 12 different faces. This was done in order to ensure that the meshing of each part of the model was done accurately. The faces numbers (before and after plate) had 30 meshes vertically and 40 meshes horizontally with a ratio of 1.2. This ensured that the mesh was fine enough at the walls in order to obtain an simulation of the boundary layers near the walls of the duct. Faces numbers (just on plate) had 15 meshes vertically and 40 meshes horizontally with a double sided ratio of 1.2. This was done so as to obtain fine meshes near the walls of the duct and near the flat plate so that boundary layers could be analysed accurately and accurate y+ values could be obtained. 2

Mihir Meetarbhan

0922031

BOUNDARY CONDITIONS 1. Inlet At the inlet, a velocity inlet boundary condition was applied and the velocity components were also specified x = 9 m/s. The temperature of the air was also set to 29C (302K). Turbulence intensity and hydraulic diameter were set to 1% and 0.5m respectively. As the flow being investigated is a flow in a pipe with low Reynolds number, the turbulence intensity is set to a relatively low value as there will be a medium of turbulence in the flow. Hydraulic diameter was used as the flow is a fully developed wall bounded flow. Hence the real diameter of the duct was used for the hydraulic diameter.

2. Wall At the wall, a wall boundary condition was set. This was set in order for the flow to be a wall bounded flow. This would then enable the boundary layers close to the wall to be studied. The wall has a roughness constant of 0.5 and a temperature of 300 k. 3. Outlet At the outlet, an outflow boundary condition was applied. 4. Flat Plate The flat plate had the boundary condition of a wall. Its temperature was set to 45 C (318 K) and the material used was aluminium. This boundary condition was set in order to analyse the heat transfer between the plate and the air flowing over it. Moreover, boundary layers forming on it can be studied.

TURBULENCE MODEL For the purpose of this study, the standard k- turbulence model was used where k is the turbulent kinetic energy and is the dissipation rate of the turbulent energy. is the variable that determines the scale of the turbulence, whereas the first variable, k, determines the energy in the turbulence. The use of this turbulence model is to predict the effect of turbulence on the mean flow. This model is a two-equation model meaning that it has two extra equations to represent the turbulent properties of the flow. Hence, this allows the model to account for effects like convection and diffusion of turbulent energy.

DISCRETISATION SCHEMES To account for transient effects, the governing equations must be discretised in time. Transient effects are usually dealt with by using a time stepping procedure, with an initial condition provided. The time dimension is divided into a set of discrete time steps, each of size t. The discretisation schemes used in solving this problem were: 3

Mihir Meetarbhan

0922031

Pressure standard Momentum second order upwind Turbulent k.e second order upwind Turbulent Dissipation Rate second order upwind Energy second order upwind The second order upwind discretisation scheme was used as this will yield an implicit system of second order accuracy with the extra storage of one additional time step meaning that the solution will be more accurate as compared to that of a first order upwind.

GRID DETAILS The grid used to solve this problem was structured grid. Structured grids have a considerable advantage over other grid methods in that they allow the user a high degree of control. It means that it enables the user to easily identify and efficiently access any point on the grid which allows for faster CFD codes. These types of grids utilize quadrilateral elements in 2D in a computationally rectangular array. Hence this enables the user to concentrate grid lines in regions where boundary layers and vortices are likely to occur and reduce the number of grid lines in regions where no major change in flow pattern take place. CONVERGENCE CRITERIA The convergence process is an iterative process which means that values are changing from one iteration to the next. As all numerical solutions contain errors, in order to minimize those errors, residual values are changed. Solver residuals represent the absolute error in the solution of a particular variable. Hence reducing their value give better convergence which in turn give better results. For this problem, the calculation was first performed with normalised residuals and was allowed to converge after 93 iterations. Following that, the residuals were reduced to 10 e-10 and the further iterations were performed in order to allow the solution to converge at 682 iterations, hence obtaining more accurate results.

Mihir Meetarbhan

0922031

DISCUSSION OF RESULTS

Boundary layer profiles

The Navier-Stokes equation deals with fluid flow which are viscous, compressible and heat conducting. The layer of fluid in which the effects of viscosity and thermal conductivity are important is known as the boundary layer. The law of the wall states that the average velocity of a turbulent flow at a certain point is proportional to the logarithm of the distance from that point to the wall, or the boundary of the fluid region. This law applies particularly to parts of the flow that are very close to the wall. Those are the regions where boundary layers are more likely to be produced. The log law states that: where

Figures show the lines set at different locations in the duct in order to analyse the boundary layer formations taking place on the lower wall of the duct according to the law of the wall. Figure shows the u+ vs y+ boundary layer profiles for the three lines placed before the plate. From figure , we can see that for Y+ values less than 10^3, the profiles follow the log law for a range of Y+ values. As the log law only applies to parts of the flow that are very close to the wall, i.e. 20% of the height of the flow, after Y+ values greater than 10^3, the profiles tend to deviate away from the log law as the ducts centerline is approached. This is so because close to the wall of the duct, the velocity tends to zero due to boundary layer formation. Hence, while approaching the ducts centerline, the log law no longer applies as the height of the flow is above 20% at the ducts centerline, the average velocity increases, thus giving rise to the divergence in the u+ v/s y+ plots. Figure show the u+ vs y+ boundary layer profiles for three lines placed at the bend. For these plots, the profiles tend to follow the log law for greater values of y+. This is so because at the 5

Mihir Meetarbhan

0922031

bend, the boundary layer is a lot thicker than before the bend. Hence there is a greater y-distance from the wall where the velocity is relatively low. After that region, where the log law no longer applies, the velocity increases as the duct centerline is approached. At that point, the u+ vs y+ profiles deviates slightly from the log law. Vector plot of the velocity field Figure show the vector plot of the velocity field on the plate. The vector plot is colour coded according to the velocity magnitude. From the plot we can see that close to the wall, the velocity tends to be zero due to the no slip boundary condition at the wall. The no slip condition for viscous flow states that at a solid boundary, the fluid will have zero velocity relative to the boundary. This is so because at the wall, the particles of the fluid tend to stick to the wall, hence creating a boundary layer. According to boundary layer equations, the flow velocity then increases rapidly within the boundary layer as shown in figure . Downstream of the leading edge of the plate, the flow is uniform and has no vorticity except in a narrow region adjacent to the plate. This narrow region containing vorticity is the viscous boundary layer. The thickness of the boundary layer increases along the length of the plate. Fluid deceleration is transferred successively from one fluid layer to the next by viscous shear stresses acting in the layers. Figure , compares the velocity profiles upstream and downstream along the plate. The velocity gradient at the wall is less downstream than upstream, indicating that the wall shear stress decreases along the plate. Temperature and Turbulence Kinetic Energy Figure , shows the variation of temperature with the distance normal to the lower wall of the duct at several locations downstream of the plate trailing edge. The graph shows that the temperature throughout the duct tends to remain constant except in the region downstream of the plate trailing edge. The highest temperature is found closer to the plate and as we move downstream, the temperature gets dissipated. This is because the highest amount of heat transfer takes place in the region close to the plate. Figure , shows the variation of turbulence k.e with the distance normal to the lower wall of the duct at several locations downstream of the plate trailing edge. The largest amount of turbulence k.e occurs at the trailing edge of the flat plate. As the boundary layer develops over the plate, the flow slowly turns turbulent. This is called boundary layer transition. Variation of friction coefficient on lower wall

Rate of Heat Transfer from the Flat Plate As air flows over the flat plate, thermal energy is transferred to the air by means of convection. As the plate is hotter than the air, a temperature gradient is formed between the two media. This gives rise to heat transfer between the air and the plate. From the temperature plot (figure), we can see that more heat transfer takes place above the flat 6

Mihir Meetarbhan

0922031

plate than below it. This is because the velocity of the air above the plate is higher than that below it. This is the case as The heat transfer from a hot plate to a laminar air stream is calculated by FLUENT using Newtons law:

Where

is the rate of heat transfer,

is the convection heat transfer coefficient, is the air temperature.

is the

area of the plate,

is the wall temperature and

Total Heat Transfer Rate (W) -------------------------------- -------------------Heated plate 323.95258 Heated plate shadow 301.66336 ---------------- -------------------Net 625.61594 The heat transfer data from FLUENT shows that the plate transfers 625.62 W of heat to the air flowing over it. Pressure loss between Inlet and Outlet The total pressure data between the inlet and the outlet by FLUENT: Total Pressure (pascal)(m2) -------------------------------- -------------------Inlet 25.310217 Outlet 22.634186 ---------------- -------------------Net 47.944405 FLUENT calculates the pressure drop by taking into consideration Reynolds number, pipe relative roughness, flow velocity, density of fluid and the dynamic viscosity of the fluid. For the flow in a pipe, Reynolds number: Re = v where v is the mean velocity of the object relative to the fluid, DH is the hydraulic diameter of the pipe and v is the kinematic viscosity. This states that there is a pressure drop of 2.676 Pa between the inlet and the outlet. The pressure drop in the pipe is a result of frictional forces acting on the fluid as it flows through the duct. The frictional forces are caused by a resistance to flow due to the fluid viscosity. Pressure drop is proportional to the frictional shear forces in the duct. Hence, the higher the wall roughness in the duct, the higher the pressure drop. 7

Mihir Meetarbhan

0922031

Pressure drop also occurs due to change in height of the duct and parasitic drag. Parasitic drag consists of form drag and skin friction drag. Those are the two types of drag that arise in the duct. Skin friction drag takes place as the wall has a certain roughness and is caused by the actual contact of the air particles against the walls of the duct. Pressure drag arises because of the shape of the object. As the duct has a flat plate midway through it and also as it has a 35 bend, those are contributing factors to the form drag. As air flows over the plate, a boundary layer is formed. Eventually the air tends to separate thus creating a wake hence opposing forward flow motion which in turn causes drag. Parasite Drag = Form Drag + Skin Friction Drag The two types of drag mentioned above follow the drag equation:

Where fluid,

is the force of drag, is the fluid density, v is the speed of the object relative to the is the drag coefficient and A is the reference area.

Frictional Drag on the flat plate

Reynolds number = Coefficient of drag ( Reference area = 1 Hence the frictional drag of the flat plate = ) = 0.001

Mihir Meetarbhan

0922031

APPENDIX
CFD model

Figure 1. CFD model with dimensions Duct with Investigation lines

Figure 2. Duct with lines to investigate several parameters a) b) c) d) e) f) g) h) i) j) k) Line 1 bottom Line 2 bottom Line 3 bottom Line before plate Midway plate End of plate Midway curve End of curve Line 1 top Line 2 top Line 3 top

Mihir Meetarbhan

0922031

Mesh From GAMBIT (Numbered faces)

Figure 3. Mesh from GAMBIT showing face numbers Turbulence Profile

Figure 4. Contours plot showing turbulence

10

Mihir Meetarbhan

0922031

Boundary layer profiles

Line 1 Bottom U+ vs Y+
3.00E+01 2.50E+01 2.00E+01 U+ 1.50E+01 1.00E+01 5.00E+00 0.00E+00 1.00E+00 U+ U+ theoretical

1.00E+01

1.00E+02 Y+

1.00E+03

1.00E+04

Line 2 Bottom U+ vs Y+
3.00E+01 2.50E+01 2.00E+01 U+ 1.50E+01 1.00E+01 5.00E+00 0.00E+00 1.00E+00 U+ U+ Theoretical

1.00E+01

1.00E+02 Y+

1.00E+03

1.00E+04

11

Mihir Meetarbhan

0922031

Line 3 Bottom U+ vs Y+
3.00E+01 2.50E+01 2.00E+01 U+ 1.50E+01 1.00E+01 5.00E+00 0.00E+00 1.00E+00 U+ U+ theoretical

1.00E+01

1.00E+02 Y+

1.00E+03

1.00E+04

Line before plate U+ vs Y+

3.00E+01 2.50E+01 2.00E+01 U+ 1.50E+01 1.00E+01 5.00E+00 0.00E+00 1.00E+00

U+ U+ theoretical

1.00E+01

1.00E+02 Y+

1.00E+03

1.00E+04

12

Mihir Meetarbhan

0922031

Midway Plate U+ vs Y+
3.00E+01 2.50E+01 2.00E+01 U+ 1.50E+01 1.00E+01 5.00E+00 0.00E+00 1.00E+00 U+ U+ theoretical

1.00E+01

1.00E+02 Y+

1.00E+03

1.00E+04

End of Plate U+ vs Y+
3.00E+01 2.50E+01 2.00E+01 U+ 1.50E+01 1.00E+01 5.00E+00 0.00E+00 1.00E+00 U+ U+ thoretical

1.00E+01

1.00E+02 Y+

1.00E+03

1.00E+04

13

Mihir Meetarbhan

0922031

Midway curve U+ vs Magnitude Y+

3.00E+01 2.50E+01 2.00E+01 1.50E+01 1.00E+01 5.00E+00 0.00E+00 1.00E+00 U+ U+ theoretical

Axis Title

1.00E+01

1.00E+02

1.00E+03

1.00E+04

Magnitude Y+

End of Curve U+ vs Magnitude Y+

3.00E+01 2.50E+01 2.00E+01 Axis Title 1.50E+01 1.00E+01 5.00E+00 0.00E+00 1.00E+00

U+ U+ theoretical

1.00E+01

1.00E+02

1.00E+03

1.00E+04

Magnitude Y+

14

Mihir Meetarbhan

0922031

Line 1 Top U+ vs Magnitude Y+

3.00E+01 2.50E+01 Axis Title 2.00E+01 1.50E+01 1.00E+01 5.00E+00 0.00E+00 1.00E+00 U+ U+ theoretical

1.00E+01

1.00E+02

1.00E+03

1.00E+04

Magnitude Y+

Line 2 Top U+ vs Magnitude Y+

3.00E+01 2.50E+01 2.00E+01 Axis Title 1.50E+01 1.00E+01 5.00E+00 0.00E+00 1.00E+00

U+ U+ theoretical

1.00E+01

1.00E+02

1.00E+03

1.00E+04

Magnitude Y+

15

Mihir Meetarbhan

0922031

Line 3 Top U+ vs Magnitude Y+

3.00E+01 2.50E+01 2.00E+01 Axis Title 1.50E+01 1.00E+01 5.00E+00 0.00E+00 1.00E+00 U+ U+ theoretical

1.00E+01

1.00E+02

1.00E+03

1.00E+04

Magnitude Y+
Line plots of Temperature

Temperature v/s Distance magnitude

3.20E+02 3.18E+02 3.16E+02 3.14E+02 Temperature 3.12E+02 3.10E+02 3.08E+02 3.06E+02 3.04E+02 3.02E+02 3.00E+02 0.00E+00 1.00E-01 2.00E-01 3.00E-01 4.00E-01 Distance Magnitude 5.00E-01 Before end of curve Temp Before midway curve temp End of curve temp End of plate temp Midway curve temp

6.00E-01

Figure 16. Plots showing difference in temperature at different locations in the duct

16

Mihir Meetarbhan

0922031

Line plots of turbulence kinetic energy

Turbulence K.E v/s Distance magnitude

8.00E-01 7.00E-01 6.00E-01 Turbulence k.e 5.00E-01 4.00E-01 3.00E-01 2.00E-01 1.00E-01 0.00E+00 0.00E+00 1.00E-01 Before end of curve turbulence Before midway curve turbulence End of curve turbulence End of plate turbulence Midway curve turbulence

2.00E-01 3.00E-01 4.00E-01 Distance Magnitude

5.00E-01

6.00E-01

Figure 17. Plots showing variation of turbulence k.e at different locations in the duct Skin Friction coefficient on lower wall

Figure 18. Line plot showing variation in skin friction coefficient along the lower wall of the duct

17

Mihir Meetarbhan

0922031

Convergence Plots

Figure 19. Plots showing initial convergence with no change in residuals

Figure 20. Diagram showing turbulence plot after initial convergence

18

Mihir Meetarbhan

0922031

Figure 21. Plots showing second convergence with change in residuals

Figure 22. Diagram showing turbulence plot after initial convergence

Velocity Profile

19

Mihir Meetarbhan

0922031

Figure 23. Velocity profile of duct

Vector plot of Velocity field

Figure 24. vector plot of velocity field

20

Mihir Meetarbhan

0922031

Figure 25. Vector plot of velocity field on lower wall of duct before bend

Figure 26. Vector plot of velocity field on plate

21

Mihir Meetarbhan

0922031

Figure 27. Vector plot of velocity field on bend

22