Conjugate Heat Transfer Simulation 

Introduction:

Conjugate Heat Transfer Simulation is the one of the key expect for the CFD simulation for various engineering application such as modeling for engine cooling jacket for engine block, modeling for cooling of battery packs with air etc. In this report one application was experience i.e. using flow through hollow pipe with particular Reynolds number.

Working conditions

The challenge was to enter the flow through pipe with Reynolds number: 7000

Inlet fluid was here for simple computation purpose was: air

According to formula for Reynolds number : R = where by definition : it is the ration of inertial force to viscous force.

Here Air is used as the primarily cooling medium and Metal Pipe for more realistic case here I used Aluminum as primarily material the case setup was similar as shown in figure below

The Diameter of the Hollow pipe was consider 0.04 m with length of 0.2 m.

Properties of Air

Temperature = 303.15 K

Density = 1.161 kg/m³

Viscosity = 1.872x10^-5 kg/m-s

Velocity from the above quantities will be 2.82 m/s.

In case of the heating the outer surface here aluminum was selected as solid material and heat flux was added with 10000 W/m².

The Base Grid was taken 0.04 m, conditions for super cycle was start time with 0.05 seconds and interval was taken 0.05 seconds. In order to do the simulation less computation sensitive converge for interference region such as fluid- solid, solid- fluid etc. Converge uses supercycle technique

What is super cycle?

CHT simulation would involve simultaneous CFD calculations for the solid and fluid phases. This approach is practical for some types of simulations but impractical for others. For example, the time-scale for the solid phase heat transfer is orders of magnitude larger than the time-scale for the fluid phase heat transfer. Thus it is not practical to run the numerical simulation entirely in a coupled transient manner. CONVERGE offers a novel approach: super-cycling to solve CHT problems in a computationally efficient manner.

Results

Simulation Time with change in the super cycle time intervals

Super Cycle Time Interval (seconds)	Computational time (seconds)
Base (0.05)	417.460124
0.01	346.195135
0.02	383.455873
0.03	355.782420

It Appears from the logfile that lower the Super Cycle time Interval affects the computational time greatly. This need to understand very deeply but from the manual it state the following In CHT until the super cycle time reaches the solver solves both part of the problem i.e. equations related to solid domain as well as fluid domain but the only difference is the solver doesn’t save the values of solid domain until the starting of the super cycle time here in our case 0.05 so when the time is reached the solver starts to record the values of the respective domain for each interval of time. Below is the graph for the temperature vs time for more detail review.

Grid Dependence Test

This test is require to check weather to check the error generated due to change in mesh size or not. This is due to the way for performing the simulation using cube as primarily size when the size are small there will be good chance of error generated. The main reason is that the model are based on RANS which are approximation of the naviour stokes equations so when the sizes are small or big there are chances that error will be great which will affect the over solution so one of the goal of the CFD Engineer to make grid in such a way that the solution doesn’t affect by the mesh generation. Here in this case I have consider the grid size of

• 04 m (Base)
• 01 m
• 02 m
• 03 m

In order to prove I considered Temperature at my outlet for the grid test here the graph between temperature vs time as follows

As you can see the results are not up to mark the solution is grid dependent. Note I have only change the gird size rest of the values of the kept the same do not get confuse the values with super cycle.

I recommend the next two grid should be taken between 0.01 and 0.02 meter for better result.

Here are the plots for the mesh generated

For base (0.05 m)	0.01 m

As seen from above the mesh generation looks clean and satisfactory.

Extra Work

Y+ value