Simulating Conjugate Heat Transfer Flow Through Solid Pipe Using CONVERGE

OBJECTIVE:

1. Simulating Conjugate Heat Transfer Flow through a solid pipe and post process using CONVERGE Studio

2. Perform Grid dependancy Test and study the effect of Supercycling Stage Interval.

INTRODUCTION:

Conjugate heat transfer is a combination of heat transfer in solids and fluids, due to which there is a variation of temperature within the solid and fluid domain. For example consider a iron ball heated to high temperature and then immersed in water for cooling down, or say a solid pipe heated from outside which in turn heats the fluid inside the pipe.

In this project a basic general flow with air as working fluid has been set up. The material for pipe is set as aluminum which forms the solid region. The solid outer wall is subjected to a heat flux of 10,000 W/m2. This simulation will be performed for three different mesh sizes and the results will be analyzed. Later we will also study the effect of super-cycling stage interval for this case setup.

CASE SETUP:

1. Geometry

Outer Radius - 20 mm

Inner Radius - 15 mm

Thickness - 5 mm

Length - 200 mm

Flow Type - General Flow (Air - 77 % N2 and 23 % O2)

Kinematic Viscosity - 1.562e-5

Material of Pipe - Aluminum

2. Simulation Parameters

In Converge case set up interface there is an option called simulation parameters. Under simulation parameters there is run parameter. Since we are solving a transient case so we will choose a transient type solver for this problem. The simulation will be run from 0-0.5 seconds with an initial time step of 1e-7.

3. Boundary Conditions

Inlet - Inlet velocity is set up as 2.87m/sec. For pressure drichlet BC is used i.e zero normal gradient.

Outlet - Outlet pressure is set up as 101325 Pa

(Temperature is 300K initially)

Solid outer wall is stationary subjected to slip condition and a heat flux of 10,000 W/m2.

4. Physical Models

Turbulence Modelling - K`epsilon`

Supercyling - Super Cycle Stage Interval (0.05 sec)

5. Grid Control

We will run the simulation for 3 base grid sizes - 0.003,0.002 and 0.001 Respectively and analyze the results. Grid dependancy test helps us analyze whether proper mesh refinement is done so as to get accurate results.

6. Output/Post-Processing

In this project we have added an extra post variable file called \'y+\'. It helps capture the complicated flow near the wall and distinguishes different regions. It is a dimensionless quantity and is scaled y-coordinate in the layer closest to the wall. In Turbulent flow, we can divide the boundary layer into three domains. The inner layer where viscous shear dominates, the outer layer where large scale outer turbulent eddy shear dominates and the overlap layer where velocity profiles have a logarithim variation. each domain requires a specified value of y+ to capture the physics in the simulation.

GRID DEPENDANCY TEST:

Now we will perform the simulation for three different mesh sizes and trace the outlet temperature for solid and fluid domains. We will try to analyze which mesize size is accurate for proper convergence.

Above images shows that suspected solid temperature is going to be around 800 - 900 K and the fluid temperature around 350-370 K.

let\'s consider mesh size 3mm for the further experiments to save optimal time.

RESULTS:

Grid size 3mm

As seen from the above plots, it can be clearly observed that there is no change in temperature in the solid domain. This is due to the fact that in that region steady state based solver was implemented using supercycling. Hence, the real physics could not be modelled.

Also it can be noted that the fluid is fully developed at the exit of the pipe and in the last quarter there is no change in fluid properties of the fluid.

EFFECT OF SUPERCYCLING STAGE PARAMETER

Now, let us change the supercycling stage interval to 0.01 and 0.02 and 0.03 to check if it cause any changes in the mean temperature of solid and fluid regions. previously it was 0.05.

From the above graphs it\'s clearly seen that super cyling stage interval doesn\'t really effect the end temperature of solid and liquid domains. Hence, we can conclude that any super-cycling stage interval would be valid in determining the final temperatures. However, the major concern here would be finding out the effect of this interval in simulation time required for convergence. Here, we have performed simulation over a coarse mesh so the time difference might not be that significant. But, if we solve for finer meshes choosing accurate supercycling stage intervals would be important to save time.