Simulation Of Buoyancy Driven Flows Inside Heated Cavities Using LES And URANS

Approaches

School Of M.A.C.E.The University Of Manchester.

Presented by: DALILA AMMOURSupervisors: Prof H. Iacovides and Dr T.J. Craft

Code_Saturne users

Table of contents

Introduction

Objectives

Methods

Results

Coclusions

Introduction

Natural convection is defined as fluid motion where the flow arises

naturally from the effect of density differences which give rise to buoyancy

forces responsible for generating the flow.

Buoyancy-driven flows in enclosures have a number of technical

applications, ranging from cooling of electronic equipments, to the thermal

design of furnaces, energy storage systems and cooling of nuclear reactors.

Convection cell

Objectives

Predict 2-D and 3-D turbulent buoyant flow inside vertical (Betts) and

inclined cavities at 60°, 15° and 165° to the horizontal, using RANS and

LES approaches within Code_Saturne.

RANS simulation: test different RANS models and compare the results

obtained with recent experimental data.

LES simulation: Validate the RANS simulations and reproduce

accurate results.

Conclude which of the models tested within Code_Saturne can produce

reliable predictions of this kind of flow.

MethodsComputational meshes: Structured meshes (uniform coarse and non-

uniform fine grids for high and low Reynolds number models

respectively).

CFD code: Unstructured finite volume code Code_Saturne.

RANS models tested: high-Re κ-ε with wall functions, SST κ-ω

scheme, Re-stress transport models (LRR and SSG), ν²-f model and

the new version φ-α model.

Spatial discretization: RANS and LES, Second-order centered

scheme.

Time discretization: RANS, Implicit First-order Euler scheme. LES:

Second-order Crank- Nicolson scheme.

Physical properties: H/L=28.6, ΔT=18°C, Pr=0.71, Ra=0.86x106.

Modifications due to buoyancy

The Boussinesq approximation is used in the present study. It states that density differences are sufficiently small to be neglected, except where they appear in terms multiplied by the gravity

Momentum equation

Thermal expansion parameter

Boussinesq approximation

Computational meshes

RESULTS2D RANS predictions of Vertical and inclined cavities at 60°

Temperature profiles

Nusselt number

Comparison of 2-D turbulent kinetic energy for the three angles of inclination (SSG prediction)

Comparison of 2-D and 3-D results (165° inclined cavity, High and Low Reynolds number models predictions)

2-D3-D

Parallel velocity profiles2-D 3-D

LES Results (Smagorinsky subgrid model)1. Inclined cavity at 165°

Averaged temperature profiles

Averaged parallel velocity profilesIso-Q contours coloured by

temperature

Temperature Normal velocities

2. Inclined cavity at 15°

Averaged temperature profiles

Temperature

Normal velocity

Local Nusselt number

Averaged parallel velocity profiles

Averaged normal velocity profiles

Conclusions

2-D vertical and inclined cavity at 60°, heated from the upper side,

gave similar results for temperature, velocity and turbulent kinetic

energy, All RANS models succeed to predict the flow pattern. However

for inclined cavity at 165° RANS models disagree with experimental

data because of a multi cellular flow.

3-D RANS computation of inclined cavity at 165° have been

performed. High-Re-number models reproduce the unsteady structures

present in the flow however low Reynolds number models capture one

recirculation cell.

LES computation of 3-D inclined cavity at 165° and 15° have been

performed. Unsteady structures present in the flow are fully captured. In

general good agreements with measurements is shown.