ik - Turbulence Mechanics/CFD Group - University of Manchester
ik - Turbulence Mechanics/CFD Group - University of Manchester
ik - Turbulence Mechanics/CFD Group - University of Manchester
Create successful ePaper yourself
Turn your PDF publications into a flip-book with our unique Google optimized e-Paper software.
ALGERIAN REPUBLIC DEMOCRATIC AND POPULAR<br />
Ministry <strong>of</strong> teaching and scientific research<br />
UNIVERSITY OF SCIENCE AND TECHNOLOGY<br />
MOHAMED BOUDHIAF ORAN<br />
U.S.T.O<br />
FACULTY OF MECHANICAL ENGINEERING<br />
DEPARTEMENT OF MARINE ENGINEERING<br />
Thesis for the degree <strong>of</strong> master in energetic<br />
SIMULATION OF TURBULENT FLOW ACROSS IN-LINE<br />
TUBE BUNDLE USING DIFFERENT URANS MODELS<br />
Presented by<br />
Miss AMMOUR Dalila<br />
Supervisors<br />
Pr. ADJLOUT Lahouari<br />
Dr. ADDAD Yacine<br />
2006-2007
Contents<br />
List <strong>of</strong> Figures.................................................................................................................4<br />
List <strong>of</strong> Tables..................................................................................................................7<br />
Abstract...........................................................................................................................8<br />
Acknowledgements......................................................................................................10<br />
Nomenclature...............................................................................................................12<br />
1 Introduction<br />
1.1 Introduction...........................................................................................................15<br />
1.2 Study objectives....................................................................................................19<br />
1.3 Outline <strong>of</strong> the thesis..............................................................................................20<br />
2 Literature Review<br />
2.1 Introduction...........................................................................................................21<br />
2.2 Literature review <strong>of</strong> tube bundles.........................................................................21<br />
2.2.1 LES <strong>of</strong> tube bundles....................................................................................21<br />
2.2.2 Heat transfer in tube bundles......................................................................22<br />
2.2.3 Pressure Fluctuations..................................................................................23<br />
2.2.4 Vortex shedding..........................................................................................24<br />
2.2.5 Vibrations………………………………………………………….……...26<br />
3 Governing Equations<br />
3.1 Introduction..........:..............................................................................................28<br />
3.2 Navier-Stokes equations......................................................................................28<br />
3.2.1 Reynolds Averaging.....................................................................................28<br />
3.3 Classes <strong>of</strong> turbulence models...............................................................................29<br />
3.3.1 Algebraic turbulence models…………………………………...…………29<br />
3.3.1.1 Baldwin-Lomax model…………………………………...……….29<br />
3.3.1.2 Cebeci-Smith model……………………………………………….30<br />
3.3.2 One-equation turbulence models………………………...………………...30<br />
3.3.2.1 Prandtl’s one-equation model……………...………………………30<br />
3.3.3 Two equation turbulence models………...…………………………………31<br />
3.3.3.1 Boussinesq eddy viscosity assumption……………………...……..31<br />
2
3.3.3.2 K-epsilon models……………………...…………………………...32<br />
3.3.3.3 K-omega models…………………………….……………………..33<br />
3.3.4 V2-f models……………………………………………………………...35<br />
3.2.5 Reynolds stress model (RSM)…………………………………………...36<br />
3.3.6 Large Eddy simulation (LES)…………………………………………....40<br />
3.3.7 Detached Eddy simulation (DES)…………………….………….……...42<br />
3.3.8 Direct numerical simulation (DNS)…………………….……………….43<br />
3.3.9 The SST- Cas<br />
model……………………………….……………………..43<br />
3.4 <strong>Turbulence</strong> modelling <strong>of</strong> unsteady flows (URANS)………….……….………44<br />
3.4.1 Introduction………………………………………….…………….……..44<br />
3.4.2 Unsteady Reynolds Navier-Stokes equations………….……….……….44<br />
3.4.2 <strong>Turbulence</strong> Modelling <strong>of</strong> Unsteady Cross Flow In-line Tube Bundle…..45<br />
4 Numerical Simulations<br />
4.1 Introduction.........................................................................................................46<br />
4.1.1 Pre-processor………………………………….……………………….....46<br />
4.1.2 Solver (Code-Saturne)……………………….…………………………..46<br />
4.1.3 Post-Processor………………………………….………………………...47<br />
4.2 The Finite Volume method.................................................................................49<br />
4.3 Time Discritisation..............................................................................................51<br />
4.4 Boundary Conditions..........................................................................................53<br />
4.4.1 Inlet.............................................................................................................53<br />
4.4.2 Outlet...........................................................................................................54<br />
4.4.3 Walls and symmetries.................................................................................54<br />
5 Results and Discussion <strong>of</strong> the simulation<br />
5.1 Introduction……………………………………………...……………………...57<br />
5.2 Case description………………………………………...………….…………...58<br />
5.3 Grid generation…………………………………………………………………59<br />
5.4 Discussion <strong>of</strong> the results………………………………..………………………60<br />
6 Conclusions and Recommendations for future work<br />
5.1 Final remarks.......................................................................................................92<br />
5.2 Recommendations for future work......................................................................93<br />
7 Bibliography.…...……………………………………………...………………….. 94<br />
3
List <strong>of</strong> Figures<br />
1.1<br />
4.1<br />
4.2<br />
5.1<br />
5.2<br />
5.3<br />
5.4<br />
5.5<br />
5.6<br />
5.7<br />
5.8<br />
5.9<br />
5.10<br />
5.11<br />
5.12<br />
5.13<br />
5.14<br />
5.15<br />
Turbulent flow around circular cylinder (Catallano et al.2003)……………………...<br />
Steps <strong>of</strong> Numerical simulation <strong>of</strong> across flow in-line tube bundles………………….<br />
Notations for the spatial discritisation……………………………………………….<br />
Tube arrangements…………………………………………………………………...<br />
Geometry <strong>of</strong> in-line tube bundles…………………………………………………….<br />
Boundary conditions <strong>of</strong> tube bundles………………………………………………..<br />
Cross sectional view <strong>of</strong> 2D grid (2X2 arrangement) N=5400 cells, y+= [13-70]……<br />
Cross sectional view <strong>of</strong> 2D grid (3X3 arrangement) N=21600 cells, y+= [13-70]…..<br />
Cross sectional view <strong>of</strong> 3D grid (3X3 arrangement) in XY, YZ and XZ sections:<br />
N=604800 cells, y+= [13-70]………………………………………………………...<br />
Evolution <strong>of</strong> pressure and velocity, Comparison between URANS models.<br />
(a) Pressure, (b) Velocity…..………………………………………………………...<br />
2D Instantaneous Pressure Contour field in a XY cross sectional view for gap ratio<br />
1.44. (a) k-ε model, (b) RSM, (c) k-ω SST, (d) SST-C as ……………………………<br />
2D Instantaneous velocity contour field in a XY cross Sectional view for gap ratio<br />
1.44. (a) k-ε model, (b) RSM, (c) k-ω SST, (d) SST-C as ……………………………<br />
2D Velocity vectors field in a XY cross Sectional view for gap ratio 1.44. (a) k-ε,<br />
(b) RSM, (c) k-ω SST, (d) SST-C as ………………………………………………….<br />
2D Vorticity field in a XY cross sectional view for gap ratio 1.44. (a) k-ε, (b) RSM,<br />
(c) k-ω SST, (d) SST-C as …………………………………………………………….<br />
3D mean pressure distribution in a XY cross view for P/D=1.44, Re=70000. (a) k-<br />
ω SST, (b) RSM, (c) SST- Cas<br />
, (d) DES, (e) LES <strong>of</strong> Imran for P/D=1.5, Re=15000..<br />
3D averaged velocity field in a XY cross view for P/D=1.44, Re=70000. (a) k-ω<br />
SST, (b) RSM, (c) SST- Cas<br />
, (d) DES, (e) LES <strong>of</strong> Imran for P/D=1.5, Re=15000…..<br />
Mean pressure distribution around centre tube, comparison between 2D Unsteady<br />
RANS for P/D=1.44, Re=70000 and LES <strong>of</strong> Imran (Star-V4) for P/D=1.5<br />
Re=15000 and Experiment <strong>of</strong> Yahiaoui et al. (2007)………………………………..<br />
Mean pressure distribution around centre tube, comparison between 2DUnsteady<br />
RANS for P/D=1.44, Re=70000 and LES <strong>of</strong> Imran (Star-V4) for P/D=1.5<br />
Re=15000 and Experiment <strong>of</strong> Yahiaoui et al. (2007)………………………………..<br />
4<br />
15<br />
48<br />
53<br />
58<br />
66<br />
66<br />
67<br />
67<br />
68<br />
69<br />
70<br />
71<br />
72<br />
73<br />
74<br />
75<br />
76<br />
76
5.16<br />
5.17<br />
5.18<br />
5.19<br />
5.20<br />
5.21<br />
5.22<br />
5.23<br />
5.24<br />
5.25<br />
5.26<br />
5.27<br />
5.28<br />
5.29<br />
5.30<br />
5.31<br />
Mean pressure distribution around centre tube, comparison between 3D<br />
UnsteadyRANS for P/D=1.44, Re=70000 and LES <strong>of</strong> Imran (Star-V4) for P/D=1.5<br />
Re=15000 and Experiment <strong>of</strong> Yahiaoui et al. (2007)………………………………..<br />
Mean velocity pr<strong>of</strong>ile, Comparison between 2D Unsteady RANS, Re=70000 and<br />
LES <strong>of</strong> Imran Re=15000 (Star-V4) and experiment <strong>of</strong> Aiba et al. (1982) in the<br />
wake <strong>of</strong> centre tubes at x=4.33cm……………………………………………………<br />
Mean velocity pr<strong>of</strong>ile, Comparison between 2D Unsteady RSM, Re=70000 and<br />
LES <strong>of</strong> Imran at Re=15000 (Star-V4) and experiment <strong>of</strong> Aiba et al. (1982) in the<br />
wake <strong>of</strong> centre tubes at x=4.33cm……………………………………………………<br />
Mean velocity pr<strong>of</strong>ile, Comparison between 2D Unsteady RSM, SST- C ,<br />
Re=70000 and LES <strong>of</strong> Imran (Star-V4), Re=15000 in the wake <strong>of</strong> centre tubes at<br />
x=4.33cm……………………………………………………………………………..<br />
Mean velocity pr<strong>of</strong>ile, Comparison between 3D URANS, Re=70000 and LES <strong>of</strong><br />
Imran (Star-V4) at Re=45000 and experiment <strong>of</strong> Aiba et al. (1982) in the wake <strong>of</strong><br />
centre tubes at x=4.33cm……………………………………………………………..<br />
Mean velocity pr<strong>of</strong>ile, Comparison between SST- C ,Re=70000 and LES <strong>of</strong> Imran<br />
(Star-V4), Re=45000 in the wake <strong>of</strong> centre tubes at x=4.33cm…………………….<br />
Fluctuating Pressure DPS at location <strong>of</strong> probe 1……………………………………..<br />
Fluctuating Pressure and DPS at location <strong>of</strong> probe 3………………………………...<br />
(a) Fluctuating Pressure and DPS at location <strong>of</strong> probe 6. (b) LES<br />
(Benhamadouche)…………………………………………………………………….<br />
Fluctuating Velocity and DPS at location <strong>of</strong> probe 1………………………………..<br />
Fluctuating Velocity and DPS at location <strong>of</strong> probe 3………………………………..<br />
Fluctuating Velocity and DPS at location <strong>of</strong> probe 6………………………………..<br />
Reynolds stresses in the wake <strong>of</strong> the centre tubes. (a) , (b) , (c) ,<br />
(d) ……………………………………………………………………………..<br />
Mean velocity pr<strong>of</strong>iles <strong>of</strong> RSM in the wake <strong>of</strong> the centre tubes. (a) , (b) ,<br />
(c) , (d) u/uo…………………………………………………………………….<br />
Iso-surface <strong>of</strong> parameter Q for the instantaneous flow across in-line tube bundles.<br />
(a) RSM, (b) k-ω SST, (c) SST- Cas<br />
, DES…………………………………………...<br />
Comparison between Code-Saturne (in right) and Star-CD (in left). k-ω SST,<br />
(a)Pressure, (b) Velocity, (c) Turbulent kinetic energy……………………………...<br />
5<br />
as<br />
as<br />
77<br />
77<br />
78<br />
78<br />
79<br />
79<br />
80<br />
81<br />
82<br />
83<br />
84<br />
85<br />
86<br />
87<br />
89<br />
90
5.32<br />
3D mean velocity vectors, (a) k-ω SST, (b) SST- C , (c) RSM, (d) DES, (e) LES<br />
(Benhamadouche)…………………………………………………………………….<br />
6<br />
as<br />
91
List <strong>of</strong> Tables<br />
3.1 Coefficients <strong>of</strong> the standard k-ε model……………………………………………..…….31<br />
3.2 Coefficients <strong>of</strong> the k-ω SST model………………………………………………..….......33<br />
3.3 Coefficients <strong>of</strong> the LRR model…………………………………………………..……….37<br />
3.4 Coefficients <strong>of</strong> the SSG model………………………………...………………..………..38<br />
5.1 Parameters <strong>of</strong> 2D and 3D grids <strong>of</strong> the present case……………………………………....60<br />
7
Abstract<br />
The flow in tube bundles is <strong>of</strong> great interest to the power generation industry, not only for the<br />
study <strong>of</strong> performance <strong>of</strong> great exchangers. Safety studies require predictions <strong>of</strong> vibrations<br />
caused by fluid-structure interaction or large temperature fluctuations that eventually lead the<br />
thermal stripping. The cross flow in a 2D and 3D square in-line tube bundle is computed for<br />
pitch ratio <strong>of</strong> P/D=T/D=1.44 and Reynolds number <strong>of</strong> 70000. The grid generated is structured.<br />
Unsteady Reynolds Navier-Stokes models are widely used for the complex unsteady flows. In<br />
the present case URANS models are used to examine the flow predictions in in-line tube<br />
bundle. URANS models tested are standard κ – ε, Menter`s shear stress transport (MSST) [37]<br />
and the Reynolds Stress Models (RSM). Other models are used, the new SST- C [18] model<br />
for 2D and 3D calculations moreover DES approach for 3D simulation. This case is computed<br />
by Code-Saturne based on the finite volume method. Quantitative and qualitative results are<br />
analyzed then compared with LES and experimental data. The 2D simulations fail to capture<br />
the complete flow physics hense 3D calculations on the other hand seem to produce better<br />
results <strong>of</strong> pressure and velocity pr<strong>of</strong>iles and agree better with LES and experiment. Good<br />
predictions are retained with the new SST- C model [18]. The three models k-ω SST, SST-<br />
Cas<br />
as<br />
and RSM seems to give similar predictions <strong>of</strong> the flow. Code Star-CD is used for<br />
comparison. It gives similar results and confirms the asymmetry <strong>of</strong> the flow. When<br />
frequencies <strong>of</strong> oscillations are given. This is done by using Density Power Spectrum (DPS)<br />
and localizing the peak values (the most energetic frequency). By applying DPS to the<br />
velocity's and pressure's signals, one clear peak is obtained around the frequency 45Hz<br />
(St=0.84) similar than the LES [13]. It means that a big recirculation coexists in the bottom <strong>of</strong><br />
the tube then the shear stress is higher in the bottom.<br />
Résumé<br />
L'écoulement dans les faisceaux de tubes est d'un grand intérêt au sein de l'industrie de<br />
production d'électricité, non seulement pour l'étude de l'exécution de grands échangeurs. Les<br />
études de sécurité exigent des prévisions de vibrations provoquées par l'interaction fluidestructure<br />
ou des grandes fluctuations de la température qui mènent par la suite au<br />
dépouillement thermique. L'écoulement dans un 2D et 3D faisceau de tubes intégré carré est<br />
simulé pour un rapport de P/D=T/D=1.44 et un nombre de Reynolds de 70000. Le maillage<br />
8<br />
as
généré est structuré. Les modèles URANS sont largement répandus pour les écoulements<br />
instables complexes. Dans le cas present les modèles URANS testés sont: κ – ε standard,<br />
(MSST) [37] de Menter et (RSM). Autres modèles sont utilisés, le nouveau model SST-<br />
C as [18] pour les calculs 2D et 3D en plus de l`approche DES pour la simulation 3D. Les<br />
conditions aux limites sont périodiques. Le cas present est simulé par Code-Saturne basé sur<br />
la méthode des volumes finis. Des résultats qualitatifs et quantitatifs sont alors analysés et<br />
comparés à la LES et aux données expérimentales. Les simulations 2D ne capturent pas<br />
complètement l'écoulement mais d`autre part les résultats des calculs 3D semblent produire de<br />
meilleurs résultats des pr<strong>of</strong>ils de pression et de vitesse et mieux conformes à la LES et à<br />
l'expérience. De bonnes prévisions sont captées avec le nouveau modèle SST- C [18]. Les<br />
trois modèles k-ω SST, SST- C et RSM semblent donner de mêmes prévisions de<br />
as<br />
l'écoulement. Le Code Star-CD est employé pour la comparaison. Il donne des résultats<br />
semblables et confirme l'asymétrie de l'écoulement. Quand les fréquences des oscillations sont<br />
indiquées. Ceci est fait en employant le spectre (DPS) et en localisant les valeurs de crête (la<br />
fréquence la plus énergique). En appliquant le DPS aux signaux de la vitesse et de la pression,<br />
une crête claire est obtenue autour de la fréquence 45Hz (St=0.84) vérifié avec la LES [13]. Il<br />
signifie qu'un grand recyclage coexiste au dessous du tube alors l'effort de cisaillement est<br />
plus grand au dessous.<br />
9<br />
as
Acknowledgements<br />
I would l<strong>ik</strong>e begin my sincere gratitude to God for his help and special thanks to<br />
Pr<strong>of</strong>essor Dominique Laurence, my supervisors Pr<strong>of</strong>essor Adjlout Lahouari and Dr.<br />
Yacine Addad and also Dr. S<strong>of</strong>iane Benhamadouche for their invaluable guidance and<br />
continuous advice throughout my present work. I wish to <strong>of</strong>fer my thanks to Dr.<br />
Alistair Revell and Dr. Juan Uribe. They have been an endless source <strong>of</strong><br />
encouragement and inspiration.<br />
I also <strong>of</strong>fer my thanks to all the <strong>CFD</strong> team in <strong>University</strong> <strong>of</strong> <strong>Manchester</strong>, School <strong>of</strong><br />
Mace and <strong>University</strong> <strong>of</strong> Oran USTO IGCMO in particular Dr. Aounallah Mohamed.<br />
There have many other people who have contributed to this work and to the fun<br />
environment in which it has be carried out. I'm grateful to Dr. Charles Moulinec,<br />
Nicolas Jarrin, and Dr. Imran Afgan. I can't forget to mention the help <strong>of</strong> Pat Shepherd<br />
for its support during the training course.<br />
My friends have been very important to me during this time, in particular Amel and<br />
Zahid, Hajira and there are many others, I <strong>of</strong>fer a general thanks to every one else.<br />
Most importantly, for their love and inspiration, I would l<strong>ik</strong>e to thank my family, in<br />
particular my parents, my brothers, my grandmother, Mounia, Sara and Fatima. They<br />
have constantly supported me taught me and encouraged me, and it is always a huge<br />
motivation to show them how grateful Iam.<br />
10
This work is dedicated to my parents<br />
11
Nomenclature<br />
Greek letters<br />
α Coefficients <strong>of</strong> the κ - ω SST model [-]<br />
1 4 ,...,α<br />
β Thermal expansion coefficient [-]<br />
Δ<br />
Filter width [m]<br />
δ ij<br />
Kronecker delta [-]<br />
Γ<br />
k<br />
Diffusion coefficient [-]<br />
Von Karman constant [-]<br />
λ Integral length scale [m]<br />
μ Molecular viscosity [N.s/m²]<br />
μ t<br />
Function <strong>of</strong> the local boundary layer velocity pr<strong>of</strong>ile [-]<br />
ν Dynamic viscosity [m²/s]<br />
ν τ<br />
Sub grid eddy viscosity [m²/s]<br />
ν t<br />
Turbulent Viscosity [m²/s]<br />
Ωij<br />
Rotation rate tensor [1/s]<br />
ρ Density [kg/m³]<br />
τ ij<br />
Viscous stress [N/m²]<br />
ε Isotropic dissipation [N/m²]<br />
ε ij<br />
Turbulent dissipation rate tensor [N/m²]<br />
Latins letters<br />
n Unit vector representing the wall-normal direction [-]<br />
A Lumley’s flatness parameter [-]<br />
A+ Van Driest damping coefficient [-]<br />
a ij<br />
bij<br />
Cij<br />
C p<br />
Anisotropy tensor [m²/s²]<br />
Normalised anisotropy tensor [-]<br />
Cross stress tensor [m²/s²]<br />
Pressure Coefficient [-]<br />
12
Cs<br />
Smagorinsky coefficient [-]<br />
D Diameter <strong>of</strong> cylinder [m]<br />
ν<br />
Dij<br />
T<br />
Dij<br />
F1<br />
F2<br />
fij<br />
Viscous diffusion <strong>of</strong> Reynolds Stresses [-]<br />
Turbulent diffusion <strong>of</strong> Reynolds Stresses [-]<br />
First blending function for the SST model [-]<br />
Second blending function for the SST model [-]<br />
Normalised redistribution tensor [-]<br />
G Kernel <strong>of</strong> spatial filter [-]<br />
g Gravity [m/s²]<br />
g ij<br />
Velocity gradient tensor [1/s]<br />
k Turbulent kinetic energy [kg.m2/s2]<br />
Lz<br />
Spanwise extrusion length (homogeneous direction) [m]<br />
l Mixing length scale [m]<br />
N Number <strong>of</strong> grid cells [-]<br />
p Pressure [N/m²]<br />
P Average pressure. ith component [N/m²]<br />
P' Pressure fluctuation [N/m²]<br />
Pk<br />
Pref<br />
Production <strong>of</strong> turbulent kinetic energy [m²/s³]<br />
Reference pressure [N/m²]<br />
R Radius <strong>of</strong> cylinder [m]<br />
Re Reynolds number [ρU D/μ]<br />
Rij<br />
Reynolds stress tensor [m²/s²]<br />
S Filtered strain rate magnitude [-]<br />
Sij<br />
Strain tensor [1/s]<br />
St Strouhal number [-]<br />
T Turbulent time scale [s]<br />
U b<br />
Bulk velocity [m/s]<br />
Uo Inlet Velocity [m/s]<br />
u`<br />
Fluctuating velocity, ith component [m/s]<br />
13
u" i<br />
ui<br />
uk<br />
xi<br />
Modelled turbulent fluctuation in RANS [m/s]<br />
Instantaneous velocity, ith component [m/s]<br />
Friction velocity based on k [m/s]<br />
Coordinate, ith component [-]<br />
y+ Nondimensional wall distance. [-]<br />
U i<br />
Average velocity, ith component [m/s]<br />
U j<br />
Filtered velocity. ith component [m/s]<br />
Acronyms<br />
CDS Central Differencing Scheme<br />
<strong>CFD</strong> Computational Fluid Dynamics<br />
DES Detached Eddy Simulation<br />
DNS Direct Numerical Simulation<br />
EVM Eddy Viscosity Model<br />
GGDH Generalised Gradient Diffusion Hypothesis<br />
LES Large Eddy Simulation<br />
LRR Launder, Reece and Rodi model<br />
RANS Reynolds Averaged Navier-Stokes<br />
RSM Reynolds Stress Model<br />
SA Spalart Allmaras<br />
SCWF Scalable Wall Function<br />
SMC Second Moment Closure<br />
SSG Speziale, Sarkar and Gatski model<br />
SST Shear Stress Transport<br />
UDS Upwind Differencing Scheme<br />
14
Chapter I Introduction<br />
Chapter 1<br />
Introduction<br />
1.1 Introduction<br />
Computational fluid dynamics or <strong>CFD</strong> is the analysis <strong>of</strong> systems involving fluid flow, heat<br />
transfer and associated phenomena such as chemical reactions by means <strong>of</strong> computer-based<br />
simulation. The technique is very powerful and spans a wide range <strong>of</strong> industrial and non-<br />
industrial application areas. Some examples are:<br />
• Aerodynamics <strong>of</strong> aircraft and vehicles: lift and drag.<br />
• Hydrodynamics <strong>of</strong> ships.<br />
• Power plant: combustion in IC engines and gas turbines.<br />
• Turbomachinery: flows inside rotating passages, diffusers…etc.<br />
• Electrical and electronic engineering: cooling <strong>of</strong> equipment including micro-circuits.<br />
• Chemical process engineering: mixing and separation.<br />
• External and internal environments <strong>of</strong> building: wind loading and heating ventilation.<br />
• Marine engineering.<br />
• Environment engineering: distribution <strong>of</strong> pollutants and effluents.<br />
• Hydrology and oceanography: flows in rivers, oceans.<br />
• Meteorology: weather prediction.<br />
• Biomedical engineering: blood flows through arteries and veins.<br />
Numerical Simulations are used for two types <strong>of</strong> purposes.<br />
The first is to accompany research <strong>of</strong> a fundamental kind. By describing the physical<br />
mechanisms governing fluid dynamics better, Numerical Simulation help to understand model<br />
and later control these mechanisms. This kind <strong>of</strong> study requires that the Numerical Simulation<br />
produce data <strong>of</strong> very high accuracy, which implies that the physical model chosen to represent<br />
the behavior <strong>of</strong> the fluid must be pertinent and that the algorithm used by the computer<br />
system, must introduce no more that a low level <strong>of</strong> error. The quality <strong>of</strong> the data generated by<br />
the numerical simulation also depends on the level <strong>of</strong> resolution chosen. For the best possible<br />
precision, the simulation has to take into account all the space-time scales affecting the flow<br />
dynamics. When the range <strong>of</strong> scales is very large, as it is in turbulent flows.<br />
15
Chapter I Introduction<br />
Numerical Simulation is also used for another purpose: engineering analysis. Where flow<br />
characteristics need to be predicted in equipment design phase. Here, the goal is no longer to<br />
produce data for analyzing the flow dynamic itself, but rather to predict certain <strong>of</strong> the flow<br />
characteristics or, more precisely, the values <strong>of</strong> physical parameters that depend on the flow,<br />
such us the stresses exerted on an immersed body, the production and propagation <strong>of</strong> acoustic<br />
waves. The purpose is to reduce the cost and time needed to develop a prototype. The desired<br />
predictions may be either <strong>of</strong> the mean values <strong>of</strong> these parameters <strong>of</strong> their extremes.<br />
In <strong>CFD</strong>, numerical algorithms are used to reach an approximate solution to the flow field,<br />
which is commonly represented by a discrete set <strong>of</strong> nodes, defined by a mesh specifically<br />
tailored to the geometry <strong>of</strong> the problem. The variation <strong>of</strong> physical values across the flow field<br />
can be expressed exactly by the differential Navier-Stokes equations which, in <strong>CFD</strong>, are<br />
replaced with sets <strong>of</strong> algebraic expressions known as discretised equations.<br />
Thus, instead <strong>of</strong> a closed-form analytical solution, the end product <strong>of</strong> <strong>CFD</strong> is a collection <strong>of</strong><br />
numbers at discrete space and time locations.<br />
<strong>Turbulence</strong> [56] is an irregular, chaotic state <strong>of</strong> fluid motion that occurs when the<br />
instabilities present in the initial or boundary conditions are amplified, and a self sustained<br />
cycle is established in which turbulent eddies (coherent region <strong>of</strong> vorticity) are generated and<br />
destroyed. <strong>Turbulence</strong> is best described through its characteristics (example <strong>of</strong> turbulent flow<br />
in figure1). The most distinguishing features <strong>of</strong> turbulent flows are:<br />
• Randomness: Turbulent flows are extremely sensitive to initial and boundary conditions:<br />
slight changes in term will make the development <strong>of</strong> flows that are otherwise identical<br />
diverge, as the differences are exponentially amplified in time. The statistical properties <strong>of</strong><br />
the flows will, however, remain unchanged. This randomness is the reason why much <strong>of</strong><br />
turbulence research has relied on statistical methods <strong>of</strong> investigation and prediction.<br />
• Vorticity: We cannot call a random flow turbulent if the curl <strong>of</strong> the velocity vector is<br />
negligibly small, even if it has some <strong>of</strong> the other characteristics <strong>of</strong> turbulence. The random<br />
motions <strong>of</strong> waves on the ocean surface as well as the irrotational fluctuations in the potential<br />
flow above the boundary layer are examples <strong>of</strong> random non-turbulent flows. All turbulent<br />
flows are rotational and exhibit high levels <strong>of</strong> fluctuating vorticity, which is usually<br />
concentrated in regions with strong coherence (coherent structure <strong>of</strong> eddies). Vortex<br />
stretching is an essential component <strong>of</strong> turbulence dynamics.<br />
16
Chapter I Introduction<br />
• Mixing: Turbulent motions greatly enhance the transport <strong>of</strong> mass, momentum and energy.<br />
The enhanced mixing <strong>of</strong> momentum in a turbulent flow results is higher skin-friction<br />
coefficient, while the more effective mixing <strong>of</strong> different species results in more rapid<br />
dispersion <strong>of</strong> contaminants. The dominance <strong>of</strong> convective effects over diffusive ones is one<br />
<strong>of</strong> the key characteristics <strong>of</strong> turbulence.<br />
• Irregularity: Turbulent flow is irregular, random and chaotic. The flow consists <strong>of</strong> a<br />
spectrum <strong>of</strong> different scales (eddy sizes) where largest eddies are <strong>of</strong> the order <strong>of</strong> the flow<br />
geometry (i.e. boundary layer thickness, jet width, etc). At the other end <strong>of</strong> the spectra we<br />
have the smallest eddies which are by viscous forces (stresses) dissipated into internal<br />
energy. Even though turbulence is chaotic it is deterministic and is described by the Navier-<br />
Stokes equations.<br />
• Diffusivity: In turbulent flow the diffusivity increases. This means that the spreading rate <strong>of</strong><br />
boundary layers, jets, etc. increases as the flow becomes turbulent. The turbulence increases<br />
the exchange <strong>of</strong> momentum in e.g. boundary layers and reduces or delays there by<br />
separation at bluff bodies such as cylinders, airfoils and cars. The increased diffusivity also<br />
increases the resistance (wall friction) in internal flows such as in channels and pipes.<br />
• Large Reynolds numbers: Turbulent flow occurs at high Reynolds number. For example,<br />
the transition to turbulent flow in pipes occurs that Re=2300, and in boundary layers at<br />
Re=10000.<br />
• Three-dimensional: Turbulent flow is always three-dimensional. However, when the<br />
equations are time averaged we can treat the flow as two-dimensional<br />
• Dissipation: Turbulent flow is dissipative, which means that kinetic energy in the small<br />
(dissipative) eddies are transformed into internal energy. The small eddies receive the<br />
kinetic energy from slightly larger eddies. The slightly larger eddies receive their energy<br />
from even larger eddies and so on. The largest eddies extract their energy from the mean<br />
flow. This process <strong>of</strong> transferred energy from the largest turbulent scales (eddies) to the<br />
smallest is called cascade process.<br />
• Continuum: Even though it has small turbulent scales in the flow they are much larger than<br />
the molecular scale and the flow can be treated as a continuum.<br />
17
Chapter I Introduction<br />
Figure1: Turbulent flow around circular cylinder (Catallano et al.2003)<br />
The difficulty in predicting turbulence arises from the nonlinearity <strong>of</strong> the Navier-Stokes<br />
equations, which generate a broad range <strong>of</strong> length and time scales, with several orders <strong>of</strong><br />
magnitude between the smallest and the largest eddy. In light <strong>of</strong> this observation various<br />
different approaches have been developed and applied, with a hierarchy <strong>of</strong> complexity, which<br />
can be broadly categorized into three groups:<br />
Direct Numerical Simulation (DNS), Large Eddy Simulation (LES) and Reynolds Averaged<br />
Navier Stokes models (RANS):<br />
• Direct Numerical Simulation: The most accurate approach to turbulence simulation is to<br />
solve the navier-Stokes equations without averaging or approximation other than numerical<br />
discretisation whose errors can be estimated and controlled. It is also the simplest approach<br />
from the conceptual point <strong>of</strong> view. In such simulations, all <strong>of</strong> the motions contained in the<br />
flow are resolved. The computed flow field obtained is equivalent to a single visualization <strong>of</strong><br />
a flow or a short duration laboratory experiment.<br />
• Large Eddy Simulation: Turbulent flows contains a wide range <strong>of</strong> length and time scale;<br />
The large scale motions are generally much more energetic than the small scale ones, their<br />
size and strength make them by far the most effective transporters <strong>of</strong> the conserved<br />
properties. The small scale is usually much weaker and provides little transport <strong>of</strong> these<br />
properties. A simulation which treats the large eddies more exactly than small ones may<br />
make sense. Large eddy simulations are three dimensional, time dependant and expensive<br />
but much less costly than DNS <strong>of</strong> the same flow, because it is more accurate, DNS is the<br />
preferred method whenever it is feasible. LES is the preferred method for flows in which the<br />
Reynolds number is too high or the geometry is too complex to allow application <strong>of</strong> DNS.<br />
18
Chapter I Introduction<br />
• RANS models: Engineers are normally interested in knowing just a few quantitative<br />
properties <strong>of</strong> a turbulent flow. In Reynolds Averaged approaches to turbulence, all <strong>of</strong> the<br />
unsteadiness is averaged out. All unsteadiness is regarded as part <strong>of</strong> the turbulence. On<br />
averaging, the non-linearity <strong>of</strong> the Navier-Stokes gives rise to terms that must be modeled,<br />
just as they did earlier. The complexity <strong>of</strong> turbulence makes it unl<strong>ik</strong>ely that any single<br />
Reynolds-Averaged model will be able to represent all turbulent flows so turbulent models<br />
should be regarded as engineering approximations rather than scientific laws.<br />
• Very Large Eddy Simulation: It appears that we have to either use RANS, which is<br />
affordable, or LES, which is more accurate but rather expensive. It is natural to ask whether<br />
there is a method that provides the advantages <strong>of</strong> both RANS and LES while avoiding the<br />
disadvantages. The method that accomplished this is called Very Large Eddy Simulation or<br />
Unsteady Reynolds-Averages Numerical Simulation, in this method, one uses a RANS<br />
model but computes an unsteady flow. The results <strong>of</strong>ten contain periodic vortex shedding.<br />
When the results <strong>of</strong> such a simulation are time-averaged, they <strong>of</strong>ten agree better with<br />
experiment than steady RANS computations.<br />
• Detached Eddy Simulation: DES is suggested for separated flows (Travin et al., 2000). In<br />
this approach, RANS is used for the attached boundary layer and LES is applied to the free<br />
shear flow resulting from separation. This requires some means <strong>of</strong> producing the initial<br />
conditions for the LES in the separation region and this are a difficulty.<br />
1.2 Study Objectives:<br />
The flow in tube bundles is <strong>of</strong> great interest to the power generation industry, not only for the<br />
study <strong>of</strong> performance <strong>of</strong> heat exchangers. Safety studies require predictions <strong>of</strong> vibrations<br />
caused by fluid-structure interaction or large temperature fluctuations that eventually lead to<br />
thermal stripping. The flow within the bundles experiences complex unsteady behaviour,<br />
making it an attractive case to be studied using different numerical model. The unsteadiness in<br />
this flow can be a result <strong>of</strong> imposed fluctuating boundary conditions. The presence <strong>of</strong> such<br />
unsteadiness in a flow can significantly after the evolution <strong>of</strong> different parameters such as<br />
Reynolds stresses u iu , turbulent kinetic energy κ, and dissipation rate ε. Despite the<br />
j<br />
existence and the utility <strong>of</strong> progressively more complex modeling.<br />
URANS models are widely used for the complex unsteady flows. In the present case URANS<br />
models are used to examine the flow predictions in the in-line tube bundles. The URANS<br />
19
Chapter I Introduction<br />
models tested are standard κ – ε, Menter`s shear stress transport (MSST) and the Reynolds<br />
Stress Models (RSM). Other models are used, the new SST-Cas model (the standard SST<br />
model is used alone for the first few time steps in order to initialize the calculation for the<br />
SST-Cas model).<br />
After obtaining results, they must be compared with LES or available experimental data and<br />
then assess which <strong>of</strong> those models is able to reproduce an unsteady flow behaviour across the<br />
tubes.<br />
1.3 Outline <strong>of</strong> the thesis:<br />
The work presented here is organized as follows. After the introduction, Chapter 2 presented<br />
a literature review <strong>of</strong> in-line tube bundle. Chapter 3 gives a review <strong>of</strong> the existing turbulence<br />
models and those which are used during the course <strong>of</strong> this project. The numerical aspects <strong>of</strong><br />
the code used and steps <strong>of</strong> the numerical study are presented in Chapter 4. In Chapter 5 a<br />
turbulent flow across in-line tube bundle studied using different URANS models, SST-Cas<br />
model and DES is presented together with results and discussion <strong>of</strong> the simulation in this<br />
Chapter.<br />
Finally, Chapter 6 includes conclusions together with suggestions for future work.<br />
20
Chapter II Literature Review<br />
Chapter 2<br />
Literature review<br />
2.1 Introduction<br />
The flow within the bundles experiences complex unsteady behaviour, Random excitation<br />
forces can cause low-amplitude tube motion that will result in-long-term-fretting-wear or<br />
fatigue. All these problems attract the attention <strong>of</strong> researchers in the whole world.<br />
2.2 Literature review <strong>of</strong> tube bundles<br />
2.2.1 LES <strong>of</strong> tube bundles<br />
Rollet-Miet et al. [58] presented the first LES calculations using the finite-element method<br />
<strong>of</strong> the turbulent flow across a staggered tube bundle. Since then, few publications have<br />
appeared although tube bundles are widely employed in cross-flow heat exchangers, as they<br />
combine case <strong>of</strong> construction with good thermal and mechanical efficiency. Rollet-Miet et al<br />
[58] pointed out the superiority <strong>of</strong> the LES technique because it is better suited the flows<br />
where the size <strong>of</strong> eddies (integral length case <strong>of</strong> the turbulence) is comparable to the size <strong>of</strong><br />
the obstacles <strong>of</strong> the flow.<br />
Benhamadouche and Laurence [13] performed similar LES calculations for the turbulent flow<br />
across the staggered tube bundle using the finite-volume method on a collocated unstructured<br />
grid. They found that the type <strong>of</strong> the subgrid scale models (whether the standard or the<br />
dynamic Smagorinsky) is not critical for this type <strong>of</strong> application. In this same year, Charles<br />
Moulinec, J.C.R Hunt and F.T.M Niuwstadt studied flow through a staggered array <strong>of</strong> parallel<br />
rigid cylinders computed with the help <strong>of</strong> a three dimensional (DNS) at Re=500 and 6000,<br />
when Re
Chapter II Literature Review<br />
development <strong>of</strong> rms level along the flow lane. A steady recirculation region consisting <strong>of</strong> a<br />
pair <strong>of</strong> counter-rotating vortices exists in the gap as found by Ziada and Oengoren [59].<br />
This regime is called reattachment regime. <strong>Turbulence</strong> intensities at this regime are small<br />
compared to the alternate vortex shedding regime. However, Liang and Papadakis [4]<br />
predicted a clear vortex shedding frequency behind the first cylinder and this is one reason<br />
that the LES calculations for the in-line tube bundle over-predicted the rms level behind that<br />
cylinder.<br />
Konstantinidis et al. [79] examined experimentally the effect <strong>of</strong> inlet flow pulsation in across<br />
flow over a tube array with an external frequency around twice that the flow pulsation<br />
activates the flow field behind the first cylinder and increases the turbulence intensities for the<br />
first three cylinders.<br />
In a later Study, Konstantinidis et al. [80] also observed a symmetrical vortex formation mode<br />
when the external frequency is around triple that <strong>of</strong> the natural alternate vortex shedding. The<br />
effect <strong>of</strong> pulsation on the flow field and heat transfer in a six row in-line tube array is<br />
investigated using 3D LES technique by Chunlei Liang and George Papadakis [4].<br />
2.2.2 Heat transfer in tube bundles<br />
Akhilech Gupta [66], an experiment <strong>of</strong> nucleate boiling heat transfer in an electrically<br />
heated 5X3 in-line horizontal tube bundle under pool and low cross flow. It is observed that<br />
the heat transfer is minimum on bottom row tubes and increases in the upward direction with<br />
maximum values on top row tubes. Also, heat transfer coefficient on central column tubes was<br />
found to be slightly higher than those on the corresponding side tubes.<br />
Chunlei Liang and George Papadakis [10] the aim is to study the effect <strong>of</strong> pulsation on the<br />
field and convective heat transfer over an in-line cylinder array at a sub critical Reynolds<br />
number Re=3400 (based on cylinder and the gap velocity across the minimum section) using<br />
LES technique. The heat transfer rate in the front part <strong>of</strong> the second cylinder is greatly<br />
enhanced due to the vortex shedding lock on behind the first cylinder. LES computations <strong>of</strong><br />
six-row cylinders demonstrated that heat transfer around the third row and downstream<br />
cylinders are not influenced much by the external pulsation.<br />
Haitham M.S. Bahaidrah and M. Ijaz and N.K. Anauds [2], two dimensional study developing<br />
fluid flow and heat transfer across five in-line tube bundle with a Prandlt number <strong>of</strong> 0.7. The<br />
tube cross-sectional shapes studied were circular, flat, oval and diamond tubes were compared<br />
22
Chapter II Literature Review<br />
with each other and with those for circular tubes. Flat and oval tubes <strong>of</strong>fered greater flow<br />
resistance and heat transfer rate when compared with circular cylinders for all values <strong>of</strong><br />
Reynolds number. Diamond tubes <strong>of</strong>fered less resistance and heat transfer rate. For Re>50,<br />
flat and oval tubes performed better.<br />
2.2.3 Pressure Fluctuations<br />
Ishigai et al. [60] investigates the flow pattern for a wide range <strong>of</strong> gap ratios. It is reported<br />
that for in-line tube bundle five distinct regions are formed. However, in case <strong>of</strong> square tube<br />
bundles only three distinct flow patters are observed; for very narrow gap ratios the free shear<br />
layer <strong>of</strong> the front <strong>of</strong> the cylinder attaches to the downstream cylinder thus stopping the<br />
Karman vortices to develop, for moderate gap ratio the Karman vortices are shed but are<br />
distorted and deflected due to downstream suppression, for very wide gap ratios regular<br />
Karman vortices are shed much l<strong>ik</strong>e in the case <strong>of</strong> a single cylinder.<br />
Aiba et al. [61] perform experimental study on square in-line tube banks for gap ratio <strong>of</strong> 1.2<br />
and 1.6. It is observed that the tube response <strong>of</strong> the downstream cylinders is quite different<br />
from the upstream ones. The pressure distribution around the cylinder surface shows highly<br />
deflected flow with a stagnation point <strong>of</strong> 45 degrees. The flow behaviour is asymmetric both<br />
these configurations which is owed to the narrow gap ratios.<br />
Traub [62] conducts open wind tunnel experiments to study the influence <strong>of</strong> turbulence<br />
intensity on pressure drop in in-line and staggered tube bundles at various Reynolds numbers.<br />
It is observed that the drag coefficient remains more or less the same for a wide range <strong>of</strong><br />
Reynolds numbers and only changes slightly for very high Reynolds numbers. It is also<br />
concluded from this study that as the Reynolds number becomes very high the circulation<br />
region shrinks due to shifting <strong>of</strong> the point <strong>of</strong> flow separation. Due to this shifting, the pressure<br />
drop decreases and hence the drag coefficient decreases. The paper provides experimental data<br />
<strong>of</strong> drag coefficient over a wide range <strong>of</strong> Reynolds number for various gap ratios.<br />
Lam and Fang [63] perform an experimental study on the effect <strong>of</strong> gap ratio on the flow over a<br />
square four cylinder in-line configuration. The paper discusses flow pattern, pressure<br />
distribution and lift and drag forces on cylinders at a Reynolds number <strong>of</strong> 12,800 based on<br />
free stream velocity. It is seen that at small gap ratios due to suppression <strong>of</strong> wake region<br />
vortex shedding is hardly formed. Moreover for these gap ratios the stagnation point is not at<br />
zero degrees rather the shift is 20-50 degrees from flow direction.<br />
23
Chapter II Literature Review<br />
Ishigai et al. [60] study fluid flow over in-line tube bundles for very low Reynolds numbers<br />
using finite element technique. The main idea behind the paper is to study the effect <strong>of</strong><br />
pressure drop on heat transfer. The research concludes that recirculation between the cylinders<br />
increases with an increase in Reynolds number. This causes the separation point to move<br />
further away from the rear stagnation point. In other words the angle <strong>of</strong> separation when<br />
measured from the front stagnation point decreases with an increase in Reynolds number.<br />
At a low Reynolds number <strong>of</strong> 200 Lam et al. [64] performed a particle image velocitymetry on<br />
a four cylinder square array where the gap ratio was 4.<br />
2.2.4 Vortex Shedding<br />
Van Atta et al. [65] studied chaotic and organized vortex shedding behind self excited<br />
cylinder wakes at fairly low Reynolds number using hot wire measurements and smoke flow<br />
visualization techniques. Their research revolves around two cases, first one being the ordered<br />
lock-in case in which only a single high order harmonic vibration frequency is excited, results<br />
indicated that the wake structure is span-wise periodic. The second case is the fully chaotic<br />
one where several high order vibration modes are simultaneously excited, results from this<br />
case show that the vortex street is disorganized and is definitely not span-wise periodic,<br />
however the statistical properties such as velocity signals are independent <strong>of</strong> the span-wise<br />
position.<br />
For a three cylinder arrangement a similar flow behavior is seen but with a switching in<br />
direction. Thus the flow pattern <strong>of</strong> the three by three arrangements is termed to be meta-stable<br />
by Zdravkovich and Stonebanks [67]. Kim and Durbin [68] suggest that this biased flow<br />
behavior is due to the turbulent perturbations in the incoming flow and is an intrinsic property<br />
<strong>of</strong> flow. Sayers [69] presents yet another experimental study for a four cylinder arrangement<br />
showing either a total suppression <strong>of</strong> vortex shedding or an asymmetrical pattern for very<br />
narrow gap ratios.<br />
Lam and Lo [70] have done extensive water tunnel experimentations on the wake formation<br />
and vortex shedding frequency <strong>of</strong> a square cylinder bundle at low Reynolds number <strong>of</strong> 2100<br />
with different angles <strong>of</strong> attack. Our interest is only the zero angle <strong>of</strong> attack that is the in-line<br />
arrangement. The bundle gap ratio varies from 1.28 to 5.96 along with the angle <strong>of</strong> attack.<br />
Interesting thing to note from this study is that a bitable state <strong>of</strong> wide and narrow wake exists<br />
for aspect ratios below 1.54. The asymmetric mode is initiated by an outward deflection <strong>of</strong> the<br />
outer shear layer <strong>of</strong> upstream cylinders which rolls up besides the downstream cylinders.<br />
24
Chapter II Literature Review<br />
Based on these observations it is thus concluded that a distinct oscillation <strong>of</strong> wake exists in<br />
the downstream flow. Finally the Reynolds number has little effect on the size and shape <strong>of</strong><br />
the wake since low spacing prohibits lengthening <strong>of</strong> the shear layers in the down stream<br />
direction. Ogengoren and Zaida [59] have done an experimental study on the vortex shedding<br />
and resonance in an in-line tube bundle. According to the authors when resonance occurs,<br />
pressure pulsations at discrete frequencies are produced. The high amplitude <strong>of</strong> these pressure<br />
pulsations causes vibrations and noise problems. The modes consisting <strong>of</strong> standing waves in<br />
normal direction to the flow is most l<strong>ik</strong>ely to cause resonance. The flow pattern <strong>of</strong> a non<br />
resonant mode and a resonant mode is entirely different. When resonance occurs vortices<br />
forming behind tubes have the same sense <strong>of</strong> rotation and are shed simultaneously from the<br />
same sides <strong>of</strong> the tubes. This means that the vortices behind all the tubes have the same sense<br />
and phase. Surface wave resonance is stated to be the reason behind this synchronization.<br />
Another interesting thing to note from this study is that when the stream-wise gaps between<br />
tube arrays is less than the tube diameters wake velocity pr<strong>of</strong>ile does not develop. Under these<br />
circumstances the gap scan is regarded as cavities bounded by shear layers which separate<br />
from the tube edges. The instability <strong>of</strong> these shears layers which is triggered and synchronized<br />
by the resonance mode causes the asymmetric flow pattern.<br />
Finally the study concludes that this asymmetric behavior is only seen when the stream-wise<br />
gap distances fairly narrow. For wider gap ratios the resonance occurs when the frequency <strong>of</strong><br />
vorticity shedding approaches the resonance frequency, the causes <strong>of</strong> this deviation from<br />
standard or classical reasoning is still under investigation and is an unsolved dilemma.<br />
Summer et al. [71] have done an extensive experimental study on two and three cylinders<br />
placed side by side for a Reynolds number range <strong>of</strong> 500-3000. It has been observed that both<br />
the two cylinder and three cylinder configurations show various flow patterns for different gap<br />
spaces. In this study the transverse ratio (T/D) is varied from 1.0 to 6.0. The three regions<br />
classified for this variation in T/D ratios are, small (T/D < 1.2), intermediate (1.2 < T/D 2.2). In the intermediate region which is also the interest region <strong>of</strong> our study<br />
an asymmetrical flow pattern is reported. This is also reported by Sumner et al. [71] and Kim<br />
&Durbin [68]. It is also observed by Kim & Durbin [68] that the biased flow pattern switches<br />
intermittently from being directed towards one cylinder to the other. Thus such a flow pattern<br />
is termed to be bistable. They concluded that this flow behavior is independent <strong>of</strong> Reynolds<br />
number and purely dependent upon the T/D ratio. Summer et al. [71] further correlates the two<br />
showing that the flow deflection decreases as the T/D ratio is increased.<br />
25
Chapter II Literature Review<br />
Wolfe and Ziada [78] used a feedback control on vortex shedding from two tandem cylinders.<br />
It was concluded that when a cylinder is placed in the wake <strong>of</strong> another cylinder then its<br />
unsteady loading is not only dependent upon the flow behavior in its own wake but also on the<br />
flow pattern in the wake <strong>of</strong> the upstream cylinder. On the basis <strong>of</strong> this, a feedback control was<br />
applied to reduce the response <strong>of</strong> the downstream cylinder to both turbulence excitations and<br />
vortex shedding. The study was based on two cases, the resonant case (lower flow speed, Re =<br />
41100) where the cylinder frequency <strong>of</strong> the vortex shedding coincides with the resonance<br />
frequency <strong>of</strong> the downstream cylinder and the non resonant case (slightly higher speed, Re =<br />
57900). The feedback control did not reduce the velocity fluctuations at the vortex shedding<br />
frequency instead it shifted the vortex shedding frequency to a higher level.<br />
Samir Ziada [77] describes the vorticity shedding excitation in tube bundles and its relation to<br />
the acoustic resonance mechanism. These phenomena are investigated by means <strong>of</strong> velocity<br />
and pressure measurements, as well as with the aid <strong>of</strong> extensive visualization <strong>of</strong> the unsteady<br />
flow structure at the presence and absence <strong>of</strong> acoustic resonance. Vorticity shedding excitation<br />
is shown to be generated by either jet, wake, or shear layer instabilities. The tube layout<br />
pattern (in-line or staggered), the spacing ratio, and Reynolds number determine which<br />
instability mechanism will prevail, and thereby the relevant Strouhal number for design<br />
against vorticity shedding and acoustic resonance excitations. Strouhal number design charts<br />
for vortex shedding in tube bundles are presented for a wide range <strong>of</strong> tube patterns and<br />
spacing ratios. Regarding the acoustic resonance mechanism, it is shown that the natural<br />
vorticity shedding, which prevails before the on set <strong>of</strong> resonance, is not always the source<br />
exciting acoustic resonance. This is especially the case for in-line tube bundles. Therefore,<br />
separate "acoustic" Strouhal number charts must be used when appropriate to design against<br />
acoustic resonances. To this end, the most recently developed charts <strong>of</strong> acoustic Strouhal<br />
numbers are provided.<br />
2.2.5 Vibrations:<br />
Weaver et al. [72] show another interesting study in which flexible cantilever in-line<br />
cylinder arrays have been experimentally tested using wind tunnel for a P/D ratio <strong>of</strong> 2.01 and<br />
3.56. The paper has two different configurations; smooth cylinders and finned cylinders. For<br />
the smooth cylinder configuration the study is based on three different models; (i) a single<br />
flexible tube amongst a set <strong>of</strong> fixed tubes, (ii) a whole array <strong>of</strong> flexible tubes and (iii) a bundle<br />
26
Chapter II Literature Review<br />
<strong>of</strong> flexible tubes. The paper presents data relating to root mean square tip amplitude at various<br />
pitch flow velocities, where the pitch flow velocity is defined in the same way as the gap<br />
velocity for a square tube bundle array. The study revolves around the fluid elastic instability<br />
which is defined as the excitation mechanism which causes the most violent vibrations leading<br />
to rapid tube failure. The flow velocity at which this failure occurs is termed as critical or<br />
threshold velocity. Price and Paidoussis [73] investigate a single flexible cylinder placed<br />
inside a rigid tube bundle configuration with a gap ratio <strong>of</strong> 1.5. Data for this case corresponds<br />
gap velocities and turbulence intensities.<br />
Feenstra et al. [74] experimentally study flow induced vibrations in a cantilevered tube bundle<br />
array with single and two phase cross-flow. For a single phase flow the study addresses two<br />
cases; a single flexible tube in an otherwise rigid tube bundle and a fully flexible tube bundle<br />
configuration. It was observed that for the single flexible tube configuration fluid elastic<br />
instability was achieved at 25% higher flow velocity and symmetric vortex shedding occurred<br />
at 50% higher flow velocities. The paper presents an excellent comparison with previous<br />
experimental studies conducted by Paidoussis [75] and Weaver & Fitzpatrick [76] at a gap<br />
ratio <strong>of</strong> 1.5 using water tunnel.<br />
27
Chapter III <strong>Turbulence</strong> modelling<br />
Chapitre 3<br />
<strong>Turbulence</strong> modelling<br />
3.1 Introduction<br />
<strong>Turbulence</strong> modeling is a key issue in most Computational Fluid Dynamics simulations.<br />
Virtually all engineering applications are turbulent and hence require a turbulence model.<br />
3.2 Navier-Stokes equations<br />
The Navier-Stokes equations are the basic governing equations for a viscous, heat<br />
conducting fluid. It is a vector equation obtained by applying Newton's Law <strong>of</strong> Motion to a<br />
fluid element and is also called the momentum equation. It is supplemented by the mass<br />
conservation equation, also called continuity equation and the energy equation. Usually, the<br />
term Navier-Stokes equations are used to refer to all <strong>of</strong> these equations.<br />
The Navier-Stokes equation is:<br />
3.2.1 Reynolds Averaging<br />
2<br />
∂ui<br />
∂ui<br />
1 ∂P<br />
∂ ui<br />
+ u j = − + v<br />
∂t<br />
∂u<br />
ρ ∂x<br />
∂x<br />
∂x<br />
j<br />
i<br />
j<br />
j<br />
(3.1)<br />
In Reynolds ensemble averaging the solution variables <strong>of</strong> the Navier-Stokes equations are<br />
decomposed into mean and fluctuating parts<br />
u = u + u'<br />
(3.2)<br />
i<br />
Where ui , ui and u'i<br />
are the instantaneous, mean and fluctuating components respectively. The<br />
scale quantities such as pressure are also decomposed the same principle.<br />
i<br />
i<br />
P = P + P'<br />
(3.3)<br />
Inserting the decomposition form equation (3.2) and (3.3) into equation (3.1) gives<br />
2<br />
( u + u'<br />
) ∂(<br />
u + u'<br />
) 1 ∂(<br />
P + P)<br />
∂ ( u + u )<br />
∂ i i<br />
i i<br />
i '<br />
+ u j = − + v<br />
∂t<br />
∂x<br />
ρ ∂x<br />
∂x<br />
∂x<br />
j<br />
i<br />
j<br />
j<br />
(3.4)<br />
Expanding equation (3.4), taking time averaging, ignoring mean <strong>of</strong> the fluctuating quantities<br />
and keeping mean <strong>of</strong> mean quantities gives<br />
2<br />
∂ui<br />
∂u<br />
u'<br />
i j ∂u'i<br />
1 ∂P<br />
∂ ui<br />
+ u j + = − +<br />
∂t<br />
∂x<br />
∂x<br />
ρ ∂x<br />
∂x<br />
∂x<br />
j<br />
j<br />
28<br />
i<br />
j<br />
j<br />
(3.5)
Chapter III <strong>Turbulence</strong> modelling<br />
The third term on the left hand side <strong>of</strong> equation (3.5) can now be further modified as<br />
u'<br />
j ∂u'i ∂u'<br />
j u'i<br />
u'i<br />
∂u'<br />
= −<br />
∂x<br />
∂x<br />
∂x<br />
j<br />
j<br />
j<br />
j<br />
(3.6)<br />
Where the 2nd term on the right hand side <strong>of</strong> equation (3.6) can be dropped out for an<br />
incompressible case. Thus the final Reynolds Averaged Navier-Stokes equations become<br />
2<br />
∂ui<br />
∂ui<br />
1 ∂P<br />
∂ u ∂u'<br />
i i u'<br />
j<br />
+ u j = − + v −<br />
∂t<br />
∂x<br />
ρ ∂x<br />
∂x<br />
∂x<br />
∂x<br />
j<br />
i<br />
j<br />
j<br />
j<br />
(3.7)<br />
Where the quantities with over-bar are mean ensemble averaged quantities. The last term in<br />
equation (3.7) is the Reynolds stresses which need to be modelled for closure <strong>of</strong> RANS<br />
equations.<br />
3.3 Classes <strong>of</strong> turbulence models<br />
3.3.1 Algebraic turbulence models<br />
Algebraic turbulence models or zero-equation turbulence models are models that do not<br />
require the solution <strong>of</strong> any additional equations, and are calculated directly from the flow<br />
variables. As a consequence, zero equation models may not be able to properly account for<br />
history effects on the turbulence, such as convection and diffusion <strong>of</strong> turbulence energy. These<br />
models are <strong>of</strong>ten too simple for use in general situations, but can be quite useful for simple<br />
flow geometries or in start-up situations. The two most well known zero equation models are:<br />
3.3.1.1 Baldwin-Lomax model<br />
Baldwin and Lomax [20] is a two-layer algebraic 0-equation model which gives the eddy<br />
viscosity, μ t as a function <strong>of</strong> the local boundary layer velocity pr<strong>of</strong>ile. The model is suitable<br />
for high speed flows with thin attached boundary-layers, typically present in aerospace and<br />
turbomachinery applications. It is commoly used in quick design iretation where robustness is<br />
more important than capturing all details <strong>of</strong> the flow physics. The Baldwin-Lomax model is<br />
not suitable for cases with large separated regions and significant curvature/rotation effects.<br />
For more information see also [21], [22], [23], [24].<br />
29
Chapter III <strong>Turbulence</strong> modelling<br />
3.3.1.2 Cebeci-Smith model:<br />
Smith and Cebeci [25] is a two-layer algebraic 0-equation model which gives the eddy<br />
viscosity, μ t as a function <strong>of</strong> the local boundary layer velocity pr<strong>of</strong>ile. The model is suitable<br />
for high-speed flows with thin attached boundary-layers, typically present in aerospace<br />
applications. L<strong>ik</strong>e the Baldwin-Lomax model, this model requires the determination <strong>of</strong> a<br />
boundary layer edge.<br />
Other even simpler models, such a models written as ( ) , are sometimes used in<br />
+<br />
μ = f y<br />
particular situations (e.g boundary layers or jets). See also [26]<br />
3.3.2 One-equation turbulence models<br />
One equation turbulence models solve one turbulent transport equation, usually the turbulent<br />
kinetic energy. The original one-equation model is Prandtl’s one-equation model [27], [28],<br />
[29].<br />
3.3.2.1 Prandtl’s one-equation model<br />
Kinematic eddy viscosity<br />
Equation <strong>of</strong> the model<br />
∂k<br />
+ U<br />
∂t<br />
j<br />
∂k<br />
− C<br />
∂x<br />
Closure coefficients and auxilary relations<br />
ε =<br />
CD<br />
σ k<br />
C D<br />
3<br />
k<br />
l<br />
2<br />
= 0.<br />
08<br />
= 1<br />
where<br />
2<br />
τ ij = 2vT Sij<br />
− kδij<br />
3<br />
j<br />
1<br />
vt D<br />
D<br />
t<br />
2<br />
k<br />
2 = k l = C<br />
(3.8)<br />
ε<br />
3<br />
k<br />
l<br />
2<br />
+<br />
30<br />
∂<br />
∂x<br />
j<br />
⎡⎛<br />
v ⎞ ⎤<br />
T ∂k<br />
⎢ ⎜<br />
⎜v<br />
+<br />
⎟ ⎥<br />
⎢⎣<br />
⎝ σ k ⎠ ∂x<br />
j ⎥⎦<br />
(3.9)
Chapter III <strong>Turbulence</strong> modelling<br />
Another popular one-equation model is the Spallart-Allmaras model [32], [33], [34], that uses<br />
a transport equation for the viscosity including eight closure coefficients and three damping<br />
functions, in a similar way to the Baldwin-Barth models [30], [31].<br />
3.3.3 Two equation turbulence models<br />
Two equation turbulence models are not <strong>of</strong> the most common type <strong>of</strong> turbulence models.<br />
Models l<strong>ik</strong>e the k-epsilon and the k-omega model have become industry standard models and<br />
are commonly used for most types <strong>of</strong> engineering problems. Two equation turbulence models<br />
are also very much still an active area <strong>of</strong> research and new refined two-equation models are<br />
still being developed.<br />
By definition, two equation models include two extra transport equations to represent the<br />
turbulent properties <strong>of</strong> the flow. This allows a two equation model to account for history<br />
effects l<strong>ik</strong>e convection and diffusion <strong>of</strong> turbulent energy.<br />
Most <strong>of</strong>ten one <strong>of</strong> the transported variables is the turbulent kinetic energy, k. The second<br />
transported variable varies depending on what type <strong>of</strong> two-equation model it is. Common<br />
choices are the turbulent dissipation, ε, or the specific dissipation, ω. The second variable can<br />
be thought <strong>of</strong> as the variable that determines the scale <strong>of</strong> the turbulent (length-scale <strong>of</strong> timescale).<br />
Whereas the first variable, k, determines the energy in the turbulence.<br />
3.3.3.1 Boussinesq eddy viscosity assumption<br />
The basis for all two equation models is the Boussinesq eddy viscosity assumption, which<br />
postulates that the Reynolds stress tensor, τ ij is proportional to the mean strain rate tensor,<br />
and can be written in the following way:<br />
τ<br />
2<br />
= 2 +<br />
(3.10)<br />
3<br />
ij μtS<br />
ij ρκδij<br />
Where μ t is a scalar property and is called the eddy viscosity which is normally computed<br />
from the two transported variables. The last term is included for modelling incompressible<br />
flow to ensure that the definition <strong>of</strong> turbulence kinetic energy is obeyed:<br />
u'i<br />
u'i<br />
κ =<br />
(3.11)<br />
2<br />
31<br />
Sij
Chapter III <strong>Turbulence</strong> modelling<br />
The same equation can be written more explicitly as:<br />
⎛ U U ⎞<br />
i j 2<br />
ρu'<br />
i u'<br />
j μ ⎜<br />
∂ ∂<br />
− = t + ⎟ + ρκδ<br />
⎜<br />
ij<br />
(3.12)<br />
x j x ⎟<br />
⎝ ∂ ∂ i ⎠ 3<br />
The boussinesq assumption is both the strength and the weakness <strong>of</strong> two equation models.<br />
This assumption is a huge simplification which allows one to think <strong>of</strong> the effect <strong>of</strong> turbulence<br />
on the mean flow in the same way as molecular viscosity effects a laminar flow. The<br />
assumption also makes it possible to introduce intiutive scalar turbulence variables l<strong>ik</strong>e the<br />
turbulent energy and dissipation and to relate these variables to even more intuitive variables<br />
l<strong>ik</strong>e turbulence intensity and turbulence length scale.<br />
The weakness <strong>of</strong> the Boussinesq assumption is that it is not in general valid. There is nothing<br />
which says that the Reynolds stress tensor must be proportional to the strain rate tensor, it is<br />
true in simple flows l<strong>ik</strong>e straight boundary layers and wakes, but in complex flows, l<strong>ik</strong>e flows<br />
with strong curvature, or strongly accelerated or decellerated flows the Boussinesq assumption<br />
is simply not valid. This gives two equation models inherent problems to predict strongly<br />
rotating flows and other flows where curvature effects are significant. Two equation models<br />
also <strong>of</strong>ten have problems to predict strongly decellerated flows l<strong>ik</strong>e stagnation flows.<br />
3.3.3.2 K-epsilon models<br />
The k-epsilon model is one <strong>of</strong> the most common turbulence models. It is a two equation<br />
model that means, it includes two extra transport equations to represent the turbulent<br />
properties <strong>of</strong> the flow. This alows a two equation model to account for history effects l<strong>ik</strong>e<br />
convection and diffusion <strong>of</strong> turbulent energy. The first transported variable is turbulent kinetic<br />
energy k. It is the variable that determines the energy in the turbulence. The second<br />
transported variable in this case is the turbulent dissipation. ε. It is the variable that determines<br />
the escale <strong>of</strong> the turbulence. The usual k-epsilons models are:<br />
• Standard k-epsilon model<br />
• Realisable k-epsilon model<br />
• RNG k-epsilon model<br />
The model used in the present work is standard k-epsilon model.<br />
32
Chapter III <strong>Turbulence</strong> modelling<br />
Transport equations for standard k-epsilon model<br />
For turbulent kinetic energy k<br />
∂<br />
∂t<br />
For dissipation ε<br />
∂<br />
∂t<br />
∂<br />
∂x<br />
∂ ∂ ⎡⎛<br />
μ ⎞ ∂k<br />
⎤<br />
ρ<br />
xi<br />
x ⎜<br />
+<br />
⎟<br />
(3.13)<br />
∂ ∂ j ⎢⎣<br />
⎝ σ k ⎠ ∂x<br />
j ⎥⎦<br />
t<br />
( k)<br />
+ ( ρkui<br />
) = ⎢⎜<br />
μ ⎟ ⎥ + Pk<br />
+ Pb<br />
− ρε − YM<br />
+ Sk<br />
∂<br />
∂x<br />
⎡⎛<br />
μ ⎞ ∂ε<br />
⎤<br />
⎜<br />
+<br />
⎢⎣<br />
σ ⎟<br />
⎝ ε ⎠ ∂x<br />
j ⎥⎦<br />
t<br />
( ρε ) + ( ρεui<br />
) = ⎢⎜<br />
μ ⎟ ⎥ + C ε ( Pk<br />
+ C3ε<br />
Pb<br />
) − C2ε<br />
ρ + Sε<br />
Modelling turbulent viscosity<br />
Production <strong>of</strong> k<br />
i<br />
j<br />
P<br />
k<br />
ε<br />
k<br />
2<br />
ε<br />
k<br />
1 (3.14)<br />
2<br />
k<br />
= ρC<br />
(3.15)<br />
ε<br />
μt μ<br />
= − '<br />
∂U<br />
j<br />
ρ u'i<br />
u j<br />
(3.16)<br />
∂xi<br />
Pk t<br />
= μ S<br />
Where S is the modulus <strong>of</strong> the mean rate-<strong>of</strong>-strain tensor, defined as:<br />
The constants have the values shown in Table (3.1)<br />
C 1ε<br />
2ε<br />
C μ<br />
2<br />
ij ij S<br />
(3.17)<br />
S S = 2<br />
(3.18)<br />
C σ k<br />
σ ε<br />
1.44 1.92 0.09 1.0 1.3<br />
3.3.3.3 K-omega models<br />
Table 3.1: Coefficients <strong>of</strong> the standard k-ε model.<br />
The k-omega model is one <strong>of</strong> the most common turbulence models. It is a two equation<br />
model that means, it includes two extra transport equations to represent the turbulent<br />
properties <strong>of</strong> the flow. This alows a two equation model to account for history effects l<strong>ik</strong>e<br />
convection and diffusion <strong>of</strong> turbulent energy. The first transported variable is turbulent kinetic<br />
energy k. It is the variable that determines the energy in the turbulence. The second<br />
33
Chapter III <strong>Turbulence</strong> modelling<br />
transported variable in this case is the turbulent dissipation, ω. It is the variable that<br />
determines th escale <strong>of</strong> the turbulence. The common used k-omega models are :<br />
• Wilcox’s k-omega model [35]<br />
• Wilcox’s modified k-omega model [36]<br />
• SST k-omega model [37], [38]<br />
The model used in the present work is SST k-omega model.<br />
SST k-omega model<br />
The SST k-ω turbulence model, Menter [37], is a two-equation eddy-viscosity model which<br />
has become very popular. The use <strong>of</strong> a k-ω formulation in the inner parts <strong>of</strong> the boundary<br />
layer makes the model directly usable all the way down to the wall through the viscous sub-<br />
layer, hense the SST k-ω model can be used as Low-Re turbulence model without any extra<br />
damping functions. The SST formulation also switches to a k-ε behaviour in the free-stream<br />
an avoid the common k-ω problem that the model is too sensitive to the inlet free-stream<br />
turbulence properties. Authors who use the SST k-ω model <strong>of</strong>ten merit it for its good<br />
behaviour in adverse pressure gradients and separating flow. The SST k-ω model does<br />
produce a bit too large turbulence levels in regions with large normal strain, l<strong>ik</strong>e stagnation<br />
regions with strong acceleration. This tendency is much less pronounced than with a normal kε<br />
model though.<br />
Kinematic eddy viscosity<br />
<strong>Turbulence</strong> kinetic energy<br />
∂k<br />
+ U<br />
∂t<br />
Specific dissipation rate<br />
j<br />
∂k<br />
∂x<br />
j<br />
v T<br />
= P<br />
k<br />
−<br />
a1k<br />
=<br />
max F<br />
β *<br />
( a ω,<br />
Ω )<br />
1<br />
∂<br />
kω<br />
+<br />
∂x<br />
j<br />
⎡<br />
⎢<br />
⎢⎣<br />
2<br />
( v + σ v )<br />
k<br />
T<br />
∂k<br />
⎤<br />
⎥<br />
∂x<br />
j ⎥⎦<br />
∂ω ∂ω<br />
⎡<br />
⎤<br />
2 2 ∂<br />
∂ω<br />
1 ∂k<br />
∂ω<br />
+ U j = αS<br />
− βω + ⎢(<br />
v + σ ωvT<br />
) ⎥ + 2( 1−<br />
F1<br />
) σ ω 2<br />
∂t<br />
∂x<br />
j<br />
∂x<br />
j ⎢⎣<br />
∂x<br />
j ⎥⎦<br />
ω ∂xi<br />
∂xi<br />
34<br />
(3.19)<br />
(3.20)<br />
(3.21)
Chapter III <strong>Turbulence</strong> modelling<br />
Closure coefficients and auxillay relation<br />
4<br />
⎧<br />
⎫<br />
⎪⎪⎧<br />
⎡ ⎛ k 500v<br />
⎞ 4σ<br />
⎤⎪⎫<br />
⎪<br />
⎨⎨<br />
⎢ ⎜ ⎟ ω 2k<br />
F1<br />
= tanh min max<br />
⎜<br />
,<br />
⎟<br />
,<br />
* 2<br />
2 ⎥⎬<br />
⎬<br />
⎪⎪⎩<br />
⎢<br />
⎥⎪⎭<br />
⎩ ⎣ ⎝ β ωy<br />
y ω ⎠ CDkω<br />
y<br />
⎦ ⎪<br />
⎭<br />
(3.22)<br />
⎛ 1 ∂k<br />
∂ω<br />
−10 ⎞<br />
CD =<br />
⎜<br />
⎟<br />
kω<br />
max 2ρσω<br />
2 , 10<br />
(3.23)<br />
⎝ ω ∂xi<br />
∂xi<br />
⎠<br />
2<br />
⎡⎡<br />
⎛<br />
⎤<br />
⎢<br />
2 k 500v<br />
⎞⎤<br />
F ⎢ ⎜ ⎟⎥<br />
⎥<br />
2 = tanh max<br />
⎢ ⎜<br />
, * 2 ⎟<br />
(3.24)<br />
⎥<br />
⎣<br />
⎢⎣<br />
⎝ β ωy<br />
y ω ⎠⎥⎦<br />
⎦<br />
⎛<br />
⎞<br />
⎜<br />
∂Ui<br />
*<br />
P<br />
⎟<br />
k = min τ β kω<br />
⎜ ij , 20<br />
(3.25)<br />
⎟<br />
⎝ ∂x<br />
j ⎠<br />
1F1 + φ2(<br />
1− F1<br />
φ = φ<br />
)<br />
(3.26)<br />
The constants have the values shown in Table 3.2<br />
α 1 α 2 β 1 β 2<br />
*<br />
β σ k1<br />
σ k 2 σ ω1<br />
σ ω 2<br />
5/9 0.44 3/40 0.0828 9/100 0.85 1 0.5 0.856<br />
3.3.4 V2-f models<br />
Table 3.2: Coefficients <strong>of</strong> the k-ω SST model<br />
2<br />
The υ − f model [39] is similar to the standard k-ε model. Additionally, it incorporates also<br />
some near-wall turbulence anisotropy as well as non-local pressure-strain. It is a general<br />
turbulence model for low Reynolds numbers. That does not need to make use <strong>of</strong> wall<br />
2<br />
functions because it is valid upto solid walls. Instead <strong>of</strong> turbulent kinetic energy k. the υ − f<br />
model uses a velocity scale<br />
2<br />
2<br />
υ for evaluation <strong>of</strong> the eddy viscosity. υ can be thougth <strong>of</strong> as<br />
the velocity fluctuation normal to the streamlines. It can provide the right scaling for the<br />
presentation <strong>of</strong> the damping <strong>of</strong> turbulent transport close to the wall. The anisotropic wall<br />
effects are modelled through the elliptic relaxation function ƒ, by solving a separate elliptic<br />
equation <strong>of</strong> the Helmholtz type. In order to improve the computational performances <strong>of</strong> the<br />
35
Chapter III <strong>Turbulence</strong> modelling<br />
2<br />
υ − f model, a variant <strong>of</strong> this eddy-viscosity model is derivied when the change <strong>of</strong> variables<br />
is introduced. Instead <strong>of</strong> using the wall-normal velocity fluctuation<br />
2<br />
υ as the velocity scale,<br />
2<br />
the normalised wall-normal velocity scale ζ = υ / k is used. This turbulence variable can be<br />
regarded as the ratio <strong>of</strong> the two time scales: scalar k/ε (isotropic), and lateral<br />
2<br />
υ /<br />
ε (anisotropic). Following the definition <strong>of</strong> ζ, the new transport equation is derivied from<br />
2<br />
2<br />
the equation for υ and k, and solved instead <strong>of</strong> the transport equation forυ<br />
. See also [40].<br />
3.2.5 Reynolds stress model (RSM)<br />
The Reynolds stress model (RSM) [41] is a higher level, elaborate turbulence model. It is<br />
usually called a second order closure. This modelling approach originates from the work by<br />
Launder. [41]. In RSM, the eddy viscosity approach has been discarded and the Reynolds<br />
stresses are directly computed. The exact Reynolds stress transport equation accounts for the<br />
directional effects <strong>of</strong> the Reynolds Stress fields.<br />
The Reynolds stress model involves calculation <strong>of</strong> the individual Reynolds stresses, ρ u'i u'<br />
j<br />
using differential transport equations. The individual Reynolds stresses are then used to obtain<br />
closure <strong>of</strong> the Reynolds-averaged momentum equation.<br />
The exact transport equations for the transport <strong>of</strong> the Reynolds stresses, u' i u'<br />
j may written as<br />
follows:<br />
Or<br />
∂<br />
∂t<br />
∂<br />
∂<br />
( ρu'<br />
u'<br />
) + ( ρu<br />
u'<br />
u'<br />
) = − [ ρu'<br />
u'<br />
u'<br />
+ P(<br />
δ u'<br />
+ δ u'<br />
) ]<br />
∂ ⎡ ∂<br />
+ ⎢μ<br />
∂xk<br />
⎣ ∂xk<br />
i<br />
j<br />
∂x<br />
⎤ ⎛ ∂u<br />
j ∂u<br />
⎞ i<br />
( u'<br />
u'<br />
) − ρ⎜u<br />
' u'<br />
+ u'<br />
u'<br />
⎟ − ρβ ( g u'<br />
θ + g u'<br />
θ )<br />
i<br />
k<br />
j<br />
⎥<br />
⎦<br />
⎛ u u ⎞<br />
i j u ∂u<br />
i j<br />
P⎜<br />
∂ ' ∂ '<br />
⎟<br />
∂ ' '<br />
+ + − 2μ<br />
− 2ρΩ<br />
⎜<br />
k<br />
x j x ⎟<br />
⎝ ∂ ∂ i ⎠ ∂xk<br />
∂xk<br />
k<br />
i<br />
⎜<br />
⎝<br />
j<br />
i<br />
k<br />
∂x<br />
∂x<br />
Local time derivative+ =<br />
ij<br />
k<br />
k<br />
j<br />
i<br />
k<br />
j<br />
k<br />
∂x<br />
k<br />
⎟<br />
⎠<br />
kj<br />
( u'<br />
j u'mε<br />
<strong>ik</strong>m + u'i<br />
u'mε<br />
jkm ) + Suser<br />
i<br />
i<br />
<strong>ik</strong><br />
j<br />
j<br />
j<br />
i<br />
+<br />
(3.27)<br />
C D T , ij + DL,<br />
ij + Pij<br />
+ Gij<br />
+ φij<br />
− ε ij + Fij<br />
(3.28)<br />
+ User defined source term.<br />
36
Chapter III <strong>Turbulence</strong> modelling<br />
Where is the convection term, equals the turbulency diffusion, stands for the<br />
Cij D T , ij<br />
D L,<br />
ij<br />
molecular diffusion, is the term for stress production, equals buoyancy production, ε<br />
Pij Gij ij<br />
stands for the dissipation and Fij<br />
is the production by system rotation.<br />
Dissipation<br />
The dissipation term can be written as the sum <strong>of</strong> the isotropic and deviatoric parts:<br />
2<br />
ε ij = εδij<br />
+ Dε ij<br />
3<br />
(3.29)<br />
In most models, the isotropic part is calculated via a transport equation and the deviatoric part<br />
is lumped into the pressure-strain correlation:<br />
φ −<br />
ij mod elled = φij<br />
Dε ij<br />
(3.30)<br />
The transport equation for the isotropic part <strong>of</strong> the dissipation rate can be derived from the<br />
fluctuating momentum equation [42]. The result is a transport equation with higher order<br />
terms that require further modelling. Although the inaccuracy <strong>of</strong> the modelling involved on<br />
the dissipation equation is an obvious deficiency, the resulting equation is widely used and is<br />
almost standard for all the models, including many two-equation models. The standard<br />
transport equation for the dissipation rate proposed by Hanjali´c and Launder [43] is:<br />
∂ε<br />
+ U<br />
∂t<br />
Where Pk is defined as:<br />
k<br />
∂ε<br />
= C<br />
∂x<br />
k<br />
Pkε<br />
− C<br />
k<br />
ε ∂ ⎛<br />
+ ⎜C<br />
∂x<br />
⎜<br />
⎝<br />
k<br />
∂ε<br />
⎞<br />
⎟<br />
∂x<br />
⎟<br />
ij ⎠<br />
ε1<br />
ε 2<br />
2<br />
k j<br />
ε u'i<br />
u'<br />
j<br />
ε<br />
(3.31)<br />
∂U<br />
P = −u<br />
'<br />
k<br />
i ' i u j<br />
(3.32)<br />
∂x<br />
j<br />
The coefficients , and C vary according to the pressure-strain closure but in general,<br />
Cε Cε1 ε 2<br />
Cε 2 is set to 1.9 to match the decay rate <strong>of</strong> isotropic turbulence; Cε<br />
is set between the values<br />
0.15 and 0.18 and C usually takes the value <strong>of</strong> 1.44 [42].<br />
Diffusion<br />
ε1<br />
The most popular way <strong>of</strong> representing the diffusive terms is the generalized gradient<br />
diffusion hypothesis, GGDH [54] which can be written as:<br />
37
Chapter III <strong>Turbulence</strong> modelling<br />
D<br />
T<br />
ij<br />
= −<br />
∂<br />
∂x<br />
k<br />
⎛<br />
⎞<br />
⎜ k ∂u'i<br />
u'<br />
j<br />
C<br />
⎟<br />
s u'k<br />
u'l<br />
⎜<br />
⎟<br />
⎝<br />
ε ∂xl<br />
⎠<br />
(3.33)<br />
Where Cs = 0.22 is a constant determined from model optimization [42]. Other models have<br />
been proposed and can be found in [43], [46] and [42]. It is important to note that some <strong>of</strong> the<br />
diffusion models do not take into account the pressure-diffusion terms. This might be seen as a<br />
source <strong>of</strong> inaccuracy and although it does not seem to be critical in the modelling; it has been<br />
called into question [45].<br />
The inclusion <strong>of</strong> the velocity-pressure correlation can be seen in the value <strong>of</strong> the empirical<br />
constant Cs. The model constant obtained without the inclusion <strong>of</strong> the pressure diffusion is<br />
about 20% greater than the usual 0.22.<br />
Pressure strain<br />
The pressure-strain correlation term is the term where most <strong>of</strong> the modelling effort has been<br />
made over the last thirty years. This term is important because it has the same order as the<br />
production term and it tends to redistribute the energy between the Reynolds stress<br />
components, diminishing the difference between them.<br />
Derived from the Navier-Stokes equations and the continuity equation, the pressure field<br />
satisfies the Poisson equation:<br />
and the fluctuating pressure satisfies:<br />
u ∂u<br />
2 ∂ i j<br />
∇ p = −ρ<br />
∂x<br />
∂x<br />
j<br />
i<br />
( u'<br />
u'<br />
−u'<br />
)<br />
i<br />
j<br />
j<br />
−<br />
∂x<br />
j<br />
2<br />
∂xi∂x<br />
j<br />
i j i u j<br />
(3.34)<br />
U u'<br />
2 ∂ ∂ ∂<br />
∇ p'= −2ρ<br />
ρ<br />
'<br />
(3.35)<br />
∂x<br />
rapid<br />
The first part is directly dependent on the mean velocity which makes it respond rapidly to the<br />
velocity changes. The second part is nonlinear and involves interaction between fluctuating<br />
velocities. This decomposition is also carried out into most <strong>of</strong> the commonly used Second<br />
Moment Closures, the pressure-strain term is divided into a slow and a rapid part and then a<br />
correction term is added such as the wall echo term:<br />
φ +<br />
slow<br />
ij = φij1<br />
+ φij<br />
2 φω<br />
(3.36)<br />
38
Chapter III <strong>Turbulence</strong> modelling<br />
One <strong>of</strong> the most popular models is the linear Launder, Reece and Rodi (LRR) [41] which<br />
models the pressure strain term as:<br />
( b Ω + b )<br />
⎛<br />
2 ⎞<br />
φ ij = −C1εbij<br />
+ C2kSij<br />
+ C3k⎜<br />
b<strong>ik</strong>S<br />
jk + bjkS<br />
<strong>ik</strong> − bmnSmnδij<br />
⎟ + C4k<br />
<strong>ik</strong> jk jkΩ<strong>ik</strong><br />
(3.37)<br />
⎝<br />
3 ⎠<br />
Where bij is the normalised anisotropy tensor, Sij is the strain tensor and Ω ij is the rotation<br />
rate tensor defined as:<br />
aij<br />
u'i<br />
u'<br />
j 1<br />
bij = = − δij<br />
(3.38)<br />
2k<br />
2k<br />
3<br />
S<br />
ij<br />
Ω<br />
ij<br />
1 ⎛ ⎞<br />
⎜ ∂U<br />
∂U<br />
i j<br />
= + ⎟<br />
2 ⎜ ⎟<br />
⎝<br />
∂x<br />
j ∂xi<br />
⎠<br />
1 ⎛ ⎞<br />
⎜ ∂U<br />
∂U<br />
i j<br />
= − ⎟<br />
2 ⎜ ⎟<br />
⎝<br />
∂x<br />
j ∂xi<br />
⎠<br />
The constants have the values shown in Table 3.3<br />
C 1<br />
2 C C 3<br />
4 C C ε1<br />
C ε 2<br />
3.0 0.8 1.75 1.31 1.44 1.90<br />
Table 3.3: Coefficients <strong>of</strong> the LRR model.<br />
(3.39)<br />
(3.40)<br />
Another popular model for the pressure-strain correlation term is the Speziale, Sarkar and<br />
Gatski model (SSG) [47] which has a quadratic behaviour included originally as a higher<br />
order correction to the slow part <strong>of</strong> the pressure-strain correlation [42]. The model for the<br />
pressure-strain term is:<br />
⎛ 1 ⎞<br />
φij<br />
= −C1εbij<br />
+ C'1<br />
ε⎜<br />
b<strong>ik</strong>bkj<br />
− bmnbnm<br />
⎟ + C2kS<br />
⎝ 3 ⎠<br />
⎛<br />
2 ⎞<br />
+ C3k⎜<br />
b<strong>ik</strong>S<br />
jk + bjkS<strong>ik</strong><br />
− bmnSmnδij<br />
⎟<br />
⎝<br />
3 ⎠<br />
+ C<br />
4)<br />
( b Ω + b Ω )<br />
<strong>ik</strong><br />
jk<br />
jk<br />
<strong>ik</strong><br />
39<br />
ij<br />
(3.41)
Chapter III <strong>Turbulence</strong> modelling<br />
The model uses the coefficients showed in Table 3.4<br />
C 1<br />
1 ' C 2 C C 3<br />
4 C C ε1<br />
C ε 2<br />
3.4+1.8 P/ε 4.2 ( ) 5 . 0<br />
0. 8 − 1.<br />
3 bijb<br />
1.25 0.4 1.44 1.83<br />
ij<br />
Table 3.4: Coefficients <strong>of</strong> the SSG model.<br />
1<br />
In the coefficient C1 , dependence on P/ ε (with P = Pii<br />
) is introduced to achieve the correct<br />
2<br />
asymptotic behaviour <strong>of</strong> the Taylor series expansion <strong>of</strong>φ ij .<br />
3.3.6 Large Eddy simulation (LES)<br />
Large eddy simulation (LES) [48] is a popular technique for simulating turbulent flows. An<br />
implication <strong>of</strong> Kolmogorov’s (1941) theory <strong>of</strong> self similarity is that the large eddies <strong>of</strong> the<br />
flow are dependant on the geometry while the smaller scales more universal. This feature<br />
allows one to explicitly solve for the large eddies in a calculation and implicitly account for<br />
the small eddies by using a subgrid-scale model (SGS model).<br />
Mathematically, one may think <strong>of</strong> separating the velocity field into a resolved and sub-grid<br />
part.<br />
The resolved part <strong>of</strong> the field represent the “large” eddies, while the sub grid part <strong>of</strong> the<br />
velocity represent the “small scales” whose effect on the resolved field is included through the<br />
subgrid-scale model. Formally, one may think <strong>of</strong> filtering as the convolution <strong>of</strong> a function<br />
with a filtering kernel G:<br />
resulting in<br />
( x)<br />
G(<br />
x ξ) u(<br />
x)<br />
dξ,<br />
u i ∫ −<br />
= (3.42)<br />
u = u + u '<br />
(3.43)<br />
i<br />
Where ui the resolvable scale is part and u' is the sub-grid scale part. However, most<br />
practical (and commercial) implementations <strong>of</strong> LES use the grid itself as the filter and perform<br />
no explicit filtering.<br />
The filtered equations are developed from the incompressible Navier-Stokes equations <strong>of</strong><br />
motion:<br />
i<br />
40<br />
i<br />
i
Chapter III <strong>Turbulence</strong> modelling<br />
∂u<br />
⎛ ⎞<br />
i ∂ui<br />
1 ∂P<br />
∂<br />
⎜<br />
∂ui<br />
+ u = − + ⎟<br />
j<br />
v<br />
∂t<br />
∂x<br />
∂ ∂ ⎜ ⎟<br />
j ρ xi<br />
x j ⎝ ∂x<br />
j ⎠<br />
(3.44)<br />
Substituting in the decomposition ui = ui<br />
+ u'i<br />
and Pi = Pi<br />
+ P'i<br />
then filtering the resulting<br />
gives the equation <strong>of</strong> motion for the resolved field:<br />
∂ui<br />
∂ui<br />
1 ∂P<br />
∂ ⎛ u ⎞ ∂<br />
i<br />
ij<br />
u<br />
⎜<br />
∂<br />
j<br />
v ⎟<br />
1 τ<br />
+ = − + +<br />
∂t<br />
∂x<br />
j ρ ∂xi<br />
∂x<br />
⎜<br />
j x ⎟<br />
⎝ ∂ j ⎠ ρ ∂x<br />
j<br />
(3.45)<br />
We have assumed that the filtering operation commute, which is not generally the case. It is<br />
thought that the errors associated with this assumption are usually small, though that commute<br />
with differentiation has been developed. The extra term<br />
advection terms, due to the fact that<br />
And hence<br />
j<br />
∂τ ij<br />
arises from the non-linear<br />
∂x<br />
∂ui<br />
∂ui<br />
ui<br />
≠ u j<br />
∂x<br />
∂x<br />
Similar equation can be derived for the subgrid-scale field.<br />
ij<br />
i<br />
j<br />
i<br />
j<br />
j<br />
j<br />
(3.46)<br />
τ = u u − u u<br />
(3.47)<br />
Subgrid-scale turbulence models usually employ the Boussinesq hypothesis, and seek to<br />
calculate the SGS stress using:<br />
where<br />
S ij is the rate <strong>of</strong> strain tensor for the resolved scale defined by:<br />
1<br />
τ ij − τ kkδ<br />
ij = −2μt<br />
Sij<br />
(3.48)<br />
3<br />
S<br />
ij<br />
1 ⎛ ⎞<br />
⎜ ∂u<br />
∂u<br />
i j<br />
= + ⎟<br />
2 ⎜ ⎟<br />
⎝<br />
∂x<br />
j ∂xi<br />
⎠<br />
(3.49)<br />
And v is the subgrid-scale turbulent viscosity. Substituting into the filtered Navier-Stokes<br />
t<br />
equations, we then have:<br />
∂u<br />
1 ⎛<br />
i ∂ui<br />
∂P<br />
∂<br />
+ u = − + ⎜<br />
j<br />
∂t<br />
∂x<br />
∂ ∂ ⎜<br />
j ρ xi<br />
x j ⎝<br />
41<br />
∂u<br />
⎞<br />
∂x<br />
⎟<br />
j ⎠<br />
i [ v + v ] ⎟,<br />
t<br />
(3.50)
Chapter III <strong>Turbulence</strong> modelling<br />
Where we have used the incompressibility constraint to simplify the equation and the pressure<br />
is now modified to include the trace term τ / 3<br />
Subgrid scale models<br />
kkδ<br />
ij<br />
• Smagorinsky model, Smagorinsky. [51]<br />
• Algebraic Dynamic model, Germano et al. [48]<br />
• Localized Dynamic model, Kim & Menon. [49]<br />
• WALE (Wall-Adapting Local Eddy-viscosity) model, Nicoud and Ducros. [50]<br />
• RNG-LES model<br />
3.3.7 Detached Eddy simulation (DES)<br />
The difficulties associated with the use <strong>of</strong> the standard LES models, particularly in near-wall<br />
regions, has lead to the development <strong>of</strong> hybrid models that attempt to combine the best aspects<br />
<strong>of</strong> RANS and LES methodologies in a single solution strategy. An example <strong>of</strong> a hybrid<br />
technique is the detached-eddy simulation (DES), Spalart et al. [52] approach. This model<br />
attempts to treat near-wall regions in a RANS-l<strong>ik</strong>e manner, and treat the rest <strong>of</strong> the flow in an<br />
LES-l<strong>ik</strong>e manner. The model was originally formulated by replacing the distance function d in<br />
the Spallart-Allmaras (S-A) model with a modified distance function d min[ d,<br />
C Δ],<br />
= DES<br />
where C is a constant and Δ is the largest dimension <strong>of</strong> the grid cell in equation. This<br />
DES<br />
modification <strong>of</strong> the S-A model, while very simple in nature, changes the interpretation <strong>of</strong> the<br />
model substantially. This modified distance function causes the model to behave as a RANS<br />
model in regions close to walls, and in a Smagorinsky-l<strong>ik</strong>e manner away from walls the<br />
model. This is usually justified with arguments that the scale-dependence <strong>of</strong> the model is<br />
made local rather than global, and that dimensional analysis backs up this claim.<br />
The DES approach may be used with any turbulence model that has an appropriately defined<br />
turbulence length scale (distance in S-A model) and is a sufficiently localized model. The<br />
Baldwin-Barth model. While very similar to the S-A model is probably not a candidate for use<br />
with DES. The standard version <strong>of</strong> this model contains several van Driest-types damping<br />
functions that make the distance function more global in nature. Menter’s SST model [37] is a<br />
good candidate, and has been used by a number <strong>of</strong> researchers. Menter’s SST model uses a<br />
turbulence length scale obtained from the model’s equations and compares it with the grid<br />
length scale to switch between LES and RANS, Streets. [53]<br />
42
Chapter III <strong>Turbulence</strong> modelling<br />
In practical, more programming is needed than simply changing the calculation <strong>of</strong> the length<br />
scale. Many implementations <strong>of</strong> the DES approach allow for regions to be explicitly<br />
designated as RANS or LES regions, overruling the distance function calculation. Also, many<br />
implementations use differencing in RANS regions (e.g. upwind differences) and LES regions<br />
(e.g. central differences).<br />
3.3.8 Direct numerical simulation<br />
A direct numerical simulation (DNS) is a simulation in Computational Fluid Dynamic in<br />
which the Navier-Stokes equations are numerically solved without any turbulence model. This<br />
means that the whole range <strong>of</strong> spatial and temporal scales <strong>of</strong> the turbulence must be resolved.<br />
All the spatial scales <strong>of</strong> the turbulence must be resolved in the computational mesh, from the<br />
smallest dissipatives scales (kolmogorov scales); up to the integral scale L, associated with the<br />
motions containing most <strong>of</strong> the kinetic energy.<br />
3.3.9 The SST- Cas<br />
model<br />
The SST- C [18] model is used in the present work. The SST model described in section<br />
as<br />
(3.2.3.3) requires only small modifications to incorporate the Cas model [18].<br />
The Cas model<br />
In contrast to eddy viscosity models, the Reynolds stress model calculates an exact<br />
production which is explicitly linear in the mean strain rate. Indeed, one can write exactly<br />
P as<br />
k = C k S where C a non-dimensional parameter is representing the degree <strong>of</strong> alignment<br />
between stresses and strains:<br />
as<br />
C<br />
as<br />
aijSij<br />
= −<br />
(3.51)<br />
S<br />
The aim <strong>of</strong> the present work is to test the performance <strong>of</strong> the SST-Cas model to reproduce an<br />
unsteady flow across in-line tube bundle.<br />
Model derivation<br />
From the definition <strong>of</strong> C in equation (3.51) its total derivative can be obtained using the<br />
as<br />
product rule ( Dφ / Dt = ∂φ<br />
/ ∂t<br />
+ U ∂φ<br />
/ ∂x<br />
). After derivation:<br />
k<br />
k<br />
43
Chapter III <strong>Turbulence</strong> modelling<br />
DC<br />
Dt<br />
+<br />
as<br />
* ( α + α A S )<br />
3<br />
∂<br />
+<br />
∂x<br />
k<br />
ε<br />
= α1<br />
C<br />
k<br />
Sija<strong>ik</strong>Ω<br />
+ α5<br />
S<br />
⎡<br />
⎢<br />
⎣<br />
3<br />
jk<br />
2<br />
as<br />
−<br />
*<br />
+ α S C<br />
1<br />
S<br />
∂C<br />
DS<br />
Dt<br />
⎤<br />
as<br />
( v + σ casvt<br />
) ⎥<br />
∂xk<br />
⎦<br />
1<br />
ij<br />
2<br />
as<br />
Sija<strong>ik</strong>a<br />
+ α4<br />
S<br />
⎛<br />
⎜a<br />
⎜<br />
⎝<br />
Sija<strong>ik</strong>a<br />
−α<br />
2<br />
η<br />
ij<br />
kj<br />
2SijC<br />
+<br />
S<br />
as<br />
kj<br />
⎞<br />
⎟<br />
⎠<br />
(3.52)<br />
The constants α1… α5 are related to those in the original underlying pressure-strain model via:<br />
*<br />
*<br />
α = ( + ) = ( − )<br />
1 1 C<br />
*<br />
3<br />
1<br />
*<br />
C3<br />
=<br />
2<br />
C = α<br />
α 1 1 C1<br />
2 2<br />
⎛ 4 ⎞<br />
⎜ − C3<br />
⎟<br />
⎝ 3<br />
α<br />
⎠<br />
3 =<br />
2<br />
α α = 2( 1−<br />
C ) α = 2( 1−<br />
C )<br />
4<br />
Tables 3.3 and 3.4 lists the values <strong>of</strong> the model constants from the LRR and SSG pressure-<br />
strain models (C1… C5)<br />
The SST model requires small modifications to incorporate theC model. Initially, the<br />
modification was intended to be applied to the production rate <strong>of</strong> turbulence kinetic energy<br />
term only, but it can be applied in a more coherent manner by means <strong>of</strong> a simple modification<br />
to the turbulent eddy viscosity in equation 3.12, as follows:<br />
⎛<br />
⎞<br />
⎜<br />
1 a1<br />
Cas<br />
v =<br />
⎟<br />
t k min ; ;<br />
⎜<br />
⎟<br />
⎝ ω S F2<br />
S ⎠<br />
4<br />
5<br />
5<br />
as<br />
(3.53)<br />
The value <strong>of</strong> C in equation (3.53) is limited to ±0.31 for the calculation <strong>of</strong> the production<br />
as<br />
terms in equations 3.20 and 3.21, while when evaluating diffusion terms, the absolute value,<br />
|Cas|, is used to avoid negative values which could lead to numerical difficulties.<br />
3.4 <strong>Turbulence</strong> modelling <strong>of</strong> unsteady flows (URANS)<br />
3.4.1 Introduction<br />
An alternative to LES for industrial flows can then be unsteady RANS (Reynolds-Averaged-<br />
Navier-Stokes), <strong>of</strong>ten denoted URANS (unsteady RANS) or TRANS (Transient RANS).<br />
44
Chapter III <strong>Turbulence</strong> modelling<br />
3.4.2 Unsteady Reynolds Navier-Stokes equations<br />
In URANS the usual decomposition is employed,<br />
1<br />
U =<br />
2T<br />
T<br />
∫<br />
−T<br />
U<br />
() t<br />
dt , U = U + u"<br />
(3.54)<br />
The URANS equations are the usual RANS equations, but with the transient (unsteady) term<br />
retained (on incompressible flow)<br />
∂ U<br />
∂t<br />
i<br />
∂<br />
+<br />
∂x<br />
j<br />
( U U )<br />
i<br />
j<br />
2<br />
1 ∂P<br />
∂ Ui<br />
= − + v<br />
ρ ∂x<br />
∂x<br />
∂x<br />
∂U<br />
∂x<br />
i<br />
i<br />
i<br />
= 0<br />
j<br />
j<br />
∂u"<br />
i u"<br />
−<br />
∂x<br />
j<br />
j<br />
(3.55)<br />
Note that the dependant variables are now not only a function <strong>of</strong> the space coordinates, but<br />
also a function <strong>of</strong> time,<br />
= U ( x,<br />
y,<br />
z,<br />
t)<br />
, P = P(<br />
x,<br />
y,<br />
z,<br />
t)<br />
, " u"<br />
= u"<br />
u"<br />
( x,<br />
y,<br />
z,<br />
t)<br />
Ui i<br />
u i j i j<br />
The results from URANS [55] can be decomposed as a time averaged part U , a resolved<br />
fluctuation u', and the modelled turbulent fluctuation u".<br />
U = U + u"<br />
= U + u'+<br />
u"<br />
(3.56)<br />
3.4.2 <strong>Turbulence</strong> Modelling <strong>of</strong> Unsteady Cross flow In-line Tube Bundle<br />
In URANS, part <strong>of</strong> turbulence is modelled (u") and part <strong>of</strong> the turbulence is resolved (u'). If<br />
we want to compare computed turbulence with experimental turbulence, we must add these<br />
two parts together.<br />
All RANS models (High Reynolds number k-ε model, k-ω SST model and the new SST-Cas<br />
model, Reynolds Stress model (RSM)) are used in this case to compute an unsteady flow. The<br />
aim <strong>of</strong> the present work is to assess which <strong>of</strong> these models is able to reproduce the cross flow<br />
in-line tube bundle which is an unsteady flow.<br />
45
Chapter IV Numerical simulation<br />
Chapter 4<br />
Numerical Simulation<br />
4.1 Introduction<br />
At Electricite De France (EDF) development <strong>of</strong> in-house codes has been a resolute strategic<br />
choice for more than fifteen years. In order to solve the Navier-Stokes equations a Finite-<br />
Volume code is used. Code -Saturne, general- purpose Computational Fluid Dynamic Code<br />
for laminar and turbulence flows in complex two and three dimensional geometries. The code<br />
is used for industrial applications and research activities in several fields related to energy<br />
production (nuclear thermal-hydraulics, gas and coal combustion, turbomachinery, heating,<br />
ventilation and air conditioning...).<br />
<strong>CFD</strong> codes are structured around the numerical algorithms that can tackle fluid flow problems.<br />
Hence all codes contain three main elements: the first is a pre-processor, the second is a solver<br />
and the third is a post-processor.<br />
4.1.1 Pre-processor<br />
Pre-processing consists <strong>of</strong> the input <strong>of</strong> a flow problem to a <strong>CFD</strong> program by means <strong>of</strong> an<br />
operator-friendly interface and the subsequent transformation <strong>of</strong> this input into a form suitable<br />
for use by the solver. In the present work the pre-processor is GAMBIT and the activities at<br />
the pre-processing stage involve:<br />
� Definition <strong>of</strong> the geometry <strong>of</strong> the region <strong>of</strong> interest: the computational domain.<br />
� Grid generation the sub division <strong>of</strong> the domain into a number <strong>of</strong> smaller, nonoverlapping<br />
sub-domain: a grid (or mesh) <strong>of</strong> cells for control volumes or elements).<br />
� Selection <strong>of</strong> the physical and chemical phenomena that need to be modelled.<br />
� Definition <strong>of</strong> fluid properties<br />
� Specification <strong>of</strong> appropriate boundary conditions at cells which coincide with or<br />
touch the domain boundary.<br />
4.1.2 Solver (Code- Saturne)<br />
The numerical methods that form the basis <strong>of</strong> the solver Code-Saturne perform the following<br />
steps:<br />
46
Chapter IV Numerical simulation<br />
� Approximation <strong>of</strong> the unknown flow variables by means <strong>of</strong> simple functions.<br />
� Discritisation by substitution <strong>of</strong> the approximations into the governing flow equations<br />
and subsequent mathematical manipulations.<br />
� Solution <strong>of</strong> the algebraic equations.<br />
Code-Saturne is well suited for two- and three-dimensional calculations <strong>of</strong> steady or transient<br />
single-phase, incompressible, laminar or turbulent flows. It supports two Reynolds Averaged<br />
Navier Stokes (RANS) models: the standard k-ε model, the launder Sharma model and a<br />
Second Moment Closure (LRR) model. It also contains the LES Smagorinsky and dynamic<br />
models. The new models which are also implemented in Code -Saturne are: The SST model,<br />
elliptic relaxation models υ²-f, the SSG model and the scalable wall function. The flow solver<br />
is based on a finite volume approach, with a fully collocated arrangement for all variables.<br />
The time discretisation is similar to the method used in other commercial codes. It is based on<br />
a predictor-corrector scheme for the Navier-Stokes equations. An important asset <strong>of</strong> Code-<br />
Saturne is relies on its ability to deal with any kind <strong>of</strong> mesh (hybrid, containing arbitrary<br />
interfaces and any type <strong>of</strong> cell).<br />
4.1.3 Post-processor<br />
As in pre-processing a huge amount <strong>of</strong> development work has recently taken place in the<br />
post-processing field. Owing to the increased popularity <strong>of</strong> engineering workstations, many <strong>of</strong><br />
which have outstanding graphics capability, the leading <strong>CFD</strong> packages are now equipped with<br />
versatile data visualisation tools. There include:<br />
� Domain geometry and grid display<br />
� Vector plots<br />
� Line and shaded contour plots<br />
� Particle traching<br />
� View manipulation (translation, rotation, scaling, etc.)<br />
� Color postscript output<br />
47
Chapter IV Numerical simulation<br />
usclim.F<br />
(Boundary conditions<br />
Symmetry and walls)<br />
usini1.F<br />
(Initialization,<br />
time step,<br />
averaging…etc)<br />
Pre-processing<br />
Grid Generation<br />
(Structured mesh)<br />
Processing<br />
Solver<br />
Code-Saturne<br />
Principal subroutines <strong>of</strong><br />
the present test case<br />
Post-processing<br />
48<br />
usproj.F<br />
(Present case)<br />
Pressure and<br />
Velocity pr<strong>of</strong>iles<br />
(Contour <strong>of</strong> pressure,<br />
velocity…, stream-lines,<br />
iso-surfaces…)<br />
ustsns.F<br />
Correction <strong>of</strong><br />
the flow<br />
(periodicity)<br />
Figure 4.1: Steps <strong>of</strong> Numerical Simulation <strong>of</strong> cross flow in-line tube bundles
Chapter IV Numerical simulation<br />
4.2 The Finite Volume method<br />
The method relies on:<br />
� Dividing the domain into control volumes.<br />
� Formal integrations <strong>of</strong> the governing equations <strong>of</strong> fluid flow over all the control<br />
volumes <strong>of</strong> the solution domain.<br />
� Discretisation involves the substitution <strong>of</strong> a variety <strong>of</strong> finite-differences-type<br />
approximations for the terms in the integrated equation representing flow processes<br />
such us convection, diffusion and sources. The converts the integral equations into a<br />
system <strong>of</strong> algebraic equations.<br />
� Solution <strong>of</strong> the algebric equations by an iterative method.<br />
The conservation <strong>of</strong> a general flow variable Φ. For example a velocity component, within a<br />
finite control volume can be expressed as a balance between the various processes tending to<br />
increase or decrease it:<br />
Rate <strong>of</strong> change <strong>of</strong><br />
Φ in the control<br />
volume with respect to<br />
time<br />
=<br />
Net flux <strong>of</strong> Φ due<br />
to convection into<br />
the control volume<br />
49<br />
+<br />
+<br />
Net rate <strong>of</strong> creation<br />
<strong>of</strong> Φ inside the<br />
control volume<br />
Net flux <strong>of</strong> Φ due<br />
to diffusion into the<br />
control volume<br />
In this chapter, (I, J) refer to the cell on either side <strong>of</strong> a face (see Figure 4.2) while the lower<br />
case letters reference the usual tensor quantities<br />
∫<br />
V<br />
∂ρφ<br />
dV<br />
∂t<br />
+<br />
∫<br />
V<br />
∂<br />
∂x<br />
j<br />
∂ ⎛ ⎞<br />
⎜<br />
∂φ<br />
( ρu<br />
⎟<br />
jφ<br />
) dV = ∫ Γ dV +<br />
∂ ⎜ ⎟ ∫ SdV (4.1)<br />
x j ⎝ ∂x<br />
V<br />
j ⎠ V<br />
Where “S” represents the source term, “V” the volume <strong>of</strong> the cell and “Г” the diffusion<br />
coefficient. Using Gauss theorem the volume integrals <strong>of</strong> the divergence can be transformed<br />
into surface integrals, which can be written as:
Chapter IV Numerical simulation<br />
∫<br />
V<br />
∂ρφ ∂φ<br />
+ ∫ ρφu<br />
j n j dA = ∫ Γ n j dA +<br />
∂<br />
∂x<br />
t A<br />
A j<br />
V<br />
n<br />
50<br />
∫<br />
SdV<br />
(4.2)<br />
Where “A” is the area <strong>of</strong> the face and “ ” is the face normal vector. The volume integrals<br />
j<br />
are approximated by the product <strong>of</strong> the value at the cell centre and the cell volume V.<br />
Face integrals<br />
∫<br />
V<br />
SdV I<br />
≅ S V<br />
S<br />
I<br />
(4.3)<br />
The surface integrals can be approximated by the mid point rule, the product <strong>of</strong> the face<br />
centre value and the area <strong>of</strong> the face. In the collocated arrangement, an interpolation is<br />
required in order to obtain the values at the face centres, since all the variables are stored at<br />
the cell centres. In Code-Saturne the convection can be calculated either by using an upwind<br />
differencing scheme (UDS) or a central differencing scheme (CDS). The code has also a slope<br />
test based on the product <strong>of</strong> the gradients at the cell centres to dynamically switch from CDS<br />
to UDS. Using the upwind scheme, the value at the face can be obtained as:<br />
φ = φ if = 0 U<br />
F<br />
F<br />
I<br />
j<br />
i i F n<br />
φ = φ if = 0 U<br />
i i F n<br />
(4.4)<br />
This scheme is robust and stable but introduces additional numerical diffusion which can<br />
become large if the grid is coarse. For a centred scheme, the value at the face can be computed<br />
as:<br />
With<br />
φ<br />
F<br />
αφ<br />
⎛<br />
⎞<br />
⎜ 1 ⎡ ∂φ<br />
∂φ<br />
⎤<br />
+ ( 1 − α ) φ<br />
⎟<br />
j +<br />
⎜<br />
⎢ I + j ⎥OF<br />
⎟<br />
(4.5)<br />
⎝<br />
2 ⎢⎣<br />
∂x<br />
j ∂x<br />
j ⎥⎦<br />
⎠<br />
= I<br />
j<br />
FJ '<br />
a = . The last term in equation (4.5) is added for non-orthogonal grids, where the<br />
I'<br />
J '<br />
centre <strong>of</strong> the face does not lie in the mid point between the cell centres. The diffusion integral<br />
can be computed as:<br />
∫<br />
A<br />
∂φ<br />
n<br />
∂x<br />
j<br />
j<br />
dA<br />
=<br />
∑<br />
Neigh<br />
∂φ<br />
Γ n<br />
∂x<br />
j<br />
j<br />
A<br />
(4.6)
Chapter IV Numerical simulation<br />
With a linear approximation for the gradient the face centre, the diffusion is computed as:<br />
The values <strong>of</strong> φJ ' and I '<br />
Gradient Reconstruction<br />
∂φ<br />
φ J ' − φ I '<br />
∑ Γ n j A = ∑ Γ n j A<br />
(4.7)<br />
∂x<br />
IJ<br />
Nei<br />
j<br />
51<br />
Neigh<br />
φ can be computed by using the gradient at the cell centre:<br />
j<br />
j<br />
∂φ<br />
φ I ' = φ I + I I ' I j<br />
(4.8)<br />
∂x<br />
The calculation <strong>of</strong> the gradients is achieved by an iterative solver in which the gradient is<br />
expressed as:<br />
∂φ<br />
1<br />
I = ∑ n jdA<br />
∂x<br />
j V ∫ φ<br />
Neigh A<br />
The surface integral can be approximated using the midpoint rule so that is becomes:<br />
∂φ<br />
1<br />
I =<br />
∂x<br />
V<br />
j<br />
∑<br />
Neigh<br />
φ A<br />
to obtain the value <strong>of</strong> φF a Taylor series expansion can be applied to obtain:<br />
∂φ<br />
∂x<br />
j<br />
O<br />
F<br />
F<br />
n<br />
j<br />
(4.9)<br />
(4.10)<br />
∂φ<br />
φ F = φ 0 + OF j 0 + O<br />
(4.11)<br />
∂x<br />
The value <strong>of</strong> φ0 can be obtained by a linear interpolation and the gradient at the same<br />
∂φ<br />
point, from an averaged between the values <strong>of</strong> I<br />
∂x<br />
solve can be written as:<br />
∂φ<br />
∂x<br />
j<br />
I<br />
1<br />
=<br />
V<br />
4.3 Time discretisation<br />
∑<br />
Neigh<br />
1<br />
j<br />
j<br />
⎛ ∂φ<br />
∂φ<br />
and J<br />
∂x<br />
{ αφ + ( 1−<br />
α)<br />
φ j + OF ⎜<br />
j I + J } AF<br />
nj<br />
j<br />
. Finally the system to<br />
1 ⎜<br />
(4.12)<br />
2 ∂x<br />
j ∂x<br />
j<br />
The time discretisation in Code-Saturne is achieved through a fractional step scheme (Euler<br />
implicit) that can be associated the SIMPLEC method. The solution algorithm consists a<br />
prediction -correction method. In the first step the momentum equation is solved using an<br />
⎝<br />
∂φ<br />
⎟ ⎞<br />
⎠
Chapter IV Numerical simulation<br />
n n<br />
explicit pressure gradient from the previous time step. With u = ρu<br />
being the momentum<br />
at time step n, the system to solve at the first step <strong>of</strong> the method is:<br />
Q<br />
−<br />
Q<br />
∂<br />
⎛<br />
* n<br />
*<br />
n<br />
i i<br />
* n<br />
i<br />
*<br />
+ ⎜ u i Q j − μ ⎟ = − + S<br />
t x ⎜<br />
j<br />
x ⎟<br />
(4.13)<br />
Δ ∂<br />
∂ j ∂x<br />
i<br />
⎝<br />
Where S includes all the source terms that can be made implicit or explicit,<br />
n n<br />
S = A + B u and Δ t is the time step. After this prediction step a new velocity field is<br />
i<br />
i<br />
i<br />
obtained (denoted by (*)) which is usually not divergence free. The second step consists <strong>of</strong><br />
calculating the pressure gradient in order to satisfy the continuity equation. By taking the<br />
divergence <strong>of</strong> the momentum equation, the Poisson equation for the pressure can be written<br />
as:<br />
52<br />
∂u<br />
* * * ( P − P )<br />
∂ ⎛ ∂ ⎞<br />
⎜<br />
∂<br />
Δt<br />
⎟ =<br />
∂x<br />
⎜<br />
j x ⎟<br />
⎝ ∂ j ⎠ ∂x<br />
⎞<br />
⎠<br />
i<br />
∂P<br />
j<br />
i<br />
*<br />
φ i<br />
(4.14)<br />
Finally, once the updated pressure (P**) has been obtained, the velocity field is corrected.<br />
This is done by neglecting convection and diffusion variations:<br />
Q<br />
* *<br />
i<br />
* * * ( P − P )<br />
* ∂<br />
− Qi<br />
= −Δt<br />
(4.15)<br />
∂x<br />
n 1 * *<br />
The velocities and the pressure are updated, that is Q = Q and<br />
i<br />
+ n + 1 * *<br />
P = P<br />
. When a<br />
turbulence model is used, the resolution <strong>of</strong> the turbulent variables takes place after the<br />
velocities are computed. The dependence <strong>of</strong> other variables is fully explicit so that each<br />
equation is solved separately.
Chapter IV Numerical simulation<br />
4.4 Boundary conditions<br />
Figure 4.2: Notations for the spatial discritisation.<br />
For the resolution <strong>of</strong> discritised equations described previously, the boundary conditions have<br />
to be prescribed for any variable Φ.<br />
4.4.1 Inlet<br />
At the inlet a Dirichlet condition is prescribed for all transport variables (velocity, scalars,<br />
n+<br />
1<br />
turbulent variables…) so the values for and Q are prescribed by the user. A<br />
homogenuous Neumann condition (zero flux) is imposed on the pressure. The convection term<br />
Q<br />
n<br />
j n j<br />
ϕ is calculated directly from the prescribed values. For the diffusion terms, the boundary<br />
value is calculated as:<br />
φ<br />
n + 1<br />
inlet<br />
53<br />
inlet
Chapter IV Numerical simulation<br />
⎛ φ ⎞<br />
⎜<br />
∂<br />
φ −<br />
Γ n ⎟<br />
j = Γ<br />
⎜ x ⎟<br />
⎝ ∂ j ⎠ I ' Fjn<br />
inlet<br />
For the source terms, the prescribed value φ n+1<br />
inlet<br />
54<br />
n+<br />
1 *<br />
inlet φI<br />
'<br />
j<br />
(4.16)<br />
is used as a boundary value. The pressure<br />
gradient normal to the face is prescribed as zero although as extrapolation from the previously<br />
obtained results is possible. The term<br />
⎛<br />
⎜<br />
⎝<br />
∂ P ⎞<br />
t n ⎟<br />
∂x<br />
⎟<br />
j ⎠<br />
δ<br />
is set to zero in the Poisson equation and<br />
Δ j<br />
the term on the right hand side is calculated as:<br />
4.4.2 Outlet<br />
Q<br />
* n+<br />
1 n+<br />
1<br />
jn j = ρinletu<br />
j(<br />
inlet)<br />
n j<br />
(4.17)<br />
For the outlet, a homogeneous Neumann condition is imposed on the velocity, scalars and<br />
n+<br />
1<br />
turbulent variables. For the pressure, a Dirichlet condition is used on P . The boundary<br />
values for the diffusion terms set to zero, and for the source terms a first order approximation<br />
is used to setφ = φ I ' . The Dirichlet condition for pressure provides the boundary value used<br />
for the computation <strong>of</strong> the pressure gradient. The Dirichlet condition is also used in δP in the<br />
Poisson equation. In the momentum correction equation, the boundary value forδ P is set to<br />
zero.<br />
4.4.3 Walls and symmetries<br />
For the walls and symmetry faces, a zero mass flux is imposed and both Dirichlet and<br />
Neumann conditions can be applied to scalars. For the tangential velocity, a homogeneous<br />
Dirichlet boundary condition is used at the wall whereas at the symmetry faces a<br />
homogeneous Neumann condition is applied. The pressure gradient normal to the face is set to<br />
zero although it can also be computed via an explicit extrapolation <strong>of</strong> the value at the<br />
boundary cell.<br />
For convections terms, the boundary value <strong>of</strong> the flux is set to zero. For diffusion terms when<br />
a Neumann condition is applied to the variable Φ; the value prescribed is used directly. If a<br />
Dirichlet condition is applied, the boundary value is calculated in the same way as in the inlet.<br />
outlet
Chapter IV Numerical simulation<br />
If the variable has a Dirichlet boundary condition and a source term requiring the gradient <strong>of</strong><br />
the variable, the prescribed value is used as boundary valueφ = φ . If a Neumann condition is<br />
used, an extrapolated value from the boundary cell is used with a first order approximation.<br />
For the velocity component normal to the wall, the boundary value is set to zero to ensure zero<br />
mass flow.<br />
For the pressure gradient calculation, if the flux <strong>of</strong> the variable at the face is prescribed, the<br />
boundary value is:<br />
∂φ<br />
φ = φ + I'<br />
F n<br />
∂δ<br />
P<br />
To solve the Poisson equation, the values <strong>of</strong> Δt<br />
n j<br />
∂x<br />
b<br />
55<br />
b<br />
I ' j b<br />
(4.18)<br />
∂x<br />
j<br />
j<br />
and Q jn j for the momentum<br />
*<br />
correction equation, the boundary value for δP is obtained from the cell value from a first<br />
∂ P<br />
order approximation using the fact that Δt n j = 0<br />
∂x<br />
δ<br />
j<br />
The value for the boundary conditions at the wall is prescribed for all velocities and<br />
turbulent variables. The way this boundary condition is treated depends on the turbulence<br />
model used. Many turbulence models have been designed under a local equilibrium<br />
assumption and use the universal logarithmic law to bridge the viscous sublayer, hence<br />
solving for the flow outside the buffer layer. To prescribe the velocity at the wall, the total<br />
shear stress is required:<br />
u uk<br />
*<br />
τω = ρ<br />
(4.19)<br />
1 / 2<br />
with u* is the friction velocity andu<br />
k = k / Cμ<br />
. In the code, the shear stress is calculated<br />
by using the wall function approach. Defining the tangential velocity at the wall as<br />
u = u − u n n<br />
tg<br />
j<br />
j<br />
i<br />
with κ =0.42 and C = 5.2.<br />
i<br />
j<br />
. The shear stress can be calculated from:<br />
u<br />
*<br />
uk<br />
=<br />
−1<br />
/ 4 1/<br />
2<br />
= Cμ<br />
k<br />
u<br />
1<br />
ln<br />
k<br />
tg<br />
j<br />
u<br />
tg<br />
j<br />
+ ( y ) + C<br />
(4.20)<br />
(4.21)
Chapter IV Numerical simulation<br />
For the turbulent variables, the boundary conditions are obtained from:<br />
56<br />
∂k<br />
= 0<br />
∂y<br />
∂ε u<br />
= −<br />
∂y<br />
ky<br />
In the case <strong>of</strong> k − ε model and for the second moment closure the conditions are:<br />
∂ u'<br />
u'<br />
i<br />
∂x<br />
j<br />
j<br />
n<br />
j<br />
= 0<br />
3<br />
k<br />
2<br />
(4.22)<br />
(4.23)<br />
if i = j (4.24)<br />
u u u u<br />
*<br />
' = (4.25)<br />
1 '2<br />
k<br />
u u = u u'<br />
= 0<br />
(4.26)<br />
1<br />
3<br />
'2 3<br />
∂ε u<br />
= −<br />
∂y<br />
ky<br />
This in a local frame where x2<br />
is normal to the wall.<br />
3<br />
k<br />
2<br />
(4.27)
Chapter V Results and Discussion <strong>of</strong> the Simulation<br />
Chapter 5<br />
Results and discussion <strong>of</strong> the simulation<br />
5.1 Introduction<br />
In the present we are looking upon clusters <strong>of</strong> density packed cylinders called tube bundles.<br />
In group <strong>of</strong> cylinders can be arranged in in-line (straight), staggered (rotate square), normal<br />
triangle or parallel (rotate normal triangular) configurations (see figure 5.1). Our particular<br />
interest is in-line configuration as this is widely used in heat exchangers <strong>of</strong> chemical and<br />
nuclear/coal power plants. The frequently parameters characterizing the tube bundle's<br />
geometry are:<br />
The gap ratio<br />
In a group <strong>of</strong> circular cylinders, the arrangement <strong>of</strong> the cylinders is important. For example,<br />
different cylinder arrays are specified by the pitch to the diameter:<br />
P<br />
D<br />
Pitch<br />
=<br />
Diameter<br />
Square array (90°)<br />
P<br />
Tube Row<br />
57<br />
P<br />
D<br />
Rotated Square array (45°)<br />
P
Chapter V Results and Discussion <strong>of</strong> the Simulation<br />
Triangular array (30°)<br />
Reynolds number (Re)<br />
P<br />
Figure 5.1: Tube arrangements<br />
Square array (60°)<br />
The Reynolds number is a dimensionless number that is significant in the design <strong>of</strong> a model<br />
<strong>of</strong> any system in which the effect <strong>of</strong> viscosity is important in controlling the velocities or the<br />
flow pattern <strong>of</strong> a fluid. For the case <strong>of</strong> circular cylinder, the characteristic length is typically<br />
taken to be the tube diameter; it can be shown that the Reynolds number is representative <strong>of</strong><br />
the ratio <strong>of</strong> inertia force to viscous force in the fluid. It is given by:<br />
Inertia force<br />
Re= (5.1)<br />
Viscous force<br />
The Reynolds numbers gives a measure <strong>of</strong> transition from laminar to turbulent flow, boundary<br />
layer thickness, and fluid field across cylinders.<br />
Strouhal number (St)<br />
The inverse <strong>of</strong> the reduced flow velocity is called the Strouhal number, provided that the<br />
frequency is the frequency associated with flow field, such as the vortex shedding, the<br />
Strouhal number is related to the oscillation frequency <strong>of</strong> periodic motion <strong>of</strong> a flow.<br />
Where ƒ is the vortex shedding frequency.<br />
D<br />
St=ƒ (5.2)<br />
U<br />
5.2 Case Description<br />
In the present work an in-line (array) tube bundle configuration is tested to understand the<br />
complex flow behaviour across tube bundles with a gap ratio P/D (aspect ratio) which is<br />
defined in section 5.1 <strong>of</strong> 1.44 and transverse length to diameter ratio T/D <strong>of</strong> 1.44 and<br />
58<br />
P
Chapter V Results and Discussion <strong>of</strong> the Simulation<br />
Reynolds number <strong>of</strong> 70000. The current configuration is square (P/D=T/D) (see figure 5.2).<br />
The present test case requires only periodic and wall boundary conditions (see Figure 5.3).<br />
The simulation configuration is reduced to a few set <strong>of</strong> cylinders as seen in Benhamadouche et<br />
al. [13]. For 3D calculation the bundle is assumed infinite in all directions but it has been<br />
observed that the extrusion length plays an important role in the development <strong>of</strong> the flow<br />
physics. Results with one diameter extrusion length and also with 2 by 2 tubes case shows that<br />
the flow did not develop an anticipated resulting. Hence a complete two points correlation<br />
study was carried out and it was concluded that one needs to take at least two diameter depths<br />
and 3 by 3 configuration for numerical simulations.<br />
5.3. Grid generation<br />
Numerical accuracy and stability <strong>of</strong> any simulation is strongly dependent on the quality <strong>of</strong><br />
the grid used hence one <strong>of</strong> the most tedious jobs in any numerical simulation is the grid<br />
generation. Numerical accuracy issue relates to mesh density and parameters such as the first<br />
+<br />
cell's width y (next to the solid wall) and moreover the adimensional distance y which is<br />
proportional to y (see equation 5.3)<br />
+<br />
y ν<br />
y = (5.3)<br />
u<br />
Where u*<br />
is the friction velocity (It was read from listing Code-Saturne)<br />
+<br />
For URANS models and especially standard k-ε model, the y must lie between 30 and 70. In<br />
the present case the grid generated at the first time is 2D mesh <strong>of</strong> the geometry 2 by 2 tubes.<br />
The number <strong>of</strong> cells in 2D indicates the number <strong>of</strong> 2D cells obtained by cutting the<br />
computational domain with a plane orthogonal to the spanwise direction. Seven blocks were<br />
first generated with different faces and were then copied 24 times to get the smaller domain <strong>of</strong><br />
2 by 2 tubes (see figure 5.4) and after were copied 96 times to get the larger domain <strong>of</strong> 3 by 3<br />
tubes (see figure 5.5). 2D grid is structured. The larger domain is used in the present<br />
simulation because if the domain size in the streamwise or spanwise directions is smaller than<br />
the size <strong>of</strong> the largest structures then errors might be occurred. It was decided to use 3 by 3<br />
configuration and to use the spanwise depth <strong>of</strong> two diameters L =2D to obtain 3D mesh (see<br />
figure 5.6). All parameters <strong>of</strong> 2D and 3D grids generated are shown in table 5.1:<br />
*<br />
59<br />
z
Chapter V Results and Discussion <strong>of</strong> the Simulation<br />
2D<br />
Grid<br />
3D<br />
Grid<br />
Number<br />
<strong>of</strong><br />
cells<br />
21600<br />
604800<br />
Number<br />
<strong>of</strong><br />
faces<br />
44400<br />
65600<br />
Size <strong>of</strong><br />
Grid<br />
(Mo)<br />
2.617<br />
34.445<br />
Number<br />
<strong>of</strong><br />
elements<br />
Quad4<br />
44400<br />
Hexa8<br />
21600<br />
Quad4<br />
65600<br />
Hexa8<br />
604800<br />
User's<br />
Time<br />
CPU<br />
(sec)<br />
1.15<br />
18.21<br />
System's<br />
Time<br />
CPU<br />
(sec)<br />
0.12<br />
2.56<br />
Total<br />
Time<br />
(sec)<br />
1.33<br />
24.45<br />
TTCPU<br />
T Time<br />
0.96<br />
0.85<br />
Table 5.1: Parameters <strong>of</strong> 2D and 3D grids in the present case.<br />
+<br />
y<br />
[13,70]<br />
[13,70]<br />
2D and 3D Grids are generated in a machine <strong>of</strong> a CPU processor Pentium4, processor's speed<br />
2.4 GHz and RAM memory 512 MB.<br />
In first time, 2D calculations with 2D grid were run in a machine's CPU processor is Pentum4;<br />
processor's speed is 2.4 GHz and RAM memory is 512 MB. About 3D grid which is so large,<br />
2 CPU processors were needed in minimum for running 3D calculations. Consequently, a<br />
cluster was used. The cluster's CPU processor is INTEL XEON, processor's speed is 3.2 GHz<br />
and Total RAM memory is 4GB (total number <strong>of</strong> CPU's is:8.) For the present work, it was<br />
used just 2 CPU processors from the 8.<br />
5.4 Discussion <strong>of</strong> the results<br />
The asymmetric flow inside tube bundle can best be understood by concentrating on the<br />
wake <strong>of</strong> the centre cylinder. In this section, the pressure distribution around centre cylinder,<br />
velocity pr<strong>of</strong>iles in the wake <strong>of</strong> tubes and turbulence intensities are discussed.<br />
Figure (5.7) shows the convergence curves. It means evolution <strong>of</strong> the pressure (see figure 5.7<br />
(a)) and velocity (see figure 5.7 (b)) according to number <strong>of</strong> iterations and comparison<br />
between various URANS models. For the pressure, URANS models show that the pressure<br />
and velocity fluctuate according to time except k-ε model which is stable and does not show<br />
any instability <strong>of</strong> pressure and velocity along time. When averaging the two quantities <strong>of</strong><br />
pressure and velocity. They become stable after 15000 iterations for a pressure and after<br />
30000 iterations for a velocity.<br />
60<br />
y<br />
(m)<br />
0.6E-03<br />
0.6E-03
Chapter V Results and Discussion <strong>of</strong> the Simulation<br />
Figure 5.8 shows the instantaneous pressure contour lines distribution in a XY cross<br />
sectional view together with a comparison between different 2D URANS models used in the<br />
present case. For accurate comparison all the plots are drawn on a same scale <strong>of</strong> -1 to1. All<br />
models show that the flow is unstable and asymmetric except standard k-ε model; it fails to<br />
capture the instability <strong>of</strong> the flow. The pressure has the same behavior around all tubes. It<br />
appears a high pressure region on the top <strong>of</strong> the cylinder with a low pressure just downstream<br />
which believed to be the correct solution according to LES. The slight <strong>of</strong> stagnation region on<br />
the top is a direct result in deflection <strong>of</strong> mean flow due to suppression and distortion <strong>of</strong> vortex<br />
shedding from downstream cylinder.<br />
Figure (5.9) shows the instantaneous velocity contour lines distribution in a XY cross section<br />
and a comparison between various 2D URANS models. For accurate comparison all the plots<br />
are drawn on a same scale <strong>of</strong> 0 to 2.5. The low velocity is seen in the stagnation point and<br />
close to the wall caused by the high viscosity turbulent. The flow seems to be going down.<br />
Shear layer separation is seen to originate from the bottom between 180 degree and 270<br />
degree. Due to small gap spacing the recirculation bubbles are not <strong>of</strong> the same size and shape,<br />
where the bottom recirculation region is substantially larger in size and bigger in intensity than<br />
the top one (see figure 5.32). Because <strong>of</strong> high viscosity at the wall, velocity is minimum and<br />
maximum in free stream flow far from cylinders and in zones <strong>of</strong> recirculation. K-ω SST, RSM<br />
and SST- C show unsteadiness <strong>of</strong> the flow but standard k-ε model does not show any<br />
as<br />
instability.<br />
Figure (5.10) shows velocity vectors distribution in a XY cross section for various 2D<br />
URANS models. The same scale than the velocity is used. This figure illustrate better zones <strong>of</strong><br />
recirculation, RSM and SST- C models show recirculation in bottom <strong>of</strong> cylinders better than<br />
as<br />
the others URANS models (see figure 5.10 (b), (d)).<br />
Figure (5.11) shows 2D Vorticity in XY sectional view for different URANS models. The kε<br />
model fails to capture the shear layer separation for the same scale <strong>of</strong> 0 to 6. Hence the other<br />
URANS models show two recirculation bubbles. The shear layer separates from the bottom <strong>of</strong><br />
the cylinder surface as well resulting in two recirculation regions behind every tube. However,<br />
the flow is still asymmetrical with one recirculation bubble larger than the other.<br />
61
Chapter V Results and Discussion <strong>of</strong> the Simulation<br />
Figure (5.12) shows distribution <strong>of</strong> mean pressure in XY cross view for pitch ratio <strong>of</strong> 1.44<br />
and Reynolds number <strong>of</strong> 70000 and comparison between various 3D URANS models, DES<br />
and LES <strong>of</strong> Imran. URANS models seem to give similar results. Pressure has a same behavior<br />
around all tubes. The peak <strong>of</strong> pressure is on the top <strong>of</strong> tube because <strong>of</strong> stagnation region. It<br />
agrees with LES and minimum values locate in the bottom <strong>of</strong> tubes caused by detachment <strong>of</strong><br />
shear layer and recirculation region. Results <strong>of</strong> RSM and SST- C agree better with DES and<br />
LES than k-ω SST.<br />
Figure (5, 13) shows distribution <strong>of</strong> velocity in the wake along a line at x=4.33cm for<br />
P/D=1.44 and comparison between 3D URANS models, present DES and LES <strong>of</strong> Imran for<br />
P/D=1.5. It appears a same sense deflection across the set <strong>of</strong> cylinders. URANS models<br />
capture the four peaks <strong>of</strong> velocity maxima and minima. Velocity is maximum far from<br />
cylinder's walls and it appears clearly minimum near the wall and in the gap spacing. RSM<br />
model is closer to LES than the other URANS models.<br />
Figure (5.14) shows the C evolution <strong>of</strong> normalized pressure coefficient along the central<br />
tube for gap ratio <strong>of</strong> 1.44, for better comparison the mean normalized C pr<strong>of</strong>ile is shown<br />
where it is defined as:<br />
p<br />
C p<br />
2<br />
( P - P ) ρ<br />
= 2 U<br />
(5.4)<br />
The angle measurement which is show in figure (5.2) is similar to a clockwise and starts from<br />
the inlet free stream direction that is from left to right. LES <strong>of</strong> Imran Afgan for aspect ratio <strong>of</strong><br />
1.5 and experimental data <strong>of</strong> Yahiaoui et al. (2007) for aspect ratio <strong>of</strong> 1.44 are used for<br />
comparisons. The effect <strong>of</strong> flow deflection is observed in term <strong>of</strong> stagnation pressure region<br />
located somewhere around 45 degrees from the flow direction. This shift <strong>of</strong> stagnation point<br />
location is also validated from experimental study <strong>of</strong> Yahiaoui (2007) and LES. It is the result<br />
in deflection <strong>of</strong> mean flow due to suppression and distortion <strong>of</strong> vortex shedding from<br />
downstream cylinder. The minimum pressure is located at around 90 degrees because <strong>of</strong> a<br />
separation <strong>of</strong> a shear layer and recirculation region. Interestingly all 2D URANS models k-ω<br />
SST, SST- C and RSM (see figure 5.14) seems to give similar predictions and agree with<br />
as<br />
experimental data <strong>of</strong> Yahiaoui et al. for the stagnation point and the pressure minima except in<br />
the gap space. RSM is closer to SST- C (see figure 5.15). There is a delay <strong>of</strong> reattachment for<br />
as<br />
2D URANS models. It can be caused by a high Reynolds number. When looking to 3D<br />
URANS results (see figure 5.16) one observes that k-ω SST, RSM, DES and experimental<br />
62<br />
ref /<br />
0<br />
as<br />
p
Chapter V Results and Discussion <strong>of</strong> the Simulation<br />
data <strong>of</strong> Yahiaoui et al. and LES <strong>of</strong> Imran are in close agreements whereas about the LES <strong>of</strong><br />
Imran the peak <strong>of</strong> the pressure around 45 degrees is over. They also slightly over predict this<br />
value to 40 degrees then they are closer to LES and experimental data (45 degrees).<br />
For an incompressible flow such as the one under consideration the bulk gap velocity is<br />
calculated by:<br />
⎡ T / D ⎤<br />
U g = U 0 ⎢<br />
⎣T<br />
/ D −1⎥<br />
(5.5)<br />
⎦<br />
For 2D URANS cases the averaged velocity along a vertical line in XY plane at X=4.33 is<br />
shown in Figure 5.17. It observed two high peaks at Y/D=1 and Y/D=4. It means that velocity<br />
is maxima in mean flow free and fully developed far from the wall in a streamwise direction.<br />
Velocity decreases while bringing closer to the wall. RSM model is closer to LES and<br />
experiment <strong>of</strong> Aiba et al. [61] than the other 2D URANS models (maximum velocity is around<br />
1.5 and minimum around zero), (see figure 5.18). Two models k-ω SST and SST- C are<br />
closest, their curves have the same pace l<strong>ik</strong>e LES and experimental data. Moreover they have<br />
negative values <strong>of</strong> velocity and a little peak (see figure 5.17). It means that velocity change<br />
direction, it might be caused by high Reynolds number and a recirculation region. k-ε model is<br />
far to predict complete flow physics and its curve is very low comparing with LES and<br />
experiment.<br />
Figure (5.20) shows the 3D mean velocity along a vertical line in the wake <strong>of</strong> tube in XY<br />
plane at X= 4.33 and length extrusion <strong>of</strong> L =2D. URANS models capture peaks <strong>of</strong> velocity<br />
z<br />
around U= 1 m/s. All curves are similar but SST-C model's prediction here is far better than<br />
the other URANS models (see figure 5.21). Since it tends to capture all the local peaks in<br />
velocity pr<strong>of</strong>ile. It agrees better with LES and experiment.<br />
Figures 5.22 to 5.27 show fluctuating velocity and pressure and their density power spectrum<br />
(DPS) in different probes (1, 3, and 6). An analytical formula for the Strouhal number is<br />
proposed in Chen’s book (see equation 5.6).<br />
as<br />
1 ⎛ P ⎞<br />
St = ⎜ −1⎟<br />
(5.6)<br />
2 ⎝ D ⎠<br />
It depends only on the tube spacing but not on the Reynolds number which is not really<br />
realistic but gives a good order <strong>of</strong> the Strouhal number. For P/D=1.44 one finds St = 1.13. By<br />
applying DPS to the pressure and velocity signals, one clear peak is obtained around the<br />
frequency 45 Hz (see figure 5.23, 5.24, 5.25, 5.26). This value corresponds respectively to the<br />
63<br />
as
Chapter V Results and Discussion <strong>of</strong> the Simulation<br />
Strouhal number <strong>of</strong> 0.84. It means that the vortex shedding detected in the shear regions with<br />
URANS models is realistic. The Strouhal numbers obtained with Star CCM and Code-<br />
Saturne, Benhamadouche [13] are 0.8 and 1.25.<br />
A thorough breakdown <strong>of</strong> Reynolds stresses is shown in Figure 5.28 for RSM model. It<br />
shows the time averaged Reynolds stresses pr<strong>of</strong>iles < u′u′ >, < v′v′ >, < w′w′ > and ,<br />
, along the same line. One notices from this figure that the peaks in the<br />
streamwise and normal Reynolds stresses (< u′u′ >, < v′v′> respectively) are quite decently<br />
captured by the RSM model. Usually has the opposite sign to the boundary layer shear<br />
strain.<br />
Figure 5.29 shows averaged velocities field , , . . Mean streamwise<br />
velocity represent sharp due to a minimum flow passage within the bundle geometry and a<br />
variation from almost 0.2 to a maximum value <strong>of</strong> around 1. Normal velocity has a same<br />
behavior but higher than mean streamwise velocity and a variation from 0 to 2. Usually mean<br />
velocity has a same pace but an opposite sign to streamwise and normal velocity.<br />
Normalized average velocity (see figure 5.29 (d)) is compared with experimental data <strong>of</strong> Aiba<br />
et al. [61] and with LES <strong>of</strong> Imran [14]. The velocity pr<strong>of</strong>ile are close to the experiment and<br />
LES when the modelled quantity u" is added to the averaged quantity u which is equal to<br />
+u'. It concludes that the URANS method produce good predictions <strong>of</strong> turbulent flow.<br />
In order to draw a comparison from the numerical results, the structure parameter, Q, is<br />
calculated, which was found by Hunt et al. (1988) to be an effective way to visualize the<br />
regions <strong>of</strong> coherent Vorticity due to rotational motion (as opposed to those from shear), and is<br />
defined as:<br />
= ij -1/2<br />
( S − Ω Ω )<br />
S<br />
Q (5.5)<br />
Where Q should take a positive value. Figure 5.30 shows instantaneous iso-surfaces for the<br />
three URANS models and DES approach. The lack <strong>of</strong> coherent structures is apparent in the<br />
results from the SST model, for the entire wake region except very close to cylinders and the<br />
side walls. For this model it is the over-predicted values <strong>of</strong> turbulent kinetic energy, leading to<br />
high values <strong>of</strong> the turbulent viscosity, that are responsible for the early damping out <strong>of</strong> these<br />
structures. As a result the vortex dislocations are not seen in the SST results (see figure 5.30<br />
(b)), whereas those from the RSM clearly show the lack <strong>of</strong> structures in the space between<br />
cylinders for the RSM results is seen clearly from the oblique angle (see figure 5.30 (a)),<br />
whereas both the standard SST and the SST- Cas<br />
model and DES approach show more<br />
64<br />
ij<br />
ij<br />
ij
Chapter V Results and Discussion <strong>of</strong> the Simulation<br />
structures in this region. The vortex dislocations can also be seen in the results from the SST-<br />
Cas<br />
model and there are significantly more spanwise structures than seen with either <strong>of</strong> the<br />
two other models. In fact, the wake predicted by DES approach (see figure 5.30 (d)) and<br />
SST-C scheme (see figure 5.30 (c)) displays much more structures.<br />
as<br />
Figure 5.31 shows comparison between results <strong>of</strong> Code-Saturne and Star-CD. For accurate<br />
comparison all the plots are drawn on a same scale. A non-symmetrical solution is observed.<br />
The behavior has been confirmed by STAR-CD calculations with P/D=1.44 and Reynolds<br />
number <strong>of</strong> 70000.<br />
Figure 5.32 represent mean velocity vectors field. Two recirculations coexist. The shear layer<br />
separates from the bottom <strong>of</strong> cylinder surface as well resulting in two recirculation regions<br />
behind every tube. However, the flow is still asymmetrical with one recirculation bubble in the<br />
bottom (see figures 5.32 (a), (b), (c), (d)) due to the acceleration <strong>of</strong> the fluid and a small<br />
recirculation on the top <strong>of</strong> the first recirculation. The shear stress in the bottom is higher than<br />
on the top. SST- C , k-ω SST, RSM and DES approach show the two recirculation bubbles;<br />
as<br />
RSM shows close agreements comparing with LES <strong>of</strong> Benhamadouche (see figure5.32 (e)).<br />
65
Chapter V Results and Discussion <strong>of</strong> the Simulation<br />
U∞ Ug<br />
Periodicity in "y" direction<br />
Figures<br />
Figure5.2: Geometry <strong>of</strong> in-line tube bundles<br />
Figure5.3: Boundary conditions <strong>of</strong> tube bundles<br />
P<br />
Periodicity in "x" direction<br />
66<br />
D<br />
T<br />
Wall
Chapter V Results and Discussion <strong>of</strong> the Simulation<br />
Figure5.4: Cross sectional view <strong>of</strong> 2D grid (2X2 arrangement) N=5400 cells, y+= [13-70]<br />
Figure5.5: Cross sectional view <strong>of</strong> 2D grid (3X3 arrangement) N=21600 cells, y+= [13-70]<br />
67
Chapter V Results and Discussion <strong>of</strong> the Simulation<br />
Figure5.6: Cross sectional view <strong>of</strong> 3D grid (3X3 arrangement) in XY, YZ and<br />
XZ sections: N=604800 cells, y+= [13-70]<br />
68
Chapter V Results and Discussion <strong>of</strong> the Simulation<br />
Figure 5.7: Evolution <strong>of</strong> pressure and velocity, Comparison between URANS models.<br />
(a) Pressure, (b) Velocity.<br />
(a)<br />
69<br />
(b)
Chapter V Results and Discussion <strong>of</strong> the Simulation<br />
(a) (b)<br />
(c)<br />
Figure 5.8: 2D Instantaneous Pressure Contour field in a XY cross sectional view<br />
for gap ratio 1.44. (a) k-ε model, (b) RSM, (c) k-ω SST, (d) SST- C<br />
70<br />
as<br />
(d)
Chapter V Results and Discussion <strong>of</strong> the Simulation<br />
(a)<br />
(c) (d)<br />
Figure 5.9: 2D Instantaneous velocity contour field in a XY cross Sectional view<br />
for gap ratio 1.44. (a) k-ε model, (b) RSM, (c) k-ω SST, (d) SST- C<br />
71<br />
as<br />
(b)
Chapter V Results and Discussion <strong>of</strong> the Simulation<br />
(a)<br />
(c) (d)<br />
Figure 5.10: 2D Velocity vectors field in a XY cross Sectional view for gap ratio 1.44.<br />
(a) k-ε, (b) RSM, (c) k-ω SST, (d) SST- C<br />
as<br />
72<br />
(b)
Chapter V Results and Discussion <strong>of</strong> the Simulation<br />
(c)<br />
(a) (b)<br />
Figure 5.11: 2D Vorticity field in a XY cross sectional view for gap ratio 1.44.<br />
(a) k-ε, (b) RSM, (c) k-ω SST, (d) SST- C<br />
73<br />
as<br />
(d)
Chapter V Results and Discussion <strong>of</strong> the Simulation<br />
(a) (b)<br />
(c) (d)<br />
(e)<br />
Figure 5.12: 3D mean pressure distributions in a XY cross view for P/D=1.44, Re=70000.<br />
(a) k-ω SST, (b) RSM, (c) SST- Cas<br />
, (d) DES, (e) LES <strong>of</strong> Imran for P/D=1.5, Re=15000.<br />
74
Chapter V Results and Discussion <strong>of</strong> the Simulation<br />
(a) (b)<br />
(c)<br />
(e)<br />
Figure 5.13: 3D averaged velocity field in a XY cross view for P/D=1.44, Re=70000.<br />
(a) k-ω SST, (b) RSM, (c) SST- Cas<br />
, (d) DES, (e) LES <strong>of</strong> Imran for P/D=1.5, Re=15000.<br />
75<br />
(d)
Chapter V Results and Discussion <strong>of</strong> the Simulation<br />
Figure 5.14: Mean pressure distribution around centre tube, comparison between 2D<br />
Unsteady RANS for P/D=1.44, Re=70000 and LES <strong>of</strong> Imran (Star-V4) for P/D=1.5<br />
Re=15000 and Experiment <strong>of</strong> Yahiaoui et al. (2007)<br />
Figure 5.15: Mean pressure distribution around centre tube, comparison between 2D<br />
Unsteady RANS for P/D=1.44, Re=70000 and LES <strong>of</strong> Imran (Star-V4) for P/D=1.5<br />
Re=15000 and Experiment <strong>of</strong> Yahiaoui et al. (2007)<br />
76
Chapter V Results and Discussion <strong>of</strong> the Simulation<br />
Figure 5.16: Mean pressure distribution around centre tube, comparison between 3D<br />
Unsteady RANS for P/D=1.44, Re=70000 and LES <strong>of</strong> Imran (Star-V4) for P/D=1.5<br />
Re=15000 and Experiment <strong>of</strong> Yahiaoui et al. (2007)<br />
Figure 5.17: Mean velocity pr<strong>of</strong>ile, Comparison between 2D Unsteady RANS, Re=70000<br />
and LES <strong>of</strong> Imran Re=15000 (Star-V4) and experiment <strong>of</strong> Aiba et al. (1982) in the wake<br />
<strong>of</strong> centre tubes at x=4.33cm.<br />
77
Chapter V Results and Discussion <strong>of</strong> the Simulation<br />
Figure 5.18: Mean velocity pr<strong>of</strong>ile, Comparison between 2D Unsteady RSM, Re=70000<br />
and LES <strong>of</strong> Imran at Re=15000 (Star-V4) and experiment <strong>of</strong> Aiba et al. (1982) in the wake<br />
<strong>of</strong> centre tubes at x=4.33cm.<br />
Figure 5.19: Mean velocity pr<strong>of</strong>ile, Comparison between 2D Unsteady RSM, SST- Cas<br />
,<br />
Re=70000 and LES <strong>of</strong> Imran (Star-V4), Re=15000 in the wake <strong>of</strong> centre tubes at<br />
x=4.33cm.<br />
78
Chapter V Results and Discussion <strong>of</strong> the Simulation<br />
Figure 5.20: Mean velocity pr<strong>of</strong>ile, Comparison between 3D URANS, Re=70000<br />
and LES <strong>of</strong> Imran (Star-V4) at Re=45000 and experiment <strong>of</strong> Aiba et al. (1982) in<br />
the wake <strong>of</strong> centre tubes at x=4.33cm.<br />
Figure 5.21: Mean velocity pr<strong>of</strong>ile, Comparison between SST- Cas<br />
, Re=70000<br />
and LES <strong>of</strong> Imran (Star-V4), Re=45000 in the wake <strong>of</strong> centre tubes at x=4.33cm.<br />
79
Chapter V Results and Discussion <strong>of</strong> the Simulation<br />
Figure 5.22: Fluctuating Pressure DPS at location <strong>of</strong> probe 1<br />
80
Chapter V Results and Discussion <strong>of</strong> the Simulation<br />
Figure 5.23: Fluctuating Pressure and DPS at location <strong>of</strong> probe 3<br />
81
Chapter V Results and Discussion <strong>of</strong> the Simulation<br />
(a)<br />
(b)<br />
Figure 5.24: (a) Fluctuating Pressure and DPS at location <strong>of</strong> probe 6.<br />
(b) LES (Benhamadouche)<br />
82
Chapter V Results and Discussion <strong>of</strong> the Simulation<br />
Figure 5.25: Fluctuating Velocity and DPS at location <strong>of</strong> probe 1<br />
83
Chapter V Results and Discussion <strong>of</strong> the Simulation<br />
Figure 5.26: Fluctuating Velocity and DPS at location <strong>of</strong> probe 3<br />
84
Chapter V Results and Discussion <strong>of</strong> the Simulation<br />
Figure 5.27: Fluctuating Velocity and DPS at location <strong>of</strong> probe 6<br />
85
Chapter V Results and Discussion <strong>of</strong> the Simulation<br />
Figure 5.28: Reynolds stresses in the wake <strong>of</strong> the centre tubes.<br />
(a) , (b) , (c) , (d) <br />
86
Chapter V Results and Discussion <strong>of</strong> the Simulation<br />
(a)<br />
(c)<br />
Figure 5.29: Mean velocity pr<strong>of</strong>iles <strong>of</strong> RSM in the wake <strong>of</strong> the centre tubes.<br />
(a) , (b) , (c) , (d) u/uo<br />
87<br />
(b)<br />
(d)
Chapter V Results and Discussion <strong>of</strong> the Simulation<br />
(a)<br />
(b)<br />
88
Chapter V Results and Discussion <strong>of</strong> the Simulation<br />
(c)<br />
(d)<br />
Figure 5.30: Iso-surface <strong>of</strong> parameter Q for the instantaneous flow across in-line<br />
tube bundles. (a) RSM, (b) k-ω SST, (c) SST-C , (d) DES.<br />
as<br />
89
Chapter V Results and Discussion <strong>of</strong> the Simulation<br />
(a)<br />
(b)<br />
(c)<br />
Figure 5.31: Comparison between Code-Saturne (in right) and Star-CD (in left).<br />
k-ω SST, (a)Pressure, (b) Velocity, (c) Turbulent kinetic energy<br />
90
Chapter V Results and Discussion <strong>of</strong> the Simulation<br />
(a) (b)<br />
(b)<br />
(e)<br />
Figure 5.32: 3D mean velocity vectors, (a) k-ω SST, (b) SST- Cas<br />
, (c) RSM, (d) DES,<br />
(e) LES (Benhamadouche)<br />
91<br />
(d)
Chapter VI Conclusions and Recommendations for future work<br />
Chapter 6<br />
Conclusions and recommendations for future work<br />
6.1 Final remarks<br />
The cross flow over 2D and 3D square in-line tube bundles <strong>of</strong> gap ratio 1.44 and Reynolds<br />
number 70000 is investigated via different 2D and 3D URANS models and DES. The results<br />
were compared to Large Eddy Simulation study for the same configuration. The complex flow<br />
for this configuration has already been shown to be asymmetric in nature with a high<br />
deflection in the mean flow direction.<br />
It is seen that the 2D and 3D results tend to capture the basic asymmetry <strong>of</strong> the flow.<br />
Moreover, the pressure predictions on the surface <strong>of</strong> tubes are also hugely under or over<br />
estimated by all 2D URANS models. 3D URANS simulations on the other hand seem to<br />
produce better results. The k-ε model failed to capture any flow physics in the spanwise<br />
direction. k-ω SST, RSM and SST-C seems to give similar predictions <strong>of</strong> the flow physics.<br />
The new SST-C model agrees better with LES and experimental data. For the present case,<br />
final remarks are highlighted:<br />
as<br />
� The flow is asymmetric and instable.<br />
as<br />
� URANS models k-ω SST, RSM and SST-C moreover DES approach show the<br />
instability <strong>of</strong> flow across tubes.<br />
� K-ε model fail to capture any instability <strong>of</strong> the flow across tubes.<br />
� Distribution <strong>of</strong> pressure around centre tube shows high pressure on the top (in<br />
stagnation region). It is a result in deflection due to suppression and distortion <strong>of</strong><br />
vortex shedding.<br />
� Distribution <strong>of</strong> velocity in a wake <strong>of</strong> centre tubes shows low velocity in free flow in<br />
streamwise direction far from cylinder's walls and also in recirculation regions.<br />
� Stagnation point is located somewhere around 45 degrees. It agrees with LES <strong>of</strong> Imran<br />
and experiment <strong>of</strong> Yahiaoui et al.<br />
� From DPS results applying to pressure's and velocity's signals, it concludes that there<br />
is a peak around 45 Hz (St=0.84) which proves existence <strong>of</strong> recirculation.<br />
� By drawing the structure parameter Q. It observes structures in the gap space between<br />
cylinders. There are more streamwise and spanwise structures for SST- C and DES<br />
than other URANS models.<br />
92<br />
as<br />
as
Chapter VI Conclusions and Recommendations for future work<br />
� Code Star-CD confirms asymmetry <strong>of</strong> a flow and gives similar results with Code-<br />
Saturne.<br />
� Two recirculations behind the centre tube are observed. One is larger than the other. It<br />
is located in the bottom and a small one in upstream. Shear stress in the bottom is<br />
higher than on the top <strong>of</strong> tube.<br />
Recommendations for the future work<br />
This work has made some significant progress towards the original objectives outlined in the<br />
Introduction, but it represents only the start <strong>of</strong> the development and validation process that<br />
would be necessary in order to confidently apply a generic form <strong>of</strong> these models to a broad<br />
range <strong>of</strong> test cases. Certain elements <strong>of</strong> the modelling described in this thesis were, <strong>of</strong><br />
necessity, not explored as fully as they could have been, in order to implement and test a<br />
robust scheme within the given time frame. There thus remain several areas within which<br />
further work would appear to be beneficial:<br />
� Compute the flow across in-line tube bundle using LES and DES.<br />
� Simulate the tube bundle case together with heat transfer.<br />
� Test the new SST-C as model [18] for different new cases.<br />
93
Bibliography<br />
[1] C. MOULINEC, J.C.R. HUNT and F.T.M. NIEUWSTADT. "Disappearing Wakes and<br />
Dispersion in Numerically Simulated Flows through Tube Bundles". Flow, <strong>Turbulence</strong> and<br />
Combustion 73: 95–116, 2004.<br />
[2] Haitham M. S. Bahaidarah and M. Ijaz and N. K. Anand. "Numerical Study <strong>of</strong> Fluid<br />
Flow and Heat Transfer over a Series <strong>of</strong> In-line Noncircular Tubes Confined in a Parallel-<br />
Plate Channel". Numerical Heat Transfer, Part B, 50: 97–119, 2006.<br />
[3] A. Gatto, K.P. Byrne, N.A. Ahmed and R.D. Archer. "Mean and Fluctuating Pressure<br />
Measurements over a Circular Cylinder In-cross Flow using Plastic Turbine". Experiments in<br />
fluids 30 (2001) 43-46.<br />
[4] Chunlei Liang & George Papadakis. "Large Eddy Simulation <strong>of</strong> Cross Flow over In-line<br />
and Staggered Tube Bundles". Flow Induced Vibration, de Langre & Axisa ed. Ecole<br />
Polytechnique, Paris, 6-9th July 2004.<br />
[5] H.R. Barsamian, Y.A. Hassan. "Large eddy simulation <strong>of</strong> turbulent cross flow in tube<br />
bundles". Nuclear Engineering and Design 172 (2004) 103-122.<br />
[6] Pierre Sagaut. "Large Eddy Simulation for Incompressible Flows". Title <strong>of</strong> the original<br />
French édition: introduction a la simulation des grandes échelles pour les écoulements de<br />
fluides incompressibles, Mathématique et applications. Springer Berlin Heidelberg 1998.<br />
[7] W. RODI and D. LAURENCE. "Engineering <strong>Turbulence</strong> Modelling and Experiments 4".<br />
Proceeding <strong>of</strong> the 4 th International Symposium on Engineering <strong>Turbulence</strong> Modelling and<br />
Measurements Ajaccio, Corsica, France, 24-26 May, 1999.<br />
[8] H.K. Versteeg and W. MALALASEKERA. "An Introduction to Computational Fluid<br />
Dynamic, The Finite Volume Method". First published 1995.<br />
[9] J.H. Ferzige and M. Peric. "Computational Method for Fluid Dynamics". 3 rd edition.<br />
Springer Verlag Berlin Heidelberg.2002.<br />
94
[10] Chunlei Liang and George Papadakis. "Study <strong>of</strong> the Effect <strong>of</strong> Flow Pulsation on the<br />
Flow Field and Heat Transfer over In-line Cylinder Array Using LES". Engineering<br />
<strong>Turbulence</strong> Modelling and Experiments 6. W. Rodi editor, 2005.<br />
[11] Aiba Shinya, Ota Ternakazn and Tuchida Hajime. "Heat Transfer <strong>of</strong> tubes closely<br />
spaced in an in-line bank". International Journal <strong>of</strong> Heat and Mass Transfer, volume 23, Issue<br />
3 March 1980, pages 311, 319.<br />
[12] JUAN C. URIBE and DOMINIQUE LAURENCE. Two Velocities Field Approach to<br />
Hybrid RANS-LES. Under consideration for publication in J.Fluid Mech.<br />
[13] S<strong>of</strong>iane Benhamadouche, Dominique Laurence, Nicolas Jarrin, Imran Afgan,<br />
Charles Moulinec. "Large Eddy Simulation <strong>of</strong> Flow Across In-line Tube Bundle", 2003<br />
[14] Imran Afgan. "Numerical Simulation <strong>of</strong> Cross Flow over Square In-Line Tube Bundle<br />
Arrays using Large Eddy Simulation". PhD. Thesis: Fluid Structure Interaction <strong>of</strong> Cylinder<br />
Bodies using LES by I. Afgan.<br />
[15] Imran Afgan. "Fluid Structure Interaction <strong>of</strong> Cylinder Bodies using Large Eddy<br />
Simulation". PhD. Thesis: Fluid Structure Interaction <strong>of</strong> Cylinder Bodies using LES by I.<br />
Afgan (Second Year Report).<br />
[16] Imran Afgan. "Large Eddy Simulation <strong>of</strong> cylindrical Bodies incorporating unstructured<br />
finite volume mesh". A thesis submitted to The <strong>University</strong> <strong>of</strong> <strong>Manchester</strong> for the degree <strong>of</strong><br />
Doctor <strong>of</strong> Philosophy in the Faculty <strong>of</strong> Engineering and Physical Sciences, March 2007.<br />
[17] Juan C. Uribe. "An Industrial Approach to Near-Wall <strong>Turbulence</strong> Modelling for<br />
Unstructured Finite Volume Method". A thesis submitted to The <strong>University</strong> <strong>of</strong> <strong>Manchester</strong> for<br />
the degree <strong>of</strong> Doctor <strong>of</strong> Philosophy in the faculty <strong>of</strong> Engineering and Physical Sciences,<br />
September 2006.<br />
[18] Alistair Revell. "A Stress-Strain Lag Eddy Viscosity Model for Mean Unsteady<br />
Turbulent Flows". A thesis submitted to the <strong>University</strong> <strong>of</strong> <strong>Manchester</strong> for the degree <strong>of</strong> doctor<br />
<strong>of</strong> philosophy in The Faculty <strong>of</strong> Engineering and Physical Sciences, June 2006.<br />
95
[19] Frederic Archambeau, Namane Michitoua and Marc Sakiz. "Code Saturne: A finite<br />
Volume Code for the Computation <strong>of</strong> turbulent Incompressible Flows Industrial<br />
Applications". International Journal <strong>of</strong> Finite Volumes<br />
[20] Baldwin, B. S. and Lomax, H. (1978), "Thin Layer Approximation and Algebraic<br />
Model for Separated Turbulent Flows", AIAA Paper 78-257.<br />
[21] Granville, P. S. (1987), "Baldwin-Lomax Factors for Turbulent Boundary Layers in<br />
Pressure Gradients", AIAA Journal, Vol. 25, No. 12, pp. 1624-1627.<br />
[22] Mavriplis, D. J. (1991), "Algebraic turbulence modeling for unstructured and adaptive<br />
meshes", AIAA Journal, Vol. 29, pp. 2086-2093.<br />
[23] Turner, M. G. and Jennions, I. K. (1993), "An Investigation <strong>of</strong> <strong>Turbulence</strong> Modeling in<br />
Transonic Fans Including a Novel Implementation <strong>of</strong> an Implicit <strong>Turbulence</strong> Model",<br />
Journal <strong>of</strong> Turbomachinery, Vol. 115, April, pp. 249-260.<br />
[24] Wilcox, D.C. (1998), <strong>Turbulence</strong> Modeling for <strong>CFD</strong>, ISBN 1-928729-10-X, 2nd Ed.,<br />
DCW Industries, Inc.<br />
[25] Smith, A.M.O. and Cebeci, T. (1967), "Numerical solution <strong>of</strong> the turbulent boundary<br />
layer equations", Douglas aircraft division report DAC 33735.<br />
[26] Wilcox, D.C. (1998), <strong>Turbulence</strong> Modeling for <strong>CFD</strong>, ISBN 1-928729-10-X, 2nd Ed.,<br />
DCW Industries, Inc.<br />
[27] Wilcox, D.C. (2004), <strong>Turbulence</strong> Modeling for <strong>CFD</strong>, ISBN 1-928729-10-X, 2nd Ed.,<br />
DCW Industries, Inc.<br />
[28] Emmons, H. W. (1954), "Shear flow turbulence", Proceedings <strong>of</strong> the 2nd U.S. Congress<br />
<strong>of</strong> Applied <strong>Mechanics</strong>, ASME.<br />
[29] Glushko, G. (1965), "Turbulent boundary layer on a flat plate in an incompressible<br />
fluid", Izvestia Akademiya Nauk SSSR, Mekh, No 4, P 13.<br />
[30] Wilcox, D.C. (2004), <strong>Turbulence</strong> Modeling for <strong>CFD</strong>, ISBN 1-928729-10-X, 2nd Ed.,<br />
DCW Industries, Inc.<br />
96
[31] Baldwin, B.S. and Barth, T.J. (1990), A One-Equation <strong>Turbulence</strong> Transport Model for<br />
High Reynolds Number Wall-Bounded Flows, NASA TM 102847.<br />
[32] Dacles-Mariani, J., Zilliac, G. G., Chow, J. S. and Bradshaw, P. (1995),<br />
"Numerical/Experimental Study <strong>of</strong> a Wingtip Vortex in the Near Field", AIAA Journal, 33(9),<br />
pp. 1561-1568, 1995.<br />
[33] Spalart, P. R. and Allmaras, S. R. (1992), "A One-Equation <strong>Turbulence</strong> Model for<br />
Aerodynamic Flows", AIAA Paper 92-0439.<br />
[34] Spalart, P. R. and Allmaras, S. R. (1994), "A One-Equation <strong>Turbulence</strong> Model for<br />
Aerodynamic Flows", La Recherche Aerospatiale n 1, 5-21.<br />
[35] Wilcox, D.C. (1988), "Re-assessment <strong>of</strong> the scale-determining equation for advanced<br />
turbulence models", AIAA Journal, vol. 26, pp. 1414-1421.<br />
[36] Wilcox, D.C. (2004), <strong>Turbulence</strong> Modeling for <strong>CFD</strong>, ISBN 1-928729-10-X, 2nd Ed.,<br />
DCW Industries, Inc.<br />
[37] Menter, F. R. (1993), "Zonal Two Equation k-ω <strong>Turbulence</strong> Models for Aerodynamic<br />
Flows", AIAA Paper 93-2906.<br />
[38] Menter, F. R. (1994), "Two-Equation Eddy-Viscosity <strong>Turbulence</strong> Models for<br />
Engineering Applications", AIAA Journal, vol. 32, pp. 269-289.<br />
[39] Durbin, P. Separated flow computations with the model, AIAA Journal,<br />
33, 659-664, 1995.<br />
[40] Popovac, M., Hanjalic, K. Compound Wall Treatment for RANS Computation <strong>of</strong><br />
Complex Turbulent Flows and Heat Transfer, Flow <strong>Turbulence</strong> and Combustion, 78, 177-202,<br />
2007.<br />
[41] Launder, B. E., Reece, G. J. and Rodi, W. (1975), "Progress in the Development <strong>of</strong> a<br />
Reynolds-Stress Turbulent Closure." Journal <strong>of</strong> Fluid <strong>Mechanics</strong>, Vol. 68(3), pp. 537-566.<br />
[42] T.B. Gatski. Turbulent flows: model equations and solution methodology. In R. Peyret,<br />
editor, Handbook <strong>of</strong> Computational Fluid Dynamics, chapter 6. Academic Press, 1996.<br />
97
[43] K. Hanjali´c, M. Popovac, and M. Had˘ziabdi´c. A robust near-wall elliptic-relaxation<br />
eddy-viscosity turbulence model for <strong>CFD</strong>. International Journal <strong>of</strong> Heat Fluid Flow, 24:1047–<br />
1051, 2004.<br />
[44] W.P. Jones and B.E. Launder. The prediction <strong>of</strong> laminarization with a two-equation<br />
model <strong>of</strong> turbulence. International Journal <strong>of</strong> Heat Mass Transfer, pages 301–314, 1972.<br />
[45] S. Parneix, D. Laurence, and P.A. Durbin. A procedure for using DNS databases.<br />
Journal <strong>of</strong> Fluid Engineering, pages 40–47, 1998.<br />
[46] S. Pope. Turbulent flows. Cambridge <strong>University</strong> Press, Cambridge, 2000.<br />
[47] C.G. Speziale, S.Sarkar, and T.B. Gatski. Modelling the pressure-strain correlation <strong>of</strong><br />
turbulence: an invariant dynamical system approach. Journal <strong>of</strong> Fluid <strong>Mechanics</strong>, pages 245–<br />
272, 1991.<br />
[48] Germano, M., Piomelli, U., Moin, P. and Cabot, W. H. (1991), "A dynamic sub-grid<br />
scale eddy viscosity model", Physics <strong>of</strong> Fluids, A (3): pp 1760-1765, 1991.<br />
[49] Kim, W and Menon, S. (1995), "A new dynamic one-equation subgrid-scale model for<br />
large eddy simulation", In 33rd Aerospace Sciences Meeting and Exhibit, Reno, NV, 1995.<br />
[50] Nicoud, F. and Ducros, F. (1999), "Subgrid-scale modelling based on the square <strong>of</strong> the<br />
velocity gradient tensor", Flow, <strong>Turbulence</strong> and Combustion, 62: pp- 183-200, 1999.<br />
[51] Smagorinsky, J (1963), "General circulation experiments with the primitive equations,<br />
in the basic experiment. Monthly Weather Review", 91: pp 99-164, 1963.<br />
[52] Spalart, P. R., Jou, W.-H., Stretlets, M., and Allmaras, S. R. (1997), "Comments on<br />
the Feasibility <strong>of</strong> LES for Wings and on the Hybrid RANS/LES Approach", Advances in<br />
DNS/LES, Proceedings <strong>of</strong> the First AFOSR International Conference on DNS/LES.<br />
[53] Strelets, M. (2001), "Detached Eddy Simulation <strong>of</strong> Massively Separated Flows", AIAA<br />
2001-0879.<br />
[54] B.J. Daly and F.H. Harlow. "Transport equations in turbulence". Physics <strong>of</strong> Fluids,<br />
pages 2634–2649, 1970.<br />
98
[55] Lars Davidson. "MTF270 <strong>Turbulence</strong> Modelling", P79-45 March 5, 2007.<br />
[56] C.Benocci, J.P.A.J Van Beak and U.Piomell. "Large Eddy Simulation and Related<br />
Techniques: Theory and Applications". Von Karman Institute For Fluid Dynamics. Lecture<br />
Series 2006-04.<br />
[57] Shoei-Sheng Chen. "Flow-Induced Vibration <strong>of</strong> Circular Cylindrical Structures".<br />
Argonne National Laboratory. Argonne, llionois.<br />
[58] Rollet-Miet, P., Laurence, D. R., Ferziger, J. 1999. "LES and RANS <strong>of</strong> turbulent flow<br />
in tube bundles". International J. <strong>of</strong> Heat and Fluid Flow, 20.241-254.<br />
[59] Ogengoren A., Zaida, S. 1992. "Vorticity shedding and acoustic resonance in an In-line<br />
tube bundle, Part II: Acoustic resonance". J. <strong>of</strong> Fluids and Structures 6. PP. 293-309.<br />
[60] Ishigai, S., Nish<strong>ik</strong>awa, E., Yagi, E. 1973. "Structures <strong>of</strong> Gas flow and vibration in tube<br />
banks with tube axes normal to flow". Inst. Sym. on Marine Engineering, Tokyo. PP. 1-5-23<br />
to 1-5-33; in Flow induced vibration <strong>of</strong> circular cylinder structures by Chen, S. S.<br />
[61] Aiba, S., Tsuchida, H., Ota, T. 1982. "Heat transfer around tubes in in-line tube banks".<br />
Bull JSME, 25. 919-926; in Flow around circular cylinders Vol 2: Applications by<br />
Zdravkovich, M. M.<br />
[62] Traub, D. 1990. "Turbulent heat transfer and pressure drop in plain tube bundles".<br />
Chem. Engineering Process, 28. 173-181.<br />
[63] Lam, K., Fang, X. 1995. The effect <strong>of</strong> interference <strong>of</strong> four equispaced cylinders in cross<br />
flow on pressure and force coefficients. J. <strong>of</strong> Fluids and Structures, 9. 195-214.<br />
[64] Lam, K., Li, J. Y., Chan, K. T., So, R. M. C. 2003. Flow pattern and velocity field<br />
distribution <strong>of</strong> cross-flow around four cylinders in a square configuration at a low Reynolds<br />
number. J. <strong>of</strong> Fluids and Structures 17. 665-679.<br />
99
[65] Van Atta, C. W., Gharib, M., Hammache, M. 1988. "Three dimensional structure <strong>of</strong><br />
ordered and chaotic vortex streets behind circular cylinders at low Reynolds numbers". Fluid<br />
Dynamics Research 3. 127-132.<br />
[66] Akhilesh Gupta. 2003. "Enhancement <strong>of</strong> boiling heat transfer in 5X3 tube bundles".<br />
[67] Zdravkovich, M. M. and Stonebanks, K. L. 1990. "Intrinsically non uniform and<br />
metastable flow in and behind tube arrays". J. <strong>of</strong> Fluids and Structures 4. 305-319.<br />
[68] Kim, H. J., Durbin, P. A. 1988. "Investigation <strong>of</strong> the flow between a pair <strong>of</strong> circular<br />
cylinders in the flopping regime". J. <strong>of</strong> Fluid <strong>Mechanics</strong> 196. 431-448.<br />
[69] Sayers, A. T. 1988. "Flow interference between four equispaced cylinders when<br />
subjected to a cross flow". J. <strong>of</strong> Wind Engineering and Industrial Applications, 31. 9-28.<br />
[70] Lam, K., Lo, S. C. 1992. "A visualization study <strong>of</strong> cross flow around four cylinders in a<br />
square configuration". J. <strong>of</strong> Fluids and Structures, 6. 109-131.<br />
[71] Sumner, D., Wong, S. S. T., Price, S. J., Paidoussis, M., P. 1999. "Fluid behavior <strong>of</strong><br />
side by side circular cylinders in steady cross-flow". J. <strong>of</strong> Fluids and Structures 13. 309-338.<br />
[72] Weaver, D. S., Zaida, S., Sun, Z., Feenstra, P. 2001. "The effect <strong>of</strong> platen fins on the<br />
flow induced vibrations <strong>of</strong> an in-line tube array". ASME, PVP-Vol 420-1, pp. 91-100.<br />
[73] Price, S. J., Paidoussis, M. P. 1989. "The flow induced response <strong>of</strong> a single flexible<br />
cylinder in an in-line array <strong>of</strong> rigid cylinders". J. <strong>of</strong> Fluids and Structures, 3. 61-82; in Flow<br />
around circular cylinders Vol 2: Applications by Zdravkovich, M. M.<br />
[74] Feenstra, P. A., Weaver, D. S., Nakamura, T. 2003. "Vortex shedding and Fluid elastic<br />
instability in a normal square tube array excited by two-phase cross-flow. J. <strong>of</strong> Fluids and<br />
Structures, 17. 793-811".<br />
[75] Paidoussis, M. P. 1982. "A review <strong>of</strong> flow induced vibrations in reactors and reactor<br />
components". Nuclear Engg. and Design, Vol 74. 31-60.<br />
100
[76] Weaver, D. S., Fitzpatrick, J. A. 1988. "A review <strong>of</strong> cross flow induced vibrations in<br />
heat exchanger tube arrays". J. <strong>of</strong> Fluids and Structures, 2. 73-93.<br />
[77] Samir Ziada. 2006. "Vorticity Shedding and Acoustic Resonance in tube bundles". J. <strong>of</strong><br />
the Braz. Soc. <strong>of</strong> Mech. Sci. & Eng.<br />
[78] Wolfe.D and Ziada.S 2003. "Feedback control <strong>of</strong> vortex shedding from two tandem<br />
cylinders". Journal <strong>of</strong> Fluids and Structures. Vol. 17, no. 4, pp. 579-592. Mar. 2003<br />
[79] E. Kanstantinidis et al. 2000. "On the Flow and Vortex Shedding Characteristics <strong>of</strong> an<br />
in-Line Tube Bundle in Steady and Pulsating Cross flow".<br />
[80] E. Konstantinidis, S. Balabani, and M. Yianneskis. 2002. "A Study <strong>of</strong> Vortex<br />
Shedding in a Staggered Tube Array for Steady and Pulsating Cross-Flow<br />
101