Catalogingin Publication Data
Division of Information and Documentation
Tomita, Jesuino Takachi
ThreeDimensional Flow Calculations of Axial Compressors and Turbines Using CFD
Techniques / Jesuino Takachi Tomita.
S˜
ao Jos´e dos Campos, 2009.
229f.
Thesis of Doctor of Science – Course of Mechanical–Aeronautical Engineering. Area of
Aerodynamics, Propulsion and Energy. – Technological Institute of Aeronautics, 2009. Advisor:
Prof. Dr. Jo˜
ao Roberto Barbosa.
1. CFD. 2. Gas Turbine. 3. Axial Compressor. 4. Numerical Methods. 5. Compressible Flows.
I. Aerospace Technical Center. Technological Institute of Aeronautics. Division of
Mechanical–Aeronautical. II. Title.
BIBLIOGRAPHIC REFERENCE
TOMITA, Jesuino Takachi. ThreeDimensional Flow Calculations of Axial
Compressors and Turbines Using CFD Techniques. 2009. 229f. Thesis of Doctor
of Science – Technological Institute of Aeronautics, S˜ao Jos´e dos Campos.
SESSION OF RIGHTS
AUTHOR NAME:
Jesuino Takachi Tomita
PUBLICATION TITLE:
ThreeDimensional Flow Calculations of Axial Compressors and
Turbines Using CFD Techniques.
TYPE OF PUBLICATION/YEAR:
Thesis of doctoral / 2009
It is granted to Aeronautics Institute of Technology permission to reproduce copies of
this thesis and to only loan or to sell copies for academic and scientific purposes. The
author reserves other publication rights and no part of this thesis can be reproduced
without the authorization of the author.
Jesuino Takachi Tomita
Rua Carneiro da Cunha, 1228
CEP 04144001 – Apto 144 – S˜ao Paulo–SP–Brasil
THREEDIMENSIONAL FLOW CALCULATIONS
OF AXIAL COMPRESSORS AND TURBINES
USING CFD TECHNIQUES
Jesuino Takachi Tomita
Thesis Committee Composition:
Prof. Dr. Nide Geraldo C. R. Fico J´
unior
Presidente

ITA
Prof. Dr.
Advisor

ITA
Prof. Dr. Edson Luiz Zaparoli
Membro Interno

ITA
Prof. Dr. Nelson Manzanares Filho
Membro Externo 
UNIFEI
Prof. Dr. Jo˜ao Batista Pessoa Falc˜ao Filho
Membro Externo 
IAE
Jo˜ao Roberto Barbosa
ITA
A minha m˜ae Izabel Tiyoka
Tomita, a minha linda esposa Thaisa Talarico Hyppolito
Tomita e ao meu orientador
Professor Jo˜ao Roberto Barbosa, pelo apoio e confian¸ca durante esses anos de estudos em
turbinas a g´as.
Agradecimentos
Durante os anos em que estudei no ITA, pude aprender e fazer muito mais do que eu
esperava, quando nesse Instituto entrei pela primeira vez. Tamb´em aprendi bastante
com os amigos que fiz no CTA. A soma de tudo que ganhei durante esse tempo ´e muito
´ conhecimento de vida.
mais do que conhecimento t´ecnicocient´ıfico. E
Agrade¸co a Deus, por permitir a conclus˜ao de mais uma etapa da minha vida.
Agrade¸co o apoio de minha m˜ae Izabel Tiyoka Tomita, que desde o primeiro dia de aula
da minha vida, me incentivou e me deu suporte para que eu pudesse concluir mais um
trabalho.
Agrade¸co a minha linda esposa Thaisa Talarico Hyppolito Tomita, que durante esses
anos, tem compreendido todo o meu esfor¸co, durante in´
umeras noites, fins de semana e
feriados, na qual fiquei gerando geometrias de compressores e turbinas, malhas e
compilando programas.
Agrade¸co a todos os turbineiros do Grupo de Turbinas do ITA (Cleverson, Franco,
Santin, Luciano, Gustavo, Dulceneia, Renato, Dora, Di Fiori, Helder, M´arcio Mendon¸ca,
e Daniel) e n˜ao poderia deixar de agradecer o Demerval, pela amizade e companheirismo.
Reconhe¸co toda a ajuda e todo o conhecimento que recebi do Professor Jo˜ao Roberto
Barbosa, que, sem d´
uvida, ´e uma referˆencia internacional em turbinas a g´as. Agrade¸co
por ter me dado a oportunidade de mergulhar no mundo das turbom´aquinas. Espero
poder fazer o mesmo com as pr´oximas gera¸co˜es.
Obrigado!
“A natureza,
´e a arte de Deus.”
— Dante Alighieri).
Resumo
Com o advento de potentes computadores, a Dinˆamica dos Fluidos Computacional
(DFC) tem sido vastamente utilizada por pesquisadores e cientistas para investigar o
comportamento de escoamentos e a varia¸c˜ao das propriedades dos mesmos. O custo de
simula¸ca˜o de DFC ´e muito pequeno comparado com o arsenal experimental como bancos
de ensaio e t´
uneis de vento. Nos u
´ltimos anos, muitos pacotes comerciais de DFC foram
desenvolvidos, alguns deles possuem proeminˆencia na ind´
ustria e na academia. Por´em,
alguns c´alculos espec´ıficos de DFC s˜ao casos muito particulares e a`s vezes necessitam de
aten¸ca˜o especial devido a complexidade do escoamento. Nesses casos, uma pesquisa meticulosa tornase necess´aria. Este ´e o caso do c´alculo de escoamentos em turbom´aquinas.
O desenvolvimento de c´odigos de DFC aplicados em simula¸c˜oes de escoamentos em turbom´aquinas e os detalhes das implementa¸c˜oes s˜ao assuntos muito reservados. Um pequeno
n´
umero de institui¸co˜es possui esse tipo de conhecimento. Cada c´odigo de DFC possue sua
particularidade. Desenvolver um c´odigo fonte particular ´e um assunto muito interessante
no senso acadˆemico.
Nesse trabalho, um c´odigo computacional escrito em FORTRAN, foi desenvolvido
para calcular escoamentos internos em turbom´aquinas usando t´ecnicas da DFC. O programa ´e capaz de calcular escoamentos tridimensionais n˜ao somente em turbom´aquinas.
Por exemplo, escoamentos internos e externos como bocais e aerof´olios podem ser cal
viii
culados. O tratamento dado no c´odigo permite o uso de malhas n˜aoestruturadas com
elementos hexa´edricos. Escoamentos envolvendo as equa¸co˜es de Euler, NavierStokes e
escoamentos turbulentos podem ser calculados, dependendo da necessidade do usu´ario.
Diferentes esquemas num´ericos foram implementados para a integra¸c˜ao no tempo e no
espa¸co. M´etodos num´ericos para melhorar a estabilidade e aumentar o passo no tempo
(passo no tempo vari´avel, suaviza¸c˜ao impl´ıcita do res´ıduo) foram tamb´em implementados
e todos os detalhes est˜ao descritos nesse trabalho.
A origem do c´odigo computacional ´e para simular escoamentos em compressores e
turbinas. Dessa forma, o sistema de referˆencia rotacional e n˜aorotacional ´e calculado
simultaneamente. Dessa forma, o proceso de verifica¸ca˜o e valida¸c˜ao do c´odigo foi realizado
para ambos os sistemas.
´ muito
Um procedimento de projeto, passoapasso, ´e apresentado nesse trabalho. E
importante mencionar que para o entendimento completo da f´ısica do escoamento em
compressores e turbinas o projetista deve possuir um s´olido conhecimento de opera¸ca˜o
dos componentes de uma turbina a g´as.
Developing a CFD code is very interest subject in academia. Each CFD code has its particularities. Computational Fluid Dynamics (CFD) has been vastly used by researches and scientists to investigate flow behavior and its properties. In the last years many CFD commercial packages were developed and some of them possess prominence in industry and academia. meticulous research becomes necessary. The cost of CFD simulation is very small compared to the experimental arsenal as test facilities and windtunnels. This is the case of turbomachinery flow calculations. A few institutions have this type of knowledge. some specific CFD calculations are particular cases and sometimes need special attention due to the complexity of the flow. was developed to calculate internal flows in turbomachines using CFD techniques.Abstract With the advent of powerful computer hardware. The solver is capable of calculating the threedimensional flows not only for turbomachines. The approach used allows the use of unstructured meshes of hexahedral elements. Different numerical schemes were implemented for time and space integration. internal and external flows of nozzles and airfoils can be calculated. Numerical tools to improve the stability . For instance. The development of CFD codes applied to turbomachinery flow simulations and its implementation issues are not available. a computational code. However. written in FORTRAN. In these cases. Euler. In this work. NavierStokes and turbulent equations can be calculated depending on the user settings.
The origin of this solver is to simulate flows in compressors and turbines. It is very important to mention that to have a complete understanding of the flow physics in compressors and turbines the designer must have a solid knowledge of the operation of gas turbine components. . A stepbystep design procedure is presented in this work.x and to increase the timestep (local timestep and implicit residual smoothing) were also implemented and all details are described in this work. Hence. the verification and validation processes were run for both inertial and noninertial systems. Therefore. both rotating and nonrotating frames of reference are calculated simultaneously.
. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .4 Author Contribution . . . . . . . . . . . . .2 The flow . .1 Objectives . . . . . . . . . . .3. . . . . . . . 32 1.1 Turbulence . . . . . . . . . . . . . . . . . .3 Previous Developments . . . . . . . . . .5 Work Organization . . . . .2. . . . . . . . . . . . . 47 2 Mathematical Formulation . . .3. . . . . . . . . . . . . . . . . . . . . . . . . . .3. . . . 34 1. . . . . . . . . . . . . . .Contents List of Figures . . 53 2. . . . . . . . . . . . . . . . . . . .1 CFD in Academic Research and Industry . . . . . . . . . 51 2. .1 The fluid . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . xxiii List of Symbols . . . . . . . . . . . . . . . . . . . . . . . . . . . . xv List of Tables . . . . . . . . . . . . . . . . . . . . 36 1. . . . . . . . . . . . . . . . . . . . . . . . . . . . 33 1. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 56 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . xxii List of Abbreviations and Acronyms . . . . . . . . .2 Motivation . . . . . . . . . . . . . . . . . xxv 1 Introduction . . . . . . . . . . . . . . . . 38 1. . . . . . . .2 CFD on Gas Turbines . . . . . . . . . . . . . . . 44 1. . 32 1. . . . . 51 2. . . . . . . . . . . . . . . . . . . . . . . . . 34 1. . . . . . . . . . . . . .3 CFD on Axial Compressors and Turbines . . . . . . . . . . . .
. . . . . . . . . . . . .4. . . 69 FiniteVolume Method . . . . . . . . . . . 84 3. . . . . . . . . . .7 Initial Conditions. . . . . .2. . . . . . . . . . . . .2. . . . . . . . . .CONTENTS 2. 99 3. . . . . . . . . . . . . . . . . . . . . .3. . . . . . . . . . . . . . . . . . .1 Initial Conditions . . . . . . . . . . . 97 3. . . . . . . . . . . . . . 85 3. . . . . . . . . . . . .5 Spatial and Time Integration of the Spalart–Allmaras Turbulence Model . 86 3. . . . . . . . . . . . . . . .2 The Upwind Scheme of Van Leer: Flux–Vector Splitting . . . .2 3 3. . . 111 . . . . . . . . . .7. .2. . . . 92 3. . . . . . .2. . . . . . . . . . . . . . . . . . 92 3. . . . . . . . . . . . . . . . . . . . .1 The Centered Scheme of Jameson . . . . . . . . . . . . . . . . . . . . .4. . . . . 76 for Conservation Laws (M U SCL) . .2 The Scheme of RungeKutta . . . . . . . . .1 3. . . . . 72 3. . . . . . . . . . . . . . . . . . . . . . . .1 The Scheme of MacCormack (1969) . . . . . . . .3. . . . . . . . . . . . . 71 3. . . 83 3. . . . . . . . . . . . . . . . . . . . . . . .2 Implicit Residual Smoothing . . . . . . . . . . .6 Unstructured Mesh Treatment . . . . .7. . . . . . . . . . . . 70 Spatial Integration . . . . . . . . 98 3. . . . 62 Numerical Formulation . . . . . .1 3. . . . . . . . . . .3 Numerical Stop Criteria . . . . . . . . . . . . . . . . . . . .1. 81 3. . . . . .7. . . . . . . . . . . .3 The Upwind Scheme of Roe: Flux–Difference Splitting 3. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 80 3. . . . . . . . 68 Discretization Models . . . . . . . . . . . . . . . . . . . 96 3. . . . . . . . . . . . . . . . . . . . . . . . . .2. . . . . . . .4 Numerical Stability and Convergence Acceleration . . . . . . . . . . . . . . . . .2 xii Turbulence Model . .2. . .1 Artificial Dissipation . . . .3 Time Integration . . . . . . .2 Boundary Conditions . . .4 Reconstruction Based on Approximate Monotone UpstreamCentered Schemes .6 Discretization of Viscous Fluxes . . . . . . . . . . . . 87 3. Boundary Conditions and Rows Interface . . . 84 3. .2. . . . .5 Venkatakrishnan’s Limiter . . . 73 3. . . . . . .
172 6. . . . . . . . . . . . . . 211 8. . . . . . . . . . .2 Laminar Case . . . . . . . . . . . . . . . . . . 170 6. . . . . . . . .4 3D Flow Calculation . . . . . . . . . . . . . . . . . . . . . . . . . . . 212 8. . . . . . . . . . . . 184 6. . . . . . . . . . . . . . .2 Multistage AxialFlow Compressor Simulation . . . . . . . . . . . . 147 6. 172 6. . . . . . . . . . . . . . 126 5. . . 143 6 6. . . . . . . . . . . . . . . . . . . . . .1 Specification of Design Parameters . . . . . . 124 5. . . . . . . . . . . . . . . . . . . . . . . . .1 8 Comments and Conclusions . . . . . . . . .1 5 xiii Computational Implementation . . . . . 207 The CFD solver as a Research and Teaching Tool . . . . 196 7 7. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .1 Results . . . . . . . . . . . . . . . . . . . . . . .2 Preliminary Design .3 Streamline Curvature Method . . . . . . . . . . . . . . . . . . . . . . .3 Turbulent Flow Case . . . . . . . . . . . 214 Bibliography . . . . . . . . 113 Code Structure and Implementation Issues . . . . . . . . . . . . . . . . . . . . . . . . .2 Single Rotor with Low Aspect Ratio . . 215 Appendix A – Artificial Dissipation Issues . . . . . . . . . . . . . . . . . . . . 134 5. . . . . . . . . . . . . . .2. . . .2 Solver . . . . . . . . 165 6. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 210 8. . . 146 Flow Simulation in Turbomachines . . . . . . .1 Inviscid Case . . . . . . . . . . . . . . . . . . . . . 225 . .2. . . . . . . . . . 146 6. . . . . . . . . . . . . . . . .2. . . . .1 SingleStage AxialFlow Turbine Simulation . . .1. . . .1 PreProcessing . . . . . . . . . . . . . .2. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .1. . . . . . 113 Code Verification and Validation . . . . . . . . 209 Future Implementations . . . .3 PostProcessing . . . . . . .CONTENTS 4 4. . . . . . . . . . . . . . .
. . . . . . . . . . . . 225 Annex A – Rotating Frame of Reference . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 228 A. . . . . . .1 Coriolis and Centrifugal Force . . . . . . . . . .CONTENTS A. . 230 . . .1 xiv Stencil Applied on Artificial Dissipation Implementation . . . 228 Glossary . . . . . . . .
129 FIGURE 5. . . . . . . . . . . . . 129 FIGURE 5. .2 – Multidisciplinary team on turbomachinery design . . . .4 – Continuity residue histories for different spatial discretization schemes . . . .2 – Experimental and numerical results for static pressure ratio 2 . . . . 128 FIGURE 5. . . . .3 – Experimental and numerical results for static pressure ratio 5 for centered and upwind schemes . . 126 FIGURE 5. . . .2 – Scheme implemented to set the boundary condition in a ghost element 98 FIGURE 3.4 – Representation of a mixingplane inlet on compressors . . .1 – Nozzle geometry and mesh (flow is from left to right) . . . . . . . . . . . . . . . . . . . . . . 37 FIGURE 1. . . . . . . . . . . .1 – Schematic velocity vector diagram . . . . . . . . . . . . . . . .6 – Mach number contours for upwind scheme . . . . .1 – Gas turbine engine and its components . . . . . . . . . . . . . 38 FIGURE 2. . . . . . . .1 – Scheme based on piecewise linear reconstruction . . .inviscid flow . 81 FIGURE 3. . . . . . . . 127 FIGURE 5. . . . . . . . . . . . . . . . . . . . 129 . . . . . . . 110 FIGURE 4. . . . . . 53 FIGURE 3. . . . 117 FIGURE 4.inviscid flow . . .1 – Scheme created to identify the number of blades . .5 – Mach number contours for centered scheme . . . . . . .List of Figures FIGURE 1. . .3 – Representation of a mixingplane outlet on compressors . . . . . . .inviscid flow . . . . . . . . . 118 FIGURE 5. . . . . . .2 – Scheme created to identify the mixingplanes . . . . . . .inviscid flow . . . 110 FIGURE 3. .
. . . . .3 on a flatplate with 15 nodes inside of the boundarylayer: sixth case .8: upwind .11 –Mach number contours for upwind scheme .8: centereddifference . 140 FIGURE 5. .3 on a flatplate with 15 nodes inside of the boundarylayer: second case . . . 136 FIGURE 5. . . . . . .9 – Pressure coefficient distribution on the NACA0012 airfoil with zero angleofattack and Mn = 0. . . 137 FIGURE 5.14 –Mesh used to calculate the flow on a flatplate . . . . . . . . . . . . . . . . 140 FIGURE 5. .18 –Analytical and numerical solutions of an flow with M∞ = 0. . 138 FIGURE 5. . 133 FIGURE 5. . . . . . . . . . . . . . . . . . 133 FIGURE 5. . .20 –Analytical and numerical solutions of an flow with M∞ = 0. . . . . . 132 FIGURE 5. . . . . . 139 FIGURE 5. . . .3 on a flatplate with 10 nodes inside of the boundarylayer: first case . . . . . . . .25 –Analytical and numerical solutions of a flow with M∞ = 0. . .3 on a flatplate with 12 nodes inside of the boundarylayer: fifth case . . . . . 139 FIGURE 5. . . . . . . . 134 FIGURE 5. . . 131 FIGURE 5. . . . .16 –Analytical and numerical solutions of flow with M∞ = 0.10 –Mach number contours for centered scheme . . . . . .17 –Continuity residue history: second case . . . . . 130 FIGURE 5. . . .3 on a flatplate with 10 nodes inside of the boundarylayer: fourth case . . . .24 –Analytical and numerical solutions of an flow with M∞ = 0.3 on a flatplate with 15 nodes inside of the boundarylayer: third case . . . . 136 FIGURE 5. . . . . . . . . .LIST OF FIGURES xvi FIGURE 5. .8 – Pressure coefficient distribution on the NACA0012 airfoil with zero angleofattack and Mn = 0. . 137 FIGURE 5. .7 – Mesh generated for NACA0012 airfoil flow calculation . . . . . . .22 –Analytical and numerical solutions of an flow with M∞ = 0. . . . . . 131 FIGURE 5. . . . . . . . . . 141 . . . . . . . . . . . . . . . . .21 –Continuity residue history: fourth case . 138 FIGURE 5. . . . . .12 –Static pressure contours for centered scheme . . . . . .13 –Static pressure contours for upwind scheme . . . . .19 –Continuity residue history: third case . . . . . .23 –Continuity residue history: fifth case . . 132 FIGURE 5. . . .15 –Continuity residue history: first case .
. . . .1 – Auxiliary curves to drawing a singlestage axialflow turbine . .29 –Comparison of experimental and numerical results for static pressure ratio 5 for centered and upwind schemes . . . .inviscid case . . . . . . 144 FIGURE 5. . . . . . . . . . . . . . . . . . . . . . . .26 –Details of the Mach number contours inside of the boundarylayer on the flatplate . . . . .32 –Velocity vectors . . . .11 –Static pressure contour of a singlestage axialflow turbine . . . . . . . . .inviscid case . . . 145 FIGURE 5. . . . . 144 FIGURE 5. . . . . . . . . . . . . . . . .6 – Singlestage axial turbine HOHgrid domain . . . . . . . . . . . . . . . . . . . . . . . . . . .10 –Singlestage axialflow turbine massflow outlet convergence history . . . . . . . . . . . . . . . . .turbulent flow . . . . .9 – Closeup of the Ogrid around the stator and rotor blades of a singlestage axialflow turbine . . 152 FIGURE 6. . . . 149 FIGURE 6. . 153 . . 143 FIGURE 5. . . . . . . . . . . . . . . 153 FIGURE 6. . . . . . . . . . . . . . . . .5 – Closeup of the leading edge of an Hgrid . . . .2 – 3D solid drawing of the NGV and rotor of a singlestage axialflow turbine . . . . . . . . .30 –Continuity residue histories for two spatial discretization schemes . . . . . . . . . .turbulent flow . . . . . . . . . . . 151 FIGURE 6. . . . .LIST OF FIGURES xvii FIGURE 5.4 – Singlestage axialflow turbine Hgrid domain . . . 142 FIGURE 5. . . . . . . . . . . . . . . . . . . . . . 145 FIGURE 6. . . .8 – Singlestage axialflow turbine Ogrid domain . .7 – Closeup of the stator trailing edge and rotor leading edge of a singlestage axialflow turbine . . . .turbulent flow . 148 FIGURE 6. . . . . . . 142 FIGURE 5. . . . . . . 150 FIGURE 6. . . . . . . . . . .turbulent flow . . . . . . . . 151 FIGURE 6. . . . . . . . 148 FIGURE 6. . . . . . .31 –Mach number contours . . . . . . . 150 FIGURE 6. . .3 – Full 3D view of an singlestage axialflow turbine . . . . . . . . . . . . . . 149 FIGURE 6. . . . .27 –Details of the velocity vectors profiles inside of the boundarylayer on the flatplate . . . . .28 –Mesh generated for nozzle flow calculation . . . . . . . . . . . . . . . . . .
. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 158 FIGURE 6. . . . .turbulent case with Hgrid . . . . . . . 155 FIGURE 6. . . . . . . . . . . . . . . . . . . . . . . . . . . . . .turbulent case with Hgrid .17 –Singlestage axialflow turbine velocity vectors distribution along the stator blade row . .21 –Detail of the gap between the turbine rotor and casing . . . . . . . . . . . . . 158 FIGURE 6. .14 –Hgrid used for a singlestage axialflow turbine . . . . .inviscid case . . . . . . . . . . . .20 –Detail of the Ogrid on the turbine casing . . . . .18 –Closeup of the reverse flow at the rotor suction side . . . . . . .13 –Closeup of the flow at rotor suction and pressure surfaces . . .turbulent case with Ogrid . . . . . . 159 FIGURE 6. . . .turbulent case with Hgrid . . . . . . . . . . . . 155 FIGURE 6. . . . . .19 –Scheme of the blocks created to generate the Ogrid . . . . . 160 FIGURE 6. 161 FIGURE 6. . . . . . . . . .24 –Singlestage axialflow turbine massflow outlet convergency history . . . . . . . . . . . . . . . . .turbulent case with Ogrid .22 –Mesh elements distribution on the tip clearance region . . . .23 –Detail of the Ogrid around the rotor blade tip and the refinement of the clearance region . . . . . . . . . . . . . 162 . 159 FIGURE 6. . . 156 FIGURE 6. . . . . . . . . . . . . . . . . . . .27 –Distribution of the velocity vectors in the turbine rotor . . . . .turbulent case with Ogrid . . . . . . . .12 –Closeup of the flow across the mixingplane of a singlestage axialflow turbine . . . . . . . . . . . . . . 157 FIGURE 6. . . . .inviscid case . . . . . . . . . 154 FIGURE 6. . . . . . . . . . . . .26 –Singlestage axialflow turbine pressure ratio monitoring . . . .15 –Mixingplane outlet (MPO) and mixingplane inlet (MPI) ratio turbulent case with Hgrid . . 156 FIGURE 6. . . . . . 161 FIGURE 6. . . . . . . . . . . . . . . . . 157 FIGURE 6. . . 160 FIGURE 6. .25 –Mixingplane outlet to mixingplane inlet ratio . . . . . . . . . . . . . . . . . . . . . . . . .turbulent case with Ogrid . . . . . . .16 –Singlestage axialflow turbine massflow outlet convergence history of the . . . . . .LIST OF FIGURES xviii FIGURE 6. . . .
. . . . .turbulent case with Ogrid . . . . . .LIST OF FIGURES xix FIGURE 6. . . . . 167 FIGURE 6. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 169 FIGURE 6.35 –3D Ogrid around the rotor blade . . . . . . . 163 FIGURE 6. . . . 162 FIGURE 6. . . . . . . . . . . . . . . . . . . .31 –Detail of the static pressure contours along the turbine stage . . .turbulent case with Ogrid . . . .41 –Increase of static pressure along the rotor blade . . 173 FIGURE 6. . . .30 –Detail of the static pressure contours along the turbine stage . . . . . . . . . . . . . . . . . 169 FIGURE 6. . . . . . . 166 FIGURE 6. . . .39 –Rotor outlet massflow convergence history . . . . . 168 FIGURE 6. . .turbulent case with Ogrid . . . . . . . . . . . . . . . . . . . . . . .turbulent case with Ogrid . . . 166 FIGURE 6.36 –3D view of the rotor . . . . . . . 170 FIGURE 6. . . . . . . . . .turbulent case with Ogrid . . . . . 165 FIGURE 6. . . . . . . . .45 –5stage axialflow compressor map . . . . . . . . . . . . 167 FIGURE 6. . . . . . . . . . . .32 –Detail of the leakage flow from pressure surface to suction surface of the turbine rotor . . . . . . . . . . . . . . . . .turbulent case with Ogrid . . . . . . . . . . . . . .29 –Global view of the gas expansion along the turbine stage . . . . . .28 –Effect of Coriolis force in the velocity field close to the turbine rotor wall . . . . 166 FIGURE 6. . .turbulent case with Ogrid . . . . . 177 . . . .40 –Rotor pressure ratio . . . . . . . . . . . . . .33 –3D view of the rotor . . . . . . .42 –Static pressure in two different planes (near of hub and near of tip) .37 –Detail of the rotor blade . . . . . . . . .34 –Detail of the rotor blade . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .turbulent case with Ogrid . 164 FIGURE 6. . . . . . . . . . . . .turbulent case with Ogrid . . . . . . . . . . . . . . . . 168 FIGURE 6. . . . . . . . . . . . . . . . . . . . . . .turbulent case with Ogrid . . . .43 –Velocity vector distribution near of the rotor leading edge . . .38 –3D Ogrid around the rotor blade . . . . . .44 –Deflexion and profile loss . . . . . . . . . . . . . 163 FIGURE 6. 167 FIGURE 6. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
.56 –Designed 5stage axialflow compressor . . . . . . . . . . . . . . . . . . . . . . . 202 FIGURE 6. . . . . . . . . . . . .46 –Compressor bleed schedule . . . . . . . . . . . . . . . . .66 –Static pressure contours in the 5stage axialflow compressor . . . .48 –5stage axialflow compressor efficiency using BOV . . . . . . . . . . . . . . . . . . . .65 –5stage axialflow compressor efficiency variation during the numerical iteration . . 177 FIGURE 6. . . . . . 183 FIGURE 6. . . . . . . . . . . . . . . . 184 FIGURE 6. .49 –Compressor VIGV schedule . . . . . . 178 FIGURE 6. . .68 –Detail of the mixingplane velocity distribution . . .54 –Compressor characteristics: pressure ratio . . . . . . . . .62 –Computational domain and mesh structure on the hub of the firststage200 FIGURE 6. . . . . . . . . 199 FIGURE 6. . . . . . . . . . . 203 FIGURE 6. . . . . . . . . 179 FIGURE 6. . 197 FIGURE 6. . . 179 FIGURE 6. .58 –Computational domain of the 5stage axialflow compressor domain 198 FIGURE 6. . . . . . . . . . 204 . . . . . . . . 199 FIGURE 6. . . . . .55 –Compressor Characteristics: efficiency . . . . . . . . . . . . . . . . . 203 FIGURE 6.67 –Detail of the velocity vectors at the compressor outlet . . . . . . . . . . .50 –5stage axialflow compressor map using VIGV . . . . .53 –Sketch of a single shaft free power turbine unit . . . . . . . . . . . . . . . . .59 –Axial view of the 5stage axialflow compressor domain . . . . . . . . 180 FIGURE 6. . . 180 FIGURE 6.60 –General view of the Ogrid generated to the 5stage axialflow compressor . . . .57 –3D view of the 5stage axialflow compressor blade profiles . . . . . . . .63 –Detail of the Ogrid mesh type around the blades . 198 FIGURE 6. . . . . . . . .52 –VIGV and bleed schedule .64 –5stage axialflow compressor outlet massflow convergence history . . . . . . . . . . .47 –5stage axialflow compressor map using BOV . . . . . . . 178 FIGURE 6. . . . . . . . 200 FIGURE 6. . . . . . .61 –Mesh on the blade surfaces .51 –5stage axialflow efficiency using VIGV .LIST OF FIGURES xx FIGURE 6. . . 202 FIGURE 6. . . . . . . . . . . . 182 FIGURE 6. . . . . . . . . . . 198 FIGURE 6. . . . . . .
. . . . . . . . . . . . . . . . . . . . . . . . . . 226 FIGURE A. . . . . . . . .3 – Second scheme to calculate the second and fourth order terms of artificial dissipation . 227 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .70 –Total pressure distribution along the 5stage axialflow compressor . . . 226 FIGURE A. .4 – Third scheme to calculate the second and fourth order terms of artificial dissipation . . . . . . 204 FIGURE 6. . . .2 – First scheme to calculate the second and fourth order terms of artificial dissipation . . . . .69 –Detail of the mixingplane velocity distribution at the compressor thirdstage . . 205 FIGURE A. . . . . . . . . 225 FIGURE A. . . . . . . . . . . . .LIST OF FIGURES xxi FIGURE 6. . . . . . . .1 – Stencil used to calculate the artificial dissipation terms . . . . . . .
. . . . . . . . . . 193 TABLE 6. .7 – Rotor loading factor distribution . . . . . . 194 TABLE 6. . 194 TABLE 6. . . .8 – Distribution of rotor incidence flow angle along the rotor height TABLE 6.11 – Distribution of stator deviation flow angle along the stator height .13 – Stator de Haller number distribution . . . . . . . .9 – Distribution of stator incidence flow angle along the stator height . . . . 195 TABLE 6. . 193 TABLE 6. . . . . . 195 TABLE 6. 193 TABLE 6. . . 205 . 195 TABLE 6. . . 196 TABLE 6. . . .12 – Rotor de Haller number distribution . . . . . . . .3 – Distribution of stator blade inlet angle along the stator height . . . . .14 – Axial Mach number for each blade row . . . . 194 TABLE 6. . . .10 – Distribution of rotor deviation flow angle along the rotor height .4 – Distribution of stator blade outlet angle along the stator height . . . . . .1 – Distribution of rotor blade inlet angle along the rotor height . 193 TABLE 6. . . . . . .15 – Comparison between streamline curvature and CFD results . . 195 TABLE 6. . . . . . . .2 – Distribution of rotor blade outlet angle along the rotor height . . 196 TABLE 6. . . . . . . . . . . . .List of Tables TABLE 6. . . . .5 – Distribution of the blade spacechord ratio along the blade height .6 – Number of blades for each row . . . . . 194 . . . .
List of Abbreviations and Acronyms 1D Unidimensional 2D Bidimensional 3D Threedimensional AFCC Axial Flow Compressor Code AIAA American Institute of Aeronautics and Astronautics BOV Bleedoff valve CONV Convective flux vector CFD Computational Fluid Dynamic CFL Courant–Friedrich–Lewy number DISS Artificial dissipation vector DP Designpoint FDS Flux–Difference Splitting FVM Finite Volume Method FVS Flux–Vector Splitting GTAnalysis Gas Turbine Analysis IRS Implicit Residual Smoothing MP MixingPlane MPO MixingPlane Outlet .
LIST OF ABBREVIATIONS AND ACRONYMS MPI MixingPlane Inlet MUSCL Monotone UpstreamCentered Schemes for Conservation Laws NGV Nozzle Guide Vane ODP Offdesignpoint RHS RightHand Side sm Surge margin SA SpalartAllmaras SLC Streamline Curvature Method VIGV Variable Inlet Guide Vane VISC Viscous flux vector VSV Variable stator vane xxiv .
List of Symbols Latin Characters a Speed of sound Ai PArameter for the artificial dissipation term b Function of inlet airangle A¯Roe Roematrix C Convection term c Blade chord Cd Drag coefficient CP Pressure coefficient cp Gas specific heat at constant pressure cv Gas specific heat at constant volume D Diffusion term D2 Second diffusion term D(Qi ) Artificial dissipation terms Dest Destruction term DISS Artificial dissipation vector .
Meridional M Mach number Mn0 Inlet absolute Mach number at the blade Mni Inlet relative Mach number at the blade Mncopt Optimum Mach number at the blade inlet ml Minimum loss condition n Normal vector. j.LIST OF SYMBOLS e Total energy per unit of volume. slope factor N Rotational speed xxvi . time integration. Euler E. G Flux vectors F~ Force ei Internal energy F~ Body forces H Specific total enthalpy i Incidence angle i. F. k Directions of the coordinate sytem (i0 )10 Zerocamber incidence angle (iD − i2D ) Correction to account for two dimensional effects k Constant used in the Venkatakrishnan’s limiter function kbl Blockage coefficient kδ Thickness correction for zero camber deviation angle ki Correction factor l0 Characteristic length L State of the controlvolume on the left m Slope factor for minimum loss deviation.
Rotor. Absolute velocity . Distance along blade edges. Transition term Tt Total or stagnation temperature t/c Blade thicknesschord ratio U Peripherical velocity uτ Friction velocity V Volume. Magnetude of vorticity. Stator t Time. xxvii State of the controlvolume on the right. Area vector. Radii Rc Radius of curvature ~r distance from the cell/elementcentroid to the face midpoint of the cell/element Re0 Reynolds number Re0 Reynolds number calculated by dimensionless variables s Blade spacing S Constant for the Sutherland law equation. Thickness T Static Temperature. Tangential vector. Riemann invariants terms.LIST OF SYMBOLS P Static pressure Pe Sum of the convective fluxes Pt Total or stagnation pressure Pr Prandtl number. pressure ratio P rt Turbulent Prandtl number P rod Production term qj Heat transfer vector Q Conserved variable vector R Gas constant.
Parameter of the implicit residual smoothing. z coordinate system z Is the calculation location along the compressor axial distance z0 Is the reference axial location y+ Turbulent dimensionless distance yp Distance between the first node on the wall and the wall surface W Relative velocity xxviii Greek Characters α Flow angle β Blade angle ∆t Timestep δ parameter used in the Harten entropy correction. y. Deviation angle δ∗ Boundarylayer displacement thickness (dδ/di)2D Slope at reference incidence (δ0 )10 Zerocamber deviation angle at reference minimum loss incidence angle deduced from low speed cascade data for 10 percent thick ε2 Parameter used in the Venkatakrishnan’s limiter function. Deflection angle ε2 Parameter of the secondorder artificial dissipation . Blade deflection angle ϕ1 Meridional streamline inclination angle relative to the axial direction ∞ Freestream condition ε Streamline slope.LIST OF SYMBOLS Vm Meridional velocity x.
LIST OF SYMBOLS ε4 Parameter of the fourthorder artificial dissipation φ Limiter γ Specific heat ratio Γ Blade sweep λ Second dynamic viscosity coefficient Λc Convective flux Jacobian κ von Karman constant µ Dynamic viscosity coefficient ~n Normal vector ν kinematic viscosity coefficient ∇ Gradient operator ∇· Divergent operator ∇2 Laplacian operator θ Blade camber angle. Flow angle relative to the boundary ν˜ Modified eddy viscosity coefficient ζ Pressure gradient sensor ψ Source term νt Turbulent kinematic viscosity coefficient ω Angular velocity ωp Total pressure loss coefficient ρ Density σ Blade Solidity θ Angle of periodicity. boundarylayer momentum thickness τij Shear stress tensor xxix .
Time integration step neig Neighbor nf aces Number of faces . k Vector components. Directions in internal k face L State of the controlvolume on the left m Meridional min Minimumm n Normal direction. j.LIST OF SYMBOLS ζ xxx Pressure gradient sensor Subscripts 0 Reference value 1 Inlet 2 Outlet B Blockage c Centrifugal CO Coriolis e Euler equation term f Fluctuation term ghost Ghost element H Hub location in Internal (adjacent neighbor) i.
Components of velocities R State of the controlvolume on the right.LIST OF SYMBOLS xxxi out Outlet x. Radial s Stall r Radii v Viscous equation term t Fluctuation term T Tip location (casing location) w Wall Superscripts 1 Inlet 2 Outlet + Positive eigenvalues − Negative eigenvalues ∗ Nominal ¯ Average 0 ” Reynolds average fluctuations Favre fluctuations ˜ Massaverage mean ˆ Dimensionless . y. z Cartesian coordinates.
In this work. was developed to solve the threedimensional NavierStokes equations in steadystate condition. the compressor designer can enhance the compressor blade geometry and rows matching aiming efficiency improvement. A long way is necessary to reach an accurate and reliable result on turbomachinery numerical simulations.1 Introduction 1. it is possible to analyze the physical aspects of the flow at near wall regions. bladetoblade and spanwise. Computational Fluid Dynamics (CFD) techniques are applied to calculate pressure. . With these results. the boundarylayer behavior and the regions with high flow losses. With the flowfield calculated along the turbomachinery streamwise. temperature and velocity distributions of axialflow compressor and turbine channels. A computational code written in FORTRAN. hence many other objectives should be reached before understanding the complex process that involves compressor and turbine design and their performance analysis.1 Objectives The goal of this work is to develop a tool to calculate the flowfield within axialflow turbomachines.
developed by Tomita [1] calculates the preliminary design of axialflow compressors and its performance. The three most important components of gas turbine engine are: compressor. Techniques to improve offdesign operation. These results can be used as input data for a streamline curvature program developed by Barbosa [2]. diameters. Any small gain in turbomachinery efficiency and performance translates into a major economic worldwide. By using a streamline curvature program it is possible to calculate all compressor geometry dimensions (annulus. A study of water injection on the compressor inlet and of the compressor poststall behavior has been developed.2 33 Motivation Gas turbines are used in various applications as: aerospace. Variable Stator Vanes (VSV) and BleedofValve (BOV) are implemented in AFCC. Industrial gas turbines to energy generation are also vastly applied. The use of turbomachines is enormous. named Axial Flow Compressor Code (AFCC). marine and land vehicles. becoming possible to obtain the compressor map.CHAPTER 1. The development of numerical tools increased dramatically as computers became more powerful in the last decades. INTRODUCTION 1. length) using streamlines along the blade spanwise. The designer should design and test each one of these components using experimental and numerical tools. Computational simulations are a promising means for alleviating the cost of the timeconsuming and expensive experimental process. The Gas Turbine Group at ITA has developed several computational codes. Experimental processes generally are very expensive. A computational code developed by Jesus [3] can be used to design and performance anal . the cost on test facilities and the components manufacturing can be significantly reduced. blades. combustion chamber and turbine. A computational program. such as Variable Inlet Guide Vanes (VIGV). If the designer starts the experimentation upon an accurate preliminary design.
major progress has been made in areas such as grid generation. Silva [5] implemented the transient module in the GTAnalysis code. named GTAnalysis. there are many CFD packages available to calculate the fluid motion for both internal and external flows. the last option was chosen. In particular. turbojet. twin or triple spool installations for both aeronautical and industrial interests considering steadystate operation. turbulence modelling. The goal of this work is to study axialflow turbomachines flowfield using CFD techniques. This engine deck is capable of calculating the engine performance of several gas turbine configurations such as turboshaft. pre. gas generator with free turbine. A robust objectoriented modelling of the engine deck framework is presented in references [6] and [7].3 1.CHAPTER 1.and postprocessing computer architec . for industry this is a very good option to obtain a fast problem solution with low cost. Considering that the historical of the Gas Turbine Group at ITA is to develop their own computational tools for academic research. Two options are possible: either to use a CFD commercial package or to develop an inhouse computational code. With an inhouse CFD code other researchers would be able to implement new techniques to improve the design process.3. 1.1 Previous Developments CFD in Academic Research and Industry Currently. with several configurations and installations. developed by Bringhenti [4]. to reduce machine time and to allow optimizations. The Gas Turbine Group at ITA also has an engine deck. Over the last decades. boundary conditions. INTRODUCTION 34 ysis of axial turbines. turbofan.
Energy conservation is one of the main goals of windtunnel testing and CFD research.CHAPTER 1. quickly sort through a large number of possible design modifications and present the best solution. mainly to racing cars. With the advance of computer hardware and software. among others. decreasing drag and increasing aerodynamic efficiency. to increase the thermal efficiency of engines and to improve the car aerodynamics [9]. jets. The computer can analyse the aerodynamic forces on the aircraft’s surface. moreover. gas and oil extraction and process. Called Computational Fluid Dynamics it uses highspeed computers to generate mathematically the flow of fluid over a computerdesigned model. In aeronautical and aerospace areas. The boundary conditions encountered in turbomachinery are among the most complex in CFD [8]. Most improvements of CFD techniques and codes are due to academic efforts or partnership between academia and industry. for example to simulate aerodynamic improvements. thermal comfort. CFD provides a complementary tool for simulation. among others. For academy and industrial purposes it is a powerful tool to study a specific flow behavior as multiphase. INTRODUCTION 35 ture. analysis of complex threedimensional flows experimentaly inaccessible. weather forecast. ablation of reentry vehicles. design optimization and. numerical simulations became widely used as well as experimental tests. fuselage and nacelles. component cooling. The last generation of CFD codes applied in aeronautical engineering are powered by an optimization technique [10] to improve the geometrical design of airfoils. highlift configurations. CFD has been used more and more. ice formation at leading edge of wings. A course of PROPESA [11] expressed this development as: ”An electronic version of a wind tunnel is emerging as a useful tool for aerodynamistis. chemical reaction. CFD has been vastly applied to the automotive industry. porousmedia. .
A typical turbofan engine with all components is show in Figure 1. a large memory storage and number of processors are necessary [12]. each component is simulated individually [12. even a one percent improvement in fuel efficiency would save millions of liters of fuel”.5 million. combustion chamber and turbine efficiencies. 13] including inlet and outlet ducts.1. Time: 150 years. To simulate all gas turbine components simultaneously. Problems in geometrical nacelle design will cause an engine performance degradation because the compressor efficiency decreases due to the flow conditions at fan inlet. 17. 1. Time: 2 years. Another point is that some flight conditions are very difficult to reproduce during the gas turbine tests in a test facility. [16. Generally. 15]. INTRODUCTION 36 Commercial airlines in the United States consume about 40 billion liters of fuel every year. • CFD Design: 800 different wing geometries analysed. such as the gas turbine nacelle [14. • Wind Tunnel Design: 800 different wing geometries tested. Cost: £0. With the advance of CFD many situations such as take . One example of the use of CFD as a design tool is the design of the A330/A340 Airbus [11]: • Stringent Design Target: 10% reduction in cruise drag compared with the A310 and 33% reduction in cruise drag compared with the MD11. 18].3.CHAPTER 1. Cost: £65 million.2 CFD on Gas Turbines A high performance gas turbine depends on compressor.
The position of the engine installation on an airplane can be carefully studied using CFD mainly to understand the problems with undesirable drag forces generation combined with parasite drag forces generated by the fuselage. decreasing the internal flow losses by components geometry optimizations. landing and cruise are calculated with high accuracy and reliability. . INTRODUCTION 37 FIGURE 1. The mixing flow in a turbofan engine at the exhaust duct or noise suppressor can be calculate and optimized to decrease the flow losses.CHAPTER 1.1 – Gas turbine engine and its components off. Tests involving high altitude and ice formation at nacelle lip surface or at wing leading edge are another examples of complex and expensive tasks to be done experimentally. CFD has been used to improve gas turbine components design. The noise from fan blade and from exhaust systems can be calculated using the CFD tools coupled with the empirical correlations. Some studies at RollsRoyce [19] involving CFD on gas turbine engines use the optimization of the guide vanes in the bypass duct to minimize excitation of the fan rotor using sensitivity gradients.
CHAPTER 1. Fluid Mechanics: Experimental Inviscid Flow Control Systems Viscous Flow Turbulent Flow Gas Dynamics Numerical Methods CFD Turbomachinery: Research Applied and Pure Mathematics Development Aero Thermodinamic and Heat Transfer Design Manufacturing Maintence Solid Mechanical and Vibrations Manufacturing Engineering Acoustic Material Science FIGURE 1.3 38 CFD on Axial Compressors and Turbines Modern multistage turbomachines are already the result of a highly complex design process based on the designer expertise.3.2 – Multidisciplinary team on turbomachinery design . aerodynamic and structural loads (pressure and temperature). INTRODUCTION 1. Modern techniques to design turbomachinery need a detailed work division as show in Figure 1. All turbomachinery components are affected by aerodynamic requirements. heat transfer mechanisms and material properties. supported by numerous computational resources (hardware and software) and still finally developed and improved by various tests. The structural loads should be satisfied by the life targets. The turbomachinery field has multidisciplinary nature [20] and a multidisciplinary engineering team becomes indispensable.2. CFD has fostered an unified approach to turbomachinery analysis and is used as design tool.
because their net power output is the reference between the turbine work and the compressor work. neural network and inverse design can be used to enhance the preliminary design. however it is possible to improve the design. Firstly. CFD is the only tool available that provides detailed flowfield .CHAPTER 1. whereby the blading is assessed using CFD codes. Advances are still possible. nowadays. But. Factors influencing efficiency are extremely complex mainly for high pressure compressor due to the difficulties during the rotor and stator rows matching. INTRODUCTION 39 Gas turbine design components are based on design by analysis. since at each stage the geometrical data for the necessary aerodynamic or stress analysis must be determined. not only in the efficiency itself but also in the amount. Reminding that. expensive and time consuming. The advent of modern numerical methods applied both whole engines and to individual components. During the design stage the designer repeatedly adjusts the blade geometry and channel until finding a suitable geometry that combines acceptable aerodynamic performance with low stress levels. This process of continual refinement of the match geometry can be tedious. A threedimensional flowfield calculation inside turbomachinery is a long and complex task. and it is practical and economic to manufacture. To obtain these parameters it is necessary to know the machine designed and its operational characteristics. Optimization techniques such as genetic algorithms. because it is necessary to previously know some important points of the compressor operation to setup correct parameters in the boundary conditions. specially for gas turbines whether used for aircraft propulsion or for landbased power plants. with CFD techniques it is impossible to design an axial compressor. in many instances. The flow through turbomachines is one of the most complicated in the field of fluid dynamics practice. and hence cost. Efficiency is probably the most important performance parameter for most turbomachines.
Decreasing these losses means to increase the compressor efficiency. specifically high bypass ratio turbofan engines. stream surface. Flow properties may be calculated for a prescribed compressor geometry and a set of boundary conditions. Certainly. endwall losses and others). al. However. The power required to increase the total pressure from the compressor inlet to the compressor outlet is provided by the turbine. in many cases. expensive and. INTRODUCTION 40 information. The flowfield within advanced axial compressors for both aircraft propulsion systems and industrial gas turbines. shock waves. this work is not trivial.CHAPTER 1. momentum and energy).. impossible. during the energy transfer process the flow losses are inevitable [23. Is difficult to quantify separately all flow loss sources (secondary flow. solving the system of differential equations usuaaly called conservation equations (mass. shock boundarylayer interactions and effects of threedimensionality and unsteadiness of the flow. friction losses. The tools used to determine . A theory of threedimensioanal flow of turbomachines using streamline. profile losses. if possible. 24]. profile losses. is characterized by the presence of mixed subsonic. transonic and supersonic regions. Actual testing of turbomachinery with detailed measurements in rotating blade rows passages is cumbersome. stream function is presented by Xu et. shock wave losses. to change the compressor flowfield by geometrical configurations. In nature. The flow within an axial compressor has some important points to be discussed related to its physical aspects. [21]. in which the compressor is attached by a shaft. decreasing the entropy generation. Current efforts from the CFD community have been very important to understand the flow for both external and internal cases and. al. the velocity field moves from a high pressure region to a low pressure region. Xu et. also present the methods of solving inverse design problem on the stream surfaces [22]. flow separation mainly at blade suction surface. flow leakage at blade tip region.
An incorrect pressure ratio can take the compressor to operate in an instability range [27]. The flow behavior when the compressor operate in an instability point is completely unsteady. In the work developed by Dawes [30]. The flowfield inside a singlestage axial compressor was calculated with and without the stator. such as compressor or turbine. the boundary conditions will be unrealistic. INTRODUCTION 41 the preliminary design should be very robust [25. Dawes [28] presents a work based on unsteady flows associated with the interblade row interactions in compressor stages. Celestina and Adamczyk . For unsteady flows in turbomachinery applications Chen. Compressor instable operation push the machine to work in surge or choke points. all numerical treatment given between the rows interface must be different compared to the numerical implementation for steadystate condition. accurate and reliable. because if the pressure ratio calculated during the compressor design is not close to the actual pressure ratio (obtained in the test facility). describe the difficulties to simulate high pressure compressor due to the ”numerical surge” caused by the complexity in start the CFD calculations for compressors because the blades can be operating near stall point. have its particularities. including the numerical treatment at inflow and outflow boundaries due to the high flow perturbations. The variation of static pressure along the outlet rotor spanwise is clearly and sometimes the imposition of a constant static pressure or simple radial equilibrium not represent correctly the actual values due to the tipclearance and static pressure drop at casing.CHAPTER 1. Hence. the calculation may fail as a result of the transient induced by the initial guess rather than because of a genuinely unstable operating point. 26]. The centrifugal and axial compressors were studied. Denton [29]. Each component. is possible to observe the difficulties in obtain boundary condition at compressor outlet due to the complex physical aspects. Hence.
36. porousmedia. incompressible or compressible. CFD codes are developed specially to solve specific problems [33. low or high Reynolds numbers. Segunpta [38] develop a new compact finitevolume scheme based on a fluxvector splitting discretization. This scheme has been analyzed by matrixspectral analysis developed by author. Some codes shown in [32] were unable to predict correctly the highly threedimensional secondary flow.CHAPTER 1. This report presented a large amount of information about the difficulties mainly to predict the pressure loss (was up to 40% in error). with or without chemical reactions as dissociations that occur in the hypersonic flows. among others. internal or external flows. Some authors [35. 37] studied specific methodologies to calculate regions as shock waves with high accuracy. INTRODUCTION 42 [31] presents two unsteady wakeblade row interaction models and a rotor. Yee [39] present a mathematical formulation for highorder schemes using explicit and implicit multidimensional compact highresolution shockcapturing methods for Euler . For complex geometries and flows the mesh generation and numerical implementation become more difficult. There are specific theoretical and numerical formulations for each case.stator unsteady interaction models. unsteady or steadystate. Several researchers on turbomachine area participated in the AGARD Propulsion and Energetics Panel set up Working Group 26 as reported in [32] to help to clarify the issues involving CFD applied to axial compressors using as wide a range of codes as possible of two representative single blade row cases: NASA Rotor 37 and an annular turbine cascade tested by DLR. These wakeblade row models had produced results that were qualitative agreement with the results from the rotorstator interaction model and quantitative agreement was reasonable and could be improved upon by further developing improvements in the inlet boundary conditions employed in the timeshift model. 34]. For example.
In this work. a threedimensional transonic flow inside an isolate compressor blade row is presented. The progress towards threedimensional flow solvers is reported by Frink [44] in detail. 49. For compressor and turbine cases some specific numerical treatment is necessary because two frames of reference must be used: rotating and nonrotating. 47. Frink presents an unstructured grid solver that upgrade its formal accuracy to second order to solve NAvierStokes on tetrahedral cells. A comparison about the use of the CFD codes developed to calculate the conservation laws using these two different reference frames based on absolute and relative . For turbomachinery flow simulations some modifications in the NavierStokes equations are necessary as discussed in forthcoming chapters [46. A method to solve turbulent flow problems on threedimensional unstructured grids with cellcentered spatial discretization and an implicity backwardEuler timestep scheme is presented by Frink [45]. CFD techniques applied to turbomachines are vastly used to study the flowfield within components such as the compressor and the turbine aiming optimization. 48. Lacor and [42]. The ENO3 scheme with the lowerupper symmetric GaussSeidel (LUSGS) algorithm is adopted to improve the computational efficiency. A rotor row is treated within the rotating frame of reference and a stator row within the nonrotating (or stationary) frame of reference. including turbulent viscous flows. Chima and Liou [43] present other spatial dicretization schemes based on highorder methods. Yang [41]. For a compressor blade row. a thirdorder accurate highresolution scheme is presented by [40].CHAPTER 1. INTRODUCTION 43 equations. 50] due to the additional terms to represent the centrifugal and Coriolis forces. The results provides good resolution of the shock system and the wakes in flow. 8. The accuracy is increased by a pseudo–Laplacian weighted averaging algorithm which produces robust convergence and permits highorder boundary conditions. In this work.
In [53] it is presented a threedimensional analysis of the flowfield within a twostage fuel turbine used on the space shuttle main engines. In the Master of Science program. The thesis might produce data for the establishment of loss models to be incorporated to the streamline curvature code.CHAPTER 1. 1. 57] was also discussed [58]. INTRODUCTION 44 velocities can be found in [51]. A big network organized by the European Community called QNETCFD [54] presented several publications of turbomachinery flowfield simulations [55]. The thesis must contribute with the improvement of the axial design procedures in use in the Gas Turbine Group at ITA. The use of the AFCC (Axial Flow Compressor Code) and GTAnalysis (Gas Turbine Analysis) . the meanline and streamline curvature methods were vastly studied and applied. 59. Turbulent modelling for turbomachinery [56. 2.4 Author Contribution The three main requirements of the supervisor are: 1. In [52] it is reported a twodimensional flow through a blade row (with rotating and stationary cascades) of a compressor fan using a formulation based on bladetoblade stream surface for one passage of a blade row. Develop the CFD code in a modular basis to serve as a platform for future studies. Some results of compressor design and performance analysis calculated using AFCC program can be found in references [1. allowing additional blade profiles. the author worked with compressor design and offdesign performance calculations. 3. 60]. Hence.
The most effective design charts for use in the preliminary design process are those that allow the engineer to rapidly identify designs that are unacceptable or provide guidance on the route towards the design optimization. turbomachinery designers must understand specific behaviors of the machine to design improvements in order to decrease loss sources. fuselage optimization and inverse problem applied to the blade design. it is of great importance to own source codes in order to form human resources capable of . However.EFEI to develop a module to optimize the design of axial compressors using Sequential Quadratic Programming (SQP).CHAPTER 1. drag minimization. The question is: why not to use a commercial package? Design tools that combine CFD and optimization techniques. need some special attention during the numerical implementation process and sometimes only with the CFD source code it is possible to implement these techniques. The developing process of this kind of computational code is very cumbersome. The AFCC program was used in the Doctoral Program in the Faculdade de Engenharia de ITAJUBA . Several problems in the applied engineer can be solved using these packages. In most cases. for an autonomous research laboratory that works with the stateofart on gas turbine engine design. INTRODUCTION 45 programs to predict the gas turbine engine performance using VIGV (Variable Inlet Guide Vanes) and BOV (BleedofValve) can be found in [61. The purpose here is to develop a tool to calculate threedimensional flowfield within axial turbomachines. The results are presented in reference [63] and the engine cycle improvement using this optimization technique in the axial compressor when it is installed in a gas turbine engine is presented in [64]. in academia. already validated and verified they are user friendly due to the Graphic User Interface (GUI). 62]. for example. Commercial packages are widely used in industry. Another option would be to use a commercial package.
reliability.CHAPTER 1. an engine deck was developed to become possible the studies involving gas turbine performance with configurations as : gas generator with free turbine. gas turbines is a strategic issue mainly due to the strong growth in energy consumption. Obtained the turbomachinery geometrical dimensions from the design tools a Computer Aided Design (CAD) softwares is used to generate the geometry. Besides. Firstly. During many years the Gas Turbine Group at ITA. has developed numerical tools to design and performance calculations of gas turbines components as compressors and turbines. A stepbystep design procedure is presented in this work. turbomachines has a complex geometry mainly on the compressor blade profiles. In Brazil. a long way is necessary to start the CFD calculation. The Gas Turbine Group at ITA worries about tools developing capable to assure success in the gas turbine components design process. The design of compressor or turbine is a stressed interactive process between the designer and the computational code due to the high number of parameters to be analyzed and that it should be considered to obtain a reliable design. the turbomachinery preliminary design was carried out and later the design refinement. This is the time to create the human resources to work in the projects involving new families of engines for the energy area aiming high efficiency. This causes many difficulties spending a long time during the mesh generation process. However. low cost and low pollutant emissions. It is very important to mention that to a complete understanding of the flow physics in compressors and turbines the designer should have a solid knowledge on gas turbine components operation. nowadays Center of Reference on Gas Turbines. INTRODUCTION 46 continuing the scientific development of the computational codes. . turboshaft. turbofan and turbojet with single or multiple spools. Generally.
A rotatingframe of reference is added to the momentum equations to quantify the additional forces that occur in the rotating region. The turbulence equation discretization is presented for both time and spatial integration. are treated in dimensionless form. This chapter starts with the ideal gas formulation. both dependent and independent. to accelerate the timemarching process are discussed. The stepbystep calculation and the implementation process involved in . An oneequation SpalartAllmaras turbulence model is also implemented and described. Chapter four describes the structure of the computational code and its implementation issues. Chapter one describes the importance and motivation of applying CFD to simulate turbomachine flows.5 47 Work Organization This work is divided in eight chapters. Damping functions as artificial dissipation models to avoid numerical instabilities for centered schemes and numerical methods such as Implicit Residual Smoothing (IRS) and variable timestep. whose authors and works are mentioned in this thesis. Chapter three describes the numerical formulation of the governing equations.CHAPTER 1. fluxvector splitting and fluxdifference splitting) and two time integration (RungeKutta and MacCormack) are presented. The general equations of fluid mechanics are obtained using both ReynoldsAverage NavierStokes (RANS) and Favre average definitions. Three discretization methods of the spatial integration (centraldifference. All variables. The research areas of the Gas Turbine Group at ITA has been previously described. The unstructured mesh treatment is briefly commented. INTRODUCTION 1. The implementation of the initial and boundary conditions and the treatment of the rotorstator interface are also described. Chapter two describes the theoretical formulation of the governing equations.
The nozzle used in the Euler equations validation was also simulated using turbulent flow. A stepbystep procedure is described to verify the convective and diffusive momentum equations terms. Euler and turbulent flow were calculated using different mesh types (Hgrid and Ogrid). The flatplate numerical results were compared with the analytical solution of Blasius. The flow on the stator outlet and rotor inlet is evaluated with the use of mixing . The results of turbulent flow simulation inside the nozzle were improved on the region after the nozzle throat when compared to the Euler equations results. considering the transonic regime with Mach number equal to 0. Cases involving turbines and compressors were simulated.8. Chapter six describes the results for turbomachinery flow simulations.CHAPTER 1. The centered and upwind spatial integration schemes were vastly tested to analyze the code numerical stability. The upwind scheme was also calculated. The mesh generation process and the mesh types as Hgrid. INTRODUCTION 48 the solver development and some comments and also recommendations of the author are presented. A NACA0012 airfoil with zero angleofattack was also simulated. A esinglestage axialflow turbine. HOHgrid and Ogrid are presented. a single compressor rotor with low aspectratio and a full multistage axialflow compressor design procedure and CFD calculations were carried out. The implementation of diffusive terms were tested with laminar flow on a flatplate for different artificial dissipation models for centered spatial integration scheme. The study case involving a singlestage axialflow turbine presents the interaction of turbine design methodology and CFD techniques to improve the flow through the turbine rotor. using two spatial integration schemes (centered and upwind) and different pressure ratio values. Chapter five describes the code verification and validation. The Euler equations were calculated for a supersonic nozzle.
losses due to Reynolds number effects) to account the flow viscous effects inside the compressor are calculated and calibrated based on the designer expertise. a multistage axialflow compressor was designed and analyzed in four steps: • Preliminary design: based on the meanline technique. friction losses. At this point. the rotor and stator blades. A CAD software is used to generate the 3 − D view of the compressor. Tools as VIGV (Variable Inlet Guide Vanes) and BOV (BleedofValve) are considered to improve the operation range in the offdesign condition. These analyzes are necessary to start the compressor design refinement using the streamline curvature method. The compressor is designed and its performance is calculated for different rotational speeds. INTRODUCTION 49 plane. Based on a 1MW gas turbine engine. After the geometrical modifications an Ogrid was generated with tipclearance between rotor tip surface and turbine casing to verify the tip leakage phenomena.CHAPTER 1. Loss models (secondary losses. The rotor blade geometrical changes are described to solve the problem of reverse flow on the rotor suction side. • Mesh generation: Once obtained the compressor geometry. such as inlet and outlet blade angles including the compressor annulus dimensions are determined. • Streamline curvature method: used to calculate all compressor geometrical dimensions based on the streamlines distribution on the blade leading and trailing edges from hubtotip. The gas turbine engine deck is used to calculate the engine running line on the compressor map. profile losses. Different types of numerical initialization to start the CFD calculations are discussed. the mesh generation .
A brief discussion is done about the importance of the CFD solver as research and teaching tool. • CFD calculation: Based on the streamline curvature results the boundary condition and numerical initialization process were carried out to start the CFD calculations. . their difficulties and numerical aspects. Chapter eight describes the future implementations to improve the code robustness from preprocessing data until postprocessing results. General aspects between design methodology and CFD results are discussed for all study cases involving turbomachines.CHAPTER 1. The techniques to improve the CFD solver robustness based on numerical aspects including the mesh generation process are described based on turbomachinery flow simulations. INTRODUCTION 50 process is started and the computational domain is presented considering the use of periodicity condition. Chapter seven describes important comments and conclusions about the cases simulated.
At the usual ranges of pressures and temperatures encountered in today’s gas turbines.1) and it follows that the internal energy (ei ) can be calculated by ei = e 1 − Wi Wi . the fluid is usually the atmospheric air or the products of combustion of oilderived fuels in air. p = ρRT . γ = cp /cv . so that the fluid properties in a gas turbine can be considered similar to those of the air.2 Mathematical Formulation 2. cp − cv = R. air behaves as a newtonian perfect gas with temperaturedependent specific heats and viscosity. It is observed that the amount of fuel burned in gas turbines is around 2% of the air mass flow entering the combustion chamber. ρ 2 (2. Bearing in mind that the total specific energy (by volume) is given by →2 1− e = ρ ei + W  . 2 (2.2) From the perfect gas equation and useful definitions.1 The fluid Aiming at the flow calculation in gas turbines parts like compressors and turbines.
it follows that p = (γ − 1)ρei . (s + T ) (2. b = 1. MATHEMATICAL FORMULATION 52 and ρei = cv T .7) where.3) assumed that the specific heat ratio γ is known.9883655 × 10−4 T 4 + 1.0126885 × 10−10 T 3 + + 4. From the calculated T . the fluid properties can be calculated [65]: cp = cp (T ) = − 8.6) The molecular dynamic viscosity is calculated by Sutherland law: 3 T2 µ=b . P r ∂xi (2. The total enthalpy is given by H = ei + →2 →2 p 1− 1− + W  = h + W  . so that ρ= p .2961648 × 10−5 T. (2.458 × 10− 6 and s = 110.9) . cv = cv (T ) = R − cp(T ).5) (2. RT (2. ρ 2 2 (2.739514 × 10−8 T 2 − 3.4. (2.8) − The heat conduction is calculated by Fourier law and is represented by the heat flux → q: µ ∂h → − q = −κ∇T = − .4) An iterative process is need to get the approximate value of γ.CHAPTER 2.
CHAPTER 2. MATHEMATICAL FORMULATION
53
where P r is the Prandtl number and κ is the thermal conductivity coefficient given by:
κ =9.0534 × 10−16 T 4 + 5.1523 × 10−12 T 3 + 2.9869 × 10−8 T 2 +
+ 8.5405 × 10−5 T + 2.6321 × 10−3
2.2
(2.10)
The flow
High performance gas turbines require very fast fluid flows inside the compressors and
turbines blade passages. Blade passage are channels that deflect the flow usually a fixed
followed by a rotating (in turbines) or a rotating followed by a fixed (in compressors).
This construction characteristic causes the flow, entering the fixed cascade that follows
a rotating one, to be locally transient, even so it may be considered steady globally, as
when the turbomachine is run at constant speed. Gaps existing at the top of the rotating
passages allow fluid flow from one passage to the other. Experience has shown that the
flow is highly 3 − D, viscous and turbulent, so that, in this work, it is modelled using
the full equations os conservation of mass, momentum and energy, written in a rotating
frame of reference. Hence, the relation of absolute and relative velocities is represented
by Figure 2.1.
FIGURE 2.1 – Schematic velocity vector diagram
~ is the vector of relative velocity, V~ is the vector of absolute velocity and U
~
where W
CHAPTER 2. MATHEMATICAL FORMULATION
54
~ =ω
is the vector of tangential velocity, given by U
~ × ~r. Hence, from 2.1 one can obtain
~ +ω
V~ = W
~ × ~r,
with
(2.11)
−ωy
ω
~ × ~r = ωx
.
0
(2.12)
Using the Einstein notation, the conservation equations are defined by
• Mass
∂
∂ρ
(ρWj ) = 0,
+
∂t ∂xj
(2.13)
∂
∂
∂p
∂τij
(ρWi ) +
(ρWi Wj ) +
−
+ Ψ = 0,
∂t
∂xi
∂xi
∂xj
(2.14)
∂
∂e
+
[(e + p)Wj − τij Wi + qj ] = 0.
∂t ∂xj
(2.15)
• Momentum
• Energy
The viscous stresses are considered linear and proportional to the rate of strain, and are
calculated using the newtonian fluid theory, hence:
τij = µ
∂Wi ∂Wj
+
∂xj
∂xi
+ λδi,j
∂Wk
∂xk
,
(2.16)
where µ is the molecular dynamic viscosity and λ is the second coefficient of viscosity. In
this work the Stokes hypothesis is used, hence λ = −2/3µl . The variable Ψ in equation
CHAPTER 2. MATHEMATICAL FORMULATION
55
2.14 represents some terms related to the rotation of the fluid flow in rotating channels.
Details of the source term are presented in the Annex A.
The complete form of the momentum equations are, therefore:
∂
∂
∂
∂
(ρWx )+ (ρWx Wx +p−τxx )+ (ρWy Wx −τyx )+ (ρWz Wx −τzx ) = ρω 2 x+2ρωWy ,
∂t
∂x
∂y
∂z
(2.17)
∂
∂
∂
∂
(ρWy )+ (ρWx Wy −τyx )+ (ρWy Wy +p−τyy )+ (ρWz Wy −τzy ) = ρω 2 y −2ρωWx ,
∂t
∂x
∂y
∂z
(2.18)
∂
∂
∂
∂
(ρWz ) +
(ρWx Wz − τzx ) +
(ρWz Wy − τyz ) + (ρWz Wz + p − τzz ) = 0, (2.19)
∂t
∂x
∂y
∂z
where ω is the angular velocity.
Analyzing the magnitudes of the variables involved in the above equations it is seen
that they differ by several order of magnitudes. Bearing in mind their numerical solution,
it is possible to avoid the numerical errors related to those differences in numerical values
by normalizing the variables. This is done through the nondimensionalization equations
ρ
,
ρ0
x
xˆ = ,
l0
y
yˆ = ,
l0
z
zˆ = ,
l0
ρˆ =
ˆ i = Wi ,
W
a0
T
Tˆ = ,
T0
e
eˆ = 2 ,
ρa0
µ
µ
ˆ= ,
µ0
κ
,
κ0
a0
tˆ = t ,
l0
ˆ= λ,
λ
µ0
P
Pˆ =
,
P0
κ
ˆ=
and the Reynolds number is calculated by Re0 = ρ0 a0 l0 /µ0 for internal flow computations
~ l0 /µ0 for external flow computations with the Mach number M∞ =
and Re0 = ρ0 W
~ /a0 , where the subscripts 0 and ∞ means referential properties from the stagnation
W
conditions and freestream conditions, respectively. These new defined parameters are
CHAPTER 2. MATHEMATICAL FORMULATION
56
called
• ρˆ = dimensionless density,
• xˆ = dimensionless coordinate of a specific point, the same rule is valid for y and z
coordinates,
ˆ i = dimensionless velocity,
• W
• Tˆ = dimensionless temperature,
• eˆ = dimensionless total energy,
• µ
ˆ = dimensionless molecular dynamic viscosity,
• κ
ˆ = dimensionless thermal conductivity,
• tˆ = dimensionless time,
ˆ = dimensionless second viscosity coefficient,
• λ
• Pˆ = dimensionless pressure,
• Re0 = Reynolds number calculated using dimensionless variables.
After the transformation of variables indicated above and collection of terms, a similar
set of equations of conservation is obtained. Details are not supplied here.
2.2.1
Turbulence
Turbulence as understood today states that the flow properties may be represented as
the sum of mean property and a fluctuation of the property:
A = A + A0 ,
(2.20)
H=H e = e˜ + e”. Wi = W ˜ + H”.24) qi = q˜i + qi0 .24 to the nondimensionalized equa . (2.23) t t+T t where ρ is the Reynolds average of ρ (density) and T is a time interval. with the definitions. the Reynolds Averaged NavierStokes (RANS) and Favre average methods are used.22) ρAdT . p = P + p0 . (2. MATHEMATICAL FORMULATION 57 A = A˜ + A” (2. In this work. T = T˜ + T ”. in this work the averaging procedures use the definitions fi + Wi ”. respectively. Applying the averages indicated by equations 2. 1 A¯Reynolds = lim T →∞ T Z eF avre = 1 lim 1 A ρ T →∞ T Z t+T AdT .21) where the overbar and tilde stands for the mean and the prime and double prime stands for the fluctuation components of the property A. (2. ρ = ρ + ρ0 .CHAPTER 2. There are different ways to extract the average values. Following successful applications reported in the literature [66].
28 the over hat stands for nondimensional variables and the overbar for the averages (RANS or Favre) as indicated by equations 2.26) ∂e ∂ ρWi ”Wj ” f f + (e + P )Wj − Wi (τij − ρWi ”Wj ”) − Wi ”(τij − ) + qj + ρWj ”h = 0 ∂t ∂xj 2 (2.25) ∂ f ∂ f f ∂P ∂ (ρWi ) + ρWi Wj + − (τij − ρWi ”Wj ”) + Ψ = 0 ∂t ∂xi ∂xi ∂xj (2.23. the following equations will result: ∂ρ ∂ fj ) = 0 + (ρW ∂t ∂xj (2. The flux vectors are made of ”nonviscous”.22 and 2. MATHEMATICAL FORMULATION 58 tions. ”viscous” and ”turbulent” parts. In equation 2. it is possible to write this ”turbulent” system of partial differential equations in matricial form ˆ ∂ Eˆ ∂ Fˆ ∂ G ˆ ∂Q ˆ = 0. F and G the flux vectors and Ψ the source term associated to the rotation of the frame of reference. 2 • qTi = ρWi ”h” = usually named the molecular transport of heat. as indi .15 except for the following terms associated with the turbulence: • ρτij = ρWi ”Wj ” = usually named the Reynolds Stress Tensor (RST).25.14 and 2. + + + +Ψ ∂ xˆ ∂ yˆ ∂ zˆ ∂ tˆ (2.13. E. 2. • 1 ρWi ”Wi ” = usually names the kinetic energy of the (turbulent) fluctuations. After the nondimensionalization and collection of the appropriate terms.26 and 2.27) Equations 2.27 are similar to equations 2. 2.28) ˆ where Q is called the vector of conserved variables.CHAPTER 2.
CHAPTER 2. MATHEMATICAL FORMULATION 59 cated by equations 2. (2.30 and 2. The ”viscous” components contain only the terms with the viscosity and are represented by the subscript v. The matricial representations of the terms are. The ”turbulent” components only the terms containing the fluctuations and are represented by subscript t.33) .32) f ˆx ρ¯ˆW f f ˆ xW ˆx + P ρ¯ˆW f f ˆ ˆ xW ˆx Ee = ρˆ¯W f f ˆ xW ˆx ρ¯ˆW f ˆx (e˜ˆ + P )W . (2.31) The convective components are represented by the subscript e. 2. (2. therefore ρ¯ˆ f ˆx ρ¯ˆW f ˆ= ˆy Q ρ¯ˆW f ˆz ρ¯ˆW eˆ˜ .29.30) G = Ge + Gv + Gt .29) F = Fe + Fv + Ft . (2. (2.31 E = Ee + Ev + Et .
(2. (2.CHAPTER 2. (2.34) . (2.35) 0 ˆ x ”W ˆ x ” − τˆxx ” ρˆW ˆ ˆ x ”W ˆ y ” − τˆyx ” Et = ρˆW ˆ x ”W ˆ z ” − τˆzx ” ρˆW ˆ x ” − τˆx (e˜ˆ + P )W f 0 ˆ y ”W ˆ x ” − τˆyx ” ρˆW ˆ y ”W ˆ y ” − τˆyy ” Fˆt = ρˆW ˆ z ”W ˆ y ” − τˆzy ” ρˆW ˆ y ” − τˆy (e˜ˆ + P )W f .37) .36) . MATHEMATICAL FORMULATION Fˆe = Gˆe = f ˆy ρ¯ˆW f f ˆ yW ˆx ρ¯ˆW f f ˆ yW ˆy + P ρ¯ˆW f f ˆ zW ˆy ρ¯ˆW f ˆy (e˜ˆ + P )W f ˆz ρˆ¯W f f ˆ zW ˆx ρ¯ˆW f f ˆ zW ˆy ρ¯ˆW f f ˆz + P ˆ zW ρ¯ˆW f ˆz (e˜ˆ + P )W 60 .
0 τˆxx 1 Eˆv = τˆyx Re0 τˆzx µ ˆ µ ˆ ∂ˆ ei l t f f f ˆ x + τˆyx W ˆ y + τˆzx W ˆz + γ τˆxx W + P rl P rt ∂ xˆ (2.39) .40) .38) 0 τˆyx 1 Fˆv = τˆyy Re0 τˆzy µ ˆ µ ˆ ∂ˆ ei l t f f f ˆ y + τˆyx W ˆ x + τˆyz W ˆz + γ τˆyy W + P rl P rt ∂ yˆ 0 τˆzx 1 Gˆv = τˆzy Re0 τˆzz µ ˆl µ ˆt ∂ˆ ei f f f ˆ ˆ ˆ τˆzz Wz + τˆzx Wx + τˆzy Wy + γ + P rl P rt ∂ zˆ .41) .CHAPTER 2. MATHEMATICAL FORMULATION 0 ˆ z ”W ˆ x ” − τˆzx ” ρˆW ˆ ˆ z ”W ˆ y ” − τˆyz ” Gt = ρˆW ˆ z ”W ˆ z ” − τˆzz ” ρˆW ˆ z ” − τˆz (e˜ˆ + P )W f 61 . (2. (2. (2.
MATHEMATICAL FORMULATION 0 f ρ¯ˆω 2 xˆ + 2ρ¯ˆWˆy ω f ˆ Ψ= ρ¯ˆω 2 yˆ − 2ρ¯ˆWˆx ω 0 0 62 . τˆyf = τˆyy W (2. (2. P rt is the turbulent Prandtl number.46) 2. µl is the molecular coefficient of viscosity.44) f f f ˆ y ”+ τˆyx W ˆ x ”+ τˆyz W ˆ z ”+ τˆyy ”W ˆ y + τˆyy ”W ˆ y ”+ τˆyx ”W ˆ x τˆyx ”W ˆ x ”+ τˆyz ”W ˆ z + τˆyz ”W ˆ z ”.2. P rl is the laminar Prandtl number. During the nondimensionalization and averaging of the variables other definition came out: µ = µl + µt .2 Turbulence Model There is not enough affordable computational power today for the flow calculation using the above equations. since to capture the fluctuations of the flow properties an .45) f f f ˆ z ”+ τˆzx W ˆ x ”+ τˆzy W ˆ y ”+ τˆzz ”W ˆ z + τˆzz ”W ˆ z ”+ τˆzx ”W ˆ x τˆzx ”W ˆ x ”+ τˆzy ”W ˆ y + τˆzy ”W ˆ y ”.42) The variables that have already appeared have the same meaning (but nondimensionalized). (2.9. assumed here equal to 0. τˆzf = τˆzz W (2.CHAPTER 2. τˆxf = τˆxx W τxy W τxz W τxx ”W τxx ”W τxy ”W τxz ”W τxz ”W (2. µt is the turbulent coefficient of viscosity. γ is the ratio of specific heats and f f f ˆ x +ˆ ˆ y τˆxy ”W ˆ z +ˆ ˆ x ”+ˆ ˆ y ”+ˆ ˆ z ”+ˆ ˆ x ”+ˆ ˆ y ”+ˆ ˆ z ”.43) where µ is the actual coefficient of viscosity.
Based on the results of published literature related to the calculation of flows in turbomachines [25. Details of the physics and of the available turbulence models. the oneequation turbulence model developed by SpalartAllmaras [73] was chosen for this work. references [66. it is practice today to model the turbulent parts of the equations. = cb1 [1 − ft2 ]S˜ν˜ + ∇ · ((ν + ν˜)∇˜ ν ) + cb2 (∇˜ ν ) − cw1 fw − 2 ft2 Dt σ κ d (2. ν˜ S˜ = S + 2 2 fv2 . 69. 68. (2.2.CHAPTER 2. Therefore. for example. 67.47) The turbulent kinematic viscosity νT is given by: νT = ν˜fv1 . can be formal in.2. Therefore S= p 2Ωij Ωij . despite not being the scope of this work. χ= ν˜ ν . 8.49) κ is the von Karman constant and S is magnitude of vorticity. to avoid this short coming. valid for high Reynolds number: i ν˜ 2 h 1 c D˜ ν b1 2 + ft1 4U 2 . 71. It is an equation developed for the eddy viscosity ν˜.48) where fv1 = χ3 χ3 + c3v1 . MATHEMATICAL FORMULATION 63 extremely refined mesh is required. and of the model. (2. 2.1 The SpalartAllmaras oneEquation Turbulence Model Reference [73] contains details of the physics behind the model. κd (2. 32]. 72].50) . 70.
1 + c6w3 g 6 + c6w3 16 g = r + cw2 (r6 − r). ft2 = ct3 exp(−ct4 χ2 ). (2.5.54) where 4xt is the grid spacing along the wall at the nominated transition point.CHAPTER 2. . dt is the distance between the local point and the transiton point and wt is the vorticity at the wall at the nominated transition point.1355 cw1 = cv1 = 7. κ = 0. cw2 = 0. (2. . ct4 = 0.52) . . fw = g ν˜ ˜ SK 2 d2 .622 cb1 (1 + cb1 ) + κ σ .1 . MATHEMATICAL FORMULATION 64 where 1 Ωij = 2 ∂Wi ∂Wj + ∂xj ∂xi .53) 4U gt = min 0. ct3 = 1. . cw3 = 2.3 . cb2 = 0. σ= ct1 = 1 .41. . The other constants are given by: 2 3 .1. wt2 2 2 2 (d + gt + dt ) ft1 = ct1 gt exp −ct2 4U 2 . wt 4xt (2.2 cb1 = 0. (2.51) and fv2 χ = 1− 1 + χfv1 r= .
νˆ˜ 1 Sˆ˜ = Sˆ + fv2 2 2 ˆ Re0 κd .56) All other variables are dimensionless by definition. − cw1 fw − 2 ft2 − ∇.(νˆ˜W ˆ κ Re 0 d (2. subscript t stand for transition points. = +W Dt ∂t (2.58. the density variation can be neglected. ˆ κ Re 0 d (2. Applying the definition of material derivative ∂( ) D( ) ~ . For freestream initial condition νˆ˜ is usually taken in the interval [0. The identity W In the equations above. MATHEMATICAL FORMULATION 65 The nondimensional form of equation 2.57) and assuming that on the near wall region the velocity is small.CHAPTER 2.47 becomes i 1 ∂ νˆ˜ 1h = cb1 [1 − ft2 ]Sˆ˜νˆ˜ + ∇ · ((ˆ ν + νˆ˜)∇νˆ˜) + cb2 (∇νˆ˜)2 + σ Re0 ∂ tˆ " #2 h cb1 i νˆ˜ 1 ~ˆ ) + ft1 4U 2 Re0 .(W ~ νˆ˜) − νˆ˜.47 is: i 1 Dνˆ˜ 1h ∇ · ((ˆ ν + νˆ˜)∇νˆ˜) + cb2 (∇νˆ˜)2 + =cb1 [1 − ft2 ]Sˆ˜νˆ˜ + σ Re0 Dtˆ " #2 h cb1 i νˆ˜ 1 − cw1 fw − 2 ft2 + ft1 δU 2 Re0 . ˆ ˜ 2 dˆ2 Re0 SK (2.∇( ). νˆ/10]. The physical process of fluid motion is governed by the NavierStokes equations where . so that the corresponding terms are set to zero outside the transition region. rˆ = νˆ˜ 1 . At walls νˆ˜ = 0. The nonconservative form of equation 2.∇W ~ was used to obtain Equation 2.55) where νˆ = ν µ0 /ρ0 .∇(νˆ˜) = ∇.58) ~ .
∂ tˆ (2. This process is called eddy diffusion.61) P rod(νˆ˜) = cb1 [1 − ft2 ]Sˆ˜νˆ˜. in the inner region the small eddies are formed and predominant. MATHEMATICAL FORMULATION 66 unsteady. The SpalartAllmaras turbulence model presented in Equation 2. (2.60) the diffusion term is D(νˆ˜) = i 1h ∇ · ((ˆ ν + νˆ˜)∇νˆ˜) + cb2 (∇νˆ˜)2 . This phenomenon is called production of turbulence. so the viscous forces become considerable and the smallest eddies are destroyed. C(νˆ˜) = −∇. Large eddies are formed in the outer region. there is a transport of fluid properties that increases the mixing process. σ (2. Due to the eddies motion in a turbulent flow. Basically. diffusion and dissipation of eddies terms including unsteadiness and convection. Each flow field problem has different physical aspects. These eddies are formed due to high velocity gradients. The smallest eddies grow and become large eddies and the inverse process occurs too.(νˆ˜W (2.59) where the convection term is ~ˆ )2 . During this mixing process the smallest eddies dominate the energy dissipation. convection and diffusion terms are considered. The turbulent fluid motion is related to production.62) the production term is .58 can be written in the following form: ∂ νˆ˜ = C(νˆ˜) + D(νˆ˜) + P rod(νˆ˜) + Dest(νˆ˜) + T.CHAPTER 2.
are presented in [73].64) and the transition term is Details of the development of all terms and the calibration of the model. such as the choice of constant values.63) T = ft1 4U 2 . . (2.CHAPTER 2. κ dˆ (2. MATHEMATICAL FORMULATION 67 the destruction term is h " # i νˆ˜ 2 c b1 Dest(νˆ˜) = cw1 fw − 2 ft2 .
. initial and boundary conditions for the rotorstator interaction are discussed. • Numerical tools as artificial dissipation. • Hardware architecture. Essential ingredients for an accurate and efficient solution of the flowfield are: • Governing equations and turbulence transport equations. In addition. consistency and convergence problems are dealt with. accurate discretization methods and proper assessment. methods for convergence acceleration. • Good initial conditions. Stability. • Enforcement of boundary conditions.3 Numerical Formulation In this chapter the equations developed in the previous chapters are prepared for the sake of numerical solution. • Adequate mesh resolutions and elements quality. mesh treatments. • Efficient numerical algorithm development.
55. Different space and time discretizations are used in this work. the socalled artificial dissipation has to be added for numerical stabilization. Upwind schemes are constructed by considering the physical properties of the convective terms (Euler equations). 68. Hence. It is based on a blend of 2nd and 4th differences scaled by the maximum eigenvalue of the convective flux Jacobian.1 69 Discretization Models Finite volume has been reported as a method which gives good results for flow bound by complex geometries using unstructured mesh. The SpalartAllmaras method is used for turbulent calculations. Central schemes cannot recognize and suppress the oddeven decoupling of the solution. the time discretization is implicit . while for the temporal discretization the explicit MacCormack (1969) and the RungeKutta (five steps) are chosen. NUMERICAL FORMULATION 3. are studied.CHAPTER 3.1. It is the method adopted in this work. For the spatial discretization the methods of Jameson (centered).4. These combination of an undivided Laplacian and biharmonic operator is employed for unstructured meshes. Two upwing schemes are implemented in this work: FluxVector Sppliting and FluxDifference Sppliting. resulting in several different methods. comparison and application of known and to be developed methods. The reader may refer to [74. the application born in mind in this work. This method was chosen based on the good results reported in the literature when applied to flows in turbomachines. The details are described in section 3. 75] if comparisons to other methods are of his interest. The diffusion and advection terms are discretized according the suggestions of [73]. aiming at a platform for the test. It is possible to combine spatial and temporal options. van Leer (Flux VectorSppliting) and Roe (FluxDifference Sppliting). These treatment distinguish the upstream and downstream influences based on wave propagation directions.
Fv and Gv are the viscous terms.1.1 on a finitevolume and applying the Gauss theorem to transform the volume integral into a surface integral: Z V ∂Q dV = − ∂t Z ~ P~ )dV = − (∇.3) G = Ge − Gv . (3. (3.5) where Ee .~n)dS. Integrating Equation 3.CHAPTER 3. matricial form have the representation ∂G ∂Q ∂E ∂F + + + = 0. NUMERICAL FORMULATION 70 and following the suggestion of the same reference (Euler implicit).1) E = Ee − Ev . (3.7) . 3. (3. (3. Fe and Ge are Euler (convection) terms and Ev . the superscripts of nondimensionalization of RANS and Favre averages. the equations in conservative.4) with (3.2) F = Fe − Fv . V Z (P~ . ∂t ∂x ∂y ∂z (3.1 FiniteVolume Method Neglecting the source term.6) S with P~ = E~i + F ~j + G~k.
All fluxes are calculated based on the face k of an i − th volume.9) V it follows that V and Z V ¯ ∂Q ∂ ¯ ∂Q dV = (QV )=V . (3.8) QdV. nf aces is the number of faces for each volume and S~k is the outward oriented normal area vector of k face. Defining the value of Q by ¯= 1 Q V Z ¯ = QV Z QdV. F and G for each k face given by Equation 3.12) where k indicates the face of i − th volume.CHAPTER 3. In this work.10 into Equation 3.11 is approximated by nf aces ∂ Q¯i 1 X ~k .12 is divided in two parts: the term involving convective flux and the term involving the .S ∂t Vi k=1 (3. =− (Ek~i + Fk~j + Gk~k). 3. the values of flow properties are considered to be located at volume centroid. (3.2 Spatial Integration The spatial integration of vectors E.11) S For the i − th control volume the Equation 3. (3.~n)dS.10) Substitution of Equation 3. NUMERICAL FORMULATION 71 S is the control surface and ~n is the outward normal vector. ∂t ∂t ∂t (3.6 yields ¯ 1 ∂Q =− ∂t V Z (P~ .
3. nf aces CON V (Qi ) = X ~k . to avoid numerical instabilities associated to the centered method caused by high gradients. 43.1 The Centered Scheme of Jameson The centered discretization [76] of flux balance for each i − th control volume is based on information from the current volume and its neighbors. Artificial dissipation acts as a damping function to limit high gradients [76] and is added to the model used in this work. centered and upwind methods are used in this work and are presented in forthcoming sections. 78. hence naturally dissipative.2. 79] do not require artificial dissipation because the flux is calculated following a specific direction. CON V (Qk ) = (Qk + Qnei ). Upwind schemes [77. The fluxes for each face can be written as 1 ~k . F and G are divided in two terms. 84. one for convective terms and other for viscous terms. Both. For each i − th volume the convective and viscous terms. 81. This numerical treatment deals with problems in high gradient regions using the vectorsplitting methodology [80. respectively. About the convective terms. 83. (Ee (Qk )~i + Fe (Qk )~j + Ge (Qk )~k). The vectors E.14) .CHAPTER 3. 67]. NUMERICAL FORMULATION 72 diffusive or viscous flux.S 2 (3. 82.13) k=1 The discretization is based on the arithmetical averages for a current i − th control volume and its neighbors nei at k − th face. The flux is calculated by summation of fluxes at all faces of i − th control volume. are represented by CON V (Qi ) and V ISC(Qi ).S (3. artificial dissipation or artificial viscosity is used.
The scheme will be presented only for vector Ee . 3. Basically the method uses a separation of the vectors Ee . thus a nonsmooth solution at the sonic region may appear within the domain. A disadvantage of this scheme is that the split fluxes are not continuously differentiable at the sonic region. (3. Steger and Warming [85] developed the method for the Euler equations splitting the eigenvalues of the Jacobian matrix. This method is an upwind method that is naturally dissipative. details of which can be found in [80]. For vectors Fe and Ge the development of terms is similar. respectively.2. For M ≥ 1 and M ≤ −1 the calculations are based on the flux for all directions.2 The Upwind Scheme of Van Leer: Flux–Vector Splitting The use of damping functions like artificial dissipation is avoided with the F V S for the convective terms. . To overcome this problem van Leer [83] proposed some modifications to the Steger and Warming scheme.17) Contributions + and − are associated with positive and negative eigenvalues. F and G are written as functions of the vector of conserved variables Q. (3.15) Fe = Fe+ + Fe− .16) − Ge = G+ e + Ge . Fe and Ge in two contributions each: Ee = Ee+ + Ee− .CHAPTER 3. (3. The vectors E. NUMERICAL FORMULATION 73 where nei stands for the neighbor volume.
(3.CHAPTER 3.20) Ee− = f. (3.22) For −1 < M < 1 the calculations are based on the forward flux f ± for all directions Ee+ = f + .19) Ee+ = {0 0 0 0 0}T .21) and for M ≤ −1: where. (3.18) Ee− = {0 0 0 0 0}T .24) .5(Wx2 + Wy2 + Wz2 ) + a2 /(γ − 1)] . (3. f= ρWx ρ(Wx2 + a2 /γ) ρWx Wy ρWx Wz ρWx [0. (3. (3. NUMERICAL FORMULATION 74 For M ≥ 1: Ee+ = f. (3.23) Ee− = f − .
(3. (3. f± = ± fmass ± fmass [(γ − 1)Wx ± 2a]/γ ± Wy fmass ± fmass Wz ± fmass {[(γ − 1)Wx + 2a]2 /[2(γ 2 − 1)] + 0. [(Ee )k~i + (Fe )k~j + (Ge )k~k]. (3.5(Wy2 + Wz2 )} .29) .S (3.28) The convective term is calculated by: nf aces CON V (Qi ) = X k=1 ~k .5(M ± 1)]2 . NUMERICAL FORMULATION 75 where.25) with ± fmass = ±ρa[0. (3.CHAPTER 3.26) and the spatial integration follows the following procedure: For positive outward normal vector at face k: (Ee )k = Ee+ (Qi ) + Ee− (Qnei ).27) and for negative outward normal vector at face k: (Ee )k = Ee− (Qi ) + Ee+ (Qnei ).
The methodology of Roe is applied quite often because of its high accuracy for boundary layers and good resolution of shocks regions.3 76 The Upwind Scheme of Roe: Flux–Difference Splitting The F DS schemes evaluate the convective fluxes at a face of control volume from left and right states based on idea of Godunov [86]. At a kface of the left (L) and right (R) states of a control volume the difference is expressed as ~R − Q ~ L ). (P~e )R − (P~e )L = (A¯Roe )k (Q (3.2. NUMERICAL FORMULATION 3. where the flow variables are replaced by Roeaveraged variables presented Jacobian. The Roematrix is identical to the convective flux ¯ c .30) where A¯Roe denotes the Roematrix. [77] and Roe [78].CHAPTER 3. In this work. It is based on the decomposition of the flux difference of convective terms over a face of the control volume into an addition of wave contributions. approximate Riemann solvers were developed by Osher et al. Λ bellow . Several numerical schemes for the solution of hyperbolic conservation laws are based on exploiting the information obtained by considering a sequence of Riemann problems. Roe’s approximate Riemann solver was implemented in the framework of the cellcentered scheme. The exact solution of the Riemann problem.
q˜2 = W ¯ c T¯−1 With some algebra. as well as the diagonal matrix of eigenvalues ¯ c ) are evaluated using equations 3.32) ¯ c  = diag(Wn . Λ (3.31) ˜ = W ˜ x nx + W ˜ y ny + W ˜ z nz .31. √ √ WxL ρL + WxR ρR . (γ − 1)(H 77 (3. W ˜ x2 + W ˜ y2 + W ˜ z2 . √ √ ρL + ρR q ˜ − q˜2 ).33) where The matrix of eigenvectors T¯ and T¯−1 . √ √ ρL + ρR √ √ WzL ρL + WzR ρR . √ √ ρL + ρR √ √ HL ρL + HR ρR . NUMERICAL FORMULATION ρ˜ = W˜x = W˜y = ˜z = W ˜ = H a ˜ = p ρ˜L ρ˜R . (P~e )R − (P~e )L = T¯Λ (3. The fluxes of convective terms . √ √ ρL + ρR √ √ WyL ρL + WyR ρR .CHAPTER 3. Wn  − a)T . the eigenvalues Λ problem and the T¯ eigenvectors are the waves themselves. Wn . The characteristic variables C ~ represent the wave (Λ ¯ c are the associated wave speeds of the approximate Riemann amplitudes. Wn  + a.30 into waves where ¯ c (C ~R − C ~ L ). Wn . inserting the diagonalisation of the Roematrix as A¯Roe = T¯Λ decomposing the Roe’s scheme given by equation 3.
NUMERICAL FORMULATION 78 are evaluated at Kfaces of a control volume as 1 ~ R ) + P~c (Q ~ L ) − A¯Roe K (Q ~R − Q ~ L )].34) The product of A¯Roe  and the difference of the L and R states is given by ~R − Q ~ L ) = ∆P~1  + ∆P~2.37) . ˜ ∆Wz − ∆W nz ˜ x ∆Wx + W ˜ y ∆Wy + W ˜ z ∆Wz − W ˜ ∆W W (3. ˜ Wz − a ˜ nz ˜ −a ˜ H ˜W (3.3.36) 0 ˜ x − ∆W nx ∆W ˜ y − ∆W ny ∆W .CHAPTER 3.4  + ∆P~5 .3.35) where ∆p − ρ˜a ˜∆W ˜ −a ∆P~1  = W ˜ 2˜ a ˜ ∆P~2. (P~e )k = [P~c (Q 2 (3.4  = W ∆p ∆ρ − a ˜ 1 ˜x W ˜ + ρ ˜ Wy ˜z W q˜2 2 1 ˜x − a W ˜ nx ˜y − a W ˜ ny . A¯Roe (Q (3.
To solve this problem. Roe’s scheme is considerably more accurate than centered scheme due to the high resolution of boundary layers and the lower sensitivity to grid distortions of the former one in comparison to central scheme. Equation 3. The discretization method of the Euler equations developed by Roe is originally of first order.2. Details of M U SCL schemes can be found in reference [87]. NUMERICAL FORMULATION 79 ∆p + ρ˜a ˜∆W ˜ +a ∆P~5  = W ˜ 2˜ a 1 ˜x + a W ˜nx ˜y + a W ˜ny .30 will produce an unphysical expansion shock in the case of stationary ex~ L 6= Q ~ R causing the socalled ”carbuncle phenomenon”. Furthermore.CHAPTER 3. pansion. The values of L and R state are used to calculate the convective flux through the face. the original scheme does not recognize the sonic point.38) Discretizations based on characteristics of the Euler equations separately interpolate flow variables from the L and R states at face using nonsymmetric formulae. ˜ ±a the modulus of the eigenvalues Λc  = W ˜ is modified using the entropy correction proposed by Harten . in which P~cL = P~cR . To increase the order of the above equations a reconstruction based on Monotone UpstreamCentered Schemes for Conservation Laws (M U SCL) approach is used to achieve secondorder accuracy as presented in the section 3. but Q where the perturbation grows ahead of a strong bow shock along the stagnation line. ˜ Wz + a ˜nz ˜ +a ˜ H ˜W (3.4.
NUMERICAL FORMULATION Λc  = Λc .39) where δ is a small value that represents some fraction of the local speed of sound. Hence. it is assumed that the solution is linearly distributed over control volume (Figure 3.4 Reconstruction Based on Approximate Monotone UpstreamCentered Schemes for Conservation Laws (M U SCL) To reach the secondorder accuracy. With this idea.05 ≤ δ ≤ 0. 2 2 Λc  + δ .2. implemented to W 3. one possibility is to use the M U SCL approach assuming that the solution to vary over the control volumes in a linear fashion. In this work a piecewise linear reconstruction was implemented. In order to calculate the L and R states.CHAPTER 3. 80 if Λc  > δ .40) where the ∇QI and ∇QJ are the gradients at the cellcenter I and J respectively. . In this work the range used is 0.1). φ denotes a limiter function.15. a reconstruction of the assumed solution variation becomes necessary. The same entropy correction was also ˜. if Λc  ≤ δ . 2δ (3. ~ri and ~rj are the distances from the cellcentroid to the facemidpoints of the cell. the L and R states are calculated for the cellcentered scheme with the following relations QL = QI + φI (∇QI · ~rL ) QR = QJ + φJ (∇QJ · ~rR ) (3.
Other limiter function widely used by CFD code developers is the Barth and Jespersen limiter presented in reference [88].1 – Scheme based on piecewise linear reconstruction This reconstruction method was presented by Barth and Jespersen [88] and corresponds to a Taylorseries expansion around the neighbouring centers of the face of a controlvolume. This is the price to be paid to use a high order upwind schemes. the Venkatakrishnan limiter function [89. in general. Reference [91] presents the Euler computations using explicit time marching for different limiter functions as minmod and superbee. The purpose of a limiter is to reduce the slopes used to interpolate a flow variable at the face of a control volume.2. need the calculation of gradients also.2. NUMERICAL FORMULATION 81 FIGURE 3.5 Venkatakrishnan’s Limiter As aforementioned. 3.5. Limiters. second and higherorder upwind spatial discretizations require the use of limiters in order to prevent the generation of oscillations and spurious solutions in regions with high gradients as shock waves. In this work.CHAPTER 3. 90] was implemented and is presented in section 3. Limiter function must be employed in order to prevent the generation of spurious oscillations close to strong discontinuities. Barth and Jespersen’s limiter enforces a monotone solution. but it is rather dissipative and it tends . It is not the scope of this work to discuss and compare limiters. requiring high computational effort. in order to constrain the solution variations. The above scheme requires the computation of gradients at cellcenters.
The formulation of this limiter function implemented in this code is presented bellow 1 (∆21.max = Qmax − Qi . minj Qj ).min ∆2 + 2 1.min φi = minj . (3. Qmin = min(Qi . maxj Qj ). In this work. ∆2 ∆21. this formulation demonstrated the effectiveness of their multidimensional limiter by computing oscillationfree transonic flow solutions on highly irregular triangular meshes. if ∆2 = 0 .max ∆2 + 2 1 (∆21.CHAPTER 3.41) where ∆1. This limiter presents superior convergence properties. 2 Qmax = max(Qi . All details of this limiter function are described in reference [89.43) where Qmax and Qmin stand for the minimum and maximum values on all neighbors of control volume i including i itself. NUMERICAL FORMULATION 82 to smear discontinuities. 90].max .min + 2 )∆2 + 2∆22 ∆1. Furthermore.max + 2∆22 + ∆1.42) and 1 ∆2 = (∇Qi · ~ri ).max + 2 )∆2 + 2∆22 ∆1. the Venkatakrisnan’s limiter function was implemented. if ∆2 < 0 . parameter 2 is used to control the amount of limiting.min = Qmin − Qi .min + 2∆22 + ∆1. if ∆2 > 0 . ∆1. . (3. ∆2 ∆21. (3.
46) where nei stands for a determined neighbor volume.2. i. 2 = (K∆h)3 .CHAPTER 3. setting 2 equal to zero results in full limiting. Otherwise.e.. dynamic viscosity. .S (3. In reference [90] the solution dependence of several k values are discussed.45) k=1 The discretization is based on the arithmetical averages for a current i−th control volume and its neighbors nei at k − th face. but this may stall the convergence. if 2 is set to a large value. (3. the limiter function will return a value around unity.S 2 (3. heat conduction coefficient and stresses terms resulting: nf aces V ISC(Qi ) = X ~k . F and G are written as function of the vector of conserved variables Q. V ISC(Qk ) = (Qk + Qnei ). 3. In practice.44) where K is a constant set by the user and it is ϑ(1) and ∆h is the cuberoot of the volume of control volume. (Ev (Qk )~i + Fv (Qk )~j + Gv (Qk )~k). it was found that 2 should be proportional to a local length scale. NUMERICAL FORMULATION 83 Hence.6 Discretization of Viscous Fluxes The viscous terms are evaluated using cellcentered scheme due to the elliptic nature of the diffusive fluxes. The fluxes for each face can be written as 1 ~k . The vectors E.
2 (3.3 84 Time Integration The used time integration scheme is explicit. The MacCormack (1969) and the RungeKutta (five steps) schemes were implemented in this work and are presented bellow. Instability of centered schemes is avoided if predictor (forward) and corrector (backward) steps like the used by Veuillot [34] for turbomachinery simulations. Vi (3. (n+1) Qi (n) = Qi − ∆ti (n) (n) (n) [CON V (Qi )f orward − V ISC(Qi )f orward − DISS(Qi )f orward ].1 The Scheme of MacCormack (1969) The MacCormack scheme for time integration is based on a special version of the LaxWendroff method.47) For the corrector step: (n+1) Qi (n) = Qi − ∆ti (n+1) (n+1) (n+1) [CON V (Qi )backward −V ISC(Qi )backward −DISS(Qi )backward ]. 3. Vi (3. from n to n + 1 time indexes: ∆ti = CF L lengthi .50) . NUMERICAL FORMULATION 3. ~  + ai ) (W (3.3. For predictor step.48) For the update: (n+1) Qi 1 (n+1) (n+1) = [Qi + Qi ].CHAPTER 3.49) Constant CourantFriedrichsLewy (CF L) number is used for the calculations of the local timestep considered.
But. The Jameson scheme is very attractive and is cited and used by many authors [8. calculated based on the distance between the i−th volume centroid and the closest face. which are not naturally damped. 92. multigrid techniques. 72. Jameson [76] proposed this time integration scheme with a centered difference scheme for spatial integration. . 39. 3. other RungeKutta schemes with different number of stages (two or four) are found in the literature [72. It allows easy incorporation of several outstanding features such as. 93. convergence acceleration.3. compact schemes and others. 83]. The variable ai is the speed of sound for the i − th control volume. 42].2 The Scheme of RungeKutta The RungeKutta scheme is a multisteps time integration.CHAPTER 3. The MacCormack time integration method presents a good convergence for 0 < CF L < 1. Jameson’s scheme used in this work is a fivestep RungeKutta time integration scheme. Additional dissipation terms are required to control the high frequency waves. 83. 47. NUMERICAL FORMULATION 85 The variable lenghti corresponds to the characteristic length of the volume.
3. Vi ∆ti (1) (1) (1) Q0i − α2 [CON V (Qi ) − V ISC(Qi ) − DISS(Qi )]. and for economy of machine time the dissipative terms can be evaluated only at every other integration step based on the author experience. In this work.51) (3. Vi ∆ti (3) (3) (3) Q0i − α4 [CON V (Qi ) − V ISC(Qi ) − DISS(Qi )]. Vi ∆ti (2) (2) (2) Q0i − α3 [CON V (Qi ) − V ISC(Qi ) − DISS(Qi )]. This numerical methodology generates oscillations near discontinuities.52) (2) = (3.1 for MacCormack scheme. In this model ∆t can be calculated using the same formulation presented in Section 3.4 Numerical Stability and Convergence Acceleration As commented before. It was preferred in place of first order . α2 = 1/6.57) where α1 = 1/4. (3. √ Good numerical stability can be obtained for CF L ≤ 2 2.56) (3. Vi (5) = Qi .3.54) (3. Vi ∆ti (4) (4) (4) Q0i − α5 [CON V (Qi ) − V ISC(Qi ) − DISS(Qi )].CHAPTER 3. a second order centered difference scheme is implemented to calculate the hyperboliclike convective term. NUMERICAL FORMULATION 86 The steps are (0) = Qi . α4 = 1/2 and α5 = 1.55) (3.53) (3) = (4) = (5) = Qi Qi Qi Qi Qi Qi (n+1) Qi (n) ∆ti (0) (0) (0) [CON V (Qi ) − V ISC(Qi ) − DISS(Qi )]. α3 = 3/8. (1) = Q0i − α1 (3. some numerical discretization methods need damping functions to avoid numerical instabilities.
A flow properties sensor is used to activate or deactivate the artificial dissipation.4. 29]. nonlinear models were implemented.1 Artificial Dissipation In this work. IRS is much used with multigrid methodology [50. can be improved in terms of convergence acceleration. Tests were carried out initially with a linear model. Some of such highorder schemes were implemented in this work and they are presented in the next subsection. . There are several mathematical models with different characteristics. Other alternative is the implementation of implicit residual smoothing (IRS) to increase the maximum operational CF L number. 94. This approach is not proper for transient solutions because the timemarching process must be compatible with the physical aspects of flow. 3. Then. using centereddifference and RungeKutta to perform the spatial and time integrations.4. This technique is presented in Section 3.2. four artificial dissipation models were implemented. what requires constant timestep for all control volumes. but very unstable in regions with high gradients like in shock waves. respectively. linear and nonlinear. NUMERICAL FORMULATION 87 schemes in which the truncation errors act as a numerical dissipation.CHAPTER 3. in order to check the quality of results. For steadystate simulations it is possible to apply a local timestep based on a linear stability analysis of the NavierStokes equations. although the CF L number may vary. The artificial dissipation or artificial viscosity is responsible for maintaining a good numerical stability. Jameson’s scheme.
treatment and formulation different from other regions of potential flow.60) and the artificial dissipation of i − th control volume is calculated by: DISS(Qi ) = ε[DISS(Qi )i + DISS(Qi )j + DISS(Qi )k ].61) where 0 < ε < 1. 3.58) DISS(Qi )j = [(Qi )j−2 − 4(Qi )j−1 + 6(Qi ) − 4(Qi )j+1 + (Qi )j+2 ].1 88 Linear Model The model used in this work is taken from Hoffmann [67]. for i. and the value used was 0. Let the i − th volume and directions i. and similarly for the others. (3. j and k be considered. the i − th control volume has a neighbor at each face. (3.4. (3. Two procedures have been implemented. The first is based on the work of Jameson [76] for twodimensional case.59) DISS(Qi )k = [(Qi )k−2 − 4(Qi )k−1 + 6(Qi ) − 4(Qi )k+1 + (Qi )k+2 ]. Therefore. the last one is dependent on the local pressure gradient. i − 1 at the i direction.2 NonLinear Models Nonlinear artificial dissipation model used in this work is a blend of second and fourth differences. j and k directions the dissipation terms are DISS(Qi )i = [(Qi )i−2 − 4(Qi )i−1 + 6(Qi ) − 4(Qi )i+1 + (Qi )i+2 ].1. The neighborhood are identified by i − 2. For each direction. Considering the directions in .08. (3. here extended for threedimensional.1. The difference between the linear and the nonlinear model is that.CHAPTER 3. NUMERICAL FORMULATION 3.4. This requires in region with severe pressure gradients.
64) DISS(Qi )k = DISS(Qi )k+1 − DISS(Qi )k−1 .CHAPTER 3. DISS(Qi ) = DISS(Qi )i + DISS(Qi )j + DISS(Qi )k . (ζi )i ]. (3. DISS(Qi )i+1 is calculated by DISS(Qi )i+1 = (Vi )i+1 (2) (4) [εi+1 ((Qi )i+1 − (Qi )i ) − εi+1 ((Qi )i+2 − 3(Qi )i+1 + 3(Qi )i − (Qi )i )]. i + 1 for i direction. j + 1 for j direction and k − 1.69) where (ζi )i = .62) For each direction of i − th control volume the neighbors i − 1. (4) (2) (3. pi+1  + 2pi  + pi−1  (3.63) DISS(Qi )j = DISS(Qi )j+1 − DISS(Qi )j−1 . j − 1.68) pi+1 − 2pi + pi−1  . and are calculated using a pressure sensor. (3. j and k for i − th control volume. NUMERICAL FORMULATION 89 the computational coordinates i. (k (4) − εi+1 )].67) εi+1 = max[0.66) where. Vi is the volume of the i − th control volume. ∆t (3. For the direction i (2) εi+1 = k (2) max[(ζi )i+1 . (3. The coefficients ε(2) and ε(4) depend on the flow behavior.65) Hence. (3. (3. k + 1 for k direction are used for the calculations: DISS(Qi )i = DISS(Qi )i+1 − DISS(Qi )i−1 .
based on the mean value properties of i−th control volume .70) Calculation starts with nneigh D(Qi ) = X (Qm − Qi ). (3. In this model the artificial dissipation terms are divided by one undivided Laplacian operator that is firstorder accurate and is used only in the regions with strong gradients to avoid numerical oscillations and one undivided biharmonic operator that is thirdorder accurate: DISS(Qi ) = DISS(Qi )Laplacian + DISS(Qi )biharmonic .73) k=1 where ak is the speed of sound.CHAPTER 3. The biharmonic operator is calculated by: nneigh DISS(Qi )biharmonic = ε (2) X m=1 [D(Qm ) − D(Qi )] Am − Ai 2 .71) m=1 where. the term ε(2) reduces the accuracy to firstorder and the term ε(4) . (3. as demonstrated by Jameson [76]. [(Wx )k S (3. (3. being turnedoff. NUMERICAL FORMULATION 90 The constants k (2) and k (4) assume the values k (2) = 1/4 and k (4) = 1/256 as in [76]. nneigh is the number of neighbors of i − th control volume. The second and less dissipative procedure is based on the work of Mavriplis [93] extended to threedimensions.72) with nf aces Ai = X ~k + (Wy )k S ~k + (Wz )k S ~k  + ak Areak ]. In regions with high gradients. The term ε(2) is secondorder accurate and the term ε(4) firstorder resulting in a thirdorder accurate scheme. For other directions there are identical procedures. tend to induce overshoots during the numerical iterations. with fourth differences.
The Laplacian operator is calculated replacing the quantities D(Qm ) and D(Qi ) in Equation 3. Other nonlinear formulation can be obtained using the formulation of the first nonlinear model presented is this subsection. but replacing the term Vi /∆t in Equation 3.73. Areak is the area at face k belonging to the i − th control volume and its neighbor. NUMERICAL FORMULATION 91 and its neighbor at face k.75) (pk − pi ) k=1 Typical values are ε(2) = 1/2 and 1/256 ≤ ε(4) ≤ 3/256.66 by the term (Am − Ai )/2 calculated in Equation 3.72 by the quantities Qm and Qi and replacing the constant ε(2) by ε(1) : nneigh DISS(Qi )Laplacian = ε (1) X [Qm − Qi ] m=1 Am − Ai 2 . The values used in this work are: ε(2) = 1/2 and ε(4) = 3/256. as indicated by Mavriplis [93]. Pulliam studied various artificial dissipation models for Euler equations.74) The value of ε(1) is calculated using the pressure sensor nf aces  ε (1) = X (pk − pi ) k=1 nf aces X . (3. Details of artificial dissipation in the boundaries are presented in Appendix A.CHAPTER 3. The treatment of the artificial dissipation models at boundaries and a fully implicity application of these models are analysed and can be found in [95]. . (3. for both. linear and nonlinear models using an implicit approximate factorization for transonic airfoils. These two last models are less dissipative than the first model presented and provide conservative dissipation using the averages given by the terms Am and Ai .
In this work. (3.2 92 Implicit Residual Smoothing The IRS method [83. The IRS is applied to the righthandside (RHS) of the governing equations.76) where ∇2 is the undivided Laplacian operator. 3. 98] is used to increase the maximum operational CF L number for numerical schemes that use explicit timeintegration. Equation 2.77 can be solved numerically using several Jacobi iterations and can be applied at alternate stages of RungeKutta timeintegration method.77) k=1 where is 1/8 ≤ δ ≤ 1/2. the IRS model implemented is the one given by Jameson and Mavriplis [92] and it is extended to threedimensions.CHAPTER 3. NUMERICAL FORMULATION 3. neglecting the terms with ft1 and ft2 associated to transition. 96. (3. Thus the smoothed residuals are calculated using: nf aces RHSi = (1 + δnf aces)RHS i − δ X RHS k . Equation 3. To apply this turbulence model in the finite volume context. indicated by subscript t. 97. The mean (RHS) or simply RHS i is calculated by RHS i = RHSi + δ∇2 RHSi . . therefore the transition terms in the turbulence model may be neglected.4.5 Spatial and Time Integration of the Spalart–Allmaras Turbulence Model The flow in several aerodynamic applications of high Reynolds number is completely turbulent.58 may be written in the integral form.
production P rod(νˆ˜) and destruction Dest(νˆ˜) terms. NUMERICAL FORMULATION Z Vi 1 ∂ νˆ˜i = ∂ tˆ Re0 σ Z 93 ∇.78) Applying Equation 3.Sk − (νˆ˜W = Vi k=1 ∂ tˆ Vi Re0 σ k=1  {z }  {z } D1 (νˆ ˜) C(νˆ ˜) " #2 νˆ˜ 1 cb2 ˆ (∇νˆ˜)2 + cb1 S˜νˆ˜ − cw1 fw .79 is nf aces nf aces X X 1 1 ∂ νˆ˜i ~ ~ˆ ).dS+ ~ ˆ ˆ [(ˆ ν + ν˜)∇ν˜].[(ˆ ν + νˆ˜)∇νˆ˜]dV − Z Vi ~ˆ )dV + ∇. {z }  Re σ dˆ Re0  0 {z } P rod(νˆ˜)  {z } ˆ D2 (ν˜) (3. for the i − th . The convection term is discretized using a firstorder upwind scheme and the diffusion term is dicretized using secondorder accurate centereddifference with second diffusion term D2 (νˆ˜).dS − (νˆ˜W V i Si Si " #2 νˆ˜ 1 cb2 (∇νˆ˜)2 + cb1 Sˆ˜νˆ˜ − cw1 fw . ˆ Re0 σ d Re0 (3.80) Dest(νˆ ˜) The spatial discretization of the convection and diffusion terms follows SpalartAllmaras work [73].S ~k + [(ˆ ν + νˆ˜)∇νˆ˜].79) The discretized form of Equation 3.78 ∂ νˆ˜i 1 = ∂ tˆ Vi Re0 σ Z Z 1 ~ ~ˆ ).(νˆ˜W Vi " #2 1 cb2 νˆ˜ ˆ 2 (∇νˆ˜) + cb1 S˜νˆ˜ − cw1 fw .CHAPTER 3.6 to Equation 3. Re0 σ dˆ Re0 (3.
2 Wk+ = Wk− (3. − [(Wk )ν˜i + (W k Vi k=1 (3.83) Since νt is the solution of the turbulence transport equation. this convection term is calculated differently: all terms are calculated at the time n except the convection term that is calculated at the time n + 1. NUMERICAL FORMULATION 94 control volume. it cannot be negative. 2 ˆ nei )k − (W ˆ nei )k (W = . and so any treatment applied to this turbulence model must be robust enough to avoid numerical instabilities and consequently divergence.81) with ˆ i )k + (W ˆ i )k (W .S ~k + D2 (νˆ˜i ) + P rod(νˆ˜i ) + Dest(νˆ˜i ).82) (3. coupled to the governing equations.84) .CHAPTER 3. nf aces X ∂ νˆ˜i 1 = ∂ tˆ Vi Re0 σ k=1 (" νˆi + (ˆ ν )nei 2 + νˆ˜i + (νˆ˜)nei 2 !# ∇νˆ˜ + (∇νˆ˜)nei 2 !) ~k + . calculated at the time n (therefore constants). the convection term in the Euler scheme is calculated by f n+1 n = f + ∆t ∂f ∂t n+1 . To avoid divergence due to the convection term calculated explicitly within the RungeKutta scheme.S nf aces 1 X ˆ+ ˆ ˆ − )(νˆ˜k )nei ]. To reduce the calculation time. How it is implemented is explained in details in the paper [73]. The machine time for implicit calculation of all turbulence terms is higher than if only the convection term is calculated implicitly. the timeintegration of the turbulence equation is included in the RungeKutta steps. The method of Spalart and Allmaras [73] was implemented for the timeintegration process using approximate factorization. For explicit timemarching. (3.
it is possible to . (3.80 as follows: nf aces nf aces X 1 1 X n+1 ~ n n ~ ˆ [(˜ ν + ν˜ )].Sk + Vi Re0 σ k=1 Vi k=1 ) +D2 (νˆ˜in ) + P rod(νˆ˜in ) − Dest(νˆ˜in ) . NUMERICAL FORMULATION 95 following the reference [74].88) ( = Replacing the variable νˆ˜n+1 following Equation 3.Sk − [(Wk+ )νˆ˜in+1 + (Wk− )νˆ˜nei ].85) ( νˆ˜n+1 =νˆ˜in + ∆ti Knowing that ∆νˆ˜in = νˆ˜in+1 − νˆ˜in .86) and. that ∆νˆ˜in = ∆ti ∂ νˆ˜ ∂t !n+1 .84. (3.Sk + Vi Re0 σ k=1 Vi k=1 ) n n n +D2 (νˆ˜i ) + P rod(νˆ˜i ) − Dest(νˆ˜i ) .S [(Wk+ )νˆ˜in+1 + (Wk− )νˆ˜nei ]. from Equation 3. (3. (3. The timemarching is applied to the turbulent eddy viscosity νˆ˜ in Equation 3.CHAPTER 3.86 and grouping terms.87) it is possible to write ∂ νˆ˜ ∂t !n+1 nf aces nf aces X X 1 n+1 ~ ~k − 1 [(˜ ν n + νˆ˜n )].
the Thomas algorithm cannot be used directly. because it is possible to know the location of the zeros out of the dominant diagonal (tridiagonal. In this work. because the locations of the nonzeroes are unknown. NUMERICAL FORMULATION 96 write ∂ νˆ˜ ∂t !n+1 ( = nf aces X 1 ~k + [(˜ ν n + νˆ˜n )]. it is also required special storage algorithms. For unstructured mesh. In order to take advantage of the large number of zeroes in the matrix. References [99] and [100] indicate several methods to solve sparse matrix. resulting in a sparse matrix. In this work only hexahedral control volumes were considered. Reference [101] gives robust method for the LU factorization of the original matrices. pentadiagonal or hexadiagonal). the Thomas algorithm may be directly used. the conjugated gradient is used. For structured grid.CHAPTER 3.S [(Wk+ )νˆ˜in + (Wk− )νˆ˜nei ]. (3.6 Unstructured Mesh Treatment Study of papers available in the literature [53] indicated that the hexahedral element has better quality.S − Vi k=1 Vi k=1 ) +D2 (ν˜ˆin ) + P rod(ν˜ˆin ) − Dest(ν˜ˆin ) .S Vi Re0 σ k=1 nf aces nf aces X 1 X 1 + − n n n ~k − ~k + [(Wk )∆νˆ˜i + (Wk )∆νˆ˜nei ]. concerning the solution in the areas of high gradients like the ones in turbomachines. 3.89) The second term in the righthandside is evaluated implicitly and all other terms representing the righthandside (involving ∆νˆ˜in ) of the turbulent transport are evaluated in time n. concerns the mesh. since . Other important remark about how to calculate the convection term implicitly.
areas. the initial conditions must be the actual flow conditions at the time when the calculations start. pyramidal elements and hybrid mesh are in mind.7 Initial Conditions. for transients. The treatment of the edges. Ghost elements are similar to real element. among which adaptative mesh refinement. NUMERICAL FORMULATION 97 turbomachinery is the application in mind. due to the different numbers of blades in each blade row. tetrahedral. The elements and its neighbors are addressed via connectivity tables.7). Boundary Conditions and Rows Interface The calculations start from the flow properties established as initial conditions set at the centroid of each element. faces.CHAPTER 3. It is worth mentioning that the computational code developed in this work must to be a platform aiming future development. For steadystate solution the initial condition set of values may not represent the actual flow but. 3. respecting the physics of the flow to avoid numerical instabilities and divergence. Elements in the boundary. outward normals and volumes follow [83]. For hexahedral element with six faces there are six neighbors. The closer the initial condition is to the solution the smaller the number of iterations to reach convergence. All flow properties for both ghost and real elements . The meshes for the rotor and for the stator must be usually independent. The elements in the boundary require boundary conditions that are set in the ghost elements. were created aiming at the boundary conditions imposition. These elements must provide correct informations about the flow in the real neighbor elements (see Figure 3. the ghost elements. prismatic.
Good initial conditions may also reduce the time machine mainly when full NavierStokes is solved and large number of elements are required. developed at Center of Reference for Gas Turbine at ITA produce very good initial conditions for the simulations carried out during this research. the streamline curvature computer code can give excellent point of departure since it calculates accurate axisymmetric solutions.2 – Scheme implemented to set the boundary condition in a ghost element Compressors or turbines stages are made of rotating and fixed elements. NUMERICAL FORMULATION 98 are cellcentred. to avoid divergence. They must be qualitatively correct and represent an approximation of the physical problem.1 Initial Conditions Initial conditions are the set up values of the variables used for starting the calculations. FIGURE 3. while the stators remain still.CHAPTER 3. Computational codes for compressors and turbines. Convergence may be difficult to achieve if adequate initial conditions are not set up appropriately. that may be used for both design and offdesign simulations. An interface between two adjacent rows must be created in order to precisely transfer information from the outcoming flow of one row to the incoming of the next row. 3. Therefore the rotor meshes move with the blades. Shapes of the ghosts elements are the same as that of the adjacent real element.7. For example. . They can be chosen from auxiliary meanline or throughflow calculations.
The second one.7.2 Boundary Conditions 3. ϕ1 is the meridional streamline inclination angle relative to the axial direction.7. for turbomachinery study which starts usually with flow directions and Mach numbers.2. r is radius of the centroid of the control volume with origin at the axis rotation. 2. NUMERICAL FORMULATION 3. two possibilities were considered: 1. . the flow velocity components are initially calculated by Wx = M1 (x cos α1 sin ϕ1 − y sin α1 )/r. e= 1 1 + [M12 − (rω)2 ]. The first one considers specified stagnation conditions as pressure and temperature and extrapolates velocities and density from adjacent interior control volumes. For subsonic inlet flow. If the inlet flow is supersonic all properties must be taken from upstream because there is no communication of the flow properties from downstream. The angles of the meridional and tangential planes should be given by the user following the turbomachinery design specifications.CHAPTER 3. Wz = M1 cos α1 cos ϕ1 . α1 is the relative flow inlet angle based on meridional plane. γ(γ − 1) 2 where M1 is the relative flow inlet Mach number. Wy = M1 (y cos α1 sin ϕ1 + x sin α1 )/r.1 Inlet 99 The inlet conditions depend on the flow regime.
The inlet flow density is calculated from static pressure and static temperature.90) where the subscript in denotes the internal control volume and ~n is the outward unit normal vector.7. Suitable nonreflecting inlet and outlet boundary conditions were vastly studied by Giles [102] and [103] to solve this problem. which uses the outgoing Riemann invariant. a0 is the stagnation speed of .7. This affect directly the mesh size. Farfield and outlet nonreflecting boundary conditions are presented in sections 3. (3.2. respectively. defined as ~ in · ~n − 2ain . ain calculated by −R(γ − 1) ain = (γ − 1) cos2 θ + 2 ( 1 + cos θ s [(γ − 1) cos2 θ + 2]a20 (γ − 1)(R− )2 ) . it is important to build an extended computational domain at the compressor inlet to avoid problems with numerical oscillations due to the flow properties fluctuations. The methodology implemented in the code is based on the work of Holmes [104] for subsonic inlet.91) where θ represents the flow angle relative to the boundary. For high performance axial flow compressors.2. The Riemann invariant is used to determine either the absolute velocity or the speed of sound at the inlet boundary. Conversely. NUMERICAL FORMULATION 100 x and y are the coordinates of the centroid of the control volume and ω is the angular velocity.3 and 3. R− = W γ−1 (3.CHAPTER 3. nonreflecting formulation may be used if the inlet is at a short distance from the blade. that involves high velocities levels. At this point it is interesting to comment that boundary conditions at the inlet interface is commonly used with success.2.
Tt Pb ρb = . density with.7.1.2 Outlet The outlet conditions depend also on the flow regime. As commented in the section 3. For subsonic exit flow the static pressure is known and the static temperature is calculated using the density from adjacent interior control volume. The velocity is extrapolated from adjacent neighbors. the static temperature. ~ in  W (3. W and velocity are evaluated at the boundary. Therefore the internal energy and the total energy can be calculated. as follows a2in . Hence. the outlet boundary condition is enforced with . It should be ~ b  according observed that the inlet velocity components are obtained by decomposing W to the two aforementioned prescribed flow angles.7.2.92) (γ − 1) ~ 2 Win  .94) The values os stagnation inlet conditions Pt and Tt are fixed by the user. Tb = Tt a20 γ Tb γ − 1 Pb = Pt .93) cos θ = − and a20 = a2in + ~ in denoting the relative velocity. NUMERICAL FORMULATION 101 sound and ~ in · ~n W . 2 (3. For supersonic outlet. represented by subscript b.2. RTb p ~ b  = 2cp (Tt − Tb ). W (3. 3. all properties are extrapolated from adjacent interior control volume. pressure.CHAPTER 3.
The static pressure profile known. 3. Care must be taken to avoid reflection of perturbations back into the computational domain. The Riemann invariant [92].CHAPTER 3. four flow variables (ρ and the three velocity components) have to be extrapolated from the adjacent interior control volume. [105] and [106] was implemented to achieve the correct propagation of the . For this case the following formulation was inplemented pb = pout . the flow can either enter or leave the domain. Depending on the sign of the eigenvalues of the convective flux. ρ 0 a0 ρb = ρin + (3. pin is the static pressure calculated with the extrapolated variables. a20 (pb − pin ) Wxb = Wxin + nx . (pb − pin ) . ρ0 a0 (pb − pin ) Wyb = Wyin + ny . In this case. the information is transported out of or into the domain along the characteristics. ρ 0 a0 (pb − pin ) Wzb = Wzin + nz . were used in this work. for example. NUMERICAL FORMULATION 102 nonreflecting condition in the case of subsonic outflow.7. In the case of subsonic inlet.3 Farfield For farfield boundary condition the stagnation pressure and stagnation temperature. there are four incoming characteristics and one outgoing.2. the velocity (or Mach number) and flow direction and turbulence level must be given. For subsonic outlet the situation is inverse.95) where pout is the outlet static pressure specified by the user. from a streamline curvature program or from experimental work. Hence.
The calculations are performed at the face of the control volumes with freestream conditions: 1 qf ace · ~n = (Rin + R∞ ).99) qin · ~n = (Wx )in nx + (Wy )in ny + (Wz )in nz .101) Rin = q∞ · ~n − 2ain . af ace (3.102) For Mf ace  > 1 there is supersonic propagation of the information at the face.100) ~n represents the outward normal vector.CHAPTER 3. (3. γ1 (3. (3. Thus R∞ = q∞ · ~n − 2a∞ . a and M the speed of sound and Mach number. NUMERICAL FORMULATION 103 disturbances. q∞ · ~n = (Wx )∞ nx + (Wy )∞ ny + (Wz )∞ nz .97) Mf ace = qf ace · ~n . Those invariants use the propagation of information among the freestream properties set up at the ghost elements and its internal neighbors. 4 (3.96) 1 af ace = (γ − 1)(Rin − R∞ ). 2 (3. so that . γ1 (3.98) with the subscript ∞ standing for freestream and in for the adjacent interior control volume. respectively.
. (Wx )f ace = (Wx )in + (qf ace · ~n − qin · ~n)nx . (Wx )f ace = (Wx )in . (Wy )f ace = (Wy )in . NUMERICAL FORMULATION 104 if qf ace · ~n > 0 the quantities at face are extrapolated from interior of control volume: ρf ace = ρin . the quantities at the face are extrapolated from freestream ρf ace = ρ∞ . For Mf ace  < 1 there is subsonic propagation of the information at that face. (Wz )f ace = (Wz )in + (qf ace · ~n − qin · ~n)nz . Therefore if qf ace · ~n > 0 the quantities at that the face are extrapolated from the interior control volume: ρf ace = ργin a2f ace pin γ 1 γ−1 . (Wx )f ace = (Wx )∞ . (Wz )f ace = (Wz )in . (Wy )f ace = (Wy )∞ . if qf ace · ~n < 0. Otherwise.CHAPTER 3. (Wz )f ace = (Wz )∞ . (Wy )f ace = (Wy )in + (qf ace · ~n − qin · ~n)ny .
so that just only one blade channel is used. NUMERICAL FORMULATION 105 Otherwise.104) pf ace = ef ace = The conserved variables at the face are calculated by Qin + Qghost . if qf ace · ~n < 0 the quantities are extrapolated from freestream: ρf ace = ργ∞ a2f ace p∞ γ 1 γ−1 .103) pf ace 1 + ρf ace [(Wx )2f ace + (Wy )2f ace + (Wz )2f ace ]. for the ghosts: 3.7. (3. the periodicity condition must be applied to the blade inlet and outlet regions.2. For simulation of rotating frame of reference. Being the flow periodic. this type of . with significant reduction in computer memory and processing time. (Wz )f ace = (Wz )∞ + (qf ace · ~n − q∞ · ~n)nz . and the static pressure and total energy are calculated respectively by ρf ace a2f ace . γ (3.4 Periodic It is assumed that the channels formed by adjacent blades are identical. 2 (3. (Wx )f ace = (Wx )∞ + (qf ace · ~n − q∞ · ~n)nx .CHAPTER 3. (Wy )f ace = (Wy )∞ + (qf ace · ~n − q∞ · ~n)ny . γ−1 2 (3.105) Qghost = 2Qf ace − Qin .106) Qf ace = so that.
because the periodicity should provide terms that include the timeshift conditions across a blade passage and to establish the dependency of the dissipation and dispersion errors on the time integration and mesh dependence.CHAPTER 3. It is evident that if inlet distortion must be taken into account it is necessary to calculate the flow over the entire compressor inlet. Attention must be given to unsteady flows. these boundary conditions are usually called lagged periodic conditions. For turbomachines. considering that the faces of the control volumes with periodic conditions have the same properties. Wz = (Wz )in . θ = 2π/nblades. The reader may refer to [107] for more complex cases. Detailed explanation is given in [52]. In this work the periodicity condition implemented was treated as follows ρ = ρin . Wx = (Wx )in cos θ + (Wy )in sin θ. not only over one blade passage. as described by [102. where. e = ein . The matrix of rotation is used to transfer information of the conserved variables of both periodic sides. 108]. the subscript in stands for adjacent interior control volumes and nblades is the number of blades. . For unsteady flow the management of space and time periodicity is calculated using a ”time inclined computational grid” and the ”phase lag” condition. Wy = −(Wx )in sin θ + (Wy )in cos θ. NUMERICAL FORMULATION 106 boundary condition is straightforward to be implemented.
5 107 Wall Walls in a turbomachine are still or rotating. like the part of the still walls in a region of rotating meshes. For inviscid case for the ghost volumes. In order to set the flow velocity equal to the wall velocity (Wx )ghost = −(Wx )in . NonRotating Walls: Two wall boundary conditions were implemented to account for inviscid and viscous flows. as the outer casing and stator blades. 1. therefore still. For the viscous case.7. Both tangential and cross product are extrapolated from adjacent interior control volume. (Wz )ghost = −(Wz )in . Walls can be associated with the inertial frame of reference. Otherwise the walls are not rotating with the shaft. (Wy )ghost = −(Wy )in . as well as density and total energy. ~t is the tangential vector and ~c = ~n × ~t. Therefore. Wghost = (Wghost )n~n + (Wghost )t~t + (Wghost )c~c. rotor blades base and drums to which rotor blades are attached.2. ~n is the unit outward normal vector.107) where. NUMERICAL FORMULATION 3. Rotating walls are formed at every piece that is rotating with the turbomachine shaft: rotor blades. .CHAPTER 3. appropriate wall conditions must be set. all quantities are extrapolated from adjacent interior control volumes. The normal velocity of the real and ghost control volumes have opposite signals. They may be also associated to rotating frames. (3.
2. Hence. for example rotor blades. where the velocity vectors stand for absolute. 3.7. respectively and U ~ × ~r and ω ~ the angular velocity of the rotating frame of reference. Wy = −(Wx )in − 2xw ω. The flow properties are calculated based on the velocity tri~ +W ~ . .2. NUMERICAL FORMULATION 108 The remaining variables were set equal to the interior neighbor volume. Wx = −(Wx )in + 2yw ω.CHAPTER 3.5 can be used. Wz = −(Wz )in . (−yw (Wx )in + xw (Wy )in ) + rin ω e = ein + 2ρin rw ω rw with subscript w meaning wall and the variables xw and yw coordinates of the wall face centroid. Fixed Walls (NonRotating) in a Rotating Frame of Reference: This category of boundary condition is applied at casing of rotor rows that nonrotates with the frame of reference. with the inclusion of the term 2ρin rw (rin − rw )ω 2 in the energy equation due to the different radii of the element and its ghost neighbor. Rotor Walls: For walls with rotating frame. the flow properties are given by ρ = ρin . the wall boundary condition presented in Section 3. tangencial angle equation V~ = U ~ =ω and relative velocities.
usually lidway between the consecutive rows. i. Other disturbances associated with pressure oscillations may have strong influence if the blade spacing is small.2. the correct transfer of information between rotor and stator is complex. causing unsteady forces. At both sides of the surface. because the flow leaving the rotor rotates. Therefore.2. is defined in the vaneless space between two consecutive rows. two sets .e.7. steady or unsteady. 110] where taken into account to calculate the flowfield in the rotorstator interface. set up at the periodic walls.7. 109. 3.8 MixingPlane A surface.7 Rotor/Stator Interface The rotorstator interface requires careful attention. In this work. 30. 3. Dawes. Approximations are made although accuracy should not be lost. NUMERICAL FORMULATION 3. The methodology must represent the ability to capture flow disturbances caused by wakes from the upstream blade row that may produce significant unsteady lift on the following blade row and eventually lead to fatigue and material failure.7.CHAPTER 3. Depending on the physics of the problem.2. Chen and Belardini [29. causing the flow to be unsteady if observed from the stator. with numerical simulations. called mixingplane.6 109 Symmetry The symmetry conditions are the same for wall conditions for inviscid case. while the stator is still. the experiences of Denton. the numerical treatment is different. The steadystate simulation is implemented in this work. Chen et al [31] discuss in more details the problem. using the mixingplane approach.
3 and 3.4 – Representation of a mixingplane inlet on compressors The mixingplane is divided radially in strips. For control volumes contained in each radial strip the conserved variables are averaged and stored as constant values for those .CHAPTER 3. NUMERICAL FORMULATION 110 of ghost volumes are created to store circumferential averages of the properties of the flow entering and leaving the mixingplane.4 show the mixingplane at a rotor row outlet and stator row inlet respectively. Geometrical properties of those mixingplane ghost volumes are calculated. FIGURE 3.3 – Representation of a mixingplane outlet on compressors FIGURE 3. Figures 3.
111].109) or where. The disadvantage of the mixingplane is that nonlinear effects and convection phenomena may be lost and additional entropy created. One of the first approach to compute the mixingplane was described by Adamczyk et. N vol is the total number of control volumes and n is the time integration iteration number. N vol i=1 (3. Details of such technique can be seen in the work of Gerolymos [112]. The residue is calculated as the logarithm of RHS of the conserved equations based on the L2 norm as presented bellow v u 2 vol u 1 N X ∂Qi t . [108. . The average flow properties in the mixingplane are recalculated at each iteration.3 Numerical Stop Criteria The stop criteria is based on numerical residue or number of iterations that the user specified. Some authors [113.7. both axially and radially. what is expected to be implemented in the future in the code developed during this research. error = log10 (RHS) = log10 N vol i=1 ∂t (3. 3. 114] compute the mixingplane with body forces to account for the ”potential” interaction between closely coupled (staged) rows using deterministic stresses for the ”average” wake blockage and mixing effects. For steadystate simulations a very robust methodology is presented in reference [115]. NUMERICAL FORMULATION 111 volumes. The average quantities are transferred to the downstream ghost volumes.CHAPTER 3.108) v u vol u 1 N X t error = log10 (RHS) = log10 (Qni − Qin−1 )2 . al.
the monitoring of the outlet massflow.111) or where. NUMERICAL FORMULATION 112 Generally only continuity equation is used to calculate the residue. error = log10 (RHS) = log10 l N vol l=1 i=1 (3. l is the mass. For turbomachinery flow simulations another stop criteria was implemented. the solution was obtained. momentum and energy equations. . v u 2 5 N vol u 1 X X ∂Qik t error = log10 (RHS) = log10 N vol k=1 i=1 ∂t (3. For turbomachinery simulation.CHAPTER 3. Generally. Another option is calculate the difference Qn − Qn−1 for each conservation equation to accompany the decaying of each conserved variable during the iterations.110) v u vol 5 N u 1 X X t (Qnil − Qin−1 )2 . pressure ratio and efficiency are good option to observe the convergence. when the outlet massflow. pressure ratio and efficiency stay unaffected during the numerical iteration.
CFD methods are based on numerical solutions of the discretized governing equations through some iterative process using powerful computers. such as the implementation of new discretization methods. accelerating techniques.1 Code Structure and Implementation Issues Usually. the CFD solver developed in this work. At the current state of development.4 Computational Implementation CFD is based on numerically solving a set of partial differential equations that describes the conservation of some primary flow variables in time and space. boundary conditions and turbulence . turbulence modeling. 4. among many others. it is necessary to verify and validate the code in order to certify the capability of code to describe the physics of the flowfield correctly. CFD codes are made of a large number of subroutines. structured in a modular way to make simpler and easier further development. can be used to simulate external and mainly internal turbomachinery flowfields fairly realistically and accurate. Thus. mesh refinement. Many issues should be better understood before starting a CFD simulation: numerical stability. Before using a specific CFD code with confidence. Most flow problems using NavierStokes equations do not have an analytic solution. discretization process.
In this work the code was written in FORTRAN 90. 7. Flow type: internal or external. Inlet condition: subsonic or supersonic. 3. 9. write output files. 10.CHAPTER 4. the complexity of the program data structure then the vectorization of variables should be carefully analyzed. Ratio of turbulent eddy viscosity to molecular viscosity (νturbulent /νlaminar ) or turbulent intensity (only for turbulent flows). Fluid properties. If the target of code developers is to optimize memory storage. CPU time. 4. 8. solve system of differential equations (inviscid or viscous). calculate meshassociated geometrical parameters. Artificial dissipation method and its coeficients (only for centeredscheme). Although the program has been written for axial turbomachines. COMPUTATIONAL IMPLEMENTATION 114 models. fluxvector splitting or fluxdifference splitting. internal and external flows can also be solved. Selection of equations to be solved: Euler. Spatial discretization: centredscheme. The program is procedural and performs the sequence: read input data. An input file contains the appropriate values for: 1. 5. CF L number. Turbulent Prandtl number (only for turbulent flows). 2. Time discretization: RungeKutta or MacCormack. 6. NavierStokes or NavierStokes with turbulence model. .
Read mesh elements nodes. 12. 22. Number of stages (for turbomachinery calculation). internal or external flows. 13. Number of divisions at mixingplane (only for turbomachinery calculation). . IRS: active or deactive. 18. COMPUTATIONAL IMPLEMENTATION 115 11. 23. 21. 17. Simulation type: turbomachinery. 2. 16.CHAPTER 4. After reading these input parameters. 19. 3. 14. Rotational speed (for turbomachinery calculation). Read boundary faces nodes. Number of blades at each blade row (only for turbomachinery calculation). Read nodes coordinates. 20. Number of blades at each blade row (only for turbomachinery calculation). Nonreflecting boundary condition: active or deactive. outlet or freestream boundary conditions. the program runs the mesh data treatment consisting of the following steps: 1. Inlet. Reverse flow control (only for turbomachinery calculation). Number of iterations to store partial solutions. Maximum number of iterations. Tolerance: stop the calculations when residual has been reduced by a specific value. 15.
Identify periodic elements. Identify rotatingframe elements. Compute the volume of each element. The initial condition depends on the type of flow to be calculated: internal or external. 9. All rotating elements are flagged for the sake of elements flux calculation based on relative motion. 87] contain further details on the mesh procedures used in this work. 6. Having calculated the meshrelated properties. Each ghost element has the same characteristics of its actual neighbor element. and it is used for the calculation of the CF L number. 12. the initial and boundary conditions are set. For turbomachinery flow simulations the initial condition must comply with the complexity of flow physics. Compute the connectivity table. 10. 5. Identify all boundary conditions within the computational domain. The characteristic length is set as the minimum distance from the element centroid to the volume faces.CHAPTER 4. COMPUTATIONAL IMPLEMENTATION 116 4. Two frames of reference are used: fixed and rota . Compute the characteristic length for each element. 8. Compute the projection of each face at each direction and its normal vectors. 7. Boundary conditions are set at the ghost elements. References [83. 11. Create the ghost elements. with or without rotating frame of reference. Identify all element faces and their boundary conditions.
they cause numerical divergence. For compressors. the coordinates of C3 and C4 are used. understanding. unsteadiness. but for elements with rotational speed.1 – Scheme created to identify the number of blades For example. to identify the periodicity imposed in the second rotor . The CFD solver searching for the mesh elements without rotational speed that is the case of stator blade. the scheme created to identify the number of blades in each blade row is based on the axial coordinates C1 to C6 shown in Figure 4. For rotors. i. Those coordinates are taken from the streamline curvature program. Initial condition can be set based on previous estimates from a streamline curvature code. Inaccurate initial condition may drive the calculations to an unrealistic solution or to instable operation such as surge or choke. for example. surge or choke points the numerical solution will diverge in the case of steadystate regime simulation.e. COMPUTATIONAL IMPLEMENTATION 117 tional.CHAPTER 4. the same scheme is used. Problems involving these operation points are extremely unsteady and. the initial condition is more critical because if the initialization means some instable operation point as stall.1. The more accurate the initial guess the faster the convergence solution is achieved. In this work. to identify the periodicity imposed in the first stator (S1). FIGURE 4.
2. Boundary conditions at each boundary face.2 shows a sketch to better understanding.2 – Scheme created to identify the mixingplanes For example. the coordinates C5 and C6 are used. COMPUTATIONAL IMPLEMENTATION 118 (R2). the coordinates of M P 2a and M P 2b are used. 4.CHAPTER 4. The next step is to identify the element faces of all mixingplanes. Convective terms. the scheme created to identify the mixingplane elements follow the same idea of the periodic elements. defining the faces that belong to each surface. in order to calculate the average properties of conserved variables. Variable timestep for each mesh element. This procedure calculates: 1. FIGURE 4. In this work. 3. Figure 4. Mixingplane. . to identify the mixingplane between the first stator (S1) and second rotor (R2). The spatial and time integration of the equations in the conservative form is then carried out. both for the outflow and the inflow.
A TECPLOTlike file is also generated. 6. Implicit residual smoothing. 10. Convergence factor. 12. the program gives the user the possibility of delaying or specifying the frequency at which it is solved. 13. 11. Gradients. 8. The time integration uses the fivestep RungeKutta scheme with secondorder accuracy. Implicit time integration of the turbulence equation. Viscous terms are often evaluated only in the first step for computational efficiency. stops the loop and computes the flow variables for the postprocessing visualization. Turbulence diffusion term. COMPUTATIONAL IMPLEMENTATION 119 5. If satisfied. Viscous terms. Concerning the turbulence model. otherwise goes to first item. Turbulence production and dissipation terms. aiming reduction of CPU time. 14. 7. . The iterations might be stopped if either convergence or divergence is achieved or after a prescribed number of iterations. Turbulence advection term. 9. Time integration of the equations. Output files are written according to the user needs. Artificial dissipation terms.CHAPTER 4. The computational domain for an axial compressor or turbine is divided in segments for each blade row in order to distinguish the type row that rotates (rotor) or stays still (stator).
e}T . ρWy . a limiter function is also used at alternate steps. Van Leer (fluxvector splitting) or Roe (fluxdifference scheme). When an upwind scheme is chosen. In [116.CHAPTER 4. The artificial dissipation procedure is introduced at alternate steps. COMPUTATIONAL IMPLEMENTATION 120 Convective terms are computed at every step. The fourthorder operators (Biharmonic) are implemented to damp highwavenumber errors and the secondorder operators (Laplacian) to improve shock capturing. IRS has the effect of smoothing the highfrequency variations of the RHS of the conservation equations.3. ρWx .2. ρWy . the user can choose either the techniques of Jameson (centereddifference). Special attention must be paid to the matrix of conserved variables of Roe scheme. This smoothing can be applied either at alternate steps or at every step. For the implementation pressure and enthalpy are used instead of ρ and e: {p. ρWz . ρWx . as well as to increase the maximum operational CF L number. as presented in Section 3. Variable timestep based on the linear stability analysis of the discretized NavierStokes equations. For spatial discretizations using F DS. the artificial dissipation procedure is automatically skipped. A IRS method can be used to introduce implicitlike character to the solver. h}T . For space integration. is used to accelerate convergence. ρWz . The F V S method by Van Leer achieved accurate solutions (and is quite robust) and perform very well in the case of Euler equations. Its original form {ρ. 117] the authors carried out this scheme with viscous simulations and the results revealed that splitting errors in the NavierStokes and energy equations smear out the boundary layers and produce inaccurate . The dissipation terms in the Roe scheme are calculated using the Roematrix. Centered scheme does not inherently damp highwavenumber disturbances and hence requires the addition of artificial dissipation when applied to hyperbolic equations.
Dt + Dh (4.16(Re )DH . In this work.CHAPTER 4.3) . 2 (4. The SpalartAllmaras turbulence model calculates the variable ν˜ presented in Equation 2.2) where the Reynolds number is calculated based on the hydraulic diameter defined by DH = Dt2 − Dh2 . The turbulent intensity can be estimated by the user or calculated using the following equation −1/8 I = 0. the F V S was used only to start the inviscid terms calculation to obtain a reasonable velocity.1) where I is the turbulent intensity and l is the turbulent length scale. For turbulent flow simulations it is necessary to supply to the CFD solver the value of νT at boundary. To avoid this sort of problem. These modifications are not implemented in the CFD solver of current work. For internal flow νT must be given at inlet. (4. For initial condition the value of ν˜ can be estimated by ν/10. pressure and temperature fields and after some number of iterations the spatial discretization is switched by another (centered or FDS). Other option is to estimate the ν˜ value using the turbulence intensity [120] r ν˜ = 3~ W Il. The turbulent kinematic viscosity coefficient is used to calculate the turbulent dynamic viscosity coefficient necessary to calculate the terms of energy equation.48. COMPUTATIONAL IMPLEMENTATION 121 stagnation and wall temperatures. For external flow νT must be given at farfield. Van Leer [118] suggested modifications in the momentum flux terms and H¨anel [119] presented a modification of the momentum and energy equations and results for transonic airfoil.
6 present the formulation used to calculate the yp value based on y + . It is required to have at least one mesh point inside the laminar sublayer. (4. low turbulent intensity is around 1% and high turbulent intensity is around 10%. For turbulent flow simulations the viscous sublayer has to be resolved in an appropriate way to represent the physics correctly. this values are know from the design. µ (4.07DH . For modern windtunnel the turbulent intensity is very small (0.4δ99 .4) for fullydeveloped duct flow. In the case of wallbounded flows I = 0. COMPUTATIONAL IMPLEMENTATION 122 where DT is the tip diameter and DH is the hub diameter.05%).CHAPTER 4.6) where uτ is the skin friction velocity and it is calculated with r uτ = τw . In the case of turboamachinery. Equation 4. The length scale can be estimated by l = 0. The νT /ν ratio is lower than 10 for low turbulent intensity and greater than 10 for high turbulent intensity. (4.5) where δ99 is the boundarylayer thickness. To calculate the distance between the first node on the wall and wall surface (yp ). the assumption of flatplate turbulent flow is made. y+ = ρuτ yp .7) . Generally. ρ (4.
037W∞ 1/5 .CHAPTER 4. (Re )∞ As implemented.8) . the code can be used for internal and external flows. COMPUTATIONAL IMPLEMENTATION 123 and τw = 2 0. (4.
Several discretization schemes in time and space aiming at testing different artificial dissipation (for cellcentered spatial discretization) and upwind schemes. The validation was carried out through the study of a convergentdivergent nozzle and a NACA aerodynamic profile aiming at testing different spatial discretization schemes. space and artificial dissipation models in order to study the flatplate.5 Code Verification and Validation The code development was carried out in the following sequence: a Inviscid part. . using the convergentdivergent NASA nozzle and the NACA0012 in the transonic regime as guidelines. introducing time. This chapter deals with the verification of the adequacy of the code developed during this research. as well as with its validation. The verification was carried out for the flow over flatplate. Results of the turbulent flow in axial turbomachines are presented in the next chapter. b Laminar (viscous) part. aiming at the application to the design and design improvement of turbomachines. using SpalartAllmaras turbulence model for the study of NASA convergentdivergent nozzle. c Turbulent part.
there are publications available showing the validation process based on experimental results. the geometry of the compressor and turbine are usually not supplied due to restrictions on publishing design information. Modern blades design have complex camber angles and chord distributions from hubtotip in order to improve machine performance by decreasing internal flow losses. High performance turbomachines are a key on aircraft and industrial powerplant installations and business competition. The implementation of the code was validated in a stepbystep manner. the NavierStokes equations and turbulent flows were simulated. To obtain this correlations experimental tests must be carried out at test facilities. However. for turbomachinery cases. It is important to highlight that the geometry and mesh generation process was very time consuming due to the geometry complexity. Considerable amount of time was spent implementing the numerical methods to solve the governing equations in a 3 − D domain considering the full NavierStokes equations. validation and calibration of computational codes. . Afterwards. CODE VERIFICATION AND VALIDATION 125 The reference [121] presents a CFD code validation process and the definition of verification. Specifically. In order to test the CFD solver robustness and accuracy of the developed code both external and internal flows study cases were solved and compared to the literature. the Euler equations were implemented and tested. First. The stateoftheart on gas turbine components design is rarely described by researches with details.CHAPTER 5. The mathematical modeling of empirical correlations is vastly used by designers during the preliminary and refining design process.
44. two test cases were simulated. The nozzle test case is based on the experimental data of a convergingdiverging nozzle at static conditions described by Mason [122].1 shows the nozzle geometry and mesh. The mesh has 241 × 61 × 3 nodes in the x.1 126 Inviscid Case For Euler equations. 103 nodes representing 28.CHAPTER 5. The first calculation was carried out with pressure ratio equal to 2 between inlet and outlet.1 – Nozzle geometry and mesh (flow is from left to right) Stagnation pressure and temperature are set at inlet boundary and static pressure is specified at outlet boundary. Hence. Figure 5. The IRS method was applied at alternate RungeKutta steps. The artificial dissipation coefficients used were: ε(2) = 1/2 and ε(4) = 3/256. The symmetry boundary condition is used on both sides and the symmetry condition is imposed at the nozzle centerline. respectively. A centereddifference scheme was chosen for the spatial integration with the artificial dissipation model described in the last paragraph of Section 3.4. CODE VERIFICATION AND VALIDATION 5. Figure 5. but based on numerical simulation. FIGURE 5. presents the same nozzle analysis. The work developed by Carlson [123].2 shows the . The CF L number was set as 2. All nozzle geometrical details can be obtained from both references. namely a supersonic nozzle and the NACA0012 airfoil. Wall boundary condition is used at the nozzle upper surface. y and z coordinates. 800 control volumes.1.
2 1 p/p0 0. In the region after the nozzle throat. respectively.3 shows the experimental and the numerical results of both spatial discretization methods aforementioned at wall surface.4 0. The centereddifference scheme was applied with the artificial dissipation model of Equation 3.4 and 3.CHAPTER 5. The fluxdifference splitting scheme achieves very similar results for both δ values (0.6 0.8 0.007 and 0. For this simulation.5 0 0.2. Minor differences between the experimental and numerical results can be observed. The fluxdifference splitting was calculated with two different δ values as described in Section 3. 1. Figure 5. CODE VERIFICATION AND VALIDATION 127 experimental data and the numerical results obtained at wall surface.5 1 x/x0 FIGURE 5.2.inviscid flow The second calculation was carried out with pressure ratio equal to 5 between inlet and outlet. The Roe discretization scheme was applied using the piecewise linear reconstruction and the Venkatakrishnan’s limiter described in Sections 3. a centered and a Roe upwind scheme were used for the spatial integration.3 in order to to analyze the influence of this parameter on the convergence rate and solution.70.2 Experimental Numerical Solution 0 −1 −0.15).2.2 – Experimental and numerical results for static pressure ratio 2 .5. these differences are some due to the interaction .
007 FDS.CHAPTER 5.2 Experimental Centered−difference FDS.3 – Experimental and numerical results for static pressure ratio 5 for centered and upwind schemes . The Mach number contours for both centered and upwind schemes are presented in Figures 5. In the section 5.6 0.5 0 0.inviscid flow between the boundarylayer and the shock waves at near wall regions.5 and 5. Note that the parameter δ has a strong influence on the numerical behavior.4 0. CODE VERIFICATION AND VALIDATION 128 1.6. Oblique shock waves can be seen at the divergent nozzle section. the same case was calculated with the full NavierStokes equations plus the turbulence transport equation and the results after the nozzle throat were improved slightly. They are not detected because the viscous (or diffusive) terms are neglected in the Euler formulation.15 0 −1 −0. .8 0.2 1 p/p0 0. Figure 5.4 shows the convergence histories of the last case described (centered and Roe upwind schemes). δ=0.5 1 x/x0 FIGURE 5. This numerical residual is calculated based on Equation 3.109. δ=0.3.
inviscid flow . δ=0.007 FDS. CODE VERIFICATION AND VALIDATION 129 0 log10(RHS) −5 −10 −15 −20 Centered−difference FDS.CHAPTER 5.6 – Mach number contours for upwind scheme .4 – Continuity residue histories for different spatial discretization schemes inviscid flow FIGURE 5.15 −25 0 10000 20000 30000 40000 iteration 50000 60000 70000 80000 FIGURE 5.5 – Mach number contours for centered scheme FIGURE 5. δ=0.
7 – Mesh generated for NACA0012 airfoil flow calculation Periodic. An OGrid mesh type was generated as presented in Figure 5. IRS was applied at alternate RungeKutta steps. Other airfoil profiles and experimental results can be found in [124]. . 000 control volumes.4.1. CODE VERIFICATION AND VALIDATION 130 A transonic NACA0012 airfoil profile was simulated with Mach number 0. The Riemann invariants option was applied to increase the numerical robustness and to avoid unphysical numerical perturbations inside of the computational domain. The mesh generated has 236. Figures 5.9 show the pressure coefficient (cp ) distribution along the airfoil chord for both spatial discretization schemes. FIGURE 5. For this simulation. centered and Roe upwind schemes were chosen for the spatial integration.8 and zero angleofattack.2 using artificial dissipation coefficients: k (2) = 1/4 and k (4) = 1/256. wall and farfield boundary condition were used.7. The computational domain has 60 times the airfoil chord.CHAPTER 5.8 and 5. The CF L number used in this simulation is 1.5. The centereddifference scheme was calculated with the artificial dissipation model proposed by Jameson [76] and presented in the Section 3.
2 0.4 0.9 – Pressure coefficient distribution on the NACA0012 airfoil with zero angleofattack and Mn = 0.5 Experimental FDS.1 −1 0 0.8 1 x/c FIGURE 5.6 0.8 – Pressure coefficient distribution on the NACA0012 airfoil with zero angleofattack and Mn = 0.4 0.6 0. CODE VERIFICATION AND VALIDATION 131 1 Cp 0.13 .10 and 5.5 0 −0.11.12 and 5.CHAPTER 5. δ=0.8: upwind The shock wave formation can be visualized in Figures 5.2 0. that show the Mach number contours for both spatial discretization schemes.8 1 x/c FIGURE 5.5 Experimental Centered−difference −1 0 0. Figures 5.5 0 −0.8: centereddifference 1 Cp 0.
11 – Mach number contours for upwind scheme 132 .10 – Mach number contours for centered scheme FIGURE 5. CODE VERIFICATION AND VALIDATION show the static pressure contours.CHAPTER 5. FIGURE 5.
12 – Static pressure contours for centered scheme FIGURE 5. CODE VERIFICATION AND VALIDATION FIGURE 5.13 – Static pressure contours for upwind scheme 133 .CHAPTER 5.
The number of control volumes inside of the boundarylayer region is very important to obtain an accurate boundarylayer resolution. The green part of the mesh shows the computational domain over the flatplate surface. y and z coordinates respectively. 444 nodes representing 99.CHAPTER 5. 134.14 shows the approach used to discretize the physical domain. This small distance is necessary to capture the correct boundarylayer thickness.6. CODE VERIFICATION AND VALIDATION 5. The mesh has 551 × 61 × 4 nodes in the x. The viscous terms were calculated based on the centered scheme presented in Section 3. The distance between the first node on wall surface is 2 · 10−6 m. Note that the mesh is finer close to the flatplate leading edge.2. Figure 5.14 – Mesh used to calculate the flow on a flatplate The part of the mesh in blue shows the computational domain in front of the flatplate leading edge. An exponential function was used to stretch the mesh smoothing along the j direction. The laminar Re number used was 5 · 105 with M∞ = 0. The numerical results are compared to the analytical solution presented in [125].2 134 Laminar Case The laminar flow test cases were carried out to validate the CFD solver. The flatplate length was calculated based on Re number and freestream Mach number. Hence. .3. The well known Blasius velocity profile of a flatplate was calculated numerically by the CF D solver. The solver verification for the laminar case was carried out for different flux computation schemes and artificial dissipation models. FIGURE 5. 000 control volumes.
The Riemann invariants option was used to increase the numerical robustness and avoid unphysical numerical perturbations inside of the computational domain. A second case was calculated with the MacCormack timeintegration method as described in Section 3.15 shows the continuity numerical residue history (convergence rate). The CF L number used in this simulation is 1.17 and 5.1.4. .1. The artificial dissipation coefficients used are the same presented in Section 3. respectively.70.7 for the fluxdifference scheme.1 without IRS and CF L = 0. The second case was solved with 15 nodes inside of the boundarylayer.3. The centereddifference scheme was calculated with artificial dissipation model proposed by Mavriplis and Jameson presented in the Equation 3. Figure 5. The IRS was applied only for centered scheme in alternate RungeKutta steps. Figures 5.16 shows the analytical and numerical solutions at 50% of flatplate distance from leading edge.2) The first case was solved with 10 nodes inside of the boundarylayer.CHAPTER 5. Figure 5.3. The centereddifference scheme was used for the spatial integration artificial dissipation model described in the last paragraph of Section 3.4. CODE VERIFICATION AND VALIDATION 135 Symmetry.18 show the continuity numerical residue and the analytical and numerical solutions at 50% of flatplate distance from leading edge. The coordinates x and y are represented by r η=y U∞ νx (5.1) f 0 (η) = u U∞ (5.5 for the centered spatial discretization scheme and 0. wall and farfield boundary conditions were used.
2 Analitical Solution Centered−difference 0 0 2 4 6 8 10 η FIGURE 5.16 – Analytical and numerical solutions of flow with M∞ = 0.8 0.20 show the continuity numerical residue and .CHAPTER 5.1.19 and 5.6 0.15 – Continuity residue history: first case 1. CODE VERIFICATION AND VALIDATION 136 −2 −3 log10(RHS) −4 −5 −6 −7 Centered−difference −8 0 5000 10000 15000 20000 25000 iteration 30000 35000 40000 45000 50000 FIGURE 5.3 on a flatplate with 10 nodes inside of the boundarylayer: first case The third case was solved with 15 nodes inside of the boundarylayer.4 0.4. Figures 5.2 1 f’(η) 0.2. The centereddifference scheme was calculated with artificial dissipation model proposed by Jameson [76] in Section 3.
1. The high order fluxdifference splitting scheme was calculated with the parameter δ = 0.17 – Continuity residue history: second case 1. The fourth case was solved with 10 nodes inside of the boundarylayer. CODE VERIFICATION AND VALIDATION 137 −2 −3 log10(RHS) −4 −5 −6 −7 Centered−difference −8 0 5000 10000 15000 20000 25000 iteration 30000 35000 40000 45000 50000 FIGURE 5.CHAPTER 5.18 – Analytical and numerical solutions of an flow with M∞ = 0.2 Analitical solution Centered−difference 0 0 2 4 6 8 10 η FIGURE 5.4 0.2 1 f’(η) 0. For this .8 0.6 0.3 on a flatplate with 15 nodes inside of the boundarylayer: second case the analytical and numerical solutions at 50% of flatplate distance from leading edge.
6 0.8 0. Figures 5.19 – Continuity residue history: third case 1.22 show the continuity numerical residue and the analytical and numerical solutions at 25%.2 Analitical Solution Centered−difference 0 0 2 4 6 8 10 η FIGURE 5.20 – Analytical and numerical solutions of an flow with M∞ = 0.2 1 f’(η) 0.CHAPTER 5.3 on a flatplate with 15 nodes inside of the boundarylayer: third case case. the mesh size is five times lower than the three previous cases.4 0. 50% and 85% of flatplate distance from leading edge. . CODE VERIFICATION AND VALIDATION 138 −2 −3 log10(RHS) −4 −5 −6 −7 Centered−difference −8 0 5000 10000 15000 20000 25000 iteration 30000 35000 40000 45000 50000 FIGURE 5.21 and 5.
4 0.3 on a flatplate with 10 nodes inside of the boundarylayer: fourth case The fifth case was solved with 12 nodes inside of the boundarylayer. CODE VERIFICATION AND VALIDATION 139 −2 −3 log10(RHS) −4 −5 −6 −7 FDS −8 0 5000 10000 15000 20000 25000 iteration 30000 35000 40000 45000 50000 FIGURE 5. The solver settings are the same as the fourth case.24 show the continuity numerical residue and the analytical and numerical solutions at 50% of flatplate distance .22 – Analytical and numerical solutions of an flow with M∞ = 0.23 and 5.2 1 f’(η) 0.CHAPTER 5.6 0.8 0.2 Analitical Solution FDS: 25% of flat−plate FDS: 50% of flat−plate FDS: 85% of flat−plate 0 0 2 4 6 8 10 η FIGURE 5. Figures 5.21 – Continuity residue history: fourth case 1.
2 1 f’(η) 0.4 0.CHAPTER 5. −2 −3 log10(RHS) −4 −5 −6 −7 FDS −8 0 5000 10000 15000 20000 25000 iteration 30000 35000 40000 45000 50000 FIGURE 5.24 – Analytical and numerical solutions of an flow with M∞ = 0.23 – Continuity residue history: fifth case 1. The solver .2 Analitical Solution FDS 0 0 2 4 6 8 10 η FIGURE 5.8 0.3 on a flatplate with 12 nodes inside of the boundarylayer: fifth case The sixth case was solved with 15 nodes inside of the boundarylayer. CODE VERIFICATION AND VALIDATION 140 from leading edge.6 0.
To make the fluxdifference splitting scheme faster per iteration .26 and 5. respectively. This case was started with the solution obtained in the second case. Comparing the results of the first and fourth cases it is possible to conclude that the fluxdifference splitting scheme is very robust and accurate to capture the boundarylayer with a smaller number of nodes (inside of the boundarylayer) compared to the centered scheme.25 show the analytical and numerical solutions at 50% of flatplate distance from leading edge. CODE VERIFICATION AND VALIDATION 141 settings are the same as the fourth case.4 0.2 Analitical Solution FDS 0 0 2 4 6 8 10 η FIGURE 5.CHAPTER 5. However.25 – Analytical and numerical solutions of a flow with M∞ = 0. Figure 5. The centered scheme is more dissipative than the fluxdifference splitting scheme.6 0. but the mesh size is the same of the second case (99.2 1 f’(η) 0. in the CFD solver developed in this work. the Roe scheme demands about 1.3 on a flatplate with 15 nodes inside of the boundarylayer: sixth case Figures 5. This happens due to the nature of the centered scheme. 1.25 times the machine time of the centered scheme.000 control volumes).27 show the details of the Mach number contours and the velocity vectors of the flow on the flatplate simulated.8 0.
. CODE VERIFICATION AND VALIDATION 142 FIGURE 5.27 – Details of the velocity vectors profiles inside of the boundarylayer on the flatplate the gradients of the limiter function can be calculated only at alternate RungeKutta timeintegration steps. For centereddifference scheme it is possible to start the solution with higher values for the artificial dissipation coefficients. However.26 – Details of the Mach number contours inside of the boundarylayer on the flatplate FIGURE 5. Each iteration of the MacCormack timeintegration is more faster than the five steps RungeKutta scheme. However. these values must be decreased as the calculation progresses in order to avoid too high numerical dissipation.CHAPTER 5. the MacCormack timeintegration can be very instable.
4 and 3. This limitation is not physical.28 shows the mesh generated for nozzle geometry.4. FIGURE 5. The fluxdifference splitting was calculated with δ = 0. 5.2. respectively.CHAPTER 5.2. 400 control volumes.1 using the piecewise linear reconstruction and Venkatakrishnan’s limiter described in Sections 3. For this simulation. The centereddifference scheme was calculated with the artificial dissipation model proposed by Jameson [76] in Section 3. .1.1.3) is relatively low for a compressible code (high velocities).turbulent flow The calculation was carried out with pressure ratio 5 between inlet and outlet.29 shows the experimental and numerical results for both spatial discretization methods aforementioned. respectively.5.2. centered and Roe upwind schemes were chosen for the spatial integration.3 Turbulent Flow Case A turbulent flow is calculated at the nozzle presented in Section 5. The mesh used has 281×91×3 nodes in the x.5. without IRS. CODE VERIFICATION AND VALIDATION 143 in the case of laminar simulation. Incompressible treatment should be used for low Mach number simulations. y and z coordinates. Figure 5. 713 nodes representing 50. 76. for CF L greater than 0. Figure 5. The numerical errors may dominate the very small variations of the density and compressibility effects.28 – Mesh generated for nozzle flow calculation . The Mach number of the flatplate (M∞ = 0. It is important to mention that the CFD solver developed in this work aims the solution of compressible flows.
CHAPTER 5. δ=0. −2 log10(RHS) −3 −4 −5 −6 Centered−difference FDS.1 −7 0 10000 20000 30000 40000 iteration 50000 60000 70000 80000 FIGURE 5. CODE VERIFICATION AND VALIDATION 144 1.5 0 0.31.5 1 x/x0 FIGURE 5.2 1 p/p0 0. Oblique shock waves can .2 Experimental Centered−difference FDS.turbulent flow Figure 5.8 0. δ=0.30 shows the convergence histories during the numerical iteration.29 – Comparison of experimental and numerical results for static pressure ratio 5 for centered and upwind schemes .4 0.30 – Continuity residue histories for two spatial discretization schemes The Mach number contours are presented in Figure 5.6 0.1 0 −1 −0.
turbulent flow The turbulent flow solutions of the nozzle presented accurate results compared to the experimental results. for both spatial discretization scheme.31 – Mach number contours .CHAPTER 5.32 shows the velocity vectors in the boundary layer region.32 – Velocity vectors . CODE VERIFICATION AND VALIDATION 145 be seen at the divergent nozzle section. including those in the region after the nozzle throat. Figure 5. . FIGURE 5.turbulent flow FIGURE 5.
A long way is necessary until start the CFD calculations. Based on a 3 − D geometry. no experimental data are available for comparison. A 5stage high performance axialflow compressor as part of an ongoing design of the gas generator of a 1 MW class turboshaft. From the results obtained from the CFD simulation. because of dimensions at leading and trailing edges being very small. it is possible to point out pitfalls and hence improve the design. surfaces.1 Flow Simulation in Turbomachines CFD for turbomachines is inherently complex and laborious due to the complexity of the geometry and physical aspects. vertices and edges to generate the respective blocks around blades and channels. . 6. creating curves. the process is more difficult. each projected geometry must be carefully analyzed through specific tools. The design process is a complex and laborious task by itself. a mesh must be generated. The mesh refinement on these regions is critical for turbulent flows.6 Results The study described in this chapter refers to the design and improvement design of axial turbomachines. points. To obtain a final 3 − D geometry. Therefore. For compressors. CFD should be used as an aid tool to find an optimal geometry.
Hgrids . Dunham and Came [127]. • Inlet total temperature = 1100 K. Kacker and Okapu [128] and Moustapha and Kacker [129] were used. • Rotational speed = 14900 rpm. but the quality and the aspect ratio of the elements when the mesh smoothing process is applied must be guaranteed to obtain a good mesh as close as possible of the orthogonal form. 6. The three mesh types most commonly used for turbomachinery domain discretizations are: Hgrid.1 SingleStage AxialFlow Turbine Simulation A singlestage axialflow turbine designed using the meanline technique was calculated. Loss corrections by Ainley and Mathienson [126].1. 000 Pa. The geometry was obtained using an inviscid formulation with the addition of loss models developed for turbine design. • Efficiency = 88%. • Massflow = 22. 6. RESULTS 147 Not only the refinement. The parameters used in the design are: • Inlet total pressure = 410. • Outlet static pressure = 220. 700 Pa.2 and 6.3. HOHgrid and Ogrid. • Pressure ratio = 1. The 3 − D views of the turbine are showed in Figures 6.1.7 kg/s.CHAPTER 6. Each one with particular characteristics.61.
Generally. Figure 6.2 – 3D solid drawing of the NGV and rotor of a singlestage axialflow turbine are vastly used due to their relative simplicity compared to both HOH and Ogrids [29].CHAPTER 6. RESULTS 148 FIGURE 6.1 – Auxiliary curves to drawing a singlestage axialflow turbine FIGURE 6. The problem of the Hgrid is that the curvature of the leading and trailing edges are partially lost. To try to overcome this problem a very refined grid should be generated in this region. HOH and Ogrids are superior because the leading and trailing edges are more accurately represented by the mesh.4 shows the Hgrid generated in .
Note that. Ogrids are generated only around the .5 shows the mesh close to the stator leading edge.3 – Full 3D view of an singlestage axialflow turbine a 3 − D singlestage axialflow turbine. FIGURE 6.CHAPTER 6.4 – Singlestage axialflow turbine Hgrid domain Figure 6. HOHgrids use a different blocking scheme. the blade leading edge radius is not well represented. RESULTS 149 FIGURE 6.
The HOHgrid type is commonly used for turbine blades. RESULTS 150 FIGURE 6. FIGURE 6.6 shows a HOHgrid block.7.CHAPTER 6. Figure 6.6 – Singlestage axial turbine HOHgrid domain A closeup of the improved curved part is presented in Figure 6. an Hgrid is generated between the blades. Note that the Ogrid is generated only on the blade surface.5 – Closeup of the leading edge of an Hgrid suction and pressure blade surfaces. Hence. For compressor blades the .
8 – Singlestage axialflow turbine Ogrid domain The Ogrid blocking has an advantage over the HOHgrid. The blade surface is . FIGURE 6. An accurate mesh smoothing process can be obtained using exponential or parabolic functions. the Ogrid is generated around the blade geometry and the distance of the elements (or nodes) is smoothly distributed along the channel between blades.7 – Closeup of the stator trailing edge and rotor leading edge of a singlestage axialflow turbine mesh type. Figure 6. For this mesh type. depends on the blade thickness and the dimension of the leading edge. RESULTS 151 FIGURE 6.8 shows an Ogrid block.CHAPTER 6.
the CFD simulations can be carried out.9 shows an Ogrid closeup. The periodic condition is applied to the mean distance between blades along the blade chord (Ogrid). Figure 6. Figure 6. An Ogrid with 167. sometimes it is not trivial to divide the blade in two different surfaces. Note that the massflow is negative at the beginning. because the blade does not need to be divided in two parts (suction and pressure surfaces). As initial conditions the velocities and the rotational speed were set equal to zero and the stagnation pressure and temperature were set equal to .10 shows the massflow behavior along the numerical iterations. Otherwise.CHAPTER 6. To verify the implementation of the mixingplane in the region between a nonrotating frame (stator) and a rotating frame (rotor) of reference. RESULTS 152 continuous. FIGURE 6. The massflow outlet were monitored and are presented in Figure 6. 756 nodes was used. This happened due to the numerical initialization process used for this case.10.9 – Closeup of the Ogrid around the stator and rotor blades of a singlestage axialflow turbine Assuming that appropriate meshes are available. For blades with high camber angle. only inlet and outlet regions would be enforced by the periodicity (H or HOHgrids). the flow within a singlestage axialflow turbine was calculated using an inviscid formulation.
It is possible to observe the gas expansion along the turbine stage. The rotor outlet static pressure was decreased during the first 2.inviscid case . 000 iterations. The converged solution presented a massflow equal to 17.98 kg/s. RESULTS 153 the inlet condition values.11 shows the static pressure contour at two planes of a singlestage axialflow turbine. 50 40 Mass−Flow (kg/s) 30 20 10 0 Outlet Mass−Flow −10 0 5000 10000 15000 20000 25000 iteration FIGURE 6.10 – Singlestage axialflow turbine massflow outlet convergence history inviscid case Figure 6. The rotational speed was linearly increased during the first 5.CHAPTER 6. FIGURE 6.11 – Static pressure contour of a singlestage axialflow turbine . 000 iterations until reaching the static pressure prescribed in the design.
This is the reason for the massflow outlet calculated in the CFD solver to be different of the turbine design massflow. following the design parameter. Figure 6.13 shows the velocity vectors at the rotor outlet and reverse flow region is observed (red cricle). note that at the stator the velocity vectors have the same directions of the blade outlet and at the rotor inlet the relative flow angle is zero. The massflow ratio between the mixingplane outlet of the stator and the mixingplane . In the rotor. However. A more detailed investigation should be done. FIGURE 6.CHAPTER 6. The figure also shows that the flow is well aligned with the stator blade as expected. 446 nodes was generated and it is presented in Figure 6. This is a consequence of the artificial viscosity addition. this viscosity is numerical and not physical.12 – Closeup of the flow across the mixingplane of a singlestage axialflow turbine .14. well distributed velocity field can be observed. the velocity vectors are incorrect due to the reverse flow presented in this simulation.12 shows the flowfield across the mixingplane from stator outlet to rotor inlet.inviscid case In the stator. The velocities angles and directions at the stator outlet and rotor inlet are in agreement. in which is inconsistent with the Euler formulation. an Hgrid with 313. To calculate the NavierStokes equations. RESULTS 154 Figure 6.
15 and 6. and the rotor outlet massflow were monitored during the iterations and are presented in Figures 6. The numerical inilialization used in this case is the same of the inviscid case.13 – Closeup of the flow at rotor suction and pressure surfaces . RESULTS 155 FIGURE 6.16 shows a suitable massflow value when the numerical convergence is reached.14 – Hgrid used for a singlestage axialflow turbine inlet of the rotor.17 shows the velocity vectors in the stator domain. Figure 6.96 kg/s. respectively. .CHAPTER 6.inviscid case FIGURE 6. Figure 6. The converged solution presented a massflow equal to 19.16.
15 – Mixingplane outlet (MPO) and mixingplane inlet (MPI) ratio . a small region with reverse flow was detected at the blade suction side.6 0. RESULTS 156 1.5 Mixing−plane mass flow ratio 0. Figure 6. .2 1.turbulent case with Hgrid At the rotor.4 0 5000 10000 15000 20000 25000 iteration FIGURE 6.7 0. close to the trailing edge.8 0.18 shows this reverse flow region in detail (red circle).1 MPO/MPI 1 0.16 – Singlestage axialflow turbine massflow outlet convergence history of the .CHAPTER 6.turbulent case with Hgrid 40 Mass−Flow (kg/s) 20 0 −20 −40 Outlet Mass−Flow 0 5000 10000 15000 20000 25000 iteration FIGURE 6.4 1.3 1.9 0.
RESULTS 157 FIGURE 6. FIGURE 6.turbulent case with Hgrid The problem of reverse flow can be solved at the turbine design process. No modification was made on the stator blade. The mesh used has 538.turbulent case with Hgrid An Ogrid was used to solve the NavierStokes equations at the singlestage axialflow turbine with the modification in the rotor blade geometry above mentioned.17 – Singlestage axialflow turbine velocity vectors distribution along the stator blade row . In this case.50◦ to −50. The size of the . Modifications on the rotor should be made to avoid this problem. 721 nodes.19.18 – Closeup of the reverse flow at the rotor suction side .CHAPTER 6. the thicknesschordratio was increased and the rotor exit blade angle was changed in two degrees (−52.50◦ ). Its mesh block is presented in Figure 6.
CHAPTER 6. Figure 6.20 – Detail of the Ogrid on the turbine casing The mesh generated on the rotor tip clearance region is presented in the Figures 6. . as shown in Figure 6. In the tip clearance region some blocks must be created between the rotor tip and the turbine casing.19 – Scheme of the blocks created to generate the Ogrid FIGURE 6.23. FIGURE 6.21 show the gap between the rotor blade and the casing.00075 m. RESULTS 158 rotor tip clearance is 0.19.22 and 6.20 shows a closeup of the mesh generated on the casing and Figure 6.
The initialization is based on the results of the design procedure. The rotor blade is considered running with design tip speed. temperature and velocity are known. In this case. but the radial equilibrium equation could be used. it is necessary to prescribe the static pressure at the hub or at the tip. an accurate distribution of pressure. RESULTS 159 FIGURE 6.22 – Mesh elements distribution on the tip clearance region The numerical initialization. follows a better methodology compared to the last cases. Hence. The static pressure at the outlet is constant at all iterations.21 – Detail of the gap between the turbine rotor and casing FIGURE 6.CHAPTER 6. . for this case.
35 30 Mass−Flow (kg/s) 25 20 15 10 5 Outlet Mass−Flow 0 0 2000 4000 6000 8000 10000 iteration 12000 14000 16000 18000 FIGURE 6.24 – Singlestage axialflow turbine massflow outlet convergency history turbulent case with Ogrid For this case. RESULTS 160 FIGURE 6. the pressure ratio was also monitored. Figure 6.24 shows the massflow outlet monitoring during the iterations.25 shows the massflow ratio between the mixingplane outlet of the stator and the mixingplane inlet of the rotor. Note that the outlet massflow is always positive. .CHAPTER 6. The design pressure ratio is 1.26.59 as shown in Figure 6.23 – Detail of the Ogrid around the rotor blade tip and the refinement of the clearance region Figure 6.62 and CFD calculated pressure ratio is 1.
Figure 6. due to geometrical improvements on the rotor blade.4 1. .08 MPO/MPI 1.1 1.2 1.26 – Singlestage axialflow turbine pressure ratio monitoring .1 pressure ratio 1 0 2000 4000 6000 8000 10000 iteration 12000 14000 16000 18000 FIGURE 6.98 0 2000 4000 6000 8000 10000 iteration 12000 14000 16000 18000 FIGURE 6.turbulent case with Ogrid The problem of reverse flow is not present anymore.5 1.06 1.7 1.04 1.02 1 Mixing−plane mass flow ratio 0.27 shows the velocity vectors in the turbine rotor region. RESULTS 161 1.3 1.25 – Mixingplane outlet to mixingplane inlet ratio .6 pressure ratio 1.CHAPTER 6.turbulent case with Ogrid 1.
turbulent case with Ogrid The static pressure along the turbine stage decreases until it reaches the static pressure prescribed at outlet boundary.CHAPTER 6.turbulent case with Ogrid The Coriolis force can be observed acting in the velocity vectors close to the rotor wall.27 – Distribution of the velocity vectors in the turbine rotor . as presented in Figure 6.28. RESULTS 162 FIGURE 6. Figure 6. FIGURE 6.28 – Effect of Coriolis force in the velocity field close to the turbine rotor wall .29 shows a global view of the static pressure .
30 – Detail of the static pressure contours along the turbine stage . FIGURE 6.turbulent case with Ogrid FIGURE 6.CHAPTER 6.turbulent case with Ogrid Figure 6. RESULTS 163 contours in two planes along the turbine stage and Figure 6.29 – Global view of the gas expansion along the turbine stage . The mixingplane .31 shows the velocity vectors across the mixingplane.30 shows the details of the static pressure contours calculated by CFD solver.
31 – Detail of the static pressure contours along the turbine stage . The transfer of information from the downstream to the upstream is accurately calculated. The leakage flow arises due to a pressure difference between the two surfaces of the blade at the tip.turbulent case with Ogrid Figure 6. Sometimes the effect of leakage flow wash . high velocity or separation zone may diffuse the leakage flow before it forms a vortex.32 shows the tip clearance effect and the flow leakage in the region between the rotor blade pressure side to the suction side. The secondary and leakage flows oppose each other. The leakage flow tends to roll up into a vortex. which does not participate in the energy conversion process. The clearance between the rotor and casing wall induces leakage flow across the gap. RESULTS 164 numerical implementation presents accurate results for all cases solved. The massflow through this gap. depends on the gap height. FIGURE 6. The formation of leakage has been observed in turbomachinery and cascades. in some cases the conditions necessary for the formation of a strong vortex may not exist because high turbulence levels.CHAPTER 6. However. Because the leakage jet and the main flow are at different angles. a flow discontinuity exists.
The parameters used in the design are: • Inlet total pressure = 101.2 Single Rotor with Low Aspect Ratio A single rotor designed using meanline technique was calculated. RESULTS 165 out the separated flow. • Outlet static pressure = 135. The leakage flow and its interaction with other flow features is a complex phenomenon.CHAPTER 6. .1. • Efficiency = 86%. 325 Pa. FIGURE 6. All geometry was obtained using an inviscid formulation with the addition of loss models developed for axial compressor design. which is beneficial.turbulent case with Ogrid 6. The Koch and Smith loss correlations were used [23]. • Massflow = 2.4 kg/s.32 – Detail of the leakage flow from pressure surface to suction surface of the turbine rotor . • Inlet total temperature = 300 K. 000 Pa.
• Rotational speed = 13950 rpm. Figure 6.39 and 6. FIGURE 6.38 shows the elements distribution along the rotor chord and from hubtotip. FIGURE 6.40 present the rotor outlet massflow and the pressure ratio along the iterations.35 – 3D Ogrid around the rotor blade Figure 6.CHAPTER 6. A 3D view of the rotor row is presented in Figures 6.35 show the 3−D mesh and Figures 6.34. respectively. RESULTS 166 • Pressure ratio = 1. . Note that in the axis and casing the mesh is refined and smoothly distributed by an exponential function.37 show detail of the mesh generated on the rotor leading edge and the mesh around the rotor hub. The mesh size is of 453543 nodes.34 – Detail of the rotor blade The Ogrid mesh type was generated around the rotor row. Figures 6.59.33 and 6.36 and 6.33 – 3D view of the rotor FIGURE 6.
CHAPTER 6. RESULTS FIGURE 6.36 – 3D view of the rotor FIGURE 6.38 – 3D Ogrid around the rotor blade 167 .37 – Detail of the rotor blade FIGURE 6.
turbulent case with Ogrid The massflow calculated by CFD is 2.5 0 5000 10000 15000 iteration 20000 25000 30000 FIGURE 6.6 2.6 pressure ratio 1.25% for the pressure ratio.56 1.52 RHS residual 1. The source of this discrepancy is the .turbulent case with Ogrid 1.7% for the massflow and 3.36kg/s and the pressure ratio is 1.4 2.54 1.8 2.39 – Rotor outlet massflow convergence history .54. RESULTS 168 3 Mass−Flow (kg/s) 2.40 – Rotor pressure ratio .58 1.CHAPTER 6.2 Outlet Mass−Flow 2 0 5000 10000 15000 iteration 20000 25000 30000 FIGURE 6.40kg/s and P r = 1.59) the discrepancy is of 1. Compared to the design values (m ˙ = 2.
To improve the results the loss model must be recalibrated.CHAPTER 6. Blockage coefficient and loss distribution from hubtotip should be changed.41 – Increase of static pressure along the rotor blade . The design was calculated using an inviscid formulation and the losses were calculated with empirical formulation.41 shows the static pressure increase along the rotor blade and Figure 6. Note that the computational domain is quite deformed in the cartesian space.42 – Static pressure in two different planes (near of hub and near of tip) turbulent case with Ogrid .turbulent case with Ogrid FIGURE 6. This makes difficult to obtain a clear cut plane parallel to the rotor axis. Despite the discrepancy.42 shows a detail of the static pressure on two different planes. FIGURE 6. RESULTS 169 formulation used. Figure 6. the design results agree with the CFD calculation.
The preliminary design of an axial compressor is the first step to obtain the operational characteristics and the preliminary dimensions.turbulent case with Ogrid 6. is described.43 shows the velocity vectors distribution at the rotor leading edge. In this work. an axial compressor design is presented.43 – Velocity vector distribution near of the rotor leading edge . . Note the effect of the Coriolis force at the rotor suction surface. basic compressor dimensions are calculated accounting internal flow losses using empirical correlations. from preliminary design to the CFD calculation. the results of the compressor performance were used to calculate the engine running line using an engine deck [4] to analyze the gas turbine behavior when equipped with this machine. The procedure. FIGURE 6. At this point. The inlet and outlet blade angles and the flow properties are calculated at the compressor blade meanline [1]. RESULTS 170 Figure 6. The main steps of a typical turbomachinery aerodynamic design are described bellow: • Specification of design parameters (based on gas turbine cycle calculation).CHAPTER 6.2 Multistage AxialFlow Compressor Simulation In this section.
the preliminary design tools remain important in the compressor aerothermodynamic design. • 3 − D flow calculation. deviation. de Haller number. • Streamline curvature method (throughflow calculation). The compressor geometry is calculated and analyzed again.CHAPTER 6. with minimum development time and cost. Several parameters must be observed to obtain an efficient turbomachinery. The final geometry is obtained transferring the results of the streamline curvature code to a CAD software to draw the 3D compressor view. degree of reaction. incidence. Hence. despite of the great simplicity of the methods used. If the values of design parameters (diffusion factor. no amount of subsequent CFD computational effort or development testing will enable the performance to be achieved. The streamline curvature code is used to refine the preliminary design based the previous results. RESULTS 171 • Preliminary design (meanline). loss mechanisms among others) are not consistent with the performance required by the customer. A compressor is usually designed to achieve optimum performance for the customer. makes it possible to investigate all of the interesting parameter options by using CFD simulations. The large range of possible turbomachinery geometry. . To simulate the flow a mesh must be generated and use as input of the CFD solver. It is very important to eliminate inadequate designs and to obtain a near optimum configuration during the preliminary design process before proceeding with refinements using streamline and CFD. particulary in multistage machines.
• Inlet Mach number: 0. • Pressure ratio: 5.2 Preliminary Design 6.40. • Inlet hubtipratio: 0.CHAPTER 6. 325 Pa. • Axial channel: Constant Outer Diameter (COD).1 172 Specification of Design Parameters The compressor to be designed must deliver a 5:1 pressure ratio in 5 stages and 8. • Number of stages: 5.15 K. • Massflow: 8. was used to calculate the main operation character . • Inlet stagnation temperature: 288. developed by Tomita [1]. • Rotational speed: 25. 3 kg/s. • Outlet Mach number: 0. the following parameters were also fixed • Inlet stagnation pressure: 101. RESULTS 6.2.2.1 Compressor Preliminary Design For the preliminary design of the axial compressor. 6.3 kg/s massflow rate.50.2. • Isentropic efficiency: 82%.26.2.0. a computational code named AFCC. 650 rpm. Taking into account the constraints generated by the gas generator.
RESULTS 173 istics of the machine based on the meanline characteristics. Clvopt . (6.44. Clvopt is the optimum . α2opt is the optimum relative outlet flow angle. ζ). Loss models were used to account for losses due to the viscous effects and the complexity of the channel flow. s/c. The δ deviation angle is defined by δ = α2 − β2 . The stall incidence was calculated using curves similar to the ones presented in Figure 6. θ. like the ones indicated below: istall = f (α2opt . iopt .CHAPTER 6.44 – Deflexion and profile loss The incidence angle is defined by i = α1 − β1 .1)where. β1 . where α2 is the relative outlet flow angle and β2 is the outlet blade angle. where α1 is the relative inlet flow angle and β1 is the inlet blade angle. FIGURE 6. The results of this calculation is a compressor map with VIGV and BOV. δopt . This initial synthesized compressor map was used in an engine deck to identify possible offdesign operation problems. The stall incidence angle is a function of several parameters.
σ is the solidity. δopt is the optimum deviation angle.2. RESULTS 174 lift coefficient. n is the slope factor. m is a slope factor for minimum loss deviation. cascade experiments have confirmed that there is only one incidence angle at which minimum losses are achieved and this incidence is named unique incidence. θ is the camber. For incidence angle: iml = ki (i0 )10 + nθ + (iD − i2D ). the flow velocity exceeds the limits of validity of the relation given by Equation 6. ki is a correction factor. Minimum loss condition was used for the design.2) where ml stands for minimum loss condition. Acceptable incidence and deviation angles and losses were based on NASA SP36 [130]. Reference [2] presents details of this correction. s/c is the blade spacechord ratio and ζ is the blade stagger angle. the design point incidence angle was set lower than at design. For these highspeed flows. (6. t/c is the thicknesschord ratio. (δ0 )10 is . the minimum loss is attained when the flow is parallel to the blade suction surface. iopt is the optimum incidence angle. For sonic and supersonic flows. (i0 )10 is the zerocamber incidence angle. (6. the exponent b is a function of inlet airangle.CHAPTER 6. that is. β is the inlet and outlet blade angle. For modern compressors. The minimum loss incidence angles were corrected to account for offdesign operation. (iD − i2D ) is a correction factor to account for two dimensional effects. For deviation angle: b δml = kδ (δ0 )10 + mσ θ + (iD − i2D ) dδ di + (δD − δ2D ).3) 2D where kδ is the thickness correction factor for zero camber deviation angle.
α1 . Hence. The stall and choke points can also be determined. RESULTS 175 the zerocamber deviation angle at reference minimum loss incidence angle deduced from low speed cascade data for 10 percent of thickness and (dδ/di)2D is the slope at reference incidence. at the designpoint operation. Mncopt is the optimum Mach number at the blade inlet. continues to be widely used although it is now clearly recognized that its mechanisms are seldom really independent. viscous effects in mixing processes. viscous effects in boundarylayers. endwall loss and leakage loss. Mn0 . Mn0 is the inlet absolute Mach number at the blade. α1 and α2 are the inlet and outlet relative flow angles. in order to eliminate some design difficulties caused by this. Mncopt .CHAPTER 6. The meanline calculation based on stagestacking method allow an accurate investigation of the incidence and deviation angles for different rotational speeds and variable geometry configurations. Mni is the inlet relative Mach number at the blade. ε) (6.4) where Cd is the drag coefficient. The mechanisms of loss sources as profile loss. i. Details can be found in [1]. respectively. The model is a function of several parameters as indicated bellow: loss = f (Cd . The area ratio at annulus of a multistage compressor is based on the compressibility . the front stages are matched closer to choke and the rear stages closer to stall. α2 . The sources of losses are. shockwaves and heat transfer across temperature differences. The tendency at partspeed for the front stages is to move the operational point into stall and the rear stage to move it towards choke. i is the incidence angle and ε is the deflection. in general. Mni .
the front stages are higher loaded and their operating point shifts towards the surge stability limit. and a sound balance between reduction in tolerance and cost reduction is to be created. This increases the massflow in the upstream annulus. The other possibility for ensuring stable flow in the front stages at partload speed is the downstream extraction of bleed. In Figure 6. The number of variable stator vanes in the compressor is an early project decision as any late correction is related to high additional cost due to the complexity of installation. The energy transforms into the bleed massflow is lost for the performance of the compressor. Figure 6. thus reducing the blade row loading. changing the overall matching of the machine. The installation of VIGV in this compressor was considered in the performance cal . To improve the compressor efficiency. In this figure is possible to observe that for partload speed as 65%N the operation range and its efficiency is very low (around 60%).54 it is possible to observe the importance and need of the VIGV and BOV installations for the engine partload speeds. All of these tolerances require careful design consideration. Minimum tolerances have to be introduced or special sealing and bearing bushes have to be used.45 shows the compressor map and its efficiency calculated using meanline technique. These accessories have a direct influence on the gas turbine efficiency.CHAPTER 6. Figure 6. Figure ?? shows the four compressor maps for each bleed schedule presented in Figure 6. The rear stages are unloaded. RESULTS 176 of the fluid and the intended stagewise work distribution. the stages operate at lower pressure ratios. first the BOV was used. Two features are common to compensate the high loading: change the stagger angle of the blade and increasing the massflow in the front by taking bleed.48 shows the improvement in the compressor efficiency at offdesign operations.46. At offdesign operation. Figure 6.46 shows the percentage of air bleed used in this simulation.
Figure 6. Many tests are necessary to decide what is the . sometimes the air bleed is necessary to avoid choke problems. is possible to define that the VIGV installation is superior when compared with the BOV instalation.50 shows three compressor maps each one with different VIGV angles schedule and Figure shows the improvement in compressor efficiency.CHAPTER 6. Figure 6. With this results. Comparing the BOV and VIGV results.45 – 5stage axialflow compressor map FIGURE 6. a study involving VIGV and BOV installations was carried out for this multistage axialflow compressor. RESULTS 177 FIGURE 6.49 shows the VIGV schedule for each rotational speed. However.46 – Compressor bleed schedule culation.
RESULTS 178 FIGURE 6.48 – 5stage axialflow compressor efficiency using BOV VIGV angle and air bleed necessary for each rotational speed.47 – 5stage axialflow compressor map using BOV FIGURE 6.CHAPTER 6. Handling the compressor parameters design the VIGV and BOV variations calculated to this compressor is presented in Figure 6.52 shows the VIGV (in degrees) and bleed (in percentage) scheduled for each rotational speed. .
50 – 5stage axialflow compressor map using VIGV 179 . RESULTS FIGURE 6.49 – Compressor VIGV schedule FIGURE 6.CHAPTER 6.
CHAPTER 6. RESULTS FIGURE 6.52 – VIGV and bleed schedule 180 .51 – 5stage axialflow efficiency using VIGV FIGURE 6.
0% • Combustor chamber pressure loss: 5. the scheme of the gas turbine engine is assembly to prepare the calculations of the engine performance. VIGV and BOV were incorporated to the compressor to use in a singleshaft free power turbine turboshaft.53 shows the engine sketch used for this calculation.0% • Gas generator shaft mechanical efficiency: 99.22 MW • Cycle efficiency: 19.2. The result of this study was the compressor performance at partload and possible adjustments to the compressor preliminary design.0% • Gas generator turbine isentropic efficiency: 85. The engine design point characteristics are: • Compressor pressure ratio: 5.0% . RESULTS 6. The running line was calculated using an engine simulation deck named GTAnalysis.2 181 Compressor Adjustment to the Gas Generator operation The adequacy of the designed compressor for the use in a gas generator was studied using previously designed gas turbine simulator [4].0% • Combustion chamber efficiency: 99.0. • Maximum cycle temperature: 1173.CHAPTER 6. Figure 6. developed by the Gas Turbine Group [4]. Aiming at foreseeing offdesign operation. Calculated the compressor.2.0 K • Shaft output = 1.4% • Cooling bleed air at station 10: 2.
calculated at design point conditions. At partload gas generator speed N1 is lowered from 100% down to the point at which the surge margin vanishes. . caused by bad component matching. mass flow splitter.0% FIGURE 6. The engine is required to operate offdesign due to load variation. massflow mixer. power turbine.0% • Free power turbine shaft mechanical efficiency: 99. so that its speed is fixed at 60Hz or 3600 rpm. from which appropriate data should be selected. and exhaust pipe. RESULTS 182 • Exhaust gas temperature: 859 K • Free power turbine isentropic efficiency: 85. compressor. Performance deteriorates because at offdesign the components operate at regions of lower efficiencies. The power turbine is directly coupled to the generator. The badmatching results from the passage areas. intake. not being adequate to accommodate the flow at those operating conditions. For each chosen offdesign point all thermodynamic parameters were calculated. turbine.CHAPTER 6. combustion chamber.53 – Sketch of a single shaft free power turbine unit The main engine functional blocks used in this analysis were: ambient.
It can be seen that VIGV and compressor bleed keep the isentropic efficiency at highest values at partload. Figure 6.54 shows the compressor map without variable inlet guide vane (IGV) and bleed (black lines) and the compressor map with variable IGV and bleed (magenta lines).55 shows the isentropic efficiency of the two compressors shown FIGURE 6. In this figure are shown the designpoint. . it can be seen that at partload operation the running line. Figure 6.3218 MW). RESULTS 183 Compressor surge margin was set to 20%.54.7202 MW). surgeline and surge margin.CHAPTER 6. It is therefore desirable to use VIGV and bleed for this engine configuration and application. at design and was monitored during the calculations at all partload conditions. When variable VIGV and compressor bleed are used this value can be decreased down to 75% (0. intercepts the surge line at approximately 88% (0.54. In Figure 6. without VIGV and compressor bleed. nearly at the same levels as at designpoint.54 – Compressor characteristics: pressure ratio in Figure 6. according to the suggestions of Walsh and Fletcher [131]. running line. this being the map corresponding to the best partload efficiency and surge margin.
CHAPTER 6. from which more realistic compressor maps can be synthesized early in the design process. mainly because it allows quick access to vital flow properties at the blades edges. a previously developed streamline curvature code was used [2].2. with addition of improved loss models. even with stators variable geometry. well in advance.3 Streamline Curvature Method To improve the meanline design. This program is still being in development at ITA. The use of this compressor code has been granted to other universities. A computer program for performance analysis was developed at ITA and has been used for many years. The result of this calculations are the whole compressor geometry and all flow properties in the compressor. Streamline curvature. Such technique has been successfully used for analysis. with applications re . In conjunction with an engine simulation program. to evaluate the engine performance and take actions aiming at better engine tuning. it is possible.55 – Compressor Characteristics: efficiency 6. from which actions may be taken to improve its performance. is a technique well suited for the design of axialflow compressors. RESULTS 184 FIGURE 6. algorithms for faster convergence and the ability to synthesize a full compressor map.
a design procedure is set up to provide information of axialflow channel geometry. fixed inner. spacechord ratios and maximum thickness ratios. axial channel type (fixed outer. The user can change these design parameters interactively to force the axial channel to have adequate shape so that the attainment to the targeted performance is feasible. Its application starts with the compressor detailed geometry already known. pressure ratio and efficiency. so that the geometrical parameters can be calculated at nodes located at the blade edges. inlet and outlet Mach numbers.CHAPTER 6. axial chord and spacing ratios. starting with the integration of the complete radial equilibrium equation at the blade leading edge based on assumed axial velocity at the hub. from which the total mass flow at the blade inlet can be calculated. tip clearances. on the streamlines. flow angles at the stators trailing edges. The calculations are standard for each row. the exchange of temperature (or pressure) increase among the stages. The exact velocity distribution is obtained by forcing the calculated mass flow to be equal . RESULTS 185 cently reported in the literature that testify its great usefulness both for compressor maps synthesis and engine performance simulation. Starting from the compressor requirements of massflow. axial velocities (or Mach numbers) at the rotors inlet. a distribution tentative of any number of streamlines is made. The integration of the radial equilibrium equation results in a spanwise axial velocity distribution. The design of a new compressor can be based also on streamline curvature techniques. Major required parameters are the inlet hubtip ratio. since the streamlines are automatically and iteratively repositioned during the calculations. maximum blade tip speed. Initial location of the streamlines is based on equal flow in the streamtubes. from hubtotip. but it could be any other criteria. fixed mean or variable mean diameter). Having obtained an acceptable axial channel. can also be set. Variation of these parameters.
converged solution is achieved. Blade after blade the calculations are repeated until the end of the compressor. RESULTS 186 to the required one. Subsonic bladings usually consist of circular arcs. All blades geometry and flow properties at the blades edges are known. Otherwise. calculated based on massaveraged enthalpy distribution. the blockage factor due to the endwall boundary layer is recalculated for each row. After the calculation of specified row is finished and the streamlines repositioned for the trailing edge. one micrometer as used for this work). or combinations thereof. The calculation proceeds to the blade trailing edge using a loss model to predict the pressure and temperature at the outlet nodes. an iterative procedure redoes all the calculations from the first blade row. If the streamlines have been repositioned to new positions that differ from the previous more than a suggested fixed limit value (for instance. The nature and type of blading employed in compressors depends on the application and Mach number range. parabolic arcs. based on both their experience and suggestions available at the program. varying the assumed hub axial velocity. the axial velocity guess is at the mean streamline. new radial positions for the streamlines are calculated. except that. The user may review their assumptions in case of mismatches and repeat the whole design procedure . The blade surfaces are designed to provide smooth entry and exit flow within minimum loss and maximum pressure rise. At this point. the losses at the bladeless space is calculated and the conditions at the next blade leading edge are obtained.CHAPTER 6. replacing the old values. The efficiency. At this point. for the sake of better convergence. is checked against the target value. The calculation then proceeds to the trailing edge similarly to what is done at inlet. based on the specified massflow for each streamtube and by inverse interpolation of massflow as a function of the radii. The user can choose among several loss models.
for each rotational speed). except that the compressor geometry is now available. rowbyrow the flow properties are calculated and the streamlines repositioned. The mass. from which all flow properties are calculated at the blade edges. Effects of viscosity are incorporated by empirical correlations and the results are enough accurate to define compressor geometry.5) . Dt ∂r r ∂ω ∂z (6. Eventually a converged solution is achieved. similarly to the design procedure. In the sequence. For each given mass flow at the compressor inlet and rotational speed. This methodology is made for the design and analysis of a 5stage axial compressor. The starting points for this design refinement were obtained from meanline calculations. at the blade edges. The inviscid momentum equation can be written as ~ ~ ~ ∂W Wω ∂ W ∂W DV~ = Vr + + Vz . The results obtained are presented in this work.CHAPTER 6. RESULTS 187 until acceptable results are obtained. the analysis is performed by the offdesign analysis. The offdesign performance calculation is carried out for specific values of massflow and compressor rotational speed. The user may set ranges for the massflow based on the designed massflow and rotational speeds. so that the compressor map can be assembled (based on pressure ratio and efficiency versus corrected massflow. momentum and energy equations for axisymmetric nonviscous flow are solved on streamlines. The offdesign module is based on the same principle used for the design. The data for the compressor map construction is calculated based on massaveraged properties. This is carried out for each massflow and rotational speed.
CHAPTER 6. Details of the streamline curvature method can be found in [2].6) ~ = Wr~er + Ww~ew + Wz~ez . The boundarylayer. energized and deenergized. Further development of the previous equation gives the following equation. Rc is the radius of curvature and S is the distance along blade edges. 134].7) ~ = ωr~eω . ∂m 1 − Mm ∂s Rc R ∂m r (6. The literature reports that the application of conventional boundarylayer theory to estimate the displacement and momentum thickness is generally inadequate since inside an axial compressor the flow is constantly: changing direction. which is not included in the previous equation. + + Vz +ω ~ ×ω ~ × ~r + 2~ω × W = Vr − ∇P ρ ∂r r ∂ω ∂z (6. skewed.9) Therefore.10) where m stands for meridional. U (6. (6.8) ~ ~ ~ Wω ∂ W ∂W ∂W 1~ ~ + F~ . The model presented by Jansen [132] uses . W (6. which is used for the calculation of the meridional velocity: Vm ∂Vm Vm2 ∂ε tg(ε + Γ) 1 ∂S 2 1 = {−sec(ε + Γ) + + − (1 + Mm ) sin(ε)}. is estimated using as reference model empirical correlations from [132. Γ is the blade sweep. that means blockage factor. ε is the streamline slope. as it progresses through the rows. RESULTS 188 and V~ = Vr~er + Vw~ew + Vz~ez . 133.
14 to the hub and to the casing ∗ (tip) walls.11. H (6.12. . dz (6. if it is assumed that the endwall boundarylayer can be treated as 2 − D and turbulent.4. that are the boundarylayer displacement thickness at hub and at casing. it is possible to obtain δH (z) and δT∗ (z). and initially can be assumed equal to 1.2 #0.11) If the boundarylayer displacement thickness is known. it can be shown that "R z θ(z) = θ(z0 ) + kbl V 4 (z)dz z0 m Vm3.8 . 6. Hence.12) where H is the shape factor.14) Calculating the equations 6. The blockage factor is defined as the decreasing in the compressor annulus height equivalent to the boundarylayer displacement thickness. in which Vm (z) is the local meridional velocity. and can be evaluated from H(z) = 1.13 and 6. 6. (6. then θ(z0 ) can be calculated by θ(z0 ) = δ∗ .CHAPTER 6. RESULTS 189 the meridional velocity near the wall region to account for the effect of the boundarylayer in restricting the flow through the compressor.13) and the displacement thickness from δ ∗ (z) = θ(z)H(z). and that the meridional velocity is the dominant characteristic. (6.5 + 30 dθ .
2. (6. At this point.1.16) kBT = RT2 − (RT − δT∗ )2 . . it is possible to define the hub and casing (tip) blockage factor due to boundarylayer as kBH = 2 ∗ 2 ) − RH (RH − δH .7.15) where RT and RH are the casing (tip) and hub radii. but a good assessment depends on the knowledge of the mechanisms through which these losses act. • for decelerating flows: H(z) < 1. 2 RT2 − RH (6. The comprehension of the interaction of all losses involved inside of the turbomachinery is very complex. RESULTS 190 respectively. Hence.18) This approach is used with the following limitations: • for accelerating flows: H(z) < 2. kBH ≤ 0. along the compressor annulus. the blockage factor is defined as the ratio of the freestream area and metal area. Then kB = ∗ 2 (RT − δT∗ )2 + (RH − δH ) .CHAPTER 6. the process of calculating the boundarylayer blockage is iterative. There are different types of losses in a compressor. only after the meridional velocity is known throughout the compressor. 2 RT2 − RH (6. 2 2 RT − RH (6.17) then kB = 1 − (kBH + kBT ).2. Knowing the kB value.25 and kBT ≤ 0. Hence. • kB ≤ 0.
it is necessary to take account of the additional losses associated with low Reynolds numbers. the losses are considered independently generated. Profile loss is usually considered to be loss generator due to the blade boundarylayers well away from the endwalls. There are several averages of tests results or their statistical curve fits as presented in NASA SP36 [130]. The detailed loss mechanisms clearly depend on whether the blades are shrouded or unshrouded. but it is due to a combination of many factors. The adjustments of each case is carried out using the designer expertise. Hence. The extra loss arising at blade trailingedge is included as profile loss. the interaction between leakage and end wall losses may be strong. The problem is that there are no expected general formulation to represent each individual compressor. Hence. The losses are accounted for using loss models. the empirical correlation cannot give good results every time that they are applied because a general model for all compressors is unrealistic. Tip leakage loss arises from the leakage of flow over the rotor tips (and sometimes of stator hub depending on the design configuration). the loss does not arise directly from the secondary flow. However. Endwall loss or secondary loss is from the secondary flows generated when the compressor annulus boundarylayers pass through a blade row. Most of the results from test facilities are published and available to the public but key information is retained as proprietary. It is possible to assume that the flow is 2 − D so. For high Reynolds numbers. RESULTS 191 in a simplified approach.CHAPTER 6. the loss may be based on 2 − D cascade tests or in boundarylayer calculations. which take account of the following . For unshrouded compressor blades. Friction and secondary losses also contribute to profile losses. some researchers have incorporated their expertise into correlations of parameters which describe the flow along the compressor annulus. some methods do not distinguish these losses separately. Fortunately.
Swan [135]. the loading factor is greater than in . Generally. using the models of references [130. • Reynolds number effect: takes account of low Reynolds numbers effect. • Friction losses: based on duct loss. secondary loss according to [139] and the Reynolds number effect according to [141].4 present blade geometrical characteristics. 140].3 and 6. from hubtotip. 6. the last stator (S5) was designed with low outlet blade angle as presented in Table 6. Hence. The user may arbitrarily combine these models and also used their own models.2. using five streamlines distributed along the blade spanwise.4. In this work.CHAPTER 6. 141]. Tables 6. Monsarrat [136]. a high performance compressor has high deflection angles in the first stages to transfer a high energy from the rotor to the air. • Shock losses: based on the NACA model of reference [138]. To deliver the flow with low swirl angle at the compressor outlet or combustion chamber inlet. 6. RESULTS 192 losses: • Profile loss: based on the models of NASA SP36 [130]. Jansen and Moffat [26]. Davis and Millar [137] and one modification of the last model mentioned is implemented [2]. calculated considering the incidence and deviation angles and accounting the flow losses using the models aforementioned.1. With the inlet and outlet blade angles it is possible to determine the blade stagger and camber angles. calculated by the streamline curvature program. • Secondary losses: based on the models developed by Griepentrog and Howell [139. profile loss was evaluated according to reference [136].
1 75.8 43.4 – Distribution of stator blade outlet angle along the Fraction of radial span (%) S1 S2 S3 S4 0 13.6 11.9 72.7 73.8 75 13.3 53.9 blade outlet angle along the R1 R2 R3 R4 13.7 50 75 73.8 63.1 56.5 rotor height R5 53.3 25 68.1 – Distribution of rotor blade Fraction of radial span (%) R1 0 59.6 11.6 11.9 76.3 10.6 50 13.4 61.0 66.7 stator height S5 1.7 57.0 71.7 25 50 43.8 56.3 – Distribution of stator blade inlet angle along the Fraction of radial span (%) S1 S2 S3 S4 0 55.8 54.7 54.6 73.7 11.7 64.7 1.7 75.4 74.5 61.1 10.6 58.1 59.9 71. The outlet flow angle can be obtained with the difference between the deviation angle and blade outlet angle.3 40.0 54. Table 6.7 62.7 10.1 70.0 52.8 48.4 1.5 11.7.9 25 13.9 10.CHAPTER 6.8 46.0 56.3 73.5 62.6 presents the number of blades for each row.8 10.2 11.3 77.4 75 100 39.1 76.5 50.6 60.8 75.3 58.2 – Distribution of rotor Fraction of radial span (%) 0 25 50 75 100 .9 55.7 42.2 TABLE 6.0 55.4 rotor height R5 73.6 55.9 57.4 29.4 58.0 53.8 53.7 54.7 10.8 TABLE 6. TABLE 6.8 70.5 50.2 1.5 11.7 77.6 stator height S5 58.4 55.7 100 13.3 63.6 68.0 74.3 TABLE 6.0 inlet angle along the R2 R3 R4 59. RESULTS 193 other stages as presented in Table 6.3 1.5 10.6 10.8 10.6 55.7 59.5 54.2 59.3 62.9 66.5 100 75.
315 0.353 0.546 0. Note that the rotor five (R5) needs some improvement near the hub region due to the high incidence angle.732 0.791 0.7 – Rotor loading factor Fraction of radial span (%) R1 R2 0 0.62 100 flow angle along the rotor height R2 R3 R4 R5 6.315 100 distribution R3 R4 0.597 0.910 0.282 0.967 0.63 6.64 6.779 0.287 S5 89 R5 0.632 0.828 0.83 6.687 0.336 0.63 6.62 6. RESULTS TABLE span (%) 0 25 50 75 100 194 6.300 0.987 0.9).728 0.359 0.681 0.327 0.830 0.66 50 6.262 0.670 0.363 0.5 – Distribution of the blade spacechord ratio along the blade height R1 S1 R2 S2 R3 S3 R4 S4 R5 S5 0.398 0.835 0.64 6.8.632 0.946 0.66 6.11 present the blade incidence and deviation angles from hubtotip calculated by streamline curvature program using 5 streamlines distributed along the blade spanwise.749 0.596 0.64 6. but no choke problems was detected during the compressor design calculation.757 0.305 0. High incidence angle causes stall problems.822 0.618 0.278 0.675 25 0.725 0.433 0.8 shows the incidence angle at the rotor leading and trailing edges.699 0.711 0.299 0.86 7.65 6.571 0.612 0.784 0.63 6.356 50 75 0.564 0.6 – Number of blades for each row row R1 S1 R2 S2 R3 S3 R4 S4 R5 blades 26 31 38 43 44 51 64 71 84 TABLE 6.536 0. TABLE 6.63 .493 0.662 0.71 6.755 0.696 0.63 6.9.589 0. Note that the incidence angle at the stator five (S5) is small. 6.563 0.643 TABLE 6.814 0.323 0.CHAPTER 6.553 0.692 0.88 25 6. the incidence angle has lower values compared with the rotors (Table 6. Table 6.537 0.648 0.65 6.79 6.65 6.422 0.10 and 6.69 6.634 0. 6.662 0.243 0.8 – Distribution of rotor incidence Fraction of radial span (%) R1 0 6.454 0.598 0.903 0. For the stators.249 Tables 6.630 0.64 75 6.742 0.10 6.62 6.529 0.775 0.701 0.62 6.
22 7.558 0.16 3.CHAPTER 6.539 0.27 2.28 2.702 R5 0.01 3.23 8.761 0.72 The de Haller number.571 0.87 1.08 75 2.72 at designpoint.97 3.36 8.709 0.01 50 6.686 0.10 – Distribution of rotor deviation flow Fraction of radial span (%) R1 R2 0 7.55 25 50 2.21 1.558 0. Hence.72 needs some attention to avoid problems with high boundarylayer growth.52 6.56 2. The relative or absolute velocity has greater values near tip region.34 1.96 2.95 1.44 2.41 9.56 4.71 2.15 angle R3 6.96 along the rotor height R4 R5 5.73 4.80 9.30 2.654 0.36 100 1. Values lower than 0. The average value of the de Haller number is 0.24 1.66 8.22 8.79 2.08 50 75 2. generally is greater on this region.19 75 6.44 along the stator height S4 S5 8.21 3.62 100 flow angle along the stator height S2 S3 S4 S5 3.46 1.23 1.647 0.60 5.70 8.11 4. RESULTS 195 TABLE 6.38 3.642 0.698 0.12 and 6.65 9.31 9.727 0.03 8.637 0.48 8.734 0.20 2.38 9.692 0.78 25 3.13 TABLE 6.684 0.9 – Distribution of stator incidence Fraction of radial span (%) S1 0 3.11 – Distribution of stator deviation flow Fraction of radial span (%) S1 S2 0 6.05 4.10 6.21 3.537 0.27 100 angle S3 8.23 8.56 8.24 5.672 0.61 3.666 0.739 .21 9.12 – Rotor Fraction of radial span (%) 0 25 50 75 100 de Haller number distribution R1 R2 R3 R4 0.88 3. the de Haller number.639 0.07 TABLE 6.74 9.91 5. The de Haller number along the rotor and stator blade spanwise are presented in Tables 6. is a parameter that quantifies the deceleration of the flow along the bladetoblade channel [1].13 1.12 3. defined as the ratio of the relative outlet and inlet flow velocities.18 2.77 4.03 1.613 0.34 9.08 3.83 3.37 25 6.19 8.18 TABLE 6.712 0.691 0.
251 3D Flow Calculation With the results from the streamline curvature program.384 0.319 0.639 High Mach number means high loading factor.632 0.57 show a 3 − D compressor picture in two different views.609 0. a solid model of the compressor channel defined by the blades was generated using a CAD software. An interpreter for the ICEMCFD unstructured hexahedral elements output files was developed to integrate the calculated mesh to the CFD solver.861 0.659 0.13 – Stator Fraction of radial span (%) 0 25 50 75 100 196 de Haller number distribution S1 S2 S3 S4 0.569 0.695 0.25. Table 6.630 0. the inlet and outlet Mach number used as input parameter to design the compressor are 0.CHAPTER 6.646 S5 0.656 0.367 0.7 shows that the first rotor blade (R1) has a greater loading factor and the last rotor (R5) has a lower loading factor due to the energy transfer for each blade row and the design criteria given by the inlet absolute flow angle for each blade row.624 0.570 0.14 – Axial Mach number for each blade row R1 S1 R2 S2 R3 S3 R4 S4 Mach number 0.679 0.636 0. Hence.605 0. .56 and 6. Figures 6. The meshes were defined based on such geometry using the version 11 of the ICEMCFD.334 0.557 0.935 0.282 6.630 0.2. High Mach number at the compressor outlet can causes some difficulties to decelerate the flow at the combustion chamber inlet. High loading factor at the compressor inlet can be causes serious structural problems for the blade. in this work.558 0.14.5 and 0. respectively.737 0.382 0. TABLE 6.494 0.294 S5 0.897 0.619 0.919 0.610 0.644 0. The distribution of the Mach number calculated through the compressor blade rows is presented in Table 6.439 0. RESULTS TABLE 6.4 R5 0.
Each blade row (rotor and stator) requires independent blocks because their number of blades are not equal. 621 nodes. Regions with high pressure gradients require special attention where finer meshing and smoothing to capture the physical aspects of the flow within the boundarylayer are necessary. The mesh used has 706. A bladetoblade framework was used with the periodic condition as describe in the boundary conditions section (3.59.60 shows a general view of the mesh generated for the multistage axialflow compressor. RESULTS 197 FIGURE 6. The geometric complexities requires the use of several mesh blocks. An hexahedral mesh was used to discretize the computational domain. The first node from wall surface has the distance of 2 · 10−5 m calculated based on the flatplate theory. Figure 6.2).2.7.CHAPTER 6.4 and 6. to guarantee the y + required by the .56 – Designed 5stage axialflow compressor The computational domain used to generate the mesh is presented in Figures 6.
59 – Axial view of the 5stage axialflow compressor domain .57 – 3D view of the 5stage axialflow compressor blade profiles FIGURE 6. RESULTS 198 FIGURE 6.58 – Computational domain of the 5stage axialflow compressor domain FIGURE 6.CHAPTER 6.
The wall regions are more refined and an exponential function was used to the nodes distributions.61 shows the mesh on flow compressor mesh the blade surfaces.6.CHAPTER 6. FIGURE 6. RESULTS 199 turbulence model as presented in Equation 4.61 – Mesh on the blade surfaces .60 – General view of the Ogrid generated to the 5stage axialflow compressor Figure 6. FIGURE 6.
CHAPTER 6. All other stages follow the same mesh fashion.62 – Computational domain and mesh structure on the hub of the firststage Figure 6.62 shows the mesh structure of the firststage of the compressor. FIGURE 6. RESULTS 200 Figure 6.63 shows a detail of the Ogrid generated close to the stator trailing edge and rotor leading edge. In this figure it is possible to observe the rotor curved surface. For an hexahe .63 – Detail of the Ogrid mesh type around the blades The elements and their neighbors are addressed via connectivity tables. FIGURE 6.
To achieve an acceptable solution. as the one used in the study of the 5stage axial compressor. what was done in this research. especially for the case of finer meshes. An accurate start may be based on the streamline program results. since it is considered periodic. an accurate initialization reduces machine time. possibly during the pseudotransient part of the calculation. stagnation temperature and flow angles) and at the outlet (static pressure). the calculation may become unstable due to numerical oscillations due to the strong unsteady flow behavior. For this case. The properties were specified at the inlet (stagnation pressure.CHAPTER 6. The problem is that for high pressure compressor. including the ghost elements. the calculation may fail as a result of the transient behavior by initial condition due to the unsteady operation point. The resulting blockage reduces the massflow and increases the flow incidence at the blade which makes the separation or the loss larger. Care has also been taken to guarantee a wellposed problem and hence numerical stability. These ghost elements are used to set the boundary conditions. Avoidance of reflection was assured by the use of nonreflecting inlet and outlet boundary conditions. it is essential to initialize the variables accurately. Besides speedingup convergence. the values of the properties were taken from the streamline curvature program calculations. The flow was calculated only in one channel. when the blade is operating near stall condition. The results of computational codes for compressors and turbines. Hence. developed at Center for Reference on Gas Turbine at ITA were used to produce the appropriate initialization. . the calculated flow separates or generates a high loss. Thus. RESULTS 201 dral element there are six neighbors.
64 – 5stage axialflow compressor outlet massflow convergence history 82 81 efficiency 80 79 78 efficiency 77 0 2000 4000 6000 8000 10000 iteration 12000 14000 16000 18000 20000 FIGURE 6.65. 10 Mass−Flow (kg/s) 9 8 7 6 Outlet Mass−Flow 5 0 2000 4000 6000 8000 10000 iteration 12000 14000 16000 18000 20000 FIGURE 6. respectively.65 – 5stage axialflow compressor efficiency variation during the numerical iteration . RESULTS 202 The convergence of the compressor outlet massflow and the compressor efficiency are presented in Figures 6.CHAPTER 6.64 and 6.
in the third stage of the axial compressor is presented in Figure 6.CHAPTER 6. Figure 6.69 shows the velocity vectors across the compressor thirdstage. RESULTS 203 Figure 6.66 – Static pressure contours in the 5stage axialflow compressor A cut of a velocity vectors distribution at the compressor outlet can be seen in the Figure 6. FIGURE 6.67 – Detail of the velocity vectors at the compressor outlet The mixingplane velocity distribution at the rotor outlet to the stator inlet.67. . The velocity vectors at rotor outlet are relative velocities. This figure shows the plane when the frame of reference was changed from a rotating frame of reference to nonrotating frame of reference.68.66 shows the static pressure increase along the compressor stages. Periodicity between suction and pressure sides can be observed too. FIGURE 6.
RESULTS 204 FIGURE 6.69 – Detail of the mixingplane velocity distribution at the compressor thirdstage .CHAPTER 6.68 – Detail of the mixingplane velocity distribution FIGURE 6.
The total pressure in the rotors region is represented by the relative total pressure. There are no general loss models to apply in the streamline curvature methodology. FIGURE 6. The calculation of these variables using CFD is proprietary information on the turbomachinery community.9 Pressure ratio 5.3 7.0 4. TABLE 6.CHAPTER 6.0 79.70 – Total pressure distribution along the 5stage axialflow compressor General results for the SLC and CFD calculations are presented in Table 6. Maybe the average values at compressor outlet are not a better option to calculate these variables.15. Some differences can be observed between streamline curvature and CFD results. The calculation of the efficiency and output power by CFD is an issue under study. CFD results are arithmetic means of the properties at each face at the compressor outlet computational plane.70. Each compressor design needs different calibration of the loss models used in the streamline curvature program.9 82.15 – Comparison between streamline curvature and CFD results Technique Streamline curvature program CFD Massflow 8.4 Efficiency (%) The efficiency was calculated by the ratio of the ideal total temperature increase across the compressor and the actual total temperature increase. Inter . Each compressor has its particular calibration and flow behavior. RESULTS 205 The total pressure distribution is presented in the Figure 6.
RESULTS 206 action between the streamline curvature program and the CFD calculations are being carried out aiming at the flow analysis at spotted parts where undesirable large gradients are present to favor flow separation. what is not aim of this work. . Interaction between the streamline curvature and the CFD results have been carried out aiming at the flow analysis at spotted parts where undesirable large gradients are present to favor flow separation.CHAPTER 6. what it is not the aim of this work.
only the minimal routines would be implemented.7 Comments and Conclusions It is important to emphasize that this thesis deals with the design and analysis of axialflow turbomachines. it was worth because a better analysis was already possible. Nevertheless. just one turbulence model and so on. The most important application was the study of the 5 to 1 pressure ratio axial compressor that will be manufactured. to be used for new students. As such. That compressor study is not validated due to the lack of the experimental data. using interactively the streamline curvature and the CFD programs. Such a platform would serve as a basis for future and continuous development. Care has been given to axial compressor analysis due to the opportunity to contribute to the compressor design of the gas turbine under development at CTA. namely just one discretization mode. besides that. the CFD procedure is the most complex part of the work. opportunities have been arisen to implement other procedures. namely the TAPP project. which will be . The constraints for this work were the need to develop a CFD software platform that would calculate the viscous flows in axial compressors. as indicated in the text. However. Some cases were used for the sake of verification and validation. sometimes at high cost in terms of time spent in the research.
to say that the utilized combination of streamline calculations and CFD verifications may lead to good compressor design. Significant part of the time was spent with grid generation and gridindependence studies. It is well know that the generation of appropriate grid for turbomachines is a timeconsuming job. . The turbine studies were less troublesome because of the nature of the flow with favorable pressure gradients. so that initialization of the compressor calculation was of most importance because of the nature of the flow in the blade passages. even though high pressure ratio stages are sought. The initialization of the parameters was simpler than the initialization for the compressor. for both external and internal flows. For example. COMMENTS AND CONCLUSIONS 208 available by the end of 2009. the final grid for the study of the 5stage axial compressor took around one month of work. Initial guess for major parameters. For this work the interest has been fixed on steadystate. The mixingplane technique was also used. pressures and temperatures. was made based on the streamline curvature calculations. The design process took several months to be accomplished. then. being possible to achieve convergence even with very rough initial guess. It is possible. The CFD solver that was developed can be used with rotating frames and nonrotating frames equally.CHAPTER 7. like velocities.
a parallel version of the CFD program is running. At the time of the writing of this work.CHAPTER 7. This extends the fields of research to the vast area of parallel processing as well. Therefore. Since the platform was implemented having in mind the modular concept. implementation of many different procedures can be made without much difficulty. a very fertile field is available for future research on methods of discretization. In this case. the CFD is used to check whether the streamline curvature design can deliver or not the specified performance. adequacy of turbulence models for flow calculation. as part of a postdoctoral research work carried out at the Center for Reference on Gas Turbine. . COMMENTS AND CONCLUSIONS 7. Although the computational capability existing now is not enough for design using CFD exclusively. the interaction with a streamline curvature program can integrate the use of CFD during the design part of the project. as well as better combination of technological parameters aiming at a turbomachine to be optimized for a specified application.1 209 The CFD solver as a Research and Teaching Tool The CFD platform developed for this work is adequate for future research and teaching activities.
The CFD platform is not userfriendly. Not only for the solver. As commented in the last chapters. but to improve the pre. the author concluded that some subjects mentioned in this chapter are strongly recommended to improve the numerical simulation of the computational code. It can be associated to an user interface that makes its use more friendly.8 Future Implementations The objective of this chapter is to describe some important future works to improve the present CFD solver robustness. many others details before discussed become this work very extensive. Furthermore. The numerical implementation process in a computational code capable to calculate the full NavierStokes equations in a 3 − D computational domain and to become it possible to simulate the rotor and stator blade rows simultaneously required a long time of programming. since it has been developed just for the author use. The developement of CFD solver was only possible because the Gas Turbine Group at ITA has turbomachinery design tools developed as meanline and streamline curvature programs.and postprocessing. turbomachinery flow simulations is very complex and needs some special treatments. during the development of this research. Hence. . The process to obtain the final turbomachinery geometry needs a long time of design and performance analysis.
Some suggestions are: 1. both by reducing the time to prepare the initialization data and reducing the number of iterations. turbulence models. The idea of this section is to discuss some important features in the sense of coupling the streamline curvature program with CFD solver. . Below follow suggestions that would facilitate the platform development. that widen the use to other applications than turbomachinery. Experience has proved that the data from the streamline curvature program is a good point of departure. Blade geometry visualization makes the work very ”hard”. as MSc or PhD research programs.1 PreProcessing The preprocessing in a turbomachine CFD simulation has many details due to the complexity of a compressor or a turbine. In the same direction. 8. It is important to develop an initialization tool using the calculations from the streamline curvature program.CHAPTER 8. mainly for high performance multistage axial flow compressors where the leading and trailing edges radii are too small. Several possible continuation work have already been mentioned in the past chapters. FUTURE IMPLEMENTATIONS 211 The modular environment used for the program development makes it simple to add extra modules incorporating other discretization methods. it is interesting to link the calculations from the streamline curvature to the CFD platform so that the boundary conditions could be set directly. in a way that could be transparent to the CFD user. It may be possible to make this initialization process transparent for the user. what would reduce significantly the time required for convergence.
new design and treatment of casing. In this work the hexahedral grid was implemented. leaving for future work other grid types to be implemented. turbomachinery loss mechanisms. unsteady effects among others. passive endwall treatments for enhancing stability. It is of utmost interest to make automatic the process of generating the compressor blade passages in a CAD environment directly from the streamline curvature program. a gridgenerator software could be developed such that the problem of making available an adequate grid could be almost transparent for the researcher. it is the author’s experience that grid generation must be one of the first works to be developed. new blade geometries. The author’s experience points that the process of generating a CAD solid and the corresponding grid is too expensive in terms of time consumed until an appropriate grid is obtained. since this program is the most appropriate for geometry calculation. FUTURE IMPLEMENTATIONS 212 2. momentum and energy equations. the extraction of the geometry data for the grid generation routine. inlet flow with distortions. Although grid generation may not be interpreted as a CFD work. inlet and outlet ducts. the firststep towards become the CFD solver faster and possible to run in various processors simultaneously was started . 8. 3. tipclearance and its influence on the machine performance. rotating stall and surge phenomenon. About the numerical improvement.2 Solver The CFD solver developed based on academic interest to become possible the study of new numerical methods applied to mass. wakes.CHAPTER 8. In the sequence. to create new techniques to improve the tip leakage. Having in mind the very specific shape of the blade passages.
Develop a parallel version of the computational platform. 7. Methods for convergence acceleration can be also implemented. 8. The use of an implicit version would improve the convergence as far as stability and CPU time are concerned. . combustion chamber and turbine) of gas turbine simultaneously. 9. 2. The first version of the CFD solver code with parallel processing was verified and publicated [142] in a ASME TurboExpo Conference. There are other flows that can be calculated using the developed software and. The current platform cannot do the calculations of reacting flows. therefore. 4. 3. Therefore. the inclusion of the equations concerning the combustion processes will make it useful for calculations in combustion chambers as well. Implement other mixingplane formulations including formulation based on unsteady flow. For the transient performance study it is required the use of appropriate techniques. Other suggestions from the author are: 1. appropriate turbulence models could be implemented accordingly. The current time integration scheme is explicit. one of which is the interpolation between grid points. In this work grid refinement is not automatic. 6. 5. Create a powerful framework to simulate the flowfield in all components (compressor. FUTURE IMPLEMENTATIONS 213 with the work involving parallel processing. Implement a convergence acceleration techniques. The implementation of higherorder spatial and time discretization methods will make the Platform best suited for transient analysis.CHAPTER 8. aiming better computational performance. so grid refinement using all the available criteria can be the focus of further research.
slices and flow path visualization in regions as tip clearance and endwalls.3 214 PostProcessing It is very adequate the development of postprocessing software to make it possible the visualization of the important phenomena occurring in turbomachines. Maybe. .CHAPTER 8. FUTURE IMPLEMENTATIONS 8. should be optimized based on its computational domain to become easy the handling of numerical results in all blade row sections. in the case of turbomachinery geometry. The postprocessing. from the streamlines to the sources of losses. this tool should be developed with the preprocessing to use the same background of the mesh visualization.
35 10 MAKINO. 192 3 JESUS. 195 2 BARBOSA.ITA.. 1996. 33. S˜ao Jos´e dos CamposSPBrazil. Haynes Publishing. 35. MarchApril 2003. dos S. 2003. Projeto e An´ alise de Desempenho de Turbinas Axiais de V´ arios Est´ agios com Geometria Vari´ avel. de. Numerical Simulation of Axial Flow Compressors. 2003. T. A. 2003. Thesis (Master of ScienceMSc) . 170. N.. 170. J.. Thesis (Doctor of PhilosophyPhD) . MORGAN.Instituto Tecnol´ogico de Aeron´autica . T. C. S˜ao Jos´e dos CamposSPBrazil. B. 2006.ITA. 2004. 34 7 REED.. 184. Journal of Aircraft. University of Wales .ITA. 2006. 35. 63. Z. 188. P. LEI. 35 11 WEATHERILL. Fluid Dynamics and Heat Transfer of Turbomachinery. 33.Cranfield Institute of Technology. 1999. 2. AFJCH. Computational Simulation of Gas Turbines . D. 1999. 34. 2003. 44.ITA.Instituto Tecnol´ogico de Aeron´autica . Computational Simulation of Gas Turbines . J. 151 p. K. 1987. Simula¸c˜ ao de Desempenho de Turbinas a G´ as em Regime Transit´ orio.Instituto Tecnol´ogico de Aeron´autica . J. 85. 174. Fuselage Shape Optimization of a WingBody Configuration with Nacelles. 40. J. 43. 34 6 REED. R. The Science of Formula 1 Design. 172. 2003. IWAMIYA. Thesis (Master of ScienceMSc) . Y. 181 5 SILVA. A Streamline Curvature Computer Programme for Performance Prediction of Axial Flow Compressors. AFJCH. S˜ao Jos´e dos CamposSPBrazil.Part I: Foundations of ComponentBased Models: INTERNATIONAL GAS TURBINE & AEROENGINE CONGRESS & EXHIBITION. John Wiley & Sons. Thesis (Master of ScienceMSc) . Thesis (Doctor of ScienceDSc) . 94GT346.Department of Civil Engineering. S. 228 9 TREMAYNE. A. 33 4 BRINGHENTI. 154 p. 36 . 94GT347. G. 2003. J. 111 p. 175. F. S˜ao Jos´e dos CamposSPBrazil. A Course Under the PROPESA 1995/96 Program: FiniteElement Methods in Computational Fluid Dynamics.Instituto Tecnol´ogico de Aeron´autica .Part II: Extensible Domain Framework: INTERNATIONAL GAS TURBINE & AEROENGINE CONGRESS & EXHIBITION. 1987.. n. Variable Geometry Gas Turbine Performance Analysis. 2003. v. 34 8 LAKSHMINARAYANA.Bibliography 1 TOMITA. 1995.
BIBLIOGRAPHY
216
12 SCHLUTER, J.; APTE, S.; KATIZIN, G.; HEINZ, P.; WEIDE, v. d.; ALONSO, J.
Unsteady CFD Simulation of an Entire Gas Turbine HighSpool. ASME Turbo Expo
2006. Proceedings of ASME. GT200690090. 36
13 GOPINATH, A. K.; WEIDE, V. D.; KALITZIN, G.; ALONSO, J. J.; JAMESON,
J. J.; EKICI, K.; HALL, K. C. ThreeDimensional Unsteady MultiStage
Turbomachinery Simulations Using the Harmonic Balance Technique AIAA report, Stanford University, Duke University. 36
14 TOMITA, J.; BRINGHENTI, C.; BARBOSA, J.; JESUS, A. Nacelle Design for
Mixed Turbofan Engines. ASME Turbo Expo 2006. ASME, 2006. GT200691212. 36
15 WING, D. Afterbody/Nozzle Pressure Distributions of a TwinTail
TwinEngine Fighter with Axisymmetric Nozzles at Mach Numbers From
0.6 to 1.2, NASA, 1995. NASATP3509. 36
16 COMPTON, W. B. Comparison of Turbulence Models for NozzleAfterbody
Flows with Propulsive Jets, 1996. Washington, DC: NASA. 1996. NASA TP3592.
36
17 WILHELM, R. Inverse Design Method for Design Isolated and WingMounted
Engine Nacelles. Journal of Aircraft, 2002. v. 39, n. 6, Nov.Dec. 2002. 36
18 FUJINO, M.; KAWAMURA, Y. WaveDrag Characteristics of an OvertheWing
Nacelle BusinessJet Configuration. Journal of Aircraft, 2003. v. 40, n. 6, Nov.Dec.
2003. 36
19 LAPWORTH, L.; SHAHPAR, S. Design of Gas Turbine Engines Using CFD.
EUROPEAN CONGRESS ON COMPUTATIONAL METHODS IN APPLIED
SCIENCES AND ENGINEERINGECOMAS 2004, 2004. 37
20 KAU, H. The Multidisciplinary Design Process. In: RTO AVT LECTURE SERIES.
”Integrated Multidisciplinary Design of High Pressure Multistage
Compression Systems”, 1998. p. 1–16.(RTO Lecture series, 211) 38
21 XU, J.; GE, M.; ZHAO, X.; WANG, Z. ThreeDimensional Flow Theory of
Turbomachinery, Part 1: Basic Methodology. Propulsion and Power, v. 14, n. 6, p.
899–906, nov 1998. 40
22 XU, J.; GE, M.; ZHAO, X.; WANG, Z. ThreeDimensional Flow Theory of
Turbomachinery, Part 1: Design and Analysis System. Propulsion and Power, v. 14,
n. 6, p. 907–915, nov 1998. 40
23 KOCH, C.; SMITH, L. Loss Sources and Magnetudes in AxialFlow Compressors.
Journal of Engineering for PowerPaper N◦ 75WA/GT6, p. 1–14, jul. 1976. 40,
165
24 DENTON, J. Loss Mechanisms in Turbomachines. Journal of Turbomachinery,
v. 115, p. 621–656, out. 1993. 40
25 CASEY, M. Computational Methods for Preliminary Design and Geometry
Definition in Turbomachinery. AGARD Lecture Series on Turbomachinery
Design Using CFD. AGARD, 1994. p. 1–22. 41, 63
BIBLIOGRAPHY
217
26 JANSEN, W.; MOFFAT, W. C. The OffDesign Analysis of AxialFlow compressors.
Journal of Engineering for Power, 1966. Paper n◦ 66WA/GT1. 41, 192
27 RIESS, W.; WALBAUM, M. Initiation and Propagation of Flow Instabilities
in MultiStage Axial Compressors: Loss Mechanisms and Unsteady Flows in
Turbomachines, 1995. (AGARD/PEP  812 may 1995). 41
28 DAWES, W. Unsteady Flow and Loss Production in Centrifugal and Axial
Compressor Stages: Loss Mechanisms and Unsteady Flows in Turbomachines, 1995.
(AGARD/PEP  812 may 1995). 41
29 DENTON, J. The Calculation of the Three Dimensional Viscous Flow Through
Multistage Turbomachines. ASME Paper. ASME, 1990. ASME 90GT19. 41, 87, 109,
148
30 DAWES, W. Towards Improved Throughflow Capability: The Use of 3D Viscous
Flow Solvers in a Multistage Environment. ASME Paper. ASME, 1990. ASME
90GT18. 41, 109
31 CHEN, J.; CELESTINA, M. A New Procedure for Simulating Unsteady Flows
Through Turbomachinery Blade Passages. ASME Paper. ASME, 1994. ASME
94GT151. 42, 109
32 AGARD, AR355., 1998, France. CFD Validation for Propulsion System
Components. 7 Rue Ancelle, 92200 NeuillySurSeine, France: AGARD, 1998. 42, 63
33 ADUMITROAIE, V.; RISTORCELLI, J. R.; TAULBEE, D. B. Progress in
Favr´eReynolds Stress Closures for Compressible Flows. n. CR19982078423, 1998. 42
34 VEUILLOT, J.; MEAUZE, G. A 3D Euler Method for Internal Transonic
Flows Computation with a MultiDomain Approach: Calculs D’Ecoulaments
Tridimensionnels Transsoniques Internes a Partir des Equations D’Euler dans Une
Approche MultiDomaine. 29 Avenue de la Division Leclerc  92320 CHATILLON, 1985.
(AGARD/PEP  Lecture Series n◦ 140, Paris, 1314 juin 1985). 42, 84
35 PULLIAM, T.; STEGER, J. Implicit FiniteDiference Simulations of
ThreeDimensional Compressible Flow. AIAA Journal, v. 18, p. 159–167, fev. 1980. 42
36 BEAM, R.; WARMING, R. An Implicit Factored Scheme for the Compressible
NavierStokes Equations. AIAA Journal, v. 16, n. 4, p. 393402, abr. 1978. 42
37 REKLIS, T.; THOMAS, P. ShockCapturing Algorithm for the NavierStokes
Equations. AIAA Journal, v. 20, p. 1212–1218, jul. 1981. 42
38 SENGUPTA, T.; GANERIWAL, G.; DE, S. Analysys of Central and Upwind
Compact Schemes. Journal of Computational Physics, v. 192, p. 677694, jul 2003.
42
39 YEE, H. C. Explicit and Implicit Multidimensional Compact HighResolution
ShockCapturing Methods: Formulation. Journal of Computational Physics, v. 131,
p. 216232, 1997. 42, 85
BIBLIOGRAPHY
218
40 YAO, J.; SHANHONG, J.; LIU, F.; JENNIONS, I. Computation of Steady Flow
in a Turbomachinery Stage by a ThirdOrder Accurate HighResolution
Scheme  Copyright 1997, American Institute of Aeronautics and Astronautics, Inc. 43
41 YANG, J. Third Order Nonoscillatory Schemes for the Euler Equations. AIAA
Journal, v. 29, n. 10, p. 1611–1618, out. 1991. 43
42 LACOR, C.; SMIRNOV, S.; BAELMANS. A finite Volume Forrmulation of
Compact Central Schemes on Arbitrary Structured Grids. Journal of Computational
Physics, v. 198, p. 535566, jan 2004. 43, 85
43 CHIMA, R.; LIOU, M. Comparison of the AUSM+ and HCUSP Schemes for
Turbomachinery Applications. Cleveland, Ohio, 2003. Glenn Research Center,
Cleveland, Ohio, 2003. NASA/TM 2003212457. 43, 72
44 FRINK, N. Recent Progress Toward a ThreeDimensional Unstructured
NavierStokes Flow Solver. AIAA, 1994. 32nd AEROSPACE SCIENCES MEETING.
Washington, DC: AIAA, 1994. AIAA Paper940061, 1994. p. 120. 43
45 FRINK, N. Assessment of an UnstructuredGrid Method for Predicting 3d
Turbulent Viscous Flows. 1996, Reno, Nevada. p. 111. AIAAPaper960292. 43
46 VEUILLOT, J.; CAMBIER, L. Computation Techniques for the Simulation of
Turbomachinery Compressible Flows: Techniques de Calcul pour la Simulation
D’Ecoulements Compressibles des Turbomachines. 29 Avenue de la Division Leclerc 92320 CHATILLON, 1991. (12th ICNMFD, Oxford (GB), july 913). 43
47 STEWART, M. E. Axisymmetric Aerodynamic Numerical Analysis of a
Turbofan Engine ASME Paper. ASME, 1988. ASME 95GT338. 43, 85
48 VAVRA, M. AeroThermodynmics and Flow in Turbomachines. John Wiley
& Sons, 1960. 43, 228
49 VKI Short Course on Flow in Turbines. Flow in turbines. von Karman Institute
for Fluid Dynamics, 1969. 43
50 DAWES, W. Development of a 3D NavierStokes Solver for Application to all Types
of Turbomachinery. ASME Paper. ASME, 1988. ASME 88GT70. 43, 87
51 CHEN, J.; GHOSH, A.; SREENIVAS, K.; WHITFIELD, D. Comparison of
Computations Using NavierStokes Equations in Rotating and Fixed Coordinates for
Flow Through Turbomachinery. AEROSPACE SCIENCES Meeting & Exhibit. AIAA,
1997. (AIAAPaper970878). 44
52 ERDOS, J. I.; ALZNER, E. Computation of Unsteady Transonic Flows
Through Rotating and Stationary Cascades. NASA Contractor Report 1997,
2900. 44, 106
53 CHIMA, R. Calculation of Multistage Turbomachinery Using Steady
Characteristic Boundary Conditions. Washington, DC: NASA, 1998. NASA
TM1998206613, 1998. 44, 96
BIBLIOGRAPHY
219
54 CASEY, M. Introduction to the QNETCFD Thematic Area 6 (TA6): CFD for
Turbomachinery Internal Flows. QNETCFD, v. 2, n. 3, dec 2003. 44
55 HIRSH, C.; DEMEULENAERE, A. State of the Art in the Industrial CFD for
Turbomachinery Flows. QNETCFD, v. 2, n. 3, p. 59, dec 2003. 44, 69
56 LAKSHMINARAYANA, B. Turbulence Modeling for Complex Shear Flows. AIAA
Journal, v. 24, n. 12, p. 19001917, may 1986. 44
57 CHITSOMBOON, T. Improved Artificial Viscosity for HighReynoldsNumber κ − ε
Turbulence Model: Institute for Fluid Computational Fluids in Propulsion. AIAA, 26,
1995, San DiegoCA. AIAA, 1995. p. 1–10. (AIAA 952166). 44
58 MENTER, F. Turbulence Modelling for Turbomachinery. QNETCFD, v. 2, n. 3,
p. 1013, abr. 2003. 44
59 TOMITA, J.; BARBOSA, J. A Model for Numerical Simulation of Variable Stator
Axial Flow Compressors. Proceedings of COBEM, 2003. 17th INTERNATIONAL
CONGRESS of MECHANICAL ENGINEERING. ABCM, 2003. COB010239. 44
60 TOMITA, J.; BARBOSA, J. Design and Analysis of an Axial Flow Compressor for a
1MW Gas Turbine. Proceedings of ENCIT, 2004. 10th Brazilian Congress of
Thermal Sciences and EngineeringENCIT. ABCM, 2004. CIT040062. 44
61 TOMITA, J.; CLEVERSON, B.; BARBOSA, J. Performance Study of a 1MW Gas
Turbine. ENCIT, 2004. 10th Brazilian Congress of Thermal Sciences and
EngineeringENCIT. Proceedings of ABCM, 2004. (CIT040089). 45
62 TOMITA, J.; BRINGHENTI C. BARBOSA, J. Study of the Air Bleed Influence in
the Industrial Gas Turbine Performance. COBEM, 2005. 18th International Congress of
Mechanical Engineering. Proceedings of COBEM, 2005. COBEM20051344. 45
63 SOUSA, F.; SILVA, R.; MANZANARES, N.; BARBOSA, J.; TOMITA, J. Design
point efficiency optimization of a multistage axialflow compressor for aero application,
applying a specially developed computer code. Proceedings of COBEM, 2005. 18th
International Congress of Mechanical Engineering. ABCM, 2005. COBEM20051802. 45
64 BRINGHENTI, C.; TOMITA, J.; SOUSA, F.; BARBOSA, J. Gas Turbine
Performance Simulation Using an Optimized Axial Flow Compressor. ASME
TURBOEXPO, 2006. ASME Turbo Expo 2006. Proceedings of ASME.
GT200691225. 45
65 BARBOSA, J. R. Turbinas a G´
as, Publica¸c˜
ao InternaITA, 2008. 52
66 WILCOX, D. C. Turbulence Modeling for CFD. 2nd. ed. California: DCW
Industries, 1998. 57, 63
67 HOFFMANN, K. A.; CHIANG, S. T. Computational Fluid Dynamics for
Engineers Vol. 1, 2 and 3. 2nd. ed. Engineering Education System, 1993. 63, 72, 88
68 CHUNG, T. Computational Fluid Dynamics. Cambridge University Press, 2002.
63, 69
. mar 1981. In: PROG.. 1981. p. D. p. 76 79 FRINK.. Journal of Computational Physics. MALALASEKERA. 14531460. v.. 50. TURKEL. N. 43. PLETCHER. 72 82 ANDERSON. n. 122. 69 76 JAMESON. 85. 85. S. 357372. California. B. 73. v. K. 30. LEER. W. 67. J. Numerical Computation of Internal and External Flows Vol. 88. Numerical Methods for Engineers and Scientists. V. 130. L. W. 97. v. AIAA Journal. AIAA. 524531. FLUID AND Plasma Dynamics Conference. Computational Fluid Mechanics and Heat Transfer. AIAA920439. n. out. In: Lecture Notes in Physics. ALLMARAS. 1990. T. 170). 1993. 136. 32. 575–624. 92. 85 73 SPALART. 1992. p. AEROSPACE SCIENCES MEETING & EXHIBIT. CHAKRAVARTHY. 1992. 1997. An Introduction to Computational Fluid Dynamics . Taylor & Francis. 70–77. J. 1996. 63. 24. WALTERS. maio 1989. 9. 72. Upwind Schemes and Boundary Conditions with Applications to Euler Equations in General Geometries. van. Analysis of FluxSplit Algorithms for Euler’s Equations with Real Gases. C. B. 94 74 HOFFMAN. THOMAS. C. Upwind Scheme for Solving the Euler Equations on Unstructured Tetrahedral Meshes. (V. P. AIAA. 507512. 63. jan. F. J. 69. S. S. set. p. p.. AEROSPACE SCIENCES MEETING & EXHIBIT. 72 83 HIRSH. B. 90. Turbulence Modelling with Application to Turbomachinery. and Difference Schemes. p. FluxVectorSplitting for the Euler Equations. 14. Numerical Solutions of the Euler Equations by Finite Volume Methods Using RungeKutta TimeStepping Schemes. Parameter Vector. v. 447481. AIAA Journal. v. H. 1981. 143 77 OSHER. 72 80 LEER. Approximate Riemann Solvers. NASATM102847. Assestment of High Order Turbulence Models for Complex TwoandThreeDimensional Flowfields. 1989. W.. 72. 63 70 BRADSHAW. B. E.. P. 27. 116 . 76 78 ROE. K.BIBLIOGRAPHY 220 69 BALDWIN. 63 72 TANNEHILL. 1982. Palo Alto: AIAA. 1995. AEROSPACE SCI VOL. R. R. 1–14.. 5. R. 1983. 73 81 GROSSMAN. A. S. n. A. D. A OneEquation Turbulence Transpor Model for High Reynolds Number WallBounded Flows. McGrawHill International Editions. AIAA.. Comparison of Finite Volume Flux Vector Splittings for the Euler Equations. H. 1992. 72.. p. p. P. 63 71 MENTER. 69. Longman Scientific & Technical. 93. AIAA Journal.. A OneEquation Turbulence Model for Aerodynamic Flows. SCHMIDT.. 1992. 72. John Wiley & Sons. p. J. 72. 95 75 VERSTEEG. NASA. NASATM103944. W. Journal of Computational Physics. 1. 1986. ANDERSON. BARTH.The FiniteVolume Method. 1: Fundamentals of Numerical Discretization. NASA..
.BIBLIOGRAPHY 221 84 GRESSIER. MAVRIPLIS. 81 92 JAMESON.. v. 611618. D. C. L. 19311940. D. 1997. 155. Improved Implicit Residual Smoothing for SteadyState Computations of FirstOrder Hyperbolic Systems. 47. D. 90.. p. 85. S. Flux Vector Splitting of the Inviscid Gasdynamic Equations with Application to Finite Difference Methods. 129–145. 133. aug 1997. AIAA. 1987. P. p.. 1986. A. Journal of Computational Physics. T. MOSHETTA. 2. 81. 79. (AIAA Paper 930880). 291296. v. 28.. Positivity of Flux Vector Splitting Schemes. 1959.. n. A Difference Scheme for Numerical Computation Discontinous Solution of Hydrodynamic Equations. Y. 24. C. Artificial Dissipation Models for Euler Equations. 102 93 MAVRIPLIS. 26. CHIANG. p. dec 1986. Multistage Schemes with Multigrid for Euler and NavierStokes Equations. J. J. The Design and Application of Upwind Schemes on Unstructured Meshes. 107. 47. v. n. AIAA Journal. Multigrid Solution of the Twodimensional Euler Equations on Unstructured Triangular Meshes. P. 91 94 MAVRIPLIS. BRIAN. abr. (AIAA Paper 890366). v. FLANNERY. T. F. V. p. 96 . 263293. 191202. T. TEUKOLSKY. 92. AIAA. 24. 76 87 HIRSH. n. PressureBased Residual Smoothing Operators for Multistage Pseudocompressibility Algorithms. R. 12. Accurate Multigrid Solution of the Euler Equations on Unstructured and Adaptative Meshes. WILLIAM. Journal of Computational Physics. On the Accuracy of Limiters and Convergence to SteadyState Solutions. D. 92 99 WILLIAM. p. 87 95 PULLIAM. 82. 2. 271306. n. p. 73 86 GODONOV. apr 1981. WARMING. 91 96 SWANSON. 1990. p. C. 2: Computational Methods for Inviscid and Viscous Flow. 130. sept 1995. K. Finite Volume Solution of the TwoDimensional Euler Equations on a Regular Triangular Mesh. apr 1993. v. VETTERLING. p. CONSTANTINESCU.. v. feb 1990. v. n. E. 4. 1993. Journal of Computational Physics. 92 98 ENANDER. v. v.. v. NASA TP3631. mar. 1997. F. Numerical Recipes in FORTRAN. SU. Journal of Computational Physics. AIAA Journal. 72 85 STEGER. AIAA Journal. P. D. p. 81. K. 1989. H. JESPERSEN. 92 97 SOTIROPOULOR. 82 90 VENKATAKRISHNAN. 824831. V. AIAA Journal. p... Journal of Computational Physics. 40. Numerical Computation of Internal and External Flows Vol. 85. TURKEL. 213221. J. 118. 83 91 TAI. 7. Convergence to SteadyState Solutions of Euler Equations on Unstructured Grids with Limiters. Explicit Time Marching Methods for the TimeDependent Euler Computations. 81 89 VENKATAKRISHNAN. Sbornik.. 116 88 BARTH. 199220. John Wiley & Sons. R. v. G. C. VILLEDIEU. H. Journal of Computational Physics. 1995. 120130.. jul 1999. Math. J. p.
ERISMAN. n. ASME. SPEAR.. C. 100 105 JAMESON.. CELESTINA. M. out. 5. BRILEY. A Model for Closing the Inviscid Form of the AveragePassage Equation System. apr 1998. 1996. R. In: ASME. dez. 102 107 HUENE. 106 103 GILES. Inviscid 2D Solutions on Unstructured. 112. J.. M. 1986. Press.. 213. J. v. 111 . T. KHAN. J.. M. M. ASME TURBO EXPO 2001.. Development and Application of a Multistage NavierStokes Solver: Part iiApplication to a HighPressure Compressor Design. Iterative Methods for Sparse Linear Systems. 1983. 3. 120. n. 180186. apr 2000.USA. A Parallel Flow Solver for Unsteady Multiple Blade Row Turbomachinery Simulations. H. M.. Journal of Turbomachinery. v. Multistage ThreeDimensional NavierStokes Computation of OffDesign Operation of a FourStage Turbine. Oxford. 120. A. v. M. Nonreflecting Boundary Conditions for Euler Equation Calculation.Louisiana. J. p. p. v. J. v. 108. Y. S.BIBLIOGRAPHY 222 100 SAAD. M. A. 243261. 215222. v. 29. ASME. M. p. MIT. 2. 110. D. C. S... M.. M. New Orleans.. PWS Publishing Company. 28. October 1988. 2226.. maio 1991. SHARP. A. I. D. p. 111 112 GEROLYMOS. 100. v. C. 20502058. apr 1998.. SPEAR. Direct Method for Sparse Matrices. M. JUNG. 1999. Transactions of the ASME. D. T.. E. p. HANISH. M. GLEIXNER. C. ZACHARIAS. R. p. IMechE. 109 111 ADAMCZYK. Adaptive Grids.. State of the Art in RotorStator Interaction for Turbines. A. 111 114 RHIE. AIAA Journal. v. G. 1990. 122. 111 113 RHIE. n. AIAA Journal.. Massively Parallel ThreeDimensional Euler/NavierStokes Method. 234246. GLEIXNER. 100 104 HOLMES. UNSFLO: A Numerical Method for Unsteady Inviscid Flow in Turbomachinery . Development and Application of a Multistage NavierStokes Solver: Part iMultistage Modelling Using Bodyforces and Deterministic Stresses.. 109 110 BELARDINI. L. Applied Mathemathics and Computation 13:327355. 1990. p. 96 102 GILES. Comparison of Different Acceleration Techniques and Methods for Periodic Boundary Treatment in Unsteady Turbine Stage Flow Simulations. Solution of the Euler Equations for TwoDimensional Transonic Flow by a Multigrid Method. Simulation of Viscous Flow Within a Multistage Turbine. J. G. REID. QNETCFD. M. A. 96 101 DUFF. 111 109 CHEN. J..internal report. BARNETT. 205214. v. A. C. Proceedings of ASME. ZACHARIAS. VKILS 198906.. J. 106. dec 2003. B. NY. R. 2001GT0348.. 106 108 ADAMCZYK. 1986. p. J. M. 2001. BEACH. FISCHBERG. 12. p. R. Journal of Turbomachinery. 102 106 LONG. J. 657666. C.. A. W. A. FISCHBERG..
AIAA Paper 910184. An Implicit FluxVector Scheme for the Computation of Viscous Hypersonic Flows. H. 126 124 WENTZ. R.. 1974. 92200 NeuillySurSeine. B. AIAA.. Germany.. 111 116 H¨aNEL. France. I. FLETCHER. v. SEETHARAM. B. BRITISH ARC. 121 121 AGARD. AIAA Paper 890274. D. L. 1951. SCHWANE. G. 2006.. A Mean Line Prediction Method for Axial Flow Turbine Efficiency. L. C. R. Numerical Influence of Upwind TVD Schemes on Transonic Airfoil Drag Prediction. On the Accuracy of Upwind Schemes for the Solution of the NavierStokes Equations. 147 127 DUNHAM. PUTNAM. K.internal report. D. 2008.. BULLOCK. FLUENT 6. Berlim. G. THOMAS. MATHIESON. R. G. G. D. ed. ed. U. V. H. J. Transactions of the ASME. 1992. W. AIAA Paper 871104. M. AIAA. 1991. New York: Springer. 147 129 MOUSTAPHA. 92. AIAA Paper 871105. 192 131 WALSH. A Nozzle Internal Performance Prediction Method. 147 128 KACKER. 1970.BIBLIOGRAPHY 223 115 HOLMES. P. 252. G. Mixing Planes Revisited: A Steady Mixing Plane Approach Designed to Combine High Levels of Conservation and Robustness. C. 1982. 1990.. An Improved Incidence Losses Prediction Method for Turbine Airfoils . GERSTEN. 174. S. 121 120 INC. 1 and 2. J. 2nd. 1989. DECEMBER. Blackwell Science.. M. L. P. A Selection of Experimental Test Cases for the Validation of CFD Codes. ASME TURBO EXPO 2008. 2000. NASA TP1704. 125 122 MASON.. 121 119 SEIDER. P. R. E. ROE. Proceedings of ASME. L... Improvements of the Ainley & Mathieson Method of Turbine Performance Prediction. GT200851296. R. 2004. AIAA. D. 1994. H. Journal of Enginnering for Power. 7 Rue Ancelle. 183 .. NASA TP3221.. D. 1987. 126 123 CARLSON. A Comparison of Numerical Flux Formulas for the Euler and NavierStokes Equations. H¨aNEL. W. P. 110. S.. France: AGARD. KACKER.User’s Guide. A.. 134 126 AINLEY. AIAA.. OKAPUU. In: ASME.. R. 130 125 SCHLICHTING. 120 118 H¨aNEL.. 1956. 1987. C.internal report. C. NASA SP36. J. TREMBLAY. The Effect of Throat Contouring on TwoDimensional ConvergingDiverging Nozzles at Static Conditions. AR303 vol. CAME. 1994. S. J. Gas Turbine Performance. NEWSOME. BoundaryLayer Theory. RE. Development of a Fowler Flap System for a High Performance General Aviation Airfoil . SCHWANE.3 . 120 117 LEER. O. SEIDER. Aerodynamic Design of AxialFlow Compressors.. 191.. p. H. p. 147 130 JOHNSEN. R & M2974. R. A Method of Performance Estimation of Axial Flow Turbines. 8th. 1980..
AGARDLS39. P. GRAF. J. M.. C. GT200851187. TOMITA. W. 1980.Single Stage Evaluation of Highly Loaded High Mach Number Compressor Stages. ASME 67WA/GT7. ASME 67WA/GT11.. The Use of BoundaryLayer Techniques to Calculate the Blockage from the Annulus BoundaryLayers in a Compressor. 192 136 MONSARRAT.. ASME. J. HARTMANN. Proceedings of ASME. 192 142 BECKER. The Application of EndWall BoundaryLayer Effects in the Performance Analysis of Axial Compressors. ed. N. D. G. ASME. S. A. 192 141 RAW. 188 133 DRANSFIELD. 1961. M. 213 143 GREITZER. C. Detailed Flow Measurements in a Four Stage Axial Compressor. 1976. C. Cambridge University Press.BIBLIOGRAPHY 224 132 JANSEN. T.. 192 137 DAVIS. L. 1st. W. A. An ObjectOriented Parallel FiniteVolume CFD Code. J. 188 134 STRAFORD. BARBOSA. Internal Flow: Concepts and Applications. 188 135 SWAN.internal report..ASME reprint . 228 . 1967. R. Eng. ASME. 1947. D. B. T. In: ASME TURBOEXPO. Through Flow Calculations Based on Matrix Inversion: Loss Prediction. Jul. C. In: Fluid Dynamics of Axial Compressors . J. Power.. M. 1969. R. S. Development of the British Gas Turbine Unit. W. 2006. 192 139 GRIEPENTROG. J. 1957. The Prediction of OffDesign Characteristics of AxialCentrifugal Compressors.. Secondary Flow Losses in Axial Compressors. SAE Paper 800628. J. E. C. 2004. ASME Turbo Expo 2006. A Practical Method of Prediction Transonic Compressor Performance. 192 140 HOWELL. KEENAN. H. D. W. WEIR. NACA RME57E30.. MILLAR. A Preliminary Analysis of the Magnitude of Shock Losses in Transonic Compressors. A. SAE. B. AGARDCP195 192 138 SCHWENK. ASME 76GT46. NASA CR196972526. TRAMM. R. R.. J. 1967. TAN. CALVERT.. Design Report .
1 Stencil Applied on Artificial Dissipation Implementation Nonlinear artificial dissipation models.Artificial Dissipation Issues A.Appendix A .3 show that the . FIGURE A.1 – Stencil used to calculate the artificial dissipation terms In this work. Figure A. The stencil of artificial dissipation calculation for each element i is presented in Figure A.2 show that two internal elements were used to calculate the values of second and fourth order terms by the matrix of conserved variables. For others directions the scheme is straightforward. At the boundaries. three different schemes was implemented in GTTeam–CFD code and it is presented bellow Figure A.1 only in one direction. generally has terms of second and fourth order. two layers of ghost elements was created to compute the artificial dissipation terms in the boundary elements.
. The connectivity table supply all neighborhood for each element i.4 show that the second and fourth order terms are calculated using the boundary condition values computed by the matrix of conserved variables.2 – First scheme to calculate the second and fourth order terms of artificial dissipation FIGURE A.APPENDIX A. the scheme presented in Figure A. In general.3 – Second scheme to calculate the second and fourth order terms of artificial dissipation ghost elements uses the same values of the matrix of conserved variables of the first adjacent internal neighbor.4 is more attractive because the terms of artificial dissipation equation are calculated with the actual condition of the problem. ARTIFICIAL DISSIPATION ISSUES 226 FIGURE A. Figure A. but some modifications in the ghost elements and its adjacent boundary elements are necessary to implement each scheme above presented.
4 – Third scheme to calculate the second and fourth order terms of artificial dissipation .APPENDIX A. ARTIFICIAL DISSIPATION ISSUES 227 FIGURE A.
z). ωz = 2πN 60 (A. 0. The Coriolis force is calculated by ~ F~CO = 2~ω × W (A. ωz ) and ~r is the radius vector i the frame of reference ~r = (x. Wy . [8] and [143].Annex A .1) ~ ~ F~c = ω × (ω × ~r) (A. ω where W ~ is the angular velocity.2) and the centrifugal force by ~ is the vector of the relative velocity components W ~ = (Wx . in this work. restrict to the component about the zaxis of the cartesian coordinate system ω ~ = (0.3) . y. Wz ).1 Coriolis and Centrifugal Force In a rotating frame of reference Coriolis and Centrifugal force should be accounted in the NavierStokes equations [48].Rotating Frame of Reference A.
and ycomponent of the momentum equations.4) where N is the rotational speed. the definition of the internal energy presented in 2.ANNEX A. Hence the sum of this forces is calculated by ρωz2 x + 2ρWy ωz F~CO + F~c = ρωz2 y − 2ρWx ωz 0 (A. Hence.2 has to be changed by ei = e 1 − (W 2 + Wy2 + Wz2 − ωz2 y 2 − ωz2 x2 ) ρ 2 x (A.6) .5) This shows that a rotation around the zaxis influences only x. ROTATING FRAME OF REFERENCE 229 and the dimensionless angular velocity is defined by ωz = ωl0 a0 (A. l0 is the characteristic length and a0 is the reference speed of sound.
Gas turbine. Turbinas a g´as. Publicada em 2009. In these cases. In the last years many CFD commercial packages were developed and some of them possess prominence in industry and academia. For instance. ˜ APRESENTAC ¸ AO: (X) Nacional ( ) Internacional ITA. 2. ´ No DE PAGINAS CTA/ITA/TD001/2009 229 T´ITULO E SUBT´ITULO: ThreeDimensional Flow Calculations of Axial Compressors and Turbines Using CFD Techniques 6. the verification and validation processes were run for both inertial and noninertial systems. Mecˆanica dos fluidos. Orientador: Jo˜ao Roberto Barbosa. The origin of this solver is to simulate flows in compressors and turbines. It is very important to mention that to have a complete understanding of the flow physics in compressors and turbines the designer must have a solid knowledge of the operation of gas turbine components. Numerical methods. GRAU DE SIGILO: (X) OSTENSIVO ( ) RESERVADO ( ) CONFIDENCIAL ( ) SECRETO . S˜ ao Jos´e dos Campos. was developed to calculate internal flows in turbomachines using CFD techniques. Different numerical schemes were implemented for time and space integration. ˜ PALAVRASCHAVE RESULTANTES DE INDEXAC ¸ AO: Dinˆ amica dos fluidos computacional. Engenharia mecˆanica 10. 11. An´alise num´erica. Fluxo tridimensional. a computational code. The development of CFD codes applied to turbomachinery flow simulations and its implementation issues are not available. The approach used allows the use of unstructured meshes of hexahedral elements. The solver is capable of calculating the threedimensional flows not only for turbomachines. written in FORTRAN. Defesa em 07/01/2009. DATA 3. PALAVRASCHAVE SUGERIDAS PELO AUTOR: CFD.FOLHA DE REGISTRO DO DOCUMENTO 1. Propuls˜ ao e Energia. Turbocompressores. Escoamento turbulento. Hence. meticulous research becomes necessary. NavierStokes and turbulent equations can be calculated depending on the user settings. However. Therefore. both rotating and nonrotating frames of reference are calculated simultaneously. A few institutions have this type of knowledge. 19 de janeiro de 2009 DOCUMENTO No 4. ˜ OES)/ ˜ ´ ˜ ˜ OES): ˜ INSTITUIC ¸ AO( ORG AO(S) INTERNO(S)/DIVISAO( Instituto Tecnol´ ogico de Aeron´ autica . Developing a CFD code is very interest subject in academia. internal and external flows of nozzles and airfoils can be calculated. Each CFD code has its particularities. Computational Fluid Dynamics (CFD) has been vastly used by researches and scientists to investigate flow behavior and its properties.ITA 8. some specific CFD calculations are particular cases and sometimes need special attention due to the complexity of the flow. AUTOR(ES): Jesuino Takachi Tomita 7. Numerical tools to improve the stability and to increase the timestep (local timestep and implicit residual smoothing) were also implemented and all details are described in this work. ˜ CLASSIFICAC ¸ AO/TIPO TD 5. 12. Area de Aerodinˆ amica. Euler. The cost of CFD simulation is very small compared to the experimental arsenal as test facilities and windtunnels. In this work. Programa de P´osGradua¸c˜ao em Engenharia Mecˆanica´ Aeron´ autica. A stepbystep design procedure is presented in this work. Streamline curvature 9. Axial compressor. Curso de Doutorado. Turbom´ aquinas. This is the case of turbomachinery flow calculations. RESUMO: With the advent of powerful computer hardware.