You are on page 1of 144

MATHEMATICS

,
NUMERICS,
DERIVATIONS
AND OPENFOAM®
The Basics for Numerical Simulations.

Tobias Holzmann
www.holzmann-cfd.de

i

Dedidcated to the OpenFOAMr community and especially to all colleagues and people who
support me. The ambition to write the book is based on my personal love to the open source
thought. Thus, my objective is to give you an introduction to computational fluid dynamics, show
interesting equations and some relations which are not given in most of the books and papers which
are famous in that area. In addition, the book should prepare you for the tasks that you may work
on during your personal career, hopefully with OpenFOAMr .

The book can be ordered as a soft-cover version. If you are interested write me an email to
Tobias.Holzmann@Holzmann-cfd.de

To get further information about my projects, developments and my personal road, you are
welcomed to follow me on Twitter, Facebook, Linkedin, XING or Youtube. If you want to get a
private email about the latest projects, you can register for free on my webpage.

Each feedback and all critics are taken into considerations. Please let me know if you find
mistakes or if you think some special topic is missing.

Cooperations are welcomed. If you want to contribute to the book, please do not hesitate to
write me an email.

ii

Copyright Tobias Holzmann, April 4, 2018

What is this book about
This book collects aspects of mathematics, numerics and derivations used in the field of compu-
tational fluid dynamics (CFD) and OpenFOAMr . The author of the book tries to keep the book
up-to-date.

Differences in the release
The release notes of the book are available at www.holzmann-cfd.de. Check out the download
section and you will find the changes that were made during a revision.

Acknowledgment
I thank Dr. Alexander Vakhrushev for the interesting discussions we had in our apartment and the
deep insight into the topics of mathematics and programming in OpenFOAMr . Furthermore, I
would like to thank Sergei (Zeppo) from the cfd-online platform, Vigneshkumar and Vishwesh Ravi
Shrimali for the useful remarks and corrections. In addition I thank my sister Michaela Holzmann
for further improvements and Andrea Jall for the beautiful cover page and all the support in my
life.

How to cite?
The citation format depends on the journal or your personal style. The following citation is just
an example based on the format style that is used in this document:

Tobias Holzmann. Mathematics, Numerics, Derivations and OpenFOAM(R), Holzmann CFD,
URL www.holzmann-cfd.de, DOI: 10.13140/RG.2.2.27193.36960.

the mathematical equations are given and a few words about the numerical stabilization is said. Finally. the shear-rate tensor τ and the pressure p is given. all kinds of energy equation are discussed and presented namely the kinetic energy. the Reynolds-Stress equation — which is fully derived in the appendix — and the analogy to the Cauchy stress tensor is shown. After that a discussion between the analogy of the Cauchy stress tensor σ. After presenting the mathematic aspects. Subsequently the incompressible equations are derived and finally the closure problem is discussed in detail. Subsequently. the general governing equation is introduced afterwards and it is demonstrated how to use the general conservation equation in order to derive other ones. Subsequently. iii Outline This book gives an introduction to the basic mathematics used in the field of computational fluid dynamics. internal energy. The last chapter is related to OpenFOAMr beginners which are seeking for tutorials and some other useful information and websites. dV . Based on the fact that engineering applications are mostly turbulent. total energy and the enthalpy equation. Based on the nature of the equations. the eddy-viscosity theory is introduced and the equation for the turbulent kinetic energy k and dissipation  are deducted. To close the subject of turbulent flows. The following chapters discuss the definition of the shear-rate tensor τ for Newtonian fluids. At the beginning the derivation of the mass and momentum equation are described. The last chapters of the book are related to the detailed explanation of the implementation of the shear-rate tensor calculation in OpenFOAMr . . Here. all conservation equations are derived using a finite volume element. a more general discussion of the different pressure-momentum coupling algorithms is given. the PIMPLE-algorithm is explained while considering an OpenFOAMr case. All equations are summed up with a one page summary at the end. During the investigation into the C++ code. The topic ends with a brief description about the derivation for the compressible Navier-Stokes-Equations equations and its difficulties and validity. the Reynolds-Averaging methods are presented and explained.

iv .

. . . . . . . . . .6 The Conserved Enthalpy Equation .2. . . . . . .5 The Conserved Thermo Energy Equation . . 18 2. . . . . . . . . . . . . . . . . . . . . . . . 24 2. . . . . .4 The Conserved Mechanical Energy Equation . . . . . . .3.2 Continuity Equation and the Total Derivative . . . . . . . . . . . .4. . 34 v . . . . . . . . . .2 Integral Form of the Conserved Momentum Equation . . . . . . . . . . .3.3. . 18 2. . . . 26 2. . . 33 2. . . . . . . . .2 Non-Conserved Mechanical Energy Equation .6. . Deviatoric and Hydrostatic Part . . .1 The Proof of the Transformation . . . . . . . . . . . . .1 The Continuity Equation . . 9 2.2 Integral Form of the Conserved Total Energy Equation . . .6. . . . . . . . . . . . . . . . . . . . . . . . . . . 24 2. . 6 1. . . . . . . . . . . . . . . . . . . . . .1 Integral Form of the Conserved Enthalpy Equation .3. . . . . . . . . .2. . . . . . . . . . . . . . 14 2. . . . .1 The Proof of the Vector Transformation . . . . . . .2. . 25 2. .1 Integral Form of the Conserved Continuity Equation . . .3. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .3 The Gauss Theorem . . .2 Non-Conserved Thermo (Internal) Energy Equation . . . . . . . . . . . . . . . . . 2 1. . . . . . . . . . . . . . . . . . . . . . . .3. . . . 2 1.3 The Conserved Enthalpy Equation (only Thermo) . . . . . . . .1. .4 Kinetic Energy and Internal Energy .1 The Total Derivative . . . . . . . . . . . .4. . . . . . . . . . . . . . . . . . . . 20 2. . . . . . .1. . . . . . . . . . . . . . . . . . . . 33 2. . . . . . . .3 Non-Conserved Total Energy Equation . . . . 12 2. 31 2. . .3 Non-Conserved Momentum Equation . . . . . . . . . . . . . . . . . . . . . . . . .5. . . . . . . . 23 2. . . . . . . . . . .3 The Conserved Total Energy Equation . . . . . . . 17 2. . . . . .2 The Conserved Momentum Equation . . . . . . . . . . . . . . . . . . . . . . 29 2.3 General Tensor Mathematics . . . .2 Matrix Algebra. . . . . . .1 Basic Rules of Derivatives . . . . . . . 32 2. . . . .6. . . . . . . . . . . . . . .2 Einsteins Summation Convention . . . . . . . . . . .1 Integral Form of the Conserved Mechanical Energy Equation . . . . . . . . . . . . . . 1 1. . . . . . 32 2. .3. . . . . . . . . . . . . . . . . . .1 Integral Form of the Thermo Energy Equation . . . . . .2 Non-conserved Enthalpy Equation . . 33 2. . . . . . . 7 2 Derivations of the Governing Equations 9 2. . 29 2. . . . . . . 6 1. . . . . . . . . .Contents 1 Basic Mathematics 1 1. . . . . . . . . . 12 2. . .5. . . . . . . . . . . . . . . .

. . . . . . . . . . . . 64 9. . . . . . . . . . . 63 9. . . . divDevReff . . . .1 Reynolds-Averaging . . . . . . . . . . 43 5. . . . 39 4 Summary of the Equations 41 5 The Shear-rate Tensor and the Navier-Stokes Equations 43 5. .1. . . . . . . . . . . . . . .2.2 Non-Newtonian fluid . . . . . . . . . . 69 9. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .3. . . . . . . . . . . . . . . . . . . . . .3 Influence of Turbulence Models . . . . . . . . . . . . . . Shear-Rate Tensor. . . . . 77 9. . . . . . . . . . . .6 Algebraic Models . . . . . . . . 48 5. . . . . . . . . . .9 The Incompressible Kinetic Energy Equation . .1. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .2 The Compr. . . . .12 Coupling of the Parameters .13 Turbulence Modeling for Compressible Fluids . .8 Incompressible Reynolds-Stress Equation . . . . . divDevRhoReff . . . . . . . . .3 Further Simplifications . . 71 9. . . . . . . . . . . . . . . . . . Shear-Rate Tensor. . . . . . . . . . . . . . . . . . . . . . .1 Common Source Terms . 77 9. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 59 9. . . .1 The Inco.1 Incompressible Mass Conservation Equation .3 Incompressible Momentum Equation . . . . . . . . . . . . . . . . . . . . . . . . . . .2. . .1 Newtonian Fluids . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .10 The Relation between  and L . . . . . . . . . . . . . . . . . . . . . . .2. 77 9. . . . . . . . 79 9. . . . . . . . . . . . . . . . . . . . .5 Eddy Viscosity Approximation . . . . . . . . . . . . . . . . . . . .1 The Proof of the Transformation . . . . . . . . . . . .4 The (Incompressible) General Conservation Equation . . . . . . . . . . . . . . . . . . . . 83 9. . . . . . . . . 38 3. . . . . . . . . . . . . . . . . . . . . . . . . . . . . 90 . . . . . . . . . . . . . . .3 The Enthalpy Equation . . . . . 83 10 Calculation of the Shear-Rate Tensor in OpenFOAMr 85 10. 49 6 Relation between the Cauchy Stress Tensor. . . . . . . . . . . . . . . . 77 9. .2 Compressible Mass Conservation Equation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 37 3. . . . . . . . . . . . . . 37 3. . .1 The Continuity Equation . . . 72 9. . . . . . . . . . . . . . . . . . . . . . . . 85 10. . . . . . . . . . . . . . . . . . . .2. . . . . . . .4 Boussinesq Eddy Viscosity . . . . . . . . .2 Reynolds Time-Averaged Equations . . . Shear-Rate Tensor and Pressure 51 7 The bulk viscosity 53 8 Collection of Different Notations of the Momentum Equations 57 9 Turbulence Modeling 59 9. 46 5. . . . 82 9. . . . . . . . 49 5. . . . . . . . .vi CONTENTS 3 The Governing Equations for Engineers 37 3. . . . . . . . . .1. .7 Turbulence Energy Equation Models . . . . . . . . . 82 9. . . . . . . . . . . . . . . . . . . . . .3 The Closure Problem . . . . . . . . . . . .2 The Term − 23 µ(∇ • U) . . . . . . . . . . .11 The Equation for the Dissipation Rate  . . . . . . . . . 88 10. . . . . . . . . . . . . . . . . . . . . . . 65 9. . 63 9. . . . .2 The Momentum Equation . .

. . . . . 117 . . . . . . . . . .5 PIMPLE working as PISO with large ∆t . . . . . . . . . . . . . . . . . . 98 11. . . . . . . 105 11. . . 102 11. . . . . . . .4. . . . . . . . . . . . . . . . . . . . . . .8 PIMPLE algorithm with under-relaxation .4. . . . . . . . . . . . . . 95 11. . . . . 94 11. . . . . . . .2 The PISO algorithm in OpenFOAMr . . . 101 11. 100 11. . . . . . . . . . . . .4. . . . . . . . . . . . . . . .1 Field relaxation . . . . . . . .7 PIMPLE algorithm further modified (add inner corrections) . . . . . . .3 The PIMPLE algorithm in OpenFOAMr . . . . . . . . . . . . . . . . . . . . . .10 PIMPLE conclusion . . . . . . . . . . . . . 113 13 OpenFOAMr tutorials 115 14 Appendix 117 14.1. . . . . . . .4. . . . . .1 The Incompressible Reynolds-Stress-Equation . . . . . . . . . . . .4.CONTENTS vii 11 SIMPLE. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .4. . . . . . . . . . . . . . . . . . . . . . 92 11. . . . . . . . . . . . PISO and PIMPLE algorithm 91 11. . . . . . . .2 Matrix relaxation . . . . . . . . . . . . . . . . . . . . . . . . .4. . . .4. . . . . . . 113 12. . . . . . .1 The SIMPLE algorithm in OpenFOAMr . . . . . . . . . . . . .5 PIMPLE algorithm flowchart . . . . . . . . 95 11.3 Run the case with the PISO algorithm . . . . . . . . . . . . . . . . 92 r 11. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 111 12 The relaxation methods 113 12. . . . . . . . . . . . . . . 98 11. . . . . . . . . . . . . . . . . . . . . . . . . . . . .9 PIMPLE algorithm speed up . . . . . . . . . . .4 PIMPLE working as PISO . . . .1 SIMPLEC in OpenFOAM . . . 109 11. . . . . . . .2 First considerations .6 PIMPLE algorithm modified (add outer corrections) . . . . . . . . . . . 99 11.4. . . 104 11. . . . . . .4 The correct usage of the PIMPLE algorithm . . . . . . .1 The test case . . . .4. . . . . . . 111 11. . . . . . . . . .

viii CONTENTS .

a brief summary of the rules that are needed to manipulating and analyzing the equations are given now. To keep things clear we use the following definition which are similar to Greenshields [2015]: Zero rank tensor T0 := scalar a First rank tensor T1 := vector a Second rank tensor T2 := tensor T (matrix of 3x3) Third rank tensor T3 := tensor Tijk If the rank of a tensor is larger than zero. The beauty of mathematics are also described in Jasak [1996]. [2015]. A field can be a scalar. In other words. it is possible to use the product rule to split the term. Considering the sum of two quantities φ and χ that are derived respectively to τ . This knowledge is required if one is going to implement. we have to keep one quantity constant while deriving the 1 . Based on that. a vector or the classical known tensor that represents a matrix (normally a 3 by 3 matrix) and is of rank two. A tensor stands for any kind of field. (1. There are a lot of ways to represent equations and thus a brief collection of the most essential mathematics are given in this chapter.1 Basic Rules of Derivatives The governing conservation equations in fluid dynamics are partial differential equations.Chapter 1 Basic Mathematics In the field of computational fluid dynamics the essential point is to understand the equations and the mathematics. Greenshields [2015] and Moukalled et al. reorder or manipulate equations within a software or toolbox. Dantzig and Rappaz [2009]. The only exception is the derivation of the Reynolds-Stress equation. we can split the derivative: ∂(φ + χ) ∂φ ∂χ = + . 1.1) ∂τ ∂τ ∂τ If we have the derivative of the product of the two quantities. the tensor is always written in bold symbols/letters. Tensors which have a rank larger than two are not needed in most of the numerics presented in this book. In the field of numerical simulations we are dealing with tensors Tn of rank n.

Assuming the sum of the following derivatives of the arbitrary variable φi (such as the mass conservation) in x. (1.4) i ∂x i ∂xi A more complex example that demonstrates the advantage of the Einsteins summation convention is the convective term of the momentum equation (it is not necessary to know the meaning of this terms right now). For that purpose the arbitrary quantities are used: a scalar φ.2 Einsteins Summation Convention For vector and tensor equations there are several options of notations. z . The longest but clearest notation is the Cartesian one. y. ∂x ∂y ∂z ∂ux uz ∂uy uz ∂uz uz + + . (1. (1. Hence.3 General Tensor Mathematics A common and easy way to deal with equations is to use the vector notation instead of the Ein- steins summation convention. the P summation sign is neglected to keep things clear and short: X ∂φi ∂φi = i = x. ∂x ∂y ∂z ∂ux uy ∂uy uy ∂uz uy + + . vectors and tensors are given now.2) ∂τ ∂τ ∂τ A constant quantity C can be taken inside or outside of a derivative without any constrain: ∂ C φχ ∂φχ =C . Due to the fact that the momentum is a vector quantity. Therefore. y and z direction. Commonly. ∂x ∂y ∂z Applying the Einsteins convention the result is as follow: X ∂ui uj ∂ui uj = i = x. z .2 Basic Mathematics other one: ∂φχ ∂φ ∂χ =χ +φ . j = x. (1.3) ∂τ ∂τ 1. The vector notation requires knowledge about special mathematics. y. it is essential to know the meaning and how it is applied. 1. the three single terms of each direction are given as: ∂ux ux ∂uy ux ∂uz ux + + . a brief description of different operations which are applied to scalars. the Einsteins summation convention can be applied.5) i ∂x i ∂xi The Einsteins summation convention is widely used in literatures. ∂x ∂y ∂z To simplify this equation. This notation can be abbreviated — if the equation contains several similar terms which are summed up — by applying the Einsteins summation convention. z. two vectors . the Cartesian form is written as: ∂φx ∂φy ∂φz + + . y.

I = 0 1 0 . e3 = ez = 0 . (1. e2 = ey = 1 . It will be indicated by the colon : sign: . 2.         0 0 1 0 0 1 Simple Operations • The multiplication of a scalar φ by a vector b results in a vector and is commutative and associative.     Tzx Tzy Tzz T31 T32 T33 Depending on the operation of interest. (1. a • T = T • a. (1.7) i=1 • The inner product of a vector a and a tensor T produces a vector b and is non-commutative if the tensor is non-symmetric:   3 X 3 T11 a1 + T12 a2 + T13 a3 X b=T•a= Tij aj ei = T21 a1 + T22 a2 + T23 a3  . z).6)     φbz φTzx φTzy φTzz The Inner Product • The inner product of two vectors a and b produces a scalar φ and is commutative.9)   i=1 j=1 a1 T13 + a2 T23 + a3 T33 A symmetric tensor is given. b = by  = b2  .1. y. φT = φTyx φTyy φTyz  . This is also valid for the multiplication of a scalar φ and a tensor T:     φbx φTxx φTxy φTxz φb = φby  . This operation is indicated by the dot sign •: 3 X φ = a • b = aT b = ai bi . one uses either the numeric indices (1. Furthermore.3.8)   i=1 j=1 T31 a1 + T32 a2 + T33 a3   3 X 3 a1 T11 + a2 T21 + a3 T31 X b = a • T = TT • a = aj Tji ei = a1 T12 + a2 T22 + a3 T32  . The Double Inner Product • The double inner product of two tensors T and S results in a scalar φ and is commutative. the unit vectors ei and the identity matrix I has to be defined:         1 0 0 1 0 0 e1 = ex = 0 . 3) or the space components (x. if Tij = Tji and hence.         az a3 bz b3     Txx Txy Txz T11 T12 T13 T = Tyx Tyy Tyz  = T21 T22 T23  . (1. GENERAL TENSOR MATHEMATICS 3 a and b and a tensor T:         ax a1 bx b1 a = ay  = a2  .

is non-commutative and is expressed by the dyadic sign ⊗:   ax bx ax by ax bz T = a ⊗ b = abT = ay bx ay by ay bz  . (1. y and z in a Cartesian coordinate system:  ∂   ∂  ∂x ∂x  ∂   ∂1  ∇ =  ∂y  =  ∂x2  . Differential Operators In vector notation.10) The Outer Product • The outer product of two vectors a and b. to be more consistent with the mathematics.11)   az bx az by az bz In most of the literatures the dyadic sign ⊗ is neglected for brevity as shown below: ab . (1.14)  ∂y y ∂y z  ∂ ∂ ∂ b ∂z x b ∂z y b ∂z z We see that this operation is actually the outer product of the Nabla operator (special vector) and an arbitrary vector b. vector or tensor) is made using the Nabla operator ∇. In this book we use the definition of equation (1. (1. also known as dyadic product. the spatial derivatives of a variable (scalar. Hence. (1. ∂ ∂ ∂z ∂x3 Gradient Operator • The gradient of a scalar φ results in a vector a:  ∂φ  ∂x grad φ = ∇φ =  ∂φ .15) . that both variants are used in literature whereas the last one is more common but the first one is more clear.4 Basic Mathematics 3 X X 3 φ = T: S = Tij Sij = T11 S11 + T12 S12 + T13 S13 + T21 S21 i=1 j=1 + T22 S22 + T23 S23 + T31 S31 + T32 S32 + T33 S33 . It contains the three space derivatives of x. it is commonly written as: ∇b . (1.11). results in a tensor.13)   ∂y  ∂φ ∂z • The gradient of a vector b results in a tensor T: ∂ ∂ ∂  b ∂x x b ∂x y b ∂x z ∂ ∂ ∂ grad b = ∇ ⊗ b =  ∂y bx b b . (1.12) Keep in mind.

we can split the term using the product rule. . (1. (1.7) If one thinks that the product rule for the inner product of two vectors is missing. (1. Note: The gradient operation increase the rank of the tensor by one and hence.7) simple multiplication • The divergence of the outer product (dyadic product) of two vectors a and b can be split as follows and results in a vector: ∇ • (a ⊗ b) = a • ∇ ⊗ b + |b∇{z• a} .18) | {z } | {z } Eqn. (1.20) | {z } | {z } Eqn.17) i=1 j=1 ∂xj ∂T 1 ∂T23 ∂T33 13 ∂x1 + ∂x2 + ∂x3 Note: The divergence operation decrease the rank of the tensor by one. Hence. After that.10) Eqn. • The divergence of the product of a vector a and a scalar φ can be split as follows and results in a scalar: ∇ • (aφ) = a • ∇φ + φ∇ • a . GENERAL TENSOR MATHEMATICS 5 In this book we use the first notation (with the dyadic sign) to be more consistent within the mathematics. (1.19) | {z } Eqn. (1.1.16) i=1 ∂x i ∂x 1 ∂x 2 ∂x 3 • The divergence of a tensor T results in a vector b:  ∂T ∂T21 ∂T31  11 3 X 3 ∂x1 + ∂x2 + ∂x3 X ∂  ∂T12 ∂T22 ∂T32  div T = ∇ • T = Tji ei =  ∂x + ∂x2 + ∂x3  . ask yourself how the divergence operator will change the rank. Based on the tensor ranks inside the divergence. we have to apply different rules. (1. ∇•: 3 X ∂ ∂b1 ∂b2 ∂b3 div b = ∇ • b = bi = + + .9) Eqn. think about the result of the inner product of the two vectors and which tensor rank the result will have. (1. The Product Rule within the Divergence Operator If we have a product within a divergence term. (1.6) • The divergence of the inner product of a tensor T and a vector b can be split as follows and results in a scalar: ∇ • (T • b) = T : ∇ ⊗ b + b •∇•T . it does not make sense to apply this operator on a scalar. Divergence Operator • The divergence of a vector b results in a scalar φ and is expressed by the combination of the Nabla operator and the dot sign.3. (1. we can apply it to any tensor field. which are given now.

2 Matrix Algebra. tensor.23) Dt ∂t ∂t | {z } | {z } non−conserved continuity = 0 The reason we multiply the continuity equation (second term on the right hand side) by the quan- tity φ comes from the product rule. In the conserved representa- tion. Deviatoric and Hydrostatic Part In the field of numerical simulations we are dealing with quantities that are represented by matrices like the stress tensor. After the momentum equation is derived and the conservative form is transformed into the non-conserved one. In literature people start to derive equations using the total derivative and using the continuity equation to extend the non-conservative equation to the conserved one.7). 1. If you have literature that start with the non-conservation equations. we have to use equation (1. and so on). . the correct mathematical expression for the second term on the right hand side (RHS) has to be applied.9).3. for non-conserved equations it is the Lagrange expression.3. Therefore.6 Basic Mathematics 1. (1.21) denotes the inner prod- uct. (1. In other words.21) Dt ∂t | {z } inner product where U represents the velocity vector. some basic mathematical expressions and manipulations are introduced now. The last term in equation (1. each conserved equation can be changed into a non-conserved equation using the continuity equation. (1. Depending on the quantity φ (scalar. we have to use equation (1. Example given: • If φ is a scalar. that is applied to the convective term. Short Outline for the Total Derivative The total derivative is used to represent non-conserved equations. Why? It is easier to understand.1 The Total Derivative The definition of the total derivative of an arbitrary quantity φ – in the field of fluid dynamics – is defined as: Dφ ∂φ = + U • ∇φ . The difference between both equations is the frame of reference. we have the Euler expression. vector. • If φ is a vector. The better way would be to derive first the conserved equation and then using the continuity equation to get the non-conserved form. this will get clear.22) Dt |∂t {z } | {z } non−conserved continuity = 0 • Compressible:     Dφ ∂φ ∂ρ ρ =ρ + U • ∇φ +φ + ∇ • (ρU) . this would help to un- derstand the following extension (at the moment it is not necessary to understand this equations): • Incompressible: Dφ ∂φ = + U • ∇φ + φ (∇ • (U)) .

it is necessary to know the Gauss theorem. (1. Note: The small dot · denotes the inner product of two vectors (1.7).3 The Gauss Theorem To transform any equation from the differential form to the integral one (or vice versa). the correct mathematical expression for the hydrostatic part of the matrix A is given by: n 1 1X Ahyd = Ahyd I = tr(A)I = (aii )I . The trace operator is zero not the diagonal elements. we use the small dot in all integrals to sign that we calculate the inner product of a vector a and the surface normal vector n. tr(Adev ) = 0. Hence. 1. If one wants to calculate the scalar.3. However. (1.24) The hydrostatic part of the matrix can be expressed as scalar or matrix and is defined by using the trace operator. dS the integration with respect to the surface and dV the integration with respect to the volume.26) 3 3 i=1 The deviatoric part Adev is given as: 1 Adev = A − Ahyd = A − tr(A)I . we use the following definition: n 1 1X Ahyd = tr(A) = (aii ) .28).1. (1. This operator is simply the sum of the diagonal elements.3. n represents the surface normal vector pointing outwards. . Keep in mind that the small dot expresses exact the same mathematical expression as the bullet. (1.27) 3 Note: The deviatoric part of a matrix is traceless. In the following book. (1. This theorem allows us to establish a relation between the fluxes through the surface of an arbitrary volume element and the divergence operator on the volume element: I Z a · ndS = (∇ • a)dV .25) 3 3 i=1 The operator tr denotes the trace operator and is applied on the matrix. GENERAL TENSOR MATHEMATICS 7 Each matrix A can be split into a deviatoric Adev and hydrostatic Ahyd part: A = Ahyd + Adev .28) In equation (1.

8 Basic Mathematics .

2. [1960]. [2015]. Bird et al. that the mass flow that enters and leaves the volume element through its surfaces has to be equal. The main references that are used within this chapter are Greenshields [2015]. total energy. momentum. The equation itself describes the mass balance of an arbitrary volume element dV . Jasak [1996]. Versteeg and Malalasekera [1995]. • Proof that the vector notation results in the Cartesian notation. Schwarze [2013] and Moukalled et al. • Manipulate the equation to get the common form.1 The Continuity Equation In the following section the derivation of the continuity equation is presented. depicted in figure 2. • Transform the Cartesian notation into the vector notation. Furthermore. The structure of this chapter is (mainly) as follows: • Express the phenomena that act on the volume element using finite differences. The equations are derived using the Cartesian coordinate system. Consider the mass flow through a small control volume element dV . That means.Chapter 2 Derivations of the Governing Equations The following chapter demonstrates how to derive the continuity.1) accumulation entering the volume leaving the volume 9 . Dantzig and Rappaz [2009]. me- chanical (kinetic) energy. a mass balance has to be fulfilled for the volume element. • Transform the finite difference equation to a partial differential equation. A complete sum- mary of all equations is given on page 41. Ferziger and Perić [2008]. we have to take the rate of mass accumulation into account: " # " # " # rate of mass rate of mass rate of mass = − . • Transform the equation into the integral and non-conserved form. thermo (internal) energy and enthalpy equation using a small volume element dV . using the constrain that mass is not transformed into energy or vice versa. (2.1.

In other words. and will only change with respect to time. . it follows: ∆ρ  ∆x∆y∆z = (ρux )|x − (ρux )|x+∆x ∆y∆z ∆t  + (ρuy )|y − (ρuy )|y+∆y ∆x∆z  + (ρuz )|z − (ρuz )|z+∆z ∆x∆y .1. The rate of mass that enters or leaves the volume element through the surface is called mass flux and is simply the density times the velocity with respect to the area of the face. the rate of change for the density ρ is related to the volume. (2.1) by using the mathematical expressions given in figure 2.1.3) z Mass inside x : (ρux )|x ∆y∆z Mass outside x : (ρux )|x+∆x ∆y∆z ∆x Mass inside y : (ρuy )|y ∆x∆z ∆z Mass outside y : (ρuy )|y+∆y ∆x∆z ux |x ux |x+∆x • dV Mass inside z : (ρuz )|z ∆x∆y y ∆y Mass outside z : (ρuz )|z+∆z ∆x∆y x Figure 2.1: Mass balance in a small volume element dV . mass per unit volume. that means the density will change.10 Derivations of the Governing Equations To make things clearer. we observe that the velocity vectors are normal to the faces. we can write the rate of change of the density as: ∆ρ Time Accumulation = . (2. Therefore. however.1 and equation (2. The single terms that describe the fluxes at the surfaces are given on the right side in figure 2. It is obvious that the mass is transported through the surface by the velocity.2). if a species or any other quantity is transported due to convection. If we have a compressible fluid.2) ∆t Rewriting equation (2. This transport phenomenon is called convection or sometimes named advection — normally if we are talking about the fluid itself. it is said to be advected — and happens for all three space directions x (ux ). Having a closer look to figure 2. we analyze figure 2. For the derivation of the mass conservation equation we have to build the balance of the fluxes at the surfaces of the volume element. everything that is going inside has to go out if we assume that there is no mass accumulation inside the volume (incompressible). Additionally a mass change inside the element can occur due to compression or expansion phenomena.1. y (uy ) and z (uz ). we refer to convection.

(2.5) and (2. only the mass flux that enters and/or leaves the volume element at its surface has to be taken into account. (2. (2. One may also try to explain it in the following way: if we assume constant density. we could assume that the density is constant and therefore the quantity ρ can be taken out of the derivatives and we are allowed to divide by ρ. (2. (2. Hence.2.7) ∆x ∆x ∂x (ρuy )|y − (ρuy )|y+∆y −∆(ρuy ) ∂ = −→ − (ρuy ) .6) to (2. (2. Remark: In many cases incompressibility means that there is no expansion and/or compres- sion phenomena.8) ∆y ∆y ∂y (ρuz )|z − (ρux )|z+∆z −∆(ρuz ) ∂ = −→ − (ρuz ) . we have to apply equation (2.5) ∆x ∆x→0 ∆x ∂x and also an infinitesimal small time range: ∆ ∆ ∂ −→ lim = . In such a case we . THE CONTINUITY EQUATION 11 Dividing the equation by the volume ∆V = ∆x∆y∆z. For that. (2.10) ∆t ∂t and the general mass conservation (continuity) equation is given by:   ∂ρ ∂ ∂ ∂ =− (ρux ) + (ρuy ) + (ρuz ) . (2. if we focus on incompressible fluids. (2.11) ∂t ∂x ∂y ∂z If we use the Nabla-Operator ∇ and the velocity vector U. the fluid density can still be temperature depended.12) ∂t However.4). It is obvious that the time derivative will vanish due to the fact that the density is a constant and will not change with respect to time.9) ∆z ∆z ∂z ∆ρ ∂ρ → . there is no expansion or compression phenomena and therefore the time derivative can be canceled to zero. However.6) ∆t ∆t→0 ∆t ∂t we can transform the finite difference equation to a partial differential equation. the equation can be rewritten in vector notation: ∂ρ = −∇ • (ρU) . It follows: (ρux )|x − (ρux )|x+∆x −∆(ρux ) ∂ = −→ − (ρux ) . we get: ∆ρ (ρux )|x − (ρux )|x+∆x = ∆t ∆x (ρuy )|y − (ρuy )|y+∆y + ∆y (ρuz )|z − (ρuz )|z+∆z + .1.4) ∆z Introducing the assumption of an infinitesimal small volume element — which means that we decrease the distance between the corners of the volume element and therefore ∆ goes to zero: ∆ ∆ ∂ −→ lim = .

This method is conservative and we can apply this method to each arbitrary volume (hexaeder.12) by applying the product rule (1.13) due to the fact that it is not possible anymore to push the density out of the derivative. (2. (2. Thus.1.13) For the simple mass conservation equation it is obvious that the vector notation results in the Cartesian form. we end up with: Compressible: Z I ∂ ρdV = − ρU · ndS .the discrete volumes are not changing during the simulation.18) to the divergence term: ∇ • (ρU) = U • ∇ρ + ρ∇ • U . tetraeder. If the density change is really small. The reason for that is based on numerics and the interaction with the momentum conservation equation. we have to evaluate more or less faces. the integral form of the mass conservation equation will be given now. (2. static meshes). For the incompressible case.21).14) ∂t Incompressible: I U · ndS = 0 . we can transform the divergence term (that acts on the volume) to a surface integral. (2. The integral form of the continuity equation leads to the so called finite volume method (FVM). the density for the fluid is constant and thus we can simplify the mass conservation equation to: ∇•U=0 . unstructured grids limit the numerical schemes to second order precision in general. Thats why the transformation is not demonstrated here. Using the Gauss theorem (1.12 Derivations of the Governing Equations have to be careful which mass conservation equation we are using (incompressible or compressible).16).1. wedges and so on) which makes this method popular and flexible. 2. However.2 Continuity Equation and the Total Derivative Using the total derivative formulation (1. if the density is not a constant value. we are able to rewrite the continuity equation (2. The accumulation of density in the element is a simple volume integral. what is going in and out. If you want to check it ourself. prisms. you just need to use equation (1.15) The surface integral means nothing more than taking the balance of the fluxes on the surfaces of the volume element. we are not allowed to use the simplified mass conservation equation (2. 2.16) .1 Integral Form of the Conserved Continuity Equation For the completeness. we are allowed to use the simplified mass conservation equation with limitations.28). the volume element itself is not changing with respect to time (fixed finite volume . In general. Depending on the shape of the volume. Furthermore.

(2. To get the fluxes. THE CONTINUITY EQUATION 13 Substituting this expression into equation (2.18) |∂t {z } Total derivative The result is: Dρ + ρ∇ • U = 0 . The flux field is named phi and is created by the call of one of the two header files in each solver: • createPhi. This calculation is done by calling the function interpolate(rho*U) and will simply calculate the product of the density and the velocity vector at the cell center and interpolate it to the face center by including the neighbor cell information with respect to the face we are going to evaluate. (2. we get: ∂ρ = −U • ∇ρ − ρ∇ • U .1.H • compressibleCreatePhi. .12).19) Dt This equation is not common but could be found in Anderson [1995].17) ∂t Finally.2. we put all terms to the LHS: ∂ρ + U • ∇ρ +ρ∇ • U = 0 . (2.H Due to the fact that we store the density and the velocity at the cell center. we need to interpolate the values to the face centers. OpenFOAMr In OpenFOAMr we are using this equation (integral one) to calculate the fluxes at the faces of each cell. the interpolated values are then multiplied by the surface normal vector (area) using the inner product of two vectors denoted by the ampersand sign & in OpenFOAMr .

the terms that transport the x-component of the momentum through the surfaces by the molecular transport effect are given. . that we have to consider more phenomena that can transport and change the momen- tum inside the volume element and that this quantity is not a scalar.1 but now showing another transport phenomenon that acts only on the surfaces. we are going to use the volume element dV again. Convection of the Momentum in x-Direction The x-component of the momentum is transported by convection into the volume element through all six faces that is also an outcome related to the vector quantity. For example we could have the gravitational acceleration and the pressure force. The main difference in the momentum equation compared to the mass conservation equation is. Generally we are allowed to say that the momentum can be transported and changed by the following aspects:         rate of rate of  rate of      sum of forces  momentum  =  momentum  − momentum +  that act on  .14 Derivations of the Governing Equations 2.2. Therefore. Other phenomena that change the momentum are given as a sum of forces acting on the volume element dV .20) accumulation the volume         volume volume Figure 2.           entering the  leaving the    (2. it is a vector (velocity in x. On the right side of figure 2.2 The Conserved Momentum Equation For the derivation of the conserved momentum equation. the convection of z τzx |z+∆z into face |x : τxx |x ∆y∆z ∆x out of face |x : τxx |x+∆x ∆y∆z ∆z τyx |y+∆y into face |y : τyx |y ∆x∆z τxx |x τxx |x+∆x out of face |y : τyx |y+∆y ∆x∆z τyx |y y ∆y into face |z : τzx |z ∆x∆y τzx |z out of face |z : τzx |z+∆z ∆x∆y x Figure 2. This molecular transport acts normal and tangential to the surface and is an outcome or property of the vector quantity.2: Molecular transport of the momentum in x-direction in an arbitrary small volume element dV .2 shows the volume element like in 2. The phenomenon transports the momentum based on molecular effects. y and z direction).

We consider these fluxes as stresses. τxx denotes the stress perpendicular to the direction we are looking at (here face |x and face |x+∆x ) and τyx .2. After combining the terms and using the face areas.2. identical to the continuity equation.2. These terms represent additional fluxes of momentum through the surface. into face |z : (ρuz )ux |z . Hence. Molecular Transport of the Momentum in x-Direction Additionally. out of face |x+∆x : (ρux )ux |x+∆x .2. But now we have to take care about the vector quantity. it is simply the velocity in x-direction multiplied by the flux through the face we are looking at (Newtons second law): into face |x : (ρux )ux |x . the x-component of the momentum is transported due to the molecular phenomenon as demonstrated in figure 2. we have different kind of terms: the normal component τxx and the tangential components τyx . But now it is also possible that the x-component of the momentum is transported through the faces in y and z direction. Therefore. The effect is based on velocity differences (velocity gradients). THE CONSERVED MOMENTUM EQUATION 15 the momentum can be derived similarly to the convective transport of the mass. All these stresses are known as shear stresses due to the fact that they are generated with respect to velocity gradients that introduce shearing. the only important forces that influence the momentum are the pressure and gravity force. we get:  (ρux )ux |x − (ρux )ux |x+∆x ∆y∆z  + (ρuy )ux |y − (ρuy )ux |y+∆y ∆x∆z  + (ρuz )ux |z − (ρuz )ux |z+∆z ∆x∆y . Thus. Additional Forces that Influence the Momentum In most problems. τzx . into face |y : (ρuy )ux |y . Hence. The pressure acts on the surface whereat the gravitational force acts on the volume of the element. we are able to derive the change of the x-momentum based on the pressure . out of face |y+∆y : (ρuy )ux |y+∆y . As we can see in figure 2. the molecular transport of the x-momentum through the surfaces can be written as:  τxx |x − τxx |x+∆x ∆y∆z  + τyx |y − τyx |y+∆y ∆x∆z  + τzx |z − τzx |z+∆z ∆x∆y . the momentum in x-direction enters the volume at the face |x and leaves the volume through the face |x+∆x . τzx denote the x-directed tangential stresses which act on the faces with respect to the indices. we can write the transport of the momentum due to convection. out of face |z+∆z : (ρuz )ux |z+∆z .

(2. we get:   ∂ ∂ ∂ ∂ ρuy = − ρux uy + ρuy uy + ρuz uy ∂t ∂x ∂y ∂z   . (2.16 Derivations of the Governing Equations and gravitational force:  p|x − p|x+∆x ∆y∆z + ρgx ∆x∆y∆z .21) Now. The other two space directions can be derived in the same way and is not shown in details. Conserved Momentum Equation After we have all terms.20) with the mathematic expressions. The x-component of the momentum is then written as:   ∂ ∂ ∂ ∂ ρux = − ρux ux + ρuy ux + ρuz ux ∂t ∂x ∂y ∂z   . by dividing the whole equation by the volume dV . (2.6) to rewrite the x-component of the momentum equation above. for the momentum in x-direction we can write: ∆  ρux ∆x∆y∆z = (ρux )ux |x − (ρux )ux |x+∆x ∆y∆z ∆t  + (ρuy )ux |y − (ρuy )ux |y+∆y ∆x∆z  + (ρuz )ux |z − (ρuz )ux |z+∆z ∆x∆y  + τxx |x − τxx |x+∆x ∆y∆z  + τyx |y − τyx |y+∆y ∆x∆z  + τzx |z − τzx |z+∆z ∆x∆y  + p|x − p|x+∆x ∆y∆z +ρgx ∆x∆y∆z .5) and time range (2. we can reconstruct equation (2. it follows: ∆ (ρux )ux |x − (ρux )ux |x+∆x (ρuy )ux |y − (ρuy )ux |y+∆y ρux = + ∆t ∆x ∆y (ρuz )ux |z − (ρuz )ux |z+∆z τxx |x − τxx |x+∆x τyx |y − τyx |y+∆y + + + ∆z ∆x ∆y τzx |z − τzx |z+∆z p|x − p|x+∆x + + + ρgx .24) ∂ ∂ ∂ ∂p − τxy + τyy + τzy − + ρgy ∂x ∂y ∂z ∂y .23) ∂ ∂ ∂ ∂p − τxx + τyx + τzx − + ρgx ∂x ∂y ∂z ∂x For the y-component of the momentum. ∆t Thus. Of course the accumulation of the momentum inside an arbitrary volume element is given by: ∆ ρux ∆x∆y∆z . (2.22) ∆z ∆x Finally we use the assumption of an infinitesimal small volume element (2.

. we get:  ∂ ρux of x − momentum ∂  ρux   ∂t ∂t  ∂  ∂  ! ∂ ρU =  ∂t ρuy  = ∂t ρuy of y − momentum .26) ∂t 2. g = gy  .25). it can be seen that the terms are equal and we end up with the same set of equations. if we introduce the definition of the shear-rate tensor τ later on.2. To evaluate the term. THE CONSERVED MOMENTUM EQUATION 17 and for the z-component we achieve:   ∂ ∂ ∂ ∂ ρuz = − ρux uz + ρuy uz + ρuz uz ∂t ∂x ∂y ∂z   . the time derivative term results in the same three terms that we have in the Cartesian formulation. Note. (2. (2. ∂ ρU = −∇ • (ρU ⊗ U) − ∇ • τ − ∇p + ρg (2. that the negative sign of the shear-rate tensor will change. For clearance.     ∂p ∂z τzx τzy τzz gz we are able to write the momentum equation in vector form. (2.11) and (1. The second term embrace the transport of momentum due to convection by the flux ρU.27) ∂t ∂  ρu  ∂ z of z − momentum  ∂t  ρuz ∂t As we see. Starting with the first term. τ = τyx τyy τyz  . the gradient of the pressure ∇p and the shear- rate tensor τ that are defined as  ∂p      ∂x τxx τxy τxz gx  ∂p  ∇p =  ∂y  .25) ∂ ∂ ∂ ∂p − τxz + τyz + τzz − + ρgz ∂x ∂y ∂z ∂z Introducing the gravitational acceleration vector g.2.23). we need the mathematics (1.2. we will focus on each term separately. ∂ ∂ ∂  ρux uz + ρuy uz + ρuz uz  ∂  ∂ ∂  ∂x ∂y ∂z − ∂x ρux uz + ∂y ρuy uz + ∂z ρuz uz Again. the time derivative.24) and (2.1 The Proof of the Transformation The following section will proof that equation (2.17):           ux ux   ux ux  ux uy ux uz   −∇ • (ρU ⊗ U) = −∇ • ρ uy  ⊗ uy  = −∇ • ρ uy ux uy uy uy uz            uz uz uz ux uz uy uz uz       ∂    ρux ux ρux uy ρux uz ∂x ρux ux ρux uy ρux uz ∂   = −∇ • ρuy ux ρuy uy ρuy uz  = −  ∂y  • ρuy ux ρuy uy ρuy uz     ∂ ρuz ux ρuz uy ρuz uz ∂z ρuz ux ρuz uy ρuz uz     ∂ ∂ ∂  − ∂ ρu u + ∂ ρu u + ∂ ρu u ρux ux + ρuy ux + ρuz ux x x y x z x   ∂x ∂y ∂z  ∂x ∂y ∂z   ∂ ∂ ∂  ! ∂ ∂ ∂ = −  ∂x ρux uy + ∂y ρuy uy + ∂z ρu z u y  = − ∂x ρu x u y + ∂y ρu y u y + ∂z ρuz uy . Now we investigate into the third term that describes shearing due to the gradients of the velocities.26) results in (2.

∂ ∂ ∂    τ + τ + τ  ∂x xz ∂y yz ∂z zz − ∂ τ +  ∂ τ + ∂ τ of z momentum ∂x xz ∂y yz ∂z zz It was already clear. For that. we get:   τxx τxy τxz −∇ • τ = −∇ • τyx τyy τyz  =   τzx τzy τzz    ∂ τ + ∂ τ + ∂ τ   − ∂ τxx + ∂ τ ∂y yx + ∂ τ ∂z zx of x momentum ∂x xx ∂y yx ∂z zx   ∂x    ∂ ∂ ∂  ! ∂ ∂ ∂ −  ∂x τxy + τ ∂y yy + τ ∂z zy  = − ∂x τxy + τ ∂y yy + τ ∂z zy of y momentum . the vector form ends up in the Cartesian form.2 Integral Form of the Conserved Momentum Equation The integral form of the momentum equation (2.12) to get a non-conservative form.     gz ρgz   ρgz of z momentum As we proofed now.17) for the shear-rate tensor τ . If there are further phenomena (forces) influencing the momentum equation. ∂ ∂p    ∂p ∂z ∂z  − ∂z of z momentum  ρg of x momentum     ρgx gx  x    ! ρg = ρ gy  = ρgy  = ρgy of y momentum . especially the treatment of the diffusion term.2. we consider equation (2. (2.13) and for the gravitational term equation (1. It follows:   ∂p   ∂p − ∂x of x momentum ∂   ∂x ∂x  ∂   ∂p  ! ∂p −∇p = −  ∂y  p = −  ∂y  = − ∂y of y momentum . that we end up with the same terms.18 Derivations of the Governing Equations Using the convention (1.3 Non-Conserved Momentum Equation We can manipulate the conserved momentum equation with the continuity equation (2. Keep in mind that equation (2. Further representations of the momen- tum equation can be found in chapter 8.26) first and split the time and .28) ∂t 2. The implementation of the momentum equation in OpenFOAMr will be discussed in chapter 10.6). it is necessary to know the shear-rate tensor τ . It follows: Z I I I Z ∂ ρUdV = − (ρU ⊗ U) · ndS − τ · ndS − pI · ndS + ρgdV .26) includes only the gravitational acceleration and pressure force. We are going to investigate into that quantity in chapter 5. If we want to solve this equation now. At last the pressure and gravitational acceleration term is analyzed.28). 2.26) can be obtained by using the Gauss theorem (1. For the pressure term we need the definition of equation (1. these terms have to be taken into account.2.

32) ∂t | ∂t   {z } continuity It is clear that the second term is zero due to the continuity equation.31) ∂t ∂t After analyzing the equation. . Applying the definition of the total derivative (1.2.33) Dt Remark: As already mentioned before. THE CONSERVED MOMENTUM EQUATION 19 convection term by using the product rule.21). (2.19). .29) ∂t ∂t ∂t and the convection term can be rewritten using equation (1. (2. It follows: ∇ • (ρU ⊗ U) = ρU • ∇ ⊗ U +U ∇ • (ρU) . we end up with: ∂ ∂ ρ U + U ρ + ρU • ∇ ⊗ U + U∇ • (ρU) = . we can write the non-conservative form of the momentum equation as: DU ρ = −∇ • τ − ∇p + ρg (2. (2. . the negative sign of the first term on the RHS will vanish after we introduced the definition of the shear-rate components τii . we see that we can take out ρ and U:   ∂ h∂ i ρ U+U•∇⊗U +U ρ+  ∇ •(ρU) = .30) | {z } | {z } gradient divergence | {z } inner product Replacing these terms into equation (2. . The time derivative becomes: ∂ ∂ ∂ ρU = ρ U + U ρ . .26) and put the convection terms to the LHS. (2.2.

[1960]. kinetic and work energy are included and therefore unsteady behavior is allowed. the change of the total energy can be described in an arbitrary volume element dV by:       rate of internal rate of internal rate of internal and kinetic  =  and kinetic energy  −  and kinetic energy         energy accumulation entering the volume leaving the volume       (2. we assume that energy cannot be transfered into mass and vice versa. e. The total energy includes the internal (thermal) and kinetic (mechanical) energy. .3 The Conserved Total Energy Equation This section will show the derivation of the total energy equation. ∆y∆zux ρ. radiative and electro- magnetic phenomena but as we already mentioned.g. In the equation above. internal. The internal energy e (per unit mass) can be interpreted as the energy associated with the random translation and internal motion of molecules plus the energy of interaction between them. the internal energy is temperature and density depended.34) net rate of net rate of net rate of + heat addition by −  work done by  +  additional  . The kinetic energy (per unit mass) is given by 1 2 ρ|U|2 where |U| denotes the magnitude of the local velocity.36) 2 2 If we take out the density. Therefore. The Accumulation of Total Energy during Time Now we write the above equation explicit for a finite volume element dV . (2. In general. The statement is not complete because no transport of energy can be done due to nuclear. (2. we simply have to multiply the internal and kinetic energy by the velocity respectively to the face it acts on (compare figure 2. we clearly see the mass flux. The accumulation in time is clear (like in the other equations before): ∂ 1 ∆x∆y∆z (ρe + ρ|U|2 ) .1):   1 1 ∆y∆z ux (ρe + ρ|U|2 )|x − ux (ρe + ρ|U|2 )|x+∆x 2 2   1 1 +∆x∆z uy (ρe + ρ|U|2 )|y − uy (ρe + ρ|U|2 )|y+∆y 2 2   1 1 +∆x∆y uz (ρe + ρ|U|2 )|z − uz (ρe + ρ|U|2 )|z+∆z .20 Derivations of the Governing Equations 2.35) ∂t 2 The Convection of Total Energy To get the net rate of total energy – that enters and leaves the volume element based on the convection phenomenon –.       conduction system on surroundings heat sources This is the first law of thermodynamics written for an open and unsteady state system with the extension of additional heat sources which was also stated by Bird et al.

(2. the rate of doing work against the components of the gravitational acceleration can be written as: − ρ∆x∆y∆z (ux gx + uy gy + uz gz ) .38) and the rate of doing work against the pressure p (static pressure) at the faces of the volume element is: ∆y∆z [−(pux )|x + (pux )|x+∆x ] +∆x∆z [−(puy )|y + (puy )|y+∆y ] +∆x∆y [−(puz )|z + (puz )|z+∆z ] . For this we introduce the heat flux vector q. qy and qz are the single components of the heat flux vector q.37) The quantities qx . The subscript S stands for Source or Sink. • (rate of doing work) = (Force) x (Velocity in the direction of the force). Hence.39) In a similar way.40) Additional Heat Source Additional heat sources or sinks can be taken into account by simply defining a source term: QS = ∆x∆y∆zρS . (2. that is used later on: ∆y∆z [qx |x − qx |x+∆x ] +∆x∆z [qy |y − qy |y+∆y ] +∆x∆y [qz |z − qz |z+∆z ] . The Change of Total Energy due to Work against its Surroundings The work done by the fluid against its surroundings can be split into two parts: • The work against the volume forces (like gravity). Some recall: • (Work) = (Force) x (Distance in the direction of the force). the rate of doing work against the viscous forces is: ∆y∆z [−(τxx ux + τxy uy + τxz uz )|x + (τxx ux + τxy uy + τxz uz )|x+∆x ] +∆x∆z [−(τyx ux + τyy uy + τyz uz )|y + (τyx ux + τyy uy + τyz uz )|y+∆y ] +∆x∆y [−(τzx ux + τzy uy + τzz uz )|z + (τzx ux + τzy uy + τzz uz )|z+∆z ] .41) Here QS denotes the heat source term. THE CONSERVED TOTAL ENERGY EQUATION 21 The Change of Total Energy due to Conduction The next term that influences the energy is based on the conduction phenomenon. (2.2.3. (2. • The work against the surface forces (like pressure or viscous forces). (2. .

we get: ∂ 1 (ρe + ρ|U|2 ) = ∂t  2      ∂ 1 ∂ 1 ∂ 1 − ux (ρe + ρ|U|2 ) − uy (ρe + ρ|U|2 ) − uz (ρe + ρ|U|2 ) ∂x 2 ∂y 2 ∂z 2 ∂ ∂ ∂ − [qx ] − [qy ] − [qz ] + ρ (ux gx + uy gy + uz gz ) ∂x ∂y ∂z ∂ ∂ ∂ ∂ − [pux ] − [puy ] − [puz ] + [−(τxx ux + τxy uy + τxz uz )] ∂x ∂y ∂z ∂x ∂ ∂ + [−(τyx ux + τyy uy + τyz uz )] + [−(τzx ux + τzy uy + τzz uz )] ∂y ∂z +ρS .43) After taking out the minus signs and sort the equation. Please keep in mind.22 Derivations of the Governing Equations Now we have all terms and are able to rewrite equation (2. (2. that we have derivatives of (φ|x −φ|x+∆x ) that end up with a negative sign and (φ|x+∆x −φ|x ) that end up with a positive sign.5).22):   ∂ 1 1 1 ∆x∆y∆z (ρe + ρ|U|2 ) = ∆y∆z ux (ρe + ρ|U|2 )|x − ux (ρe + ρ|U|2 )|x+∆x ∂t 2 2 2   1 1 +∆x∆z uy (ρe + ρ|U|2 )|y − uy (ρe + ρ|U|2 )|y+∆y 2 2   1 1 +∆x∆y uz (ρe + ρ|U|2 )|z − uz (ρe + ρ|U|2 )|z+∆z 2 2 +∆y∆z [qx |x − qx |x+∆x ] + ∆x∆z [qy |y − qy |y+∆y ] +∆x∆y [qz |z − qz |z+∆z ] +ρ∆x∆y∆z (ux gx + uy gy + uz gz ) −∆y∆z [−(pux )|x + (pux )|x+∆x ] −∆x∆z [−(puy )|y + (puy )|y+∆y ] −∆x∆y [−(puz )|z + (puz )|z+∆z ] −∆y∆z [−(τxx ux + τxy uy + τxz uz )|x + (τxx ux + τxy uy + τxz uz )|x+∆x ] −∆x∆z [−(τyx ux + τyy uy + τyz uz )|y + (τyx ux + τyy uy + τyz uz )|y+∆y ] −∆x∆y [−(τzx ux + τzy uy + τzz uz )|z + (τzx ux + τzy uy + τzz uz )|z+∆z ] +∆x∆y∆zρS . we get the conserved total energy equation . (2.42) As before. we divide everything by the volume dV and use the assumption (2. Hence.

This will not be discussed here.45) − ∇ • (pU) − ∇ • [τ • U] + ρS | {z } | {z } | {z } pressure viscous forces heat source All terms on the RHS denote the inner product of two vectors.45) will be used and transformed into (2.8) and (1. the potential energy is not of interest or is in a neglect-able range. If you need information about that. (2. equation (2. THE CONSERVED TOTAL ENERGY EQUATION 23 as:      ∂ 1 ∂ 1 ∂ 1 (ρe + ρ|U|2 ) = − ux (ρe + ρ|U|2 ) + uy (ρe + ρ|U|2 ) ∂t 2 ∂x 2 ∂y 2     ∂ 1 ∂ ∂ ∂ + uz (ρe + ρ|U|2 ) − qx + qy + qz + ρ (ux gx + uy gy + uz gz ) ∂z 2 ∂x ∂y ∂z    . 314.1 The Proof of the Vector Transformation The next sites will convert the vector form of the total energy equation back into the Cartesian coordinate system.3.8). The total energy equation then can be written in the following form and the terms can be described more precisely:   ∂ 1 1 (ρe + ρ|U|2 ) = −∇ • ρU(e + |U|2 ) −∇•q + ρ (U • g) ∂t 2 2 | {z } | {z } | {z } convection conduction gravity . The reason why we do not mention this.7). It follows that the gravity term can be changed as:      ux gx      ! ρ(U • g) = ρ uy  • gy  = ρ (ux gx + uy gy + uz gz ) . This is done step by step but not for each term. For the viscous force term we need equation (1.3. we use the vector notation. To manipulate the gravity term we need the mathematic law of equation (1. for a implementation into a software toolbox we need to use equation (1.44) ∂ ∂ ∂ ∂ − pux + puy + puz − (τxx ux + τxy uy + τxz uz ) ∂x ∂y ∂z ∂x  ∂ ∂ + (τyx ux + τyy uy + τyz uz ) + (τzx ux + τzy uy + τzz uz ) + ρS ∂y ∂z To get a more visible and readable equation. Remark: Till now no word is said about the potential energy. is the fact that in most of the engineering cases. Hence. Hence. you will find all necessary information in Bird et al. The terms of interest are the gravity and viscous force term.46) uz gz .17). (2.44). (2. 2.2.7) and (1. [1960] p.

2. Hence. the mass conservation equation can be used to manipulate the conserved total energy equation. we will not demonstrate it here again.48) + ρU • gdV − pU · ndS − (τ • U) · ndS + ρSdV 2. It follows: Z I I ∂ 1 1 ρ(e + |U|2 )dV = − ρU(e + |U|2 ) · ndS − q · ndS ∂t 2 2 Z I I Z . (2. .51) ∂t 2 2 ∂t 2 2 .24 Derivations of the Governing Equations The viscous force term can be changed as follows:     τxx τxy τxz ux − ∇ • [τ • U] = −∇ • τyx τyy τyz  • uy      τzx τzy τzz uz    ∂    τxx ux + τxy uy + τxz uz ∂x τxx ux + τxy uy + τxz uz ∂  = −∇ • τyx ux + τyy uy + τyz uz  = −  ∂y  • τyx ux + τyy uy + τyz uz      ∂ τzx ux + τzy uy + τzz uz τzx ux + τzy uy + τzz uz ∂z  ! ∂ ∂ =− (τxx ux + τxy uy + τxz uz ) + (τyx ux + τyy uy + τyz uz ) ∂x ∂y  ∂ + (τzx ux + τzy uy + τzz uz ) . The transformation lead to the non-conservative form. the terms are equal of both notations.3. the time derivative can be rewritten as:   ∂ 1 ∂ 1 ∂ 1 1 ∂ (ρe + ρ|U|2 ) = ρ(e + |U|2 ) = ρ (e + |U|2 ) + (e + |U|2 ) ρ .3 Non-Conserved Total Energy Equation As before.49) ∂t 2 ∂t 2 ∂t 2 2 ∂t and the convective term will be transformed to: 1 1 1 ∇ • ρU(e + |U|2 ) = ρU • ∇(e + |U|2 ) +(e + |U|2 ) ∇ • (ρU) .54). The other terms that are not discussed above are similar to the momentum equation and it is easy to demonstrate that each term of the vector notation represents the corresponding terms in the Cartesian equation. Due to that. we first need to break the time and convective term of equation (2.45) by using the product rule. (2. (2.3. For that.45) and put the terms that are related to the convective part to the LHS: ∂ 1 1 ∂ 1 1 ρ (e + |U|2 ) + (e + |U|2 ) ρ + ρU • ∇(e + |U|2 ) + (e + |U|2 )∇ • (ρU) = .47) ∂z As we may already might had the feeling.50) 2 | 2 {z } 2 | {z } gradient divergence | {z } inner product The next step is replacing the above terms into equation (2. (2. (2. we use the Gauss theorem.2 Integral Form of the Conserved Total Energy Equation To obtain the integral form of the total energy equation (2. .

After we have either the kinetic or internal energy equation. (2. (2.53) ∂t 2 2 By using the definition of the total derivative (1. we can simply subtract that equation from the total energy equation to get the other one. .54) Dt 2 2. we get:   ∂ 1 1 1 h∂ i ρ (e + |U| ) + U • ∇(e + |U| ) + (e + |U|2 ) 2 2 ρ+  ∇ •(ρU) = .3. (2. we can simplify the equation and end up with:   ∂ 1 1 ρ (e + |U|2 ) + U • ∇(e + |U|2 ) = . In your case we will derive the mechanical (kinetic) energy equation first and get the internal energy equation by subtracting the kinetic energy from the total energy equation. . Based on that. The answer why we derive the mechanical (kinetic) energy equation instead of the internal energy equation is simple. The derivation of the kinetic energy equation can be done simply by multiplying the momentum equation by the velocity again.3.52) ∂t 2 2 2 | ∂t   {z } continuity We can observe that the second term on the LHS is equal to zero due to the continuity equation. we finally are able to rewrite the equation in the non-conservative form: D 1 ρ (e + |U|2 ) = −∇ • q + ρU • g − ∇ • pU − ∇ • [τ • U] + ρS .2.21). THE CONSERVED TOTAL ENERGY EQUATION 25 After taking out the density and the term (e + 12 |U|2 ). . .4 Kinetic Energy and Internal Energy The total energy equation is the sum of the kinetic energy and internal energy. .

2 1 into face |z : (ρuz ) |U|2 |z . At that stage it is easier to analyze the meaning of the different terms. To be consistent with the derivations before.57) ∆t ∆t 2 The Convection of Kinetic Energy The kinetic energy that enters or leaves the volume element is transported due to fluxes and can be derived analogous to the convective transport of the total energy or momentum: 1 into face |x : (ρux ) |U|2 |x . 2 1 into face |y : (ρuy ) |U|2 |y .4 The Conserved Mechanical Energy Equation As mentioned before. In general we can define the change of kinetic energy in an arbitrary volume element dV as: " # " # " # rate of kinetic rate of kinetic energy rate of kinetic energy = − energy accumulation entering the volume leaving the volume " # (2. 2 1 out of face |x+∆x : (ρux ) |U|2 |x+∆x . (2.55) sum of additional + source terms Using the definition of the kinetic energy per unit mass (divided by ρ): 1 ekin = |U|2 . 2 1 out of face |z+∆z : (ρuz ) |U|2 |z+∆z . (2. 2 . we will derive the mechanical (kinetic) energy equation using the momentum equation.56) 2 we can simply derive the accumulation of the kinetic energy as: ∆ ∆ 1 ∆x∆y∆z ρekin = ∆x∆y∆z ρ |U|2 . After that we use the source terms of the momentum equation to get the source terms for the mechanical (kinetic) energy equation. The first part will use the finite volume element dV to derive the first part of the conserved kinetic energy equation without explaining the meaning of the source terms. 2 1 out of face |y+∆y : (ρuy ) |U|2 |y+∆y .26 Derivations of the Governing Equations 2. we will split the derivation into two parts.

(2.4.61) ∂t 2 2 Equation (2. Source Terms of the Kinetic Energy The kinetic energy can be changed by several phenomena.5) and (2. 2 2 Defining the sum of additional source terms that act on the volume by the quantity eS and put the new evaluated terms into equation (2. it is possible to rewrite the transport of the kinetic energy due to convection as:   1 1 (ρux ) |U|2 |x − (ρux ) |U|2 |x+∆x ∆y∆z 2 2   1 1 + (ρuy ) |U|2 |y − (ρuy ) |U|2 |y+∆y ∆x∆z 2 2   1 1 + (ρuz ) |U|2 |z − (ρuz ) |U|2 |z+∆z ∆x∆y .2.59) ∆z The next step is to use the assumption of equation (2. Thus.60) ∂t 2 ∂x 2 ∂y 2 ∂z 2 The vector notation for this equation is: ∂ 1 1 ρ |U|2 = −∇ • (ρU |U|2 ) + eS .61) can also be derived using the momentum equation and forming the scalar product of the local velocity and equation (2. (2.58) To get to a partial differential equation. we get:   ∂ 1 1 1 ∆x∆y∆z ρ |U|2 = (ρux ) |U|2 |x − (ρux ) |U|2 |x+∆x ∆y∆z ∂t 2 2 2   1 1 + (ρuy ) |U| |y − (ρuy ) |U|2 |y+∆y ∆x∆z 2 2 2   1 1 + (ρuz ) |U|2 |z − (ρuz ) |U|2 |z+∆z ∆x∆y 2 2 +eS ∆x∆y∆z . THE CONSERVED MECHANICAL ENERGY EQUATION 27 With the expressions above. Therefore.26). To get the different source terms we will use the momentum equation (2. we will divide the whole equation by the volume of the element dV . These sources (terms) act on the volume element. (2. we get the first part of the kinetic energy equation without any explicit mentioned source term: ∂ 1 ∂ 1 ∂ 1 ∂ 1 ρ |U|2 = − (ρux |U|2 ) − (ρuy |U|2 ) − (ρuz |U|2 ) + eS .6).26). (2.55). we get: ∂ 1 (ρux ) 12 |U|2 |x − (ρux ) 21 |U|2 |x+∆x ρ |U|2 = ∂t 2 ∆x (ρuy ) 12 |U|2 |y − (ρuy ) 21 |U|2 |y+∆y + ∆y (ρuz ) 21 |U|2 |z − (ρuz ) 12 |U|2 |z+∆z + + eS . The terms .

(2.66) U • ∇p = ∇ • (pU) − p∇ • U .65) |{z} gradient divergence | {z } inner product Now we rearrange these equations: U • (∇ • τ ) = ∇ • [τ • U] − τ : (∇ ⊗ U) . • Shear-rate (surface) .28 Derivations of the Governing Equations eS are simply the source terms in the momentum equation (2. we will replace these terms by manipulating both by applying the product rule. we can replace the quantity eS by: eS = −(∇ • τ ) • U − (∇p) • U + (ρg) • U . If we insert the source term eS into equation (2. ∂ 1 1 |U|2 = −∇ • (ρU |U|2 ) − (∇ • τ ) • U − (∇p) • U + (ρg) • U . • Gravity (volume) . Using this information.61). Hence. The term that denotes the viscous force.26) multiplied by the velocity. we already know the meaning of the term on the LHS and the the first and last term on the RHS. (2.62) | {z } work done by gravity Note: The multiplication with the velocity results in the inner product.64) | {z } | {z } gradient divergence | {z } | {z } double inner product inner product The pressure term can be manipulated like: ∇ • (pU) = U • ∇p +p ∇ • U} | {z . (2. (2. we get for the source terms the following .62). (2. Therefore. The resulting equation gives us the possibility to get a better physical base for the meaning of the single terms.67) and insert both into equation (2.63) ∂t 2 2 Analyzing the new equation. Recall: The source terms of the equation of motion are: • Pressure (surface) . we get the conserved mechanical (kinetic) energy equation with the source terms. can be rewritten as: ∇ • [τ • U] = τ : (∇ ⊗ U) + U • (∇ • τ ) . Thinking about the second and third term on the RHS is not so clear till now. (2.

28) as before. This term will heat up the fluid internally.2. (2. • p(−∇ • U): this term will cool or heat the fluid internally due to sudden expansion or compression phenomena.g.71) ∂t 2 ∂t 2 2 ∂t and the convection term to:   1 1 1 ∇ • ρU |U|2 = ρU • ∇ |U|2 + |U|2 ∇ • (ρU) .4. to change the conserved kinetic energy equation into the non-conservative form.72) 2 2 2 .4. Therefore. (2. e.2 Non-Conserved Mechanical Energy Equation As before. The heating due to this term will only be measurable if the speed of the fluid is very high (large velocity gradients). Hence. we use the new evaluated sources eS to get a more understandable and readable kinetic energy equation: ∂ 1 1 ρ|U|2 = −∇ • (ρU |U|2 ) − ∇ • [τ • U] − (−τ : (∇ ⊗ U)) ∂t 2 2 .69) −∇ • (pU) − p(−∇ • U) + (ρg) • U The Meaning of Some Terms • (−τ : ∇U): as stated by Bird et al.68) − p(−∇ • U) + (ρg) • U | {z } | {z } reversible conversion work done to internal energy by gravity Finally. turbines or shock-tubes.70) − (−τ : (∇ ⊗ U))dV − pU · ndS − p(−∇ • U)dV + (ρg) • UdV 2. (2. In addition we have to put the convective term to the LHS of equation (2.g. the time derivative will change to: ∂ 1 ∂ 1 1 ∂ ρ |U|2 = ρ |U|2 + |U|2 ρ . high-speed flight or rapid extrusion.69) can be achieved by using the Gauss theorem (1. we are able to use the continuity equation. we get: Z I I ∂ 1 1 ρ|U|2 dV = − ρU |U|2 · ndS − [τ • U] · ndS ∂t 2 2 Z I Z Z . e.4. (2. [1960]. we have to split the time derivative and convection term using the product rule again. this term is always positive for Newtonian fluids and describes that motion energy is irreversibly exchanged into thermal energy and therefore no real processes are reversible.1 Integral Form of the Conserved Mechanical Energy Equation The integral form of the kinetic energy equation (2. (2. 2. THE CONSERVED MECHANICAL ENERGY EQUATION 29 expression: eS = −∇ • [τ • U] − (−τ : (∇ ⊗ U)) −∇ • (pU) | {z } | {z } | {z } work done irreversibe conversion work done by by viscous force to internal energy pressure of sourroundings (shear−heating) . Thus.69).

74) ∂t 2 2 2 | ∂t   {z } continuity As we can see. (2. . (2. . . (2. . . the second term on the LHS is zero due to continuity and therefore we get the non-conserved kinetic energy equation by using the definition of the total derivative (1.21): D 21 |U|2 ρ = −∇ • [τ • U] − (−τ : (∇ ⊗ U)) − ∇ • (pU) − p(−∇ • U) + (ρg) • U .75) Dt Remark: It should be obvious that we can use equation (2.30 Derivations of the Governing Equations We end up with: ∂ 1 1 ∂ 1 1 ρ |U|2 + |U|2 ρ + ρU • ∇ |U|2 + |U|2 ∇ • (ρU) = .64) and (2.73) ∂t 2 2 ∂t 2 2 After we exclude ρ and 1 2 |U|2 from the equations.65) to change/eliminate some terms again. we get:   ∂ 1 1 1 h∂ i ρ |U| + U • ∇ |U| + |U|2 2 2 ∇ ρ+ •(ρU) = .

(2. (2. . (2. we get the following equation for the kinetic (internal) energy equation: ∂ (ρe) − (−τ : (∇ ⊗ U)) − p(−∇ • U) = −∇ • (ρUe) − ∇ • q + ρS .5. fourth. fifth and sixth term on the RHS.   ∂ 1 2 (ρe) − ∇ • ρU |U| − ∇ • [τ • U] − (−τ : (∇ ⊗ U)) − ∇ • (pU) ∂t 2   1 − p(−∇ • U) + (ρg) • U = −∇ • (ρUe) − ∇ • ρU |U|2 2 − ∇ • q + ρU • g − ∇ • pU − ∇ • [τ • U] + ρS . . THE CONSERVED THERMO ENERGY EQUATION 31 2. we can simply get the equation for the thermo (internal) energy equation by subtracting equation (2.69).45). we get:     ∂ 1 1 ρe + ρ|U|2 = −∇ • ρU(e + |U|2 ) + . .78) We can see that the second. (2. third.45) first.77) ∂t ∂t 2 2 The next step is to replace the underlined term by the conserved kinetic energy equation (2.69) from (2. Hence. To get the internal energy equation.80) ∂t | {z } | {z } | {z } | {z } |{z} transport by irreversible reversible energy input heat source convection energy by energy by by conduction viscous dissipation compression shear−heating .5 The Conserved Thermo Energy Equation After we have the total energy and the kinetic energy equation.76) ∂t 2 2     ∂ ∂ 1 1 (ρe) + ρ|U|2 = −∇ • (ρUe) − ∇ • ρU |U|2 + . Note that ρU • g = (ρg) • U. we will split the time and convection term of equation (2. Hence. fifth and seventh term on the LHS cancel with the second.2.79) ∂t After sorting the equation we get the final form: ∂ (ρe) = −∇ • (ρUe) −(τ : (∇ ⊗ U)) −p(∇ • U) −∇ • q +ρS . This is done to separate the single quantities. . (2.

.28): Z I Z ∂ ρedV = − ρUe · ndS − (τ : (∇ ⊗ U))dV ∂t Z I Z . As before. The time derivative will be manipulated to: ∂ ∂e ∂ρ (ρe) = ρ +e .85) ∂t | ∂t   {z } | {z } total derivative continuity Due to the continuity equation.83) After inserting the two terms into equation (2. (2. (2.5.1 Integral Form of the Thermo Energy Equation The integral form of the internal energy equation (2.80). .81) − p(∇ • U)dV − q · ndS + ρSdV 2.82) ∂t ∂t ∂t and the convective term to: ∇ • (ρUe) = ρU • ∇e + e∇ • (ρU) .2 Non-Conserved Thermo (Internal) Energy Equation As before. we can cancel out the second term and write the non-conserved internal energy equation as: De ρ = −(τ : (∇ ⊗ U)) − p(∇ • U) − ∇ • q + ρS .80) is observed using the Gauss theorem (1. . we can rewrite the conserved internal energy equation into a non-conserved form using the continuity equation.32 Derivations of the Governing Equations 2.5.84) ∂t ∂t Now we extract ρ and e:   ∂e h ∂ρ  i ρ + U • ∇e +e ∇ + • (ρU) = . . (2. (2. we get: ∂e ∂ρ ρ +e + ρU • ∇e + e∇ • (ρU) = . the terms look always similar.86) Dt . For that we split the time and convective terms again. (2. (2.

93) .2.1 Integral Form of the Conserved Enthalpy Equation The integral form of the conserved enthalpy equation (mechanical energy included) is constructed by using the Gauss theorem. that is simply the sum of the internal energy plus the kinematic pressure: p h=e+ . (2.45) with the new expression.6. (2.6.2 Non-conserved Enthalpy Equation As for each conserved equation. Hence.90) to a non-conservative form by using the continuity equation. the time derivation can be re-ordered as: ∂ ∂ ∂ ρh = ρ h + h ρ . (2.90) +ρ (U • g) − ∇ • [τ • U] + ρS 2.87) ρ If we replace e in equation (2. (2. we split the time and convective term:     ∂ ∂ p ∂ 1 p 1 ρh − ρ + ρ|U|2 = − ∇ • (ρUh)+∇ • ρU − ∇ • ρU |U|2 − ∇ • q ∂t ∂t ρ ∂t 2 ρ 2 (2. we get:         ∂ p 1 p 1 ρ h− + ρ|U|2 = − ∇ • ρU h − + |U|2 − ∇ • q + ρ (U • g) ∂t ρ 2 ρ 2 (2.90) can be rewritten like: Z Z Z I Z ∂ ∂ ∂ 1 ρhdV − pdV + ρ|U|2 dV = − (ρUh) · ndS + ρ (U • g) dV ∂t ∂t ∂t 2 I   I I Z . For that we will use the total energy equation (2. equation (2. THE CONSERVED ENTHALPY EQUATION 33 2. (2. Hence.89) + ρ (U • g) −∇ • (pU) − ∇ • [τ • U] + ρS . For further simplification.45) and the definition of the enthalpy h.91) 1 − ρU |U|2 · ndS − q · ndS − [τ • U] · ndS + ρSdV 2 2.6. we get the conserved enthalpy equation as:   ∂ ∂ ∂ 1 1 ρh − p + ρ|U|2 = −∇ • (ρUh) − ∇ • ρU |U|2 − ∇ • q ∂t ∂t ∂t 2 2 . it is possible to change equation (2. Thus. To manipulate the conserved equation. we first have to split the time and convection terms of the enthalpy equation.88) − ∇ • (pU) − ∇ • [τ • U] + ρS .92) ∂t ∂t ∂t The convection term will be manipulated to: ∇ • (ρUh) = ρU • ∇h + h∇ • (ρU) .6 The Conserved Enthalpy Equation The next equation that we are going to derive is the conserved enthalpy equation. The terms that are underlined are equal and cancel out as well as the density in the second term of the LHS.

The equation we get is very common and can be found in many literatures. . −∇ • (ρUh) − ∇ • ρU |U|2 − ∇ • q + ρ (U • g) − ∇ • [τ • U] + ρS 2  1 − −∇ • (ρU |U|2 ) − ∇ • [τ • U] − (−τ : (∇ ⊗ U)) 2 −∇ • (pU) − p(−∇ • U) + (ρg) • U} .65):     ∂ ∂ ∂ 1 ∂ 1 ρh − p + ρ|U|2 − ρ|U|2 = .99) ∂t ∂t | {z } Total derivative .69) and use (2. that the mechanical energy is removed and we only have the thermo energy included. . we need to subtract equation (2.90) with (2. we get the non-conserved enthalpy equation:   Dh ∂ 1 ∂ 1 ρ =− ρ|U|2 + p − ∇ • ρU |U|2 − ∇ • q + ρ (U • g) Dt ∂t 2 ∂t 2 .34 Derivations of the Governing Equations Finally. that a lot of terms can be canceled out. . Furthermore. it is possible to modify this equation by putting the second term of the LHS to the RHS: ∂ ∂ ρh = −∇ • (ρUh) − ∇ • q + ρS + (−τ : (∇ ⊗ U)) + p + U • ∇p . ∂t ∂t ∂t 2 ∂t 2 continues on next site     1 .95) ∂t | ∂t {z } ∂t 2 ∂t continuity and sort the equation. .96) −∇ • [τ • U] + ρS 2..94) ∂t ∂t ∂t ∂t 2 By taking out the density ρ and enthalpy h of the terms of interest:   ∂ h∂ i ∂ 1 ∂ ρ h + U • ∇h+ + h ρ+  ∇•(ρU) + ρ|U|2 − p = . (2.90) and move the convective term to the LHS.6. . The difference is.97) The outcome of the subtraction is. we are allowed to rewrite this term as −U • ∇p. the 10th and 11th term on the RHS can be combined using the product rule. . ∂ ∂ ρh − p = −∇ • (ρUh) − ∇ • q + ρS + (−τ : (∇ ⊗ U)) + U • ∇p .98) ∂t ∂t Furthermore. (2.. (2. The result is: ∂ ∂ ∂ ∂ 1 ρ h+h ρ− p+ ρ|U|2 + ρU • ∇h + h∇ • (ρU) = . Therefore. we replace the split term in equation (2. (2. (2.3 The Conserved Enthalpy Equation (only Thermo) In many literatures we find another enthalpy equation. (2. Thus.

6. Ferziger and Perić [2008]. [1960].100) ∂t Dt Note: The equation above can be found in the following literatures Bird et al. .2.21). Schwarze [2013]. (2. we have a negative sign in front of τ . Therefore. Keep in mind that we still did not introduce the definition of the shear-rate tensor τ . THE CONSERVED ENTHALPY EQUATION 35 This lead to the the total derivative on the RHS for the pressure and we can apply the rule given by equation (1. The modified equation is then given by: ∂ Dp ρh = −∇ • (ρUh) − ∇ • q + ρS + (−τ : (∇ ⊗ U)) + .

36 Derivations of the Governing Equations .

2) ∂t Of course. In addition. if we have an incompressible fluid we get equation (2. The general (governing) conservation equation of any quantity φ is given by: ∂ ρφ =− ∇ • (ρUφ) + ∇ • (D∇φ) + Sφ . we have to replace φ by 1. Now we are able to simply derive the mass. no source terms.2 The Momentum Equation To get the momentum equation we replace φ by U. we need to know the diffusion term and all other source terms that influence the momentum in the volume element.1) ∂t | {z } | {z } | {z } | {z } time accumulation convective transport diffusive transport source terms In the equation above.Chapter 3 The Governing Equations for Engineers Normally. 3.3) ∂t 37 . 3. we get the continuity equation (2. The diffusion term determines the transport of momentum due to molecular effects (τ ). Thus. (3. Later on. momentum and other conservative equations out of this by replacing the quantity φ by the quantity of interest.1 The Continuity Equation To derive the mass conservation equation. that can be a scalar or a vector and Sφ stands for any kind of sources or sinks that influence the quantity φ. it is sufficient enough (for engineers) to know the general conservation equation for an arbitrary quantity φ. we are able to – probably – derive any kind of equation. D stands for the diffusion coefficient.13). we see a more general form of these equation. Once the meaning of this equation is understood. (3. we have to know that the mass is not transfered by diffusion and we suggest that the mass is not transfered into energy or vice verse. ∂ ρU = −∇ • (ρU ⊗ U) + ∇ • τ − ∇p + ρg (3.12): ∂ ρ = −∇ • (ρU) . The source terms are: gravitational acceleration and the pressure force. Furthermore.

Solver crashes in OpenFOAMr are sometimes related to wrong coupled equations.38 The Governing Equations for Engineers Note: As we see. pressure work or the kinetic energy accumulation are really influencing the enthalpy equation. that it is possible to use the governing conservation equation to derive other – more complex – equations. the temperature equation was derived with the assumptions of incompressibility and constant heat capacity. the shear-rate tensor τ has a positive sign in this equation. If we get more familiar with the equations especially with the stress-tensor and the source terms. as already mentioned above. ∂ ρh = −∇ • (ρUh) + ∇ • (λ∇T ) (+Sh ) . Doing so. In addition. The sign change will be understood after we introduce the definition of the shear-rate components (all components are negative). However. In each equation before we had a negative sign. 3. That can be also analyzed from the OpenFOAMr toolbox. Temperature equation The temperature equation can be derived using the thermodynamic relation: ∂h cp = .3 The Enthalpy Equation To derive the enthalpy equation. we can manipulate the enthalpy equation to get to the following temperature equation: ∂ cp ρT = −cp ∇ • (ρUT ) + ∇ • (λ∇T ) (+Sh ) . we recalculate the enthalpy based on the temperature fluid and the fluid properties. we have to take care about different phenomena that influence the temperature in our system. friction. no temperature or enthalpy equation is used whereas for compressible fluid we solve the enthalpy equation. Example given: if friction. we have to replace φ by h. if we are solving compressible fluids. In the equation above we already applied the definition of the shear-rate components and hence the sign has to change. For incompressible fluids. This will lead to the internal energy equation.4) ∂t | {z } neglected It should be mentioned that the enthalpy equation has a special characteristic because is the necessity to know the temperature field T .5) ∂t | {z } neglected Depending on the field we are working on. ∂T Assuming constant heat capacity and incompressibility. . These terms are given in the chapter before and are neglected now. One nice example would be. The momentum equation shows. we always have to know which source terms are important and how the diffusion term looks like. The enthalpy equation is of interest. we have to take these phenomena into account. Therefore. (3. it is very easy to use this equation and derive the one that is needed. However we set the temperature field. (3. we should be aware if the equation is valid in the case we are trying to solve. The diffusion term (−∇ • q) can be expressed by the Fourier law q = −λ∇T . and so on. the energy of a fluid can be changed by other sources like the pressure work.

3. that the temperature equation looks different if the above mentioned assumptions are not fulfilled. This can be expressed as.68): Sapw = ∇ • (Up) .8) If we are dealing with incompressible fluids. we have to add the term that describes the shear-heating. The term was introduced in OpenFOAMr 2.6) • Pressure work Pressure can also increase the enthalpy of a fluid during time.7) ∂t Note: This term can be turned on and off in the enthalpy equation by using the dpdt keyword within the thermophysicalProperties file in OpenFOAMr . | {z (3. • Additional pressure work There is also an additional pressure work done by the divergence of the pressure and velocity. compare equation (2. If follows: Sapw = ∇ • (Up) = U • ∇p + p ∇ • U} = U • ∇p .0.3.2. (3. compare (2. (3. THE ENTHALPY EQUATION 39 solving a fluid for incompressible fluids but using a temperature depended density. (3. Therefore. An extension to that topic might come in some new release.90): ∂p Spw = . we are allowed to say that the additional pressure work is only done by the gradient of the pressure ∇p because we can split Sapw using the product rule. 3.9) continuity Other source terms can be found in chapter 2 or in the literature that was given at the beginning of this chapter.1 Common Source Terms • Shear-heating – viscous dissipation Shear-heating can be included to the enthalpy equation. compare equation (2.80): Ssh = τ : (∇ ⊗ U) . It can be shown — mathematically —.3. In addition it is worth to mention. . that this case can cause troubles if the implementation is not done correct. By default it is set to yes.

40 The Governing Equations for Engineers .

Depending on the problem we are focusing on. they are not presented here. for example the equation for solid mechanics (stress calculation) and/or magneto hydrodynamics (Maxwell-equations).Chapter 4 Summary of the Equations On the next site. Due to the fact that we did not discuss these special kind of equations. There are more equations that could be included here. we should be familiar with the toolbox — which equation and which terms are solved — and the equation respectively. In addition — and again — voluntary contributors are welcomed to work on that book and extend it with their topics. Thus. all derived equations are given in a summary for a fast look-up. However. 41 . it is worth to mention that this section will be extended based on the time I can spend on the book. special terms can be neglected or has to be taken into account.

∇p = ρ̄g Bird Free Convection ∂ ρU = −∇ • (ρU ⊗ U) − ∇ • τ − ρβg(T − T ) 0 ∂t et al. 42 Table 4. [1960]   Total Energy ∂ (ρe + 1 ρ|U|2 ) = −∇ • ρU(e + 1 |U|2 ) − ∇ • q + ρ (U • g) − ∇ • (pU) − ∇ • [τ • U] + ρS Sum of thermo and mechanical energy ∂t 2 2 Kinetic Energy ∂ 1 ρ|U|2 = −∇ • (ρU 1 |U|2 ) − ∇ • [τ • U] − (−τ : (∇ ⊗ U)) − ∇ • (pU) − p(−∇ • U) + (ρg) • U mechanical energy ∂t 2 2 Internal Energy ∂ (ρe) = −∇ • (ρUe) − (τ : (∇ ⊗ U)) − p(∇ • U) − ∇ • q + ρS thermo energy ∂t   Total Enthalpy ∂ ρh − ∂ p + ∂ 1 ρ|U|2 = −∇ • (ρUh) − ∇ • ρU 1 |U|2 − ∇ • q + ρ (U • g) − ∇ • [τ • U] + ρS ∂t ∂t ∂t 2 2 p Energy Enthalpy (Only Thermo) ∂ ρh − ∂ p = −∇ • (ρUh) − ∇ • q + (U • ∇p) + [−τ : (∇ ⊗ U)] + ρS h = e+ ∂t ∂t ρ   ∂ ρc T = −∇ • (ρUc T ) − ∇ • q − (τ : ∇ ⊗ U) − T ∂p Dcp in terms of cv Bird et al.1: Conserved equations for pure fluids Continuity – ∂ρ For incompressible fluids we get = −∇ • ρU ∂t ∇•U = 0 Forced Convection ∂ ρU = −∇ • (ρU ⊗ U) − ∇ • τ − ∇p + ρg For τ = 0 we get the Euler equa- ∂t tion Momentum Approximate. Temperature v v (∇ • U) + ρT ∂t ∂T ρ Dt [1960]   in terms of Bird ∂ ρc T = −∇ • (ρUc T ) − ∇ • q − (τ : ∇ ⊗ U) + ∂ ln V Dp Dcp cp Temperature p p + ρT ∂t ∂ ln T ρ Dt Dt et al. [1960] Summary of the Equations .

if we analyze the equations we figure out that some quantities are not known. 5. These unknown quantities have to be expressed by known one.5) ∂z ∂y   ∂uz ∂ux τzx = τxz = −µ + . This enables the possibility to get a better insight into the physics and lead to a better understanding about the phenomena in the flow field which can be used to optimize designs or increase the efficiency of a special device. (5.Chapter 5 The Shear-rate Tensor and the Navier-Stokes Equations The equations that we derived till now allow us to calculate the flow fields numerically. In addition.3) ∂z  3  ∂ux ∂uy τxy = τyx = −µ + . it is possible to extract quantities that are not measurable in reality – just imagine a liquid metal and measuring the pressure or velocity in a simple way.4) ∂y ∂x   ∂uy ∂uz τyz = τzy = −µ + .1 Newtonian Fluids If we investigating into Newtonian fluids. In this chapter we introduce the shear-rate tensor τ for Newtonian fluids. (5.1) ∂x 3 ∂uy 2 τyy = −2µ + ( µ − κ)(∇ • U) . However. Bird et al. Dantzig and Rappaz [2009]. Further notes and information can be found in Ferziger and Perić [2008]. (5. It was shown that the nine components can be described as: ∂ux 2 τxx = −2µ + ( µ − κ)(∇ • U) . Here we distinguish between Newtonian and Non-Newtonian fluids. (5.6) ∂x ∂z 43 . we use the Newtonian law for the shear-rate (viscous stress) tensor τ . (5. [1960]. (5. The shear-rate tensor is expressed by different equations which depend on the behavior of the liquid.2) ∂y 3 ∂uz 2 τzz = −2µ + ( µ − κ)(∇ • U) . for example the shear-rate components τij .

we will get the momentum equations for Newtonian fluids also known as Navier-Stokes equations. Further information can be found in chapter 7 or directly in Gurtin et al. The other term in the normal stress components takes the value 3 µ. Remark In general it is sufficient to know how the shear-rate tensor is defined. we will keep κ for now in the equations. However. 2 again. later on we see why. Therefore. In addition we should make clear statements about the quantities pressure and equilibrium pressure. However.7) ∂ ∂uz ∂ux + −µ + ∂z ∂x ∂z ∂p − + ρgx ∂x For the y-component of the Navier-Stokes equation we get:   ∂ ∂ ∂ ∂ ρuy = − ρux uy + ρuy uy + ρuz uy ∂t ∂x ∂y ∂z       ∂ ∂ux ∂uy ∂ ∂uy 2 − −µ + + −2µ + ( µ − κ)(∇ • U) ∂x ∂y ∂x ∂y ∂y 3    . [2010]. dilatation term or first Lamés coefficient that is again not correct. we see that this definition is not correct and we should describe the bulk viscosity in a different manner. [2010]. (2. Everything about that can be found in Gurtin et al. [1960] describes this variable as the bulk viscosity. Nevertheless.23). which many authors refers to the secondary viscosity.44 The Shear-rate Tensor and the Navier-Stokes Equations The normal stress components have the quantity κ included.25). [1960] mentioned. The main reason for that misleading definition is based on the fact that we can make the following correlation λ = − 32 µ. if we are going to implement new models or establish new models. Continued If we insert the above mentioned definitions of the shear-rate components into the momentum equations (2. (5. as Bird et al.24) and (2. For the x-component of the Navier-Stokes equation we get:   ∂ ∂ ∂ ∂ ρux = − ρux ux + ρuy ux + ρuz ux ∂t ∂x ∂y ∂z       ∂ ∂ux 2 ∂ ∂ux ∂uy − −2µ + ( µ − κ)(∇ • U) + −µ + ∂x ∂x 3 ∂y ∂y ∂x    .8) ∂ ∂uz ∂uy + −µ + ∂z ∂y ∂z ∂p − + ρgy ∂y and finally the z-component of the Navier-Stokes equation can be written as: . (5. the quantity κ is not really important for dense gases and liquids and can be neglected. Bird et al. in chapter 7. it is necessary to understand the different terms and their meaning. it is a good choice to know why and how we get to the shear rate tensor. However.

(5.13) ∂t 3 | {z } viscous stress tensor τ The viscous stress tensor also named shear-rate or deformation rate tensor τ can be defined as:     2 τ = 2µD + − µ + κ (∇ • U)I . (5. (5. we are able to write the vector form of this equation:     ∂ 2 ρU = −∇ • (ρU ⊗ U) − ∇ • −2µD + µ − κ (∇ • U)I − ∇p + ρg . (5.14) 3 If we use κ = 0. we take the negative sign into the brackets of the second term on the RHS to get the general Navier-Stokes equation in vector notation:     ∂ 2 ρU + ∇ • (ρU ⊗ U) = ∇ • 2µD + − µ + κ (∇ • U)I −∇p + ρg .10) 2 ∂ui ∂p + µ−κ − + ρgi 3 ∂xi ∂xi Introducing the deformation rate (strain-rate) tensor D. 1h i D= ∇ ⊗ U + (∇ ⊗ U)T . (5.9) ∂ ∂uz 2 + −2µ + ( µ − κ)(∇ • U) ∂z ∂z 3 ∂p − + ρgz ∂z The three equations can be put together by using the Einsteins summation convention:     ∂ ∂ ∂ 1 ∂ui ∂uj ρui = − (ρuj ui ) − −2µ + ∂t ∂xj ∂xi 2 ∂xj ∂xi    . which comes from the Stokes. that means Aij → Aji .15) ∂t 3 | {z } viscous stress tensor τ Another form of the momentum equation (5.13) can be achieved after pushing the pressure gradient .12) ∂t 3 Finally. NEWTONIAN FLUIDS 45   ∂ ∂ ∂ ∂ ρuz = − ρux uz + ρuy uz + ρuz uz ∂t ∂x ∂y ∂z        ∂ ∂ux ∂uz ∂ ∂uy ∂uz − −µ + + −µ + ∂x ∂z ∂x ∂y ∂z ∂y   . (5.1. we get the common Navier-Stokes equation as shown in almost all literatures:   ∂ 2 ρU + ∇ • (ρU ⊗ U) = ∇ • 2µD − µ(∇ • U)I −∇p + ρg .11) 2 where T stands for the transpose operation. (5.5.

16) ∂t 3 | {z } Cauchy stress tensor σ The result of the bracket is called the Cauchy stress tensor σ. we will investigate only into the terms that are not discussed till now. 3 The first step is to split the terms:    2 ∇ • (2µD) + ∇ • − µ + κ (∇ • U)I − ∇p . For that.1 The Proof of the Transformation The following section discusses the transformation of the vector form into the Cartesian one. The terms that we are going to investigate are the viscous stress tensor τ and the pressure gradient of equation (5. A discussion about this quantity is given in chapter 6. A short overview of some constitutive relations are given in chapter 7.8) and (5. Additionally it is easy to show that the expression of ∇ • (pI) is equal to ∇p.17)     ∂ ∂p 0 0 1 0 0 p   ∂z ∂z . 5. it has to be mentioned that there are much more ways to define this equation.46 The Shear-rate Tensor and the Navier-Stokes Equations into the viscous stress tensor:      ∂ 2 ρU + ∇ • (ρU ⊗ U) = ∇ • 2µD + − µ + κ (∇ • U) − p I +ρg . More information about why we can use this equation for solids too. The momentum equation can also be defined for solids.13).1. Especially if we distinguish between total pressure p and the equilibrium pressure peq .7). Finally.9). Implementation information and stability terms for solid stress simulations can be found in Jasak and Weller [1998]. [2010].     2 ∇ • 2µD + − µ + κ (∇ • U)I − ∇p . (5. See chapter 7. Therefore we are using the Lamés coefficients. can be found in Gurtin et al. (5. (5.17):    ∂     ∂p   1  0 0  ∂x p 0 0 ∂x ∂    !  ∂p  − ∇ • (pI) = −∇ • p 0 1 0 = −  ∂y •  0 p 0 = −  ∂y  = −∇p . we are using the mathematic operation (1. 3 It is easy to demonstrate that the pressure gradient is equal to the terms in equation (5. As before.

24) ∂y ∂y ∂x ∂z ∂x ∂z taking the minus sign into the brackets.17):  i 1h ∇ • (2µD) = ∇ • 2µ ∇ ⊗ U + (∇ ⊗ U)T (5. needs the mathematic operations (1. The investigation into the first term that includes the deformation rate tensor D.1.14) and (1. we end up with:     ∂ ∂ux ∂ 2 2µ − ( µ − κ)(∇ • U) ∂x ∂x ∂x 3       ∂ ∂ux ∂uy ∂ ∂uz ∂ux + µ + + µ + . we use the term within the brackets {.. of course equation (5. split the x-derivative.   ∂x ∂z ∂x ∂y ∂z Now we need to check if the terms are correct.25) ∂y ∂y ∂x ∂z ∂x ∂z . For that. (5.23) already shows that the terms have to be similar but we will demonstrate why. .} of equation (5.    ∂ ∂ux 2 − −2µ + ( µ − κ)(∇ • U) ∂x ∂x 3       ∂ ∂ux ∂uy ∂ ∂uz ∂ux + −µ + + −µ + . (5.5..22)   ∂y  ∂y ∂x ∂y ∂y ∂z  ∂u ∂ ∂z µ[ ∂u ∂z x + ∂uz ∂x ] µ[ ∂zy + ∂uz ∂y ] µ[ ∂u ∂z z + ∂uz ∂z ]   ∂u    ∂ 2µ ∂u ∂ µ[ y + ∂ux ] + ∂ µ[ ∂uz ∂ux x  ∂x ∂x + ∂y + ]  of x − mom. NEWTONIAN FLUIDS 47 As we can see. the terms are equal for the pressure. (5.23)    ∂y   ∂u  ∂ µ[ ∂u + ∂u ∂ µ[ ∂zy + ∂u ∂ 2µ ∂uz x z  z ] + ∂y ] + ∂z of z − mom.19)         ∂ ∂ u u   ∂z z ∂z z   ∂u x ∂uy ∂uz   ∂u x ∂ux ∂ux   ∂x  ∂x ∂x ∂x ∂y ∂z   ∂uy  ∂uy ∂uy ∂uy   = ∇ • µ  ∂u ∂uz  + (5.20)   x ∂y ∂y ∂y   ∂x ∂y ∂z   ∂uy   ∂ux ∂uz ∂uz ∂uz ∂uz   ∂z ∂z ∂z ∂x ∂y ∂z   ∂u x ∂ux ∂uy ∂ux ∂uz ∂ux  ∂x + ∂x ∂x + ∂y ∂x + ∂z ∂uy ∂uy ∂uy = ∇ • µ  ∂u ∂uz ∂ux  (5.18) 2         T  ∂ ∂ ux ux    ∂x  ∂x    ∂  ∂    = ∇ • µ  ∂y  ⊗ uy  +  ∂y  ⊗ uy  (5.7).21)   x ∂y + ∂x ∂y + ∂y ∂y + ∂z  ∂ux ∂uz ∂uy ∂uz ∂uz ∂uz ∂z + ∂x ∂z + ∂y ∂z + ∂z ∂uy µ[ ∂u ∂ux ∂ux µ[ ∂u ∂ux  ∂    x z ∂x ∂x + ∂x ] µ[ ∂x + ∂y ] ∂x + ∂z ] ∂uy ∂u ∂uy • µ[ ∂u µ[ ∂u ∂ux  ∂  = x + ] µ[ ∂yy + ] z + ] (5.  ∂x  ∂y  ∂z  ∂x ∂z       ∂u ∂u  ! µ[ ∂u µ[ ∂uz ∂ux ∂ ∂ ∂ =  ∂x ∂y x + ∂xy ] + ∂y 2µ ∂yy + ∂z + ∂z ]  = of y − mom.

we will use the continuity equation to modify this equation.32) ∂t .18). This term comes from the last term that we neglected till now. all terms of the shear-rate tensor are similar and therefore the vector form is identical to the three single Cartesian ones. The mass conservation equation is given by: ∂ρ + ∇ • (ρU) = 0 . To demonstrate the meaning of the term and the correlation to expansion and compression phenomena. (5. (5. That means. the second term − 23 µ is referred in many liter- atures to be the dilatation viscosity.27) ∂z ∂z ∂z 3 The underlined terms occur in equations (5.30) ∂ 2  − ∂z [ 3 µ − κ](∇ • U)    of z mom As demonstrated.9). (5.8) and (5. However. we get the analyzed term of the y-momentum to:      ∂ ∂ux ∂uy ∂ ∂uy µ + + 2µ ∂x ∂y ∂x ∂y ∂y      ∂ 2 ∂ ∂uz ∂uy − ( µ − κ)(∇ • U) + µ + .1.31) ∂t After applying the product law (1. the choice of this nomenclature is a bad one.48 The Shear-rate Tensor and the Navier-Stokes Equations Apply the same procedure on the terms of equation (5. (5. As already said.23). we get: ∂ρ + ρ∇ • U + U • ∇ρ = 0 . this term represents expansion and compression phenomena.28)   3 3 0 0 1  ∂   2   ∂x 3 µ − κ (∇ • U) 0 0 ∂   2  = −  ∂y  • 0 µ − κ (∇ • U) 0 (5.2 The Term − 23 µ(∇ • U) As mentioned at the beginning of this chapter. (5. The last term can be manipulated to:         1 0 0 2  2 −∇ • µ − κ (∇ • U)I = −∇ •  µ − κ (∇ • U) 0 1 0 (5. 5. that the vector form and Cartesian one are similar for these term but there is one term missing in each derivative.29)  3 ∂ 2   ∂z 0 0 3 µ − κ (∇ • U)  of x mom  ∂ 2   − ∂x [ 3 µ − κ](∇ • U)     ∂ 2  ! = − ∂y [ 3 µ − κ](∇ • U) = of y mom .26) ∂y 3 ∂z ∂y ∂z and the term of the z-momentum is equal to:       ∂ ∂ux ∂uz ∂ ∂uy ∂uz µ + + µ + ∂x ∂z ∂x ∂y ∂z ∂y     ∂ ∂uz ∂ 2 + 2µ − ( µ − κ)(∇ • U) .

Hence. the dilatation term.5. Therefore we get the two parameter equation: τ = kγ̇ n (5.34).36) The underlined term results in a tensor that can be simplified by the continuity equation. They assume that τ 0 = 0. For that we can use the equation suggested by Herschel-Bulkley. the shear-rate tensor has to be treated in another way due to the fact that the Newtonian law is not valid anymore. we are allowed to take out the density of all remaining derivatives. we can divide by this quantity to get rid µ of the density.1.2.2 Non-Newtonian fluid Considering non-Newtonian fluids. Further simplifications are made by Ostwald and de Waele. (5.38) This equation can be used in modelling Non-Newtonian fluids. we get the famous Laplace equation: ∇ • τ = ν∇2 U = ν∆U .39) k denotes a consistency factor. the second term in equation (5. γ̇ the shear-rate and n the potential factor. this term will vanish based on the fact that the density will not change during time and the gradient of a constant number is zero.13).37) 5. (5. . NON-NEWTONIAN FLUID 49   1 ∂ρ ∇•U=− + U • ∇ρ . The outcome is the following: We clearly see that the second term on the RHS is only related to the density change and thus.33) ρ ∂t This expression is now insert into the shear-rate tensor τ of equation (5. (5. it is related to expansion and compression phenomena:    2 1 ∂ρ τ = 2µD − µ − + U • ∇ρ I . we further can simplify the equation by taking out the viscosity of the divergence operator: ∇ • τ = ν∇ • (∇ ⊗ U + (∇ ⊗ U)T ) . we can use the continuity equation for a simplifying the shear-rate tensor.35) If the dynamic viscosity ν can be assumed as constant. Thus. In addition. (5. depends only on the density. They assumed the shear-rate tensor with a power law equation: τ = τ 0 + kγ̇ n (5. Thus. ρ = constant. Using the following parameters k = 2µ.3 Further Simplifications If we assume incompressibility of the fluid.34) 3 ρ ∂t | {z } expansion and compression 5. (5. γ̇ = D and n = 1 lead to the shear-rate tensor for incompressible Newtonian fluids. The result of the shear-rate tensor is as follows (ν = ρ ): τ = 2νD .

my colleagues are modeling non-Newtonian fluids with a different viscosity model rather than a different strain-rate tensors. . However. There is no time to update this section right now.50 The Shear-rate Tensor and the Navier-Stokes Equations Remark: This section was build in the first edition and is not added without proofing.

Shear-Rate Tensor and Pressure In equation (5. That means.25) or (1.2):   1 σ = −pI + σ − tr(σ)I . (6. First we start with the introducing of the Cauchy stress tensor σ:   σxx σxy σxz σ = σyx σyy σyz  . (6.26) can be related to the pressure as: 1 − p = Ahyd = tr(A) .4) 3 Using the definition of the deviatoric part (1. we are able to split each matrix into a deviatoric and hydrostatic part. The definition is given in equation (1.6) 51 . (6. shear-rate stress and pressure are related is briefly discussed in this chapter.Chapter 6 Relation between the Cauchy Stress Tensor. we can rewrite equation (6. (6. equation (1.3) 3 1 − pI = Ahyd I = tr(A)I . How the total stress.24) and can be applied to the Cauchy stress tensor: σ = σ hyd + σ dev . (6.2) The hydrostatic part of an arbitrary matrix A has the special meaning of the negative pressure p.1)   σzz σzy σzz As we know from chapter 1.5) 3 | {z } shear−rate tensor τ The deviatoric part is defined as the shear-rate stress and can be expressed by the shear-rate tensor τ : σ dev = τ .27) and the above expression. Hence.16) we introduced the Cauchy stress tensor σ. shear and pressure forces because both can related to stresses. This stress tensor includes all stresses that act on the volume element dV . (6.

Shear-Rate Tensor and Pressure σ = −pI + τ . . In solid mechanics it is common to work with the full stress tensor σ. Note: In fluid dynamics it is common to use the split Cauchy tensor to get the shear-rate tensor and pressure. compare Jasak and Weller [1998].7) Depending on the fluid we are using. τ has to be replaced with the correct expressions.52 Relation between the Cauchy Stress Tensor. (6.

we are able to write the viscous (deviatoric) part of the Cauchy stress tensor as: σ vis (ρ. First we have to know that an isotropic linear tensor function T(A) is isotropic. dilatation viscosity. 0) has to be an isotropic tensor. All of these names represent the same. All necessary informations can be found in Gurtin et al. (7. The bulk viscosity is also named. D) . (7.1) Gurtin et al. 0) . D) = σ(ρ. Why we have the density included can be seen in equation (7. Bird et al.3) Based on the fact that σ(ρ. In order to make a clear statement about the bulk viscosity. (7. It follows: σ = σ(ρ. volumetric viscosity or volume viscosity.4) 53 . 0) that represents the stress in the fluid in the absence of flow (no velocity gradients and therefore no shearing). [2010] proofed that the shear-rate tensor (also named stretch tensor in their book) is an isotropic tensor which can be shown using frame-indifferences and thus τ has to follow the equation above. we have to go deeper into the basics of the mechanics and thermodynamics. D) − σ(ρ. [2010] which are summarized and compressed in the following chapter for compressible fluids that show a linear function in the viscosity for the shearing. if and only if there are two quantities (scalars) µ and λ (do not mix with first and second Lamé coefficient or molecular viscosity here) such that we can define the tensor function as: T(A) = 2µA + λ tr(A)I . (7. it has the specific form to be equal to the negative equilibrium pressure peq : σ(ρ. we have to start with the Cauchy stress tensor σ which is depended on the density and the strain-rate tensor D.2) Defining a new quantity σ(ρ.Chapter 7 The bulk viscosity As we already pointed out in chapter 5. 0) = −peq (ρ)I . To demonstrate which quantity defines the bulk viscosity. [1960] referred the quantity κ which is included in the normal components of the shear-rate tensor as bulk viscosity.9).

we get: σ vis (ρ. volume viscosity or dilatation viscosity and is equal to: 2 κ(ρ) = µ(ρ) + λ(ρ) . D). these components try to deform the volume element while keeping the volume of the element constant (no compression or expansion). namely the isotropic one.5 is similar but not equal to 6.15) 3 . (7. we end up with: σ(ρ.11) 3 1 = 2µ(ρ)D0 + 2µ(ρ) tr(D)I + λ(ρ) tr(D)I .6) 3 we are able to express the total pressure by the equilibrium pressure and the viscous stress tensor: 1   p(ρ) = − tr −peq (ρ)I + σ vis (ρ. D) = 2µ(ρ)D + λ(ρ) tr(D)I . the equilibrium pressure and a part. To answer this. Using the already known definition of the total pressure p.12)  3  2 = 2µ(ρ)D0 + µ(ρ) + λ(ρ) tr(D)I . (7. volumetric viscosity. Thus. (7. (7. D) . of the viscous part which is based on internal friction.54 The bulk viscosity Finally. Thus. In other words. The quantity D0 represents the deviatoric part of the strain-rate tensor which include only friction which acts angular. (7. (7.8) 3 Now it is obvious that the total pressure for compressible fluids is based on two different contri- butions. D)] .7 based on the the equilibrium pressure and the viscous term. which is given by: 1 p(ρ) = − tr(σ) . (7. (7.9) The next step is to split the deformation (strain-rate) tensor in its deviatoric and hydrostatic (spherical) part: 1 D = D0 + tr(D)I . after we sum up the latest results.7) 3 It is trivial that this lead to: 1 p(ρ) = peq (ρ) − tr [σ vis (ρ. D) − peq (ρ)I . (7.5) Equation 7. The question that may arise is: Where is the equilibrium pressure in all equations that we had before.13) 3 | {z } =κ(ρ) = 2µ(ρ)D0 + κ(ρ) tr(D)I .1) for the viscous part. Later we see why both equations are equal for some particular cases – in which we are working commonly. D) = 2µ(ρ) D0 + tr(D)I + λ(ρ) tr(D)I .10) 3 and insert this expression into the first term on the RHS of equation (7. (7. we have to use the relation (7. (7.14) κ represents the bulk viscosity. σ vis (ρ.9). D) = σ vis (ρ. it follows:   1 σ vis (ρ.

Thus.17) and thus the Cauchy stress tensor can be written as: σ(ρ.15).14): tr [σ vis (ρ. As Bird et al. Further information can be found in Gurtin et al. both equations are equivalent. this was just a brief summary for compressible linearly viscous fluids. (7. [1960] already mentioned. Comparing this with Bird et al. (7.17). In addition the above constitutive equation for the stress tensor σ is only valid for compressible fluids which have a linear viscosity behavior. Doing that. Remark In this chapter we used a form of the equations which included the functionality dependency of the quantities.15).16) and put this into equation (7. considering equation (7.9). This was done just in order to show the dependency which is not really necessary. we directly see that λ has to be equal to − 32 µ. this equation looks similar to equation (6. . [1960]. keep in mind that the variables λ(ρ) and µ(ρ) in this section are not related to the molecular viscosity or the Lamés coefficients and represent two arbitrary scalar functions which depend on the density. that the total pressure p is equal to the equilibrium pressure peq in the case when we assume the Stokes relation. Now. we get: p(ρ) = peq (ρ) − κ(ρ) tr(D) .7) but is in fact not equivalent. In addition we see. D) = 2µ(ρ)D0 − p(ρ)I . we see that there is a difference in the nomenclature. Furthermore. if we take the trace of equation (7. If we insert the outcome related to the Stokes hypothesis into equation (7. the quantity κ is not too important and can be neglected which is also stated by Gurtin et al. Based on that.18) Again. 55 Now we stated the correct bulk viscosity definition. it is common to use the Stokes relation given by the definition κ = 0. In addition. cf. equation (5. and express the quantity µ with the molecular viscosity of the fluid. D)] = 3κ(ρ) tr(D) . [2010]. [2010]. we get the expression for the shear-rate tensor τ .8). (7. cf (7. To get the equivalence of both equations we have to go further.

56 The bulk viscosity .

26): ∂ ρU = −∇ • (ρU ⊗ U) − ∇ • τ − ∇p + ρg . (8.Chapter 8 Collection of Different Notations of the Momentum Equations In literatures we find a lot of different notations for the momentum equation for Newtonian fluids. (8.2) ∂t 3 • Conserved momentum equation as used in chapter 7:   ∂ 2 ρU = −∇ • (ρU ⊗ U) + ∇ • 2µD0 + (λ + µ)(∇ • U)I − ∇p + ρg . • Common conserved momentum equation (2. (8.1) ∂t • Conserved momentum equation with the shear-rate tensor:   ∂ 2 ρU = −∇ • (ρU ⊗ U) + ∇ • 2µD − µ(∇ • U)I − ∇p + ρg . (8.4) ∂t 57 . It is obvious that we can change the stress tensors as we want and play around with the mathematic laws to manipulate the equation as we like to have it.3) ∂t 3 • Conserved momentum equation with bulk viscosity: ∂ ρU = −∇ • (ρU ⊗ U) + ∇ • (2µD0 + κ(∇ • U)I) − ∇p + ρg .

That the operation tr(D) results in ∇ • U is shown now: ∇ • U = tr(D) . The Proof that the Trace Operator replaces the Divergence Operator In one equation above we replaced the divergence operator by the trace operator.58 Collection of Different Notations of the Momentum Equations • Conserved momentum equation with trace operator:   ∂ 2 ρU = −∇ • (ρU ⊗ U) + ∇ • 2µD − µ tr(D)I − ∇p + ρg .10) 2 ∂x ∂y ∂z The divergence operator is evaluated by equation (1.5) ∂t 3 • General conserved momentum equation with Cauchy stress tensor: ∂ ρU = −∇ • (ρU ⊗ U) + ∇ • σ + ρg . (8. (8. . (8.16). (8.8) ∂t It should be obvious that we can transform all equations above into the non-conservative or integral form. (8. (8.7) Dt • Integral form of momentum equation: Z I I I Z ∂ ρUdV = − (ρU ⊗ U) · ndS − τ · ndS − pI · ndS + ρgdV .9) The demonstration is very simple:  h i 1 tr(D) = tr ∇ ⊗ U + (∇ ⊗ U)T 2   1 ∂ux ∂uy ∂uz = 2 +2 +2 =∇•U .6) ∂t • Non-conserved momentum equation with Cauchy stress tensor: DU ρ = ∇ • σ + ρg .

x)dt .1. with all physics. This behavior is valid for each quantity we focus on like ux . That lead to the derivative of the Reynolds stress equation. we discuss the main problem if we want to average the compressible mass. the concept is defined by: Z t+T 1 φ̄(x)T = lim φ(t.Chapter 9 Turbulence Modeling In this chapter we focus on turbulent flow fields and the Reynolds-Averaging approach which was introduced by Osborne Reynolds. After that. That means. Then we are going to derive the incompressible mass and momentum equation to show the closure problem. First we will investigate into different averaging approaches. Finally. • Ensemble averaging . c and so on. uy . Wilcox [1994]. momentum and energy equation and introduce the Favre averaging concept.1 Reynolds-Averaging The investigation into flow fields are generally turbulent and hence. T. The main literature that is used in this chapter is Ferziger and Perić [2008]. φ would be the velocity u (in one direction) and could be expressed as: φ(t. (9. 9.1. The time averaging method can be used for a statistic stationary turbulent flow (left figure of 9. it is a challenging task to resolve the flow with all details – in other words. x) and the time averaged one by φ̄(x)T . Observing an arbitrary flow field.2) T →∞ T t 59 . x) = φ̄(x) + φ0 (t. we can analyze that the flow has a deterministic character. the flow is chaotic and can be prescribed using a time independent mean value φ̄ and its fluctuation φ0 that is oscillating around the mean value. [1960]. figure 9. x) . we discuss some hypothesis that are used to get rid of the closure problem.1). for figure 9. • Spacial averaging . Defining the instantaneous flow variable by φ(t. Bird et al. h.1) Osborne Reynolds introduced several averaging concepts that are presented now: • Time averaging . (9. uz . The outcome of this equation is the analyze of the analogies to the Cauchy stress tensor and the derivation of the turbulent kinetic energy.

1).   φ̄(x) = lim (9. As we observe in the figure and in the equation. The reason for that is based on the fact that things can be described easy and clear. x) . The concept can be defined as follows. the averaged value is no longer time depended. The spacial averaging method is appropriate for homogeneous turbulent flows. (9. T should be chosen large enough compared to the time scale of the fluctuation φ0 . x) = φ(t.60 Turbulence Modeling In order to get good results. x) at the nth series. Let us focus on the left part of figure 9. Mean value Mean value Real value Real value Velocity Velocity T N N T1 T2 Time Time Figure 9. Think about a series of measurement with the number of N identical experiments where φn (t. By replacing the instantaneous variable in equation (9.1: Averaging of a statistic stationary (left) and statistic non-stationary flow (right).3) V→∞ V The ensemble averaging method is the most general method. we get: Z t+T 1 φ̄(x) + φ0 (t. x)E : N 1 X φ̄(t. x) dt .4) N →∞ N n=1 For turbulent flow fields that are stationary and homogeneous. x)E = lim φn (t.1 first. that the time average of an already averaged mean quantity . the averaged value is denoted by φ̄(t. x)dV . we can write: Z Z Z 1 φ̄(x)V = lim φ(t. That is the reason why we are interested in the case when T goes to ∞. all three concepts are similar and lead to the same result. Here we average over the volume. This is also known as the ergodic hypothesis. (9. Renaming the averaged quantity to φ̄(x)V .2) by the definition (9. The averaging method that we choose for the further investigation is the time averaging method.5) T →∞ T t We observe that the time averaging of the mean and fluctuation quantity leads to the mean quantity again. This yields to a uniform turbulence in all space directions. Thus we can show.

figure 9. x) + g 0 (t. x) + f 0 (t. x) dt   g(t. x) = f (t. x) + g 0 (t. Linear Terms Applying the average method (9. For now we will use an overline to indicate that we use the Reynolds time-averaging concept instead of writing the integrals.2) to: φ(t. x) | {z } | {z } ¯(t. we have to modify equation (9.1. x) and g(t. we use equation (9. x)dt − lim φ̄(x)dt T →∞ T t T →∞ T t Z t+T Z t+T 1 1 φ̄(x) + φ0 (t.1. x) = f¯(t. x) = lim φ(t.9) to linear terms lead to linear averaged terms.x) =0 Z t+T Z t+T 1 1 ḡ(t.1) and replace the mean quantity on the RHS of equation by (9.x) ḡ =0 .8) Z t+T 1 φ̄(t. If we think about flows where the mean value of the instantaneous quantity is changing during time (non-stationary flows.9. (9. x) dt − lim   = lim φ̄(x)dt T →∞ T t T →∞ T t = φ̄(x) − φ̄(x) = 0 . (9. REYNOLDS-AVERAGING 61 is the mean quantity again: Z t+T 1 φ̄(x) = lim φ̄(x)dt . x) = φ̄(t. we will use two linear functions f (t. x) | {z } | {z } f¯(t. To demonstrate that. To demonstrate this. x) + φ0 (t. right). x) dt   f (t.9) T →∞ T t Some remarks and limits to equation (9. x)dt = T t T t = f¯(t.1) and (9.7) Some remarks of the validity of this method can be found in Wilcox [1994].6) T →∞ T t In addition we can see that the time averaging of the fluctuation is zero. x) + f 0 (t. x)dt = T t T t = ḡ(t. x) = ḡ(t. (9. x)dt . x) = g(t. Correlation for the Reynolds-Averaging For the derivations of the Reynolds-Averaged conservation equations.9) are given in Wilcox [1994]. (9. x)dt = lim φ(t.6): Z t+T Z t+T 1 0 1   lim φ (t. we need some mathematic rules. x): Z t+T Z t+T 1 1 f¯(t. x) . x) − φ̄(x) dt T →∞ T t T →∞ T t Z t+T Z t+T 1 1 = lim φ(t.

 x) + ḡ(t. x).  x) = f¯(t. x) + g(t. x) = f¯(t. x)  = T t +f 0 (t. x) + f 0 (t. The fluctuation terms will vanish based on the proof we did above. x) + g 0 (t. x) + f 0 (t. This behavior is related to the problems we are looking at. x)dt + g(t. ḡ(x). x)g(t. If we have the sum of two linear terms f (t. x)g 0 (t. x)g 0 (t. x) (9. (9. x)g 0 (t. For the last term. To demonstrate this. x)ḡ(t. x) + g(t. x) + g(t. x) +  f¯(t. x) +  f 0 (t. x)] dt T t Z t+T Z t+T 1 1 = f (t. there is no reason for the product of the fluctuations to vanish. x)] dt T t Z t+T 1 f¯(t. x) dt + ḡ(t. x) x)g    f 0 (t. x)ḡ(t. we will use the above term and average it. x)ḡ(t. x) + f 0 (t. For stationary problems we will end up with f¯(x). . we get: f (t. x) = [f (t. x)g(t. x) = [f (t. x).7) is valid.62 Turbulence Modeling Note: The result can be time dependent or not.  0 (t. x)g(t. x) dt   = T t T t = f¯(t. x)ḡ(t.10) Averaging linear terms end up with the same linear terms but now with the mean quantities. x) + g 0 (t. x)ḡ(t. whereas for non-stationary problems we get the terms we derived above. x) + ḡ(t. x) ḡ(t. we produce new additional terms during the averaging procedure. x) + f¯(t. (9. x) . x) . Non-Linear Terms If we focus on non-linear terms like f (t. we get: Z t+T 1 f (t. x)g(t. x) + f 0 (t. +   x) + f 0 (t. x) dt   = f¯(t.11) Rewriting the whole equation. x) dt    = T t Z t+T 1 f¯(t. x) + g 0 (t. x)g 0 (t. x)dt T t T t 1 t+T  ¯ 1 t+T  Z Z f (t. What we get is: Z t+T 1 f (t.12) | {z } additional terms The reason why the second and third term cancels out is due to the fact that the fluctuation is linear in this terms and hence equation (9.

in other words. (9. x) + f 0 (t.2. (9. Furthermore.16) ∂x ∂y ∂z and apply the Reynolds time-average concept.1). Hence. it is obvious that the time derivative will vanish and we end up with: ∂ux ∂uy ∂uz ρ +ρ +ρ =0. x) + a f 0 (t. x)dt T t 1 t+T  ¯ Z f (t. (9. Applying the equations to turbulent flow fields is a hard and challenging topic that requires extreme fine meshes and time steps and lead to high computational costs.13) 9. ∂(ūx + u0x ) ∂(ūy + u0y ) ∂(ūz + u0z ) ρ +ρ +ρ =0.12) in the form of Cartesian coordinates: ∂ρ ∂(ρux ) ∂(ρuy ) ∂(ρuz ) + + + =0.2. REYNOLDS TIME-AVERAGED EQUATIONS 63 Constants Finally. we are allowed to put the density out of the derivatives. Defining an arbitrary constant a. (9. there is no need to resolve all details. Thus.17) ∂x ∂y ∂z . (9. x) = af (t.1 Incompressible Mass Conservation Equation The start point for the derivation is the compressible mass conservation equation (2.2 Reynolds Time-Averaged Equations The Navier-Stokes equations give us the possibility to resolve each vortex and hence all flow phenomena. This will be discussed using the mass conservation equation now. we get: Z t+T 1 af (t.14) ∂t ∂x ∂y ∂z Assuming incompressibility of the fluid.9. ∂(ūx + u0x ) ∂(ūy + u0y ) ∂(ūz + u0z ) ρ +ρ +ρ =0. x) . we use the Reynolds-Averaging concept to simplify the flow equations. the Reynolds time-averaging concept does not affect constant quantities. x) dt  =a T t = af¯(t. Considering incompressibility.  x) = af¯(t. engineers are commonly only interested in some averaged values and on some special physics. the Reynolds time-averaged equations can be derived relatively easily compared to compressible flow fields. the whole turbulence behavior is approximated with models.15) ∂x ∂y ∂z Replacing the values ui by the assumption (9. 9.

2 Compressible Mass Conservation Equation Doing the same average procedure with the compressible mass conservation equation lead to a more complex form because we also have to consider the density as a varying quantity.25) u0z we can rewrite the Reynolds time-averaged compressible mass conservation equation in vector notation: ∂ ρ̄   + ∇ • ρ̄Ū + ρ0 U0 = 0 . 9. (9. (9.19) ∂t ∂x ∂y ∂z ∂ (ρ̄ + ρ0 )(ūy + u0y )   ∂(ρ̄ + ρ0 ) ∂ [(ρ̄ + ρ0 )(ūx + u0x )] + + ∂t ∂x ∂y ∂ [(ρ̄ + ρ0 )(ūz + u0z )] + =0.2.20) ∂z   ∂(ρ̄ + ρ0 ) ∂ [(ρ̄ + ρ0 )(ūx + u0x )] ∂ (ρ̄ + ρ0 )(ūy + u0y ) + + ∂t ∂x ∂y ∂ [(ρ̄ + ρ0 )(ūz + u0z )] + =0. we will investigate into the incompressible . u0y .18) ∂x ∂y ∂z Of course we are allowed to divide the whole equation by the density. (9.  0 ux 0  0 U =  uy  .22) ∂z ∂ ρ̄ ∂(ρ̄ūx + ρ0 u0x ) ∂(ρ̄ūy + ρ0 u0y ) ∂(ρ̄ūz + ρ0 u0z ) + + + =0. that lead to new unknown ρ0 u0i . (9.21) ∂z 0 + ρ0 ū 0 0 ∂(ρ̄ūy +  0 + ρ0 ū ρ̄u 0 0 ∂ ρ̄ ∂(ρ̄ūx +  ρ̄u x  x + ρ ux ) y  y + ρ uy ) + + ∂t ∂x ∂y ∂(ρ̄ūz +  ρ̄u 0 + ρ0 ū 0 0  z  z + ρ uz ) + =0. Due to this behavior. (9. u0z . we get non-linear terms.26) ∂t Due to the fact that we have two quantities that have to be averaged.24) ∂t ∂xi Introducing a vector that contains the fluctuations of the velocities u0x .23) ∂t ∂x ∂y ∂z ∂ ρ̄ ∂ + (ρ̄ūi + ρ0 u0i ) = 0 . (9. (9.64 Turbulence Modeling we get the time-averaged incompressible mass conservation equation: ∂ ūx ∂ ūy ∂ ūz ρ +ρ +ρ = ρ∇ • Ū = 0 . If follows: ∂ρ ∂(ρux ) ∂(ρuy ) ∂(ρuz ) + + + =0. (9. (9.

REYNOLDS TIME-AVERAGED EQUATIONS 65 flows fields first.13.29) ∂x ∂x ∂ ∂ h i ρ (ūy + u0y )(ūx + u0x ) = ρ ūy ūx +  ūy0 + u0  u x y ū  x + u 0 u0 y x ∂y ∂y  ∂ ∂ = ρ ρ(ūy ūx ) + ρ (u0y u0x ) . The x component is now analyzed and averaged in detail. the equations (5.27) ∂x ∂x ∂y ∂y ∂x    ∂ ∂uz ∂ux ∂p + −µ + − + ρgx . For the derivation. Starting with the term on the LHS. The first step is to replace the velocity quantities by equation (9. Therefore we get: ∂ ∂ h i ρ (ūx + u0x )(ūx + u0x ) = ρ ūx ūx +  ūx0 + u0  u x 0 0 x ūx + ux ux  ∂x ∂x  ∂ ∂ =ρ (ūx ūx ) + ρ (u0x u0x ) . ∂x ∂y ∂z To simplify the terms. For the y and z components.2.9.1) and apply the Reynolds time-averaging concept. Averaging the compressible equations will be discussed in section 9.28) ∂t ∂t The first term on the RHS ends up as:   ∂ 0 0 ∂ 0 0 ∂ 0 0 − ρ (ūx + ux )(ūx + ux ) + ρ (ūy + uy )(ūx + ux ) + ρ (ūt + ut )(ūx + ux ) .30) ∂y ∂y ∂ ∂ h i ρ (ūz + u0z )(ūx + u0x ) = ρ ūz ūx +  ūz0 + u0  u x 0 0 z ūx + uz ux  ∂z ∂z  ∂ ∂ =ρ ρ(ūz ūx ) + ρ (u0z u0x ) .9) have to be used. we will examine each term separately. (9. (9.2.8) and (5. we get: ∂ ∂ ρ (ūx + u0x ) = ρ ūx . Due to the fact that the derivations are identical. we will use equation (5. (9. For clearance. we only give the final equation for y and z without all steps. x-Component of Momentum The incompressible momentum equation for the x-component is given by:   ∂ ∂ ∂ ∂ ρ ux = − ρ ux ux + ρ uy ux + ρ uz ux ∂t ∂x ∂y ∂z       ∂ ∂ux ∂ ∂ux ∂uy − −2µ + −µ + (9.7) and assume incompressibility of the fluid. (9. ∂z ∂x ∂z ∂x This is the start point for the procedure. 9.3 Incompressible Momentum Equation The derivation of the time-averaged x-component of the momentum equation will be discussed now. we focus on each term separately.31) ∂z ∂z .

the first term on the RHS can be written as:  ∂ ∂ ∂ − ρ (ūx ūx ) + ρ (u0x u0x ) + ρ (ūy ūx ) ∂x ∂x ∂y  ∂ 0 0 ∂ ∂ +ρ (uy ux ) + ρ (ūz ūx ) + ρ (u0z u0x ) . we get for the first one the following expression:     ∂ ∂(ūx + u0x ) ∂ ∂ ūx −2µ = −2µ . after combining all three parts. (9.27) as Reynolds time-averaged .33) ∂y ∂y ∂x ∂y ∂y ∂x and for the third term this one:       ∂ ∂(ūz + u0z ) ∂(ūx + u0x ) ∂ ∂ ūz ∂ ūx −µ + = −µ + . It follows: ∂p ∂(p̄ + p0 ) ∂ p̄ − + ρgx = − + ρgx = − + ρgx .32) ∂x ∂x ∂x ∂x for the second one this expression: ∂(ūy + u0y )       ∂ ∂(ūx + u0x ) ∂ ∂ ūx ∂ ūy −µ + = −µ + . ∂y ∂z ∂z After sorting the terms.34) ∂z ∂x ∂z ∂z ∂x ∂z Finally.27):          ∂ ∂ ūx ∂ ∂ ūx ∂ ūy ∂ ∂ ūz ∂ ūx − −2µ + −µ + + −µ + . The term is given by:           ∂ ∂ux ∂ ∂ux ∂uy ∂ ∂uz ∂ux − −2µ + −µ + + −µ + ∂x ∂x ∂y ∂y ∂x ∂z ∂x ∂z | {z } | {z } | {z } Term 1 Term 2 Term 3 By analyzing term one to term three step by step. we get the following expression for the second term on the RHS of equation (9. Reynolds−Stress The second term on the RHS of equation (9.66 Turbulence Modeling Finally.27) will be discussed now. (9. ∂x ∂x ∂y ∂y ∂x ∂z ∂x ∂z After we analyzed the first two terms. we will investigate into the last two terms on the RHS of equation (9. (9.35) ∂x ∂x ∂x Now we can rewrite the x-component of the momentum equation (9. (9.27). we end up with:   ∂ ∂ ∂ − ρ (ūx ūx ) + ρ (ūy ūx ) + ρ (ūz ūx ) ∂x ∂y ∂z | {z } identical convective terms   ∂ ∂ ∂ − ρ (u0x u0x ) + ρ (u0y u0x ) + ρ (u0z u0x ) . ∂x ∂y ∂z | {z } additonal terms.

9.38) ∂ ∂ ūz ∂ ∂ ūx ∂ ūz − −2µ + −µ + ∂z ∂z ∂x ∂z ∂x    ∂ ∂ ūy ∂ ūz ∂ p̄ + −µ + − + ρgz ∂y ∂z ∂y ∂z If we are using the vector of fluctuations U0 (9.2.) . (9.11) and taking the convective terms to the LHS.37) ∂ ∂ ūy ∂ ∂ ūx ∂ ūy − −2µ + −µ + ∂y ∂y ∂x ∂y ∂x    ∂ ∂ ūz ∂ ūy ∂ p̄ + −µ + − + ρgy ∂z ∂y ∂z ∂y z-Component of Momentum   ∂ ∂ ∂ ∂ ρ ūz = − ρ ūx ūz + ρ ūy ūz + ρ ūz ūz ∂t ∂x ∂y ∂z   ∂ 0 0 ∂ 0 0 ∂ 0 0 − ρ (ux uz ) + ρ (uy uz ) + ρ (uz uz ) ∂x ∂y ∂z       (9. we get the Reynolds time-averaged momentum equation for the y and z components respectively. REYNOLDS TIME-AVERAGED EQUATIONS 67 x-momentum equation:   ∂ ∂ ∂ ∂ ρ ūx = − ρ ūx ūx + ρ ūy ūx + ρ ūz ūx ∂t ∂x ∂y ∂z   ∂ 0 0 ∂ 0 0 ∂ 0 0 − ρ (ux ux ) + ρ (uy ux ) + ρ (uz ux ) ∂x ∂y ∂z       ∂ ∂ ūx ∂ ∂ ūx ∂ ūy . we can rewrite the averaged momentum equation in vector form as: ∂ −ρ∇ • (U0 ⊗ U0 ) .36) − −2µ + −µ + ∂x ∂x ∂y ∂y ∂x    ∂ ∂ ūz ∂ ūx + −µ + ∂z ∂x ∂z ∂ p̄ − + ρgx ∂x If we apply the same procedure to the y and z component of the momentum equation. the definition of the deformation (strain) rate tensor D (5. y-Component of Momentum   ∂ ∂ ∂ ∂ ρ ūy = − ρ ūx ūy + ρ ūy ūy + ρ ūz ūy ∂t ∂x ∂y ∂z   ∂ ∂ ∂ − ρ (u0x u0y ) + ρ (u0y u0y ) + ρ (u0z u0y ) ∂x ∂y ∂z       .  ρ Ū + ρ∇ • (Ū ⊗ Ū) = ∇ • 2µD̄ −∇p̄ + ρg (9.25). These three Reynolds time-averaged equations are then called Reynolds-Averaged-Navier-Stokes equa- tions (RANS).13) (with ρ=const.39) ∂t | {z } | {z } τ̄ Reynolds−Stresses σ̄ t | {z } Same as equation (5. (9.

7).44)  ∂ ∂z u0z u0x u0z u0y u0z u0z  h ∂  ∂  ∂ i  − ρ ∂x u0x u0x + ρ ∂y u0x u0y + ρ ∂z u0x u0z    of x − mom. Hence. the shear-rate tensor and the pressure (6.   ∂  ∂  ∂   !  0 0 0 0 0 0  − hρ ∂x uy ux  + ρ ∂x uy uy  + ρ ∂x = uy uz i  = of y − mom. In most literatures we will find the Reynolds time-averaged momentum equations in Cartesian form using the Einsteins summation convention. the terms are equal.45)   ∂ ∂ ∂  − ρ ∂y u0z u0x + ρ ∂x u0z u0y + ρ ∂z u0 u0  of z − mom. (9. (9.  z z  As we can see – and already knew –. we will only investigate into that one. sometimes the Reynolds-Stress term is put into the convective term on the LHS.68 Turbulence Modeling D̄ defines the Reynolds-Averaged (mean) deformation rate tensor. This lead to the following equation: ∂ ∂ ∂ τ̄ij ∂ p̄ ∂  0 0 ρ ūi + ρ (ūi ūj ) = − + ρgi − ρ ui uj . The Reynolds-Stress tensor σ̄ t is defined as:     −ρu0x u0x − ρu0y u0x − ρu0z u0x σ̄txx σ̄tyx σ̄tzx σ̄ t = −ρu0i u0j = −ρu0x u0y − ρu0y u0y − ρu0z u0y  = σ̄txy σ̄tyy σ̄tzy  . we will use the relation between the Cauchy stress tensor.41) ∂t Finally. τ̄ the mean shear-rate tensor and the last term the Reynolds-Stresses. (9. Hence. denoted as Reynolds-Stress tensor σ̄ t . (9.46) ∂t ∂xj ∂xj ∂xi ∂xj Furthermore.2 we already showed and discussed that the vector form results in the Cartesian one. we end up with the following equation: ∂ ρ Ū + ρ∇ • (Ū ⊗ Ū) = ∇ • σ̄ + ρg + ∇ • σ̄ t .40)     −ρu0x u0z − ρu0y u0z − ρu0z u0z σ̄txz σ̄tyz σ̄tzz After we introduced the Reynolds-Stress tensor. we can rewrite the momentum equation in a more general form: ∂ ρ Ū + ρ∇ • (Ū ⊗ Ū) = ∇ • τ̄ − ∇p̄ + ρg + ∇ • σ̄ t . (9.43) 0 ∂ ∂z uz u0z  ∂    ∂x u0x u0x u0x u0y u0x u0z  ∂   0 0 = −ρ  ∂y  • uy ux u0y u0y u0y u0z  (9. The Reynolds-Stress term can be rewritten as:   0   0  ∂ u u   x0   x0   ∂x − ρ∇ • (U0 ⊗ U0 ) = ∂  −ρ  ∂y  •  u y  ⊗  uy    (9. In the above equation there is only one term left that we should transformed to demonstrate that each term of the vector form represents the corresponding term in the Cartesian equation.42) ∂t In section 2. . in many literatures we will find the greek symbol τ̄ t to express the Reynolds-Stress tensor – this is omitted here because otherwise we are not able to show the analogies between the real stress tensor σ (Cauchy stress tensor) and the Reynolds-Stress tensor σ t clearly.

Hence. we will derive the Reynolds time-averaged governing conserved equation (3. Note: It should be obvious that we can put the density in or out of the derivatives (it is a constant if we use the assumption of incompressibility). REYNOLDS TIME-AVERAGED EQUATIONS 69 Hence.4 The (Incompressible) General Conservation Equation After we derived the RANS equations. the derivation of all other conserved equations like the enthalpy. The dynamic viscosity will become the kinematic viscosity ν and the pressure is divided by the density. this formulation is also valid: ∂ ∂   ∂ τ̄ij ∂ p̄ ρūi + ρūi ūj + ρu0i u0j = − + ρgi . (9. As we already know.2. like different thermal diffusivity coefficients in the three space directions (otherwise the derivation get simplified and is not worth do show):   ∂ ∂ ∂ ∂ ρ φ = − ρ (ux φ) + ρ (uy φ) + ρ (uz φ) ∂t ∂x ∂y ∂z       ∂ ∂φ ∂ ∂φ ∂ ∂φ + Dx + Dy + Dz .2. If we still want to have a buoyancy term within the incompressible equations. that we are allowed to divide the equations by the density ρ.1) for incompressible fluids without any source terms.49) ∂t | {z } | {z } | {z } time accumulation convective transport diffusive transport To show the transformation to the Reynolds time-averaged equation. we could use a general conservation equation to derive other equations. the starting point is: ∂ ρ φ = −ρ ∇ • (Uφ) + ∇ • (D∇φ) . (9. this term gets constant and can be neglected because it will not change the momentum in any case. we get: ∂ ∂   ∂ τ̄ij ∂ p̄ ρ ūi + ρ ūi ūj + u0i u0j = − + ρgi . The boxed equations above are known as Reynolds-Averaged-Navier-Stokes equations (RANS).50) ∂x ∂x ∂y ∂y ∂z ∂z . 9. we will switch this equation into the Cartesian form first by using the mathematics (1. temperature or species equation can be done with the same procedure but is not demon- strated now. Furthermore we can think about the gravitational acceleration term ρgi . we need to use some models like the Boussinesq approximation. (9.47) ∂t ∂xj ∂xj ∂xi The derivation of the Reynolds-Averaged momentum equations are done.48) ∂t ∂xj ∂xj ∂xi In addition it is clear. Therefore. If the density is constant. For that.16) and assume that the diffusion coefficient D represents a vector.9. Hence. (9. we just have to be sure to have the right quantities for the pressure and the dynamic viscosity µ.

54) ∂y ∂y ∂   ∂ h i ρ (ūz + u0z )(φ̄ + φ0 ) = ρ (ūz φ̄ + ūz φ0 + u0z φ̄ + u0z φ0 ) ∂z ∂z ∂ ∂ =ρ (ūz φ̄) + ρ (u0z φ0 ) . the time term results in: ∂ ∂ ρ (φ̄ + φ0 ) = ρ φ̄ . ∂z will be split to enable analyzing term by term. (9. (9. ∂x ∂y ∂z ∂x ∂y ∂z The second.1) and apply it to φ. we will analyze each term separately of equation (9. the first term on the RHS after sorting is:   ∂ ∂ ∂ ∂ ∂ ∂ − ρ (ūx φ̄) + ρ (ūy φ̄) + ρ (ūz φ̄) + ρ (u0x φ0 ) + ρ (u0y φ0 ) + ρ (u0z φ0 ) .53) ∂x ∂x ∂   ∂ h i ρ (ūy + u0y )(φ̄ + φ0 ) = ρ (ūy φ̄ + ūy φ0 + u0y φ̄ + u0y φ0 ) ∂y ∂y ∂ ∂ =ρ (ūy φ̄) + ρ (u0y φ0 ) . (9.51). (9.70 Turbulence Modeling The next step is to use the expression from equation (9. It follows: ∂   ∂ h i ρ (ūx + u0x )(φ̄ + φ0 ) = ρ (ūx φ̄ + ūx φ0 + u0x φ̄ + u0x φ0 ) ∂x ∂x ∂ ∂ =ρ (ūx φ̄) + (ρu0x φ0 ) . uy and uz – keep in mind that D are constant values and are not influenced by the averaging procedure.55) ∂z ∂z Hence.56) ∂x ∂x ∂x ∂x .  ∂  ∂  (ūx + u0x )(φ̄ + φ0 ) + ρ (ūy + u0y )(φ̄ + φ0 )   − ρ ∂x ∂y  ∂  (ūz + u0z )(φ̄ + φ0 )  +ρ . Therefore. third and fourth term on the RHS will end up as:     ∂ ∂ ∂ ∂ φ̄ Dx (φ̄ + φ0 ) = Dx .51) ∂y ∂y ∂z ∂z To discuss the Reynolds-Averaging procedure. ux . (9.52) ∂t ∂t The first term on the RHS. It follows:  ∂ ∂  ∂  ρ (φ̄ + φ0 ) = − ρ (ūx + u0x )(φ̄ + φ0 ) + ρ (ūy + u0y )(φ̄ + φ0 )   ∂t ∂x ∂y    ∂  ∂ ∂ (ūz + u0z )(φ̄ + φ0 ) + (φ̄ + φ0 )  +ρ Dx ∂z ∂x ∂x     ∂ ∂ 0 ∂ ∂ + Dy (φ̄ + φ ) + Dz (φ̄ + φ0 ) . (9.

(9. (9.58) ∂z ∂z ∂z ∂z To sum up. Till today this problem is still not solved and we do not have a set of equations to get .3. we can put the convective and the turbulent scalar flux terms together. (9.60) allows us to derive each Reynolds-Averaged conservation equation by replacing φ with the quantity of interest (analogy to chapter 3).62) ∂t 9. The resulting general Reynolds-Averaged conservation equation is then written as: ∂   ρφ̄ + ∇ • ρŪφ̄ + ρU0 φ0 = ∇ • D∇φ̄ + Sφ . we need approximations that correlate the unknown with known quantities.61) |∂t {z } | {z } same as before turbulent scalar flux Manipulating the equations should be familiar now.9. Hence.3 The Closure Problem The Reynolds-Averaged procedure lead to the problem. As we already realized.59)   ∂ 0 0 ∂ 0 0 ∂ 0 0 − ρ (ux φ ) + ρ (uy φ ) + ρ (uz φ ) ∂x ∂y ∂z | {z } turbulent scalar flux and is given in vector form by: ∂ ρ∇ • (U0 φ0 )   ρ φ̄ = −ρ∇ • Ūφ̄ + ∇ • D∇φ̄ − . Therefore. In other words. THE CLOSURE PROBLEM 71     ∂ ∂ ∂ ∂ φ̄ Dy (φ̄ + φ0 ) = Dy .60) | ∂t {z } | {z } same as before turbulent scalar flux Equation (9.57) ∂y ∂y ∂y ∂y     ∂ ∂ ∂ ∂ φ̄ Dz (φ̄ + φ0 ) = Dz . the set of equations are not enough to close our problem and we cannot solve our system. we can put the density inside the derivations and we end up with: ∂ ∇ • (ρU0 φ0 )   ρφ̄ = −∇ • ρŪφ̄ + ∇ • D∇φ̄ − . after deriving the Reynolds-Averaged momentum equations. Finally – and if we want –. (9. the terms −ρu0i u0j and −ρu0i φ0 are not known and cannot be calculated. we get the same equations but with additional terms. Thus.  (9. In addition we add the arbitrary source term of φ. that we create additional unknown quan- tities and no further equations. the general Reynolds-Averaged conservation equation can be written as:   ∂ ∂ ∂ ∂ ρ φ̄ = − ρ (ρūx φ̄) + ρ (ūy φ̄) + ρ (ūz φ̄) ∂t ∂x ∂y ∂z       ∂ ∂ φ̄ ∂ ∂ φ̄ ∂ ∂ φ̄ + Dx + Dy + Dz ∂x ∂x ∂y ∂y ∂z ∂z . This is known as closure problem. (9. These additional terms are called Reynolds- Stresses for the momentum equations and are named additional turbulent scalar flux for all other quantities.

64) 3 3 2 σ̄ t = τ̄ t − ρIk . the flow gets more chaotic and we get a lot of vortexes that can transport for example heat in addition to the already existing transport phenomena. In addition we can say that the turbulent Reynolds-Stress tensor σ̄ t equals to the shear-rate tensor τ¯t and an additional term 23 ρδij k. Bird et al.14) with the difference that we use the turbulent eddy viscosity µt instead of the molecular viscosity µ. Further information about these theories (concept of higher viscosity) can be found in Ferziger and Perić [2008]. Keywords: energy cascade. we also could describe the theory vice versa: the higher the eddy viscosity. Wilcox [1994]. In other words. cf. 9. we are forced to use approximations.4 Boussinesq Eddy Viscosity The most used hypothesis is the theory postulated by Joseph Boussinesq that simply relates the turbulence of a flow to a higher fluid viscosity. Therefore. (9. we try to correlate the unknown quantities with known one. The equations that are introduced by authors to get rid of the closure problem are known as turbulence models. It is also possible to use the higher viscosity to describe or characterize the dissipation of kinetic energy (per unit mass) of the turbulence into heat – the higher the viscosity of the fluid. [1960]. it is clear and of humans nature to say.65) 3 In the equations above. the higher the turbulence of the flow field. Joseph Boussinesq related the Reynold-Stresses −ρu0i u0j to the mean values of the velocities and the kinetic energy of the turbulence k as:   ∂ ūi ∂ ūj 2  2 − ρu0i u0j = µt + − ∇ • Ū δij − ρδij k . (9. higher viscosity concept. Hence. The most popular methods are the Boussinesq’s eddy viscosity.63) ∂xj ∂xi 3 3 | {z } | {z } | {z } σ̄ t ¯ 2 ρIk 2D̄− 2 3 tr(D)I 3 2 2 σ̄ t = 2µt D̄ − µt tr(D̄)I − ρIk . the higher the shearing and therefore we get higher mixing rates (additional transport) but although a larger dissipation of the kinetic energy into heat. This assumption give us the possibility to model the smallest vortexes by using correlations and approximations and only resolve the larger eddies.72 Turbulence Modeling rid of the closure problem and therefore. hence we mark it with the subscript t (τ̄ t ). we can see. if we use the Reynolds time-averaging procedure. if we increase the diffusion coefficient (the viscosity in the momentum equation) and keep the rest as it is. The thought behind is as follows: If we have a higher turbulence flow. For the Reynolds-Stresses and turbulent scalar fluxes we can use several theories that try approximate the unknown terms. if we set the bulk viscosity κ to zero. eddy viscosity. Within this assumptions. the molecular viscosity is increased by the so called eddy or turbulent viscosity. (9. the Prandtl’s mixing length or the Von-Kármán’s similarity hypothesis. This term is necessary to guarantee the proper trace of the Reynolds-stress tensor σ̄ t as mentioned by Ferziger and Perić [2008] and Wilcox . equation (5. that the underlined term is identical to the shear-rate tensor τ̄ . that we could achieve that. dissipation and turbulent viscosity.

we simply have to take the trace of the Reynolds-Stress tensor σ̄ t and the shear-rate tensor τ̄ t . the trace of the RHS of equation (9. BOUSSINESQ EDDY VISCOSITY 73 [1994].11) with respect to the mean quantities . For that.4. (9. that the trace of the Reynolds-Stress σ̄ t tensor has to be equal to −2ρk.65) has to be equal to −2ρk too.68) | {z } = −2ρk The result that we get is the following: The trace of the Reynolds-Stress tensor is twice the density multiplied by the kinetic energy of the turbulence. Thus we get:   2 tr(σ̄ t ) = tr τ̄ t − ρIk = −2ρk . (9. It follows:   −ρu0x u0x − ρu0y u0x − ρu0z u0x tr(σ̄ t ) = tr(−ρu0i u0j ) = tr −ρu0x u0y − ρu0y u0y − ρu0z u0y  . Therefore. Where does it come from and why is this term necessary? As mentioned by Ferziger and Perić [2008] and Wilcox [1994].67)   −ρu0x u0z − ρu0y u0z − ρu0z u0z = (−ρu0x u0x ) + (−ρu0y u0y ) + (−ρu0z u0z ) . 2 One may think about the term 3 ρδij k now.70) 3   2 2 tr(σ̄ t ) = tr 2µt D̄ − µt tr(D̄)I − ρIk = −2ρk .66) 2 i i 2 New we need to take the trace of the Reynolds-Stress tensor σ̄ t (9. (9. (9. (9.9. (9. this term has to be added to get the proper trace of the Reynolds-stress tensor.71) 3 3 If we use the definition of the deformation rate tensor (5. To understand the meaning of the additional term. Otherwise the equation would not be correct in the mathematical point of view.66):   1 0 0  − 2ρk = −2ρ ux ux + u0y u0y + u0z u0z = (−ρu0x u0x ) + (−ρu0y u0y ) + (−ρu0z u0z ) . we need to know the definition of the kinetic energy of the turbulence: 1 0 0 1 0 0  k= uu = ux ux + u0y u0y + u0z u0z . and can be validated by substituting k by its definition (9. −2ρk.69) 2 | {z } tr(σ̄ t ) We demonstrated.40).

74 Turbulence Modeling and apply the transformation (8.11) to term 1.9). add both matrices and multiply everything by the eddy viscosity. it follows:   h   1 T i 2  2 tr(σ̄ t ) = tr 2µt ∇ ⊗ Ū + ∇ ⊗ Ū − µt ∇ • Ū I − ρIk . If we would remove the term − 23 ρδij k on the RHS of equation (9. Hence.63).63) is equal. the trace of the RHS and LHS of equation (9. The matrices of term 1. (9. term 2 and term 3 have to be summed up and the trace operator has to be applied. chapter 6.73) 3 | {z } Term 3 The result of the trace operator to the Boussinesq hypothesis is −2ρk.  ∂y ∂y ∂z ∂u µt ∂u ∂z x + µt ∂u ∂x z µt ∂zy + µt ∂u ∂y z µt ∂u ∂z z + µt ∂u ∂z z Due to the fact that we are only interested in the main diagonal elements (trace). . It follows:       2 ∂ ūx ∂ ūy ∂ ūz 2 ∂ ūx ∂ ūy ∂ ūz tr τ̄ t − ρIk = 2µt + + − 3 µt + + 3 ∂x ∂y ∂z 3 ∂x ∂y ∂z | {z } | {z } Term 1 Term 2 | {z } = 0 2 − 3 ρk = −2ρk . the Boussinesq eddy viscosity assumption would be wrong because τ̄ t is traceless. the trace of the RHS would not be equal to the trace of the Reynolds-Stress tensor σ̄ t and hence. 2 3 3 | {z } | {z } | {z } Term 1 Term 2 Term 3    ∂     ∂   T     ∂x ūx  ∂x  ūx   ∂    ∂     = tr µt  ⊗ ū + ⊗ ūy   y     ∂y      ∂y     ∂  ∂ ūz ūz    ∂z ∂z | {z } Term 1  ∂ ū x ∂ ūy ∂ ūz  ∂x + ∂y + ∂z 0 0 2  ∂ ūx ∂ ūy ∂ ūz − µt  0 + + 0  3 ∂x ∂y ∂z  ∂ ūx ∂ ūy ∂ ūz 0 0 ∂x + ∂y + ∂z | {z } Term 2    − 32 ρk 0 0   − 0 − 32 ρk 0  . we end up with term 1 as: ∂uy µt ∂u + µt ∂u + µt ∂u µt ∂u + µt ∂u   x x x z x ∂x ∂x µt ∂x ∂y ∂x ∂z  ∂ux ∂u ∂u ∂uy ∂uy µt ∂y + µt ∂xy µt ∂yy + µt µt ∂u z + µt  . cf. we just consider these terms for now.72)  0 0 − 23 ρk   | {z } Term 3 Applying the dyadic product rule (1. (9.

Keep in mind that this thread can cause confusion because only the last posts are correct and as Gerhard Holzinger mentioned. (9. the term is put into a modified pressure and is not neglected in OpenFOAMr .79) ∂t | {z3 } = p∗ I Introducing a modified pressure p∗ = p̄ + 32 ρk and replacing the shear-rate tensors by their defi- nitions (for the shear-rate tensor τ̄ . (9.42) and replacing the Reynolds-Averaged Cauchy stress tensor σ̄ and the Reynolds- Stress tensor σ̄ t by their definitions (6. we can highlight the similarities better:   ∂ 2 ρ Ū + ρ∇ • (Ū ⊗ Ū) = ∇ • τ̄ + ∇ • (−p̄I) +ρg + ∇ • τ̄ t + ∇ • − ρkI . (9.7) and (9.com/Forums/openfoam-solving/58214-calculating-divdevreff. html.7) and (9. you can go to: www.65).com. Even I made wrong statements at the beginning in a way that within the OpenFOAMr toolbox.4.65).9. Analogy to the Cauchy Stress Tensor. part (=trace) If we compare the terms.77) ∂t 3 | {z } | {z } ∇•σ̄ ∇•σ̄ t   ∂ 2 ρ Ū + ρ∇ • (Ū ⊗ Ū) = ∇ • τ̄ + ∇ • τ̄ t + ρg + ∇ • (−p̄I) + ∇ • − ρkI . we get the following equation:   ∂ 2  ρ Ū + ρ∇ • (Ū ⊗ Ū) = ∇ • 2µl D̄ − µl ∇ • Ū I ∂t 3   2 + ρg − ∇ • (p∗ I) . (9. (9. Analyzing equation (6.cfd-online. we can evaluate the same kind of behavior: σ = τ + −pI . this term is neglected. Shear-Rate Tensor and Pressure Comparing the last derived equations with those of chapter 6. term (=trace) dev hyd A |{z} = A | {z } + A | {z } . BOUSSINESQ EDDY VISCOSITY 75 Forums Discussion It is worth to mention that there were a lot of people asking about the term − 23 ρδij k in public forums.80) 3 . it is obvious that there are similar- ities. Finally. How this is working. we can observe that the term − 32 ρk seems to behave like a pressure. is given on the next page. Using equation (9.78) ∂t 3   ∂ 2 ρ Ū + ρ∇ • (Ū ⊗ Ū) = ∇ • (τ̄ + τ̄ t ) + ρg − ∇ • p̄I + ρkI . For those who are interested in the discussion on cfd-online.76) complete matrix deviatoric part (traceless) hydro. we cannot find this term in OpenFOAMr which is related to a very simple correlation that is given below.74) |{z} |{z} |{z} pressure (=trace) (Cauchy)−Stress tensor shear−rate tensor (traceless) 2 σ̄ t = τ̄ t + − ρIk .  + 2µt D̄ − µt ∇ • Ū I (9. we mark the molecular viscosity µ by the subscript l ).75) |{z} |{z} | 3{z } (Reynolds)−Stress tensor (RA)−shear−rate tensor (traceless) add. (9.

26).81) 3    ∂ ¯ − 2 ∇ • Ū I + ρg − ∇ • (p∗ I) . Instead we have the modified pressure p∗ . If we introduce an effective Cauchy-Stress tensor σ̄ eff . Only if we are using some modified equations in OpenFOAMr where we need the real pressure.  ρŪ + ∇ • (ρŪ ⊗ Ū) = ∇ • [µl + µt ] 2D (9. we can simplify the equation to: ∂ ρŪ + ∇ • (ρŪ ⊗ Ū) = ∇ • τ̄ eff − ∇ • (p∗ I) + ρg . the only unknown in that equation is the eddy viscosity µt . The differences within the equations are. (9.83) we can rewrite the Reynolds-Averaged-Navier-Stokes equations. we are not calculating the real pressure field p. it should be obvious that we have mean quantities here.88) ∂t Note: If we solve the incompressible Reynolds-Averaged Navier-Stokes equations.  + µt 2D (9. we are able to build the general form of the momentum equation.  ρŪ + ∇ • (ρŪ ⊗ Ū) = ∇ • µeff 2D (9. . The reason for introducing the Boussinesq eddy hypothesis and the advantages are given in the next section.  ρŪ + ∇ • (ρŪ ⊗ Ū) = ∇ • 2µeff D̄ − µeff ∇ • Ū I (9. (9.84) ∂t 3   ∂ 2 + ρg − ∇ • (p∗ I) . that include the effective viscosity and a modified pressure field p∗ as:    ∂ ¯ − 2 ∇ • Ū I + ρg − ∇ • (p∗ I) . After that. For most of the problems this is not a big deal and we do not have to consider this. (9.76 Turbulence Modeling    ∂ ¯ − 2 ∇ • Ū I  ρ Ū + ρ∇ • (Ū ⊗ Ū) = ∇ • µl 2D ∂t 3    ¯ − 2 ∇ • Ū I + ρg − ∇ • (p∗ I) . we have to recalculate the real pressure field by subtracting the kinetic part.86) ∂t It is clear that this equation is similar to equation (2. It follows: σ̄ eff = τ̄ eff − p∗ I .85) ∂t | 3 {z } τ̄ eff After introducing the effective shear-rate tensor τ̄ eff . (9.87) ∂ ρŪ + ∇ • (ρŪ ⊗ Ū) = ∇ • σ̄ eff + ρg . that we use a modified pressure p∗ and a new viscosity µeff field.82) ∂t 3 Introducing an effective viscosity µeff that is simply the sum of the molecular and turbulent (eddy) viscosity: µeff = µl + µt .

or even flow separation). To get the equation of the kinetic energy of the turbulence k. no source terms and incompressibility (dilatation term is zero).70). the characteristic length L and the eddy viscosity. the kinetic energy of the turbulence √ can be related to a velocity q = k. it is logical to use two-equation models. (9. these models will fail and produces non-physical values for the eddy viscosity. we will use the Navier-Stokes equation (5. we get new unknown quantities like the eddy viscosity µt and the kinetic energy of the turbulence k. we simply have to take the trace of the Reynolds-Stress equation. The challenge now is to relate the characteristic length L and the velocity q to known quantities.6 Algebraic Models At the beginning of computational fluid dynamics the power of personal and super computers were restricted and therefore it was necessary to have simple models that approximate the Reynolds- Stresses −ρu0i u0j . 9.5. Commonly the turbulent (eddy) viscosity is characterized with the kinetic energy of the turbulence k and a characteristic length L.9. Furthermore. 9. where each equation models one parameter.8 Incompressible Reynolds-Stress Equation To derive the Reynolds-Stress equation. To evaluate k we can make use of the already know relation between the trace of the Reynolds-Stress tensor σ̄ t and k. This is done by using turbulence models. we focus only on this kind of approximations for now. How- ever.89) The parameter Cµ is a dimensionless constant. In general the velocity q is calculated using the kinetic energy of the turbulence k. The estimation of the eddy viscosity µt is done by using algebraic expressions. The assumption that was invented is: µt ≈ Cµ ρqL .7 Turbulence Energy Equation Models The most common approximations for the Reynolds-Stresses (finally to calculate the characteristic length scale L and the kinetic energy of the turbulence k) are called turbulence energy equation models. A few models are described in Wilcox [1994] chapter 3. How we get this equations are discussed in the following sections. This two values enable us to derive a correlation between the velocity q (kinetic energy of the turbulence k). 9. These models commonly use the introduced Boussinesq eddy viscosity theory. There are one-equation and two-equation models. Therefore. (9. The start . cf.5 Eddy Viscosity Approximation The Boussinesq theory allows us to eliminate the Reynolds-Stresses with known quantities. EDDY VISCOSITY APPROXIMATION 77 9.10) with bulk viscosity equals to zero. Algebraic models can be used for simple flow patterns but hence the flow is getting complex (imagine geometries in combustion. Due to the fact that we need the √ values of the velocity q (= k) and the characteristic length L. Wilcox [1994] listed plenty of theories and models that relates the eddy viscosity to known quantities.

we will give a brief summary of the operations and relations and present the Reynolds-Stress equation without any derivation.9). (9.92) ∂t ∂xj ∂xj 2 ∂xj ∂xi ∂xi It is clear that the Navier-Stokes operator N has to be equal to zero and thus we can write: N (ui ) = 0 . Operations and relations For the derivation of the Reynolds-Stress equation we need to build the following tensor equation with formula (9.78 Turbulence Modeling point for the derivation is:     ∂ui ∂ ∂ 1 ∂ui ∂uj ∂p ρ +ρ (uj ui ) = 2µ + − .93) With the new information. . we are able to derive the Reynolds-Stress equation. put everything to the LHS and introduce the Navier-Stokes operator N . Hence. (9.1.1) to the Navier-Stokes operator N (ui ) and average the whole equation by using the Reynolds time-averaging method (9. (9.95): u0x N (ūx + u0x ) + u0y N (ūz + u0z ) + u0x N (ūy + u0y ) + u0y N (ūx + u0x ) +u0x N (ūz + u0z ) + u0y N (ūy + u0y ) + u0y N (ūx + u0x ) + u0z N (ūz + u0z ) +u0y N (ūy + u0y ) + u0z N (ūx + u0x ) + u0y N (ūz + u0z ) + u0z N (ūy + u0y ) +u0z N (ūx + u0x ) + u0x N (ūz + u0z ) + u0z N (ūy + u0y ) + u0x N (ūx + u0x ) +u0z N (ūz + u0z ) + u0x N (ūy + u0y ) = 0 . The full derivation of this tensor equation is given in the appendix in section 14.95) This equation has to be evaluated to get the Reynolds-Stress equation. we need to be familiar with the mathematics.90) ∂t ∂xj ∂xj 2 ∂xj ∂xi ∂xi To make life easier. We get: ∂ ∂ui ∂u   j ρ (uj ui ) = ρuj + ρui .91) ∂xj ∂xj ∂xj | {z } continuity Replacing the convective term with the new form. (9.94) The next step is to apply the expression (9. In the following. The derivation itself is not a big deal but we have to have the feeling for different behaviors of the terms and hence. In order to form this equation we multiply the Navier-Stokes operator by the fluctuation with respect to the different space directions: u0i N (uj ) + u0j N (ui ) = 0 . (9. This leads to the following equation: u0i N (ūj + u0j ) + u0j N (ūi + u0i ) = 0 . we get:     ∂ui ∂ui ∂ 1 ∂ui ∂uj ∂p ρ + ρuj − 2µ + + = N (ui ) . we will split the convective term by using the product rule. we can remove one term due to the continuity equation (non-conserved equation). (9.

we are able to derive the equation for the kinetic energy of the turbulence k. we have to take the trace of equation (9.66).9. the Reynolds-Stress equation is given as: ∂ σ̄tji ∂ σ̄tij ∂ σ̄tij   ∂ ūi ∂ ūj ∂ + ūk = −σ̄tjk − σ̄tik + ν + Cijk + ij − Πij . If we use these assumptions. ∂u0i • The derivative ∂xi =0. In order to get the equation.98) ∂t | {z } ∂t (9.97) ∂xk ∂xk ∂xk ∂xk After applying the trace operator to each term we get: For the time derivation we get: ∂ σ̄tji ∂u0y u0y   ∂ σ̄txx ∂ σ̄tyy ∂ σ̄tzz ∂u0 u0 ∂u0 u0 tr = + + = −ρ x x − ρ −ρ z z ∂t ∂t ∂t ∂t ∂t ∂t ∂t ∂ ∂k = −ρ (u0x u0x + u0y u0y + u0z u0z ) = −2ρ .96) ∂t ∂uk ∂xk ∂xk ∂xk ∂xk After knowing the Reynolds-Stress equation. (9.96).2) . THE INCOMPRESSIBLE KINETIC ENERGY EQUATION 79 For the derivation we further use the following relations. Hence. The result of the derivation procedure is a more or less complex equation. g(x) = g(x) + f (x) − f (x) .66)→2k . we can derive the Reynolds-Stress equation.9. rules and tricks: • Reynolds time-averaged terms that are linear in the fluctuation are zero . The Reynolds-Stress equation also gives insight into the nature of the turbulent stresses and can be used to understand the turbulence in more detail or it can be used for further investigations in deriving more accurate turbulence models.9 The Incompressible Kinetic Energy Equation The derivation of the kinetic energy equation of the turbulence (per unit mass) for incompressible flows is simple after knowing the Reynolds-Stress equation due to the fact of the relation given by equation (9. (9. (9. • Product rule (1. It follows: ∂ σ̄tji ∂ σ̄tij   tr + ūk ∂t ∂uk ∂ σ̄tij     ∂ ūi ∂ ūj ∂ = tr −σ̄tjk − σ̄tik + ν + Cijk + ij − Πij . 9. • Adding and subtracting terms to be able to use the product rule.

103)   ρu0x u0z u0x ρu0x u0z u0y ρu0x u0z u0z In a similar way to Cijk . (9. Cijk . (9. (9. the term ij can be manipulated. results in:  i ∂ 0 0 0 ∂ h 0 0 0 ∂ ∂ 0 0 tr ρu u u + 0 p uj δik + p ui δjk = ρu0 u0 u0 + 2 p uj .104) ∂xk ∂xk ∂xk ∂xk . The first term is a third rank tensor T3 and the trace results in the underlined term on the RHS.100) ∂xk ∂xk ∂xk ∂xk ∂xk The first part of the third term results in: ∂ σ̄tij          ∂ ∂ ∂ σ̄txx ∂ ∂ σ̄tyy ∂ ∂ σ̄tzz tr ν = ν + ν + ν ∂xk ∂xk ∂xk ∂xk ∂xk ∂xk ∂xk ∂xk       ∂ ∂u0x u0x ∂ ∂u0x u0x ∂ ∂u0x u0x =− ρν − ρν − ρν ∂xk ∂xk ∂xk ∂xk ∂xk ∂xk   ∂ ∂ =− µ (u0x u0x + u0y u0y + u0z u0z ) ∂xk ∂xk   ∂ ∂k = −2 µ .99) ∂uk ∂uk The first and second term of equation (9. (9.80 Turbulence Modeling If we apply the trace operator to the convective term. (9.96) lead to:   ∂ ūi ∂ ūj ∂ ūi ∂ ūi ∂ ūi tr −σ̄tjk − σ̄tik = −σ̄tik − σ̄tik = 2ρu0i u0k . we get: ( ) ∂u0 ∂u0j ∂u0i ∂u0i tr 2µ i = 2µ . It is simply twice the trace of one of the terms.101) ∂xk ∂xk The second part of the third term. we get: ∂ σ̄tij   ∂ σ̄txx ∂ σ̄tyy ∂ σ̄tzz tr ūk = ūk + ūk + ūk ∂uk ∂uk ∂uk ∂uk ∂u0x u0x ∂u0y u0y ∂u0 u0 = −ρūk − ρūk − ρūk z z ∂uk ∂uk ∂uk ∂ ∂k = −ρūk (u0x u0x + u0y u0y + u0z u0z ) = −2ρūk .102) ∂xk i j k ∂xk ∂xj j i i ∂xj The evaluation of the second term that includes the pressure can be done in an easy way. Thus. (9. This can be demonstrated by analyzing the first entries of the third rank tensor:     ρu0x u0x u0x ρu0x u0x u0y ρu0x u0x u0z tr ρu0x u0j u0k = tr  ρu0x u0y u0x ρu0x u0y u0y ρu0x u0y u0z  = ρu0x u0i u0i .

Prt denotes the turbulent Prandtl number and is assumed to be one.9. (9.107) ∂xk 2 ∂xj ∂xk ∂xk | {z } | {z } turbulent diffusion dissipation  If we merge the terms that include the turbulent diffusion. we end up with the common kinetic energy equation as:   ∂k ∂k ∂ ∂k ∂ hρ 0 0 0 i ρ + ρūk = µ + Pk − uj ui ui + p0 u0j −  . we get the kinetic energy equation of the turbulence. For the turbulent diffusion we use the assumption that the diffusion is based on the gradients: hρ i µt ∂k − u0j u0i u0i + p0 u0j ≈ .108) ∂t ∂uk ∂xk ∂xk ∂xk 2 The production rate and turbulent diffusion term has to be modeled. z y (9. THE INCOMPRESSIBLE KINETIC ENERGY EQUATION 81 ∂u0i The last term of equation (9. is zero due to the fact that ∂xi = 0: (  ) ∂u0j  0 ∂u0i ∂u0x  0 ∂ux tr p + = p0 + ∂xj ∂xi ∂x ∂x " # 0 0  0 0  ∂u ∂u + p0  +  + p0 ∂uz + ∂uz = 0 . (9.96). The production rate is modeled with the assumption given by equation (9. Recall: This term was just added to equilibrate both sides. (9.9. use the term Pk . Due to the fact that we use sigma to describe any kind of stress tensor. In the literature we it is also common to denote the turbulent Prandtl number by σk . k.105) ∂y ∂y ∂z ∂z If we sum up all terms. (9.63) but with the difference. Hence. that denotes the production rate of the kinetic energy and the acronym . we avoid the usage of another sigma quantity here. Πij . we divide the whole equation by −2 to get the final form of the incompressible kinetic energy equation:   ∂k ∂k ∂ ūi ∂ ∂k ρ + ρūk = −ρu0i u0k + µ ∂t ∂uk ∂xk ∂xk ∂xk | {z } Pk ∂ ρ 0 0 0 ∂ 0 0 ∂u0 ∂u0i − uj ui ui − p uj − µ i . for incompressible fluids:   ∂k ∂k ∂ ūi ∂ ∂k − 2ρ − 2ρūk = 2ρu0i u0k −2 µ ∂t ∂uk ∂xk ∂xk ∂xk ∂ ∂ 0 0 ∂u0 ∂u0i + ρu0j u0i u0i + 2 p ui + 2µ i . that we do not need the term −2ρk.109) 2 Prt ∂xj Here.110) ∂xj ∂xj ∂xi ∂xj . the production rate term is given by:   ∂ ūi ∂ ūi ∂ ūj ∂ ūi Pk = −ρu0i u0j ≈ µt + .106) ∂xk ∂xk ∂xk ∂xk Finally. (9.

the dissipation .112) ∂t ∂xj k k ∂xj σ ∂xj As we can see. the length scale L and the dissipation : 3 k2 ≈ . is coupled to a characteristic length scale L. both quantities can be related.82 Turbulence Modeling The dissipation . we eliminated one unknown (q = k) by using the kinetic energy k but we also introduced a new unknown quantity. the length scale L and the dissipation . But the equation allows us to estimate the dissipation . Thus. The good thing is. Hence. This equation can be derived by using the Navier-Stokes equation (like we did for the Reynolds-Stress equation) but due to the fact that most terms on the RHS have to be modeled. The quantity can be used for the kinetic energy equation and allows us to estimate the length scale L and hence. it is possible to derive a relation between the kinetic energy k. 9. (9.89) with the new quantities: √ √ √ k3 k2 µt = ρCµ kL = ρCµ k = ρCµ . we need the equation for the dissipation .11 The Equation for the Dissipation Rate  To calculate the length scale L and close the equation for the turbulent energy k. the whole right side is more like a playground of parameters and assumptions than a real fundamental equation. the viscous effect will transfer the energy of motion into internal energy. we still have two unknown. Up √ to now. Now we can rewrite the Boussinesq eddy hypothesis (9. the complete derivation is not shown.10 The Relation between  and L The most common equation that is used to estimate the length scale L is based on the observation that the dissipation phenomena can also be observed in the energy transport and thus in its equation. In general we are using the following equation for the dissipation : 2   ∂ ∂  ∂ µt ∂ ρ + ρūj = C1 Pk − ρC2 + (9. we are able to approximate the the eddy viscosity µt .111) L The idea behind this relation is the so called energy cascade for high turbulent flow fields (high Reynolds numbers). To get this value we introduced the Boussinesq eddy viscosity hypothesis and related the eddy viscosity µt to a characteristic length scale L and a velocity q. 9. we should describe this equation more like a model than an exact equation. that describe the transfer of the turbulence into internal energy (a better description given in the next section). In fluid flows which are in a so called turbulent equilibrium. This phenomena is called dissipation.113)   . The concept can be described as follows: The kinetic energy of the turbulence is transformed from big scale eddies to small scale eddies. Recall: After we derived the RANS equations. If we reach the smallest scale (this vortexes are named Kolmogorov vortexes). we figured out that we need to calculate the Reynolds-Stress tensor σ̄ t . (9.

9. Another example would be the turbulence modeling of flow separation. x)dτ . In the literature we will find different equations that give reasonable results for a special kind of problem(s). the length scale L. we can imagine. The easiest equations for the turbulence modeling were derived. The section about the turbulence model gave us a feeling that the topic about turbulent flows are extreme complex. x)ui (t.9. the varying density has to be taken into account during for compressible fluids. compare the already derived Reynolds time-averaged compressible mass conservation equation (9. we observed that all parameters are coupled. Therefore. that are vanishing for incom- pressible fluids. Furthermore. that things get even worse.114) This lead to more unknown terms that will make the problem even more complex. Good references for further investigations into the turbulence modeling are the books of Ferziger and Perić [2008]. (9. If we go further and think about the momentum equations. (9. ρ̄ denotes the Reynolds time-averaged density and the tilde above the velocity ui marks . Bird et al. To get rid of the additional correlation between ρ0 and u0i . we introduce a mathematical method suggested by Favre. we get: ρ = ρ̄ + ρ0 .12 Coupling of the Parameters As we could see in the last sections.09 . COUPLING OF THE PARAMETERS 83 The model parameters of the equations above are: Cµ = 0. C1 = 1.92 .115) ∂t ∂xi Here we need approximations for the correlation between ρ0 and u0i .44 . 9.12. that is defined as: Z t+T 1 ũi = lim ρ(t. Just think about the turbulence behavior close to the walls compared to the far field. σ = 1.116) ρ̄ T →∞ t Here. The kinetic energy of the turbulence k. (9. There are a lot of more considerations that have to be taken into account if turbulence modeling is used. What we do is simple.24) which is given again: ∂ ρ̄ ∂ + (ρ̄ūi + ρ0 u0i ) = 0 . the turbulence modeling is a complex topic. the dissipation  and the eddy viscosity µt . This concept is based on mathematics and therefore not physical correct.3 . C2 = 1.13 Turbulence Modeling for Compressible Fluids As already discussed during the Reynolds averaging procedure for the incompressible mass conser- vation equation. A lot of research was done and till today the turbulence has still to be modeled and can only be applied and resolved with all details for a couple of problems. [1960] and Wilcox [1994]. We introduce a mass-averaged velocity field ũi .

.84 Turbulence Modeling the quantity to be Favre averaged instead of Reynolds averaged.119) ∂t ∂xi This equation looks just similar to the laminar mass conservation or the Reynolds time-averaged equation. as the depended variable rather than the velocity.118) If we use this expression for equation (9. Hence. This is a sensible thing to do from a physical point of view.. In terms of the Reynolds time- averaging procedure we are allowed to say: ρ̄ũi = ρui . Note: The key argument to use the Favre averaging procedure is to simplify the averaging procedure and get rid of additional correlations that have to be modeled. ρ0 (9. [1960]. a lot of information can be found in Wilcox [1994] and Bird et al.115). Wilcox [1994] explained this kind of averaging as follows:  What we have done is treat the momentum per unit volume. ρui . (9. we end up with: ∂ ρ̄ ∂ + (ρ̄ũi ) = 0 ..117) To show what happens here. we end up with the same set of equations for incompressible turbulent flow fields but now we use the Favre- weighted quantities. (9. . If you are interested into that kind of averaging. we will expand the RHS: ρ̄ũi = (ρ̄ + ρ0 )(ūi + u0i ) = ρ̄ūi + ρ̄u0i +  ūi + ρ0 u0i = ρ̄ūi + ρ0 u0i .

Chapter 10 Calculation of the Shear-Rate Tensor in OpenFOAMr In this chapter we will discuss the implementation of the calculation of the shear-rate tensor τ . Reynolds-averaged or Favre-averaged quantities. 10.1) 85 . For now we will focus on the term (IV) turbulence->divDevReff(U). U ) ( II ) 5 + MRF . you have to understand the concept of hacking. some additional correction term due to MRF (III) is added.> divDevReff ( U ) ( IV ) 7 == 8 fvOptions ( U ) (V) 9 ). Shear-Rate Tensor. The first one is the time derivation (I). real. τ̄ t or τ̃ t .H Lets analyze the code. The next one is the the shear-rate tensor (IV) and finally we have a term (V) that handles additional sources within the fvOptions dictionary. If you are asking yourself for that reason. the momentum equation that is going to be constructed in the UEqn.1. For those who are interested you can read the book of Erickson [2014]. The following discussion is based on the OpenFOAMr 2. DDt ( U ) ( III ) 6 + turbulence . First of all. Listing 10. divDevReff For incompressible fluids.1: $FOAM SOLVERS/incompressible/pimpleFoam/UEqn.H file looks equal to the following code (snippet from pimpleFoam): 1 tmp < fvVectorMatrix > UEqn 2 ( 3 fvm :: ddt ( U ) (I) 4 + fvm :: div ( phi .3. In OpenFOAMr we calculate the shear-rate tensor by calling the functions divDevReff or divDevRhoReff. Be prepared that the code can look different for other versions. we will discuss the meaning of the name divDevReff: ∇ • τ = ∇ • σ dev (10. The second term is the convective term (II). After that. First of all we observe different terms.1 The Inco.

Furthermore. U ) 6 .. Therefore.C First.2: . 8 } Listing 10. a new function named dev() is called.. which indicates incompressible equations. The R comes from the Reynolds-Average approach. Finally we are interested in the effective transport which includes laminar and turbulent transport phenomena. The analyze of the function is very simple and can be checked with the code source guide named Doxygen. we get the name divDevReff for incompressible and divDevRhoReff for compressible fluids. The second thing that can be observed is the fact. 86 Calculation of the Shear-Rate Tensor in OpenFOAMr The divergence of the shear-rate tensor is equal to the divergence of the deviatoric part of the stress tensor σ. we can see that we have the kinematic viscosity included. The object named turbulence. will call the function divDevReff(U)./turbulenceModel/incompressible/RAS/laminar/laminar. An additional Rho defines if we are using a density based or non density based solver.fvc :: div ( nuEff () * dev ( T ( fvc :: grad ( U ) ) ) ) 7 ). which is based on the general turbulenceModel class and further derived from the incompressibleTurbulenceModel class. that we calculate one part implicit (fvm) and another part explicit (fvc). .fvm :: laplacian ( nuEff () . The corresponding function that is called and implemented is: 1 tmp < fvVectorMatrix > laminar :: divDevReff ( volVecto rField & U ) const 2 { 3 return 4 ( 5 .

we can rewrite the code into a mathematic form. cf. we can write: 1 Adev = A − Ahyd = A − tr(A)I . Ū) = −∇ • νeff (∇ ⊗ Ū) . DIVDEVREFF 87 The new function dev() is implemented as: 1 template < class Cmpt > 2 inline Tensor < Cmpt > dev ( const Tensor < Cmpt >& t ) 3 { 4 return t .7) 3 After we push the first term into the second one:   1 − ∇ • τ̄ eff = −∇ • νeff (∇ ⊗ Ū) + νeff (∇ ⊗ Ū)T − νeff tr((∇ ⊗ Ū)T )I . (10.1. Hence. Nabla ∇ and the velocity vector U. we are able to put everything together and end up with:    1 − ∇ • τ̄ eff = −∇ • νeff (∇ ⊗ Ū) − ∇ • νeff (∇ ⊗ Ū)T − tr((∇ ⊗ Ū)T )I  . the function dev() needs an argument. (10. Thus.SphericalTensor < Cmpt >:: oneThirdsI * tr ( t ) .8) 3 and rewrite the last term by using the relation (8.4) the second term is equal to:   − fvc :: div(νeff ∗ dev(T(fvc :: grad(Ū)))) = −∇ • νeff ∗ dev((∇ ⊗ Ū)T ) .5) and the dev-function is similar to: 1 dev((∇ ⊗ Ū)T ) = (∇ ⊗ Ū)T − tr((∇ ⊗ Ū)T )I . (10. the first term of the divDevReff function can be written as:  − fvm :: laplacian(νeff . 5 } Listing 10. Therefore. 10. we end up with:   1 − ∇ • τ̄ eff = −∇ • νeff (∇ ⊗ Ū) + νeff (∇ ⊗ Ū)T − νeff  • (∇ Ū) I .2) 3 In addition. THE INCO. The argument is the transposed matrix that we get. (10.3: $FOAM SRC/OpenFOAM/primitives/Tensor/TensorI.H Analyzing the C++ Functions Starting with the function dev().9) 3 | {z } continuity . if we build the dyadic product of the two vectors. we can write:  T  T A = grad(Ū) = ∇ ⊗ Ū .27).6) 3 Finally. we can see that the function simply calculates the deviatoric part of the matrix.9). (1. (10.3) If we express the C++ code with the expressions above. (10. SHEAR-RATE TENSOR. (10. (10.

The con- structed momentum equation looks similar to the one we had for incompressible fluids but now we have the density included and we call a new function named divDevRhoReff. we have the time derivation (I). the convective (II). U ) ( II ) 5 + MRF .1 because the derived equation is similar to equation (5. U ) (V) 9 ). some additional correction due to MRF (III). the calculation of the incompressible shear-rate tensor is correct im- plemented into the version 2. If we are using dev2() instead of dev() for incompressible solvers will lead to an even better stabilization and convergence rate.10) 2 | {z } deformation rate tensor D̄ As we demonstrated now. In newer OpenFOAMr versions like 4. Listing 10. This is based on the discretization. 1 tmp < fvVectorMatrix > UEqn 2 ( 3 fvm :: ddt ( rho . we figured out that the term 31 νeff tr(∇⊗U) is kept. U ) ( III ) 6 + turbulence . The difference of dev2 and dev will be shown later and is simple a difference of the factor two at some position of the code.> divDevRhoReff ( U ) ( IV ) 7 == 8 fvOptions ( rho . (10. The term simply stabilizes the calculation because the continuity equation is never 100% zero. the shear-rate tensor (IV) and the term (V) that handles additional sources within the . interpolation and accuracy of the machine. The code snippet is based on rhoPimpleFoam. the incompressible solvers will call dev2() that is normally used for compressible flows.x. The sign difference of the term corresponds to the position at the LHS in OpenFOAMr whereas in equation (5. the effective shear-rate tensor τ̄ eff for incompressible fluids is calculated as:   − ∇ • τ̄ eff = −∇ • νeff (∇ ⊗ Ū) + νeff (∇ ⊗ Ū)T | {z } τ̄ eff   n o  1 = −∇ • 2νeff (∇ ⊗ Ū) + (∇ ⊗ Ū)T . However.H As before. 88 Calculation of the Shear-Rate Tensor in OpenFOAMr The last term is zero due to the continuity equation.35). due to the continuity equation. The reason for that is based on numerics. the equation is also valid.35) the shear-rate tensor stands on the RHS.4: $FOAM SOLVERS/compressible/rhoPimpleFoam/UEqn. 10. the shear-rate tensor has a different formulation. Shear-Rate Tensor.3. U ) (I) 4 + fvm :: div ( phi . This is the reason why it was introduced in OpenFOAMr 3. Therefore. Stability During the analyze of the function.0. we can cancel the additional term again and hence. divDevRhoReff If we focus on compressible fluids.2 The Compr.0. DDt ( rho .

12) the second term to:   − fvc :: div(µeff ∗ dev2(T(fvc :: grad(Ũ)))) = −∇ • µeff ∗ dev2((∇ ⊗ Ũ)T ) .. THE COMPR. SHEAR-RATE TENSOR.27) with the already mentioned difference that we subtract the hydrostatic part twice. This indicates that we calculate the shear-rate tensor based on the theory for compressible fluids.fvm :: laplacian ( muEff () .5: .SphericalTensor < Cmpt >:: twoThirdsI * tr ( t ) . 1 template < class Cmpt > 2 inline Tensor < Cmpt > dev2 ( const Tensor < Cmpt >& t ) 3 { 4 return t .11) 3 The argument of the function dev2() is equal to the one we had in the incompressible case. DIVDEVRHOREFF 89 fvOptions dictionary. The shear-rate tensor that is calculated for a compressible fluid is given below. It is somehow calculating the deviatoric part of a tensor but subtraction twice the hydrostatic part instead of once.. (10.C The function is similar to dev() but now we have the molecular instead of the kinematic viscosity and call a new function named dev2. 8 } Listing 10. 1 tmp < fvVectorMatrix > laminar :: divDevRhoReff ( volV ectorFi eld & U ) const 2 { 3 return 4 ( 5 . it can be evaluated by (10. U ) 6 .fvc :: div ( muEff () * dev2 ( T ( fvc :: grad ( U ) ) ) ) 7 ). 10. the dilatation term is included due to expansion and compression phenomena which can be related to the non- constant density.H Analyzing the C++ Functions The argument that is return by the function dev2() represents equation (1. twoThirdsI: 2 Adev = A − 2Ahyd = A − tr(A)I (10.2. Hence. Thus. we proceed like before. Rewriting the C++ code into the different equations. (10. Ũ) = −∇ • µeff (∇ ⊗ Ũ) . The code of the new function is presented below. 5 } Listing 10./turbulenceModel/compressible/RAS/laminar/laminar.13) .6: OpenFOAM/primitives/Tensor/TensorI. First we investigate into the call of the function turbulence->divDevRhoReff(U). The reason for that is obvious after we analyze the code snippet.3). For the analysis of the function. Note: The keyword Rho is now included in the name of the function. we are able to rewrite the first term as:   − fvm :: laplacian(µeff .

18) 3 | {z } effective shear−rate tensor τ̄ eff To get a more familiar equation. (10. Hence. it follows:   T 2 T − ∇ • τ̃ eff = −∇ • µeff ∇ ⊗ Ũ + µeff (∇ ⊗ Ũ) − µeff tr((∇ ⊗ Ũ) )I .15) 3 After we pushed the divergence operator out. we get the known shear-rate tensor by using equation (8. the calculation of the compressible shear-rate tensor is implemented in OpenFOAMr correctly.9) as:   T 2 − ∇ • τ̃ eff = −∇ • µeff ∇ ⊗ Ũ + µeff (∇ ⊗ Ũ) − µeff (∇ • Ũ)I .16) 3 By eliminating the brackets inside. we do some simple mathematics and end up with:   n o 2  1 − ∇ • τ̃ eff = −∇ • 2µeff (∇ ⊗ Ũ) + (∇ ⊗ Ũ)T − µeff (∇ • Ũ)I .90 Calculation of the Shear-Rate Tensor in OpenFOAMr and the dev2-function like: 2 dev2((∇ ⊗ Ũ)T ) = (∇ ⊗ Ũ)T − tr((∇ ⊗ Ũ)T )I (10.19) 2 3 | {z } deformation rate tensor D̃ As we demonstrated. The question now is. The only difference is related to the viscosity.3 Influence of Turbulence Models If we use a compressible based solver in OpenFOAMr and simulate a laminar flow pattern. Reynolds-Averaged or Favre-Averaged flow fields are identical.85) with the difference that we use Favre averaged quantities here. (10. the difference in the sign is based on to the fact that the term stands on the LHS in OpenFOAMr . we get:    2 − ∇ • τ̃ eff = −∇ • µeff (∇ ⊗ Ũ) + µeff (∇ ⊗ Ũ)T − tr((∇ ⊗ Ũ)T )I .14) 3 After combining these terms. 10. what happens if we do so (not use a turbulence model)? As we saw in the chapters above. if we do not use a turbulence model. the equations for full resolved eddies. (10. the contribution of the eddy viscosity µt is zero.17) 3 and finally. The equation above is equal to the averaged shear-rate tensor τ̄ in equation (9.20) . the momentum equation will not change based on the fact that it is hard coded. As before. It follows: µeff = µl +  µt . (10. (10. it follows:      2 − ∇ • τ̃ eff = −∇ • µeff (∇ ⊗ Ũ) − ∇ • µeff (∇ ⊗ Ũ)T − tr((∇ ⊗ Ũ)T )I . (10.

This is done by the well known algorithms named: SIMPLE.1) ∂t Note: In the above equation we did not introduce the shear-rate tensor. one can see that we have four unknowns and only three equations (momentum in x. Further information about these algorithms can be found in Ferziger and Perić [2008] and Moukalled et al. PISO and PIMPLE.35) and thus it follows: ∂U p = −∇ • (U ⊗ U) − ∇ • τ inco − ∇ + g .13). Although.26) and apply the incompressibility char- acter — the density is constant and can be taken out of the derivatives —. we only discretized the time derivative while the space derivatives are kept in partial differential form.Chapter 11 SIMPLE. Hence. 91 . we will understand these algorithms better after we introduced the difficulties that come into handy when we solve the Navier-Stokes equations or any other coupled system. That’s why the sign is negative in this particular case. we need some special techniques to solve the coupled pressure-momentum system. there are four unknown quantities. As we can see. This is also known as the pressure-momentum coupling problem. is to apply the divergence operator onto the momentum equation. However. (11. Let us consider the general momentum equation (2. there is no need to solve any energy equation and we only need to solve the momentum equation. The idea — to get rid of the problem — is to use the mass conservation equation somehow. (11. What we finally do. PISO and PIMPLE algorithm Solving the Navier-Stokes equations requires numerical techniques for solving the coupled pressure- momentum system. that means. we can use the mass conservation equation to eliminate terms and end up with the well known Poisson equation for the pressure p. τ has to be expressed by equation (5. the pressure p and the three velocity components denoted by U. [2015].2) ∂t ρ Based on the fact that the density is a constant. Now. y and z). Therefore. The different algorithms are based on different problems and therefore. we need an additional equation which is the mass conservation equation (2. derivation is not shown in this book. we divide the whole equation by the density ρ. After doing a semi-discretization. we get: ∂U ρ = −ρ∇ • (U ⊗ U) − ∇ • τ − ∇p + ρg . this equation does not has the pressure included.

For the SIMPLE algorithm it is very important to estimate the relaxation factors for the fields and equations for good stability and fast convergence rate. In OpenFOAMr we are using this algorithm for steady-state analyzes. • PISO:= Pressure-Implicit-of-Split-Operations. That means. the SIMPLE algorithm is not consistent. Furthermore. This algorithm combines both algorithms and allows us to use bigger time-steps (Co >> 1). SIMPLEC and so on.1. the solver just blow up and gives a floating point exception (dividing by zero). As a simple rule (not valid all the time). that we solve for Ux while we keep all other variables constant. Note: There is a family of different algorithms available.92 SIMPLE. In OpenFOAMr we are using this algorithm for transient calculation. • PIMPLE:= Merged PISO–SIMPLE.0. changing ∆t can affect the results but only if we will do not reach the steady state solution. the time will indicate the number of iterations that we did within the SIMPLE loop. A very good overview is given in Moukalled et al. That means. PISO and PIMPLE algorithm Now we have an equation for the momentum and the pressure. [2015]. SIMPLE and PIMPLE. then for Uy . It simply let us reach the pseudo end time faster or not. we are only interested in the steady-state behavior and based on the missing natural limiter ∆t and the fact that the SIMPLE algorithm is not consistent (missing term).1 The SIMPLE algorithm in OpenFOAMr If we use any kind of SIMPLE based solver in OpenFOAMr . the SIMPLEC algorithm can be used in all SIMPLE and PIMPLE operating algorithms. That means. For that purpose. The calculation is limited in the time step based on the Courant number. 11. The strategy now is to find a pressure and momentum field that fulfill the mass conservation and of course the case conditions related to the boundary conditions and time. Therefore. the time step ∆t should be always set to 1. we need to under-relax the equations to achieve stability. for a special time interval ∆t. different investigations were done to calculate the missing term and are known under the terms of SIMPLER. For example. Otherwise. then for Uz and then for p. that during the derivation of the pressure equation. we have to add some special keyword to the SIMPLE or PIMPLE control . 11. Of course. In general these equations are solved sequential. the solution can only go on by this time step and not further. The sequential solving procedure is been achieved by using the aforementioned algorithms namely: PISO. A time derivation is normally a natural limiter for the solution.1 SIMPLEC in OpenFOAMr In the release version 3. we can say: • SIMPLE:= Semi-Implicit-Method-Of-Pressure-Linked-Equations. Changing the time step to other values will not influence the solution. Doing that. SIMPLEM.0. Based on the fact that we do not have the time derivation within that algorithm. In other words. In other word. we do not have a time derivation. we first solve the equation for Ux . we do more or less iterations. we neglect one term.

Activating the SIMPLEC method can be done by adding the following keyword (next side). .11.1. THE SIMPLE ALGORITHM IN OPENFOAMr 93 dictionary in the fvSolutions file.

3) ∆x The Courant number depends on the local cell velocity U. we can derive the following aspects: • The higher the local cell velocity U. Using the SIMPLEC method will require more iterations for each single segregated calculation step but the convergence rate will increased. U∆t Co = . will be discussed later. larger values for the under-relaxation factors can be choosen. 11.3).1: SIMPLE operating in SIMPLEC mode As already mentioned in the previous section. 9 } Listing 11.3). this keyword will not affect the numerical procedure. the SIMPLEC algorithm include the missing pressure term. Based on the simulation type. Therefore. we do not need to under-relax the fields and equations but need to fulfill a stability criterion. increase the velocity or the time step. To fulfill the criteria in equation (11. the larger the Courant number. if we operate in the PIMPLE mode. 4 } 5 6 PIMPLE 7 { 8 consistent true . The Courant number can be visualized as follows: If the dimensionless number is smaller than one. Otherwise. Based on this two additional criteria. the calculation is based on the cell volume and not on the distance ∆x. the information from one cell can only reach the next neighbor cell within one time step. the Courant number will increase.2 The PISO algorithm in OpenFOAMr The two main differences to the SIMPLE algorithm are the included time derivation term and the consistency of the pressure-velocity coupling equation. Note: The consistency keyword can be added to the PIMPLE dictionary too. The main aspect here is. The release notes report a speed-up of three times. the Courant number has to be smaller than one. that if we refine the mesh. the time step has to be adjusted based on the mesh size and the velocity. Furthermore. the information can reach a second or third neighbor cell which is not allowed based on some explicit aspects. (11. the time step ∆t and the distant between the cells ∆x. but will only affect the algorithm. the larger the Courant number. The added character C   stands for consistency. Otherwise. How to use the PIMPLE algorithm correct. • The smaller the distance ∆x. Based on formula (11. a small value has to be considered at the beginning of a simulation. we have to make sure that the so called Courant number is not larger than one. PISO and PIMPLE algorithm 1 SIMPLE 2 { 3 consistent true . In general. In OpenFOAMr . • The larger the time step ∆t. the larger the Courant number. 94 SIMPLE. . which can then be increased to some – case depended – value.

2: The output of the pimple algorithm 11.3 The PIMPLE algorithm in OpenFOAMr The PIMPLE algorithm is one of the most used one if we have transient problems because it combines the PISO and SIMPLE (SIMPLEC) one. The advantage is.3.Settings that we can made 4 } Listing 11. After we found the solution. to ensure that explicit parts of the equations are converged. we leave the outer correction loop and move on in time. we need to define the algorithm control dictionary named PIMPLE.3: The control dictionary within the fvSolution file The keyword has to be added to the fvSolution file. The value of zero just ignores the whole pimple loop (no calculation). 11. That means. First of all. we search a steady-state solution with under-relaxation. that one bad cell can limit the whole simulation. . If we do so. we go on in time. Here. The output should be as follow: 1 Create mesh for time = 0 2 3 PIMPLE : Operating solver in PISO mode Listing 11. we need the so called outer correction loops. otherwise the solver will throw out an error. After we reach a defined tolerance criterion within the steady-state calculation. 10 nCorrPIMPLE_ (0) .4 The correct usage of the PIMPLE algorithm The usage of the PIMPLE algorithm is explained and discussed within this section. that we can use larger Courant numbers (Co >> 1) and therefore. 6 const word & dictName 7 ) 8 : 9 solutionContr ol ( mesh . it is sufficient just to create an empty dictionary. This can be checked. 11. Nevertheless. if we start any PIMPLE solver. THE PIMPLE ALGORITHM IN OPENFOAMr 95 Note: The criteria has to be fulfilled for each cell. The principal of the algorithm is as follows: Within one time step. 1 PIMPLE 2 { 3 // . 11 nCorrPISO_ (0) . This is done till we reach the end time of the simulation. dictName ) . if we set the nOuterCorrectors to one. the settings that can be set for controlling the algorithm has to be written into the fvSolution dictionary. the code will use default values based on the constructor of the class. For this. 1 // * * * * * * * * * * * * * C o n s t r u c t o r s * * * * * * * * * * * // 2 3 Foam :: pimpleControl :: pimpleControl 4 ( 5 fvMesh & mesh . Note: The PIMPLE algorithm in OpenFOAMr can also work in PISO mode. the time step can be increased drastically.

we go into the read function. After that. 21 } Listing 11. Here. we solve the turbulence equation within each outer loop. If there is no entry to read. we initialize other essential control parameters for the algorithm such as the nNonOrthogonalCorrectors. consistent keywords as well as stuff for the residualControl (not shown in the code). the default values are set. momentumPredictor. 14 nCorrPISO_ = pimpleDict . 1 // * * * * * * * * * P r o t e c t e d Member F u n c t i o n s * * * * * * * * // 2 3 void Foam :: pimpleControl :: read () 4 { 5 solutionControl :: read ( false ) . we will read the PIMPLE dictionary in the fvSolution file. The same is valid for the nCorrectors. PISO and PIMPLE algorithm 12 corrPISO_ (0) . For the nOuterCorrectors a value of one is used. 1) . 14 converged_ ( false ) 15 { 16 read () . 12 1 13 ). and set two booleans. lookupOrDefault < label > 10 ( 11 " nOuterCor r e c t o r s " . The read() function will also call another read(argument) function out of the solutionControl class. 9 nCorrPIMPLE_ = pimpleDict . 96 SIMPLE. 13 t u r b O n F i n a l I t e r O n l y _ ( true ) . 19 true 20 ). Finally the turbOnFinalIterOnly is set to true. lookupOrDefault < label > . Listing 11. transonic.4: The constructor of the pimpleControl class As we can see.5: The read function of the pimpleControl class In addition.> dict () . lookupOrDefault < Switch > 17 ( 18 " turbOnFinalIterOnly ". One to true and the other one to false. we see that the nOuterCorrectors are related to the PIMPLE and the nCorrectors (inner loops) to the PISO algorithm. the constructor will initialize all values with zero first. otherwise we solve it only once at the last outer iteration. 15 turbOnFinalIterOnly_ = 16 pimpleDict . 1 // * * * * * * * * * P r o t e c t e d Member F u n c t i o n s * * * * * * * * // 2 3 void Foam :: solution C on tr ol :: read ( const bool absTolOnly ) 4 { 5 const dictionary & solutionDict = this . 6 7 // Read soluti o n c o n t r o l s 8 const dictionary & pimpleDict = dict () . In this particular function. If that switch is turned on. lookupOrDefault < label >( " nCorrectors " . 6 7 // Read soluti o n c o n t r o l s 8 nNonOrthCorr_ = 9 solutionDict .

l o ok up Or D ef au l t ( " m o m e n t u m P r e d i c t o r " . true means using SIMPLEC • No residual control information is set . l o ok up Or D ef au lt ( " transonic " .6: The read function of the solutionControl class To sum up. Listing 11. • transonic is set to false . lo ok u pO rD ef a ul t ( " consistent " . 16 transonic_ = solutionDict . false ) . 11. 12 0 13 ). 24 25 // Residual c o n t r o l s not shown 26 .4. • nNonOrthogonalCorrectors (corrPiso) is set to 0 . false ) . 17 consistent_ = solutionDict . • momentumPredictor is set to true . • nCorrectors (nCorrPiso) is set to 1 . false means using SIMPLE. subO rEmptyDi ct ( " r es id ua l Co nt ro l " ) 23 ). • turbOnFinalIterOnly is set to false .. we will have the following set-up: • nOuterCorrectors (nCorrPimple) is set to 1 . 18 19 // Read resid u a l i n f o r m a t i o n 20 const dictionary residualDict 21 ( 22 solutionDict . THE CORRECT USAGE OF THE PIMPLE ALGORITHM 97 10 ( 11 " nNonOrthogonalCorrectors ". a simple non-steady test case is provided and the key-ideas of the PIMPLE algorithm are demonstrated. true ) .. . If we use the PIMPLE algorithm without any specific keyword. 14 m om en t um P re d i c t o r _ = 15 solutionDict . For the further analyze. • consistent is set to false .

98 SIMPLE, PISO and PIMPLE algorithm

11.4.1 The test case
To demonstrate the behavior of different settings within the fvSolution file, a simple 2D tran-
sient pipe flow will be considered; this is not the best case because the big advantage of the
PIMPLE method comes with complex geometries.
The pipe is contracted in the middle in order to accelerate the fluid and to create some vortex
after that; compare figure (11.1).
The kinematic viscosity ν is set to 1e−5 and the extrusion of the 2D mesh in z-direction is 0.01m.
The OpenFOAMr version that is used is 4.x and the case is available on www.holzmann-cfd.de/
pimpleCase/pimpleCase.tar.gz.

∆x = 0.45m

∆x = 0.05m ∆x = 0.2m

noSlip noSlip

∆y = 0.025m

U = 0.2 m
s
∆y = 0.1m inlet outlet

∆y = 0.05m

noSlip noSlip

Figure 11.1: The 2D pipe flow domain for the analyze of the PIMPLE algorithm.

Another example and further explanations about the equations that are solved can be found in
Holzmann [2014] and an even more detailed introduction into the method and how the pressure-
momentum coupling is done in OpenFOAMr is given in Moukalled et al. [2015].

11.4.2 First considerations
In the following, we are using the above domain with the mentioned boundary conditions to
demonstrate the usage of the PIMPLE algorithm. First of all, we investigate into some basic
behavior of the problem.
By using the continuity equation, one can estimate the maximum time step for the simulation.
The area at the inlet is A = 0.1m · 0.01m = 0.001m2 . Therefore, we get a volumetric flux
3
φ = U A = 0.2 ms 0.001m2 = 0.0002 ms . This flux has to cross the small section in the middle of
the domain and thus, the velocity has to increase, while the pressure will drop. The cross section
has an area of Across = 0.025m · 0.01m = 0.00025m2 . Hence, the velocity in the cross section
became around Ucross = 0.8 ms . Based on the fact that the flow will be contracted in the cross
section, the velocity will further rise. So let us assume that we will get the maximum velocity of
1.2 ms somewhere in the mesh for the first guess. The resulting time step, with a cell distance of

11.4. THE CORRECT USAGE OF THE PIMPLE ALGORITHM 99

∆x = 0.002m, can then be evaluated with equation (11.3):

1 · 0.002
∆t = s = 0.0016s .
1.2
The generated mesh is a pure hexaeder mesh and thus we do not need any orthogonal corrections.
First, we will use the pisoFoam solver and the evaluated time step as a reference for the
residuals and velocity contour plot. After that, we will use the pimpleFoam solver and apply
different keywords and settings.
To make the simple problem more complex, we use the Gauss linear discretization scheme,
that tends to produce non physical results, if the stability criterion is not strictly fulfilled.

11.4.3 Run the case with the PISO algorithm
The first step is to generate a reference case. This is done using the pisoFoam solver.
Due to the fact, that we solve the flow without a turbulence model, we will naturally get some
vortexes which result in high transient behavior. This will make the system more stiff and harder
to solve. Furthermore, we can expect, that the residual plots for the time steps look similar for
each case, based on the transient character. At last, based on the high transient behavior, we will
see some very interesting numerical phenomenon at the end of the discussion.
After the solver finishes, we get the simulation results shown in figure 11.2 and 11.3. Here,
we used a fixed time step (not adjustable). That means, based on the fact that the velocity will
change during the simulation, the Courant number will change too, cf. (11.3).
During the simulation a maximum Courant number of 1.4 and a maximum magnitude of the
velocity of around 1.3 ms are achieved. This is the first indication, that our time step approximation

PISO algorithm
1e+00
p
Ux
Uy
1e-01
Residuals [-]

1e-02

1e-03

1e-04

1e-05
0 0.2 0.4 0.6 0.8 1 1.2 1.4
Time [s]

Figure 11.2: Residual plot for the pisoFoam solver; fixed ∆t = 0.0016s.

100 SIMPLE, PISO and PIMPLE algorithm

Figure 11.3: Velocity contours generated with pisoFoam and a fixed ∆t = 0.0016s.

works fine for this case. Although we reach higher Courant number than one, the simulation is
still stable. It should be clear that the time step approximation based on equation (11.3) can only
be used for very simple geometries; moreover it is common to adjust the time step based on the
Courant number than using a fixed ∆t.
Another thing that we can observe in front of the contraction area are velocity stripes. This
is based on the Gauss linear scheme and indicate that we are close to the instability limit and is
a common habit of that scheme.

11.4.4 PIMPLE working as PISO
Now we are using the same case without any changes and run the pimpleFoam solver on it. Based
on the fact that the PIMPLE entry within the fvSolution is empty, OpenFOAMr will use the
default values and hence, the pimpleFoam solver is run in PISOmode. Thus the execution is equal
to the pisoFoam solver. That can be proofed by checking the residuals and contour plots of this

PIMPLE algorithm
1e+00
p
Ux
Uy
1e-01
Residuals [-]

1e-02

1e-03

1e-04

1e-05
0 0.2 0.4 0.6 0.8 1 1.2 1.4
Time [s]

Figure 11.4: Residuals of the calculation with the PIMPLE algorithm that is working in PISO mode.

On the right top and bottom we can see areas where the velocity is already larger than 60 ms (total PISO algorithm 1e+00 p Ux Uy 1e-01 Residuals [-] 1e-02 1e-03 1e-04 1e-05 0 0.4 0. calculation and the last one.7 the qualitative result of the time step is given. In figure 11.6 0.5: Velocity contours of the PIMPLE algorithm that is working in PISO mode.11.2 0. to reach the end time of the simulation much faster. figure 11. the simulation should crash.2 1. This will increase the Courant number based on equation (11. THE CORRECT USAGE OF THE PIMPLE ALGORITHM 101 Figure 11.8 1 1. One can discover that both results are equal.5 PIMPLE working as PISO with large ∆t Now we want to increase the time step to ∆t = 0.4 and 11. we observe what we expect and after a few time steps the solver crashes with a Floating Point Exception (division by zero). cf.4.5. critical velocities are reached and the algorithm will give an error.6: Residuals of the calculation with the PIMPLE algorithm that is working in PISO mode using a large time step. that after the evaluation of the flow. the comparison is given in figure 11.3) and hence.4.025s. 11.6. when the instability is initiated. The residual plot demonstrates. .4 Time [s] Figure 11. Finally.

re-building the velocity matrix with the new flux field. that stabilizes the whole solution procedure. 102 SIMPLE.7: The control dictionary within the fvSolution file The nOuterCorrectors will set the nCorrPIMPLE variable to five and hence.000 ms . Within the next two time steps. it is not like the real SIMPLE algorithm because we have the limiting time step. gray areas). we know that the SIMPLE algorithm is not stable without relaxation. we would think that we will improve our calculation and make the algorithm more robust and stable but in fact the solver crashes. Finally.6 PIMPLE algorithm modified (add outer corrections) Up to know. the procedure can be unstable. That include. The outer corrector loops can be considered as a SIMPLE loop and thus without under-relaxation.M o m e n t u m C o r r e c t i o n ) 4 nOuterCorrecto r s 5. We can do this by adding the following keywords to the PIMPLE directory.4.8. Furthermore. This can be seen in the residual and contour plot. . 11. Respectively the pressure will blow up too. the PIMPLE algorithm is a combination of PISO and SIMPLE. till. This happens until the accuracy and the delineation of the computer results into a zero value.7: Velocity contours of the PIMPLE algorithm that is working in PISO mode with a large time step. Moreover. 5 } Listing 11. Finally. 1 PIMPLE 2 { 3 // Outer Loops ( Pressure . we run the PIMPLE algorithm in PISO mode. First of all. Now we will use the merged PISO- SIMPLE method by manipulating the algorithm. we see. compare residual plot 11. as before. In figure 11. PISO and PIMPLE algorithm Figure 11. The reason for that is the usage of the outer corrector loop.9. As we already said. the whole right part has already incredible large velocities which will move on to the left. correct the pressure with the new velocity matrix and correct the fluxes based on the new pressure. that the crash happens already earlier as before. the solver crashes. the instability moves further to the left side and the velocities increases to values larger than 50. we make five outer corrections (pressure-momentum correction loop). we correct the velocities and go back to the re-guessing step till we reached five times.

the solver fails faster than in the case before. The difference with nOuterCorrectors As we already discussed. The reason for the crash can be explained as follow: If we go on in time. Figure 11. pressure and momentum more often within one time step.11. All in all. As we can see. the flow field will further develop and within the small gap. the solution will diverge.4. The fact that we are looping five times more over one time step will speed up the divergence after it is initiated and hence. THE CORRECT USAGE OF THE PIMPLE ALGORITHM 103 PIMPLE algorithm 1e+00 p Ux Uy 1e-01 Residuals [-] 1e-02 1e-03 1e-04 1e-05 0 0.8 1 1.4 Time [s] Figure 11. the difference with the usage of the outer loop correction is. If we would put a line through the highest peaks of each quantity.9: Velocity contours of the PIMPLE algorithm with the usage of five nOuterCorrectors.6 0. we have five more iteration steps (SIMPLE).2 0. we do this five times here.2 1. we will get higher velocities.10.4 0. we would get figure 11.8: Residuals of the calculation with the PIMPLE algorithm and nOuterCorrectors = 5. within one time step.8. we get the plot shown in figure 11. After some critical velocity is reached. . Doing a more detailed analyze of the residuals. that we recalculate the fluxes.

we calculate five outer loops and within one outer loop. that within the first time step. What we observe is.1 0.6 1e+00 p Ux Uy 1e-01 Residuals [-] 1e-02 1e-03 1e-04 1e-05 0 20 40 60 80 100 120 Accumulated Outer Correction Loop [-] Figure 11.4 0.4.12. Recall: For one time step. 1 PIMPLE 2 { 3 // Outer Loops ( Pressure . 11. 8 } Listing 11. the solution tends to diverge. 104 SIMPLE. For that purpose we have to add a new keyword to the PIMPLE dictionary as shown in the code snipped below. Therefore. Increasing the outer loops would lead to an even faster abort. that within the first time step. that we recalculate the pressure with the new updated fluxes but with the old matrix of the momentum (PISO corrections . the solution tends to diverge already.M o m e n t u m C o r r e c t i o n ) 4 nOuterCorrecto r s 5.8: The control dictionary within the fvSolution file As before.pressure correction loop).5 0. we cannot analyze the general residual plot but we can evaluate the curvatures of the outer and inner iterations. This trend is continued within the second time step till the solver crashes. the solver already crashes in the second time step. we calculate twice the pressure. Figure 11. Using more outer iterations will help to establish the wrong solution and lead to a failure of the algorithm already in the first time step.11 shows the inner and outer loops of the first two time steps. 5 6 // Inner Loops ( P r e s s u r e C o r r e c t i o n ) 7 nCorrectors 2. This is because we can already observe. . The result of the first time step calculation is given in figure 11.3 0.7 PIMPLE algorithm further modified (add inner corrections) Commonly it is helpful and suggested.10: Residuals of the outer loop iterations. PISO and PIMPLE algorithm PIMPLE algorithm Time [s] 0 0. instead of getting a more stable algorithm.2 0.

THE CORRECT USAGE OF THE PIMPLE ALGORITHM 105 PIMPLE algorithm Accumulated Inner Correction Loops [-] 0 2 4 6 8 10 12 14 16 18 20 1e+00 p Ux Residuals [-] 1e-01 Uy 1e-02 1e-03 1e-04 1e-05 0 2 4 6 8 10 Accumulated Outer Correction Loops [-] Figure 11. Up to now. nOuterCorrectors = 5 and nCorrectors = 2.8 PIMPLE algorithm with under-relaxation To make the PIMPLE algorithm work fine. 11.4.M o m e n t u m C o r r e c t i o n ) 4 nOuterCorrect o r s 100. it is essential to tell OpenFOAMr the specific under-relaxation factors that should be used for the fields and/or equations. stable and more robust. we are forced to use the under-relaxation technique. Figure 11. 1 PIMPLE 2 { 3 // Outer Loops ( Pressure .11: Residuals of the iterations of the inner and outer correction loops. nOuterCorrectors = 5 and nCorrectors = 2.1. the method is not consistent and hence. it seems that the PIMPLE algorithm only offers disadvantages.4. we have to think about the SIMPLE method (outer correction loop). The reason for that is based on the wrong usage of the method. 11. . note that under- relaxation is not always necessary especially for Co < 1 and non-stiff problems. As we already mentioned in section 11. Therefore.12: Velocity contours. The first step for the correct usage of the algorithm is discussed in the next section.

Starting the simulation with the above mentioned relaxation factors and the increase of the outer correction loops (to see what happens). it should be fine to use a value smaller than one for the Final iteration. However. First of all. If we do not specify any relaxation factor. the fact. As we can see. This will stabilize the whole simulation because the relaxation factor of one in the last iteration can produce diverging results in some cases. 8 } 9 r e lax ati onF acto rs 10 { 11 fields 12 { 13 p 0.3 and 11. 106 SIMPLE. The question is. If we specify only the the relaxation factor for p and U without the Final one. figure 11. that the Final under-relaxation factor was not set to one or that the time step within the PIMPLE mode was way too large (Co ≈ 20) and hence. lead to the results shown in figure 11. // Last outer loop 15 } 16 17 equations 18 { 19 U 0. the final relaxation factors are set to one by default.3). cf.4. It seems that the PIMPLE algorithm smooths the whole solution. we observe big differences. Within one time step we have much more iterations to find the correct solution. PISO and PIMPLE algorithm 5 6 // Inner Loops ( P r e s s u r e C o r r e c t i o n ) 7 nCorrectors 2. we get similar plots than using the PISO mode but now we have one essential difference.6. To understand the cut of the information more. if we have a lot of outer iterations already done. the default value of one is used for all outer iterations. Checking the time depended residuals. are we allowed to under-relax the final iteration like we did it for the previous outer loops? Actually. And thats why we are allowed to increase the time step without taking care that Co < 1.13 and 11.14.4. that we calculate for one time step. That means. the different relaxation methods are discussed in more detail in section 12. 14 pFinal 0. Comparing the contour plots of figure 11. One for all outer iterations except the last one and one that is used just for the last one.15. 20 UFinal 0. Note: The usage of the SIMPLEC algorithm can be activated by using the consistent keyword.9: The relaxation dictionary within the fvSolution file The relaxation information is put into another dictionary within the fvSolution file. In the following case the Courant number is larger than 20 and the simulation is still stable.14. There is a big discussion about the value of the Final one. // Last outer loop 21 } 22 } Listing 11.6. The reasons for that could be related to the too large time step that we were using within the PISO mode (Co ≈ 1. if we under-relax the final outer loop we might loose some information and we are not consistent. we have two relaxation factors. 100 momentum-pressure and 200 pressure correction calculations. we over jump some important and essential transient information . for each time step we are doing 100 outer corrections and within one outer loop two inner corrections.

that it is important to keep the time step in a range where all important phenomena that influence the flow field can be resolved. nOuterCorrectors = 100 and nCorrectors = 2.4. it is related to the last mentioned hypothesis.11. The current investigated case can be changed in a way that the time step is adjusted by the highest Courant number. THE CORRECT USAGE OF THE PIMPLE ALGORITHM 107 PIMPLE algorithm 1e+00 p Ux Uy 1e-01 Residuals [-] 1e-02 1e-03 1e-04 1e-05 0 0.16. The new result is very close to the PISO calculation compared to the result observed with a fixed time step.4 0.14: Velocity contours. you should keep in mind. . Note: For high transient calculations. nOuterCorrectors = 100 and nCorrectors = 2. In our case.4 Time [s] Figure 11.13: Residuals of the iterations of the inner and outer correction loops. we get the result given in figure 11.8 1 1.6 0.2 1. Based on the large time step.2 0. By using Co = 4 and re-run the simulation without changing any other setting. Figure 11. we do not recognize the establishment of a back flow vortex. which will influence the flow significantly.

nOuterCorrectors = 100 and nCorrectors = 2. nOuterCorrectors = 100 and nCorrectors = 2. Figure 11. PISO and PIMPLE algorithm PIMPLE algorithm Accumlation Inner Correction Loops [-] 0 200 400 600 800 1000 1200 1400 1600 1e+00 p 1e-01 Ux Uy 1e-02 Residuals [-] 1e-03 1e-04 1e-05 1e-06 1e-07 0 100 200 300 400 500 600 700 800 Outer Correction Loops [-] Figure 11.16: Velocity contours. additional residual control added and Courant number controlled .15: Residuals of the iterations of the inner and outer correction loops.108 SIMPLE.

11.4. OpenFOAMr will run the outer corrector loop until the residual criterion for each quantity is fulfilled. // Last outer loop 40 } 41 } Listing 11. This will reduce the calculation time extremely and ensure good accuracy within each time step. 34 pFinal 0.4. // Last outer loop 35 } 36 equations 37 { 38 U 0. 39 UFinal 0. Doing that. The added residual control will speed up the whole PIMPLE procedure. 25 } 26 } 27 } 28 29 r e lax atio nFa cto rs 30 { 31 fields 32 { 33 p 0. 5 6 // Inner Loops ( P r e s s u r e C o r r e c t i o n ) 7 nCorrectors 2.4.6. 14 15 // If this initial t o l e r a n c e is reached . For that purpose.6. 17 } 18 19 U 20 { 21 relTol 0. 22 23 // If this initial t o l e r a n c e is reached . 1 PIMPLE 2 { 3 // Outer Loops ( Pressure . The next listing shows the correct usage of the PIMPLE algorithm. leave 16 tolerance 5e -5. you can also turn on or off the different switches we already know.9 PIMPLE algorithm speed up Instead of looping over the nOuterCorrectors till we reach the set value for each time step. leave 24 tolerance 1e -4. THE CORRECT USAGE OF THE PIMPLE ALGORITHM 109 11. Of course.4. 8 9 residualContr ol 10 { 11 p 12 { 13 relTol 0.10: The residualControl dictionary within the fvSolution file .M o m e n t u m C o r r e c t i o n ) 4 nOuterCorrect o r s 100. we should also add the residual control sub dictionary to the PIMPLE dictionary. it would be much better to leave the loop after we a defined residual limit is fulfilled for each quantity.

. after each residual criterion is fulfilled.19 because within the same amount of outer loops.17: Residuals of the iterations of the inner and outer correction loops. additional residual control added The decrease of the calculation effort can also be seen in figure 11. The corresponding residual and contour plots are given below.8 1 1.4 Time [s] Figure 11. additional residual control added .2 1. the solution is already further with respect to the simulation time than in the case before – here we have eleven time steps whereas before we had only seven within the same amount of outer loop iterations.110 SIMPLE. nOuterCorrectors = 100 and nCorrectors = 2.4 0. PIMPLE algorithm 1e+00 p Ux Uy 1e-01 Residuals [-] 1e-02 1e-03 1e-04 1e-05 0 0.6 0. PISO and PIMPLE algorithm We can now set-up 1000 outer correctors and the algorithm will automatically leave the loop.2 0. Figure 11. nOuterCorrectors = 100 and nCorrectors = 2.18: Velocity contours.

For simple cases and flow pattern the PIMPLE method does not provide too much advantages. PIMPLE ALGORITHM FLOWCHART 111 PIMPLE algorithm Accumlation Inner Correction Loops [-] 0 200 400 600 800 1000 1200 1400 1600 1e+00 p 1e-01 Ux Uy 1e-02 Residuals [-] 1e-03 1e-04 1e-05 1e-06 1e-07 0 100 200 300 400 500 600 700 800 Outer Correction Loops [-] Figure 11.5 PIMPLE algorithm flowchart In the following flowchart. the PIMPLE algorithm has to be applied correctly and the time step cannot be chosen as large as we want.5.4. skewed. if we do not take care about the numerics.x. the given results might be different or not reproducible with other version. non-orthogonal meshes that include different kinds of cells and complex flow patterns or if we have to solve stiff systems.10 PIMPLE conclusion As we saw in the last sections. the PIMPLE algorithm will provide much more advantages and can stabilize the simulation whereas the case would fail with PISO all the time or at least require extreme small time steps. nOuterCorrectors = 100 and nCorrectors = 2. flowchart will be added in the next release. It was presented and discussed how the algorithm should be used and which advantages it offers. In advance. In addition. one can see how the PIMPLE algorithm is working within the OpenFOAMr version 5. Therefore. Different settings will lead to different solutions. 11.19: Residuals of the iterations of the inner and outer correction loops.11. . Once the idea of the algorithm is understood. keep in mind that OpenFOAMr is continuously updated. For more complex geometries. We should always know which time scale can be achieved or which phenomena we are interested in. additional residual control added. 11. the whole pressure-momentum coupling solving procedure will be understood as well as the topics. the PIMPLE algorithm can be used to enlarge the time step.

112 SIMPLE. PISO and PIMPLE algorithm .

we would cut off information. As we can see.∗ n+1 n denotes the value of the last iteration. we need to use relaxation methods. Diagonal dominance means the following: If we would go through the matrix.2 Matrix relaxation In the field of computational fluid dynamics we are solving matrix systems given as Ax = b.g. we do not take care about numerical limits like the Courant number. There are two different ways to do that known as field relaxation and matrix relaxation.Chapter 12 The relaxation methods Relaxation methods are used for steady-state simulations or if one uses a large time step and the PIMPLE algorithm. The matrix A is in general a sparse matrix. That means that we have a lot of zero entries. 12. Imagine a scenario where we calculate a quantity φ for each time step only once while using a relaxation factor of α = 0. That is why OpenFOAMr offers the usage of the Final flag in order to change the relaxation factor for the final outer corrector within each time step. the new value and α is the relaxation factor. Solving this system in an iterative way requires linear solvers. each row has to be at least diagonal equal and one of the rows has to be 113 . (12. The field relaxation is mainly used for the pressure field. This would mean that we are not time consistent because we cut off almost 50% of the information. The performance of the solver is based on the diagonal dominance of the matrix A. E. To stabilize or even get a solution.1) n+1. the relaxation is limiting the change of the value of the quantity we are interested. the actual calculated value. On the other hand.∗ − φn ) . The problem in such cases is that the solution might be not stable because the quantities change to much.1 Field relaxation The field relaxation is simple to understand and limits the new values of the field as follows: φn+1 = φn + α(φn+1.5.∗ α = 1. if we set the relaxation factor to zero. The implementation is given in the GeometricField class. If we use n+1 n+1. 12. If we would make two corrections within one time step. the new value is identically to the one we calculated . the new values will always be the old one. then we would reach a better accuracy but still.

If we would only change the values of the diagonal elements. we replace the diagonal value with the sum of the off-diagonals. To be consistent. that we do not cut off too much information. . Pre-requirement for matrix relaxation The matrix relaxation requires the matrix to be at least diagonal equal in each row. the whole procedure is not consistent. at least all rows that are not diagonal equal or dominant are manipulated to become diagonal equal. we have to add the changes to the source vector b in order to keep consistency. we could make use of the matrix relaxation with the relaxation factor of α = 1. Relaxing the matrix means to make it more diagonal dominant and therefore the linear solvers are happier and will find the solution easier and faster. The main difference between matrix and field relaxation is based on the fact. we first check each row of the matrix and examine if we have an diagonal equal or dominant row. For example in cases where we have shock-waves. the loose of information should me much less than for the field relaxation because the matrix A should have only a few non-diagonal dominant entries if ever. If the diagonal element is less than the sum of the magnitude of the off-diagonals. The result is that we need more outer corrections to get all terms converged. While the field relaxation would just do nothing. However. it can happen that we loose information. the cell in front of the shock-wave will have a lower pressure value than the neighbor cell in which the shock-wave already exists. the diagonal elements increase its value.114 The relaxation methods diagonal dominant. To get an at least diagonal equal matrix A. Thats why we have to manipulate the matrix before relaxing which introduces errors. if ever . Thats why for transonic cases the pressure equation is relaxed by α = 1. Here. see rhoPimpleFoam cases with transonic behavior. Relaxing the matrix means actually to divide all diagonal elements by the relaxation factor α. However. The implementation is given in the fvMatrix class. Important is the pre-requirement for the matrix relaxation which introduce inconsistency. To make the linear solver happier. But this is case depended (see example below). What does that mean? • Diagonal equality is given if the diagonal element is equal to the sum of the magnitude of the off-diagonals. The whole procedure makes a better matrix system for the linear solvers but on the other hand makes it more explicit. • Diagonal dominant is given if the diagonal element is larger than the sum of the magnitude of the neighbor elements. the matrix relaxation with might change the matrix. Based on the fact that α is between 0 < α < 1. That will lead to non-diagonal dominant or equal entry in the matrix. Special matrix treatment with α = 1 Using a relaxation factor of α = 1 is different to the field relaxation.it is based on the method how we relax. there are some things that we have to keep in mind.

Here you find a new wiki that offers a lot of information.openfoam.com.Chapter 13 OpenFOAMr tutorials For those who want to deal with OpenFOAMr and search for special tutorials. For OpenFOAMr beginners If you are not familiar with OpenFOAMr I recommend you to check out the following website. • Using the dynamic mesh library.Holzmann-cfd. • Coupling DAKOTAr with OpenFOAMr . • Generating own boundary conditions using the codedFixedValue. the User-Guide of OpenFOAMr should be read. that offers a lot of additional cases related to the following topics: • Meshing with snappyHexMesh. wiki.de. • Setting up boundary conditions for AMI and ACMI. Furthermore. In addition you will find some three weeks series in which you will learn a lot of stuff. you can thank Jószef Nagy for the good work and all contributers mentioned on the site. Furthermore. you can find different libraries and publications an my website. 115 . • Coupling OpenFOAMr with Blenderr . you can checkout my website. • Solving and meshing (different scenarios). www.

116 OpenFOAMr tutorials .

Chapter 14

Appendix

14.1 The Incompressible Reynolds-Stress-Equation

The derivation of the Reynolds-Stress tensor is structured as follows:
• The derivation of the time derivative is shown completely for all terms ,

• The derivation of the convective derivative is shown completely for term a) ,

• The derivation of the shear-rate derivative is shown completely for term a) ,

• The derivation of the pressure term is shown completely for all terms .
The derivation is given in all details to demonstrate how we get to the equation. Furthermore, we
will be able to understand the terms and the reason why we have to add terms in order to apply
the product rule.
Recall: In section 9.8 we introduced the way how we will use the momentum equation in
order to build the Reynolds-Stress equation. Therefore, we multiplied the momentum equation
with respect to the different fluctuations and set the sum of all terms to zero. Here, we introduced
the Navier-Stokes operator N . Finally, we build the equation that has to be evaluated which is
given for completeness again:

u0x N (ūx + u0x ) + u0y N (ūz + u0z ) + u0x N (ūy + u0y ) + u0y N (ūx + u0x )
| {z } | {z }
a) b)

+ u0x N (ūz + u0z ) + u0y N (ūy + u0y ) + u0y N (ūx + u0x ) + u0z N (ūz + u0z )
| {z } | {z }
c) d)

+ u0y N (ūy + u0y ) + u0z N (ūx + u0x ) + u0y N (ūz + u0z ) + u0z N (ūy + u0y )
| {z } | {z }
e) f)

+ u0z N (ūx + u0x ) + u0x N (ūz + u0z ) + u0z N (ūy + u0y ) + u0x N (ūx + u0x )
| {z } | {z }
g) h)

+ u0z N (ūz + u0z ) + u0x N (ūy + u0y ) =0.
| {z }
i)

Furthermore, we introduced the rules and tricks we are using during the derivation procedure:

117

118 Appendix

• Reynolds time-averaged terms that are linear in the fluctuation are zero ,
∂u0i
• The derivative ∂xi
=0,

• Product rule (1.2) ,

• Adding and subtracting terms to be able to use the product rule; g(x) = g(x) + f (x) − f (x) .

With that information, we will now demonstrate the derivation of the Reynolds-Stress equation
in the order given above.

The Time Derivative
a)

∂(ūx + u0x ) ∂(ūz + u0z )
u0x ρ + u0y ρ
∂t ∂t
∂
ū
x ∂u0 ∂
ū
z ∂u0 ∂u0 ∂u0
= u0x
ρ + u0x ρ x + u0y 
ρ + u0y ρ z = u0x ρ x + u0y ρ z . (14.1)
 ∂t ∂t  ∂t ∂t ∂t ∂t
b)

∂(ūy + u0y ) ∂(ūx + u0x )
u0x ρ + u0y ρ
∂t ∂t
∂
ū
y ∂u0y ∂
ū
x ∂u0 ∂u0y ∂u0
= u0x
ρ + u0x ρ + u0y 
ρ + u0y ρ x = u0x ρ + u0y ρ x . (14.2)
 ∂t ∂t  ∂t ∂t ∂t ∂t
c)

∂(ūz + u0z ) ∂(ūy + u0y )
u0x ρ + u0y ρ
∂t ∂t
∂
ū
z ∂u0 ∂
ū
y ∂u0y ∂u0 ∂u0y
= u0x
ρ + u0x ρ z + u0y 
ρ + u0y ρ = u0x ρ z + u0y ρ . (14.3)
 ∂t ∂t  ∂t ∂t ∂t ∂t
d)

∂(ūx + u0x ) ∂(ūz + u0z )
u0y ρ + u0z ρ
∂t ∂t
∂
ū
x ∂u0 ∂
ū
z ∂u0 ∂u0 ∂u0
= u0y 
ρ + u0y ρ x + u0z 
ρ + u0z ρ z = u0y ρ x + u0z ρ z . (14.4)
 ∂t ∂t  ∂t ∂t ∂t ∂t
e)

∂(ūy + u0y ) ∂(ūx + u0x )
u0y ρ + u0z ρ
∂t ∂t
∂
ū
y ∂u0y ∂
ū
x ∂u0 ∂u0y ∂u0
= u0y 
ρ + u0y ρ + u0z 
ρ + u0z ρ x = u0y ρ + u0z ρ x . (14.5)
 ∂t ∂t  ∂t ∂t ∂t ∂t
f)

∂(ūz + u0z ) ∂(ūy + u0y )
u0y ρ + u0z ρ
∂t ∂t
∂
ū
z ∂u0 ∂
ū
y ∂u0y ∂u0 ∂u0y
= u0y 
ρ + u0y ρ z + u0z 
ρ + u0z ρ = u0y ρ z + u0z ρ . (14.6)
 ∂t ∂t  ∂t ∂t ∂t ∂t

14.1. THE INCOMPRESSIBLE REYNOLDS-STRESS-EQUATION 119

g)

∂(ūx + u0x ) ∂(ūz + u0z )
u0z ρ + u0x ρ
∂t ∂t
∂
ū
x ∂u0 ∂
ū
z ∂u0 ∂u0 ∂u0
= u0z 
ρ + u0z ρ x + u0x
ρ + u0x ρ z = u0z ρ x + u0x ρ z . (14.7)
 ∂t ∂t  ∂t ∂t ∂t ∂t
h)

∂(ūy + u0y ) ∂(ūx + u0x )
u0z ρ + u0x ρ
∂t ∂t
∂
ū
y ∂u0y ∂
ū
x ∂u0 ∂u0y ∂u0
= u0z 
ρ + u0z ρ + u0x
ρ + u0x ρ x = u0z ρ + u0x ρ x . (14.8)
 ∂t ∂t  ∂t ∂t ∂t ∂t
i)

∂(ūz + u0z ) ∂(ūy + u0y )
u0z ρ + u0x ρ
∂t ∂t
∂
ū
z ∂u0 ∂
ū
y ∂u0y ∂u0 ∂u0y
= u0z 
ρ + u0z ρ z + u0x
ρ + u0x ρ = u0z ρ z + u0x ρ . (14.9)
 ∂t ∂t  ∂t ∂t ∂t ∂t
Sorting the terms,
∂u0x ∂u0 ∂u0y ∂u0y ∂u0 ∂u0
u0x ρ + u0x ρ x + u0y ρ + u0y ρ + u0z ρ z + u0z ρ z
∂t ∂t ∂t ∂t ∂t ∂t
∂u0y ∂u 0
x ∂u 0
z ∂u 0
x ∂u 0
z ∂u0y
+u0x ρ + u0y ρ + u0x ρ + u0z ρ + u0y ρ + u0z ρ
∂t ∂t ∂t ∂t ∂t ∂t
∂u 0
x ∂u0y ∂u0y ∂u 0
z ∂u 0
x ∂u 0
+u0y ρ + u0x ρ + u0z ρ + u0y ρ + u0z ρ + u0x ρ z .
∂t ∂t ∂t ∂t ∂t ∂t
and using the product rule, we end up with:

∂u0x u0x ∂u0y u0y ∂u0 u0 ∂u0x u0y ∂u0 u0
ρ +ρ +ρ z z +ρ +ρ x z
∂t ∂t ∂t ∂t ∂t
∂u0y u0z ∂u0y u0x ∂u0z u0y ∂u0z u0x
+ρ +ρ +ρ +ρ .
∂t ∂t ∂t ∂t
The above expression can be written in one single term by using the Einsteins summation con-
vention:
∂u0j u0i ∂ρu0j u0i
ρ = . (14.10)
∂t ∂t

Hence. we will end up always with the last line of equation (14.11) result with respect to the used quantities.120 Appendix The Convective Term First we will focus on the convective term of part a)   ∂  ∂ ∂ u0x ρ (ūx + u0x ) (ūx + u0x ) + ρ ūy + u0y (ūx + u0x ) + ρ (ūz + u0z ) (ūx + u0x ) ∂x ∂y ∂z   ∂  ∂ ∂ + u0y ρ (ūx + u0x ) (ūz + u0z ) + ρ ūy + u0y (ūz + u0z ) + ρ (ūz + u0z ) (ūz + u0z ) ∂x ∂y ∂z ∂  ∂ = (u0x ρūx + u0x ρu0x ) (ūx + u0x ) + u0x ρūy + u0x ρu0y (ūx + u0x ) ∂x ∂y ∂ + (u0x ρūz + u0x ρu0z ) (ūx + u0x ) ∂z  ∂  ∂ + u0y ρūx + u0y ρu0x (ūz + u0z ) + u0y ρūy + u0y ρu0y (ūz + u0z ) ∂x ∂y  ∂ + u0y ρūz + u0y ρu0z (ūz + u0z ) ∂z  ∂ūx + u0x ρūx ∂ u0x + u0x ρu0x ∂ ūx + u0x ρu0x ∂ u0x + u0x ρūy∂ūx + u0x ρūy ∂ u0x  = u0x ρūx  ∂x ∂x ∂x ∂x   ∂y ∂y ∂ ∂ 0 ∂  ∂ 0 ∂ ∂ 0 + u0x ρu0y ūx + u0x ρu0y u + u0x ρūz ūx + u0x ρūz u + u0x ρu0z ūx + u0x ρu0z u ∂y ∂y x  ∂z ∂z x ∂z ∂z x ∂  ∂ 0 ∂ ∂ 0 ∂ ∂ 0 + u0y  ρūx ūz + u0y ρūx uz + u0y ρu0x ūz + u0y ρu0x uz + u0y  ρūy ūz + u0y ρūy u  ∂x ∂x ∂x ∂x  ∂y ∂y z ∂ ∂ 0 ∂  ∂ 0 ∂ ∂ 0 + u0y ρu0y ūz + u0y ρu0y u + u0y  ρūz ūz + u0y ρūz u + u0y ρu0z ūz + u0y ρu0z u ∂y ∂y z  ∂z ∂z z ∂z ∂z z ∂ 0 ∂ ∂ 0 ∂ 0 ∂ ∂ 0 = u0x ρūx u + u0x ρu0x ūx + u0x ρu0x u + u0x ρūy u + u0x ρu0y ūx + u0x ρu0y u ∂x x ∂x ∂x x ∂y x ∂y ∂y x | {z } | {z } | {z } | {z } | {z } | {z } 1 2 3 4 5 6 ∂ 0 ∂ ∂ 0 ∂ 0 ∂ ∂ 0 + u0x ρūz u + u0x ρu0z ūx + u0x ρu0z u + u0y ρūx u + u0y ρu0x ūz + u0y ρu0x u ∂z x ∂z ∂z x ∂x z ∂x ∂x z | {z } | {z } | {z } | {z } | {z } | {z } 7 8 9 10 11 12 ∂ 0 ∂ ∂ 0 ∂ 0 ∂ ∂ 0 + u0y ρūy u + u0y ρu0y ūz + u0y ρu0y u + u0y ρūz u + u0y ρu0z ūz + u0y ρu0z u . Thus.11) ∂y z ∂y ∂y z ∂z z ∂z ∂z z | {z } | {z } | {z } | {z } | {z } | {z } 13 14 15 16 17 18 The same procedure can be done with the terms marked as b) to i). (14. we end up with: .

∂y ∂y ∂y ∂z ∂z ∂z | {z } | {z } | {z } | {z } | {z } | {z } 67 68 69 70 71 72 e) ∂ 0 ∂ ∂ 0 ∂ ∂ ∂ u0y ρūx uy + u0y ρu0x ūy + u0y ρu0x uy + u0y ρūy u0y + u0y ρu0y ūy + u0y ρu0y u0y | {z∂x } | {z ∂x } | {z ∂x } | ∂y ∂y ∂y {z } | {z } | {z } 73 74 75 76 77 78 ∂ ∂ ∂ ∂ 0 ∂ ∂ 0 + u0y ρūz u0y + u0y ρu0z ūy + u0y ρu0z u0y + u0z ρūx ux + u0z ρu0x ūx + u0z ρu0x ux | {z∂z } | {z∂z } | {z∂z } | {z∂x } | {z ∂x } | {z ∂x } 79 80 81 82 83 84 ∂ ∂ ∂ ∂ ∂ ∂ + u0z ρūy u0x + u0z ρu0y ūx + u0z ρu0y u0x + u0z ρūz u0x + u0z ρu0z ūx + u0z ρu0z u0x .1. ∂y ∂y ∂y ∂z ∂z ∂z | {z } | {z } | {z } | {z } | {z } | {z } 49 50 51 52 53 54 d) ∂ 0 ∂ ∂ 0 ∂ ∂ ∂ = u0y ρūx ux + u0y ρu0x ūx + u0y ρu0x ux + u0y ρūy u0x + u0y ρu0y ūx + u0y ρu0y u0x | {z∂x } | {z ∂x } | {z ∂x } | ∂y ∂y ∂y {z } | {z } | {z } 55 56 57 58 59 60 ∂ ∂ ∂ ∂ 0 ∂ ∂ 0 + u0y ρūz u0x + u0y ρu0z ūx + u0y ρu0z u0x + u0z ρūx uz + u0z ρu0x ūz + u0z ρu0x uz | {z∂z } | {z∂z } | {z∂z } | {z∂x } | {z ∂x } | {z ∂x } 61 62 63 64 65 66 ∂ ∂ ∂ ∂ ∂ ∂ + u0z ρūy u0z + u0z ρu0y ūz + u0z ρu0y u0z + u0z ρūz u0z + u0z ρu0z ūz + u0z ρu0z u0z . ∂y ∂y ∂y ∂z ∂z ∂z | {z } | {z } | {z } | {z } | {z } | {z } 85 86 87 88 89 90 . THE INCOMPRESSIBLE REYNOLDS-STRESS-EQUATION 121 b) ∂ 0 ∂ ∂ 0 ∂ ∂ ∂ u0x ρūx uy + u0x ρu0x ūy + u0x ρu0x uy + u0x ρūy u0y + u0x ρu0y ūy + u0x ρu0y u0y | {z∂x } | {z ∂x } | {z ∂x } | ∂y ∂y ∂y {z } | {z } | {z } 19 20 21 22 23 24 ∂ ∂ ∂ ∂ 0 ∂ ∂ 0 + u0x ρūz u0y + u0x ρu0z ūy + u0x ρu0z u0y + u0y ρūx ux + u0y ρu0x ūx + u0y ρu0x ux | {z∂z } | {z∂z } | {z∂z } | {z∂x } | {z ∂x } | {z ∂x } 25 26 27 28 29 30 ∂ ∂ ∂ ∂ ∂ ∂ + u0y ρūy u0x + u0y ρu0y ūx + u0y ρu0y u0x + u0y ρūz u0x + u0y ρu0z ūx + u0y ρu0z u0x .14. ∂y ∂y ∂y ∂z ∂z ∂z | {z } | {z } | {z } | {z } | {z } | {z } 31 32 33 34 35 36 c) ∂ 0 ∂ ∂ 0 ∂ ∂ ∂ u0x ρūx uz + u0x ρu0x ūz + u0x ρu0x uz + u0x ρūy u0z + u0x ρu0y ūz + u0x ρu0y u0z | {z∂x } | {z ∂x } | {z ∂x } | ∂y ∂y ∂y {z } | {z } | {z } 37 38 39 40 41 42 ∂ ∂ ∂ ∂ 0 ∂ ∂ 0 + u0x ρūz u0z + u0x ρu0z ūz + u0x ρu0z u0z + u0y ρūx uy + u0y ρu0x ūy + u0y ρu0x uy | {z∂z } | {z∂z } | {z∂z } | {z∂x } | {z ∂x } | {z ∂x } 43 44 45 46 47 48 ∂ ∂ ∂ ∂ ∂ ∂ + u0y ρūy u0y + u0y ρu0y ūy + u0y ρu0y u0y + u0y ρūz u0y + u0y ρu0z ūy + u0y ρu0z u0y .

∂y ∂y ∂y ∂z ∂z ∂z | {z } | {z } | {z } | {z } | {z } | {z } 139 140 141 142 143 144 i) ∂ 0 ∂ ∂ 0 ∂ ∂ ∂ u0z ρūx uz + u0z ρu0x ūz + u0z ρu0x uz + u0z ρūy u0z + u0z ρu0y ūz + u0z ρu0y u0z | {z∂x } | {z ∂x } | {z ∂x } | ∂y ∂y ∂y {z } | {z } | {z } 145 146 147 148 149 150 ∂ ∂ ∂ ∂ 0 ∂ ∂ 0 + u0z ρūz u0z + u0z ρu0z ūz + u0z ρu0z u0z + u0x ρūx uy + u0x ρu0x ūy + u0x ρu0x uy | {z∂z } | {z∂z } | {z∂z } | {z∂x } | {z ∂x } | {z ∂x } 151 152 153 154 155 156 ∂ ∂ ∂ ∂ ∂ ∂ + u0x ρūy u0y + u0x ρu0y ūy + u0x ρu0y u0y + u0x ρūz u0y + u0x ρu0z ūy + u0x ρu0z u0y . ∂y ∂y ∂y ∂z ∂z ∂z | {z } | {z } | {z } | {z } | {z } | {z } 103 104 105 106 107 108 g) ∂ 0 ∂ ∂ 0 ∂ ∂ ∂ = u0z ρūx ux + u0z ρu0x ūx + u0z ρu0x ux + u0z ρūy u0x + u0z ρu0y ūx + u0z ρu0y u0x | {z∂x } | {z ∂x } | {z ∂x } | ∂y ∂y ∂y {z } | {z } | {z } 109 110 111 112 113 114 ∂ ∂ ∂ ∂ 0 ∂ ∂ 0 + u0z ρūz u0x + u0z ρu0z ūx + u0z ρu0z u0x + u0x ρūx uz + u0x ρu0x ūz + u0x ρu0x uz | ∂z {z } | {z∂z } | {z∂z } | {z∂x } | {z ∂x } | {z ∂x } 115 116 117 118 119 120 ∂ ∂ ∂ ∂ ∂ ∂ + u0x ρūy u0z + u0x ρu0y ūz + u0x ρu0y u0z + u0x ρūz u0z + u0x ρu0z ūz + u0x ρu0z u0z . we figure out that there are different kind of terms: . ∂y ∂y ∂y ∂z ∂z ∂z | {z } | {z } | {z } | {z } | {z } | {z } 121 122 123 124 125 126 h) ∂ 0 ∂ ∂ 0 ∂ ∂ ∂ u0z ρūx uy + u0z ρu0x ūy + u0z ρu0x uy + u0z ρūy u0y + u0z ρu0y ūy + u0z ρu0y u0y | {z∂x } | {z ∂x } | {z ∂x } | ∂y ∂y ∂y {z } | {z } | {z } 127 128 129 130 131 132 ∂ ∂ ∂ ∂ 0 ∂ ∂ 0 + u0z ρūz u0y + u0z ρu0z ūy + u0z ρu0z u0y + u0x ρūx ux + u0x ρu0x ūx + u0x ρu0x ux | {z∂z } | {z∂z } | {z∂z } | {z∂x } | {z ∂x } | {z ∂x } 133 134 135 136 137 138 ∂ ∂ ∂ ∂ ∂ ∂ + u0x ρūy u0x + u0x ρu0y ūx + u0x ρu0y u0x + u0x ρūz u0x + u0x ρu0z ūx + u0x ρu0z u0x . ∂y ∂y ∂y ∂z ∂z ∂z | {z } | {z } | {z } | {z } | {z } | {z } 157 158 159 160 161 162 Analyzing the sum of terms.122 Appendix f) ∂ 0 ∂ ∂ 0 ∂ ∂ ∂ u0y ρūx uz + u0y ρu0x ūz + u0y ρu0x uz + u0y ρūy u0z + u0y ρu0y ūz + u0y ρu0y u0z | {z∂x } | {z ∂x } | {z ∂x } | ∂y ∂y ∂y {z } | {z } | {z } 91 92 93 94 95 96 ∂ ∂ ∂ ∂ 0 ∂ ∂ 0 + u0y ρūz u0z + u0y ρu0z ūz + u0y ρu0z u0z + u0z ρūx uy + u0z ρu0x ūy + u0z ρu0x uy | {z∂z } | {z∂z } | {z∂z } | {z∂x } | {z ∂x } | {z ∂x } 97 98 99 100 101 102 ∂ ∂ ∂ ∂ ∂ ∂ + u0z ρūy u0y + u0z ρu0y ūy + u0z ρu0y u0y + u0z ρūz u0y + u0z ρu0z ūy + u0z ρu0z u0y .

(14. (14. Lets consider the terms that contains the fluctuation quantities for now: ∂ 0 ∂ ∂ ∂ 0 ∂ ∂ u0x ρu0x ux + u0x ρu0y u0x + u0x ρu0z u0x + u0y ρu0x uz + u0y ρu0y u0z + u0y ρu0z u0z | {z ∂x } | ∂y ∂z ∂x ∂y ∂z {z } | {z } | {z } | {z } | {z } 3 6 9 12 15 18 ∂ 0 ∂ ∂ ∂ 0 ∂ ∂ + u0x ρu0x uy + u0x ρu0y u0y + u0x ρu0z u0y + u0y ρu0x ux + u0y ρu0y u0x + u0y ρu0z u0x | {z ∂x } | ∂y ∂z ∂x ∂y ∂z {z } | {z } | {z } | {z } | {z } 21 24 27 30 33 36 ∂ 0 ∂ ∂ ∂ 0 ∂ ∂ + u0x ρu0x uz + u0x ρu0y u0z + u0x ρu0z u0z + u0y ρu0x uy + u0y ρu0y u0y + u0y ρu0z u0y | {z ∂x } | ∂y ∂z ∂x ∂y ∂z {z } | {z } | {z } | {z } | {z } 39 42 45 48 51 54 ∂ 0 ∂ ∂ ∂ 0 ∂ ∂ + u0y ρu0x ux + u0y ρu0y u0x + u0y ρu0z u0x + u0z ρu0x uz + u0z ρu0y u0z + u0z ρu0z u0z | {z ∂x } | ∂y ∂z ∂x ∂y ∂z {z } | {z } | {z } | {z } | {z } 57 60 63 66 69 72 ∂ 0 ∂ ∂ ∂ 0 ∂ ∂ + u0y ρu0x uy + u0y ρu0y u0y + u0y ρu0z u0y + u0z ρu0x ux + u0z ρu0y u0x + u0z ρu0z u0x | {z ∂x } | ∂y ∂z ∂x ∂y ∂z {z } | {z } | {z } | {z } | {z } 75 78 81 84 87 90 ∂ 0 ∂ ∂ ∂ 0 ∂ ∂ + u0y ρu0x uz + u0y ρu0y u0z + u0y ρu0z u0z + u0z ρu0x uy + u0z ρu0y u0y + u0z ρu0z u0y | {z ∂x } | ∂y ∂z ∂x ∂y ∂z {z } | {z } | {z } | {z } | {z } 93 96 99 102 105 108 ∂ 0 ∂ ∂ ∂ 0 ∂ ∂ + u0z ρu0x ux + u0z ρu0y u0x + u0z ρu0z u0x + u0x ρu0x uz + u0x ρu0y u0z + u0x ρu0z u0z | {z ∂x } | ∂y ∂z ∂x ∂y ∂z {z } | {z } | {z } | {z } | {z } 111 114 117 120 123 126 ∂ 0 ∂ ∂ ∂ 0 ∂ ∂ + u0z ρu0x uy + u0z ρu0y u0y + u0z ρu0z u0y + u0x ρu0x ux + u0x ρu0y u0x + u0x ρu0z u0x | {z ∂x } | ∂y ∂z ∂x ∂y ∂z {z } | {z } | {z } | {z } | {z } 129 132 135 138 141 144 ∂ 0 ∂ ∂ ∂ 0 ∂ ∂ + u0z ρu0x uz + u0z ρu0y u0z + u0z ρu0z u0z + u0x ρu0x uy + u0x ρu0y u0y + u0x ρu0z u0y .14. we will realize that not all terms can be combined. THE INCOMPRESSIBLE REYNOLDS-STRESS-EQUATION 123 • Terms that only include the fluctuation quantities . • Terms that include the mean quantity inside the derivation . Applying the product rule to the terms. One example would be: ∂ 0 0 0 ∂ 0 ∂ 0 ∂ 0 ρ uy uz ux = ρu0z u0x uy + ρu0y u0x uz + ρu0y u0z ux .13) ∂x k .1. Therefore we have to add 27 terms of the following kind: ∂ 0 ρu0i u0j uk = 0 . three terms can be combined to one term. | {z ∂x } | ∂y ∂z ∂x ∂y ∂z {z } | {z } | {z } | {z } | {z } 147 150 153 156 159 162 We get 54 terms that can be combined using the product rule.12) ∂x ∂x ∂x ∂x In other words. • Terms that include the mean quantity outside the derivation .

we end up with the convective term as: ∂ ∂ ρ u0 u0 u0 = ρu0i u0j u0k .14) ∂xk i j k ∂xk . ∂x ∂y ∂z Now it is obvious that we can rewrite this equation using the Einsteins summation convention.124 Appendix After adding these terms we end up with 81 terms that can be reduced to 27. In addition. (14. we will put the density inside the derivative due to the fact that it is constant. Finally we get: ∂ 0 0 0 ∂ ∂ ∂ ∂ ∂ ρ ux ux ux + ρ u0x u0x u0y + ρ u0x u0x u0z + ρ u0x u0y u0x + ρ u0x u0y u0y + ρ u0x u0y u0z ∂x ∂y ∂z ∂x ∂y ∂z ∂ 0 0 0 ∂ ∂ ∂ ∂ ∂ +ρ ux uz ux + ρ u0x u0z u0y + ρ u0x u0z u0z + ρ u0y u0x u0x + ρ u0y u0x u0y + ρ u0y u0x u0z ∂x ∂y ∂z ∂x ∂y ∂z ∂ 0 0 0 ∂ ∂ ∂ ∂ ∂ +ρ uy uy ux + ρ u0y u0y u0y + ρ u0y u0y u0z + ρ u0y u0z u0x + ρ u0y u0z u0y + ρ u0y u0z u0z ∂x ∂y ∂z ∂x ∂y ∂z ∂ 0 0 0 ∂ ∂ ∂ ∂ ∂ +ρ uz ux ux + ρ u0z u0x u0y + ρ u0z u0x u0z + ρ u0z u0y u0x + ρ u0z u0y u0y + ρ u0z u0y u0z ∂x ∂y ∂z ∂x ∂y ∂z ∂ 0 0 0 ∂ ∂ +ρ uz uz ux + ρ u0z u0z u0y + ρ u0z u0z u0z . After applying the Reynolds time-averaging.

THE INCOMPRESSIBLE REYNOLDS-STRESS-EQUATION 125 Now. Doing so. (14. (14. we will consider all terms that contain the mean quantity inside the derivative: ∂ ∂ ∂ ∂ ∂ ∂ u0x ρu0x ūx + u0x ρu0y ūx + u0x ρu0z ūx + u0y ρu0x ūz + u0y ρu0y ūz + u0y ρu0z ūz | {z ∂x } | ∂y ∂z ∂x ∂y ∂z {z } | {z } | {z } | {z } | {z } 2 5 8 11 14 17 ∂ ∂ ∂ ∂ ∂ ∂ + u0x ρu0x ūy + u0x ρu0y ūy + u0x ρu0z ūy + u0y ρu0x ūx + u0y ρu0y ūx + u0y ρu0z ūx | {z ∂x } | ∂y ∂z ∂x ∂y ∂z {z } | {z } | {z } | {z } | {z } 20 23 26 29 32 35 ∂ ∂ ∂ ∂ ∂ ∂ + u0x ρu0x ūz + u0x ρu0y ūz + u0x ρu0z ūz + u0y ρu0x ūy + u0y ρu0y ūy + u0y ρu0z ūy | {z ∂x } | ∂y ∂z ∂x ∂y ∂z {z } | {z } | {z } | {z } | {z } 38 41 44 47 50 53 ∂ ∂ ∂ ∂ ∂ ∂ + u0y ρu0x ūx + u0y ρu0y ūx + u0y ρu0z ūx + u0z ρu0x ūz + u0z ρu0y ūz + u0z ρu0z ūz | {z ∂x } | ∂y ∂z ∂x ∂y ∂z {z } | {z } | {z } | {z } | {z } 56 59 62 65 68 71 ∂ ∂ ∂ ∂ ∂ ∂ + u0y ρu0x ūy + u0y ρu0y ūy + u0y ρu0z ūy + u0z ρu0x ūx + u0z ρu0y ūx + u0z ρu0z ūx | {z ∂x } | ∂y ∂z ∂x ∂y ∂z {z } | {z } | {z } | {z } | {z } 74 77 80 83 86 89 ∂ ∂ ∂ ∂ ∂ ∂ + u0y ρu0x ūz + u0y ρu0y ūz + u0y ρu0z ūz + u0z ρu0x ūy + u0z ρu0y ūy + u0z ρu0z ūy | {z ∂x } | ∂y ∂z ∂x ∂y ∂z {z } | {z } | {z } | {z } | {z } 92 95 98 101 104 107 ∂ ∂ ∂ ∂ ∂ ∂ + u0z ρu0x ūx + u0z ρu0y ūx + u0z ρu0z ūx + u0x ρu0x ūz + u0x ρu0y ūz + u0x ρu0z ūz | {z ∂x } | ∂y ∂z ∂x ∂y ∂z {z } | {z } | {z } | {z } | {z } 110 113 116 119 122 125 ∂ ∂ ∂ ∂ ∂ ∂ + u0z ρu0x ūy + u0z ρu0y ūy + u0z ρu0z ūy + u0x ρu0x ūx + u0x ρu0y ūx + u0x ρu0z ūx | {z ∂x } | ∂y ∂z ∂x ∂y ∂z {z } | {z } | {z } | {z } | {z } 128 131 134 137 140 143 ∂ ∂ ∂ ∂ ∂ ∂ + u0z ρu0x ūz + u0z ρu0y ūz + u0z ρu0z ūz + u0x ρu0x ūy + u0x ρu0y ūy + u0x ρu0z ūy .1. Finally we have to consider all terms that contain the mean of the quantities outside of the .16) ∂xk ∂xk Note: If you want to check if everything is fine with the above equation.14.15) ∂xk ∂xk and the second 27 terms could be written as: ∂ ∂ ρu0j u0k ūi = ρu0j u0k ūi . | {z ∂x } | ∂y ∂z ∂x ∂y ∂z {z } | {z } | {z } | {z } | {z } 146 149 152 155 158 161 Again. we end up with 54 terms. just build the sum of the two last equations and you will see that you get the 52 terms. we will see that we can simplify the first 27 terms using the Einsteins summation convention to: ∂ ∂ ρu0i u0k ūj = ρu0i u0k ūj . This terms can be sorted and rearranged into twice 27 terms.

126 Appendix

derivative:

∂ 0 ∂ ∂ ∂ 0 ∂ ∂
u0x ρūx ux + u0x ρūy u0x + u0x ρūz u0x + u0y ρūx uz + u0y ρūy u0z + u0y ρūz u0z
| {z∂x } | ∂y ∂z ∂x ∂y ∂z
{z } | {z } | {z } | {z } | {z }
1 4 7 10 13 16

∂ 0 ∂ ∂ ∂ 0 ∂ ∂
+ u0x ρūx uy + u0x ρūy u0y + u0x ρūz u0y + u0y ρūx ux + u0y ρūy u0x + u0y ρūz u0x
| {z∂x } | ∂y ∂z ∂x ∂y ∂z
{z } | {z } | {z } | {z } | {z }
19 22 25 28 31 34

∂ 0 ∂ ∂ ∂ 0 ∂ ∂
+ u0x ρūx uz + u0x ρūy u0z + u0x ρūz u0z + u0y ρūx uy + u0y ρūy u0y + u0y ρūz u0y
| {z∂x } | ∂y ∂z ∂x ∂y ∂z
{z } | {z } | {z } | {z } | {z }
37 40 43 46 49 52

∂ 0 ∂ ∂ ∂ 0 ∂ ∂
+ u0y ρūx ux + u0y ρūy u0x + u0y ρūz u0x + u0z ρūx uz + u0z ρūy u0z + u0z ρūz u0z
| {z∂x } | ∂y ∂z ∂x ∂y ∂z
{z } | {z } | {z } | {z } | {z }
55 58 61 64 67 70

∂ 0 ∂ ∂ ∂ 0 ∂ ∂
+ u0y ρūx uy + u0y ρūy u0y + u0y ρūz u0y + u0z ρūx ux + u0z ρūy u0x + u0z ρūz u0x
| {z∂x } | ∂y ∂z ∂x ∂y ∂z
{z } | {z } | {z } | {z } | {z }
73 76 79 82 85 88

∂ 0 ∂ ∂ ∂ 0 ∂ ∂
+ u0y ρūx uz + u0y ρūy u0z + u0y ρūz u0z + u0z ρūx uy + u0z ρūy u0y + u0z ρūz u0y
| {z∂x } | ∂y ∂z ∂x ∂y ∂z
{z } | {z } | {z } | {z } | {z }
91 94 97 100 103 106

∂ 0 ∂ ∂ ∂ 0 ∂ ∂
+ u0z ρūx ux + u0z ρūy u0x + u0z ρūz u0x + u0x ρūx uz + u0x ρūy u0z + u0x ρūz u0z
| {z∂x } | ∂y ∂z ∂x ∂y ∂z
{z } | {z } | {z } | {z } | {z }
109 112 115 118 121 124

∂ 0 ∂ ∂ ∂ 0 ∂ ∂
+ u0z ρūx uy + u0z ρūy u0y + u0z ρūz u0y + u0x ρūx ux + u0x ρūy u0x + u0x ρūz u0x
| {z∂x } | ∂y ∂z ∂x ∂y ∂z
{z } | {z } | {z } | {z } | {z }
127 130 133 136 139 142

∂ 0 ∂ ∂ ∂ 0 ∂ ∂
+ u0z ρūx uz + u0z ρūy u0z + u0z ρūz u0z + u0x ρūx uy + u0x ρūy u0y + u0x ρūz u0y .
| {z∂x } | ∂y ∂z ∂x ∂y ∂z
{z } | {z } | {z } | {z } | {z }
145 148 151 154 157 160

To simplify these terms, we use the product rule (1.2) to combine two terms to one. An example
would be:
∂ 0 0 ∂ ∂
ρūz ux uy = u0y ρūz u0x + u0x ρūz u0y . (14.17)
∂z ∂z ∂z
After combining the terms, we can rewrite the sum by using the Einsteins summation convention
because the derivatives are always with respect to the mean quantities. Hence, we end up with 27
terms that can be expressed as:

∂ 0 0 ∂
ρūk u u = ūk ρu0 u0 . (14.18)
∂xk i j ∂xk i j

Now, the convective term is manipulated and derived. Combining all terms, we end up with:

∂ρu0i u0j ∂ ūi ∂ ūj ∂
ūk + ρu0j u0k + ρu0i u0k + ρu0 u0 u0 . (14.19)
∂xk ∂xk ∂xk ∂xk i j k

14.1. THE INCOMPRESSIBLE REYNOLDS-STRESS-EQUATION 127

The Shear-Rate Term
a)

∂(ūy + u0y )
( " !#
∂ ∂(ūx + u0x ) ∂(ūx + u0x ) ∂ ∂(ūx + u0x )
  
u0x − µ + − µ +
∂x ∂x ∂x ∂y ∂y ∂x

∂ ∂(ūx + u0x ) ∂(ūz + u0z ) ∂ ∂(ūz + u0z ) ∂(ūx + u0x )
      
− µ + + u0y − µ +
∂z ∂z ∂x ∂x ∂x ∂z
∂(ūy + u0y ) ∂(ūy + u0y )
" !# " !#)
∂ ∂(ūz + u0z ) ∂ ∂(ūz + u0z )
− µ + − µ +
∂y ∂y ∂z ∂z ∂z ∂z

∂ ∂ ū ∂ ∂u0x ∂ ∂ ū ∂ ∂u0x ∂ ∂ ū ∂ ∂u0x
           
x x x
= − u0x µ − u0x − u0x µ − u0x − u0x  − u0x
µ
 µ µ µ
 ∂x ∂x ∂x ∂x  ∂x ∂x ∂x ∂x  ∂y ∂y ∂y ∂y
∂u0y
!
∂ ∂ ū ∂ ∂ ∂ ū ∂ ∂u0x ∂ ∂ ū ∂ ∂u0z
          
0 y 0 x z
− u0x  0 u0x  − u0x
− ux  − ux
 
 ∂y

∂x ∂y
µ
∂x  ∂z µ ∂z − ux ∂z µ ∂z −  ∂z
µ
∂x ∂z
µ
∂x

∂ ∂ ū ∂ ∂u0z ∂ ∂ ū ∂ ∂u0x ∂ ∂ ū ∂ ∂u0z
             
z x z
− u0y µ − u0y − u0y µ − u0y − u0y  − u0y
 µ
 µ µ µ
 ∂x ∂x ∂x ∂x  ∂x ∂z ∂x ∂z  ∂y ∂y ∂y ∂y
∂u0y
!
∂ ∂ ū ∂ ∂ ∂ ū ∂ ∂u0z ∂ ∂ ū ∂ ∂u0z
            
y z z
− u0y  0 u0y  − u0y − u0y  0
  
 ∂y µ ∂z − uy ∂y µ ∂z −  ∂z
µ
∂z ∂z
µ
∂z  ∂z µ ∂z − uy ∂z µ ∂z

∂u0y
!
∂ ∂u0x ∂ ∂u0x ∂ ∂u0x ∂ ∂ ∂u0x ∂ ∂u0z
         
= − u0x µ − u0x µ − u0x µ − u0x µ − u0x µ − u0x µ
∂x ∂x ∂x ∂x ∂y ∂y ∂y ∂x ∂z ∂z ∂z ∂x

∂u0y
!
∂ ∂u0z ∂ ∂u0x ∂ ∂u0z ∂ ∂ ∂u0z ∂ ∂u0z
         
− u0y µ − u0y µ − u0y µ − u0y µ − u0y µ − u0y µ
∂x ∂x ∂x ∂z ∂y ∂y ∂y ∂z ∂z ∂z ∂z ∂z

For the parts b) to i) we will just write the final terms:
b)

∂u0y ∂u0y ∂u0y ∂u0y
! ! ! !
∂ ∂ ∂u0x ∂ ∂ ∂ ∂ ∂u0z
   
− u0x µ − u0x µ − u0x µ − u0x µ − u0x µ − u0x µ
∂x ∂x ∂x ∂y ∂y ∂y ∂y ∂y ∂z ∂z ∂z ∂y

∂u0y
!
∂ ∂u0x ∂ ∂u0x ∂ ∂u0x ∂ ∂ ∂u0x ∂ ∂u0z
         
− u0y µ − u0y µ − u0y µ − u0y µ − u0y µ − u0y µ
∂x ∂x ∂x ∂x ∂y ∂y ∂y ∂x ∂z ∂z ∂z ∂x

c)

∂u0y
!
∂ ∂u0z ∂ ∂u0x
∂u0z ∂ ∂ ∂u0z ∂ ∂ ∂u0z
        
− u0x µ − u0x µ µ − u0x
− u0x µ − u0x
µ − u0x µ
∂x ∂x ∂x ∂z ∂y ∂y ∂y ∂z ∂z ∂z ∂z ∂z

∂u0y ∂u0y ∂u0y ∂u0y
! ! ! !
∂ ∂ ∂u0x ∂ ∂ ∂ ∂ ∂u0z
   
− u0y µ − u0y µ − u0y µ − u0y µ − u0y µ − u0y µ
∂x ∂x ∂x ∂y ∂y ∂y ∂y ∂y ∂z ∂z ∂z ∂y

d)

∂u0y
!
∂ ∂u0x ∂ ∂u0x ∂ ∂u0x ∂ ∂ ∂u0x ∂ ∂u0z
         
− u0y µ − u0y µ − u0y µ − u0y µ − u0y µ − u0y µ
∂x ∂x ∂x ∂x ∂y ∂y ∂y ∂x ∂z ∂z ∂z ∂x

∂u0y
!
∂u0z
∂ ∂ ∂u0x ∂ ∂u0z ∂ ∂ ∂u0z ∂ ∂u0z
         
− u0z µ − u0z µ − u0z µ − u0z µ − u0z µ − u0z µ
∂x ∂x ∂x ∂z ∂y ∂y ∂y ∂z ∂z ∂z ∂z ∂z

e)

∂u0y ∂u0y ∂u0y ∂u0y
! ! ! !
∂ ∂ ∂u0x ∂ ∂ ∂ ∂ ∂u0z
   
− u0y µ − u0y µ − u0y µ − u0y µ − u0y µ − u0y µ
∂x ∂x ∂x ∂y ∂y ∂y ∂y ∂y ∂z ∂z ∂z ∂y

∂u0y
!
∂ ∂u0x ∂ ∂u0x ∂ ∂u0x ∂ ∂ ∂u0x ∂ ∂u0z
         
− u0z µ − u0z µ − u0z µ − u0z µ − u0z µ − u0z µ
∂x ∂x ∂x ∂x ∂y ∂y ∂y ∂x ∂z ∂z ∂z ∂x

128 Appendix

f)

∂u0y
!
∂ ∂u0z ∂ ∂u0x ∂ ∂u0z ∂ ∂ ∂u0z ∂ ∂u0z
         
− u0y µ −u0y µ − u0y µ − u0y µ − u0y µ − u0y µ
∂x ∂x ∂x ∂z ∂y ∂y ∂y ∂z ∂z ∂z ∂z ∂z
| {z }

∂u0y ∂u0y ∂u0y ∂u0y
! ! ! !
∂ ∂ ∂u0x ∂ ∂ ∂ ∂ ∂u0z
   
− u0z µ − u0z µ − u0z µ − u0z µ − u0z µ − u0z µ
∂x ∂x ∂x ∂y ∂y ∂y ∂y ∂y ∂z ∂z ∂z ∂y

g)

∂u0y
!
∂ ∂u0x ∂ ∂u0x ∂ ∂u0x ∂ ∂ ∂u0x ∂ ∂u0z
         
− u0z µ − u0z µ − u0z µ − u0z µ − u0z µ − u0z µ
∂x ∂x ∂x ∂x ∂y ∂y ∂y ∂x ∂z ∂z ∂z ∂x

∂u0y
!
∂ ∂u0z ∂ ∂u0x ∂ ∂u0z ∂ ∂ ∂u0z ∂ ∂u0z
         
− u0x µ − u0x µ − u0x µ − u0x µ − u0x µ − u0x µ
∂x ∂x ∂x ∂z ∂y ∂y ∂y ∂z ∂z ∂z ∂z ∂z

h)

∂u0y ∂u0y ∂u0y ∂u0y
! ! ! !
∂ ∂ ∂u0x ∂ ∂ ∂ ∂ ∂u0z
   
− u0z µ −u0z µ − u0z µ − u0z µ − u0z µ − u0z µ
∂x ∂x ∂x ∂y ∂y ∂y ∂y ∂y ∂z ∂z ∂z ∂y
| {z }
∗∗

∂u0y
!
∂ ∂u0x ∂ ∂u0x ∂ ∂u0x ∂ ∂ ∂u0x ∂ ∂u0z
         
− u0x µ − u0x µ − u0x µ − u0x µ − u0x µ − u0x µ
∂x ∂x ∂x ∂x ∂y ∂y ∂y ∂x ∂z ∂z ∂z ∂x

i)

∂u0y
!
∂ ∂u0z ∂u0x
∂ ∂ ∂u0z ∂ ∂ ∂u0z ∂ ∂u0z
         
− u0z µ − u0z µ − u0z µ − u0z µ − u0z µ − u0z µ
∂x ∂x ∂x ∂z ∂y ∂y ∂y ∂z ∂z ∂z ∂z ∂z

∂u0y ∂u0y ∂u0y ∂u0y
! ! ! !
∂ ∂ ∂u0x ∂ ∂ ∂ ∂ ∂u0z
   
− u0x µ − u0x µ − u0x µ − u0x µ − u0x µ − u0x µ
∂x ∂x ∂x ∂y ∂y ∂y ∂y ∂y ∂z ∂z ∂z ∂y

After the manipulation we find 102 terms. Again we want to put the fluctuation terms together,
hence we need the product rule. Analyzing the sum, we can figure out that there is no way to
apply the product rule. The trick is simply to add the missing 204 terms. One example is given
now. Taking the term (∗ ) of f) and (∗∗ ) of h), we get:

∂u0y
   
∂ ∂u0 ∂
− u0y µ z − u0z µ . (14.20)
∂x ∂x ∂x ∂x

This two terms cannot be merged, hence we need two terms in addition:

1 2 3 4
z }| { z }| { z }| { z }| {
0 0 0
 0
 0 0
 
∂ ∂u ∂u y ∂u ∂u y ∂u ∂ ∂u y
− u0y µ z −µ z
+µ z
− u0z µ
∂x ∂x | ∂x ∂x{z ∂x ∂x} ∂x ∂x
added
5 6
z
}| { z }| {
∂u0z ∂u0y ∂u0 ∂u0y
−µ +µ z . (14.21)
| ∂x ∂x {z ∂x ∂x}
added

Furthermore term 3 and 6 are similar and can be combined too: ∂u0y ∂u0 ∂u0y     ∂ 0 ∂u0 ∂ 0 − uy µ z − uz µ +2µ z .26) ∂xk ∂xk ∂xk ∂xk The pressure derivative a) ∂(p̄ + p0 ) ∂(p̄ + p0 ) ∂ p̄ ∂p0 ∂ p̄ ∂p0 ∂p0 ∂p0 u0x + u0y = u0x + u0x + u0y + u0y = u0x + u0y .25) ∂xk ∂xk ∂xk ∂xk Summing up. (14.23) ∂x ∂x ∂x ∂x Repeating this procedure for all terms. (14. (14.28) ∂y ∂x ∂y ∂y ∂x ∂x ∂y ∂x c) ∂(p̄ + p0 ) ∂(p̄ + p0 ) ∂ p̄ ∂p0 ∂ p̄ ∂p0 ∂p0 ∂p0 u0x + u0y = u0x + u0x + u0y + u0y = u0x + u0y . can be expressed as: ∂u0i ∂u0j ∂u0 ∂u0j 2µ = 2µ i . (14.27) ∂x ∂z ∂x ∂x ∂z ∂z ∂x ∂z b) ∂(p̄ + p0 ) ∂(p̄ + p0 ) ∂ p̄ ∂p0 ∂ p̄ ∂p0 ∂p0 ∂p0 u0x + u0y = u0x + u0x + u0y + u0y = u0x + u0y .14. (14. (14.1.29) ∂z ∂y ∂z ∂z ∂y ∂y ∂z ∂y d) ∂(p̄ + p0 ) ∂(p̄ + p0 ) ∂ p̄ ∂p0 ∂ p̄ ∂p0 ∂p0 ∂p0 u0y + u0z = u0y + u0y + u0z + u0z = u0y + u0z . (14.31) ∂y ∂x ∂y ∂y ∂x ∂x ∂y ∂x . the shear-rate terms can be written as: ! ∂ ∂u0i u0j ∂u0i ∂u0j − µ + 2µ .24) ∂xk ∂xk ∂xk ∂xk The new introduced terms (due to the trick).22) ∂x ∂x ∂x ∂x ∂x ∂x | {z } | {z } 7 8 Now we see that the term 7 and 8 can be combined using the product rule again. THE INCOMPRESSIBLE REYNOLDS-STRESS-EQUATION 129 Now we are able to combine term 1 – 2 and 4 – 5 using the product rule. we will reduce the already existing 102 terms of a) to i) to 54. (14. At the end we would realize that we can rewrite the sum of the 54 terms as: ! ∂u0i u0j ∂u0i u0j   ∂ ∂ − µ =− µ . (14.30) ∂x ∂z ∂x ∂x ∂z ∂z ∂x ∂z e) ∂(p̄ + p0 ) ∂(p̄ + p0 ) ∂ p̄ ∂p0 ∂ p̄ ∂p0 ∂p0 ∂p0 u0y + u0z = u0y + u0y + u0z + u0z = u0y + u0z . (14. Finally we end up with: ∂u0y u0z ∂u0z ∂u0y   ∂ − µ + 2µ .

35) ∂z ∂y ∂z ∂z ∂y ∂y ∂z ∂y Summing up and sorting: ∂p0 ∂p0 ∂p0 ∂p0 ∂p0 ∂p0 ∂p0 ∂p0 ∂p0 u0x + u0x + u0x + u0y + u0y + u0y + u0z + u0z + u0z ∂x ∂y ∂z ∂x ∂y ∂z ∂x ∂y ∂z ∂p0 ∂p0 ∂p0 ∂p0 ∂p0 ∂p0 ∂p0 ∂p0 ∂p0 + u0x + u0y + u0z + u0x + u0y + u0z + u0x + u0y + u0z . (14. (14. (14.41) ∂xj ∂xi ∂xk Now the derivation of the Reynolds-Stress equation is done. By using the Kronecker delta function (due to the fact that the pressure is only in the main diagonal of a matrix).34) ∂y ∂x ∂y ∂y ∂x ∂x ∂y ∂x i) ∂(p̄ + p0 ) ∂(p̄ + p0 ) ∂ p̄ ∂p0 ∂ p̄ ∂p0 ∂p0 ∂p0 u0z + u0x = u0z + u0z + u0x + u0x = u0z + u0x .32) ∂z ∂y ∂z ∂z ∂y ∂y ∂z ∂y g) ∂(p̄ + p0 ) ∂(p̄ + p0 ) ∂ p̄ ∂p0 ∂ p̄ ∂p0 ∂p0 ∂p0 u0z + u0x = u0z + u0z + u0x + u0x = u0z + u0x . (14. we get: ∂u0i ∂u0j   ∂ h 0 0 i −p0 + + p uj δik + p0 u0i δjk . (14. (14.37) ∂xj ∂xi Finally we use the product rule.130 Appendix f) ∂(p̄ + p0 ) ∂(p̄ + p0 ) ∂ p̄ ∂p0 ∂ p̄ ∂p0 ∂p0 ∂p0 u0y + u0z = u0y + u0y + u0z + u0z = u0y + u0z .33) ∂x ∂z ∂x ∂x ∂z ∂z ∂x ∂z h) ∂(p̄ + p0 ) ∂(p̄ + p0 ) ∂ p̄ ∂p0 ∂ p̄ ∂p0 ∂p0 ∂p0 u0z + u0x = u0z + u0z + u0x + u0x = u0z + u0x . Putting all terms together. (14.36) ∂x ∂x ∂x ∂y ∂y ∂y ∂z ∂z ∂z leads to the following expression: ∂p0 ∂p0 u0i + u0j . ∂p0 ∂u0 ∂p0 u0i u0i = −p0 i + .39) ∂xi ∂xi ∂xi ∂u0i ∂p0 u0i ∂u0j ∂p0 u0j − p0 + − p0 + . (14.38) ∂xj ∂xj ∂xj ∂p0 ∂u0j ∂p0 u0j u0j = −p0 + . we can write the Reynolds-stress equation in the following form: ! ∂ρu0j u0i ∂ρu0i u0j 0 0 ∂ ūi 0 0 ∂ ūj ∂ 0 0 0 ∂ ∂u0i u0j + ūk + ρuj uk + ρui uk + ρu u u − µ ∂t ∂xk ∂xk ∂xk ∂xk i j k ∂xk ∂xk 0 0 ∂u0 ∂uj ∂u0i ∂uj   ∂ h i + 2µ i − p0 + + p0 u0j δik + p0 u0i δjk = 0 . (14. (14.40) ∂xj ∂xj ∂xi ∂xi to get the final form.42) ∂xk ∂xk ∂xj ∂xi ∂xk . (14.

we can write the common Reynolds-Stress equation: ∂ σ̄tji ∂ σ̄tij ∂ σ̄tij   ∂ ūi ∂ ūj ∂ + ūk = −σ̄tjk − σ̄tik + ν + Cijk + ij − Πij .44) ∂t ∂xk ∂xk ∂xk ∂xk ∂xk . Hence.40).1. we end up with the following form. (14.14. re-order the terms and multiply the whole equation by −1. recall: In almost all literatures we find the definition of the Reynolds-Stress tensor denoted by τ .43) ∂xj ∂xi ∂xk ∂xk | {z } |  {z  } Πij 0 0 0 0 0 0 0 ∂ ∂xk ρui uj uk + p uj δik + p ui δjk | {z } Cijk Finally. THE INCOMPRESSIBLE REYNOLDS-STRESS-EQUATION 131 Applying the definition of the Reynolds-Stress tensor (9. (14. we get: ∂ σ̄tji ∂ σ̄tij ∂ σ̄tij ∂u0 ∂u0j   ∂ ūi ∂ ūj ∂ + ūk = −σ̄tjk − σ̄tik + ν + 2µ i ∂t ∂xk ∂xk ∂xk ∂xk ∂xk ∂xk ∂xk | {z } ij ∂u0i ∂u0j   ∂ ∂ h 0 0 i − p0 + + ρu0i u0j u0k + p uj δik + p0 u0i δjk . In addition we use the relation between the kinematic and dynamic viscosity: µ = νρ.

132 Appendix .

first edition. Switzer- land. CO. Transport Phenomena. Anderson. The pimple algorithm in openfoam. August 1998. 2009. Open∇FOAMr Programmers Guide. doi: 10. dpunkt. 2014. volume 113 of Fluid Mechanics and its applications. Madison. Springer-Vieweg. The Finite Volume Method in Com- putational Fluid Dynamics An Advanced Introduction with OpenFOAM(R) and Matlab (Fluid Mechanics and Its Applications). PhD thesis.researchgate. Springer- Verlag. Ferziger and Milovan Perić. 2014. CH-1015 Lausanne. 1995.. Christopher J.1002/(SICI)1097-0207(20000520)48:23. Numerische Strömungsmechanik. Hrvoje Jasak. Computational Fluid Dynamics .0. Erickson. Error Analysis and Estimation for the Finite Volume Method with Applications to Fluid Flows. Morton E.net/ index. The Mechanics and Thermodynamics of Con- tinua. URL https://www.php/OpenFOAM guide/The PIMPLE algorithm in OpenFOAM.The Basics With Applications. and Luca Mangani. Gurtin. first edition. Tobias Holzmann. Wisconsin. second edition. 2013. doi: DOI10. Marwan Darwish. CFD Modellierung. Jonathan A. Lightfoot. June 1960. Berlin. 2015. Springer-Verlag.net/publication/2294055 Application of the Finite Volume Method and Unstructured Meshes to Linear Elasticity. R. 1996. Weller. and Lallit Anand. first edition. Heidelberg. University of London. Byron Bird. Greenshields. December 2015. Hrvoje Jasak and Henry G. Solidification. 2008. John Wiley & Sons. URL http://foam.1007/978-3-319-16874-6. Joel H. OpenFOAM Foundation Ltd. In International journal for Numerical Meth- ods in Engineering.verlag GmbH. J.1 edition.Bibliography J.2-Q. Heidelberg. Fadl Moukalled.pdf. Hacking. Berlin. EPFL Press. Warren E. Eliot Fried. and Edwin N.0. Dantzig and Michel Rappaz. 133 . 2010. Berlin.sourceforge. URL https://openfoamwiki.net/docs/Guides-a4/ ProgrammersGuide. Application of the finite volume method and un- structured meshes to linear elastics. Heidelberg. version 3. Rüdiger Schwarze. Stewart.

Wilcox. An Introduction to Computational Fluid Dynamics.134 BIBLIOGRAPHY H. Turbulence Modeling for CFD. . David C. Malalasekera. 1995. first edition. California 91011. DCW Industries. Longman Group. La Canada. November 1994. Versteeg and W. K. 5354 Palm Drive. first edition.

D. my name is Tobias Holzmann from Germany (Bavaria). I started to compare the quantitative results of different phenomena using ANSYSr CFX and OpenFOAMr . I am also active in the bug- tracking system as well as generating feature patches. Good luck and all the best to you.About the Author Hello everybody. The main topics were thermal stress analysis. mainly in the conjugated heat transfer section. The topic was the investigation into local heat treatments for aluminum alloys while coupling different phenomena and toolboxes. After that. in 2014. I started to published my knowledge in the field of numerical simulations and OpenFOAMr on my private website in order to help other people. However. matrix algebra. I finished my master study by writing a thesis about biomass combustion using the flamelet model in OpenFOAMr . The result was a new developed framework that handles all these three topics in an automatic way. mont.de as well as in the known cfd-online. I am a moderator in the German OpenFOAMr forum namely cad. Additionally. derivations and advanced programming in C++. Since 2009 I am working in the field of numerical simulations. I decided to investigate more into the field of numerical mathematics. Additionally. Tobias Holzmann . After I started my Ph.com forum. In 2014 I started my Ph.D.holzmann-cfd. During my studies at the Fachhochschule Augsburg. my personal focus was set on all topics including numerical simulations. Dr. In 2017 I became an official contributor to the OpenFOAM Foundation toolbox. I always tried to go beyond the limits of OpenFOAMr while publishing script based tutorials on my personal website. During my Master study. During that time.de. I decided to write my Bachelor thesis in the field of heat transfer while us- ing numerical tools namely OpenFOAMr and vali- dating the results against measurements. at the Montanuniversität Leoben. All my recent projects and investigations can be found on my websit www. material calculation for local heat treatment in 3D and optimization.