ME 654 COMPUTATIONAL FLUID

DYNAMICS

Prof. Dr. Haşmet Türkoğlu

Çankaya University
Faculty of Engineering
Mechanical Engineering Department

Spring, 2025
INTRODUCTION

What is Computational Fluid Dynamics (CFD)?

Computational Fluid Dynamics (CFD) = Hesaplamalı Akışkanlar Dinamiği (HAD)

Computational fluid dynamics (CFD) is a technique for analysis of problems involving

fluid flow using computers and numerical methods.
Why use CFD?
CFD is a valuable tool for,
• Analysis and Design
Simulation-based design instead of “build and test”
- More cost effective and more rapidly than experiments
- Provides high-fidelity database for analysis of flow fields
Simulation of flow phenomena that are difficult to be measured by experiments
- Full scale simulations (e.g., full-scale ships, airplanes)
- Hazards (e.g., explosions, radiation, pollution)
• Gaining knowledge and exploration of flow physics

CFD has been successfully applied in many areas of fluid mechanics:

Aerospace
Appliances
Automotive
Biomedical
Chemical Processing
HVAC&R
Hydraulics
Marine
Oil & Gas
Power Generation
Sports

1 Introduction and
2
Mathematical Formulation
 Methods Used in Research
1. Experimental Method: A physical model of the problem is set up in the
laboratory and important parameters characterizing the problem are measured.

2. Theoretical (Analytical) Method: The problem is modelled mathematically, and

mathematical equations are solved using analytical approaches.

3. Numerical Method: The problem is modelled mathematically, and

mathematical equations are solved using numerical techniques.

General Aim of Solution of a Flow Problem is to predict the distribution of primary

variables such as velocity, pressure, temperature, species concentration, and other
relevant variables, i.e., stress, heat flux, force, heat transfer rate, etc.

u=?, v=?, w=?, P=?, T=?, C=? → t=?, FD=?, CL=?, f, CD=?, q=?, Nu=?, ….

Method Advantages Disadvantages

Experimental Method Capable of producing the The cost is high and may
most realistic answers for take long time
many problems.
Theoretical (Analytical) Answers are obtained in Simplifying assumptions
Method the form of clear general are made.
expressions
Restricted to simple
geometries and physical
phenomenon
Numerical Method No restriction to the Results contain error
equations (equations
could be linear or For some cases may be
nonlinear). costly

Complicated physics can

be treated.

Time evolution of the

process can be obtained.

1 Introduction and
Mathematical Formulation 3
Basic Feature of CFD
We seek a solution in terms of values of the dependent (primary) variables (u, v, w, p,
T, ….) at a finite number of grid points called discretization points.

Continuous curve Discrete approximation

Basic Steps of Solution of a Flow Problem Using CFD

Followings are common basic steps in solution of a flow problem using CFD:

1. Mathematical formulation (modelling) of the problem: In this step, the physical

phenomena (flow, heat transfer, etc.) under consideration is expressed in terms
of mathematical terms. Using the basic laws (conservation of mass, Newton's
second law, conservation of energy, etc.), equations governing the problem, and
the boundary and initial conditions are expressed in terms of mathematical
terms. These are generally partial differential equations.
   2 
 u u u u  p   u  2
u  2
u 
Example :   u v  w    gx   2  2  2 
 t x 0 y 0 z  x  x y z 
 0   0 0 
0
0

BC 1: at y=0, u=0 BC 2: At y=a, u=0

2. The flow domain is discretized (Mesh generation): Problem domain is

approximated with finite number of points (nodes, grid points, finite control
volume, finite elements, etc.).

3. Discretization of Equations: Mathematical equations are discretized and

approximated by algebraic equations at discrete points (node, grid points). This
step yields a set of linear algebraic equations for all the dependent variables of
the problem.

4. Solution of set of algebraic equations: The discretization equations (algebraic

equations) are solved, and the values of the dependent variables are obtained at
all grid points.

5. Post processing: Evaluation of the results of the variables obtained at the grid
points and representing the results as graphs, tables, etc.
1 Introduction and Mathematical Formulation 4
 MATHEMATICAL FORMULATION
Numerical solution of any problem starts with a mathematical formulation of the
problem.

Mathematical equations of flows are obtained by applying the basic laws

(conservation of mass, Newton’s second conservation of energy) to a control or a
system volume.

To complete the mathematical formulations, boundary and initial conditions should

also be expressed in terms of mathematical terms.

Flow equations are generally non-linear partial differential equations:

Continuity Equation
Continuity equation is derived from law of conservation of mass.
𝜕𝜌 𝜕    
+ 𝜌𝑢𝑘 = 0   u     v     w  0
𝜕𝑡 𝜕𝑥𝑘 t x y z
𝜕𝜌
+ 𝛻. 𝜌𝑢 = 0
𝜕𝑡

Momentum Equations – Navier Stokes Equations

Momentum equations are derived from Newton’s second law. Momentum equations
can be written in two different forms:
1. Non-conservative form
2. Conservative form

Non-conservative form:
𝛛𝒖𝒋 𝛛𝒖𝒋 𝛛𝒑 𝛛 𝛛𝒖𝒌 𝛛 𝛛𝒖𝒊 𝛛𝒖𝒋
𝝆 + 𝝆𝒖𝒌 =− + 𝝀 + 𝝁 + + 𝝆𝒇𝒋
𝛛𝒕 𝛛𝒙𝒌 𝛛𝒙𝒋 𝛛𝒙𝒊 𝛛𝒙𝒌 𝛛𝒙𝒊 𝛛𝒙𝒋 𝛛𝒙𝒊

Conservative form:
𝛛 𝛛 𝛛𝒑 𝛛 𝛛𝒖𝒌 𝛛 𝛛𝒖𝒊 𝛛𝒖𝒋
(𝝆𝒖𝒋 ) + (𝝆𝒖𝒋 𝒖𝒌 ) = − + 𝝀 + 𝝁 + + 𝝆𝒇𝒋
𝛛𝒕 𝛛𝒙𝒌 𝛛𝒙𝒋 𝛛𝒙𝒊 𝛛𝒙𝒌 𝛛𝒙𝒊 𝛛𝒙𝒋 𝛛𝒙𝒊

Equations in conservative form contains divergence of a physical quantity.

1 Introduction and Mathematical Formulation 5

Energy Equation
Energy equation is mathematical expression for the law of conservation of energy.

𝝏𝒆 𝝏𝒆 𝝏𝒖𝒌 𝝏 𝝏𝑻
Non-conservative form: 𝝆 + 𝝆𝒖𝒌 = −𝒑 + 𝒌 +𝜱
𝝏𝒕 𝝏𝒙𝒌 𝝏𝒙𝒌 𝝏𝒙𝒋 𝝏𝒙𝒋

𝝏 𝝏 𝝏𝒖𝒌 𝝏 𝝏𝑻
Conservative form: (𝝆𝒆) + (𝝆𝒖𝒌 𝒆) = −𝒑 + 𝒌 +𝜱
𝝏𝒕 𝝏𝒙𝒌 𝝏𝒙𝒌 𝝏𝒙𝒋 𝝏𝒙𝒋

Generalized Equation for a general variable, 𝝋

The above three basic equations can be in terms of a general variable 𝝋

as follows:
𝝏 𝝏 𝝏 𝝏𝝋
(𝝆𝝋) + (𝝆𝒖𝒌 𝝋) = 𝜞 +𝑺
𝝏𝒕 𝝏𝒙𝒌 𝝏𝒙𝒋 𝝏𝒙𝒋

From the generalized equation, all governing equations can be obtained by the
following settings.

For continuity equation, set 𝝋 →1, S →0

𝛛𝒑 𝛛 𝛛𝒖𝒌
For momentum equation, 𝝋 →u, v, w, 𝜞→, S →− + 𝝀 + 𝝆𝒇𝒋
𝛛𝒙𝒋 𝛛𝒙𝒊 𝛛𝒙𝒌

𝝏𝒖𝒌
For energy equation, 𝝋 →T, 𝜞→k/Cp S → −𝒑 +𝜱
𝝏𝒙𝒌

The generalized equation show that all the governing equations consists of the
same type of terms, i.e., transient term, convective terms, diffusion terms and
source term.

Because of this, the discretization of each type of these terms are going to be
formulated for the general variable 𝝋 and then it will be adapted to all the
governing equations, i.e., continuity, momentum and energy equations.

1 Introduction and Mathematical Formulation 6

BOUNDARY CONDITIONS

1. Solid-fluid boundary: There is no slip between the fluid and boundary

interface. The velocity of fluid is the same as the velocity of the boundary.

ufluid = usolid

Temperatures of fluid and solid are the same, Tfluid = Tsolid

𝜕𝑇 𝜕𝑇
Heat flux is continuous, i.e. qʺ=-k ) = -k )
𝜕𝑛 fluid 𝜕𝑛 solid

2. Interface of two different fluids: Velocity and shear stress are continuous.

𝜕𝑢 𝜕𝑢
ufluid 1 = ufluid 2 and  )fluid 1 =  )
𝜕𝑛 𝜕𝑛 fluid 2

Heat flux and temperature are continuous, i.e.

𝜕𝑇 𝜕𝑇
Tfluid 1 = Tfluid 2 and k )fluid 1 = k )
𝜕𝑛 𝜕𝑛 fluid 2

3. At Inlet: Velocity and temperature are generally prescribed.

u = Uinlet T=Tinlet

3. At Outlet: Generally, exit values of the variables are not known prior to the
solution. They are outcome of the solution. Depending on the physics of the
problem, variation of the variables in the flow direction may be considered
negligible.
1 Introduction and Mathematical Formulation 7
 Mathematical Classification of Differential Equations

Consider the general differential equation as below:

a𝝋𝒙𝒙+b𝝋𝒙𝒚 + 𝒄𝝋𝒚𝒚 + 𝒅𝝋𝒙 + 𝒆𝝋𝒚 + 𝒇𝝋 = 𝒈(𝒙, 𝒚)

Where a, b, c, d, e, and f are coefficients, 𝝋 is a dependent variable, and x and y are

independent variables.

Linear Differential Equation: If coefficients a, b, c, d, e, and f are constant or

function of only x and y, the equation is a linear differential equation.

Nonlinear Differential Equation: If coefficients a, b, c, d, e, and f are function of 𝝋,

𝝋𝒙 and/or 𝝋𝒚, the equation is a non-linear differential equation.

𝜕2𝜑 𝜕2𝜑
𝜕𝜑 𝜑𝑥𝑥 = 𝜑𝑥𝑦 =
𝜑𝑥 = 𝜕𝑥 2 𝜕𝑥𝜕𝑦
𝜕𝑥

Elliptic Equation: If b2 – 4ac < 0, equation is an elliptic differential equation.

Parabolic Equation: If b2 – 4ac = 0, equation is a parabolic differential equation.

Hyperbolic Equation: If b2 – 4ac > 0, equation is a hyperbolic differential equation.

Physical Implication

Elliptic Parabolic Hyperbolic

1 Introduction and Mathematical Formulation 8

 TURBULENT FLOWS
What is Turbulence? A “random”, 3-D, time-dependent eddying motion with many
scales, superposed on a mean flow.

Characteristics of Laminar and Turbulent Flows

Motion of fluid particles in laminar and turbulent flows are quite different. In laminar
flows, fluid particles move in an orderly manner. However, in turbulent flows there
isn’t any order in the motion of fluid particles, the motion of the particles are
random.

For steady laminar and turbulent flows, the velocity variation in time would be as
follows:

Time averaged value

Time
Momentum Transfer in Laminar and Turbulent Flows:
In laminar flows, adjacent layers of fluid slide past each other without mixing.
Transfer of momentum between layers moving at different speeds occurs because of
viscous stresses.

In turbulent flows, adjacent layers continually intermingle. A net transfer of

momentum occurs because of viscous stresses and the mixing of fluid elements
from layers with different mean velocity. This mixing is a far more effective means of
transferring momentum than viscous stresses. Consequently, the mean-velocity
profile tends to be more uniform in turbulent flows.
Note: Reynolds number is the criteria we use to determine if a flow is laminar or
turbulent.
For internal flows: Vd h If Re ≤ 2300, flow is laminar
Re  If Re > 2300, flow is turbulent

Vx If Re ≤ 500,000, flow is laminar
For external flows: Re  If Re > 500,000, flow is turbulent

1 Introduction and Mathematical Formulation 9
TURBULENT FLOW EQUATIONS
As explained above, in turbulent flows all instantaneous flow variables fluctuate
around a mean value i.e.

u=U+u’ , v=V+v’ , w=W+w ’ , p=P+p’

These expressions can be written using a generic variable 𝜑, i.e. 𝜑 =Φ+ 𝜑’

Where Φ is the time averaged value, 𝜑’ is the fluctuating value.

1 ∆𝑡
The time averaged values of flow properties are obtained as Φ= ‫ 𝜑 ׬‬t dt
∆𝑡 0
To illustrate the influence of turbulence fluctuations on the mean flow we will
consider the instantaneous continuity and Navier-Stokes equations for an
incompressible flow with constant viscosity.

These equations are valid for both laminar and turbulent flows instantaneously. For
turbulent flows, because of fluctuations, velocity and pressure terms are rapidly
varying functions of time and space. At present, our mathematics can not handle such
instantaneous fluctuating variables.

To investigate the effects of fluctuations, in above equations, we replace the variables

u, v, w and p by the sum of a mean and fluctuating component. Then, take the time
average of the equations applying the following rules (Osborne Reynolds, 1895),

'  0             ' '

 '  0  
  ds   ds 
s s

1 Introduction and Mathematical Formulation 10

Then, the Reynolds averaged Navier-Stokes (RANS) equations are obtained as follows:

𝜕𝑈 1 𝜕𝑃 𝜕𝑢′2 𝜕𝑢′ 𝑣 ′ 𝜕𝑢′ 𝑤 ′

+ 𝑑𝑖𝑣 𝑈𝑉 = − + 𝑣 𝑑𝑖𝑣 (𝑔𝑟𝑎𝑑 𝑈) + − − −
𝜕𝑡 𝜌 𝜕𝑥 𝜕𝑥 𝜕𝑦 𝜕𝑧

𝜕𝑉 1 𝜕𝑃 𝜕𝑢′ 𝑣 ′ 𝜕𝑣 ′2 𝜕𝑣 ′ 𝑤 ′
+ 𝑑𝑖𝑣 𝑉𝑉 = − + 𝑣 𝑑𝑖𝑣 (𝑔𝑟𝑎𝑑 𝑉) + − − −
𝜕𝑡 𝜌 𝜕𝑦 𝜕𝑥 𝜕𝑦 𝜕𝑧

𝜕𝑊 1 𝜕𝑃 𝜕𝑢′ 𝑤 ′ 𝜕𝑣 ′ 𝑤 ′ 𝜕𝑤 ′2
+ 𝑑𝑖𝑣 𝑊𝑉 = − + 𝑣 𝑑𝑖𝑣 (𝑔𝑟𝑎𝑑 𝑊) + − − −
𝜕𝑡 𝜌 𝜕𝑧 𝜕𝑥 𝜕𝑦 𝜕𝑧

(turbulence stress terms, Reynolds stresses)

Turbulent stress terms represent the transported quantities, due to the

fluctuations. For instance momentum, energy and species concentration transport
resulting from the fluctuations in velocities, temperature, etc..

Hence, additional six stress terms form in the equations,

Turbulent normal stress 𝜎𝑥𝑥 = −𝜌𝑢′2 , 𝜎𝑦𝑦 = −𝜌𝑣 ′2 , 𝜎𝑧𝑧 = −𝜌𝑤 ′2

Turbulent shear stress 𝜏𝑥𝑦 = 𝜏𝑦𝑥 = −𝜌𝑢′ 𝑣 ′ , 𝜏𝑥𝑧 = 𝜏𝑧𝑥 = −𝜌𝑢′ 𝑤 ′ ,

𝜏𝑦𝑧 = 𝜏𝑧𝑦 = −𝜌𝑣 ′ 𝑤 ′

A similar transport equation can be derived for any scalar quantity.

Time averaged transport equation for a scalar variable 𝛷 is,

𝜕𝛷 𝜕𝑢′ 𝜑 ′ 𝜕𝑣 ′ 𝜑 ′ 𝜕𝑤 ′ 𝜑 ′
+ 𝑑𝑖𝑣 𝛷𝑉 = 𝑑𝑖𝑣 𝛤𝛷 𝑔𝑟𝑎𝑑𝛷 + − − − + 𝑠𝛷
𝜕𝑡 𝜕𝑥 𝜕𝑦 𝜕𝑧

1 Introduction and Mathematical Formulation 11

Closure problem: The need for turbulence modelling.

The instantaneous continuity and Navier-Stokes equations form a set of four

equations with four unknowns u, v, w and p (u=U+u’ , v=V+v ’ , w=W+w’ , p=P+p’). In
turbulent flows, these instantaneous variables consist of fluctuating components.
These fluctuations are small scale and high frequency.

For the direct solution of instantaneous equations for u, v, w, and p, very small space
and time intervals must be used. Hence, the direct solution of turbulent flows is very
expensive and requires powerful computational resources.

The Reynolds averaged N-S (RANS) equations are time averaged equations. By time
averaging, the small-scale fluctuations are eliminated. The resulting reformed set of
equations are computationally less expensive. As a result of time average, we end up
with six additional unknowns, Reynolds stresses, in momentum equations.

1 Introduction and Mathematical Formulation 12

Turbulence Models

To solve the modified (RANS) equations, models are required to determine Reynolds
stress terms. There are many turbulence models which approximately expresses the
Reynolds stress terms in terms of the average variables.

Turbulence models:
- Mixed length model
- One-equation models
- Spalart-Allmaras model
- Two-equation models
- k-Ɛ model
- k-w model
- Reynolds stress model (RSM)
- Large Eddy Simulation (LES) model

k-ɛ Turbulence Model:

In this model, Reynolds stress terms are represented by turbulent viscosity and
average velocities. Turbulence viscosity is calculated from the turbulence kinetic
energy and turbulence kinetic energy dissipation rate.

Continuity equation: 𝑑𝑖𝑣 𝜌𝑉 = 0

N-S equation in x-direction:

𝝏𝑼
Total stress can be written as 𝝉=𝝁 − 𝝆𝒖′ 𝒗′
𝝏𝒚

𝜕𝑈 𝜕𝑃 𝜕 𝜕𝑈 𝜕 𝜕𝑈 𝜕 𝜕𝑈
𝜌 + 𝑑𝑖𝑣 𝜌𝑈𝑉 = − + 𝜇𝑒𝑓𝑓 + 𝜇𝑒𝑓𝑓 + 𝜇𝑒𝑓𝑓 + 𝜌𝑔𝑥
𝜕𝑡 𝜕𝑥 𝜕𝑥 𝜕𝑥 𝜕𝑦 𝜕𝑦 𝜕𝑧 𝜕𝑧

N-S equation in y-direction: ……..

N-S equation in x-direction: ………..

1 Introduction and Mathematical Formulation 13

 Where,
𝜇𝑒𝑓𝑓 = 𝜇 + 𝜇𝑡
𝜇𝑒𝑓𝑓 = 𝑒𝑓𝑓𝑒𝑐𝑡𝑖𝑣𝑒 𝑣𝑖𝑠𝑐𝑜𝑠𝑖𝑡𝑦
𝜇 = 𝑚𝑜𝑙𝑒𝑐𝑢𝑙𝑎𝑟 𝑙𝑎𝑚𝑖𝑛𝑎𝑟 𝑣𝑖𝑠𝑐𝑜𝑠𝑖𝑡𝑦
𝜇𝑡 = 𝑡𝑢𝑟𝑏𝑢𝑙𝑒𝑛𝑡 𝑣𝑖𝑠𝑐𝑜𝑠𝑖𝑡𝑦
𝑘2
𝜇𝑡 = 𝐶𝜇 𝜌
𝜀
k: turbulent kinetic energy
𝜀: 𝑡𝑢𝑟𝑏𝑢𝑙𝑒𝑛𝑡 𝑘𝑖𝑛𝑒𝑡𝑖𝑐 𝑒𝑛𝑒𝑟𝑔𝑦 𝑑𝑖𝑠𝑠𝑖𝑝𝑎𝑡𝑖𝑜𝑛 𝑟𝑎𝑡𝑒

Turbulence Kinetic Energy Equation

𝜕 𝜕 𝜕 𝜇𝑒𝑓𝑓 𝜕𝑘 𝜕 𝜇𝑒𝑓𝑓 𝜕𝑘
𝜌𝑈𝑘 + 𝜌𝑈𝑘 = + + 𝑠𝑘
𝜕𝑥 𝜕𝑦 𝜕𝑥 𝜎𝑘 𝜕𝑥 𝜕𝑦 𝜎𝑘 𝜕𝑦
1 2
𝑘= 𝑈 + 𝑉2 + 𝑊2
2
Turbulence Kinetic Energy Dissipation Rate Equation
𝜕 𝜕 𝜕 𝜇𝑒𝑓𝑓 𝜕𝜀 𝜕 𝜇𝑒𝑓𝑓 𝜕𝜀
𝜌𝑈𝜀 + 𝜌𝑉𝜀 = + + 𝑠𝜀
𝜕𝑥 𝜕𝑦 𝜕𝑥 𝜎𝜀 𝜕𝑥 𝜕𝑦 𝜎𝜀 𝜕𝑦

Where 𝑠𝑘 𝑎𝑛𝑑 𝑠Ɛ are kinetic energy and dissipation rate source terms.
𝑆𝑘 = 𝐺 − 𝜌𝜀
𝜀
𝑆𝜀 = 𝐶 𝐺 − 𝐶2 𝜌𝜀
𝑘 1
2 2 2
𝜕𝑈 𝜕𝑉 𝜕𝑈 𝜕𝑉
𝐺 = 𝜇𝑡 + + 2𝜇𝑡 +
𝜕𝑦 𝜕𝑥 𝜕𝑥 𝜕𝑦

Model constants:

𝐶1 = 1.44, 𝐶2 = 1.92, 𝐶𝜇 = 0.09, 𝜎𝑘 = 1.00, 𝜎𝜀 = 1.30

1 Introduction and Mathematical Formulation 14