Université de Saad Dahleb Blida 1. Module : Technique de visualisation d’ecoulement.

Institut d’Aéronautique et des Etudes Spatiales. Semestre: 03.

Département de Construction. Année Scolaire: 2023/2024.

2ème année Master Propulsion.

Mini-Projet de
Technique de visualisation d’ecoulement

Fait par :

 SAIDANI Ali.
 M2 propulsion .

Dirigé par :

Mr.Sbaa.
Sommaire :
1. General numerical method ............................................................................................................. 3
2. General classification by discretization scheme.............................................................................. 4
2- Density definition in the code: ............................................................................................................ 5
3- Relation between pressure field and velocity field:............................................................................ 5
Preconditioning treatment: ................................................................................................................. 6
Segregated style: ................................................................................................................................. 6
Treatment for unsteady characteristics in simulation: ..................................................................... 8
4- Grid system. ...................................................................................................................................... 10
Dimension type of system. ............................................................................................................... 14
Unstructured grid ? ........................................................................................................................... 20
Method for body motion:................................................................................................................. 21
Overset grid method:........................................................................................................................ 21
Grid deformation: ............................................................................................................................. 22
Regeneration of grid: ........................................................................................................................ 22
Immersed boundry method (IBM): .................................................................................................. 23
the number of grids in a computational domain: ........................................................................... 24
The non-dimensional length of the first grid size (Y+) from the model in...................................... 25
5- Numerical schemes: ......................................................................................................................... 25
Discretization of convection term .................................................................................................... 25
Order of accuracy of convection term discretization: ..................................................................... 27
Discretization of temporal term:...................................................................................................... 28
Conserved quantities:........................................................................................................................ 29
6- Boundary conditions: ........................................................................................................................ 30
7- Turbulence modeling: ....................................................................................................................... 31
Type of turbulence model : .............................................................................................................. 32
DNS (Direct Numerical Simulation): ................................................................................................. 35
Transition treatment: ....................................................................................................................... 35
Wall treatment: ................................................................................................................................ 36
6- Computations: .................................................................................................................................. 38
7- Verification and validation: ............................................................................................................... 40
a- Comparison against experimental data: ........................................................................................... 41

2
 1. General numerical method

Numerical methods are techniques to approximate mathematical processes

(examples of mathematical processes are integrals, differential equations, nonlinear
equations).

Approximations are needed because

1) we cannot solve the procedure analytically, such as the standard normal

cumulative distribution function

or

2) the analytical method is intractable, such as solving a set of a thousand

simultaneous linear equations for a thousand unknowns for finding forces in a truss
(Figure 1.1.1.11.1.1.1).

Figure 1.1.1.11.1.1.1. A truss holding a roof.

In the case of Equation (1), an exact solution is not available for Φ(x) other than
for x=0 and x→∞. For other values of x where an exact solution is not available,
one may solve the problem by using approximate techniques such as the left-hand
Reimann sum you were introduced to in the Integral Calculus course.

In the truss problem, one can solve 10001000 simultaneous linear equations
for 10001000 unknowns without using a calculator. One can use fractions, long
divisions, and long multiplications to get the exact answer. But just the thought of
such a task is laborious. The task may seem less laborious if we are allowed to use a
calculator, but it would still fall under the category of an intractable, if not an
impossible, problem. So, we need to find a numerical technique and convert it into a
computer program that solves a set of n equations and n unknowns.

3
Again, what are numerical methods? They are techniques to solve a mathematical
problem approximately. As we go through the course, you will see that numerical
methods let us find solutions close to the exact one, and we can quantify the
approximate error associated with the answer. After all, what good is an
approximation without quantifying how good the approximation is?

The name of numerical code used :

Commercial codes:

 Ansys-Fluent ( https://www.ansys.com/products/fluids/ansys-fluent )
 Ansys-CFX ( https://www.ansys.com/products/fluids/ansys-cfx )
 COMSOL Multiphysics ( https://www.comsol.com/ )
 EasyCFD (http://www.easycfd.net/)
 SimFlow ( https://sim-flow.com/ )
 CONVERGE ( https://convergecfd.com/ )

Open source:

 OpenFlower ( http://openflower.sourceforge.net/ )
 OpenFoam (https://www.openfoam.com/ )
 Code_Saturne (https://www.code-saturne.org/cms/web/ )
 OpenLB (https://www.openlb.net/ )
 SLFCFD (http://slfcfd.sourceforge.net/)

2. General classification by discretization scheme

• Finite element method:

The finite element method (FEM) is a numerical technique used to approximate solutions of partial
differential equations (PDEs) using linear combinations of basis functions. The method is
characterized by a variational formulation, a discretization strategy, one or more solution algorithms,
and post-processing procedures . The FEM has been widely used in various fields such as structural
analysis, fluid mechanics, electromagnetics, and heat transfer .

• Finite differential:

The finite difference method is a numerical technique used to approximate solutions of partial
differential equations (PDEs) using linear combinations of basis functions. The method is
characterized by a variational formulation, a discretization strategy, one or more solution algorithms,
and post-processing procedures . The finite difference method is used to solve differential equations
by approximating derivatives with finite differences .

• Finite volume method:

The finite volume method (FVM) is a numerical technique used to solve partial differential equations
(PDEs) by representing them in the form of algebraic equations . In this method, volume integrals in a
PDE that contain a divergence term are converted to surface integrals using the divergence theorem.
These terms are then evaluated as fluxes at the surfaces of each finite volume. Because the flux

4
entering a given volume is identical to that leaving the adjacent volume, these methods are
conservative . The finite volume method is used in many computational fluid dynamics packages .

• Spectral method.
The spectral method is a numerical technique used to solve partial differential equations (PDEs) by
representing them in the form of algebraic equations . In this method, the solution of the differential
equation is written as a sum of certain “basis functions” (for example, as a Fourier series which is a
sum of sinusoids) and then the coefficients in the sum are chosen to satisfy the differential equation
as well as possible 1. Spectral methods and finite element methods are closely related and built on
the same ideas; the main difference between them is that spectral methods use basis functions that
are generally nonzero over the whole domain, while finite element methods use basis functions that
are nonzero only on small subdomains (compact support) . Spectral methods have excellent error
properties, with the so-called “exponential convergence” being the fastest possible, when the
solution is smooth .

• vortex method.
The vortex method is a numerical technique used to simulate fluid flows by representing the flow
field as a collection of discrete vortices . The method is based on the Biot-Savart law, which states
that the velocity at a point in a fluid due to a vortex element is proportional to the strength of the
vortex and inversely proportional to the distance between the point and the vortex element . The
vortex method has been used to simulate a wide range of fluid flows, including those with complex
geometries and unsteady behavior .

2- Density definition in the code:

density is a physical property of a fluid that describes the amount of mass per unit volume of the
fluid . In CFD simulations, the density of a fluid is often assumed to be constant, but it can also be
modeled as a function of temperature and pressure . The density of a fluid is an important parameter
in CFD simulations because it affects the flow behavior of the fluid .

• Incompressible type:
Incompressible flow is a type of fluid flow where the density of the fluid is assumed to be constant.
This assumption is valid when the Mach number of the flow is low (less than 0.3) and the fluid is not
subjected to large pressure changes. In CFD simulations, the incompressible flow assumption
simplifies the governing equations and reduces the computational cost of the simulation.

• Compressible type.

Compressible flow is a type of fluid flow where the density of the fluid changes significantly due to
changes in pressure and temperature. This type of flow is characterized by the presence of shock
waves, which are abrupt changes in the flow properties that occur when the flow speed exceeds the
speed of sound. Compressible flow is important in many applications, such as supersonic aircraft,
rocket engines, and gas pipelines.

3- Relation between pressure field and velocity field:

In fluid dynamics, the relationship between pressure and velocity is governed by the Navier-Stokes
equations, which describe the motion of fluids. The equations are a set of partial differential
equations that relate the pressure and velocity fields of a fluid to its density, viscosity, and other

5
properties. The Navier-Stokes equations are used to model a wide range of fluid flows, from the flow
of air over an airplane wing to the flow of blood through the human circulatory system.

• Coupled style:
the coupled solver is a numerical technique used to solve fluid flow problems by simultaneously
solving the governing equations for pressure and velocity. In contrast, the segregated solver solves
the equations for pressure and velocity separately. The coupled solver is generally more accurate
than the segregated solver, but it is also more computationally expensive. The choice of solver
depends on the specific problem being solved and the available computational resources.

Preconditioning treatment:
preconditioning is a technique used to improve the convergence rate of iterative solvers for the
solution of linear systems of equations that arise in the discretization of partial differential equations
. The preconditioner is a matrix that is used to transform the original system of equations into a new
system that is easier to solve. The preconditioner is chosen so that the new system has a better-
conditioned matrix than the original system. Preconditioning can be used to accelerate the
convergence of iterative solvers for both steady-state and time-dependent problems .

• WEISS-Smith method:
The WEISS-Smith method is a preconditioning technique used to improve the convergence rate of
iterative solvers for the solution of linear systems of equations that arise in the discretization of
partial differential equations . The method is used to analyze low Mach number flow over a body . In
this method, both Euler and viscous terms are multiplied with the preconditioning matrix . The
steady-state solution will be the same regardless of whether preconditioning was used or not .

• Merkle method:
Certainly! The Merkle method is a numerical simulation technique that is used to solve fluid flow
problems. It is based on the finite element method (FEM), which is a numerical method for solving
problems in engineering and mathematical physics. The FEM is used to discretize a continuous
problem into smaller, more manageable parts, which can then be solved using numerical methods .

The Merkle method is specifically used to solve nonlinear problems in fluid mechanics. Nonlinear
problems are those where the output is not directly proportional to the input. In fluid mechanics, this
can occur when the fluid flow is turbulent or when the fluid properties change with temperature or
pressure .

Segregated style:
the segregated solver is a numerical technique used to solve the equations for pressure and velocity
separately . In contrast, the coupled solver solves the equations for pressure and velocity
simultaneously. The segregated solver is generally faster than the coupled solver, but it can be less
accurate in some cases . The choice of solver depends on the specific problem being solved and the
available computational resources .

• SIMPLE:
The SIMPLE algorithm is a widely used numerical procedure to solve the Navier-Stokes equations in
computational fluid dynamics (CFD) . The algorithm is used to produce steady flow solutions in CFD .
The method is characterized by a semi-implicit method for pressure-linked equations . The algorithm

6
is iterative and solves the equations for pressure and velocity separately . The SIMPLE algorithm is
generally faster than the coupled solver, but it can be less accurate in some cases . The choice of
solver depends on the specific problem being solved and the available computational resources .

• SIMPLER:
- SIMPLE - Revised

The revised algorithm consist of solving the pressure equation to obtain the pressure field and solving the
pressufre-correction equation only to correct the velocities. The sequence of operations can be stated as:

1. Start with a guessed velocity field.

2. Calculate the cofficients for the momentum equations and hence calculate from momentum
equations by substituting the value of the neiboughbor velocities

3. Calculate the cofficients for the pressure equation, and solve it to obtain pressure field

4. Treating this pressure field as , solve the momentum equation to obtain

5. Calculate the mass source and hence solve the p^{'} equation

6. Correct the velocity field by use, by not do not correct the pressure

7. Solve the discretization equations for other if necessary

8. Return to step 2 and repeat until convergence

• SIMPLEC:
The SIMPLEC algorithm is a modified form of the SIMPLE algorithm and is a widely used numerical
procedure to solve the Navier-Stokes equations in computational fluid dynamics (CFD). The method
is used to produce steady flow solutions in CFD. The SIMPLEC algorithm is characterized by a semi-
implicit method for pressure-linked equations. The algorithm is iterative and solves the equations for
pressure and velocity separately. The SIMPLEC algorithm is generally faster than the coupled solver,
but it can be less accurate in some cases. The choice of solver depends on the specific problem being
solved and the available computational resources.

• Fractional step method:

The numerical integration of the Navier-Stokes equations for incompressible flows demands
efficient and accurate solution algorithms for pressure-velocity splitting. Such decoupling was
traditionally performed by adopting the Fractional Time-Step Method that is based on a formal
separation between convective-diffusive momentum terms from the pressure gradient term. This
idea is strictly related to the fundamental theorem on the Helmholtz-Hodge orthogonal
decomposition of a vector field in a finite domain, from which the name projection methods
originates. This note briefly addresses the steps of the method and highlights the issue. Any
notification of error in the note is welcome.

• PISO:

7
PISO algorithm (Pressure implicit with splitting of operator) was proposed by Issa in 1986 without
iterations and with large time steps and a lesser computing effort. It is an extension of the SIMPLE
algorithm used in CFD (computational fluid dynamics) to solve the Navier-Stokes equations. PISO is a
pressure-velocity calculation procedure for the Navier-Stokes equations developed originally for non-
iterative computation of unsteady compressible flow, but it has been adapted successfully to steady-
state problems.

PISO involves one predictor step and two corrector steps and is designed to satisfy mass
conservation using predictor-corrector steps.

Algorithm Steps:

The algorithm can be summed up as follows:

1. Set the boundary conditions.

2. Solve the discretized momentum equation to compute an intermediate velocity field.

3. Compute the mass fluxes at the cells faces.

4. Solve the pressure equation.

5. Correct the mass fluxes at the cell faces.

6. Correct the velocities on the basis of the new pressure field.

7. Update the boundary conditions.

8. Repeat from 3 for the prescribed number of times.

9. Increase the time step and repeat from 1.

Treatment for unsteady characteristics in simulation:

Unsteady characteristics in a simulation can be treated by using time-dependent numerical methods.
These methods can be used to solve the governing equations of fluid dynamics, which describe the
motion of fluids. The governing equations can be solved using finite difference, finite volume, or
finite element methods. These methods can be used to simulate the unsteady behavior of fluids in a
variety of applications, such as aerodynamics, hydrodynamics, and heat transfer.

In addition, there are several techniques that can be used to improve the accuracy of simulations of
unsteady flows. These include adaptive mesh refinement, higher-order numerical schemes, and
parallel computing. Adaptive mesh refinement is a technique that allows the mesh to be refined in
regions where the solution changes rapidly, while keeping the mesh coarse in regions where the
solution changes slowly. Higher-order numerical schemes can be used to reduce numerical errors in
the solution, while parallel computing can be used to speed up the simulation.

• Steady simulation:
Steady-state simulation is a type of simulation that is used to determine the operating behavior of a
system at a specific point in time or under steady-state conditions . This type of simulation is useful
when the system is in equilibrium and the flow rates and hydraulic grades remain constant over time.

In contrast, dynamic simulation refers to a trajectory where history matters . It is beneficial when the
system is not in equilibrium and the flow rates and hydraulic grades change over time .

8
• Unsteady simulation:
Unsteady simulation is a type of simulation that is used to determine the behavior of a system over
time. This type of simulation is useful when the system is not in equilibrium and the flow rates and
hydraulic grades change over time .

In contrast, steady-state simulation is a type of simulation that is used to determine the operating
behavior of a system at a specific point in time or under steady-state conditions. This type of
simulation is useful when the system is in equilibrium and the flow rates and hydraulic grades remain
constant over time .

To simulate unsteady behavior of fluids, time-dependent numerical methods can be used. These
methods can be used to solve the governing equations of fluid dynamics, which describe the motion
of fluids. The governing equations can be solved using finite difference, finite volume, or finite
element methods. These methods can be used to simulate the unsteady behavior of fluids in a
variety of applications, such as aerodynamics, hydrodynamics, and heat transfer .

In addition, there are several techniques that can be used to improve the accuracy of simulations of
unsteady flows. These include adaptive mesh refinement, higher-order numerical schemes, and
parallel computing. Adaptive mesh refinement is a technique that allows the mesh to be refined in
regions where the solution changes rapidly, while keeping the mesh coarse in regions where the
solution changes slowly. Higher-order numerical schemes can be used to reduce numerical errors in
the solution, while parallel computing can be used to speed up the simulation.

• Constant time step:

In a simulation, a time step is the amount of time that passes between each calculation of the
simulation. In a constant time step approach, the time step remains the same throughout the
simulation. This approach is useful when the simulation is stable and the solution does not change
rapidly over time .

To set up a constant time step in a simulation, you can use the simulation control panel and set the
adjustable time step to false . This will ensure that the time step remains constant throughout the
simulation.

• Dynamic time step (Target Courant number):

a time step is the amount of time that passes between each calculation of the simulation. In a
dynamic time step approach, the time step changes over time to ensure that the simulation is
accurate and stable .

The Courant number is a non-dimensional number that can be used to evaluate the time step
requirements of a transient simulation for a given mesh size and flow velocity. The Courant number
is linked to the Courant–Friedrichs–Lewy (CFL) stability condition of numerical schemes. The Courant
number formula is defined as:

C=ΔhUΔt
where U indicates the flow velocity, Δt is a representative time step of the simulation, and Δh is the
characteristic size of the mesh cell (in 1 dimension).

9
The target Courant number is the desired value of the Courant number for a given simulation . The
target Courant number is typically set to a value less than or equal to 1 to ensure that the simulation
is stable and accurate .

To set up a dynamic time step with a target Courant number, you can use the simulation control
panel and set the adjustable time step to true . This will allow the time step to change over time to
ensure that the simulation is accurate and stable . You can also set the target Courant number to the
desired value.

4- Grid system.

 Mesh generation in CFD simulations plays the same role as meshing in finite
element simulations, where discretization will determine the accuracy and
computation time in the simulation.
 The grid generation method that is used in a problem will try to match the mesh to
the geometry of the system being simulated.
 More advanced meshing techniques do not use Cartesian or structured grids,
instead, they apply polynomial curves and interpolation schemes to generate the
mesh and results.

Numerical methods for solving partial differential equations are advanced topics not
normally taught in undergraduate engineering classes. However, as a working
engineer, you can’t work out every problem by hand. Therefore, it’s important to have
some knowledge of numerical methods as well as how they are executed in
computational simulation. When generating meshes for these simulations, it’s your

10
job to understand how discretization in a numerical model influences accuracy and
what you’ll be able to observe in your results.

Grid generation in CFD refers to a set of techniques for defining a numerical mesh
throughout the system to be simulated. The grid that is selected for CFD simulations
will define the accuracy and resolution of the simulation results, both of which will
affect the computation time and level of detail in the results. Using the right grid
generation tools, automation and templatization are put in place to ensure that the
required expertise and time commitment can be minimal while still ensuring the
utmost accuracy in your mesh.

Grid Generation in CFD Simulations

Grid generation in CFD simulations involves choosing a mathematical technique to
represent the arrangement and spacing between each node in the numerical grid for
your system. The technique that is chosen can then be used to define grid points along
the surface (for 2D or boundary element method problems) or inside the volume (for
3D problems) of the system.

Why is meshing needed in CFD problems? In addition to converting complex

differential equations into simpler arithmetic problems, discretization allows a
simulation to account for changes in continuous physical properties across the solution
domain. An example would be compressible flow, where the density of the fluid varies
in space and with flow rate.

The discretization resolution can also be tuned to the problem at hand to ensure the
system can be solved with reasonable computational effort. When the generated mesh
matches the system geometry, a coarser mesh can be used so that computational cost is
reduced. After the simulation converges to a solution, the results can be interpolated
for the entire domain.

The five common geometries used in grid generation for CFD problems are detailed
below.

Cartesian Grid Generation

This is the simplest type of grid generation technique. In this method, a rectangular
grid is used to represent the boundary and interior of a system for use in a CFD
simulation, with grid points constrained to planes aligned with the Cartesian axis
system. Cartesian grids are simple systems, with each cell being shaped like a cube or
brick, that can provide very high accuracy as long as the geometry only consists of
orthogonal surfaces. Systems with curved surfaces or slanted boundaries may require
higher meshing density to ensure accuracy, which increases the computation time in
the simulation. Another technique for handling curved surfaces is for the grid cells to
be cut by the boundaries into non-cubical shapes. Because of these complications, an
alternative grid approach is normally used in most systems.

11
Structured Grid Generation
A structured grid creates an arrangement of quad (2d) or brick (3d) grid cells that are
arranged in a simple matrix-like structure. Unlike a Cartesian grid in which the cells
are always regularly shaped, a structured grid is warped to follow the boundaries of
the system. The node arrangement in the grid follows the same shape as the boundary
surface, effectively scaling its interpolated nodes into the interior of the system. This
type of grid generation provides much higher accuracy than Cartesian grid generation
in CFD problems, as it will closely follow the surface of the curve along the boundary.

Example showing a structured grid conforming to the surface of a jet turbine housing

Unstructured Grid Generation

An unstructured grid is a tessellation across the surface and interior volume of the
system, seen as triangles (in 2D problems) or tetrahedra (in 3D problems). These grids
can provide comparable accuracy as a structured grid, but they require similarly high
node density in regions with high flow gradients. Therefore, the use of an unstructured
grid does not always guarantee more efficient computation. However, if the grid
resolution can still be tuned across the solution domain, computational complexity can
be reduced for certain problems.

12
Hybrid Grid Generation
This grid generation method for CFD simulations applies a multi-structured approach
to complex shapes that could conform to different coordinate systems. For example,
consider square-shaped housing with a round hole; hybrid grid generation would use a
structured grid around the curved portion and a Cartesian grid along the orthogonal
boundaries. Somewhere between the two regions, the two grid styles would converge
on each other and make a smooth transition. The challenge in this grid generation
method is to define the transition between different regions.

This hybrid grid for an airfoil contains unstructured regions and structured regions
around the edge of the structure.

Adaptive Meshing
This grid generation method for CFD simulations is applied as part of any of the other
grid generation methods shown above. As the grid is applied, the spatial node density
is increased near specific regions in the system, or near curves with higher curvature,
in order to capture data with higher resolution in specific regions in the system. For
electromagnetics problems, this allows regions in the system where higher frequency
waves are prominent to be captured with higher accuracy, whereas lower frequency
regions can have a coarser grid. The same idea applies to CFD simulations.

13
 Adaptive meshing applies higher density grids near curves with high gradients,
including in the transition between two surfaces

High-Order Meshing
High-order meshing is one technique that can be used to apply polynomial meshes to
system boundaries with complex shapes. In some systems with curved surfaces, the
surface cannot be easily described by analytic equations, so these systems require a
parametric approach, interpolation, spline representation, or polynomial approximation
to determine the appropriate grid points. Applying a polynomial is a simple way to
approximate the surface as a curve, and a standard interpolation approach can then be
easily applied between mesh points. Resolving the geometry with curved elements in
adaptive meshing, rather than using piecewise linear or piecewise quadratic sections,
can provide very high accuracy in regions with large gradients without requiring
higher grid density.

Matching a Solution Method

Technically, any numerical computational scheme can be used in the above grid
geometries to calculate the solution to the Navier-Stokes equations in a CFD problem.
The finite volume method (FVM) is preferred, particularly for problems that involve
convection, as it can handle the nonlinear convective term in the Navier-Stokes
equations. FDTD is also useful in CFD simulations and is sometimes discussed in the
literature as an FDTD-CFD method. Commercial solvers that can operate directly from
your physical layout data will typically use FDM, FEM, or FVM to solve CFD
problems.

Our aim is to ensure grid generation is as seamless as possible without sacrificing

fidelity. It’s vital that whether you’re implementing Cadence into a pre-existing
workflow or working to rebuild your CFD workflow entirely, you get the fidelity you
need at the speed you want.

Dimension type of system.

• Tetrahedral grid:
a tetrahedral grid is a type of unstructured meshing that is used to represent complex geometries
using an unstructured grid 1. This method is useful when the geometry of the system being simulated
is complex and cannot be represented using a structured grid.

14
A tetrahedral grid is composed of tetrahedrons, which are four-faced polyhedrons that are similar to
pyramids. Tetrahedrons are used to represent the volume of the fluid being simulated, and the
vertices of the tetrahedrons are used to represent the nodes of the grid.

Tetrahedral grids are commonly used in CFD simulations of complex geometries, such as those found
in aerodynamics, hydrodynamics, and heat transfer. They can be generated using software tools such
as ANSYS ICEM CFD, Pointwise, and Gmsh.

A shape is represented with a signed distance function (SDF) encoded with a deformable
tetrahedral grid [1]. The grid fully tetrahedralizes a unit cube, where each cell is a
tetrahedron with 4 vertices and faces. Moreover, the grid vertices can deform to
represent the geometry of the shape more efficiently. The use of SDF, instead of
occupancy values [1], provides more flexibility in representing the underlying surface.

Deformable tetrahedral mesh. The deformable tetrahedral mesh is denoted with

(𝑉𝑇 , 𝑇), where 𝑉𝑇 is the set of vertices and 𝑇 = {𝑇𝑘 }𝐾 𝑘=1 is the set of tetrahedra. 𝑇𝑘 is
representated with four vertices {𝑣𝑎𝑘 , 𝑣𝑏𝑘 , 𝑣𝑐𝑘 , 𝑣𝑑𝑘 } .We represent the SDF by interpolating
the SDF values defined on the vertices with the barycentric interpolation.

Volume subdivision. Shapes are represented in a coarse to fine manner for efficiency.
Surface tetrahedra are defined as those with vertices of different SDF signs. We subdivide
surface tetrahedra, as well as their immediate neighbors, by adding a mid point to each
edge.

Marching tetrahedra for converting between an implicit and explicit

representation. We use Marching Tetrahedra [2] to convert the SDF into an explicit
triangular mesh. Based on the signs of the SDF values, MT determines the surface
typology inside the tetrahedron. Once the surface typology is identified, the vertex
location of the iso-surface is computed at the zero corssings of the linear interpolation
along the tetrahedron’s edges.

15
Surface subdivision. Having a surface mesh as output allows us to increase the
representation power and the visual quality with a differentiable subdivision module. The
authors follow the Loop Subdivision method [3,4] and used a fixed set of learnable
parameters, including positions 𝑣𝑖′ and smoothness 𝛼𝑖 .

• Hexahedral grid:
A hexahedron, a topological cube, has 8 vertices, 12 edges, bounded by 6 quadrilateral faces. It is
also called a hex or a brick. For the same cell amount, the accuracy of solutions in hexahedral meshes
is the highest.

The pyramid and triangular prism zones can be considered computationally as degenerate
hexahedrons, where some edges have been reduced to zero. Other degenerate forms of a
hexahedron may also be represented.

• H.type:
One of the classical gridding approaches is the H-type pattern. It has finely clustered cells at the
leading and trailing edges as shown in figure 2. This pattern is simple to construct and holds good
for a biconvex airfoil with sharp leading and trailing edges. However, when applied to an airfoil
with a curved leading edge, the H-pattern creates a singular point (figure 3a). This can partly be
alleviated by splitting the singularity into two weaker singularities as shown in figure 3b. But still, it
falls short of capturing the leading edge curvature accurately and also propagates the boundary
layer in the transverse direction (figure 4).

16
Alongside the wasted use of point clusters and high aspect ratio cells, the high clustered cells
both parallel and perpendicular in regions where the flow accelerates can result in significant time
step reductions due to CFL conditions, leading to slowing of solver convergence.

Figure 5: C-type pattern for a blunt leading edge airfoil with boundary layer clustering.

 C-Type
An improved variant of H-type topology is the C-type pattern, which captures the leading edge
curvature without any singularities. Though the C-type pattern avoids the propagation of
boundary layer fineness upstream, it fails to do so downstream of the airfoil trailing edge. In a
way, this downstream fineness proves beneficial as it helps to capture the shear layer for solver
runs at low angles of attack.

In practice, however, most applications of the C-type pattern, have not taken advantage of the
wonderful alignment of the grid lines along with the shear layer. To take full advantage of this
pattern, the CFD analysis would have to continually shift the grid curves to dynamically align with
the shear layer and this would result in greater conformity to the flow physics. While not doing this
alignment results in algorithmic simplicity, it does result in the wastage of grid points downstream

17
and falls short of being an efficient grid type.

Figure 6: O-type pattern for a blunt leading edge airfoil with boundary layer clustering.

 O-Type
The next classical blocking strategy is the O-type, which almost overcomes the disadvantages of
the H-type and C-type grids. In O-type the entire grid sheet is wrapped up around the airfoil
without any propagation of the boundary-layer fineness into the field. This helps in getting an
optimal cell count, eliminating the redundant propagation of cells as seen in the H-type and C-
type patterns.

However, this pattern has its inherent limitations. It creates highly skewed cells for zero thickness
trailing edge airfoils. The grid cell on the upper airfoil surface directly connects to the cell on the
lower airfoil surface across the horizontal grid line emanating from the trailing edge. For a single
cell-to-cell step, it nearly creates a 360-degree turn in one step. Thus, each cell will have an angle
of slightly less than 180 degrees which is not exactly of good quality.

The level of skewness created in such cases is acceptable for Euler grids but is extremely high in
viscous grids. Singular points like the trailing edge are critical for accurate CFD computations.
High skewness will affect the solver’s robustness and also the quality of the solutions. For this
reason, many prefer to go with the C-type grid and live with the excessive cells in the wake
region.

18
 Figure
7a: Cartesian grid. Image source – hanleyinnovations, 7b – Unstructured grid

• Hybrid:
Traditional approaches to mesh generation have been via block-structured
(usually composed of hexahedra) or unstructured (tetrahedra) methods.
There are pros and cons to both approaches and thus the strategy chosen
has been dependent on the particular application. CENTAUR combines the
best of both methods via a hybrid
(prismatic/hexahedral/pyramidal/tetrahedral) meshing strategy. The
prismatic and hexahedral elements are used in regions of high solution
gradients, and tetrahedra are used elsewhere. Pyramids are used in some
locations to allow for the transition between the prisms/hexahedra and the
tetrahedra.

Hybrid Grid Advantages:

19
 The prismatic or hexahedral layers close to wall surfaces exhibit good orthogonality and clustering capabilities, characteristics of structured mesh generation approaches.

The structured nature allows for the implementation of multigrid convergence acceleration schemes, implicit methods, and also results in memory savings.

The use of tetrahedra to fill the rest of the domain allows single-block generation for extremely complex geometries since the tetrahedron is the simplest 3-D element.

The tetrahedral elements are well suited for grid refinement.

The flexibility of the hybrid mesh generation approach automates an otherwise highly interactive procedure.

The hybrid prismatic/hexahedral/tetrahedral strategy requires no solution interpolation or grid interfacing techniques as in the traditional block-structured approach.

The prismatic and hexahedral portions of the grid require less memory than a tetrahedral grid of the same resolution.

Generation of the prisms and hexahedra is faster than the generation of tetrahedra.

Unstructured grid ?
• Polyhedral grid:

In Computational Fluid Dynamics (CFD), a polyhedral grid is a type of unstructured meshing that is
used to represent complex geometries using an unstructured grid. This method is useful when the
geometry of the system being simulated is complex and cannot be represented using a structured
grid.

A polyhedral grid is composed of polyhedrons, which are three-dimensional shapes with flat faces
and straight edges. Polyhedrons are used to represent the volume of the fluid being simulated, and
the vertices of the polyhedrons are used to represent the nodes of the grid.

20
Polyhedral grids are commonly used in CFD simulations of complex geometries, such as those found
in aerodynamics, hydrodynamics, and heat transfer. They can be generated using software tools such
as Layoutit Grid, GE GridOS Orchestration Software, and ETAP Wind Turbine Generator.

• Cartesian grid:
a Cartesian grid is a type of structured meshing that is used to represent the boundary and interior
of a system using a rectangular grid. This method is useful when the geometry of the system being
simulated is simple and can be represented using a single structured grid.

A Cartesian grid is composed of rectangular cells that are arranged uniformly in a matrix. The vertices
of the cells are used to represent the nodes of the grid.

Cartesian grids are commonly used in CFD simulations of simple geometries, such as those found in
heat transfer, combustion, and fluid flow . They can be generated using software tools such as ANSYS
ICEM CFD, Pointwise, and Gmsh .

Method for body motion:

There are several methods for simulating body motion in computational fluid dynamics (CFD). One
such method is the FloatStepper algorithm. It is a non-iterative algorithm that couples the motion of
a rigid body and an incompressible fluid in CFD simulations. The algorithm is designed to remove the
so-called added mass instability problem, which may arise when a light floating body interacts with a
heavy fluid. The algorithm is implemented and released as an open source extension module to the
widely used CFD toolbox, OpenFOAM, as an alternative to the existing body motion solvers.

Another method is to use SimScale’s cloud-based CFD2. It allows performing a CFD simulation and
evaluation for problems involving solid body motion with 6 DoF (Degrees of Freedom) movement.
Such simulations find applications in understanding the effect of moving fluid on solid bodies and
vice-versa. Typical use includes valves and pipes design or shipbuilding.

Overset grid method:

The overset grid method is a technique used in computational fluid dynamics (CFD) to simulate
complex fluid flow problems where not every geometry can be well represented using a single,
contiguous (structured or unstructured) grid . In this method, a grid system is constructed by blocks
of overlapping structured grids, which are used to represent different geometrical features. The
Chimera or overset grid approach is a popular implementation of this method. In a full Chimera grid
system, a complex geometry is decomposed into a system of geometrically simple overlapping grids.
Boundary information is exchanged between these grids via interpolation of the flow variables, and
many gridpoints may not be used in the solution. The overset grid method has found applications in
various fields such as shipbuilding, valves and pipes design, and global geomagnetic induction.

21
Grid deformation:

Grid deformation is a technique used in computational fluid dynamics (CFD) to simulate complex
fluid flow problems where not every geometry can be well represented using a single, contiguous
(structured or unstructured) grid . There are several methods for grid deformation, such as the
square grid deformation approach , elastic deformation, and deformation method for moving grid
generation. The square grid deformation approach is generally employed to forming processes
solved with the aid of slip-line field in which a ductile body is supposed to be made with arrays of
square grids of finite sizes before the deformation process. After the application of the load, as the
deformation occurs, square grids are deformed. The square grid deformation has three utilities—(1)
visualization of actual deformation process, (2) prediction of trajectory at any point and (3)
computation of strain. Elastic deformation generates a coarse displacement grid with a random
displacement for each grid point. This grid is then interpolated to compute a displacement for each
pixel in the input image. The input image is then deformed using the displacement vectors and a
spline interpolation. The deformation method for moving grid generation can generate a moving grid
on the physical domain such that its cell size distribution is proportional to 0ÐBß>Ñ at all time >. This
method was used in a finite element scheme for a convection-diffusion problem and in a finite
difference scheme for a hyperbolic problem.

Regeneration of grid:
Grid regeneration refers to the process of updating or modifying the computational grid used for
simulating fluid flow over a particular domain. The grid, or mesh, in CFD is a discretization of the
physical domain into a collection of elements or cells, and it plays a crucial role in accurately
representing the geometry and capturing the flow physics.

Grid regeneration becomes necessary in various scenarios, including:

22
1. Geometry Changes: If there are changes to the physical geometry of the system being simulated,
such as the addition or removal of components, or changes in shape, the grid may need to be
regenerated to reflect these alterations.

2. Refinement or Coarsening: Based on the requirements of the simulation, the grid may need to be
refined (increased in resolution) or coarsened (decreased in resolution). This is often done in regions
of interest where detailed flow features are expected, or in areas where lower resolution is
acceptable to reduce computational costs.

3. Optimizing Grid Quality: The quality of the grid can affect the accuracy and stability of the
simulation. Grid regeneration may involve optimization techniques to ensure that the grid elements
are well-shaped and do not contain highly skewed or distorted elements that can lead to numerical
issues.

4. Adaptive Mesh Refinement: Some CFD simulations use adaptive mesh refinement strategies where
the grid resolution is adjusted dynamically during the simulation based on the evolving flow field.
This adaptive refinement may involve both refining and coarsening the grid in different regions.

5. Boundary Conditions Changes: If there are changes in the boundary conditions, such as the
introduction of new inlets, outlets, or wall conditions, the grid near these boundaries may need to be
regenerated to accurately represent the updated conditions.

6. Transient Simulations: In simulations involving transient flows, where the flow conditions change
with time, the grid may need to be regenerated at certain time steps to account for evolving flow
features.

Grid regeneration is a critical step in ensuring the accuracy and reliability of CFD simulations.
However, it also introduces challenges, as the process can be computationally expensive, and care
must be taken to avoid introducing errors during the regeneration process. Efficient grid generation
tools and techniques are essential for successful CFD simulations.

Immersed boundry method (IBM):

The Immersed Boundary Method (IBM) is a numerical technique used in Computational Fluid
Dynamics (CFD) to simulate the interaction between a fluid and immersed objects or boundaries
within that fluid. It is particularly useful for problems involving complex geometries, irregular shapes,
and moving boundaries.

Here's a basic overview of how the Immersed Boundary Method works:

1. Representation of Immersed Boundaries:

- Instead of explicitly conforming the computational grid to the complex geometry of immersed
objects (such as solid structures), the IBM uses a separate representation for these boundaries.

- The immersed boundaries are often described by a set of marker points or a distribution of forces
located within the fluid domain.

2. Interpolation:

- The forces or marker points associated with the immersed boundaries are used to impose
boundary conditions on the fluid flow.

23
 - Interpolation techniques are employed to transfer information between the markers and the fluid
grid. This allows the fluid solver to account for the presence and effects of immersed boundaries.

3. Force Imposition:

- Forces or constraints are applied to the fluid at the grid points near the immersed boundaries
based on the interpolated information.

- These forces account for the effect of the immersed boundaries on the fluid flow.

4. Coupling with Fluid Solver:

- The fluid solver solves the Navier-Stokes equations to compute the flow field in the entire domain,
including the regions near immersed boundaries.

- The forces imposed by the immersed boundaries influence the fluid flow, and the fluid solver
accounts for these effects in a coupled manner.

The Immersed Boundary Method offers several advantages:

- Flexibility: It allows for the simulation of complex geometries without the need for a highly
conforming grid.

- Simplicity: The representation of immersed boundaries can be done independently of the grid
structure, simplifying the mesh generation process.

- Applicability to Moving Boundaries: IBM is well-suited for problems involving moving objects or
dynamic boundary conditions.

One of the early developments of the Immersed Boundary Method was by Charles S. Peskin, who
introduced the concept in the context of simulating the motion of the heart. Since then, IBM has
found applications in various fields, including biofluid dynamics, aerodynamics, and other fluid-
structure interaction problems.

the number of grids in a computational domain:

The number of grids in a computational domain, also known as the mesh size or grid resolution, can
vary widely depending on the complexity of the simulation, the desired accuracy, and the available
computational resources. In Computational Fluid Dynamics (CFD) simulations, the grid is typically
composed of a mesh of cells, elements, or grid points that discretize the physical domain.

For a simple simulation, you might have a relatively coarse grid with, for example, tens of thousands
to a few hundred thousand grid points. In more complex simulations, especially those involving
intricate geometries or capturing fine details in the flow field, the grid size may extend to millions or
even billions of grid points.

The appropriate grid resolution is often determined through a process of grid sensitivity analysis,
where the simulation results are examined at different grid resolutions to ensure that the solution
converges to a stable and accurate result. This process helps balance the need for accuracy with the
computational resources available.

24
The non-dimensional length of the first grid size (Y+) from the model in
• wall normal direction:
The non-dimensional length of the first grid size in the wall-normal direction is often expressed using
the dimensionless wall distance, commonly denoted as .The non-dimensional length of the first
grid size is an important parameter in the context of boundary layer simulations, especially in
computational fluid dynamics (CFD). It is related to the near-wall grid resolution and is used to assess
the accuracy of capturing the flow near solid surfaces.

The value is calculated as the distance from the wall to the first grid point (y1) normalized by the
molecular viscosity of the fluid and the wall shear stress :

where:

The goal in CFD simulations is often to achieve a specific target value to properly resolve the near-
wall flow physics. Different values of are suitable for different turbulence models and simulation
goals. In general, lower values indicate a finer grid resolution near the wall, providing better
accuracy in capturing the viscous sublayer and the buffer layer.

To determine ,you need information about the specific simulation, the turbulence model used,
and the desired target. If you have more details, I can provide a more specific calculation or
explanation.

5- Numerical schemes:
Numerical schemes, also known as numerical methods or numerical algorithms, are computational
techniques used to approximate solutions to mathematical problems that may not have analytical
solutions or are difficult to solve using traditional methods. These methods are particularly important
in fields such as physics, engineering, finance, and computer science, where complex mathematical
models often need to be solved numerically.

Discretization of convection term

• Centered:

25
 In a centered discretization scheme, the value of the convection term is evaluated at the center of
the control volume or grid cell. For a one-dimensional convection term, the discretization can be
expressed as:

This approach is symmetric, providing a balanced consideration of values on both sides of the point
i. While it can maintain accuracy, it may suffer from numerical oscillations, especially in situations
with rapidly changing gradients or discontinuities.

• Upwind:
In the upwind discretization scheme, the convection term is approximated by considering the
direction of flow. It uses information from the upstream (upwind) side of the grid cell. For the one-
dimensional case, the upwind scheme can be expressed as:

The upwind scheme is more stable in the presence of strong gradients or discontinuities but may
introduce numerical diffusion, causing a loss of accuracy.

• Hybrid (Centered.Upwind):

The hybrid scheme is a combination of the centered and upwind schemes, aiming to balance
accuracy and stability. The choice of using the centered or upwind scheme depends on the local flow
characteristics. For example, it might use the upwind scheme in regions of high gradients and the
centered scheme in smoother regions.

Here, is a parameter that can be adjusted based on the local flow conditions.

• Explicit artificial viscosity:

Artificial viscosity is a technique used to dampen numerical oscillations and stabilize the solution,
particularly in the presence of shocks or discontinuities. For the convection term, an explicit artificial
viscosity term is added to the discretized equation. The viscosity term helps smooth out abrupt
changes in the solution.

Artificial viscosity term

26
 The artificial viscosity term is designed to be active only in regions with high gradients, where it is
needed to stabilize the solution.

Order of accuracy of convection term discretization:

The order of accuracy in numerical schemes refers to the rate at which the numerical solution
approaches the exact solution as the discretization parameters (such as grid size or time step) are
refined. A higher order of accuracy generally implies that the numerical method converges to the
exact solution more rapidly as the resolution is increased. Let's explore first, second, and third-order
accuracy in the context of discretizing the convection term.

In summary, the order of accuracy is a measure of how rapidly a numerical method converges to the
exact solution as the discretization is refined. Higher-order accuracy is generally desirable as it allows
for more accurate simulations with fewer computational resources. However, achieving higher order
accuracy often requires more complex numerical methods and may be challenging in certain
scenarios.

a- First:
A discretization method is said to be first-order accurate if the error in the numerical solution is
proportional to the size of the discretization parameter, typically denoted as ( Delta x ) for spatial
discretization or ( Delta t) for temporal discretization. Mathematically, the error ( E ) can be
expressed as:

In the context of the convection term, a first-order accurate scheme might involve simple finite
difference approximations, such as the first-order upwind scheme:

b- Second:
A discretization method is said to be second-order accurate if the error is proportional to the
square of the discretization parameter. Mathematically:

In the context of the convection term, a second-order accurate scheme might involve more
sophisticated finite difference approximations, such as the second-order central difference scheme:

c- Third:

27
 A discretization method is said to be third-order accurate if the error is proportional to the cube of
the discretization parameter. Mathematically:

Achieving third-order accuracy often involves even more advanced finite difference or finite
element methods. For example, a third-order accurate scheme for the convection term might use a
combination of higher-order central differences and possibly artificial viscosity to achieve stability.

Discretization of temporal term:

The discretization of the temporal term in numerical methods involves approximating the time
derivatives of a system to enable the simulation of dynamic processes over time. Here's an
explanation of several temporal discretization methods:

a- Euler explicit:
Euler's explicit method is a simple and straightforward approach where the time derivatives are
approximated using forward differences. For a first-order ordinary differential equation (ODE), the
Euler explicit method can be expressed as:

Here, represents the solution at time is the time step, and is the derivative of u with
respect to time at the current time step.

b- Euler implicit:
Euler's implicit method, on the other hand, involves using backward differences to approximate the
time derivatives. The update equation becomes:

This results in an implicit equation that often requires solving a nonlinear system of equations at
each time step.

c- Runge-kutta:
Runge-Kutta methods are a family of iterative methods that use multiple evaluations of the
derivative function to improve accuracy. The fourth-order Runge-Kutta method, commonly denoted
as RK4, is widely used and is expressed as:

Here, and are approximations of the derivative at different points within the time step.

28
d- Adam-bashforth:
Adams-Bashforth methods are explicit schemes that use multiple previous time steps to approximate
the derivative. The second-order Adams-Bashforth method is given by:

These methods are known for their simplicity and computational efficiency.

e- Multi-stage explicit:
Multi-stage explicit methods, such as the explicit Runge-Kutta methods, involve multiple stages of
calculations within a single time step. They offer higher accuracy compared to simple explicit
methods.

f- Three point backward implicit:

The three-point backward implicit method is an implicit scheme that uses information from three
previous time steps. For a first-order ODE, it can be expressed as:

This implicit scheme is designed to provide better stability properties compared to some explicit
methods.

Conserved quantities:
Conserved quantities in physics are properties of a system that remain constant over time, regardless
of the internal processes or interactions within the system. These quantities are crucial in
understanding the behavior of physical systems and form the basis for conservation laws. Let's
explore the conservation of mass, momentum, kinetic energy, and total energy:

a- Mass:
Conservation of Mass: In a closed system (one that does not exchange matter with its surroundings),
the total mass of the system remains constant over time. The principle of conservation of mass is a
fundamental concept in physics and chemistry. Mathematically, for a closed system:

This means that the rate of change of mass with respect to time is zero, indicating that mass is
conserved.

b- Momentum:

29
Conservation of Momentum: In the absence of external forces, the total linear momentum of an
isolated system is conserved. Linear momentum (p) is defined as the product of mass and velocity for
each object in the system. Mathematically:

The conservation of momentum is particularly important in understanding the motion of objects and
collisions.

c- Kinetic energy:
Conservation of Kinetic Energy: In isolated systems without external forces doing work, the total
kinetic energy of the system is conserved. Kinetic energy KE is the energy associated with the motion
of an object and is given by the formula ,where m is mass and v is velocity.

This conservation principle is often applied in problems involving the interaction of objects in motion.

d- Total energy:
Conservation of Total Energy: In closed systems where only conservative forces are at play, the total
mechanical energy (sum of kinetic and potential energy) remains constant. Mathematically:

This conservation law is based on the work-energy principle and is a key concept in classical
mechanics.

6- Boundary conditions:
In Computational Fluid Dynamics (CFD), boundary conditions play a critical role in simulating fluid
flow over a specified domain. The choice of boundary conditions is crucial for obtaining accurate and
physically meaningful solutions. CFD simulations typically involve solving sets of partial differential
equations (PDEs) that describe the behavior of fluid flow. Here are some common types of boundary
conditions used in CFD:

1. Inlet Boundary Conditions:

- Velocity Inlet: Specifies the inflow velocity of the fluid at the boundary.

- Mass Flow Inlet: Prescribes the mass flow rate entering the domain.

- Pressure Inlet: Sets the pressure at the boundary.

2. Outlet Boundary Conditions:

- Pressure Outlet: Prescribes the pressure at the outlet, allowing the fluid to freely exit.

- Velocity Outlet: Specifies the outflow velocity.

30
 - Mass Flow Outlet: Sets the mass flow rate leaving the domain.

3. Wall Boundary Conditions:

- No-Slip Wall: Assumes that the fluid velocity at the wall is zero, simulating the effect of fluid
sticking to the wall.

- Slip Wall: Allows fluid to slip along the wall without friction.

- Temperature Wall: Prescribes the temperature at the wall.

4. Symmetry Boundary Conditions:

- Symmetry: Assumes symmetry across the boundary, simplifying the simulation domain.

- Periodic: For problems with periodic structures, simulates a periodic continuation of the domain.

5. Far-Field Boundary Conditions:

- Far-Field: Prescribes the flow conditions at a boundary located far from the region of interest,
allowing the simulation to represent an unbounded domain.

6. Special Boundary Conditions:

- Inlet/Outlet Mass Flow: Prescribes the mass flow at both the inlet and outlet, ensuring a specific
mass flow profile.

- Fan/Blower: Simulates the effect of a rotating fan or blower at the boundary.

- Porous Jump: Represents a porous medium at the boundary.

7. Custom or User-Defined Boundary Conditions:

- Users may define custom boundary conditions based on the specific requirements of their
simulation.

7- Turbulence modeling:
In turbulence modeling, different regimes describe the flow characteristics based on factors such as
viscosity and turbulence intensity. The terms viscous, inviscid, laminar, and turbulent are used to
classify flows based on their behavior. Let's explore each regime:

a- Viscous regime:
In the viscous regime, the effect of viscosity dominates the flow behavior. Viscosity is the internal
friction of a fluid that tends to resist the relative motion between fluid layers. In flows characterized
as viscous, the fluid layers near solid surfaces experience significant shear stresses, leading to well-
defined layers with smooth, ordered flow patterns. This regime is typical for flows with low
velocities, small length scales, or high viscosity fluids (e.g., honey or oil).

b- Inviscid:
In the inviscid regime, the effects of viscosity are negligible, and the flow is considered to be
frictionless. Inviscid flows are often characterized by high velocities and large length scales where the
viscous forces have minimal impact. The flow patterns in inviscid flows are typically less constrained,

31
and the fluid moves more freely. Ideal gases and flows with low viscosity can often be approximated
as inviscid.

c- Laminar :
Laminar flow is characterized by smooth, ordered layers of fluid that move parallel to each other
without significant mixing. In laminar flows, the motion of the fluid is well-organized, and the flow
patterns remain stable over time. Laminar flows are common in low Reynolds number situations,
where viscous forces dominate. The transition from laminar to turbulent flow often occurs as the
Reynolds number increases.

d- Turbulent :
Turbulent flow is characterized by chaotic and irregular fluid motion, involving vortices, eddies, and
intense mixing. Turbulent flows have high Reynolds numbers, where inertial forces dominate over
viscous forces. Turbulence is a complex phenomenon that leads to enhanced mixing and transport of
momentum, heat, and other properties within the fluid. Turbulent flows are common in many
practical engineering applications, such as rivers, pipes, and atmospheric flows.

Reynolds Number:

The Reynolds numberRe is a dimensionless parameter that helps determine the dominant flow
regime. It is defined as the ratio of inertial forces to viscous forces and is given by:

The Reynolds number helps in predicting the transition from laminar to turbulent flow. At low
Reynolds numbers, flows tend to be laminar, while at higher Reynolds numbers, turbulence becomes
more prevalent.

Type of turbulence model :

Turbulence models are mathematical formulations used in Computational Fluid Dynamics (CFD) to
simulate the effects of turbulence in fluid flows. These models aim to predict the turbulent flow
characteristics, such as velocity fluctuations and turbulent energy, without resolving all the turbulent
scales explicitly. Here's an explanation of various turbulence models:

a- RANS: (Reynolds-Averaged Navier-Stokes):

The RANS model solves time-averaged equations to predict mean flow properties and turbulence
quantities. It involves decomposing variables into mean and fluctuating components, leading to
Reynolds-averaged equations. RANS models are computationally efficient but may not capture
unsteady or large-scale turbulent structures.

32
b- URANS :(Unsteady Reynolds-Averaged Navier-Stokes):
URANS extends RANS by considering time-dependent mean flow properties. It captures some level of
unsteadiness but still relies on time averaging. URANS is useful for simulating moderately unsteady
flows, such as oscillating flows or periodic phenomena.

c- Zero equation model:

Zero equation models, such as the Eddy Viscosity Model, assume a constant eddy viscosity
throughout the flow field. They are computationally efficient but provide limited accuracy, especially
for complex flows with variations in turbulence intensity.

d- 1 equation model:
1 equation models, like the Spalart-Allmaras model, solve a single transport equation for a turbulent
variable (e.g., eddy viscosity). They are simpler than 2 equation models but offer better accuracy
compared to zero equation models.

e- Spalart -Allmaras model:

The Spalart-Allmaras model is a 1 equation model that solves for the eddy viscosity using a transport
equation. It is widely used for a range of applications and is known for its robustness and efficiency.

f- 2 equation model:
2 equation models, like the k-epsilon model, solve separate transport equations for turbulent kinetic
energy (k) and its dissipation rate (epsilon). They are more accurate than 1 equation models and
suitable for a wide range of turbulent flows.

g- 2 equation model K-w:

The k-omega (k-w) model is another 2 equation model that uses different transport equations for
turbulent kinetic energy (k) and specific turbulent dissipation (w). It is known for better accuracy in
adverse pressure gradient flows and near-wall regions.

h- 2 equation model K-W SST :

The Shear-Stress Transport (SST) model is a blend of the k-epsilon and k-omega models. It combines
the advantages of both models, offering good performance in a variety of flow situations, including
near-wall regions and adverse pressure gradients.

i- Reynolds stress model:

Reynolds Stress Models (RSM) solve transport equations for individual Reynolds stresses. They
provide more detailed information about the turbulence structure and are suitable for complex flows
but require more computational resources.

j- Hybrid RANS/LES method:

Hybrid models combine RANS and Large Eddy Simulation (LES) techniques. They use RANS in regions
with attached boundary layers and LES in regions with separated or unsteady flows. This approach
aims to capture both large and small-scale turbulent features.

33
k- DES (Detached Eddy Simulation) :
DES is a hybrid modeling approach that transitions from RANS to LES in different regions of the flow.
It is particularly effective for simulating separated and unsteady flows, providing a compromise
between computational cost and accuracy.

l- LNS (Limited Numerical Scales):

LNS is a hybrid RANS/LES approach that combines elements of RANS and LES based on the numerical
resolution. It dynamically selects the modeling approach depending on the resolved turbulent scales,
allowing for improved accuracy.

m- SAS (Scale Adaptive Simulation) :

SAS is a scale-adaptive modeling approach that dynamically adjusts the turbulence model based on
the local flow conditions. It aims to capture a broad range of turbulence scales and is effective for
simulating flows with varying turbulence intensity.

n- LES (Large Eddy Simulation):

Large Eddy Simulation (LES) is a computational approach used in fluid dynamics to simulate turbulent
flows. In LES, the large-scale turbulent structures are explicitly resolved, while the smaller-scale
structures are modeled. This allows for the simulation of a wide range of turbulent scales and
provides more detailed information about the flow compared to Reynolds-averaged methods.

• Smagorinsky model.

The Smagorinsky model is a sub-grid scale (SGS) model used in LES to account for the effects of
unresolved turbulent scales. It introduces a eddy viscosity term into the governing equations, which
is proportional to the square of the grid size and the rate of strain tensor. The model is simple and
widely used but has limited performance in flows with complex geometries and adverse pressure
gradients.

• Dynamic model.

The dynamic model is an extension of the Smagorinsky model that dynamically adjusts the model
coefficient based on the local flow conditions. The model coefficient is computed during the
simulation based on the resolved and sub-grid scale flow properties. This dynamic adaptation
improves the model's performance in various flow situations.

• Scale similarity model.

The scale similarity model, often referred to as the "Lilly model," is designed to provide a more
physically accurate representation of the sub-grid scale stresses. It employs a scale similarity
hypothesis to relate the sub-grid scale stress to the resolved scales. This model aims to capture the
anisotropy of turbulence and is particularly useful for wall-bounded flows.

• Structure function model:

The structure function model is a statistical approach used in LES to model the sub-grid scale
stresses. It is based on the concept of structure functions, which quantify the statistical differences
between resolved and sub-grid scale velocities. This model is particularly useful in complex turbulent
flows where local anisotropy is significant.

34
• SGS (Sub-Grid Scale) model:

The term "SGS model" is a general category that includes various approaches to model the effects of
unresolved turbulent scales. This can encompass models like the Smagorinsky model, dynamic
models, and other methodologies that aim to represent the sub-grid scale stresses and turbulence
characteristics.

DNS (Direct Numerical Simulation):

Direct Numerical Simulation (DNS) is a computational method in fluid dynamics that aims to provide
a highly detailed and accurate simulation of fluid flows by solving the governing equations without
any turbulence modeling. In DNS, the Navier-Stokes equations are solved directly for all scales of
motion, from the largest to the smallest, without relying on turbulence models to represent the
effects of unresolved scales.

Key features of DNS include:

1. Resolution of All Scales: DNS resolves the entire range of scales present in a turbulent flow, from
the energy-containing large eddies down to the dissipative small eddies. This allows for a detailed
understanding of the underlying physics of turbulence.

2. High Computational Cost: DNS is computationally expensive because it requires a very fine spatial
and temporal grid to capture all relevant scales. The number of grid points needed increases rapidly
with Reynolds number, making DNS particularly demanding for high Reynolds number flows.

3. Accuracy and Realism: Since DNS directly solves the Navier-Stokes equations, it provides accurate
results and captures complex phenomena, making it a valuable tool for fundamental research in fluid
dynamics.

4. No Need for Turbulence Models: Unlike Reynolds-averaged methods or Large Eddy Simulation
(LES), DNS does not rely on turbulence models. Instead, it explicitly resolves all turbulent scales,
making it a "first-principle" approach.

5. Applications: DNS has been used to study various turbulent flows, such as boundary layers,
channel flows, and isotropic turbulence. It is a valuable tool for validating and improving turbulence
models used in less computationally demanding simulations.

Despite its computational cost, DNS remains a fundamental tool for advancing our understanding of
turbulence and validating other turbulence modeling approaches. Ongoing developments in
computational resources and algorithms continue to extend the capabilities of DNS and make it
applicable to increasingly complex and practical flow scenarios.

Transition treatment:
The term "transition treatment" typically refers to strategies or methods used in fluid dynamics and
aerodynamics to simulate or control the transition from laminar to turbulent flow. Understanding
and controlling the transition process is important in various engineering applications, particularly in
aerospace, where the performance of aircraft and other aerodynamic components is influenced by
the flow characteristics.

Here are some common transition treatments:

1. Surface Roughness:

35
 - Introducing controlled surface roughness is one way to trigger the transition from laminar to
turbulent flow. Certain roughness elements can disturb the laminar boundary layer and promote the
onset of turbulence.

2. Trip Wires:

- Trip wires are small protrusions or disturbances intentionally placed on the surface of a wing or
other aerodynamic components. They disturb the flow and encourage the transition to turbulence.

3. Plasma Actuators:

- Plasma actuators use electrical discharges to energize and manipulate the airflow. They can be
used to control the transition process by inducing perturbations in the boundary layer.

4. Piezoelectric Actuators:

- Piezoelectric actuators are devices that can deform surfaces in response to electrical signals. They
can be used to introduce controlled disturbances to the boundary layer, promoting the transition to
turbulence.

5. Blowing or Suction:

- Controlled blowing or suction of air through small holes in the surface can be employed to
manipulate the boundary layer and induce turbulence.

6. Hybrid Laminar Flow Control (HLFC):

- HLFC systems use a combination of passive and active methods to delay the onset of turbulence.
Passive methods include smooth surfaces and suction panels, while active methods involve boundary
layer suction or blowing.

7. Surface Coatings:

- Special coatings with specific properties can be applied to the surface to promote or delay
transition. For example, porous coatings can absorb or release fluids, influencing the boundary layer
behavior.

The choice of transition treatment depends on the specific requirements of the aerodynamic design,
the desired performance characteristics, and the operational conditions. In many cases, designers
aim to delay the transition to turbulent flow to benefit from the lower skin friction associated with
laminar flow. However, there are also scenarios where controlled transition to turbulence is desired
for better aerodynamic performance or improved heat transfer characteristics.

These transition treatments play a crucial role in the design and optimization of aerospace vehicles,
ensuring that the aerodynamic characteristics are well-understood and meet the desired
performance goals.

Wall treatment:
While wall functions offer computational efficiency, it's essential to recognize their limitations. They
are generally valid for fully turbulent flows or for flows where the viscous sub-layer is well-resolved
by the mesh. In cases of transitional or laminar-turbulent flows, or in regions where the flow
undergoes significant changes, higher mesh resolution near the wall may still be necessary for
accurate predictions.

36
a- No slip condition:
The no-slip condition is a fundamental boundary condition in fluid dynamics, particularly near solid
surfaces. It stipulates that the fluid velocity at the boundary of a solid surface is equal to the velocity
of the solid surface itself. In other words, the fluid does not slip past the solid boundary, and there is
zero relative velocity between the fluid and the solid surface.

Mathematically, the no-slip condition can be expressed as:

Here,

The no-slip condition is crucial for understanding the behavior of fluid near solid surfaces. It has
significant implications for the formation of boundary layers and affects the overall flow patterns in
various engineering applications.

b- Use wall Function:

In computational fluid dynamics (CFD), simulations of turbulent flows often involve resolving the flow
close to the walls where viscous effects are important. However, resolving the near-wall region with
fine grids throughout the computational domain can be computationally expensive. To address this
challenge, wall functions are employed.

- Near-Wall Region: The region near a solid boundary, where the viscous effects dominate, is known
as the near-wall or viscous sub-layer.

- Wall Functions: Wall functions are empirical models or algorithms that approximate the velocity
and turbulence profiles in the near-wall region without explicitly resolving the viscous sub-layer.
These functions are often used in conjunction with the Reynolds-averaged Navier-Stokes (RANS)
models.

- Two-Layer Approach: In a typical wall function approach, the flow is divided into two layers:

1. Inner Layer: Modeled using a logarithmic law of the wall, capturing the viscous sub-layer.

2. Outer Layer: Modeled using an equilibrium logarithmic profile, which extends into the main part
of the flow.

- Benefits:

- Computational Efficiency: Wall functions allow coarser grids away from the wall, leading to
significant computational savings.

37
 - Less Sensitivity: The need for very fine near-wall grids is reduced, making simulations less sensitive
to grid resolution.

- Usage: Wall functions are commonly used in turbulence models such as the k-epsilon and k-omega
models.

6- Computations:
a- Process type:
In the context of computations, "process type" typically refers to the nature of the computational
tasks or algorithms being executed. Processes can be categorized based on their characteristics, such
as whether they are sequential, parallel, or distributed. Different process types impact the design
and implementation of computational solutions.

- Sequential Process: Tasks or computations are executed one after the other, with each task
depending on the completion of the previous one. This is a traditional, linear approach.

- Parallel Process: Computations are divided into smaller tasks that can be executed simultaneously.
Parallel processing aims to improve performance by dividing the workload among multiple
processors.

- Distributed Process: Computations are distributed across multiple computers or nodes, typically
connected over a network. Each node may perform a part of the overall computation.

b- Number of processors
The number of processors refers to the quantity of central processing units (CPUs) or processing
cores available in a computing system. More processors can lead to increased computational power
and the ability to handle parallel tasks more efficiently.

c- Processor speed
Processor speed, measured in gigahertz (GHz) or terahertz (THz), represents how quickly a CPU can
execute instructions. Higher processor speeds generally result in faster computations, but other
factors, such as architecture and parallelization, also play significant roles.

d- Total memory:
Total memory, often referred to as RAM (Random Access Memory), is the amount of volatile memory
available for a computing system. It impacts the system's ability to handle large datasets and
complex computations. Higher total memory allows for more substantial data storage and faster
access.

e- Compute systems

38
Compute systems encompass the hardware and software infrastructure used for computational
tasks. This can include high-performance computing (HPC) clusters, supercomputers, workstations,
and cloud computing platforms.

f- Distributed memory scalable parallel processing system:

This refers to a type of parallel processing system where each processor has its own local memory,
and communication between processors is achieved through message passing. Distributed memory
systems are often used in large-scale parallel computing, allowing for scalability and efficient
handling of massive datasets.

g- PC cluster:

A PC cluster is a type of parallel computing system that consists of multiple interconnected personal
computers (PCs). These systems use distributed processing to divide tasks among the individual PCs,
providing a cost-effective way to achieve parallel computing capabilities.

h- PC:
A personal computer (PC) is a general-purpose computing device designed for individual use. PCs
come in various forms, including desktops, laptops, and workstations. While not traditionally
associated with high-performance computing, multiple PCs can be combined into clusters to create
parallel computing environments.

39
In computational contexts, understanding these factors is essential for optimizing performance,
efficiency, and scalability. The choice of process type, the number and speed of processors, available
memory, and the type of compute system all influence the overall computational capabilities and
performance of a system.

7- Verification and validation:

a- Verification:
Verification in the context of computational simulations refers to the process of ensuring that the
numerical solution obtained from a simulation accurately represents the mathematical model it
intends to simulate. It involves checking the correctness of the implementation of the numerical
methods, algorithms, and codes used in the simulation. The goal is to confirm that the simulation
software performs as intended and that the mathematical model is appropriately solved.

Verification activities may include:

1. Code Verification: Ensuring that the numerical code accurately implements the mathematical
model, including checking coding errors, solving algorithms, and discretization schemes.

2. Order of Accuracy: Verifying that the numerical method used in the simulation converges to the
correct order as the grid is refined.

3. Consistency: Confirming that the numerical discretization is consistent with the governing
equations.

Verification is an essential step to build confidence in the reliability of a simulation code and to
ensure that the numerical solution is a faithful representation of the underlying physics.

b- Grid convergence study:

A Grid Convergence Study is a specific verification technique used to assess the convergence
behavior of a numerical solution concerning the grid or mesh refinement. The purpose is to
determine the sensitivity of the solution to changes in the grid resolution. This involves running
simulations with different grid sizes and comparing the results.

Key steps in a grid convergence study include:

1. Selecting Grids: Choosing a set of grids with varying levels of resolution (coarse to fine).

2. Running Simulations: Performing simulations using each grid and obtaining numerical solutions.

3. Error Analysis: Calculating the error between solutions obtained from different grids. Common
measures include comparing results at specific points or using global error norms.

4. Convergence Rate: Assessing how the error decreases with grid refinement. A higher convergence
rate indicates faster convergence to a more accurate solution.

c- Time-step convergence study:

Similar to grid convergence, a Time-Step Convergence Study assesses the sensitivity of the numerical
solution to changes in the time-step size. It helps ensure that the temporal discretization is
appropriate for capturing the dynamic behavior of the system accurately.

Steps in a time-step convergence study include:

40
1. Selecting Time-Steps: Choosing a sequence of time-step sizes.

2. Running Simulations: Conducting simulations using each time-step size and obtaining numerical
solutions.

3. Error Analysis: Calculating the error between solutions obtained with different time-step sizes.

4. Convergence Rate: Evaluating how the error decreases as the time-step size decreases. A higher
convergence rate indicates better convergence with smaller time steps.

d- Validation:
Validation involves assessing the accuracy and reliability of a simulation by comparing its results with
experimental or observational data. The objective is to determine how well the numerical model
represents the real-world physical phenomena it is intended to simulate.

Validation activities include:

1. Experiment Comparison: Comparing simulation results with experimental measurements or other

reliable data sources.

2. Quantitative Metrics: Using quantitative metrics, such as error norms, correlation coefficients, or
statistical measures, to assess the agreement between simulation and experimental data.

3. Sensitivity Analysis: Examining the sensitivity of the simulation results to variations in input
parameters and initial/boundary conditions.

4. Uncertainty Analysis: Evaluating the uncertainties associated with both the simulation and
experimental data.

Validation is a critical step to ensure that a simulation model is trustworthy and can be used with
confidence for predictive purposes in real-world applications. It provides evidence that the
simulation captures the relevant physics and reproduces observed behaviors.

a- Comparison against experimental data:

Comparison against experimental data is a fundamental step in validating computational simulations
and assessing the accuracy of numerical models. This process involves comparing the results
obtained from a simulation with experimental measurements or observations collected from real-
world experiments. The goal is to verify the reliability of the simulation and ensure that it accurately
predicts physical phenomena. Here are key aspects of the comparison against experimental data:

1. Quantitative Metrics:

- Error Analysis: Calculate the differences (errors) between the simulated values and the
corresponding experimental data. Common error metrics include mean absolute error (MAE), root
mean square error (RMSE), or percentage error.

- Correlation Coefficients: Assess the correlation between simulated and experimental data using
coefficients such as Pearson correlation coefficient or Spearman rank correlation coefficient.

2. Spatial and Temporal Comparisons:

- Spatial Agreement: Evaluate how well the spatial distribution of simulation results matches the
experimental data. This involves comparing profiles, contours, or other spatial representations.

41
 - Temporal Agreement: Assess the accuracy of temporal trends and dynamic behaviors predicted by
the simulation compared to experimental data.

3. Uncertainty Analysis:

- Uncertainty in Experimental Data: Consider uncertainties associated with the experimental

measurements. Experimental data often come with uncertainties, and understanding these
uncertainties is crucial for a meaningful comparison.

- Uncertainty in Simulation: Assess the sensitivity of the simulation results to variations in input
parameters and model assumptions.

4. Visualization:

- Graphical Comparison: Create graphical representations, such as plots, charts, or graphs, to

visually compare simulation results with experimental data. Visualization aids in identifying trends,
discrepancies, and areas of agreement.

5. Statistical Analysis:

- Statistical Tests: Perform statistical tests to determine whether any observed differences between
simulation and experimental data are statistically significant. Common tests include t-tests or
analysis of variance (ANOVA).

6. Qualitative Comparison:

- Qualitative Observations: Consider qualitative aspects of the comparison, such as the ability of the
simulation to capture physical trends, phenomena, or important features observed in the
experiments.

7. Model Calibration and Adjustment:

- Model Refinement: If discrepancies are identified, consider refining or calibrating the simulation
model. Adjust model parameters, boundary conditions, or other aspects to improve agreement with
experimental data.

8. Validation and Confidence:

- Validation: The comparison against experimental data is a crucial part of the validation process,
providing evidence that the simulation model is capable of accurately representing physical reality.

- Confidence Assessment: Assess the level of confidence in the simulation results based on the
quality of agreement with experimental data and the thoroughness of the comparison process.

Comparing simulation results against experimental data is an iterative process that may involve
refining the simulation model based on feedback from the comparison. This iterative refinement
enhances the accuracy and reliability of the numerical model, making it a valuable tool for predicting
real-world phenomena.

42
 Sources:

Wikipedia.org

Comsol.com

math.libretexts.org

www.researchgate.net

mathworld.wolfram.com

www.math.uci.edu

kth-nek5000.github.io

www.cambridge.org

vortexapproach.org

www.mdpi.com

www.simscale.com

byjus.com

collegedunia.com

www.cfd-online.com

link.springer.com

doc.cfd.direct

www.merkle-partner.de

ieeexplore.ieee.org

iccfd.org

www.siplec.leclerc

ntrs.nasa.gov

hec.usace.army.mil

symscape.com

simscale.com

forum.ansys.com

idealsimulations.com

www.robertschneiders.de

resources.system-analysis.cadence.com

learncax.com

www.cs.jhu.edu

blog.gridpro.com

cfd.ninja

www.ljll.math.upmc.fr

43