Computational Fluid Dynamics for Incompressible Flows

Professor Amaresh Dalal

Department of Mechanical Engineering
Indian Institute of Technology, Guwahati
Module 01: Introduction to Computational Fluid Dynamics
Lecture 01: Application of CFD

Hello everyone. So, today we will study the application of CFD. So, it is a introductory
lecture. CFD is the short form of Computational Fluid Dynamics.

(Refer Slide Time: 00:45)

So, fluid dynamics you know that when we study the fluid in motion then that is known as
fluid mechanics. So, fluid dynamics is the science of fluid motion. And CFD is the, this fluid
dynamics when we study numerically then that is your CFD, Computational Fluid Dynamics,
Computational Fluid Dynamics. Fluid flow is commonly studied in one of the three ways,
Experimental fluid dynamics, theoretical fluid dynamics and computational Fluid Dynamics.

So, first one is experimental fluid dynamics, so you can study the fluid dynamics doing some
experiments, so you can measure the temperature with thermocouple or the velocity with
pitot tube and you can study what is the velocity profile or the temperature profile using
experimental techniques.

Other way is that theoretical fluid mechanics or fluid dynamics, you have the governing
partial differential equations, which represent the fluid flow, heat transfer or other multi
physics problems. So, when you write the partial differential equations, we with some
assumptions you can write it to ordinary differential equations and with the boundary
conditions if you can solve those equations, then you can have the exact solution.

So, theoretical also you can study the fluid dynamics, but you can study theoretical under
certain assumptions, because you have to bring down this partial differential equation to
ordinary differential equation. Like if you are studying the fully developed fluid flow
problem, then fluid fully developed means there is no variation of velocity in axial direction.

So, it essentially boils down to one dimensional problem. And with boundary conditions you
can solve and get the velocity profile theoretically. The other way is to study this fluid
dynamics numerically. So, you have the partial differential equations using these
computational fluid dynamics techniques, you can solve the partial differential equation and
get the solution using computational fluid dynamics.

So, computational fluid dynamics is the science of predicting fluid flow, heat transfer, mass
transfer, chemical reactions and related transport phenomena by solving the mathematical
equations which govern these processes using numerical approach. So, you can see that CFD
is the calculation of properties of a flowing fluid.

So, it is not only limited to fluid dynamics but you can extend it to heat transfer, you can
mass transfer, then some other multi physics problems, electro hydro dynamics flow or
magneto hydro dynamics flows or with chemical reaction and many more.

(Refer Slide Time: 04:26)

CFD provides a qualitative and sometimes even quantitative predication of fluid flow by
means of three ways, first one is mathematical modelling, then numerical methods, then
software tools. So, if you want to solve some problem, then first you find or you write the
partial differential equations which govern the flow, so that is your mathematical modelling,
then you can choose suitable numerical methods, like discretization techniques and some
solution techniques you can choose.

Then you can use some software tools like you can solve using some solvers those discretize
equations and get some solutions in terms of data, so only numbers will be there as output.
So, to visualize this data you can use some post-processor and in the pre-processor you need
to give required boundary conditions and initial conditions if required. So, in post processing
tools you will be able to visualize those data and you can see the velocity profile or
temperature profile or some contours using some post-processing tool.

(Refer Slide Time: 05:50)

So, we will now see the history of this CFD. So, you can see that earlier CFD work started in
1910. And Richardson used human computer to solve Laplace equation using finite
difference method and he solved flow over cylinder only for the potential flow. For inviscid
flow he solved using finite difference method. So, in 1910 using human computer.
(Refer Slide Time: 06:24)

Then Courant and his research groups like students, Friedrich and Lewy they solved the
hyperbolic equations in 1928.

(Refer Slide Time: 06:35)

Von Neumann in 1950 proposed the stability criteria for parabolic problems, which is known
as Von Neumann stability analysis and in this course we will study also this stability analysis,
Von Neumann stability analysis.
(Refer Slide Time: 06:53)

Harlow and Fromm computed unsteady vortex street using digital computer in 1963. And
later Harlow and Welch published a scientific American article which ignited interest in
modern CFD and the idea of computer experiments. So, they solved first time the free surface
flow, which is a two component fluid flow problems in 1965 using digital computer.

(Refer Slide Time: 07:26)

Professor Brian Spalding is known as the founding father of CFD. So, he and his research
group developed this boundary layer codes which is known as GENMIX and in they
developed these codes in the years 1960 to 1970s and 1972, they actually those code they put
as a software as known as GENMIX. And Patankar is student of Professor Spalding and they,
their solution techniques for incompressible flows is published through the 1960s and they
first time proposed the simple family of algorithms, which we will study in this course for
solving the full Navier-Stokes equations.

(Refer Slide Time: 08:15)

Jameson computed Euler flow over complete aircraft, so this is for compressible flows Euler
equation they solved and they published almost 14 papers in 1981 on this using the Euler
flow solver.

(Refer Slide Time: 08:37)

Later this there was a problem because in the structured mesh it is very difficult to generate
for a complicated geometry, so the unstructured grid you can easily fit in a complex
geometry. So, unstructured mesh methods developed in 1990s used for first used for
aerodynamic calculation in NASA. So, later professor Murthy and her group developed this
unstructured grid solvent for ANSYS Fluent. And she and her co-workers developed different
modules in fluent.

And you can see the publications during the development and now this fluent is known as
ANSYS Fluent and most of you I think you use this commercial structured ANSYS Fluent
for solving any problem numerically.

(Refer Slide Time: 09:37)

Now, who are interested in CFD? So, practical problems may include multi-physics like
energy flow, chemical reaction, phase change et cetera. And that domain or computational
domain mostly these are three dimensional and very complicated. So, it is relevant and so it is
easier to use the numerical technique to solve the governing equations.

So, relevant industries include automotive, chemical processing, aerospace, heating

ventilation and AV conditioner; even nowadays in biomedical applications we use CFD. So,
the results of CFD analysis is relevant engineering data used in conceptual studies of new
designs, detailed product development, troubleshooting and redesign.
(Refer Slide Time: 10:40)

CFD enables scientist and engineers to perform numerical experiments in a virtual laboratory.
So, you can see this is a solution for flow over a circular cylinder, you have a circular
cylinder and you have a fluid flow over it, so if you do the real experiment the you can see
the visualization. Now, if you do the CFD simulations, you will get this type of simulation
result. So, you can see that you are doing actually numerical experiments in a virtual flow
laboratory.

(Refer Slide Time: 11:25)

CFD gives an insight into flow pattern that are difficult, expensive or impossible to study
using traditional experimental techniques. In experiment, it is very difficult to get the velocity
profile or temperature profile at desired or different locations. But, when you use this CFD
technique, you discretise these governing equations at a discrete points. So, you can get easily
any value, velocity, temperature or species at those discrete point.

So, it is easy to visualize the results in terms of velocity contours or velocity vector or
temperature profile. CFD does not replace the measurements completely, but the amount of
experimentation and the overall cost can be significantly reduced. The doing experiment is
costly, because you need to fabricate the setup and also you need different instrument to
measure velocity, temperature; so it is very costly. So easily you can use this numerical
techniques to solve those partial differential equations for a particular problem and get the
solutions.

And equipment and personnel difficult to transport and CFD software is portable, easy to use
and modify. So, we can see that real experiments are expensive whereas CFD simulations are
cheaper; real experiments are slow, CFD simulations are faster; real experiment is sequential,
CFD simulations are parallel; real experiments are single purpose, CFD simulation are
multiple purpose. What does it mean? That, when you are doing some experiment, so you
have made the setup, so you are doing the experiment in the laboratory. So, at a time you can
do only single experiments and obviously doing the experiment, fabricating the setup it is
very expensive.

Whereas if you develop a numerical solver for solving that problems, fluid flow or heat
transfer problems, then once the code is ready you can solve the problem for different
conditions parallely. So, that means you can run that solver in different computers for
different parameters.

So, parallelly you can run and obviously you can see that it is very portable because you can
take the solver with you to somewhere else, but it is very difficult to shift the experimental
setup from one location to other.
(Refer Slide Time: 14:40)

The result of CFD simulations are not always 100 percent reliable. The input data may
involve too much guessing or imprecision. The mathematical model of the problem at hand
may be inadequate. The reliability of the CFD simulations is greater for laminar flows than
for turbulent ones. For single-phase flows than for multi-phase flows and chemically inert
system than for reactive system.

When we solve these equations, we have some assumptions, so obviously, when it becomes
more multi-physics then you have lesser reliability. CFD is a highly interdisciplinary research
area, which lies at the interface of physics, applied mathematics and computer science.

(Refer Slide Time: 15:29)

Now, let us see few examples or applications of CFD applications, sorry, applications of
CFD. So, first, we see in the Aerospace. So, we can see that when you are actually designing
this aeroplane, obviously you can solve the governing equations and you can design such a
way that you can have the minimum drag while flying.

So, obviously, in designing the exterior even for interior design also, you need the safe
dissimulations for the passenger comfort, to design the combustor then pumps, missile
systems. So, you can use this CFD to design these external aerodynamics, propulsion, missile
systems and pumps.

(Refer Slide Time: 16:31)

CFD application in automobile, so obviously, here also you can have the you can use CFD
for exterior design to reduce the drag, even for the interior design you can use CFD so that
the air form the air conditioning reaches to all the passengers to fill the comfort. So, for
interior design you can use the CFD, even in the combustion chamber you can use the CFD
techniques and also for engine cooling and external aerodynamics you can use the CFD
application in automobile.
(Refer Slide Time: 17:14)

So, you can see some in this animations one car is there and you can see how the flow
physics loose behind this car. So, it is very complex you can see, so it is a numerical
simulation.

(Refer Slide Time: 17:35)

So, some CFD application in process engineering. So, you can use it for reactor design, heat
exchanger, so here fluid flow and heat transfer you can solve; mixtures, boiler pumps,
compressor and diffuser design you can use this CFD.
(Refer Slide Time: 17:52)

This is some fun is moving and you can see how the flow looks like.

(Refer Slide Time: 18:04)

CFD also you can apply in heating, ventilation and air conditioning. So, when you are
designing some room for comfort stay, you can use the CFD technique. So, air flow around
building, burner design, environmental control system, heating system design, room flow
distributions, all this you can use, to solve this problem you can use CFD.
(Refer Slide Time: 18:36)

CFD is also having applications in electronics cooling, so you can see that when you use any
computer or laptop you have the processor and if processor is having high temperature, so
earlier days in your desktop you will find a fun mounted on the processor and it is cooling.
So, these kind things you can actually solve using CFD, you can see this picture where you
can see the temperature distribution.

So, this is the fun and here you have the processor with fin mounted on it, so the cooling is
taking place due to the force convection, here you can see it is the application in heat pipe, so
you can do the CFD analysis for component level flow and cooling, electronic chip cooling,
magnetic storage devices, telecommunication equipment.

(Refer Slide Time: 19:39)

So, some piezo electric device you can see how the fluid flow is taking place, vortex are
generated.

(Refer Slide Time: 19:54)

You can also use CFD in sports, so you can see when the ball is moving how the flow physics
looks behind the ball you can see from here or if you are cycling then how the flow physics
behind you, you can solve using the CFD technique, even for car racing and swimming you
can use CFD techniques.

(Refer Slide Time: 20:22)

Now, there are many applications in biomedical, so blood flows through arteries, so we can
have the deformable arteries as well using different advance technique you can simulate this
problems. Heart pumps where you have a moving boundary problems and for tumours you
can have the solution of the heat generation inside the tumour and that you can model using
CFD.

(Refer Slide Time: 20:58)

You can see this is the blood flow inside the heart, so one-simulation results.

(Refer Slide Time: 21:07)

So, there are many applications of CFD in many different industrial in industry and you can
see in that power generations, power pants you can have the application of CFD; in
hydraulics, so hydraulic turbine all those things you can model using CFD; oil and gas
industries, in marine industries, so there are many applications in different kind of industries.

(Refer Slide Time: 21:39)

So, the governing equations you can write; with certain assumptions you can apply this
governing equation to some problem. So, asssumptions may be like incompressible flow,
unsteady flow, laminar flow, Newtonian fluids, single phase, constant properties. So, for this
you can write the governing equations. So, obviously, all the equations you can have the
conservation loss, conservation of mass where continuity equation you can write,
conservation of momentum where Navier-Stokes equation you will write; conservation of
energy, energy equation; conservation of species, diffusion equation.

(Refer Slide Time: 22:15)

So, you can see that this is your continuity equation in general but if it is incompressible
flow, obviously it will be diversions of v will be 0. Then, you have general transport
equations; for any species you can write this equation, especially for s momentum equation if
you write this equation then it is the temporal term, this is your convective term and this is
your viscus term, this is the pressure gradient term and if you have any source term that you
can write, energy equations in terms of enthalpy if you write, then this is the energy equation
where k is the thermal conductivity.

Species transport equation in terms mass fraction, Y i is the mass fraction, so that you can
write, so you can see you can see all these equations you can write in a general transport
equation for a general variable, phi where s may be different or the diffusion coefficient
gamma will be different.

(Refer Slide Time: 23:17)

So, when you have the governing equation you need to use some discretization techniques, so
mainly in CFD we use three different techniques, one is finite difference method, then finite
volume method and finite element method. In this course, we will study only finite difference
method and finite volume method.

Finite difference method, generally we use Taylor series expansion and find the
approximation of any derivative, first or second derivative and we discretize the governing
equations and write the final algebraic equations. When we use finite volume method then we
integrate the governing equation over a control volume and write the discretize equation.
But, in finite element method we integrate the governing equation with some waiting function
in a particular element and we write the discretise equations. So, obviously, in this course we
will study only finite difference method and finite volume method.

(Refer Slide Time: 24:19)

To solve these governing equations you need to discretise this domain into grid. At those
discrete point, you need to solve the discretised algebraic equations. So, now grid can be
classified into two; structure grid and unstructured grid. So, as I told before that structure grid
are easy to generate in a simple geometry, but if we have a complex geometry then it is very
difficult to generate structure grid, so you need to use, divide the domain into blocks, and in
the block you need to generate the mesh.

But, unstructured grid is it is very easy to fit into a complex geometry. So, structured grid can
be further classified as regular grid, block structured grid and curvilinear grid.
(Refer Slide Time: 25:14)

So, in the regular grid, regular structured grid you can see these are actually orthogonal in the
coordinate system. So, if it is we use Cartesian grid then this grids are orthogonal to each
other, it can be uniform spacing or non-uniform spacing; you can see here we have used non-
uniform spacing. Cylindrical grid, so in cylindrical coordinate system these are orthogonal to
each other. And in spherical grid, in spherical coordinated the grids are orthogonal to each
other. So, these are regular structured grid.

(Refer Slide Time: 25:47)

Then block structured grid. So, you have a complex geometry, then you can divide the
domain into zones and generate the mesh and at the interface you keep the continuity, make
the continuity. So, we can see it is some cylinder is there in this here, so it is cylinder is there.
So this is divided into 1, 2, 3, 4, 5, 6, 7, 8, 8 subdomains and each domains the grids are
generated.

And it is a continuity is maintained at the interface. So, this is one the representation of the
grid for a flow over circular cylinder. So, these are known as block structured grid because
you are dividing the domain into blocks and in each block, you are generating the mesh.

(Refer Slide Time: 26:42)

Then curvilinear grid. So, curvilinear grid. So, curvilinear grid are known as body fitted grid,
so you can see that the grids are following the body, surface of the body. So, here you can see
if it is this is the body the grids this is the grid so it is following the body. So, these are
curvilinear structured grid, so obviously, these girds are not orthogonal you can see. So, these
are non-orthogonal grids. And for different geometry you can see that how the curvilinear
grids are generated. In curvilinear grids, as I told that, grids will follow the boundary.

(Refer Slide Time: 27:17)

In unstructured grid, so we can have regular grid and hybrid grid. In regular grid, you can
have same type of cells, like hexahedral, tetrahedral, prism or pyramid. In 3D and triangular
cell or quadrilaterals in 2D. And hybrid grids are grids where you can have more than one
type of cells; if you have hexahedral cells and tetrahedral cells are mixed then obviously you
can have hybrid grid.

(Refer Slide Time: 27:44)

So, you can see here, so you can see that in this case this first figure, you can see the grids are
fine near to the boundary to capture the gradient mode correctly or mode accurately. So, local
refinement you can do using unstructured grid. But, in structured grid, it is very difficult to do
local refinement because to maintain the continuity this will be extended in other directions.

So, in unstructured grid that is why it is very easy to use the local refinement. And you can
see you can use hybrid grid, hybrid grid mean more than one type of cells if you use. In this
case, you can see near to the boundary you have quadrilateral cells and away from the
boundary, you have triangular cells.

So, that means near to the boundary you have almost structure and orthogonal mesh so that
you can capture the gradient correctly. And away from the surface where you do not have
much gradient you can use triangular cell, so this is kind of hybrid cell. So, in 3D, you can
have hexahedron type cells or tetrahedron or pyramid or prism.

Now, it is in commercial softwares they use any type of poly hydra, but Ansys Fluent earlier
they used use only these four types of cells for the three dimensional domain. And for surface
grid, triangular or quadrilateral mesh are used.
(Refer Slide Time: 29:30)

So, you can see there are some advantages and disadvantages of this structured and
unstructured grid. So, you can see the structured grid, what are the advantages, efficiency in
CPU time and computer memory, because in structured grid you can easily find the
neighbours, because you have the indices I, J, K and I plus 1, I minus 1, or J plus 1, or J
minus 1, will give you the neighbours.

So, less storage memory is required and it is good environment for multigrid technique.
Whereas, in the unstructured grid, advantages are flexible for very complex geometry region
and it permits automatic adaptive refinement based on regions of interest, so at the particular
region if you need very fineness then you can use actually unstructured grid.

The disadvantages of the structured grid is that lack of total flexibility for very complex
regions because you cannot do local refinement and it cannot be distorted to increase
resolution in a localised region. The unstructured grid disadvantages are there, that requires
more memory as compared to structured grid to store the connectivity.

So, in unstructured grid you cannot find the neighbours with the indices I plus1, I minus 1,
because there is no structured way these grids are oriented. So, in pre-processing stage, you
need to find the neighbours or connectivity so that you can use it later while calculating the
fluxes.

So, obviously, you need more memory requirement to store this connectivity and the
neighbours’ informations. And this another disadvantage for unstructured grid is not
necessarily amenable to the implementation of multigrid.
(Refer Slide Time: 31:25)

So, now we will talk about the grid terminology. So, when you discretise the domain for
generally, for finite difference method we solve the equations, discretised algebraic equation
is solved at this grid node, which are known as vertex.

But, in case of finite volume method we integrate over this cell, so the value is stored at the
cell centre or cell centroid and this is the cell and these are known as faces, so in faces in two
dimension it is line, but it is three dimensional it will be a surface. So, you can have node
based finite volume scheme where phi the variable stored at vertex, or cell based finite
volume method where the variable stored at the cell centroid.

(Refer Slide Time: 32:22)

Based on the variable storage you can have two different types of grid arrangement, one is
staggered grid and another is co-located grid, we will learn more detail when we will study
the finite volume method, but here just I will just introduces that, when all the variables you
store at the cell centre or at the same node then it is known as co-located grid. So, you can see
the velocities, pressure and any species let us say temperature or any species you can store at
the cell centre at the same point then this is known as co-located grid.

But, you can see that it is very easy to have the data structure, because all the variables are
stored at the cell centre, but it is having some disadvantages, so that we will learn later, that is
known as velocity pressure decoupling. So, if you use co-located grid, it may be possible that
your velocity and pressure are not talking each other because your velocity there will be due
to the pressure difference there will be velocity.

So, if there is no pressure difference then obviously there will be no velocity, so this pressure
and velocity if it is decoupled then there will be problem, so that is known as pressure
velocity decoupling. And when you have this checker boarding kind of distribution of
velocity or pressure then you will get this type of problem in co-located grid. So, that we will
discuss in detail later.

And in staggered grid to avoid this problem, velocity pressure recoupling problem, we use
staggered grid, where we store the variable at different places. So, we solve the pressure and
the scalers like temperature or species at the cell centre and the phase centre in staggered way
we will solve the velocities. So, we can see in this figure, so if this is the figure where we are
solving at the cell centre only the pressure and any scaler like temperature or species, whereas
we solve the velocities at the at this cell.

So, this is velocity, U velocity and at the cell we solve the B velocity. So, you can see that U
and B are solved at staggered way and pressure and any scaler are solved at the cell centre of
the main control volume. So, this is the main control volume for pressure and scaler but this
is the control volume for U velocity and this is the control volume for B velocity.

So, in staggered grid, actually you can avoid this pressure velocity decoupling, but you can
see that storage requirements are different. Because as and U or B are served and different
places, so you need to write the code carefully so that you can take care about the data
structure.
(Refer Slide Time: 35:41)

So, you can see that in co-located arrangements store all the variables at the same set of grid
points and to use the same control volume for all the variables. Advantage, obviously, it is to
code, because all the variables are stored at the same point. Disadvantage is, pressure velocity
decoupling. And in staggered arrangement, not all variables share the same grid, so advantage
is strong coupling between pressure and velocities. Disadvantage is higher order numerical
schemes with order higher than second order will be difficult.

(Refer Slide Time: 36:16)

For the unsteady problems, also if you are discretising and using forward time then this is
known as explicit method where you have only one unknown, where n plus 1 is the current
time step and n is the previous time step. So, this is known as explicit method. So, only one
unknown is there and you can have in the right hand side all known terms at the time level n.

But if you use backward time discretization method for the temporal gradient then it is known
as implicit method, where you have more than one unknowns. You can see it is n plus 1 and
here also n plus 1, so obviously you have more than one unknowns at n plus 1 time level,
which is the present time level. So, this is known as implicit method.

So, in explicit method as it is only one unknown it is easy to solve, but where in implicit
method where you have more than unknowns so you have to use some numerical techniques,
sorry numerical solvers to solve these equations.

(Refer Slide Time: 37:24)

So, for solving this discretization equation, you can use these iterative methods, one is Jacobi
method, other is Gauss Seidel method, we will discuss in detail later, Successive Over-
Relaxation, Alternative direction implicit method, so these are some iterative methods. We
have some other methods where which are known as conjugate gradient methods, biconjugate
gradient methods and you can also use multigrid.
(Refer Slide Time: 37:49)

So, you can see that when you are solving some problems using CFD, first you identify the
right approximation and write down the governing equations. So, identification of right
approximation means, whether it is Viscous or inviscous inviscid, because you can write the
governing equation accordingly, whether laminar or turbulent, incompressible or
compressible, single-phase or multi-phase, so accordingly you first identify the right
approximation and write the governing equations.

Then you identify the right solution method. So, which discretization scheme you want to
use, say finite difference method, finite element method or finite volume method, whether
you can use want to use structured or unstructured grid. And what is the order of accuracy
you will use, temporal and spatial both. So, accordingly you need to discretise these partial
differential equations.

Then in pre-processing stage, generate the computational grid, because at discrete points you
need to solve this discretized equation, so depending on your choice whether structured or
unstructured grid you generate the computational grid, then in pre-processing stage you
assign the boundary conditions. If it is time varying then with solutions you need to apply the
boundary conditions.

Then set initial conditions for the unsteady problem. If it is a steady problem then you need to
have the guess solution at starting while starting the iterative method. Then once all these are
done then you compile the code, so if there are some errors then you fix the bug, then prepare
input parameters, then you solve this, run the code, monitor the solutions. So, whether your
error is decreasing with iteration or time that you check and monitor.
Once you get the conversions then you get the results. But, result you will get in terms of
numbers only. So, at discrete points you will have the values of particular variables. So, those
you collect and organize the data, then you use some post processing software and analyse
the results.

Now, once you post process it, first you need to verify the solution, what is verification?
Whether these are physically correct or not. Because if you are solving a fluid flow problem
in a channel, let us say the flow is taking place from left to right, but if you are getting the
solution from right to left, the velocity is coming from right to left , then obviously it is not
physically correct. So, based on your problem whatever you have chosen, so you check
whether the results are physically correct or not.

So, do the results make sense? Are the trends right? Does it agree with previous calculation
on similar configurations? So, that you verify. Once you get it, then you solve a known
problem, which already the solution is available in literature, so that is known as validation.
So, when you write the solver first time, you need to know whether the solver is giving
correct result or not, so to test it you need to solve a known problem which is available in the
literature.

So, does the results or the aspect of the results agree with theory or experiments? So, you can
have some numerical or experimental results available in literature. So, qualitatively you can
check first whether these are matching or not, but you need to verify also quantitatively. That
means you need find the velocity distribution at a particular line or a particular area then you
can compare it with the available results in literature, whether it is numerical or experimental
but need to compare, or any other values like drag coefficient or lift coefficient or the
moment coefficient or the shear stresses, distribution along a wall.

So, all those things quantitatively you need to compare with the with your results and the
results available in the literature to verify your solver so that you will be confident that your
code is correctly written and there is no bugs.
(Refer Slide Time: 42:55)

So, there are different types of CFD codes available like commercial CFD code ANSYS
Fluent, Star CD, CFX, then COMSOL it is a very multi physics software, so ANSYS Fluent,
Star CD these are written in using finite volume method, but COMSOL is a written using
finite element method. You have some public domain software like PHI3D, HYDRO, Open
FOAM. And for this solving the multi-phase flow you have Gerris and Basilisk, so that are
also available and these are free, you can download and you can use it.

And other CFD software includes the grid generation software, obviously you need to
generate the grid before solving the equations. And there are some software like Gridgen,
Gambit, ICEM CFD, these are some commercial softwares available but also some open
source software is there like Salome, so you can generate structured and unstructured grid
and it is having the graphical user interphase also to generate the grid.

And to post process the results you have flow visualization software, so Tecplot, FieldView,
you can use to post process the results and visualize the results, so these are some commercial
and open source software we discussed. So, in today’s lecture, we have seen why we need to
use the computational fluid dynamics and what are the procedures to solve any problem using
CFD.

We have also discussed some history of CFD, then we have seen some application of CFD in
particular areas and later we have seen the different type of grid, structured grid and
unstructured grid; we have also seen different grid arrangement like co-located grid and
staggered grid. After that, also, when you discretize the equation, you will get the discretized
algebraic equations.
But you need to solve it, so for solving it you need some solver, so we have seen some
iterative solvers, like Jacobi, Gauss Seidel or Successive-Over-Relaxation and also some
direct method like conjugate gradient method or biconjugate gradient method. Those things
we will discuss more detailed in other modules.

And we have seen that when you solve a problem, what are the steps you need to follow for
solving any problems. Thank you.