Computational Fluid

Dynamics (CFD)
BTD3233 / BHA4532
Topic 2
Related Course Outcomes

BTD BHA
Lecture Outline

CFD Solution Procedure:

• Introduction
• Problem setup-preprocess
• Numerical solution—CFD solver
• Result report and visualization—postprocess
 CFD
CFD Codes Commercial CFD Packages
Opensource CFD Packages
• FORTRAN • Fluent (Ansys)
• OpenFOAM
• C++ • CFX (Ansys)
• Nek5000
• Python • Star-CCM+
• etc
• etc • COMSOL
• etc

CFD users need to be able to be:

• analyze and interpret the computational solution
• is it physically realistic?
• is it numerically accurate?
Main steps in CFD

• Generic for all CFD analyses

• Three main steps in CFD:
1. Preprocessing
2. Solver
3. Postprocessing

Tu, J., Yeoh, G. H., & Liu, C. (2018). Computational fluid dynamics: a practical approach. Butterworth-Heinemann.
Overall procedure
Problem Setup
Step 1: Creation of geometry
Pre-processing
Step 2: Mesh generation
Step 3: Selection of physics and fluid properties
Step 4: Specification of initial and boundary conditions

CFD Solver
Solver
Step 5: Initialization and solution control Example: FLUENT general
Step 6: Monitoring convergence procedure using Ansys
Workbench

Result Analysis
Post-processing
Step 7: Visualization of numerical results
Step 8: Analysis and reporting of results
 Problem Setup
Actual 3D physical problem
Step 1: Creation of geometry
Step 2: Mesh generation
Step 3: Selection of physics and fluid properties
Step 4: Specification of initial and boundary conditions

• define and create geometry of the flow

region – the computational domain
• important to first understand the physical wall (no flow here)
2D computational domain
problem
• In this example:
• if W is sufficiently large, the flow
variation in this direction is negligible
• the computational domain can be 2D
instead of 3D
• this will require less computational
resource but produce a similar final Example 1: Fluid flow between two stationary plates
result

Tu, J., Yeoh, G. H., & Liu, C. (2018). Computational fluid dynamics: a practical approach. Butterworth-Heinemann.
 Problem Setup Actual 3D physical problem
Step 1: Creation of geometry
Step 2: Mesh generation
Step 3: Selection of physics and fluid properties
Step 4: Specification of initial and boundary conditions
open boundary (fluid
• In this example: can flow in or out)
• if the length of the cylinder W is
sufficiently large, the flow variation in
this direction is negligible
• the computational domain can be 2D
instead of 3D 2D computational domain
• this will require less computational
resource but produce a similar final
result
• since the boundary of the problem is
open surrounding, the computational
domain needs to take this factor into
consideration (it is different Example 2: Fluid flow through two stationary
cylinders in an open surrounding
compared to example 1)

Tu, J., Yeoh, G. H., & Liu, C. (2018). Computational fluid dynamics: a practical approach. Butterworth-Heinemann.
 Problem Setup
Step 1: Creation of geometry
Step 2: Mesh generation
Step 3: Selection of physics and fluid properties
Step 4: Specification of initial and boundary conditions

• Usually the creation of geometry is

performed in a CAD software (Solidworks,
Fusion 360 etc)

Example: Geometry creation in Ansys Fluent Design Modeler

 Problem Setup
Step 1: Creation of geometry
Step 2: Mesh generation
Step 3: Selection of physics and fluid properties
Step 4: Specification of initial and boundary conditions

• Meshing – subdivision of the domain into a

smaller subdomains
• The subdomains cannot overlap
• Mesh is necessary to solve the flow physics
within the geometrical domain
• Also called grid or cells
• The solver will calculate essential fluid flow
parameters at each cell:
• velocity Example: Mesh generation in Ansys ICEM CFD
• pressure
• temperature
• The size and number of cells are really
important
• determine the accuracy of the solution
 Problem Setup
Step 1: Creation of geometry
Step 2: Mesh generation
Step 3: Selection of physics and fluid properties
Step 4: Specification of initial and boundary conditions

• More number of cells = increased accuracy

• More number of cells = longer simulation
time
• There are different types of mesh:
• accuracy
• stability
• simulation time (resource)

https://www.manchestercfd.co.uk/post/all-there-is-to-know-about-different-mesh-types-in-cfd
 Problem Setup
Step 1: Creation of geometry
Step 2: Mesh generation
Step 3: Selection of physics and fluid properties
Step 4: Specification of initial and boundary conditions

https://www.manchestercfd.co.uk/post/all-there-is-to-know-about-different-mesh-types-in-cfd
 Problem Setup
Step 1: Creation of geometry
Step 2: Mesh generation
Step 3: Selection of physics and fluid properties
Step 4: Specification of initial and boundary conditions

• Basic terminologies in meshing:

• cell
• cell center
• face
• node
• edge
• structured mesh
• unstructured mesh
• Majority time spent in CFD is in generating
a high quality mesh that can produce
reliable result in a reasonable amount of
time

https://www.manchestercfd.co.uk/post/all-there-is-to-know-about-different-mesh-types-in-cfd
https://www.nas.nasa.gov/Software/FAST/RND-93-010.walatka-clucas/htmldocs/chp_16.surferu.html
 Problem Setup
Step 1: Creation of geometry
Step 2: Mesh generation
Step 3: Selection of physics and fluid properties
Step 4: Specification of initial and boundary conditions

• Generally fluid flow (and heat transfer) is a

very complicated process
• Various flow physics
• There are various models (simplifications)
that can be used to make it less
complicated → less computational resource
• Specific cases dan influence flow behaviour
drastically:
• combustion can strongly influence the
local and global heat transport

Tu, J., Yeoh, G. H., & Liu, C. (2018). Computational fluid dynamics: a practical approach. Butterworth-Heinemann.
 Problem Setup
Step 1: Creation of geometry
Step 2: Mesh generation
Step 3: Selection of physics and fluid properties
Step 4: Specification of initial and
boundary conditions

• User needs to define appropriate

conditions that mimic the real physical
representation of the fluid flow into the
computational domain
• Basic types of boundary condition (b.c.):
• inlet
• outlet
• wall
• symmetry plane
• many others
Example 1: Fluid flow between two stationary plates

Tu, J., Yeoh, G. H., & Liu, C. (2018). Computational fluid dynamics: a practical approach. Butterworth-Heinemann.
 Problem Setup
Step 1: Creation of geometry
Step 2: Mesh generation
Step 3: Selection of physics and fluid properties
Step 4: Specification of initial and
boundary conditions

Example 3: Fluid flow between two walls with a

symmetric boundary condition (Example 1)
Example 2: Fluid flow through two stationary
cylinders in an open surrounding

Tu, J., Yeoh, G. H., & Liu, C. (2018). Computational fluid dynamics: a practical approach. Butterworth-Heinemann.
 CFD Solver
Step 5: Initialization and solution control
Step 6: Monitoring convergence

• Iterative approach is used to solve the

defined problem
• Initialization
• Flow parameters (velocity, pressure,
temperature etc) are intelligently
guessed
• Good initial condition → quicker
solution
• Bad initial condition → longer time to
solve, may even fail

Overview of the solution procedure

Tu, J., Yeoh, G. H., & Liu, C. (2018). Computational fluid dynamics: a practical approach. Butterworth-Heinemann.
 CFD Solver
Step 5: Initialization and solution control
Step 6: Monitoring convergence

• Finite volume method (FVM)

• Finite difference method (FDM)
• Solution control • Finite element method (FEM)
• etc.
• Selection of appropriate
• discretization schemes • first order upwind
• interpolation schemes • second order upwind
• solution procedure • QUICK
• etc.

• SIMPLE
• SIMPLEC
• PISO
• etc.

Overview of the solution procedure

Tu, J., Yeoh, G. H., & Liu, C. (2018). Computational fluid dynamics: a practical approach. Butterworth-Heinemann.
 • Convergence – the solution that is obtained by the CFD algorithm has
CFD Solver a residual value that is accepted by the user
Step 5: Initialization and solution control
Step 6: Monitoring convergence

Example of convergence monitoring based on the designated

convergence criteria (in Ansys CFX)

Tu, J., Yeoh, G. H., & Liu, C. (2018). Computational fluid dynamics: a practical approach. Butterworth-Heinemann.
 CFD Solver
Step 5: Initialization and solution control
Step 6: Monitoring convergence

Example of convergence monitoring based on the designated

convergence criteria (in Ansys Fluent)

Tu, J., Yeoh, G. H., & Liu, C. (2018). Computational fluid dynamics: a practical approach. Butterworth-Heinemann.
 CFD Solver
Step 5: Initialization and solution control
Step 6: Monitoring convergence

Example of good and bad convergence

Example of convergence plot for different

regions of the domain Example of convergence plot

https://www.simscale.com/knowledge-base/how-to-check-convergence-of-a-cfd-simulation/
 CFD Solver • Number of cells very important to accuracy
• Not enough cells → inaccurate result
Step 5: Initialization and solution control • Too many cells → simulation too long
Step 6: Monitoring convergence • Mesh dependent study proves that the simulation result is not affected by
the mesh

Same geometry with different number of

cells Example of mesh dependence study
 Result Analysis
Step 7: Visualization of numerical results
Step 8: Analysis and reporting of results

• Visualization of the obtained CFD result is

very important
• Helps the used to interpret the result
• CFD algorithm will produce millions of
values for various parameters
• Various types of visualization
• x-y plot
• streamline plot
• contour plot
• vector plot

https://www.afs.enea.it/project/neptunius/docs/fluent/html/cfd/copyright.htm
 Result Analysis
Step 7: Visualization of numerical results
Step 8: Analysis and reporting of results

Velocity vector plot Normalized velocity contour plot

Velocity profile plot Dynamics pressure contour plot

Tu, J., Yeoh, G. H., & Liu, C. (2018). Computational fluid dynamics: a practical approach. Butterworth-Heinemann.
 Result Analysis
Step 7: Visualization of numerical results
Step 8: Analysis and reporting of results

Velocity vector plot Contour plot of dynamics

pressure distribution

Streamline plot Line contour plot of pressure

coefficient
X-Y plot of the normalized x-direction velocity Tu, J., Yeoh, G. H., & Liu, C. (2018). Computational fluid dynamics: a practical approach. Butterworth-Heinemann.
 Result Analysis Various commercial graphical software
Step 7: Visualization of numerical results
Step 8: Analysis and reporting of results

• The raw CFD simulation data (millions of

points / values) is visualized using various
plots
• The physical meaning of the flow behaviour
can be analyzed using the plots
• The relationships between the observed
parameters are then reported by including
the visualized data

Tu, J., Yeoh, G. H., & Liu, C. (2018). Computational fluid dynamics: a practical approach. Butterworth-Heinemann.
 Result Analysis
Step 7: Visualization of numerical results
Step 8: Analysis and reporting of results