Introduction To Computational Fluid Dynamics: Basics of CFD Numerical Model in CFD Steps in A CFD Simulation

Introduction to
Computational Fluid Dynamics

Dr. Kumar Bappaditya Salui
Somenath Chatterjee
1 © 2018 ANSYS, Inc. November 26, 2018
Contents
Overview
Basics of CFD
Numerical Model in CFD
Steps in a CFD Simulation
Questions Welcome!

Overview
What is CFD?
Computational Fluid Dynamics (CFD) is the numerical

study of how things flow.
CFD is used as a tool by many engineers (mechanical,
chemical, etc.) across a broad range of industries.
CFD can provide detailed information about what is
happening in a process where fluid is in motion.

History of CFD
Mid 1950’s - began to understand mathematics

Early 1980’s - first general purpose CFD codes, needed very
large computers, required extensive knowledge of fluid
dynamic and were used for research only
Present day- recent advances in computing power,
advanced computer graphics and robust solvers make CFD
a cost-effective engineering tool
Uses of CFD
Process/Food Industry
Automotive/Mechanical
 Architectural
 Environmental
Power Generation Safety/Health

The Building Block
The most basic building block in CFD is the

conservation volume
conservation volume
what goes in what goes out
what is
made/destroyed
Conservation volume:
what goes in + what is made/destroyed = what goes out
Assembling the Blocks
 The conservation volumes are assembled together to fill

space.
Simple algebraic
equations describe how
the conservation
volumes are connected.
These simple equations
conserve mass,
momentum and energy.

Conservation Volumes Assembled Around A Car
Once the conservation volumes are assembled, a flow field

can be calculated.
The Final Solution
The final flow field may look like this.

Basics of CFD
CFD Basic Concepts
The Control Volume

Conservation Laws
Governing Equations of CFD
Linear and Non-Linear Partial Differential Equations
Boundary Conditions
Discrete Methods
Simple Heat Transfer Example

The Control Volume
control surface
Control • control volume refers to a region in space

volume • region is bounded by a control surface
• size and shape is completely arbitrary
• could be a pipe, room, turbomachinery blade row, etc.

• applicable to finite or infinitesimal size regions
• basic conservation principles are applied to this defined region
Conservation Principle
Fluid flow in most devices can be described using the principles of conservation of mass,
momentum and energy (also species etc.)
Mass conservation says that matter is neither created or destroyed, but rather
conserved
Momentum conservation is really just an expression of Newton’s First Law
Energy conservation is a statement of First Law of Thermodynamics
CFD is based on these fundamental laws!

Conservation Principle
Applied for a finite control volume yields integral equations

• Starting point for the finite volume method
Applied for a infinitesimal fluid element results in partial

differential equations (differential form)
• Starting point for the finite difference method
Mass Conservation
control surface
Mass
Mass in Mass out
storage
control volume
In words
Rate of increase
of mass in volume = Mass flow out - Mass flow in
Integral Equation

0 cv  d  cs  v  dA

t

Example: Uniform, 1D, Steady Flow

0   AV1   AV 2

Momentum Conservation
Momentum in Momentum Momentum

storage out
Net Force
In words
Net Force = Rate of increase + Momentum outflow
of linear momentum - Momentum inflow
Integral Equation
 d ( mv ) 
 
F  cv  vd  cs  vv  dA
 
t

dt
Example: Uniform, steady, 1D Flow
Fx  P1 A 1  P2 A 2  ( mV)
 2  ( mV)

 1
m
 = AV
Energy Conservation
Heat In Work out

Total Energy Total
Enthalpy storage Enthalpy
flow in flow out
In words
Rate of increase + Enthalpy outflow
Heat in - Work out = of internal energy - Enthalpy inflow
Integral Equation

 W
(Q  )   e d + cs H v  dA
 
 t cv
v2 Example: 1D Steady Flow
e= u+  gz
2
 W
Q   0  (mH)
 2  (mH)
H eP   1
 = AV
m
Governing Equations in Differential Form
Conservation of mass
 

 t  xj
 u j   0
Conservation of momentum

t
 u i  

 xj
 u j u i   
 P  ij

 xi  x j
 Sui
Conservation of energy
 u j H    j   u    SE
 P  q 
 H   
t  t  xj  x j  x j i ij
xi  coord dir' ns H  total enthalpy
ui  velocities  ij  viscous stress tensor
P = static pressure q j  diffusive heat flux
 = density S = source terms
Governing Equations in Differential Form
Auxiliary quantities:
ui ui
Total Enthalpy H  h
2
  ui u j  2  ul
 ij        
 xj  xi  3  xl ij
Stress-Strain Relation
T  Yk
q j     k hk
n
 xj k  xj
Heat Flux
h  static enthalpy k  species diffusion coefficient

  dynamic viscosity hk  species static enthalpy
 = thermal conductivity Yk  species mass fraction

Governing Equations
Mathematical characteristics of the equations is based on

the differential form
• Elliptic
• Parabolic
• Hyperbolic
General solutions are available for only a few simple cases,
with simple boundary conditions
• example:
–Flow through circular pipe
–linear heat conduction equation on a rectangle, with temperature specified boundary
conditions
Nature of the Equations
Equations of fluid flow are complex and non-

linear
•solving not just one equation, but a set of coupled (or related)
equations
•equations are very non-linear
•simple, analytic methods are not available
•must resort to approximate discrete methods such as the finite
element, finite volume or finite difference methods

Linear and non-linear equations
Linear equations are relatively easy to solve

•solutions are usually guaranteed
•examples are simple (constant thermal conductivity) heat
conduction, potential flow equation, and elastic stress analysis
Example: 2D Heat Conduction
 q  0 T=50

q = -kT

2T  0
for k = constant
T=100 T=0
2T  0
 2T  2T
or  0
 x 2  y2 q=0
Non-linear equations
Non-linearity means that

• coefficients depend on the solution, therefore:
• iteration is required, to update the non-linearities
• direct solutions of the discrete equations is not possible, since general, direct
algebraic solutions methods are not available
Procedure for solving non-linear equations
• linearize the equations
• start with an initial guess
• solve the algebraic equations
• update the non-linearities
• repeat until convergence

Boundary conditions
Boundary conditions are needed to completely define

the problem
Boundary conditions must be realistic and consistent
Examples include:
• fluid has zero velocity relative to a wall, at the wall
• inlet total pressure or velocity
• exit static pressure, or mass flow
Usually specified explicitly, through the GUI, in a CFD
code
Discrete methods
Basic discretization methods

–finite element
–finite difference
–finite volume
–“hybrid” (some combination of the above)
•even within each category, differences can be large
•finite volume or hybrid finite volume methods most popular
because of an attractive feature: conservation
•classical finite difference and finite element methods seldom
used in CFD, since they are non-conservative

Finite Volume Method
•Identify the region of interest

•Divide domain into discrete regions (control volumes)
•Apply conservation principles to each control volume:
–integrate equations over control volume
–convert volume integrals to surface integrals using Gauss divergence theorem
–approximate all volume and surface flux terms
–linearize, if necessary
–assemble and apply boundary conditions
–solve the resulting set of algebraic equations
–if non-linear, iterate until convergence
Heat conduction example
1. Identify the region of 2. Divide into discrete

interest regions
T=50
T=100 2T  0 T=0
q=0
region is identified and region is divided into 12 control

boundary conditions are volumes
shown

3. Integrate the differential 4. Convert volume integrals

equations over each to surface integrals
control volume using Gauss divergence
T T theorem
  q = 0, q  ( , )
 
x y
control
volume
   qd  0  q  dn = 0
  
cv cs
n = unit surface normal vector

q = surface heat flux vector


5a. Approximate all volume and 5b. Approximate all volume and
surface flux terms surface flux terms
k n
 q  dn =   qn  n 0    Toutside  Tinside   0
N
  N
cs n 1 i  1  s i
control control
volume volume
q is the representative k
where q  - (T+  T- )
heat flux on each face s
N = number of faces
n = surface area

6. Simplify by defining 7. Assembled algebraic equation

coefficients a p Tp  a e TE  a w TW  a n TN  a s TS  b
  a i  Toutside  Tinside   0
N N
n
i 1
Pe E
kn 
W w
where a i   
s
 s  S
a p  a e  a w  a n  as
b = source term (neglected earlier)
8. Final Details
Apply boundary conditions
• specify temperature where known
• set heat flux where known j
i
Assemble set of algebraic equations
in this case, a 4 x 3 matrix (12 equations), one for each control

volume
a ijP Tij  a ije Ti 1j  a wij Ti 1j  a nij Tij1  a sij Tij1  b ij

9. Solve the equations

• for small set, direct solutions
• for CFD, iterative solution
The way in which the equations are solved is crucial.

• affects efficiency, robustness and viability
• should not affect accuracy, but often does, when iterative methods cannot
reach convergence, and hence solution is incorrect
Hybrid Method
Earlier sections established that:

• the governing, conservation equations for fluid flow and heat transfer are
complex, non-linear p.d.e’s
• we have no hope of analytical solutions (in general), therefore must resort to
numerical methods
The next step:

• apply the finite volume numerical methods that were discussed earlier
• now more complex than previous simple case

Element Based Finite Volume Method
• Advantages to the method are:

–flexibility and versatility of the finite element method
–fully conservative because implemented within a finite volume framework
• Some details
–we have both “elements” and “volumes”
–elements just a convenient building block for constructing volumes and
governing equations
–many differences from standard finite element methods
Finite Volume/ Hybrid Method
Divide domain into volumes
Form exact integral conservation

equation for each volume
Approximate algebraic conservation

equation for each volume
Iteration
Required
Solve set of algebraic equations
Output

Domain Discretization
Fill domain with

elements (triangle,
quad, hex, wedge, tet,
pyramid etc.)
Nodes at corners of
elements
Control volumes
formed by sub-dividing
elements
Element
Nodes
Hexahedral Tetrahedral Wedge (Prism) Pyramid

Building the Volumes
control volume is formed

by adding together all
quadrants adjacent to any
given node
conservation equations are
integrated over the control
volume (not the element)

Integrating the Equations
conservation equations
contain some divergence
terms: the convection and
diffusion
use Gauss divergence
theorem to convert volume
integrals to surface integrals
for these terms only

   j d     j dn j
 xj S Control Surface, S
1st Level Discrete Approximations
1st level approximation: surface

integrals are evaluated at discrete
locations only: the “integration
points’’
X X
n
X
X
X
X
X
X
  j dn j 
ip
  j n j ip
S
Integration points (ip)

1st Level Discrete Equations
    0 
Transient  
 t 
 ip v ip  nip  ip

n X X
Advection n
ip1
X
X
 q  nip  dsip
n
X
X
Diffusion X
X
ip
ip1
Sources S
Integration points (ip)
Advection- 2nd Level Approximations
Upstream differencing:  ip   UDS

• physical
• stability
Along streamline:  ip   SUD
• results become grid orientation X X
insensitive
• captures sharp transverse
X
X
solution gradients
• accuracy
X
X
Correction: X
X
 ip   SUD   physical ip
• diffusion
X
• source terms  UDS

 SUD
X
• improves resolution of
streamwise gradients
• second order accuracy
Accurate Gradient Resolution
X X
X X X
X X
X X X
Element based
X
X
X
X X X
X
X
Standard
More accurate gradient prediction due to

more integration points per control volume
face
Solution of the Discrete Equations
Nature of the discrete equations

• result of the discretization process is a set of algebraic equations that must be
solved
–one equation for each unknown (u,v,w,P,H,k,e) at each node: the equation set
is large
–equations are coupled: more than one variable appears in some of the
equations, thereby inter-connecting the equations (as we know they should
be, from a physical point of view)
–the equations are linearized: the non-linearity is embedded within the
coefficients: the coefficients depend on the solution
–therefore, repeated solutions using updated coefficients is required

Nature of the discrete equations (cont’d)

• local: the equations are strongly coupled
–suggests using a “coupled” approach, where equations are solved together,
instead of a “segregated” approach, where each set of equations (say the u-
momentum equation set, etc.) are solved separately
• global: propagation of long wave-length information
–boundary conditions must fully communicate with each other
• for example for total pressure at inlet, static pressure at exit, the two must
communicate to establish the overall flow
–algebraic multi-grid methods provide this facility
use a coupled algebraic multi-grid (AMG) solver!
Importance of the Solution Method

• equation set is large, therefore must use iterative methods
–direct methods would result in enormous solution times, for practical cases
• “direct solution” at every iteration not required anyway
–coefficients are a function of the solution, so why solve, the inexact equations
precisely?
• solution method would be a non-issue if direct solver could be used
• in practice solver is a big issue: getting fast, accurate, dependable linear
solutions is essential

Final points
•AMG solver gives scaleable performance
–solution time is linearly proportional to number of nodes
• double the grid, solution time doubles
–other solvers do not have this desirable property (some give quadratic
performance
• doubling the number of nodes increases solution time by a factor of four!)
•handles “stiff” equation sets
–an internal “adaptive” algorithm addresses “stiffness” issues, caused by grid
aspect ratio, for instance
Basic AMG Concepts
Example coarsening of a 2D grid:
Fine grid
1st coarse grid
2nd coarse grid
3rd coarse grid

Validation - 1
Grid Refinement
0.00
Curved duct with incompressible, -1.00

turbulent flow. Demonstrates
25K Nodes
100K Nodes
constant (linear) performance -2.00

400K Nodes
with grid refinement.
log 10 (RMS Residual)

-3.00
-4.00
-5.00
-6.00
0 10 20 30 40 50 60 70 80 90 100
Work Unit
Validation - 2
Sensitivity to Grid Aspect Ratio 0.00
Curved duct solved with

average aspect ratio of
-1.00
3, 30 and 300.
-2.00
log 10 (RMS Residual)
-3.00
Comparison of algebraic
multigrid (anisotropic coarsening)
-4.00
with geometric multigrid

3 GMG
30 GMG
-5.00 300 GMG
(isotropic coarsening)
3 AMG
30 AMG
300 AMG
-6.00
0 20 40 60 80 100 120
Work Unit

Step I: Pre-processing step - define the problem - CFX-Build

• Define the geometry
• Define the domain(s)
• Define boundary and initial conditions
• Define the mesh
• Define solver parameters
Step II: Solution step - CFX-Solve
• solve the governing equations
Step III: Post-processing step CFX-Post
• analyse results

Step I: Pre-processing
Define the geometry

• Domain in which the governing
equations will be solved and
solution obtained
• Create a B-rep Solid
Define the fluid domain.
• Create fluid regions (fluid, solid,
conducting solid, porous media)
• Assign fluid properties
(viscosity, thermal conductivity,
specific heat, etc…)
• Select the physical models
(turbulence models,
compressibility, buoyancy, two-
phase flow, etc…)
Define the boundary conditions

• Needed on all external surfaces of geometry

Grid Generation
• Process of generating finite volumes or cells
• Tetrahedral/Pyramidal/Prismatic/Hexahedral cells created
• Surface mesh and volume mesh make up the grid
Step II: Solving
Solve the governing equations

• Set the flow solver options
• Iteratively solve the governing
equations
• Obtain convergence

Step III: Post-processing
Qualitative
• Graphical format
• Pressure, Temperature distribution
• Velocity field
Step III: Post-processing

Quantitative
• Quantitative post-processing is the reduction of data to produce
performance coefficients
• Simplifies comparison of designs

• Application dependent
• Examples:
–1. Pressure drop in a duct
–2. Lift, drag coefficients of airfoil
–3. Head, Efficiency etc. in pump
–4. Torque, Efficiency etc. in a torque converter
–5. Average NOx emission at a furnace exit

CFD Modelling
Introduction to ANSYS Workbench
ANSYS Workbench
ANSYS Workbench is a project-management tool. It can be considered as the top-level

interface linking all ANSYS software tools.
Workbench handles the passing of data between ANSYS Geometry / Mesh / Solver / Post-
Processing tools.
This greatly helps project management. You do not need worry about the individual files
on disk (geometry, mesh etc). Graphically, you can see at-a-glance how a project has
been built.
Because Workbench can manage the
individual applications AND pass data
between them, it is easy to automatically
perform design studies (parametric
analyses) for design optimization.

Workbench Overview
“Analysis Systems” are ready-made

The options visible in the stencils that include all the
Toolbox show all the individual systems (applications)
products (systems) you have needed for common analyses (for
licenses for. example Geometry + Mesh + Solver
+ Post-Processor)
TIP: If this list appears empty,
you have a problem with your
licensing! “Component Systems” are the
individual building-blocks for each
stage of the analysis
“Design Exploration”
provides tools for
optimizing designs and
understanding the
parametric response.
Basic Workflow
Dragging an Analysis System onto

the Project Schematic lays out a
workflow, comprising all the steps
needed for a typical analysis.
Workflow is from top to bottom.

As each stage is complete, the icon
at the right-hand side changes

Alternative Workflow
However, an analysis could equally well be prepared by selecting the

individual Component Systems that are needed for this analysis, and then
linking them together with connectors.
TIP: There are two ways to create the connectors between the systems:
1) Use the mouse to draw a line (eg A2 to B2, B3 to C2 etc)
2) Or, simply drop the new system on the cell of the upstream one, and
the link will be generated automatically.
Cell States
As each stage in the model-build is completed, the state of the

cell changes.
Icon Meaning Status after creating
Up to Date Geometry in A2, not yet
opened mesh in A3
Refresh required. Upstream data has changed
Update required. Local data has changed
Unfulfilled. Upstream data does not exist
Attention Required
Solving
Update Failed
Update Interrupted Status after model has
Changes pending (was up-to-date, but upstream data has changed) converged, waiting for
post-processing

Sharing Data between Different Solvers
Workbench can be used to transfer data between solvers. In this 1-way FSI (fluid-
structure-interaction) example, we transfer the loads from a Fluent CFD simulation
over to a Mechanical system to perform a stress analysis
The square connector shows that The round connector shows that the
the geometry created in cell A2 CFD results are being transferred as a
(CFD model) is being shared with Setup (input) condition to be used for
cell B3 (FEA model). FEA stress analysis.
File Location on Disk
Should you need to identify the individual files on your disk for each stage of the project,
these can be found by enabling View > Files. The resulting table will cross-reference the
directory and filename with the project cells.
Filenames Directory

Use of Archive / Restore
The workbench project comprises many files and directories. If you need to either
archive the project, or bundle it to send to us for a Technical Support query, use the
‘Archive’ tool. This generates a single zipfile of the entire project.
When archiving, you can choose

whether to include the
computed result files or not
(omitting these may make it
small enough to send by email)
Working with Parameters / Refresh and Update

To make changes, you can manually open up a component cell (eg. geometry)
Or: most Workbench applications will let you specify key quantities as a parameter. A new object
‘Parameter Set’ appears on the Project schematic.
You then need to update your model. From Workbench you can choose to then update the entire
project, or just a single cell.
Refresh: Reads upstream data, but will not do any lengthy operation like solving or meshing.
Update: Performs both a Refresh, AND generates the new output
2] “Update Project” will then work through each
component in turn (geometry > mesh > solver ) to compute
the new design point.
3] However you may want to update an individual

component (eg to preview the new geometry before
proceeding). Right-click on a individual cell.
1] Clicking here will let you modify the parameters centrally,

without having to open the individual application.

Solution Strategies Through Workbench
Different models can be set-Up parametrically through Workbench
CFD Modelling
Introduction to SCDM

What is SpaceClaim Direct Modeler?
Sketching Modeling
• Industry leading direct modeling tool for rapid
concept design and geometry manipulation
• Analysis-focused tools to repair, prepare, and

optimize models
• Computer-Aided Design (CAD) – like approach to

create new models OR Import CAD models
without CAD connection
CAD
• Dead model parameterization Import &
– No need of native CAD data for parameterization Clean-up
• Freedom to explore solutions without relying on CAD team
– More flexibility to make unplanned and local changes
• No features + No constraints = No regeneration failures
• Short learning curve for engineers without CAD

background
Launching SpaceClaim Direct Modeler
Standalone Session From Geometry cell of any System
ANSYS SpaceClaim
Direct Modeler is
supported only on the
‘Windows’ platform
• ANSYS SCDM is launched within

Workbench
• Right click on “Geometry” and select “New
SpaceClaim Geometry”

SpaceClaim Interface
Quick Access Toolbar

Ribbon
Toolbar
Structure Panel
Options Panel
Properties Panel
Graphics
Window
Status Bar
Ribbon Toolbar
• Familiar ribbon UI design

Display Tab
• Tools categorized in a series
of Tabs
– File handling
– Designing Assembly
– Displaying Tab
– Repairing
– Etc.
Repair Tab
• Each Tab displays relevant
tools in organized sections
Prepare Tab

Structure Panel
SpaceClaim Direct Modeler is not a feature-based modeler

• Operations are not stored under Structure Tree
– This is characteristic of history-based or feature-based modelers
– Functionality similar to feature based modelers can be achieved using
geometry scripting with automated replay
• Structure Tree shows the objects/entities
– Bodies (solid, surface)
– Curves (sketch curves, 3D curves)
– Assembly constraints
– Origin
– Plane
Options Panel and Properties Panel
Options Panel
• Displays options for modifying functions of active SpaceClaim
tool
– E.g., Pull tool contains option for add material, subtract
material, create fillet, chamfer, etc.
Properties Panel
• Displays properties of selected entity in Graphics window or
Structure panel
• Modify property values
– Color

Tool Guide and Mini Toolbar
Mini
Toolbar
Mini Toolbar
• Contains frequently used options of active tool
Tool Guide
• Contains different options to change behavior of active tool
– E.g., Tool guide of “Pull” tool contains option for Revolve,
Sweep, Scale, etc.
Tool Guide
Status Bar
Quick Selection Selection

Measurement List options
Status Error/Warning View

Message Message controls
Status Message: Displays message and progress information about current tool
Quick Measurement: Displays simple measurement of selected entities (distance, angle)
Error/Warning Message: Displays error and warning messages
Selection List: Displays list of currently selected objects to the right and being hovered over at the left
Selection Options: Hosts various options for selection
View Controls: Controls for spin, pan, zoom, and switch to previous or next views

Selection
• You can select vertices (including centers of circles and ellipses, midpoints of lines, and points on splines),
edges, planes, axes, faces, surfaces, rounds, solids, and components.
• Most commonly used selection methods are:
– Click to select an object.
– Double-click to select an edge loop. (Double-click again to cycle through alternate loops.)
– Triple-click to select a solid.
– Drag to create a selection Box (can also use Lasso, Polygon, and Paint). If you draw the box from left to right, all
objects fully enclosed within the box will be selected. If you draw the box from right to left, all objects touching the
box will be selected.
– Press CTRL+A to select all similar objects, such as faces, edges, or points on the same solid or surface part.
– Hold CTRL and select to add or remove items from the selection. Ctrl with box-selection toggles the selection; Shift
with box-selection adds to the selection.
Selection Selection
List options
Selection
Selecting entities
• Select vertices (including centers of circles and
ellipses, midpoints of lines, and points on
splines), edges, planes, axes, faces, surfaces,
rounds, solids, and components.

Mouse and View Controls
• Easy-to-use Mouse and View controls to speed up operations

and manipulate graphics
• Controls listed in “Quick Reference Card”
• “Quick Reference Card” can be accessed during SCDM launch
Notice: The controls shown here are true after doing a ‘Reset
All’ in File/SpaceClaim Options/Navigation. They can be changed
to match those of your CAD Software by changing the Theme or
can be customized.
Display (1)
Display tab
• Hosts several features for managing display
Change Transparency
of geometry
– Cut, Copy, Paste objects
– Standard and custom view
– Change object color
– Manage layers
– Change graphics mode
• Wireframe, shaded etc.
– Rendering options
– Change transparency
– Arrange graphics windows
– Etc.
Opaque Transparent

Display (2)
Face hide and face visibility

• By right clicking on a face you can:
– Hide it
– Hide other faces
– Reverse its visibility
– Show all faces
File Handling
File Formats
SpaceClaim Options
• Supports import from major CAD
packages (CATIA, Pro/E, NX, Solid
Works, etc.)
- Separate license not required
except for CATIA V5 and V6, JT, and
PDF3D
• Neutral file formats like STEP and

Parasolid are also supported
• Additional options for controlling

import/export of file formats
available in SpaceClaim Options panel

SpaceClaim Demo
CFD Modelling
Introduction to ANSYS Meshing

What is ANSYS Meshing
ANSYS Meshing is a component of ANSYS Workbench

• Meshing platform
• Combines and builds on strengths of preprocessing offerings from ANSYS:
– ICEM CFD, TGRID (Fluent Meshing), CFX-Mesh, Gambit
Able to adapt and create Meshes for different Physics and Solvers
• CFD: Fluent, CFX and POLYFLOW
• Implicit, Explicit, Hydrodynamics, Rigid Dynamics ( Hydrodynamics for AQWA and
Rigid Dynamics for RBD)
• Electromagnetic
Integrates directly with other WB systems
Meshing Fundamentals
Purpose of the Mesh

• Equations are solved at cell/nodal locations
– Domain is required to be divided into
discrete cells (meshed)
Mesh Requirements
• Efficiency & Accuracy
– Refine (smaller cells) for high solution gradients and fine geometric detail.
– Coarse mesh (larger cells) elsewhere.
• Quality
– Solution accuracy & stability deteriorates as mesh cells deviate from ideal
shape

Launching ANSYS Meshing
ANSYS Meshing is launched within Workbench

– 2 ways:
From Analysis Systems From Component Systems
Fluid Flow (Fluent), Static Structural,… Mesh
Double click
Mesh in the
System
or right click and

select Edit
Graphical User Interface
Toolbars
Outline Graphics window
Worksheet
Details view
Message window Mesh Metrics
Section Planes
Entity Details Bar Units Bar

Outline
Three default sections

• Geometry
– Bodies
• Coordinate Systems
– Default global & user defined systems
• Mesh
– Meshing operations (controls & methods)
• displayed in the order in which they are inserted
In the tree
• Right clicking on any object
– launches a context sensitive menu
– Example: contains commands to generate, preview, clear mesh etc.
Details View
Accessing Object Details

• Select an object (in the Outline)
– Related information to that object are displayed in the Details View below
– Ex: Select a body (“Fluid”) in the Outline
• Details of “Fluid” : contains graphical and geometric details
– To access meshing details
• Click the Mesh object or any of the inserted objects
• The Details View provides options to

– review,
– edit, or values for every object in the Tree
– input

Which method to choose?
High aspect ratio cells Cells refined around

(Inflation) near wall to small geometric details
Why Multiple Methods? capture boundary layer and complex flow
gradients
• Choice depends on:
– Physics
– Geometry
– Resources
• Mesh could require just one or a

combination of methods
Hex (3d) or Quad (2d)

cells used to mesh
simple regions
Tet (3d) or Tri (2d) cells used
here to mesh complex region
Hexahedral versus Tetrahedral Elements (1)
• Advantages of hexahedral over tetrahedral:

– Less elements = Faster solution time with better
accuracy
• Naturally anisotropic: Fewer elements required as mesh is
aligned with the physics
– Fewer elements for given number of nodes
– 3 mostly parallel sets of faces (improves solution accuracy)
However, this assumes the geometry is such that the hex mesh is more
efficient and that the structured mesh aligns to the physics

Hexahedral versus Tetrahedral Elements (2)
• Advantages of tetrahedral over hexahedral:

– Easier to mesh more complex geometry:
• Mesh quality is often easier to achieve with tetrahedral
(or poly) mesh
• 12 tets can be converted to one hex
• Mesh transitioning with hex mesh can be problematic
Mesh refinement
need to the
continued through
all the model All Hex
Choosing the proper mesh element type will improve

the mesh generation efficiency
Generating the Mesh
Two options to generate the mesh

• Generate Mesh icon in toolbar
• Select Mesh in the Outline Tree, then the Update icon in the toolbar
Generate Mesh meshes the geometry using the current mesh settings.
Update meshes the geometry using the current mesh settings but also
first checks if the upstream geometry has changed and afterwards
updates the meshing cell status if it is attached to any downstream
cells.
If all you are trying to do is generate a new mesh, there is no difference

– pick either one
• Alternatively, right click on Mesh and use context menu

Named Selections (1)
Named Selections are groups of geometric or finite element entities:

• Named selections can be created either by selecting the desired items and clicking the “Named Selection”
icon in the context toolbar or RMB > Named Selection OR using the named selection worksheet (shown
later).
RMB
• Named selections must be composed of “like” entities (all surfaces or all edges, all nodes, etc.).
Section Planes (1)
Displays internal elements of the mesh

• Elements on either side of plane can be displayed
• Toggle between cut or whole elements display
• Elements on the plane
Edit Section Plane button can be used to drag section plane to new location
• Clicking on “Edit Section Plane” button will make section plane’s anchor to appear
Multiple section planes are allowed
For large meshes, it is advisable to switch to

geometry mode (click on geometry in the Tree
Outline), create the section plane and then go
back to mesh model

Section Planes (2)
• Shaded section planes

• Shaded or hollow section plane
• Plot by body color or same color for section plane
Mesh Statistics & Mesh Metrics
• Displays mesh information for Nodes and Elements

• List of quality criteria for the Mesh Metric
• Different physics and different solvers have different requirements for mesh quality
Mesh metrics available in Workbench Meshing include:
• Element Quality
• Aspect Ratio
• Jacobian Ratio (MAPDL, Corner Nodes, or Gauss Points)
• Warping Factor
• Parallel Deviation
• Maximum Corner Angle
• Skewness
For Multi-Body Parts, go to corresponding body in Tree Outline
• Orthogonal Quality
to get its separate mesh statistics per part/body
• Characteristic Length

Mesh Metric Graph
• Displays Mesh Metrics graph for the element

quality distribution
• Different element types are plotted with
different color bars
• Can be accessed through menu bar using
Metric Graph button
• Axis range can be adjusted using controls

button (details next slide)
• Click on bars to view corresponding elements

in the graphics window
– Use to help locate poor quality elements
Mesh Metric Graph Controls
• Elements on Y-Axis can be plotted with two

methods;
• Number of Elements
• Percentage of Volume/Area
• Options to change the range on either axis
• Specify which element types to include in graph
• Tet4 = 4 Node Linear Tetrahedron
• Hex8 = 8 Node Linear Hexahedron
• Wed6 = 6 Node Linear Wedge (Prism)
• Pyr5 = 5 Node Linear Pyramid
• Quad4 = 4 Node Linear Quadrilateral
• Tri3 = 3 Node Linear Triangle
• Te10, Hex20, Wed15, Pyr13, Quad8 & Tri6 non-linear
elements

Mesh Quality
Mesh quality recommendations

Low Orthogonal Quality or high skewness values are not recommended
Generally try to keep minimum orthogonal quality > 0.1, or maximum skewness < 0.95. However these values
may be different depending on the physics and the location of the cell
Fluent reports negative cell volumes if the mesh contains degenerate cells
Skewness mesh metrics spectrum
Orthogonal Quality mesh metrics spectrum
Meshing Demo

Summary
Basics concepts in CFD

Numerical model in CFD
Steps involved in a CFD simulation
Geometry & Meshing
THANK YOU!

Introduction To Computational Fluid Dynamics: Basics of CFD Numerical Model in CFD Steps in A CFD Simulation

Caricato da

Informazioni sul documento

Titolo originale

Copyright

Formati disponibili

Condividi questo documento

Condividi o incorpora il documento

Opzioni di condivisione

Hai trovato utile questo documento?

Questo contenuto è inappropriato?

Copyright:

Formati disponibili

Introduction To Computational Fluid Dynamics: Basics of CFD Numerical Model in CFD Steps in A CFD Simulation

Caricato da

Copyright:

Formati disponibili

Introduction to

Computational Fluid Dynamics

1 © 2018 ANSYS, Inc. November 26, 2018

2 © 2018 ANSYS, Inc. November 26, 2018

3 © 2018 ANSYS, Inc. November 26, 2018

Computational Fluid Dynamics (CFD) is the numerical

4 © 2018 ANSYS, Inc. November 26, 2018

Mid 1950’s - began to understand mathematics

5 © 2018 ANSYS, Inc. November 26, 2018

Power Generation Safety/Health

The most basic building block in CFD is the

what goes in what goes out

7 © 2018 ANSYS, Inc. November 26, 2018

Assembling the Blocks

 The conservation volumes are assembled together to fill

8 © 2018 ANSYS, Inc. November 26, 2018

Once the conservation volumes are assembled, a flow field

9 © 2018 ANSYS, Inc. November 26, 2018

The Final Solution

The final flow field may look like this.

10 © 2018 ANSYS, Inc. November 26, 2018

11 © 2018 ANSYS, Inc. November 26, 2018

CFD Basic Concepts

The Control Volume

12 © 2018 ANSYS, Inc. November 26, 2018

Control • control volume refers to a region in space

• could be a pipe, room, turbomachinery blade row, etc.

13 © 2018 ANSYS, Inc. November 26, 2018

14 © 2018 ANSYS, Inc. November 26, 2018

Applied for a finite control volume yields integral equations

Applied for a infinitesimal fluid element results in partial

15 © 2018 ANSYS, Inc. November 26, 2018

Example: Uniform, 1D, Steady Flow

16 © 2018 ANSYS, Inc. November 26, 2018

Momentum in Momentum Momentum

17 © 2018 ANSYS, Inc. November 26, 2018

Heat In Work out

Governing Equations in Differential Form

h  static enthalpy k  species diffusion coefficient

20 © 2018 ANSYS, Inc. November 26, 2018

Mathematical characteristics of the equations is based on

21 © 2018 ANSYS, Inc. November 26, 2018

Nature of the Equations

Equations of fluid flow are complex and non-

22 © 2018 ANSYS, Inc. November 26, 2018

Linear equations are relatively easy to solve

23 © 2018 ANSYS, Inc. November 26, 2018

Non-linearity means that

24 © 2018 ANSYS, Inc. November 26, 2018

Boundary conditions are needed to completely define

25 © 2018 ANSYS, Inc. November 26, 2018

Basic discretization methods

26 © 2018 ANSYS, Inc. November 26, 2018

•Identify the region of interest

27 © 2018 ANSYS, Inc. November 26, 2018

Heat conduction example

1. Identify the region of 2. Divide into discrete

T=100 2T  0 T=0

region is identified and region is divided into 12 control

28 © 2018 ANSYS, Inc. November 26, 2018

3. Integrate the differential 4. Convert volume integrals

q = surface heat flux vector

29 © 2018 ANSYS, Inc. November 26, 2018