Introduction to Computational
Fluid Dynamics
Adapted from notes by:
Tao Xing and Fred Stern
The University of Iowa
Outline
What is CFD?
Why use CFD?
Where is CFD used?
Physics
Modeling
Numerics
CFD process
Resources
2
What is CFD?
What is CFD and its objective?
– Computational Fluid Dynamics
– Historically Analytical Fluid Dynamics (AFD) and EFD
(Experimental Fluid Dynamics) was used. CFD has become
feasible due to the advent of high speed digital computers.
– Computer simulation for prediction of fluid-flow phenomena.
– The objective of CFD is to model the continuous fluids with
Partial Differential Equations (PDEs) and discretize PDEs into
an algebra problem (Taylor series), solve it, validate it and
achieve simulation based design.
3
Why use CFD?
Why use CFD?
– Analysis and Design
Simulation-based design instead of “build & test”
– More cost effectively and more rapidly than with experiments
– CFD solution provides high-fidelity database for interrogation of
flow field
Simulation of physical fluid phenomena that are difficult to be
measured by experiments
– Scale simulations (e.g., full-scale ships, airplanes)
– Hazards (e.g., explosions, radiation, pollution)
– Physics (e.g., weather prediction, planetary boundary layer,
stellar evolution)
– Knowledge and exploration of flow physics
4
Where is CFD used? (Aerospace)
• Where is CFD used?
– Aerospace
– Appliances
– Automotive
F18 Store Separation
– Biomedical
– Chemical Processing
– HVAC&R
– Hydraulics
– Marine
– Oil & Gas
– Power Generation
– Sports Wing-Body Interaction Hypersonic Launch Vehicle
5
Where is CFD used? (Appliances)
• Where is CFD used?
– Aerospace
– Appliances
– Automotive
– Biomedical
– Chemical Processing
– HVAC&R
– Hydraulics
– Marine
Surface-heat-flux plots of the No-Frost
– Oil & Gas
refrigerator and freezer compartments helped
– Power Generation BOSCH-SIEMENS engineers to optimize the
location of air inlets.
– Sports
6
Where is CFD used? (Automotive)
• Where is CFD used?
– Aerospace
– Appliances
– Automotive
– Biomedical
– Chemical Processing External Aerodynamics Undercarriage
Aerodynamics
– HVAC&R
– Hydraulics
– Marine
– Oil & Gas
– Power Generation
– Sports
Interior Ventilation
Engine Cooling 7
Where is CFD used? (Biomedical)
• Where is CFD used?
– Aerospace
– Appliances
– Automotive
– Biomedical
Medtronic Blood Pump
– Chemical Processing
– HVAC&R
– Hydraulics
– Marine
– Oil & Gas
Temperature and natural
– Power Generation convection currents in the eye
following laser heating.
– Sports
Spinal Catheter 8
Where is CFD used? (Chemical Processing)
• Where is CFD used?
– Aerospace
– Appliances
– Automotive
– Biomedical Polymerization reactor vessel - prediction
of flow separation and residence time
effects.
– Chemical Processing
– HVAC&R
– Hydraulics
– Marine
– Oil & Gas
Twin-screw extruder
– Power Generation modeling
– Sports
Shear rate distribution in twin-
screw extruder simulation 9
Where is CFD used? (HVAC&R)
• Where is CFD used?
– Aerospace
– Appliances
– Automotive
Particle traces of copier VOC emissions
– Biomedical colored by concentration level fall
Streamlines for workstation behind the copier and then circulate
– Chemical Processing ventilation through the room before exiting the
exhaust.
– HVAC&R
– Hydraulics
– Marine
– Oil & Gas
– Power Generation
– Sports Flow pathlines colored by
pressure quantify head loss
Mean age of air contours indicate
location of fresh supply air
in ductwork 10
Where is CFD used? (Hydraulics)
• Where is CFD used?
– Aerospace
– Appliances
– Automotive
– Biomedical
– Chemical Processing
– HVAC&R
– Hydraulics
– Marine
– Oil & Gas
– Power Generation
– Sports
11
Where is CFD used? (Marine)
• Where is CFD used?
– Aerospace
– Appliances
– Automotive
– Biomedical
– Chemical Processing
– HVAC&R
– Hydraulics
– Marine
– Oil & Gas
– Power Generation
– Sports
12
Where is CFD used? (Oil & Gas)
• Where is CFD used?
– Aerospace
Volume fraction of gas
– Appliances
– Automotive
– Biomedical
Flow vectors and pressure Volume fraction of oil
– Chemical Processing distribution on an offshore oil rig
– HVAC&R
– Hydraulics
Volume fraction of water
– Marine
Analysis of multiphase separator
– Oil & Gas
– Power Generation
– Sports
Flow of lubricating
mud over drill bit 13
Where is CFD used? (Power Generation)
• Where is CFD used?
– Aerospace
– Appliances
– Automotive
– Biomedical
Flow in a burner
Flow around cooling towers
– Chemical Processing
– HVAC&R
– Hydraulics
– Marine
– Oil & Gas
– Power Generation
– Sports Pathlines from the inlet
Flow pattern through a water turbine.
colored by temperature
during standard operating
14
conditions
Where is CFD used? (Sports)
• Where is CFD used?
– Aerospace
– Appliances
– Automotive
– Biomedical
– Chemical Processing
– HVAC&R
– Hydraulics
– Marine
– Oil & Gas
– Power Generation
– Sports
15
Physics
CFD codes typically designed for representation of
specific flow phenomenon
– Viscous vs. inviscid (no viscous forces) (Re)
– Turbulent vs. laminar (Re)
– Incompressible vs. compressible (Ma)
– Single- vs. multi-phase (Ca)
– Thermal/density effects and energy equation (Pr, , Gr, Ec)
– Free-surface flow and surface tension (Fr, We)
– Chemical reactions, mass transfer
– etc…
16
Physics
Fluid Mechanics
Inviscid Viscous
Laminar Turbulence
Internal External
Compressibl Incompressible (airfoil, ship)
(water) (pipe,valve)
e
(air, acoustic)
Components of Fluid Mechanics
17
Governing Equations
(Equations based on “average” velocity)
ux u y uz 0
t x y z
Continuity
u u u u p
x u x x u y x u z x xx yx zx g x
t x y z x x y z
Equation of motion
18
Navier-Stokes Equations
Claude-Louis Navier George Gabriel Stokes
C.L. M. H. Navier, Memoire sur les Lois du Mouvements des Fluides, Mem. de l’Acad. d. Sci.,6, 398 (1822)
C.G. Stokes, On the Theories of the Internal Friction of Fluids in Motion, Trans. Cambridge Phys. Soc., 8, (1845)
Navier-Stokes Equations
(constant and )
u x u x u x u x p 2 ux 2 ux 2 ux
u x uy uz 2 2 2 g x
t x y z x x y z
yu u u u y p 2
u 2
u 2
uy
u x y
uy
y
uz 2 2 2 g y
y y
t x y z y x y z
u z u z u z u z p 2 u z 2 u z 2 u z
u x uy uz 2 2 2 g z
t x y z z x y z
D
v p 2v g
Dt 20
Navier–Stokes Example
u y u y u y u y p 2 u y 2 u y 2 u y
u x uy u z 2 2 2 g y
t x y z y x y z
dp d 2uy
0 2 g
dy dx
Integrate
duyx dp
g C1
Fluid
dx dy
x2 d p L
Integrate u y g C1 x C2
2 dy
B.C. u y 0 at x 0, u y 0 at x L
y
L dp
C1 g C2 0
2 dy
x
1 dp Laminar Flow
Final Expression u y g ( Lx - x 2
)
2 dy Static Parallel Plates
21
Modeling
Mathematical representation of the physical problem
– Some problems are exact (e.g., laminar pipe flow)
– Exact solutions only exist for some simple cases. In these cases nonlinear
terms can be dropped from the N-S equations which allow analytical solution.
– Most cases require models for flow behavior [e.g., K-, K-, Reynolds
Averaged Navier Stokes equations (RANS) or Large Eddy Simulation (LES)
for turbulent flow]
Initial —Boundary Value Problem (IBVP), include: governing Partial
Differential Equations (PDEs), Initial Conditions (ICs) and Boundary Conditions
(BCs)
22
Turbulent Flow Representation
(K- as an example)
u i u u' Where : u' deviating velocity, u constant net velocity
in the direction of flow, and u i instantaneous velocity
23
Turbulent Boundary Layer
y
Bulk Stream
x
U0 Edge of boundary layer
Outer layer
Fully turbulent layer
Sublayer + buffer layer
Wall
24
Wall Shear Stress Friction Velocity Viscous Length Scale
d U w
w
u
dy y 0 u
y u y y+ is similar to a local Reynolds number.
y Small y+ - Viscous effects dominate
Large y+ - Turbulence dominates
25
y+ and Turbulence Models
COMSOL has many turbulent models available
Low-Re models require a y+ resolution of < 1 to guarantee
accuracy
Low-Re models are necessary to accurately estimate skin
friction and flow separation
High-Re models use wall functions to approximate averaged
turbulent flow properties
Less accurate, but more computationally efficient
In COMSOL, a minimum y+ of 11.06 is enforced. To
maintain accuracy, ensure cells meet this requirement
26
Numerics / Discretization
Computational solution of the IBVP
Method dependent upon the model equations and
physics
Several components to formulation
– Discretization and linearization
– Assembly of system of algebraic equations
– Solve the system and get approximate solutions
27
Finite Differences
u u i 1, j u i , j 2u x 3 u x 2
2 3
x i, j x x i, j 2 x i, j 6
Finite difference Truncation error
representation
Methods of Solution
Direct methods Iterative methods
Cramer’s Rule, Gauss elimination Jacobi method, Gauss-Seidel
LU decomposition Method, SOR method
28
Numeric Solution
(Finite Differences)
u 2u x 2 3 u x 3
u i 1, j u i , j x 2 3
x i, j x i, j 2 x i, j 6
x
jmax
j+1
j
y
j-1 Taylor’s Series Expansion
u i,j = velocity of fluid
o i-1 i i+1 imax x
Discrete Grid Points 29
Finite Difference Truncation Error
f 2 f x 2 n f x n
f x x f ( x) x 2 n
x x i , j 2 x i , j n !
f ( x) sin 2 x
at : x 0.2 f ( x) 0.9511
f (0.22) ???? x 0.02
f
f x x f ( x) x
x
f (0.22) f (0.2) 2 cos[ 2 (0.2)](0.02) 0.9899
Exact solution for f (0.22) 0.9823 Error 0.775 percent
30
CFD process
Geometry description
Specification of flow conditions and properties
Selection of models
Specification of initial and boundary conditions
Grid generation and transformation
Specification of numerical parameters
Flow solution
Post processing: Analysis, and visualization
Uncertainty assessment
31
Geometry description
Typical approaches
– Make assumptions and
simplifications
– CAD/CAE integration
– Engineering drawings
– Coordinates include Cartesian
system (x,y,z), cylindrical system (r,
θ, z), and spherical system(r, θ, Φ)
32
Flow conditions and properties
Flow conditions and properties required are
unique for each flow code and application
– FlowLab requires all variables in dimensional
form
– Because of focused application, research codes
often use non-dimensional variables.
33
Selection of models for flow field
Direct Numerical Simulations (DNS) is to solve the N-S equations
directly without any modeling. Grid must be fine enough to
resolve all flow scales. Applied for laminar flow and rare be used
in turbulent flow.
Reynolds Averaged Navier-Stokes (NS) equations (RANS) is to
perform averaging of NS equations and establishing turbulent
models for the eddy viscosity. Too many averaging might
damping vortical structures in turbulent flows
Large Eddy Simulation (LES), Smagorinsky’ constant model and
dynamic model. Provide more instantaneous information than
RANS did. Instability in complex geometries
Detached Eddy Simulation (DES) is to use one single formulation
to combine the advantages of RANS and LES.
34
Initial and boundary conditions
For steady/unsteady flow
IC should not affect final solution, only convergence path, i.e.
iteration numbers needed to get the converged solution.
Robust codes should start most problems from very crude IC, .
But more reasonable guess can speed up the convergence.
Boundary conditions
– No-slip or slip-free on the wall, periodic, inlet (velocity
inlet, mass flow rate, constant pressure, etc.), outlet
(constant pressure, velocity convective, buffer zone,
zero-gradient), and non-reflecting (compressible flows,
such as acoustics), etc.
35
Grid generation
Grids can either be structured (hexahedral) or
unstructured (tetrahedral). Depends upon
type of discretization scheme and application
– Scheme
Finite differences: structured
Finite volume or finite element:
structured or unstructured
– Application
Thin boundary layers best resolved
with highly-stretched structured grids
Unstructured grids useful for complex
geometries
Unstructured grids permit automatic
adaptive refinement based on the
pressure gradient, or regions of
interest (FLUENT)
36
Grid Resolution
37
Grid generation and transformation
Grids designed to resolve important
flow features which are dependent
upon flow parameters (e.g., Re)
Commercial codes such as Gridgen,
Gambit
For research code, grid generated by
one of several methods (algebraic vs.
PDE based, conformal mapping)
For complex geometries, body-fitted
coordinate system will have to be
applied (next slide). Grid Sample grid established by
transformation from the physical Gambit of FLUENT
domain to the computational domain
will be necessary
38
Grid transformation
y
o x o
Physical domain Computational domain
Transformation between physical (x,y,z) f f f f f
x x
and computational () domains, x x x
important for body-fitted grids. The partial
f f f f f
derivatives at these two domains have the y y
y y y
relationship (2D as an example)
39
Numerical parameters & flow
solution
Numerical parameters are used to control flow solution.
– Under relaxation factor, tridiagonal or pentadiagonal solvers
– CFD Labs using FlowLab
Monitor residuals (change of results between iterations)
Number of iterations for steady flow or number of time steps for unsteady flow
Flow solution
– Solve the momentum, pressure Poisson equations and get flow field quantities, such as velocity, turbulence intensity,
pressure and integral quantities (drag forces)
40
Numerical parameters & flow
solution
Typical time
history of
residuals
The closer the
flow field to the
converged
solution, the
smaller the speed
of the residuals
decreasing.
Solution converged, residuals do
not change after more iterations
41
Post-processing
Analysis, and visualization
– Calculation of derived variables
Vorticity
Wall shear stress
– Calculation of integral parameters: forces, moments
– Visualization (usually with commercial software)
Simple X-Y plots
Simple 2D contours
3D contour carpet plots
Vector plots and streamlines (streamlines are the lines
whose tangent direction is the same as the velocity
vectors)
Animations (dozens of sample pictures in a series of
time were shown continuously)
42
Post-processing (Parallel Plates)
43
Post-Processing (example)
Pressure contour and
velocity vectors .
Note the locations of
the highest and lowest
pressure regions.
44
Uncertainty assessment
Rigorous methodology for uncertainty assessment using
statistical and engineering concepts
– Verification: process for assessing simulation numerical uncertainty
Iterative convergence: monitoring point & integral quantities should
change within the convergence criterions
Grid independent studies: 3-grids and Richardson Extrapolation
– Validation: process for assessing simulation modeling uncertainty by
using benchmark experimental data
Certification: full Verification and Validation done for a
certain range of geometries & parameters which are well
known and then extrapolated, qualitatively as well as
quantitative
– Simulating flows for which experiments are difficult (e.g., full-scale
Reynolds numbers, hypersonic flows, off-design conditions)
– Objective: Simulation-based design
45
CFD Example
Sulzer Chemtech
250 Y Plastic
Structured Packing
46
Geometry
• CT > STL > CFD
• CT = 0.322 mm
Min Resolution
• Copy/Pasted 2x
• Surface Wrapping
• Adaptive Meshing
• Tetrahedral Mesh
• Polyhedral Mesh
47
Mess Dimensions
48
Experiment vs. Simulation
49
Velocity Map
50
Software and resources
CFD software was built upon physics, modeling, numerics.
Two types of available software
– Commercial (e.g., FLUENT, CFX, Star-CCM, COMSOL)
– Research (e.g., CFDSHIP-IOWA, U2RANS)
More information on CFD can be got on the following website:
– CFD Online: http://www.cfd-online.com/
– CFD software
FLUENT: http://www.fluent.com/
COMSOL http://www.comsol.com/
CD-adapco: http://www.cd-adapco.com/
– Grid generation software
Gridgen: http://www.pointwise.com
GridPro: http://www.gridpro.com/
– Visualization software
Tecplot: http://www.amtec.com/
Fieldview: http://www.ilight.com/
51
