Você está na página 1de 31

(Adv) Introduction to Computational Fluid

Dynamics

Section #1: Introduction

Reading Assignments:
Section #1 Chapter 1 in Anderson
Section #2 Chapter 2 in Anderson

Dr. Chris Roy
Associate Professor
Aerospace and Ocean Engineering Department
Virginia Tech
What is CFD?
CFD stands for Computational Fluid Dynamics
Computational: suggests numerical
methods, computers
Fluid: a gas or a liquid continuous
deformation by shear stress
Dynamics: motion

What this course does:
Provide a basic introduction to techniques
for numerically solving fluid flow problems
Explain the theory underlying many of the
options found in modern commercial CFD
codes
What this course does NOT do:
Focus on training you how to run specific
commercial CFD codes

Intro CFD
Outline
Motivation
History of CFD
Role of CFD in engineering
Limitations of CFD
CFD methodology
Governing equations
CFD sub-models
Some CFD applications

Motivation
Computational Fluid Dynamics (CFD) has enormous
potential to impact the design, analysis, and
optimization of engineering systems
Microelectromechanical systems (MEMS)
Cooling of electronic circuits
Industrial processing
Road vehicle aerodynamics
Contaminant transport
Aircraft aerodynamics/propulsion
Spacecraft aerodynamics/propulsion
CFD History: 1900-1950s
Up to 1900, aerodynamics / fluid dynamics problems
addressed with experiment and theory
Numerical methods used by just a few researchers
Had to be computed by hand!
1910: Richardson computed numerical solution to Laplaces
equation to predict the stresses in a masonry dam
1928: Courant, Friedrichs, and Lewy developed a stability
requirement for hyperbolic PDEs (CFL number)
1940s: von Neumann developed method for evaluating
stability of numerical methods for time-marching problems
1945: first programmable electronic computer (ENIAC)
CFD History: 1950s-1960s
1955: Allen and Southwell computed incompressible,
viscous flow over a circular cylinder by hand
Digital computers now become widely available
Developments in numerical solution algorithms
1950: Successive Over Relaxation (SOR) method
1955: Alternating Direction Implicit (ADI) method
First numerical methods books begin to appear
Richtmyer (1957) and Richtmyer and Morton (1967):
numerical methods for marching problems
Forsythe and Wasow (1960): elliptic problems
First CFD book by Patrick Roache (1972)
CFD History (1960s)
Hypersonic blunt-body problem
ICBMs (1950s)
Manned space program (1960s)
Aerodynamic theory couldnt predict shock shape or
stand-off distance (needed for aerodynamic forces)
Institute for Aeronautical Sciences (later AIAA) devoted
entire sessions of technical meetings to this problem
Breakthrough in 1966 by Moretti and Abbett
One of the first applications of CFD
Finite-difference numerical method
Time marching approach to steady-state
Now a homework problem in graduate CFD courses!
Growing Role of CFD
Pre-CFD
(-1960)
Experiment Theory
Cost of conducting a
given CFD calculation
is dropping (order of
magnitude /decade)
Speed of computers
doubles every 18 mos.
(Moore s law)
Cost of experiments is
generally increasing
However, there will
always be a role for
experiments!
Experiment
CFD
Theory
Post-CFD
(1960-1990)
Experiment
CFD
Theory
Now
(1990+)
Growing Role of CFD
There is an increased emphasis on CFD-based design
and analysis

Reduced design time and time to market
Improved optimization of designs
Potential to reduce costs compared to the traditional test-
fail-fix procedure
Only option when tests are not feasible or allowed
Limitations of CFD
CFD results can only be trusted after the numerical
errors are quantified and shown to be small
CFD cant reproduce physics not properly included in
the formulation or models (experiments do not suffer
from this problem!)
CFD has difficulties accounting for uncertainty
CFD solutions are generally deterministic
Sometimes there are inherent uncertainties: e.g.,
freestream conditions for aircraft (gusts, clouds)
CFD is just beginning to incorporate uncertainty
Overview of CFD Methodology

(Ref: Oberkampf, DeLand, Rutherford, Diegert, and Alvin, 2002)


Governing Equations
Correlations/tables
Panel methods
Potential methods
Subsonic
Supersonic
Transonic
Euler equations
Coupled Euler/boundary layer (BL) approach
Navier-Stokes equations
Parabolized Navier-Stokes (PNS)
Viscous Shock Layer (VSL)
Full Navier-Stokes (N-S)
Governing Equations
The following governing equations form the
underlying basis or framework for CFD
Conservation of mass (continuity)
Conservation of momentum (F = ma)
Conservation of energy (1
st
law of thermo)
These equations generally take the form of nonlinear
partial differential equations (PDEs) or integral
equations
The form of the governing equations can be very
important in CFD

2D Euler Eqns.:
0
) ( ) (
=
c
c
+
c
c
+
c
c
y
v
x
u
t

( )
0
) ( ) (
2
=
c
c
+
c
+ c
+
c
c
y
uv
x
p u
t
u
( )
0
) ( ) (
2
=
c
+ c
+
c
c
+
c
c
y
p v
x
vu
t
v
( )
0
) ( ) (
=
c
+ c
+
c
+ c
+
c
c
y
pv ve
x
pu ue
t
e
t t t

, RT p =
2 1
1
2 2
v u
RT e
t
+
+

Auxiliary Eqns.:
Example: 2D Euler Equations


CFD Complexity
Issues affecting the complexity of CFD simulations

Temporal character: steady-state vs. unsteady
Spatial character: 1D, 2D, axisymmetric, or 3D
Geometric complexity, moving boundaries
Physico-chemical complexity
Radiation
Chemical reactions
Electrical/magnetic fields
Laminar vs. turbulent vs. transitional flow
Multi-phase flow
CFD Sub-Models

Molecular transport
Turbulent transport
Laminar to turbulence transition
Chemical reactions
Multiphase flow
Surface tension
Nonequilibrium thermodynamics
Rarefied flow
Turbulence

I am an old man now, and when I die and go to Heaven there are
two matters on which I hope for enlightenment. One is quantum
electrodynamics and the other is the turbulent motion of fluids.
And about the former I am really rather optimistic.
Horace Lamb (1849-1934)
Turbulence can be characterized by a cascade of vortical
structures (turbulent eddies) from large scales to small scales
Direct Numerical Simulation (DNS) all scales resolved
Large Eddy Simulation (LES) largest scales resolved, smaller
scales modeled
Reynolds-averaged N-S (RANS) all scales modeled
Hybrid RANS/LES
RANS in some regions (e.g., attached BLs)
LES in other regions (e.g., separated regions)

C. B. da Silva, G. Balarac, and O. Metais, Transition in high velocity ratio coaxial jets analysed from
direct numerical simulations, J ournal of Turbulence, Vol. 4, 2003.
Incompressible, Coaxial Jets
2D, Transitional, Subsonic Jets

Comparison of Transitional Free Jet and Wall Jet, Submitted by S. Gogineni, C. Shih, and A.
Krothapalli, Physics of Fluids, Vol. 5, No. 9, 1993.
Supersonic Turbulent Jet*
*Mach Waves Radiating from a Supersonic
Jet, N. T. Clemens and P. H. Paul, Physics
of Fluids, Vol. 5, No. 9, 1993.
Direct Numerical Simulation of Turbulent Channel Flow

Japan Aerospace Exploration Agency: http://www.ista.jaxa.jp/aet/cfd/0cfd-f04.html


Re = 5,600 Re = 41,400
Mach 2 Round Jet Flow

Center for Turbulence Research, Stanford, CA, J. Freund, P. Moin, and S. Lele: http://ctr.stanford.edu/
Examples of CFD Applications

Examples found at:
http://www.cfdreview.com/randpic_all.shtml
Streamlines Colored by Velocity

Web Site: http://www.ansys.com/magazine/issues/06-12-2008-ansys-advantage/01-sports.pdf


Pressure Contours
Fluent Analysis
of Speedo LZR
RACER Suit


Surface Shear Stress
CFD of Automotive
Flows from Corvid
Technologies
(Cadillac CTS-V)

Web Site: http://www.corvidtechnologies.com


CFD Simulations
of Oxygen Distribution
in the Human Lung

Web Site:
http://www.personal.psu.edu/rfk102/PUBS/BIOMED2003Paper.pdf
X
Y
Z
Ground Transportation System (GTS)
L = 2.48 m(97.5 in)
W = 0.324 m(12.75 in)
Re
W
= 2 million
M
inf
= 0.27
u
inf
= 91.6 m/s (205 mph)
Tractor/Trailer Aerodynamics

GTS in Ames 710 tunnel
(NASA/TM-2001-209621)
GTS Computational Mesh
(2.5 M Grid Points)
C. J. Roy, J. L. Payne, and M. A. McWherter-Payne, RANS Simulations of a Simplified Tractor/Trailer Geometry, ASME
J ournal of Fluids Engineering, Vol. 128, No. 5, 2006, pp. 1083-1089.
Tractor/Trailer Aerodynamics

Reynolds-Averaged Navier-Stokes
(Modeling of All Turbulent Scales)
Detached Eddy Simulation
(Modeling Largest Turbulent Scales)
H. A. Ghuge and C. J. Roy, Preliminary Detached Eddy Simulations of a Generic Tractor/Trailer, AIAA Paper 2006-3858,
San Francisco, CA, June 2006.
Tractor/Trailer Aerodynamics

Detached Eddy Simulation
Modeling Largest
Turbulent Scales
Z- (out-of-plane)
vorticity shown: measure
of rotation of the flow
Blue is clockwise
Red is counter-
clockwise
Animations are slowed
down by factor of ~100
x/W
C
p
0 2 4 6 8
-0.5
-0.4
-0.3
-0.2
-0.1
0
0.1
0.2
0.3
0.4
0.5
Experiment
Menter k-e
Spalart-Allmaras
Top of Tractor/Trailer
z/W = 0
Surface Pressure: Top
Tractor/Trailer Aerodynamics

Cp
y
/
W
-2 -1 0 1
0
0.25
0.5
0.75
1
1.25
1.5
Experiment
Menter k-e
Spalart-Allmaras
Front of Tractor
z/W = 0
Front Back
Surface Pressure
Tractor/Trailer Aerodynamics

C
D
Expt. 0.25
Menter 0.298
Spalart 0.413
DES 0.244
Drag Coefficient
C
p
y
/
W
-0.2 -0.15 -0.1 -0.05 0
0
0.2
0.4
0.6
0.8
1
1.2
1.4
1.6
Medium Mesh
Medium Mesh
Upper and Lower Limit
Experiment
RANS
Comparison of the pressure distribution along truck
base: z/W = 0