AIAA 2002-5642

TRANSONIC AXIAL-FLOW BLADE SHAPE OPTIMIZATION

USING EVOLUTIONARY ALGORITHM
AND THREE-DIMENSIONAL NAVIER-STOKES SOLVER
Akira Oyama*
NASA Glenn Research Center, Brook Park, OH 44135, USA
(440) 962-3148, [email protected]
and
Meng-Sing Liou=
NASA Glenn Research Center, MS 5-11, Brook Park, OH 44135, USA
(216) 433-5855, [email protected]
and
Shigeru Obayashi#
Tohoku University, Katahira 2-1-1, Sendai, Japan
+81-22-217-5265, [email protected]
ABSTRACT
A reliable and efficient aerodynamic design
optimization tool using evolutionary algorithm has
been developed for transonic compressor blades. A
real-coded adaptive-range genetic algorithm is used
to improve efficiency and robustness in design
optimization. To represent flow fields accurately and
produce reliable designs, three-dimensional NavierStokes computation is used for aerodynamic analysis.
To ensure feasibility of the present method,
aerodynamic redesign of NASA rotor67 is
demonstrated. Entropy production is considered as
the objective function to be minimized. The
computation is parallelized on the SGI ORIGIN2000
cluster at Institute of Fluid Science, Tohoku
University, by distributing flow analyses of design

candidates to 64 processing elements. The present

method successfully obtained a design that reduced
entropy production by more than 19% compared with
the rotor67 while satisfying constraints on the mass
flow rate and the pressure ratio. The use of the
present tool for turbomachinery blade design is
demonstrated to allow designers to achieve higher
aerodynamic efficiency, while shortening design
cycle and reducing design costs significantly.
INTRODUCTION
Aircraft industry is increasingly exposed to
considerable commercial competitions to reduce
operation costs and to increase safety. Key factors for
success in developing an aircraft are reduction in
cost, timeliness, and improvement in quality of the

NRC Research Associate, Turbomachinery and Propulsion System Division, Member AIAA. Located at Ohio
Aerospace Institute ICOMP, 22800 Cedar Point Rd., Cleveland, OH 44142, USA
=
Senior Scientist, Turbomachinery and Propulsion System Division, Associate Fellow AIAA
#
Associate Professor, Institute of Fluid Science, Associate Fellow AIAA
Copyright 2002 by the American Institute of Aeronautics and Astronautics, Inc. No copyright is asserted in the
United States under Title 17, U.S. Code. The U.S. Government has a royalty-free license to exercise all rights under
the copyright claimed herein for governmental purposes. All other rights are reserved by the copyright owner.

1
American Institute of Aeronautics and Astronautics

product. Among many components of an aircraft,

design of jet engines is critical because a small
improvement in efficiency can result in huge savings
for commercial airlines in yearly fuel costs.
Computational design tool offers a feasible approach
to solving a very complex nonlinear optimization
problem involving a multitude of design variables
and constraints in a systematic and efficient manner
that is impossible to do otherwise. Application of this
computational design optimization approach to
compressor and turbine blade designs can reduce
design cost, design cycle, as well as increase
efficiency of jet engines.
With advances in Computational Fluid
Dynamics (CFD) and computer hardware, CFD has
become an integral part of the blade design process.
CFD has been employed to cut aerodynamic design
cost and time scales by reducing the number of
required experiments. However, the current design
process is, by and large, still based on trial and error,
and the success of the final design depends on the
designers expertise and companys proprietary
database. CFD technology will be able to display its
ability to the full extent when it is coupled with a
numerical optimization method and when any human
interactions in the design procedure are minimized.
Numerical optimization methods have been
successfully used for a variety of design problems.
However, application to aerodynamic blade shape
optimization problem still remains as a formidable
challenge. First of all, flow field inside a transonic
compressor or turbine is highly three-dimensional
and extremely complex. Therefore three-dimensional
Navier-Stokes computations are essential for
aerodynamic blade shape optimization. The sectionby-section or quasi-3D technique is more efficient
computationally, but has limitations in capturing 3D
effects. On the other hand, design optimization based
on three-dimensional Navier-Stokes is computationintensive and currently still expensive.
Second reason is that aerodynamic design
optimization problem of a blade itself is very hard to
solve. Because aerodynamic performance of a
transonic blade is very sensitive to its shape, a blade
shape must be parameterized with a large number of
parameters to be optimized. In addition, objective
function landscape of an aerodynamic design
optimization problem is often multimodal and
nonlinear because the flow field is governed by a
system of nonlinear partial differential equations.
Finally, aerodynamic blade shape optimization
problem is usually subject to some required
constraints, such as mass flow rate, pressure ratio,
and others.
The gradient-based methods are a well-known
optimization algorithm in which the optimum is

probed by calculating the local gradient information.

These methods are efficient in searching an optimum,
especially when it is isolated. The optimum obtained
from these methods will be a global one, if the
objective and constraints are differentiable and
convex. Therefore, this approach has been widely
used for many design problems including
aerodynamic designs such as wing design1, scramjet
nozzle design2, supersonic wing-body design3, and
more complex aircraft configurations4,5. This
approach has been also applied to aerodynamic
design optimization problems of turbomachinery
such as vaneless diffuser for a centrifugal
compressor6 and compressor airfoils7. However,
distribution of an objective function of an
aerodynamic design problem is usually multimodal,
and thus, one could only hope for a local optimum
neighboring the initial design point by using the
gradient-based methods. To find a global optimum,
one must start the optimization process repeatedly
from a number of initial points and check for
consistency of the optima obtained. In this sense, the
gradient-based methods are neither efficient nor
robust.
Evolutionary Algorithms (EAs) are emergent
optimization algorithms mimicking mechanism of the
natural evolution, where a biological population
evolves over generations to adapt to an environment
by selection, recombination and mutation. When EAs
are applied to optimization problems, fitness,
individual and genes usually correspond to an
objective function value, a design candidate, and
design variables, respectively. One of the key
features of EAs is that they search simultaneously
from multiple points in the design space, instead of
moving from a single point like gradient-based
methods do. Furthermore, these methods work on
function evaluations alone and do not require
derivatives or gradients of the objective function.
These features lead to the advantages such as
robustness, suitability to parallel computing and
simplicity in coupling the CFD code with other
disciplines codes. Owing to these advantages over the
analytical methods, EAs have become increasingly
popular in a broad class of design problems (for
example, see [8,9]). EAs have been also successfully
applied to aerodynamic shape optimization problems
such as airfoil shape design10-12, Multi-element airfoil
shape design13, subsonic to supersonic wing shape
designs14-16, vaned diffuser design for centrifugal
compressor17, compressor airfoil design18 and turbine
airfoil design19.
The objective of the present study is to develop
a reliable and efficient design optimization tool for
transonic
compressor
blade
shape
design
optimization problems. A real-coded Adaptive-Range

2
American Institute of Aeronautics and Astronautics

Genetic Algorithm (real-coded ARGA) is used for

design optimization. To represent flow fields
accurately and produce reliable designs, threedimensional Navier-Stokes computation is used for
aerodynamic analysis. To reduce turn-around time,
the computation is parallelized on the SGI
ORIGIN2000 cluster at the Institute of Fluid Science,
Tohoku University of Japan, by distributing flow
analyses of design candidates to 64 processing
elements. The present method is applied to
aerodynamic redesign of NASA rotor67 20.
THREE-DIMENSIONAL NAVIER-STOKES
SOLVER FOR CASCADE FLOW
Flow
field
inside
high-speed
axial-flow
turbomachinery is highly three-dimensional and
involves significant viscous effects, such as
boundary-layer separations and shock wave/boundary
layer interactions. Therefore three-dimensional
Navier-Stokes computations are essential for blade
shape optimization because further improvement in
the aerodynamic performance requires detailed
knowledge of the flow structure such as secondary
flows and tip clearance flow.
In this study, the three-dimensional NavierStokes code TRAF3D21,22 is used for aerodynamic
analysis of blade designs. Capability of the present
code has been validated by comparing the computed
results to some experiments such as the Goldman
annular vane with and without end wall contouring,
the low speed Langston linear cascade21 as well as
the NASA rotor67 22.
The present code solves the three-dimensional
full Reynolds-averaged Navier-Stokes equations. The
present code uses a central-differencing scheme
including artificial dissipation terms introduced by
Jameson, Schmidt, and Turkel23 to maintain stability
and to prevent oscillations near shocks or stagnation
points. In order to minimize the amount of artificial
diffusion inside the shear layer, the eigenvalues
scaling of Martinelli24 and Swanson and Turkel25 are
used. The two-layer eddy-viscosity model of Baldwin
and Lomax is used for the turbulence closure. The
system of the differential equations is advanced in
time using an explicit four-stage Runge-Kutta
scheme. In order to accelerate convergence of
calculations, local time-stepping, implicit residual
smoothing26, and the Full Approximation Storage
(FAS) multigrid technique27 are used.
At the subsonic axial inlet, the flow angles, total
pressure and total enthalpy are specified according to
the theory of characteristics while the outgoing
Riemann invariant is taken from the interior. At the
subsonic axial outlet, the average value of the static
pressure at the hub is prescribed and the density and
components of velocity are extrapolated together

with the circumferential distribution of pressure. The

radial equilibrium equation is used to determine the
spanwise distribution of the static pressure. On
sidewalls, the momentum equation, the no-slip
condition, and the temperature condition are used to
compute pressure and density. For the calculations
presented in this paper, all the walls have been
assumed to be adiabatic. The periodicity from blade
passage to blade passage is imposed by setting
periodic phantom cell values. At the wake, where the
grid is not periodic, the phantom cells overlap the
real ones. Linear interpolations are then used to
compute the value of the dependent variables in
phantom cell.
The three-dimensional grids are obtained by
stacking two-dimensional grids generated on the
blade-to-blade surface. These two-dimensional grids
are of C-type and are elliptically generated, with
controlled grid spacing and orientation at the wall.
The problem of grid skewness due to high stagger or
large camber is addressed by allowing the grid to be
non-periodic on the wake28. By adding lines near the
wall, viscous grids are obtained from the inviscid
grids. The wall normal spacing scaled with the axial
chord is 10-4. In the spanwise direction a standard Htype structure has been adopted. Near the hub and tip
walls geometric stretching is used for a specified
number of grid points, after which the spanwise
spacing remains constant. The number of the grid
points is 201 chordwise x 53 tangential x 57
spanwise. Among the 201 chordwise grid points, 149
grid points are distributed along the blade shape. The
computational grid for NASA rotor67 is shown in
Fig. 1.
BLADE SHAPE PARAMETERIZATION
Here a rotor blade shape is represented by four blade
profiles, respectively at 0%, 31%, 62%, and 100%
spanwise stations (all spanwise locations discussed
here are measured from the hub) and linearly
interpolated. Each of these sectional profiles can be
uniquely defined by using a mean camber line and a
thickness distribution and they are parameterized by
the third-order B-Spline curves. Parameterization
using B-Spline curves is one of the most popular
approaches for airfoil designs.
When B-Spline curves are used for shape
parameterization, positions of control points of the BSpline curves are often considered as the design
parameters. Here, five control points are used for the
mean camber line as illustrated in Fig 2. For the
thickness distribution, two control points are added at
the leading edge and the trailing edge so that these
points represent leading edge and trailing edge radii,
respectively. Chordwise locations of the control
points at leading edge and trailing edge are frozen to

3
American Institute of Aeronautics and Astronautics

zero and one, respectively. As a result, 14 design

parameters are required to represent a sectional
shape. Each blade shape is then represented with 56
design parameters.
EVOLUTIONARY ALGORITHM
EAs mimic mechanism of natural evolution, where a
biological population evolves over generations to
adapt to an environment by selection according to
fitness, recombination and mutation of genes (Fig. 3).
In EAs, a design candidate, objective function values,
and design variables usually correspond to an
individual, fitness, and genes, respectively.
Starting with an initial population of design
candidates that is often generated by random
sampling from the design space, EAs select good
design candidates in terms of fitness, which is
assigned on the bias of their objective function
values. Typically, fitness of a design candidate is its
objective function value itself for a single objective
problem. Recombination is applied, where new
population is generated by exchanging features of the
selected designs with the intent of improving the
fitness of the next generation. Then, mutation is
applied to design parameters of the new population to
maintain diversity in the population.
One of the key features of EAs is that it searches
from multiple points in the design space in contrast to
the traditional methods that usually move from a
single design point. In addition, EAs use objective
function values alone to determine a search direction
and do not require gradients of the objective function
while the traditional methods use local gradient
information of an objective function. These features
also lead to advantages such as,
1) Robustness: Deterministic methods, such as the
gradient-based methods, typically start with a
single design point and use the local gradient
information to determine a search direction. As a
result, they generally lead to a local, not
necessarily a global optimum near the starting
point. In contrast to them, EAs determine their
search direction globally and probabilistically
but efficiently using their unique operators socalled recombination and mutation that give EAs
capability of finding global optimums.
Compared with other probabilistic methods such
as the simulated annealing method29 that is
similar to the gradient-based methods but tries a
random step according to the so-called
Boltzmann probability distribution, EAs are
more robust because they maintain a population
of design candidates and they dont use function
gradients that direct the search toward a local
optimum. In addition, EAs have a capability to
handle any design problems that may involve

non-differentiable objective function and/or a

mix of continuous, discrete, and integer design
parameters.
2) Suitability to parallel computing: Because EAs
are population-based search algorithms, all
design candidates in each generation can be
evaluated in parallel by using the simple masterslave concept. Parallel efficiency is extremely
high, if objective function evaluations consume
most of the computational time. Aerodynamic
design optimization is a typical case.
3) Simplicity in coupling evaluation codes: Because
EAs use only objective function values of design
candidates, EAs do not need substantial
modification or sophisticated interface to
evaluation codes. If an all-out re-coding were
required to every optimization problem,
extensive validation of the new code would be
necessary every time. EAs can save such
troubles.
4) Straightforward application to multiobjective
optimization problems: Because EAs maintain
multiple designs, EAs can find compromised
optimum designs, so-called Pareto-optimal
solutions, by introducing Pareto-optimal concept.
In the present study, the real-coded AdaptiveRange Genetic Algorithm30 (real-coded ARGA) is
used. The real-coded ARGA is an EA that can solve
large-scale design optimization problems very
efficiently by promoting the population toward
promising design regions during the optimization
process.
To represent design parameters of design
candidates, the floating-point representation31 is used
where an individual is characterized by a vector of
real numbers. It is natural to use the floating-point
representation for real parameter optimization
problems instead of binary representation, because it
is conceptually closest to the real design space, and
moreover, the string length is reduced to the number
of design variables.
The parental selection consists of the
stochastic universal sampling32 and the ranking
method31. To handle design constraints, the
constrained domination approach33 is used. Blended
crossover34 (BLX-0.5) is used for recombination.
Mutation takes place at a probability of 10% and then
adds a random disturbance to the corresponding gene.
The present EA adopts the elitist strategy35 where the
best and the second best individuals in each
generation are transferred into the next generation
without any recombination or mutation. Population
size is set to 64.
The main concern related to the use of a threedimensional Navier-Stokes solver for aerodynamic
shape design is the required computational effort.

4
American Institute of Aeronautics and Astronautics

Fortunately, powerful parallel computers are

increasingly made available in many institutions and
universities. In addition, EAs are intrinsically
amenable to parallel algorithms and the computation
can be easily parallelized. Furthermore, the PC
clusters are emerging as a powerful and affordable
alternative. Hence, the issue of computational cost is
rapidly diminishing and yet, the ability of applying
the EAs to complex problems is increasing. In the
present study, all computations are performed on the
SGI Origin2000 cluster consisting of 640 processing
elements located at the Institute of Fluid Science,
Tohoku University in Japan. The total scalar
performance and the total memory size are
384GLOPS and 640GB, respectively.
Here aerodynamic evaluation of design
candidates at each generation is parallelized using the
simple master-slave concept; the grid generations and
the flow calculations associated to the design
candidates of a generation are distributed into 64
processing elements of the SGI Origin2000 cluster.
This makes the corresponding turnaround time
almost 1/64 of that needed on a single processor
alone, because the computational time used for EA
operators are negligible compared with that of
Navier-Stokes computations.
DESIGN OPTIMIZATION PROBLEM
The optimization problem considered here is to seek
a redesign of NASA rotor6720. The rotor is a lowaspect-ratio transonic axial-flow fan rotor and is the
first-stage rotor of a two-stage fan. The rotor design
pressure ratio is 1.63 at a mass flow of 33.25 kg/sec.
The design rotational speed is 16043 rpm, which
yields a tip speed of 429 m/sec and an inlet tip
relative Mach number of 1.38. The rotor has 22
blades and aspect ratio of 1.56 (based on average
span/root axial chord). The rotor solidity varies from
3.11 at the hub to 1.29 at the tip. The inlet and exit
hub/tip radius ratios are 0.375 and 0.478,
respectively. Reynolds number is 1.797M based on
the blade axial chord at the hub.
The objective of aerodynamic rotor shape
design optimization problem is to minimize the flow
loss manifested via entropy generation. To achieve
this goal, the isentropic efficiency is often considered
as a design objective function to be maximized. From
our experience, however, a numerical design
optimization using the evolutionary algorithm
coupled with the three-dimensional Navier-Stokes
solver resulted in an optimum design that maximized
its isentropic efficiency by maximizing the total
pressure ratio rather than minimizing flow loss.
Therefore, mass-weighted sum of entropy production
from inlet to exit at the design point of rotor67 is
considered as the objective function to be minimized.

Because an optimized rotor design should meet

the required mass flow rate and pressure ratio, they
are maintained by specifying constraints on them:
massflowra tedesign massflowra terotor 67
massflowra terotor 67

pressureratiodesign pressureratiorotor67
pressureratiorotor67

0.005

(1)

0.01

(2)

These constraints are satisfied by using the

constrained domination approach33.
RESULTS
The first step of the EA is to properly define the
initial design space. The existing design (rotor67) is
used as a baseline around which the initial candidates
are populated. Specifically, the central values of the
initial design space are made to correspond to the
design parameter values representing the rotor67
geometry. These values are found by minimizing
geometry difference from rotor67 by using the EA
without any flow computation. Unbiased initial
population is generated by randomly spreading
solutions over the entire initial design space.
Population size and number of generation are 64 and
200, respectively. The computation is parallelized on
64 processing elements of the SGI Origin2000
cluster. The computational time is about 7 hours,
where most of the computational time is spent on grid
generation.
Optimization history in terms of objective
function value (entropy production) is shown in Fig.
4. Entropy production was reduced from the original
design by more than 19% after 100 generations.
Better designs may be obtained if the computation is
further continued. At each generation, 64 NavierStokes computations were performed in parallel
using 64 processing elements of the SGI
ORIGIN2000 cluster. Parallelization efficiency was
almost 1 because computational time necessary for
the real-coded ARGA is negligible. Each NavierStokes computation took about 16 hours of
computational time on one SGI ORIGIN2000
processing element while the real-coded ARGA used
less than one second per each generation. The total
turn around time was about 1550 hours (about two
months). Table 1 presents performance of the
optimum design and rotor67. The constraints on the
mass flow rate and pressure ratio are satisfied. The
isentropic efficiency is improved by 1.783%,
resulting in a higher pressure ratio across the rotor
than the baseline design. The blade profiles of the
optimized design and rotor67 are shown in Fig. 5.
Figure 6 compares spanwise entropy production

5
American Institute of Aeronautics and Astronautics

distributions of rotor67 and the optimum design. The

figure shows that the optimized design reduced
entropy production in the regions between the hub
and the midspan and near the tip.
Figures 7-10 compare blade profiles and surface
static pressure distributions at 10%, 33%, 67%, and
90% spanwise stations, respectively. Excessive flow
accelerations near the leading edge at 10% and 33%
spans are diminished by increasing the incidence
angles. In addition, at the 67% and 90% stations, the
shock on the suction side moves toward aft and
considerably weakens due to the aft movement of the
maximum camber position. Figures 11-14 present the
corresponding relative Mach number contours. The
optimized design avoids supersonic bubble on the
suction side near the leading edge at 10% span. The
supersonic bubble at 33% is also minimized. This
explains the reduction in entropy production between
the hub and the midspan. At 67% and 90% stations,
the bow shock impinging the blade suction side and
its reflection shock have become more oblique and
are significantly weakened to reduce entropy
production though the shocks. Also, flow separation
is decreased due to a weakened shock, thus
contributing to a reduction of entropy generation.
Figures 15 and 16 show the oil flow patterns
and static pressure contours of rotor67 and the
optimized design on pressure and suction surfaces,
respectively. The figures show that the shock wave
on the suction side of the optimized design is
weakened and more oblique than that of rotor67 in
the meridional plane as well as in the tangential plane
as shown in the Figs. 13 and 14, to reduce the shockgenerated entropy.
Figure 17 shows the performance maps of the
optimum design and rotor67. Although optimization
is carried out for the designed operating condition
(33.774kg/sec), it is remarkable that the optimized
design still maintains higher isentropic efficiency
over the entire range of operating conditions, from
the choke to stall limits.
CONCLUSIONS
A reliable and efficient aerodynamic design
optimization tool for transonic compressor blade has
been developed. The real-coded ARGA was used for
efficient and robust design optimization. To represent
flow fields accurately and produce reliable designs, a
three-dimensional Navier-Stokes solver was used for
aerodynamic analysis.
To ensure feasibility of the present method,
aerodynamic redesign of the NASA rotor67 was
demonstrated. Entropy production was considered as
the objective function to be minimized in order to
reduce flow losses at the rotor. The code was
parallelized and the computation was run on the SGI

ORIGIN2000 cluster at the Institute of Fluid Science

by using the simple master-slave concept. The total
turn around time was about 1550 hours. The present
method successfully obtained a design that reduced
entropy production by more than 19% compared with
the rotor67 while satisfying constraints on the mass
flow rate and the pressure ratio. This study
demonstrated that the present method offers a
promising approach to turbomachinery designer to
design a better machine, while shortening design
cycle and reducing design costs.
ACKNOWLEDGEMENTS
The authors thank Drs. L. M. Larosiliere and R. V.
Chima at NASA Glenn Research Center for their
valuable advices in defining the present design
optimization problem, and Dr. E. R. McFarland for
managing the project. The authors also thank Dr. M.
Marconcini at University of Florence for his help in
using the TRAF3D code. The generosity of
Advanced Fluid Information Research Center at
Tohoku University for providing computational time
on the SGI ORIGIN2000 cluster is greatly
appreciated.
REFERENCES
1) Hicks, R. M. and Henne, P. A., Wing Design by
Numerical Optimization, Journal of Aircraft,
Vol. 15, 1978, pp.407-412.
2) Baysal, O. and Eleshaky, M. E., Aerodynamic
Design Optimization Using Sensitivity Analysis
and Computational Fluid Dynamics, AIAA
Journal, Vol. 30, No. 3, 1992, pp. 718-725.
3) Reuther J. J. and Jameson, A., Supersonic wing
and wing-body shape optimization using an
adjoint formulation, Technical report, The
Forum on CFD for Design and Optimization,
(IMECE95),
San
Francisco,
California,
November 1995.
4) Reuther, J. J., Jameson, A., Alonso, J. J.,
Rimlinger, M. J. and Saunders, D., Constrained
Multipoint Aerodynamic Shape Optimization
Using an Adjoint Formulation and Parallel
Computers, Part 1, Journal of Aircraft, Vol.36,
No. 1, January-February 1999, pp.51-60.
5) Reuther, J. J., Jameson, A., Alonso, J. J.,
Rimlinger, M. J. and Saunders, D., Constrained
Multipoint Aerodynamic Shape Optimization
Using an Adjoint Formulation and Parallel
Computers, Part 2, Journal of Aircraft, Vol.36,
No. 2, January-February, 1999, pp.61-74.
6) Lee, Y.-T., Luo, L., and Bein, T. W., Direct
Method for Optimization of a Centrifugal
Compressor Vaneless Diffuser, Journal of
Turbomachinery, Vol.123, January 2001, pp.7380

6
American Institute of Aeronautics and Astronautics

7) Koller, U, Monig, R., Kusters, B., and Schreiber,

H.-A., Development of Advanced Compressor
Airfoils for Heavy-Duty Gas Turbines Part I:
Design and Optimization, Journal of
Turbomachinery, Vol. 122, July 2000, pp.397405.
8) Miettinen, K., Makela, M. M., Neittaanmaki, P.
and Periaux, J. (eds.), Evolutionary Algorithms
in Engineering and Computer Science, John
Willey & Sons Ltd, Chichester, U.K., 1999,
Chaps.17-24.
9) Dasgupta, D. and Michalewicz, Z. (eds.),
Evolutionary Algorithms in Engineering
Applications,
Springer-Verlag,
Berlin,
Heidelberg, 1997.
10) Quagliarella, D. and Cioppa, A. D., Genetic
Algorithms applied to the Aerodynamic Design
of Transonic Airfoils, AIAA-94-1896-CP, June
1994.
11) Yamamoto, K. and Inoue, O., Applications of
Genetic Algorithm to Aerodynamic Shape
Optimization, AIAA Paper 95-1650-CP, A
collection of technical papers, 12th AIAA
Computational Fluid Dynamics Conference,
CP956, San Diego, CA, June 1995, pp. 43-51.
12) Poloni, C., Mosetti, G. and Contessi, S., Multi
Objective Optimization by GAs: Application to
System and Component Design, Proceedings of
the Third ECCOMAS Computational Fluid
Dynamics Conference, John Wiley & Sons, Ltd,
Chichester, U.K., 1996, pp. 258-264.
13) Cao, H. V. and Blom, G. A., NavierStokes/Genetic Optimization of Multi-Element
Airfoils, AIAA 96-2487, June 1996.
14) Obayashi, S. and Oyama, A., ThreeDimensional Aerodynamic Optimization with
Genetic Algorithms, Proceedings of the Third
ECCOMAS Computational Fluid Dynamics
Conference, John Wiley & Sons, Ltd,
Chichester, U.K., 1996, pp.420-424.
15) Oyama, A., Multidisciplinary Optimization of
Transonic Wing Design Based on Evolutionary
Algorithms Coupled with CFD solver, CDROM Proceedings of the ECCOMAS 2000,
2000.
16) Sasaki, D., Obayashi, S., Sawada, K. and
Himeno, R., Multiobjective Aerodynamic
Optimization of Supersonic Wings Using
Navier-Stokes
Equations,
CD-ROM
Proceedings of the ECCOMAS 2000, 2000.
17) Benini, E. and Tourlidakis, A., Design
Optimization of Vaned Diffusers for Centrifugal
Compressors Using Genetic Algorithms, AIAA
Paper 2001-2583, 2001.
18) Uelschen, M. and Lawerenz, M., Design of
Axial Compressor Airfoils With Artificial

19)

20)

21)

22)

23)

24)

25)

26)

27)

28)

29)

30)

31)

32)

Neural Networks and Genetic Algorithms,

AIAA Paper 2000-2546, 2001.
Akmandor, S. and Oksuz, O, Aerodynamic
Optimization of Turbomachinery Cascades
Using Euler/Boundary-Layer Coupled Genetic
Algorithms, AIAA Paper 2001-2577, 2001.
Walter, S. C., William, S., and Donald, C. U.,
Design and Performance of a 427-Meter-PerSecond-Tip-Speed Two-Stage Fan Having a 2.40
Pressure Ratio, NASA TP-1314, October 1978.
Arnone, A., Liou, M.-S., and Povinelli, L. A.,
Multigrid Calculation of Three-Dimensional
Viscous Cascade Flows, NASA TM-105257,
ICOMP-91-18, 1991.
Arnone, A., Viscous Analysis of ThreeDimensional Rotor Flow Using a Multigrid
Method, ASME Journal of Turbomachinery,
Vol. 116, July 1994, pp. 435-445.
Jameson, A., Schmidt, W., and Turkel, E.,
Numerical Solutions of the Euler Equations by
Finite Volume Methods Using Runge-Kutta
Time-Stepping Schemes, AIAA paper 81-1259,
1981.
Martinelli, L. and Jameson, A., Validation of a
Multigrid Method for the Reynolds Averaged
Equations, AIAA paper 88-0414, 1988.
Swanson, R. C., and Turkel, E., Artificial
Dissipation and Central Difference Schemes for
the Euler and Navier-Stokes Equations, AIAA
paper 87-1107, 1987.
Jameson, A., The Evolution of Computational
Methods in Aerodynamics, Journal of Appl.
Mech., Vol. 50, 1983.
Jameson, A., Transonic Flow Calculations,
MAE Report 1651, MAE Department, Princeton
University, July 1983.
Arnone, A., Liou, M.-S., and Povinelli, L. A.,
Transonic Cascade Flow Calculations Using
Non-Periodic C-Type Grids, Computational
Fluid Dynamics Symposium on Aeropropulsion,
NASA Lewis Research Center, Cleveland, April,
1990.
Aarts, E. and Korst, J., Simulated Annealing and
Boltzmann Machines, John Wiley & Sons,
Chichester, U.K., 1989.
Oyama, A., Obayashi, S., and Nakahashi, K.,
Wing Design Using Real-Coded Adaptive
Range Genetic Algorithm, 1999 IEEE
International Conference on Systems, Man and
Cybernetics, Tokyo, Japan, 1999.
Michalewicz, Z., Genetic Algorithms + Data
Structures = Evolution Programs, third revised
edition, Springer-Verlag, Berlin, 1996.
Baker, J. E., Reducing Bias and Inefficiency in
the Selection Algorithm, Proceedings of the
Second International Conference on Genetic

7
American Institute of Aeronautics and Astronautics

camber line
0.2

z/c

0.15
0.1

blade profile
0.25

0.05

control points
B-Spline curves

0
0

0.2

0.4

0.6

0.2

0.8

x/c
z/c

0.15
0.1

thickness distribution

0.06

0.05

0.05
0.04

z/c

Algorithms, Morgan Kaufmann Publishers, Inc.,

San Mateo, California, 1987, pp 14-21.
33) Deb, K., Pratap, A. and Moitra, S., Mechanical
Component Design for Multiple Objectives
Using Elitist Non-Dominated Sorting GA,
Lecture Notes in Computer Science 1917
Parallel Problem Solving from Nature PPSN
VI, edited by Schoenauer, M., Deb, K., Rudolph,
G., Yao, X., Lutton, E., Merelo, J. J., and
Schwefel, H.-P., Springer, Berlin, Germany,
2000, pp.859-868.
34) Eshelman, L. J. and Schaffer, J. D., Real-Coded
Genetic Algorithms and Interval Schemata,
Foundations of Genetic Algorithms.2, Morgan
Kaufmann Publishers, Inc., San Mateo,
California, 1993, pp 187-202.
35) De Jong, K. A., An Analysis of the Behavior of
a Class of Genetic Adaptive Systems, Doctoral
Dissertation, University of Michigan, Ann
Arbor, 1975.

0.03

0.2

0.4

0.6

0.8

x/c

0.02
control points
B-Spline curves

0.01
0
0

0.2

0.4

x/c

0.6

0.8

Figure 2. B-Spline curves for mean camber line and

thickness distribution and the resultant
blade profile.

initial population
evaluation of designs
Next
generation

selection according to fitness

recombination of selected designs

mutation applied to new designs

Figure 3. Flowchart of typical evolutionary
algorithms.

entropy production

0.01
rotor67
optimized designs

0.0095
0.009
0.0085
0.008

Figure 1. Computational grid over NASA rotor67.

0.0075
0.007

20

40

60

80

100

generation
Figure 4. Optimization history in terms of entropy
production.

8
American Institute of Aeronautics and Astronautics

0.16

1.2

0.14
0.12

static pressure

Table 1. Computed performance of rotor67 and the

optimum design.

y/c

0.1
0.08
0.06

mass flow isentropic

[kg/sec]
efficiency

pressure
ratio

entropy
production

33.774

0.91890

1.6758

0.0090467

Optimum 33.929

0.93528

1.6859 0.0073263

0.04
0.02

Rotor67

0.4

x/c

0.6

0.8

rotor67
optimized design
0

0.2

0.4

0.6

0.8

x/c

Figure 7. Comparison between the optimum design

and rotor67 at 10% span.

90% span
67%span

0.8

0.2

0.6
0.4

rotor67
optimized design

0.8

0.12

1.4

0.1

1.2

33%span

0.08

static pressure

0.6

y/c

0.06
0.04

10% span

0.4

1
0.8

0.02

0.6
0
-0.02
0

0.2
rotor67
optimized design

0
0

0.2

0.4

0.6

0.8

0.4
0.2

0.4

x/c

0.6

0.8

0.2

0.4

0.6

0.8

x/c

Figure 8. Comparison between the optimum design

and rotor67 at 33% span.

Figure 5. Blade profiles of the optimum design and

rotor67.

0.06

1.8

0.05

1.6

static pressure

0.04

y/c

0.03
0.02

0.1

rotor67
optimized design

1.4
1.2
1

0.01

0.8
0

rotor67
optimized design

0.08

rotor67
optimized design

0.6

-0.01
0

0.2

0.4

0.6

0.8

0.2

0.4

0.6

0.8

x/c

x/c

Figure 9. Comparison between the optimum design

and rotor67 at 67% span.

0.06
0.04

0.04

0.02

2.2

y/c

0.02
0.01

-0.02

rotor67
optimized design

0.03

static pressure

entropy production

rotor67
optimized design

rotor67
optimized design

0.2

0.4

0.6

0.8

span
Figure 6. Comparison of spanwise entropy
production distribution.

1.6
1.4
1.2
1

0
-0.01
0

1.8

rotor67
optimized design

0.8
0.6

0.2

0.4

0.6

0.8

x/c

0.2

0.4

0.6

0.8

x/c

Figure 10. Comparison between the optimum design

and rotor67 at 90% span.

9
American Institute of Aeronautics and Astronautics

Figure 11. Relative Mach number contours of the optimum design and rotor67 at 10% span.

Figure 12. Relative Mach number contours of the optimum design and rotor67 at 33% span.

Figure 13. Relative Mach number contours of the optimum design and rotor67 at 67% span.

Figure 14. Relative Mach number contours of the optimum design and rotor67 at 90% span.
10
American Institute of Aeronautics and Astronautics

rotor67

optimized design
Figure 15. Oil flow patterns and static pressure contours on pressure surfaces.

optimized design
rotor67
Figure 16. Oil flow patterns and static pressure contours on suction surfaces.
0.94

adiabatic efficiency

0.92
0.9
0.88
0.86
0.84
0.82
0.8
30

rotor67
optimized design
31

32

33

34

35

mass flow rate [kg/sec]

Figure 17. Performance map comparison between rotor67 and the optimum design.
11
American Institute of Aeronautics and Astronautics