The focus of this work was to develop and apply
high-fidelity numerical design optimization techniques
to the aerodynamic design of High Speed
Civil Transport (HSCT) configurations for the purpose
of substantially improving their aerodynamic performance,
in an effort to develop an economically
feasible concept. It was sponsored by the High Speed
Research (HSR) program in direct support of Goal 2,
Objective 6: "Reduce the travel time to the Far East
and Europe by 50% within 20 years,...." A direct
consequence of this effort was the rapid acceleration
of the development of high-fidelity aerodynamic
design optimization methods within NASA and
industry, applicable to the entire spectrum of flight
vehicles. Such methods represent the next generation
of design tools and hence directly support Goal 2,
Objective 8: "Provide next-generation design
tools...."
Single-point and multi-point aerodynamic-shape
optimization methods at Ames were developed and
demonstrated for the design of an advanced HSCT
referred to as the Technology Concept Airplane
(TCA). A large variety of software methods were
integrated to yield an efficient and productive design/
optimization process. This integrated set of tools is
referred to as the Aerodynamic Shape Optimization
(ASO) Library. The tools consisted of flow solvers,
grid-generation and perturbation tools, design
variable and geometric constraint implementation
tools, gradient computation methods, and numerical
optimization methods. Some of the tools were
commercially available, others were developed
in-house, and still others were modifications of
commercial software.
Both single and multiple grid-block Euler optimization
methods were developed and applied. The
single-block method was used solely for the cruise-point
design of wing-body-nacelle configurations,
and the multi-block method was used for the multi-point
designs of full configurations. Both optimization
methods were coupled to constrained and unconstrained
optimization algorithms, which employ
sequential quadratic programming methods. These
methods allow for a single objective function and
multiple linear and nonlinear constraints. The
objective function gradients are with respect to a
user-specified set of design variables. The gradients
are calculated using either finite differences or, more
efficiently, with the adjoint method. The adjoint
method results in over an order-of-magnitude reduction
in computational time relative to the use of finite
differences. Consequently, a much larger set of
design variables can be employed with the adjoint
approach. The single-block method was run exclusively
on the CRAY C-90; the multi-block method
was parallelized and run on a variety of platforms.
A number of analytical tools were used to
provide higher fidelity analysis of the optimized
configurations than provided by the design methods.
Intermediate and final configuration analyses were
carried out by use of the AIRPLANE Euler code and
two Navier-Stokes codes. AIRPLANE uses an unstructured
tetrahedral mesh and is therefore capable of
computations about arbitrarily complex configurations.
Navier-Stokes analyses were performed with
OVERFLOW and UPS. OVERFLOW was used to
analyze the wing/body/nacelle/diverter configurations,
and UPS was restricted to wing/body configurations.
The UPS code uses an upwind marching
scheme and was used to validate previous Ames HSR
designs; it was found to produce very accurate
solutions.
The TCA baseline configuration, which was the
subject of the optimization effort and is shown in the
figure, was developed by the Boeing Company in
support of the NASA HSR program, using linear-theory-
based methods and extensive multi-disciplinary
system analysis. The design was subject
to an extensive set of geometric constraints generated
as part of the conceptual design of the aircraft.
The configuration consisted of a wing, body, four
nacelles and boundary-layer diverters for the single
cruise-point (Mach 2.4) design. For the multi-point
design, the configuration was expanded to include
leading and trailing edge wing flaps for improving
transonic performance and a canard and empennage
for longitudinal trim. Multi-point optimization was
performed at three design points: supersonic cruise,
transonic cruise, and transonic acceleration, corresponding
to Mach 2.40, 0.90, and 1.10, respectively.
Cruise-point optimization using the single-block
approach was followed by two forms of multi-point
design: (1) sequential design (design at the cruise-flight
condition followed by flap and canard/tail
incidence angle optimization at the two transonic
conditions), and (2) multi-point design (simultaneous
design at the three flight conditions via a composite
objective function). The single-point approach
produced the bulk of the performance benefits,
lowering the weighted-composite required thrust by
4.28 counts after trimming the full configuration at
the three design points. (A 7.0 count drag reduction
was achieved at cruise for the vehicle without
trimming surfaces.) The sequential and multi-point
methods achieved 6.03 and 7.55 counts of thrust
reduction, respectively. These performance gains are
significant, since a single drag count reduction at
cruise reduces the takeoff weight of current designs
by approximately 8,000 pounds.
Point of Contact: S. Cliff
(650) 604-3907
scliff@mail.arc.nasa.gov
Back To Top
Previous Paper
Return to Global Civil Aviation
Next Paper 
