aerospace
Article
A Novel Direct Optimization Framework for Hypersonic
Waverider Inverse Design Methods
Jiwon Son 1 , Chankyu Son 2
Citation: Son, J.; Son, C.; Yee, K. A
Novel Direct Optimization
Framework for Hypersonic
Waverider Inverse Design Methods.
Aerospace 2022, 9, 348. https://
doi.org/10.3390/aerospace9070348
Academic Editor: Sergey Leonov
Received: 27 May 2022
Accepted: 25 June 2022
Published: 29 June 2022
Publisher’s Note: MDPI stays neutral
with regard to jurisdictional claims in
published maps and institutional affil-
iations.
Copyright: © 2022 by the authors.
Licensee MDPI, Basel, Switzerland.
This article is an open access article
distributed under the terms and
conditions of the Creative Commons
Attribution (CC BY) license (https://
creativecommons.org/licenses/by/
4.0/).
Aerospace 2022, 9, 348. https://doi.org/10.3390/aerospace9070348
and Kwanjung Yee 1, *
1 Department of Aerospace Engineering, Seoul National University, Seoul 08826, Korea; forscing@gmail.com
2 Department of Unmanned Aircraft Systems, Cheongju University, Cheongju 28503, Korea; cson@cju.ac.kr
* Correspondence: kjyee@snu.ac.kr; Tel.: +82-2-880-4151
Abstract: Waverider is a hypersonic vehicle that improves the lift-to-drag ratio using the shockwave
attached to the leading edge of the lifting surface. Owing to its superior aerodynamic performance, it
exhibits a viable external configuration in hypersonic flight conditions. Most of the existing studies
on waverider employ the inverse design method to generate vehicle configuration. However, the
waverider inverse design method exhibits two limitations; inaccurate definition of design space
and unfeasible performance estimation during the design process. To address these issues, a novel
framework to directly optimize the waverider is proposed in this paper. The osculating cone theory
is adopted as a waverider inverse design method. A general methodology to define the design
space is suggested by analyzing the design curves of the osculating cone theory. The performance
of the waverider is estimated accurately and rapidly via combining a high-fidelity computational
fluid dynamics solver and a surrogate model. A comparison study shows that the proposed direct
optimization framework enables a more accurate design space and efficient performance estimation.
The framework is applied to the multi-objective optimization problem, which maximizes internal
volume and minimizes aerodynamic drag. Finally, general characteristics for waverider are presented
by analyzing the optimized results with data mining methods such as K-means.
Keywords: waverider; hypersonic vehicle; osculating cone theory; inverse design; optimization
1. Introduction
Hypersonic vehicles, which have the advantage of prompt global reach, are being
actively researched as next-generation transport systems worldwide, especially in the
United States and China [1]. In a hypersonic environment, the vehicle experiences extreme
aerodynamic heating, dynamic instability, and a decrease in the lift-to-drag ratio (L/D) due
to the wave drag [2]. This decrease is a major factor that reduces the cruising efficiency of
the vehicle in terms of range and endurance. Since a hypersonic vehicle spends most of its
flight time in a cruising state, the overall performance may be critically reduced by its low
L/D. Therefore, it is essential to investigate the external configuration of the hypersonic
vehicle that significantly affects the L/D.
A waverider with a high L/D under the hypersonic condition was developed by
Nonweiler as a hypersonic vehicle concept in 1959 [3]. The waverider is a supersonic
lifting body characterized by an attached bow shockwave along its leading edge [4], which
eliminates the flow spillage from the lower surface to the upper surface. Consequently, high
pressure generated by the shockwave affects only the lower surface of the vehicle. Therefore,
the waverider exhibits a higher L/D than other lifting bodies. Owing to its superior
aerodynamic performance, the waverider is regarded as the best among hypersonic aircraft.
Along with the new hypersonic vehicle concept, the waverider suggested by Non-
weiler [3] introduced a design methodology to generate the waverider configuration.
Herein, the shape of the shockwave is determined by the vehicle configuration and flow
conditions. Therefore, to identify a vehicle configuration with an attached shockwave
https://www.mdpi.com/journal/aerospace
Aerospace 2022, 9, 348 2 of 22

along the leading edge, it is necessary to repeat the process of fabricating the shape of the
aircraft and verifying whether the generated shockwave is attached to the leading edge.
To address this issue, Nonweiler suggested the inverse design method for the waverider,
which comprises the following three steps:
(1) Generating the basic flow field
The first step in designing a waverider is to generate a flow field that occurs around a
vehicle before deciding the waverider configuration. The flow properties, including the
shape of the shockwave, are determined in this step;
(2) Defining the leading edge
In the second step, the leading-edge shape is defined to derive the waverider configura-
tion. Since the shockwave must be attached along the leading edge of the vehicle, the shape
of the leading edge is drawn on the shockwave plane. Additionally, the surface generated
by moving the leading-edge shape parallel to the freestream direction is determined as
the upper surface of the vehicle. This ensures that the shockwave is not generated on the
upper surface;
(3) Deriving the waverider configuration using the streamline tracing technique
In the final stage of the design, the lower surface of the vehicle is defined. Initially, the
leading edge is discretized. Considering that each discretized point is within the basic flow
field defined in step (1), the streamline is traced for each point. As the streamline must be
parallel to the wall because of the supersonic boundary condition, a shockwave identical
to that of the basic flow field is generated when these streamlines are combined and used
as the lower surface. Therefore, streamlines compose the lower surface of the waverider.
The designing of the waverider is completed when the rear of the vehicle is defined to be
perpendicular to the free flow.
The three-step inverse design method has been improved through several studies [5–8].
Jones developed a new waverider design method that improves volumetric efficiency by
employing a conical flowfield as the basic flowfield to a conical flow field [5]. Rasmussen
designed a waverider with the basic flowfield around the elliptic cone [6]. Sobieczky
developed the osculating cone theory to enable waverider design even if the basic flowfield
is not an actual flow field [7]. Osculating cone theory has paved the way for waverider
inverse design to have a higher degree of design freedom by greatly expanding the range
of the shockwave shape. Subsequently, Chen extended the osculating cone theory and
presented a methodology that can be designed for variable shock angles, and Zheng
presented a design methodology that can define the shape of the shockwave surface in
three dimensions [8,9]. With these research achievements, the inverse design method
became the standard for waverider design.
Nevertheless, the inverse design method exhibits two inherent problems [10,11],
namely the design space and performance estimation. The inverse design method begins
with the resultant physical phenomena, such as the shape of the shockwave. Thus, the
design method cannot derive the solution if an impractical shockwave shape is defined.
Therefore, the design space is defined by establishing whether the shape of the shockwave
is realistic. Since it is difficult to determine the reality in advance, the definition of the
design space is highly complicated.
Furthermore, the performance of the designed waverider is not directly estimated with
the inverse design method. For instance, in the case of a waverider, vehicle configuration
satisfies the design requirement of a leading edge attached shockwave. However, the
performance of the vehicle, such as lift or drag, is not estimated. Consequently, the inverse
design method cannot be employed directly on the optimization as design objectives and
constraints cannot be imposed.
The aforementioned problems result in certain limitations, such as using only a part
of the design freedom and incorrectly predicted performance, in the waverider design.
Moreover, the combination of the two problems prevents optimization, which restricts the
waverider from exhibiting its optimal performance.
Aerospace 2022, 9, 348 3 of 22

To address the design space and performance estimation problems of the inverse
design method, direct optimization of the waverider was suggested by Takashima [12],
wherein the osculating cone theory, developed by Sobieczky [7], was used to generate
the waverider configuration. For the design space problem, Takashima suggested an
osculating cone theory-based design space equation formulated by determining design
variables empirically, which generate an unrealistic shockwave shape. The performance
estimation problem was addressed by predicting the aerodynamic forces acting on the
vehicle using the tangent-cone method, which is a type of surface inclination method.
After addressing the two problems of the inverse design method, Takashima performed
optimization to identify the best performing waverider configuration. Thus, an optimal
design of the waverider that maximizes the L/D was obtained. However, the empirically
obtained design space equation is valid only for the limited design parameters used by
Takashima. Thus, designers must determine a new design space with each change in the
design conditions, such as the range of design variables and the shape of design parameters.
Moreover, using the tangent-cone method to predict the high-pressure flow characteristics
that occur in large, curved shapes is challenging.
Considering the design space problem, Kontogiannis et al. [13] suggested paramet-
ric waverider geometry models for the osculating cone theory, which can be applied to
waveriders with a higher degree of freedom (DOF). The investigations reported that the
design space of the waverider is constrained by the vertical position of the leading edge,
and the inverse design cannot be executed if the position of the leading edge is extremely
low. Although Kontogiannis et al. considered the effect of the leading edge on the design
space, the effect of the shape of the shockwave was excluded. As unrealistic shockwave
shapes can also cause design space problems, the shape of the shockwave needs to be
investigated to define an accurate design space.
Furthermore, Lobbia [14,15] developed an aerodynamics analysis tool to evaluate
the performance of the vehicle by calculating the aerodynamic forces acting on a vehicle
with high fidelity through the integration of modified Newtonian, tangent-cone, and
shock/expansion methods. Although the aerodynamic force was calculated accurately
unlike previous studies, the calculation time was not reduced and the design required
numerous computational resources.
In summary, the design space and performance estimation problems of the inverse
design method can be addressed by the direct optimization approach. Although efforts
have been made to improve this approach, such as the performance estimating technique
of Lobbia and design space analysis of Kontogiannis, they have certain limitations.
Therefore, we propose a new direct optimization framework for a waverider inverse
design. The direct optimization framework consists of two modules that deal with the
issues of inverse design and waverider configuration generating module. To address the
design space problem, a general methodology that can derive a design space for a waverider
based on the osculating cone theory was proposed. We used a discriminant formula to
determine whether the design parameter is realistic. The simultaneous consideration of the
effect of the leading edge and shockwave generated a more accurate design space equation.
Furthermore, the performance estimation problem is addressed using a high-fidelity flow
analysis solver based on the Euler equation that can calculate the aerodynamic forces
acting on a waverider. Additionally, a surrogate model evaluated the performance of the
designed waverider to decrease the required number of computational resources. The
direct optimization framework was completed by applying the above processes to the
modules. Finally, we performed optimization to identify the best performing waverider
configuration. The Pareto design solutions obtained by the completed design process
achieved a high internal volume with low drag. Moreover, the significant factors influencing
the waverider performance and general characteristics of configuration were identified
using data mining methods, such as K-means clustering, analysis of variance (ANOVA)
and decision tree analysis.
 Aerospace 2022, 9, x FOR PEER REVIEW 4 of 22

significant factors influencing the waverider performance and general characteristics of

Aerospace 2022, 9, 348 configuration were identified using data mining methods, such as K-means clustering,
4 of 22
analysis of variance (ANOVA) and decision tree analysis.

2. Methodology
2. Methodology
To overcome
To overcome thethe two
two limitations
limitations of
of the
the waverider
waverider inverse
inverse design,
design, aa design
design space-
space-
defining methodology and performance estimation technique are presentedin
defining methodology and performance estimation technique are presented inthis
thispaper.
paper.
Section2 2presents
Section presents a generalized
a generalized application
application process
process of theof the design
direct direct framework,
design framework,
wherein
awherein a design space-defining
design space-defining methodology
methodology and performance
and performance estimationestimation
technique technique are
are applied.
applied. Section 2.1 describes the implementation process of the direct design
Section 2.1 describes the implementation process of the direct design framework. The framework.
The theory
theory and validation
and validation for establishing
for establishing thespace-defining
the design design space-defining methodology
methodology are
are described
described
in in Section
Section 2.2. 2.2. The
The detailed detailed
methods usedmethods used in the estimation
in the performance performance estimation
technique are
technique are explained
explained in Section 2.3. in Section 2.3.

2.1. Direct
2.1. Direct Design
Design Approach
Approach
Figure 11 illustrates
Figure illustrates the
the three
three modules
modules inin the
the direct
direct optimization
optimization framework
framework for for the
the
waverider,namely
waverider, namelythe: the:(1)
(1) problem
problem definition;
definition; (2)
(2) waverider
waverider inverse
inverse design;
design; and
and (3)
(3) model
model
construction. The
construction. The design
design space
space for
for the
the waverider
waverider is is defined
defined inin the
the problem
problem definition
definition
module. The model construction module evaluates the performance of the waverider and
module. The model construction module evaluates the performance of the waverider
generates
and a surrogate
generates a surrogatemodel
model with
withthe
theperformance
performanceestimation
estimationtechnique.
technique. Waverider
Waverider
configuration is
configuration is generated
generated in in the
the waverider
waverider inverse
inverse design
design module.
module. TheThe osculating
osculating cone
cone
theory,which
theory, whichcan canensure
ensurea ahigh
highdesign
design DOF,
DOF, was
was employed
employed in this
in this study.
study. TheThe details
details of
of the
the osculating
osculating conecone theory
theory are described
are described in Appendix
in the the AppendixA. A.

Start

(1) Problem definition

Identify the optimum design
Optimal design
using the converged model
Objectives and constraints
(3) Model construction
Parameterization Construction of surrogate model Performance analysis

Design space analysis Check the criterion

Add sample points

Waverider geometry
(2) Waverider inverse design

Generate flowfield Define leading edge Streamline tracing

Figure1.1. Direct
Figure Direct optimization
optimization framework
framework for
for waverider.
waverider.

The direct
The direct design
design approach
approach begins
begins with
with the
the problem
problem definition
definition module.
module. Initially,
Initially, the
the
designobjectives
design objectivesand
and constraints
constraints areare formulated
formulated in terms
in terms of performance
of performance variables
variables to
to reflect
reflect the purpose of the design. Subsequently, design parameters for the waverider are
the purpose of the design. Subsequently, design parameters for the waverider are defined
by representing
defined the shape the
by representing of the shockwave
shape and the waverider
of the shockwave and thewith design curves,
waverider such
with design
as Bezier curves. Finally, the design space is derived using the proposed
curves, such as Bezier curves. Finally, the design space is derived using the proposed design space-
defining methodology, which
design space-defining involves which
methodology, deriving the discriminant
involves deriving the formula by analyzing
discriminant formula the
osculating cone theory and applying the discriminant formula to the
by analyzing the osculating cone theory and applying the discriminant formula to thedesign parameters.
The lattice
design space expressed
parameters. in space
The lattice the problem
expresseddefinition moduledefinition
in the problem of Figuremodule
1 represents the
of Figure
design space.
1 represents the design space.
The
The model
model construction
construction module
module begins after deriving
begins after deriving the
the waverider
waverider configurations.
configurations.
Computational fluid dynamics (CFD) is used to calculate the performance
Computational fluid dynamics (CFD) is used to calculate the performance (such as (such as the
the lift
lift
and drag) of the waverider. However, as CFD analysis requires extensive computational re-
sources, it cannot be used directly in the design process owing to the numerous estimations.
To address this problem, a surrogate model is employed to represent the performance of the
waverider. This surrogate model enables fast computation by expressing the relationship
between the result of the CFD and the design parameters as a numerical formula. The
Aerospace 2022, 9, 348 5 of 22

generation of the surrogate model requires design parameters and a performance estima-
tion dataset. The design of experiment (DOE) is performed to select the dataset efficiently.
Initially, the design parameters are selected, as represented by circles in the design space
(Figure 1). After selecting the design parameters, the waverider configuration of each
parameter is generated in the waverider inverse design module. Subsequently, the perfor-
mance of each configuration is estimated using the CFD analysis. The surrogate model
is constructed with the datasets that comprise these results and the design parameters.
The accuracy of the constructed model is enhanced by data addition, which is iteratively
conducted identically to the DOE. The model construction is terminated when the model
surpasses a certain accuracy level after repeated addition. The squares in the design space
(Figure 1) represent the added points.
The problem definition and model construction modules are followed by optimization.
In this study, a genetic algorithm is used as the optimization method [16], which applies
natural selection to optimize and robustly identify the non-linear maximum and minimum
values of the model. A rank-based fitness assignment technique developed by Fonseca was
used for selection in the genetic algorithm [17]. The fitness of each solution is estimated by
calculating the number of dominated solutions in this method.

2.2. Design Space Defining Methodology

This section explains the design space-defining methodology. The design parameter
of the inverse design is defined in terms of resultant physical phenomena, which is the
shape of the shockwave in the case of waveriders. Therefore, the result cannot be obtained
using the inverse design if an impractical design parameter is imposed. The design space of
the inverse design is the feasible region obtained considering a practical design parameter.
To derive this design space, we identified the design parameters that cause the failure
of the inverse design. Subsequently, the design space was expressed in the form of a
discriminant formula based on mathematical analysis, and the design space was compared
with previously reported design spaces in the literature for performance verification.
Initially, the undesignable design parameters that limit the design space should be
identified to derive the design space. The design parameters that cause the failure of the
inverse design were expressed through geometric relationships based on design curves
reported in the studies of Takashima and Kontogiannis [12,13]. The geometric relationships
can be classified as: (a) the upper surface curve located above the center of the osculating
cone, and (b) the osculating planes intersecting under the upper surface. Figure 2 illustrates
the two constraints expressed using design curves on the osculating plane and base plane.
Figure 2a depicts the relationship between the center of the local osculating cone and the
upper surface of the waverider on the osculating plane. The solid line P1 − P2 − P3 denotes
a feasible design case, whereas the dashed line P4 − P5 indicates an infeasible design case.
In the case of the feasible design, the upper surface curve P1 − P2 is below the center of
the local osculating cone and the lower surface curve P2 − P3 is formed in a downward
direction. Conversely, the upper surface curve P4 − P5 is located above the center of the
local osculating cone in the case of the infeasible design. Herein, a streamline that begins
from point P5 is generated in an upward direction, and it cannot encompass the lower
surface of the vehicle. Therefore, the upper surface curve must be located below the center
of the osculating cone. Figure 2b depicts the osculating planes, upper surface curve, and
shockwave curve on the base plane. The filled circle indicates the intersection point of the
two osculating planes, wherein each osculating plane has its flow field. If the intersection
point is below the upper surface curve, it exhibits the physical quantity of both osculating
planes simultaneously. As two physical quantities cannot concurrently exist at one point,
the waverider inverse design cannot derive the resultant configuration in this scenario.
Therefore, the osculating planes must not intersect below the upper surface curve.
 exhibits the physical quantity of both osculating planes simultaneously. As two physical
quantities cannot concurrently exist at one point, the waverider inverse design cannot
Aerospace 2022, 9, 348 derive the resultant configuration in this scenario. Therefore, the osculating planes6 must
of 22
not intersect below the upper surface curve.

(a) (b)
Figure 2. Geometric relationships in the osculating cone theory: (a) Geometric constraint based on the
Figure 2. Geometric relationships in the osculating cone theory: (a) Geometric constraint based on
position of the cone center; (b) Geometric constraint based on the intersected planes.
the position of the cone center; (b) Geometric constraint based on the intersected planes.

Weanalyzed
We analyzedthe theaforementioned
aforementionedtwo twogeometric
geometricrelationships
relationshipsby byrepresenting
representingthe the
shockwave curve as a two-dimensional (2D) coordinate system on the base
shockwave curve as a two-dimensional (2D) coordinate system on the base plane to derive plane to derive
thedesign
the designspace
spaceformula.
formula. As As illustrated
illustrated inin Figure 2b,xxdenotes
Figure 2b, denotesthethespanwise
spanwisedistance
distance
between the center axis of the waverider and the discretized point of the
between the center axis of the waverider and the discretized point of the shockwave. The shockwave. The
vertical position of the discretized point of the shockwave is expressed as a function
vertical position of the discretized point of the shockwave is expressed as a function f ( x ). 𝑓𝑓(𝑥𝑥).
Thedistance
The distancebetween
betweenthe theshockwave
shockwaveand andthetheupper
uppersurface
surfacewithin
withinthe
theosculating
osculatingplane
planeisis
denotedas
denoted ash.ℎ.To
Toavoid
avoidthethefirst
firstgeometric
geometricrelationship,
relationship,thethecenter
centerofofthe
thelocal
localosculating
osculating
conemust
cone mustbebeabove
abovethetheupper
uppersurface.
surface.ItItcan
canbebeexpressed
expressedusing
usingEquation
Equation(1) (1)considering
considering
the radius of curvature
the radius of curvature R. 𝑅𝑅.
  2  332
df
1 + dx𝑑𝑑𝑑𝑑�2 2
�1+�𝑑𝑑𝑑𝑑 �
R = 𝑅𝑅 = 2 𝑑𝑑2𝑓𝑓 >
>ℎ.h. (1)
(1)
d f
2
dx2𝑑𝑑𝑥𝑥
Forthe
For thesecond
secondgeometric
geometricrelationship,
relationship,the theosculating
osculatingplanes planesmustmustnot notintersect
intersectbelow
below
theupper
the uppersurface
surfacecurve.
curve.We Weidentified
identifiedthe theintersection
intersectionpoint pointofofthe thetwo
twoosculating
osculatingplanes
planes
byassuming
by assumingthatthatthe
theplanes
planescoexist.
coexist. Considering
Considering that that each
each osculating
osculating plane
plane does
doesnotnot
intersectwith
intersect withthe
theneighboring
neighboringplane planeunder underthe theupper
uppersurface,
surface,all allplanes
planesdo donot
notintersect
intersect
underthe
under theupper
uppersurface.
surface. Further,
Further, we weselected
selectedarbitrary
arbitraryneighboring
neighboring osculating
osculating planes;
planes;
theirshockwave
their shockwave points
points werewere represented
represented as as x𝑥𝑥11 and
and x𝑥𝑥2 2. . Finally,
Finally, thethe coordinate
coordinate x𝑥𝑥 ofofthe
the
intersectionpoint
intersection pointofofthe
thetwotwoosculating
osculatingplanesplaneswas wasderived,
derived,asaspresented
presentedininEquation
Equation(2).(2).
   
1 1 𝑥𝑥2 𝑥𝑥1
� −
 h 1i �𝑑𝑑𝑑𝑑h� 1i �𝑑𝑑𝑑𝑑 � 𝑥𝑥 = � x𝑑𝑑𝑑𝑑 − 𝑑𝑑𝑑𝑑 x1� + 𝑓𝑓(𝑥𝑥2 ) − 𝑓𝑓(𝑥𝑥1 ). (2)
� x =  h � 2i�
 df
−𝑑𝑑𝑑𝑑 𝑥𝑥2 𝑑𝑑𝑑𝑑
df 𝑥𝑥1  𝑑𝑑𝑑𝑑 𝑥𝑥2 − �𝑑𝑑𝑑𝑑
df
h �𝑥𝑥1i 
df
 + f ( x2 ) − f ( x1 ). (2)
dx dx dx dx
Equation (3) indicates
x2 the
x1 constraint formulated
x2 x1for the vertical coordinate of the
intersection point to be above the upper surface. To satisfy this condition, 𝑥𝑥2 was
Equation (3) indicates the constraint formulated for the vertical coordinate of the inter-
represented as 𝑥𝑥1 + 𝑑𝑑𝑑𝑑, and certain transformations were performed.
section point to be above the upper surface. To satisfy this condition, x2 was represented as
1
x1 + dx, and certain transformations were performed. 2 2 𝑑𝑑𝑑𝑑
�1+� � �
(𝑥𝑥1− 𝑥𝑥) h 𝑑𝑑𝑑𝑑i2  12 > ℎ.
𝑑𝑑𝑑𝑑 𝑥𝑥1 (3)
d�f𝑑𝑑𝑑𝑑�𝑥𝑥
1 + dx 1
x
The left-hand side of Equation
( x1 − x(3) h i 1 to >
) is identical that
h. of Equation (1), which implies
(3)
df
that the design space constraints described dx asx1 geometric relationships are redundant.
Therefore, the design constraints treated as separate in the literature must be unified. We
The left-hand side of Equation (3) is identical to that of Equation (1), which implies that
the design space constraints described as geometric relationships are redundant. Therefore,
the design constraints treated as separate in the literature must be unified. We used
Equation (1) that unifies the entire design constraints as a “discriminant formula” for the
osculating cone theory.
Aerospace 2022, 9, x FOR PEER REVIEW 7 of 22

Aerospace 2022, 9, 348 7 of 22

used Equation (1) that unifies the entire design constraints as a “discriminant formula”
for the osculating cone theory.
The
The design
design space
space can can bebe formulated
formulated by by applying
applying the the discriminant
discriminant formula
formula toto each
each
osculating
osculating plane. When Equation (1) is applied, the minimum radius of curvature of each
plane. When Equation (1) is applied, the minimum radius of curvature of each
osculating
osculating plane
plane point
point is is determined
determined as as aa function
function of of the
the shockwave.
shockwave. The The most
most restrictive
restrictive
condition
condition obtained
obtained by by the
the discriminant
discriminant formula formula represents
represents the
the required design space.
required design space.
We compared the design space-defining methodology
We compared the design space-defining methodology with the design space constraint with the design space
constraint
reported by reported
Takashima by Takashima
[12]. The design [12]. The design
curves andcurves and the
the design design parameters
parameters defined by
defined by Takashima are depicted on the left-hand side
Takashima are depicted on the left-hand side of Figure 3. The upper surface curve of Figure 3. The upper surface
was
curve was defined as a linear segment and a third- order
defined as a linear segment and a third- order Bezier curve, whereas the shockwave curveBezier curve, whereas the
shockwave curve was defined as a linear segment and
was defined as a linear segment and a fourth-order polynomial. Furthermore, xshock anda fourth-order polynomial.
Furthermore,
yshock denote the widthand
𝑥𝑥𝑠𝑠ℎ𝑜𝑜𝑜𝑜𝑜𝑜 and 𝑦𝑦height
𝑠𝑠ℎ𝑜𝑜𝑜𝑜𝑜𝑜 ofdenote the widthrespectively,
the shockwave, and heightand of both
the variables
shockwave, are
respectively,
normalized toand both length
the entire variables of thearevehicle.
normalized to the xentire
Furthermore, f lat length
indicates of
the the vehicle.
length of the
Furthermore,
linear segment𝑥𝑥𝑓𝑓𝑓𝑓𝑓𝑓𝑓𝑓 indicatestothe
normalized xshocklength
. of the linear segment normalized to 𝑥𝑥𝑠𝑠ℎ𝑜𝑜𝑜𝑜𝑜𝑜 .

leading edge shape symmetry plane

X3

X4

shock wave shape

yshock X2

xflat X1

xshock width

Reference model parameter Parameter of present study

Figure
Figure 3.
3. Waverider
Waverider design
design curves
curves and
and design
design parameters
parameters at
at the
the base
base plane.
plane.

In
In the
the study
study of of Takashima,
Takashima, the the design
design space space of of the
the waverider
waverider was was constrained
constrained by by
defining
defining the maximum maximumvalue valueofof yshock𝑦𝑦𝑠𝑠ℎ𝑜𝑜𝑜𝑜𝑜𝑜 as indicated
as indicated in Equation
in Equation (4). zHere,
(4). Here, 𝑧𝑧 denotes
waverider𝑤𝑤𝑤𝑤𝑤𝑤𝑤𝑤𝑤𝑤𝑤𝑤𝑤𝑤𝑤𝑤𝑤𝑤
denotes the normalized
the normalized waverider waverider length affecting
length affecting the distance the distance
betweenbetween the upper thesurface
upper surface
and the
and the shockwave of the waverider.
shockwave of the waverider. As x shock As 𝑥𝑥𝑠𝑠ℎ𝑜𝑜𝑜𝑜𝑜𝑜 increases
increases or z waverider or 𝑧𝑧𝑤𝑤𝑤𝑤𝑤𝑤𝑤𝑤𝑤𝑤𝑤𝑤𝑑𝑑𝑑𝑑𝑑𝑑 decreases, the
decreases, the feasible range of
feasible
the y shockrange of the 𝑦𝑦𝑠𝑠ℎ𝑜𝑜𝑜𝑜𝑜𝑜shock,max
between 0 and y between 0 and 𝑦𝑦𝑠𝑠ℎ𝑜𝑜𝑜𝑜𝑜𝑜,𝑚𝑚𝑚𝑚𝑚𝑚 , which represents the design space,
, which represents the design space, widens.
widens.
yshock,max = 0.75( xshock − 0.11) − 0.111(zwaverider − 0.31) + 0.0275. (4)
𝑦𝑦𝑠𝑠ℎ𝑜𝑜𝑜𝑜𝑜𝑜,𝑚𝑚𝑚𝑚𝑚𝑚 = 0.75(𝑥𝑥𝑠𝑠ℎ𝑜𝑜𝑜𝑜𝑜𝑜 − 0.11) − 0.111(𝑧𝑧𝑤𝑤𝑤𝑤𝑤𝑤𝑤𝑤𝑤𝑤𝑤𝑤𝑤𝑤𝑤𝑤𝑤𝑤 − 0.31) + 0.0275. (4)
In this study, we applied the discriminant formula [Equation (1)] to the design curves
In this study, we applied the discriminant formula [Equation (1)] to the design curves
and design parameters reported by Takashima, which are illustrated in Figure 3, to derive
and designspace.
the design parameters As thereported by Takashima,
linear segment which arehas
of the shockwave illustrated
an infiniteinradius
Figureof 3,curvature,
to derive
the design space. As the linear segment of the shockwave
the osculating planes that intersect the linear segment do not violate the constraint. In the has an infinite radius of
curvature, the osculating planes that intersect the
case of the fourth-order polynomial region, the radius of curvature of each osculating plane linear segment do not violate the
constraint.
point changes In the along casethe of spanwise
the fourth-order direction. polynomial
Since it isregion, difficult theto radius
compare of curvature
the radius of of
each
curvature and the distance between the upper surface and shockwave before difficult
osculating plane point changes along the spanwise direction. Since it is generating to
compare
the waverider the radius
configuration,of curvature the lowestand the distance
position of the between the upper
local osculating cone surface
amongand all
shockwave before generating the waverider configuration,
osculating planes was identified and compared with the highest position of the the lowest position of theupper
local
osculating cone amongthealldesign
surface. Consequently, osculating
space of planes
this study was was identified
derivedand comparedin with
as presented Equationthe
highest
(5), where position
β denotes of the theupper
shockwavesurface. Consequently,
angle of the osculating the design cone.space of this
Equation (5)study was
constrains
derived
the design as space
presented in Equation
by limiting the maximum (5), where value β of denotes the shockwave
yshock , similar to that inangleEquation of the (4).
osculating cone. Equation (5) constrains
Additionally, the effect of x f lat was added to Equation (5). the design space by limiting the maximum value
of 𝑦𝑦𝑠𝑠ℎ𝑜𝑜𝑜𝑜𝑜𝑜 , similar to that in Equation (4). Additionally, the effect of 𝑥𝑥𝑓𝑓𝑓𝑓𝑓𝑓𝑓𝑓 was added to
Equation (5).
 4
x 4 1 − x
7 shock f lat
4
yshock,max = 4
7 3𝑥𝑥𝑠𝑠ℎ𝑜𝑜𝑜𝑜𝑜𝑜 �1−𝑥𝑥𝑓𝑓𝑓𝑓𝑓𝑓𝑓𝑓 � . (5)
64
𝑦𝑦𝑠𝑠ℎ𝑜𝑜𝑜𝑜𝑜𝑜,𝑚𝑚𝑚𝑚𝑚𝑚 = waverider
z 3
tan 3β
3 .
(5)
64 𝑧𝑧𝑤𝑤𝑤𝑤𝑤𝑤𝑤𝑤𝑤𝑤𝑤𝑤𝑤𝑤𝑤𝑤𝑤𝑤 tan 𝛽𝛽

The design spaces obtained from Equations (4) and (5) are depicted in Figure 4. To
visualize the difference between the two design spaces, we plotted the isosurfaces for
yshock,max at 0.01 and 0.07 based on the three variables, namely xshock , x f lat , and zwaverider ,
in the three-dimensional (3D) space. The design space of Takashima is represented by
 The design spaces obtained from Equations (4) and (5) are depicted in Figure 4. To
visualize the difference between the two design spaces, we plotted the isosurfaces for
𝑦𝑦𝑠𝑠ℎ𝑜𝑜𝑜𝑜𝑜𝑜,𝑚𝑚𝑚𝑚𝑚𝑚 at 0.01 and 0.07 based on the three variables, namely 𝑥𝑥𝑠𝑠ℎ𝑜𝑜𝑜𝑜𝑜𝑜 , 𝑥𝑥𝑓𝑓𝑓𝑓𝑓𝑓𝑓𝑓 , and
Aerospace 2022, 9, 348 𝑧𝑧𝑤𝑤𝑤𝑤𝑤𝑤𝑤𝑤𝑤𝑤𝑤𝑤𝑤𝑤𝑤𝑤𝑤𝑤 , in the three-dimensional (3D) space. The design space of Takashima is
8 of 22
represented by surfaces 𝑆𝑆1 and 𝑆𝑆3 , and the region surrounded by surfaces 𝑆𝑆2 and 𝑆𝑆4
indicates that of the present study. The design space defined by Takashima cannot
consider
surfaces Sthe variable 𝑥𝑥 the region
1 and S3 , and 𝑓𝑓𝑓𝑓𝑓𝑓𝑓𝑓
[Equation (4)]. Therefore,
surrounded by surfacesa substantial
S2 and S4 differenceindicates that existsof
between the design spaces reported by Takashima
the present study. The design space defined by Takashima cannot consider the variable and in this study. The design space of
this study expands
x f lat [Equation (4)]. Therefore, as 𝑧𝑧 increases, whereas that reported by
𝑤𝑤𝑤𝑤𝑤𝑤𝑤𝑤𝑤𝑤𝑤𝑤𝑤𝑤𝑤𝑤𝑤𝑤a substantial difference exists between the design spacesTakashima does not
change.
reported Figures by Takashima 4a,b illustrate
and in thisthe cross-sections
study. of theof𝑧𝑧this
The design space study axis
𝑤𝑤𝑤𝑤𝑤𝑤𝑤𝑤𝑤𝑤𝑤𝑤𝑤𝑤𝑤𝑤𝑤𝑤 expandsat 0.3as zand 0.4,
waverider
respectively.
increases, whereas The filledthatcircles
reported denote the designdoes
by Takashima points that
not cause Figure
change. the failure4a,bof the inverse
illustrate the
design,
cross-sections and theofempty circles represent
the zwaverider axis at 0.3 the andfeasible design points.
0.4, respectively. The In Figure
filled 4a,denote
circles the valuesthe
of 𝑦𝑦𝑠𝑠ℎ𝑜𝑜𝑜𝑜𝑜𝑜
design of the
points design
that cause points
the failureindicated byinverse
of the empty circles
design,are and0.001,
the empty 0.02, and 0.06;represent
circles in Figure
4b, they are 0.005, 0.01, 0.04, and 0.06. Although
the feasible design points. In Figure 4a, the values of yshock of the𝑠𝑠ℎ𝑜𝑜𝑜𝑜𝑜𝑜 the points with 𝑦𝑦 values of 0.001
design points indicated and
0.005
by empty are not circlesincluded
are 0.001, between
0.02, and the0.06;
points with values
in Figure 4b, they ofare0.01 to 0.07,
0.005, 0.01,the0.04,points are
and 0.06.
included in
Although thethe design
points withspace yshock because
values ofthe 0.001design spaceare
and 0.005 is not definedincluded considering
between the the
maximum
points withvalue values forof𝑦𝑦𝑠𝑠ℎ𝑜𝑜𝑜𝑜𝑜𝑜
0.01 . to Figure
0.07, 4a thedepicts
pointsaaredesign spacein
included reported
the design by Takashima
space because that
does
the design not overlap space iswith definedthe design
considering spacetheof maximum
this study.valueA design
for yshock point included
. Figure in this
4a depicts
design
a designspace spacecauses
reported the by failure
Takashima of the that
design.
doesTherefore,
not overlap this with design space must
the design space be of
excluded.
this study. Figure A design 4b point
illustrates
included the indesign point space
this design that cancausesbe designed
the failuredespite it not
of the design.
belonging to
Therefore, thisthe design
design space
space must of Takashima.
be excluded.This design
Figure point must
4b illustrates thebedesign
includedpoint inthat
the
design
can be designed space. In summary,
despite it not Figures 4a,b verify
belonging to thethat the design
design space space
of Takashima. derived in thisdesign
This study
is more
point must accurate than that
be included in thereported
designby Takashima.
space. Therefore,
In summary, Figurethe 4a,b proposed design
verify that space-
the design
defining methodology can be applied to arbitrary design parameters to formulate the
space derived in this study is more accurate than that reported by Takashima. Therefore, an
accurate design
proposed designspace.
space-defining methodology can be applied to arbitrary design parameters
to formulate an accurate design space.

(a) (b)
Figure 4.
4. Design
Designspace
spacecomparison: (a)(a)
𝑧𝑧𝑤𝑤𝑤𝑤𝑤𝑤𝑤𝑤𝑤𝑤𝑤𝑤𝑤𝑤𝑤𝑤𝑤𝑤
zwaverider==0.3; 0.3;(b)
(b)𝑧𝑧z𝑤𝑤𝑤𝑤𝑤𝑤𝑤𝑤𝑤𝑤𝑤𝑤𝑤𝑤𝑤𝑤𝑤𝑤 = 0.4.
Figure comparison: waverider = 0.4.

2.3. Performance
2.3. Performance Estimation
Estimation Technique
Technique
As explained
As explainedin inSection
Section2.1,
2.1,thethe DOE,
DOE, CFD
CFD analysis,
analysis, surrogate
surrogate model
model construction
construction and
and adaptive
adaptive sampling
sampling techniques
techniques are required
are required to represent
to represent theperformance
the vehicle vehicle performance
based on
based
the on the configuration.
configuration. This
This section section describes
describes each methodology
each methodology used in thisused in this study.
study.
The first
The first step
step in
in constructing
constructing the the model
model isis selecting
selecting the
the initial
initial sample
sample points.
points. Sample
Sample
points should
points should be be selected
selected such
such that
that they
they are
are evenly
evenly distributed
distributed in
in the
the design
design space.
space. In
In this
this
study, the
study, the initial
initial sample
sample points
points were
were selected
selected using
using optimal
optimal Latin-hypercube
Latin-hypercube sampling
sampling
(OLHS), which enables uniform sampling sampling forfor multivariate
multivariate models
models [18].
[18].
KFLOW, a compressible solver based on a structured grid, was used to calculate the
aerodynamic force
aerodynamic acting on
force acting on the
the waverider
waverider [19,20].
[19,20]. The
The Euler equation was
Euler equation was applied
applied toto
analyze the
analyze the inviscid steady-state flow.
inviscid steady-state flow. The
The HLLE+
HLLE+ flux scheme [19]
flux scheme [19] as
as the
the convective
convective term
term
and the Van Leer limiter of second-order accuracy were used for accurate and robust flow
analysis. To ensure a steady-state calculation, the backward Euler method was applied
with local time-stepping as a temporal integration scheme. The diagonalized alternating
direction implicit method was used to calculate the inverse matrix. The Courant–Friedrichs–
Lewy (CFL) number was set at 0.5.
 and the Van Leer limiter of second-order accuracy were used for accurate and robust flow
analysis. To ensure a steady-state calculation, the backward Euler method was applied
Aerospace 2022, 9, 348 with local time-stepping as a temporal integration scheme. The diagonalized alternating 9 of 22
direction implicit method was used to calculate the inverse matrix. The Courant–
Friedrichs–Lewy (CFL) number was set at 0.5.
Figure 5 shows a grid system used in this study. The grid independency analysis was
Figure 5 shows a grid system used in this study. The grid independency analysis was
conducted on this grid system The corresponding results were listed in Table 1. The lift
conducted on this grid system The corresponding results were listed in Table 1. The lift
coefficients are sufficiently accurate even in the coarse grid. However, the drag coefficient
coefficients are sufficiently accurate even in the coarse grid. However, the drag coefficient
shows a considerable error between the coarse grid and the fine grid. In particular, since
shows a considerable error between the coarse grid and the fine grid. In particular, since
the drag coefficient will be used as the objective function of this study, 7,180,000 grids
the drag coefficient will be used as the objective function of this study, 7,180,000 grids were
were used for the grid system.
used for the grid system.

Figure
Figure5.
5.Grid
Gridsystem
systemof
ofthe
thewaverider.
waverider.

Table 1. Grid independency analysis results.

Table 1. Grid independency analysis results.

Cases
Cases Grid
GridSize
Size 𝐂𝐂C𝐋𝐋L 𝐂𝐂C𝐃𝐃
D
−1
Coarse
Coarse11 1,709,960
1, 709, 960 1.091
1.091 2.195
2.195 ×
× 10
10−1
−1
10− 1
Coarse
Coarse22 3,578,542
3, 578, 542 1.091
1.091 2.177
2.177 ×
× 10
Moderate 7, 180, 000 1.091 2.165 × 10−1
Moderate 7,180,000 1.091 2.165 × 10−−11
Fine 14, 372, 820 1.091 2.151 × 10
Fine 14,372,820 1.091 2.151 × 10−1
The CFD code used herein was validated against the experimental results of an all-
bodyThe CFD codeaircraft
hypersonic used herein
[21]. Thewasexperimental
validated against the experimental
conditions results
were as follows: Mof an all-
∞ = 7.4,
body hypersonic aircraft [21]. The experimental conditions were
Re∞ = 15 × 10 , T∞ = 62K, Tw = 300K, and α = 0 , 5 , 10 , and 15 . Based on each𝑀𝑀angle
6 ◦ ◦ ◦ ◦ as follows: ∞ =
7.4, 𝑅𝑅𝑅𝑅∞ =(AOA),
of attack 15 × 10the 6
, 𝑇𝑇∞windward
= 62𝐾𝐾, 𝑇𝑇𝑤𝑤and
= 300K, and centerline
leeward 𝛼𝛼 = 0°, 5°, 10°, 15°. Based
and pressures
surface wereon each
obtained
angle of attack (AOA), the windward and leeward centerline surface
through experiments. Figure 6 illustrates the experimental and numerical results. Since the pressures were
obtained
Euler solverthrough experiments.
was employed in thisFigure
study,6 the
illustrates the experimental
viscous effect could not be and numerical
included in the
results. Since the Euler solver was employed in this study, the viscous
results. Therefore, CFD results were slightly underpredicted compared to the experiments effect could not be
included
owing toin thethe results. Therefore,
boundary layer effect.CFD resultsthe
However, were slightly results
numerical underpredicted
for two lowcompared to
AoA cases
the experiments
concurred well. Thisowing to the
study boundaryfocused
is primarily layer effect.
on theHowever,
waveriderthe thatnumerical
cruises at results
an AOAfor of
two low aAoA
0◦ with slender cases concurred
body shape. Inwell. This
other studythe
words, is flow
primarily focused on
characteristics the waverider
around that
the waverider
cruises
correspondat an AOA
to the of 0° area
physical withwhena slender
the AOA bodyis lowshape.
amongIn the
other words, theresults.
experimental flow
characteristics around the waverider correspond to the physical
Therefore, sufficiently accurate aerodynamic data can be obtained by applying the Euler area when the AOA is
low among the
solver used herein. experimental results. Therefore, sufficiently accurate aerodynamic data
can beTheobtained
aerodynamicby applying the Eulerissolver
performance used for
calculated herein.
sample points selected through the
OLHS with Euler solver. The surrogate model was generated using the performance data.
In this study, we used a kriging model suitable for a deterministic computer model [22].
The kriging model is expressed as indicated in Equation (6) for the input variable vector x.

y ( x ) = β + Z ( x ), (6)

where β denotes a global model, which is the trend of the entire domain, and Z ( x ) repre-
sents a local deviation, correcting the kriging model value based on each sample point. A
correlation function is defined as a function of distance to determine the global model and
local deviation. The model parameter that best predicts the value of the sample points is
determined through maximum likelihood estimation.
Aerospace 2022, 9, x FOR PEER REVIEW 10 of 22
Aerospace 2022, 9, 348 10 of 22

(a) (b) (c)

Figure
Figure6.6.Effect
Effectofofangle
angleofof
attack on on
attack centerline surface
centerline pressures:
surface (a) windward
pressures: cases; (b)cases;
(a) windward leeward cases;
(b) leeward
experimental model (c) all-body model [21].
cases; experimental model (c) all-body model [21].

The
Theaerodynamic performance
constructed model can beisefficiently
calculatedenhanced
for sample bypoints
addingselected
certainthrough the
data points.
OLHS with Euler solver. The surrogate model was generated using
Expected improvement (EI) is a method of selecting sample points to be added to improve the performance data.
In this study,
model we [23].
accuracy usedThea kriging
modelmodel
accuracy suitable
can be foreffectively
a deterministic
improved computer model
by adding [22].
a small
The kriging model is expressed as indicated in Equation (6) for the
number of sample points with EI. Equation (7) is used to select additional sample points in input variable vector
𝑥𝑥.
EI. The first term of Equation (7) increases when the distance between the design point and
optimum point reduces. The second term 𝑦𝑦(𝑥𝑥)exhibits a high value when the uncertainty of (6)
= 𝛽𝛽 + 𝑍𝑍(𝑥𝑥), the
model is large. The combination of these two values can generate the design point with a
where 𝛽𝛽 denotes
high model a global
uncertainty model,
around which is point.
the optimum the trend Thus,ofthe
theaccuracy
entire domain, and can
of the model 𝑍𝑍(𝑥𝑥)
be
represents a local deviation,
improved significantly correcting
by selecting the kriging
this point model value
as an additional sample based on each sample
point.
point. A correlation function is definedas a function  of distance todetermine the global
model and local deviation.
E[ I ] =The model parameter that+best
−
sφ predicts the. value of the sample
−

f min ŷ f min ŷ
( f min − ŷ)Φ (7)
points is determined through maximum likelihood estimation. s s
The constructed model can be efficiently enhanced by adding certain data points.
The surrogate model is modified by the performance estimation data of the added
Expected improvement (EI) is a method of selecting sample points to be added to improve
sample points. This process is repeated until the generated model attains a certain level of
model accuracy [23]. The model accuracy can be effectively improved by adding a small
accuracy. Finally, optimization is performed using the completed model.
number of sample points with EI. Equation (7) is used to select additional sample points
in
3. EI. The first
Results andterm of Equation (7) increases when the distance between the design point
Discussion
and
3.1. Progress of Directreduces.
optimum point The second term exhibits a high value when the uncertainty
Optimization
of the model is large. The combination of these two values can generate the design point
This section explains the application of the proposed direct design optimization
with a high model uncertainty around the optimum point. Thus, the accuracy of the model
method to the design problem. Section 3.1 presents the problem definition, which includes
can be improved significantly by selecting this point as an additional sample point.
defining the objectives, constraints and design parameters. Section 3.2 describes the process
𝑓𝑓 −𝑦𝑦� 𝑓𝑓 −𝑦𝑦�
of model construction to𝐸𝐸[𝐼𝐼] be used for optimization.
= (𝑓𝑓𝑚𝑚𝑚𝑚𝑚𝑚 − 𝑦𝑦�)𝛷𝛷 � 𝑚𝑚𝑚𝑚𝑚𝑚 � + 𝑠𝑠𝑠𝑠 � 𝑚𝑚𝑚𝑚𝑚𝑚 �.
𝑠𝑠 𝑠𝑠
(7)
3.1.1.The surrogate
Problem model is modified by the performance estimation data of the added
Definition
sample Thepoints. Thisconcern
primary process of
is repeated until
hypersonic the generated
flight is the highmodel
waveattains a certain
drag and level of
aerodynamic
accuracy. Finally,tooptimization
heating. Owing aerodynamicisheating,
performed using the
hypersonic completed
vehicles model.
require additional volume for
internal components, such as the thermal protection system. Since the drag and internal
3.volume
Resultsofand
the Discussion
vehicle are in a trade-off relationship, it is crucial to identify a concurrence
3.1. Progress of Direct
between them. Optimization
In this regard, two objectives, namely minimizing the drag coefficient
and maximizing the internal volume of the waverider, were considered herein. The lift
This section explains the application of the proposed direct design optimization
coefficient was selected as the constraint to support the mission payload. In the case of
method to the design problem. Section 3.1 presents the problem definition, which includes
design conditions, the altitude was set at 25 km, Mach number at 5, and shockwave angle β
defining the objectives, constraints and design parameters. Section 3.2 describes the
at 15◦ . The other atmospheric quantities, such as temperature or density, were set based on
process of model construction to be used for optimization.
the US standard atmosphere data [24]. The design problem can be summarized as follows:
Bulleted lists look like this:
3.1.1. Problem Definition
• Obj 1 Minimize CD ;
• Obj 2 Maximize Internal Volume;
Aerospace 2022, 9, 348 11 of 22

• Subject to CL ≥ CL,baseline ;
• under : M∞ = 5, β = 15◦ , altitude = 25 km.
Owing to the inverse design problem of the waverider, it is difficult to set a design
parameter that achieves a balance between over-parameterizing and over-restricting. To
address this problem, Kontogiannis investigated the DOF of various parameterization
methods for waverider design curves based on the osculating cone theory [13]. Herein, we
employed an efficient design parameterization method of Kontogiannis, which exhibits
a high DOF of design while using a small number of design parameters. The right-hand
side of the symmetry plane in Figure 3 depicts the design curves and design parameters of
this study. The height is the largest vertical distance between the shockwave and the upper
surface, and the width is the horizontal distance from the symmetry plane to the end of
the span direction. The height was fixed at 1.34 m, which set the length and width of the
vehicle at 5 m and 3 m, respectively. The shockwave curve comprises a straight line and a
fourth-order Bezier curve. The horizontal length of the straight line was expressed as X1
normalized to the width. The Bezier curve used five control points, wherein four points
were positioned parallel to the straight line to ensure second-order differential continuity
required for the osculating cone theory. Additionally, the five control points were evenly
positioned horizontally. Based on these two conditions, only one variable was used for the
shockwave curve, namely the height of the rightmost control point, which is expressed
as X2 normalized with height. The upper surface curve comprised only a fourth-order
Bezier curve, and four control points were used to define the upper surface of the Bezier
curve. All control points were evenly distributed in the horizontal direction. The positions
of the leftmost and rightmost control points were determined by the height, width, and
X2. Therefore, we used only two design variables that define the vertical position of the
inner two control points. The vertical distances between the two points and the center of
the upper surface are expressed as X3 and X4. Both these variables are normalized with
(1 − X2), which indicates the vertical distance between the rightmost control point and the
center of the upper surface. Table 2 summarizes the design variables, wherein the range of
each design variable is set to 0 to 1.

Table 2. Design variables.

Design Variable Description

X1 Flat region of the shockwave curve normalized with the width
X2 Height of the shockwave curve normalized with the height
Vertical distance from the upper surface normalized with the
X3
height (1 − X2)
Vertical distance from the upper surface normalized with the
X4
height (1 − X2)

Figure 7 illustrates examples of waverider configurations viewed from the top and
rear. The leading-edge shape viewed from the top and the upper surface shape viewed
from the rear vary from sharp to curved and flat, respectively. The rearview indicates that
the backside of the vehicle can demonstrate various design shapes. Therefore, various types
of waverider shapes can be derived with only four design variables because the efficient
parameter was employed in this study.
 Figure 7 illustrates examples of waverider configurations viewed from the top and
rear. The leading-edge shape viewed from the top and the upper surface shape viewed
from the rear vary from sharp to curved and flat, respectively. The rearview indicates that
the backside of the vehicle can demonstrate various design shapes. Therefore, various
Aerospace 2022, 9, 348
types of waverider shapes can be derived with only four design variables because the
12 of 22

efficient parameter was employed in this study.

Figure7.7.Examples
Figure Examplesofof waverider
waverider configurations
configurations with
with various
various design
design parameters:
parameters: M∞ =𝑀𝑀5,
∞ =β =5,15𝛽𝛽 ◦=,
15°, 𝑤𝑤𝑤𝑤𝑤𝑤𝑤𝑤ℎ = 3 m, and ℎ𝑒𝑒𝑒𝑒𝑒𝑒ℎ𝑡𝑡
width = 3 m, and height = 1.34 m. = 1.34 m.

Weformulated
We formulatedthe thedesign
designspace
spaceby byapplying
applyingthethedesign
designspace-defining
space-definingmethodology
methodology
presented in Section II.A.1. Equation (8) presents the design space used
presented in Section 2.2. Equation (8) presents the design space used in this study. in this study.The
The
largevalue
large valueon onthe
theleft-hand
left-handside
sideofofthe
theequation
equationimplies
impliesthatthatthe
thedistance
distancebetween
betweenthe the
uppersurface
upper surfaceand
andthe
thecenter
centerofofthe
thelocal
localosculating
osculatingcone
coneisisshort.
short.The
Theinfluence
influenceofofX3
X3and
and
X4was
X4 wasignored
ignoredbecause
becausethe thedesign
designspace
spacewaswasdefined
definedbased
basedon onthe
thecenter
centerof
ofthe
theupper
upper
surface.In
surface. Inthe
theoptimization
optimizationprocess,
process,10%10% ofof the
the offset
offset was
was applied
applied to the
to the right-hand
right-hand sideside
of
of Equation (8) to avoid the tolerance problem, which can occur owing
Equation (8) to avoid the tolerance problem, which can occur owing to computing errors. to computing
errors.
7 width 4 4
 
X2
X2 < 7 𝑤𝑤𝑤𝑤𝑤𝑤𝑤𝑤ℎ . (8)
(1 − X1 )4 4 <64 �height� . (8)
(1−𝑋𝑋1) 64 ℎ𝑒𝑒𝑒𝑒𝑒𝑒ℎ𝑡𝑡

Thebaseline
The baseline waverider
waverider configuration
configuration was generated
was generated using theusing
same the
designsame design
parameter
parameter with the median value of the design variable range. The range of
with the median value of the design variable range. The range of each design variable is seteach design
variable 0isand
between set 1.
between 0 and 1. of
The constraint The constraint
Equation ofviolated
(8) is Equationif (8)
theisvalues
violated if the
of the fourvalues
designof
the four design
variables are set variables
to 0.5. To are set tothis,
prevent 0.5. To
theprevent
value ofthis, the value
the design of the X1
variable design
wasvariable X1
set at 0.25,
was set at
whereas the0.25, whereas
remaining the remaining
values were set at values
0.5. were set at 0.5.

3.1.2.Model
3.1.2. ModelConstruction
Construction
Surrogatemodels
Surrogate modelswere
wereconstructed
constructedconsidering
consideringthe thedrag
dragforce,
force,internal
internalvolume,
volume,and
and
liftforce
lift forceto
tooptimize
optimizethe
thedesign.
design.The
Thedrag
dragand
andlift
lift forces
forceswere
wereestimated
estimatedusing
usingthe
theKFLOW
KFLOW
based on
based onthe
theEuler
Eulerequation.
equation. The
The internal
internal volume
volume was was calculated
calculated using
using the
thegenerated
generated
waverider configuration. We developed an initial kriging model using 34 initialsample
waverider configuration. We developed an initial kriging model using 34 initial sample
points generated based on the OLHS. Furthermore, EI was applied to the initial kriging
model to improve its accuracy by adding the data points selected using Equation (7). Three
sample points were added at each iteration, wherein two points maximize the EI values
for each of the two objective functions used in this study. The other point maximizes
the sum of the two EI values of each objective function. The process of sample point
addition was repeated until the error for all three added points was within 3%. Figure 8
illustrates the error of the model for each objective function and its averaged value based
on the refinement process. The drag coefficient error gradually decreased with the decrease
in the number of EI refinement processes. Although the volume error did not decrease
significantly and remained at a certain level, the average of the two values decreased to
3.24%. We performed nine iterations with three sample points added in each iteration. The
final 60 sample points, excluding one point because of the grid generation problem, were
used to construct the kriging model.
 with the decrease in the
value based onnumber of EI refinement
the refinement process. processes. Although the
The drag coefficient errorvolume errordecreased
gradually
did not decrease
with thesignificantly andnumber
decrease in the remained at refinement
of EI a certain level, the average
processes. Althoughof the
the two
volume error
values decreased
did not to 3.24%. We
decrease performedand
significantly nineremained
iterationsatwith three sample
a certain points
level, the addedof the two
average
in each iteration. The finalto603.24%.
values decreased sampleWepoints, excluding
performed one point
nine iterations because
with of the grid
three sample points added
Aerospace 2022, 9, 348 generationinproblem, were used
each iteration. Thetofinal
construct the kriging
60 sample points,model.
excluding one point because of the13grid of 22
generation problem, were used to construct the kriging model.
25
25 𝑎𝑎𝑣𝑅𝑅𝑟𝑟𝑎𝑎𝑒𝑒𝑅𝑅
average

20 cd 𝐶𝐶𝐷𝐷 𝑎𝑎𝑣𝑅𝑅𝑟𝑟𝑎𝑎𝑒𝑒𝑅𝑅
average

20 𝑉𝑜𝑎𝑎𝑎𝑎𝑚𝑚𝑅𝑅
vol cd 𝐶𝐶𝐷𝐷
Average error (%)
15 𝑉𝑜𝑎𝑎𝑎𝑎𝑚𝑚𝑅𝑅
vol

Average error (%)

15

10
10

5
5

0
1 2 3 4 5 6 7 8 9
0
1 2 3 4 5 6 7 8 9
Number of refinements
Number of refinements
Figure 8. Model error with the refinement process.
Figure 8.
Figure Modelerror
8. Model error with
with the
the refinement
refinement process.
process.
We performed cross-validation of the constructed kriging model to calculate the
We performed cross-validation of the constructed kriging model to calculate the
goodness of fitWebetween the model
performed and the flowof
cross-validation solver [23,25]. Figure
the constructed 9 depicts
kriging the to
model result
calculate the
goodness of fit between the model and the flow solver [23,25]. Figure 9 depicts the result
of cross-validation
goodness with
of fit the entirethe
between sample
modelpoints
and theindicated in the[23,25].
flow solver figure.Figure
The horizontal
9 depicts the result
of cross-validation with the entire sample points indicated in the figure. The horizontal
axis represents the calculated with
of cross-validation valuethe
derived
entirefrom the flow
sample points solver, whereas
indicated thefigure.
in the verticalThe
axishorizontal
axis represents the calculated value derived from the flow solver, whereas the vertical axis
representsaxis
therepresents
value predicted by the kriging
the calculated model. The
value derived frommodel is considered
the flow accurate
solver, whereas if
the vertical axis
represents the value predicted by the kriging model. The model is considered accurate if
sample points are closer
represents to the predicted
the value diagonal line of symmetry.
by the kriging model. The model is considered accurate if
sample points are closer to the diagonal line of symmetry.
sample points are closer to the diagonal line of symmetry.

(a) (b) (c)

(a)
Figure 9. Cross-validation data required for thefor(b)
kriging model; (a) (a)
lift lift
coefficient; (c) drag
Figure 9. Cross-validation data required the kriging model; coefficient;(b)
(b) drag coefficient;
coefficient;Figure
(c) internal volume.
9. Cross-validation data required for the kriging model; (a) lift coefficient; (b) drag
(c) internal volume.
coefficient; (c) internal volume.
The EI method selects the sample points that are used to improve the accuracy of the
model based on the value of the objective functions and constraints. Therefore, the accuracy
of the model varies according to the predicted value of the model. For instance, the lift
coefficient model with a value greater than 0.8 can be considered accurate, as illustrated
in Figure 9a. This is because the additional sample points were selected to satisfy the
constraint defined by a lower lift limit. Conversely, in the drag coefficient model, only
the sample points around the 0.2 value are close to the diagonal line. This result can be
attributed to two reasons. First, the EI process selects the sample points with a low drag
value because the design objective is defined as drag minimization. These sample points
improve the accuracy of the low-value region of the model. Second, the lower lift limit
constraint prevents the additional sample points from exhibiting a low drag value because
the lift and drag share a positive correlation.
As the additional sample points are selected to adhere to the optimization problem,
the regions of the lift and drag model with high accuracy obtained through the EI process
contain the solution of the optimal problem. Therefore, the optimization process can
successfully determine the optimal solution with the constructed lift and drag models.
 lower lift limit constraint prevents the additional sample points from exhibiting a low
drag value because the lift and drag share a positive correlation.
As the additional sample points are selected to adhere to the optimization problem,
the regions of the lift and drag model with high accuracy obtained through the EI process
Aerospace 2022, 9, 348 contain the solution of the optimal problem. Therefore, the optimization process can 14 of 22
successfully determine the optimal solution with the constructed lift and drag models.
Furthermore, the volume model exhibited high accuracy over all areas, as plotted in
Figure 9c. Therefore,
Furthermore, the the entiremodel
volume constructed
exhibited modelhigh is accuracy
sufficiently accurate
over to beasused
all areas, for in
plotted
optimization.
Figure 9c. Therefore, the entire constructed model is sufficiently accurate to be used
We
for investigated the effect of each design variable on the constructed surrogate
optimization.
model. We performed the
We investigated the effect
ANOVA to design
of each identify the effect
variable on theofconstructed
the designsurrogate
variable model.
by
decomposing the variance of the model based on the design variables
We performed the ANOVA to identify the effect of the design variable by decomposing [26]. The results of the
the variance
ANOVAofthat the quantify
model based the proportion
on the design of variables
the influence [26]. of
The each design
results variable
of the ANOVA arethat
plotted in Figure 10. As indicated in the figure, the design variable X2
quantify the proportion of the influence of each design variable are plotted in Figure 10. is dominant in all As
three models. As X2 defines the height of the shockwave, it exhibits the most
indicated in the figure, the design variable X2 is dominant in all three models. As X2 defines pronounced
effect
theon the curvature
height of the shockwave
of the shockwave, it exhibitsamong
the most thepronounced
four designeffect
variables.
on the The curvature
curvature of the
of the shockwave determines the location of the center of the cone
shockwave among the four design variables. The curvature of the shockwave determines on the osculating plane,
which
the is a crucial
location offactor in determining
the center of the conethe onbasic flow field. plane,
the osculating Therefore,
whichtheisflow properties
a crucial factor in
around the vehicle
determining theare primarily
basic affected
flow field. by X2. Additionally,
Therefore, as variables
the flow properties around X3 and
the X4 were are
vehicle
normalized
primarilytoaffected
X2, X2 significantly influences
by X2. Additionally, as all models.X3
variables Inand
the X4
casewere
of the drag coefficient
normalized to X2, X2
model, the effectinfluences
significantly of X3 is stronger
all models. compared
In the caseto ofthattheofdrag
X4. coefficient
This resultmodel,
indicates
the that
effectthe
of X3
shape of the central area of the vehicle is associated with the drag coefficient.
is stronger compared to that of X4. This result indicates that the shape of the central area of In contrast,
X4 affects the volume
the vehicle modelwith
is associated more thethan
dragX3, which implies
coefficient. that the
In contrast, X4side areathe
affects of volume
the vehicle
model
influences
more than the X3,
volume.
which implies that the side area of the vehicle influences the volume.

(a) (b) (c)

Figure
Figure10. 10.
Analysis of of
Analysis variance
variance(ANOVA)
(ANOVA)results
results for the constructed
for the constructedkriging
krigingmodel;
model;(a)(a) drag
drag coeffi-
coefficient; (b) internal volume; (c) lift coefficient.
cient; (b) internal volume; (c) lift coefficient.

3.2.3.2.
Design Optimization
Design Optimization
In this section,
In this optimal
section, solutions
optimal obtained
solutions from
obtained fromthe genetic
the geneticalgorithm
algorithmare
arediscussed
discussed in
in terms
terms ofofobjectives
objectivesand
and constraints.
constraints. Furthermore,
Furthermore, flow flow characteristics
characteristics aroundaround the
the waverider
are analyzed by re-performing the flow analysis on the optimal solutions. Finally, the
decision tree analysis is performed to identify the general characteristics of the waverider.

3.2.1. Optimized Results

Optimization was performed through a genetic algorithm using the constructed krig-
ing model. The population size and generation size were 256 and 100, respectively. The
crossover rate and mutation rate were fixed at 1 and 0.1. Through the optimization process,
111 non-dominated solutions were obtained. Figure 11 illustrates the relationship between
the drag coefficient and a geometric constraint of each optimal solution. The geometric
constraint is the value of the left-hand side in Equation (8), which determines whether
the design point is included in the design space. The dashed line indicates the boundary
of the design space, and the shaded area represents the region outside the design space.
As plotted in Figure 11, the geometric constraint value increases with the increase in the
drag coefficient and remains constant after encountering the boundary of the design space.
Although optimal solutions after the boundary encounter exhibit a constant geometric
constraint value, these solutions present a wide range of drag. This is because the values of
design variables X3 and X4 can be different here, as these two design variables were not
considered in Equation (8). The optimal solutions with a constant geometric constraint
 boundary of the design space, and the shaded area represents the region outside the
design space. As plotted in Figure 11, the geometric constraint value increases with the
increase in the drag coefficient and remains constant after encountering the boundary of
the design space. Although optimal solutions after the boundary encounter exhibit a
Aerospace 2022, 9, 348 constant geometric constraint value, these solutions present a wide range of drag. This 15 ofis
22
because the values of design variables X3 and X4 can be different here, as these two design
variables were not considered in Equation (8). The optimal solutions with a constant
geometric constraint
exhibit a large exhibit
internal a largewhich
volume, internal volume,
is the secondwhich is the second
objective. Designobjective.
points for Design
these
points
optimal solutions are close to the boundary of the design space. If the design spaceIfis the
for these optimal solutions are close to the boundary of the design space. not
design space
accurate, the is not accurate,
boundary the and
changes, boundary changes,
the design and
points theliedesign
that on thepoints that can
boundary lie on
be the
lost.
boundary
In this case,canthe
beoptimization
lost. In this case, the optimization
process cannot determineprocess
thecannot
optimal determine
solutions.the
Tooptimal
prevent
solutions. To the
this situation, prevent
designthis situation,
space the designdefined.
must be accurately space Therefore,
must be accurately
the proposed defined.
design
Therefore,
space-definingthe proposed
methodology design
thatspace-defining methodology
can accurately derive thatspace
the design can accurately
is necessaryderive
when
the design space is necessary when designing
designing an osculating cone theory-based waverider. an osculating cone theory-based waverider.

Infeasible region
2.5
Feasible region
Geometric constraint

1.5

0.5

0
0.21 0.22 0.23 0.24 0.25
Obj. 1 - 𝐶𝐶𝐷𝐷
Figure11.
Figure Dragcoefficient
11.Drag coefficientand
andgeometric
geometricconstraint
constraintfor
foroptimal
optimal designs.
designs.

The Pareto front for the non-dominated solutions is shown in Figure 12. The horizontal
The Pareto front for the non-dominated solutions is shown in Figure 12. The
and vertical axes are drag coefficient and internal volume, respectively. The areas with low
horizontal and vertical axes are drag coefficient and internal volume, respectively. The
performance compared to the baseline were shaded. Twelve representative solutions out
areas with low performance compared to the baseline were shaded. Twelve representative
of the 111 solutions were obtained by sorting the entire solutions in the order of the drag
solutions out of the 111 solutions were obtained by sorting the entire solutions in the order
coefficient and selecting the solutions to have nine solution intervals. In order to identify the
of the drag coefficient and selecting the solutions to have nine solution intervals. In order
characteristics of the waverider configuration by classifying the 12 representative solutions,
to identify the characteristics of the waverider configuration by classifying the 12
the K-mean clustering method, a famous data mining method, was used for this study [27].
representative solutions, the K-mean clustering method, a famous data mining method,
As a result of K-mean clustering, 12 solutions were classified into four groups. The top and
was used for this study [27]. As a result of K-mean clustering, 12 solutions were classified
back view of each solution was plotted in Figure 12. The design variables and geometric
constraint—defined by the left-hand side of Equation (8)—values of the 12 solutions are
shown in Table 3.

Table 3. Design variables and geometric constraint for representative optimal solutions.

X1 X2 X3 X4 Constraint
OPT 1 0.002 0.828 0.306 0.183 0.83
Group 1 OPT 2 0.051 0.824 0.351 0.198 1.02
OPT 3 0.088 0.826 0.393 0.184 1.19
OPT 4 0.119 0.826 0.418 0.194 1.37
OPT 5 0.142 0.826 0.430 0.200 1.52
Group 2 OPT 6 0.166 0.825 0.442 0.202 1.70
OPT 7 0.186 0.828 0.431 0.191 1.88
OPT 8 0.217 0.829 0.419 0.186 2.21
OPT 9 0.230 0.849 0.418 0.084 2.41
Group 3
OPT 10 0.225 0.881 0.317 0.020 2.44
OPT 11 0.226 0.885 0.168 0.044 2.47
Group 4
OPT 12 0.226 0.882 0.034 0.001 2.46
Aerospace 2022, 9, x FOR PEER REVIEW 16 of 22

into four groups. The top and back view of each solution was plotted in Figure 12. The
Aerospace 2022, 9, 348
design variables and geometric constraint—defined by the left-hand side of Equation
16 of 22

(8)—values of the 12 solutions are shown in Table 3.

65

60

OPT 3
55

OPT 9 OPT 10 OPT 11 OPT 12

50

OPT 2
Obj. 2 - 𝑉𝑜𝑎𝑎𝑎𝑎𝑚𝑚𝑅𝑅
45
Group 3 Group 4
40
OPT 4 OPT 5 OPT 6 OPT 7 OPT 8
OPT 1
35

30
Group 1 Group 2

25
Baseline

20
0.21 0.215 0.22 0.225 0.23 0.235 0.24 0.245 0.25

Obj. 1 - 𝐶𝐶𝐷𝐷

Figure 12.
Figure 12. Optimal
Optimal points
points and
and Pareto
Pareto front
front estimated
estimated by
by kriging
kriging model.
model.

TableAs3. listed
Designinvariables
Table 3,and
thegeometric
value of X1 in solutions
constraint of group 1 isoptimal
for representative lower solutions.
compared to those
in the other groups. A low X1 value reduces the curvature of the shockwave attached to
the waverider. Therefore, a geometric constraint value of group 1 is lower than those of the
X1 X2 X3 X4 Constraint
other groups because the distance between the center of the local osculating cone and the
shockwave on the osculating plane is large. The X1 value in group 2 gradually increases
OPT 1 0.002 0.828 0.306 0.183 0.83
from OPT 4 to OPT 8, which in turn increases the geometric constraint values for group 2.
Groupthe
Finally, OPT 2 constraint
1 geometric 0.051 value is0.824
maximized 0.351at group 3 and0.198
group 4 due 1.02to high X1
and X2 values.OPT 3 0.088 0.826 0.393 0.184 1.19
Geometric constraint values for groups 3 and 4 are maximized owing to the high X1
OPT 4 0.119 0.826 0.418 0.194 1.37
and X2 values. The X3 and X4 values of group 4 are the lowest among all groups. As the
X3 and X4 areOPT 5
decreased, 0.142
the 0.826of the waverider
upper surface 0.430 0.200 higher. The
is formed 1.52higher
Group 2
upper surface OPT 6
configuration 0.166
causes the 0.825
center of the 0.442
local 0.202
osculating cone to 1.70 the
approach
upper surface.
OPT 7 0.186 0.828 0.431 0.191 1.88
Figure 13 shows the lateral cross-sections of OPT1 and OPT12. In sections (a) and (d),
the curvature OPTof the8lower 0.217
surface of the0.829 0.419 is similar.
two configurations 0.186 However,2.21 the lower
surface of OPT12 OPT 9partially protrudes
0.230 at section
0.849 (b) and
0.418protrudes very
0.084 large in section
2.41 (c).
Group
This 3 effect of a high geometric constraint value, given in groups 3 and 4. A high
is the
OPT 10 value0.225
geometric constraint implies that0.881
the distance0.317
between the0.020 2.44
center of the osculating
cone and the OPT 11
shockwave is0.226
short. Hence, 0.885
the initial 0.168 0.044
point of the streamline 2.47which
tracing,
isGroup 4
the leading-edge
OPT 12 point, is
0.226close to the
0.882 center of the
0.034 cone. When
0.001 the initial point of
2.46
the streamline approaches the center of the cone, the angle between the streamline and
the centerline of the cone increases considerably. Consequently, the lower surface of the
As listed in Table 3, the value of X1 in solutions of group 1 is lower compared to those
waverider defined by the streamline approaches the shockwave. Owing to this process,
in theends
both otherofgroups. A low X1
the waverider value reduces
configuration the curvature
of groups 3 and 4 of the shockwave
protrude, attachedthe
which expands to
the waverider. Therefore, a geometric constraint value of group 1 is
volume of the waverider. To conclude, a high curvature of the shockwave increases the lower than those of
the othervolume.
internal groups because the distance between the center of the local osculating cone and
the shockwave on the osculating plane is large. The X1 value in group 2 gradually
increases from OPT 4 to OPT 8, which in turn increases the geometric constraint values
for group 2. Finally, the geometric constraint value is maximized at group 3 and group 4
due to high X1 and X2 values.
 is the leading-edge point, is close to the center of the cone. When the initial point
streamline approaches the center of the cone, the angle between the streamline an
centerline of the cone increases considerably. Consequently, the lower surface o
waverider defined by the streamline approaches the shockwave. Owing to this pr
both ends of the waverider configuration of groups 3 and 4 protrude, which expand
Aerospace 2022, 9, 348 17 of 22
volume of the waverider. To conclude, a high curvature of the shockwave increas
internal volume.

Figure
Figure 13. Lateral 13. Lateral cross-sections
cross-sections diagram of OPTdiagram of OPT 112;
1 and and(a)OPT
z =12; 𝑧 =z0=𝑚;−(b)
(a) (b)
0 m; 1.4𝑧 m;
= −1.4 𝑚; (
−1.8
(c) z = − 1.8 m; (d) z=𝑚;−(d)
2.2𝑧m.
= −2.2 𝑚.

The upper
The upper surface shape issurface
formed shape is formed
upwards due toupwards
the low due to the
X3 and X4low X3 and
of OPT X4 of OPT 12
12, and
the leading edge
the moves
leadingforward.
edge moves As aforward.
result, theAslength of OPT
a result, 12 is longer
the length of OPTthan that
12 is of than t
longer
OPT 1, as shownOPTin Figure 13b–d. At
1, as shown in the same13b–d.
Figure time, the
At higher
the sameupper surface
time, the also decreases
higher upper surface
the distance between
decreases the the
localdistance
osculating cone vertex
between and osculating
the local the upper surface. Therefore,
cone vertex and thelowupper su
X3 and X4 values induce low
Therefore, a large
X3 internal volume.
and X4 values induce a large internal volume.
The performance The analysis
performancewas repeated
analysis wasfor the baseline
repeated for shape, OPT 1,
the baseline and OPT 12
shape, 1, and OPT
to calculate the objective
calculate thevalues
objectiveaccurately and identify
values accurately andthe detailed
identify thecharacteristics of
detailed characteristics o
the flow field.flow
Figure 14Figure
field. depicts 14the flowthe
depicts analysis results results
flow analysis for the for
baseline, OPT 1,OPT
the baseline, and1, and OP
OPT 12. According
Accordingto thetocharacteristics of the
the characteristics of waverider,
the waverider, the the
shockwave
shockwave is attached
is attachedto to the le
the leading edge of the vehicle in all three cases. In the case of OPT 1, the drag coefficient
edge of the vehicle in all three cases. In the case of OPT 1, the drag coefficient decr
decreased by by3.0%,
3.0%,andandthe volume
the volume increased by 51.5%
increased from from
by 51.5% the baseline. In OPT
the baseline. In 12,
OPT 12, w
which maximized the
maximized internal
the volume,
internal the volume
volume, the increased
volume by 159.9%.
increased by Although
159.9%. the
Although the
Aerospace 2022, 9, x FOR PEER REVIEW 18 of 22
drag coefficient increasedincreased
coefficient by 5.65% in bycomparison with that ofwith
5.65% in comparison the baseline, the increase
that of the baseline,inthe increa
performance was drastic. was drastic.
performance

(a) (b) (c)

Figure
Figure 14. Computational fluid
14. Computational fluid dynamics
dynamics(CFD) (CFD)results
results
for for optimal
optimal points
points and baseline
and baseline config-
configuration (𝑀𝑀∞ = 5, 𝑎𝑎𝑎𝑎𝑎𝑎𝑎𝑎𝑎𝑎𝑎𝑎𝑎𝑎𝑎𝑎 = 25 𝑘𝑘𝑘𝑘); (a) baseline; (b) optimum 1-minimal drag; (c) optimum
uration (M∞ = 5, altitude = 25 km ); (a) baseline; (b) optimum 1-minimal drag; (c) optimum 12-
12-maximal volume.
maximal volume.

Figure
Figure1515illustrates
illustrates the
the distribution
distribution ofof the
thepressure
pressurecoefficient
coefficientatatthe
the lower
lower surface
surface for
for OPT 1 and OPT 12. The pressure values around the plane of symmetry
OPT 1 and OPT 12. The pressure values around the plane of symmetry in the two cases in the two cases
are
are identical.
identical. However,
However, thethe pressure
pressure at both
at both endsends of the
of the vehicle
vehicle is higher
is higher at OPT
at OPT 12,12,
andand
the
the pressure acting on the area between the ends and the plane of symmetry
pressure acting on the area between the ends and the plane of symmetry is higher at OPT is higher at
OPT
1. The1. The shaded
shaded area area is where
is where the pressure
the pressure of OPT of 12
OPT 12 is higher
is higher than thethan the pressure
pressure of OPTof1,
OPT 1, and the striped area is the opposite. As optimized waveriders
and the striped area is the opposite. As optimized waveriders have a shape in which both have a shape in
which both ends protrude, the front projected area, which significantly affects
ends protrude, the front projected area, which significantly affects the drag value, is large the drag
value,
at both is ends.
large at both ends.
Therefore, theTherefore, the and
shading area shading area and
the striped theare
area striped area
similar, butarethesimilar,
drag is
but the drag is very different. In summary, the waverider has
very different. In summary, the waverider has a lower drag when the protruding a lower drag when the
region
protruding
is small. Thisregion is small.
is opposite to This is opposite
the shape to the shape
of the waverider of the for
required waverider
volume required
increase, for
and
volume increase, and it can be seen that the waverider has an inherent trade-off
relationship between internal volume and drag.
 are identical. However, the pressure at both ends of the vehicle is higher at OPT 12, and
the pressure acting on the area between the ends and the plane of symmetry is higher at
OPT 1. The shaded area is where the pressure of OPT 12 is higher than the pressure of
OPT 1, and the striped area is the opposite. As optimized waveriders have a shape in
which both ends protrude, the front projected area, which significantly affects the drag
Aerospace 2022, 9, 348
value, is large at both ends. Therefore, the shading area and the striped area are similar,
18 of 22

but the drag is very different. In summary, the waverider has a lower drag when the
protruding region is small. This is opposite to the shape of the waverider required for
volume
it can beincrease,
seen that and it can be has
the waverider seen
an that the trade-off
inherent waverider has an inherent
relationship betweentrade-off
internal
relationship
volume and between
drag. internal volume and drag.

Figure 15. Lower surface pressure coefficient distribution for OPT 1 and OPT 12.
Figure 15. Lower surface pressure coefficient distribution for OPT 1 and OPT 12.
3.2.2. Decision Tree Analysis
3.2.2. In
Decision Treewe
this study, Analysis
analyzed the design points using the decision tree analysis, which
evaluates specific problems
In this study, we analyzed through statistical
the design pointsanalysis
using theofdecision
multivariate data [28].
tree analysis, The
which
proposed design space-defining methodology facilitates this analysis because
evaluates specific problems through statistical analysis of multivariate data [28]. Theit corresponds
to an exploration
proposed design ofspace-defining
the design space. The application
methodology of this method
facilitates aids in identifying
this analysis because ita
set of designto
corresponds variables and a design
an exploration of therule thatspace.
design maximize or minimizeofthe
The application objective
this methodfunction.
aids in
The decision tree analysis repeatedly divides the data into two subgroups
identifying a set of design variables and a design rule that maximize or minimize to obtain results.
the
The criteria for the classification of the two groups are determined by a specific design
objective function. The decision tree analysis repeatedly divides the data into two
reference value. After several repetitions, a resultant subgroup that represents the design
subgroups to obtain results. The criteria for the classification of the two groups are
rule can be obtained.
Figure 16 illustrates the results of the decision tree analysis based on each objective
function and the results of 60 CFD analyses. In the figure, r av denotes the ratio of the mean
of the current group objective function to the mean of the population objective function,
and n indicates the number of data points in the group. Furthermore, the criteria used to
classify the group are indicated in each box. The classification was conducted until the
number of data points in the subgroup reached 10% of those in the population. The first
design rule expressed for the volume is calculated using Equation (9).

i f (X2 > 0.4338) and (X4 ≤ 0.4834), then (r av = 1.5329). (9)

Identical to the ANOVA result, variables X2 and X4 classified the data with the highest
accuracy. Limiting these two variables can facilitate the derivation of design results that
increase the volume significantly. According to the expressed design rule, the values of
X2 and X4 should be higher and lower than 0.4338 and 0.4834, respectively. As discussed
above, when the value of X2 increases, the curvature of the shockwave and volume increase.
When the value of X4 decreases, the distance between the center of the osculating cone and
the upper surface decreases, which in turn increases the volume. OPT 1 and OPT 12 of this
study strictly follow this design rule.

i f (X2 ≤ 0.1147) and (X1 ≤ 0.5) and (X4 > 0.4358), then (r av = 0.7445) (10)
 variables X2 and X1 should be low, and X4 should be high to exhibit a low drag. These
values correspond to the large distance between the center of the osculating cone and the
upper surface of the waverider. Therefore, the results of the two design rules indicate that
the internal volume and drag of the waverider rely on the location of the center of the cone
and upper surface. Unlike the optimization results which have lift coefficient constraints,
Aerospace 2022, 9, 348 decision tree analysis was applied for the entire design space. To conclude, the 19 above
of 22

results can be regarded as a general characteristic of waverider.

(a) (b)
Figure
Figure 16.
16. Decision
Decisiontree
treeanalysis: (a)(a)
analysis: Geometric constraint
Geometric based
constraint on the
based position
on the of the
position conecone
of the center; (b)
center;
Geometric constraint based on the intersected planes.
(b) Geometric constraint based on the intersected planes.

4. Conclusions
Equation (10) represents the design rule for minimizing drag. The design variable
To address
X2 was selectedthe inherent
as the issues in the
first criterion. inverse
Unlike the waverider
first designdesign, a novel direct
rule (Equation (9)), design
design
method for a waverider is proposed. We performed multi-objective design optimization
variables X2 and X1 should be low, and X4 should be high to exhibit a low drag. These
for the correspond
values drag and internal volume.
to the large The conclusions
distance between theofcenter
the analysis of the proposed
of the osculating cone anddesign
the
method and design
upper surface of theresults can be
waverider. summarized
Therefore, as follows:
the results of the two design rules indicate that
the internal
(1) volume and
We developed drag of
a general the waverider
approach relythe
to define on design
the location
space,ofwhich
the center of the
can be cone
applied
and upper surface. Unlike the optimization results which have lift coefficient
to an arbitrary set of design parameters of a waverider using the osculating cone constraints,
decision tree analysis was applied for the entire design space. To conclude, the above
theory. To this end, the failure conditions of the waverider inverse design used in the
results can be regarded as a general characteristic of waverider.
literature were classified into two geometric relationships, which were
4. Conclusions
To address the inherent issues in the inverse waverider design, a novel direct design
method for a waverider is proposed. We performed multi-objective design optimization
for the drag and internal volume. The conclusions of the analysis of the proposed design
method and design results can be summarized as follows:
(1) We developed a general approach to define the design space, which can be applied to
an arbitrary set of design parameters of a waverider using the osculating cone theory.
To this end, the failure conditions of the waverider inverse design used in the literature
were classified into two geometric relationships, which were mathematically analyzed
and determined to be redundant. Based on the analysis, we obtained a discriminant
formula that unifies the two conditions. The design space for the waverider can be
derived by applying the discriminant formula to the design curves. We observed that
the obtained design space was more accurately represented than the reference model
based on the ad hoc relation;
(2) Further, general characteristics of the waverider were derived using data mining
methods, such as K-means clustering and decision tree analysis. The aerodynamic
performance and the shape of the waverider were primarily affected by the curvature
of the shockwave. The large curvature of the shockwave reduced the distance between
the cone vertex of the osculating plane and the upper surface point. Consequently, the
lower surface tends to approach the shockwave. As a result, the ends of the waverider
protrude, and the internal volume increases. This protruding region increases the
drag acting on the waverider. In summary, the waverider has an inherent geometric
trade-off relationship between internal volume and drag;
(3) In the proposed design framework, the aerodynamic performance of the aircraft was
directly considered an objective or constraint during the design process, which is one
of the primary strengths of the direct design method. The computational efficiency
required for the direct design method was achieved using surrogate models derived
from high-fidelity flow analyses.
Aerospace 2022, 9, 348 20 of 22

Author Contributions: Conceptualization, J.S. and K.Y.; methodology, J.S. and K.Y.; validation, J.S.;
formal analysis, J.S.; writing—original draft preparation, J.S.; writing—review and editing, C.S. and
K.Y.; visualization, J.S.; supervision, K.Y.; project administration, K.Y.; funding acquisition, K.Y. All
authors have read and agreed to the published version of the manuscript.
Funding: This research was funded by Scramjet Combined Propulsion System Project (No.16-106-
501-035) of Republic of Korea.
Institutional Review Board Statement: Not applicable.
Informed Consent Statement: Not applicable.
Data Availability Statement: Not applicable.
Acknowledgments: The authors especially thank Soo Hyung Park at Konkuk University for provid-
ing the source code of KFLOW for the current study.
Conflicts of Interest: The authors declare no conflict of interest.

Appendix A. Osculating Cone Theory

The osculating cone theory is used to generate the basic flow field, which is the
initial step in the waverider inverse design. The flow is approximated using a spanwise
combination of the conical flow field. To generate the basic flow field, the desired 2D
shockwave shape is initially defined on the base plane, as depicted in Figure A1 The
shockwave is then discretized into a set of shockwave points. The plane that passes
through the shockwave point and is perpendicular to the shockwave and the base plane
is identified as the osculating plane. If the radius of curvature is calculated at each point
of the shockwave, and the center of the cone is located at a certain distance based on the
size of the radius of curvature within the osculating plane, the flow of the cone and the
flow generated by the shockwave defined in the base plane can be considered similar. If the
change in curvature of the shockwave is not large and the conical flow of each osculating
plane is derived from the same shockwave angle, the shockwave of each osculating plane
exhibits similar strength. In this case, the flow independence between the osculating planes
Aerospace 2022, 9, x FOR PEER REVIEW 21 of 22
is ensured, as the spanwise gradient is not large. The actual flow field can be approximated
by combining the flows on the osculating planes.

Osculating plane
Base plane

𝑀𝑀∞ Upper surface of waverider

Conical shockwave

Desired shockwave

Figure A1.
Figure A1. Osculating
Osculating cone
cone theory
theory based
based waverider.
waverider.

Subsequently,the
Subsequently, theprocess
process of waverider
of waverider shape
shape derivation
derivation is performed
is performed on eachon each
osculat-
osculating
ing plane. Herein,
plane. Herein, the upperthesurface
upper onsurface on plane
the base the base plane istodefined
is defined to waverider
derive the derive the
waverider
shape. shape.
After After identifying
identifying the intersection
the intersection point with point withsurface
the upper the upper surface
in each in each
osculating
osculating plane, the point in the upstream direction of free flow is moved, and the
intersection point with the shockwave generated by the cone can be determined. This
intersection is considered the leading edge of the waverider shape, and the path created
by the movement of the point serves as the upper surface of the waverider. This is
followed by tracing the streamline of the osculating plane flow at each leading edge, and
Aerospace 2022, 9, 348 21 of 22

plane, the point in the upstream direction of free flow is moved, and the intersection point
with the shockwave generated by the cone can be determined. This intersection is con-
sidered the leading edge of the waverider shape, and the path created by the movement
of the point serves as the upper surface of the waverider. This is followed by tracing the
streamline of the osculating plane flow at each leading edge, and the generated streamline
is used as the bottom surface of the waverider. As the streamline and the wall surface must
be parallel according to the boundary conditions of the flow after the shockwave, the wall
surface generated by this streamline and the cone of the osculating plane must produce
identical shockwaves. Accordingly, when the upper and bottom lines of all osculating
planes are combined, a shockwave is attached along the leading edge of the vehicle. In
this process of deriving the waverider shape, the flow analysis is performed only for the
conical flow of a single shockwave angle, which considerably reduces the time required for
the design in comparison with other methods. However, the osculating cone theory can
be effectively applied to the design because it can generate sufficiently diverse basic flow
field shapes.

References
1. Jackson, K.; Gruber, M.; Barhorst, T. The HIFIRE flight 2 experiment: An overview and status update. In Proceedings of the 45th
AIAA/ASME/SAE/ASEE Joint Propulsion Conference & Exhibit, Denver, CO, USA, 2–5 August 2009. [CrossRef]
2. Küchemann, D. Hypersonic aircraft and their aerodynamic problems. Prog. Aerosp. Sci 1965, 6, 271–353. [CrossRef]
3. Nonweiler, T.R. Aerodynamic problems of Manned Space Vehicles. RAeS 1959, 63, 521–528. [CrossRef]
4. Ding, F.; Liu, J.; Shen, C.-b.; Liu, Z.; Chen, S.-h.; Fu, X. An overview of research on Waverider Design methodology. Acta Astronaut.
2017, 140, 190–205. [CrossRef]
5. Jones, J.G.; Moore, K.C.; Pike, J.; Roe, P.L. A method for designing lifting configurations for high supersonic speeds, using
axisymmetric flow fields. Ing. Arch. 1968, 37, 56–72. [CrossRef]
6. Rasmussen, M.L. Waverider Configurations Derived from Inclined Circular and Elliptic Cones. J. Spacecr. Rocket. 1980, 17, 537–545.
[CrossRef]
7. Sobieczky, H.; Dougherty, F.C.; Jones, K.D. Hypersonic Waverider Design from Given Shock Waves. In Proceedings of the First
International Hypersonic Waverider Symposium, College Park, MD, USA, 17–19 October 1990.
8. Chen, L.; Guo, Z.; Deng, X.; Hou, Z.; Wang, W. Waverider configuration design with variable shock angle. IEEE Access 2019, 7,
42081–42093. [CrossRef]
9. Zheng, X.; Li, Y.; Zhu, C.; You, Y. Multiple osculating cones’ waverider design method for ruled shock surfaces. AIAA J. 2019, 58,
854–866. [CrossRef]
10. Liebeck, R.H. Subsonic airfoil design. In Applied Computational Aerodynamics; Henne, P.A., Ed.; Progress in Astronautics and
Aeronautics: New York, NY, USA, 1990; Volume 125, pp. 133–165. [CrossRef]
11. Labrujere, T.E.; Slooff, J.W. Computational methods for the aerodynamic design of aircraft components. Annu. Rev. Fluid Mech.
1993, 25, 183–214. [CrossRef]
12. Takashima, N. Optimization of Waverider—Based Hypersonic Vehicle Designs. Ph.D. Thesis, University of Maryland, College
Park, MD, USA, 1997.
13. Kontogiannis, K.; Sóbester, A.; Taylor, N. Efficient Parameterization of Waverider Geometries. J. Aircr. 2017, 54, 890–901.
[CrossRef]
14. Lobbia, M.A. Multidisciplinary Design Optimization of Waverider-Derived Crew Reentry Vehicles. J. Spacecr. Rocket. 2017, 54,
233–245. [CrossRef]
15. Lobbia, M.A. Rapid Supersonic/Hypersonic Aerodynamics Analysis Model for arbitrary geometries. J. Spacecr. Rocket. 2017, 54,
315–322. [CrossRef]
16. Goldberg, D.E. Genetic Algorithms in Search, Optimization and Machine Learning; Addison-Wesley Longman Publishing Co., Inc.:
Boston, MA, USA, 1989.
17. Fonseca, C.M.; Fleming, P.J. Genetic Algorithms for Multiobjective Optimization: FormulationDiscussion and Generalization.
ICGA 1993, 93, 416–423. [CrossRef]
18. Park, J.-S. Optimal latin-hypercube designs for computer experiments. J. Stat. Plan. Inference 1994, 39, 95–111. [CrossRef]
19. Park, S.H.; Lee, J.E.; Kwon, J.H. Preconditioned Hlle method for flows at all Mach numbers. AIAA J. 2006, 44, 2645–2653.
[CrossRef]
20. Hong, Y.; Lee, D.; Yee, K.; Park, S.H. Enhanced high-order scheme for high-resolution rotorcraft flowfield analysis. AIAA J. 2022,
60, 144–159. [CrossRef]
21. Lockman, W.K.; Lawrence, S.L.; Cleary, J.W. Flow over an all-body hypersonic aircraft—Experiment and Computation. J. Spacecr.
Rocket. 1992, 29, 7–15. [CrossRef]
22. Martin, J.D.; Simpson, T.W. Use of Kriging models to approximate deterministic computer models. AIAA J 2005, 43, 853–863.
[CrossRef]
Aerospace 2022, 9, 348 22 of 22

23. Jones, D.R.; Schonlau, M.; Welch, W.J. Efficient Global Optimization of Expensive Black-Box Functions. J. Glob. Optim. 1998, 13,
455–492. [CrossRef]
24. Pasquale, M.S. Commercial Airplane Design Principles, 1st ed.; Elsevier: Amsterdam, The Netherlands, 2014; pp. 487–495. [CrossRef]
25. Lee, S.; Yee, K.; Rhee, D.-H. Optimization of the array of film-cooling holes on a high-pressure turbine nozzle. J. Propuls. Power
2017, 33, 234–247. [CrossRef]
26. Jeong, S.; Chiba, K.; Obayashi, S. Data mining for aerodynamic design space. J. Aeros. Compt. Inf. Commun 2005, 2, 452–469.
[CrossRef]
27. Lloyd, S. Least squares quantization in PCM. IEEE Trans. Inf. Theory 1982, 28, 129–137. [CrossRef]
28. Sugimura, K.; Obayashi, S.; Jeong, S. Multi-objective optimization and design rule mining for an aerodynamically efficient and
stable centrifugal impeller with a vaned diffuser. Eng. Optim. 2010, 42, 271–293. [CrossRef]