Aerospace Science and Technology 34 (2014) 3544

Contents lists available at ScienceDirect

Aerospace Science and Technology

www.elsevier.com/locate/aescte

Drag reduction on aircraft congurations with adaptive lifting surfaces

Aaron A. Cusher 1 , Ashok Gopalarathnam ,2
Department of Mechanical and Aerospace Engineering, Box 7910, North Carolina State University, Raleigh, NC 27695-7910, USA

a r t i c l e

i n f o

Article history:
Received 26 June 2013
Received in revised form 30 January 2014
Accepted 30 January 2014
Available online 15 February 2014
Keywords:
Conguration aerodynamics
Drag reduction
Adaptive wings
Aircraft aerodynamics
Aircraft design

a b s t r a c t
An approach, applicable to multiple-lifting-surface xed-wing aircraft operating at subcritical Mach
numbers, is presented for minimizing induced and prole drag with a constraint on the pitching moment.
The approach allows the designer to select surface incidence, twist, and ap angles as variables for the
optimization. The numerical formulation uses superposition to construct the spanwise lift distribution
from basic and additional loadings, and decomposes the ap-angle distributions for each surface into
mean and variation distributions. Together, these elements enable the solution of the problem using
semi-analytical methods that also provide insight. Results are presented for a three surface aircraft which
highlights low drag possibilities with positive static margins, presents the trade-offs between induced
and prole drag, and provides insight into the aerodynamics of multiple lifting surface congurations.
2014 Elsevier Masson SAS. All rights reserved.

1. Introduction
Aircraft congurations with adaptive lifting surfaces offer the
capability for in-ight alteration of the aerodynamics to suit different ight conditions. The resulting drag-reduction and loadalleviation possibilities could translate to potential benets in reduced fuel burn and emissions. Adaptive, or morphing, wing and
aircraft technologies therefore is an active area of research interest within aircraft aerodynamics and design [5,36,37]. Multiple trailing-edge aps distributed along the aircraft surfaces offer a simple method of aerodynamic adaptation, as it allows for
the in-ight redistribution of spanwise lift for minimum drag.
Benets of such approaches have been shown in several studies
for drag reduction and load alleviation [33,35,7,34,25], and adaptive trailing-edge aps are used frequently on high performance
sailplanes today. Recent research at North Carolina State University has shown benets from multiple trailing-edge aps on both
tailed [14] and tailless [9] congurations with constraints. The current study presents a similar optimization method for use on multiple surface congurations with trim constraints. As an example
application, a three surface conguration with trailing edge aps
and twist-distribution control sections is shown in Fig. 1.
Multiple surface congurations have been studied extensively.
Prandtls biplane equation [29], which assumes elliptically loaded

surfaces, served as an early approach for induced drag prediction.

However, subsequent studies have shown that elliptical distributions on each surface, particularly those operating in the downwash eld of another surface, do not necessarily result in minimum induced drag. Several researchers have provided modied

Corresponding author.
E-mail address: ashok_g@ncsu.edu (A. Gopalarathnam).
1
Graduate Research Assistant; currently Senior Aerodynamics Engineer, UTC
Aerospace Systems, Burnsville, MN, USA.
2
Associate Professor.
http://dx.doi.org/10.1016/j.ast.2014.01.012
1270-9638/ 2014 Elsevier Masson SAS. All rights reserved.

Fig. 1. Planform for example aircraft. The left side shows sections used for dening
twist relative to wing root. The right side shows ap locations.

36

A.A. Cusher, A. Gopalarathnam / Aerospace Science and Technology 34 (2014) 3544

versions of Prandtls equation [24,20,21] for multiple surface congurations, each presenting steps forward in understanding optimal spanwise lift and surface lift ratios. The study by Kroo and
McGeer [18] on conventional, tailed and canard congurations has
provided much inspiration for this work, as comparisons of surfaces twisted for minimum induced drag and trim have highlighted
shortcomings of canard congurations. Three surface congurations provide unique opportunities for aerodynamic analysis, as the
availability of three surface-lift variables opens up the possibilities
for satisfying longitudinal trim, thus allowing for optimal lift and
trim to occur simultaneously. This redundancy in design variables
has increased interest in three surface aircraft, as indicated by several references [17,26,31,13].
The current methodology builds on earlier work by the authors [9] and differs from earlier approaches for multiple-surface
optimization [24,20,21,18,17,26,31,13] in the use of basic and additional loading to formulate the optimization problem. Also, prole drag is considered. The formulation allows for minimization of
induced drag with or without trim constraints, with design variables that include incidence, twist, and ap angles. In addition,
the method allows for some ap angles to be set for prole drag
reduction. The method is very ecient and requires only the solution of simple matrix equations. Further, the information for these
equations can be obtained using standard vortex-lattice type methods.
2. Background
The following three subsections briey provide background material for the methodology developed in the paper.
2.1. Basic and additional lift distributions
The concept of basic and additional lift distributions is described in several references [4,1,19]. As explained in detail in
a recent article on multiple TE aps [14], the use of this concept enables the determination of ap angles using a simple,
semi-analytical approach. In this subsection, the highlights of this
concept are briey presented. For further details and an example showing the accuracy of the approach, the reader is referred
to [14].
Within the assumption of linear aerodynamics (linear C l
variation and linear C l relationship) and so long as the component of the induced velocities along the freestream direction are
small compared to V , the spanwise distribution of the section
lift coecient, C l ( y ), can be written as a sum of two contributions: (1) basic distribution, C lb ( y ), and (2) additional distribution,
C la ( y ). The basic distribution is the C l distribution at C L = 0, and
is the result of spanwise variations in geometric twist, aerodynamic twist due to camber, and ap deections. The additional C l
distribution, C la , is due to changes to for the wing with zero
geometric and aerodynamic twist.
The advantage of using the linear superposition concept is that
the net C l distribution for a particular conguration with N independent variables, (which can be TE ap angles, f , or twist
angles, t ), at a given C L can be posed in terms of the unknown
ap/twist angles:

C l = C L C la,1 + C lb,0 +

N


C lb, j j

(1)

j =1

where, C lb,0 is the basic C l distribution of the conguration due to

given baseline spanwise distributions of twist and camber with all
aps set to zero, and C lb, j is the basic C l distribution due to a unit
deection of ap j or unit angle for twist variable j. It is noted
that twist at any section, as used in this paper, is dened relative

to the wings root chord. Thus, twist of sections on the canard or

tail includes the incidence of that surface.
For a given bound-circulation distribution the induced drag,
D ind , can be obtained by integration along the wake trace in the
Trefftz plane (see Refs. [30] or [16]). In non-dimensional form, the
induced drag coecient, C D ind , can be written as:
b

C D ind =

2S ref

2 (k)
M


c ( y )C l ( y )

k =1

b2 (k)

w ( y)
V

dy

(2)

where the summation accounts for an M-surface conguration,

b
(k) is the half-span of lifting surface k, and w ( y ) is the Trefftz2
plane downwash distribution. As shown in Ref. [14], the total induced drag coecient for a conguration with N aps/twist variables can be expressed as follows:

C D ind = f T Df

(3)

where f , the transpose of f, is written as:

fT = [ C L

1 1

(4)

. . . N ]

and the drag-coecient matrix, D, is written as:

C D a0

C D a1

...

C D aN

C D 0a

C D 1a
D=

..
.

C D 00

C D 01

...

C D 0N

C D 10

C D 11

...

..
.

..
.

C D Na

C D N0

C D N1

D aa

C D 1N

..
.

(5)

. . . CD NN

in which, for each element of the D matrix, the rst subscript indicates the source of the c ( y )C l ( y ) distribution and the second
subscript indicates the source of the w ( y )/ V distribution. As an
illustration, C D a0 is written as:
b

C D a0 =

1
2S ref

2 (k)
M


c ( y )C la,1 ( y )

k =1

b2 (k)

w b,0 ( y )
V

dy

(6)

The elements of the D matrix can be pre-computed using any

wing analysis method such as a panel, vortex-lattice, Weissingertype, or lifting-line method that outputs the spanwise C l and
w / V distributions at a specied C L .
2.2. TE ap for drag-bucket control
For ight at low subsonic Mach numbers, wing prole drag is
dominated by skin-friction drag when there is no separated ow
and associated pressure drag. To minimize prole drag, airfoils are
often designed to have signicant regions of favorable pressure
gradient on both upper and lower surfaces to support laminar ow.
Such a natural laminar ow (NLF) airfoil typically has a distinct
low-drag range, or drag bucket, which is the range of lift coecients over which low drag is achieved. To extend the range of
lift coecients over which low drag is achieved, a trailing-edge
cruise ap is often used [27,28,22]. First introduced by Pfenninger [27,28], it has since been used on several airfoil designs [32,
23,3,11,6,2], especially on airfoils for high-performance sailplanes
[6,3].
Fig. 2 illustrates the effect of a trailing-edge ap on the lowdrag range of the NASA NLF(1)-0215F airfoil [32], chosen here
purely for illustration. It is seen that low prole C d is thus achieved
over a larger C l range using a TE ap than without the ap. This
benet is limited to a small range of ap angles, typically close

A.A. Cusher, A. Gopalarathnam / Aerospace Science and Technology 34 (2014) 3544

C M NP =

37

N

(C M sections, j + C M basic, j ) j
j =1

+ C M sections,0 + C M basic,0

(9)

The subscript sections refers to the airfoil section pitching moment, and the subscript basic refers to the pitching moment on
the conguration caused by a basic lift distribution. The sections
and basic terms that are subscripted with 0 relate to those contributions when no aps are deected and the twist is xed during
the design; these two terms, when added up, will be referred to as
C M 0 . C M sections, j and C M basic, j , which together will be referred to as
C M j , are the pitching moment contributions from a unit ap/twist
angle of variable j, which can be precomputed using Eqs. (10)
and (11):

y j ,end

C M sections, j =
Fig. 2. Effect of TE ap angle on the drag polar for the NASA NLF(1)-0215F airfoil
from wind-tunnel experiments [32] at a Reynolds number of 6 million. The division
of the drag polar into three regions, illustrated for the 0-deg ap, is discussed later
in Section 3.6.

to 10 deg, beyond which ow separation near the ap hinge results in large pressure drag, which negates the benet from the
reduction in skin-friction drag. These limiting values of ap angle,
when the ap is used for drag reduction, are referred to in this paper as the most-negative ap angle, f min , and the most-positive
ap angle, f max . Using the results for the NLF(1)-0215F airfoil in
Fig. 2 as an example, low C d can be achieved over a C l range from
C l low of 0.2, achieved with f min of 10 deg ap, to C l up of 1.3,
achieved with f max of +10 deg ap. For any value of operating C l
that lies between C l low and C l up , using a ap angle that is determined by linear interpolation between f min and f max will result
in a smooth variation of the TE ap angle with operating C l , while
ensuring low C d .
2.3. Static margin and pitching moment constraint
For longitudinal static stability, the aircraft center of gravity
should be forward of the neutral point, resulting in a positive
value for the static margin (SM), where SM = (xNP xcg )/c with
the x-axis pointing toward the rear of the aircraft. The pitching
moment coecient about the CG for an aircraft for a given lift coecient can be expressed as:

C M cg = C M NP C L SM

(7)

For an aircraft that is trimmed longitudinally, C M cg = 0, giving:

C M NP = C L SM

(8)

2
S c

(C m f ) j c ( y )2 dy

(10)

y j ,start
b

2 (k)
M 
2 
C M basic, j =
c ( y )C lb, j ( y )x( y ) dy
S c

(11)

k =1 0

The (C m f ) j term in Eq. (10) is the increment in section C m

due to a unit deection of ap j, and the C lb, j ( y ) term in Eq. (11)
is the basic loading resulting from a unit angle of ap/twist j.
The x( y ) term in Eq. (11) refers to the longitudinal difference
between the aircrafts neutral point and the local airfoil section
aerodynamic center, and is equal to (xNP xac ( y )).
3.2. Decomposition of the ap-angle distribution
Previous studies [14,9] on adaptive wings (single lifting surface
congurations) have shown that it is advantageous to decompose
the ap angles on a wing into a full-span deection angle, denoted
by f and labeled the mean angle, and a vector of variation an-

gles about the mean, denoted by f and labeled the variation

ap-angle distribution, as follows:

f = f + f

(12)

As shown in [14,9], for a particular lifting surface, the mean ap

affects the lift in a manner similar to surface angle of attack, and
can be used to change the surface C L without affecting the basic
loading. And conversely, the variation angles can be used to alter
the basic loading without changing surface C L . It will be shown
that the mean angle can be used to directly target prole drag
reduction while the variation angles play a critical role in induced
drag reduction via spanwise lift redistribution.

3. Methodology

3.3. Flap/twist angles for minimum trimmed induced drag

3.1. Pitching moment decomposition for a apped conguration

For minimum C D ind , the rst derivative of C D ind with respect

to each ap/twist variable is set equal to zero, as shown here for
ap/twist variable j:

A deection of a TE ap will result in a change to the pitching moment about the aircrafts neutral point from two sources:
(1) the change in the airfoil section properties, resulting from a
change in the airfoil camber line, and (2) the change in the basic lift distribution. A change in section twist will result in only
the second contribution as there is no change in section pitching
moment. Note that the neutral point of the aircraft is located at
the centroid of the additional loading. Therefore, the additional lift
distribution does not cause any pitching moment about the neutral point. Eq. (9) displays how the pitching moment coecient is
decomposed for a series of ap/twist angles:


C D ind
= (C D aj + C D ja )C L + (C D 0 j + C D j0 ) +
(C D i j + C D ji )i
j
N

i =1

=0

(13)

A constraint is introduced on the pitching moment about the aircraft NP which satises the trim requirement for a desired SM and
C L . Building on Eqs. (8) and (9), the constraint equation, g () = 0,
is:

38

A.A. Cusher, A. Gopalarathnam / Aerospace Science and Technology 34 (2014) 3544

g () =

N


C M j j + C M 0 (SM C L ) = 0

(14)

j =1

For a conguration with N ap/twist variables, the ap/twist

angles resulting in minimum trimmed induced drag can be obtained by solving the system of N + 1 linear equations expressed
in matrix form as:

C D 11

.
.
.

C D N1

. . . C D 1N
..
.
. . . C DNN

C M1

..

C MN

C M1

1
..
.

/2

. . . C MN
0

C D a1 C L + C D 01

..

.
=

C D aN C L + C D 0N

( S M C L ) + C M 0

(15)

In the above equation the j variables can be twist (relative to

wing root) or ap angles. The set of ap angles for one or more
lifting surfaces can be written as a sum of mean and variation
ap angles (Eq. (12)). In this paper, this ap-angle decomposition is done only for the wing. The details for the more general
situation are presented in [8]. Here the four wing ap angles in
Eq. (15), f ,1 f ,4 , are instead written as four variation ap angles,

f ,1 f ,4 , and a mean angle, f , w , making the system size N + 2.

The requirement that the choice of variation ap angles does not

result in a change to the C L of the wing gives the ( N + 3)rd equation:

W 1 f ,1 + + W 4 f ,4 = 0

(16)

where W 1 W 4 are weighting factors that account for the relative

contributions to the wing C L from the four aps. The resulting
overdetermined system can be solved either by eliminating one
of the four equations for f ,1 f ,4 or by using a least-squares approach.
3.4. Flap angles for prole drag reduction
The deection of the mean ap for a surface is similar to using
a cruise ap for shifting the low-drag range of the sections, and
allows for prole drag reduction. Determination of the best mean
ap f for a given wing C L is achieved by linear interpolation between (C l low , f min ) and (C l up , f max ), as follows:

f = f min +

C L C llow
C l up C llow

( f max f min )

(17)

Thus, the mean ap for a surface can be set to ensure that the
surface will operate with low prole drag. Considering Eq. (15), if
one of the ap angles is xed at a value for drag-bucket control,
the order of the system reduces by one and the remaining aps
can still be optimized for minimum C D ind . Displayed in Eq. (18) is
the example where ap j has been xed at a value of X f , j .

..
.

... CD
j 1 j 1

. . . C D j +1 j 1

..
.

..
.
C D j 1 j +1
C D j +1 j +1

..
.

..

. . . f , j 1

...
f , j +1

..

Fig. 3. Surface lift variation for a three surface aircraft from the current approach
compared with LOTS [13]. Optimization variables are the canard and tail incidence
angles.

CD

..

C
+
C
L
D
aj 1
0 j 1

C D aj+1 C L + C D 0 j+1

..
.

.
..

CD

j 1 j

C D j +1 j

..
.

[X f , j]

(18)

Thus, it is possible to x ap j to some value which ensures

low prole drag for that surface, and still achieve the minimum
C D ind value by optimum selection of the remaining ap angles.
3.5. Surface lift validation study
For validation purposes, a study was performed to mimic that
of Goodrich, et al. [13], which optimally adjusts the incidence angles of a three surface aircraft in order to minimize induced drag
with trim. Fig. 3 compares the results from the current method
with those from LOTS (linear optimum trim solution) [13] for a
static margin of 10% and a C M 0 = 0.1. The correlations for the
surface loadings are excellent, validating the effectiveness of the
current optimization method. As shown in Fig. 3, both methods
call for the wing to carry the primary loading while maintaining a
slightly negative load on the aft tail.
3.6. Drag polar approximation routine
The methodology presented in Section 3.4 assumes that the resulting ap angle will enable operation of the airfoil in its low-drag
range. Situations in which the airfoil ends up operating outside
the low-drag range are not considered in that methodology. To
account for such situations in the ap optimization calculations,
a rapid calculation procedure for the airfoil drag polar (for any
given Reynolds number and ap angle) is needed. The process
of producing a drag polar using an airfoil analysis code, such as
XFOIL [10], can take several minutes given several ap sections
with different ap angles and Reynolds numbers. A drag polar
approximation routine was developed for rapid calculation of approximate prole drag for the airfoil sections with any given ap
angle and Reynolds number. The drag polar approximation routine
considers three distinct drag polar regions: (1) the upper region,
(2) the drag bucket, and (3) the lower region. These regions, displayed earlier in Fig. 2 using the 0-deg ap as an example, are

A.A. Cusher, A. Gopalarathnam / Aerospace Science and Technology 34 (2014) 3544

39

dependent not only on airfoil shape, but also on Reynolds number

and ap angle.
In this procedure, input les are created to dene how the airfoil drag polar changes with Reynolds number and ap angle. The
input les are produced using several XFOIL analyses to span the
operating range of Reynolds numbers and ap angles for the airfoil.
Typically, ve input les are needed to dene the airfoil characteristics. A polynomial curve t is used to approximate each region
for each of these ve polars. Linear interpolation is used to produce
polynomial coecients based on changes in Reynolds number and
ap angle. Once the necessary input les for an airfoil are created
and stored, this procedure allows for rapid calculation of approximate prole drag for use in the ap optimization process. Details
of the procedure, including the choice of the ve drag polars and
the accuracy of the resulting polars, are provided in Section 2.2.7
and Figs. 2.7 and 2.8 of [8].
4. Drag reduction schemes

Fig. 4. Polar plots for the wing and canard airfoils at Re C l = 3 106 .

4.1. Scheme A: Unconstrained induced drag minimization using twist

In scheme A, the induced drag coecient of the conguration
is minimized at a specied C L with no constraint on the pitching
moment. The twist values are used as variables with all aps set to
zero degrees, and Eq. (15) is used without the constraint equation.
The lack of any constraint on the pitching moment means that the
method is free to alter the fore-and-aft distribution of lift among
the three surfaces. The resulting lift distribution may well be such
that pitch equilibrium is satised about an undesirable CG location,
such as with a negative static margin.
4.2. Scheme B: Trim-constrained induced drag minimization using twist

Fig. 5. Variation of C D ind and span eciency, e with static margin (SM) for the threesurface conguration at C L = 0.5, showing results from schemes A and B.

In scheme B, twist variables are again used with aps set to

zero degrees, but the conguration C D ind is minimized subject to a
constraint on the pitching moment coecient about a specied CG
location using Eq. (15). Thus pitch trim is satised with a desired
static margin.

Fig. 6. Comparison of span eciency, induced, prole, and total C D for the three-surface conguration from schemes A, B, and C.

40

A.A. Cusher, A. Gopalarathnam / Aerospace Science and Technology 34 (2014) 3544

Fig. 7. Spanwise C l , drag polars, and twist distributions from scheme B compared
with those from scheme A for C L = 0.3 and C L = 0.7. Flap angles are zero for both
schemes.

Fig. 8. Spanwise C l , drag polars, ap angles, and twist distributions from scheme C
for C L of 0.3 and 0.7. Spanwise C l distributions from scheme A are shown for comparison.

4.3. Scheme C: Trim-constrained induced drag minimization using aps

variables for trim-constrained C D ind minimization. Twist variables

remain unchanged in this scheme.

Scheme C is similar to scheme B, except that ap angles on all

the surfaces are used as the optimization variables to minimize
conguration C D ind subject to a trim constraint about a specied
CG location using Eq. (15). The twist variables are kept unaltered.
4.4. Scheme D: Trim-constrained induced drag minimization using aps
with wing mean ap set for prole drag reduction
In scheme D, Eq. (17) is used to set the wings mean ap for the
desired C L to enable operation of the wing within the drag bucket.
With the mean ap thus xed, the remaining aps are used as

4.5. Scheme E: Trim-constrained total drag reduction using constrained

nonlinear optimization
Schemes C and D are designed to minimize one of the two
components of total drag, induced (scheme C) or prole (scheme
D), while leaving the other component free to enable meeting the
trim constraint. Decomposition of the ap angles into mean and
variation ap angles enables the use of well-posed optimization
problems for these two schemes with closed-form solutions for
the ap angles. However, they do not guarantee that total drag is

A.A. Cusher, A. Gopalarathnam / Aerospace Science and Technology 34 (2014) 3544

41

Fig. 9. Comparison of span eciency, induced, prole, and total C D for the three-surface conguration from schemes C, D, and E.

in fact minimized. Total drag, dened as the sum of induced and

prole drag, can suffer in either case because of requirements to
trim or owing to large ap angles.
Scheme E uses constrained nonlinear optimization to minimize
total C D at any given C L . Twist values are held constant, and the
ap angles are used as variables for the optimization. An objective
function is used, which outputs total C D for the conguration for
any given set of ap angles. In this function, C D ind is calculated using Eq. (3) and prole C D is calculated by integrating airfoil C d for
each section of the wing at the correct C l , ap angle and Reynolds
number. The C d for each section is determined using the drag polar approximations described in Section 3.6, rather than with direct
analysis using XFOIL or a similar code. This approach avoids several problems with the use of XFOIL in the optimization process:
increased computational time, possibility of non convergence of
XFOIL, and non-smooth C d variations from XFOIL for small perturbations to or ap angles. A constraint function is also used
in the optimization, which sets the trim constraint using Eqs. (8)
and (9).
The optimizer used in this work is the Matlab function fmincon
which nds minimum of a nonlinear multivariable function with
equality constraints. While this scheme, in contrast to schemes
C and D, enables minimization of total C D , it comes with standard limitations accompanying the use of such an optimizer. One
notable limitation is the lack of aerodynamic insight that results
when using an optimizer, as the solutions are largely produced by
a black box type routine. Secondly, there is no guarantee that the
optimizer has not converged to a local minimum, and the solution
can depend on the starting values for the aps. In the current research, this problem is mitigated by using the solutions from both
scheme C and scheme D as starting points for each C L , and choosing the one that results in lower C D as the nal solution.
5. Results
This section presents the results from the drag minimization
schemes of Section 4 applied to the example aircraft conguration shown in Fig. 1. With any given scheme, once the optimum

ap/twist variables were determined, the aerodynamics were determined from full analysis of the conguration. For this full analysis, the AVL vortex lattice method [12] was used to determine
the lift, induced drag, C l distributions, and pitching moment. The
XFOIL code [10] was then used several times to determine the
drag polars for the apped sections at the appropriate Reynolds
numbers. This information was used in spanwise integration to determine the prole drag. Thus, even though approximations (superposition approach for lift distributions and curve ts for the section
drag polars) were used in the drag reduction schemes for determining the optimum ap/twist variables, the results presented in
this section are from full analysis without those approximations.
The airfoils used on the conguration were a custom natural
laminar ow (NLF) airfoil on the wing, the MH201 airfoil [15] on
the canard, and an NACA 0012 airfoil on the tail. Information about
the wing and canard airfoils is presented in Fig. 4. Other aircraft
parameters include the aircraft weight ( W ) = 18 000 N, reference
chord (c ref ) = 3.9 ft, and reference area ( S ref ) = 177.9 ft2 .
5.1. Induced drag minimization: Comparison of results from schemes A,
B, and C
In schemes A, B, and C, induced drag is minimized without
any consideration given to prole drag. When scheme A is used,
there is no constraint on the pitching moment. Thus the method
in scheme A is free to choose the optimum distribution of lift between the three lifting surfaces. To achieve pitch trim with the lift
distribution resulting from scheme A, the CG has to be placed aft
of the neutral point, corresponding to a static margin of 35%, and
resulting in a statically unstable aircraft. When a pitch-trim constraint is enforced with scheme B for a static margin of 10%, there
is a very small increase in C D ind (of 0.0003) compared to that of
scheme A. This additional induced drag is associated with an upward loading on the canard. Fig. 5 shows the variation of C D ind and
span eciency with static margin for C L = 0.5.
Another approach is to take the conguration with xed twist
values, but use aps to minimize the induced drag using scheme C.

42

A.A. Cusher, A. Gopalarathnam / Aerospace Science and Technology 34 (2014) 3544

The advantage with aps, in comparison to twist, is that they can

be altered in ight to adapt the wings for the C L of the ight condition. Using a xed twist from scheme B for C L of 0.5, the ap
angles for minimum induced drag were determined for a range of
C L values from 0.1 to 0.9. Fig. 6 compares the variation of span
eciency, induced, prole and total drag coecients vs. C L for
the example aircraft from schemes A, B, and C. It is seen that all
three schemes result in nearly the same variation of C D ind with
C L . The improved span eciency at low C L with scheme A compared to the other two schemes does not translate to noticeable
improvement in C D ind because C D ind is small at these C L values. It is seen that prole drag considerations are important when
considering the three schemes. While scheme C shows benet at
low C L , this scheme also results in large prole drag penalties at
high C L .
The reasons for this behavior can be understood by examining Figs. 7 and 8. In Figs. 7(a) and (b), the spanwise C l distributions on the three surfaces from schemes A and B are compared
for C L of 0.3 and 0.7. Because the ap angles are all zero, there
is no change in the section pitching moments between the two
schemes. Thus the wing C l distributions from the two schemes are
nearly the same. To compensate for the more forward CG location in scheme B, the canard has a higher incidence angle and
higher loading, while the tail has a smaller incidence angle and
lower loading. These differences do not contribute to any signicant difference in either induced or prole C D over the entire C L
range.
In Figs. 8(a) and (b), the spanwise C l distributions on the three
surfaces from scheme C are compared with those from scheme A
for C L of 0.3 and 0.7. Scheme C uses the ap angles to achieve
minimum induced drag while satisfying pitch-trim constraint for
SM = 10%. The method in scheme C was able to nd ap angles resulting in nearly optimal lift distributions by making use of
ap-section pitching moments to achieve pitch trim. In doing so,
however, it is seen that a majority of the wing sections end up operating within the low-drag ranges for C L of 0.3, but well outside
the low-drag ranges for C L of 0.7. Thus, lower prole C D at low
C L and higher prole C D at high C L result with scheme C when
compared to schemes A and B, as seen in Fig. 6.
5.2. Prole drag considerations: Comparison of results from schemes C,
D, and E
Schemes D and E were developed to include prole drag considerations in the ap optimization. This is done in scheme D by
setting the wings mean ap angle to enable operation of most of
the wing within the low-drag range, and in scheme E by using
constrained nonlinear optimization to minimize total drag. Fig. 9
compares the variation of span eciency, induced, prole and total
drag coecients vs. C L for the example aircraft from schemes C, D,
and E. It is seen that all three schemes result in nearly the same
variation of C D ind with C L , although scheme C achieves the lowest C D ind for any C L . Schemes D and E achieve signicantly lower
prole C D than scheme C, resulting in signicantly lower total C D
than with scheme C at all C L values greater than 0.2. Although it
is only slightly better than scheme D by a maximum C D reduction
of 0.0006 at C L = 0.1, scheme E achieves the lowest total C D at
any C L .
The fact that scheme E, in comparison to all the schemes,
achieves the lowest total C D at any C L is only to be expected
because scheme E minimizes total drag. These results were computed using the full analysis without the approximations used
in the drag-reduction schemes. That this expected result was
achieved even when the input optimum ap/twist variables were
determined using the approximations made in the drag-reduction
schemes (superposition approach for the loadings in Section 2.1

Fig. 10. Spanwise C l , drag polars, ap angles, and twist distributions from scheme D
for C L of 0.3 and 0.7. Spanwise C l distributions from scheme C are shown for comparison.

and curve-ts for the drag polars in Section 3.6) supports the argument that those approximations are accurate enough to be used
in the optimization schemes.
The reasons for the lower prole C D with schemes D and E can
be seen by studying Figs. 10 and 11, which show the spanwise C l
distributions and section drag polars on the three surfaces for C L
of 0.3 and 0.7 from schemes D and E. For scheme D, it is seen from
Fig. 10 that adjusting the wings mean ap angle has successfully
resulted in the wing operating within the drag bucket at both lift
coecients. Thus, for a small penalty in induced drag compared
to scheme C, this scheme achieves a signicant benet in prole
drag. As seen from Fig. 11, the results from scheme E for the two
lift coecients are nearly the same as that from scheme D, except

A.A. Cusher, A. Gopalarathnam / Aerospace Science and Technology 34 (2014) 3544

43

method. For this example aircraft it is seen that when the surfaces
are twisted for minimum induced drag without any pitching moment constraint minimum induced drag occurs at a negative static
margin. Using twist to minimize induced drag with a trim constraint for a positive static margin results in only a slight increase
in induced drag. As opposed to twist which cannot be controlled
in ight, trailing-edge aps distributed along the span of the lifting surfaces may be used for in-ight adjustment of the spanwise
lift distribution to suit the ight condition. When aps are used
to minimize the induced drag with a trim constraint there may
be signicant prole drag penalties, as with the example conguration used in this paper. If the ap optimization is done with a
further constraint that the wing mean ap is set to operate the
wing sections within their low-drag ranges, then the small penalty
in induced drag that results is far outweighed by the prole drag
reduction. The results show the importance of considering all components of drag when designing adaptive lifting surfaces. With the
constrained minimization of total drag using a nonlinear optimization, the total drag was minimized, but the solution was found
to be sensitive to initial values used to start the minimization. In
the current work, the detrimental effect of initial-value sensitivity
was mitigated by using the ap angles from the other closed-form
methods as initial values.
The example airfoils, conguration geometries, ap numbers
and sizes used in this paper were chosen only to illustrate the
methods developed in this effort and are not intended as recommendations for any particular application. While there are certainly several other design considerations for multiple-surface aircraft, drag reduction is an important goal. The current method,
therefore, can form an important part of a designers toolbox for
complex congurations and for adaptive lifting surfaces.
References

Fig. 11. Spanwise C l , drag polars, ap angles, and twist distributions from scheme E
for C L of 0.3 and 0.7. Spanwise C l distributions from scheme C are shown for comparison.

that the total-drag minimization approach has adjusted some ap

angles to achieve a slight additional prole drag reduction.
6. Conclusions
A method has been presented for optimizing multiple surface
congurations for both trimmed induced and prole drag, with
design variables that can include incidence, twist, and ap angles.
The different implementations of the optimization formulation allow for the study of various ways to reduce drag and gain insight
into the aerodynamic design of multiple lifting surface congurations.
Results are presented for a three surface aircraft used as an
example in this paper for illustration of the capabilities of the

[1] I.H. Abbott, A.E. von Doenhoff, Theory of Wing Sections, Dover, New York, 1959.
[2] D. Althaus, Niedrig-geschwindigkeits-prole, Friedr. Vieweg & Sohn Verlagsgesellschaft mbH, Braunschweig/Wiesbaden, 1996.
[3] D. Althaus, F.X. Wortmann, Stuttgarter Prolkatalog I, Friedr. Vieweg & Sohn,
Braunschweig, 1981.
[4] R.F. Anderson, Determination of the characteristics of tapered wings, NACA
Rept. 572, 1936.
[5] S. Barbarino, O. Bilgen, R.M. Ajaj, M.I. Friswell, D.J. Inman, A review of morphing aircraft, Journal of Intelligent Material Systems and Structures 22 (2011)
823877.
[6] L.M.M. Boermans, A. van Garrel, Design and windtunnel test results of a
apped laminar ow airfoil for high-performance sailplane applications, in:
ICAS 94-5.4.3, 1994.
[7] A. Bolonkin, G.B. Gilyard, Estimated benets of variable-geometry wing camber
control for transport aircraft, NASA TM 1999-206586, 1999.
[8] A.A. Cusher, A design and analysis approach for drag reduction on aircraft with
adaptive lifting surfaces, Ph.D. thesis, North Carolina State University, 2008,
http://repository.lib.ncsu.edu/ir/handle/1840.16/5063.
[9] A.A. Cusher, A. Gopalarathnam, Drag reduction methodology for adaptive tailless aircraft, Journal of Aircraft 49 (2012) 161172.
[10] M. Drela, XFOIL: an analysis and design system for low Reynolds number airfoils, in: T.J. Mueller (Ed.), Low Reynolds Number Aerodynamics, in: Lecture
Notes in Engineering, vol. 54, Springer-Verlag, New York, 1989, pp. 112.
[11] M. Drela, Elements of airfoil design methodology, in: P.A. Henne (Ed.), Applied Computational Aerodynamics, in: Progress in Astronautics and Aeronautics, vol. 125, AIAA, Washington, DC, 1990, pp. 167189.
[12] M. Drela, H. Youngren, AVL 3.26 user primer, MIT Department of Aeronautics and Astronautics, Cambridge, MA 02139, http://web.mit.edu/drela/
Public/web/avl/, accessed 23 May 2013.
[13] K.H. Goodrich, S.M. Silwa, F.J. Lallman, A closed-form trim solution yielding
minimum trim drag for airplanes with multiple longitudinal-control effectors,
NASA TP 2907, 1989.
[14] A. Gopalarathnam, R.K. Norris, Ideal lift distributions and aps angles for adaptive wings, Journal of Aircraft 46 (2009) 562571.
[15] M. Hepperle, Design of two airfoils for a canard airplane, MH AeroTools, http://www.mh-aerotools.de/company/paper_3/yaka.html, accessed 23
May 2013.
[16] J. Katz, A. Plotkin, Low-Speed Aerodynamics, second edition, Cambridge
Aerospace Series, Cambridge University Press, Cambridge, UK, 2001.

44

A.A. Cusher, A. Gopalarathnam / Aerospace Science and Technology 34 (2014) 3544

[17] E.R. Kendall, The theoretical minimum induced drag of three-surface airplanes
in trim, Journal of Aircraft 22 (1985) 847854.
[18] I. Kroo, T. McGeer, Optimization of canard congurations, ICAS 82-1498, 1982.
[19] A.M. Kuethe, C.Y. Chow, Foundations of Aerodynamics: Bases of Aerodynamic
Design, fth edition, John Wiley and Sons, Inc., 1997.
[20] E. Laitone, Positive tail loads for minimum induced drag of subsonic aircraft,
Journal of Aircraft 15 (1978) 837842.
[21] E. Laitone, Prandtls biplane theory applied to canard and tandem aircraft, Journal of Aircraft 17 (1980) 233237.
[22] C.W. McAvoy, A. Gopalarathnam, Automated cruise ap for airfoil drag reduction over a large lift range, Journal of Aircraft 39 (2002) 981988.
[23] R.J. McGhee, J.K. Viken, W. Pfenninger, W.D. Beasley, W.D. Harvey, Experimental
results for a apped natural-laminar-ow airfoil with high lift/drag ratio, NASA
TM 85788, 1984.
[24] M. McLaughlin, Calculations and comparison with an ideal minimum of
trimmed drag for conventional and canard congurations having various levels
of static stability, NASA TN-D 8391, 1977.
[25] H.P. Monner, E. Breitbach, T. Bein, H. Hanselka, Design aspects of the adaptive
wing the elastic trailing edge and the local spoiler bump, The Aeronautical
Journal (2000) 8995.
[26] C. Ostowari, D. Naik, Experimental study of three-lifting-surface conguration,
Journal of Aircraft 25 (1988) 106112.

[27] W. Pfenninger, Investigation on reductions of friction on wings, in particular by

means of boundary layer suction, NACA TM 1181, 1947.
[28] W. Pfenninger, Experiments on a laminar suction airfoil of 17 per cent thickness, Journal of the Aeronautical Sciences (1949) 227236.
[29] L. Prandtl, Induced drag of multiplanes, NACA TN 182, 1924.
[30] A. Robinson, J.A. Laurmann, Wing Theory, Cambridge Aerospace Series, Cambridge University Press, Cambridge, UK, 1956.
[31] K. Rokhsaz, B.P. Selberg, Three-surface aircraft optimum vs typical, Journal of
Aircraft 26 (1989) 699704.
[32] D.M. Somers, Design and experimental results for a apped natural-laminarow airfoil for general aviation applications, NASA TP 1865, 1981.
[33] J.J. Spillman, The use of variable camber to reduce drag, weight and costs of
transport aircraft, Aeronautical Journal 96 (1992) 19.
[34] E. Stanewsky, Aerodynamic benets of adaptive wing technology, Aerospace
Science and Technology 4 (2000) 439452.
[35] S.V. Thornton, Reduction of structural loads using maneuver load control on
the advanced ghter technology integration (AFTI)/F-111 mission adaptive
wing, NASA TM 4526, 1993.
[36] J. Valasek (Ed.), Morphing Aerospace Vehicles and Structures, John Wiley &
Sons, 2012.
[37] A.M. Wickenheiser, E. Garcia, Extended nonlinear lifting-line method for aerodynamic modeling of recongurable aircraft, Journal of Aircraft 48 (2011)
18121817.