CEAS Aeronautical Journal

[Link]

ORIGINAL PAPER

Control surface allocation based on offline handling quality

simulations for a flying wing aircraft
Salvatore Asaro1 · Direnc Atmaca1 · Erik‑Jan van Kampen1 · Roelof Vos1

Received: 19 June 2025 / Revised: 22 August 2025 / Accepted: 1 October 2025

© The Author(s) 2025

Abstract
Commercial applications of flying wing aircraft, such as the Flying-V considered herein, can contribute to reducing carbon
and nitrogen emissions produced by the aviation sector. However, because of the lack of a tail, all flying wing aircraft have
reduced controllability. For this reason, the placement and sizing of the control surfaces along the wing is a nontrivial problem.
The paper focuses on solving this problem using offline handling quality simulations based on certification requirements. In
different flight conditions, the aircraft must be able to perform a set of maneuvers as defined by the certification specifications.
First, offline simulations calculate the minimum control authority required from the elevator, aileron, and rudder to perform
each maneuver. Then, based on the global minimum for all maneuvers, the control surfaces are sized and placed along the
wings. The aerodynamic model employed uses a combination of Reynolds-averaged Navier–Stokes (RANS) and vortex lattice
method (VLM) simulations. The control authority of the control surfaces is estimated with VLM and VLM calibrated with
RANS simulations, showing significant differences between the two.

Keywords Flying wing · Control surface sizing · Offline handling quality simulations

List of symbols I moment of inertia, kg⋅m2

Ax , Ay , Az acceleration along X, Y, Z axis, m/s2 J cost function
b wing span, m K gain
c mean aerodynamic chord, m m aircraft mass, kg
CG center of gravity, m n number of peaks
CX , CY , CZ force coefficient along x, y, z M∞ freestream Mach number
Cl , Cm , Cn moment coefficient around X, Y, Z p, q, r angular velocity around x, y, z, rad/s
Cl,𝛿a rolling moment coef. due to 𝛿a pt,∞ freestream total pressure, Pa
Cm,𝛿e pitching moment coef. due to 𝛿e p̂ , q̂ , r̂ dimensionless p, q r
Cn,𝛿r yawing moment coef. due to 𝛿r ṗ , q̇ , ṙ angular acceleration around X, Y, Z rad/s2
g gravitational acceleration, m/s2 Rec mean chord Reynolds number
h altitude, m s Laplace variable (for continuous-time
H transfer function systems)
t time, s
T thrust, N
* Salvatore Asaro
[Link]@[Link] Tt,∞ freestream total temperature, K
Uapproach approach velocity, m/s
Direnc Atmaca
[Link]@[Link] Udes desired freestream velocity, m/s
U∞ freestream velocity, m/s
Erik‑Jan van Kampen
[Link]@[Link] Urotation rotation velocity, m/s
Ustall stall velocity, m/s
Roelof Vos
[Link]@[Link] Utrim freestream trim velocity, m/s
x axis and coordinate along chord direction, m
1
Faculty of Aerospace Engineering, Delft University y axis and coordinate along span direction, m
of Technology, Kluyverweg 1, Delft 2629HS,
The Netherlands

Vol.:(0123456789)
 S. Asaro et al.

z axis and coordinate along vertical direction, 1 Introduction

m
xhinge hinge line The Flightpath 2050 report issued by the Advisory Council
𝛼 angle of attack, ◦ for Aeronautics Research in Europe (ACARE) sets an ambi-
𝛽 sideslip angle, ◦ tious goal of reducing greenhouse gas emissions [5]. Particu-
𝛿a , 𝛿e, 𝛿r aileron, elevator, rudder deflection angle, ◦ larly, a reduction of 75% for CO2 and 90% for NOx is defined
𝛿f high-lift flap deflection angle, ◦ as targets with respect to conventional aircraft from 2000s.
𝛿̇ control surface deflection rate, ◦/s Novel aircraft designs, such as flying wing aircraft, could
𝜔n natural frequency, rad/s reduce drag, contributing to reducing the aviation sector’s
𝜌∞ freestream density, kg/m3 footprint. Several flying wing designs are being investigated
𝜙 bank angle, ◦ [4]; most of them feature highly swept inner wings to replace
𝜃 pitch angle, ◦ the fuselage, such as the Flying-V [3]. An artist’s impression
𝜁 damping ratio of the Flying-V is depicted in Fig. 1a and a top view with the
𝜔 natural frequency, rad/s cabin layout in Fig. 1b. The airplane has an inner wing with
aft aft CG location a leading-edge sweep angle of 64◦ and an outer wing with a
act actuator leading-edge sweep angle of 39◦ . The wing span is b = 65
eng engine m, and the mean aerodynamic chord c = 18 m. The current
forw forward CG location design, without engines, landing gear fairing, or roughness,
BPR Bypass ratio shows a maximum lift-to-drag ratio of 24.2 in cruise condi-
CG Center of Gravity tions, with angle of attack 𝛼 = 3.6◦ and freestream Mach
CV VLM calibrated with experiment number M∞ = 0.85 [11].
HQ handling quality The initial design of the airplane features an elevon
MTOW Maximum take-off weight, N along the trailing edge of the outer wing that acts as
MLW Maximum landing weight, N both elevator and aileron for pitch and roll control,
OEW Operative empty weight, N respectively. The yaw motion is controlled with rudders
RANS, R Reynolds-averaged Navier-Stokes spanning the height of the winglets. The current design
SLS Sea-level static of the control surfaces has been tested by pilots in a
SQP Sequential quadratic programming simulator [2], in which the aerodynamic model is used to
VLM Vortex Lattice Method describe the aircraft relied solely on vortex lattice method
simulations combined with wind tunnel data to partially
consider the pitch break occurring at an angle of attack of
approximately 20◦ . The tests showed an unstable Dutch
roll response of the aircraft, and to test the requirements
outlined by EASA CS-25 [7] the lateral-directional
dynamics were decoupled from the longitudinal dynamics.

Fig. 1  Artistic impression and top view of Flying-V aircraft

Control surface allocation based on offline handling quality simulations for a flying wing…

The Dutch roll response of the aircraft is discussed in Their combination defines the control surface layout for the
Sect. 2.6. Flying-V aircraft.
Furthermore, the simulator tests featured various
maneuvers as the ones tested in this work and outlined
in Sect. 2.1. The tests showed a lack of rudder’s control 2 Methodology
authority and some slight pitch dropout during the
bank-to-bank maneuver due to elevator saturation. Most The control surfaces are sized according to the maneuvers
longitudinal results were between level 1 and level 2. This described in Sect. 2.1. The maneuvers are simulated in a
classification is achieved by converting Cooper–Harper framework in which the aerodynamic model (Section 2.2)
ratings into handling quality levels from 1 to 3, as outlined and the rigid-body dynamics of the aircraft (Section 2.3)
in the equivalency chart defined in the military standard are integrated. For each maneuver, simulations are run
1797 A [14]. Overall, the simulation tests showed that iteratively to identify a range of control surface derivatives
the design of control surfaces should be improved to that can satisfy its requirements. The minimum control
meet all certification requirements, and a more complex surface derivative, in absolute value, is then selected as
aerodynamic model would be required to identify the sizing case and is used to find a control surface on
nonlinearities occurring in the various flight phases. the aircraft that can provide the requested control surface
Conventional aircraft employ the aileron, elevator, and derivative. The same process is then repeated for all the
rudder to control and maneuver the aircraft. However, selected maneuvers, leading to the new control surface
alternative control devices could serve similar purposes, layout.
potentially being more effective for aircraft like the
Flying-V. For example, split drag-rudders can create a 2.1 Criteria for sizing control surfaces
yawing moment by creating a drag difference between the
left wing and the right wing [17] - [20]; spoilers located The aileron, the elevator, and the rudder control the moment
on the inner part of the wing can also induce yawing and around the body x-, y-, and z-axes, respectively. The
rolling moments [16]. To limit the scope of this work, only moments around the three axes are here indicated as Cl , Cm ,
aileron, elevator, and rudder are considered in this study. and Cn . The body axis reference frame has its origin at the
However, if certification requirements cannot be met, it is Center of Gravity (CG) which is defined in Sect. 2.3.
stipulated whether these alternative control surfaces could In the certification documentation [7], a series of criteria
potentially be used to solve this problem. are defined that allow the sizing of the control surface
In this study, the maneuvers are simulated offline, required for each function. Particularly, the elevator for
i.e., not in the flight simulator. This allows the testing pitch control, the aileron for roll control, and the rudder
of new control surface layouts with low workload and for yaw control. However, the control surface on the
computational efforts. Previous studies have sized trailing-edge area can potentially act as both elevator and
control surfaces similarly based on offline handling aileron, i.e., an elevon. To discriminate between high-rate
quality evaluations in unconventional aircraft [6]-[19]. A and low-rate actuators which have a direct impact on the
similar approach is also pursued in this paper, with an power consumption and the weight, the elevon is sized to
aerodynamic model of the aircraft which incorporates provide the required roll control, and if the elevon cannot
Reynolds-averaged Navier–Stokes (RANS) simulations provide enough pitch control an additional control surface
and hence nonlinear effects in the early phase of the is designed to serve as a pure elevator and it is located more
control surface sizing process [1]. inboard than the elevon.
The contribution of this paper is to size, position, and Herein, the functions used to size the different control
define the control surfaces required for a Flying-V aircraft. surfaces are summarized:
The control surface layout is driven by the maneuvers
that the Flying-V must be able to perform inside its flight 2.1.1 Elevator
envelope. The maneuvers are defined by handling quality
requirements, and compliance is demonstrated through • Longitudinal trim: the elevon and the elevator are sized
offline simulations. to longitudinally trim the aircraft. The procedure to
The aerodynamic model of the Flying-V is introduced distinguish between elevon and elevator is discussed
in Sect. 2.2, together with the flight conditions considered. in detail in Sect. 3.1. The longitudinal trim procedure
The maneuvers tested are presented in Sect. 2.1. For each allows determining the required angle of attack 𝛼 and
maneuver, the required control authority of the elevator, the thrust T at a certain flight condition. The identified
aileron, and the rudder is estimated, and then, the control longitudinal trim condition is then used as the starting
surfaces are each sized and placed on the wing accordingly. point for the remaining cases. Furthermore, the thrust
 S. Asaro et al.

lapse with the operating conditions is estimated for the 2.1.3 Aileron

engine considered in this work, for which the sea-level
static thrust-to-weight ratio is TSLS ∕WMTOW = 0.27. A • Bank-to-bank: The aircraft is required to bank from −30◦
detailed description of the engine can be found in [10]. to 30◦ in 7 s or less. During this maneuver, the flight path
The thrust is assumed to scale with the Mach number and angle should be kept between 0 ◦ and 5 ◦.
the altitude according to [13] as
pt,∞ � �
Additionally, the aileron is also used to compensate for the
√
Tmax = TSLS 1 − (0.43 + 0.014 ⋅ BPR) M∞ , (1)
p 0 rolling moment occurring during the one-engine inoperative
where pt,∞ is the freestream total pressure at cruise and steady-heading-sideslip maneuvers.
altitude, p0 the sea-level pressure of 101 kPa, and
BPR = 10 is the bypass ratio of the engine. For the 2.1.4 Coordinated turn
operating conditions considered in this paper, Tt,∞/T0 is
always lower than 1.06, and hence, Eq. 1 is sufficient After sizing the control surfaces, a coordinated turn is
to describe the thrust lapse. The total pressure and considered to combine longitudinal and lateral–directional
temperature are calculated with the isentropic equations. motions. This maneuver allows testing the combination of the
The influence of the CG on the control surface size is three control surfaces, i.e., studying the aircraft’s capability
described in Section 2.3 and the trimming algorithm is to coordinate yaw, pitch, and roll simultaneously. A turn is
discussed in Section 2.5. performed at a constant speed while minimizing the sideslip
• Pull-up and push-over maneuver: the aircraft has to angle. The requirement is to perform the maneuver for 10 s
pull-up to at least 1.5g and push-over to 0.5g at the while keeping the bank angle between 40◦ and 45◦ and the
different flight conditions. The limits are reduced to 1.3g flight path angle between 0◦ and 5 ◦ , to prevent altitude loss.
and 0.5g at landing speed. In the worst case scenario with a flight path angle of 5 ◦ and
• Take-off rotation: the aircraft is supposed to provide the highest speed tested in this work (Table 3), the change in
a rotation rate of 3 ◦ /s at rotation speed and Maximum altitude corresponds to 219 m which does not invalidate the
Take-off Weight. operating condition tested because of the limited change in the
aerodynamic coefficients.

2.1.2 Rudder 2.2 Aerodynamic model

• Steady-heading sideslip: The aircraft should keep a The parameters influencing the aerodynamic forces and
constant heading while facing a certain sideslip angle moments are determined with a combination of Reynolds-
𝛽 . This maneuver assesses the capabilities of the aircraft averaged Navier–Stokes (RANS) and Vortex Lattice Method
during crosswind. The certification documentation EASA (VLM) simulations. These models are thoroughly discussed in
CS-25 (AMC 25.177(c)) [7] defines 𝛽 = asin(30∕U∞ ), [1]. A summary of the parameters that influence the forces and
where the velocity is expressed as calibrated airspeed in moments acting on the aircraft is presented in Table 2, which
knots. The sideslip values in absolute value are listed in also indicates if the parameter is determined with RANS (R) or
Table 1 for the flight conditions considered here. VLM (V). The variables in this table are the angle of attack (𝛼),
• One-engine-inoperative: A second function of the rudder the sideslip angle (𝛽 ), the elevator deflection (𝛿e), the aileron
is to compensate the Cn induced by one engine when the deflection (𝛿a), the rudder deflection (𝛿r), and the high/lift flap
other engine fails. Although the precise location of the deflection (𝛿f ). Furthermore, it contains the dimensionless
engines is still under investigation [10], it is assumed to angular rates, i.e. the roll rate ( p̂ = pb∕(2U∞ )), the pitch rate
be located at y = 5.7 m, z = 0 m. Furthermore, it is tested (q̂ = qc∕(2U∞ )), and the yaw rate (r̂ = rb∕(2U∞ )), where p,
the one-engine-inoperative case in combination with the q, and r are the angular rates with respect to the x-, y-, and
crosswind cases. z-axes, respectively.
The RANS simulations determine the forces and moments
generated by a single control surface design for the aileron,
elevator, high-lift flap, and rudder. The results are then used to

Table 1  Sideslip angle 𝛽 at the

considered flight conditions M∞ 0.2 0.25 0.3 0.4 0.6 0.7 0.85
𝛽 13.1 ◦
10.5 ◦
8.7 ◦
8.9 ◦
7.0 ◦ 6.9◦ 6.2◦
Control surface allocation based on offline handling quality simulations for a flying wing…

Table 2  Force and moment 𝛼 𝛽(𝛼) p̂ q̂ r̂ 𝛿e (𝛼) 𝛿a (𝛼) 𝛿r (𝛼) 𝛿f (𝛼)

coefficient dependencies,
marked with “R” when Cx R V-R V-R V-R V-R
determined through RANS
Cy R V V-R V-R
simulations and with “V” when
determined through VLM Cz R V V-R V-R
simulations Cl R V V V-R V-R
Cm R V V-R V-R
Cn R V V V-R V-R

Table 3  Flight conditions h/m U∞/(m/s)

𝜌∞/(kg/m 3) M∞ Rec/10 7 m/(10 3 kg)

MLW 0 1.22 68 0.20 8.4 202

0 1.22 85.1 0.25 10 202
0 1.22 102.1 0.30 13 202
MTOW 0 1.22 102.1 0.30 13 266
5450 0.70 127.5 0.40 10 266
7650 0.55 185.7 0.60 12 266
9750 0.43 210.4 0.70 11 266
11225 0.35 250.8 0.85 11 266

calibrate the VLM simulations, which are less time-consum-

ing. For this reason, the results presented in Sect. 3, are divided
into "VLM" and "VLM calibrated with RANS". More details
on the control surfaces simulated with RANS are introduced
in [1].
The maneuvers, presented in Section 2.1, are tested at
several flight conditions, as summarized in Table 3. The
selected operating conditions are chosen to represent differ-
ent phases of the flight envelope. The different conditions are
divided into two groups, those at maximum landing weight
(MLW) where mMLW = 202⋅103 kg, and those at maximum
take-off weight (MTOW) where mMTOW = 266⋅103 kg. The
masses are derived from [15], assuming a family-optimized
Flying-V-1000 aircraft. The relationship between the CG
range and the weights is depicted in Fig. 2. The control sur-
face sizing is conducted for the most forward CG location Fig. 2  CG range of the aircraft and CG locations highlighted
of the aircraft indicated as forw0 . A detailed description of
the CG diagram can be found in [1].
As discussed above, three different types of control Table 4  Control surfaces' 𝛿/◦ 𝛿̇ /(◦/s)
surfaces are considered: elevator, elevon, and rudder. characteristics
Table 4 indicates their maximum and minimum deflection Elevator ± 25 40
angles and maximum deflection rates, based on [9]. The Aileron/elevon ± 25 55
indicated limits are used during the simulations. However, Rudder ± 25 50
new maximum and minimum deflections are suggested in
Section 3 to satisfy specific requirements when needed.
on five assumptions: the aircraft is rigid and has a constant
2.3 Rigid‑body dynamics mass m, the Earth is flat and non-rotating, the gravitational
acceleration g is constant, the aircraft has a plane of symmetry
The aerodynamic force and moment coefficients are expressed in the body-fixed reference frame which leads to Ixy = Iyz = 0,
in the body reference frame and converted into linear and rota- and finally, when both engines are operative, the resultant
tional accelerations. The system of equations is derived based thrust vector is on the symmetry plane and hence only affects
 S. Asaro et al.

the force along x. Based on these assumptions, the system of The inertia terms are calculated by dividing the aircraft
equations can be expressed as follows: mass into 100 point masses along the span, being located at
the mid-chord of each section. The masses distribution along
Cx
Ax = − g sin 𝜃 the aircraft is detailed in [18], where higher masses are used
m in the inner zone to indicate the passenger and cargo contri-
Cy
Ay = + g sin 𝜙 cos 𝜃 bution as in Fig. 1b. The mass distribution is then adjusted
m to obtain the required CG location, by employing a pump-
Cz ing system to distribute the kerosene between the different
Az = + g cos 𝜃 cos 𝜙
m fuel tanks [18]. As expected for both weights, when moving
from forward to aft CG location, the inertia term increases
( )
Izz Ixz Ixx − Iyy + Izz Ixz
ṗ = ∗ Cl + ∗ Cn + pq
I ( I I∗ in absolute value. The increase is higher for the MLW case,
(2) because the CG range is higher than at MTOW.
2
)
(Iyy − Izz )Izz − Ixz
+ qr The impact of the moments and products of inertia on
I∗
C I (r2 − p2 ) Izz − Ixx the Dutch roll characteristics of the aircraft is considered
q̇ = m + xz + pr in Section 2.5. For the control surface sizing in Section 3,
Iyy Iyy Iyy
( ) initial tests show a limited impact of the inertia, with the
Ixz Ixx −Ixx + Iyy − Izz Ixz CG location being the main driver. For a more conservative
ṙ = ∗ Cl + ∗ Cn + qr
I ( I I∗ design, the most forward CG location of the aircraft at any
weight (forw0 ) is considered for the sizing process of the
2
)
(Ixx − Iyy )Ixx + Ixz
+
I∗
pq, control surfaces.

where I ∗ = Ixx Izz − Ixz

2
. All linear and angular body velocities 2.4 Actuator and engine dynamics
can be obtained by integrating these equations. In addition,
the kinematic equations that allow converting body The control surfaces or actuators (act) are modeled as a
velocities into attitude angles can be expressed as follows: second-order system based on [12], and the engines (eng)
𝜙̇ = p + q sin 𝜙 tan 𝜃 + r cos 𝜙 tan 𝜃 are modeled as a first-order system. Their transfer functions
are as follows:
𝜃̇ = q cos 𝜙 − r sin 𝜙
(3)
sin 𝜙 cos 𝜙 4000 𝜔2n
𝜓̇ = q +r . Hact (s) = =
cos 𝜃 cos 𝜃 s2 + 140s + 4000 s2 + 2𝜁𝜔n s + 𝜔2n
(4)
The moments and products of inertia of the aircraft used 1
and Heng (s) = .
in Eq. 2 are summarized in Table 5. The inertia terms are 0.2s + 1
calculated at maximum take-off (MTOW) and maximum For the actuators, this leads to a natural frequency
landing (MLW) weights, for each weight they are calculated 𝜔n = 63.25 rad/s and a damping ratio 𝜁 = 1.11.
at the most forward (forw1 for MLW and forw2 for MTOW)
and aft Center of Gravity (CG) positions. The location of the
2.5 Simulation model
different CG considered are summarized in Fig. 2. Because
of the presence of the plane of symmetry, Ixy = Iyz = 0.
Figure 3 shows the layout of the simulation model, which
encapsulates the aerodynamic model, the rigid-body dynam-
ics, and the control law used to command the aircraft.
Based on the figure, two types of control laws act as
Table 5  The moments and products of inertia at different aircraft bridges between the pilot commands and control surface
weights and CG locations deflections. In direct law, the pilot inputs are manually
Ixx/ Iyy/ Izz/ Ixz/ (Ixy = Iyz)/ mapped over the control surfaces, such that the pilot’s
(kg ⋅ m 2 ) (kg ⋅ m 2 ) (kg ⋅ m 2 ) (kg ⋅ m 2 ) (kg ⋅ m 2 ) sidestick and pedal inputs directly command the con-
trol surfaces. Normally, this would be sufficient to study
MLW forw1 3.4 ⋅ 10 7 2.9 ⋅ 10 7 6.2 ⋅ 10 7 −0.09 ⋅ 10 7 0 the aircraft’s handling qualities in a real-time simulation
CG
environment. However, since simulations are conducted
MLW aft CG 4.1 ⋅ 10 7 3.2 ⋅ 10 7 7.2 ⋅ 10 7 −0.11 ⋅ 10 7 0 offline, a basic automatic control augmentation is neces-
MTOW forw 2 4.8 ⋅ 10 7 3.9 ⋅ 10 7 8.7 ⋅ 10 7 −0.13 ⋅ 10 7 0 sary to perform maneuvers without a pilot. It is designed
CG
to have a proportional-derivative structure. The control
MTOW aft CG 5.3 ⋅ 10 7 4.2 ⋅ 10 7 9.5 ⋅ 10 7 −0.14 ⋅ 10 7 0
deflections computed through either direct law or basic
Control surface allocation based on offline handling quality simulations for a flying wing…

Fig. 3  Model structure for

offline simulations

to minimize the difference between the desired airspeed

(Udes ) and the trim airspeed (Utrim ). This cost function is
given as follows:

J(u,
̇ w,
̇ q, ̇ 𝛾)
̇ V, ̇ = 100 ⋅ u̇ 2 + 1 ⋅ ẇ 2
(5)
+ 100 ⋅ q̇ 2 + 100 ⋅ (Udes − Utrim )2 + 100 ⋅ 𝛾 2 .
Fig. 4  PD controller used in basic augmentation The variables of this cost function can all be expressed in
terms of aerodynamic forces and moments using Equation 2.
augmentation serve as inputs to the aerodynamic model Hence, the problem of minimizing the accelerations
to calculate the forces and moments acting on the aircraft. indirectly becomes a net force/moment minimization
These are then used in dynamical equations given in Equa- problem. Given that longitudinal forces and moments are
tion 2. The final states of the aircraft are calculated by related to elevator deflection 𝛿e, angle of attack 𝛼 , and engine
integrating the linear and rotational accelerations of the thrust T, the minimization algorithm tries to iteratively
aircraft, and using the kinematic transformations given in change these parameters to find the global minimum of the
Equation 3. The structure of the PD controller is shown cost function. The algorithm is implemented in Matlab using
in Fig. 4. the "fmincon" function, employing a sequential quadratic
When the controller is active, it is applied only to the programming (SQP) approach.
roll and sideslip channels. Several maneuvers in this study
require the aircraft to reach and maintain a specific sideslip 2.6 Dutch roll analysis
and/or bank angle. The pitch channel uses the direct law
in all cases, controlling the symmetric deflection of the In a recent simulator flight test conducted on an aerodynamic
elevons, which corresponds to an elevator input. The model of the Flying-V which relied only on VLM simula-
resulting motion causes changes in the angle of attack, tions, the open loop Dutch roll is found to be unstable [2].
pitch angle, flight path angle, and altitude. The gains of the Consequently, this is a point of concern for the new aero-
controller are obtained by trial and error, separately for dynamic model used in this study. For this reason, as a pre-
roll and sideslip, resulting in Kp𝛽 = Kd𝛽 = Kp𝜙 = 5000 and check before running the actual sizing simulations, an offline
Kd𝜙 = 1000 . The trial-and-error process aims to find a fast Dutch roll simulation is conducted for all flight conditions
control response without excessive overshoot. Although using the direct law. Control augmentation for the Dutch
the gains might seem high at first glance, this is because roll is avoided to prevent masking the fundamental response
the sideslip and beta are defined in radians, whereas the of the aircraft. To this end, the nonlinear simulation model
aileron and rudder deflections are in degrees. is excited using a doublet rudder input and then allowed to
For the sizing maneuvers explained in Section 2.1, evolve without interference to observe whether the sideslip
the controller is active for steady-heading sideslip, one- oscillations dampen out over time. In Dutch roll simula-
engine-inoperative, bank-to-bank, and coordinated turn. tions, the aircraft is initialized from a trimmed steady-level
The longitudinal trim condition is calculated statically condition, where it stays for 10 s until the rudder doublet
using a cost function and minimization algorithm. The excites the lateral–directional dynamics and the simulation
direct law is used without any control augmentation to test is then run for 300 s. Figure 5a shows the rudder input vary-
the calculated trim point in a steady-level flight scenario. ing between 𝛿r = ± 30◦ , and the time histories of the sideslip
The aim of the cost function and the minimization angles at two different operating conditions, M∞ = 0.25 and
algorithm is to find a steady state for the aircraft such that MLW, and M∞ = 0.85 and MTOW. For the sake of demon-
the longitudinal acceleration terms, u̇ , ẇ , q̇ and the flight stration, the figure discards the first 10 s of the simulation
path angle 𝛾 are zero. In addition, the cost function seeks where the steady flight takes place. Each condition is tested
at two CG locations with their related inertia presented in
 S. Asaro et al.

Fig. 5  Dutch roll simulation results and damping–frequency product

Table 5. Moving from the forward to the aft CG locations at linearized aircraft models. However, one could argue that
both M∞, the frequency increases and the oscillations damp using a nonlinear simulation model is more accurate and
out more slowly. To analyze these time histories in more allows for the possibility of revealing undesired phenom-
detail, the logarithmic decrement method is used to calcu- ena. Nevertheless, due to the nonlinear nature of the model,
late the natural frequency 𝜔d and the damping ratio 𝜁d of the the handling quality ratings for the Dutch roll serve as an
oscillations, which are given as follows: approximation.
( n ) Using Equations 6 and 7, and comparing the results with
2𝜋 ∑ the Dutch roll requirements, it can be observed that the
𝜔d = t −t (6)
n k=0 k+1 k Dutch roll is stable and damped under all flight conditions.
The results are summarized in Fig. 5b and Table 7. The
( n ) figure only depicts the damping-frequency product, because
1 ∑ x(tk ) it serves as the primary driving factor for the Handling
𝜁d = log , (7)
n k=0
x(tk+1 ) Qualities (HQ) classifications in this application. For M∞ ≥
0.3 and MTOW, Fig. 5b shows that the Dutch roll is within
where n is the number of peaks, x is the amplitude of the Level 2. At MLW, the Dutch roll is at Level 1 for M∞ = 0.25
peaks, and tk is the time at which the peaks occur. Hence, the and 0.3, and at Level 2 for M∞ = 0.2. As will be discussed in
damping ratio is calculated by taking the sum of consecutive Section 3, at M∞ = 0.2, a high-lift flap could be required, and
peaks and averaging them, whereas the frequency comes when simulating the Dutch roll of high-lift configuration,
from averaging the consecutive periods of oscillation. 𝜔d the HQ diminishes to Level 3 at the forward CG and drops
and 𝜁d are calculated considering the peaks with positive below 3 at the aft CG. Further investigations are underway
sideslip angle. to identify the source of the drop in HQ level.
The final values of 𝜁d and 𝜔d determine the handling quali- With MLW at the forward CG, the aircraft is more sta-
ties based on the MIL-STD-1797A standards [14]. Assum- ble than at the aft CG. Differently, the opposite occurs at
ing that the Flying-V model falls under class III aircraft, the MTOW with the exception of M∞ = 0.85. Table 7 shows
classifications in Table 6 are considered for the terminal and that for all the cases, 𝜁d decreases and vice versa 𝜔d increases
nonterminal flight phases. An important point to mention when moving from forward to aft CG. The main trend that
is that these Dutch roll classifications are mainly used for can be observed with increasing M∞ is that 𝜔d increases at

Table 6  Dutch roll Category C - terminal flight phases Category B - nonterminal flight phases
classifications based on MIL-
STD-1797A 𝜔d/(rad/s) 𝜁d 𝜁d 𝜔d/(rad/s) 𝜔d/(rad/s) 𝜁d 𝜁d 𝜔d/(rad/s)

Level 1 ≥ 0.4 ≥ 0.08 ≥ 0.10 ≥ 0.4 ≥ 0.08 ≥ 0.15

Level 2 ≥ 0.4 ≥ 0.02 ≥ 0.05 ≥ 0.4 ≥ 0.02 ≥ 0.05
Level 3 ≥ 0.4 ≥0 - ≥ 0.4 ≥0 -
Control surface allocation based on offline handling quality simulations for a flying wing…

Table 7  Dutch roll analysis and Flight conditions Flight phase CG 𝜔d/(rad/s) 𝜁d 𝜁d 𝜔d/(rad/s) HQ level
classification results
M = 0.2 (MLW) Terminal forw1 0.42 0.20 0.087 2
aft 0.47 0.17 0.081 2
M = 0.25 (MLW) Terminal forw1 0.45 0.27 0.125 1
aft 0.82 0.13 0.105 1
M = 0.3 (MLW) Terminal forw1 0.46 0.22 0.101 1
aft 0.90 0.11 0.100 1
M = 0.3 (MTOW) Nonterminal forw 2 0.49 0.18 0.089 2
aft 0.84 0.12 0.103 2
M = 0.4 (MTOW) Nonterminal forw 2 0.55 0.17 0.093 2
aft 0.80 0.12 0.100 2
M = 0.6 (MTOW) Nonterminal forw 2 0.69 0.13 0.091 2
aft 0.88 0.12 0.104 2
M = 0.7 (MTOW) Nonterminal forw 2 0.69 0.14 0.096 2
aft 0.81 0.13 0.102 2
M = 0.85 (MTOW) Nonterminal forw 2 0.70 0.17 0.116 2
aft 0.77 0.13 0.103 2

the aft CG location. Additionally, some differences in trends span and four along the chord. Initially, for the elevator siz-
can be noticed between subsonic and high subsonic/tran- ing, the most inboard sections are considered to leave the
sonic cases as it will also be seen in Section 3. outer sections to place the aileron. For example, the first
elevator considered is the most inboard panel with the hinge
line closer to the trailing edge, xhinge ∕c = 0.95. After that,
3 Results the hinge line is moved upstream, up to xhinge ∕c = 0.65. The
control authority of the four different described elevators
Section 3.1 introduces how the initial trim conditions are is shown in Fig. 6b as the blue line closer to zero in terms
obtained that are required for all subsequent simulations. of Cm,𝛿e . The same procedure is then applied when increas-
Section 3.2 presents the aileron dimensions and ing the number of panels along the span where each panel
deflection needed to satisfy the certification requirements. has an extension of 2.8%b, and with the simplification that
Subsequently, the sized aileron is also assigned a pitch the hinge line is fixed, to avoid steps in the control surface
control function, making it an elevon. If this elevon is or “hollowed” control surfaces. Figure 6b shows the con-
unable to satisfy a pitch control requirement by itself, an trol authority of the considered elevators, which are 32 in
additional elevator is sized next to the elevon on the inboard total, and Fig. 6c shows the corresponding deflection angle
side (Section 3.3). Section 3.4 presents the sizing of a required to satisfy the considered trim condition.
high-lift flap and discusses why it is needed at low speed. The black horizontal line in Fig. 6b indicates the mini-
Section 3.5 investigates the dimension and deflection of the mum requirement in absolute value to satisfy the longitu-
rudder to satisfy the certification requirement. The newly dinal trim conditions. This requirement is determined in
defined control surfaces are tested with the coordinated turn the different flight conditions introduced in Table 3, with
in Section 3.6. Finally, Section 3.7 defines the stall speed a maximum possible deflection of 20◦ as shown in Fig. 6c,
and related speeds as the rotation speed, indicating which are the limit is lower than in Table 4 to avoid saturation of the
the maneuver that the aircraft can perform at these speeds. control surface for other maneuvers. The minimum require-
ment is determined by running ten simulations introduced
3.1 Initial trim conditions in Section 2.5 and selecting the smaller control authority
that can satisfy the deflection angle constraint. A similar
An elevator is initially sized to find longitudinal trim condi- procedure is then repeated with the other control surfaces
tions. The trim conditions are required to define the angle for the selected maneuvers.
of attack 𝛼 and the thrust T, which are needed for all the The output of the longitudinal trim simulations is also
remaining simulations. the angle of attack 𝛼 and the thrust T as depicted in Fig. 7a
Figure 6a shows the discretization of the outer wing trail- and b, respectively. As expected, at fixed weight, 𝛼 tends to
ing-edge area used to test different elevator designs. The decrease with M∞, because the dynamic pressure increases.
considered area is discretized into eight panels along the However, at MTOW, this is not the case between M∞ = 0.3
 S. Asaro et al.

Fig. 6  Discretization of the elevator in the trailing-edge area, and corresponding pitching moment coefficient and related deflection due to eleva-
tor for trimming

Fig. 7  Angle of attack and thrust required for longitudinal trim at various flight conditions and at xCG = forw0
Control surface allocation based on offline handling quality simulations for a flying wing…

and 0.4, because also the altitude is varied between the two The deflections required from the aileron for the bank-to-
cases, leading to a reduction of density, which is higher bank maneuver and steady-heading sideslip maneuver are
than the increase due to the freestream velocity squared. At shown in Fig. 10a and b, respectively. The results for the
MLW, also between M∞ = 0.2 and 0.25, 𝛼 increases, in this one-engine-inoperative condition are not included for the
case, because a high-lift flap is used as it will be discussed aileron, because the deflections are much lower than for the
in Section 3.4. depicted cases, being at most 𝛿a = 1.1◦ at M∞ = 0.2.
The thrust-to-weight ratio is shown in Fig. 7b, where the Figure 10a and b comprises the deflection required in the
weight is the maximum take-off weight (MTOW). T is the various flight conditions introduced in Table 3. Bold lines
required trim-thrust for level flight, i.e., to compensate the indicate MTOW and dashed lines MLW. Additionally, for all
component of the weight and the force along the x axis due the cases, the deflections are calculated with VLM and with
to 𝛼 and 𝛿e . The elevator considered here is the final one VLM calibrated with RANS, as introduced in Section 2.2.
sized in this work. Figure 7b considers the engine introduced Particularly, the aileron is sized so to obtain a maximum
in Section 2.1.1. The results show that the thrust available is deflection of 25◦ , which occurs at M∞ = 0.2, MLW, xCG =
at least 8% higher than the thrust required. forw0 , with a high-lift flap deflection of 𝛿f = 40◦ to satisfy
𝛼 and T are the initial conditions for the remaining the sideslip requirement. Further details on the high-lift flap
simulations and they are kept constant for all of them. can be found in Section 3.4.
The center of gravity (CG) used in all the considered The difference between the VLM and VLM calibrated
simulations is located at 28.1 m from the aircraft nose with RANS results is related to the different scaling applied
(Fig. 9) and indicated as forw0 (Fig. 2), corresponding to the to take into account the 𝛼 and Mach effects. For both VLM
most forward CG location considered for this aircraft. The and RANS simulations, the increase of 𝛼 leads to a decrease
combination of MTOW and the most forward CG location in control authority. Conversely, the increase in M∞ leads
of the aircraft represents an extreme combination that would to an increase in control authority for VLM due to the
likely not occur in reality, as it is shown in Fig. 2. compressibility correction, while for RANS, the change in
control authority is nonlinear due to the formation of shocks
and local flow separation [1]. As a result, the deflections for
3.2 Aileron sizing the aileron predicted by the VLM are between 31 and 44%
lower than those for the VLM-calibrated-with-RANS. This
As introduced in Section 2.1.3, the aileron is required for difference is mainly due to the higher reduction in control
three maneuvers. First, during the bank-to-bank maneuver, authority with 𝛼 as predicted by RANS.
where the aircraft has to bank from −30◦ to 30◦ in 7 s or less. It should be noted that, as introduced in Section 2.2, the
Second, to maintain a bank angle of ±2◦ during the steady- forces and moments due to 𝛼 and 𝛽(𝛼) are always calculated
heading sideslip maneuver. Finally, to compensate for the with RANS simulations. Hence, if the aerodynamic model
rolling moments induced by the rudders when the aircraft had been derived only with VLM simulations, the obtained
faces the one-engine-inoperative condition (Section 2.1.2). deflections could have been even further from the VLM-
In this scenario, the rudders help maintain a sideslip angle calibrated-with-RANS simulations.
of ±2◦ , while the ailerons ensure wings-level flight. As mentioned before, the most critical condition,
Starting from the trim conditions identified in Section 3.1, in terms of deflection, for the aileron occurs at M∞
the aileron is sized with a procedure similar to the initial = 0.2 during the approach; being for the sideslip
elevator sizing, but differently than for the elevator the maneuver 𝛿a = 25◦ , and for the bank-to-bank maneuver,
aileron sizing starts from the aircraft tip moving inboard 𝛿a ≃< spanclass =� crossLinkCiteEqu� > 19 < ∕span >◦ .
(Section 3.1). Figure 8a shows the discretization of the This is in line with the aileron deflection limits implemented
trailing-edge area, Fig. 8b shows the rolling moment in modern aircraft.
generated by each aileron considered, and Fig. 8c shows Up to M∞ = 0.4, the aileron deflection required during
the corresponding deflection required to satisfy the sideslip a steady-heading-sideslip maneuver is higher than the one
requirement at low speed which is the most demanding required for the bank-to-bank maneuver. Differently, for
for the aileron. The last two figures include the minimum M∞ ≥ 0.6 , the bank-to-bank maneuver requires higher
requirement in absolute value in term of Cl,𝛿a which deflections.
corresponds to a maximum deflection of 15◦.
The final aileron geometry considered is depicted in 3.3 Elevator sizing
Fig. 9, where the aileron is located in the outer part of the
aircraft. In the figure, the surface here sized is labeled as The aileron sized in Section 3.2 can also be employed as a
elevon, because it serves both as elevator and aileron, but pitch control device. Therefore, from now on, the aileron
its dimension is driven by the requirement for the aileron. is used as an elevon. However, the moment created by this
 S. Asaro et al.

Fig. 8  Discretization of the aileron in the trailing-edge area, and corresponding rolling moment coefficient and related deflection due to aileron
for sideslip requirement

Fig. 9  Control surfaces sized in this study positioned on the aircraft. The drag rudder is part of the elevon
Control surface allocation based on offline handling quality simulations for a flying wing…

Fig. 10  Maximum aileron deflections for bank-to-bank and steady-heading sideslip requirements at various flight conditions and at xCG = forw 0

maneuvers. The certification specifications require that they

should be performed at specified speeds, such as the rotation
speed and the landing speed, which will be discussed in
Section 3.7, together with the consequent requirements for
the elevator.

3.4 High‑lift flap

One of the challenges of the planform geometry of the

Flying-V is an unstable pitch break at high angle of attack.
This limits the effective maximum lift coefficient and
therefore determines the stall speed of the airplane. To
increase the lift coefficient without increasing the angle
of attack of stall, a trailing-edge flap is proposed over
the inboard wing of the Flying-V (see Fig. 9. The flap is
Fig. 11  Elevator and elevon trim deflection angles at various flight located inboard of the inner wing, and the position near the
conditions and at xCG = forw 0
supposed CG range minimally impacts the aircraft’s pitching
moment. The exact location of the surface along the span
elevon is not sufficient to trim the aircraft at low speed. and, therefore, the extension along the chord can depend on
Therefore, the elevator initially sized in Section 3.1 is the location of the engine, which is still under investigation
resized to provide the additional elevator power required and [10]. The high-lift surface has been simulated as a plain
it is located next to the elevon on the inboard side, keeping flap. However, due to the presence of the fuel tank [15] and
the same chord ratio at the interface between the two control to reduce interference with the flow at the engine, the high-
surfaces, as depicted in Fig. 9. lift flap would realistically be installed on the lower side
Figure 11 shows the deflection required by the selected of the aircraft in the form of a split flap, which is actuated
elevon and the elevator under different flight conditions. similarly to a spoiler. This split flap implementation has been
As expected, the figure shows that the elevator deflection experimentally characterized in [8], proving the ability to
requirement decreases with M∞. The maximum deflection increase the lift for a given angle of attack with minimum
occurs at M∞ = 0.2, MLW, xCG = forw0 , with a high-lift impact on the pitching moment.
flap deflection of 𝛿f = 40◦ (discussed in Section 3.4). Similar Figure 12 shows the effect of the high-lift flap deflected
to the aileron results, the elevator deflections predicted by over 40◦ on the effectiveness of the various control sur-
VLM are between 29% and 39% lower than those predicted faces. As can be seen, the required control surface deflec-
by VLM-calibrated-with-RANS simulations. tions decrease due to the deployment of the high-lift flap
As introduced in Section 2.1.1, the elevator is also because of the higher effectiveness of the control sur-
required for the pull-up, push-over, and the rotation faces at lower angles of attack. The high-lift condition
 S. Asaro et al.

3.5 Rudder sizing

As introduced in Section 3.5, the rudder is sized to comply

with the steady-heading sideslip and one-engine-inoperative
conditions.
The rudder is sized with a similar but simpler procedure
with respect to the the other control surfaces considered
so far. Figure 13a shows the discretization of the trailing-
edge area of the winglet where the span extension of the
different rudder is always the same and only the hinge line
varies. Figure 13b shows the yawing moment generated by
each rudder considered and the corresponding deflection
required to satisfy the one-engine-out requirement at low
speed which is the most demanding for the rudder. In this
case, Fig. 13b does not feature a maximum requirement in
Fig. 12  Impact of high-lift flap on the angle of attack and con-
trol surface deflections at M∞ = 0.2, h = 0 m, at xCG = forw 0 , and
terms of deflection or a minimum in term of coefficient,
W = WMLW because the maximum possible rudder is considered for the
continuation of the study, and the reason is further discussed
hereafter.
considered herein leads to a reduction in the required The sized rudder with xhinge ∕c = 0.5 is depicted in Fig. 9,
deflections of 25% for the elevator, 26% for the aileron, and the deflections required for the two maneuvers are
and 39% for the rudder. depicted in Fig. 14.
As done previously, the VLM is calibrated with RANS The maximum rudder deflection in Fig. 14a occurs at
simulations, leading to a flap deflection of ca. 40◦ to obtain M∞ = 0.25, being 22◦ , which is lower than what is typically
the results in Fig. 12. The forces and moment generated by achieved by rudders on vertical tails of modern transport
the high-lift flap lead to an 8.5◦ angle of attack decrease at aircraft. As for the other control surfaces, the VLM predicts
M∞ = 0.2 . The size of the flap and its deployment angle between 30% and 43% lower deflections than VLM
have been selected to obtain a 𝛿e < 25◦ for longitudinal calibrated with RANS. As explained in Section 2.2, the
trim and 𝛿a < 25◦ during the steady-heading sideslip force and moment coefficients induced by 𝛽(𝛼) are always
maneuver. In future studies, based on an updated engine determined with RANS simulations.
location, the high-lift flap will be design such as to create a Before proceeding, it should be noticed that the sideslip
zero-pitching moment flap at the most forward CG, leading requirements vary with flight conditions as described in
to a smaller requirement for the elevator/elevon. Section 3.5. In particular, the maximum value of 𝛽 decreases
with increasing freestream velocity and, therefore, with M∞.

Fig. 13  Discretization of the rudder in the winglet, and corresponding yawing moment coefficient and related deflection due to rudder for one-
engine-inoperative requirement
Control surface allocation based on offline handling quality simulations for a flying wing…

Fig. 14  Rudder deflections for sideslip and one-engine-inoperative requirements at various flight conditions and at xCG = forw 0

However, the required rudder deflection to compensate the On military airplanes, a combination of an aileron and drag
sideslip angle does not decrease with M∞, as depicted in rudder is sometimes referred to as a "splitteron". In the
Fig. 14a. The reason for this trend is related to two main current implementation, this surface acts as a drag rudder,
effects. First, at fixed 𝛼 and 𝛽 , the forces and moments elevator, and aileron at the same time. The drag rudder
acting on the aircraft increase with M∞ . However, this has a span of 0.05b. Supposing that the maximum rudder
effect is largely compensated by the lower required 𝛽 with deflection is 𝛿r = 30◦ , the drag rudder should be deflected
increasing M∞ (see Table 1). The second effect is related to at ca. 30◦ to compensate the one-engine-inoperative yawing
the difference in aerodynamic behavior below and above moment; at sea level and M∞ = 0.2 with a flap deflection
M∞ = 0.6, resulting from transonic effects. In fully subsonic of 𝛿f = 40◦, which leads to 𝛼 = 10.5◦ . Here, it is supposed
conditions and with fixed 𝛽 , Cn decreases with 𝛼 . Conversely, that the effects of the drag rudder and of the rudder are
in transonic conditions, yawing moment coefficient increases superposable; however, this hypothesis will be tested in
with 𝛼 . For this reason, in transonic conditions, the required future studies.
deflections increase (Fig. 14a). It should also be noted that When combining the sideslip and the one-engine-
the control authority of the rudder decreases with 𝛼 , which inoperative requirements, the maximum rudder deflections
explains the large peak in required rudder deflection at needed are occurring at M∞ = 0.2 and 0.25 amounting to
M∞ = 0.25. ca. 32◦ , supposing the use of the drag rudder at M∞ = 0.2.
Figure 14b depicts the rudder deflections for the one- Overall, these results show a different trend of the rudder
engine-inoperative requirement, showing for subsonic cases deflection with Mach number compared to those for the
higher rudder deflections than for the sideslip cases with the elevon/elevator and the aileron. For elevon/elevator and
exception of M∞ = 0.4. Differently, the rudder deflections aileron, the requirements in terms of deflection angles tend
are lower for the one-engine-inoperative requirement than to decrease with M∞. Conversely, the requirement for the
sideslip cases at higher M∞. Once again, the requirement at rudder increases with M∞ to satisfy the sideslip requirement,
M∞ = 0.2 is the highest and surpasses the limits considered while showing a similar trend to elevon/elevator and aileron
in Table 4. for the one-engine-inoperative condition.
The rudder requirement for the one-engine-inoperative
condition at M∞ = 0.2 of 𝛿r ≃ 40◦ is considered too large. 3.6 Coordinated turn
The rudder requirement could be reduced by employing
an additional control surface as the drag rudder. Here, the Based on the results shown in Section 2.1.1, the coordi-
drag rudder is not sized as the other control surfaces, but the nated turn is utilized as a verification maneuver to assess
dimension and deflection required to unload the rudder are the final sizing of the control surfaces. As mentioned in
estimated based on [20]. The drag rudder is located on the Section 2.1.4, executing this maneuver requires reaching
outer part of the elevon, as visible in Fig. 9. It comprises two and maintaining a bank angle 𝜙 between 40◦ and 45◦ while
surfaces that move as one and in combination with the rest keeping the sideslip angle 𝛽 as close to zero as possible,
of the elevon, but the two surfaces can open, inducing drag with a flight path angle 𝛾 between 0 ◦ and 5 ◦ . The maneu-
and hence a yawing moment when required at low speed. ver is considered successful if these margins are satisfied
 S. Asaro et al.

for 10 s. Figure 15a shows the time history of the required when using an aerodynamic model, which includes nonlin-
angles. The angles are nondimensionalized for ease of com- earities, such as the pitch break.
parison, where 𝛽 ∗ = 𝛽∕(max(𝛽) − min(𝛽)), 𝛾 ∗ = 𝛾∕max(𝛾) However, the offline simulations also allows to
and 𝜙∗ = 𝜙∕max(𝜙) . The max and min are the indicated determine speeds which are important to characterize the
margins above. aircraft. The rotation speed, approach speed, and landing
Figure 15a shows, for M∞ = 0.85, that with the layout speed are a function of the stall speed. The stall speed is
of the selected control surfaces, it is possible to perform a calculated here as the lowest speed at which the aircraft can
coordinated turn within the limits defined by the certification be trimmed. Hence, the stall speed considered here is not
specifications. Similar results are obtained in the other directly comparable with the stall speed of a conventional
operating conditions considered in this study (Table 3). aircraft, because the limit factor here is not the maximum lift
The most demanding requirements are for the elevon coefficient but the occurrence of the pitch break [1].
and elevator, which is shown in Fig. 15b. The required The stall speed is calculated at the two CG locations
deflections are higher than those for the longitudinal trim discussed in Section 2.3. Additionally, for each CG
in Section 3.3. In particular, as seen for all the cases so far, position, two cases are tested, with and without high-lift
the highest 𝛿e is required at M∞ = 0.2, leading to 𝛿e = −32◦ , flap, and finally, the calculations are conducted using MLW
which is higher than the deflections used in conventional and MTOW. The stall speeds for the different cases are
aircraft. summarized in Table 8.
Identifying the lowest speed at which the aircraft could
3.7 Discussion be trimmed corresponds to finding the highest CL before
the pitch break region. Table 8 shows that the stall speed
The study so far conducted has focused on operating condi- increases with weight and decreases when the high-lift
tions, which should fall inside the flight envelope, helping flap is employed, as expected. At MLW, moving from the
to analyze how the requirements for control surfaces change most forward to the most aft CG location, the stall speed

Fig. 15  Angles evolution for coordinated turn and elevator deflection for coordinated turn at various flight conditions and at xCG = forw 0

Table 8  Trimmed stall speed xCG 𝛿f 𝛿e Ustall/(m/s) Ustall/(knots) Mstall

with and without high-lift at
forwards and aft CG position MLW forw1 0◦ −9◦ 71 139 0.21
MLW aft 0 ◦
−4◦ 75 146 0.22
MLW forw1 40◦ −8 ◦ 68 132 0.2
MLW aft 40 ◦
−3 ◦
71 139 0.21
MTOW forw 2 0◦ −7◦ 82 159 0.24
MTOW aft 0◦ −4◦ 82 159 0.24
MTOW forw 2 20 ◦
−6 ◦
78 152 0.23
MTOW aft 20◦ −3◦ 78 152 0.23
Control surface allocation based on offline handling quality simulations for a flying wing…

increases. This is counterintuitive, but can be explained by rudder deflection is limited at 30◦ assuming that a drag rud-
the interaction between the elevator deflection and the stall der is added to increase the yawing moment further.
over the outboard wing. At forward CG, the elevator and Finally, it should be noted that because the elevon is
elevon are deflected trailing-edge up, which postpones stall used at the same time as aileron and elevator, when two
over the outer wing. Therefore, the pitch break occurs at a maneuvers are combined, such as during longitudinal trim
higher angle of attack, corresponding to a higher lift coef- and a bank-to-bank maneuver, the deflection of the elevon
ficient. At MTOW, the stall speed between the forward and would exceed 40◦ which is unrealistic. Apart from the
aft CG locations is shown to be the same. This is because the take-off rotation and the pull-up at approach speed cases,
difference in speed is lower than M∞ = 0.005, which is the the results are shown at stall speed. However, the approach
interval in Mach number for which the Mach trim routines speed or landing speeds are always greater than the stall
are performed. For both the MLW and MTOW cases, the speed, and as seen throughout this work, an increase in speed
elevator deflection decreases when the CG is moving aft, greatly decreases the maximum deflection required as does
as is expected. the deployment of a high-lift flap at low speeds.
As seen during the control surface sizing, an increase
in speed leads to a lower deflection required by the control
surfaces. For this reason for a more conservative analysis, 4 Conclusion
the two lowest speeds identified in Table 8 are used as stall
speeds, Ustall = 68 m/s at MLW and Ustall = 78 m/s at MTOW, The present work proposes a control surface layout for
to calculate the related speeds a flying wing aircraft based on offline handling quality
simulations. The simulations employ an aerodynamic
• At MLW, Uapproach = 1.23 ⋅ Ustall = 84 m/s = 162 knots model based on Reynolds-Averaged Navier–Stokes
( Mapproach = 0.25) (RANS) and Vortex Lattice Method (VLM) simulations.
• at MTOW, Urotation = 1.2 ⋅ Ustall = 94 m/s = 182 knots The control authority of the elevator, aileron, and rudder is
( Mrotation = 0.28). estimated both with VLM and VLM calibrated with RANS
simulations.
At the identified rotation speed, the take-off rotation Initial simulations show that the Dutch roll is stable for
requirement leads to an additional elevator and elevon most flight conditions, resulting in Level 2 handling qualities
deflection of ca. −8◦ , leading to 𝛿e > −25◦, when using the based on MIL-STD-1797A. These results allow simulations
high-lift flap deflected at 20◦. to be performed that would excite the aircraft simultaneously
Regarding the pull-up and push-over maneuvers, longitudinally and laterally without causing divergent
the former is the critical one, because it requires a more behavior. Several simulations are then conducted to size
negative elevator and elevon deflection than the trim each control surface. In particular, longitudinal trim, pull-up
condition. Particularly, to obtain a load factor of 1.3 at and take-off rotation maneuver for the elevator, steady-
Uapproach in combination with the high-lift flap, the additional heading sideslip and a one-engine-inoperative condition for
requirement leads to 𝛿e > −25◦. the rudder, and steady-heading sideslip and bank-to-bank
Table 9 summarizes for all tested requirements which for the aileron. Finally, coordinated turns are performed to
is the most critical condition in terms of speed and weight verify whether the sized control surfaces could also comply
with the control surfaces as sized in this paper (Fig. 9). For with this requirement.
all cases, a high-lift flap is required, to obtain deflections The defined control surface layout comprises an elevator,
smaller than 32◦ . The highest deflections are always reached an elevon and a rudder. Where the elevon acts simultane-
with the elevator and elevon used together as elevator. The ously as elevator and aileron, having the same deflection

Table 9  Summary of various Certification req. Speed Weight Max |𝛿| Addition
requirements at most critical
conditions and at xCG = forw 0 Longitudinal trim Stall speed MLW 𝛿e > - 25◦ High-lift 𝛿f = 40◦
Pull-up Approach speed MLW 𝛿e > - 25◦ High-lift 𝛿f = 40◦
Coordinated turn Stall speed MLW 𝛿e > - 32 ◦
High-lift 𝛿f = 40◦
Take-off rotation Rotation speed MTOW 𝛿e > - 25 ◦
High-lift 𝛿f = 20◦
Bank-to-bank Stall speed MLW - 20◦ < 𝛿a < 20◦ High-lift 𝛿f = 40◦
Sideslip Stall speed MLW - 25◦ < 𝛿a < 25◦ High-lift 𝛿f = 40◦
OEI Stall speed MLW - 30 < 𝛿r < 30
◦ ◦
High-lift 𝛿f = 40◦,
Drag rudder 𝛿dr ≃ 30◦
 S. Asaro et al.

rates of the aileron, which are ca. 40% higher than the rate References
of the elevator.
The control surfaces are sized to satisfy the certification 1. Asaro, S., Vos, R.: Synthesis of the aerodynamic model of a fly-
requirements even at stall speed, which is always the most ing wing aircraft. In: AIAA SciTech, Orlando, United States of
America, [Link] (January 2025)
demanding case. However, at stall speed, a high-lift flap and 2. Atmaca, D., Stroosma, O., van Kampen, E.J.: Piloted evaluation
a drag rudder are required to satisfy the certification require- of flying-v with incremental nonlinear dynamic inversion and
ments. Overall, the simulations showed that the certification envelope protection. In: AIAA SciTech, Orlando, United States
requirements can be satisfied at stall speed with elevator and of America, [Link] (January 2025)
3. Benad, J.: The flying v - a new aircraft configuration for commer-
elevon deflections 𝛿e > - 32◦ , elevon deflection −25◦ < 𝛿a < cial passenger transport. In: Deutscher Luft-und Raumfahrtkon-
25◦ , and rudder deflection −30◦ < 𝛿r < 30◦ . In a more realis- gress, Rostock, Germany, [Link] oi.o​ rg/1​ 0.2​ 5967/3​ 70094 (2015)
tic scenario, when the aircraft flies faster than the stall speed, 4. Bravo-Mosquera, P.D., Catalano, F.M., Zingg, D.W.: Unconven-
the required deflections of the control surfaces are reduced tional aircraft for civil aviation: a review of concepts and design
methodologies. Prog. Aerosp. Sci. 131, 1–29 (2022). [Link]
and get closer to the maximum deflections of the control org/​10.​1016/j.​paero​sci.​2022.​100813
surfaces of conventional aircraft. 5. Darecki, M., Edelstenne, C., Enders, T., et al.: Flightpath 2050 -
The control authority of control surfaces is estimated Europe’s vision for aviation. Tech. Rep. EUR 098 EN, European
with VLM and VLM calibrated with RANS, showing that Commission (2011)
6. Denieul, Y., Bordeneuve, J., Alazard, D., et al.: Multicontrol sur-
without RANS calibration, VLM would underestimate the face optimization for blended wing–body under handling qual-
deflections needed to satisfy the certification requirements ity constraints. J. Aircr 55, 638–651 (2018). [Link]
up to 43%. This difference is mainly driven by the inability 2514/1.​C0342​68
of VLM to adequately predict the reduction in control 7. EASA. Acceptable means of compliance for large aeroplanes
CS-25. Tech. Rep. Amendment 24, EASA, Europe (2007)
effectiveness with angle of attack. 8. Eftekhar, S.: High lift split flaps for the flying-v. Master thesis,
Technische Universiteit Delft (2024)
Acknowledgements This publication is part of the project Flying-V 9. Hanke, C.R., Nordwall, D.R.: The simulation of a jumbo jet trans-
Flight Control with project number 19511 of the research program port aircraft. Volume 2: Modeling data. Tech. Rep. D6-30643,
Open Technology Programme, which is (partly) financed by the Dutch NASA (1970)
Research Council (NWO). The authors would like to acknowledge the 10. Kumar, A., Asaro, S., Vos, R.: Aerodynamic analysis of the fly-
in-kind contribution of KLM Royal Dutch Airlines and ADSE. ing v subsonic transport to enable engine integration. In: AIAA
SciTech, Orlando, United States of America, [Link]
Author contributions S.A. and D.A. contributed to the study 2514/6.​2025-​0666 (2025)
conception and design. Material preparation, data collection, and 11. Laar, Y., Atherstone, D., Benad, J., et al.: Aerodynamic design
analysis were performed by S.A. and D.A. The first draft of the of a flying v aircraft in transonic conditions. In: AIAA SciTech,
manuscript was written by S.A. and all authors commented on previous Orlando, United States of America, [Link]
versions of the manuscript. All authors read and approved the final 2024-​2669 (January 2024)
manuscript. 12. Matamoros, I., de Visser, C.C.: Incremental nonlinear control
allocation for a tailless aircraft with innovative control effectors.
Funding This study was funded by the Dutch Research Council (NWO) In,: Aiaa guidance. Navigation, and Control Conference (2018).
under the Open Technology Programme with the project Flying-V [Link] (2018)
Flight Control under Project No. 19511. 13. Mattingly, J.D., Heiser, W.H., Boyer, K.M., et al.: Aircraft Engine
Design. American Institute of Aeronautics and Astronautics (2018)
Data availability No datasets were generated or analyzed during the 14. MIL-STD. Flying qualities of piloted aircraft. Tech. Rep. MIL-STD-
current study. 1797A, Department of Defense, United States of America (1990)
15. Oosterom, W., Vos, R.: Conceptual design of a flying-v aircraft
Declarations family. In: AIAA SciTech, Virtual event, [Link]
2514/6.​2022-​3200 (January 2021)
Conflict of interest The authors declare no competing interests. 16. Paul, M., Rein, M.: Transonic numerical and experimental evalu-
ation of unconventional lambda wing control surfaces. J. Aircr.
Ethical approval Not applicable. 54, 1139–1149 (2017). [Link]
17. Qu, X., Zhang, W., Shi, J., et al.: A novel yaw control method for
Open Access This article is licensed under a Creative Commons Attri- flying-wing aircraft in low speed regime. Aerosp. Sci. Technol.
bution 4.0 International License, which permits use, sharing, adapta- 69, 636–649 (2017). [Link]
tion, distribution and reproduction in any medium or format, as long 18. van der Toorn, M.: Flying-v family - design for stability and con-
as you give appropriate credit to the original author(s) and the source, trol. Master thesis, Technische Universiteit Delft (2022)
provide a link to the Creative Commons licence, and indicate if changes 19. Varriale, C.: Applying control allocation methods to a staggered
were made. The images or other third party material in this article are box-wing aircraft configuration. PhD thesis, Technische Univer-
included in the article’s Creative Commons licence, unless indicated siteit Delft (2022)
otherwise in a credit line to the material. If material is not included in 20. Vos, R., Johnson, N., Nolet, S.: Flying v directional flight ccontrol:
the article’s Creative Commons licence and your intended use is not an experimental investigation. In: ICAS, Florence, Italy (2024)
permitted by statutory regulation or exceeds the permitted use, you will
need to obtain permission directly from the copyright holder. To view a Publisher's Note Springer Nature remains neutral with regard to
copy of this licence, visit [Link] jurisdictional claims in published maps and institutional affiliations.