Professional Documents
Culture Documents
A R T I C L E I N F O A B S T R A C T
Keywords: Aircraft directional stability and control are related to vertical tail design. The safety, performance, and flight
Aircraft aerodynamic design qualities of an aircraft also depend on a correct empennage sizing. Specifically, the vertical tail is responsible for
CFD the aircraft yaw stability and control. If these characteristics are not well balanced, the entire aircraft design may
Directional stability fail. Stability and control are often evaluated, especially in the preliminary design phase, with semi-empirical
Vertical tail sizing methods, which are based on the results of experimental investigations performed in the past decades, and oc-
casionally are merged with data provided by theoretical assumptions. This paper reviews the standard semi-
empirical methods usually applied in the estimation of airplane directional stability derivatives in preliminary
design, highlighting the advantages and drawbacks of these approaches that were developed from wind tunnel
tests performed mainly on fighter airplane configurations of the first decades of the past century, and discussing
their applicability on current transport aircraft configurations. Recent investigations made by the authors have
shown the limit of these methods, proving the existence of aerodynamic interference effects in sideslip conditions
which are not adequately considered in classical formulations. The article continues with a concise review of the
numerical methods for aerodynamics and their applicability in aircraft design, highlighting how Reynolds-
Averaged Navier-Stokes (RANS) solvers are well-suited to attain reliable results in attached flow conditions,
with reasonable computational times. From the results of RANS simulations on a modular model of a represen-
tative regional turboprop airplane layout, the authors have developed a modern method to evaluate the vertical
tail and fuselage contributions to aircraft directional stability. The investigation on the modular model has
permitted an effective analysis of the aerodynamic interference effects by moving, changing, and expanding the
available airplane components. Wind tunnel tests over a wide range of airplane configurations have been used to
validate the numerical approach. The comparison between the proposed method and the standard semi-empirical
methods available in literature proves the reliability of the innovative approach, according to the available
experimental data collected in the wind tunnel test campaign.
1. Introduction directional (i.e. around the vertical axis) equilibrium, stability, and
control. The concept of equilibrium is inherent to the absence of accel-
The empennages in traditional aircraft configurations (Fig. 1) perform erations on the aircraft. Directional stability is the aircraft tendency to
three fundamental functions: (i) they provide static and dynamic stability; return to the initial equilibrium condition, if perturbed. Directional
(ii) through their movable parts, they enable aircraft control; (iii) they control is the aircraft ability to maintain equilibrium at a desired sideslip
allow to reach a state of equilibrium in each flight condition. angle, i.e. the angle between the relative wind and the aircraft longitu-
Tail surfaces sizing and shaping are almost exclusively determined by dinal axis [1]. From the dynamic point of view, the role of the vertical tail
stability and control considerations. Both horizontal and vertical tail- is to provide yaw damping, that is to reduce the oscillations around the
planes usually operate at only a fraction of their lift capability, since stall vertical axis (dynamic directional stability). If the aircraft directional
conditions should never be achieved. The vertical tail provides stability is too small with respect to its lateral stability (i.e. around the
* Corresponding author.
E-mail addresses: danilo.cilberti@unina.it (D. Ciliberti), pierluigi.dellavecchia@unina.it (P. Della Vecchia), fabrizio.nicolosi@unina.it (F. Nicolosi), agostino.demarco@unina.it (A. De
Marco).
https://doi.org/10.1016/j.paerosci.2017.11.001
Received 23 August 2017; Accepted 3 November 2017
Available online 23 November 2017
0376-0421/© 2017 Elsevier Ltd. All rights reserved.
D. Ciliberti et al. Progress in Aerospace Sciences 95 (2017) 140–172
Nomenclature to sideslip
CNβh horizontal tail yawing moment coefficient derivative due
to sideslip
Acronyms CNβv vertical tail yawing moment coefficient derivative due
AGILE Aircraft 3rd Generation MDO for Innovative Collaboration to sideslip
of Heterogeneous Teams of Experts CNβw wing yawing moment coefficient derivative due to sideslip
CFD Computational Fluid Dynamics CYβ airplane sideforce coefficient derivative due to sideslip
CNC Computer Numerical Control Df fuselage max diameter
DES Detached Eddy Simulation KFv coefficient to account for fuselage interference on
DII Dipartimento di Ingegneria Industriale vertical tail
DNS Direct Navier-Stokes KHf coefficient to account for horizontal tail interference
ESDU Engineering Science Data Unit on fuselage
F fuselage KHv coefficient to account for horizontal tail interference on
FV fuselage – vertical tail combination vertical tail
H horizontal tail KVf coefficient to account for vertical tail interference
LES Large Eddy Simulation on fuselage
m.a.c. mean aerodynamic chord KWf coefficient to account for wing interference on fuselage
MDO Multidisciplinary Design Optimization KWv coefficient to account for wing interference on vertical tail
NACA National Advisory Committee for Aeronautics Lf fuselage length
OAD Overall Aircraft Design Ln fuselage nose length
RANS Reynolds-Averaged Navier-Stokes Lt fuselage tail-cone length
SCoPE Sistema Cooperativo Per Elaborazioni scientifiche S wing planform area
multidisciplinari Sr rudder planform area
USAF DATCOM Sfront fuselage frontal area
United States Air Force Data Compendium Sh horizontal tailplane area
V vertical tail Sv vertical tailplane area
W wing VsTO take-off stall speed
WFV wing – fuselage – vertical tail combination V∞ asymptotic velocity
WFVH wing – fuselage – vertical tail – horizontal tail combination b wing span
Notation bv vertical tailplane span
A wing aspect ratio bv1 vertical tailplane span extended to the fuselage centerline
Ah horizontal tailplane aspect ratio cr rudder m.a.c.
Av vertical tailplane aspect ratio df fuselage equivalent diameter
Aveff vertical tailplane effective aspect ratio dfv fuselage height at vertical tail aerodynamic center
pffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
ffi lv vertical tail directional moment arm
B compressibility parameter 1 M 2
Ch hinge moment coefficient rf fuselage max radius
CL airplane rolling moment coefficient y+ dimensionless wall distance
CLαv vertical tail lift curve slope zftc height of the fuselage tail-cone
CL β airplane rolling moment coefficient derivative due zh position of the horizontal tail on the vertical tail
to sideslip zw wing position in fuselage
CN airplane yawing moment coefficient α angle of attack
CNf fuselage yawing moment coefficient β angle of sideslip
CNv vertical tail yawing moment coefficient λ taper ratio
CNβ airplane yawing moment coefficient derivative due Δ difference
to sideslip Λ sweep angle
CNβf fuselage yawing moment coefficient derivative due
longitudinal axis), the aircraft tends to oscillate in yaw as the pilot gives sideslip β. Such design is not a simple task, since it involves the prediction
rudder or aileron inputs. This tendency is called dutch roll and makes of nontrivial phenomena, such as the asymmetrical flow behind the
precise directional control difficult. wing-fuselage combination, and the solution of lateral cross-control is-
Extreme flight conditions usually set design requirements for tail sues (due to side force on the fin causing a rolling moment).
surfaces, as minimum control speed with one engine inoperative (Fig. 2) The following design requirements can be formulated for vertical
or maximum cross-wind capability (Fig. 3). Stability and control must be tailplanes, as suggested in Ref. [2–5]:
ensured even in large angles of sideslip as 25 [2]. The design of a vertical
tailplane depends mainly on the type of airplane (configuration layout, 1. The vertical fin must provide a sufficient contribution to static and
flow regime, aesthetics, costs), and on engines number and position. For dynamic stability, which is function of the vertical tail lift curve slope
a given layout, the vertical tail design should take into account the and planform area (Fig. 4) or volume coefficient. These are ensured
relative size and position of other elements in the whole aerodynamic by a sufficiently high value of the derivative
configuration, such as wing, fuselage, and horizontal tail [3]. These
factors affect the aircraft stability derivatives, i.e. the variation of aero- Sv lv
CNβv ¼ f CLαv ; (1)
dynamic coefficients with the main flight variables, and, in particular, the S b
vertical fin design influences all derivatives with respect to the angle of
141
D. Ciliberti et al. Progress in Aerospace Sciences 95 (2017) 140–172
1
M h ¼ ρV 2 Ch Sr cr (2)
2
which depends on rudder planform area Sr, mean aerodynamic chord cr,
dynamic pressure 12 ρV 2 , and hinge moment coefficient Ch. This is
important for the sizing of control surface actuators and pedal con-
trol forces.
5. Generally, the tail weight must be as low as possible, and this calls for
low aspect ratios. Moreover, for T-tail configurations higher aspect
ratios might make the flutter phenomenon even more critical.
6. A tapered planform, of taper ratio λv, leads to lower fin weight. On the
other hand, excessive taper ratios may lead to premature tip stall.
142
D. Ciliberti et al. Progress in Aerospace Sciences 95 (2017) 140–172
Fig. 5. Aircraft models investigated by NACA in 1939 and 1950. Reproduced from Report 730 [12] (left) and Report 1049 [13] (right).
143
D. Ciliberti et al. Progress in Aerospace Sciences 95 (2017) 140–172
Fig. 6. Effects of wing position on vertical tail yawing moment coefficient derivative
provided by DATCOM and ESDU.
Fig. 8. Rudder equilibrium angle versus sideslip angle for two vertical tailplanes.
due to sideslip coefficient CYβv and the yawing moment derivative CNβv
by moving the wing from high to low position in fuselage, except for the
fin-off combination where there is a minimum for the mid wing position. aerodynamic center (shortly: tail span per body depth), for various
This trend holds at different wing dihedral angles, flap deflections, and tail-fuselage combinations. The vertical tail effective aspect ratio in-
angles of attack, although different values have been achieved for each creases as the ratio bv/dfv decreases, that is, as the vertical tail becomes
combination. In other words, it is found that a low wing is beneficial to small compared to the fuselage. The theory based on the assumption that
directional stability. The rolling moment derivative CL βv decreases and the body acts as an infinite plate at the base of the vertical tail suggested a
changes sign as the wing is moved downward. This is the well-known non-linear increase of the effective aspect ratio with the
dihedral effect due to wing position, where a high wing provides an above-mentioned tail span over body height ratio. The scatter in exper-
increased lateral stability [1]. This work did not include the horizontal imental data is such that best fitting lines between theory and experi-
tail, therefore the interference effect on lateral-directional stability is due ments are drawn as a function of both the bv/dfv ratio and the vertical tail
to the wing – body – vertical tail combination. geometric aspect ratio Av, whereas in USAF DATCOM [9] the dependency
To evaluate the effect of the fuselage on the vertical tail, without from the aspect ratio disappears and the effect of taper ratio is included.
horizontal tail, Queijo and Wolhart [13] in 1950 defined an effective The effect of size and position of horizontal tail on vertical tail
aspect ratio Aveff as function of the ratio of vertical tail span bv to the characteristics was studied by Brewer and Lichtenstein [14] in 1950, who
fuselage diameter dfv at the longitudinal location of the vertical tail also found that previous results [15,16] could not be extended to swept
Fig. 7. Effects of horizontal tail position on vertical tail yawing moment coefficient de-
rivative provided by DATCOM and ESDU. Fig. 9. Minimum control speed as function of engine yawing moment and vertical tail
planform area.
144
D. Ciliberti et al. Progress in Aerospace Sciences 95 (2017) 140–172
Fig. 10. Computational time versus CPUs number for RANS simulations.
planforms. Results of their work have shown that the effect of the hori- Table 1
zontal tail is advantageous to directional stability, at low angles of attack, Geometric data of the aircraft simulated with CFD.
if it is positioned upward and rearward on the vertical tail (T-tail Parameter Symbol Values
configuration) or if it is located on the fuselage. At high angles of attack, Wing aspect ratio A 6 to 14, step 2
flow separation and wing downwash can highly affect the directional Wing position zw/rf 1, 0, 1
stability, although positive effects can be obtained by moving the hori- Horizontal tail position zh/bv1 0, 0.38, 0.62, 0.82, 1.0
zontal tail forward and upward. The vertical tail effective aspect ratio Fuselage tail-cone ratio zftc/rf 0, 0.5, 1.0
Fuselage slenderness ratio Lf/Df 7.0, 8.0, 8.7, 9.5, 11.0, 12.0
Fuselage nose slenderness ratio Ln/Df 1.1 to 1.7, step 0.1
Fuselage tail-cone slenderness ratio Lt/Df 2.3 to 3.0, step 0.1
Fig. 11. Layout of the aircraft model for the numerical analyses.
145
D. Ciliberti et al. Progress in Aerospace Sciences 95 (2017) 140–172
Table 2
Vertical tail planform data. See Fig. 12.
Aspect ratio Av 0.5, 2.0, step 0.25, 0.5 to 2.5, 0.40, 0.80, 1.59, 2.39,
0.5 step 0.5 3.19, 3.98
Taper ratio λv 0.86, 0.73, 1.0 0.62
0.62, 0.53
Leading edge 26.6 0 13.7 , 16.9 , 22.1 ,
sweep angle Λv 31.4 , 50.6 , 67.7
Table 3
Number of configurations in the numerical analyses.
F 3 zftc/rf
V 18 Av; λv; Λv
FV 30 Av; λv; Λv; zftc/rf
WFV 45 λv; zftc/rf; A; zw/rf
WFVH 135 Av; λv; zftc/rf; zw/rf; zh/bv1
< 1.5 and this leads to the conclusion that, for most vertical tail surfaces
at a given angle of sideslip, the side force is only dependent on the tail
Fig. 14. Effects of Reynolds number on sideforce aerodynamic coefficient.
span bv, whereas planform is of secondary importance [2].
3. Semi-empirical methods
Results of NACA wind tunnel tests previously described were orga- [17]. This method assumes an almost circular fuselage and a constant
nized and used to formulate a set of semi-empirical procedures presented sidewash. It is a synthesis of experimental analyses performed by NACA,
in USAF DATCOM [9]. Semi-empirical methods are simple mathematical British Aerospace, SAAB, and others, from the ‘40s to the ‘70s, with
models of a physic phenomenon, based on both theoretical assumptions application of the potential flow theory in highly scattered data zones.
and on experimental evidence. They provide a valuable aid in the con- The theory at the base is found in the work of Weber and Hawk [18],
ceptual and preliminary aircraft design stages, even if the aircraft layout dating back to 1954, who assumes that a fin-body-tailplane combination
has not been sketched yet. They have the advantage to be simple and at incidence (or sideslip) develops a vortex system that induces a constant
rapid to compute, especially on electronic spreadsheets, allowing quick velocity distribution along the wing span. Here the term wing is used as
design of experiments and optimization. They are categorized as low fi- generic lifting surface, since the vertical fin in fuselage is considered as a
delity models, because they are typically used by designers when the wing with a cylinder at its root. Load distribution is computed in the
exact shape of the aircraft is not well defined, as the aerodynamics, Trefftz-plane (located far downwind for high aspect ratio wings) once the
structures, and performance. induced velocity is known. The induced velocity is calculated from the
It is here remarked that the DATCOM method for the evaluation of wing characteristics as planform, sweep angle, and wing section lift curve
aircraft directional stability has been mainly derived from the results on slope. The definitions of vertical tail planform area and aspect ratio, as
the geometries previously presented, i.e. elliptical bodies, swept wing
and tails (see again Fig. 5). Apart from the NACA, in the UK an alternative
method to compute the vertical tailplane contribution to directional
stability in presence of body, wing, and horizontal tailplane, was pro-
posed and described in the Engineering Science Data Unit (ESDU) report
146
D. Ciliberti et al. Progress in Aerospace Sciences 95 (2017) 140–172
aircraft directional stability may lead to a different vertical tail sizing and
performance. This is due to the different estimation of the aerodynamic
interference effects, which will be described in Section 5.4. Fig. 6 shows
that, while DATCOM and ESDU methods agree for a high wing airplane,
the latter predicts a vertical tail yawing moment coefficient derivative
that is 25% higher than the former for a low wing aircraft. Fig. 7 shows
that the two methods disagree on the most significant empennage con-
figurations, with ESDU giving a vertical tail contribution to directional
stability that with respect to DATCOM is 25% higher in the case of
body-mounted horizontal tail, 10% higher for a cruciform tail, and 7%
lower in the case of a T-tail configuration.
In another recent work [20], the authors have highlighted the
importance of a correct vertical tail sizing to avoid stability and control
issues. Perkins and Hage [1] stated that the airplane has better flying
qualities with a high directional stability, with typical values (per degree)
of CNβ from 0.0015 to 0.0020 for normal aircraft and from 0.0008 to
0.0010 for World War II fighters. A high wing loading should increase the
aircraft directional stability because a reduced wing span should lead to
weaker directional perturbations. For this reason, they suggested an
expression of the desirable aircraft directional stability as a fraction of the
Fig. 16. Effect of the fuselage on the vertical tail: acceleration of the flow at the root of the
vertical tail. span loading
rffiffiffiffiffi
W
CNβ ¼ 0:0005 (3)
b2
well as the aerodynamic interference factors, are quite different from
those of DATCOM. which holds true as long as linearity of CN with β is valid. In Ref. [20] the
Recently, the authors have found that the results of DATCOM and authors have also highlighted the importance of non-linear effects in
ESDU methods for aircraft directional stability evaluation can lead to aircraft directional stability and control. For instance, an oversized ver-
significant discrepancies for some configurations [19]. For instance, tical tail planform area can lead to insufficient control capability, since
considering a typical turboprop aircraft geometry, a sensitivity study on the aircraft becomes directionally too stable and it is unable to counteract
the effects of wing position (Fig. 6) and of horizontal tail position (Fig. 7) crosswinds and lateral gusts. Fig. 8 shows the required rudder deflections
has shown that the predicted values of vertical tail contribution to to fly with given sideslip angles, for two identical regional turboprop
147
D. Ciliberti et al. Progress in Aerospace Sciences 95 (2017) 140–172
Fig. 18. Effect of the horizontal tail on the vertical tail: the end-plate effect.
airplanes with different vertical tail planform areas and same rudder
chord ratios. The baseline aircraft with a tail area of 13 m2 can maintain
control linearity, with good approximation, up to 15 of sideslip angle;
the same aircraft is able to keep 24 of sideslip at full rudder deflection.
On the other hand, the aircraft with an increased tail area of 20 m2
maintains a linear control up to 12 of sideslip, then a further rudder
deflection stalls the vertical tail, which now needs 25 of rudder
deflection to counteract less than 14 of sideslip angle. This leads to the
rule of thumb of keeping the derivative dδr/dβ < 1, that is the value of
directional stability derivative CNβ should be lower than the value of the
directional control derivative CNδr. Moreover, an oversized tail is heavier,
produces more parasite drag, and may result in an airborne minimum
control speed that is below the minimum airspeed value (usually 1.13
times the stall speed at take-off VsTO) stated by aviation authorities
Fig. 19. Isolated vertical tail lift curve slope versus aspect ratio. Planform set of Fig. 12a,
M ¼ 0, Λv ¼ 26.6 .
Fig. 21. Effect of fuselage slenderness ratio, non-dimensioned on max fuselage diameter df
Fig. 20. Fuselage reference system. and frontal area Sfront. Each marker represents a CFD analysis.
148
D. Ciliberti et al. Progress in Aerospace Sciences 95 (2017) 140–172
Fig. 24. Definition of vertical tail span over fuselage diameter bv/dfv.
Fig. 22. Effect of fuselage nose slenderness ratio, non-dimensioned on max fuselage
diameter df and frontal area Sfront. Each marker represents a CFD analysis.
4. Numerical methods
Fig. 23. Effect of fuselage tail-cone slenderness ratio, non-dimensioned on max fuselage
diameter df and frontal area Sfront. Each marker represents a CFD analysis. Fig. 26. Aerodynamic interference factor KFv. Effect of the fuselage on the vertical tail.
The markers are results of CFD analyses.
149
D. Ciliberti et al. Progress in Aerospace Sciences 95 (2017) 140–172
Fig. 29. Aerodynamic interference factor KWv. Effect of the wing on the vertical tail. The markers are results of CFD analyses.
150
D. Ciliberti et al. Progress in Aerospace Sciences 95 (2017) 140–172
Fig. 30. Aerodynamic interference factor KWf. Effect of the wing on the fuselage. The markers are results of CFD analyses.
decreasing lift gradient, and flow separation. Neglecting viscosity gives a shown next.
fair good approximation only in attached flow regimes at high Reynolds Thus, the limits of wind-tunnel investigation and panel codes can be
numbers. Panel codes can be coupled with boundary layer codes to es- overcome by modern CFD. The available Navier-Stokes solvers are so
timate the friction drag: a displacement thickness, due to the loss of powerful to offer significant benefits as companions to the experimental
momentum in the boundary layer, is calculated on the body walls by the methods when designers need reasonably accurate aerodynamic pre-
viscous code; this displacement thickness is added to the previous ge- dictions. Since 1985 the increased computational power and the devel-
ometry, providing the new input surface of the panel code; the process is opment of new solution techniques and turbulence modeling have
repeated until a convergence criterion is finally satisfied [23]. However, permitted an extensive application of CFD in aircraft aerodynamic
a fundamental limitation remains: panel methods solve only linear dif- analysis and design [21,27,28]. Panel methods have been discarded in
ferential equations, i.e. they can simulate only attached flows. Also, since the numerical investigations performed by the authors, because of their
panel methods do not compute the pressure in the separated flow regime, misprediction of the aerodynamic interference.
prediction of pressure drag is not possible. Finally, by their own nature, Although other more complex numerical techniques and tools are
panel codes can hardly provide wake-body interference, since the shape available, as Large Eddy Simulations (LES) or Detached Eddy Simulations
of the vortex sheet is unknown a priori [24]. It is possible to determine (DES) solvers, which are naturally unsteady, and even Direct Navier-
the pressure acting on bodies in the region of wing-body interference in Stokes (DNS) codes, which are practically applicable only to very sim-
subsonic flow with a hybrid panel method [25], but viscosity is neglec- ple geometries, the authors focused on the applications of RANS. This is
ted again. motivated by the nature of the topic and by the encouraging results
The complexity and costs of wind tunnel tests and the increasing attained in several years of research in numerical and experimental
viscosity effects at high angles of incidence led to the development of aerodynamics, which have led to the evaluation of performance, high lift,
Navier-Stokes solvers [26], which represent, in principle, the true flight loads, stability and control, and to the overall design of various
simulation of the physics of the viscous flow. Compared to the panel general aviation and regional turboprop airplanes. In recent years, the
codes, they are much more computationally expensive, also due to the aerodynamics characteristics of a new commuter aircraft have been
generation of an adequate computational mesh, and the problem of deeply investigated, the satisfying agreement between numerical pre-
turbulence modeling is still open. However, adequate turbulence models dictions and wind tunnel tests both in longitudinal and lateral-directional
do exist for aeronautical applications and the possibility to simulate and axes are highlighted in Refs. [29–31]. Aerodynamic design, analysis, and
visualize a viscous flow pay back these disadvantages, as will be optimization of regional turboprop airplanes have also deeply exploited
the numerical tools [32]. The numerical high lift prediction on the NASA
common research model [33] and the parametric investigations on
aircraft dorsal fin [34] enabled further confidence in the available nu-
merical capabilities in the non-linear range of aerodynamics.
The analysis on the aerodynamic interference effects involved in
sideslip conditions [19] stimulated further investigations to develop an
alternative method to evaluate aircraft directional stability [35–37] and
to provide a more reliable tool for the preliminary design of the vertical
stabilizer of a regional turboprop aircraft. The same approach has been
applied to fuselage aerodynamics [38] and control power estimation of a
directional rudder [39]. The numerical analyses that allowed to highlight
the aerodynamic interference issues in aircraft directional stability are
described in the next section.
151
D. Ciliberti et al. Progress in Aerospace Sciences 95 (2017) 140–172
Fig. 32. Aerodynamic interference factor KHv. Effect of the horizontal tail on the vertical tail. The markers are results of CFD analyses.
simulate many configurations in a reasonably short amount of time. for attached boundary layers and flows with mild separation (such as
Fig. 10 shows the time necessary to solve a RANS simulation versus CPUs flow past a wing). It has been found that this model is reliable for external
number for two configurations: a body-tail combination with 1.8 million aerodynamics, even at high angles of attack and in high lift configura-
cells and 2000 iterations and a complete aircraft in high lift configuration tions [33,42,43], provided that a properly refined mesh is chosen and the
with 18 million cells and 10000 iterations. Simple configurations, such as level of refinement is determined on the basis of a grid conver-
isolated fin or fuselage-fin, can be analyzed on desktop workstations gence study.
within hours, while more complex geometries, such as wing-body with
flap and slat, require a fine mesh and many iterations to achieve nu-
merical convergence. In these cases, the typical computational time is 1 5.1. The aircraft model and sampling plan
or 2 days per angle of attack.
The numerical investigation campaign has been performed assuming The aircraft investigated is the modular model shown in Fig. 11.
a subsonic, incompressible, stationary, and fully turbulent flow regime. Several wing positions, wing aspect ratios, vertical tailplane planforms,
STAR-CCMþ applies the finite volume method to convert the continuous horizontal tailplane positions, and three fuselages have been arranged in
system of flow equations to a set of discrete algebraic equations [40]. many different combinations to perform a parametric investigation. The
The Spalart-Allmaras (SA) turbulence model [41] has been applied. wing is straight and untapered with an aspect ratio A varying from 6 to
The original model was developed primarily for the aerospace industry 14, with step 2. The fuselage is a narrow body with a circular section and
with three different after-bodies (tail-cones). These parameters are
152
D. Ciliberti et al. Progress in Aerospace Sciences 95 (2017) 140–172
Fig. 33. Aerodynamic interference factor KHf. Effect of the horizontal tail on the fuselage. The markers are results of CFD analyses.
summarized in Table 1. The horizontal stabilizer is straight and unta- simulations have been performed to write a database on the regional
pered, with a fixed aspect ratio (Ah ¼ 4.5). The vertical tailplane aspect turboprop aircraft directional stability. The geometries have been kept
ratio Av is varied from 0.4 to 4, the sweep angle Λv from 0 to 60 , and the simple to automate the mesh generation process and to realize a modular
taper ratio λv from 0.3 to 1. Three different vertical tail planforms fam- model to be tested in the wind tunnel. Particular attention has been given
ilies have been analyzed. The main set, which has been used throughout to the vertical tailplane and the fuselage, since these are the most
the investigation, is made up of four planforms with the same sweep important components in aircraft directional stability. The analyses have
angles, but different aspect ratios, areas and taper ratios. A single vertical shown that wing and horizontal stabilizer do not directly affect direc-
tail in this set has been cut in distinct parts by 3 horizontal planes tional stability, but rather influence the vertical tailplane and fuselage
(Fig. 12a). The choice of the second set follows the same principles of the behavior in sideslip.
previous one, except that the fin planforms are unswept and untapered
(Fig. 12b). The third set is made up of planforms with a constant taper 5.2. Mesh set-up
ratio, but different areas and aspect ratios (Fig. 12c). The last two families
have been used only in the investigation on the fuselage effect described The computational domain is made up of polyhedral cells, whose
in Section 6.2. The vertical tail parameters are summarized in Table 2. number changes according to the configuration analyzed and increases
The use of different planform areas is not an issue, because, for the up to 10 million for a complete aircraft configuration. A careful grid
evaluation of the aerodynamic interference, the effect of vertical tail area optimization study enabled to find a convenient expansion rate of the
is canceled, as shown in the following sections. The number of simula- volume of cells from the no-slip walls to the external boundaries of the
tions for each configuration is reported in Table 3. More than 200 computational domain. This allowed to save up memory and
153
D. Ciliberti et al. Progress in Aerospace Sciences 95 (2017) 140–172
154
D. Ciliberti et al. Progress in Aerospace Sciences 95 (2017) 140–172
ID Description Length (m) Width (m) The effects of the Reynolds number on the stability derivatives is
2 Fuselage nose 0.293 0.222 negligible at low sideslip angles, as shown in Fig. 14. Test cases for
3, 21, 22 Fuselage tail-cone 0.733 0.222 several models in different conditions of Reynolds and Mach numbers
4 Fuselage cabin 0.978 0.222 have been presented in Ref. [19]. All the numerical analyses, whose re-
sults have been organized in the metamodel introduced in Section 6,
have been performed at angle of attack α ¼ 0 and angle of sideslip
Table 5
Dimensions of the lifting surfaces of the aircraft model. Items ID are shown in Fig. 35. β ¼ 5 , with a Reynolds number based on the wing mean aerodynamic
chord Re 1000000, which is close to the value attainable in low-speed
ID Description A λ S (m2) croot (m) ctip (m) b (m)
wind tunnels.
26 Wing 8.3 1.0 0.273 0.182 0.182 1.5
5 to 8 Hor. tail 4.1 1.0 0.066 0.128 0.128 0.521
5.4. Aerodynamic interference
11 Ver. tail 1.0 0.73 0.044 0.242 0.177 0.210
12 Ver. tail 1.5 0.62 0.058 0.242 0.151 0.295
13 Ver. tail 2.0 0.53 0.068 0.242 0.128 0.370 A typical aircraft yawing moment coefficient breakdown is shown in
Fig. 15. The fuselage contribution is unstable (dCN/dβ < 0), whereas the
vertical tail contribution is stabilizing (dCN/dβ > 0) and about twice the
fuselage contribution, which is a typical result for regional turboprop
aircraft. Straight wing surfaces give negligible direct contribution to
directional stability, whereas their aerodynamic interference on the
vertical tail may be significant, as described in the following.
155
D. Ciliberti et al. Progress in Aerospace Sciences 95 (2017) 140–172
Fig. 41. Loads path with the vertical tail load cell.
Fig. 39. Fuselage tail-cone detail.
Table 6
PROLAB 65 properties.
156
D. Ciliberti et al. Progress in Aerospace Sciences 95 (2017) 140–172
Table 7 in the pressure loading of the latter. This effect is more significant if the
Wind tunnel main characteristics. horizontal panel is mounted where the vertical tail exhibits an appre-
Test section dimensions 2.0 m 1.4 m ciable gradient, i.e. at a relatively high or low position, as shown in
Maximum wind speed 50 m/s Fig. 18. Test data (see Ref. [9], x5.3.1.1) highlight the greater effective-
Turbulence level 0.10%
ness of horizontal panels in these positions and the relative ineffective-
Maximum power 150 kW
ness of a horizontal panel at the midspan position on the vertical tail
(cruciform empennage). At subsonic speeds, the aircraft fuselage and
(pressure þ shear) and pressure coefficient distribution at the vertical tail horizontal tail affect the flow on the vertical tail in such a way as to in-
mean aerodynamic chord section are also shown. The low wing config- crease the effectiveness of the vertical tail. This phenomenon, known as
uration has a higher peak of pressure coefficient, although the difference the end-plate effect, is represented in DATCOM [9] by an effective change
with the high wing configuration is small. The increase (low wing) or in vertical tail aspect ratio required to give the same lift effectiveness as
decrease (high wing) of vertical tail effectiveness due to the wing the actual planform in the presence of the other airplane components.
mounting position in fuselage is around 10%. The magnitude of this effect depends also from vertical tail aspect ratio,
wing position, and fuselage after-body shape. The increase of vertical tail
5.4.3. Effect of the horizontal tail effectiveness in sideslip may be up to 45% for a T-tail configuration,
The presence of a horizontal panel near a vertical tail causes a change whereas the fuselage unstable contribution may be reduced up to 10% for
157
D. Ciliberti et al. Progress in Aerospace Sciences 95 (2017) 140–172
a body-mounted horizontal tail. Cruciform empennage configurations 6.1. Isolated vertical tail and fuselage
slightly decrease the effectiveness of vertical tail in sideslip.
The lift curve slope CLαv of the isolated vertical tailplane is a function
6. A modern semi-empirical method for the evaluation of aircraft of its planform (Fig. 4), airfoil shape, and Mach number. It can be pre-
directional stability dicted by the Helmbold-Diederich [44,45] formula for vertical tail, also
illustrated in Fig. 19 at M ¼ 0
By combining the geometries presented in the previous sections, a
metamodel has been generated from the results of the numerical ana- 2πAv
CLαv ¼ rffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
(7)
lyses. The aircraft yawing moment coefficient derivative due to sideslip B2 A2v tan2 Λ
2þ κ2
1 þ B2v;c=2 þ 4
can be calculated as the sum of the contributions of vertical tail, fuselage,
wing, and horizontal tail
where Av p isffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
the vertical
ffi tail aspect ratio, B is the compressibility
CNβ ¼ CNβv þ CNβf þ CNβw þ CNβh (4) parameter 1 M 2 , κ is the ratio of section lift-curve slope to theoretical
thin-section value that for thin airfoils can be approximated to 1/B, Λv,c/2
where each term includes the aerodynamic interference, which will be is the vertical tail sweep angle at half chord.
calculated in the following. For a regional turboprop airplane, with a The yawing moment coefficient derivative CNβf of the isolated fuse-
straight tapered wing, the vertical tailplane and the fuselage give the lage may be evaluated as described in Ref. [38].
main contributions, defined as follows S
front df
CNisolated
β
¼ CNβf þ ΔCNβf þ ΔCNβf (8)
lv Sv f slend nose tail S b
CNβv ¼ KFv KWv KHv CLαv (5)
b S
where the reference system of the isolated fuselage is shown in Fig. 20;
CNf slend is the yawing moment coefficient as function of fuselage slen-
CNβf ¼ KVf KWf KHf CNisolated (6)
β f derness ratio Lf/Df (see Fig. 21); ΔCNf nose is the yawing moment
The vertical tail contribution is stable (and usually reported with correction factor due to the nose slenderness ratio Ln/Df (see Fig. 22);
positive sign), whereas the fuselage contribution is unstable. The lift ΔCNf tail is the yawing moment correction factor due to the tail-cone
curve slope of the isolated vertical tailplane CLαv and the yawing moment slenderness ratio Lt/Df (see Fig. 23). These data are referred to the
coefficient derivative due to sideslip of the isolated fuselage CNβf are fuselage frontal area Sfront and cabin equivalent diameter df, which are
corrected by the interference factors K(∙), which are ratios of aero- related as follows
dynamic coefficients between two configurations differing for only one
df2
component. For instance, the effect of the fuselage on the vertical tail Sfront ¼ π (9)
contribution to directional stability is given by the ratio CNβv (FV)/CNβv 4
(V), where F stands for fuselage and V stands for vertical tailplane. The In the case of a cabin with circular cross-section one has df ¼ Df.
effect of the wing (W) on the vertical tail is given by the ratio CNβv
(WFV)/CNβv (FV), that is wing – fuselage – vertical tail combination over 6.2. Fuselage - vertical tail interference factors KFv and KVf
fuselage – vertical tail combination. Similarly, for the horizontal tailplane
(H) the effect is measured by the ratio CNβv (WFVH)/CNβv (WFV). In this The interference factor KFv is defined as the ratio between the yawing
way, the global effect of the aerodynamic interference can be obtained by moment coefficient of the fuselage - vertical tail combination to the
multiplying these factors. Each interference factor K(∙) is defined such yawing moment coefficient of the isolated vertical tailplane
that K(∙) > 1 means that the aerodynamic interference increases the value
of the derivative CNβ, while K(∙) < 1 reduces it.
158
D. Ciliberti et al. Progress in Aerospace Sciences 95 (2017) 140–172
parameterized with fuselage tail-cone shape, defined in Fig. 25. The ef-
fect of the fuselage on the vertical tailplane in sideslip is represented in
Fig. 26. The markers represent the combinations analyzed with CFD.
There is an increase of vertical tail effectiveness in sideslip (i.e. KFv > 1) in
almost all the configurations analyzed, although each fuselage exhibits a
different trend.
The effect of the vertical tailplane on the fuselage in sideslip is rep-
resented in Fig. 27. The interference factor KVf is defined as the ratio
between the yawing moment coefficient of the fuselage - vertical tail
combination to the yawing moment coefficient of the isolated fuselage
CNβf ðFVÞ
KVf ¼ (11)
CNβf ðFÞ
159
D. Ciliberti et al. Progress in Aerospace Sciences 95 (2017) 140–172
has shown that results do not change with vertical tail aspect ratio Av. If
an aircraft configuration is not directly represented, it is always possible
to interpolate values between two charts. The reference system for wing
position is shown in Fig. 28, whereas that describing the fuselage
tail-cone shape has been reported in Fig. 25.
The effect KWv of the aerodynamic interference of the wing on the
vertical tail is reported in Fig. 29. The markers represent the results of
CFD analyses. The mid-low position is the most favorable, increasing the
vertical tail contribution to directional stability up to 13%, whereas the
high wing reduces this contribution from 7% to 17% according to wing
position and fuselage tail-cone shape. The effect of the wing aspect ratio
is to slightly modify the KWv factor, especially at the high and low wing
positions, zw/rf ¼ 1 and 1 respectively. The typical effect of larger wing
aspect ratios is to reduce both the positive and negative perturbations of
the wing, with the exception of the configurations with low-wing (zw/
rf ¼ 1) and fuselage tail-cones with zftc/rf ¼ 0.5 to 1, where the KWv
factor is greater for high aspect ratio wings. This is probably due to the
longer distance between wing wake and wingtip vortices from the ver-
tical tail surface.
The charts representing the interference effect of the wing on the
fuselage KWf are shown in Fig. 30. Their trend is the opposite of the KWv
factor. Since the fuselage is an aerodynamically unstable body, an
interference factor K > 1 indicates an increase in directional instability.
The numerical analyses show variations from 14% to 16% of fuselage
directional instability.
CNβv ðWFVHÞ
KHv ¼ (14)
CNβv ðWFVÞ
CNβf ðWFVHÞ
KHf ¼ (15)
CNβf ðWFVÞ
160
D. Ciliberti et al. Progress in Aerospace Sciences 95 (2017) 140–172
161
D. Ciliberti et al. Progress in Aerospace Sciences 95 (2017) 140–172
the direct measure of the aerodynamic force on the empennage has balance, obtained by hollowing the central, lower part. It allows the wing
been provided. to be placed in low, mid, and high position.
The model main dimensions are reported in Fig. 34. It is 2.0 m long The wing airfoil section is a NACA 23015, typical for regional tur-
and 1.5 m wide. The wing span is limited by the wind tunnel test section bopropeller transport airplane. The wing has a straight untapered plan-
width. The aspect ratio is held to A ¼ 8.3. The model total height (with form of 1.5 m span and 8.3 aspect ratio. The effects of the latter have not
the longest vertical tail span) is about 0.6 m. An exploded view of the been investigated experimentally. As previously stated, the wing span,
model is reported in Fig. 35. The dimensions of the main components are and hence the aspect ratio, is limited by the wind tunnel test section
reported in Tables 4 and 5. Some details about these components are width. Four countersunk M6 screws are used to attach the wing in high
given in the following. and low positions in fuselage.
The fuselage central body (Fig. 36) is the component to which all the Three vertical tail planforms, shown in Fig. 37, are provided. The
other main parts are linked. It provides housing for the wind tunnel
Fig. 52. Comparison of numerical and experimental data on the isolated fuselage with
zftc/rf ¼ 1.0. Re ¼ 470000, α ¼ 0 .
Fig. 51. Comparison of numerical and experimental data on the isolated vertical tail with
Av ¼ 1.5. Re ¼ 470000, α ¼ 0 .
162
D. Ciliberti et al. Progress in Aerospace Sciences 95 (2017) 140–172
Fig. 53. Comparison of numerical and experimental data on the body – vertical tail Fig. 55. Non-linear effects on the vertical tail yawing moment coefficient of the body –
combination. Re ¼ 470000, α ¼ 0 . vertical tail combination in the wind tunnel at Re ¼ 470000 and α ¼ 0 .
airfoil section is the NACA 0012. The root chord and the sweep angle is to protect the load cell and its supporting plates from the aerodynamic
the same for the three planforms. The aspect ratios Av investigated are flow. A slot on the bottom fuselage surface has been designed to
1.0, 1.5, and 2.0. Each planform presents pairs of holes at several span accommodate the data acquisition cable from the load cell to the wind
stations. These are used to insert the horizontal tail spars to get different tunnel acquisition system. The slot continues through the fuselage cabin,
empennage configurations. allowing the cable to be safely taken to the wind tunnel balance sting,
Three fuselage tail-cones, shown in Fig. 38, are provided to investi- avoiding aerodynamic interference. The tail-cone is a crucial component,
gate the effects of the wing-body wake on the empennage. To get because it must: (i) support the empennage; (ii) be stiff enough to not be
experimental data about the vertical tail in various aircraft configura- deformed under aerodynamic loads; (iii) be easily handled during the
tions, the vertical stabilizer has been attached to a separate load cell, change of configuration; (iv) not weigh too much on the fuselage cabin.
which, in turn, is attached to the fuselage. The space to allow load cell The only constraint between the vertical tail and the fuselage is a
and empennage mounting is shown in Fig. 39. It provides enough volume mono-axial load cell, which is linked to the fuselage on one side and to
the vertical stabilizer on the other side. Thus, the latter is suspended on
Fig. 54. Non-linear effects on the yawing moment coefficient of the body – vertical tail
combination in the wind tunnel at Re ¼ 470000 and α ¼ 0 . Fig. 56. Non-linear effects on the fuselage yawing moment coefficient of the body –
vertical tail combination in the wind tunnel at Re ¼ 470000 and α ¼ 0 .
163
D. Ciliberti et al. Progress in Aerospace Sciences 95 (2017) 140–172
Fig. 57. Comparison of numerical and experimental data on the wing – body – vertical tail
Fig. 58. Comparison of numerical and experimental data on complete aircraft configu-
combination. Re ¼ 470000, α ¼ 0 .
ration. Re ¼ 470000, α ¼ 0 .
the load cell, since no contact can be allowed with the fuselage to avoid
the direct transmission of the aerodynamic loads from the tail to the complete model weighs 50 kg, with a center of gravity close to the wing
body, to measure the aerodynamic forces acting on the vertical tail in leading edge.
each aircraft configuration. For this reason, an empty space, few milli-
meters thick, between the vertical tail root section and the fuselage has
7.2. The wind tunnel facility
been provided. The empennage assembly is shown in Fig. 40 and a
description of the load path is reported in Fig. 41. The idea is to get
The experimental tests campaign has been performed in the main
aerodynamic data from both the vertical tail (through the load cell) and
subsonic wind tunnel facility of the Department of Industrial Engineering
the entire aircraft (through the wind tunnel balance) at the same time, to
(DII). This is a subsonic, closed-circuit tunnel shown in Fig. 42, with a
evaluate the aerodynamic interference following the approach of the
tempered rectangular cross section. The main wind-tunnel characteristics
metamodel described in Section 6.
are summarized in Table 7. The main components of the closed-circuit
The aircraft model has been mainly realized by computer numerical
tunnel are described in detail in Ref. [47].
control (CNC) milling. The non-metal parts are made from a high density
The measurement instrumentation consists of an internal strain gage
machinable polyurethane slab named PROLAB 65, which properties are
balance for the measurement of aerodynamic forces and moments, a
described in Table 6. The total milled surface is about 3.90 m2. The
Venturi system to measure the dynamic pressure, a potentiometer to
164
D. Ciliberti et al. Progress in Aerospace Sciences 95 (2017) 140–172
Fig. 60. The effect of the horizontal tail on the vertical tail. Comparison between the curves derived from numerical analyses and the experimental data.
165
D. Ciliberti et al. Progress in Aerospace Sciences 95 (2017) 140–172
Table 8
Comparison of semi-empirical methods with wind tunnel data for zftc/rf ¼ 1.0 (high tail-cone).
speed, using the Bernoulli incompressible equation, and to obtain the corresponding distances for flight operations. In addition, the flow
mass density through the equation of state. properties in the test section may not be the same in space and time. To
The control instrumentation used in this test campaign consists of a include appropriate corrections, the effects of horizontal buoyancy, solid
kinematic mechanism (handled by the operator) with a crank handle and wake blockages, streamline curvature, and normal, spanwise, and
fixed at the end of a horizontal shaft acting as worm-screw. This shaft tail downwash have been considered as described in Ref. [48]. For a
transmits the rotatory motion to the vertical axis of a small diameter gear typical configuration, as shown in Fig. 43, the solid blockage is 0.0115,
wheel. The rotatory motion is then transmitted to a larger gear wheel whereas for the fuselage alone it is 0.0101.
through a steel chain reducing the angular velocity of the model. A steel Since the Reynolds number of scaled aircraft models in low-speed
plate, which is at level with the floor, is fixed to the axis of the second wind tunnels cannot be as large as those occurring in real flight condi-
gear wheel. The steel plate allows the whole assembly sting-balance- tions, the boundary-layer transition on wind tunnel models must be
model to rotate. promoted artificially. In the experimental campaign presented here this
The instrumentation for acquisition and elaboration consist of: a 16 has been achieved by means of trip strips, made of adhesive tape with
channels SPARTAN system (produced by Imc DataWorks, LLC) for the triangular edges, placed on all components of the aircraft. The thickness
acquisition and conversion into 16 bit of output data coming from the and the right position of the trip strips have been estimated by arranging
measurement instrumentation; a desktop PC, provided with an interface ad hoc flow visualizations using fluorescent oil. Results led to the
software for the A/D converter; a software for the elaboration and conclusion that 2 layers of tape are sufficient to get the boundary layer
visualization of the acquired data. The software, named WT6, has been transition at the desired place. The location of the trip strips is at about
developed by researchers of the DII laboratory. 5% local chord for wing and horizontal tail, even closer to the leading
edge for the vertical tail, whereas they have been placed at 20% nose
7.3. Wind tunnel corrections, transitional strips, and uncertainty of length on the fuselage.
measures Effects of uncertainty of measurements on the lateral-directional
static stability derivatives have been estimated below 3%. Details
In wind tunnel testing there are some constraints due to the nature of about the wind tunnel preliminary tests, including the effects of transi-
the tunnel itself. While there is no difference in having the model at rest tion, the assessment of the uncertainty, as well as further details on the
and the air moving around it, the distances of some or all the stream CAD model are available in Ref. [46].
boundaries from the model under test are usually less than the
166
D. Ciliberti et al. Progress in Aerospace Sciences 95 (2017) 140–172
Table 9
Comparison of semi-empirical methods with wind tunnel data for zftc/rf ¼ 0.5 (mid tail-cone).
7.4. Wind tunnel results between β ¼ 6 and þ6 (Cxβ represents the generic directional stability
derivative). The Reynolds number based on wing mean aerodynamic
The reference system adopted, shown in Fig. 44, has the origin of the chord is 470000. Features of the aircraft configurations are shown
axes at the balance center, with the x-axis parallel to the fuselage from Figs. 46–49.
waterline and positive towards the fuselage stern, the y-axis perpendic- Comparisons between numerical and wind tunnel tests are here pre-
ular to the symmetry plane and positive towards the right wing, and the sented for two configurations with the high after-body, the vertical tail
z-axis perpendicular to the other two and positive upward. The sideslip planform with Av ¼ 1.5, high and low wing, and T-tail horizontal stabi-
angle is considered positive when both the wind components are posi- lizer, shown in Fig. 50. Results for the isolated components (vertical tail
tive, i.e. when the side wind is coming from the left wing. With this and fuselage) are reported in Figs. 51 and 52, where the significant dif-
reference system, for a positive sideslip angle, a yawing moment N > 0 ference between numerical and experimental data at high sideslip angles
means the model is directionally stable, whereas a rolling moment L < 0 is worthy of further investigations.
means the model is laterally stable. Fig. 53 shows the results for the body - vertical tail combination (BV).
In other words: The total and the vertical tail contributions, BV and V respectively, have
been directly measured, whereas the fuselage contribution B has been
CNβ > 0 is required for directional stability; measured as the difference (CN CNv). The numerical results for the
CL β < 0 is required for lateral stability. vertical tail contribution are overlapped by experimental measurements,
therefore the numerical prediction of the vertical tail aerodynamics in the
Yawing, rolling moments, and their derivatives have been reduced to presence of other components is correct. As stated in Section 5.4.1, the
the mid wing aerodynamic center, which is also the reference point in the fuselage in sideslip conditions exhibits a crossflow similar to a two-
numerical analyses of Section 5.4. Its location related to the balance dimensional flow field around a cylinder, where the maximum local
center is shown in Fig. 45. The quantities of interest are the model total velocity occurs at the top of the cylinder and decays further downstream
yawing moment coefficient derivative CNβ and the vertical tail yawing to the freestream cross-flow value V∞ sinβ at distance from the body
moment coefficient derivative CNβv, approximated as the incremen- surface. This phenomenon tends to increase the effectiveness of the
tal ratio vertical tail: the fuselage directly alters the vertical tail incidence because
of the cross-flow around the body. At the same time, the vertical tail
Cx ðβ1 Þ Cx ðβ2 Þ reduces the fuselage instability in sideslip. In fact, the experimental re-
Cxβ ¼ (16)
β 1 β2 sults reported from Figs. 54–56 show that, as the sideslip angle increases
167
D. Ciliberti et al. Progress in Aerospace Sciences 95 (2017) 140–172
Table 10
Comparison of semi-empirical methods with wind tunnel data for zftc/rf ¼ 0 (low tail-cone).
over a certain value, the total yawing moment coefficient CN increases its by the significant agreement between numerical and experimental data
slope, while the slope of the vertical tail contribution CNv remains almost about the vertical tail in Fig. 58. The maximum deviation from wind
constant up to the stall, and the fuselage instability, represented by the tunnel results of the stability derivative CNβ predicted by CFD is about
negative slope of the CNf curve, is decreased. While the standard semi- 15% on the isolated fuselage and from 2% to 6% on the vertical tail in all
empirical methods consider the fuselage contribution to aircraft direc- the configurations.
tional stability unaffected by the presence of the vertical stabilizer [9], The novel method of Section 6 has also been verified with wind
these wind tunnel tests have shown the true repartition of the aero- tunnel data. Some comparison between the curves of the proposed
dynamic forces. method and the data points extracted from several wind tunnel tests are
The wing gives a direct and favorable contribution to airplane presented here. Fig. 59 shows that the effect of the fuselage on the ver-
directional stability if it has a positive sweep angle [1]. For any sweep tical tail is captured with acceptable approximation. The difference be-
angle, the wing has always an indirect effect due to the aerodynamic tween results at zftc/rf ¼ 0.5 is due to the arrangement of the tested
interference. The vortex system developed by the wing-fuselage combi- assembly, designed to avoid the direct contact of the surfaces of the
nation in sideslip, named sidewash and analogous to the downwash in the vertical tail and the fuselage, to allow a direct measure of the aero-
longitudinal plane, indirectly affects the incidence of the vertical tail. dynamic forces on the fin by means of the load cell installed in the
This effect is such to increase the vertical tail contribution to directional fuselage tail-cone. Such arrangement provides a gap at the intersection
stability if the wing is positioned low in fuselage, the contrary happens if between the vertical tail and the fuselage. This gap has been somewhat
the wing is high. The comparison between numerical and experimental reduced by using tape to extend the vertical tail root downward to pre-
results is reported in Fig. 57, for the WBV configuration with high and vent cross-flow at the intersection, yet complete sealing of the assembly
low wing respectively. Again, the total and the vertical tail contributions, could not be achieved. The effect of the horizontal tail is shown in Fig. 60
WBV and V respectively, have been directly measured, whereas the for several possible configurations. This is well predicted numerically,
wing-body contribution WB has been measured as the difference (CN especially for the high-winged, high tail-cone configuration, which rep-
CNv). resents the typical regional turboprop layout. Some data points are out of
With the horizontal tail mounted on the vertical tail, the off-center the trend highlighted from numerical analyses. This is probably due to
load cell measures the aerodynamic forces on the entire empennage misalignment during assembly and gaps between the empennage
(VH). However, the horizontal tail provides only aerodynamic interfer- components.
ence and not a direct contribution to directional stability, as also proved
168
D. Ciliberti et al. Progress in Aerospace Sciences 95 (2017) 140–172
Fig. 61. Relative error of semi-empirical methods with respect to wind tunnel data for high tail-cone configurations.
Fig. 62. Relative error of semi-empirical methods with respect to wind tunnel data for mid tail-cone configurations.
169
D. Ciliberti et al. Progress in Aerospace Sciences 95 (2017) 140–172
Fig. 63. Relative error of semi-empirical methods with respect to wind tunnel data for low tail-cone configurations.
8. Application of the proposed modern method and comparison ○ WL - wing low zw/rf ¼ 1
with other semi-empirical methods and experimental data Vertical tail is indicated by a letter and two digits (Fig. 48)
○ V10 - Av ¼ 1.0
The proposed semi-empirical method for the evaluation of aircraft ○ V15 - Av ¼ 1.5
directional stability has been implemented in JPAD [49], a Java-based ○ V20 - Av ¼ 2.0
computing library for aircraft designers, developed by the authors at Horizontal tail is indicated by a letter and a digit (Fig. 49)
the University of Naples Federico II. A standalone Java application ○ H1 – body-mounted tailplane zh/bv1 ¼ 0
dedicated to the aircraft directional stability has been provided for this ○ H2 – low cruciform tailplane zh/bv1 ¼ 0.38
work and it has been called within a MATLAB script for each of the ○ H3 – cruciform tailplane zh/bv1 ¼ 0.62
complete aircraft configurations investigated in the wind tunnel test ○ H4 – high cruciform tailplane zh/bv1 ¼ 0.82
campaign. The aerodynamic interference factors predicted by numerical ○ H5 – T-tail configuration zh/bv1 ¼ 1.0
analyses have been collected as multi-dimensional arrays and stored into
hierarchical data format HDF5 files. The Java application reads an XML Experimental data for the V10_H2 and V15_H2 combinations are not
input file, interrogates the database, if necessary it performs a linear available due to manufacturing reasons, hence they are not presented.
interpolation among the available data, provides the interference factors It can be observed that DATCOM and ESDU results do not change with
and the aerodynamic coefficients, then writes the results in an output fuselage after-body shape, as the aerodynamic interference is calculated
XML file. Each call to the Java application returns a point of a linear as function of the vertical tail planform, vertical tail size with respect to
surface response of the metamodel described in Section 6. Future the fuselage after-body, and wing position with respect to the fuselage
development of the code may include other polynomial surface re- [9,17], while the relative position of the vertical tail with respect to the
sponses. The MATLAB scripts written for this work calculates the vertical fuselage central body and wing is not considered.
tail directional stability derivatives with DATCOM [9] and ESDU [17] To get a clearer view of the reliability of the three methods,
methods, then calls the Java application that applies the alternative Figs. 61–63 show the relative error with respect to the wind tunnel data,
method, and compare the results. Data are reported from Tables 8–10, highlighting that the alternative method performs better than the clas-
assuming wind tunnel results as reference, with the following sical semi-empirical methods. DATCOM and ESDU methods well predict
nomenclature: configurations with high wing (WH) and high tail-cones (BH), achieving
an error below 15%. The relative error from wind tunnel data increases
Aircraft components IDs are separated by underscores _ up to 40% for mid wing (WM) configurations and up to 65% for low wing
Body is indicated by two letters (Fig. 46) configurations (WL). The proposed method exhibits an error below 20%
○ BH - body high zftc/rf ¼ 1.0 for all the configurations. For the body-mounted and T-tail configurations
○ BM - body mid zftc/rf ¼ 0.5 the error is below 5%, except for the low wing (WL), mid tail-cone (BM)
○ BL - body low zftc/rf ¼ 0 configuration. As expected, the new method is generally more reliable
Wing is indicated by two letters (Fig. 47) than the standard methods on the regional turboprop aircraft configu-
○ WH - wing high zw/rf ¼ 1 rations, because it has been developed from high fidelity analyses.
○ WM - wing mid zw/rf ¼ 0
170
D. Ciliberti et al. Progress in Aerospace Sciences 95 (2017) 140–172
9. Conclusion [13] M.J. Queijo, W.D. Wolhart, Experimental Investigation of the Effect of the Vertical-
tail Size and Length and of Fuselage Shape and Length on the Static Lateral Stability
Characteristics of a Model with 45 Sweptback Wing and Tail Surfaces, Report
The authors have reviewed the main semi-empirical methods avail- 1049, National Advisory Committee for Aeronautics, 1950.
able in literature for the evaluation of aircraft directional stability and [14] J.D. Brewer, J.H. Lichtenstein, Effect of Horizontal Tail on Low-speed Static Lateral
assessed their applicability in aircraft design with computational aero- Stability Characteristics of a Model Having 45 Sweptback Wing and Tail Surfaces,
Technical Note 2010, National Advisory Committee for Aeronautics, 1950.
dynamics examples. This review has highlighted the limits of the avail- [15] H.R. Pass, Analysis of Wind-tunnel Data on Directional Stability and Control,
able approaches, based on the results of wind tunnel tests performed in Technical Note 775, National Advisory Committee for Aeronautics, 1940.
the past decades on aircraft shapes quite different from modern transport [16] H.E. Murray, Wind-tunnel Investigation of End-plate Effects of Horizontal Tails on a
Vertical Tail Compared with Available Theory, Technical Note 1050, National
airplanes, and the need of a more reliable preliminary design method for Advisory Committee for Aeronautics, 1946.
airplane vertical stabilizers. This has led to the development of an up-to- [17] R.W. Gilbey, Contribution of Fin to Sideforce, Yawing Moment and Rolling Moment
date semi-empirical method, based on the results of CFD RANS simula- Derivatives Due to Sideslip, (Yv)F, (Nv)F, (Lv)F, in the Presence of Body, Wing and
Tailplane, Item 82010, ESDU, 1982.
tions on hundreds of aircraft configurations obtained by varying the [18] J. Weber, A.C. Hawk, Theoretical Load Distributions on Fin-body-tailplane
layout of a modern representative regional transport airplane. The Arrangements in a Side-wind, Reports and memoranda no. 2992, Ministry of
method has been validated with wind tunnel results and performs better Supply, London, UK, 1954.
[19] F. Nicolosi, P. Della Vecchia, D. Ciliberti, An investigation on vertical tailplane
than the other semi-empirical methods. This will hopefully reduce the contribution to aircraft sideforce, Aerosp. Sci. Technol. 28 (1) (2013) 401–416.
sizing uncertainties in the aircraft preliminary design phase. The meth- [20] F. Nicolosi, P. Della Vecchia, D. Ciliberti, S. Corcione, V. Cusati, A comprehensive
odology has also been successfully implemented in the AGILE framework review of vertical tail design, Aircr. Eng. Aerosp. Technol. 89 (Issue 4) (2017).
[21] J.B. Vos, A. Rizzi, D. Darracq, E.H. Hirschel, Navier-Stokes solvers in European
for the design and optimization of turboprop aircraft. Following the
aircraft design, Prog. Aerosp. Sci. 38 (8) (2002) 601–697.
AGILE paradigm, the method can be continuously enhanced. For [22] R.M. Hall, et al., Computational Methods for Stability and Control (COMSAC): the
instance, further analyses can extend it to account for the effect of wing Time Has Come, Tech. rep. 6121, American Institute of Aeronautics and
sweep (DATCOM provides a small effect on the sidewash, while ESDU Astronautics, 2005.
[23] L.L. Erickson, Panel Methods: an Introduction, vol. 2995, National Aeronautics,
does not account for sweep angle) and to include directional control Space Administration, Office of Management, Scientific, and Technical Information
derivatives. It has to be highlighted that the approach adopted to develop Division, 1990.
the proposed method is also applicable to other types of research. In such [24] Terzi A, Chiu TW. Modern panel method techniques for modeling wake body
interference. In: 28th Fluid Dynamics Conference, Snowmass Village (Colorado),
cases a convenient database of numerical results will be generated using 1997.
high fidelity tools and, successively, some meaningful flow conditions [25] H. Dutt, S. Rajeswari, Wing-body interference using a hybrid panel method, Acta
will be verified in wind tunnel experiments. This could be useful in Mech. 106 (3–4) (1994) 111–126.
[26] F.T. Johnson, E.N. Tinoco, N.J. Yu, Thirty years of development and application of
assessing the stability derivatives of innovative aircraft configurations, CFD at Boeing commercial airplanes, Seattle, Comput. Fluids 34 (10) (2005)
which are not provided in standard semi-empirical methods, and fully 1115–1151.
complies with the AGILE paradigm. [27] J.S. Shang, Three decades of accomplishments in computational fluid dynamics,
Prog. Aerosp. Sci. 40 (3) (2004) 173–197.
[28] K. Fujii, Progress and future prospects of CFD in aerospace – wind tunnel and
Appendix A. Supplementary data beyond, Prog. Aerosp. Sci. 41 (6) (2005) 455–470.
[29] F. Nicolosi, S. Corcione, P. Della Vecchia, Commuter aircraft aerodynamic design:
wind-tunnel tests and CFD analysis, in: ICAS 2014 Proceedings, International
Supplementary data related to this article can be found at https://doi.
Council of the Aeronautical Sciences. Optimage Ltd., 2014.
org/10.1016/j.paerosci.2017.11.001. [30] F. Nicolosi, P. Della Vecchia, S. Corcione, Design and aerodynamic analysis of a
twin-engine commuter aircraft, Aerosp. Sci. Technol. 40 (2015) 1–16.
References [31] F. Nicolosi, S. Corcione, P. Della Vecchia, Commuter aircraft aerodynamic
characteristics through wind tunnel tests, Aircr. Eng. Aerosp. Technol. 88 (4)
(2016) 523–534.
[1] C.D. Perkins, R.E. Hage, Airplane Performance Stability and Control, Wiley, New [32] P. Della Vecchia, F. Nicolosi, Aerodynamic guidelines in the design and
York, 1949, ISBN 9780471680468. optimization of new regional turboprop aircraft, Aerosp. Sci. Technol. 38 (2014)
[2] E. Obert, Aerodynamic Design of Transport Aircraft, Ios Press, Amsterdam, The 88–104.
Netherlands, 2009, ISBN 9781586039707. [33] P. Della Vecchia, D. Ciliberti, Numerical aerodynamic analysis on a trapezoidal
[3] E. Torenbeek, Synthesis of Subsonic Airplane Design, Delft University Press, Delft, wing with high lift devices: a comparison with experimental data, in: 22th AIDAA
1982, ISBN 9024727243. Conference, Associazione Italiana di Aeronautica e Astronautica, Naples (Italy),
[4] D.P. Raymer, in: Aircraft Design: a Conceptual Approach. AIAA Education Series, 2013.
fifth ed., American Institute of Aeronautics and Astronautics, Reston, VA, 2012, [34] Nicolosi F, Ciliberti D, Della Vecchia P. Aerodynamic design guidelines of aircraft
ISBN 9781600869112. dorsal fin. In: 34th AIAA Applied Aerodynamics Conference, Washington D.C.,
[5] S. Gudmundsson, General Aviation Aircraft Design: Applied Methods and 2016.
Procedures, Butterworth-Heinemann, 2013, ISBN 9780123973290. [35] D. Ciliberti, F. Nicolosi, P. Della Vecchia, A new approach in aircraft vertical
[6] Ciampa PD, Nagel B. The AGILE Paradigm: the next generation of collaborative tailplane design, in: 22th AIDAA Conference, Associazione Italiana di Aeronautica e
MDO. In: 18th AIAA/ISSMO Multidisciplinary Analysis and Optimization Astronautica, Naples (Italy), 2013.
Conference, Denver, Colorado, 2017. [36] F. Nicolosi, P. Della Vecchia, D. Ciliberti, Development of new preliminary design
[7] Lefebvre T, Bartoli N, Dubreuil S, Panzeri M, Lombardi R, D'Ippolito R, Della methodologies for regional turboprop aircraft by CFD analyses, in: ICAS 2014
Vecchia P, Nicolosi F, Ciampa PD. Methodological enhancements in MDO process Proceedings, International Council of the Aeronautical Sciences. Optimage Ltd,
investigated in the AGILE European project. In: 18th AIAA/ISSMO Multidisciplinary 2014.
Analysis and Optimization Conference, Denver, Colorado, 2017. [37] Della Vecchia P, Nicolosi F, Ciliberti D. Aircraft directional stability prediction
[8] Della Vecchia P, Stingo L, Corcione S, Ciliberti D, Nicolosi F, De Marco A, Nardone method by CFD. In: 33rd AIAA Applied Aerodynamics Conference, Dallas (Texas)
G. Game theory and evolutionary algorithms applied to MDO in the AGILE 2015.
European project. In: 18th AIAA/ISSMO Multidisciplinary Analysis and [38] F. Nicolosi, P. Della Vecchia, D. Ciliberti, V. Cusati, Fuselage aerodynamic
Optimization Conference, Denver, Colorado, 2017. prediction methods, Aerosp. Sci. Technol. 55 (2016) 332–343.
[9] R.D. Finck, USAF Stability and Control DATCOM. AFWAL-TR-83–3048, McDonnell [39] F. Nicolosi, P. Della Vecchia, D. Ciliberti, Aerodynamic interference issues in
Douglas Corporation, Wright-Patterson Air Force Base, Ohio, 1978. aircraft directional control, J. Aerosp. Eng. 28 (1) (2015) 1–7.
[10] E.N. Jacobs, K.E. Ward, Interferences of Wing and Fuselage from Tests of 209 [40] STAR-CCMþ version 11.06 User Guide. Siemens.
Combinations in the NACA Variable Density Wind Tunnel, Technical Report 540, [41] Spalart PR, Allmaras SR. A one equation turbulence model for aerodynamic flows.
National Advisory Committee for Aeronautics, 1935. AIAA Journal 94.
[11] R.O. House, A.R. Wallace, Wind-tunnel Investigation of Effect of Interference on [42] C.L. Rumsey, et al., Summary of the first AIAA CFD high-lift prediction workshop,
Lateral-stability Characteristics of Four NACA 23012 Wings, an Elliptical and J. Aircr. 48 (6) (2011) 2068–2079.
Circular Fuselage and Vertical Fins, Report 705, National Advisory Committee for [43] A.P. Antunes, R.G. da Silva, J.F.L. Azevedo, On the effects of turbulence modeling
Aeronautics, 1940. and grid refinement on high lift configuration aerodynamic simulations, in: ICAS
[12] R.J. Bamber, R.O. House, Wind-tunnel Investigation of Effect of Yaw on Lateral- 2012 Proceedings, vol. 23, International Council of Aeronautics and Astronautics.
stability Characteristics. II – Rectangular NACA 23012 Wing with a Circular Optimage Ltd., 2012, p. 28.
Fuselage and a Fin, Report 730, National Advisory Committee for Aeronautics, [44] F.A. Diederich, Plan-form Parameter for Correlating Certain Aerodynamic
1939. Characteristics of Swept Wings, Technical Note 2335, National Advisory Committee
for Aeronautics, 1951.
171
D. Ciliberti et al. Progress in Aerospace Sciences 95 (2017) 140–172
[45] P.D. Chappell, R.W. Gilbey, Lift-curve Slope and Aerodynamic Centre Position of [48] J.B. Barlow, W.H. Rae, A. Pope, in: Low-speed Wind Tunnel Testing, third ed.,
Wings in Inviscid Subsonic Flow, Item 70011, Engineering Science Data Unit, 1970. Wiley, New York, 1999, ISBN 0471557749.
[46] D. Ciliberti, An Improved Preliminary Design Methodology for Aircraft Directional [49] F. Nicolosi, A. De Marco, L. Attanasio, P. Della Vecchia, Development of a Java-
Stability Prediction and Vertical Tailplane Sizing, PhD Thesis, YouCanPrint, 2016, based framework for aircraft preliminary design and optimization, J. Aerosp. Inf.
ISBN 9788892608825. Syst. 13 (6) (2016) 234–242.
[47] S. Corcione, Design Guidelines, Experimental Investigation and Numerical Analysis
of a New Twin Engine Commuter Aircraft, PhD thesis, 3D Tech, 2015, ISBN
889838212X.
172