AIAA SciTech Forum 10.2514/6.

2020-0020
6-10 January 2020, Orlando, FL
AIAA Scitech 2020 Forum

Control Surface Design Analysis and Actuation

Requirements Development for Munitions

Joshua T. Bryson1, Joseph D. Vasile2, Benjamin C. Gruenwald3, and Frank E. Fresconi4

U.S. Army Combat Capabilities Development Command - Army Research Laboratory, Aberdeen Proving Ground,
MD, 21005

Enhancing munition maneuverability is a key enabling technology for long range precision
munitions which enables both range extension and terminal maneuvers. One approach to
munition maneuverability is to deflect control surfaces to achieve body attack angles which
Downloaded by UNIVERSITY OF NEW SOUTH WALES on May 11, 2020 | http://arc.aiaa.org | DOI: 10.2514/6.2020-0020

generate lift from the body and other aerodynamic surfaces. This work presents a
methodology to analyze aerodynamic control surfaces to determine effectiveness at generating
a pitching or yawing moment on a projectile to attain an angle of attack and generate lateral
accelerations in flight. A methodology is then presented to develop actuator requirements to
ensure the control actuation system has sufficient torque to overcome aerodynamic loading
and sufficient bandwidth to stabilize the projectile dynamics. These methodologies are
demonstrated in this paper on a gun-launched, aerodynamically stabilized, fin-controlled
projectile, but are applicable to other control surface design problems.

Nomenclature
𝛼 = body angle of attack in pitch plane
𝛼̅ = total body angle of attack, √𝛼 2 + 𝛽 2
𝛽 = body angle of sideslip in yaw plane
𝐶𝐷 = drag coefficient
𝐶𝐿 = lift coefficient
𝐶𝑙0 = zeroth order roll moment coefficient
𝐶𝑙𝑝 = roll damping coefficient
𝐶𝑚 = coefficient of pitching moment
𝐶𝑚𝛼 , 𝐶𝑚 , 𝐶𝑚 = first, third, and fifth order fit coefficients for aerodynamic pitching moment
𝛼3 𝛼5
𝐶𝑚𝑞 = pitch damping coefficient
𝐻
𝐶𝑚 = aerodynamic hinge moment coefficient
𝐶𝑛𝛽 , 𝐶𝑛 , 𝐶𝑛 = first, third, and fifth order fit coefficients for aerodynamic yaw moment
𝛽3 𝛽5
𝐶𝑛𝑟 = yaw damping coefficient
𝐶𝑋0 , 𝐶𝑋 ̅ 2 = zeroth, and second order fit coefficients for X-axis aerodynamic force
𝛼
𝐶𝑌𝛼 , 𝐶𝑌 3 , 𝐶𝑌 = first, third, and fifth order fit coefficients for Y-axis aerodynamic force
𝛼 𝛼5
𝐶𝑍𝛼 , 𝐶𝑍 3 , 𝐶𝑍 5 = first, third, and fifth order fit coefficients for Z-axis aerodynamic force
𝛼 𝛼
𝛿1 , 𝛿2 , 𝛿3 , 𝛿4 = deflection angles for control surface 1,2,3,4
𝛿𝑞 = deflection of virtual/combined control surface for pitch
D = reference diameter
M = Mach number
𝑝 = roll rate
𝑞 = pitch rate

1
Mechanical Engineer, Weapons and Materials Research Directorate, FCDD-RLW-LE, AIAA Member.
2
Aerospace Engineer, Weapons and Materials Research Directorate, FCDD-RLW-LE, Senior AIAA Member.
3
Mechanical Engineer, Weapons and Materials Research Directorate, FCDD-RLW-LE, AIAA Member.
4
Mechanical Engineer, Weapons and Materials Research Directorate, FCDD-RLW-LE, Senior AIAA Member.
1
American Institute of Aeronautics and Astronautics

This material is declared a work of the U.S. Government and is not subject to copyright protection in the United States.
 Q = ½ ρV2 , dynamic pressure
𝑟 = yaw rate
S = D2π/4, aerodynamic reference area
𝑤 = Z-axis velocity
𝑤̇ = Z-axis acceleration

I. Introduction

R ECENTLY, there has been emphasis on investigating technologies and methodologies that can extend the range
of guided munitions in order to provide better coverage of the battlefield. Range extension improvements can be
achieved through rocket propulsion and gun propellant technology research to increase the projectile energy. Interior
ballistics research and propulsion technology research are being leveraged to extend range through increased launch
energy [1].
For a fixed launch energy, the munition range can be extended by enhancing the projectile maneuverability to
Downloaded by UNIVERSITY OF NEW SOUTH WALES on May 11, 2020 | http://arc.aiaa.org | DOI: 10.2514/6.2020-0020

generate lift through a positive body angle of attack, enabling range extension through gliding flight [2-4] as well as
terminal maneuvers. Projectile maneuverability is influenced by the airframe design and aerodynamics, as well as the
design and implementation of control surfaces [2-4]. Active research into low drag, high lift airframes for both
supersonic and subsonic flight regimes is improving understanding of the driving features and trends with airframe
design, and reducing design cycle iteration time to rapidly evolve capabilities [5].
This research presents a methodology to analyze control surface designs and their effect on the projectile,
quantifying the ability to generate lift through positive body angle of attack, along with the associated actuator
bandwidth and torque requirements. This research is illustrated on a generic fin-controlled projectile, but the approach
is applicable to other control surface design problems. The first goal for this projectile illustration problem is to size
the control surface to achieve lift-to-drag ratio of 3 across the supersonic and subsonic flight regimes with less than
20° required deflection on each control surface. A second goal is to estimate torque and bandwidth requirements for
the actuator to drive the control surface by 1) analyzing the aerodynamic hinge load that must be overcome by the
actuator and 2) by sweeping the natural frequency of a 2nd order actuator model and analyzing the effect on the
performance of the flight controller.

II. Airframe
The characteristics of the projectile outer mold line were shaped through a series of optimization analyses which
identified design candidates with low drag and high lift to drag ratios [5]. The optimization study focused on fin-
stabilized designs to improve maneuverability of the projectile. The projectile is designed to be sabot-launched from
an 8” diameter gun, using a smooth bore barrel or slip-band obturator. The projectile has no aerodynamic surfaces
deploying after launch. This 8” diameter gun requirement constrains the optimization to limit the fin span to 8” tip-
to-tip. The optimized control surface design for a given body baseline configuration with a 105mm diameter, 10
caliber length, and ogive length of 30% of the overall length of the projectile is shown in Figure 1. The design was
optimized to maximize lift to drag, minimize drag and meet a desired static margin value (i.e. 0.3) across supersonic
Mach regime (M = 1-4) at an 8° body angle attack. This 8° body angle was selected based on preliminary trim angle
predictions for the vehicle with trailing edge flap deflections. Table 1 shows the projectile mass properties.

Figure 1. Illustration of the projectile flight body. Dimensions are given in mm.

2
American Institute of Aeronautics and Astronautics
 Table 1. Mass properties for projectile
Mass Properties
Mass 14.8 kg
CGX 630mm from nose
CGY, CGZ on center line
𝑰𝑿𝑿 0.0273 kg-m2
𝑰𝒀𝒀 , 𝑰𝒁𝒁 1.17 kg-m2

A trailing edge flap on each of the four fins is rotated about the leading edge to provide control of the projectile
during flight. Figure 1 shows the trailing edge flap as 12cm, but a variant with 8cm trailing edge flaps is also
considered in this study. The aerodynamic forces and moments of the configuration were obtained using both semi-
empirical aerodynamic prediction code Missile DATCOM (release 2014) [6], as well as NASA’s Cartesian Euler CFD
analysis package Cart3D (1.5.5) [7]. Only the drag and dynamic derivatives for the airframe were predicted using
Downloaded by UNIVERSITY OF NEW SOUTH WALES on May 11, 2020 | http://arc.aiaa.org | DOI: 10.2514/6.2020-0020

DATCOM, whereas all other forces and moments (e.g., pitching moment of airframe, normal force of trailing edge
flap, etc.) were predicted by Cart3D. The aerodynamic data for the trailing edge flap components was found by
simulating a single trailing edge flap (i.e. flap 3) in the cruciform orientation at multiple deflections (i.e. δ = 0° , 5°,
10°, 15°, 25°, and 30°) across all flight conditions. The trailing edge flap data was then applied to the other flaps (i.e.
flap 1, 2, and 4) and combined with the rest of the airframe as discussed is Section III. The numbering scheme of the
control flaps as well as the deflection sign convention is presented in Figure 2.

Figure 2. Numbering scheme and deflection sign convention for the trailing edge control surfaces compared
to the body-fixed coordinate system. View is from projectile base.

III. Aerodynamic Model

The aerodynamic model provides the aerodynamic forces and moments at a given angle of attack and Mach number
using aerodynamic coefficient data [8, 9, 10]. Aerodynamic data describing the forces and moments due to the
movable fin tabs, termed movable aerodynamic surfaces (MAS) is applied separately from the aerodynamic data for
the assembly of the body and fixed fin surfaces, referred to as fixed aerodynamic surfaces (FAS).
The aerodynamic forces and moments from the FAS component are [𝐹𝑋 𝐹𝑌 𝐹𝑍 ]𝑇 and [𝑀𝐿 𝑀𝑀 𝑀𝑁 ]𝑇 ,
respectively, and are given in (1)-(6).
𝐹𝑋 = −𝑄𝑆 [𝐶𝑋0 (𝑀) + 𝐶𝑋 ̅ 2 (𝑀) sin2 𝛼̅] (1)
𝛼

𝐹𝑌 = −𝑄𝑆 [𝐶𝑌𝛽 (𝑀) sin 𝛽 + 𝐶𝑌 3 (𝑀) sin3 𝛽 + 𝐶𝑌 5 (𝑀) sin5 𝛽] (2)

𝛽 𝛽

𝐹𝑍 = −𝑄𝑆 [𝐶𝑍𝛼 (𝑀) sin 𝛼 + 𝐶𝑍 3 (𝑀) sin3 𝛼 + 𝐶𝑍 5 (𝑀) sin5 𝛼] (3)

𝛼 𝛼

3
American Institute of Aeronautics and Astronautics
 𝑝𝐷
𝑀𝐿 = 𝑄𝑆𝐷 [𝐶𝑙0 (𝑀) + 𝐶𝑙𝑝 (𝑀) ] (4)
2𝑉
𝑞𝐷
𝑀𝑀 = 𝑄𝑆𝐷 [𝐶𝑚𝛼 (𝑀) sin 𝛼 + 𝐶𝑚 3 (𝑀) sin3 𝛼 + 𝐶𝑚 5 (𝑀) sin5 𝛼 + 𝐶𝑚𝑞 (𝑀) ] (5)
𝛼 𝛼 2𝑉
𝑟𝐷
𝑀𝑁 = 𝑄𝑆𝐷 [−𝐶𝑛𝛽 (𝑀) sin 𝛽 − 𝐶𝑛 3 (𝑀) sin3 𝛽 − 𝐶𝑛 5 (𝑀) sin5 𝛽 + 𝐶𝑛𝑟 (𝑀) ] (6)
𝛽 𝛽 2𝑉

Where: 𝛼 is the body angle of attack, 𝛽 is the body angle of sideslip, 𝛼̅ = √𝛼 2 + 𝛽 2 is the total body angle of attack,
1 𝜋
D is the projectile diameter, V is the projectile velocity, 𝑄 = 𝜌𝑉 2 is the dynamic pressure, and 𝑆 = 𝐷 2 is the
2 4
aerodynamic reference area. The MAS aerodynamic model is given in (7)-(12), which sums the force and moment
contributions of the 4 movable fin flap surfaces arrayed around the body.
Downloaded by UNIVERSITY OF NEW SOUTH WALES on May 11, 2020 | http://arc.aiaa.org | DOI: 10.2514/6.2020-0020

4
𝐹𝑋𝑀𝐴𝑆 = −𝑄𝑆 ∑ [𝐶𝑋𝑖 0 (𝑀, 𝛿𝑖 ) + 𝐶𝑋𝑖 2 (𝑀, 𝛿𝑖 ) sin2 𝛼̅] (7)
̅
𝛼
𝑖=1
4
𝐹𝑌𝑀𝐴𝑆 = −𝑄𝑆 ∑ [𝐶𝑌𝑖 𝛽 (𝑀, 𝛿𝑖 ) sin 𝛽 + 𝐶𝑌𝑖 3 (𝑀, 𝛿𝑖 ) sin3 𝛽 + 𝐶𝑌𝑖 5 (𝑀, 𝛿𝑖 ) sin5 𝛽] (8)
𝑖=1 𝛽 𝛽
4
𝐹𝑍𝑀𝐴𝑆 = −𝑄𝑆 ∑ [𝐶𝑍𝑖 𝛼 (𝑀, 𝛿𝑖 ) sin 𝛼 + 𝐶𝑍𝑖 3 (𝑀, 𝛿𝑖 ) sin3 𝛼 + 𝐶𝑍𝑖 5 (𝑀, 𝛿𝑖 ) sin5 𝛼] (9)
𝑖=1 𝛼 𝛼

𝑀𝐿𝑀𝐴𝑆 = 𝑄𝑆𝐷 [𝐶𝑙1𝛼 (𝑀, 𝛿1 ) sin 𝛼 + 𝐶𝑙2𝛽 (𝑀, 𝛿2 ) sin 𝛽 + 𝐶𝑙3𝛼 (𝑀, 𝛿3 ) sin 𝛼 + 𝐶𝑙4𝛽 (𝑀, 𝛿𝑖 ) sin 𝛽] (10)
4
𝑀𝐴𝑆 𝑖 (𝑀, ) 𝑖
𝑀𝑀 = 𝑄𝑆𝐷 ∑ [𝐶𝑚𝛼
𝛿𝑖 sin 𝛼 + 𝐶𝑚 3
(𝑀, 𝛿𝑖 ) sin3 𝛼 + 𝐶𝑚
𝑖
5
(𝑀, 𝛿𝑖 ) sin5 𝛼] (11)
𝛼 𝛼
𝑖=1
4
𝑀𝑁𝑀𝐴𝑆 = 𝑄𝑆𝐷 ∑ [−𝐶𝑛𝑖 𝛽 (𝑀, 𝛿𝑖 ) sin 𝛽 − 𝐶𝑛𝑖 3 (𝑀, 𝛿𝑖 ) sin3 𝛽 − 𝐶𝑛𝑖 5 (𝑀, 𝛿𝑖 ) sin5 𝛽] (12)
𝛽 𝛽
𝑖=1

While the FAS model coefficients are only dependent on Mach number, the MAS aerodynamic model is
populated with coefficients that depend on both Mach number and the deflection angle of the 𝑖 𝑡ℎ control surface, 𝛿𝑖 .
Both the FAS and MAS coefficients are calculated in the aerodynamic body coordinate frame, with +X out the tail,
+Y right, +Z up, as shown in Figure 3a, and are converted to force and moments in the standard flight dynamics
coordinate frame with +X out the nose, +Y right, +Z down, as shown in Figure 3b, through the formulation of (1)-
(12).

Figure 3. Comparison of a) the aerodynamic coordinate frame for the aero coefficients and b) the flight
dynamics coordinate frame for the calculated forces and moments.

IV. Control Surface Design Methodology

Once the coefficients have been developed to populate the aerodynamic model of the projectile FAS and MAS, an
aerodynamic trim analysis is performed to determine the body angle of attack and lateral acceleration due to varying
MAS deflections. This analysis is performed in the pitch plane for vertical accelerations, but represents sideways
acceleration capabilities as well due to the pitch and yaw symmetry of the projectile. In this pitch-plane analysis, the
individual deflections of flap 1 and flap 3 are combined together following the convention shown in Figure 2 to
produce a pitch deflection, 𝛿𝑞 , as defined in (13) below:

4
American Institute of Aeronautics and Astronautics
 𝛿𝑞 = −𝛿1 + 𝛿3 (13)
The total pitching moment coefficient, 𝐶𝑚 , is calculated from the FAS and MAS models for varying 𝛼 at each
deflection, 𝛿𝑞 . The equilibrium points (trim points) where 𝐶𝑚 = 0 indicates the steady state 𝛼 for a given 𝛿𝑞 . This
process is illustrated in Figure 4, which shows projectile trim angles of [0°, 11°, 19°, 23.5°] for 𝛿𝑞 of [0°, 10°, 20°,
30°] at Mach 2, as indicated with red circles.
Downloaded by UNIVERSITY OF NEW SOUTH WALES on May 11, 2020 | http://arc.aiaa.org | DOI: 10.2514/6.2020-0020

Figure 4. Total pitching moment, Cm, as well as the FAS and MAS Cm components vs 𝜶 for varying 𝜹𝒒 at
Mach 2.0. The pitch trim 𝜶 for each 𝜹𝒒 is found where the total Cm curves cross the Cm= 0 line, indicated
with red circles.

The total 𝐶𝑚 and total 𝐶𝑁 data for varying 𝛼 and 𝛿𝑞 for both the 8cm and 12cm control surface is presented using
a series of plots in Figure 5, with constant pitch deflection shown using solid lines and constant attack angle shown
using dashed lines. As before, the 𝐶𝑚 = 0 line indicates the steady state trim condition for the projectile during flight.
Inspecting the intersection between the lines of constant 𝛼 and constant 𝛿𝑞 along the 𝐶𝑚 = 0 line shows where the
projectile will trim for various deflections, or conversely, what deflection is required to trim at a particular angle of
attack. Both control surfaces are sufficient to generate high angles of attack (𝛼 > 10𝑜 ) with modest deflections in
both supersonic and subsonic flight (6𝑜 − 10𝑜 at Mach 2, 8𝑜 − 12𝑜 at Mach 0.6) Compared to the 8cm control
surface, the larger 12cm control surface generates a larger normal force and pitching moment for a given deflection,
and as a result requires smaller deflection values to trim at the same angle of attack, as expected.

5
American Institute of Aeronautics and Astronautics
Downloaded by UNIVERSITY OF NEW SOUTH WALES on May 11, 2020 | http://arc.aiaa.org | DOI: 10.2514/6.2020-0020

Figure 5. Total Cm and CN for 8cm (a, c, e) and 12cm (b, d, f) fin flap data as a function of body angle of
attack for, 𝜶, and pitch deflection, 𝜹 for Mach 0.6 (a, b), 2.0 (c, d), and 3.0 (e, f). Lines of constant deflection
are solid, while lines of constant angle of attack are dashed.

The lift coefficient, 𝐶𝐿 and drag coefficient, 𝐶𝐷 , at 𝛼 trim is calculated for various 𝛿𝑞 across Mach for the
different control surface sizes. Figure 6 shows plots of 𝐶𝐿 , 𝐶𝐷 , and the lift-to-drag ratio, 𝐶𝐿 /𝐶𝐷 . Maximizing 𝐶𝐿 /𝐶𝐷
is used as a metric for evaluating performance in this analysis, and the optimal 𝐶𝐿 /𝐶𝐷 values across Mach are shown
in black on the 𝐶𝐿 /𝐶𝐷 plots. The 𝐶𝐿 and 𝐶𝐷 components comprising the optimal 𝐶𝐿 /𝐶𝐷 ratio are also plotted in black
on their respective plots. Both control surface designs achieve similar maximum 𝐶𝐿 /𝐶𝐷 ratios of about 3 across the
supersonic and subsonic flight regimes.

6
American Institute of Aeronautics and Astronautics
Downloaded by UNIVERSITY OF NEW SOUTH WALES on May 11, 2020 | http://arc.aiaa.org | DOI: 10.2514/6.2020-0020

Figure 6. Comparison of 𝑪𝑳 and 𝑪𝑫 at trim for various pitch deflections across Mach. The optimal 𝑪𝑳 /𝑪𝑫
ratio across Mach is plotted in black in a) and b), with the corresponding 𝑪𝑳 and 𝑪𝑫 values plotted in black
on c), d) and e), f), respectively.

Figure 7 shows how the trim 𝛼 changes across Mach number for a set of given deflection angles. Note that the
lateral force and moment data is calculated using an inviscid flow solver and therefore does not predict flow separation
(i.e. onset of stall) accurately; the higher body angles of attack in Figure 6 and Figure 7 are likely not achievable.
Figure 8 facilitates comparisons between the 8cm and 12cm flap designs by plotting each pitch deflection schedule
across Mach to achieve the optimal 𝐶𝐿 /𝐶𝐷 ratio and the resulting body 𝛼. Both the 8cm and 12cm designs achieve
similar 𝛼 for optimal 𝐶𝐿 /𝐶𝐷 , but the 12cm control surface generally requires smaller deflections than the 8cm design,

7
American Institute of Aeronautics and Astronautics
 as expected. At Mach 4, the undeflected flight vehicle exhibits marginal stability, therefore, the flight vehicle
experiences large body angle of attack with small trailing edge flap deflection.
Downloaded by UNIVERSITY OF NEW SOUTH WALES on May 11, 2020 | http://arc.aiaa.org | DOI: 10.2514/6.2020-0020

Figure 7. Trim 𝜶 across Mach for varying pitch deflection angles for both the 8cm and 12cm control surface.
The 𝜶 corresponding to the maximum lift-to-drag is plotted in black.

Figure 8. Pitch deflection schedule across Mach to optimize lift-to-drag ratio for the 8cm and 12cm control
surface designs. Both designs achieve similar 𝜶 to optimize lift-to-drag, but the 12cm control surface
generally requires smaller deflections than the 8cm design.

V. Actuator Requirements: Torque

The deflection of a control surface requires an actuator with sufficient mechanical torque to overcome the
aerodynamic hinge moment induced by the airflow over the control surface. This hinge moment is calculated from
the aerodynamic normal force on the control surface and the moment arm between the mechanical hinge and the
aerodynamic CP. Component-level aerodynamic analysis provides hinge moment coefficient for the 𝑖 𝑡ℎ control
surface, 𝐻𝐶𝑚
𝑖
, which is a function of Mach, 𝛼, 𝛽, and 𝛿𝑖 . The aerodynamic hinge moment can be obtained from the
𝐻 𝑖
𝐶𝑚 value as shown below:
𝐻
𝑀𝑖 = 𝑄𝑆𝐷[ 𝐻𝐶𝑚
𝑖 (𝑀,
𝛼, 𝛽, 𝛿𝑖 )] (14)
Figure 9 plots 𝐻𝐶𝑚1
for control surface 1 across Mach for the 8cm and 12cm designs. In each case, the control surface
held at the optimal 𝛿𝑞 with the projectile at the optimal trim angle, 𝛼, as previously shown in Figure 8.

8
American Institute of Aeronautics and Astronautics
Downloaded by UNIVERSITY OF NEW SOUTH WALES on May 11, 2020 | http://arc.aiaa.org | DOI: 10.2514/6.2020-0020

Figure 9. Coefficient of aerodynamic hinge load experienced by actuator 1 at optimal 𝜹𝒒 and projectile trim
𝜶 across Mach for the 8cm and 12cm control surface.

Assuming constant density and speed of sound at a constant altitude, the dynamic pressure, 𝑄, can be calculated
across Mach for the 105mm diameter projectile. Using these terms, and assuming the 𝛿𝑝 and 𝛼 schedule given in
Figure 8, the aerodynamic moment experienced at the actuator can be recovered from 𝐻𝐶𝑚 1
. Figure 10 plots this hinge
moment for control surface 1 at three different altitudes: Sea level, 5km, and 10km. Both the 8cm and 12cm control
surface require less than 2Nm of actuator torque to overcome the aerodynamic forces at subsonic speeds. CP
movement and higher dynamic pressure at high Mach number cause the 12cm control surface to generate significantly
higher aerodynamic hinge moments at Mach 2-3. Overall, the hinge moment generated from the 8cm control surface
is flatter across Mach, regardless of altitude. The large magnitude destabilizing moments (show as negative in Figure
8) at higher Mach numbers are of particular concern for actuator design, because they tend to cause deflection angles
to increase, potentially destabilizing the projectile if the actuator is underpowered.

Figure 10. Aerodynamic hinge moment experienced at the actuator for control surface 1 at optimal 𝜹𝒒 and
optimal projectile trim 𝜶 across Mach for the 8cm and 12cm control surface designs at sea level, 5km, and
10km altitude.

Designing an actuator capable of enough output torque to overcome the aerodynamic hinge moments given in
Figure 10 is challenging, and the realized projectile maneuverability may be limited in part by the output torque
capability of the actuator, particularly in the supersonic flight regimes. Figure 11 shows the impact to performance in

9
American Institute of Aeronautics and Astronautics
 the case of an actuator limited to 2Nm output torque for both the 8cm and 12cm control surface designs. Figure 11 (a)
and (b) plot the 2Nm torque-limited maximum deflection angles at different altitudes across Mach, along with the
optimal deflection angle goal in black. Figure 11 (c) and (d) plot the resulting body trim 𝛼 achieved by the limited 𝛿
for various flight altitudes, along with the optimal 𝛼 in black. Figure 11 (e) and (f) plot the 𝐶𝐿 /𝐶𝐷 ratio for the projectile
for the achievable 𝛿 and 𝛼, along with the optimal 𝐶𝐿 /𝐶𝐷 in black. The 2Nm actuator is able to drive both the 8cm
and 12cm control surface designs to meet the optimal 𝐶𝐿 /𝐶𝐷 in the subsonic flight regime, but limits the achievable
𝐶𝐿 /𝐶𝐷 in the transonic flight regime for both the 8cm and 12 cm control surface. In the supersonic Mach 2-3 flight
regime, both the 8cm and 12cm control surface designs are limited by the 2Nm actuator, but the achievable 𝐶𝐿 /𝐶𝐷 is
better for the 8cm design, especially as altitude is increased above sea level.
Downloaded by UNIVERSITY OF NEW SOUTH WALES on May 11, 2020 | http://arc.aiaa.org | DOI: 10.2514/6.2020-0020

Figure 11. Achievable performance estimates for 2Nm actuator torque limit for both the 8cm and 12cm control
surface across Mach for several flight altitudes. The limited actuator torque limits the achievable 𝜹, (a)-(b),
which limits the achievable 𝜶 trim, (c)-(d), resulting in sub-optimal 𝑪𝑳 /𝑪𝑫 , (e)-(d).

VI. Actuator Requirements: Bandwidth

In addition to the torque requirements for the actuator, the actuator bandwidth requirements can also be estimated
for each of the control surface designs from the flight control design for the projectile. Since the actuator is the physical
channel between the flight control algorithm and the projectile, it is necessary to first understand the flight control
requirements. The actuator bandwidth requirements can then be determined such that the flight control algorithm can
be properly applied to the projectile. For this purpose, a linearized plant model of the projectile in the pitch plane is
represented here as
𝑥̇ p (𝑡) = 𝐴p 𝑥p (𝑡) + 𝐵p 𝑢(𝑡), (15)
with 𝑥p (𝑡) = [𝑞(𝑡), 𝑤̇ (𝑡)]T being the plant state vector, 𝑢(𝑡) = [𝛿1 (𝑡), 𝛿3 (𝑡)]T being the pitch control input vector

10
American Institute of Aeronautics and Astronautics
 𝑄𝑆𝐷2 𝑚𝐷 𝐶𝑚𝛼
𝐶   𝑄𝑆𝐷 𝑄𝑆𝐷
2𝐼𝑧𝑧 𝑉 𝑚𝑞 𝐼𝑧𝑧 𝐶𝑍𝛼 − 𝐶 𝐶
𝐴p = , and 𝐵p = [ 2𝐼𝑧𝑧 𝑚𝛿𝑞 2𝐼𝑧𝑧 𝑚𝛿𝑞 ].
𝑄𝑆 𝑄𝑆
0 0
[ 𝑚 𝐶𝑍𝛼 − 𝐶
𝑚𝑉 𝑍𝛼 ]
Investigation of the airframe dynamics shows the roll dynamics are slower than the pitch plane dynamics for this
projectile. Additionally, the yaw dynamics are the same as the pitch dynamics due to symmetry, so the pitch dynamics
will dictate the overall actuator requirements.
The desired objective for the flight control is to track a desired pitch acceleration 𝑤̇ d (𝑡) while regulating the pitch
rate 𝑞(𝑡) to zero. This can be achieved by first augmenting the integrator dynamics given by
𝑧̇ (𝑡) = 𝑤̇ (𝑡) − 𝑤̇ d (𝑡), (16)
with the linearized plant dynamics (15), which can be written as
Downloaded by UNIVERSITY OF NEW SOUTH WALES on May 11, 2020 | http://arc.aiaa.org | DOI: 10.2514/6.2020-0020

𝑥̇ (𝑡) = 𝐴𝑥(𝑡) + 𝐵𝑢(𝑡) + 𝐵r 𝑐(𝑡), (17)

T
where 𝑥(𝑡) = [𝑥pT (𝑡), 𝑧(𝑡)] is the augmented state vector, 𝑐(𝑡) = 𝑤̇ d (𝑡) is the desired tracking command,
𝑄𝑆𝐷2 𝑚𝐷 𝐶𝑚𝛼
𝐶𝑚𝑞   0 𝑄𝑆𝐷 𝑄𝑆𝐷
2𝐼𝑧𝑧 𝑉 𝐼𝑧𝑧 𝐶𝑍𝛼 − 𝐶𝑚𝛿𝑞 𝐶 0
𝐴 = 𝑄𝑆 ,    𝐵 = [ 2𝐼𝑧𝑧 2𝐼𝑧𝑧 𝑚𝛿𝑞 ] , and 𝐵 = [ 0 ].
𝑄𝑆 0 0
r
𝐶𝑍𝛼 − 𝐶𝑍𝛼 0 −1
𝑚 𝑚𝑉 0 0
[ 0 1 0]
The control law can then be designed as
𝑢(𝑡) = −𝐾𝑥(𝑡), (18)
where linear quadratic regulator (LQR) theory can be used to design an appropriate gain 𝐾. The closed-loop controlled
projectile model can now be written using (18) in (17) as
𝑥̇ (𝑡) = (𝐴 − 𝐵𝐾)𝑥(𝑡) + 𝐵r 𝑐(𝑡). (19)
The use of an LQR theory to design the feedback gain ensures all the states in the state vector 𝑥(𝑡) are regulated to
zero. Since this includes the integral error state 𝑧(𝑡) = ∫ (𝑤̇ (𝑡) − 𝑤̇ d (𝑡))d𝑡, it implies the desired commands are
tracked as well.
For the designed control architecture, the method of obtaining an appropriate gain matrix 𝐾 and then analyzing
the requirements for the actuator bandwidth can now be discussed. First, an operating point is chosen, where the model
will be linearized. We illustrate this approach in this paper for both a subsonic and supersonic case at Mach 0.8 and
2.5 respectively. For each flight speed we then consider three altitudes at sea level, 5km, 10 km, and calculate a linear
model for each flight speed/altitude pair using (15). The feedback gain 𝐾 is then tuned using the linearized models for
each Mach and altitude pairing for both the 8cm and 12cm control surfaces (i.e., 12 different cases). The tuning
consists of placing weighting terms on the states and control to obtain the desired level of command following, while
regulating the remaining states to zero, and minimizing the control effort.
The desired performance can be captured by analyzing the step response for the different cases. Figure 12 shows
the achievable performance for the tuned controllers. It is evident the controller is tuned to achieve similar performance
for the different Mach, altitude, and control surface length combinations. The rise times for each was desired to be
approximately 0.8 seconds, which can be seen from the time necessary to reach 0.9 in the amplitude. This dictates
how quickly the flight controller causes the projectile is to track a vertical acceleration command. By tuning the
controllers to perform consistently between the different cases, we can then investigate how the demands on the
actuator for proper application of the flight control law will change across the flight envelope and for different control
surface lengths.

11
American Institute of Aeronautics and Astronautics
Downloaded by UNIVERSITY OF NEW SOUTH WALES on May 11, 2020 | http://arc.aiaa.org | DOI: 10.2514/6.2020-0020

Figure 12. Example step response for flight controllers after tuning across Mach, altitude, and control surface
size demonstrating desired 0.8s rise time.

To determine what is necessary from the actuator to achieve the desired performance, we use a frequency response
analysis of the different cases in the form of Bode diagrams as shown in Figure 13. These are obtained by breaking
the closed loop at the input and then looking separately at both fin control surfaces (i.e., 𝛿1 and 𝛿3 ). Since there is
symmetry between the two fins in controlling the pitch dynamics, the frequency responses are identical for each
control channel, and hence, we only consider one control channel. The frequency responses include the stability
margins which are a measure of how robust the flight controllers are to uncertainties and unmodeled dynamics such
as the actuator dynamics being considered. In particular, the phase margin (labeled Pm in the plots) is a measure of
how much additional phase lag can be introduced into the system before instability occurs. In addition, the time delay
margin can be computed from the phase margin and crossover frequency. It is a fairly common rule of thumb that the
actuator dynamics should have a natural frequency five times the crossover frequency. This ensures the actuator
dynamics are fast enough to allow for correct application of the flight control law through the actual physical control
surfaces. It should also be noted from Figure 13 how the stability margins change between different Mach and altitude
conditions. Specifically, in order to obtain the same performance shown in Figure 12, the crossover frequency
increases as the altitude decreases from 10km to sea level. This implies the actuator will need to be faster at lower
altitudes as the higher air densities at lower altitudes increase the dynamic pressure, requiring faster application of the
flight control law. The speed of the projectile also plays a role in the actuator bandwidth requirement. The crossover
frequencies are smaller for Mach 2.5 cases compared to the Mach 0.8 cases. Lastly, the size of the control surface also
plays a role in the actuator bandwidth requirement as seen by the increase in crossover frequencies for increased
control surface size from 8cm to 12cm.

12
American Institute of Aeronautics and Astronautics
Downloaded by UNIVERSITY OF NEW SOUTH WALES on May 11, 2020 | http://arc.aiaa.org | DOI: 10.2514/6.2020-0020

Figure 133. Frequency response for tuned flight controllers. The gain and phase margin, as well as the
crossover frequencies are included for each condition.

For the cases considered, the largest crossover frequency occurs at 57.8 rad/s for the Mach 0.8, sea level, and 12cm
flap flight condition. This results in the required actuator bandwidth being approximately 300 rad/s or about 48 Hz.
Figure 14 provides the frequency response for the Mach 0.8, sea level, and 12cm flap case with the actuator dynamics
included. We consider second order actuator models with a damping ratio 𝜁 = 1 and decreasing natural frequencies,
𝜔𝑛 . It can be seen from the figure that at 𝜔𝑛 = 300 rad/s there is still sufficient phase margin to ensure robustness to
other unmodeled dynamics and uncertainties not considered. In addition, as the actuator dynamics become slower the
phase margin becomes negative implying the system is unstable, as seen for the 𝜔𝑛 = 100 rad/s case. As a final
remark, it should be noted that the reverse of this process can also be considered. One can specify an actuator
bandwidth and then design the flight controller to work with that specification. In this case, the flight controller is
tuned such that the crossover frequency is a fifth of the natural frequency of the specified actuator dynamics. The
tradeoff is that the achievable performance of the flight controller could then deteriorate.

13
American Institute of Aeronautics and Astronautics
Downloaded by UNIVERSITY OF NEW SOUTH WALES on May 11, 2020 | http://arc.aiaa.org | DOI: 10.2514/6.2020-0020

Figure 144. Frequency response for flight controller tuned to Mach 0.8 with the altitude at sea level and a 12cm
control surface. The actuator dynamics included with three decreasing natural frequency values.

VII. Summary
A methodology was presented to analyze an aerodynamic control surface design and evaluate the effect on a
projectile, quantifying the ability to generate lateral accelerations to improve maneuverability, and enabling range
extension through glide and terminal maneuvers at both supersonic and subsonic speeds. This control surface
analysis methodology facilitates the evaluation of the surface size and effectiveness at achieving a desired angle of
attack to generate lift early in the design process, and enables design iterations adjusting aerodynamic stability, static
margin, and control surface sizing. A methodology was also presented to estimate actuator requirements to ensure
the control actuation system has sufficient torque to overcome in-flight aerodynamic loading and sufficient
bandwidth to stabilize the projectile dynamics. These methodologies were demonstrated in this paper on a gun-
launched, aerodynamically stabilized, fin-controlled projectile.
Future work on this topic will include investigation of flight control performance with consideration for as-built
actuator characteristics, along with expansion of the linear model to include sensor dynamics for state estimation.
Additional aerodynamic studies are also planned, with Navier-Stokes CFD simulations and wind tunnel testing to
augment current estimates before beginning flight test evaluations.

VIII. References
[1] Minnicino M, Sorensen B, Hepner DJ, Horst AW, Schmidt EM, Brandon FJ, “Artillery Range Extension”, ARL-TR-5221,
Aberdeen Proving Ground, MD, 2010.
[2] Costello M, “Extended Range of a Gun Launched Smart Projectile Using Controllable Canards”, Shock and Vibration, 8, pp.
203-213, 2001.
[3] Fresconi FE, “Range Extension of Gun-Launched Smart Munitions”, International Ballistics Symposium, New Orleans, LA,
2008.
[4] Bryson, Joshua T., Vasile, Joseph D., Celmins, Ilmars, Fresconi, Frank E., “Approach for Understanding Range Extension of
Gliding Indirect Fire Munitions”, AIAA Scitech 2018-3158.
[5] Vasile, Joseph D., Bryson, Joshua T., Fresconi, Frank E., “Aerodynamic Design Optimization of Long Range Projectile using
Missile DATCOM”, submitted for publication in AIAA Scitech 2020.
[6] Rosema, C., Doyle, J., and Blake, W., “Missile Datcom User’s Manual – 2014 Revision”. Flight Dynamics Directorate, Wright
Patterson Air Force Base (OH), Air Force Research Laboratory (US), Technical Report AFRL-RQ-WP-TR-2014-0281,
December 2014.
[7] Aftosmis, M. J., Berger, M. J., and Adomavicius, G., “A Parallel Multilevel Method for Adaptively Refined Cartesian Grids
with Embedded Boundaries,” 38th Aerospace Sciences Meeting and Exhibit, AIAA Paper 2000-0808, 2000.
[8] McCoy RL, Modern Exterior Ballistics, Schiffer, Atglen, PA, 2012.

14
American Institute of Aeronautics and Astronautics
 [9] Fresconi FE, Celmins I, Silton S, Costello M, “High Maneuverability Projectile Flight using Low Cost Components”, Aerospace
Science and Technology, 41, pp. 175-188, 2015.
[10] Zipfel, P. H., “Modeling and Simulation of Aerospace Vehicle Dynamics”, AIAA Education Series, Reston, VA, 2014.
[11] Fresconi, FE, Celmins I, Maley J, Nelson B, “Experimental Flight Characterization of a Canard-Controlled, Subsonic Missile”,
ARL-TR-8086, Aberdeen Proving Ground, MD, 2017.
[12] Gruenwald, B. C., Bryson, J. T., “Adaptive Control for a Guided Projectile using an Expanded Reference Model”, submitted
for publication in AIAA Scitech 2020
[13] Vasile, J. D., Bryson, J. T., Gruenwald, B. C., Fairfax, L., Strohm, L., and Fresconi, F. E., “A Multi-Disciplinary Approach to
Evaluate Long Range Guided Projectiles,” submitted for publication in AIAA SciTech Forum and Exposition 2020.
Downloaded by UNIVERSITY OF NEW SOUTH WALES on May 11, 2020 | http://arc.aiaa.org | DOI: 10.2514/6.2020-0020

15
American Institute of Aeronautics and Astronautics