Load Control in Highly-Deforming
Aeroelastic Systems
Rafael Palacios
Department of Aeronautics
Seminar at Vibration UTC, MED
30 April 2014
http://www.imperial.ac.uk/aeroelastics
Load Control and Aeroelastics: Research Topics
Dynamics & control Load control in Computational
of flexible aircraft large wind turbines methods for FSI
Load Control - Overview
• Motivation: ever more efficient airframes → perpetual flight
• Approach: A framework for Simulation of High-Aspect Ratio Planes
o Aeroelasticity with large-displacements
o Coupling with flight dynamics
• Model reduction and control
• Examples:
o Dynamic stability of very flexible aircraft
o Gust alleviation strategies – wake encounter
o On-going work
Quick Quiz
Q: What do this companies have in common?
A: They all announced a solar-powered aircraft project in April 2014
The challenge of very high efficiency
Extreme efficiency: Solar-powered planes for “perpetual flight”
Solar Impulse 2
QinetiQ Zephyr 7
• Span: 22.5 m
• Weight: 53 kg
• Flight altitude: 60,000 ft
• Payload: 2.5 kg
• Power: Solar panels
Li-S batteries
see movie
Large (and flexible) wings on transport aircraft
• Who is not interested in higher efficiency?
[source:USPTO] B787 (www.aviationweek.com
Boeing SUGAR Volt
NASA X-56A
The challenge of very high efficiency
The Great Flight Diagram* Conventional aircraft
Solar-powered
Power management:
Weight
Birds
Wing loading *Noth (2006). Design of Solar Powered
Airplanes for Continuous Flight, PhD thesis, ETH
Modelling of (very) flexible aircraft dynamics
• Some major modelling challenges:
1. Design for jig shape
2. Validity of linear methods even more restricted in flight envelope (e.g., gust loads)
3. Blurred boundaries between aeroelasticity, flight dynamics, and control
• Our objective:
“Apparent stiffness” by means of dynamic load control systems
Modelling of (very) flexible aircraft dynamics
FLUID
MECHANICS
Flow Control
Aeroelasticity Flight Mechanics
CONTROL
Vibration Navigation
suppression
STRUCTURAL RIGID-BODY
MECHANICS DYNAMICS
Flexible Body Dynamics
...and this is geometrically nonlinear
Modelling of (very) flexible aircraft dynamics
von Flotow (1989): Daedalus
• Linear beams
• Unsteady thin aerofoil
• No controls
Drela (1999): ASWing
• Nonlinear beams
• Unsteady lifting line
• PID control
Patil (1999) & Cesnik (2002)
• Nonlinear composite beams
• Unsteady thin aerofoil
• LQG control
(Patil & Hodges, 2006)
Wang (2010)
• Nonlinear beams
• Unstedy Vortex Lattice
• No controls
Overview
• Motivation: ever more efficient airframes → perpetual flight
• Approach: A framework for Simulation of High-Aspect Ratio Planes
o Aeroelasticity with large-displacements
o Coupling with flight dynamics
• Model reduction and control
• Examples:
o Dynamic stability of very flexible aircraft
o Gust alleviation strategies – wake encounter
o On-going work
Flexible Aircraft Dynamics Simulation (SHARP)
SHARP: Structural Dynamics
SHARP.Cells: Homogenisation of periodic structures*
*Dizy et al. Homogenization of slender periodic composite structures, IJSS 50 (2013)
Condensation of large FE models*
Stiffness model Lumped mass model
Solve 1-D
Modal basis
• “Tree” of master • Modal velocities and nonlinear
nodes (ASET) along internal forces
wings, fuselage, etc. •Guyan reduction • In modal coordinates
• Identify coefficients in
• Lumped masses •Nodal nonlinear modal EoM • Back to 3-D through
displacements/ Guyan transformation
rotations
Modal
3-D FEM
Intrinsic
( −ω M
2
j a )
+ K a φaj = 0 q = Ωq + L(q)q + Q
*Wang et al. A Method for Normal‐Mode‐Based Model Reduction in Nonlinear Structural Dynamics, under review
SHARP.Beams: Geometrically-nonlinear composite beams
Rigid-body DoF Structural DoF
⎧ va ⎫ ⎧ Ra ⎫
β =⎨ ⎬ η =⎨ ⎬
⎩ωa ⎭ Ψ
⎩ ⎭ at FE nodes
Nonlinear equations of motion
Linearized equations
⎧η ⎫
M (η ) ⎨ ⎬ + Qgyr (η ,η , β ) + Qstif (η ) = Qext (η ,η , β , ζ )
⎩β ⎭
⎧⎪ β ⎫⎪ ⎧β ⎫ ⎧0⎫
M (η0 ) ⎨ ⎬ + C gyr (η0 , β 0 ) ⎨ ⎬ + K stiff (η0 ) ⎨ ⎬ =Qext
⎪⎩η ⎪⎭ ⎩η ⎭ ⎩η ⎭
Propagation of body-
attached FoR
SHARP.Beams: Geometrically-nonlinear composite beams
Free-flying flexible beam in vacuum (no gravity)*
(Simo et al, 1988)
*Hesse, Palacios. Consistent structural linearization in flexible‐body dynamics with large rigid‐body motion, Comp&Struct 110 (2012)
Flexible Vehicle Structural Dynamics
Adding the unsteady aerodynamics
Structure to Aerodynamics Aerodynamics to structure
Unsteady Vortex Lattice Method (UVLM)
• Vortex-ring discretization (Katz & Plotkin, 2001)
• Potential flow, thin wing → Low speed flight, attached flow
• 3-D, unsteady, free-wake, interference, large (but slow) displacements
Propagation step:
Output step*:
δ Fst = ρ∞ Γ ( U × δ l )
∂Γ k
Funst ,k = ρ ∞
∂t
(
ˆ ×δ l
c U )
see movie
*Simpson et al. Induced Drag Calculations in the Unsteady Vortex Lattice Method, AIAA Journal 61 (2013)
Monolithic Coupling (linear problem)
• Discrete-time state-space formulation*
• Linearization around given configuration (usually trim)
• Frozen geometry assumption → Prescribed wake
*Murua et al. Applications of the unsteady vortex‐lattice method in aircraft aeroelasticity and flight dynamics JPAS 55 (2012)
Flexible aircraft dynamics (linear/nonlinear)
Flexible aircraft dynamics (linear/nonlinear)
Stability of very light HALE UAV*
HALE model characteristics (Patil, 2001)
Aspect ratio 16
Elastic axis (from le) 50 %
Center of gravity (from le) 50 %
Mass per unit length 0.75 kg/m
Torsional rigidity 1σ×104 N·m2
Bending rigidity 2σ×104 N·m2
σ: stiffness parameter
• Flexible main wing
• Rigid fuselage and T-tail
• Total mass: 75.4kg
• 20km altitude
Trim at V∞=30m/s
*Hesse et al. (2014) Consistent Structural Linearization in Flexible Aircraft Dynamics with Large Rigid‐Body Motion, AIAA Journal
Stability analysis of full aircraft
TRIM AIRCRAFT DISCRETE-TIME EIGENVALUE
(Nonlinear) LINEARIZE SYSTEM MATRIX ANALYSIS
Phugoid Short period Spiral
Rigid -0.011±0.27i -4.53±1.67i -0.064
V∞=30m/s
Flexible -0.0044±0.30i -2.18±1.57i -0.088 σ =2
Dynamic stability vs. stiffness
Dynamic stability directly from physical degrees of freedom
Closing the loop
Closing the loop
Dynamic aeroelasticity of manoeuvring aircraft
uA
Linear
ΔΓ n +1 = AΔΓ n + BS ΦΔuΦn + BAu An
subsystem Δy An = ΦT C ΔΓ n
yA
uΦ AERODYNAMICS
FLEXIBLE-BODY
DYNAMICS
⎧ q ⎫ ⎧ q ⎫ ⎧q ⎫
ΦT M (η0 ) Φ ⎨ ⎬ + ΦT C (η0 , β ) Φ ⎨ ⎬ + ΦT K (η0 , β ) Φ ⎨ ⎬ = ΦT Qext ( q, q , β , ζ )
⎩β ⎭ ⎩β ⎭ ⎩0 ⎭
Dynamic aeroelasticity of manoeuvring aircraft*
• Small number of inputs and outputs (structural modes + rigid body)
• Model reduction through balanced truncation
o Balance aerodynamic states using controllability/observability Gramians
o Truncate least controllable and observable states
uA
ΔΓ n +1 = AΔΓ n + BS ΦΔuΦn + BAu An
Δy An = ΦT C ΔΓ n
uΦ yA
Γ = T ΓB
AERO-
AERODYNAMICS
DYNAMICS
ΓΓ B
ΔΓ nB+1 = T −1 AT ΔΓ nB + T −1 BS ΦΔuΦn + T −1 BAu An
Δy An = ΦT CT ΔΓ nB
*Hesse et al. Reduced‐Order Aeroelastic Models for the Dynamics of Maneuvering Flexible Aircraft, AIAA Journal (2014)
Load Alleviation for a HALE UAV
Composite Nonlinear Dynamic
beam Trim Stability
Hinf Linear Linear
Synthesis ROM Plant
Load
Alleviation
Load Loads Loads
Simulation ROM Plant
Wake Vortex Encounters (WVE)
• Generating wake modelled following Kier (2011)
• Wake vortices modelled as vortex filaments from
bG= 50 m, WG=2800 kg, VG= 8 m/s
• Biot-Savart Law and exponential decay around
viscous core radius
Kier, T, 2011. An integrated loads analysis model including unsteady aerodynamic effects for position and attitude dependent gust fields. IFASD 2011.
Are wake encounters important?
Open-Loop results*
*Hesse and Palacios. Dynamic Load Alleviation in Wake Vortex Encounters, under review
Hybrid control for load alleviation
• Linear ROM for control synthesis (50 states)
• H∞ controller
• Robustness vs. performance (actuator constraints
not in the model)
bending strains S
aileron P flap P flap S aileron S
elevator P elevator S
Control surface inputs to root bending strain
Hybrid control for load alleviation
• Load envelopes for open and close loop responses
Left wing Right wing
Effective approach for load alleviation → “less structure” → less weight
Towards Predictive Control
• “1-cos” gust on cantilever wing
• Control via trailing edge flap
• MPC vs. LQR with same weights
see movie
Root bending Root torsion
Flap deflection
time
time
*Simpson et al. (2014). Predictive Control for Alleviation of Gust Loads on Very Flexible Aircraft. AIAA SciTech, Washington, DC
Applications to wind energy*
• NREL 5-MW test case in storm (gusty wind and waves)
• Flaps actuated to reduce blade loads
see movie see movie
*Ng et al. (2014) Efficient Aeroservoelastic Modeling for the Control of Trailing‐Edge Flaps on Wind Turbines, UKACC
Conclusions
• Physics-based state-space realizations of 3-D aeroelasticity with
potential-flow unsteady aerodynamics
• Coupling of rigid-body, structural, unsteady aerodynamics, and
control system. Monolithic approach.
• integration of load alleviation strategies in Flight Control System
• Demonstration on Wake Encounter Loads
• On-going work:
o Coupling with RANS (no longer monolithic!)
o MPC implementation for simultaneous flight & load control