WT2: the Wind Turbine in a Wind Tunnel Project C. L. Bottasso, F WT2: the Wind Turbine in a Wind Tunnel Project C.L. Bottasso, F. Campagnolo Politecnico di Milano, Italy Spring 2010

Outline Project goals The wind tunnel at the Politecnico di Milano Wind turbine model scaling and configuration Aerodynamics Blade manufacturing Simulation environment Data acquisition, control and model management system Conclusions and outlook

Project Goals Goals: design, manufacture and test an aeroelastically-scaled model of the Vestas V90 wind turbine Applications: Testing and comparison of advanced control laws and supporting technologies (e.g. wind and state observers) Testing of extreme operating conditions (e.g. high speed high yawed flow, shut-down in high winds, etc.) Tuning of mathematical models Testing of system identification techniques Aeroelasticity of wind turbines … Possible extensions: Multiple wind turbine interactions Aeroelasticity of off-shore wind turbines (with prescribed motion of wind turbine base) Effects of terrain orography on wind turbines

The Politecnico di Milano Wind Tunnel 1.4MW Civil-Aeronautical Wind Tunnel (CAWT): 13.8x3.8m, 14m/s, civil section: turbulence < 2% with turbulence generators = 25% - 13m turntable 4x3.8m, 55m/s, aeronautical section: turbulence <0.1% open-closed test section

The Politecnico di Milano Wind Tunnel Turn-table 13 m Turbulence (boundary layer) generators Low speed testing in the presence of vertical wind profile Multiple wind turbine testing (wake-machine interaction) High speed testing Aerodynamic characterization (Cp-TSR-β & CF-TSR-β curves)

Outline Project goals The wind tunnel at the Politecnico di Milano Wind turbine model scaling and configuration Aerodynamics Blade manufacturing Simulation environment Data acquisition, control and model management system Conclusions and outlook

Model Scaling Criteria for definition of scaling (using Buckingham Π Theorem): Best compromise between: Reynolds mismatch (quality of aerodynamics) Speed-up of scaled time (avoid excessive increase of control bandwith) Aeroelastic effects: correct relative placement of frequencies wrt rev harmonics, correct Lock number Quantity Scaling factor Length Ratio 1/45 Time Ratio 1/22.84 Velocity Ratio 1/1.97 Power Ratio 1/15477 Rotor Speed Ratio 22.84 Torque Ratio 1/353574 Reynolds Ratio 1/88.64 Froude Ratio 11.6 Mach Ratio V2 V90 Rotor Diameter 2 [m] 90 [m] Blade Length 977.8 [mm] 44 [m] Rotor Overhang 75.1 [mm] 3.38 [m] Hub Height 1.78 [m] 79.94 [m] Rotor Speed 367 [rpm] 16 [rpm] Nominal Power 193.8 [W] 3 [MW] Nominal Torque 5.06 [Nm] 1790 [KNm] Average Reynolds  5÷6 e4  4÷5 e6 Reynolds mismatch: Use low-Re airfoils (AH79 & WM006) to minimize aerodynamic differences Keep same chord distribution as original V90 blade, but Adjust blade twist to optimize axial induction factor

Electronic board for blade strain gages (6 force/moment components) V2 Model Configuration Electronic board for blade strain gages CONICAL SPIRAL TOOTHED GEARS Rotor radius = 1m Height = 2.8 m Up-tilt = 6 deg Balance (6 force/moment components)

Main shaft with torque meter V2 Model Configuration Pitch actuator control units: Faulhaber MCDC-3003 C 30 V – 10 A Max Position and speed Cone = 4 deg Main shaft with torque meter Slip ring Moog AC6355: 36 Channels 250 V – 2 A Max Conical spiral gears Torque actuator: Portescap Brushless B1515-150 Pn = 340 W Planetary gearhead Torque and speed control Pitch actuator: Faulhaber 1524 Zero backlash gearhead Built-in encoder IE 512

V2 Model Configuration

V2 Model Configuration Wind turbine model shown without nacelle and tower covers, for clarity

Outline Project goals The wind tunnel at the Politecnico di Milano Wind turbine model scaling and configuration Aerodynamics Blade manufacturing Simulation environment Data acquisition, control and model management system Conclusions and outlook

BEM Predicted Aerodynamic Performance Good agreement in full load region III Poorer agreement in partial load regions II and II1/2, due to higher drag of V2 airfoils Region II1/2 P<Pr Ω=Ωr Region II CPopt λopt βopt CP Region III P=Pr Ω=Ωr TSR

BEM Predicted Aerodynamic Performance Good agreement between thrust coefficients in the entire working region, due to good lift characteristics of V2 airfoils CF Region II CPopt λopt βopt Region II1/2 P<Pr Ω=Ωr Region III P=Pr Ω=Ωr TSR Filippo Campagnolo

Aerodynamic Identification Goal: identification of airfoil aerodynamic characteristics Application: blade redesign, choice of airfoils, understanding of rotor aerodynamics Approach: use wind tunnel measurements of the wind turbine response Pros: Avoid testing of individual airfoils Include 3D and rotational effects Procedure: Measure power and thrust coefficients Parameterize airfoil lift and drag coefficients Identify airfoil aerodynamic parameters that best match wind turbine performance, using a BEM model of the rotor (Work in progress, results expected summer 2010)

Aerodynamic identification Constrained optimization: Goal: match CP & CF at tested TSR & β Unknowns: parameters describing airfoil CL & CD characteristics Rotor model: BEM Aerodynamic identification Experimental CP & CF coefficients 2.Experimental power CP and thrust CF coefficients 3. Maximum Likelihood identification 4. Identified aerodynamic coefficients of airfoils 5. Redesign blade to improve matching wrt V90 1. Wind tunnel testing Trim at varying pitch β and TSR Measure power CP and thrust CF CL Design data Identified data a CD a

Outline Project goals The wind tunnel at the Politecnico di Milano Wind turbine model scaling and configuration Aerodynamics Blade manufacturing Simulation environment Data acquisition, control and model management system Conclusions and outlook

Blade Manufacturing Rigid blades: Easier and faster to manufacture than aero-elastically scaled blades Used for initial testing and verification of suitable aerodynamic performance Implemented two manufacturing solutions: 1. CNC machining of light aluminum alloy 2. UD carbon fiber Carbon blades (will include blade-root strain gage in 2nd blade set – May 2010) FEM verification of strain gage sensitivity CAD model for CNC machining, with support tabs (+resin support)

Blade Manufacturing Aero-elastically scaled blades: Solution: Need accurate aerodynamic shape: classical segmented solution is unsuitable Structural requirements: match at least lower three modes Very challenging problem: only 70g of weight for 1m of span! Solution: Rohacell core with carbon fiber spars and film coating Sizing using constrained optimization (Work in progress, expected completion of blade set by end of 2010)

Design of the V2 Aero-elastically Scaled Composite Blade Objective: size spars (width, chordwise position & thickness) for desired sectional stiffness within mass budget Cost function: sectional stiffness error wrt target (scaled stiffness) Constraints: lowest 3 frequencies Carbon fiber spars for desired stiffness Structural optimization Optimization Cross sectional analysis Equivalent beam model Sectional optimization variables (position, width, thickness) Span-wise shape function interpolation Rohacell core with grooves for the housing of carbon fiber spars Width Chordwise Position ANBA (ANisotropic Beam Analysis) FEM cross sectional model: Evaluation of cross sectional stiffness (6 by 6 fully populated matrix) Thickness Thermo-retractable film

Design of the V2 Aero-elastically Scaled Composite Blade Solid line: scaled reference values Dash-dotted line: optimal sizing Mass gap can be corrected with weights Modes Reference [Hz] Optimization procedure [Hz] 1st Flap-wise 23.2 23.1 2nd Flap-wise 59.4 59.1 1st Edge-wise 33.1 Filippo Campagnolo

Design of the V2 Aero-elastically Scaled Composite Blade Approach: Demonstration of technology on simple specimen: Design specimen (uniform cross section, untwisted) of typical properties (mass, stiffness) Characterize material properties Manufacture specimen Characterize specimen (mass, stiffness, frequencies, shape) Verify accuracy wrt design Status: completed Demonstration of technology on blade-like specimen (twist, variable chord) Status: in progress Manufacture wind turbine model blade Status: to be done (expected end 2010) Filippo Campagnolo

Demonstration of Technology on Simple Specimen Characterization of material properties: Specimen of uniform properties: Results: Good matching of lowest natural frequencies Acceptable repeatability Good shape and finishing Dynamic testing Static testing Temperature–dependent characterization Carbon fiber spars Airfoil cross section Modes (specimen A/B) Percent Error (specimen A/B) 236/246 Hz 4.5/0.3 % 329/339 Hz 3.1/6.1 % 545/570 Hz 1.9/6.3 % 604/627 Hz 5.1/1.2 %

Outline Project goals The wind tunnel at the Politecnico di Milano Wind turbine model scaling and configuration Aerodynamics Blade manufacturing Simulation environment Data acquisition, control and model management system Conclusions and outlook

Simulation Environment Comprehensive aero-elastic simulation environment: supports all phases of the wind turbine model design (loads, aero-elasticity, and control) Sensor models Virtual plant Cp-Lambda model Measurement noise Wind Supervisor Start-up, power production, normal shut-down, emergency shut-down, … Pitch-torque controller Controller

Simulation Models Cp-Lambda highlights: Geometrically exact composite-ready FEM beam models Generic topology (Cartesian coordinates+Lagrange multipliers) Dynamic wake model (Peters-He, yawed flow conditions) Efficient large-scale DAE solver Non-linearly stable time integrator Fully IEC 61400 compliant (DLCs, wind models) Cp-Lambda (Code for Performance, Loads, Aero-elasticity by Multi-Body Dynamic Analysis): Global aero-servo-elastic FEM model Compute sectional stiffness ANBA (ANisotropic Beam Analysis) cross sectional model Rigid body Geometrically exact beam Revolute joint Flexible joint Actuator Recover cross sectional stresses/strains

Simulation Environment Example: verify adequacy of model for the testing of control laws Question: does testing of control laws on V2 lead to similar conclusions than V90 testing, notwithstanding differences in aerodynamics (Reynolds)? Approach: Choose comparison metrics Simulate response of scaled and full-scale models Compare responses upon back-scaling Draw conclusions Example: LQR controller outperforms PID by similar amount on V2 and V90 Aeroelastic Simulation Performance Model Parameters Scaling Laws Aeroelastic Simulation Inverse Scaling Laws

Outline Project goals The wind tunnel at the Politecnico di Milano Wind turbine model scaling and configuration Aerodynamics Blade manufacturing Simulation environment Data acquisition, control and model management system Conclusions and outlook

Data Acquisition, Control and Model Management System Wind tunnel control panel Pitch demand Torque demand Remote Control Unit: Management of experiment (choice of control logic, choice of trim points, etc.) Data logging, post-processing and visualization Emergency shut-down Wind turbine sensor readings: Shaft torque-meter Balance strain gages Blade strain gages (May 2010) Rotor RPM and azimuth Blade pitch Nacelle accelerometer Wind tunnel sensor readings: Wind speed Temperature, humidity Ethernet Control PC: Real time Linux OS (RTAI) Supervisory control Control logic: - Normal mode: pitch-torque control law - Trimming mode: RPM regulation and pitch setting

Outline Project goals The wind tunnel at the Politecnico di Milano Wind turbine model scaling and configuration Aerodynamics Blade manufacturing Simulation environment Data acquisition, control and model management system Conclusions and outlook

Conclusions and Outlook Work is in progress on many fronts, no meaningful conclusions can be drawn at the moment Work plan: Initial entry in the wind tunnel by April 2010 (rigid blades, trimming control mode) - Verification of functionality of all systems, troubleshooting, software debugging - Verification of aerodynamic performance (measurement of CP-TSR-β & CF-TSR-β curves) Second entry in May 2010 after fixes/improvements (rigid blades with root strain gages, trimming and normal control modes) Aerodynamic identification: possible redesign of rotor blades to improve aerodynamic model fidelity (airfoils, transition strips, flaps, etc.) Blade design and manufacturing: Implement strain gages in composite rigid blades Continue development of flexible composite blades Add strain gages and/or fiber optics to flexible composite blades Control and management system: complete and improve GUI and functionalities Full model capabilities: expected end 2010

Acknowledgements Research funded by Vestas Wind Systems A/S The authors gratefully acknowledge the contribution of S. Calovi and S. Cacciola, G. Galetto, L. Maffenini, P. Marrone, M. Mauri, M. Monguzzi, D. Rocchi, S. Rota, G. Sala of the Politecnico di Milano