Tec

hn

isc

he

Un

ive

rsit

ät

Mü

nc

he

nW

ind

En

erg

yIn

sti

tute

Integrated Design Optimization of Wind Turbines:

Challenges, Methods, Applications

Carlo L. BottassoTechnische Universität München

Funktionsleichtbau für die Windkenergie – Anforderungen, Möglichkeiten, Nutzen DLR, Braunschweig, 20 September 2016

Desig

n O

pti

miz

ati

on o

f W

ind T

urb

ines

Motivation: the Need for Design Tools

Size

Weig

ht (C

ost)

Cubic law of growth

Technological innovation

Desig

n O

pti

miz

ati

on o

f W

ind T

urb

ines

Motivation: the Need for Design Tools

Active and passive load

reduction techniques

(source: Timber Tower)

(source: Alstom)

Electromechanical conversionRotor aerodynamics

Materials and structural design

Sensing &

advanced controls

Construction technology

(source: Risø-DTU)

Each innovation will come with pros and cons

Crucial role of design -the final judge-: “Nice idea, but does it reduce the CoE?”

Desig

n O

pti

miz

ati

on o

f W

ind T

urb

ines

Pitch-torque control laws:- Regulating the machine at different set points depending on wind conditions- Reacting to gusts- Reacting to wind turbulence- Keeping actuator duty-cycleswithin admissible limits- Handling transients: run-up, normal and emergency shut-down procedures- …

- Annual Energy Production (AEP)- Noise- Transportability-…

- Loads: envelopecomputed from large number of Design Load Cases (DLCs, IEC-61400) - Fatigue (25 year life), Damage Equivalent Loads (DELs)- Maximum blade tip deflections - Placement of natural frequencies wrt rev harmonics- Stability: flutter, LCOs, low damping of certain modes, local buckling- Complex couplingsamong rotor/drive-train/tower/foundations (off-shore: hydro loads, floating & moored platforms)- Weight: massive size, composite materials (but shear quantity is an issue, fiberglass, wood, clever use of carbon fiber)- Manufacturingtechnology, constraints

- Generator (RPM, weight, torque, drive-train, …)- Pitch and yaw actuators- Brakes- …

GE wind turbine (from inhabitat.com)

Wind Turbine Design

Desig

n O

pti

miz

ati

on o

f W

ind T

urb

ines

Pitch-torque control laws:- Regulating the machine at different set points depending on wind conditions- Reacting to gusts- Reacting to wind turbulence- Keeping actuator duty-cycleswithin admissible limits- Handling transients: run-up, normal and emergency shut-down procedures- …

- Annual Energy Production (AEP)- Noise- Transportability-…

- Loads: envelopecomputed from large number of Design Load Cases (DLCs, IEC-61400) - Fatigue (25 year life), Damage Equivalent Loads (DELs)- Maximum blade tip deflections - Placement of natural frequencies wrt rev harmonics- Stability: flutter, LCOs, low damping of certain modes, local buckling- Complex couplingsamong rotor/drive-train/tower/foundations (off-shore: hydro loads, floating & moored platforms)- Weight: massive size, composite materials (but shear quantity is an issue, fiberglass, wood, clever use of carbon fiber)- Manufacturingtechnology, constraints

- Generator (RPM, weight, torque, drive-train, …)- Pitch and yaw actuators- Brakes- …

GE wind turbine (from inhabitat.com)

Wind Turbine Design

One load assessment:107÷108 time steps

Desig

n O

pti

miz

ati

on o

f W

ind T

urb

ines

Holistic Design of Wind Turbines

There is a need for multi-disciplinary optimization tools, which must:• Be fast (hours/days) (on standard hardware!)

• Provide workable solutions in all areas (aerodynamics, structures, controls) for specialists to refine/verify

• Account ab-initio for all complex couplings (no fixes a posteriori)

• Use fully-integrated tools (no manual intervention)

They will never replace the experienced designer! … but would greatly speed-up design, improve exploration/knowledge of design space

Current approach to design: discipline-oriented specialist groups

Different simulation modelsLengthy loops to satisfy all requirements/constraints

(months)

Data transfer/compatibility among groups

Desig

n O

pti

miz

ati

on o

f W

ind T

urb

ines

Motivation: the Need for Combined Aero-Structural Design

Cost model (Fingersh at al., 2006):

𝑪𝒐𝑬 =𝑭𝒊𝒙𝒆𝒅𝑪𝒉𝒂𝒏𝒈𝒆𝑹𝒂𝒕𝒆 ∗ 𝑰𝒏𝒊𝒕𝒊𝒂𝒍𝑪𝒂𝒑𝒊𝒕𝒂𝒍𝑪𝒐𝒔𝒕 𝒑

𝑨𝑬𝑷 𝒑+ 𝑨𝒏𝒏𝒖𝒂𝒍𝑶𝒑𝒆𝒓𝒂𝒕𝒊𝒏𝒈𝑬𝒙𝒑𝒆𝒏𝒔𝒆𝒔 𝒑

where 𝒑 = design parameters

Strong couplings between aerodynamic shape (AEP) and structural sizing (blade cost)

Example:

Spar cap thickness ⇧

Blade thickness ⇩

Reduce solidity to increase efficiency (AEP ⇧)

Consequence: effect on weight ⇧

Desig

n O

pti

miz

ati

on o

f W

ind T

urb

ines

Increase solidity

Consequence: effect on weight ⇧

Motivation: the Need for Combined Aero-Structural Design

Cost model (Fingersh at al., 2006):

𝑪𝒐𝑬 =𝑭𝒊𝒙𝒆𝒅𝑪𝒉𝒂𝒏𝒈𝒆𝑹𝒂𝒕𝒆 ∗ 𝑰𝒏𝒊𝒕𝒊𝒂𝒍𝑪𝒂𝒑𝒊𝒕𝒂𝒍𝑪𝒐𝒔𝒕 𝒑

𝑨𝑬𝑷 𝒑+ 𝑨𝒏𝒏𝒖𝒂𝒍𝑶𝒑𝒆𝒓𝒂𝒕𝒊𝒏𝒈𝑬𝒙𝒑𝒆𝒏𝒔𝒆𝒔 𝒑

where 𝒑 = design parameters

Strong couplings between aerodynamic shape (AEP) and structural sizing (blade cost)

Example:

Skin buckling critical load ⇩

Skin core thickness ⇧

Desig

n O

pti

miz

ati

on o

f W

ind T

urb

ines

Motivation: the Need for Combined Aero-Structural Design

Example: INNWIND 10 MW HAWT (class 1A, D=178.3, H=119m)

Baseline design by INNWIND consortium

1. Perform purely aerodynamic optimization for max(AEP)

2. Follow with structural optimization for minimum weight

Dramatic reduction in solidity to improve AEP leads to large increase in weight

⇨ CoE increases (+2.6%)

Spar cap ▼Chord ▼

Desig

n O

pti

miz

ati

on o

f W

ind T

urb

ines

Cp-Max Rotor Design Environment

First release: 2007, improved and expanded since then

Applications: academic research and industrial blade design

Cost:AEPAerodynamic parameters:chord, twist

Cost:Blade weight (or cost model if available)Structural parameters:thickness of shell and spar caps, width and location of shear webs

Cost:Physics-based CoEParameters:Aerodynamic and structural

Controls:model-based (self-adjusting to changing design)

Desig

n O

pti

miz

ati

on o

f W

ind T

urb

inesCp-Lambda highlights:

• IEC 61400 compliant (DLCs, wind models)

• Geometrically exact composite-ready beam models

• Generic topology (Cartesian coordinates+Lagrange multipliers)

• Dynamic wake model (Peters-He, yawed flow conditions)

• Efficient large-scale DAE solver with index 3 pre-conditioning

• Non-linearly stable time integrator₋ Energy decaying/preserving scheme₋ Zero-work constraints

• Rigid body

• Geometrically exact beam

• Revolute joint

• Flexible joint

• Actuator

Multibody Dynamics Technology

Desig

n O

pti

miz

ati

on o

f W

ind T

urb

ines

• Rigid body

• Geometrically exact beam

• Revolute joint

• Flexible joint

• Actuator

ANBA (Anisotropic Beam Analysis) cross sectional model (Giavotto et al., 1983):• Evaluation of cross sectional stiffness (6 by 6 fully populated) • Recovery of sectional stresses and strains

Compute cross sectional stresses and strains

Compute sectional stiffness of equivalent

beam model

Sectional Analysis

Desig

n O

pti

miz

ati

on o

f W

ind T

urb

ines

Some Blades Designed with Cp-Max

16 m ▶

45 m ▶

◀ 24 m

Desig

n O

pti

miz

ati

on o

f W

ind T

urb

ines

Design Environment

Aero-Structural

Optimization

Structural

Optimization +

Controls

Aerodynamic

Optimization

• Variables: chord, twist and airfoil positioning

• Merit figure: max AEP

• Constraints on max chord, tip speed, geometry

• Variables: blade components thickness, tower structure

• Merit figure: min ICC

• Load freezing for reduced computational time and handling of solution space roughness

• Multi-level coarse-fine iterations

• Variables: rotor diameter, hub height, nacelle uptilt, rotor cone, blade aero-structure

• Merit figure: min CoE

• Complexity/cost trade-offs

Desig

n O

pti

miz

ati

on o

f W

ind T

urb

ines

Cp-Max

Aero-Structural

Optimization

Structural

Optimization +

Controls

Aerodynamic

Optimization

• Variables: chord, twist and airfoil positioning

• Merit figure: max AEP

• Constraints on max chord, tip speed, geometry

• Variables: blade components thickness, tower structure

• Merit figure: min ICC

• Load freezing for reduced computational time and handling of solution space roughness

• Multi-level coarse-fine iterations

• Variables: rotor diameter, hub height, nacelle uptilt, rotor cone, blade aero-structure

• Merit figure: min CoE

• Complexity/cost trade-offs

Desig

n O

pti

miz

ati

on o

f W

ind T

urb

ines

“Fin

e”

level: 3

D F

EM

“Coars

e”

level: 2

D F

EM

secti

on &

beam

mod

els

Update complete HAWT Cp-Lambda multibody model- DLCs simulation- Campbell diagram- AEP

DLC post-processing: load envelope, DELs, Markov, max tip deflection

Blade: - Geometrically exact beam model- Span-wise interpolation

Blade: - ANBA 2D FEM sectional analysis- Computation of 6x6 stiffness matrices

Blade: definition of aerodynamic & structural design parameters

Cost function:- Min initial capital cost ICC

Automatic 3D CAD model generation

Automatic 3D FEM meshing

Update of blade mass (cost)

Analyses:- Max tip deflection- Max stress/strain- Fatigue- Buckling

Verification of design constraints

Constr

ain

t/m

odel up

date

heuri

sti

c (to

rep

air

constr

ain

t vio

lati

ons)

When SQP converged

Tower: definition of structural design parameters

Tower: - Geom. exact beam model- Height-wise interpolation

Tower:- Computation of stiffness matrix

Constraints:- Max tip deflection- Natural frequencies- Max stresses and strains- Fatigue (ANBA)- Tower buckling- Manufacturing constraints

SQP optimizer

min ICC subject to constraints

Automatic 3D FEM meshing Root detailed analysis:Geometry parameterization

Sizing of bolted joint and blade root laminate

- Bolt preload calculation- Max stress/strain- Fatigue

Root analysis

Desig

n O

pti

miz

ati

on o

f W

ind T

urb

ines

The Importance of Multi-Level Blade Design

Stress/strain/fatigue/frequency/max TD:- Constraints not satisfied at first

iteration on 3D FEM model- Modify constraints based on 3D

FEM analysis- Converged at 2nd iteration

Fatigue damage constraint satisfied

Buckling:- Buckling constraint not satisfied at first iteration- Update skin core thickness- Update trailing edge reinforcement strip - Converged at 2nd iteration

Peak stress on initial model

Increased trailing edge strip

ITERATION 1

ITERATION 0

ITERATION 1

ITERATION 0

ITERATION 1

ITERATION 0

ITERATION 1

ITERATION 0

No

rmalized s

tress

Fati

gu

e d

am

ag

e in

dex

Th

ickness

Tra

ilin

g e

dg

e s

trip

th

ickness

Increased skin core thickness

Desig

n O

pti

miz

ati

on o

f W

ind T

urb

ines

Cp-Max

Aero-Structural

Optimization

Structural

Optimization +

Controls

Aerodynamic

Optimization

• Variables: chord, twist and airfoil positioning

• Merit figure: max AEP

• Constraints on max chord, tip speed, geometry

• Variables: blade components thickness, tower structure

• Merit figure: min ICC

• Load freezing for reduced computational time and handling of solution space roughness

• Multi-level coarse-fine iterations

• Variables: rotor diameter, hub height, nacelle uptilt, rotor cone, blade aero-structure

• Merit figure: min CoE

• Complexity/cost trade-offs

Desig

n O

pti

miz

ati

on o

f W

ind T

urb

ines

Integrated Preliminary-Detailed Design

Preliminary design:

- Overall size (rotor radius, hub height)

- Other macro parameters (uptilt, cone-prebend)

Detailed design:

- Aerodynamic shape

- Structural sizing

- Tower (diameters, thicknesses)

- Systems and components

Strong couplings:

Combined preliminary-detailed design (Wind Energy Science, Bortolotti et al. 2016)

𝐻

𝑅

𝜑

𝛾

Desig

n O

pti

miz

ati

on o

f W

ind T

urb

ines

Definition of global parameters:

- Rotor radius R- Hub height H- Cone angle ϒ- Uptilt angle φ- Solidity σc,t

- Tapering τc,t

- Pitch offset δ

Static and turbulent AEPwithaeroelasticeffects

Cost function:Min cost of energy CoE

Constraints:- Ultimate & fatigue loads- Manufacturing constraints

SQP optimizer

min CoE subject to constraints

CoE models

NREL

INNWIND(offshore)

SANDIA(blade cost)

Blade: - Geom. exact beam model- Span-wise interpolation

Blade:- ANBA 2D FEM sectional analysis- Compute 6x6 stiffness matrices

Blade: definition of structural design parameters

Tower: - Geom. exact beam model- Height-wise interpolation

Tower:Compute stiffness matrices

Tower: definition of structural design parameters

CompleteCp-Lambda multibody model

Blade: definition of aero designparameters Cost function:

- Max AEP

Constraints:- Max chord- Max tip speed- σ / τ - Blade geom.- ...

SQP optimizer

max AEP subject to constraints

Parametrization:- Chord- Twist- Thickness

Aerodynamic optimization

“Fin

e”

level: 3

D F

EM

“Coars

e”

level: 2

D F

EM

secti

on &

beam

mod

elsStructural

optimizationBlade: - ANBA 2D FEM sectional analysis- 6x6 stiffness matrices

Blade: definition of aero & structural design parameters

Tower: definition of structural design parameters

Tower:- Computation of stiffness matrix

Update complete HAWTCp-Lambda multibody model- DLCs simulation- Campbell diagram- AEP

DLC post-processing: load envelope, DELs, Markov, max tip deflection

Blade: - Geom. exact beam model- Span-wise interpolation

Tower: - Geom. exact beam model- Height-wise interp.

Automatic 3D FEM shell element meshing

Load application and FE solver

Post processing:- Max tip deflection- Max stress/strain- Fatigue- BucklingVerification of design constraints

Automatic 3D FEM meshing

Root detailed analysis:geometry parameterization

Sizing of bolted joint and blade root laminate- Bolt preload calculation- Max stress/strain- Fatigue

Cost function:- Min ICC

Constraints:- Max tip deflection- Natural frequencies- Max stress and strain- Fatigue- Tower buckling- Manufacturing constr.

SQP optimizer

min ICC subject to constraints

Constr

ain

t/m

od

el

up

date

heuri

sti

c (

to r

ep

air

constr

ain

t vio

lati

ons)

R

Desig

n O

pti

miz

ati

on o

f W

ind T

urb

ines

Applications

10 MW – 1A INNWIND Cp-Max Opt Difference

P rated 10 MW 10 MW --

Rotor Diameter 178.3 m 223.2 m + 25.2 %

Hub Height 119.0 m 138.3 m + 16.2 %

Rotor Cone 4.65 deg 5.51 deg + 18.5 %

Nacelle Uptilt 5.00 deg 5.25 deg + 5.0 %

Rotor Solidity 4.66 % 4.08 % - 12.3 %

Blade Tapering 0.43 0.40 - 7.0 %

Blade Mass 42.5 ton 75.6 ton + 77.9 %

Blade Cost 316 k$ 544 k$ + 72.2 %

Tower Mass 597 ton 886 ton + 43.6 %

Tower Cost 2.0 M€ 2.94 M€ +44.1 %

AEP 48.8 GWh 57.2 GWh +17.2 %

TCC 9.1 M€ 12.2 M€ +34.2 %

ICC 30.0 M€ 34.3 M€ +14.3 %

Cost of Energy 70.73 €/MWh 65.79 €/MWh - 7.0 %

INNWIND 10 MW HAWT (class 1A, D=178.3, H=119m)

Baseline design by INNWIND consortium

Desig

n O

pti

miz

ati

on o

f W

ind T

urb

ines

IEA Task 37International Energy Agency (IAE) Wind Agreement:

Founded in 1977, it sponsors the cooperation in the research, development, and deployment of wind energy systems

Task 37:

Wind Energy Systems Engineering: Integrated research, design and development (RD&D)

Work Packages

• WP0: Management and Coordination

• WP1: Guidelines for a common framework for integrated RD&D at different fidelity levels (both turbines and plants)

• WP2: Reference wind energy systems (both turbines and plants)

- 3.4 MW onshore wind turbine

- 10 MW offshore wind turbine

• WP3: Benchmarking MDAO activities at different system levels (both turbines and plants)

Link: http://www.ieawind.org/

Contacts:

K. Dykes, NREL, USA, P. E. Rethore and F. Zahle, DTU Wind Energy, Risø, Denmark

K. Merz, SINTEF Energy Research, Norway

Desig

n O

pti

miz

ati

on o

f W

ind T

urb

ines

Applications: Passive Load AlleviationOptimal blade design with Bend-Twist Coupling (BTC) by fiber rotation

Note: all designs satisfy exactly the same requirements

Assume fiber angle as design variable in skin and spar caps

Spar-caps: steep increase

Skin: milder increase

Design trade-offs

+ Decreased load envelope and fatigue+ Decreased actuator duty cycle

+ Decreased weightBut same power

Spar-cap/skin synergy

Desig

n O

pti

miz

ati

on o

f W

ind T

urb

ines

Integrated Passive and Active Load Alleviation

Active load alleviation: Individual Blade pitch Control (IBC) Significant life-time improvement,

but large duty cycle increase

Integrated Passive-Active load alleviation: IBC+BTC

Decreased pitch activity for BTC blade

IBC+BTCLifetime ADC: +5% wrt baselineLifetime DEL: circa BTC + IPC effects

Desig

n O

pti

miz

ati

on o

f W

ind T

urb

ines

Applications: Passive Load Alleviation

Optimal blade design with Bend-Twist Coupling (BTC) by spar cap offset

Coupling coefficient 𝜶 =𝑲𝒇𝒍𝒂𝒑−𝒆𝒅𝒈𝒆𝟐

𝑲𝒇𝒍𝒂𝒑𝑲𝒆𝒅𝒈𝒆

Note: all designs satisfy exactly the same requirements

Flap up –> bend edgewise back (sweep)

Optimal offset: 20 cm

Desig

n O

pti

miz

ati

on o

f W

ind T

urb

ines

Applications: Passive Load Alleviation

Optimal combination of spar fiber rotation and offset

Note: all designs satisfy exactly the same requirements

Optimal combination: offset 20 cm + angle 5 deg

Desig

n O

pti

miz

ati

on o

f W

ind T

urb

ines

Optimal combination of spar fiber rotation and offset

Rotor resizing:

◀ Significant benefits

Applications: Passive Load Alleviation

Optimum: R+5%(similar hub loads as baseline)

Desig

n O

pti

miz

ati

on o

f W

ind T

urb

ines

Composite Optimization

Automatic selection of composite laminates along the blade for the different structural components (in collaboration with Owens Corning)

Material catalogue: Multi-parametric composite material model(e.g. fiber fraction, technological parameter, etc.)

Application

INNWIND 10 MW wind turbine, equipped with E-GFRP

Costs estimated from SANDIA model

Blade cost breakdown

Blade materials cost breakdown

Desig

n O

pti

miz

ati

on o

f W

ind T

urb

ines

Composite Optimization

Redesign of the spar caps laminate

Redesign of the shell skin laminate

The optimum laminate is between H-GFRP and CFRP

The optimum laminate is between Bx-GFRPand Tx-GFRP

Combined optimum: Blade mass -9.3%, blade cost -2.9%

Desig

n O

pti

miz

ati

on o

f W

ind T

urb

ines

In

te

g

ra

te

d

a

p

p

ro

a

c

h

fo

r a

e

ro

-stru

c

tu

ra

l o

p

tim

iza

tio

n

o

f

W

T

B

la

d

e

s

POLITECNICO di MILANO POLI-Wind Research Lab

Case Study Results: Optimal blade 3D

Three-dimensional view with detail of thick trailing edge and flatback airfoils.

Free-Form 3D Aero-Structural Optimization

Design airfoils together with blade:

• Bezier airfoil parameterization

• Airfoil aerodynamics by Xfoil + Viterna extrapolation

Additional constraints:

• CL max (margin to stall), geometry

Automatic appearance of flatback airfoil!

Desig

n O

pti

miz

ati

on o

f W

ind T

urb

ines

Some ReferencesP. Bortolotti, A. Croce, and C.L. Bottasso: Combined preliminary–detailed design of wind turbines. Wind Energ. Sci., 1, 1–18, 2016, doi:10.5194/wes-1-1-2016

C.L. Bottasso, P. Bortolotti, A. Croce, and F. Gualdoni: Integrated Aero-Structural Optimization of Wind Turbine Rotors. Multibody Syst. Dyn., doi: 10.1007/s11044-015-9488-1

C.L. Bottasso, F. Campagnolo, A. Croce, S. Dilli, F. Gualdoni, M.B. Nielsen: Structural Optimization of Wind Turbine Rotor Blades by Multi-Level Sectional/Multibody/3DFEM Analysis, Multibody System Dynamics, 32:87-116, 2014

C.L. Bottasso, F. Campagnolo, C. Tibaldi: Optimization-Based Study of Bend-Twist Coupled Rotor Blades for Passive and Integrated Passive/Active Load Alleviation, Wind Energy, 16:1149-1166, 2013

C.L. Bottasso, A. Croce, F. Campagnolo: Multi-Disciplinary Constrained Optimization of Wind Turbines, Multibody System Dynamics, 27:21-53, 2012

O.A. Bauchau, A. Epple, C.L. Bottasso: Scaling of Constraints and Augmented Lagrangian Formulations in Multibody Dynamics Simulations, ASME Journal of Computational and Nonlinear Dynamics, 4:021007, 2009

• P. Bortolotti, G. Adolphs, C.L. Bottasso: A methodology to guide the selection of composite materials in a windturbine rotor blade design process

• A. Croce, L. Sartori, M.S. Lunghini, L. Clozza, P. Bortolotti, C.L. Bottasso: Lightweight rotor design by optimalspar cap offset

• L. Sartori, P. Bortolotti, A. Croce, and C.L. Bottasso: Integration of prebend optimization in a holistic windturbine design tool

Desig

n O

pti

miz

ati

on o

f W

ind T

urb

ines

Conclusions

• Strong couplings between aero and structural design variables

• Multi-level approach to marry high fidelity and computational effort

• Integrated design optimization allows for fast exploration of design space, leading to potential significant CoE improvements

Open issues/outlook:• CoE: solutions are highly sensitive to cost model, need detailed

reliable models that truly account for all significant effects, problem partially alleviated by Pareto solutions (in progress)

• Uncertainties everywhere (aero, structure, wind, …), move away from deterministic design (but what about certification standards?), currently working on UQ and robust design P

DF

Input

PD

F

Output

Simulation

model

Desig

n O

pti

miz

ati

on o

f W

ind T

urb

ines

Acknowledgements

Work in collaboration with:

• Pietro Bortolotti (TUM)

• Alessandro Croce, Luca Sartori (Politecnico di Milano)

Desig

n O

pti

miz

ati

on o

f W

ind T

urb

ines

Email: info@torque2016.org

Web: www.torque2016.org

TORQUE 2016Munich, Germany, 5-7 October 2016

The Science of Making Torque from Wind