Objectives and Constraints

for Wind Turbine Optimization

S. Andrew Ning
National Renewable Energy Laboratory
January, 2013

NREL is a national laboratory of the U.S. Department of Energy, Office of Energy Efficiency and Renewable Energy, operated by the Alliance for Sustainable Energy, LLC.

 Overview

1. Introduction
2. Methodology (aerodynamics,
structures, cost, reference model,
optimization)
3. Maximum Annual Energy
Production (AEP)
4. Minimum m/AEP
5. Minimum Cost of Energy (COE)

6. Conclusions

NATIONAL RENEWABLE ENERGY LABORATORY 2

Introduction

 Wind Turbine Design

Aerodynamic
Performance
Capital Structural
Costs Design
Optimization
Maintenance and Uncertainty Control
Costs Quantification Strategy

Farm Site
Layout Selection

NATIONAL RENEWABLE ENERGY LABORATORY 4

 Wind Turbine Optimization

Objectives Design Vars

• Max P/min Mb • Blade shape
• Max AEP • Rotor/nacelle
• Min COE • Turbine

Fidelity Optimization
• Analytic • Gradient
• Aeroelastic • Direct search
• 3D CFD • Multi-level

NATIONAL RENEWABLE ENERGY LABORATORY 5

 Wind Turbine Optimization

Objectives Design Vars

• Max P/min Mb • Blade shape
• Max AEP • Rotor/nacelle
• Min COE • Turbine

Fidelity Optimization
• Analytic • Gradient
• Aeroelastic • Direct search
• 3D CFD • Multi-level

NATIONAL RENEWABLE ENERGY LABORATORY 6

Methodology

 Model Development

1. Capture fundamental trade-offs (physics-based)

2. Execute rapidly (simple physics)

3. Robust convergence (reliable gradients)

NATIONAL RENEWABLE ENERGY LABORATORY 8

 Aerodynamics

• Blade-element momentum theory

‣ hub and tip losses, high-induction factor

correction, inclusion of drag

‣ 2-dimensional cubic splines for lift and drag

coefficient (angle of attack, Reynolds number)
• Drivetrain losses incorporated in power curve

• Region 2.5 when max rotation speed reached

• Rayleigh distribution with 10 m/s mean wind
speed (Class I turbine)
• Availability and array loss factors

NATIONAL RENEWABLE ENERGY LABORATORY 9

 Blade Element Momentum

0
⌦r(1 + a )
plane of rotation
U1 (1 - a)
W

NATIONAL RENEWABLE ENERGY LABORATORY 10

 Blade Element Momentum

0
⌦r(11 + a )
plane of rotation
1 - a)
U1 (1
W

NATIONAL RENEWABLE ENERGY LABORATORY 11

 Blade Element Momentum

0
⌦r(1 + a )
plane of rotation
U1 (1 - a)
W

NATIONAL RENEWABLE ENERGY LABORATORY 12

 Blade Element Momentum

sin ¢ cos ¢

f (¢) = - = 0

1 - a(¢) Ar (1 + a0 (¢))
NATIONAL RENEWABLE ENERGY LABORATORY 13
 Blade Element Momentum

S.Andrew Ning, “A Simple Solution Method for the Blade Element Momentum
Equations with Guaranteed Convergence,” Wind Energy, to appear. CCBlade
NATIONAL RENEWABLE ENERGY LABORATORY 14
 Figure 1: Example of composites layup at a typical blade section. Principal material direction of laminas,

Composite Sectional Analysis

shown in brown, is skewed with respect to the blade axis. This causes bend-twist coupling as evident
at the blade tip.

Sector 2

Sector 3
Sector 1
AA
BB

Laminas schedule at AA Laminas schedule at BB

Figure 2: Example of spar-cap-type layup of composites at a blade section

PreComp

NATIONAL RENEWABLE ENERGY LABORATORY 15

 Beam Finite Element Analysis

Z 1
1 00 00
Kij = 3 [EI](⌘)fi (⌘)fj (⌘)d⌘
L 0
pBEAM
NATIONAL RENEWABLE ENERGY LABORATORY 16
 Panel Buckling

⇣ ⇡ ⌘2 Z E(⌧ )⌧ 2

Ncr = 3.6 d⌧

b b 1 ⌫(⌧ )2

NATIONAL RENEWABLE ENERGY LABORATORY 17

 Fatigue (gravity-loads)

NATIONAL RENEWABLE ENERGY LABORATORY 18

 Cost Model

• Based on NREL Cost and Scaling Model

• Replaced blade mass/cost estimate

• Replaced tower mass estimate
• New balance-of-station model

NATIONAL RENEWABLE ENERGY LABORATORY 19

 Reference Geometry

• NREL 5-MW reference design

• Sandia National Laboratories

initial layup
• Parameterized for optimization
purposes

NATIONAL RENEWABLE ENERGY LABORATORY 20

 Design Variables

Description Name # of Vars

chord distribution {c} 5
twist distribution {θ} 4
spar cap thickness distribution {t} 3
tip speed ratio in region 2 λ 1
rotor diameter D 1
machine rating rating 1

NATIONAL RENEWABLE ENERGY LABORATORY 21

 Constraints

minimize J(x)
x
subject to (rf rm ✏50 i )/✏ult < 1, i = 1, . . . , N ultimatetensile
ultimate tensilestrain
strength
(rf rm ✏50 i )/✏ult > -1, i = 1, . . . , N ultimatecompressive
ultimate compressive strength
strain
(✏50 j rf - ✏cr )/✏ult > 0, j = 1, . . . , M sparcap
spar capbuckling
buckling
6/60 < 1.1 tipdeflection
tip deflectionatatrated
rated
speed
!1 /(3⌦rated ) > 1.1 bladenatural
blade naturalfrequency
frequency
0root-gravity /Sf < 1 fatigue
fatigueatatblade
bladeroot
rootdue to gravity
(gravity loads
loads)
0root-gravity /Sf > -1 fatigue
fatigueatatblade
bladeroot
rootdue to gravity
(gravity loads
loads)
Vtip < Vtip max maximum tip speed (imposed directly in
maximum tip speed

cset (x) < 0

NATIONAL RENEWABLE ENERGY LABORATORY 22

Maximum Annual

Energy Production

 Why a single-discipline objective?

1. No structural model
2. No cost model
3. Organizational structure
4. Computational limitations

NATIONAL RENEWABLE ENERGY LABORATORY 24

 Maximize AEP at Fixed Mass

maximize AEP (x)

with respect to x = {{c}, {✓}, }
subject to cset < 0
mblade = mc

NATIONAL RENEWABLE ENERGY LABORATORY 25

 Maximize AEP at Fixed Mass

NATIONAL RENEWABLE ENERGY LABORATORY 26

 AEP First

maximize AEP (x)

with respect to x = {{c}, {✓}, A}
subject to Vtip < Vtip max

NATIONAL RENEWABLE ENERGY LABORATORY 27

 AEP First

maximize AEP (x)

with respect to x = {{c}, {✓}, A}}
subject to Vtip < Vtip max
✓ Z ◆
Mb
iaero < iaero 0 iaero ⌘
t
dr

Splan < Splan 0

(root < (root 0

NATIONAL RENEWABLE ENERGY LABORATORY 28

 AEP First

maximize AEP (x)

with respect to x = {{c}, {✓}, A}
}
subject to Vtip < Vtip max
iaero < iaero 0

0
Splan < Splan 00

Jroot < (
( Jroot 0

minimize m(x)
with respect to x = {t}
subject to cset (x) < 0

NATIONAL RENEWABLE ENERGY LABORATORY 29

 Comparison Between Methods

AEP mass
COE

0.5

0
% change

-0.5

-1

-1.5

AEP 1st
mass 1st min COE

NATIONAL RENEWABLE ENERGY LABORATORY 30

 Mass First

minimize m(x)
with respect to x = {{c}, {t}}
subject to cset (x) < 0

maximize AEP (x)

with respect to x = {{✓}, }
subject to Vtip < Vtip max

minimize m(x)
with respect to x = {{c}, {t}}
subject to cset (x) < 0

NATIONAL RENEWABLE ENERGY LABORATORY 31

 Comparison Between Methods

AEP mass
COE

0.5

0
% change

-0.5

-1

-1.5

AEP 1st mass 1st

min COE

NATIONAL RENEWABLE ENERGY LABORATORY 32

 Minimize COE

minimize COE(x)
with respect to x = {{c}, {✓}, {t}, }
subject to cset (x) < 0

NATIONAL RENEWABLE ENERGY LABORATORY 33

 Comparison Between Methods

AEP mass
COE

0.5

0
% change

-0.5

-1

-1.5

AEP first mass first min COE

NATIONAL RENEWABLE ENERGY LABORATORY 34

 Conclusions

1. Similar aerodynamic performance

can be achieved with feasible
designs with very different masses.
2. Sequential aero/structural
optimization is significantly inferior
to metrics that combine
aerodynamic and structural
performance.

NATIONAL RENEWABLE ENERGY LABORATORY 35

Minimum Turbine Mass /
AEP
 Cost of Energy

FCR (TCC + BOS) + O&M mturbine

COE = ⇡
AEP AEP

NATIONAL RENEWABLE ENERGY LABORATORY 37

 Vary Rotor Diameter

minimize COE(x; D) or m(x; D)/AEP (x; D)

with respect to x = {{c}, {✓}, {t}, }

subject to cset (x) < 0

NATIONAL RENEWABLE ENERGY LABORATORY 38

 Maximize AEP at Fixed Mass

NATIONAL RENEWABLE ENERGY LABORATORY 39

 Tower Contributions to Mass/Cost

mass cost

14% 17% 12%

53%
33% 24%
38%

9%

rotor nacelle tower rotor nacelle tower

bos o&m

NATIONAL RENEWABLE ENERGY LABORATORY 40

 Maximize AEP at Fixed Mass

NATIONAL RENEWABLE ENERGY LABORATORY 41

 Maximize AEP at Fixed Mass

NATIONAL RENEWABLE ENERGY LABORATORY 42

 Fixed Mass

mf ixed = mblades + mother

NATIONAL RENEWABLE ENERGY LABORATORY 43

 Fixed Mass

NATIONAL RENEWABLE ENERGY LABORATORY 44

 Fixed Mass

NATIONAL RENEWABLE ENERGY LABORATORY 45

 Conclusions

1. m/AEP can work well at a fixed

diameter but is often misleading
for variable diameter optimization
2. Problem must be constructed
carefully to prevent over-
incentivizing the optimizer to
reduce tower mass

NATIONAL RENEWABLE ENERGY LABORATORY 46

Minimum Cost of Energy

 Robust Optimization

Wind Power Class Wind Speed (50m)

3 6.4
4 7.0
5 7.5
6 8.0
7 11.9

NATIONAL RENEWABLE ENERGY LABORATORY 48

 Robust Optimization

minimize < COE(x; V hub ) >

where V hub ⇠ U (6.4, 11.9)
with respect to x = {{c}, {✓}, {t}, , D, rating}
subject to cset (x) < 0

NATIONAL RENEWABLE ENERGY LABORATORY 49

 Robust Design

Vhub

NATIONAL RENEWABLE ENERGY LABORATORY 50

 Conclusions

1. Optimization under uncertainty is

important given the stochastic
nature of the problem
2. Fidelity of the cost model can
dramatically affect results

NATIONAL RENEWABLE ENERGY LABORATORY 51

Conclusions

 Conclusions

1. Sequential (or single-discipline)

optimization is significantly inferior
as compared to integrated metrics
2. m/AEP can be a useful metric at a
fixed diameter if tower mass is
handled carefully
3. High-fidelity cost modeling and
inclusion of uncertainty are
important considerations

NATIONAL RENEWABLE ENERGY LABORATORY 53

 Acknowledgements

• Rick Damiani
• Pat Moriarty
• Katherine Dykes

• George Scott

NATIONAL RENEWABLE ENERGY LABORATORY 54

Thank You