Integrated Design and eg a ed es g a dSystem Analysis at

S diSandia2010 Sandia Wind Turbine Blade Workshopp

Brian ResorT h i l St ff D i T l L dTechnical Staff, Design Tools LeadWind & Water Power Technologies

Sandia National Laboratoriesbrresor@sandia.gov

Sandia is a multiprogram laboratory operated by Sandia Corporation, a Lockheed Martin Company,for the United States Department of Energy’s National Nuclear Security Administration

under contract DE-AC04-94AL85000.

@ g(505) 284-9879

Modeling and Simulation MotivationModeling and Simulation Motivation

Smaller turbines → Larger turbines

$ → $$$

B ild B k R d i → Ad d Si l iBuild, Break, Redesign → Advanced Simulation

Paper airplaneCommercial Aircraft

Wind Turbine Design Elementsg

Blade Analysis

Full System

Material Properties & Layup

Full 3D Blade Structural Analysis

Full System SimulationAerodynamics

Shapes

Blade Flutter

Aerodynamic Loads

Advanced Performance

Data

Shapes

ControlsInflow

Recently Supported Research Projectsy pp j

The use of design tools at Sandia provides a bridge between research and application by supporting our projects:research and application by supporting our projects:

SMART rotor technology• System simulations with active reduction of loads • System simulations and blade analysis to support SMART

Blade designbl d i i i CX-100 sensor blade activities

• Blade test definitions• Sensor placements• Sensor placements

“Certification” of the 100m blade design concept Blade fatigue failure modelingg g

Wind Turbine Design Elementsg

Blade Analysis

Full System

Material Properties & Layup

Full 3D Blade Structural Analysis

Full System SimulationAerodynamics

Shapes

Blade Flutter

Aerodynamic Loads

Advanced Performance

Data

Shapes

ControlsInflow

Blade Design with NuMADANSYS FE ModelNuMAD ANSYS Analysis

1

X

Y

Z

*

Modal

ANSYS FE ModelNuMADNuMAD:Numerical ManufacturingAnd Design Tool

ANSYS Analysis

pseudoSERI8

Blade Geometry

Buckling

Materials & Layups

Stress &Strain

2 000

3,000

4,000

5,000

6,000

7,000

8,000

9,000

FlpStff (kN/m^2)

EdgStff (kN/m^2)

Stack Placement

Beam Properties

0

1,000

2,000

0 0.2 0.4 0.6 0.8 1

BlFract

Use of Offset-Thickness Shell Nodesf ff

Offset-thickness nodes are most desirable for wind turbine blade FE models

The outer blade surface is the specified surface

Offset Node Shell Element Problemsff

Work by Daniel Laird, and others, documented in 2005 uncovered significant problems with the use of offset-node shell elementsproblems with the use of offset-node shell elements

The wind industry sought other solutions for blade models: The wind industry sought other solutions for blade models:• Restrictions to mid-thickness node shell elements • Entire wind turbine blades modeled with solid elements

h f d l l l h• High fidelity cross sectional analysis, such as VABS

New shell formulation in ANSYS 12.0Very good news!Very good news!

Redo the AIAA 2005 investigations using new

90 Sh ll Mid2

Results including old formulation

Results not including old formulation

formulation: SHELL281

Shear:2030405060708090

or (%

Diff

eren

ce)

Shell MidShell ExteriorShell Exterior, ImprovedSolid

1

0

1

ror (

% D

iffer

ence

)

Shell MidShell Exterior, ImprovedSolid

• Element aspect ratio-10

01020

0.01 0.10 1.00 10.00

Erro

Aspect Ratio (L/W)

-2

-1

0.01 0.10 1.00 10.00

Err

Aspect Ratio (L/W)100 5

Sh ll id

• Number of Elements 20

40

60

80

Erro

r (%

Diff

eren

ce)

Shell Mid

Shell Exterior

Shell Exterior, Improved

Solid-1

0

1

2

3

4

Erro

r (%

Diff

eren

ce)

Shell MidShell Exterior, ImprovedSolid

-20

0

0 5,000 10,000 15,000 20,000

E

Number of Elements

-3

-2

0 5,000 10,000 15,000 20,000

E

Number of Elements

Error as high as Error mostly belowError as high as 80-90%

Error mostly below 1%

Blade Structural Model Si lifi ti Wind turbine blades include Simplification• Variable section shapes

with twist, • Multiple materials and

composite layups (glass, carbon, balsa, foam, epoxy, adhesives)

• One or more shear• One or more shear webs

Beam Model:Up to 6 DOF per node

(Colors represent composite stacks)

Beam Propertiesp

Motivation: Efficient aeroelastic analysis for design and certificationcertification• Time marching system response simulation• Stability analyses• Blade test design and setup

i l i f h f ll i di ib i Typical outputs consist of the following distributions• Bending, torsion and axial stiffness• Coupled stiffness• Coupled stiffness• Shear center coordinates• Tension center coordinates• Inertial properties: masses and center of mass

Property Distribution ComputationsProperty Distribution ComputationsTwo-Dimensional Approach Pros

Three-Dimensional Approach Pros

• Readily and freely available• Computationally efficient

Cons

• Includes three dimensional effects Cons

• Requires creation of the finite element d l• Limited to 2D understanding

• Simple examples below:

model

0

1

2

Undisplaced nodesDisplaced nodes (exaggerated)

= dxdyxyxEflapEI 2),(_ ,3

4

5

6

7-0 4-0.2

00.2

Blade Span= dxdyyyxEedgeEI 2),(_ ,

+= dxdyyxyxGGJ ))(,( 22 and

= dxdyyxEEA )(

Useful Tool: PreComp• Created by Gunjit Bir, NREL

Useful Tool: Beam Property Extraction (BPE)

• Created by David Malcolm GEC

7

-0.2 0 0.2 0.4 0.6 0.8

-0.4

Chord

= dxdyyxEEA ),(

• Created by David Malcolm, GEC• Distributed with NuMAD (Sandia Labs)

BSDS Model2D & 3D b i2D & 3D beam properties

Flapwise bending stiffness Edgewise bending stiffness*10

8 10

8

106

107

Nm

2 )

3D Approach (BPE)2D Approach (PreComp)

105

106

107

m2 )

3D Approach (BPE)2D Approach (PreComp)

104

105

EI_

Edg

e (N

2

103

104

10

EI_

Flap

(N

0 2 4 6 8 1010

3

blade span

0 2 4 6 8 10

101

102

blade span

*Note: bugs were identified and resolved

BPE Property ~50% of PreComp

Note: bugs were identified and resolved in PreComp as a result of this work. See Resor 2010 AIAA-SDM paper for details.

BSDS Model2D & 3D b i2D & 3D beam properties

Torsional stiffness10

8

106

107

)3D Approach (BPE)2D Approach (PreComp)

103

104

105

GJ

(Nm

2

0 2 4 6 8 1010

2

10

blade span

Note discrepanciesNote discrepancies along entire span

Current Sandia Classical Flutter CapabilityCurrent Sandia Classical Flutter Capability Current capability utilizes:

• MSC.Nastran 2005• FAST2NAST.m (Matlab routine)( )

Required inputs: lift curve slope and pitch axis location along with information taken from ad.IPT and blade.DAT files utilized by FAST

• Fortran executable Determines necessary mass stiffness and damping matrix additions due to aerodynamic Determines necessary mass, stiffness, and damping matrix additions due to aerodynamic

effects (Theodorsen) Generates additional Nastran decks for the complex eigenvalue solve

The analyst iterates on operating speed, following the complex modes, to find the fl tt dflutter speed

1010Flutter Mode Shape

( )[ ] ( ) ( )[ ] ( ) ( ) ( )[ ] 0,,, 0 =Ω+Ω++Ω+Ω+Ω+Ω+ uKKKuKuCCuMM acstcaCa ωω

Matrix Description

-5

0

5

10

(m)

-5

0

5

10

(m)

pM, C, K Conventional matrices

(with centrifugal stiffening)

Ma(Ω), Ca(ω, Ω), Ka(ω, Ω) Aeroelastic matrices

0

10

20

30 -10 -5 0 5 10

-10

(m)

(m)

0

10

20

30 -10 -5 0 5 10

-10

(m)

(m)

CC(Ω) Coriolis

Kcs(Ω) Centrifugal softening

Ktc Bend-twist coupling

Wind Turbine Design Elements

Blade Analysis

g

Full System

Full 3D Blade Structural Analysis

Material Properties & Layup

Full System Simulation

Blade FlutterAerodynamics

ShapesAerodynamic

Loads

Advanced Performance

Data

Shapes

ControlsInflow

Blade Aerodynamics Modelingy g• Blade aerodynamic model requires

computational models of airfoil performance(h d d ) f l i i ll i d

Blade Aerodynamic Model

SNL Thunderbird Computing Cluster

• Many (hundreds) of solutions typically required

Advanced 3D Airfoil Simulationp g Simulation

Wind Turbine Design Elements

Blade Analysis

g

Full 3D Blade Structural Analysis

Material Properties & Layup

Full SystemBlade Flutter

Aerodynamics

Shapes

Full System Simulation

PerformanceData

ShapesAerodynamic

Loads

Advanced

InflowControls

Wind Turbine System Modeling

Aerodynamics All disciplines come together to y All disciplines come together to

enable assessment of system effects

Changes to component technology

ControlsStructures

g p gyaffect system behavior resulting in both costs and benefits

Design Criteria ExamplesDesign Criteria Examples

Design Requirements Example Load CasesDesign Requirements Conditions:

• 20 year minimum design life• Normal wind conditions

Example Load Cases Normal production: Fatigue and/or ultimate

loads due to • Normal turbulence

• Normal wind conditions• Extreme wind conditions• Wind defined by average wind speed

and turbulence intensity

• Extreme turbulence• Extreme gust

Extreme wind speed Extreme direction change

Loads• Ultimate loads – can the system

withstand the largest expected loads?• Fatigue – can it withstand the

Extreme wind shear

• Start up and shut down

Normal production with faults• Yaw system fault• Fatigue – can it withstand the

combination of all loads?• Functional requirements – deflections

(tower clearance)

• Pitch system fault• Loss of electrical load, etc.

Parked Turbine• Extreme loads• Extreme loads• Normal loads

Transportation loads

System Analysis with Wind Turbine Aeroelastic Simulation

Aerodynamic PerformanceAeroelastic Simulation

Aeroelastic System

Turbulent Wind Input

Dynamics ModelSystem Response

Includes ControlsStructure and Materials Includes Controls Implementation

Tools: FAST and ADAMSFAST ADAMS

Available from NREL MSC Corporation

Up to 2 each of blade Unlimited; depends onStructural modes included Up to 2 each of blade flap/edge and tower F-A/S-S

Unlimited; depends on discretization

Aerodynamic forces AeroDyn AeroDyn

Very fast computations; Code verification;

108

/Hz

UsesVery fast computations;Adequate for most work,

especially certification

Code verification; Simulations requiring several

dynamic structural modes

106

g M

omen

t, (k

N)2 /

102

104

se R

oot B

endi

ng

FAST - RigidFAST - FlexibleADAMS - RigidADAMS - Flexible

10-1

100

101

100

Flap

wis

Frequency, Hz

Aeroelastic Tool Computation Times

8 1.5MW VSVP Simulation Times

p Project goals dictate the system dynamics tool used

6.826

7

8

e/

10min Simulated Turbulent Wind:Intel Core2 6700@2.66 GHz

4 GB DDR2 RAM; Win 7 64-bit OS

4

5

puta

tion

Tim

em

ulat

ed T

ime

0 16 0.421.04

1

2

3

Com

pSi

m

0.160

FAST FAST&Simulink ADAMS ADAMS&Simulink

Typical Structurally Nonlinear ypSystem

Certification

yTailored

Blades/TowerControls

Research

Example: Fatigue Analysis of System R D tResponse Data

Percent Change in Equivalent Fatigue Load

9m/s

11m/s

18m/s

Avg.Wind5.5m/s

Avg.Wind7m/s

q gLow Speed Shaft Torque -1.7 -4.9 -33.5 -3.1 -7.3

Blade Root Edge Moment 1.7 1.9 -2.5 0.8 0.8Blade Root Flap Moment -31.2 -27.1 -30.4 -23.1 -26.3

Tower Base Side-Side Moment -0.1 -8 -7.2 -0.9 -2.9Tower Base Fore-Aft Moment -18.6 -16.5 -13.8 -5 -8

Tower Top Yaw Moment -53.2 -42.9 -43.4 -25.1 -32.2

Importance of System Analysisp f y y Return at this point to the original motivation:

Bridging research and application

A philosophy that seems to always apply for modifications to properly designed wind turbine systems:

There are no free handouts from Mother nature

It is just as important to understand and report the cost of an innovation as well as the benefit; Common system costs include

• Increased forces and moments elsewhere in the systemd l• Increased complexity

• Decreased energy capture

LLMOLRCICCFCRCOE + )&(* LLMOAEP

COE ++= )&(

Wind Turbine Design Tools in Use at SandiaMaterialMaterial

Properties & Layup:-NuMAD

Full 3D Blade Mechanics:

-ANSYS

StructuralProperties:

Blade Flutter Stability:

-MSC.NastranProperties:-PreComp

-ANSYS/BPE-BModes

Full System Simulation:

-FASTAirfoils

FAST -ADAMSAerodynamic

Results Postprocessing:

-CrunchPerformance

Data:

Shapes

Corrections:Loads:

-AeroDyn

Advanced Inflow Model:

TurbSim

Crunch-Matlab-ARC2D

-XFOIL-AirFoilPrep

Controls:-Simulink

-TurbSim-IECWind

Summaryy Blade design tools have been shown

Success with new shell element formulations has been shown

Full aeroelastic models are critical for assessment of system effects of innovations on overall response and on the cost of

l i ielectricity

Thank you!