REAL-TIME SHAPE ESTIMATION WITH FIBER OPTIC SENSORS DISTRIBUTED IN ROTOR BLADES

Smart Systems & Structures Lab.

New concepts and Methods : Hardware

1

REAL-TIME SHAPE ESTIMATION WITH FIBER OPTIC SENSORS DISTRIBUTED IN ROTOR

BLADES

Hong-Il Kim1, Lae-Hyong Kang1, Jae-Hung Han1*, Hyung-Joon Bang2

2010.04.23. 09:00~10:30

1Department of Aerospace Engineering, KAIST, Republic of Korea2Wind Energy Research Center, KIER, Republic of Korea



2

Outline

Experiments

Conclusion

Introduction

Numerical Study



Introduction - Research Backgrounds (Condition monitoring with Shape estimation) - Why Fiber Bragg Grating sensors? - Shape Estimation based on Measured Strains Using FBG Sensors - Research objectives

3



4

Condition Monitoring for Reliability

Rumsey, 2009, “Condition Monitoring and Wind Turbine Blades,” Wind Turbine Reliability Workshop

Full-scale Testing

O & MData Base

Designed-in

Maintainability

Accurate Loads-Design Requirements

Appropriate Environmental

Conditions

Design

ed-in

Reliab

ility

Reliability

Analysis

High-ReliabilityWT Blade

Conditio

n

Mon

itorin

g

• Strains• Loads• Cracks• Dry-spots• Voids• Operational Dynamics• Temperature gradients• Lightening

Sense What?

Blade shape(Deformation)



5

Why blade shapes are important?The “Blades”

The shapes of the “Blades” influence the whole systems’ status

Status Monitoring

Design Validation

Active control for blades

Blade Shape Information - Bending => Flapping motion- Torsion => Pitching motion



6

Why blade shapes are important?

The real-time shape estimation techniques based on embeddable sensors

Shape estimation On operation

It is difficult to directly monitor the shape changes on operation.

Marker(DNW)

SPR(Stereo Pattern Recognition)

Optical image processing techniques

PMI (Projection Moire Interferometry)

Pattern (NASA Langley)

Direct Shape Measurement



• Typical embeddable sensors (Strain gauge, accelerometer..)- Complex electric-wiring (Slip ring) + Significant measurement noise

• FBG (Fiber Bragg Grating) sensor– Small, lightweight, High sensitivity, Electro-magnetic immunity – No hygro-effects and easily installable onto/into host structures.– Multiplexing– Real time strain acquisition

– FBG sensors are already applied to the load monitoring

Why Fiber Bragg grating sensor?

7

(1 )Be s f

B G

p T

[1] A fibre Bragg grating sensor system monitors operational load in a wind turbine rotor blade[2] Advanced Wing Turbine Controls Input Based on Real Time Loads Measured with Fibre Optical Sensors embedded in Rotor Blades

[1] [2]

Optical fiberBragg grating

B

I

I

B

I

Input spectrum

Reflected signal Transmitted signal

L 10mmIndex of refraction of

fiber core

zz2z1

ne

Δn = 10-5 to 10-3

2B en

Slip ringOptical Rotary Joint



8

Shape Estimation based on Measured Strains Using FBG Sensors – previous works

[DST]Discrete strains

full state vectordisplacement field

Estimation model using• modal approach• FEM data

0200

400600

8000

200400

600

-0.02

-0.01

0

0.01

0.02

0.03

y [mm]x [mm]

disp

lace

men

t [m

m]

real displacement fieldestimated displacement field

4 Laser Sensors

Shaker

16 FBG Sensorsy

x

(0,0)

Distributed FBG sensors

Real time Shape Estimation of a Two-Dimensional Structure

Fr

Kk

C

kx̂

kky ̂ˆ

C’F1ˆ kx

RQ ,2 kP

0x

0P

wState matrixOutput matrix

Weighting matrix

Error covariances

State Space

Integration of the filtering technologies

Real-time shape estimation of the Rotating structures



9

Research objectives• Primary objectives

– Development and validation of a real-time shape estimation technique for Wind Tur-bine blades using FBG sensors

• Research steps① Numerical study on the shape estimation method for the rotating beams

• Rotating beam dynamics are simulated. (displacement fields, a few strain data)• Displacement is reconstructed using strains • Shape estimation method is evaluated through the comparison between original dis-

placements and the estimated displacements. • Sensor location is optimized.

② Experimental Demonstration of the real-time shape estimation for the rotating structures• FBG sensors are used to measure multi-point strains of the beam.• Structural deformation shape of the rotating beam is estimated. • The estimated shapes are compared with the directly measured shapes using pho-

togrammety.



10

Numerical Study - Simulation Steps - Simulation Results - Optimization of Sensor locations



FMD

Virtual experiments – simulation steps

DST

N MT

( )ey t( )s t

Beam model

Mode shapes

11

Rotating beam motions are simulated - Full-field Displacement & strain

( )y t( )x t

M : # of sensors, N : # of disp. Points, n : # of used modes

Discrete strainsDST matrix constructed Shape estimation

Sensor locationOptimization

Full-field Strain & Displacement

Sensorlocation

Evaluation



12

• Rotating beam dynamics are simulated– Full-field displacement & distributed strain

Simulation Results

Full-field Displacement Discrete strains

Estimated Deflection Strains at a few points are used for reconstruction of full-field displacement via DST matrix.

Comparison

Directly Simulated Deflection



13

Simulation ResultsRotating beam displacement at the Tip of the beam(Numerical simulation vs. Shape estimation results)

- Shape estimation using simulated strains are performed- Full-field displacement from numerical simulation are compared with Estimated shape using

strains

Numerical simulation

Shape estimation

Reconstructed fromstrains

Directly Simulated



14

Optimization of Sensor locations

Initial Seed Sensor 1: 0~4cm Sensor 2: 5~16cm Sensor 3: 19~31cm Sensor 4: 33~38cm

Condition Number of DSTSensor position CN=19, (4.0,15.0,21,33)

ˆ( ) ( )DSTy t T tEstimated displacement Measured strain

DST matrix (Displacement Strain Transformation)

1C T T Condition number

- Used as the objective function for sensor location optimization - Small condition number indicates good information conservation during matrix operations



15

Experiments –rotating beam - Rotating beam test setup - Test measurand/DST matrix - Results



16

Test setup – Demonstration of the rotating beam

fbg1fbg2

fbg3

fbg4Reconstructed shape (DST)Photo-grammetry

Images taken by High-speed camera

Optical rotary joint



17

Test measurand

• Measurand– Four Strains (FBG sensors)– 13 Marker positions (Photogrammety)– Angular position

Strain by FBG

Rotating angle60RPM case

In. Volt. Ang. Vel

Case 1 0.1V 15 RPM

Case 2 0.2V 30 RPM

Case 3 0.3V 45 RPM

Case 4 0.4V 60 RPM



18

DST matrix

fbg1

fbg2

fbg3

fbg4

point pt1 pt2 pt3 pt4 pt5 pt6 pt7 pt8 pt9 pt10 pt11 pt12 pt13[mm] 0 40 80 120 160 200 240 280 320 360 400 440 500

Marker position

　 FBG1 FBG2 FBG3 FBG4wavelength

[nm] 1533 1541 1547 1556[mm] 40 150 210 310

FBG position

DSTmatrix

13 4

TDSTT

　　　 -1.20 -4.14 -7.94 -11.99 -16.04 -20.06 -24.16 -28.36 -32.59 -36.73 -40.72 -46.54

　 0.66 1.35 0.58 -2.18 -6.38 -10.90 -14.81 -17.95 -20.89 -24.40 -28.84 -36.56

　 -0.38 -0.79 -0.51 0.35 0.90 0.01 -2.89 -7.35 -12.16 -16.12 -18.74 -21.15

　 0.07 0.15 0.11 -0.02 -0.12 -0.22 -0.61 -1.90 -4.77 -9.52 -15.91 -26.94

- Acrylic beam (500mm×20mm×1.9mm) was used for denstrating large deflection in low speed

Optimized sensor locations

Marker positions



19

Results – qualitative aspects

30 RPM rotation 60 RPM rotation



20

Results -Shape comparison between DST vs. Images

Directly Measured(High Speed Cam-era)

Estimated(from strains using FBG)



21

Results – quantitative aspects

Pole effect

15RPM 30RPM 45RPM 60RPM

MAC (median) 0.993 0.997 0.999 0.998Skewed distribu-

tion

Time [s]



Conclusion

• Development of the shape estimation technique for a rotating structure– A real-time deflection of the rotating beam is successfully estimated based displacement -strain

transformation - Sensor location optimization is executed.

- From the test results, it is clear that beam shape estimation of the rotating beam is successfully performed based on DST method and strain data obtained by FBG sensors.

– FBG (Fiber Bragg grating) sensor is selected as a strain sensor because of many inherent advan-tages of fiber optic sensors and multiplexing capability.

22



Hong-Il Kim ([email protected]) Ph. D. candidateAerospace Engineering, KAIST

Jae-Hung Han ([email protected]) Associate Prof.Aerospace Engineering, KAISTSmart Systems and Structures Lab. : Design & Control

Visit our website: http://sss.kaist.ac.kr

23

THANK YOU!

Acknowledgments

This work was supported by the Korea Institute of Energy Research through the research project (grant No. NT2009-0008).

mailto:[email protected]

mailto:[email protected]

http://sss.kaist.ac.kr/

REAL-TIME SHAPE ESTIMATION WITH FIBER OPTIC SENSORS DISTRIBUTED IN ROTOR BLADES

Documents

new concepts

shape estimation techniques

shape changes

hardware7shape estimation

fiber optic sensors

fibre optical sensors

measured strains

wind turbine blades