the impact of ice formation on wind turbine performance and aerodynamics

The Impact of Ice Formation on Wind Turbine Performance

and Aerodynamics

S. Barber, Y. Wang, S. Jafari, N. Chokani and R.S. Abhari

barbers@ethz.ch

European Wind Energy Conference, Warsaw21st April 2010

21.04.10 2

Overview

• Motivation• Research objectives• Experimental approach• Results and discussion

– Experiment (performance)– CFD (aerodynamics)

• Conclusions

• Wind energy is world’s fastest growing source of electricity production− 160 GW installed wind capacity reached in 2009

• Wind-rich sites must be effectively taken advantage of– Many wind-rich sites are in cold, wet regions

Icing a Global Challenge for Wind Energy

21.04.10 3

Northern USA & Canada

Scandinavia & Russia

ChinaAlps

Decreasing temperatureIncreasing humidity

Icing Dependent on Altitude• Ice formation dependent on many factors,

including:– Air humidity– Air density– Air temperature– Wind velocity– Object size on which ice formed– Cloud water droplet concentration

• Rate of ice formation therefore highly altitude-dependent:– Altitude 800-1,500m: high risk of ice formation– Altitude > 1,500m: lower risk of ice formation

21.04.10 4

Pow

er (

kW)

Velocity (m/s)

• Results from Alpine Test Site Gütsch, Switzerland: 2,300 m altitude– 10-min average power and velocity measurements over a year (Meteotest)*– Corrected for density and hub height

• Measured Annual Energy Production 20% less than predicted• Possible reasons:

– Icing: investigated here– Gusts and turbulence in complex terrain: being investigated in ETH sub-scale test facility

Measured Energy Yield 20% Less Than Predicted

21.04.10 5*Barber et al, “Assessment of wind turbine performance in alpine environments,” submitted to J. Wind Eng. Ind. Aero

Power curve

Annual average of measurements

Research Objectives

• Quantify performance of wind turbines with specified icing on rotor blades in a systematic, parametric study

• Detail impact of icing on aerodynamics

21.04.10 6

21.04.10 7

Specification of Simulated Icing

2D profile 2D ice accretion code (LEWICE), atmospheric

conditions at Gütsch

Span-wise distribution1000s of photographs from Alpine Test Site

Gütsch

r/R = 0.90 = 8.8o

Vrel = 31.6 m/s

r/R = 0.63 = 8.8o

Vrel = 22.2 m/s

r/R = 0.30 = 6.9o

Vrel = 11.2 m/s

2D profile + spanwise distribution ≅ simulated icing

21.04.10 8

Specified Ice Shapes

high-altitude, Gütsch conditions = non-“extreme”

low altitude, Bern Jura conditions = “extreme”

5% chord 5% chord 5% chord 5% chord 5% chord 10% chord

21.04.10 9

ETH Sub-Scale Model Wind Turbine Test Facility

• Velocity and acceleration of turbine can be precisely specified: arbitrary velocity profiles• Turbulence intensity can be controlled with grids• Systematic and parametric studies can be carried out: not possible in field

Salient characteristics of facility• For given model & flow velocity, advantage in Reynolds number of factor 15 gained using water as test medium, compared to air• Free-stream turbulence intensity is zero: reliable baseline conditions• Controlled test conditions: accurate assessment of performance due to ice shapes.

Summary of test conditions

Tip speed ratio = 3 - 8

Re0.75 = 1.4 x 105

21.04.10 10

Model and InstrumentationRotor geometry:• Blade geometry matches NREL

S809• Interchangeable hub, 2 or 3

bladed

Instrumentation:• Torque measured with in-line

torquemeter• Torquemeter installed between

motor & shaft• Series of tare measurements

undertaken to remove drive & seal resistances

• Power coefficient:

CP Trotor

0.5u3Arotor

Max. relative errors3.0% in CP

1.1% in tip speed ratio

Turbulent skin friction:

Reynolds number correction:

ETH Sub-Scale Model Matches NREL

21.04.10 11

corrected uncorrected

21.04.10 12

Effect of Ice on Performance

• Ice on outboard 5% of span has most significant effect on performance

• Ice removal / prevention systems can be substantially more efficient if their effectiveness is tailored to outboard 5% span of blades

No ice

21.04.10 13

Effect of Ice on Performance

• Sawtooth shapes do not have significantly different effect on CP compared to smooth shapes

• No power generated for Case F (“extreme”) at tip speed ratio ≥ 6

No ice

21.04.10 14

“Extreme” Icing Has Large Impact on Annual Energy Production

Gütsch conditions / non-“extreme”

Bern Jura conditions / “extreme” − Predicted loss is in good agreement with Gütsch data− Non-”extreme” icing has small impact− “Extreme” icing has large (15% loss) impact

Annual Energy Production (AEP) • Estimated using IEC standard bins method • Optimal tip speed ratio• Measured wind speeds & atmospheric conditions at Gütsch; icing in 2 months per year

Gütsch measurements

21.04.10 15

CFD ModelANSYS CFX• Commercial, implicit flow solver• One blade, periodic boundaries, k- turbulence model with scalable

wall function• Computational grid: 4 million cells

Blade surface

Periodic boundary Periodic boundary

4R

4R

R = rotor radius

x

y

z

CFD Results Match Experiments

21.04.10 16

Tip speed ratio = 6

Cp

,wit

hou

t ic

e –

CP

, w

ith

ice

(C

P)

21.04.10 17

“Extreme” Ice Causes Extensive Flow Separation

• Flow separation limited to root for non-“extreme” ice

• No separation on blade

Clean Non-“extreme” “Extreme”

• Flow separation over ¾ of blade for “extreme” ice

3.0

2.0

1.0

0.0

Total Velocity (m/s) z-y plane, x = -0.1R

Blade rotation

Incidence ≈ 15o

Incidence ≈ 5o

Incidence ≈ 5o

Incidence ≈ 15o

Incidence ≈ 5o

Incidence ≈ 5o

Incidence ≈ 30o

Incidence ≈ 15o

Incidence ≈ 15o

21.04.10 18

Conclusions

• For icing at high altitudes > 1,500 m: non-”extreme” ice on outboard 5% of the blade has most significant impact on performance → tailor removal systems for outboard 5% of blade

• For icing at lower altitudes, 800 – 1,500 m: Annual Energy Production can be reduced up to 15% due to “extreme” ice

• At the Alpine Test Site Gütsch, icing does not explain the losses of 20% in Annual Energy Production

• Gusts and turbulence are being investigated in the sub-scale model wind turbine test facility at ETH Zurich, which allows testing of dynamically scaled models at near full-scale non-dimensional parameters

21.04.10 19

Acknowledgements• Financial support: Swiss Federal Office of Energy (BFE)• LEC workshop: H. Suter, T. Künzle, C. Troller and C.

Reshef

barbers@ethz.ch