NEXT-GENERATION ACOUSTIC WIND PROFILERS

Stuart Bradley1,2

Sabine Von Hünerbein2

1Physics Department, University of Auckland, New Zealand2Acoustics Research Centre, University of Salford, UK.

THE UNIVERSITY OF AUCKLANDNEW ZEALAND

The Two Existing Technologies

LIDAR• Scatters light from natural

particles

• Lowest height 40m• Highest reading 150m• Averaging time 10 min• Cost €150 k • Maintenance: currently

high, longer term unknown

SODAR• Scatters sound from

turbulent temperature variations

• Lowest height 10m• Highest reading 1000m• Averaging time 10 min• Cost €35 k• Maintenance: currently

low and well-established

THE UNIVERSITY OF AUCKLANDNEW ZEALAND

A careful, unbiased field trial

Calibration uncertainty 0.1%Variation with height 0.3%

0

20

40

60

80

100

120

-0.40 -0.30 -0.20 -0.10 0.00 0.10 0.20 0.30 0.40

Percentage calibration variation

Hei

ght z

[m]

AeroVironment calibration vs height

Cup vs SODAR residuals

Cup vs Cup residuals

Cup-cup variation equivalentto SODAR-cup variation

THE UNIVERSITY OF AUCKLANDNEW ZEALAND

LIDAR vs SODAR Performance?

0

20

40

60

120

140

-5 -4 -3 -2 -1 0 1

Slope Error [%]

Height [m]

SODAR(Bradley)

ZephIR (Albers)

ZephIR (Kindler)

ZephIR(Courtney)

ZephIR(Mikkelsen)

Difficult to find independent comparisons which compare LIDARs and SODARs under similar conditions !

THE UNIVERSITY OF AUCKLANDNEW ZEALAND

Are Comparisons Reliable?

• Many comparisons and field tests are done by companies or institutions which have major financial commitments (ownership of instruments, dependency on grants, development contracts, etc), particularly with the more expensive LIDARs. This can mean they concentrate on the product which has the greatest financial impact on them

• Many comparisons and field tests do not give enough statistical information to allow solid judgement to be made

THE UNIVERSITY OF AUCKLANDNEW ZEALAND

E. g. Correlation: R2 = 2(1- R2)/N1/2

R2 N = 200 N = 400

0.97 0.9658- 0.9742 0.967- 0.973

0.98 0.9772- 0.9828 0.978- 0.982

0.99 0.9886-0.9914 0.989- 0.991

THE UNIVERSITY OF AUCKLANDNEW ZEALAND

E. g. Typical remote vs mast output

Really important• Fit a straight line through the origin (why expect non-zero remote

wind when actual wind speed is zero ?) • Number of data points N (affects uncertainties)• Averaging time• Height

0 5 10 15 200

5

10

15

20

Mast wind speed [m/s], Height=80 m

SO

DA

R w

ind

sp

ee

d [m

/s]

N=358, Average 600 s, Slope =1.05+/- 0.02, r2=0.9866+/-0.001

THE UNIVERSITY OF AUCKLANDNEW ZEALAND

Root-mean-square deviation

0 5 10 150

0.01

0.02

0.03

0.04

0.05

0.06

0.07

0.08

Wind speed [m/s]

RM

S d

evi

atio

n fr

om

fit /

win

d s

pe

ed

• Is correlation coefficient R2 important? (what will you do with it?)• Are instrument calibration details important? (as long as they are stable)• The important parameter is the most likely size of wind speed error

THE UNIVERSITY OF AUCKLANDNEW ZEALAND

What does the extra €110k buy you ?

A LIDAR has• A smaller (but heavier) box• Quiet operation (but with turbine noise and distance from

housing, who cares?)• Marginal increase in R2 (perhaps)• Marginally improved accuracy of wind speed ??• Increased maintenance costs (probably)• Decreased performance in rain and fog• Higher power consumption• No data below 40m or above 150m• Large height steps (poor vertical resolution)

THE UNIVERSITY OF AUCKLANDNEW ZEALAND

History and What is Next

• Big advance in SODARs in 1980s with multi-speaker/microphone phased-array technology

• Big advance in LIDARs in 2000s with fibre-based lasers: directly impacting on wind energy

• SODAR developers are only now responding to the impact of LIDARs in wind energy: expectation from new, intense, R&D is another “technology leap”

THE UNIVERSITY OF AUCKLANDNEW ZEALAND

SNODAR

• 1m resolution• Autonomous operation through Antarctic winter

THE UNIVERSITY OF AUCKLANDNEW ZEALAND

SONDE

• Long (several seconds) FM-CW acoustic pulses

• Huge increase in signal strength (30 dB)

• Wind speed estimates every few meters, and every few seconds

• Some remaining technological challenges

• Current work on real-time tracking of aircraft wing-tip vortices (during landing and take-off)

THE UNIVERSITY OF AUCKLANDNEW ZEALAND

New Acoustic Wind Profilers

Triton• Similar acoustic design to

long-established ASC (AeroVironment) and ART SODARs

• Solar powered

AQ500• Innovative acoustic

design (3 dishes, 3 tilted beams, small footprint, reduced diffraction)

• Generally needs additional baffles

THE UNIVERSITY OF AUCKLANDNEW ZEALAND

Data availability (example from an AQ500)

0 20 40 60 80 100

20

35

50

65

80

95

110

125

140H

eig

ht m

Data availability %

THE UNIVERSITY OF AUCKLANDNEW ZEALAND

Where will Improvements be Made?

For SODARs:• Clutter (ability to operate near masts etc, quieter)

• Smart signals (to give shorter averaging times, lower rms errors)

• Lower power

• Compact (smaller footprint)

• Better operation in rain

THE UNIVERSITY OF AUCKLANDNEW ZEALAND

Acoustic camera picture of diffraction

THE UNIVERSITY OF AUCKLANDNEW ZEALAND

Beam design

Plot of measured and modelled correlation between winds measured by different beam combinations on the same 5-beam SODAR

At 60m, the time taken for air to travel from one beam to another (at the wind speed during these measurements) is equal to the time between acoustic pulses

The shape of this curve is a measure of the spatial correlation for wind

020406080

100120140160180200

0.60 0.70 0.80 0.90

R2

Hei

ght [

m]

THE UNIVERSITY OF AUCKLANDNEW ZEALAND

Bistatic SODARs

• Useful to separate transmitter and receiver (‘bistatic’) so that turbulent velocity fluctuations are also measured

• The ‘break-even’ angle is about 11: this would require separating the transmitter and receiver by about 20m

10

20

30

40

30

210

60

240

90

270

120

300

150

330

180 0

Power dB.Blue=velocity fluctuations,Red=temperature fluctuations

THE UNIVERSITY OF AUCKLANDNEW ZEALAND

“Piccolo” Bi-Static + Smart signals

• Continuous cycle step-chirp• Unfortunately, the direct signal saturated all channels so all useful

data is lost during those periods (it needs acoustic baffles)• However, beam-forming for z = 32m, picks up the gaps at 0.3s,

0.4s, …,0.9s. Adding resulting power spectra gives 8 dB improvement for SNR of the 3000 Hz signal. This is a very useful increase!

0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1-0.5

0

0.5

Time [s]

Am

plit

ud

e 3000Hz 3150Hz 3300Hz 3450Hz 3600Hz 3750Hz 3900Hz 4050Hz 4200Hz 4350Hz

0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1-10

0

10

Time [s]

Am

plitu

de

[V

]

Transmitted

Received

THE UNIVERSITY OF AUCKLANDNEW ZEALAND

The Next-Generation SODARs

• Will look different (smaller, new acoustic design)

• Will be quiet, except when alongside

• Will sound different (smart signals)

• Will be solar-powered

• Will clearly surpass current LIDAR performance

THE UNIVERSITY OF AUCKLANDNEW ZEALAND

Summary

• Next-generation SODARs, with very superior performance will emerge in the next 2-3 years

• The design challenges are not extreme

• In an increasingly recession and risk-averse financial environment, the business case for SODARs is strong