Renewable energy consultants

Banking on Wind: gResource

AssessmentAssessment

Graham Slack

Quantum Leap in Wind Power in Asia Forum

June 2011

GL Garrad Hassan825 staff in 41 locations across 22 countries and 5 continents825 staff, in 41 locations, across 22 countries and 5 continents

HeerenveenSint Maarten

Kaiser-Wilhelm-Kaiser WilhelmKoog

Bristol

CopenhagenHinnerupOldenburgHamburg

GlasgowLondon

Slough

VancouverOttawa

PortlandSan Diego

BeijingTokyoShanghaiMumbai

BristolDublin

Paris

IzmirCairo

gWarsaw

LisbonBarcelona

Imola

San DiegoMontreal

PeterboroughAustin

MumbaiBangaloreNewcastleMelbourne

CairoZaragozaMadrid Bangkok

Ho Chi Minh

SingaporeManila

MonterreyPorto Alegre

Santiago

Wellington

Locations in red denote GL partner offices in the region.

Based On ExperienceSince 1984 - over 25 years experience in the sector

Wind Farm Energy AssessmentWind Farm Energy Assessment• analysing 20,000 MW of new projects per year

• 25% of all projects worldwide

Operational Assessment• 15% of the world’s installed capacity

Due DiligenceDue Diligence• over 25% of the world’s project financed wind farms

• world’s largest wind farm portfolio acquisition

Independent Engineer• the world’s five largest wind farm financings

• the first project financed offshore wind farm• the first project financed offshore wind farm

Short Term Forecasting• over 20% of the world’s operational wind capacity

Wind ResourceWind Resource

Wind resource is fundamentali ti l t V3- energy is proportional to V3

- Small wind error > large energy error

Mean wind speed AND distribution- Mesoscale wind maps for site ID

Reference wind data crucial- Reference wind data crucial

Assessment of energy production is a iti l j t i kcritical project risk

Wind is critical input to energy prediction

So good wind monitoring is a critical project need

Methodology Overview

Analyse and predict the long-term wind regime at site masts

gy

Predict the wind speed variations over the site

Predict gross energy output of all turbines

Predict likel energ lossesPredict likely energy losses

Result: Predicted long-term net energy output of the wind farm

At each step quantify the mean value AND the uncertainty

2. Correlate to a long-term reference

Wind farmsite masts

R f

3. Predict long-term wind regime1. Measure wind climate

400m

Reference station

20%15%10%5%

50 km

4. Wind flow modelling

Steps to long-

g

Steps to longterm energy

7. Uncertainty analysis

6. Estimate losses → Net energy80%

90%

100%

eeda

nce P90 (10 year)

5. Predict gross energy

AvailabilityTurbine WakeTopography

60%

70%

80%

Pro

babi

lity

of e

xce

P50 central estimate

P75 (10 year)

Power [kW]CpFrequency [hrs]

TopographyElectrical network

Other40%

50%

60 70 80 90 100 110

Annual average production [GWh]

P50 central estimate

wind speed

1. Measure wind climate

Masts in red Turbines in blue

Good quality instrumentsDifferent heightsClose to hub height

Mounting on mast> One year of data

Different heights for wind shear

M t t t ti l ti 1k

Calibration & records

Care for Masts at representative locations <1km

Care for local features & flow separation issues

Care for shadowing and icing

Uncertainty can be minimised through well designed measurement plans

2. Correlate to a long-term reference

Short-term measurementSite Layout

g

Site data

Long-term measurementWind Farm

1000mReference station

Reference stationAbsolute accuracy not vitalConsistency is vital

400m

50 km

Often there is no reference stationInspect reference site

MethodologyMeasure Correlate Predict (MCP)

3. Wind flow modelling

WAsP model is most common

g

s ode s os co oCFD is appearing but be careful –

different and complicated does not mean better

Wi d dWind speed map

Human Inter entionIntervention

This process has

Mast

Terrain Map

process has uncertainty

4. Predict Gross Energy

1,500

gy

900

1,100

1,300

W]

10-min mean power

0.5 m/s bin mean power 10 % 20% 30% 40%

500

700

Pow

er [k

W

Reliable power curve

-100

100

300

0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 200-3 3-6 6-9 >9 m/s

Reliable power curve is important

Wind Speed [m/s]

Combine:Power curve – power output at each wind speed (MW)

Wind rose

Wind distribution – time spent at each wind speed over one year (hours)

Gives total MWh in each year

Losses: minimise by modellingake and topographical losses

If turbine spacing too small, d d d hi h

wake and topographical losses

reduced energy and high turbulence: -ve turbine life

S f d l t iSo preferred layout is a compromise between highest wind areas, low wake effects

Analytical tools exist for wind farm modeling, including wake losses

This allows optimised layouts to increase energy capture

WindFarmer is one example

Losses: Turbine availability

99%

100% 250

y

97%

98%

99%

200

95%

96%

97%

Ava

ilabi

lity 150

WFs

in d

atab

ase

93%

94%

Win

d Fa

rm A

100

Num

ber o

f W

91%

92% 50

Monthly Avg 3 Monthly Avg

Annual Avg Number of WFs

90%0.0 1.0 2.0 3.0 4.0 5.0 6.0 7.0 8.0 9.0 10.0

Years of operation

0

g

Result: Net energy prediction

Rated Power 50 MW

gy p

Gross Output 170 GWh/annum Wake effect 98.7% Calculated Electrical efficiency 97 0% CalculatedElectrical efficiency 97.0% CalculatedAvailability 97.0% GH assumption Icing and blade degradation 99.5% GH assumption High wind hysteresis 99.2% CalculatedHigh wind hysteresis 99.2% CalculatedSubstation maintenance 99.8% Typical value Utility downtime 100.0% GH assumption Power curve adjustment 98.5% GH assumptionj pColumnar control loss 100.0% GH assumption Cold weather shut down 100.0% GH estimate Wake effect of future projects 100.0% GH assumption to be covered in

h Fi Athe Finance Agreement

Net output 153.2 GWh/annum

Result: Uncertainty

Source of uncertainty Wind speed Energy output [%] [ / ] [%] [GWh/ ]

y

[%] [m/s] [%] [GWh/annum]Anemometer accuracy 2.5% 0.21 1.4Consistency of reference 1.0% 0.08 0.6Correlation accuracy Mast A to Mast B 1.4% 0.12 0.8Shear 40 m to 60 m 2.0% 0.17 1.1Variability of 8.8 year period 2.0% 0.17 1.1y y p

Overall historical wind speed 0.35 2.3Wake and topographic calculation 4.0% 1.3Performance and availability 1 0% 0 3Performance and availability 1.0% 0.3Substation metering 0.3% 0.1Future wind variability (1 year) 6.0% 0.50 3.4Future wind variability (10 years) 1.9% 0.16 1.1Overall energy uncertainty (1 year) 4.3Overall energy uncertainty (10 years) 2.9

Probability Distribution• Mean = 50 GWh/year • Standard deviation = 3.5 GWh/year (in this example)

y

0.1

0.12

Normal distribution

P90 = 45.5 GWh

0 06

0.08P90 = 45.5 GWh

0.04

0.06

0

0.02

35 37.5 40 42.5 45 47.5 50 52.5 55 57.5 60 62.5 65

Net energy production (GWh/year)

35 3 5 0 5 5 5 50 5 5 55 5 5 60 6 5 65

Financing IssuesFinancing IssuesUncertainty in energy production dependant on several issues- Monitoring plan for site, grid reliability, supply warranties, O+MMonitoring plan for site, grid reliability, supply warranties, O M

contracts, turbine reliability and availabilityUncertainty reflected in probability of exceedence values:

Validation of methodology – UK dataAdj sted for indiness and a ailabilit sing c rrent methodsAdjusted for windiness and availability, using current methodsActual/predicted P50 values = 100.2%

Other benefits of measurement Extreme wind speedsOther benefits of measurement - Extreme wind speedsTyphoons frequent over much of north Pacific (2009 data):

Conclusions

Accurate wind assessment is critical- Good quality instrumentsGood quality instruments- Well mounted to avoid flow distortion - Adequate number of tall masts

C lib ti d d- Calibration and records

Combine with reference data to assess the long-term wind resource

Uncertainty assessment is critical toUncertainty assessment is critical to assess technical risks

Enough operational data to demonstrateEnough operational data to demonstrate the analytical process is robust: confidence for investors and lenders