micro wind turbine paper

18
Micro wind turbine performance under real weather conditions in urban environment A Glass BEng(Hons) and G Levermore BSc ARCS PhD DIC FCIBSE School of Mechanical, Aerospace and Civil Engineering, University of Manchester, Manchester, UK The aim of this article is to evaluate the performance of micro wind turbines in a built-up environment. For this purpose, five independent micro wind turbine systems, consisting of two distinctly different models, were tested and evaluated under real life conditions over a period of 12 months. This article provides an overview of the experimental set-up used to test the two different micro wind turbines and then goes on to present the basic background theory for horizontal axis micro wind turbines and the variation of coefficient of performance with wind speed. The wind potentials at the test site were assessed to determine the theoretical outputs of the turbines which were compared with the measured outputs over a year. The measured outputs were disappointingly low. One reason for this is turbulence, for which directional turbulence (lateral turbulence) has been shown to be a key indicator, better than the standard wind speed (longitudinal) turbulence. Another factor is the inverter efficiency and power consumption, which is not negligible. Finally the theoretical paybacks under the 2010 Feed-in Tariffs were calculated along with estimated carbon savings. Practical application: Renewables such as wind turbines are increasingly being designed and installed to help achieve lower carbon buildings. The output of micro turbines, however, can be disappointing due to lateral turbulence and inverter consumptions. These factors are explained so that designers can be aware and assess the likely outputs more accurately. Symbols ¼ air density (kg/ms 1 ) A ¼ swept area of turbine (m 2 ) U 0 ¼ air speed (m/s) P 0 ¼ power contained by wind (W) P T ¼ power generated by turbine (W) C p ¼ coefficient of performance R ¼ gas constant T ¼ temperature (8C) P ¼ pressure (Pa) (...) v ¼ vapour (...) d ¼ dry air x ¼ mean of data SD ¼ standard deviation of data n ¼ number of data points 1 Introduction After agreeing to the Kyoto protocol, the UK government has accepted targets to lower its greenhouse gas emissions by 80% until 2050. As a result, targets were set to generate 10% of electricity demand from renewable sources by 2010, and 20% by 2020. 1 In 2003, UK Address for correspondence: Geoffrey Levermore, School of Mechanical, Aerospace and Civil Engineering, University of Manchester, Manchester, UK. E-mail: [email protected] Building Serv. Eng. Res. Technol. 32,3 (2011) pp. 245–262 ß The Chartered Institution of Building Services Engineers 2010 10.1177/0143624410389580

Upload: alex-glass

Post on 30-Jul-2015

104 views

Category:

Documents


3 download

TRANSCRIPT

Micro wind turbine performance under real weatherconditions in urban environmentA Glass BEng(Hons) and G Levermore BSc ARCS PhD DIC FCIBSE

School of Mechanical, Aerospace and Civil Engineering, University of Manchester, Manchester, UK

The aim of this article is to evaluate the performance of micro wind turbines in a built-upenvironment. For this purpose, five independent micro wind turbine systems, consisting oftwo distinctly different models, were tested and evaluated under real life conditions over aperiod of 12 months. This article provides an overview of the experimental set-up used totest the two different micro wind turbines and then goes on to present the basicbackground theory for horizontal axis micro wind turbines and the variation of coefficient ofperformance with wind speed. The wind potentials at the test site were assessed todetermine the theoretical outputs of the turbines which were compared with the measuredoutputs over a year. The measured outputs were disappointingly low. One reason for this isturbulence, for which directional turbulence (lateral turbulence) has been shown to be a keyindicator, better than the standard wind speed (longitudinal) turbulence. Another factor isthe inverter efficiency and power consumption, which is not negligible. Finally thetheoretical paybacks under the 2010 Feed-in Tariffs were calculated along with estimatedcarbon savings.Practical application: Renewables such as wind turbines are increasingly being designedand installed to help achieve lower carbon buildings. The output of micro turbines,however, can be disappointing due to lateral turbulence and inverter consumptions. Thesefactors are explained so that designers can be aware and assess the likely outputs moreaccurately.

Symbols

�¼ air density (kg/ms�1)

A¼ swept area of turbine (m2)

U0 ¼ air speed (m/s)

P0 ¼ power contained by wind (W)

PT ¼ power generated by turbine (W)

Cp ¼ coefficient of performance

R¼ gas constant

T¼ temperature (8C)P¼ pressure (Pa)

(. . .)v¼ vapour

(. . .)d¼ dry air

�x¼mean of data

SD¼ standard deviation of data

n¼ number of data points

1 Introduction

After agreeing to the Kyoto protocol, the UKgovernment has accepted targets to lower itsgreenhouse gas emissions by 80% until 2050.As a result, targets were set to generate 10%of electricity demand from renewable sourcesby 2010, and 20% by 2020.1 In 2003, UK

Address for correspondence: Geoffrey Levermore, School ofMechanical, Aerospace and Civil Engineering, University ofManchester, Manchester, UK.E-mail: [email protected]

Building Serv. Eng. Res. Technol. 32,3 (2011) pp. 245–262

� The Chartered Institution of Building Services Engineers 2010 10.1177/0143624410389580

housing was responsible for 30% of the totalenergy consumption within the country.2

In order to tackle the environmental issuesin the UK domestic sector, the UK govern-ment has issued the Code for SustainableHomes3 (CSH), and demands that all newhomes as of 2016 should be built to CSH level6 (zero-carbon) standards.4

In an attempt to investigate how muchenergy can be generated from on-site microrenewable energy systems in order to satisfyCSH level 6 requirements, Barratt PLC con-structed the EcoSmart Show Village inChorley, Lancashire, in 2006. It consisted ofseven test homes which featured 2006 energy-efficiency and renewable energy technologies,including micro wind turbines. At that timethere were only few investigations into microwind turbines in the urban environment, forexample by Clausen et al.,5 who saw potentialfor micro wind but concluded that technologyhad not reach the maturity of larger turbines.Some of the problems associated with windgenerating in urban environments had beenexplored in greater detail, in particular theeffect of turbulence in urban canyons.Eliasson et al.6 measured counter-rotating

vortices within the canyons, wind shearalong canyon edges and high degrees ofturbulence even at low wind speeds. Windtunnel simulations7 showed strong evidencethat sharp flow accelerations develop aroundroof tops, causing high fluctuations in hori-zontal velocities.

To gain a better understanding about theperformance of the micro renewable energysystems, they were evaluated over a 15-monthtest period under real weather conditions.Weather conditions were monitored andrecorded using an on-site weather station.Figure 1 shows a model of the test site.

Parallel to this investigation, several otherstudies were conducted to test micro windturbines in the urban environment, includingthe Warwick wind trials and the WINEURproject. From these studies it was concludedthat urban wind turbines faced several prob-lems, such as turbulence, which was found toreduce output by 15–30%.8 It was furthershown that the capacity factor was onlyaround 4–6.4%, compared to around 10%for rural sites. The WINEUR project specif-ically suggested9 minimum requirements tomake urban wind generation viable, including

1b

2b

2a

1a

WS

2c

W N

S E

Figure 1 Photograph showing a model of the test site, where WS refers to weather station

246 Micro wind turbine performance

average wind speeds above 5.5m/s, the tur-bine to be mounted on a building 50% higherthan surroundings and at a hub height at least30% greater than building height. This is alsoconfirmed by a CFD (computational fluiddynamics) analysis conducted by Heathet al.,10 showing that for a typical urbanlayout of buildings a hub height of at least50% above building height is required tocapture wind that is not significantly affectedby surrounding buildings. Further studieshave been conducted to show the viability ofurban micro wind turbines. Financial pay-back estimates range from 170 to 240 years11

for a range of wind data from Turkey, to30–90 years12 in the UK using a model thataccounts for wind shear and terrain correc-tion. A different approach to life-cycle anal-ysis was taken by Allen et al., who calculatedthe energy payback to be 9 years using amicro wind turbine system model includinginverter.

In November 2008 the Planning andEnergy Act13 set out a series of requirementsfor the UK Government to meet its commit-ments to combat climate change, in particularby encouraging the use of renewable energysystems to generate power. As a result of thePlanning and Energy Act, renewable energyFeed-in tariffs14 (FIT) have been introducedin April 2010, which are incentives forinstalling renewable energy systems such asWind Turbines.

1.1 Micro wind turbines at EcoSmart

show village

The experimental set-up consisted of fivemicro wind turbines. All turbines weremounted on the roof edges of test homeswithin the EcoSmart Show Village. Theturbines were installed with around 1.5–2mclearance from the roof top, which is similarto any private micro wind turbine arrange-ment. The effective hub height of the turbinesis around 10m. The weather station used to

record wind data was mounted in a similarposition on the roof of one of the test homes.

Two of the five turbines were type 1turbines, a 1 kW rated 3-blade turbine. Theother three turbines were type 2 turbines, a0.4 kW rated 5-blade turbine. Specificationsfor both turbines are summarised in Table 1below. The turbines were used in conjunctionwith an inverter, which had similar powerratings to the turbines.

In addition to the parameters in Table 1,power curves, which show the variation ofenergy generation for different wind speeds,have also been supplied by the manufacturer.These are shown in Figure 2 for both turbine1 and turbine 2. The cut-in speed, dependingon the required start-up torque,15 is 3m/s forboth turbines.

2 Wind turbine theory

2.2 Power coefficient

Assuming the turbine is constantly point-ing into the wind, linear momentum theorystates that the power of the wind movingthrough the turbine rotor is given byEquation (1):

P0 ¼1

2�AU3

0 ð1Þ

However, the power that can be generatedby the rotor differs from that contained by thewind. This is shown by simple observation

Table 1 Wind turbine specifications

Turbine 1 Turbine 2

Diameter (m) 1.75 1.1Area (m2) 2.4 0.95Rated power (kWh) 1.0 0.4No. turbine blades 3 5Cut-in speed (m/s) 3.0 2.5Cut-off speed (m/s) 12.5 16Capital Cost 2006 £1500 £2250Warranty (yrs) 10 1

A Glass and G Levermore 247

that the air is still moving away from the rotorafter it has passed through it. This means thatthere is still some energy left in the air,allowing it to carry on moving. Hence,another term needs to be introduced to theabove equation. This term is called the poweror performance coefficient (Cp) and essen-tially determines the efficiency at whichenergy is extracted from the wind. Thepower generated by the turbine is thereforegiven by Equation (2):

PT ¼ CpP0 ð2Þ

This Cp value typically varies with windspeed and is different for each turbine design.According to Betz’ law, the Cp may achieve amaximum value of 0.59, assuming perfectlyefficient machinery.

The power coefficient, Cp, can be deducedfrom the power curve of the wind turbine.Equations (1) and (2) can be used to relate Cp

to the wind speed, as shown in Equation (3):

CP ¼2PT

�AU30

ð3Þ

where A is the rotor area stated in Table 1The power values can be determined from

the power curves shown in Figure 2, where

the wind speed U0 acts as a control variable.Results for CP variation with wind speed areshown in Figure 3.

Turbine 1 shows a fairly consistent CP

value around 0.4, which only shows someslight variations between 3m/s and 7m/s.Turbine 2 on the other hand shows a consid-erably higher CP value for 3m/s to 7m/s,beyond which it begins to fall off continu-ously, until reaching a value of around 0.2 atthe cut-off speed.

This difference in CP variation is largely aresult of the number of blades of the differentturbine models. The CP value mainly dependson the tip-speed ratio of the turbine blades,which is the ratio of rotational speed overwind speed. If the blades move too slowly incomparison to wind speed, a large part ofthe wind will pass through the turbineblades without losing any of its energy.If on the other hand blades move too fast,then the turbulent air from one blade willaffect the next blade, reducing aerodynamicefficiency. The tip-speed ratio is a functionof the number of blades. Hence a turbine witha large number of blades, such as turbine 2,will work best at low wind speeds, while aturbine with fewer blades, such as turbine 1,will perform better at higher wind speeds.However, the controller of the turbines as well

1200

1000

800

600

400Out

put (

W)

200

00 1 2 3 4 5 6 7 8

Wind speed (m/s)9 10 11 12 13 14 15 16 17 18

Turbine 1

Turbine 2

Figure 2 Power curve for 1 kW rated turbine 1 and 400W rated turbine 2

248 Micro wind turbine performance

as blade design16 also plays a part in deter-mining the CP value, as the actual tip-speed ratio may vary depending on thesefactors.

2.3 Wind potentials at test site

Before finding the theoretical output of thewind turbines, the wind potential at the testsite must first be established. Weather data,including wind speed, wind direction, atmo-spheric pressure and humidity have beenrecorded directly at the site at 10-min inter-vals in accordance with BS EN 61400-21:2002. After testing was completed, therevised BS 61400-21:2008 was released, whichrecommends using measurements at 1-minintervals. While this would have vastlyimproved the analysis it could not be antic-ipated at the outset of the research in 2006.Data were acquired using the Davis VantagePro 2 Plus weather station, which is acombined package including an anemometer,wind vane, solar panel, temperature andhumidity sensors. The accuracy of the ane-mometer and wind vane is given by themanufacturer as 2% for wind speed and to78 for wind direction. However it must benoted that the rather compact mountingarrangement is likely to result in more

significant inaccuracies due to wind distortionby the mounting pole. Additionally, theweather station will experience different tur-bulence patters to those experienced by tur-bines 2a, 2b and 2c, which is a result of thedifferent roof orientation. The wind data forthe test site is as shown in Table 2, derivedfrom wind software.17

Figure 4 provides a monthly breakdown ofthe recorded wind speeds, giving mean, dailyhigh, daily low, as well as maximum andminimum wind speeds recorded for eachparticular month. Figure 5 provides thewind speed frequency distribution.Interestingly, the most common wind speedthat was recorded at the test site is 0m/s. Thisleads to suggest that the anemometerresponse is poor at very low wind speeds.This has a negligible effect on the presentresearch, as the affected range of wind

Cp

0.6

0.5

0.4

0.3

0.2

0.1

0

0 1 2 3 4 5 6 7 8Wind speed (m/s)

9 10 11 12 13 14 15 16 17

Turbine 1

Turbine 2

Figure 3 CP Variation with wind speed

Table 2 Results from weather data analysis

Variable Value

Mean wind speed (m/s) 3.1Min. wind speed (m/s) 0Max. wind speed (m/s) 18.3Hours of peak wind speed 22Mean power density (W/m2) 53

A Glass and G Levermore 249

speeds is well below the turbine cut-in speedof 3m/s.

The wind rose plot in Figure 6 appears toshow some evidence of a prevailing winddirection from the south and west directions.However, there were buildings situated to thesouth, to the north and at some distance tothe east of the weather station. All buildingsare of approximately the same height.

2.4 Density adjustment

Air density was not measured directly. As itforms a part of the wind power Equation (1),it must be found using atmospheric pressure,humidity and temperature readings.

The ideal gas law can be combined with themolecular density relationship to formEquation (4):

� ¼P

RTð4Þ

However, when determining the density ofair, the water vapour contained by the airmust also be considered. This essentiallyforms a mix of two gasses, dry air andvapour. Hence, the density of air can beexpressed by Equation (5):

� ¼Pd

RdT

� �þ

Pv

RvT

� �ð5Þ

Fre

quen

cy

12000

10000

8000

6000

4000

2000

00 1 2 3 4 5 6 7 8 9

m/s10 11 12 13 14 15 16 17 18

Figure 5 Frequency distribution of wind speeds recorded at the test site

20

15

10

Ave

rage

val

ue (

m/s

)

5

0J F M A M J J A S O N D A

Max

Daily highMean

Daily lowMin

Figure 4 Monthly variation of wind speeds (Windographer)

250 Micro wind turbine performance

The pressure values can then be determinedusing relative humidity (RH), which is definedas the ratio (expressed as a percentage) of theactual vapour pressure to the saturationvapour pressure at a given temperature.18

2.5 Theoretical energy generation

Having established the wind potentials andwind turbine properties, it is now possible toestimate the theoretical energy generation ofthe turbines over the 12-month test period.

To account for the variation in powercoefficient CP, the method outlined previ-ously has been used to find average CP valuesfor different wind speed bands. Results areshown in Table 3.

Using the adjusted CP values and Equation(2), the theoretical output was calculatedusing the following steps:

(1) Calculate air density for each interval ofweather data

(2) Calculate the power output of bothturbines for each interval of weatherdata

(3) Using the power output, find the totalannual energy generation for bothturbines

Table 4 provides a summary of the theo-retical total annual energy output, as well astheoretical average daily output for bothturbines.

As expected, the estimated output of thelarger turbine (turbine 1) is considerablygreater than the estimated output of the

90°

45°

135°

180°

225°

270°

315°

Figure 6 Wind rose plot of wind speeds recorded at the test site

Table 3 Summary of Cp values at given wind speeds bands

Cp Turbine 1 Turbine 2

Cut-in – 4m/s 0.21 0.544m/s to 7m/s 0.34 0.487m/s to 10m/s 0.36 0.3610m/s – Cut-off 0.35 0.17

Table 4 Estimated theoretical wind energy generation

Manually calculated estimate

Annual output (kWh) Avg. daily output (kWh)

Turbine 1 203 0.56Turbine 2 100 0.27

A Glass and G Levermore 251

smaller turbine (turbine 2). However, thisdifference is disproportionate to the ratedpower of the systems. While turbine 2 hasonly 40% of the power rating and swept areaof turbine 1, its high efficiency at low windspeeds means that in theory it is able togenerate around 50% of the energy that isbeing generated by the larger turbine 1.

This theoretical output does not accountfor any losses that may occur due to turbu-lence, or while the turbine is turning into thewind.

2.6 Inverters

Nearly all small-scale wind turbine systemsuse a permanent-magnet generator, theoutput of which is rectified to give a DCvoltage which varies with speed. Even in thelarger sizes the trend is toward permanent-magnet machines. This is because, comparedto induction motors, they are more efficientand effective over a wider speed range, andthis is especially so on the smaller scale.Inverters are required to change the DCpower generated by the generator of theturbine to AC power that can either beexported to the grid, or be used directly byappliances.

In its simplest form, an inverter uses atransformer and a switch on the primary coilto allow current to flow in opposite direc-tions, causing the induction of alternatingcurrent in the secondary coil.

The switching mechanism in inverters,which is required to change the direction ofDC current, is called ‘commutation’. This canbe controlled in two ways, by self-commuta-tion or forced commutation. The main dif-ference is that forced commutation, or line/network-commutation, allows the switch tocontrol the ‘on’ setting by using a device suchas a thyristor, while the ‘off’ setting iscontrolled by a supplementary circuit.19

A self-commutated inverter on the otherhand can control both ‘on’ and ‘off’ settings.With modern semi-conductor switching

devices, such as IGBT (Insulated GateBipolar Transistor) or MOSFET (MetalOxide Semiconductor Field EffectTransistor), high switching frequenciesexceeding several kHz are reached, whichmakes it much easier to filter harmonics,resulting in low network disturbances.20 Thisproperty makes self-commutated invertersmore applicable to small-scale, grid-con-nected renewable energy systems such as PVand Micro Wind Turbines. They can either bevoltage commutated or current commutated,meaning the switching is controlled by eithervoltage or current levels. A survey has shownthat practically all inverters used for peakloads of �1 kWh are self-commutated, volt-age type inverters.9

Any inverter, whatever its type, requires alow-voltage control system, and most modernsystems will employ a microprocessor. Thepower needed to drive the electronics isusually obtained from the AC mains bystepping down and rectifying using furtherdevices to give a stabilised low-voltage powersupply. It is unlikely that this can be donewithout consuming at least 5W, so that, evenif the inverter is not switching (i.e. is notpassing power into the mains) it will consumeabout 120Wh per day if it is left connectedand operational. If the inverter starts toswitch, then losses occur in each switchingoperation additional to those already men-tioned. It is impossible to generalise but somefeel that if operating at its rated output, theinverter is unlikely to be more than about90% efficient, with most of the losses beingattributable to switching. Hence it is quitepossible that the DC link needs to input apower of about 2% of the inverter ratingbefore any measurably significant power isfed into the mains.

3 Measured output

Before the energy generation of the windturbine systems can be determined accurately,

252 Micro wind turbine performance

the energy consumption of the inverters mustbe considered. The electricity meters installedat the EcoSmart show village were designedto measure electricity flowing both ways,hence recording both power consumptionand power generation. In order to determinethe energy consumption of the inverters,3 days were analysed, which were known tohave extremely low wind speeds. After ana-lysing the weather data, the 8th, 19th and 21stof July 2007 were found to have wind speedsconsistently below 2.5m/s, which is the cut-inspeed of the smaller turbine (turbine 2). Thetotal daily energy readings for the windturbine meters on those particular days arepresented in Table 5.

The values in Table 5 were then used tofind the annual inverter energy consumptionby multiplying the daily value by the numberof days during which the turbines wereoperating, in order to extrapolate the resultsover the entire test period. Table 6 shows themeasured output as well as the estimatedannual power consumption of the inverterunit, and recorded system downtime.

Table 6 shows that the energy output of thewind turbines over the 12-month periodbetween 24/10/2006 and 27/10/2007 wasmuch lower than the expected theoreticalvalue shown in Table 4. Only two systemsshowed a significant output, one turbine 1system which generated 36.1 kWh and oneturbine 2 system which generated 38.0 kWhannually. However, when considering theinverter energy consumption, all systemsinstalled at the EcoSmart show village

showed a negative net output, that is, inevery case the inverter consumed more energythan the wind turbine was able to generate.All systems experienced considerable down-time where the turbines were non-operational,ranging between 10 and 66 days.

4 Discussion

When compared to theoretical energy gener-ation, the values for measured energy gener-ation seem very disappointing. While onlytwo turbines were able to generate a signifi-cant amount of energy, none of the fiveturbines were able to generate a positive netoutput including inverter energy consump-tion. The underperformance of the microwind turbine systems in general can be largelyattributed to two factors; turbulence in urbanenvironment and the use of inverters. In somecases, such as for turbine 2c, there can beadditional effects from a blocked wind flowpath, in this case from the western direction.With reference to Table 6 and Figure 1,turbine 2c has a significantly lower outputthan turbine 2b, which does not suffer from ablocked wind flow path. However, in com-parison to the other turbines the lack ofperformance of turbine 2c does not stand out.

4.1 Inability to deal with turbulence

One main problem that was observedduring the operation of all wind turbines istheir inability to adequately deal withturbulence.

Table 5 Inverter energy consumption on days with extremelylow wind speeds

Daily Inverter consumption (Wh)

System 08/07/2007 19/07/2007 21/07/2007 Average

Turbine 1a 169.5 170.5 170.0 170.0Turbine 1b 181.0 180.5 180.5 180.7Turbine 2a 159.5 159.5 159.5 159.5Turbine 2b 119.5 120.5 118.0 119.3Turbine 2c 122.5 130.0 130.0 127.5

Table 6 Summary of wind turbine performance over 12 months

System Energygeneration(kWh)

Inverterconsumption(kWh)

Netoutput(kWh)

Downtime

Turbine 1a 36.1 55.9 �19.8 36 daysTurbine 1b 3.9 64.1 �60.2 10 daysTurbine 2a 1.8 47.7 �45.9 66 daysTurbine 2b 38.0 41.5 �3.5 18 daysTurbine 2c 5.6 41.1 �35.5 43 days

A Glass and G Levermore 253

Turbulence is inevitably created around thesharp edges of the roof of the buildings. Thisdisrupts the smooth airflow over the turbineblades by causing sharp and frequent changesin both wind speed and wind direction. Theturning mechanism of the wind turbines isrelatively sensitive with a large wind vane,causing the turbines to react promptly to achange in wind direction. However, while theturbine changes its position, the airflow isdisrupted even further for some time, causingthe blades to slow down dramatically evenwhen there is an adequate amount of wind.

Further analysis was conducted using theweather data that was recorded during thetrial period. The weather data was recorded at10-min intervals. However, the logging soft-ware of the weather station received samplesevery second and then averaged the valuesover the recording interval. While the intervalof 10min is deemed to be accurate enough forwind data (BS EN 61400-21:2002), it isslightly too long for measuring turbulence.Other research21 confirms that the samplingperiod has a significant effect on measuredwind speed frequencies. The 10-min intervallimits the accuracy of the assessment, but willstill provide valuable trends and indications.

The standard deviation of wind data can beused as a measure for turbulence, whereturbulence must be considered in threeplanes, longitudinal, lateral and vertical.12

To find longitudinal turbulence, the hourlystandard deviation of measured wind speedwas found. An indication for lateral turbu-lence was gained by analysing measured winddirection. Wind direction was measured indegrees, ranging from 08 to 3608. Differencein wind direction was assessed by subtraction,but when the wind direction went from say3508 to 108, an algorithm was written to givethis as a difference of 20 as opposed to theunlikely value of 340. The algorithm pre-vented changes of greater than 180 which wasconsidered unlikely over the period of 10minsampling time. The standard deviation of

differences in wind direction was calculatedfor every hour using the 10-min intervalreadings. No measurements were availablefor vertical wind speed.

For micro wind turbine performance, it isexpected that lateral turbulence will have thegreatest effect. Both longitudinal and verticalturbulence will effectively cause a variation ofthe wind angle of incidence on the turbineblades, resulting in less efficient aerodynam-ics, hence reduced turbine efficiencies. Lateralturbulence on the other hand will cause theentire turbine to turn, attempting to redirectinto the changed wind direction. It has beenobserved that turbine blades almost come to astandstill while the turbine redirects itself, sothat during lateral turbulence very littleenergy generation is possible.

Turbines 1a and 2b have shown someenergy output, so these turbines are used toidentify the effect of turbulence. Interestingly,both turbines showed very similar generationpatterns. Between April and June 2007 sixtime periods where identified independentlywhere noticeable energy generation was mea-sured. Five out of the six generating periods,ranging from 12 h to approximately 2 dayseach, coincided exactly. The differencesbetween theoretical and actual AC generationover these periods are shown in Table 7.Average measured inverter consumptionshown in Table 5 is used for calculations.

Having established that turbines 1a and 2btend to generate energy at the same time, andhaving pin-pointed these times, the energygeneration can now be correlated to turbu-lence estimates based on weather data.

When looking at Figures 7 and 8, showinglateral and longitudinal turbulence through-out the coinciding generating periods ofturbines 1a and 2b, it becomes apparent thatlateral turbulence has a greater effect onenergy generation. Lateral turbulence wasmeasured to be noticeably lower duringperiods of energy generation compared toperiods where no generation was measured.

254 Micro wind turbine performance

23/04/07 01/05/07 07/05/07 14/05/07 21/05/07

1

Sta

ndar

d de

viat

ion

2 3 4 52.5

2

1.5

1

0.5

0

Figure 8 Hourly standard deviation of wind speed (longitudinal turbulence) during generating times of both turbines over a 4-weeksample period in 2007

1120

100

80

60

Sta

ndar

d de

viat

ion

40

20

023/04/07 01/05/07 07/05/07 14/05/07 21/05/07

2 3 4 5

Figure 7 Hourly standard deviation of wind direction differences (lateral turbulence) during generating times of both turbines over a4-week sample period in 2007

Table 7 Turbine generation and inverter losses during coinciding generating periods

Case Duration(hours)

Turbine TheoreticalDC (kWh)

Total inverterlosses (kWh)

TheoreticalAC (kWh)

MeasuredAC (kWh)

% oftheoretical

1 28 1a 0.761 0.511 0.250 0.214 862b 1.475 0.571 0.904 0.749 83

2 59 1a 1.591 1.368 0.223 0.215 962b 2.631 1.503 1.128 0.431 38

3 42 1a 2.973 1.309 1.664 1.058 642b 6.132 1.446 4.687 2.224 47

4 13 1a 0.932 0.325 0.607 0.248 412b 2.112 0.352 1.760 1.296 74

5 38 1a 2.051 0.892 1.159 0.673 582b 4.586 0.989 3.597 2.492 69

A Glass and G Levermore 255

Periods where no generation was measuredbut lateral turbulence was low generallycoincided with periods of low wind speeds.

Figure 8 on the other hand does not appearto show any direct correlation between lowlongitudinal turbulence and turbine genera-tion. The average measured standard devia-tion of wind speed at a height ofapproximately 10m above the ground is0.49m/s. Turbulence intensity, which is ratioof standard deviation to the mean value22 hasalso been calculated for the entire data set. Thelongitudinal turbulence intensity over 12months was found to be 0.52. For comparison,the longitudinal turbulence intensity in rural,unsheltered areas at a height of 10m above theground is around 0.18.23

While the average standard deviation ofnon-generating periods is 0.44m/s, the aver-age standard deviation during generatingperiods is actually higher, at 0.59m/s. It canbe deduced that longitudinal turbulence,meaning frequent changes in wind speed,does not appear to show a significantimpact on turbine performance.

To investigate this further, Figures 9 and10 show plots of lateral turbulence for periodsof energy generation (cases 1 to 5 combined),and periods of no generation (combined‘gaps’ between cases 1 and 5) respectively.

For this analysis all wind speeds below thecut-in speed of the turbine 2 (2.5m/s) havebeen neglected to avoid any skewing of thedata, as the wind was found to changenaturally more frequently at lower windspeeds. The mean of the standard deviationsof wind direction during periods of energygeneration is 6.18� 8.18, while the mean ofthe standard deviations of wind direction ofspeeds above 2.5m/s during non-generatingperiods is 10.98� 11.58. To compare thesevalues, the t-test is used. This statisticalfunction is able to compare two means inrelation to the variation by finding the stan-dard deviation of the difference. The expres-sion used to calculate the t-value (t) is given inEquation (6)24:

t ¼�x1 � �x2ffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiSD2

1

n1þ

SD22

n2

� �r ð6Þ

For the two datasets shown in Figures 9and 10 the t-value is 5.36. This is a very hight-value, meaning that the difference betweenthe two means is statistically very significant,well beyond a 0.0001 significance. In relationto the wind turbines that were tested thisindicates that energy generation is highly

120

01 63 125 187 249 311 373 435 497 559 621 683 745 807 869 931 993 1055 1117 1179 1241 1303

No. of samples

20

40

60

Sta

ndar

d de

viat

ion 80

100

Figure 9 Combined hourly standard deviation of wind direction differences (lateral turbulence) for periods of energy generationonly, over a 4-week sample period in 2007

256 Micro wind turbine performance

dependent on the amount of lateralturbulence.

The turbulence experienced at the locationof the turbines may even be enhanced by theirmounting position relative to the building,and by the building geometry that surroundsthem. The weather station is mounted in asimilar position as turbines 1a and 1b, hencethis analysis is very representative for thosetwo cases. For all turbines of type 2, however,the turbulence profile will be very different.Nonetheless it can still be expected to have asimilarly significant effect.

If micro wind turbines are to become aviable option for the urban environment inthe future, the technology must be improveddramatically to overcome this problem.Vertical axis turbines25 could provide apotential solution if their efficiency can beimproved. It may be possible to reduce theeffect of turbulence by installing a fixedturbine, which faces into the direction ofprevailing wind, or by channelling the airflowinto the turbine, using building geometry orotherwise.

4.2 Inverter inefficiencies

The inverter of the wind turbine is anothersource of inefficiency. As explained earlier,inverters require a certain amount of power

for control circuits as well as internal relayswitching. In this case, the power is takenfrom the AC side of the inverter, that is, thepower grid. In addition to this, inverterefficiencies generally drop off under partialinverter loads.26

Figure 11 shows that the efficiency dropsoff dramatically when the inverter load isbelow 10%. For the case of the wind turbines,this would include steady wind speeds below5m/s, assuming that the inverter has a ratedpower equivalent to the wind turbines.During the 1-year period, 85% of all windspeeds that were measured at the EcoSmartshow village were below 5m/s.

Finding the appropriate size of an inverterwhich is to be used in conjunction with windturbines can be difficult, as the turbines havelarge power output range, and the frequencyof low power generation is much higher thanthe frequency of high power generation. Onepossible way of avoiding this problem wouldbe to use an onsite DC battery storage systeminstead of converting the electricity to an ACoutput. If AC output is required, then the useof two linked inverters with different powerratings might be considered, where oneinverter is used for small loads and thesecond one comes during periods of highloads, that is, high wind speeds. The downside

120

01 129 257 385 513 641 769 897 1025 1153 1281 1409 1537 1665 1793 1921 2049 2177 2305 2433 256112689

No. of samples

20

40

60

Sta

ndar

d de

viat

ion

80

100

Figure 10 Combined hourly standard deviation of wind direction differences (lateral turbulence) for periods of zero energygeneration only, over a 4-week sample period in 2007

A Glass and G Levermore 257

to this approach is the increased capital costand additional power consumption for asecond inverter control circuit.

4.3 Reliability

System reliability is another importantcause for concern. Table 6 has already pro-vided an overview of system downtime, whichwas found to range between 3% and 18%over the 1-year period. These downtimes werelargely caused by wind speeds that exceededthe rated ‘cut-off’ speeds of the turbines.While average wind speeds over 10min thatexceed the cut-off speed were rarely mea-sured, gust speeds may well have been higheron several occasions. It is also possible thatthe power curve supplied by the manufactureris not 100% accurate, as was found in otherresearch projects such as the Warwick windtrials.8 The excessive wind speeds caused theturbines to shut down and, on many occa-sions, they failed to automatically start upagain. The turbines had to be reset manually.

A major issue for concern is the systems’apparent lack of ability to deal with extreme

wind speeds that far exceed the rated cut-offspeeds. On one occasion during extremewinds in January 2007, where average windspeeds above 18m/s were measured, a bladefrom one of the turbines detached and waslater recovered far from the installation site.Figure 12 shows a photograph of the blade asit was recovered.

While this is unlikely to have occurredduring standard operation, it is possible thatthe shut-down mechanism was responsible forthis. When the wind speeds exceed the ratedcut-off speeds of the turbine, a brakingmechanism is applied, which forces theblades to come to a standstill. However, atvery high wind speeds the transient loadsduring turbine shut-down and the aerody-namic forces acting on a stationary bladeapply an extreme amount of stress to the rootof the blades. As blades are designed to belight and thin to maximise turbine efficiency,safety margins in blade design may not haveanticipated and accounted for the forcesunder such extreme wind speeds as experi-enced on the 18 January 2007.

100

80

60

40

20

00.0 0.2 0.4 0.6 1.0 1.2 1.4 1.60.8

Ppv /Pinv,rated

Inve

rter

effi

cien

cy (

%)

Figure 11 Efficiency variation with power load of typical inverter14

258 Micro wind turbine performance

4.4 Payback rate and carbon savings

Feed-in Tariffs for electricity generatingrenewable energy systems, such as wind tur-bines, will be introduced in April 2010. TheFITs are fixed rates at which energy generatedby renewable energy systems will be valued,and do not require electricity to be ‘exported’to the national grid. However, if electricity isexported, an additional 3 p/kWh will be paidon top of the nominal tariff. The model of theFIT’s is such that initial tariffs are availablefor installations commissioned in or afterApril 2010. For wind turbines, the tariff willbe paid over a 20-year period. For anyinstallation commissioned after April 2012,there will be a ‘degression’ value for theinitial tariff, which is a reduction of thetariff to be received over the entire period.

The degression is linked to inflation, andTable 8 provides an extract6 of tariffs forwind turbine systems until April 2018.

As none of the micro wind turbine systemswere able to generate a positive electricalenergy output, the actual payback periodsand carbon offset were not calculated.Instead, the theoretical values will be consid-ered for this system. Using the theoreticaloutput from manual calculations and theWindographer estimates, the payback ratesand carbon offset for the two systems areshown in Table 9.

For the purpose of these calculations thefollowing has been assumed:

– The systems are used with either a DCbattery and charge controller, eliminating

Figure 12 Damaged turbine blade which had detached during the storm on 18 January 2007

Table 8 2010 Feed-in tariffs until April 2018 for wind turbines

Size Annual tariff (pence/kWh), starting in

April 2010 April 2011 April 2012 April 2013 April 2014 April 2015 April 2016 April 2017

Wind �1.5 kW 34.5 34.5 32.6 30.8 29.1 27.5 26 24.6Wind 41.5–15 kW 26.7 26.7 25.5 24.3 23.2 22.2 21.2 20.2Wind 415–100 kW 24.1 24.1 23 21.9 20.9 20 19.1 18.2Wind 4100–500 kW 18.8 18.8 18.8 18.8 18.8 18.8 18.8 18.8Wind 4500 kW–1.5MW 9.4 9.4 9.4 9.4 9.4 9.4 9.4 9.4Wind 41.5MW–5MW 4.5 4.5 4.5 4.5 4.5 4.5 4.5 4.5

A Glass and G Levermore 259

the need for an inverter, or a perfectlyefficient inverter that consumes no energy.

– All of the energy that is generated isconsumed on-site.

– 2010 feed-in tariffs are applicable for theseinstallations, providing a maximum possi-ble contribution to annual savings.

– 1 kWh of electricity taken from power gridequates to approximately 1 kg CO2 ofcarbon.27

It must be emphasised that this is a ‘best-case’ scenario of purely theoretical nature,and not a realistic representation of what wasfound during the 12-month experimentalperiod. All systems from the experimentalset-up consistently showed negative annualelectricity generation, hence a negative carbonoffset. Any potential annual savings andrelated payback periods would depend onwhat the system actually generates, and aregreatly improved by Feed-in Tariffs.

5 Conclusion

– Theoretical wind generation for this partic-ular site is around 203 kWh and 100 kWhannually for turbine 1 and turbine 2 respec-tively, while measured generation wasbelow 40 kWh for both turbines.

– Inverters consume a considerable amountof energy, measured to be between 41 kWhand 64 kWh annually.

– Considering inverter consumption, the netenergy output of all systems was measuredto be negative, that is, in all cases moreenergy was consumed than generated.

– The use of an inverter inevitably causes highefficiency losses for generation at low windspeeds.

– Turbulence, in particular lateral turbulencebrought about by changes in wind direction,is a major concern for the performance ofmicro wind turbines in an urban environ-ment. This needs to be overcome byimproved technology or buildingintegration.

– At present, micro wind turbines can posesevere safety problems by detaching bladesduring extreme wind speeds.

– During the experiment it was found that allfive systems showed an overall negativecarbon offset over a 12-month period.

– During the experiment no financial savingswere achieved as net generation was nega-tive. In theory, if the turbines were to workat design efficiency and without theobserved lateral turbulence and a perfectlyefficient inverter was used, optimistic pay-back periods for the turbine 1 model are 17–18 years, and 51–55 years for turbine 2models assuming 2010 FITs.

Acknowledgments

The work described in this article was supported

by a grant from Barratt Development PLC, whoalso provided equipment and testing facilities, and

an EngD grant from the Engineering and Physical

Sciences Research Council. Thanks are given to

Dr Tony Sung who initiated and supervised the

project for 2 years. Thanks are also given to

Dr Alan Williamson Senior Research Fellow atthe University of Manchester for advice on the

inverter theory and to Emeritus Professor Patrick

Laycott University of Manchester, who advised on

aspects of the statistical data analysis.

Table 9 theoretical simple payback rate and carbon offset ofmicro wind turbine systems

Turbine 1 Turbine 2

Theoretical generation (kWh) 203 100April 2010 to March 2012 FIT £70.40 £34.50April 2012 to March 2013 FIT £66.18 £32.60April 2013 to March 2014 FIT £62.52 £30.80Annual savings £20.30 £10.00Theoretical payback rate (yrs) 16.5–18.1 50.6–55.1Carbon offset (kgCO2) 203 100

260 Micro wind turbine performance

References

1 DTI Energy Group. Our energy future –creating a low-carbon community. UKGovernment Energy White paper, 2003.

2 DTI Energy Group. UK energy end users,2004.

3 Department of Communities and LocalGovernment. Code for sustainable homes,2007. Available at: http://www.planningportal.gov.uk/uploads/code_for_sust_homes.pdf (accessed January 2010).

4 Department of Communities and LocalGovernment. What standards may be in 2013& 2016, 2007. Available at: http://www.communities.gov.uk/documents/planningandbuilding/doc/br-energyefficiency.doc (accessedJanuary 2010).

5 Clausen PD, Wood DH. Recent advances insmall wind turbine technology small windturbines. Wind Engineering 2000; 24(3):189–201.

6 Eliasson I, Offerle B, Grimmond CSB,Lindqvist S. Wind fields and turbulencestatistics in an urban street canyon.Atmospheric Environment 2006; 40: 1–16.

7 Kastner-Klein P, Fedorovich E, Rotach MW.A wind tunnel study of organised and turbu-lent air motions in urban street canyons.Journal of Wind Engineering and IndustrialAerodynamics 2001; 89: 849–861.

8 Encraft Warwick Wind Trials Final Report.Available at: http://www.warwickwindtrials.org.uk (accessed 8 July 2010).

9 Cace J, Horst E, Syngellakis K, Niel M,Clement P, Heppener R, Peirano E. Urbanwind turbines: Guidelines for small windturbines in the built environment. Available at:www.urbanwind.org (accessed 8 July 2010).

10 Heath MA, Walshe JD, Watson SJ.Estimating the potential yield of smallbuilding-mounted wind turbines. Wind Energy2007; 10(3): 271–287.

11 Celik AN, Muneer T, Clarke P. An investiga-tion into micro wind energy systems fortheir utilization in urban areas and their lifecycle assessment. Proceedings of theInstitution of Mechanical Engineers Part A:Journal of Power and Energy 2007; 221(8):1107–1117.

12 Bahaj AS, Myers L, James PAB. Urban energygeneration: Influence of micro-wind turbineoutput on electricity consumption in buildings.Energy and Buildings 2007; 39(2): 154–165.

13 Office of Public Sector Information. Planningand Energy Act 2008. Available at: http://www.opsi.gov.uk/acts/acts2008/pdf/ukpga_20080021_en.pdf (accessed February 2010).

14 Department of Energy & Climate Change.Feed-in Tariffs – Government’s response to2009 consultation. Available at: http://www.decc.gov.uk/Media/viewfile.ashx?FilePath¼Consultations\Renewable%20Electricity%20Financial%20Incentives\1_20100204120204_e_@@_FITsconsultationresponseandGovdecisions.pdf&filetype¼4(accessed January 2010).

15 Wood DH. A blade element estimation of thecut-in wind speed of a small turbine. WindEngineering 2001; 25: 249–255.

16 Wood DH. Dual purpose design of small windturbine blades. Wind Engineering 2004; 28(5):511–527.

17 Mistaya Engineering Inc., Windographer.Availabale at: http://www.mistaya.ca.

18 Brutsaert W. Moist air. Evaporation into theatmosphere: Theory, history and applications.Dordrecht, Holland: D. Reidel PublishingCompany, Chapter 3.1, pp. 37–42, 1991.

19 Ishikawa T (2002). Grid-connected photovol-taic power systems: survey of inverters andrelated protection equipments. ReportIEA-PVPS T5-05, Central Research Instituteof Electric Power Industry Japan.

20 Grauers A (1994). Synchronous generatorand frequency converter in wind turbineapplications: system design and efficiency.Technical Report no. 175 L, ISBN91-7032-968-0, Chalmers Universityof Technology.

21 Makkawi A, Celik AN, Muneer T. Evaluationof micro-wind turbine aerodynamics,wind speed sampling interval and its spa-tial variation. Building ServicesEngineering Research and Technology 2009;30(1): 7–14.

22 Roth M. Review of atmospheric turbulenceover cities. Quarterly Journal of the RoyalMeteorological Society 2000; 126(564):941–990.

A Glass and G Levermore 261

23 Holmes JD. Wind loading of structures.London: Spon Press, Chapter 3, pp. 46–68,2001.

24 Gravetter FJ, Wallnau LB. Essentials of sta-tistics of the behavioural sciences. Belmont CA:Thomson Wadsworth, Chapter 10,pp. 266–267, 2008.

25 Revolutionary wind turbine. Building ServicesJournal, CIBSE, June 2007.

26 Mondol J. Sizing of grid-connected photovoltaicsystems: SPIE (International Society for OpticalEngineering). 10.1117/2.1200704.0612, 2007.

27 Parliamentary Office of Science andTechnology. Carbon footprint of electricitygeneration - postnote, October 2006. Availableat: http://www.parliament.uk/documents/upload/postpn268.pdf (accessed November2009).

262 Micro wind turbine performance