Académique Documents
Professionnel Documents
Culture Documents
Department of Computer Science and Engineering CHALMERS UNIVERSITY OF TECHNOLOGY UNIVERSITY OF GOTHENBURG Gteborg, Sweden, September 2010
THESISFORTHEDEGREEOFMASTEROFSCIENCE
Controlofawindturbineequippedwithavariable rotorresistance
HCTORA.LPEZCARBALLIDO
DepartmentofEnergyandEnvironment DivisionofElectricPowerEngineering CHALMERSUNIVERSITYOFTECHNOLOGY Gteborg,Sweden,May2009
DepartmentofEnergyandEnvironment DivisionofElectricPowerEngineering CHALMERSUNIVERSITYOFTECHNOLOGY Gteborg,Sweden,May2009
Abstract
Inthisthesisthecontrolofawindturbineequippedwithaninductiongeneratorwith variable rotor resistances was investigated. Analysis, modelling and control of the induction generatorsystemwasconducted.Inparticularthe focuswasputonthereductionoftorque fluctuations,inordertoreducethestressesinthegearboxandinthemechanicalstructureas wellasreducingtheflickeremission. Different controlling methods were studied in order to find an appropriate choice. Finally,theinductionmachinewiththevariablerotorresistancecontrollerwascomparedwith thesameinductionmachinewithoutcontrollerforthewindapplication,inordertostudythe improvementofimplementingthecontroller. The first thing to be emphasized is that by utilising the controller the flicker contributioncanbereducedbetween35%60%comparedtotheuncontrolledsystem.Itwas also found that the reduction of the flicker contributions was strongly related with the turbulenceintensityinthewind.Thereductionintheflickeremissionwasstrongerformore turbulent winds. Furthermore, a reduction in the magnitude of the electrical torque componentsforthefrequenciesabove1Hzwasfound.Thosehighfrequencycomponentsare theonesthatmorecontributetothemechanicalstressesinthegearboxandstructureofthe turbine, this means that by utilising the controller there will be less tear and wear on the mechanicalpartsoftheturbine.
iii
iv
Acknowledgement
First of all, I would like to thank my supervisor at Chalmers University of Technolgy, Assoc.Prof.TorbjrnThiringerforhisencouragingandinspiringattitude.Ialsowouldliketo thanktoProf.StefanLundberg,fortheoreticalhelpanddiscussions. Finally, I would like to thank the whole department, especially to the master thesis students that were preparing their thesis at the same time as me, for a nice working atmosphereandkindtreatmentduringmystayinSweden. HctorLpez May,2009
vi
Contents
Abstract.......................................................................................................................................iii Acknowledgement........................................................................................................................v Chapter1 Introduction............................................................................................................... 1 1.1 1.2 1.3 1.4 Background.................................................................................................................... 1 Previouswork................................................................................................................ 2 Goaloftheproject ........................................................................................................ 2 Thesislayout.................................................................................................................. 2
Chapter3 Inductionmachine..................................................................................................... 9 3.1 3.2 3.3 3.4 3.5 Inductionmachineaswindturbinegenerator.............................................................. 9 Inductionmachinemodelling...................................................................................... 10 Linearizationoftheinductionmachinemodel ........................................................... 14 Inductionmachinewithextrarotorresistance........................................................... 17 Parametersofageneric2MWinductionmachine ..................................................... 20
4.1.1 4.2
vii
5.1 5.2
Chapter6 Finalspecificationsofthecontrollerbasedontheevaluations ............................. 56 6.1 6.2 6.3 Selectionofthecutofffrequencyandtheproportionalgain .................................... 56 Selectionofthedefaultexternalrotorresistance ...................................................... 59 Finalspecifications ...................................................................................................... 60
viii
ix
Chapter1 Introduction
1.1 Background
Wind power is without any doubt becoming an important energy source. Today a massive expansion is going on. However, the possible sites on land are starting to get fewer andaccordinglythefocusonsealocationsaregrowing.Herethereismuchmorespace,and notonlythat,ahigheraveragewindspeedisavailable,becausewindspeedsareaffectedby thefrictionagainsttheearthssurface,andsincethesurfacesoftheseaisverysmoothand theobstaclestothewindarefew,theaveragewindspeedishigherthanonland. Forthesereasons,offshorewindenergyisapromisingsolutionforcountrieswithhigh populationdensityandwithoutsuitablesitesonlandtobuildwindparks.Ofcourse,thereare somelimitationsfortheconstructionofoffshorewindfarmslikethatcostsarehigherthanon land,buttheenergyproductionisalsohigher. However,cablesareneededandthenonly100kmofenergytransportationispossible if50HzACisused.Sincelargescaleelectricpowerhastobetransmittedoverlongdistances, then HVDC lines are needed, and the first such park is today being built in the North Sea. Furthermore,HVDCtransmissionsystemshavebeeninsuccessfuloperationinpowersystems allovertheworldforabout40years,andtherearesomewellknownandacceptedfactsthat canbeveryusefulforoffshorewindparks.Themostimportantisthattheoffshoregridand onshoregridareisolatedbytheHVDClink,thisgivesnewpossibilitiesregardingthelocalAC systemwhichcanbeutilised. In this local ACsystem the voltage and frequency can be changed by the ACDC converterintheoffshoreside.Thisopenthepossibilityofremovingtheconverterconnected to the generator of each wind turbine, but still having variable rotor speed in the turbines thankstoacentralspeedcontrolforthewholewindfarmthatcanchangesthefrequencyof thegrid,hencethespeedofthewindturbines,whichareequippedwithinductiongenerators directlyconnectedtotheelectricalgrid.Thiswillreducethecostforthewindturbinesystem, and in addition increase the reliability. Here the system with a variable rotor resistance becomes of high interest, since it can reduce mechanical stresses on the turbine just as a variablespeedsystem. Averyinterestingissueforthissystemwithvariablerotorresistancesistowhatextent this system can be used for this application, and the selection of a suitable control. With a controllable rotor resistance it will be possible to absorb incoming power variations by changingtherotorspeedmuchfasterthanthepitchcontroller,whichwillgiveabetterpower quality.
1.2 Previouswork
The control ofanexternalrotorresistancehasbeenstudied before,forexample the
OptiSlip concept by the Danish manufacturer Vestas. But there is not too much available
literature about this subject in particular for the situation where voltage an frequency can change.In[1]aderivedcontrollawforawindturbineusingvariablerotorresistancesisfound, butinthatpaperonlytheflickercontributionofthewindturbineisstudied.
1.3 Goaloftheproject
The main purposeofthis thesisistoinvestigate howawindturbine usinga variable rotorresistancecanbecontrolledforawindturbineapplication.Forthispurposeaninduction machinewillbemodelledandlinearized.Moreover,anobjectiveistostudyhowthissystem behavesforasituationwithrealisticwindinputs.
1.4 Thesislayout
Thebodyofthethesisisorganizedasfollows: Chapter2,descriptionofcommonwindturbinessystemsandoffshorewindfarms. Chapter3,presentationoftheinductionmachinemodelanditlinearization. Chapter4,presentsthedesignofthecontroller. Chapter 5, presentation of the results for different shaft torque curves, synthetic and real data. Chapter6,describesthefinaldesignofthecontrolleranditsfinalcharacteristics. Chapter7,givesasummaryoftheconclusionthatcouldbeextractedfromtheresultsofthe evaluations. Chapter8,containstheproposedfuturework.
Chapter2 Windturbines&offshorewindparks
Thischaptergivesabriefoverviewofcommonwindturbinesystems.Italsointroduces the advantages of combine offshore wind parks with HVDC lines. The interested reader can find more information about wind turbine theory in [16], about offshore wind farms in [17] andaboutHVDClinesin[15].
2.1 Windturbines
2.1.1 Aerodynamicconversion Awindturbinesystemgetsitsinputpowerbyconvertingsomeofthekineticenergyin the wind into torque acting on the rotor blades. This process is called the aerodynamic conversion,anddependsonthewindspeed,therotorarea,thepitchangleoftheblades,and thedensityoftheair.Althoughtherearedifferenttypesofwindturbinessystems,allofthem workinasimilarway. InordertocalculatethemechanicalpowerfromawindturbinetheCp(,)curvecan beused.Withthisknowledgethemechanicalpowercanbedeterminedby (2.1)
Where Cp(,) describes how much of the available energy in the wind that can be convertedintomechanicalpowerthatdependsonthetipspeedratioandinthepitchangle oftheblades,istheairdensity,Aristheareasweptbytherotorandwisthewindspeed.A typical Cp(,) is presented in the figure 2.1, where it is possible to see the Cp() curves for differentvaluesinthepitchangleoftheblades.
0.5
0.4
Cp
0.3
0.2
0.1
10
12
14
16
18
20
Lambda
Figure2.1,typicalCp(,)curve. The wind turbine starts to generate energy when the wind speed is above Vcutin and stops when the wind speed is above Vcutoff. Figure 2.2, shows an example of how the mechanicalpowervarieswiththewindspeed.
2000
Power (kW)
1500
1000
500
10
12
14
16
18
20
thefrequencyofthegrid,admittingonlyverysmallspeedvariationfromthenominalvalue. Figure2.3showstheprincipallayoutofsuchasystem.
Gear box
IM
Soft starter
Grid
Bank capacitor
Figure2.3,fixedspeedwindturbinesystem Variablespeedsystem
Thevariablespeedwindturbinesystem[10]usesaninverterconnectedtotherotorof thegenerator.Byusingavariablerotorspeedatlowwindspeedsitispossibletooperateat idealthatresultsinmaximumCpvalue.Sovariablespeedwindturbinesmaximizetheenergy captured in weaker winds. The possibility to control the rotor speed also reduce the mechanical stresses by better torque control and reduce the power fluctuations, hence increasethepowerquality.Infigure2.4canbeseenatypicalconfigurationforthissystemthat consistsofawindturbinewithadoublyfedinductiongenerator. Disadvantages of this method are the additional cost and the power losses in the converter.Otherdisadvantageisthatextrafilteringoftheoutputcurrentisneededtoobtaina good power quality due to the harmonics that are added with the power electronic equipment.
Gear box
IM
Inverter
Grid
Figure2.4,variablespeeddoublyfedinductiongeneratorsystem
2.2 Offshorewindfarms
Theoffshorewindsector iscurrentlyboomingwith severalnew offshorewindfarms under construction. The offshore wind energy is expected to be a major contributor in renewablegeneration. Thefactsthatmotivatesthesethoughtsarethatnewrestrictionstotheonshorewind farmshaveappear,suchas,limitedamountofavailablelandordifficultiesforobtaininglocal permitsandpublicacceptance,andalsoenvironmentalaspectsandthenoisesfromthewind turbinesthatdisturbpeopleintheareaaroundthewindfarm. Withoffshorewindfarmsmostoftheseproblemsaresolved.Inoffshorewindfarms largeinstallationcapacitiesarepossible,whichminimizethenumberoffactorsthatincrease the cost of offshore wind farms to be above the onshore ones. The attraction that it has minimalenvironmentaleffectsand,windspeedsaregenerallyhigherthanonshoreandwith reduced turbulence. However, it has to be considered that offshore wind farms are without anyenvironmentalimpact,buttheseareconsiderablysmallerthaninonshorewindfarms. DevelopmentofwindturbinesisnowfocusedonturbinesizesofseveralMW.These largesizeturbinesuseconverterstoconnectthemtothegrid,controlthevaryinginputand alsoreducetheharmoniccontentoftheoutput.Iftheconvertercouldbeavoided,sincethe windturbinesareconnectedtoalocalgridthatisisolatedfromthepublicgrid;soitcanwork atvariablevoltagefrequencyandwithoutveryhighpowerqualityrequirements.Furthermore, the wind farm already has one/several HVDC converter that can be used as a central speed controlforthewholewindfarmbychangingthefrequencyofthegrid,hencethespeedofthe windturbine,whichhastheinductiongeneratordirectlyconnectedtothegrid.Then,itwillbe possible to remove the converters from all the wind turbines, and only keep the HVDC converter.Nowthewindturbinescanbeequippedwithvariableslipresistancescontrollerto improvethe powerqualityinsteadof oneconverterineachwindturbine.Thiswillleadtoa cheaperandalsorobustnesssystem. InthiscasethecontrolofthewindfarmwillbecarriedoutbytheHVDCconverterin the offshore side, that will set the frequency and the voltage that optimize energy captured from the wind. In figure 2.5 it is presented the configuration of an offshore wind farm with HVDCtransmission[14],whereeachturbineisdirectlyconnectedtotheoffshoregridandis equippedwithexternalrotorresistancescontrol.
HVDC line
Grid
Offshore platform
Onshore platform
Figure2.5,ConfigurationofanoffshorewindfarmwithHVDCtransmission.
2.3 HVDClines
Toconnecttheoffshorewindfarmswithland,theHVDClinesarethebestalternative. Thisisduetothefactthatlargescalepowertransmissionsoverlongdistancesarenotfeasible withtraditionalACtransmissionsystems,soHVDChasbeenproventobeabetteroption. Ithasbeenproven,accordingto[15],thatwhentheamountofpowertobe transmitted,andthedistancearehighenough,theHVDCtransmissionsystemischeaperthan theACsystem.AlsotheDCtransmissionsystemcauseslowertransmissionlosses. UsinganHVDCtransmissionprovidesthesystemwithmorestabilityandcontroldue tothefactthatthepowerflowcanbefastandcontrolledbytheHVDClink.Also,thegrids offshoreandonshoreareisolatedduetotheexistenceoftheHVDClinkwhichprotectsboth gridsfromfaultsintheotherside.Ifthewindturbinesareisolatedfromtheothergrid,the offshoregridwillbeabletoworkatthefrequencythatallowsthewindfarmtoobtainthe higherefficiencyfromthewindalsoifturbineswithoutconvertersareused.
2.4 Powerqualitycharacteristicsofwindturbines
Power quality characteristics of grid connected wind turbines are becoming more importanteverydayduetothedevelopmentoflargewindfarmsthatmayformasignificant partofthepowersystem.Nowadaysthepowerqualitystandardsofwindturbinesareissued by the International Electrotechnical Commission (IEC), IEC6140021: Measurement and assessmentofpowerqualitycharacteristicsofgridconnectedwindturbines,Ed1,2001[7] defined the parameters that are characteristic of the wind turbine behavior in terms of the qualityofpower.
With the development of IEC6140021, it was possible to identify the factors and characteristics with highest influence on the power quality of wind turbines and the parameters then became more adapted to their quantification, to act as normalized quality indicators.Theseparametersareusedtoestimatethepowerqualityofawindturbine. Thetypicalbehaviorofawindparkbasedoninductiongeneratorsdirectlyconnected tothegrid,deliversafairlyvariablepowertothegrid.Thispowerflowcancontributetoflicker emissions and affect the mean voltage profile. Certainly, this can be compensated by installation of reactive power compensation. The use of doublyfed induction generators or generators with fully rated frequency converters generally offers smaller fluctuations in the active power output. But the disadvantage of using power electronic converters may be a higherharmonicdistortion,increasedcostandlossesintheconverter As mentioned before, the publication of the IEC 6140021 standard enabled the determinationofsystematicparameterstocharacterizethequalityofpowerofgridconnected wind turbines. In the chapter 5 Evaluation of the controller, improvement of the power quality that is achieved with the use of the variable rotor resistances controller will be determined.Themainparameterthatwillbeusedforthispurposewillbetheflickeremission. Thisparametergivesanideaofthewindpowerfluctuationsinsteadystateoperation.Another parameter that can be used is the emission of current harmonics, but in this case the controller will have no effect on this parameter since it is not using power electronic convertersthataretheequipmentthatmostlyiscausingthecurrentharmonicsemission.
Chapter3 Inductionmachine
Inthis chapter,asuitablemodelofaninduction machinewillbepresented. Further, thelinearizationoftheobtainedmodelwillbeshown.Attheend,itwillbedemonstratedhow theinductionmachinebehavesdependingonthevalueoftherotorresistance.
3.1 Inductionmachineaswindturbinegenerator
The induction generators that are used in the wind turbine industry [12] have two maintypesofrotor:squirrelcagerotororwoundcagerotor.Thislastcanhaveslipringsthat can be connected to an external circuit. When it is desired to control the turbine, a wound rotorwithslipringshastobeused. Theslipofaninductiongeneratorisusuallyverysmall(forefficiencyreasons),butthe slipdependsontheresistanceoftherotorwindings.Thus,itispossibletoincreasetherotor resistance by increasing external resistances connected to the slip rings. The fact that the generator will increase or decrease its speed slightly if the torque varies is a very useful mechanical property; because this means less stresses in the gearbox and in the induction machine,soitwillbepossibletousesmaller(accordinglycheaper)gearboxes. Running wind turbine at variable speed has several advantages, one is that it allows the rotor to speed up while a wind gust is happening, storing the excess of energy into rotational energy until the wind dust is finished. In addition, the conversion of wind energy intoshaftenergywillincreaseifthewindturbinecanoperateatitsoptimalspeeddepending onthewindspeed.Although,thecomplexityofthevariablespeedsystemleadstoincreased costareducedreliabilityduetouseofpowerelectronicsandamorecomplicatedcontrol. But new offshore wind farms connected to land through HVDC lines, open new possibilitiesregardingthelocalACsystemwhichcanbeutilisedinordertoreducethecostfor a wind turbine system, and in addition increase the reliability. Removing the converter connected to the stator of the wind turbine is still possible to have variable rotor speed systems by changing the frequency of the offshore grid with the HVDC converter. Here is wherethesystemwithavariablerotorresistancesystembecomesofhighinterest.
3.2 Inductionmachinemodelling
In this section, the equations that are needed to make a model of an induction machinewillbepresented.Inthisprojectonlytheinductionmachinemodelwillbetakeninto accountinthewindturbinemodel,sothedrivetrainwillnotbeconsideredinthemodelof thewindturbinesystemwherethecontrollerwillbetested. The objective of this section is to transform the induction machine into a separately magnetized dc machine and presenting the mathematical model of it. The purpose of transforming the IM into a separately magnetized dc machine is to make the subsequent designofthecontrollereasier.Tomakethistransformationtwostepsareneeded,first,make the Threephase to twophase transformation, and second transform from stationary to a rotatingcoordinatesystemorientedwiththerotorfluxinxdirection. Totransformthethreephasesystemintoatwophasesystemthetransformation willbedone.Thesystemwillhavetwoperpendicularaxes,and,thatcanbeconsidered as the real and complex axes in a complex plane. The interested reader can consult Park transformation[18]forfurtherinformation. The transformation is given by the following matrix for the amplitude invariant transformation:
(3.1)
Theinversetwophasetothreephasetransformationisgivenby:
Electricaldynamics
(3.2)
Nowthatthetransformationneededtogetthetwophasesystemhasbeenshowed, the electrical equations that govern the dynamics of the stator and rotor of the induction machineinstationarycoordinatesystemswillbestatedbelow: (3.3)
(3.4)
10
(3.5) (3.6)
The equations above are the representation of dynamic model of the induction machine with a stationary coordinates ( system), where is is the applied phase stator voltage to the induction machine, is the stator current, is the rotor current, is the statorflux, istherotorflux, isthestatorresistance, istherotorresistance, isthe statorinductance, istherotorinductance, isthemagnetizinginductanceand isthe rotorspeed.Theequations3.3to3.6willbeusedtocreatethemodelinMATLAB.Fromthose equationsispossibletoderivetheequivalentelectricalcircuitoftheinductionmachine.
is
Rs
Lsl
ir
Lrl
Us
Lm
Rr
Then, substituting these terms in the equations 3.5 and 3.6 and separating the quantities into real and imaginary parts, the electrical equations of the rotor and the stator become: (3.9)
(3.10)
(3.11)
(3.12)
If we arrange these equations into the matrix form, the matrixes below will be obtained:
11
(3.13)
(3.17) (3.18)
istheelectromechanicaltorqueand
isapplied
From this equations are derived the fifth and sixth states of the statespace representationoftheinductionmachinethatare:
(3.19) (3.20)
12
Thevaluesofthestatesandoutputsofthepreviousmodelaresinusoidal.But,likeit wassaidbefore,thesevaluesarebetterrepresentedinarotatingsystemwheretheybecome dc quantities, which simplifies the design of the controller and also the analysis of the simulations. For this reason is necessary to transform the values from the stationary to a rotatingsystemperfectlyalignedwiththerotorfluxwhichisreferredtoasadqsystem. This transformation is made by multiplying the quantities with angle of the rotor flux that can be obtained like transformationcanbeexpressedlike: , where is the
(3.21) andtheinversetransformationwillbe:
(3.22)
Thesamematrixes3.21and3.22areusedtotransformthecurrentsoftherotorand thestatorfromansystem,intoadqsystem. The mathematical equations can be also transformed and solved in a rotating referenceframe.Theseequationsarepresentedbelow. (3.23)
(3.25) (3.26)
(3.24)
(3.27) (3.28)
where
isthestatorangularfrequency.
These equations will be used in the next point to linearize the induction machine model, and to obtain the transfer functions that will be used to design the external rotor resistancescontroller. Losscomponentsintheinductiongenerator
13
The losses in the induction machine mainly consist of two components, the copper lossesandtheironlosses,buthereitwillbeonlythecopperlossesconsidered.Theyoccurin thestatorandrotorwindings,soincreasingtherotorresistancewillhaveaneffectonthem. Thecopperlossesaredeterminedas: Where: arethelossesinthestator arethelossesintherotor (3.29)
3.3 Linearizationoftheinductionmachinemodel
Mostoftheexistingtheoryforcontrolsystemuselinearizedmathematicalmodelsof theprocesstocontroltheminaclosedloopmanner.Butinthiscase,thesystemisnonlinear, for this reason, it is needed to transform the nonlinear system into a linear one. Then the linearized model will be used in chapter 4 Design of the controller for the extra rotor resistancetodevelopthecontrolsystemoftheoriginalnonlinearmodel. Apossiblewaytogetacontrolsystemisthefollowing:Thefirststepistoobtainanon linear model of the system, like it was made in the previous section. Next the model is transformed into a linear one, as will be explained in this point. Later the control system is designedforthelinearmodel.Finally,thecontrollerisdevelopedusingthenonlinearmodel. Thelasttwopointswillbeexplainedinfollowingchapters. In this point it is presented how the linearization of the induction machine model describedbynonlineardifferentialequationsisperformed.Theprocedurethatwillbeusedto linearizetheequationsystemisbasedonTaylorsseriesexpansion[13]. The first step is to calculate the equilibrium point. The equilibrium points are those points where all the derivatives are simultaneously zero, and also the points which we are going to operate around with small variations. To find the equilibrium point the equation belowhastobesolved. (3.30) Oncetheequilibriumpointshavebeencalculated,themotionofthenonlinearsystem isintheneighborhoodofthenominalsystemtrajectory,thatis: (3.31)
14
TherighthandsidecanbeexpandedintoaTaylorseriesexpansion,asfollows:
wherethepartialderivativesrepresenttheJacobianmatrixesgivenby
(3.38)
15
Taking the equations 3.23 to 3.28 and rearranging the terms, the following equations to be linearizedareobtained:
(3.40)
(3.41)
(3.42)
(3.43)
(3.44)
andaccordingtoequations3.38and3.39,theAandBmatrixeslooklike:
16
Theabovepartialderivativeshavetobeevaluatedattheequilibriumpointaboutthe systemisbeinganalyzedwithsmallperturbation. Transferfunction The transfer functions of the statespace model showed before can be obtained by takingtheLaplacetransformof hence, (3.46) Then,dividingby ,giving (3.47) (3.48) thisissubstitutedfor intheoutputequation3.37giving (3.49) Thedefinitionoftransferfunctioncanbefoundastheratiooftheoutputtotheinput ofasystem, (3.45)
3.4 Inductionmachinewithextrarotorresistance
17
The idea of this project is to control an extra rotor resistance in an induction generator.Whenanextrarotorresistanceisaddedtotheinductionmachine,thepossibilityof having more control over the stator current is open. In this section it will be showed how differentvaluesoftherotorresistanceaffectthebehaviorofthemachine. Inthefigures3.2to3.5thedifferentresponsesofaninductionmachinewhenastepin the shaft torque is made are shown for three different values of the extra rotor resistance. One of the induction machines has no extra rotor resistance; another has the extra rotor resistancesettoaconstantvaluethatisthesameasthenominalrotorresistance;andthelast one has the extra rotor resistance set to a constant value that is the double of the nominal rotorresistance.
6700 6600 6500
Torque (Tm)
2.5
3.5
4.5
5.5
Time(s)
18
Current (A)
2.5
3.5
4.5
5.5
Time(s)
318
317.5
317
316.5
316
315.5
315 1
1.5
2.5
3.5
4.5
5.5
Time(s)
19
9000
8000
7000
6000
5000
4000
3000
2000
2.5
3.5
4.5
5.5
Time(s)
Figure3.5,Cooperlossesofaninductionmachinewhenastepintheshafttorqueoccursfor differentvaluesintherotorresistance;inblueRrisequaltothenominalvalue,ingreentheRris thedoubleofthenominalRrandinredtheRristhreetimesthevalueoftheRr. From figure 3.3, it can be noted that the higher the resistance, the smoother the responseofthestatorcurrentiswhenachangeintheshafttorqueoccurs.Thismeansthat increasingtherotorresistanceleadsthesystemtoadampedresponse,buthigherresistances leadtohighercopperlossesforthesameoperatingpoint.Thiscanbeseeninthefigure3.5, wherethecooperlossesduetotherotorresistanceincreaseproportionallytothisvalue,since therotorcurrentdoesnotvarytoomuchduetoachangeintherotorresistance. Theselosses releasesmoreheatfromthegenerator,whichoperateslessefficiently.Finally,infigure3.4itis noted how varying the rotor resistance, the rotor speed is varying as well, so as it was said beforetherotorresistancecanbeusedtocontroltherotorspeedinalimitedrange.
3.5 Parametersofageneric2MWinductionmachine
Theparametersusedinthisthesisforallthesimulationsaretakenfroma2MWIM. Theseparametersareshowedinthetablebelow:
Stator resistance Rotor resistance Stator leakage inductance Rotor leakage inductance Magnetizing inductance Machine and rotor inertia Rated voltage Frequency Number of pole pair Rs Rr Ls Lr Lm J Un f n 2.2 m 1.8 m 0.12 mH 0.05 mH 2.9 mH 460 kgm2 690 50 Hz 2
Table3.1,typicalparametersofa2MWinductionmachine.
20
21
4.1 Modelofthelinearized2MWinductionmachine
In the section 3.3 linearization of the induction machine, it was explained how a linearizedmodelofaninductionmachinecouldbeobtained,inordertouseitforthedesignof acontroller.FollowingthestepsgiveninthatsectionandusingMATLAB,alinearizedmodelof a2MWinductionmachinewascreated. For the spacestate model of a linearized induction machine model, which was presented in section 3.3 linearization of the induction machine, there are 20 transfer functions,relatingthefivestatesofthemodeltothefourinputsofthemodel.Butonlyoneof themisofinterestforthedesignofthecontroller.Thisisthetransferfunctionthatlinksthe variationintheqcomponentofthestatorcurrenttothevariationintheextrarotorresistance. This is due to the fact that the extra rotor resistance is the only parameter that is possible to control (in a very small range), and the q component of the stator current is the parameter that will be used as reference for the control system, since the stator current is relatedwiththepowerqualityparameters. After that, the nonlinear model and the linearized model of the induction machine were compared in order to see if they behave in a similar way, so that the linearized model couldbeusedforthelaterdevelopofthecontroller.Bothsystems(nonlinearandlinearized)
22
were compared for a step in the extra rotor resistance and in the shaft torque, for different valuesofthesesteps. Thenextfiguresshowthestatorcurrentresponse tostepsintherotorresistanceof differentvaluesandalsototorquesteps.Itwasmadefordifferentequilibriumpointstoseeif thebehaviorofbothmodelswasthesameinthewholerangeofinputshafttorque.
440
430
420
Current (A)
410
400
390
380
370 1.5
2.5
3.5
4.5
Time(s)
Current (A)
500 450 400 350 300 250 200 1.5 2 2.5 3 3.5 4 4.5 5
Time(s)
23
1230 1220 1210 1200
Current (A)
1190 1180 1170 1160 1150 1140 1130 1.5 2 2.5 3 3.5 4 4.5 5
Time(s)
Current (A)
2.5
3.5
4.5
Time(s)
24
2040
2020
2000
Current (A)
1980
1960
1940
1920
1900 1.5
2.5
3.5
4.5
Time(s)
2200
2000
Current (A)
1800
1600
1400
1200
1000 1.5
2.5
3.5
4.5
Time(s)
Figure4.6,statorcurrentresponse(fortheequilibriumpointofTm=10.000Nm)toresistance stepof0.002(100%ofthenominalvalue)att=2sandtorquestepof1200Nm (approximately10%ofthenominalvalue)att=3.5s,inblackthenonlinearmodel,inred dashedthelinearizedmodel. Inthefigures4.1to4.6ispossibletoconcludehowforsmallincrementsinboth,the rotorresistanceandtheshafttorque,thebehaviorisverysimilarforthenonlinearinduction machinemodelandthelinearizedinductionmachinemodel. Now the response of the linearized model will be plotted when the rated rotor resistanceisincreased10%(0.2m)after2sec,andafter3.5sectheinputtorqueisincreased
25
a10%oftheratedtorqueofthemachine(1.2kNm)forthreedifferentequilibriumpointsfor thevalueoftheshafttorque.
300 250
Time(s)
Figure4.7,statorcurrentresponseofthelinearizedmodelforthreedifferentequilibrium points,indarklineequilibriumpointwhentheshafttorqueis2kNm,inreddashedline equilibriumpointwhentheshafttorqueis6kNmandingreenpointedlineequilibriumpoint whentheshafttorqueis10kNm. Theresponseforthethreeequilibriumpointsforavariationintheinputshafttorque isalmostthesameinthethreecases,butnotforthevariationintheextrarotorresistance. Fromnowonthelinearizedmodelfortheequilibriumpointwiththevalueoftheshaft torquesetto6000Tmwillbeused. 4.1.1 Orderreductionofthelinearizedmodel Nowthatithasbeendemonstratedthatthelinearizedmodelhasasuitablebehavior, itistimetoanalysisandstudythemodelforlaterdesignofacontroller. Thetransferfunctionofthelinearizedmodel,relatingtheqcomponentofthestator currenttotheextrarotorresistance,fortheequilibriumpointof6kNmis:
(4.1)
26
300
200
-100
-200
-300
-400 -14
-12
-10
-8
-6 Real Axis
-4
-2
Figure4.8,PoleZeromapofthetransferfunctionrelatingstatorcurrenttotheextra rotorresitancevalue. Wherethedominantspolesare(5.5114.26i),andthereisapolezerocancellation ofp5andz4. Reductionoftheordersystemusingthemainpoles Thetransferfunctionofthelinearizedmodelthatwasobtainedisafifthordersystem. It is dificoult to design a controller for such kind of system. For this reason, it will be approximatedforareducedordermodel. Thefirstapproximationthatcanbedoneisthecancellationofonepoleandonezero that can be seen in the figure 4.8. By doing this the following fourth order system was obtained:
27
(4.2)
The Bode diagram presented in figure 4.9, demonstrates that the fifth order system andthefourthordersystembehaveinthesameway.
Bode Diagram 120 110 Magnitude (dB) Phase (deg) 100 90 80 70 90 45 0 -45 -90 -135 10
0
10
10 Frequency (rad/sec)
10
Figure4.9,bodediagramofthefifthordersystem(blackline)andthefourthordersystem(red dashedline). But it is still not useful to have a fourth order system, for this reason, it has to be reducedtoafirstorsecondordersystem.Nowlookingagaintothezeropolemapofthefifth ordersystem,figure4.8,itispossibletoseethatthereisazerointheoriginoftherealaxe, andthedominantpolesareplacedin5.5114.3i(thedominantpolesarethoseclosertothe imaginaryaxe).Ifthiszeroandthosepolesareusedtoobtainasecondordertransferfunction,
28
Bode Diagram 150 100 Magnitude (dB) Phase (deg) 50 0 -50 -100 90 45 0 -45 -90 -135 10
0
10
10 Frequency (rad/sec)
10
andthebodediagramlookslike
(4.3)
Bode Diagram 120 110 Magnitude (dB) Phase (deg) 100 90 80 70 90 45 0 -45 -90 -135 10
0
10
10 Frequency (rad/sec)
10
Figure4.11,bodediagramofthefifthordersystem(blackline)andthesecondordersystem obtained(reddashedline).
29
Current (A)
0.1
0.2
0.3
0.4
0.5
0.6
0.7
0.8
0.9
Time (s)
Figure4.12,statorcurrentresponsetoastepof0.1mintherotorresistanceforthefifth ordersystem(blackline)andthesecondordersystemobtained(reddashedline). Inthemethodjustdemonstrated,thereductionofthefifthordersystemwasdoneby choosing the dominant poles and zeros of the original fifth order system. There are some mathematical methods for reducing the order of a transfer function. One of these methods willbepresentednext. Reductionoftheordersystemusingtheclusteringtechnique AmixedmethodforfindingstablereducedordermodelsistheonethatusesthePade approximation and the clustering technique. The denominator polynomial of the reduced ordermodelisdeterminedbyformingtheclustersofthepolesoftheoriginalsystem,andthe coefficients of numerator polynomial are obtained by using the Pade approximation technique.Furtherinformationcanbefoundin[4]. Thetransferfunctionfoundusingthismethodwas
(4.4)
andthebodediagramandtheresponseofthesystemtoastepintherotorresistanceis
30
Bode Diagram 130 120 Magnitude (dB) Phase (deg) 110 100 90 80 70 90 45 0 -45 -90 -135 10
0
10
10 Frequency (rad/sec)
10
Figure4.13,bodediagramoftheoriginalfifthordersystem(blueline)andthesystemobtained withtheclusteringmethod(greenline).
0.1
0.2
0.3
0.4
0.6
0.7
0.8
0.9
31
4.2 Designofthecontroller
The idea behind the controller is to highpass filter the q component of the stator current,toobtainthehighpassfrequencycomponentsofthesignal,thepartthatispursued toeliminate.
Gear box
IM
Power electronics
Grid
32
isqref
Controller
RR
Induction Machine
isq
High-pass Filter
Figure4.16,possibleblockdiagramofthesystemwiththerotorresistancecontroller. The reference for the stator current will be zero, since the stator current from the induction machine is filtered in a highpass frequency filter in order to eliminate the low frequencycomponentsinthesignal. 4.2.1 Highpassfilter To highpass filter the stator current a second order filter will be used. The general transferfunctionforasecondorderhighpassfilteris:
where isthehighfrequencygainand
(4.5)
isthecutofffrequency.
33
Current (A)
2.2
2.4
2.6
2.8
3.2
3.4
3.6
3.8
Time(s)
Figure4.17,statorcurrentoftheinductionmachinewhenastepintheshafttorqueoccurs.
60 50 40 30
Current (A)
20 10 0 -10 -20 -30 -40 2 2.2 2.4 2.6 2.8 3 3.2 3.4 3.6 3.8 4
Time(s)
Figure4.18,highpassfilteredstatorcurrentoftheinductionmachinewhenastepintheshaft torqueoccurs,inbluewithcutofffrequency=1Hz,ingreenwithcutofffrequency=5Hzand inredwithcutofffrequency=10Hz. Abigdifferencecanbeseenbetweenthepeaksofthedifferentsignalsdependingon the cutoff frequency. For this reason, the cutoff frequency selected for the secondorder highpassfilterwillbeadecidingfactorinthedesignofthecontroller.Atfirstsight,itcanbe saidthatthegainofthecontrollerwillbedirectlyrelatedwithtothecutofffrequencyofthe filter. This fact is even more pronounced for higher values of the cutoff frequency, for valuesofthecutofffrequencyover10Hzitcanbesaidthattheonlydifferencebetweenthe highpass filtered signals of the q component of the stator current with different cutoff frequencyaretheirgains,asitcanbeobservedinthefigure4.19.
34
1.5
Current(A)
0.5
-0.5
-1 10
11
12
13
14
15
16
17
18
19
20
Time(s)
(4.6)
Insection6.1Selectionofthecutofffrequencyandproportionalgaincarefullyhow thebehaviorofthewholesystemdependingonthecutofffrequencywillbestudied,including thecontrollerandthehighpassfilteredcurrentfeedback. 4.2.2 Proportionalcontroller The first controller that was implemented was a simple proportional controller. The actionofthecontrolleris proportionaltothe controlerror.That meansthatinthis casethe action will be only proportional to the highpass filtered stator current since the reference valueofthestatorcurrentiszero.
35
2.5 2 1.5 1
Current (A)
Time(s)
Figure4.20,highpassfilteredstatorcurrentoftheinductionmachinewhenastepintheshaft torqueoccurs.
2.05 2.04 2.03 2.02 2.01 2 1.99 1.98 1.97 1 1.5 2 2.5 3 3.5 4 4.5 5 5.5 6 x 10
-3
Rr(Omh)
Time(s)
Figure4.21,rotorresistanceoftheinductionmachinewithproportionalcontrollerwhenastep intheshafttorqueoccurs. Inthefigures4.20and4.21,itcanbeobservedwhatwassaidbefore,thatthevalueof therotorresistanceisdirectlyrelatedwiththehighpassfilteredstatorcurrent. With a proportional controller the system becomes unstable for values of Kp bigger than1e2.Simplyturningupthegain,leadtoinstabilityofthesystem.Thiscanbeseeninthe nextfigures.
36
Current (A)
1.5
2.5
3.5
4.5
5.5
Time(s)
Current (A)
1.5
2.5
3.5
4.5
5.5
Time(s)
Figure4.23,statorcurrentoftheinductionmachinewhenastepof500Nmisintroduceat t=3,5sec,inblacktheresponseofthesystemwithaproportionalcontroller(kp=1e5).Thered dashedlineistheresponseofthesystemwithoutcontroller. For higher values of Kp the system is becoming more and more unstable until it is criticallystableforKp=1e2.Thiscanbeexplainedbystudyingthenextfigures.
37
15
10
Imaginary Axis
-5
-10
-15 -250
-200
-150
-100
-50
Real Axis
Figure4.24,PoleZeromapofthesystemwithproportionalcontroller,Kp=1e5.
15
10
Imaginary Axis
-5
-10
-15 -160
-140
-120
-100
-80
-60
-40
-20
Real Axis
Figure4.25,PoleZeromapofthesystemwithproportionalcontroller,Kp=1e8. AfterstudyingthesystemfordifferentvaluesofKp,itwasfoundthat:forvaluesofKp higherthan1e2 thesystemwasunstable.Thesystemwasbecomingmorestableforsmaller valuesofKp,untilitwassosmallthatitdidnotaffectthebehaviorofthesystem,asitisshown infigure4.22,wheretherearetwozeropolecancellation,andtheotherspolesandzeroare placed in the same position as the 2nd order transfer function of the linearized induction machinethatrelatesthestatorcurrenttotherotorresistance,equation4.3.Thismeans,that it is not possible to improve the response of the system with just a proportional controller, becausethebestresponsewillbeobtainedwhentheKpissosmallthatitisthesamelikeifthe controllerwerenotthere.
38
4.2.3 ProportionalIntegralcontroller Once the proportional controller has been ruled out, it was implemented the most common controller in the control industry, the classic PI. The ProportionalIntegral (or PI control) combines proportional gain in parallel with an integrator. A block diagram of the currentcontrolwiththePIcontrollercanbeseenbelow.
isq
ref
PI Controller
RR
Induction Machine
isq
High-pass Filter
Figure4.26,BlockdiagramofthesystemwithaPIcontroller. The most complicated thing for the PI controller is tuning the values of the proportional gain and the integrator gain. There are a lot of literature and also a lot of methods for obtaining the values of these parameters. But these methods are not useful in thiscase,sinceinthetransferfunction(equation4.3)thereisazerointheoriginofthereal axe, and also in the feedback loop there is a second order transfer function (the highpass filter). Therefore, the tuning of the PI controller was made by trying different values for Kp andKi,andobservingtheresponseofthesystem.
1340 1320 1300
Current (A)
1.5
2.5
3.5
4.5
Time(s)
Figure4.27,statorcurrentwhenastepof500Nmoccursintheshafttorque,inblackthe responsewiththePIcontroller(Kp=1e5,Ki=5e4).Reddashedlineistheuncontrolledsystem.
39
Current (A)
1.5
2.5
3.5
4.5
Time(s)
Figure4.28,statorcurrentwhenastepof500Nmoccursintheshafttorque,inblackthe responsewiththePIcontroller(Kp=1e7,Ki=2e4).Reddashedlineistheuncontrolledsystem. As it can be seen in the figure 4.28, the results obtained are a little bit better than usingonlytheproportionalcontroller,buttheyarestillnotasdesirableasitwasexpectedat thebeginningoftheproject. Whathappenswiththeproportionalintegralcontroller,isthatasmuchasthevalues ofKpandKiareincreasethemoreunstablethesystembecomes,inthesamewayasoccurwith theproportionalcontroller. 4.2.4 Doubleintegratorcontroller AftertheProportionalcontrollerandtheProportionalIntegralcontroller,anonevery usual transfer function for a controller was tried, a double integrator. In the same way as it wasdonefortuningtheothercontrollers,differentvaluesofthegainwereproved.Butinthis case,therewillbeonlyoneparameterthathastobeset.
(4.7) AgoodresponseofthesystemtoastepintheinputtorquewasobtainedforKp=5e3. The plots of the stator current response and the extra rotor resistance, with the double integratorcontrollerarepresentedinthenextpage.
40
Current (A)
10
Time(s)
Rer (Omh)
8 6 4 2 0 -2
10
Time(s)
41
Current (A)
-1
-2
10
Time(s)
Figure4.32,blockdiagramofthedoubleintegratorcontroller. Iftheblockdiagramissimplified,thenthefollowingtransferfunctionisobtained:
(4.8) Then, if the controller described above (equation 4.8) is implemented, the stator currentwillhavethesamebehaviorasinfigure4.29,buttheextrarotorresistancewillbehave like in figure 4.33. In this figure it is also possible to see that the value of the extra rotor resistancewhenitgoestoinfinite,ittendtoacertainnegativevaluethatdependsonthedc component that the highpass filtered signal has. And of course, this is not possible because theresistancecannotbenegative. In any case, a constant value that will be related with the stator current could be addedafterthecontroller,tocompensateit.
42
12 10 8
x 10
-5
Rer (Omh)
6 4 2 0
-2
10
Time(s)
Table4.1,valuesofthehighpassfilteredstatorcurrentoffsetfordifferentoperating points. Fromthetableaboveitcouldbenotedthatthesevaluesarereallysmall,iftheyare compared with the variations that can suffer the extra rotor resistance due to the torque variationsduringthenormaloperationofthewindturbine.Forthisreasonthisoffsetwillbe not taken into account for the selection of the average rotor extra resistance that has to be addedforthecompensationoftheinstantaneousrisesanddecreasesinthetorque,whichwill bestudiedcarefullyinthechapter6inthesection6.2Selectionofthedefaultexternalrotor resistance.
43
Chapter5 Evaluationofthecontroller
The stator current controller that was developed in the previous chapter has been implemented in Matlab/Simulink to see the response of the whole system and analyze the powerqualityimprovementaswellasthemechanicalstressesreduction.Inthischapter,first the behavior of the system when it was exposed to synthetic shaft torque curves will be presented.Afterthat,theresponseofthecontrolledsystemwhenitwasexposedtorealshaft torquedatawillbeshown. In this chapter, the values selected for the parameters to be tuned in the controlled system were Kp= 5e3 for the controller gain, a cutoff frequency of 10 Hz for the highpass filter and a average value of 1.2 m for the external rotor resistance. These values were proved to work fairly well for different shaft torque curves. In the next chapter it will be presentedhowdifferentcombinationsoftheseparametersaffectthebehaviorofthesystem.
5.1 Responseofthesystemtosyntheticcurves
In the previous chapter it was shown how a step in the load torque affects the response of the IM with the stator current controller. In this section the response of the systemfordifferentsyntheticloadtorquecurveswillbeanalyzed. Responsetosteppedsignal First the controller was analyzed with a signal where a step of 400 Nm occurs every second,toseeifthesystemcanfollowpositiveloadtorquevariationsaswellasnegative.The electrical torque generated, the stator current and the rotor resistance are presented in the nextpage.
44
Tg,Ts (Nm)
10
Time(s)
Figure5.1,inreddashedlinetheinputshafttorque,inblackthegeneratedelectricaltorquefor thecontrolledsystem.
1220 1200 1180
Current (A)
10
Time(s)
Figure5.2,statorcurrent,inreddashedlinetheuncontrolledsystem,inblackforthecontrolled system.
3.8 3.6 3.4 3.2 x 10
-3
Rr(Omh)
Time(s)
Figure5.3,rotorresistanceforthesystemwiththecontrollablerotorresistance.
45
Fromfigure5.2itcanbeseenthattheresponseofthecontrolledsystemissmother thanthesystemwithoutcontroller.Italsocanbenotedthattheovershootsthatexistinthe noncontrolledsystemareeliminated.Finallyitcanbementionedthatthecontrolsystem followapositivevariationintheshafttorqueaswellasanegative. Responseofthesystemtosinusoidalcurves In this case, the input load torque was composed for 3 sinusoidal signal of 1 Hz 2Hz and 3 Hz, with amplitudes of 250 Nm125 Nm and 75Nm respectively added to a constant valueof6000Nm.Withthiskindofcurveitwasofinteresttoseehowthecontrollerbehaves forsharpchangesintheloadtorque.Infigure5.4,itcanbeseenthatthepeaksinthestator current signal have been reduced to the half size of the peaks of the signal without the controllerimplemented.
7000 6800 6600 6400
Tg,Ts (Nm)
Time(s)
Figure5.4,inreddashedlinetheinputshafttorque,inblackthegeneratedelectricaltorquefor thecontrolledsystem.
1300 1280 1260 1240
Current (A)
Time(s)
Figure5.5,statorcurrent,inreddashedlinetheuncontrolledsystem,inblacklinewiththe controller.
46
x 10
-3
Rr(Omh)
Time(s)
Figure5.6,rotorresistanceofthesystemwiththerotorresistancecontroller. After testing the controller with synthetic curves, and check that the response was acceptable,itwastimetotestthecontrollerwithrealshafttorquedata.
5.2 Responseofthesystemtorealshafttorquedata
Inthissectionthe controllerhasbeen exposedtorealshafttorqueloads.Thiscurve wasobtainedfromarealwindturbineexposedtorealconditionsofwind,andhasaduration of180s.Theshafttorquecurvewhichthesystemwastestedisshowninfigure5.7.
8500
Torque (Nm)
20
40
60
80
100
120
140
160
180
Time (s)
Figure5.7,realshafttorquecurve.
Below this paragraph the detailed response of the system with the stator current controllerimplementedandwithoutthecontrollerwillbepresented.Theshafttorque,stator currentandrotorresistancevariations,fordifferent20secondgapsofthe180secondssignal willbeplotted.
47
8200
Tg,Ts (Nm)
8000
7800
7600 20
22
24
26
28
30
32
34
36
38
40
Time(s)
Figure5.8,inreddashedlinetheinputshafttorque,inblackthegeneratedelectricaltorquefor thecontrolledsystem.
Current (A)
22
24
26
28
30
32
34
36
38
40
Time(s)
Figure5.9,statorcurrent,inreddashedlinetheuncontrolledsystem,inblacklinewiththe controller.
2.6 x 10
-3
Rr(Omh)
2.4 2.2 2 20
22
24
26
28
30
32
34
36
38
40
Time(s)
Figure5.10,rotorresistanceofthesystemwiththerotorresistancecontroller.
48
7800
Tg,Ts (Nm)
122
124
126
128
130
132
134
136
138
140
Time(s)
Figure5.11,inreddashedlinetheinputshafttorque,inblackthegeneratedelectricaltorque forthecontrolledsystem.
Current (A)
1550
1500
1450 120
122
124
126
128
130
132
134
136
138
140
Time(s)
Figure5.12,statorcurrent,inreddashedlinetheuncontrolledsystem,inblacklinewiththe controller.
2.6 x 10
-3
Rr(Omh)
122
124
126
128
130
132
134
136
138
140
Time(s)
Figure5.13,rotorresistanceofthesystemwiththerotorresistancecontroller.
In the figures above, it can be noted that the response of the controlled system is smootherthanfortheuncontrolledsystemaswasexpected.Alsothevariationofthevariable rotor resistance is about 0.8 m, which represent 46 % of the nominal value of the rotor resistance.Thisvalueisonlyroughvaluethat hasastrongrelationwiththewindconditions
49
and also with the proportional gain of the controller; the variation of the external rotor resistancewillbestudiedcarefullyin6.2Selectionofthedefaultexternalrotorresistance. The figures above are just illustrative, the developed controller will be analyzed carefullyinthenextsectionbymeasuringdifferentparameterssuchastheflickeremissionor thelossesintheinductionmachine.
5.2.1 Flickerreduction Asitwassaidinthesection2.4Powerqualitycharacteristicsofwindturbines,one wayofevaluatethepowerqualityismeasuringtheflickercontributionofthewindturbineto thegrid.TheflickercontributioncanbedescribedwiththedimensionlessparametersPstand Plt.Thefirstoneisreferredtotheshortterminfluence,andthesecondonetothelongterm influence.Inthiscaseonlytheshortterminfluencewillbestudied. Accordingto IEC6140021[7]theflickerevaluationshouldbedeterminedindirectly, basedonasimulatedvoltageonafictitiousgrid.Tosimulatethevoltageonthefictitiousgrida methodbasedonvoltageandcurrentmeasurements(currentflicker)canbeapplied.
Rfic
Lfic
im(t)
U0 (t)
Ufic(t)
Figure5.14,circuitofthefictitiousgrid.
The fictitious grid is represented as in figure 5.14, and it is composed for an ideal phasetoneutralvoltagesourceU0(t)andagridimpedancerepresentedbyaresistanceRficin serieswithaninductanceLfic.Thecurrentsourceim(t)representsthewindturbine,whichisthe measured instantaneous value of the phase current. With this simple model, the fluctuating voltageUfic(t)inthepowersystemisgivenby: (5.1)
ThefluctuatingvoltageUfic(t)isthenusedasinputtoavoltageflickeralgorithmthatis described in IEC 61000415 [8] to generate the flicker emission value Pst,fic. The simulated flickerPst,fic valuewilldependontheshortcircuitpowerofthegrid,Sk,fic,andtheangleofthe gridimpedance,k.Thegridangleandshortcircuitapparentpowerofthefictitiousgridare givenbythefollowingequations:
50
(5.2)
(5.3)
foreachPst,ficvalue,aflickercoefficientc(k)couldbedeterminedbyutilizing:
Theresultsofthec(k)valuesforthestatorcurrentcontrollerarepresentedintable 5.1.Inthetableappeardifferentvaluesfordifferentgapsof60secondfromrealshafttorque data. Theresultsofthestatorcurrentcontrollerhasbeencomparedwiththesamesystem withoutanycontroller,withtherotorresistancevaluesettothedoubleofthenominalrotor resistance, and with the system with a fixed rotor resistance set to three times the nominal valueoftherotorresistance. Withstatorcurrent controller Withoutcontroller Fixedrotorresistance to3.6m Fixedrotorresistance to5.4m 1stgap 0,48 0,895 0,69 0,61 Table5.1,c(k)valuesfordifferentshafttorquecurves Accordingtotable5.1,thepowerqualityofthecontrolledsystemisbetterthaninany othersystemstudied.Thisisbecausethecontrolledsystemhasthelowerc(k)valuesinthe table.Thereductionintheflickercontributioncangofrom35%tomorethan60%withregard to the uncontrolled system, depending on the shaft torque curves. If the average c(k) is calculatedforallthegaps,thenthefollowingresultsareobtained: Withcontroller:c(k)=0,395 Withoutcontroller:c(k)=0,675 Fixedrotorresistanceto3.6m:c(k)=0,525 Fixedrotorresistanceto5.4m:c(k)=0,48 2ndgap 0,4 0,575 0,475 0,44 3rdgap 0,41 0,585 0,495 0,46 4thgap 0,335 0,52 0,415 0,385 5thgap 0,395 0,59 0,5 0,46
51
The finding is that the average reduction of the flicker contribution by utilizing the controller is almost the 50% with regard to the uncontrolled system, for this shaft torque curves.Thesystemwiththerotorresistancefixedtothreetimesthenominalrotorresistance (5.4 m) has the closest response to the controlled system. But as it was said before, the higher the rotor resistance is, the higher the losses in the rotor resistance will be. Later the lossesinthemachinewillbestudiedinitsownsection. Itmustbesaidthatsincetheexternalrotorresistancecanonlyvaryinasmallrange, thereductionintheflickercontributionwillbeverydependentontheshafttorquecurves,a higherreductionontheflickeremissionwasobservedforahigherturbulenceinthewind. 5.2.2 Mechanicalstressesreduction Inthissectionhowimplementingthestatorcurrentcontrollerreducesthemechanical stresses on the turbines shaft will be presented, results from several simulations will be evaluated. Talking about mechanical stresses, the highfrequency components of the electrical torquearetheworstforthemachine.Sotoanalyzethemechanicalstressesitwillbedonethe FFT (Fast Fourier Transform) of the electrical torque signal. It will give the magnitude of the torquecomponentforeachspecificfrequency.Itcanbeseeninfigure5.15.
10
2
Torque (Nm)
10
10
10
-1
10
-2
10
-1
10
10
Frequency (Hz)
Figure5.15,Periodogramoftheelectricaltorqueoftheinductionmachine,inreddashedline withoutthecontrollerandindarklinewiththecontroller. What is possible to see in figure above, is that the magnitude of the highfrequency components of the electrical torque for the system without controller is always higher than thecontrolledsystemafterf=1Hz.Then,thecurvethatrepresentsthenoncontrolledsystem isalwaysabovethecurvethatrepresentsthecontrolledsystem.Thiswasexpectedbecauseby implementingthiscontrolleritwaspursuedtoreducethehighfrequencycomponentsofthe torque,whicharetheworstforthemechanicalstressesinthegearbox.Thesehighfrequency
52
(5.5)
whereT(n)isthemagnitudeofthetorqueforthefrequency . Byusingthisparameter,thehigherfrequenciesarepunishedbymultiplyingthetorque magnitudebythefrequency. Ifthisparameteriscalculatedforthesystemwiththecontrollerimplementedandfor the system without the controller, then the results obtained are 22.1156e3 and 29.6226e3 respectively.Thesevaluesagreewiththecurvesthatappearinfigure5.15,wherethevaluefor thecontrolledsystemismuchsmallerthanthevalueforthenoncontrolledsystem. 5.2.3 Energylossesintheinductionmachine Inthissectionitwillbepresentedhowthevariationsintherotorresistanceaffectthe energylossesintheinductionmachine.Forthispurposethecopperlosseswillbedetermined. Thecopperlossesarecalculatedbecausetheyarestraightrelatedwiththevalueoftherotor resistance,sothehighertherotorresistanceis,thehigherthecopperlosseswillbe. To compare the losses of the system running with the stator current controller, the copperlossesofthesystemwithoutcontrollerandwiththerotorresistancefixedto3.6m willbealsoplotted.
10000 9000
3000
20
40
60
80
100
120
140
160
180
Time(s)
53
Figure5.16,cooperlossesoftheinductionmachine,inbluewiththecontrollerimplemented,in greenwithoutexternalrotorresistances,andinredwiththeexternalrotorresistancesfixedto 1.8m. Fromfigure5.16,itcanbeseenthatthecontrolledsystemhasmorelossesthanthe system without external rotor resistances, but not much higher, however, much less losses than the system with the rotor resistance fixed to 3.6 m like it could be expected. This is because the average of the rotor resistance selected is 2.4 m, so the copper losses, in average,willbethesameasistherotorresistancesoftheinductionmachinewerefixedto2.4 m. Itistruethatrunningageneratorathigherslipgeneratesmoreheatfromthegenerator, whichrunslessefficientlyathasbeenshownbefore.Butsincetheincrementoftheselossesis notasmuchsignificantastheimprovementinthepowerquality,itismorejustifiedimplement this controller. But it always must be present that as higher these losses are, it will be necessarytoincreasethecoolingofthegenerator.
54
55
6.1 Selectionofthecutofffrequencyandtheproportionalgain
As it was presented in the section 4.2.1 highpass filter, the cutoff frequency influences in the proportional gain that have to be selected, so indirectly, when the cutoff frequencyisselected,thegainofthecontrollerisalsoselected. Theparameterselectedforcomparingtheresponseofthesystemfordifferentvalues ofthecutofffrequencyandtheproportionalgainwasthePst.Becausethisparametergivesan ideaofthepowerqualitygeneratedforthewindturbinethatisoneoftheaimsthatarefound withthiscontroller.Fromthedifferentsimulationsconducteditwasobservedthatthecutoff frequencyshouldbeatleast5Hztogetagoodresponsefromthecontroller. Whenthecutofffrequencyissmallerthan5Hz,thenaflickeringeffectappearsinthe signallikeitispresentedinfigure6.1.
56
Current (A)
Time(s)
Figure6.1,statorcurrentforacutofffrequencyinthehighpassfilterof3Hz. Infigure6.2,therelationbetweenthecutofffrequencyandthemaximumgainthat can be applied to the controller is presented, which is also the value that gives the best response.
Figure6.2,relationbetweenthecutofffrequencyandthemaximumKpselectable. The Pst values for the different combinations of cutoff frequency and Kp are fairly similarforthefrequenciesfrom7.5Hzto20Hz.Asitwassaidinthesection4.2.1Highpass filter,thisisbecausethemaindifferencebetweenthehighpassfilteredstatorcurrentsignals forthedifferentcutofffrequencyhigherthan10Hzistheamplitudeofthesignal,whichcan be compensated by adjusting the proportional gain. This can be observed in the figure 6.3, wheretheminimumflickeremissionsarealmostthesameforcutofffrequencieshigherthan 5Hz.
57
Figure6.3,relationbetweenthecutofffrequencyandthePstobtained. Forthedifferentcombinationsitwasseenthatwhenthecutofffrequencywas20Hz andtheKpwas0.3,theminimumPstcoefficientswereobtained.Inthenextfiguretherelation betweentheproportionalgainofthecontrollerforagivencutofffrequencywillbepresented. In figure 6.4 can be noted how for higher values of Kp the flicker contribution obtained was smaller, until the controller reach the maximum value of Kp that makes the system critically stable.Inthiscasethatvaluewas0.33.
Figure6.4,relationbetweenproportionalgainofthecontrollerandthePstobtainedwhenthe cutofffrequencyofthehighpassfilteris20Hz.
58
It has to be mentioned, that this behavior was the same for the different cutoff frequencies studied, were the higher the proportional gain is, the smaller the flicker contributionis,untilthemaximumKpvaluethatmakethesystemunstableisreached. Now, how the intensity in the wind gusts affect the flicker contribution will be presented. For this purpose the flicker contribution of the controlled system and the non controlledsystemfordifferentintensityintheshafttorquevariationwillbestudied.
Figure6.5,relationbetweenshafttorquevariationintensityandthePstobtainedwhenthecut offfrequencyofthehighpassfilteris20Hz. In figure 6.5 it can be observed that for higher fluctuating shaft torque curve, the higher the flicker reduction of the controlled system respect the uncontrolled system. This means that in presence of high turbulence intensity in the wind the flicker reduction will be considerablysmallerthanforthesystemwithoutthecontroller.
6.2 Selectionofthedefaultexternalrotorresistance
Thedefaultvalueoftherotorresistanceshouldbeassmallaspossible,butstillhigh enoughtoallowthepossibilitytoreduceitwhenadecreaseintheshafttorqueoccurs.This parameterwillberelatedtothemaximumvariationsthatoccurintheextrarotorresistance. Then it will be set according to the behavior of the system when it is exposed to high turbulencesinthewindload,whichisthemomentwhenthecontrollerisaskedtheextrarotor resistancetovarymore.
59
Rr(Omh)
3 2 1
20
40
60
80
100
120
140
160
180
Time(s)
Figure6.7,rotorresistancesample.
After studying different curves of the extra rotor resistance like figure x, then it was noted that the maximum variation from the average rotor resistance was 1.2 m. Since the valueoftherotorresistancehasalwaystobeabove1.8m(thenominalvalueoftherotor resistance), that is the one of the requirements that must be fulfilled to select the default externalrotorresistance,andifasecuritymarginof0.6mischosen,thentheaveragevalue of the external rotor resistance will be 1.8 m, that is the same as the nominal rotor resistance. This means that average total rotor resistance will be the double of the nominal rotorresistance,andthetotalrotorresistancewillbeabletovarybetween1.8mto5.4m.
6.3 Finalspecifications
Accordingwiththeprevioussectionsinthischapter,thefollowingvaluesforthestator currentcontrollerhavebeenselected: Itmustbesaidthatsincetheinductionmachinedoesnothavealinearbehavior,the differentparametersthatappearabovearenotthebestoptionforallthesituations,butthey worksfairlywellforthedifferentcurvesthatwereevaluated. Cutofffrequency=20Hz Kp=0.3 Rr0=1.8e3
60
61
Chapter7 Conclusions
Thepurposeofthisthesishasbeentoevaluatethebehaviourofawindturbinewith aninductiongeneratorequippedwithavariablerotorresistance.Acontrollerforthevariable rotor resistance generator has been developed with the objective of minimizing the torque variations that reduces the mechanical stresses onthe shaft of the turbine and also reduces theflickeremission,whichleadstoabetterpowerqualityimpact. In this thesis an induction generator equipped with a variable rotor resistance was analysedandmodelled.Afterthatdifferentconfigurationsforthecontrollerwerestudiedto findthebeststructureforthecontroller.Finally,theinductionmachinewiththevariablerotor resistancecontrollerwascomparedwiththesameinductionmachinewithoutanycontroller, tostudytheimprovementofimplementingthecontroller. What was found is that reduction in the flicker contribution by utilising the variable rotor resistance can be between 35%60% compared to the uncontrolled system, depending on the shaft torque curves. A higher reduction on the flicker emission was observed for a higherturbulenceinthewind. Fromthisthesis,itcanbesaidthatthissystemgeneratesasmootherelectricpowerin thepresenceofwindgustswithreductionofthemechanicalstressandthemechanical.Witha traditional fixedspeed system, without a variable rotor resistance, in the presence of wind gusts,thepowerdeliveredisfluctuatingstronglywithanincreaseofthepressuresandstress appliedtoitsmechanicalelements.
62
63
Chapter8 Proposedfuturework
In this thesis an induction machine with a variable rotor resistance has been investigated. As always there are many more interesting aspects that can be taken in consideration.Forcompletionofthemodelofthe windturbine,adrivetrainrepresentation should be included in the model. Also a pitch control system should be added to the wind turbine, to ensure that the two control systems cooperate efficiently, since pitching the bladesisamechanicalprocess,whichmeansthatthereactiontimeforthepitchmechanism willbemuchhigherthanforthevariablerotorresistance. Otheraspectsthatcouldbeofinterestistoinvestigatehowseveralofthesewind turbinesusingavariableslipresistancebehavesworkingtogether.Anotherinterestingissueis theevaluationofawindfarmwiththeseturbinesthatisconnectedtothegridoveranHVDC line.
64
65
References
[1]A.Petersson,T.ThiringerandL.Harnefors,Flickerreductionofstallcontrolledwind turbinesusingvariablerotorresistances;Nordicwindpowerconference,ChalmersUniversity ofTechnology,12March,2004. [2]B.Kahinpour,Modelling,controlandinvestigationofanHVDCtransmissionforan offshorewindfarm/Thecentralizedcontrolstrategyforanoffshorewindfarm;Licenciate thesis,ChalmersUniversityofTechnology,2009. [3]B.C.KuoandF.Golnaraghi,Automaticcontrolsystems;JohnWiley&Sons,2003. [4]CB.VishwakarmaandR.Prasad,ClusteringMethodforReducingOrderofLinearSystem usingPadeApproximation.IETEJRes2008;54:32630. [5]R.S.Burns,Advancedcontrolengineering;ElsevierLtd,2001. [6]J.O.TandeandJ.A.PeasLopes, AssessmentofPowerQualityCharacteristicsofWind Farms;IEEE,2007. [7]IEC6140021:2001;WindturbinegeneratorsystemsPart21:Measurementand assessmentofpowerqualitycharacteristicsofgridconnectedwindturbines,IECStandard, 2001. [8] IEC61000415,Electromagneticcompatibility(EMC)Part4:Testingandmeasurement techniquesSection15:FlickermeterFunctionalanddesignspecification,1997+A1:2003. [9] T.Thiringer,J.PaixaoandM.BonGiorno,Monitoringofridethroughabilityofa2MW WindturbineinTvker,Halland;TechnicalReport,ChalmersUniversityofTechnology,2009. [10]A.PeterssonandT.Thiringer,ControlofaVariableSpeedPitchRegulatedWind Turbine,TechnicalReport,ChalmersUniversityofTechnology,Gteborg,Sweden,2005. [11]T.ThiringerandAndreasPetersson,Controlofafixedpitchvariablerotorresistance windturbine;TechnicalReport,ChalmersUniversityofTechnology,2009. [12]S.SoterandR.Wegener,DevelopmentofInductionMachinesinWindPower Technology;InstituteofElectricalDrivesandMechatronicsUniversityofDortmund, Germany,2003. [13]G.FranklinJ.D.PowellandA.EmamiNaeini,Feedbackcontrolofdynamicsystems;5th Edition,Hardcover,2005. [14]S.Lundberg,WindFarmConfigurationandEnergyEfficiencyStudiesSeriesDCversusAC Layouts;Ph.Dthesis,ChalmersUniversityoftechnology,2006. [15]ABBPowerTechnologiesAB,ItstimetoconnectTechnicaldescriptionofHVDCLight technology;Brochure,2005.
66
67