Photovoltaic plant with reduced output current harmonics using generation-side active power...

www.ietdl.org

IE

d

Published in IET Renewable Power GenerationReceived on 29th July 2013Revised on 4th December 2013Accepted on 19th February 2014doi: 10.1049/iet-rpg.2013.0251

T Renew. Power Gener., 2014, Vol. 8, Iss. 7, pp. 817–826oi: 10.1049/iet-rpg.2013.0251

ISSN 1752-1416

Photovoltaic plant with reduced output currentharmonics using generation-side active powerconditionerMuhammad Imran Hamid1,2, Awang Jusoh1, Makbul Anwari3

1Faculty of Electrical Engineering, Universiti Teknologi Malaysia, Malaysia2Electrical Engineering Department, Andalas University, Padang, Indonesia3Electrical Engineering Department, Umm Al-Quro University, Mecca, Saudi Arabia

E-mail: [email protected]

Abstract: This study presents a study on a scheme to reduce the current harmonics from a photovoltaic (PV) plant. The schemeuses a power conditioner unit, which is placed parallel with the plant and works in feed-forward mode, and behaves to compensatethe PV plant’s output current distortion, so that the total current flows to the grid is sinusoidal. In addition, this scheme includes aneffective control structure, which combines a simple sine-wave generation for synchronisation, as well as a distortion current’scomponents extraction, and a conditioner current control. In this study, the control system was implemented to a voltagesource inverter with LCL output filter as conditioner power circuit. The workability of the scheme and the power qualityimprovement achieved were evaluated by using a TMS320F28335 DSP-based control conditioner proto-type; using realsignal from PV plant operation. Then, the effect on the conditioner installation to the power quality improvement for differentpower operation of PV-inverter was analysed. Finally, the improvement on the utilisation factor of PV-inverter and plant wasanalysed and described.

1 Introduction

Today, the world’s electricity supply is still highly dependenton fossil-based power generation. The International EnergyAgency report stated that by 2010, approximately 45.2% ofthe world’s electricity production is generated from thistype of fuel [1]. The dependence on this source causeselectricity prices likely to increase as the price of fossilfuels increases every year. Encouraged by this factor, inrecent years, there is a trend to look for alternative energyfor electricity. Among the alternative sources, renewableenergy such as photovoltaic (PV) generation has receivedmuch attention. This is because the availability of PV isabundant, the generation plant can be built easily andinstantly, flexible in location, and the plant capacity can beadjusted to follow the demand [2]. In addition, the rapiddevelopment of PV generation technology has resulted insignificant reduction of cost per electricity generation.With regard of its advantages, the application of PV for

electricity has increased both in capacity and in itsimplementation at various places. However, PV energy isgenerated and integrated into the grid through aPV-inverter, which is power electronics-based equipment. Ithas a non-linear characteristic, which may generate powerquality problems and affect other generation anddistribution apparatus and consumer equipment [3].PV-inverter(s) harmonics can trigger further consequencessuch as resonance, DC injection to AC system, voltage

distortion, transformer overheating, overload in neutral wire,power factor degradation and many other power qualityproblems.To suppress the deterioration of power quality because of

harmonics from PV-inverter integration, various approacheshave been introduced. One of them is suppression strategythat is applied to the inverter controller. Hatziadoniu et al.[4] proposed a control scheme which consisted of asteady-state loop to eliminate the low-order harmonicsproduced by an inverter. Ko et al. [5] implemented a directcoupled power quality controller which used an inner andouter current control loop to improve the grid power qualityof grid-connected PV inverter. Wu et al. [6] demonstratedhow the non-linear behavioural of the components thatcaused distortion on the output current of the inverter wastaken into account in the design of the controller togenerate a better current output. Improving the quality ofthe output current has also been carried out with the use ofthe converter topology, which is classified as the lowharmonic’s type converter. In this category, the use ofpulse-width modulation (PWM) and multilevel inverter in[7–9] are good examples. Another method that has beenused is by paralleling two sub-converters, which build a PVinverter system. In [10], parallel association of twovoltage-source inverters in different topologies and workingin different switching frequency were employed to obtainthe sinusoidal output from the inverter. The first converterwas a neutral point clamped multilevel inverter that worked

817& The Institution of Engineering and Technology 2014

www.ietdl.org
in low frequency (near to power frequency) and producedquasi-square waveform output, and the second one was atraditional inverter operated with a synchronous hysteresisband controller (high switching frequency). Parallelassociation principle to produce a summation current to thedesired shape resulting from multiple inverters had alsobeen conducted in [11], where in this case, twoPV-inverters were employed as a harmonic’s compensatorin a load-side compensation.In addition to improving the inverter design, PV-inverter
integration into the grid has also been regulated with avariety of stringent standards. Some standards have beenintroduced, such as IEEE519-1992, IEC1000-3-2,IEC1000-3-4 and IEC610003-2. These standards areintended to protect the grid from the use and integration ofPV-inverters that can degrade the quality of power in thesystem. However, the approaches have not shown fullyoptimum results; harmonics produced by PV-inverter stillappear in the real operation of PV plant.The difficulty to suppress the existence of harmonics from

PV-inverter is related to their real operation condition.PV-inverters are employed in an environment where factorsthat cause them to produce harmonics and power qualityproblems are uncontrollable. Fig. 1a shows a histogrambuilt from the observation on a small scale PV plant with2.1 KW PV-inverter operation. The plant was dominantlyoperated at 100 W level, and as shown by the cumulativecurve, in most operation time (about 70% operating time) itoperated at the power levels of 100–400 W range, or withina range of 4.76–19.04% of the plant installed capacity (0.4/2.1 K). Fig. 1b shows the trend line of current distortion asthe function of the power level operation of the PV-inverterused at the plant. The trend line indicates that to operatewith output currents distortion less than 20% of totalharmonic distortion (THD) according to Institute ofElectrical and Electronics Engineering std 519 [12], thePV-inverter must be employed at least 16.6% of its overallrating. The value of the PV plant operating level and theminimum value of inverter operation in order to operate instandard-meet power quality (19.04 and 16.6%) as

Fig. 1 PV Plant and inverter operation

a Occurrence of various power levels operation in a PV plant with totally 2.1 KWb Trend line to describe the relation between inverter’s load factor and distortion l


described in this case, indicate that the plant often produceda power with a level that caused the inverter to operate inpoor condition. This case confirms why harmonics and poorpower quality still appeared even when the plant wasequipped with standard-meet PV-inverter.Level and power range operations of a PV plant are highly

dependent on the condition of energy sources, for example,solar irradiance and array temperature [2]. PV-energydensity is influenced by many factors: fluctuation inirradiance intensity because of the changes in the sun’sposition, weather changes, partial shading and movingclouds. Some of these factors occur extremely and abruptly,causing energy density to fluctuate suddenly within awide-range level. All these are natural factors that aredifficult to be avoided; as implication, the plant wouldoperate at a power that fluctuates in a wide range. As aresult, power quality problems would arise and alwaysexist. To always fulfil the power quality standard undersuch conditions, further efforts and approaches in the PVplant design and operation are very much needed.This paper proposes a schema which utilises power

conditioner to reduce the current distortion and improve thepower quality of PV-inverter at a PV plant. It is differentfrom the other improvement methods applied toPV-inverter, in which the conditioning method isimplemented on the functions and equipment embedded inthe PV-inverter unit. In this scheme, conditioning functionis realised as a separated unit, as a generation-side powerconditioner installed at the output side of a PV plant.Conditioner works in parallel with the PV-inverters fromPV plant, but unlike the parallel working mode as in [10]and [11] where the control system of the paralleledconverter is performed in a coordinated manner, in thiscase, the conditioner’s controller is operated independentlyand separated from the PV-inverters control at the PV plant.The unit is placed at a point where the output of an inverteror accumulation output of a number of inverters in plant iscreated [point of common coupling (PCC)]. At this point,compensation current from the conditioner is added tocompensate the distorted-current from PV-inverter(s). A

PV inverterevel produced

IET Renew. Power Gener., 2014, Vol. 8, Iss. 7, pp. 817–826doi: 10.1049/iet-rpg.2013.0251

www.ietdl.org
sinusoidal current then will be produced and be injected intothe grid even when PV plant operates under wide-rangepower fluctuation. The proposed conditioner in this schemeworks in feed forward mode. A voltage source inverter(VSI) is used as power circuit; where a simple and fastconditioner controller consists of reference generation forcompensation and PWM current control is realised usingdigital signal processor (DSP) TMS320F28335. Referencefor the compensator is generated by sensing the distortedcurrents from the PV plant to obtain the fundamentalcomponents of the active, reactive and harmonics of thecurrent through Fourier’s components extraction method.Meanwhile, the grid-synchronised sine-wave used in theextraction process is obtained from a synchronisationmechanism based on lookup table method. For theworkability, we had tested the conditioner and the schemeunder real operation at a PV plant.The next section in this paper will discuss this scheme and
proposed conditioner more deliberately.

2 Operation principle of proposedimprovement schema

The structure of the proposed conditioner is shown in theimprovement scheme in Fig. 2. The conditioner consists ofa single-phase H-bridge VSI working in current-controlmode and an LCL output filter. Connection to the grid isdone through an isolation transformer. For three-phaseconditioning, the conditioner can be set up from three unitsof single-phase H-bridge VSI, supplied from a common DCsource. Each single-phase VSI that is connected to theappropriate phase can be controlled and operatedindependently. Instead of using only one unit of tree-phaseconditioners, the implementation of three units ofsingle-phase H-bridge for three-phase conditioning systemswould be more beneficial for independent operation. It isvery much needed to deal with the imbalance compensationcondition that is likely to occur as an effect of imbalanceinjection current from inverters in each phase or as theeffect of imbalance loading in the grid side. It is generally

Fig. 2 Structure and configuration of proposed conditioner withPV plant in the electrical Grid


known that single-phase PV-inverter has become the trendin PV-inverter application [13].As shown in the figure, the PV-inverters produces an

output current ipv(t), in which is assumed be distorted byharmonics and contains reactive power. Thus, the currentipv(t) represents the produced active current ia(t), reactivecurrent ir(t) and harmonics contents ih(t), and is written as

ipv(t) = ia(t)+ ir(t)+ ih(t) (1)

Without conditioner, these power and all distortioncomponents will enter the grid as grid injection current

ig(t) = ipv(t) (2)

By installing a conditioner, the reactive and harmonicscomponents are compensated and blocked to flow and enterthe grid, where these components are forced to flow intothe conditioner circuit. To enable this, the conditioner’scontroller should generate a reference current (iref) oppositeto these reactive and harmonics currents.The reference then is used to produce conditioner current ic

(t) as

ic(t) = − ir(t)+ ih(t)( )

(3)

Then, with the compensation current added, the currententering the grid becomes

ig(t) = ipv(t)+ ic(t) (4)

ig(t) = ia(t) (5)

It can be seen that with the addition of a conditioner, with acompensation current equivalent to the negative of theharmonics and reactive currents generated by the PV plant,the reactive and harmonics current components can beeliminated, leaving the active component to flow into the grid.Reference current is generated through a calculation

process using the instantaneous value of the PV plant’scurrent and grid voltage. The process extracts the active,reactive and harmonics components. Reference current iReffor harmonics and reactive compensation can then beobtained by subtracting the active current components fromthe distorted current ipv(t).

iRef (t) = − ipv(t)− ia(t)( )

(6)

2.1 Current component extraction andcompensator reference generation

In this study, a simple extraction method based on the Fourierdecomposition of components contained in a distorted currentwas applied. In this method, at first, the magnitudes of activeand reactive components are determined, and then thesevalues are multiplied by the unity synchronised-sine andcosines wave obtained from a synchronisation process. Theresults are the estimated value of the fundamental activeand reactive components of the distorted current. Theharmonics’ component then is obtained by subtracting theseestimated values from the distorted current. This method issimpler than other extraction methods, which are generally


www.ietdl.org
based on the use of instantaneous power theory [14, 15] andits extension [16], where the value of the power componentsmust firstly be calculated. The extraction method used isdescribed as follows.At the PCC, the voltage vpcc(t) and PV plant distorted
current ipv(t) can be written as

vpcc(t) =��2

√V Sin(vt) (7)

ipv(t) =∑1n=1

��2

√In Sin n vnt + un

( )(8)

In these equations, PCC voltage is assumed as ideal, subscriptn represents harmonics order and θn denotes the phase-shiftbetween fundamental and the nth order harmonicscomponent.Distorted current equation can be composed as

ipv(t) =��2

√I1 Cos u1

( )Sin v1t

( )+

��2

√I1 Sin u1

( )Cos v1t

( )

+∑1n=2

��2


( )(9)

The first and second terms at the right side of (9) are thefundamental components of active and reactive currentcontained in ipv(t), respectively. They are sinusoidal waveswhich oscillate as the power frequency ω1. Their peakvalues are

Ia1 =��2

√I1Cos u1

( )(10)

Ir1 =��2

√I1 Sin u1

( )(11)

Equations (10) and (11) indicate that Ia1 and Ir1 are DCquantities, in which their magnitudes depend on thephase-shift θ1. These quantities represent the magnitude ofthe fundamental active and reactive components containedin the output current of the PV plant. By using thisnotation, (8) can also be expressed as

ipv(t) = Ia1 Sin v1t( )+ Ir1 Cos v1t

( )

+∑1n=2

��2


( )(12)

Ia1 and Ir1 in (12) can be written as a separate DC componentin ipv(t) equation by multiplying (12) with factor 2 Sinω1t and2 Cosω1t, respectively, to give

2 Sin v1t( )

.ipv(t) = Ia1 − Ia1 Cos 2v1t( )+ Ir1Sin 2v1t

( )

+ 2Sin v1t( )

.∑1n=2

��2

√In Sin nvnt + un

( )

(13)

2 Cos v1t( )

.ipv(t) = Ia1 Sin 2v1t( )+ Ir1 + Ir1 Cos 2v1t

( )

+ 2 Cos v1t( )

.∑1n=2

��2

√In Sin nvnt + un

( )

(14)

As referred to (13) and (14), it can be said that by multiplying


the distorted current ipv(t) with 2 Sinω1t and 2 Cosω1t, theresulting component will consist of DC and oscillatingcomponent (AC). DC components which are contained in asignal can then be separated physically by using filteringmethod, that is, low-pass filter (LPF).Passing the signals as expressed in left side of (13) and (14)

through an LPF functions to extract the peak value offundamental active and reactive current componentIa1 and Ir1( )

from the PV plant’s output current (ipv(t)).Meanwhile, as referred to the first and the second term of(12), by multiplying these filtering results with Sinω1t andCosω1t, respectively, then the estimated instantaneous valueof the fundamental active and reactive current componentsof the ipv(t) can be obtained. Fig. 3 shows the structure ofthe extraction and estimation process. The figure alsoillustrates the current component waveforms resulting froma distorted input current and voltage. The shown waveformswere obtained through Matlab/SIMULINK simulation.When the instantaneous values of the active and reactive

are known, the harmonics component in ipv(t) can becalculated as

ih(t) = ipv(t)− Ia1 Sin v1t( )+ Ir1 Cos v1t

( )( )(15)

In the physical implementation, although (15) explicitly listsonly the harmonics components, with a subtraction operationat the right side of the equation, the obtained componentswould not only be the harmonics but also inter-harmonicscomponents that are in ipv(t).Finally, with all values of active, reactive and harmonics

current components having been obtained, thecompensation type can be determined. The type ofcompensation is reflected from the form of signal referenceused: if the harmonics and reactive power components areto be compensated, the reference current should becalculated according to (6), whereas if the compensation isonly for harmonics components, the reference becomes

iRef (t) = − ipv(t)− ia(t)− ir(t)( )

(16)

Determining the type of compensation that must be handledby the conditioner is an important step in the operation ofconditioners, especially in the reactive power compensationto maintain the voltage stability. With current controloperation, and the use of compensation current reference as(6) and (16), the conditioner can be programmed and usedas a harmonics compensator and as a reactive powercontroller during plant operation.

2.2 Grid voltage-synchronised sinusoidal-wavegeneration

Extraction of the active, reactive and harmonics component inthe distorted current equation that has been described requiresmultiplication signals of sine and cosine-wave with unityamplitude and in-phase with the grid voltage. In manyactive filters or grid-interactive converters, the signals areobtained from a synchronisation process using a phaselocked-loop (PLL). Various research and methods havebeen reported concerning to this topic, which were mainlyintended to create a fast and accurate structure, and useseffective calculation step.The common structure of PLL is the d–q

synchronous-reference frame based PLL (SRF-PLL) [17].This structure uses the abc to αβ and αβ to dq


Fig. 3 Reference current generation process

Synchronised sinusoidal-wave generation block based on lookup table was used to provide the sine and cosine wave for the current components extraction block.

www.ietdl.org

transformations to present the three-phase voltage system asorthogonal voltage (Vd and Vq). A regulator, usuallyPI-regulator, is used to track the obtained direct axis voltageVd. Grid frequency as the regulator’s output then is passed toan integration operator for obtaining the voltage’sphase-angle. This angle is further used to generate theinternal sine and cosine-wave as well as phase-angleinformation for the αβ to dq transformation. For applicationin single-phase, in which less information of the gridparameters are available, SRF-PLL needs an artificialorthogonal voltage that must be generated using certainmethods (transport delay, Hilbert transforms).SRF-PLL will work optimally in conditions of a pure

sinusoidal signal input, but in distorted conditions, thismethod will show poor performance [18, 19]. Therefore,different improvement methods have been proposed.Methods presented in [20] and [21] are two methods whichare able to reject the disturbance effects that worsen theperformance of synchronisation system. However, the useof various functions (transformation, regulator andintegrator) in conventional SRF-PLL and in itsimprovement, as well as the process of generating artificialorthogonal system, may increase the operating timerequired, which may reduce performance system as a whole.In this study, a synchronisation method, in which

generation of sine and cosine-waves is done by referring toa lookup-table, was implemented. It was done withoutcalculation of grid frequency to obtain instantaneous phaseangle of the input signal. Through seeing the benefit fromdiscrete system that works in sampling by sampling, thegenerated value of the synchronised sine-wave in eachsampling time was fetched from a lookup-table thatcontained series of sine-function values. The elements wereidentified and called by an index number. By using thismethod, a sine-wave can be generated faster because it isnot derived from a calculation process. It is also suitable tobe applied on hardware controller (DSP) that usually comeswith an accompanying memory for locating the table. Wave


generation process does not require a PI-regulator,transformation and artificial orthogonal voltage, and thesine-wave obtained is a pure sinusoidal, which is notaffected by distortion that may accompany the input signal.This synchronisation method works based on the principle

that if the PCC voltage signal equation is multiplied by a sineor cosine-wave, it will produce an equation that consists of aDC and oscillations component. It has also been shown thatthe magnitude of the DC component obtained isproportional to the phase-shift of the multiplied waves.Thus, an algorithm can be created which will keep the DCcomponent always within a pre-determined limit byupdating the value of sine or cosine-wave. A part in Fig. 3shows the structure of the method, in which thecosine-wave is used as multiplication factor.In this method, the algorithm will start in an instance when

the input signal crosses the zero point at the first time. Anindex number which represents the element of a lookup-tablewill be created, which then is used to call the correspondingcosines-function value which resides in the table. This valueis taken as the internal synchronised cosine-wave in thatsampling period. In the next sampling period, the indexnumber is increased by one and the next value of thecosine-function in the table is called as the next value of thesynchronised cosine-wave. If a jump happens in the inputsignal that causes the multiplication result beyond thepre-determined limit, the algorithm will recalculate andcorrect the value of the index so that multiplication result isback within the limit. This process continues until the nextzero crossing is detected. The sine-wave is generated with thesame way, except that the called table’s element is theelement with index number corresponding to 90° shifting, assine and cosine-wave phase-shifting.

2.3 DC-voltage for the VSI

In order to work properly, VSI must be supplied by a DCvoltage with magnitude greater than peak value of the grid


www.ietdl.org
voltage. The voltage can be provided from an externalregulated-DC source or from the DC capacitor bank withself-regulated mechanism.In operation with self-regulated DC capacitor, during
conditioner operation, the energy flow occurs in bothdirections; between the capacitor and the grid through theVSI. The capacitor will experience a charging-dischargingcondition, and as a result, the capacitor’s voltage willfluctuate, following the grid frequency. This fluctuationdepends on the amount of the current and capacity of thecapacitor

Vc(t) =1

CDC

∫idc dt (17)

For an intended average voltage, the amount of currentflowing into and out from the capacitor can be controlledusing a current control, whereas the capacity of thecapacitor must be designed properly so that no excessivevoltage fluctuations would occur. A method of determiningthe capacity of capacitor for grid-interactive VSI converterbased on the principle of energy-flow balance has beenpresented in [22].Although the capacitor in is charging-discharging

condition, a portion of power will be released as operationlosses. As the voltage and current at the PCC are denotedas vpcc(t) and ipv(t), the losses can be obtained based onpower flow calculation at that point. The instantaneouspower at the PCC is calculated as

Ppcc(t) = vpcc(t).ipv(t)

Ppcc(t) = 2VI1 Sin2(v1t) Cosu1

+ 2 VI1 Sin(v1t) Cos(v1t) Sinu1

+��2

√V Sin(v1t).

∑1n=2

��2

√In Sin nvnt + un

( )

(18)

Ppcc(t) = Ppcc−a(t)+ Ppcc−r(t)+ Ppcc−h(t) (19)

where PPCC−a(t) is active power, PPCC−r(t) is reactive powerand PPCC−h(t) represents the harmonics power. If losses in thepower conditioner are assumed to be supplied only from thefundamental component, that is, a portion of firstcomponent at the right side of (19), then it can beexpressed as a current il(t) in-phase with the PCC voltage.The losses then can be calculated as

Pcl(t) =��2

√V Sin v1t

( ).

��2

√Il Sin v1t

( )= VIl − VIl Cos 2v1t

( ) (20)

Pcl(t) = �Pcl + Pcl (21)

�Pcl and Pcl are the DC and oscillating components flowinginto and out from the conditioner circuit, which causes thecapacitor voltage (Vc0) to change as ΔVc0. In the capacitorwith capacitance CDC, the energy change because of thispower flow is calculated as

DE = 1

2CDC Vc0 + DVc0

( )2−V 2c0

( )(22)

From the above equations, �Pcl correlates with component


(1/2)CDC(Vc0 + ΔVc0)2, whereas the Pcl correlates to

(1/2)CDCV2c0.

With the assumption that ΔVc0 is small enough comparedwith Vc0, then (22) can be represented as

DE = CDC Vc0 DVc0 (23)

This leads to the notion that the current losses component��2

√Il Sin v1t

( )is absorbed dominantly to become �Pcl. With

regard to that, the relation between the capacitor voltageand energy change for one period T can be formulated as

DE = �Pcl.T (24)

Then, from (19) and (22), the following equations are derived

�Pcl.T = CDC Vc0 DVc0

DVc0 =�Pcl.T

CDCVc0= V .Il.T

CDCVc0Sinv1t

(25)

The last equation presents the relation between voltage drop,average voltage and capacity of capacitor, effective value ofgrid voltage and the operating current. This equation is thenused to design the capacity of capacitor and DC buscontroller for the conditioner.As the DC bus controller, the proportional plus integral (PI)

regulator is used. Constants KP and KI are determined basedon (25). The input signals for the controller are the desiredDC-bus voltage value as the controller reference (Vdc−ref)and the measured value of DC-bus. The output of regulatoris a signal representing the amount of current needed tokeep the DC-bus voltage equal to Vdc−ref. After beingmultiplied with Sin(ω1t), this signal becomes the referencefor current control. The presence of this current implies theaddition of total current that flows in and out from theconditioner circuit. Thus, the reference current for currentcontrol as (6) and (16) must be corrected by consideringthis portion of losses current.

2.4 LCL filter and current controller

After being generated within the VSI circuit, thecompensation current will be further passed to an outputfilter. Filter is used because the current generated by VSIwould still have considerable ripples. Besides that,impedance of the filter and isolation transformers will limitthe current out and enter the VSI circuit.Traditionally, output filtering of a VSI is done using L-type

or LC-type filter, but along with the increasing size of theconditioner, the use of these types has become inefficient.In this study, the VSI circuit was paired with an LCL-typefilter. This filter has more advantage in size andperformance compared with the traditional ones [23, 24]. Itsonly weakness is the prospect of resonance if the filteroperates around the resonance frequency. However,resonance can actually be reduced by using dampingresistors [25].Injection of compensation current into the grid is done

using current control mechanism. In this study, a linearpulse-width modulation with PI-regulator (PWM-PI) currentcontrol was used. Configuration of all transfer functions ofthe current control system gives the close and open loop


www.ietdl.org
transfer functions as
GCL(s)

= GPI−CC(s).GPWM(s).GVSI(s).GLCL(s)

1+ GPI−CC(s).GPWM(s).GVSI(s).GLCL(s).HAD(s).HCT(s)

(26)

GOL(s) = GPI−CC(s).GPWM(s).GVSI(s).GLCL(s).HAD(s).HCT(s)

(27)

These transfer functions are then used to design the current

Fig. 4 Controller structure of the conditioner

a Structure of PWM-PI current controlb Overall diagram of conditioner control


controller parameter of the conditioner. The current controlblock diagram is shown in Fig. 4a, whereas the overallcontrol system of the conditioner is shown in Fig. 4b.

3 Hardware implementation and experimentresults

To show the workability of the scheme, an experimental setupwhich consisted of conditioner prototype and a PV plant wasconstructed and operated. The diagram and photograph of thesetup are shown in Fig. 5.The plant consisted of 24 PV modules, each with 24 V DC

output. Modules were configured as an array to give a total of5 kW-peak array installed capacity. The array was thendivided into three sub-arrays, feeding three unitssingle-phase PV-inverters of 700 W. Each PV-inverter wasconnected to each phase of the grid to form three-phaseconnection. The output side of PV-inverters was fed to thePCC, where the grid integration, measurement andconditioner prototype were connected. The TMS320F28335DSP from Texas Instrument was used to implement theconditioner controller. The control algorithm was createdusing C language program and was loaded into the DSPusing integrated development environment (IDE) softwarecode composer studio.The signals for control algorithm were: PV plant current

Ipv(t),conditioner output Ic(t), PCC voltage Vpv(t) and DCbus voltage VDC(t). The signals were sensed usinghall-effect sensors and then pre-filtered with anti-aliasingfilter. The signals were then fed to the ADC to be processedfor finally generating the gating signals for the VSI switches.A 300 MHz four-channel oscilloscope was used to measurethe value and display the waveform of currents and voltagesfor the purposes of analysis. For current measurement,voltage probes of the oscilloscope were used to measure thevoltage at the output of the sensor that changed themeasured current into voltage with ratio of 1:4 V/A,whereas the voltage was measured by sensors with a ratioof 1.767 mV/V.The experiment was conducted as follows: In the first step,

the inverter in PV plant was operated using the array powerwithout connecting the conditioner. The grid injectioncurrent and voltage at the PCC were then measured anddisplayed on the oscilloscope. Power level operation,represented by the injection current, was noted and itswave-shape was then recorded as both picture andspreadsheet format data. In the second step, the gridinjection current and voltage at the PCC with the samegenerated power were recorded but with conditioner wasoperated to generate compensation current. The injectioncurrent as the sum of PV plant and conditioner current thenwere measured, displayed and recorded, also both in pictureand spreadsheet data format. The obtained data was thenprocessed using a Matlab tool to calculate the grid injectioncurrent distortion of both operation steps. The distortionlevel of the current was then presented as THD and theharmonics spectrum contained was analysed and compared.

Fig. 6a shows the measurement results, whereas the PVplant’s inverter was operated at the level of 1.8 A, or 60%of the inverter rating. The PCC voltage and generatedcurrent appeared as sinusoidal and distorted wave-shape.The grid injection current and conditioner current were alsoobserved. As observed, the PV plant’s current became theinjection current into the grid, indicated by similarity of thewaveform of PV plant’s current and the grid injection


Fig. 5 Experiment setup

a Schematicb Photograph: (1) PCC; (2) Cap. Bank; (3) VSI; (4) LCL Filter; (5) isolation transformer; (6) gate driver and interface; (7) DSP TMS320F28335; (8) sensors; and(9) computer with CCS-IDE

www.ietdl.org

current. In this condition, the conditioner was not inoperation, indicated by its zero current conditions.Fast-Fourier transform (FFT) analysis was then performedto obtain the THD value of the injection current. As seen,the injection current into the grid contained harmonicsdistortion of 12.21% THD. Harmonics spectrum showedthat the content of harmonics was dominated by thelow-order harmonics with high magnitude.

Fig. 6b shows the results of measurements in the samecondition, in which the PV plant’s inverter was operated atthe level of 1.8 A; however, in this condition, theconditioner was operated. Based on the PV plant currentand voltage at the PCC, the conditioner generatedcompensation current and caused the waveform of injectioncurrent into the grid to change and be repaired. Thewaveform also became more symmetrical and sinusoidal.The FFT analysis showed that the harmonics distortionlevel was 7.9% THD. This value indicated a reduction of35.29% distortion compared with the condition if theconditioner was not operated. The harmonics spectrum alsoindicated a reduction in magnitude of low-order harmonicscompared with the previous case.Measurement results in operation of 1.1 A is shown in

Fig. 6c. In this condition, conditioner operation causes adistortion reduction of 17.14%; PV plant produced a currentwith 16.80% THD, and after being compensated, thedistortion became 13.92% THD on the grid injection current.Further, to evaluate the effect of conditioner operation at

various power level operations of PV-inverter, a number ofmeasurements were performed at different levels ofPV-inverter power conversion. As seen in Fig. 7a, themeasurement data of PV-inverter’s output currentdistortions were in range from 0.4 to 1.9 A. At each level, itwas observed that the reduction of distortion occurred whenthe conditioner was operated. This was an indication thatthe improvements with the use of conditioner affected theoverall levels of PV-inverter operation. Significantimprovement occurred when the PV-inverter was operated


at low power conversion. This was because in these levels,the wave distortion was very high. In contrary, at highconversion levels, the output current waveform tended to bemore sinusoidal so that the effect of connecting theconditioner would not be too significant.To illustrate the significance and benefit of the distorted

waveform reduction by using this conditioner in the powersystem, the measurement data is presented as trend linecurves of the PV-inverter’s conversion levels againstresulted distortion in conditions with and withoutcompensation, shown in Fig. 7b. A maximum limit ofcurrent distortion from a source that allowed to beconnected was also drawn in the figure. As outlined inIEEE standard 519 [12], the value denoted 20% THD. Thecrossing point between each trend line curve (with andwithout conditioner) with the allowable-limit linecorresponded to minimum power level operation, where thePV-inverter met the required power quality standard.From the graphs, it can be seen that the use of conditioner

caused a shift of the minimum power operation to a lowerlimit than when the inverter was operated withoutconditioner. If it was required that the PV-inverter couldonly be connected if the power quality standard was met,then the shift from minimum limit power level to a lowervalue would mean improving the utility factor of aninverter. From the power-generation aspect, it means morepower can be produced. This condition would become moresignificant when the number of inverters used is increased,as can be seen at a PV plant with a large number ofPV-inverters.

4 Conclusion

This paper has presented a method to improve the powerquality of a PV plant, using a generation-side conditionerunit working in a feed-forward scheme. The proposedmethod’s workability and performance have also been


Fig. 7 Effects of conditioner installation on PV plant

a Reduction of THD at the injection current in various power conversion levelsb Effect of installing a conditioner at the PV plant; the minimum level of PV inverter power operation to fulfil a certain power quality standard shifted to a lowervalue

Fig. 6 Experiment results

a Waveform of PCC voltage, PV-inverter’s, injection current, compensation current and FFT analysis of the grid injection current without conditioner during theplant operated at 1.8 AbWaveform of PCC voltage, PV-inverter’s, injection current, compensation current and FFT analysis of the grid injection current with conditioner during the plantoperated at 1.8 AcWaveform of PCC voltage, PV-inverter’s, injection current, compensation current and FFT analysis of the PV plant current and grid injection current during theplant operated at 1.1 A

www.ietdl.org



www.ietdl.org
demonstrated. The proposed conditioner control method,which is a combination of a lookup table-based ofsynchronised sinusoidal-wave generation and method forextracting the fundamental components of active andreactive power, as well as the harmonics from a distortedcurrent, possesses a number of advantages, because of itssimplicity in algorithm and resources. The main advantageof using this scheme is its ability to perform compensationand power quality improvement in wide range operation ofPV plant, which cannot be achieved by using otherimprovement methods which are commonly utilised at PVinverter units. In the experiments conducted, although anon-ideal compensation condition was obtained because ofthe delay factor, in general, the proposed method has shownan alternative to improve the quality of power generated bythe PV plant and reduce the distortion content of theinjection current to the grid. Non-ideal compensation couldbe reduced through the implementation of faster algorithmand utilisation of faster DSP equipment.
5 Acknowledgment

The authors would like to express gratitude to UniversitiTeknologi Malaysia (UTM) for providing financial supportunder Fundamental Research Grant Scheme (FRGS) (VotQ.J130000.2523.03H40).

6 References

1 International Energy Agency. 2012 Key World Energy Statistics. 2012.Available from: www.iea.org

2 Hegedus, S.S., Luque, A.: ‘Achievement and challenges os solarelectricity from photovoltaics’, in: Luque, A., Hegedus, S.S., (Eds.),‘Handbook of photovoltaic science and engineering’ (John Wiley &Sons, 2011, 2nd edn.), pp. 1–38

3 Thomson, M., Infield, D.G.: ‘Impact of widespread photovoltaicsgeneration on distribution systems’, IET Renew. Power Gener., 2007,1, (1), pp. 33–40

4 Hatziadoniu, C.J., Nikolov, E.N., Pourboghrat, F.: ‘Power conditionercontrol protection for distributed generators and storage’, IEEE Trans.Power Syst., 2003, 18, (1), pp. 83–90

5 Ko, S., Lee, S., Dehbonei, H., Nayar, C.V.: ‘A grid-connectedphotovoltaic system with direct coupled power quality control’. Proc.IEEE Industrial Electronics, IEC – 32nd Annual Conf., Paris, French,November 2006, pp. 5203–5208

6 Wu, T., Nien, H., Shen, C., Chen, T.: ‘A single-phase inverter system forPV power injection and active power filtering with nonlinear inductorconsideration’, IEEE Trans. Ind. Appl., 2005, 41, (4), pp. 1075–1083

7 Takeshita, T., Matsui, N.: ‘Current waveform control of PWM convertersystem for harmonic suppression on distribution system’, IEEE Trans.Ind. Electron., 2003, 50, (6), pp. 1134–1139

8 Palanivel, P., Dash, S.S.: ‘Analysis of THD and output voltageperformance for cascaded multilevel inverter using carrier pulse widthmodulation techniques’, IET Power Electron., 2011, 4, (8), pp. 951–958


9 Altin, N., Ozdemir, S.: ‘Three-phase three-level grid interactive inverterwith fuzzy-logic based maximum power point tracking controller’,Energy Convers. Manage., 2013, 69, pp. 17–26

10 Milanés-Montero, M.I., Miñambres-Marcos, V., Guerrero-Martínez, M.Á., Romero-Cadaval, E.: ‘Cooperative converter for improving theperformance of grid-connected photovoltaic power plants’, IETRenew. Power Gener., 2013, 7, (2), pp. 110–117

11 Macken, K.J.P., Vanthournout, K., Van den Keybus, J., Deconinck, G.,Belmans, R.J.M.: ‘Distributed control of renewable generation unitswith integrated active-filter’, IEEE Trans. Power Electron., 2004, 19,(5), pp. 1353–1360

12 IEEE Std 519-1992: ‘IEEE recommended practices and requirements forharmonic control in electrical power systems’. 1992

13 Blaabjerg, F., Chen, Z., Kjaer, S.B.: ‘Power electronics as efficientinterface in dispersed power generation systems’, IEEE Trans. PowerDeliv., 2004, 19, (5), pp. 1184–1194

14 Akagi, H., Kanazawa, Y., Nabae, A.: ‘Instantaneous reactive powercompensators comprising switching devices without energy storagecomponents’, IEEE Trans. Ind. Appl., 1983, I, (3), pp. 625–630

15 Akagi, H., Watanabe, E.H., Aredes, M.: ‘Instantaneous power theoryand applications to power conditioning’ (Hoboken, New Jersey, 2007)

16 Tan, P.C., Jusoh, A., Salam, Z.: ‘A single-phase hybrid active-powerfilter using extension p–q theorem for photovoltaic application’. Proc.IEEE International Conf. on Power Electronics and Drives Systems,PEDS, Kuala Lumpur, Malaysia, 2005, pp. 1250–1255

17 Timbus, A., Liserre, M., Teodorescu, R., Blaabjerg, F.: ‘Synchronizationmethods for three phase distributed power generation systems. anoverview and evaluation’. Proc. IEEE 36th Conf. on PowerElectronics Specialists, Recife, Brazil, 2005, pp. 2474–2481

18 Carugati, I., Maestri, S., Donato, P.G., Carrica, D., Benedetti, M.:‘Variable sampling period filter PLL for distorted three-phasesystems’, IEEE Trans. Power Electron., 2012, 27, (1), pp. 321–330

19 Guo, X.-Q., Wu, W.Y.: ‘Simple synchronisation technique forthree-phase grid-connected distributed generation systems’, IETRenew. Power Gener., 2013, 7, (1), pp. 55–62

20 Garcia-reteguit, R., Gonza, S.A., Funes, M.A., Maestrit, S.:‘Implementation of a novel synchronization method using slidingGoertzel DFT’. Proc. IEEE Int. Symp. on Intelligent SignalProcessing, Alcala de Hanare, October 2007, pp. 1–5

21 Rodríguez, P.Ç., Teodorescu, R.Ë., Candela, I.Ç., Timbus, A.V.Ë.,Liserre, M.Ï., Blaabjerg, F.Ë.: ‘New positive-sequence voltage detectorfor grid synchronization of power converters under faulty gridconditions’. Proc. IEEE 37th Power Electronics Specialists Conf., Jun2006, pp. 1–7

22 Wu, J.-C., Jou, H.-L.: ‘Simplified control method for the single-phaseactive-power filter’. Proc. Electric Power Applications, May 1996,pp. 219–224

23 Routimo, M., Tuusa, H.: ‘LCL type supply filter for active power filter –comparison of an active and a passive method for resonance damping’.Proc. IEEE Power Electronics Specialists Conf., Orlando, FL, June2007, pp. 2939–2945

24 Tong, Y., Tang, F., Chen, Y., Zhou, F., Jin, X.: ‘Design algorithm ofgrid-side LCL-filter for three-phase voltage source PWM rectifier’.Proc. IEEE Power and Energy Society General Meeting – Conversionand Delivery of Electrical Energy in the 21st Century, Pittsburgh, PA,Jul 2008, pp. 1–6

25 Zhao, W., Chen, G.: ‘Comparison of active and passive dampingmethods for application in high power active power filter withLCL-filter’. Proc. Int. Conf. on Sustainable Power Generation andSupply, Nanjing, China, April 2009, pp. 1–6


Photovoltaic plant with reduced output current harmonics using generation-side active power...

Documents

Transcript of Photovoltaic plant with reduced output current harmonics using generation-side active power...