Photovoltaic Maximum PowerPoint Tracker with ZeroOscillation and Adaptive StepbyFrancisco PazIng., Universidad Nacional del Comahue, 2012A THESIS SUBMITTED IN PARTIAL FULFILLMENT OFTHE REQUIREMENTS FOR THE DEGREE OFMASTER OF APPLIED SCIENCEinThe Faculty of Graduate and Postdoctoral Studies(Electrical & Computer Engineering)THE UNIVERSITY OF BRITISH COLUMBIA(Vancouver)August 2014c© Francisco Paz 2014AbstractMaximum Power Point Tracking (MPPT) strategies in Photovoltaic (PV) systems ensure ef-ficient utilization of PV arrays. Among different strategies, the Perturb and Observe (P&O)algorithm has gained wide popularity due to its intuitive nature and simple implementation.However, such simplicity in P&O introduces two inherent issues, an artificial perturbationthat creates losses in steady-state operation and a limited ability to track transients in chang-ing environmental conditions. This work develops and discusses in detail an MPPT algorithmwith zero oscillation and slope tracking to address those technical challenges. The strategycombines three techniques to improve steady-state behavior and transient operation: 1) idleoperation on the Maximum Power Point (MPP), 2) identification of the irradiance changethrough a natural perturbation and 3) a simple multi-level adaptive tracking step. Two keyelements, which form the foundation of the proposed solution, are investigated: the suppres-sion of the artificial perturbation at the MPP and the indirect identification of irradiancechange through a current-monitoring algorithm which acts as a natural perturbation. TheZero-oscillation, Adaptive step Perturb and Observe (ZA-P&O) MPPT strategy builds onthese mechanisms to identify relevant information and produce efficiency gains. As a result,the combined techniques achieve superior overall performance while maintaining simplicity ofimplementation. Simulations and experimental results are provided to validate the proposedstrategy and illustrate its behavior in steady state and transient operation.iiPrefaceThis work is based on research performed at the Electrical and Computer Engineering de-partment of the University of British Columbia by Francisco Paz, under the supervision ofDr. Martin Ordonez.A first version of this work was presented in the 4th IEEE International Symposium onPower Electronics for Distributed Generation (PEDG), 2013 [1].An extended version of this work was published in IEEE Transaction on Industrial Elec-tronics [2].As first author of the above-mentioned publications, the author of this thesis developedthe theoretical concepts and wrote the manuscripts, receiving advice and technical supportfrom Dr. Martin Ordonez, and developed simulation and experimental platforms, receivingcontributions from Dr. Ordonez’s research team.iiiTable of ContentsAbstract . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . iiPreface . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . iiiTable of Contents . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . ivList of Figures . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . viAcknowledgements . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . viiiDedication . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . ix1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 11.1 Motivation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 11.2 Literature Review . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 21.2.1 Constant Relationship Algorithms . . . . . . . . . . . . . . . . . . . 31.2.2 Hill Climbing Algorithms . . . . . . . . . . . . . . . . . . . . . . . . 41.2.3 Ripple Correlation Control . . . . . . . . . . . . . . . . . . . . . . . 61.2.4 Model Based Algorithms . . . . . . . . . . . . . . . . . . . . . . . . 61.2.5 Heuristic-Based Algorithms and Other Special Algorithms . . . . . . 61.2.6 Summary . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 71.3 Contribution of the Work . . . . . . . . . . . . . . . . . . . . . . . . . . . . 71.4 Thesis Outline . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 9ivTable of Contents2 System Background and Energy Harvesting Losses In PV Systems . . . 122.1 PV Panel Background . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 132.2 Basic P&O Algorithm Background . . . . . . . . . . . . . . . . . . . . . . . 162.3 Steady-State Power Losses Estimation . . . . . . . . . . . . . . . . . . . . . 182.4 Irradiance Transient Power Losses Estimation . . . . . . . . . . . . . . . . . 212.5 Control System . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 242.6 Summary . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 253 Zero Oscillation Adaptive Perturb and Observe MPPT . . . . . . . . . . 263.1 Idle Mode Operation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 273.2 Perturbation Direction and Magnitude Estimation . . . . . . . . . . . . . . 283.3 Flowchart of the ZA-P&O MPPT . . . . . . . . . . . . . . . . . . . . . . . 293.4 Summary . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 304 Simulations . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 325 Experimental Results . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 396 Conclusions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 466.1 Summary . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 466.2 Future Work . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 47Bibliography . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 48vList of Figures1.1 Conceptual representation of the ZA-P&O MPPT . . . . . . . . . . . . . . . 91.2 Issues with P&O technique . . . . . . . . . . . . . . . . . . . . . . . . . . . . 92.1 Block diagram of the PV system . . . . . . . . . . . . . . . . . . . . . . . . . 122.2 PV cell model . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 142.3 VI/PV curves . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 152.4 Fill factor (FF) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 162.5 Flowchart of the basic P&O MPPT algorithm. . . . . . . . . . . . . . . . . . 172.6 Tracking losses on different operating conditions . . . . . . . . . . . . . . . . 182.7 Tracking efficiency for a transient . . . . . . . . . . . . . . . . . . . . . . . . 233.1 Idle mode operation illustrated . . . . . . . . . . . . . . . . . . . . . . . . . 273.2 Irradiance change identification . . . . . . . . . . . . . . . . . . . . . . . . . 303.3 Flow chart ZA-P&O . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 314.1 Trapezoidal irradiance profile . . . . . . . . . . . . . . . . . . . . . . . . . . 324.2 Simulations standard P&O with transient on raising edge . . . . . . . . . . . 334.3 Simulations standard P&O with transient on falling edge . . . . . . . . . . . 344.4 Simulations ZA-P&O with same transient . . . . . . . . . . . . . . . . . . . 354.5 Efficiency of the PV panel for the ZA-P&O vs. standard P&O . . . . . . . . 374.6 Transient analysis in the phase V-I plane . . . . . . . . . . . . . . . . . . . . 38viLIST OF FIGURES5.1 Picture of the experimental set-up . . . . . . . . . . . . . . . . . . . . . . . . 395.2 Exp. capture of the standard P&O on rising edge . . . . . . . . . . . . . . . 405.3 Exp. capture of the standard P&O on falling edge . . . . . . . . . . . . . . . 415.4 Exp. capture of the ZA-P&O with the same transient . . . . . . . . . . . . . 415.5 Steep/gradual trapezoidal irradiance profile 1 . . . . . . . . . . . . . . . . . 435.6 Steep/gradual trapezoidal irradiance profile 2 . . . . . . . . . . . . . . . . . 435.7 Exp. capture of the ZA-P&O with steep/gradual transient 1 . . . . . . . . . 445.8 Exp. capture of the ZA-P&O with steep/gradual transient 2 . . . . . . . . . 44viiAcknowledgementsI would like to thank Dr. Martin Ordonez for his supervision. He established an outstandingexample of leadership, mentoring, research, supervision and instruction.I would like to thank my parents Cesar and Rosa, my sisters Carmen, Dolores and Carolaand all of my family for their support, both moral and financial, during all my life and inparticular during the past two years. Their constant presence in my life led me to this place.I feel specially grateful for my girlfriend Celeste who supported me during this two yearsabroad with her constant encouragement. Her support in my decision to go for graduatestudies, even when it would mean a long time apart, made this work possible. My specialthanks and love go to her.I would like to thank all the members of Dr. Ordonez’s research group during the timeof my Master’s program (in order of position around the table): Matias Anun, Juan Galvez,Peter Ksiazek, Robert Cove, Jason Forbes, Ignacio Galiano, Navid Shafiei, Rafael Pen˜a-Alzola and Ion Isbasescu. Their comments, discussion, suggestions, corrections, help andsupport made this work possible and always pushed me forward. If I don’t take anything butthe time shared with them, I would already be a better person.I would also like to thank the University of British Columbia, the Faculty of Graduateand Postdoctoral Studies and the Electrical and Computer Engineering Department for theopportunity and support received during all this time. To all the faculty, administration andstaff I feel grateful.Finally, thank you for taking the time to read my thesis.viiiTo all those I love.ixChapter 1Introduction1.1 MotivationWith the rising costs of oil and the concern surrounding global warming, the use of photo-voltaic (PV) energy has been growing an average of 48% annually between the years 2000and 2012 [3], with an cumulative a growth above 7,300%. PV array installations have beenpopularized in a variety of application and power levels, and can provide electricity either toisolated locations or contributing to the global grid through grid-tie inverters. Developmentsin power electronics topologies and control schemes that target PV applications present aspecial interest. In particular, a key element of the power conversion system is the MaximumPower Point Tracking (MPPT) algorithm. This control algorithm is responsible for findingthe operating condition of the PV panel the yields the maximum possible power (known asMaximum Power Point, MPP). The MPP depends on the characteristics of the particularPV panel as well as the environmental conditions, such as the temperature (T ) and the ir-radiance (G). On the one hand, the inherent characteristics of the PV panel change veryslowly (mainly by aging or breaking) so they are not considered a concern in most MPPTalgorithm; on the other hand, T and G change constantly just by the day/night cycle as wellas by additional external events such as wind or shadows.The most popular MPPT algorithms in industry are implemented by measuring the PVpanel current and voltage and using that information to look for the MPP. The use of T and11.2. Literature ReviewG sensors is usually avoided due to the complexity and cost of such measurements. Withthis limited information, the changes in T and G present a challenge for the algorithms.Once the MPPT algorithm has found the MPP, at the current T and G, the algorithmkeeps the scanning perturbation in order to be able to identify a change in the PV charac-teristics. Most MPPT algorithms will make the PV panel work 50% the time outside theoptimal condition causing some of the available energy not to be transferred to the load.A transient in G occurs when the amount of light that arrives to the PV panel changes,therefore changing the characteristic curve and available power. Transients are produced bythe natural day/night cycle (which produces slow transients) or by artificial object obstructingthe path of the light (fast transients). During this periods, the MPPT algorithm needs tofollow the MPP as the transient occurs and find it quickly once the transient has beenextinguished. The industry-standard algorithms do not provide such feature without theneed of adding a sensor to measure G.Although the principal issues of the industry-standard MPPT algorithms are well docu-mented in the literature, the impact of this effects is not treated with enough depth and asimple solution that mitigates this issues is not presented.The work presented in this thesis provides a new improvement to MPPT techniques,which features a simple implementation and includes advanced characteristics such as theelimination of the steady state oscillation and the accurate tracking of transients.1.2 Literature ReviewMaximum Power Point Tracking (MPPT) for Photovoltaic (PV) applications is an activeresearch in both academia and industry with more than 250 journal papers and 1600 confer-ence papers available in IEEEXplore, and several commercial products in the market haveincluded active MPPT algorithms. The existing literature discusses a variety of MPPT tech-21.2. Literature Reviewniques, circuits, applications and issues that help minimize the energy lost by not transferringit to the load but some issues remain unsolved.As pointed out by the early works in the field [4–9], the MPPT process is based onfinding the load impedance (as seen from the PV panel terminals) that matches the equivalentimpedance of the PV panel. Before these early works, direct connection from the PV panel tothe load was the common method, without the use of any impedance adaptation device. Theeconomical advantage of using MPPT algorithms for PV applications was discussed in [10],showing that the energy transferred was highly incremented using power converters with thistechnology.MPPT algorithms have a wide range of characteristics depending on the core application,ranging from very simple implementation to highly complex, from low to high precision.Simple algorithms are compact and can be implemented directly with analog circuits withvery low consumption while high complexity algorithms are implemented using costly DSPfor very large power applications. The principal contributions as well as the latest advancesto the field are presented in the following paragraphs, highlighting the known issues that givea place to the present work.1.2.1 Constant Relationship AlgorithmsSome very simple algorithms, such as the Constant Voltage (CV), Fractional Open-CircuitVoltage (FOCV) [11–17] and Fractional Short-Circuit Current (FSCI) [4, 14–18] have thesimplest implementation. The algorithm works by assuming a constant relationship betweenthe open-circuit voltage (Voc) or short-circuit current (Isc) and the MPP voltage (Vmpp) orcurrent (Impp). Such relationship is usually between 0.7 and 0.8 for a silicon solar cell.The algorithm then proceed to periodically measure the Voc (or Isc) and computing thecorresponding optimal operating point. The precision of this algorithm is low comparedwith other advanced techniques and has the disadvantage of having to open the circuit31.2. Literature Reviewat the terminals of the PV panel in order to measure Voc periodically (or short circuit it tomeasure the Isc), which disturbs the operation of the load [14, 19]. Since the algorithms needsto constantly interrupt the operation of the PV panel these algorithms are not suitable toaccurately track changing environmental conditions, since they would require a high frequencyin the interruption conditions.These techniques have been successfully implemented in extremely low power circuits like[20] and [21] boosting the energy harvested for autonomous equipment.1.2.2 Hill Climbing AlgorithmsThe hill-climbing techniques are the most popular family of MPPT algorithms for industryapplications due to the good balance between complexity, accuracy [22] and reliability [23].These algorithms are based on scanning the Voltage-Current-Power characteristic curves insearch for a determined condition that signals the MPP. One of these MPPT algorithms isthe Incremental Conductance [22, 24–28], which periodically changes the operating point ofthe PV panel and compares the incremental conductance (∆I/∆V ) to the DC conductance(I/V ). These two variables have the same magnitude and opposite sign in the MPP, signalingthe match of the impedance between load and source.The other popular popular hill-climbing technique is the Perturb & Observe [27, 29–32]. This technique constantly changes the operating point and measures the change in theextracted power. It moves the operating point in the direction which causes the power toincrement.Both techniques offer a simple implementation but have some critical issues that makeroom for several improvements. The main disadvantages of this techniques have to do withthe constant oscillation around the MPP [33, 34] even in steady state, the confusions thathappen when the environmental conditions change [22, 28, 31, 35], the inability to adapt thetracking speed to different conditions and the presence of multiple local maxima. The basic41.2. Literature Reviewimplementation of these algorithms are unable to stop oscillating in steady state (once theMPP has been found) since they need to keep scanning in case the environmental conditionschange and a new MPP must be found. When the amount of light changes the amount ofpower yielded at a given operating point changes, the hill-climbing algorithms are unableto separate this change from the one produced by their own perturbation and often makewrong decisions leading to drifts from the MPP. For the basic algorithm the step-size of theperturbation is fixed, causing the scanning process to take very long when the MPP is faraway from the current operating point. In large arrays, global maxima and local maximacan be different due to the effect of shadows cast over the panel. Since this algorithms usemethods based on derivatives equal to zero, they are good only to find local maxima.Some modified algorithms have been presented based on the hill-climbing algorithms,however these usually target one or more drawbacks while penalizing the others. Someexamples of these modified strategies include the Adaptive Step variants of the P&O [29, 36–40] or InCond [41, 42], which apply some algorithm to change the step-size for fast trackingwhen away from the MPP and low oscillation when close to the maxima. This algorithmhowever, penalizes the detection of a change in the environmental conditions and can amplifythis errors due to the step-size adaptation algorithm. The Plane Division (PD) method[43, 44] improves the speed of the standard InCon algorithm by creating a forbidden region,where the MPP cannot be located based on historical environmental data on the site andmanufacturer information about the PV panel. Some modified versions of the hill-climbingalgorithms [45–50] are able to find global maxima by modifying the algorithm, usually addinga second tracking stage. In [49], in addition to a second scanning stage, each cell’s voltage ismeasure individually allowing for the identification of shaded cells. Recently, some strategieshave been developed to overcome the confusion that results from changes in G [51, 52].Opportunities to further improve steady-state behavior and provide an accurate trackingunder changing G exist and are explored in this work.51.2. Literature Review1.2.3 Ripple Correlation ControlSince power converters have switching elements, their action introduces fluctuations in theextracted power causing the voltage and current to oscillate in each cycle. Some MPPTalgorithms use this information in order to characterize the PV panel and find the MPP, thismethods are known as Ripple Correlation Control (RCC) algorithms [53–58]. This algorithmsprovide accurate results without further oscillation, but the sensors implied need to be veryprecise since the oscillations from the point of view of the PV panel are small and fast,since the oscillation is present as speeds higher than the switching frequency. The cost ofimplementation of such algorithms is very high.1.2.4 Model Based AlgorithmsA different approach to MPPT algorithms is used by the Model-Based algorithms. These al-gorithms use a complex model of the PV panel and the measurement of voltage, current (andsome times temperature and irradiance) to compute the optimal operating point. These usu-ally leads to a very high precision (since an accurate model is being used) and tracking speed(since only a few steps are needed) but the complexity of the hardware and software becomealso very high (since it usually implies the resolution of implicit non-linear equations) [19, 59–62] or the use of heuristic methods to estimate the curve is needed [63]. These algorithmsusually need a large amount of computational power per-step and a lot of information fromthe PV panel to be used.1.2.5 Heuristic-Based Algorithms and Other Special AlgorithmsHeuristic algorithm use principles derived from natural behavior to look for the MPP. Exam-ples of these MPPT algorithms include the ones based in Artificial Neural Networks [29, 64–67], Fuzzy Logic [38, 66, 68–73], Particle Swarm Optimization [73–78],61.3. Contribution of the WorkHeurist-based algorithms are very popular when dealing with local maxima produced bypartial shading [67, 79, 80]. Due to the training characteristics of these algorithms, a learningroutine can be implemented that adapts in order to find the global maximum in a limitednumber of steps. However, these methods usually require resources and energy that mightnot be applicable to small-scale systems.Finally, some applications allow for specific solutions that make use of the characteristicsof the power converter [81, 82], such as the One Cycle Control in [83–85] for Grid con-nected systems. These strategies deeply connect with the nature of the controller to use asa evaluation criteria for the MPP. These algorithms are very specific of the topology understudy.1.2.6 SummaryAs discussed in this literature review, there are a number of MPPT algorithms presentedin the literature in the past 30 years. Many applications have benefited with the incrementin processing power available and new techniques became available. However, most MPPTalgorithms reported focus on constant environmental conditions and many challenges remainopen. In particular an simple algorithm, comparable with hill-climbing, that is able toremove the oscillation in steady-state without compromising the tracking speed and accuracyand including a dynamical environmental conditions tracking is lacking in the literature.These technical challenges are addressed in this thesis and a zero oscillation adaptive trackis investigated.1.3 Contribution of the WorkThe work presented in this thesis introduces a valuable MPPT strategy that contributes tomitigate the characteristic extraction losses present in PV electrical systems, as well as theo-71.3. Contribution of the Workretical contributions to the quantization of the losses in steady state and transient conditions.The presented technique is simple and requires a small increment in the complexity of thecode to implement it, making it a practical option for industrial applications. The followinglist summarizes the contributions of this work:• The theory for a MPPT strategy referred to as Zero-oscillation, Adaptive-step Perturband Observe (ZA-P&O), which reduces losses in steady-state and improves trackingunder speed-varying changes in G is developed. The losses are reduced by suppressingthe artificial perturbation around the MPP in steady-state, and are made possible byindirectly estimating a change in G through the natural perturbation introduced bythe error on a Proportional-Integral (PI) controller and the change in current in theoperating point. Estimating the change enables the perturbation step to be adjusted inorder to accurately track the changes in the MPP. The main advantages of this combinedstrategy are conceptually presented in Fig. 1.1 and compared with the standard P&O inFig. 1.2. The improvements of the proposed strategy are summarized as follows: Ê theefficiency in steady state is improved by suppressing the oscillation, Ë the confusioncaused by a change in G is eliminated and Ì the step is adjusted for accurate tracking.• A study of the nature of the extraction losses both in steady state and transient isperformed and an estimation of those values is performed. The losses are quantized insteady state as a function of the PV cell parameters and the step size of the MPPTalgorithm. The losses during the transient are quantized, for a fixed step size, as afunction of the slope of the change in G, between two fixed levels. This figures allowthe visualization of the need for the ZA-P&O MPPT strategy.81.4. Thesis OutlinevmpptvpvFigure 1.1: Conceptual representation of the ZA-P&O combined MPPT strategies: efficiencymaximized with no oscillation Ê, correct decision Ë and step adjustment Ì under changesin irradiance.vmpptoscillation around the MPPwrong decision due to irradiance changenot enough speedvpvFigure 1.2: Issues with P&O technique: oscillation in steady state, wrong tracking step andconfusion during irradiance change.1.4 Thesis OutlineThe present work is organized in the following way:91.4. Thesis Outline• In Chapter 2, the characteristics of a PV powered system are presented as well asthe different conditions that cause some of the energy harvested by the PV panel notto be transferred to the load. The basic concepts of a PV panel are introduced inSection 2.1. The basic P&O algorithm is explained in Section 2.2 as a reference tool tocompare the proposed algorithm. The losses in steady state, which are inherent to theMPPT strategy, are studied in Section 2.3. The amount of energy lost is quantized asa function of the step size selected for the tracking process. In Section 2.4, the powerlosses during transient conditions in the irradiance are studied and the effect of themismatch between the speed at which the MPP moves and the speed at which thestrategy tracks it is quantized. The basic concepts of the implemented control systemand its impact on the proposed MPPT algorithm are explained in Section 2.5.• In Chapter 3, the ZA-P&O MPPT strategy is introduced. The proposed strategyreduces the losses in steady state by eliminating the constant oscillation, characteristicof the regular MPPT algorithms producing a clean behavior once the MPP was reached. The losses that are produced during transients in the irradiance are reduced byindirectly estimating this change and adjusting the tracking step to match the directionand slope of the transient. The resulting transient closely tracks the MPP duringirradiance changes.• The simulation validation of the ZA-P&O is presented in Chapter 4 using simulationsand an experimental set-up. The simulations compare the ZA-P&O with the standardP&O algorithm in order to point out the advantages of the proposed strategy. Theseresults show the elimination of the losses in steady state and the reduction of the lossesduring the transient.• Further validation of the proposed strategy is presented with several captures of theexperimental results in Chapter 5. The different features are tested using the same101.4. Thesis Outlineprofile for G and both algorithms and compared with the simulation results. Additionprofiles are presented and tested using the ZA-P&O algorithm to illustrate the stepadaptation features.• Lastly, in Chapter 6 a summary and conclusion of this work is presented, along withsome details of the future research opportunities.11Chapter 2System Background and EnergyHarvesting Losses In PV SystemsA typical PV power conversion system includes a PV array connected to a power converterthat charges a battery or is connected to the grid. The objective of the converter is both tochange the characteristics of the PV voltage/current to the needs of the load and to adapt theimpedance seen from the PV panel to ensure maximum power extraction. The block diagramof the implemented system is presented in Fig. 2.1. It includes a PV panel supplying energythrough a power converter to a load, formed by a battery bank. The batteries are assumedPV Module+-vpveZA-P&OPI controllerMPPTipvPower Converter Battery Packvpv* ipv*s+KpKivpv+-i pv+-e’Gidvo+-Figure 2.1: Block diagram of the PV system, the PV panel is connected to a battery bankthrough a DC/DC power converter, the control system regulates the PV voltage to matchthe instructions of the MPPT block.122.1. PV Panel Backgroundto be discharged, and therefore able to absorb all the available power without influencing theMPPT process. The PV panel voltage and current (vpv,ipv) are regulated by the controllerto achieve the maximum power extraction determined by the ZA-P&O MPPT strategy. TheDC/DC converter has a boost topology that ensures continuous current from the PV panel,minimizing the losses due to the current ripple.In this chapter the model of the PV panel is presented. A summary of the basic P&O andits limitations is discussed. The losses due to the inaccuracy of MPPT are derived, showingthe need to develop an improved MPPT strategy that minimizes the losses in steady stateand boosts the accuracy of the tracking under changing environmental conditions. Finally,the concepts behind the implemented control loops are presented, introducing the naturalperturbation concept that allows the elimination of the steady state oscillations and andenables the dynamic adjustment of the step size in proportion to the change in G. Theproposed ZA-P&O strategy builds on this foundation.2.1 PV Panel BackgroundThe equivalent circuit for a PV cell is presented in Fig. 2.2 [86], including the effects of theseries resistor (Rs) and the shunt resistor (Rsh). The diode D is characteristic of the P-Njunction of the cell structure, while the photocurrent (iG) is produced by the light photonsarriving at the junction. The basic PV panel equations are reviewed to provide a clearbackground for the estimation of losses due to the MPPT strategy operation. For a PVpanel built with M parallel strings of N cells connected in series, the panel’s current (ipv) atany given voltage (vpv), temperature (T ) and irradiance (G), neglecting the resistors, is givenby [61]ipv = MiG −MI0(exp(vpvNnkT/q)− 1). (2.1)132.1. PV Panel Backgroundvpv+-i pviG DRSRShiD i ShFigure 2.2: Photovoltaic (PV) cell model including the parasitic effects of the series resistor(Rs) and the shunt resistor (Rsh), the photocurrent (iG) is proportional to the irradiance(G).where I0 is the reverse saturation current of D, q is the electron charge, n is the diode factor,k is Boltzmann’s constant (in joules per kelvin), T is the PV panel temperature (in kelvin)and iG is proportional to GiG = γG. (2.2)This relationship determines the influence of the environment variables (G,T ) in thenonlinear cell characteristics and is fundamental to develop a MPPT strategy that can trackthis changes. Two other basic magnitudes are of special interest when describing a PV celland will be used to quantify the MPPT strategy behavior: the open-circuit voltage (Voc) andthe short-circuit current (Isc). The Voc is defined as vpv when isc is null:Voc = vpv|ipv=0=NnkTqln(iGI0+ 1). (2.3)On the other hand, Isc is defined as the ipv when vpv is null:Isc = ipv|vpv=0= MiG. (2.4)142.1. PV Panel Backgroundvpvi pv ppvpmppvmppimppG0>G0G1Figure 2.3: Characteristic I-V and P-V curves of the a PV cell under different irradianceconditions.The characteristic I-V and P-V curves for a PV cell are shown in Fig. 2.3, the operatingcondition (vmpp,impp) that yields the maximum available power (pmpp) is called the MaximumPower Point (MPP). When the environmental conditions (G, T ) change, the characteristiccurves of the PV panel change and the MPP moves. The MPPT algorithm must not only beable to find the MPP in stationary environmental conditions, but to track it while it changes.A final magnitude to characterize the PV panel is useful, the fill-factor (FF). The FF isdefined as the ratio between the MPP power and the product of Voc and IscFF =ImppVmppIscVoc, (2.5)indicating how much the actual PV panel differs from an ideal Voltage/Current source. TheFF is illustrated in Fig. 2.4, a good PV panel will have a Fill factor between 0.75 and 0.90.Aging and damages can reduce this value.Both the change in G and in T influence the characteristics of the PV panel, but they doso in different ways. From the basic expressions (2.1)-(2.4) each influence can be quantifiedand studied. A change in G mostly affects Isc, which increases proportionally to G, whileVoc remains almost the same as in Fig. 2.3 (for example, doubling G increases Isc by 100%while Voc only increases by 3%). On the other hand, changes in T mostly affect Voc while152.2. Basic P&O Algorithm Backgroundvpvi pvvmppimppIscVocPmpp = impp vmppPideal = Isc VocFigure 2.4: The Fill Factor (FF) of the PV panel is defined as the ratio between the maximumpower (pmpp) and the ideal power that could be extracted from the cell (VocIsc).Isc is less affected (for example 1 K variation increases Isc by 0.06% and Voc is decreased by0.4%). However, large gradients are expected from G due to clouding and shades, while Tis expected to have a smaller gradient. The present work will focus only on the change in Gwith different dynamics (fast and slow) during the transient.2.2 Basic P&O Algorithm BackgroundThe flowchart of the basic P&O MPPT algorithm is presented in Fig. 2.5. The basic P&Oscans the P-V curve of the panel in search for the MPP by changing the operating point(v∗pv or i∗pv), which is known as perturbation step, and then measuring the change in P (∆P ),known as observation step. If ∆P is greater than zero, then a new perturbation is introducedin the same direction. If ∆P is lower than zero, the direction of the perturbation is changed.The P&O keeps searching for the MPP until it has found an operating point such that ∆Pis lower than zero in any direction; this condition is called steady-state. The P&O keepsperturbing the system in order to detect a change in the MPP (caused by a change in theenvironmental conditions), which triggers a new scan. An illustration of this process can beobserved in Fig. 1.2. This steady-state perturbation drifts the operating point away from the162.2. Basic P&O Algorithm BackgroundMPPTvpv*ipv > pLastv*pv = v*pv + dir*Vstepdir = -dirYesNoNopLast = vpv * ipviLast = ipvipv, vpvv*pv = Vinitialdirection = -1Figure 2.5: Flowchart of the basic P&O MPPT algorithm.MPP; this introduces losses. In theory, this perturbation can be reduced, reduced in orderto keep the detection feature but minimize power losses. However, such a small perturbationwould require extremely precise sensors to measure the change in power. Therefore, theperturbation in steady state (if kept) has a minimum amplitude that depends on the sensors.Every time there is a change in the environmental conditions, there is a change in theP at the established operating point that masks the change caused by the perturbation. Inthis condition, the P&O algorithm might be induced to respond as though the perturbationintroduced produced an effect different than the true one. As observed in Fig. 1.2, duringthe transient on the right (MPP moves to a lower voltage), since G is reduced, the overallP is reduced, regardless of the direction of the perturbation. In this condition, the P&Oalgorithm changes direction in each step, trapping the system until the transient finishes.Finally, the classic P&O algorithm has a fixed step size, and therefore can only accuratelytrack the change in the MPP when it moves at a given rate (providing it made the decisionfor the correct direction). The quantification of these losses is estimated in this thesis.172.3. Steady-State Power Losses Estimationvpvtt1 t2steady statet0transientΔtVstepta ta+Tvmpp@ G=G0vmpp@ G=G1VeVe Vstep+Ve Vstep-Figure 2.6: The tracking losses depend on the different operating conditions (steady state orirradiance change).The above mentioned issues present serious drawbacks to the P&O algorithm; the ZA-P&O MPPT tackles those limitations, as will be explained in Chapter 3.2.3 Steady-State Power Losses EstimationThe proposed MPPT strategy removes the oscillation around the MPP in steady state that isintroduced by the artificial perturbation, and adapts the steps during a transient to accuratelytrack the MPP in order to reduce the available power lost. A typical operation situation ofa PV panel is displayed in Fig. 2.6, including a period in which G remains constant and atransient. The power losses in different stages of the MPPT (steady state or transient) aregiven by different expressions.The power losses (Pr) in relationship with the available power (Pmpp) in steady state forthe MPPT algorithms can be estimated as [33]PrPmpp≈((∆vpv)RMSvmpp)2(1 +vcell2nkT/q), (2.6)182.3. Steady-State Power Losses Estimationwhere (∆vpv)RMS is the RMS value of the voltage oscillation and vcell is the MPP voltage ofeach cell in the panel (around 0.5 V). As shown in the following equations, derived as part ofthis thesis, (2.6) can be manipulated to compare the losses in steady state for the traditionalP&O with the ZA-P&O. A cycle of the voltage oscillation around the MPP for the P&O isgiven by∆vpv(t) =Ve + Vstep if 0 < t ≤ T/4Ve if T/4 < t ≤ T/2Ve − Vstep if T/2 < t ≤ 3T/4Ve if 3T/4 < t ≤ T,(2.7)where Ve is the difference between Vmpp and the closest vpv set by the MPP tracker, due tothe step size (see Fig. 2.6, in steady state). As can be observed, Ve is related to Vstep byVe = bVstep, (2.8)where b is a number between −0.5 and 0.5. Therefore (2.7) can be expressed as∆vpv(t) =(b+ 1)Vstep if 0 < t ≤ T/4bVstep if T/4 < t ≤ T/2(b− 1)Vstep if T/2 < t ≤ 3T/4bVstep if 3T/4 < t ≤ T.(2.9)192.3. Steady-State Power Losses EstimationThe RMS value of the voltage oscillation for the P&O is(∆vpv)P&ORMS =√1T∫ T0∆v(t)2dt= Vstep√14((b+ 1)2 + 2b2 + (b− 1)2)= Vstep√12+ b2. (2.10)If the voltage is kept as close as possible to the MPP (no oscillation), the RMS value of thevoltage is(∆vpv)0RMS = bVstep. (2.11)Replacing (2.10) in (2.6) gives the relative losses for the P&O algorithm in steady state(PrPmpp)P&O≈(12+ b2)(Vstepvmpp)2(1 +vcell2nkT/q). (2.12)Plugging (2.11) in (2.6) gives the relative losses when the algorithm does not oscillate.(PrPmpp)0≈ b2(Vstepvmpp)2(1 +vcell2nkT/q). (2.13)The ratio between (2.12) and (2.13) indicates the extra power lost by keeping the oscillationin steady state(Pr/Pmpp)P&O(Pr/Pmpp)0≈1/2 + b2b2= 1 +12b2. (2.14)As just demonstrated, (2.14) indicates that the losses in steady state are incremented by1/2b2 because of the oscillation. In the best case, with b = 0.5, the losses are three timeslarger if the oscillation is maintained.For a PV panel with Voc = 200 V and Isc = 1 A, vmpp is around 170 V, the cell voltageat the MPP is around 0.5 V and with Vstep = 2 V (1% of Voc). The relative losses (expressed202.4. Irradiance Transient Power Losses Estimationin percentage) using the P&O are obtained evaluating (2.12) for these specific values(PrPmpp)P&O≈ 0.09%. (2.15)That means the oscillation around the MPP causes a 0.09% power losses. For the same PVpanel, if the oscillation is removed in steady state, the losses and can be estimated as(PrPmpp)0≈ 0.03%. (2.16)This configuration losses only 0.03% of the available power, three times less than does thetraditional P&O. Projected in a 25 years life cycle of the PV setup this means a tangiblebenefit in overall energy production.2.4 Irradiance Transient Power Losses EstimationThe following equations are derived in this thesis to quantify the losses due to dynamicMPPT error. During a transient in G, the MPP will move. The standard P&O algorithmhas a fixed Vstep and sampling time ∆t; therefore it is able to accurately track only one slopeof G, as seen in Fig. 2.6. If G changes more rapidly, the tracking would be inaccurate duringthe transient and would have to reach the MPP after the ramp stops. If G changes moreslowly, the operating point would drift away from the MPP until the ramp stops, after whichit would have to reach the true MPP (see Fig. 2.6).212.4. Irradiance Transient Power Losses EstimationDuring a ramp change in G from G0 at t0 to G1 at t1, G can be expressed asG(t) =G0 if t ≤ t0G0 + (t− t0)δG if t0 < t ≤ t1G1 if t1 < t.(2.17)Then, the power extracted from the PV panel during the transient is given byppv(t) = vpv(t)(γG(t)− I0(exp(vpv(t)nkT/q)− 1)). (2.18)A comparison with the maximum power (obtained while operating constantly at vmpp)pmpp(t) = vmpp(t)(γG(t)− I0(exp(vmpp(t)nkT/q)− 1)), (2.19)shows the loss due to inaccurate tracking. The instantaneous tracking efficiency is given byηMPPT (t) =ppv(t)pmpp(t), (2.20)however, sometimes it is more important to have a more general overview of the losses.Computing the energy obtained through the transient (Epv) and comparing it with the energythat would be obtained if the panel had always operated at the MPP (Empp) , enables theaverage tracking efficiency (η¯MPPT ) to be quantifiedη¯MPPT =EpvEmpp=∫ t2t0ppv(t)dt∫ t2t0pmpp(t)dt. (2.21)Numerically integrating these expression for a given PV panel using a P&O with a fixed ∆tand Vstep shows how η¯MPPT is minimal for a certain δG.222.4. Irradiance Transient Power Losses Estimationirradiance ramp slope [W m-2 s-1]efficiency [%]100 200 300 400 500 600 7009999.299.499.699.8100100.2efficiency losses (1-ηMPPT )t ttv v vFigure 2.7: Tracking efficiency (η¯MPPT ) for a transient between G0 = 600 Wm−2 to G1 =1000 Wm−2, for a P&O algorithm with ∆t = 1 s and Vstep = 2 V showing the losses whenthe MPPT slope does not match the slope produced by δG.These equations derived in this thesis provide valuable insight into the transient losses.The results for the example PV panel when G is changing from G0 = 600 Wm−2 to G1 =1000 Wm−2 with several different δG are displayed in Fig. 2.7. The small sketches in Fig. 2.7illustrate the tracking process with different slopes of G and a fixed step size: when theslope is slower than the steps, the tracking point overshoots; when the slope is higher thanthe steps, the tracking point takes time to catch up. There is one slope where the trackingpoint closely follows MPP. The power losses are minimal for a slope of 300 Wm−2s−1; thisis the slope that the P&O can accurately track. If the slope is lower than the optimal, forexample δG = 100 Wm−2s−1, vpv will become larger than vmpp leading to larger losses in theorder of 3% (since the PV curves steeper for higher voltages as seen in Fig. 2.3). When theslope is higher and the MPPT lags behind, for example for δG = 600 Wm−2s−1, the lossesare around 0.1%. It is clear that a MPPT strategy that has a fixed step for tracking is not232.5. Control Systemoptimal during transients, and an adaptive strategy will ensure a closer tracking of the MPPin this condition, independent of the change in G.2.5 Control SystemUnlike most power converters, where the controller aims to regulate the output voltage, forMPPT purposes the controller regulates the input of the converter (vpv). The controller isshown in Fig. 2.1. It uses a dual loop to regulate both ipv and vpv, this is then used toobtain additional information regarding the change in environmental conditions, a naturalperturbation to the system. The inner loop (faster) will regulate ipv by setting the duty cycle(d) of the converters switches. The outer loop, slower regulates vpv to the level determinedby the MPPT strategy (v∗pv) by setting the reference current (i∗pv).The use of a dual-loop controller serves two purposes. On the one hand, it improves thestability of the system. On the other hand, it helps isolate the change in G acting as a naturalperturbation to lead the MPPT. This avoids the use of a continuous artificial perturbationin steady state to track this change.Since v∗pv is updated at the sampling speed of the MPPT (slower than the control system),during a sudden change in G, the outer loop will keep vpv constant by changing ipv. Thischange in ipv indicates the change in G and acts as a natural perturbation. Moreover, it isshown in [52] that when G changes with a slope δG during the sample period (∆t) and thevoltage loop has a PI controller the tracking error (e) is proportional to δGe ∝δGKi. (2.22)242.6. SummaryUsing this information, the proposed MPPT strategy can stop the artificial perturbation insteady state and monitor the change in ipv and e (natural perturbation) to determine thechange in the environment and adjust the step size (Vstep).2.6 SummaryIn this chapter, the block diagram of the system was studied. The characteristics of thePV panel and a industry-standard MPPT algorithm were presented in order to introducefurther details. The characteristic losses produced by MPPT algorithms in steady state andtransient conditions were presented and a means to evaluate them was introduced. Finally,the control technique and its impact on the proposed MPPT strategy was presented. Theeffects of a transient in G are seen as a natural perturbation, that allows the removal ofthe artificial perturbation characteristic of the P&O. In the following chapter, an advancedMPPT strategy will be introduced that uses this characteristics to eliminate the oscillationin steady-state (reducing the losses) and uses the controller signals to infer the direction andmagnitude of the change in G and produce an accurate tracking of transients.25Chapter 3Zero Oscillation Adaptive Perturband Observe MPPTThe traditional hill-climbing algorithm (P&O) discussed in the previous chapter has a num-ber of disadvantages such as an oscillation in steady state (due to the continuous scanningprocess), it gets confused during changes in G and is unable to identify the rate of changeof the irradiance and adapt the step-size correctly, this introduces losses to the system thatwere also identified in the previous chapter. The ZA-P&O MPPT strategy is developed basedon two key aspects: 1) detection of the steady-state operation and 2) determination of thedirection and magnitude of the perturbation. In steady-state, the standard P&O algorithmwill oscillate around the closest possible voltage (due to quantization of the voltage step) inthree levels, as in Fig. 1.2. This introduction of an artificial perturbation allows the MPPTto scan the curve for a change in the characteristics caused by environmental conditions.The ZA-P&O identifies this situation and establishes the operating point (v∗pv) in the clos-est value, as shown in Fig. 1.1. In this operating mode, called idle mode, the losses areminimized as shown in Section 3.1. The environmental change is identified by monitoringthe error in the PI controller (e) and the change in current (ipv), which generate a naturalperturbation clearly correlated to the change in the conditions. The use of this natural per-turbation enables a cleaner operation, removing redundant oscillations that are convolutedwith the information of environmental change and can cause confusion. This allows tracking263.1. Idle Mode Operationiddle mode activetppvipvvpvppvvpvtoggle count (TgC) incrementedscanning steady stateFigure 3.1: Illustration of the Idle mode operation. On the left, the V-I plane shows thecharacteristic curve and the tracking process; on the right, the time domain correspondenceis shown. The idle mode detects the oscillations and activates the idle mode eliminating thelosses.to reactivate when necessary and to establish an accurate step size based on the known slope,instead of toggling continuously as is usually done.3.1 Idle Mode OperationConventional MPPT strategies search for the MPP by periodically changing vpv and mea-suring the effect over ppv or some other parameter. Since no way of identifying a change inG (and the corresponding displacement of the MPP) is included in the MPPT strategies, itmust keep perturbing the operating point even when the MPP has been found (steady-state).This is reflected in a three-level operating condition shown in Fig. 1.2, where the operatingpoint toggles around the closest voltage allowed by the discrete steps. This is an artificialperturbation that reduces the efficiency of the energy extraction.The ZA-P&O MPPT strategy uses a natural perturbation, native to the control of thesystem: the change in ipv and the error in the PI controller (with constant v∗pv), making itpossible to eliminate the artificial perturbation. The proposed method identifies the oper-273.2. Perturbation Direction and Magnitude Estimationation in steady state by counting consecutive changes in direction with a common middlepoint (vmid). When a step is introduced in v∗pv, if the step is the second in the same directionand the power increased, the algorithm displaces vmid. The software computes one toggleand registers it in the toggle counter (TgC) each time the set point returns to vmid. Themaximum number of toggles around vmid before activating the idled mode and removing theperturbation is set as a parameter (TgM) and is used to avoid confusion caused by noise.The operation of the idle mode is illustrated in Fig. 3.1. On the left side, the V-I planecurves are shown; the corresponding time domain progression is displayed on the right.During the tracking phase, when the operating voltage is away from the MPP, the steps arealways in the same direction and therefore the toggle counter is not incremented. When vpvreaches the MPP it starts oscillating around a middle point. The MPPT algorithm detectsthis condition and increments the counter. Once the maximum number of counts allowed isreached, the Idle mode is activated and the tracking process stops. In a lighter shade, it canbe observed how the MPPT would continue oscillating if the Idle mode was not implementedleading to losses in the system.3.2 Perturbation Direction and Magnitude EstimationAs explained in Section 2.5, the implemented controller can be used to monitor the changesin the environmental conditions. This leads to accurate knowledge of the magnitude of thechange in G and the slope δG. This can be directly correlated to the displacement of vmppand impp. This information provides the correct direction to the ZA-P&O.In case of a slope change in G it is not enough to know the direction of the displacement.If the slope is too small, the MPPT may detect the change and introduce a step that willlead the operating point far away from the MPP. If the slope is too steep, the steps maynot be large enough to track the MPP. Since the magnitude of the slope can be identified283.3. Flowchart of the ZA-P&O MPPTfrom (2.22), the magnitude of the change in ipv that results from the change in G is knownas∆ipv = Kie∆t (3.1)where ∆t is the sample time of the MPPT strategy. As can be seen, from the measurementof e and the knowledge of Ki and ∆t, it is possible to know the change in ipv, proportionalto the change in G. In this condition, we can adjust the step in proportion to the change inthe G. The value of the proportionality constant depends on the PV panel.The identification of the change in G is illustrated in Fig. 3.2. In the upper part of the theillustration, target voltage v∗pv and the actual PV panel voltage vpv are presented. The PVpanel is commanded to change the operating point in the characteristic three-level operation.In the bottom of the graph, a profile in G is shown: in the first part G keeps constant andthen it starts increasing with a certain slope δG; finally it starts decreasing with a differentslope. Observing v∗pv and vpv it can be seen that there is a tracking error during the slopechanges in G and the amplitude of this error is proportional to the slope. Since the errorsignal is automatically available from the implementation of the controller, this signal can befed to the MPPT and used to determine the correct direction.3.3 Flowchart of the ZA-P&O MPPTThe flow chart for the ZA-P&O is presented in Fig. 3.3. The strategy includes tunable param-eters, such as the thresholds for the current change and error (iTh and eTh) and the numberof toggles around the MPP required to establish the idle mode (TgM). This thresholds areincluded in order to prevent the algorithm to get confused by the noise in the sensors. Out-side the special conditions established of Idle mode and identification of the change in G, the293.4. SummarytGv*pvvpve ∝ dG/dte ∝ dG/dtdG/dt>0dG/dt<0e = dG/dt=0dG/dt=0dG/dt=0e = dG/dt=0ΔtFigure 3.2: The change in irradiance is identified via the tracking error in steady state causedby the slope of the irradiance change.algorithm works as a P&O. As can be observed, the blocks added to the algorithm are basedon comparisons and simple operations and do not add major complexity to the algorithmwhich leads to a simple implementation comparable with the standard P&O.3.4 SummaryIn this chapter, an advanced MPPT strategy was proposed. The ZA-P&O MPPT strategycombines two key elements discussed previously to enhance the standard P&O algorithmwith new features. The Idle mode was introduced, that allows the MPPT to stop once it hasreached the MPP instead of constantly oscillating, reducing the losses in steady-state. This ispossible due to the implementation of an indirect estimation of the change in environmentalconditions via the control law of the converter. An additional benefit of this estimation isthe ability to identify the occurrence of a change in G, its direction and the magnitude ofthe slope allowing for accurate tracking of the MPP during the transients. Validation of theZA-P&O MPPT is provided in the next two chapters, first with simulations and then withexperimental captures.303.4. SummaryMPPTvpv*ipv > pLastTgC = 0v*pv = v*pv + dir*VstepTgC = TgC+1dir = -dirTgC>TgM and v*pv = vmidYesNoIdle = 1No|ipv-iLast|>iThIdle = 0YesYes|e| > eThNov*pv = v*pv+ k*e YesToggleC = 0Idle = 0dir = sign(k*e)Idle = 1pLast = vpv * ipviLast = ipvYesNoipv, vpv, ev*pv = VinitialIdle = 0TogleC = 0direction = -1dir = sign(ipv-iLast)v*pv = vmidYesNov*pv = v*pv + dir*VstepFigure 3.3: Flow chart for the ZA-P&O MPPT strategy.31Chapter 4SimulationsIn the previous chapter, the ZA-P&O MPPT was introduced. This algorithm adds newfeatures to the standard P&O algorithm and creates an important improvement in terms ofsteady-state oscillation and transient tracking. To validate these improvements, a computermodel and a profile of G were generated to test the known issues of the regular P&O againstthe ZA-P&O MPPT algorithm.The results of the computer simulation for both the ZA-P&O MPPT and the standardP&O for a trapezoidal irradiance (G) profile are presented in this chapter. The modelconsists of a PV panel that can be configured to perform with the desired characteristics,time [s]irradiance [W m-2] 10 15 20 25 30 35 40 45 506008001000400∆t1=4 s ∆t2=11 s ∆t3=2 shigh dG/dt (decreasing)low dG/dt (increasing)high Glow GFigure 4.1: Trapezoidal irradiance (G) profile used to simulate the P&O algorithm and theproposed ZA-P&O strategy.32Chapter 4. Simulationspv panel voltage [V] 150160170180190vmppvpvpv panel current [A] 23456imppipvtime [s]pv panel power [W] 10 15 20 25 30 35 40 45 504006008001000pmppppvthe MPPT winds-up and then it has to come backconfusion due to irradiance changeoscillation around the MPPlarge drift fromthe MPP currentlarge drop in efficiencyFigure 4.2: Standard P&O issues when the irradiation (G) slope starts during a raising stepof the algorithm for vpv, ipv and ppv.a PI controller to regulate the voltage of the PV panel, and the MPPT to determine theoperating point. The output of the PI controller sets the current in the PV panel.33Chapter 4. Simulationspv panel voltage [V] 150160170180190vmppvpvpv panel current [A] 23456imppipvtime [s]pv panel power [W] 10 15 20 25 30 35 40 45 504006008001000pmppppvoscillation around the MPPthe MPPT goes in the wrong direction and has to come backconfusion due to irradiance changedrift from the MPP currentdrop in efficiencydrops in the efficiencyfor oscillationsFigure 4.3: Standard P&O issues when the irradiation (G) slope starts during a falling stepof the algorithm for vpv, ipv and ppv.The PV panel is configured to have 200 V open-circuit voltage and 5 A short-circuitcurrent, with a Fill Factor (FF) of 0.8 at standard test conditions (STC, 1 kW/m2 and25 ◦C). The MPPTs are configured with the same voltage step for the P&O part and thesame sampling period. The sampling period of the MPPT is established in 0.5 s and the34Chapter 4. Simulationspv panel voltage [V] 150160170180190vmppvpvpv panel current [A] 23456imppipvtime [s]pv panel power [W] 10 15 20 25 30 35 40 45 504006008001000pmppppvno oscillation around the MPPstep adjusted to follow the slopecorrect decision under irradiance changeaccurate tracking of the MPP currentFigure 4.4: ZA-P&O MPPT strategy improvements in steady state and transient for thevoltage vpv, ipv and ppv.fixed voltage step is 3 V. The testing profile is presented in Fig. 4.1. G starts at 0.6 kW/m2;at t = 19 s, it starts increasing with a slope of 0.1 kW/m2s. When it reaches 1.0 kW/m2,it stops and waits for 11 s and then it starts decreasing with a slope of 0.3 kW/m2s until itreaches 0.4 kW/m2. Then it remains constant until the end of the simulation. To illustrate35Chapter 4. Simulationsthe error in tracking for the standard P&O, two cases were studied, a) slope was startedduring a falling edge and b) during a raising edge of the MPPT by introducing a smalldisplacement in the profile. The input signals to the MPPT are vpv, ipv and the error of thePI controller as indicated in Fig. 2.1. When the standard P&O is tested, the error signal isnot used as one of the inputs. For the simulations, the threshold level of the current and theerror (iTh and eTh) are set to zero, since there is no noise in this environment.The results of the simulation for the standard P&O are displayed in Figs. 4.2 and 4.3,while the results for the ZA-P&O MPPT are presented in Fig. 4.4. When the transient profilestarts, the standard P&O keeps going in the direction in which it was going before G changed(since it detects an increase in power) even when this direction is incorrect. In Fig. 4.2 thisdirection is correct, however, the tracking is not accurate since it is not able to detect thatit is leaving the MPP voltage behind. Fig. 4.3 is an example of a bad decision: even whenthe tracking step is given in the incorrect direction, the MPPT algorithm keeps going in thesame direction until it reaches the minimum operating voltage and has to return. Whenthe irradiance decreases, it starts toggling in the same position, since any step produces adecrement in power. This deviation from the correct direction leads performance losses, sincereal profiles can have slopes for extended periods of time. Moreover, the standard MPPTalgorithm is unable to adjust the tracking step to different δG; this leads to an algorithmthat, even when it goes in the correct direction, may drift from the MPP because of thewrong step selection. This issue is shown in Fig. 4.2 where the operating point is increaseduntil the drop is very large and it has to return.The ZA-P&O MPPT strategy resolves those issues effectively as can be seen in Fig. 4.4.The idle operating mode allows the PV panel to operate in a smooth way when there isno need to keep tracking. When a change occurs in G, the strategy clearly identifies thecorrect direction to move the operating point and adjusts the step-size to provide a close36Chapter 4. Simulations 98.099.0100.0ZA−P&Oefficiency [%] 98.099.0100.0P&O ↓time [s] 10 15 20 25 30 35 40 45 5098.099.0100.0P&O ↑Figure 4.5: Efficiency of the PV panel for the ZA-P&O and the standard P&O when theslope starts in the falling edge (P&O ↓) and in the raising edge (P&O ↑).tracking of the MPP. The effectiveness of the identification does not depend on the momentthe irradiance slope starts.The efficiency of the tracking for the three cases is shown in Fig. 4.5. It is evident thatthe ZA-P&O improves the overall performance: 1) in steady-state, the efficiency remainsconstant and close to 100% instead of oscillating periodically, 2) the correct direction isdetermined and 3) the step is adjusted; thus the efficiency remains high even during thetransient, whereas the standard P&O leads to drops in efficiency.Lastly, the transient trajectory is picture in the I-V plane in Fig. 4.6. This shows thetrajectory of the tracking process for the P&O and for the ZA-P&O. The P&O deviates fromthe optimal trajectory, while the proposed strategy keeps very close track of it.The experimental validation of the proposed strategy is presented in the next chapter.Using a similar profile, the promising results of the simulations will be tested.37Chapter 4. Simulationspv panel voltage [V]pv panel current [A]150 155 160 165 170 175 180 185 190 1952.02.53.03.54.04.55.0G = 1000 W/m2G = 600 W/m2P&O ↑ZA-P&OoptimalFigure 4.6: P&O undesirable drifting behavior versus ZA-P&O direct trajectory in the V-Iplane. The proposed strategy closely follows the locus of MPP.38Chapter 5Experimental ResultsThe simulation results in the previous chapter showed how the standard P&O algorithm hasissues in steady state and while tracking changing environmental conditions and how the ZA-P&O MPPT mitigates the problem by using an indirect estimation of change in G. To furthervalidate the algorithm, an experimental set-up was developed and the same algorithms weretested using the same irradiance profile. Additional validation of the characteristic featuresare presented using other profiles of G. These additional profiles include steeper and gradualramps in G both decreasing and increasing from the initial state. This validation show howthe adaptation of the step size provides accurate tracking of the MPP.The experimental setup of Fig. 5.1 was developed with the same parameters than thesimulation to validate the ZA-P&O MPPT strategy. The prototype was implemented usingFigure 5.1: Picture of the experimental set-up.39Chapter 5. Experimental Resultsoscillation oscillation oscillation wind-up duringirradiance changewrong decision due to irradiance changepower drop during transientipvppvvpvFigure 5.2: Standard P&O experimental capture when the irradiance transient starts in araising edge of the MPPT showing the issues of steady state oscillations and inaccuratetracking of transients.an industry-standard microcontroller (TI C2000 core) typically employed to control powerconverters. The experimental results are shown in Figs. 5.2 and 5.3 for the standard P&Owhen the slope starts at a falling edge and a raising edge respectively and in Fig. 5.4 forthe proposed strategy. The experimental captures closely match the simulation results. Theexperimental captures for the standard P&O (Figs. 5.2 and 5.3) show the characteristicissues: oscillation in steady state, wrong direction and wind-up. The benefits of the ZA-P&O are clearly shown in Fig. 5.4, when compared with the standard P&O. The oscillationis removed and the step is given in the correct direction and magnitude, as shown in the fastre-establishment of the idle mode after the irradiance slope ends. Calculating the total powerproduced during the transient for the three cases shows that the ZA-P&O produces 0.4%more energy comparing Fig. 5.4 and 5.3 and 0.7% more energy comparing Fig. 5.4 and 5.2.The overall efficiency of the ZA-P&O MPPT for this transients is 99.3%.40Chapter 5. Experimental Resultswrong decision due to irradiance changeoscillation vpvoscillation power drop during transientipvppvFigure 5.3: Standard P&O experimental capture when the irradiance transient starts ina falling edge of the MPPT showing the issues of steady state oscillations and inaccuratetracking of transients.no oscillationadjusted speedright decisionFigure 5.4: ZA-P&O experimental test; the improvements are shown with the steady stateoperation and the accurate and fast tracking during transients.41Chapter 5. Experimental ResultsIt can be seen in the experimental Figs. 5.2 and 5.3 that the standard P&O strategy driftsaway from the MPP when G changes with a slope δG. Keeping the oscillation around theMPP in the idle condition increases the probability of making a mistake due to the noise inthe measurement. This is demonstrated in Fig. 5.3, where the P&O process reaches the MPPduring the first stage and works with three levels but suddenly drifts away and returns. TheZA-P&O MPPT strategy benefits from the removal of this perturbation to enable a clearermeasure of the change in G, which is represented by Ë in Fig. 5.4. The use of an indirectestimation though the control loop allows this estimation to be done without increasing thecomplexity of the system, since the signals are already available inside the micorcontroller.In order to test the irradiance slope identification and multi-level step-size adaptation, twodifferent profiles are introduced and tested in the experimental set-up. The profile in Fig. 5.5shows the PV panel operating at 1.0 kWm−2 followed by a gradual drop in irradiance during5 s period of time. After staying at 0.5 kWm−2 during 5 s the profile returns to 1.0 kWm−2 in1 s. The profile in Fig. 5.6 a reflected version of the previous one, starting at low irradianceand gradually reaching the 1.0 kWm−2 level. Using this profile it is possible to test theZA-P&O algorithm for a range of slopes both positive and negative showing the step sizeadaptation feature.The captures in Fig. 5.7 and Fig. 5.8 show the ZA-P&O MPPT for the different G-profilesstated before. The captures show details of the behavior under this changing environmentalconditions, with a closer time scale and vertical scales compared with the one in Fig. 5.4. Theexperimental set shows how the ZA-P&O reacts to a very fast transient and a very slow one.As can be seen, the estimation of the new position ensures the correct direction and placesthe operating point close to the MPP in such a way that the local optimizations (activated42Chapter 5. Experimental Resultstime [s]irradiance [W m-2] 0 2 4 6 8 10 12 14 166008001000400∆t1=5 s ∆t2=5 s ∆t3=1 s500 W m-21000 W m-2-100 W m-2 s-1500 W m-2 s-1Figure 5.5: Irradiance profile to test the adaptive step feature of the ZA-P&O MPPT strategy.A gradual transition from a high-irradiance level to a low-irradiance level followed by a fastreturn is implemented.time [s]irradiance [W m-2] 0 2 4 6 8 10 12 14 166008001000400∆t1=5 s ∆t2=5 s ∆t3=1 s500 W m-21000 W m-2-100 W m-2 s-1 500 W m-2 s-1Figure 5.6: Irradiance profile to test the adaptive step feature of the ZA-P&O MPPT strategy.A gradual transition from a low-irradiance level to a high-irradiance level followed by a fastreturn is implemented.after the transient) can locate the MPP in a few cycles and reactivate the Idle mode reducingthe losses to a minimum.The ZA-P&O was tested against the standard P&O algorithm in an experimental set-upusing the same profile of G than the simulations and obtaining equivalent results. Additionalprofiles were tested in order to further validate the adaptive-step feature of the proposed43Chapter 5. Experimental Resultsno oscillationadjusted speedright decisionFigure 5.7: ZA-P&O experimental test for a step-down change from 1 kW m−2 to 500 W m−2in 5 s and then back to 1 kW m−2 in 1 s.no oscillationadjusted speedright decisionFigure 5.8: ZA-P&O experimental test for a step-up change from 500 W m−2 to 100 W m−2in 5 s and then back to 500 W m−2 in 1 s.strategy. The captures show the indirect identification of the change G allows the MPPTalgorithm to avoid making mistakes and tracking in the correct direction even with different44Chapter 5. Experimental Resultsslopes. Besides the increment in extracted energy, the ZA-P&O creates a cleaner operatingcondition for the converter with no unnecessary oscillations and long tracking transients. Theaddition of these features creates a more reliable MPPT algorithm when compared with theP&O and allow for more energy to be harvested from the solar panel.45Chapter 6Conclusions6.1 SummaryPhotovoltaic (PV) energy systems use power electronics converters to interface the energysource with either the load or the grid. This is done in order to modify the characteristicsof the voltage/current extracted from the PV panel to match the needs of the load and toadapt the impedance connected to the panel to extract the maximum power. This is calledMaximum Power Point Tracking (MPPT). In the industry, the standard MPPT algorithm isthe Perturb and Observe (P&O) or the incremental conductance (InCond). Both of them arebased on the hill-climbing method and have the same issues that introduce losses: oscillationsin steady-state, errors during changing environmental conditions and inability to detect therate of change of the irradiance and adjust the step-size to accurately track it.This thesis introduced the Zero-oscillation, Adaptive-step Perturb and Observe (ZA-P&O) MPPT strategy for PV panels. This combined strategy reduced steady-state lossesand improved transient behavior during slope changes irradiance, while maintaining a simi-lar implementation complexity compared with industry-standard algorithms. The enhancedbehavior resulted from the combination of three techniques: 1) idle operation when steady-state is reached, 2) correct irradiance change identification and 3) multi-level adaptive track-ing step. The idle operation was possible due to the identification of the irradiance slopethrough a current monitoring algorithm. The adaptive tracking speed minimized error duringa fast change in irradiance.466.2. Future WorkThe proposed combined techniques were studied with simulations and validated throughexperimental results implemented in a low cost microcontroller. The overall performanceimprovements, both in steady-state and with different irradiance change profiles show thebenefits of the combined techniques.The contribution investigated in this thesis has been published IEEE PEDG 2013 [1] andIEEE Transactions on Industrial Electronics [2].6.2 Future WorkThe algorithm developed in this work provides an original contribution to the field of MPPTfor photovoltaic applications. The extension of this concept to other renewable and alterna-tive energy sources (wind, hydro, fuel cells, etc.) is being studied and a research paper isbeing prepared. A research paper involving advanced control strategies for power convertersinvolved in MPPT for PV applications was accepted for PEDG 2014. A research paper onhigh speed MPPT strategies for electric vehicles is under development.The use of indirect environmental changes identification to improve the MPPT capabilitiesof the converter presents a useful tool for all renewable energy sources in the future.47Bibliography[1] F. Paz and M. Ordonez, “Zero-oscillation adaptive-step solar maximum power pointtracking for rapid irradiance tracking and steady-state losses minimization,” in IEEEInt. Symp. Power Electronics for Distributed Generation Systems (PEDG), Rorgers, AR,Jul. 2013, pp. 1–6, c© 2013 IEEE.[2] F. Paz and M. Ordonez, “Zero oscillation and irradiance slope tracking for photo-voltaic mppt,” IEEE Trans. Ind. Electron., vol. 61, no. 11, pp. 6138 – 6147, Nov. 2014,c© 2014 IEEE.[3] “Photovoltaics report,” FRAUNHOFER INSTITUTE FOR SOLAR ENERGYSYSTEMS ISE, Tech. Rep., Nov. 2013. [Online]. Available: www.ise.fraunhofer.de[4] G. Hart, H. Branz, and C. C. III, “Experimental tests of open-loop maximum-power-point tracking techniques for photovoltaic arrays,” Solar Cells, vol. 13, no. 2, pp. 185 –195, 1984.[5] J. Applebaum, “The quality of load matching in a direct-coupling photovoltaic system,”IEEE Trans. Energy Convers., vol. EC-2, no. 4, pp. 534–541, Dec 1987.[6] J. Dunlop, “Analysis and design optimization of photovoltaic water pumping systems,”in IEEE Photovoltaic Specialists Conf. (PVSC), Las Vegas (NV), Sep. 1988, pp. 1182–1187 vol.2.48BIBLIOGRAPHY[7] J. Appelbaum and M. Sarma, “The operation of permanent magnet dc motors poweredby a common source of solar cells,” IEEE Trans. Energy Convers., vol. 4, no. 4, pp.635–642, Dec 1989.[8] J. Appelbaum and S. Singer, “Magnification of starting torques of dc motors by maxi-mum power point trackers in photovoltaic systems,” in Intersociety Energy ConversionEngineering Conference (IECEC), Aug. 1989, pp. 749–754 vol.2.[9] J. Appelbaum, “The operation of loads powered by separate sources or by a commonsource of solar cells,” IEEE Trans. Energy Convers., vol. 4, no. 3, pp. 351–357, Sep.1989.[10] J. H. R. Enslin, “Maximum power point tracking: a cost saving necessity in solar energysystems,” in IEEE Annu. Conf. Industrial Electronics (IECON), Nov. 1990, pp. 1073–1077 vol.2.[11] A. Pandey, N. Dasgupta, and A. Mukerjee, “A simple single-sensor mppt solution,”IEEE Trans. Power Electron., vol. 22, no. 2, pp. 698–700, Mar. 2007.[12] D. Patterson, “Electrical system design for a solar powered vehicle,” in IEEE Annu.Power Electronics Specialists Conf. (PESC), Jan. 1990, pp. 618–622.[13] W. Li, Y. Zheng, W. Li, Y. Zhao, and X. He, “A smart and simple pv charger for portableapplications,” in IEEE Annu. Applied Power Electronics Conf. and Expo. (APEC), Feb.2010, pp. 2080–2084.[14] M. Masoum, H. Dehbonei, and E. Fuchs, “Theoretical and experimental analyses ofphotovoltaic systems with voltageand current-based maximum power-point tracking,”IEEE Trans. Energy Convers., vol. 17, no. 4, pp. 514 – 522, Dec. 2002.49BIBLIOGRAPHY[15] J. Appelbaum, “Discussion of ”theoretical and experimental analyses of photovoltaicsystems with voltage and current-based maximum power point tracking”,” IEEE Trans.Energy Convers., vol. 19, no. 3, pp. 651–652, Sep. 2004.[16] J. Martynaitis, “Discussion of ”theoretical and experimental analyses of photovoltaicsystems with voltage and current-based maximum power point tracking”,” IEEE Trans.Energy Convers., vol. 19, no. 3, pp. 652–, Sep. 2004.[17] M. Masoum, H. Dehbonei, and E. Fuchs, “Closure on ”theoretical and experimentalanalyses of photovoltaic systems with voltage and current-based maximum power pointtracking”,” IEEE Trans. Energy Convers., vol. 19, no. 3, pp. 652–653, Sep. 2004.[18] A. Murtaza, H. Sher, M. Chiaberge, D. Boero, M. De Giuseppe, and K. Addoweesh,“Comparative analysis of maximum power point tracking techniques for pv applica-tions,” in Int. Multi Topic Conference (INMIC), Dec 2013, pp. 83–88.[19] B. Subudhi and R. Pradhan, “A comparative study on maximum power point trackingtechniques for photovoltaic power systems,” IEEE Trans. Sustain. Energy, vol. 4, no. 1,pp. 89 –98, Jan. 2013.[20] O. Lopez-Lapena, M. Penella, and M. Gasulla, “A closed-loop maximum power pointtracker for subwatt photovoltaic panels,” IEEE Trans. Ind. Electron., vol. 59, no. 3, pp.1588–1596, Mar. 2012.[21] A. Weddell, G. Merrett, and B. Al-Hashimi, “Photovoltaic sample-and-hold circuit en-abling mppt indoors for low-power systems,” IEEE Trans. Circuits Syst. I, Reg. Papers,vol. 59, no. 6, pp. 1196 –1204, Jun. 2012.50BIBLIOGRAPHY[22] S. Kjaer, “Evaluation of the ”hill climbing” and the ”incremental conductance” maxi-mum power point trackers for photovoltaic power systems,” IEEE Trans. Energy Con-vers., vol. 27, no. 4, pp. 922 –929, Dec. 2012.[23] G. Petrone, G. Spagnuolo, R. Teodorescu, M. Veerachary, and M. Vitelli, “Reliabilityissues in photovoltaic power processing systems,” IEEE Trans. Ind. Electron., vol. 55,no. 7, pp. 2569–2580, Jul. 2008.[24] E. Roman, R. Alonso, P. Ibanez, S. Elorduizapatarietxe, and D. Goitia, “Intelligent pvmodule for grid-connected pv systems,” IEEE Trans. Ind. Electron., vol. 53, no. 4, pp.1066–1073, Jun. 2006.[25] A. Ridge and G. Amaratunga, “Photovoltaic maximum power point tracking for mobileapplications,” Electronics Letters, vol. 46, no. 22, pp. 1520–1521, Oct. 2010.[26] A. Safari and S. Mekhilef, “Simulation and hardware implementation of incrementalconductance mppt with direct control method using cuk converter,” IEEE Trans. Ind.Electron., vol. 58, no. 4, pp. 1154–1161, Apr. 2011.[27] D. Sera, L. Mathe, T. Kerekes, S. Spataru, and R. Teodorescu, “On the perturb-and-observe and incremental conductance mppt methods for pv systems,” IEEE J. Photo-volt., vol. 3, no. 3, pp. 1070–1078, Jul. 2013.[28] M. Elgendy, B. Zahawi, and D. Atkinson, “Assessment of the incremental conductancemaximum power point tracking algorithm,” IEEE Trans. Sustain. Energy, vol. 4, no. 1,pp. 108 –117, Jan. 2013.[29] N. Femia, G. Petrone, G. Spagnuolo, and M. Vitelli, “Optimization of perturb andobserve maximum power point tracking method,” IEEE Trans. Power Electron., vol. 20,no. 4, pp. 963 – 973, Jul. 2005.51BIBLIOGRAPHY[30] N. Femia, G. Petrone, G. Spagnuolo, and M. Vitelli, “A technique for improving p&omppt performances of double-stage grid-connected photovoltaic systems,” IEEE Trans.Ind. Electron., vol. 56, no. 11, pp. 4473–4482, Nov. 2009.[31] M. Elgendy, B. Zahawi, and D. Atkinson, “Assessment of perturb and observe mppt al-gorithm implementation techniques for pv pumping applications,” IEEE Trans. Sustain.Energy, vol. 3, no. 1, pp. 21 –33, Jan. 2012.[32] M. de Brito, L. Galotto, L. Sampaio, G. de Azevedo e Melo, and C. Canesin, “Evaluationof the main mppt techniques for photovoltaic applications,” IEEE Trans. Ind. Electron.,vol. 60, no. 3, pp. 1156 –1167, Mar. 2013.[33] C. Sullivan, J. Awerbuch, and A. Latham, “Decrease in photovoltaic power output fromripple: Simple general calculation and the effect of partial shading,” IEEE Trans. PowerElectron., vol. 28, no. 2, pp. 740 –747, Feb. 2013.[34] R. Faranda, S. Leva, and V. Maugeri, “Mppt techniques for pv systems: Energetic andcost comparison,” in IEEE Power and Energy Society General Meeting, Pitsburg, PA,Jul. 2008, pp. 1–6.[35] M. Berrera, A. Dolara, R. Faranda, and S. Leva, “Experimental test of seven widely-adopted mppt algorithms,” in IEEE Bucharest PowerTech (BPT), Bucharest, Romania,Jun. 2009, pp. 1–8.[36] A. Abdelsalam, A. Massoud, S. Ahmed, and P. Enjeti, “High-performance adaptiveperturb and observe mppt technique for photovoltaic-based microgrids,” IEEE Trans.Power Electron., vol. 26, no. 4, pp. 1010–1021, Apr. 2011.52BIBLIOGRAPHY[37] N. Khaehintung, T. Wiangtong, and P. Sirisuk, “Fpga implementation of mppt usingvariable step-size p o algorithm for pv applications,” in Int. Symp. Communications andInformation Technologies (ISCIT), Oct. 2006, pp. 212–215.[38] M. Mohd Zainuri, M. Mohd Radzi, A. Soh, and N. Rahim, “Development of adaptiveperturb and observe-fuzzy control maximum power point tracking for photovoltaic boostdc-dc converter,” IET Renew. Power Gen., vol. 8, no. 2, pp. 183–194, Mar. 2014.[39] S. Kollimalla and M. Mishra, “Variable perturbation size adaptive p amp;o mppt algo-rithm for sudden changes in irradiance,” IEEE Trans. Sustain. Energy, vol. 5, no. 3, pp.718–728, Jul. 2014.[40] W. Xiao and W. Dunford, “A modified adaptive hill climbing mppt method for pho-tovoltaic power systems,” in IEEE Annu. Power Electronics Specialists Conf. (PESC),vol. 3, Aachen, Germany, Jun. 2004, pp. 1957 – 1963 Vol.3.[41] F. Liu, S. Duan, F. Liu, B. Liu, and Y. Kang, “A variable step size inc mppt methodfor pv systems,” IEEE Trans. Ind. Electron., vol. 55, no. 7, pp. 2622–2628, Jul. 2008.[42] Q. Mei, M. Shan, L. Liu, and J. Guerrero, “A novel improved variable step-sizeincremental-resistance mppt method for pv systems,” IEEE Trans. Ind. Electron.,vol. 58, no. 6, pp. 2427 –2434, Jun. 2011.[43] H. Koizumi and K. Kurokawa, “Plane division maximum power point tracking methodfor pv module integrated converter,” in IEEE Int. Symp. Industrial Electronics (ISIE),vol. 2, Quebec, Canada, Jul. 2006, pp. 1265–1270.[44] H. Koizumi, “A summary of plane division maximum power point tracking methods,”in IEEE Int. Conf. on Sustainable Energy Technologies (ICSET), Singapore, Singapore,Nov. 2008, pp. 306–311.53BIBLIOGRAPHY[45] H. Patel and V. Agarwal, “Maximum power point tracking scheme for pv systems op-erating under partially shaded conditions,” IEEE Trans. Ind. Electron., vol. 55, no. 4,pp. 1689–1698, Apr. 2008.[46] A. Zbeeb, V. Devabhaktuni, and A. Sebak, “Improved photovoltaic mppt algorithmadapted for unstable atmospheric conditions and partial shading,” in Int. Conf. CleanElectrical Power, Jun. 2009, pp. 320–323.[47] E. Koutroulis and F. Blaabjerg, “A new technique for tracking the global maximumpower point of pv arrays operating under partial-shading conditions,” IEEE J. Photo-volt., vol. 2, no. 2, pp. 184–190, Apr. 2012.[48] J. Bastidas-Rodriguez, E. Franco, G. Petrone, C. Ramos-Paja, and G. Spagnuolo, “Max-imum power point tracking architectures for photovoltaic systems in mismatching con-ditions: a review,” IET Power Electron., vol. 7, no. 6, pp. 1396–1413, Jun. 2014.[49] K. Chen, S. Tian, Y. Cheng, and L. Bai, “An improved mppt controller for photovoltaicsystem under partial shading condition,” IEEE Trans. Sustain. Energy, vol. 5, no. 3,pp. 978–985, Jul. 2014.[50] K. S. Tey and S. Mekhilef, “Modified incremental conductance algorithm for photovoltaicsystem under partial shading conditions and load variation,” IEEE Trans. Ind. Electron.,vol. 61, no. 10, pp. 5384–5392, Oct. 2014.[51] D. Sera, R. Teodorescu, J. Hantschel, and M. Knoll, “Optimized maximum power pointtracker for fast-changing environmental conditions,” IEEE Trans. Ind. Electron., vol. 55,no. 7, pp. 2629 –2637, Jul. 2008.54BIBLIOGRAPHY[52] R. Kadri, J.-P. Gaubert, and G. Champenois, “An improved maximum power pointtracking for photovoltaic grid-connected inverter based on voltage-oriented control,”IEEE Trans. Ind. Electron., vol. 58, no. 1, pp. 66 –75, Jan. 2011.[53] P. Midya, P. Krein, R. Turnbull, R. Reppa, and J. Kimball, “Dynamic maximum powerpoint tracker for photovoltaic applications,” in IEEE Annu. Power Electronics SpecialistsConf. (PESC), vol. 2, Jun. 1996, pp. 1710–1716 vol.2.[54] T. Esram, J. Kimball, P. Krein, P. Chapman, and P. Midya, “Dynamic maximum powerpoint tracking of photovoltaic arrays using ripple correlation control,” IEEE Trans.Power Electron., vol. 21, no. 5, pp. 1282–1291, Sep. 2006.[55] S. Jain and V. Agarwal, “Comparison of the performance of maximum power pointtracking schemes applied to single-stage grid-connected photovoltaic systems,” ElectricPower Applications, IET, vol. 1, no. 5, pp. 753–762, Sep. 2007.[56] J. Kimball and P. Krein, “Discrete-time ripple correlation control for maximum powerpoint tracking,” IEEE Trans. Power Electron., vol. 23, no. 5, pp. 2353–2362, Sep. 2008.[57] A. Bazzi and P. Krein, “Concerning maximum power point tracking for photovoltaic op-timization using ripple-based extremum seeking control,” IEEE Trans. Power Electron.,vol. 26, no. 6, pp. 1611–1612, Jun. 2011.[58] R. Khanna, Q. Zhang, W. Stanchina, G. Reed, and Z.-H. Mao, “Maximum power pointtracking using model reference adaptive control,” IEEE Trans. Power Electron., vol. 29,no. 3, pp. 1490–1499, Mar. 2014.[59] J. Blanes, F. Toledo, S. Montero, and A. Garrigos, “In-site real-time photovoltaic i-vcurves and maximum power point estimator,” IEEE Trans. Power Electron., vol. 28,no. 3, pp. 1234 –1240, Mar. 2013.55BIBLIOGRAPHY[60] T. Esram and P. Chapman, “Comparison of photovoltaic array maximum power pointtracking techniques,” IEEE Trans. Energy Convers., vol. 22, no. 2, pp. 439–449, Jun.2007.[61] N. Mutoh, M. Ohno, and T. Inoue, “A method for mppt control while searching for pa-rameters corresponding to weather conditions for pv generation systems,” IEEE Trans.Ind. Electron., vol. 53, no. 4, pp. 1055 –1065, Jun. 2006.[62] L. Cristaldi, M. Faifer, M. Rossi, and S. Toscani, “An improved model-based maximumpower point tracker for photovoltaic panels,” IEEE Trans. Instrum. Meas., vol. 63, no. 1,pp. 63–71, Jan. 2014.[63] X. Li, Y. Li, and J. Seem, “Maximum power point tracking for photovoltaic systemusing adaptive extremum seeking control,” IEEE Trans. Control Syst. Technol., vol. 21,no. 6, pp. 2315–2322, Nov. 2013.[64] W.-M. Lin, C.-M. Hong, and C.-H. Chen, “Neural-network-based mppt control of astand-alone hybrid power generation system,” IEEE Trans. Power Electron., vol. 26,no. 12, pp. 3571–3581, Dec 2011.[65] L. Elobaid, A. Abdelsalam, and E. Zakzouk, “Artificial neural network based maxi-mum power point tracking technique for pv systems,” in IEEE Annu. Conf. IndustrialElectronics (IECON), Montreal, QC, Oct. 2012, pp. 937–942.[66] M. Veerachary, T. Senjyu, and K. Uezato, “Neural-network-based maximum-power-point tracking of coupled-inductor interleaved-boost-converter-supplied pv system usingfuzzy controller,” IEEE Trans. Ind. Electron., vol. 50, no. 4, pp. 749–758, Aug. 2003.56BIBLIOGRAPHY[67] Syafaruddin, E. Karatepe, and T. Hiyama, “Artificial neural network-polar coordinatedfuzzy controller based maximum power point tracking control under partially shadedconditions,” IET Renew. Power Gen., vol. 3, no. 2, pp. 239–253, Jun. 2009.[68] A. Al Nabulsi and R. Dhaouadi, “Efficiency optimization of a dsp-based standalone pvsystem using fuzzy logic and dual-mppt control,” IEEE Trans. Ind. Electron., vol. 8,no. 3, pp. 573 –584, Aug. 2012.[69] C.-U. Lee, J.-S. Ko, T.-Y. Seo, D.-K. Kim, and D.-H. Chung, “The mppt control ofphotovoltaic system using fuzzy-pi controller,” in Int. Conf. Electrical Machines andSystems (ICEMS), Oct. 2013, pp. 303–305.[70] C.-S. Chiu, “T-s fuzzy maximum power point tracking control of solar power generationsystems,” IEEE Trans. Energy Convers., vol. 25, no. 4, pp. 1123–1132, Dec 2010.[71] C.-S. Chiu and Y.-L. Ouyang, “Robust maximum power tracking control of uncertainphotovoltaic systems: A unified t-s fuzzy model-based approach,” IEEE Trans. ControlSyst. Technol., vol. 19, no. 6, pp. 1516–1526, Nov. 2011.[72] B. Alajmi, K. Ahmed, S. Finney, and B. Williams, “A maximum power point trackingtechnique for partially shaded photovoltaic systems in microgrids,” IEEE Trans. Ind.Electron., vol. 60, no. 4, pp. 1596 –1606, Apr. 2013.[73] Y. Hu, J. Liu, and B. Liu, “A mppt control method of pv system based on fuzzy logic andparticle swarm optimization,” in Int. Conf. Intelligent System Design and EngineeringApplication (ISDEA), Jan. 2012, pp. 73–75.[74] Y. Liu, D. Xia, and Z. He, “Mppt of a pv system based on the particle swarm op-timization,” in Int. Conf. Electric Utility Deregulation and Restructuring and PowerTechnologies (DRPT), Jul. 2011, pp. 1094–1096.57BIBLIOGRAPHY[75] M. Miyatake, M. Veerachary, F. Toriumi, N. Fujii, and H. Ko, “Maximum power pointtracking of multiple photovoltaic arrays: A pso approach,” IEEE Trans. Aerosp. Elec-tron. Syst., vol. 47, no. 1, pp. 367–380, Jan. 2011.[76] M. Abdullah, A. Yatim, C. Tan, and A. Samosir, “Particle swarm optimization-basedmaximum power point tracking algorithm for wind energy conversion system,” in IEEEInt. Conf. on Power and Energy (PECon), Dec 2012, pp. 65–70.[77] K. Ishaque, Z. Salam, M. Amjad, and S. Mekhilef, “An improved particle swarm op-timization (pso) and based mppt for pv with reduced steady-state oscillation,” IEEETrans. Power Electron., vol. 27, no. 8, pp. 3627–3638, Aug. 2012.[78] K. Ishaque and Z. Salam, “A deterministic particle swarm optimization maximum powerpoint tracker for photovoltaic system under partial shading condition,” IEEE Trans. Ind.Electron., vol. 60, no. 8, pp. 3195–3206, Aug. 2013.[79] A. Bidram, A. Davoudi, and R. Balog, “Control and circuit techniques to mitigate partialshading effects in photovoltaic arrays,” IEEE J. Photovolt., vol. 2, no. 4, pp. 532–546,Oct. 2012.[80] F. Giraud and Z. Salameh, “Analysis of the effects of a passing cloud on a grid-interactivephotovoltaic system with battery storage using neural networks,” IEEE Trans. EnergyConvers., vol. 14, no. 4, pp. 1572–1577, Dec 1999.[81] M. Boztepe, F. Guinjoan, G. Velasco-Quesada, S. Silvestre, A. Chouder, andE. Karatepe, “Global mppt scheme for photovoltaic string inverters based on restrictedvoltage window search algorithm,” IEEE Trans. Ind. Electron., vol. 61, no. 7, pp. 3302–3312, Jul. 2014.58BIBLIOGRAPHY[82] E. Mamarelis, G. Petrone, and G. Spagnuolo, “Design of a sliding-mode-controlled sepicfor pv mppt applications,” IEEE Trans. Ind. Electron., vol. 61, no. 7, pp. 3387–3398,Jul. 2014.[83] N. Femia, D. Granozio, G. Petrone, G. Spagnuolo, and M. Vitelli, “Optimized one-cyclecontrol in photovoltaic grid connected applications,” IEEE Trans. Aerosp. Electron.Syst, vol. 42, no. 3, pp. 954–972, Jul. 2006.[84] M. Fortunato, A. Giustiniani, G. Petrone, G. Spagnuolo, and M. Vitelli, “Maximumpower point tracking in a one-cycle-controlled single-stage photovoltaic inverter,” IEEETrans. Ind. Electron., vol. 55, no. 7, pp. 2684–2693, Jul. 2008.[85] E. Sreeraj, K. Chatterjee, and S. Bandyopadhyay, “One-cycle-controlled single-stagesingle-phase voltage-sensorless grid-connected pv system,” IEEE Trans. Ind. Electron.,vol. 60, no. 3, pp. 1216–1224, Mar. 2013.[86] A. Luque and S. Hegedus, Handbook of Photovoltaic Science and Engineering. Wiley,2003.59
- Library Home /
- Search Collections /
- Open Collections /
- Browse Collections /
- UBC Theses and Dissertations /
- Photovoltaic maximum power point tracker with zero...
Open Collections
UBC Theses and Dissertations
Featured Collection
UBC Theses and Dissertations
Photovoltaic maximum power point tracker with zero oscillation and adaptive step Paz, Francisco 2014
pdf
Page Metadata
Item Metadata
Title | Photovoltaic maximum power point tracker with zero oscillation and adaptive step |
Creator |
Paz, Francisco |
Publisher | University of British Columbia |
Date Issued | 2014 |
Description | Maximum Power Point Tracking (MPPT) strategies in Photovoltaic (PV) systems ensure efficient utilization of PV arrays. Among different strategies, the Perturb and Observe (P&O) algorithm has gained wide popularity due to its intuitive nature and simple implementation. However, such simplicity in P&O introduces two inherent issues, an artificial perturbation that creates losses in steady-state operation and a limited ability to track transients in changing environmental conditions. This work develops and discusses in detail an MPPT algorithm with zero oscillation and slope tracking to address those technical challenges. The strategy combines three techniques to improve steady-state behavior and transient operation: 1) idle operation on the Maximum Power Point (MPP), 2) identification of the irradiance change through a natural perturbation and 3) a simple multi-level adaptive tracking step. Two key elements, which form the foundation of the proposed solution, are investigated: the suppression of the artificial perturbation at the MPP and the indirect identification of irradiance change through a current-monitoring algorithm which acts as a natural perturbation. The Zero-oscillation, Adaptive step Perturb and Observe (ZA-P&O) MPPT strategy builds on these mechanisms to identify relevant information and produce efficiency gains. As a result, the combined techniques achieve superior overall performance while maintaining simplicity of implementation. Simulations and experimental results are provided to validate the proposed strategy and illustrate its behavior in steady state and transient operation. |
Genre |
Thesis/Dissertation |
Type |
Text |
Language | eng |
Date Available | 2014-08-08 |
Provider | Vancouver : University of British Columbia Library |
Rights | Attribution-NonCommercial-NoDerivs 2.5 Canada |
DOI | 10.14288/1.0165912 |
URI | http://hdl.handle.net/2429/49955 |
Degree |
Master of Applied Science - MASc |
Program |
Electrical and Computer Engineering |
Affiliation |
Applied Science, Faculty of Electrical and Computer Engineering, Department of |
Degree Grantor | University of British Columbia |
GraduationDate | 2014-09 |
Campus |
UBCV |
Scholarly Level | Graduate |
Rights URI | http://creativecommons.org/licenses/by-nc-nd/2.5/ca/ |
AggregatedSourceRepository | DSpace |
Download
- Media
- 24-ubc_2014_september_paz_francisco.pdf [ 11.41MB ]
- Metadata
- JSON: 24-1.0165912.json
- JSON-LD: 24-1.0165912-ld.json
- RDF/XML (Pretty): 24-1.0165912-rdf.xml
- RDF/JSON: 24-1.0165912-rdf.json
- Turtle: 24-1.0165912-turtle.txt
- N-Triples: 24-1.0165912-rdf-ntriples.txt
- Original Record: 24-1.0165912-source.json
- Full Text
- 24-1.0165912-fulltext.txt
- Citation
- 24-1.0165912.ris
Full Text
Cite
Citation Scheme:
Usage Statistics
Share
Embed
Customize your widget with the following options, then copy and paste the code below into the HTML
of your page to embed this item in your website.
<div id="ubcOpenCollectionsWidgetDisplay">
<script id="ubcOpenCollectionsWidget"
src="{[{embed.src}]}"
data-item="{[{embed.item}]}"
data-collection="{[{embed.collection}]}"
data-metadata="{[{embed.showMetadata}]}"
data-width="{[{embed.width}]}"
async >
</script>
</div>
Our image viewer uses the IIIF 2.0 standard.
To load this item in other compatible viewers, use this url:
https://iiif.library.ubc.ca/presentation/dsp.24.1-0165912/manifest