Photovoltaic Power Systems: A Review of Topologies ... - IEEE Xplore

Photovoltaic Power Systems: A Review of Topologies, Converters and Controls N.D Marks1 , T.J. Summers2 and R.E. Betz3 School of Electrical Engineering and Computer Science University of Newcastle, Australia, 2308 Email:1 [email protected]; 2 [email protected]; 3 [email protected]

Abstract—Renewable energy resources will likely be an integral part of future electrical systems. Photovoltaics in particular are receiving significant research attention in terms of developing the photovoltaic (PV) system to improve efficiency and to meet grid codes, as well as developing intelligent methods of control. This paper presents a review of the main established PV system topologies as well as the emerging multilevel converter based topologies. The Model Predictive Control (MPC) method for converter control is also presented, with particular emphasis on the key features for PV applications.

significantly on a day to day basis, as well as on a minute to minute basis. The control system must be able to track these changes and manage the power output of the system effectively. Section III of this paper will discuss Model Predictive Control (MPC) which is emerging as an excellent converter control scheme with considerable flexibility and features for PV applications.

II. PV POWER SYSTEM TOPOLOGIES I. INTRODUCTION Renewable energy is viewed as the alternative to our classical dependence on fossil fuels and harmful environmental emissions [1], [2]. Renewable energy includes wind, solar, hydro and fuel cells and over the last decade, these technologies have formed a significant part of the content throughout the literature [3]. This trend is unlikely to change over the next decade as a plethora of challenges remain with respect to the cost and controllability of these energy resources [4]. Photovoltaic power research in particular has garnered significant attention due to renewable energy initiatives and reduction in cost. In contrast, the installed capacity of photovoltaic power is much lower than that of wind and hydro power [1]. However, as the cost continues to decrease and the efficiency increases, it is expected that photovoltaic power will become a significant part of the renewable energy sector throughout the world [1], [3], [5]. Power electronic converters and their control schemes are the backbone of photovoltaic power systems [4]. The way in which the photovoltaic cells and the converters are arranged varies among the literature, with each topology offering something over the others in terms of controllability, efficiency or cost. A large amount of research continues in the area of photovoltaic power system topologies along with their controls [1], [2], [4]–[10]. Section II of this paper presents a review of the key topologies, outlining their benefits and drawbacks, as well as identifying where the technology is currently being developed. One of the challenges for photovoltaic power systems is the variability in the available power. The available power depends entirely on the irradiation conditions which can vary

A variety of photovoltaic power system topologies exist owing to the range of situational requirements and rapidly advancing state of the art. PV topologies have evolved in both practicality and complexity, to become a highly active area of research, particularly with respect to high power applications. A. GENERAL TOPOLOGIES There are four general classes of PV topologies, with the multilevel topologies building from these. The four classes of topologies are: 1) Centralised, 2) String, 3) AC Modules and 4) Multistring [5], [6]. 1. C ENTRALISED T OPOLOGY Fig. 1 shows the centralised topology. A single inverter interfaces PV strings to the grid [4]. These strings are created by connecting individual panels in series to provide a larger DC voltage so that a boost converter is not required. Several strings are placed in parallel with a common DC bus to increase the available power [5]. Diodes are required in the strings to prevent reverse current flow. The centralised topology is an attractive option due to its simplicity [6]. However there are a number of significant drawbacks that have rendered this topology obsolete [3]–[6], [9], [10]: • Diode conduction losses reduce the efficiency of the topology • Poor power output and power loss since all the PV panels are controlled by a single MPPT scheme – Partial shading and module mismatches cause power loss

Figure 2.

Figure 1.

• •

Centralised Topology [6], [10]

The centralised inverter is rated for a certain number of panels and strings, limiting future addition Strings are connected via high voltage DC cables

2. S TRING T OPOLOGY Fig. 2 shows the string topology. The strings are constructed in the same way as in the centralised inverter case but without the diode. In this topology, each string has its own inverter (typically single phase) to interface with the grid (common AC bus). A boost converter can be used in the string if the number of panels is insufficient to produce the required DC voltage. The string topology is very common in grid connected PV power systems since a string and its inverter is how a household system is structured. The topology is widely used due to the benefits it provides over the centralised topology [3], [5], [6]: •

• • •

String Topology [6]

Each string has its own MPPT scheme which improves the power output – Partial shading and module mismatches are less severe but remain present Adding new strings is easier since there is no central inverter to limit the power No diode losses Generally smaller distance of high voltage DC cable (if any)

3. AC M ODULES Fig. 3 shows the AC module topology. Each PV module has a dedicated inverter to interface with a common AC bus (typically the grid) [5]. AC modules are most suited to small power applications, including domestic use. The topology offers the best MPPT since there is one inverter per panel [6]. More modules can easily be added to the system in a “plug and play” fashion [3]. The main disadvantage of the AC module topology is that voltage elevation is required, whether that be a boost converter or transformer,

Figure 3.

AC Modules Topology [6]

adding cost and reducing the efficiency [3], [5], [6]. 4. M ULTISTRING T OPOLOGY The multistring topology is shown in Fig. 4. It aims to achieve a better configuration with the advantages of the centralised and string topologies [6]. In general, PV strings are interfaced to a common DC bus with DC-DC converters which control the MPPT of the string. A centralised inverter interfaces with the grid (single or three phase) [2], [3], [6], [9]. The power losses due to partial shading and mismatches are reduced due to the individual string MPPT [2], [3], and the DCDC converters ensure the central inverter operates with a set voltage DC link [10]. Extra strings can be added to the system, provided the central inverter has the capacity [5], [6], [9]. Strings can also be removed from the system in case of failure or damage, as well as for maintenance purposes due to the modularity of the system. The multistring topology has some disadvantages such as [6], [9], [10]: • High voltage DC cables may still be required depending on the physical arrangement of the PV strings and converters • Increased system complexity – Extra converters and sensors – More control objectives and problems • Poor utilisation of central inverter if too few strings are attached • Expandability limited by the central inverter capacity B. MULTILEVEL SPECIFIC TOPOLOGIES The limiting factor for the multistring topology is the power rating of the central inverter. As the PV technology improves and the size of PV generation plants increases, the central

Figure 4.

Multistring Topology [6], [10]

Figure 5.

inverter will need to process larger levels of power [6], [9], [11]. The centralised inverter has generally been a simple two level voltage source inverter [1], [9], [10]. Higher power requirements cannot be achieved by a single inverter; as such, multiple inverters may be required to interface the power [9]. Multilevel inverters can be connected to larger voltages, up to about 6.6kV with current technology [9], [11]. A higher voltage level allows significantly more power to be transferred with a single converter. The higher voltage connection can also eliminate the need for a transformer on the AC side, thereby increasing efficiency and reducing the cost of the system [2], [4], [10], [11]. Total Harmonic Distortion (THD) and common mode voltages are also reduced when using multilevel inverters [1], [4], [6], [10]– [12] due to the redundancy in the switching states when a higher number of levels are used [11]. A number of multilevel inverter topologies exist; however for PV applications, two in particular have received continued attention throughout the literature; namely the Neutral Point Clamped (NPC) and the Cascaded H-Bridge (CHB) inverters [1], [2], [6], [8]–[11], [13]. The main feature of these topologies is that they have multiple DC links available for the connection of either single PV modules or PV strings. In this way, the efficiency and performance of MPPT schemes can be preserved while also increasing the power processing capability of the system. Multilevel inverter applications are receiving increased attention throughout the literature in an attempt to extract larger amounts of power from PV applications with increased efficiency. Furthermore, power quality issues can more easily be solved with multilevel inverters due to redundant switching states and a higher number of voltage levels. 1. C ASCADED H-B RIDGE T OPOLOGIES The CHB inverter has an ideal structure for use with PV systems since each of the cells requires an isolated DC power source. The CHB has the nice feature that extension to more levels requires only a linear increase in the number of components. Two applications in particular have received attention in the literature: 1) Single String/Single Module CHB [2], [6], [13], [14] and 2) Multistring CHB [9].

CHB Single String Topology [2], [6], [13]

i) Single String/Single Module The single string/module topology is shown in Fig. 5. They have been developed primarily for single phase systems where either strings of PV modules or single PV modules are connected to the isolated DC sources of the CHB cells. MPPT control is performed using the control scheme for the inverter. The inverter transfers the power by switching in each of the cells in an appropriate fashion so that power is extracted from the DC link capacitor which is charged by the real power produced by the PV. The control scheme is more complex than for the other topologies due to the common AC grid current for each of the CHB cells and the inherently unbalanced DC link voltages [2], [6], [13], [14]. ii) Multistring The multistring CHB topology is shown in Fig. 6. The multistring topology has been developed for three phase systems to increase the amount of power the system can generate and transfer. The DC link of each CHB cell is a common DC bus to which multiple strings of PV modules are connected via DC-DC converters [9]. The MPPT scheme is performed by the DC-DC converters for each string while regulating the common DC bus voltage [9]. The coupling of a DC-DC converter to each string increases modularity so that more strings can be added if capacity is available, or strings can be removed due to faults or maintenance. Ideally the DC links of the individual cells are regulated to the same voltage. The control scheme for this topology is more complex again due to the three phase nature. The control scheme must be able to manage the power transfer across three phases but also within each phase [9]. Each phase is likely to have a different amount of power available leading to unbalanced currents if this is not compensated for. The different power available in a single phase’s cells also needs to be accounted for in the modulation of the inverter [9]. A compensation method for both of these issues and good simulation results are presented in [9].

Figure 8.

Figure 6.

CHB Multistring Topology [9]

NPC Multistring Topology [10]

increased since the PV strings are independent of one another. The NPC inverter is controlled to deliver the real power to the grid or load, without the need for MPPT. Some form of unbalance compensation is required since the two DC buses are likely to be transferring different amounts of power [10]. III. CONTROL

Figure 7.

NPC Single String Topology [1]

2. N EUTRAL P OINT /D IODE C LAMPED T OPOLOGIES The NPC inverter also has a suitable structure for PV systems. Individual DC voltage levels are available for the connection of PV strings or modules in a similar way to the CHB. The disadvantage of the NPC is that extension to more levels requires a geometric increase in the number of components. i) Single String The single string/module topology is shown in Fig. 7. A PV string or module is connected across each capacitor in the NPC topology to provide the DC power. The topology can be utilised for single or three phase applications although three phase is preferred to transfer as much power as possible. The three phases share the DC link in this converter so the main control issue is the correct balancing of the DC link capacitor voltages [1]. A voltage balancing strategy that provides MPPT for three PV arrays with four panels in each is implemented in [1]. ii) Multistring A multistring topology is shown in Fig. 8. where there are two common DC buses corresponding to the two capacitor voltages of the NPC inverter. Multiple strings of PV modules are connected to these DC buses via DC-DC boost converters to maintain the voltage on the bus at a constant level. The DC converters also control the MPPT of the PV strings. The dual DC bus configuration allows a larger DC link voltage without adding more levels to the inverter. Modularity is also

The complexity of the control scheme in PV power systems can vary considerably depending on the size and structure of the system. The fundamental elements of control for these systems are the converter control schemes and the PV control schemes. The control schemes for converters are well established. The primary method used is Voltage Oriented Control (VOC) with some form of modulation scheme. The main modulation schemes used with multilevel converters are Phase Shifted PWM, Level Shifted PWM and Space Vector Modulation. These control schemes are covered comprehensively in [8], [11], [12]. The power output of PV arrays varies according to solar irradiation conditions. To ensure that the maximum power is always extracted from the PV array, Maximum Power Point Tracking (MPPT) schemes are used. MPPT schemes are covered in detail throughout the literature and continue to be an active area of research. A. MODEL PREDICTIVE CONTROL (MPC) MPC is a control method that uses a model of the system under control to predict the effect of control actions and select the control action that minimises some cost function [15]. MPC is a very powerful and useful tool for power electronic converters as it allows the discrete nature of the converter to be exploited and used to minimise distortion, reduce switching losses, reduce common mode voltage and effectively balance capacitor voltages in converters [15]–[20]. MPC for inverters is termed Finite Control Set MPC (FCSMPC) since there are finite, discrete control actions that can be applied [15]. The control problem thus reduces to choosing the appropriate switching combination to minimise the cost function. Fig. 9 shows a three phase two level voltage source inverter connected to the grid via an R-L load. The most important part of predictive control schemes is the model of the system since all the control decisions are based on the model. In three phase systems this modelling is usually performed by making use of space vectors.

The inverter switching states can be represented as a vector [19]: S=

2 (Sa + aSb + a2 Sc ) 3

(1)

√

Where, a = (− 12 ) + (j 23 ) and Sa , Sb , Sc are the states of the phase leg switches. They can be 1 or 0, with 1 corresponding to the top switch on (bottom off) and vice versa [19]. The output voltage vector is then given by [19]: v = VDC S

(2)

Using KVL, the expression for the system using space vectors is [19]: di +e (3) dt The system must be converted to discrete time to be implemented in reality. This is achieved by approximating the derivative in (3) to obtain the discrete time current expression [19]: v = Ri + L

i(k) =

1 [Li(k − 1) + Ts v(k) − Ts e(k)] RTs + L

(4)

The model expressions are used to predict the current at the end of the next control interval for each possible switching state of the inverter. The combination that minimises the cost function is chosen as the control action. Typical cost functions can have the forms [19]: g = |x∗ − x|

(5)

g = |x∗ − x|2

(6)

1 g= T

ZT

[x∗ − x]dt

(7)

This method is relatively simple for a two level inverter with 8 different states. However, FCS-MPC is applicable to multilevel converters with a significantly larger number of possible switching combinations that would need to be evaluated in the model. This problem rapidly becomes a computational burden, with the number of evaluations requiring too much time to be implemented in real time at reasonably fast control frequencies [18], [20], [21]. This issue has been addressed in a CHB in [20] where only the switching vectors adjacent to the last applied vector are evaluated in each control interval. In real power grid connected applications the grid voltage (and hence the required inverter output voltage and current) are reasonably uniform and adjacent vectors are likely to be the ones selected via the cost function anyway. This method is used primarily for accurately tracking the reference current. To include other conditions in the cost function such as common mode voltage,

switching losses and the switching frequency, the adjacent vectors may not be the most appropriate. A different approach to reducing the number of vectors to be evaluated is addressed in [18] for a 19 level CHB StatCom. The switching states are considered on a per phase basis rather than as a three phase space vector. This method has been used as a key control objective was to tightly regulate the DC capacitor voltages in each H-Bridge cell. A dead-beat current controller was used to regulate the three phase currents and the MPC is applied in each phase to select the appropriate HBridges to produce the desired voltage while also balancing the capacitors and reducing switching transitions. The key feature of the MPC control scheme used in [18] is that the capacitor voltages can be regulated to any desired voltage level while simultaneously retaining the harmonic performance of traditional PS-PWM and LS-PWM schemes for multilevel converters. This has excellent implications for PV applications as MPPT can be integrated into the converter control scheme and variable power avilability among the HBridge cells is easily managed. Furthermore, loss of power in individual cells due to shading or module failure can be handled elegantly with the ability to regulate the capacitor voltage and maintain the full stack voltage. FCS-MPC applied to inverter control has primarily focused on one step ahead predictions. Reference [22] has presented results for multiple prediction steps ahead applied to controlling variable speed drives. The results show that an increase in performance is achieved relative to other control methods when the longer horizon is used. However, the significant drawback of predicting further ahead is the rapid increase in the number of switching combinations that must be evaluated [22], [23]. The computational burden becomes too large if all possibilities are to be evaluated. Instead, [23] constrains switching events by only considering transitions when the output is close to its bound and using a search tree to find the optimal switching sequence. The required computational effort of FCS-MPC for higher level inverters may be addressed in the future by parallel processing methods. FCS-MPC can easily be adapted for parallel processing which may lead to superior performance and longer prediction horizons [18]. The CHB topology has an ideal structure for PV power systems due to isolated DC links. Higher level number increases power output and also the ability to simultaneously achieve multiple control objectives due to significant switching state redundancy. Developments in MPC for multilevel converters have demonstrated the ability to independently control H-Bridge capacitor voltages without sacrificing harmonic performance. The need for extra converters or transformers in PV systems is thus removed, offering superior performance at lower cost. These features are key to the future of renewable power systems to ensure cost is minimised, efficiency is maximised and system integrity is maintained. The application of MPC to PV power systems using CHB inverters will be the topic of future

Figure 9.

Three Phase Grid Connected Inverter [19]

research to investigate fully integrated applications for large scale PV power systems and realise the benefits offered. IV. CONCLUSION A review of PV power system topologies has been presented. Multilevel converter topologies offer several advantages for PV power systems such as higher voltage and power capability, as well as improved harmonic performance to meet grid codes. Intelligent control systems for multilevel converters also allow integration of MPPT schemes to remove extra converters and increase the efficiency. PV power system research in the future is likely to focus on multilevel converter based systems, especially for high power installations. Control development has focused on minimising distortion, increasing efficiency, capacitor voltage balancing and has recently moved to fault detection and fault tolerant control. MPC has started to find application in power converter control due to its usefulness and simplicity. MPC has the potential to be used in PV power systems as mentioned, and future research in this area will likely result in superior performance with respect to traditional converter control methods. This will be the topic of future research to be undertaken in extension to this paper. R EFERENCES [1] S. Busquets-Monge, J. Rocabert, P. Rodriguez, S. Alepuz, and J. Bordonau, “Multilevel diode-clamped converter for photovoltaic generators with independent voltage control of each solar array,” Industrial Electronics, IEEE Transactions on, vol. 55, no. 7, pp. 2713 –2723, july 2008. [2] E. Villanueva, P. Correa, J. Rodriguez, and M. Pacas, “Control of a single-phase cascaded h-bridge multilevel inverter for grid-connected photovoltaic systems,” Industrial Electronics, IEEE Transactions on, vol. 56, no. 11, pp. 4399 –4406, nov. 2009. [3] F. Blaabjerg, F. Iov, T. Terekes, R. Teodorescu, and K. Ma, “Power electronics - key technology for renewable energy systems,” in Power Electronics, Drive Systems and Technologies Conference (PEDSTC), 2011 2nd, feb. 2011, pp. 445 –466. [4] J. Carrasco, L. Franquelo, J. Bialasiewicz, E. Galvan, R. Guisado, M. Prats, J. Leon, and N. Moreno-Alfonso, “Power-electronic systems for the grid integration of renewable energy sources: A survey,” Industrial Electronics, IEEE Transactions on, vol. 53, no. 4, pp. 1002 –1016, june 2006. [5] S. Kjaer, J. Pedersen, and F. Blaabjerg, “A review of single-phase gridconnected inverters for photovoltaic modules,” Industry Applications, IEEE Transactions on, vol. 41, no. 5, pp. 1292 – 1306, sept.-oct. 2005.

[6] S. Kouro, B. Wu, A. Moya, E. Villanueva, P. Correa, and J. Rodriguez, “Control of a cascaded h-bridge multilevel converter for grid connection of photovoltaic systems,” in Industrial Electronics, 2009. IECON ’09. 35th Annual Conference of IEEE, nov. 2009, pp. 3976 –3982. [7] R. Faranda, S. Leva, and V. Maugeri, “Mppt techniques for pv systems: Energetic and cost comparison,” in Power and Energy Society General Meeting - Conversion and Delivery of Electrical Energy in the 21st Century, 2008 IEEE, july 2008, pp. 1 –6. [8] S. Kouro, M. Malinowski, K. Gopakumar, J. Pou, L. Franquelo, B. Wu, J. Rodriguez, M. Pe? andrez, and J. Leon, “Recent advances and industrial applications of multilevel converters,” Industrial Electronics, IEEE Transactions on, vol. 57, no. 8, pp. 2553 –2580, aug. 2010. [9] S. Rivera, S. Kouro, B. Wu, J. Leon, J. Rodriguez, and L. Franquelo, “Cascaded h-bridge multilevel converter multistring topology for large scale photovoltaic systems,” in Industrial Electronics (ISIE), 2011 IEEE International Symposium on, june 2011, pp. 1837 –1844. [10] S. Kouro, K. Asfaw, R. Goldman, R. Snow, B. Wu, and J. Rodriguez, “Npc multilevel multistring topology for large scale grid connected photovoltaic systems,” in Power Electronics for Distributed Generation Systems (PEDG), 2010 2nd IEEE International Symposium on, june 2010, pp. 400 –405. [11] J. Rodriguez, L. Franquelo, S. Kouro, J. Leon, R. Portillo, M. Prats, and M. Perez, “Multilevel converters: An enabling technology for highpower applications,” Proceedings of the IEEE, vol. 97, no. 11, pp. 1786 –1817, nov. 2009. [12] J. Rodriguez, J.-S. Lai, and F. Z. Peng, “Multilevel inverters: a survey of topologies, controls, and applications,” Industrial Electronics, IEEE Transactions on, vol. 49, no. 4, pp. 724 – 738, aug 2002. [13] O. Alonso, P. Sanchis, E. Gubia, and L. Marroyo, “Cascaded h-bridge multilevel converter for grid connected photovoltaic generators with independent maximum power point tracking of each solar array,” in Power Electronics Specialist Conference, 2003. PESC ’03. 2003 IEEE 34th Annual, vol. 2, june 2003, pp. 731 – 735 vol.2. [14] S. Khajehoddin, A. Bakhshai, and P. Jain, “The application of the cascaded multilevel converters in grid connected photovoltaic systems,” in Electrical Power Conference, 2007. EPC 2007. IEEE Canada, oct. 2007, pp. 296 –301. [15] S. Kouro, P. Cortes, R. Vargas, U. Ammann, and J. Rodriguez, “Model predictive control: A simple and powerful method to control power converters,” Industrial Electronics, IEEE Transactions on, vol. 56, no. 6, pp. 1826 –1838, june 2009. [16] C. Townsend, T. Summers, and R. Betz, “Optimisation of switching losses and harmonic performance using model predictive control of a cascaded h-bridge multi-level statcom,” in Energy Conversion Congress and Exposition (ECCE), 2011 IEEE, sept. 2011, pp. 3159 –3166. [17] R. Vargas, P. Cortes, U. Ammann, J. Rodriguez, and J. Pontt, “Predictive control of a three-phase neutral-point-clamped inverter,” Industrial Electronics, IEEE Transactions on, vol. 54, no. 5, pp. 2697 –2705, oct. 2007. [18] C. Townsend, T. Summers, and R. Betz, “Multigoal heuristic model predictive control technique applied to a cascaded h-bridge statcom,” Power Electronics, IEEE Transactions on, vol. 27, no. 3, pp. 1191 – 1200, march 2012. [19] J. Rodriguez, J. Pontt, C. A. Silva, P. Correa, P. Lezana, P. Cortes, and U. Ammann, “Predictive current control of a voltage source inverter,” Industrial Electronics, IEEE Transactions on, vol. 54, no. 1, pp. 495 –503, feb. 2007. [20] P. Cortes, A. Wilson, S. Kouro, J. Rodriguez, and H. Abu-Rub, “Model predictive control of multilevel cascaded h-bridge inverters,” Industrial Electronics, IEEE Transactions on, vol. 57, no. 8, pp. 2691 –2699, aug. 2010. [21] J. Rodriguez, J. Pontt, P. Cortes, and R. Vargas, “Predictive control of a three-phase neutral point clamped inverter,” in Power Electronics Specialists Conference, 2005. PESC ’05. IEEE 36th, june 2005, pp. 1364 –1369. [22] T. Geyer, “Generalized model predictive direct torque control: Long prediction horizons and minimization of switching losses,” in Decision and Control, 2009 held jointly with the 2009 28th Chinese Control Conference. CDC/CCC 2009. Proceedings of the 48th IEEE Conference on, dec. 2009, pp. 6799 –6804. [23] T. Geyer, “Computationally efficient model predictive direct torque control,” Power Electronics, IEEE Transactions on, vol. 26, no. 10, pp. 2804 –2816, oct. 2011.