IEEE JOURNAL OF EMERGING AND SELECTED TOPICS IN POWER ELECTRONICS, VOL. 2, NO. 4, DECEMBER 2014
1081
Microgrids Operation Based on Master–Slave Cooperative Control Tommaso Caldognetto, Student Member, IEEE, and Paolo Tenti, Fellow, IEEE Abstract— Low-voltage microgrids can be seen as the basic tiles of the smart grid patchwork owing to their capability to efficiently manage the distributed energy resources (DERs) in aggregate form. They can support the grid in terms of demand response, power quality, ride through capability, and at the same time, they can ensure electrical continuity to the loads, even in case of grid failure. This paper describes a simple and effective approach to manage microgrids by synergistic control of the power electronic interfaces acting therein, i.e., the utility interface (UI), installed at the point of common coupling with the utility and the energy gateways (EGs), interfacing the DERs with the distribution grid. The proposed master–slave control uses the UI as control master for the EGs. In grid-connected operation, the UI performs as a grid-supporting unit and dispatches active and reactive power references to the EGs so as to improve energy efficiency and power quality; in islanded operation, the UI performs as a grid-forming voltage source and ensures the power balance by exploiting every power source and energy storage unit available in the microgrid. This paper discusses the theoretical background, architecture, and algorithms of the proposed master–slave control and demonstrates the resulting microgrid performance by means of simulation and experimental results. Index Terms— Energy gateway (EG), master–slave control, microgrid, power sharing, utility interface (UI).
I. I NTRODUCTION OW-VOLTAGE microgrids will play a major role in future smart grids [1]. The presence of distributed microgeneration (MG) and energy storage (ES) owned by end users (prosumers) creates a new paradigm for electrical grids and a potentially huge new market for technology manufacturers, service providers, energy traders, distributors, and regulatory boards. Several challenges must be tackled; these challenges are in the area of technology, organization, standards, rules, and economy [2]–[4]. According to the new paradigm, the distribution grid can be seen as a patchwork where microgrids act as basic tiles, supporting the utility in terms of demand response, power quality, network dynamics management, and other ancillary services, also ensuring electrical continuity to the loads even in case of grid failure. A major goal of microgrids is to integrate and manage effectively every kind of distributed energy resource (DER), either MG or ES [5]. In fact, the increasing pervasiveness
L
Manuscript received January 31, 2014; revised April 28, 2014; accepted July 22, 2014. Date of publication July 31, 2014; date of current version October 29, 2014. Recommended for publication by Associate Editor Yan-Fei Liu. The authors are with the Department of Information Engineering, University of Padova, Padua 35131, Italy (e-mail:
[email protected];
[email protected]). Color versions of one or more of the figures in this paper are available online at http://ieeexplore.ieee.org. Digital Object Identifier 10.1109/JESTPE.2014.2345052
Fig. 1.
Structure of the considered microgrid.
of renewable energy sources [mainly photovoltaic (PV)] can cause misbehavior of the distribution grid due to overproduction around daytime [6], and also affects some economic assets of the electrical market [7]. The capability to control and plan the energy in- and out-flow of microgrids, seen as aggregate entities, will play a major role in ensuring stability, efficiency and cost-effectiveness of future smart grids [8]–[10]. For this aim, the distributed units must be driven cooperatively, and the control architecture must be flexible and scalable to integrate every type and number of DERs and dynamically adapt to load power demand and energy supply [11]–[15]. This paper proposes a microgrid control architecture which allows full exploitation of DERs, optimization of steadystate performance, both on-grid and off-grid, and effective management of transients. As observed from the utility terminals, the microgrid behaves as an aggregate prosumer with extended power capacity and wide control functions. Moreover, multiple microgrids can plug into the same electrical distribution system without affecting voltage or frequency stability. These features open new technical and economical scenarios, where communities of prosumers, aggregated in microgrids or pools of microgrids, may gain an increasing role in the electrical market by taking advantage of their cumulative energy resources and control capabilities. II. M ICROGRID S TRUCTURE The considered structure of low-voltage microgrid is shown in Fig. 1 with K nodes connecting end-users.
2168-6777 © 2014 IEEE. Personal use is permitted, but republication/redistribution requires IEEE permission. See http://www.ieee.org/publications_standards/publications/rights/index.html for more information.
1082
IEEE JOURNAL OF EMERGING AND SELECTED TOPICS IN POWER ELECTRONICS, VOL. 2, NO. 4, DECEMBER 2014
At the point of common coupling (PCC) with the utility (node 0), the microgrid is equipped with a utility interface (UI) with ES (UI-ES). The UI is driven as a voltage source and interacts, via bidirectional communication links (e.g., on power lines or wireless), with the other nodes of the microgrid. N grid nodes are active, i.e., linkup power sources and/or ES devices interfaced to the grid by energy gateways (EGs). EGs are driven as current sources [16] and have bidirectional communication with the UI. The remaining M nodes are passive, i.e., linkup loads only, and are equipped with smart meters (SMs), which perform local measurements and handle one-way communication to UI. The UI hosts the master control unit (MCU) and includes a power section equipped with three-phase grid-interactive inverter, ES device (e.g., battery or super-cap), and possibly, backup generator (e.g., microturbine, fuel cell, diesel gen-set). The EGs host the local control units (LCUs) and have a structure similar to that of UI. However, their grid-interactive inverters are generally single-phase, and ES is optional. III. M ASTER –S LAVE C ONTROL P RINCIPLE As mentioned before, with the proposed control approach, the UI permanently performs as a voltage source, while EGs are driven as current sources. In grid-connected operation, the UI behaves as a gridsupporting voltage source and can implement ancillary control functions (e.g., management of UI-ES, compensation of residual load unbalance and distortion). Since the power balance is ensured by the mains, the local control needs may prevail, and each active node makes available only its residual power and energy capacity for microgrid control. In spite of this limitation, the power flow from EGs can be adjusted by the MCU to meet global needs, e.g., grid voltage stabilization, power loss minimization, peak power shaving, demand response, day-ahead planning, and low-voltage ride through. A different scenario occurs in islanded operation, during transitions from on-grid to off-grid, and under black start. In these cases, the UI acts as a grid-forming voltage source, and the MCU manages the entire energy reserve of the microgrid to ensure power balance. The EGs keep behaving as current sources, but the whole energy generated and stored locally is made available for microgrid needs. The EGs can also be driven to a controlled overload condition to meet temporary energy needs. In all cases, the distributed units perform cooperatively. However, while in grid-connected operation, the local needs may prevail; in any other operating conditions, the microgrid needs are given higher priority. This change of priority does not require modification of control algorithms; it is simply determined by the MCU, which knows the power capacity of each DER by properly assigning the power commands to EGs. Compared with droop control approaches [17]–[20], the solution presented here has the drawback to require communication among microgrid nodes. On the other hand, it allows plug-and-play integration of several DERs without affecting microgrid voltage and frequency. Moreover, the
master–slave architecture allows concurrent operation of multiple microgrids, resulting in a higher level of coordination toward a multilayer organization of the entire smart grid [21]. In fact, if the UI is driven by droop control, the entire microgrid adapts to its voltage and frequency and appears, at the utility terminals, as a single droop-controlled voltage source with extended power rating. This makes it possible to plug in multiple microgrids to the same distribution grid while ensuring their seamless operation, according to droop control principles. IV. C ONTROL H IERARCHY The proposed control approach does not directly reflect the control hierarchy adopted for droop-based control of microgrids [17]–[22], and requires some clarifications to fit in. 1) Primary Control: In droop-controlled microgrids, the distributed units perform as voltage sources, and the primary control aims to adjust the amplitude and frequency of local voltage references, thus avoiding circulation of unwanted currents among DERs. It also allows plug-and-play connection of DERs. In our proposal, the EGs perform as current sources and automatically adapt to existing grid voltage and frequency. The principle to control the active and reactive currents to avoid useless circulation of power among DERs is still valid, as well as contributing to voltage and frequency stabilization by properly managing the power exchanges within the microgrid. Our primary control definition includes every control function that can be done locally, without inputs from the rest of the microgrid. Examples are compensation of reactive power and currents generated by local loads, management of local ES, support of local voltage if limits are exceeded, and emergency supply to local loads in case of microgrid failure. 2) Secondary Control: The meaning of secondary control in droop-controlled microgrids is to compensate for the amplitude and frequency deviations caused by droop control [22], [23]. This allows voltage stabilization, regular power flow, and generally, better operation of the microgrid. With our approach, frequency stabilization is not an issue; however, microgrid operation can still be improved by adjusting the set points of LCUs. To this end, the MCU can process the data collected in the entire microgrid and feed optimized references into EGs. Our secondary control definition includes every cooperative control function that can be implemented to improve the global operation of the microgrid. Examples are stabilization of voltage profiles, reduction of distribution and conversion losses, effective load power sharing among active nodes, prevention of potential instabilities caused by saturation of EGs power limits or storing capacity. 3) Tertiary Control: Commonly, tertiary control is committed to manage the interaction between microgrid and utility in grid-connected mode, ensure effective control
CALDOGNETTO AND TENTI: MICROGRIDS OPERATION BASED ON MASTER–SLAVE COOPERATIVE CONTROL
of the power flow at the utility terminals and provide smooth transitions from grid-connected to islanded mode and vice versa. This meaning is consistent with our control approach too. V. P OWER M ANAGEMENT This section describes how the microgrid is driven in gridconnected and islanded operations. Observe first that the number of active and passive nodes may change during time, depending on how many end-users are actually connected to the microgrid. Moreover, the active nodes may turn to passive node operation if their generated power is not enough to fulfill local needs. The control algorithm must therefore be devised to allow dynamic adjustment of microgrid parameters. The control performs as follows. At the beginning of each control period TS (lasting few line cycles), the UI, as control master, polls all the nodes of the microgrid. The active nodes return the values of active and reactive power, which are available for microgrid control, while passive nodes return their active and reactive power consumption. More in detail, the data packet sent by the nth EG (LCU) to UI (MCU) includes: 1) kilovoltampere power rating of local inverter (Sn ); 2) active and reactive power (PLn , Q Ln ) absorbed by local loads; 3) active power PGn generated by local power sources; 4) maximum additional energy that can be stored in the ES max ) and maximum energy that can be extracted unit (E in max ); from the ES unit (E out min , P max ) of the active 5) upper and lower limits (PGn Gn power deliverable by local energy resources, including min , P max ). Since P max > 0 (fed by ES power limits (PSn Sn Sn min ES) and PSn < 0 (absorbed by ES), we have min min = sat (PGn + PSn ) PGn
(1.a)
max max PGn = sat (PGn + PSn ).
(1.b)
−An An
The saturation function sat(.) points out that the actual power that can be delivered is also limited by the power rating An of the inverter. The same data structure can be used for passive nodes by disregarding the fields related to power generation. After collecting data from all nodes, the control master computes the total power consumed and generated by the microgrid as PLtot = PGtot =
K k=1 N n=1
PLk =
M
PLm +
m=1 min PGn , PGtot =
N
PLn
(2.a)
n=1 N n=1
min max PGn PGtot =
N
1083
A. Grid-Connected Operation In this case, power control is noncritical since the mains ensure the power balance. Thus, the LCUs can choose the power to deliver, their actual choice being dependent on the kind of local power source. In fact, renewable energy sources (e.g., wind or PV) should be fully exploited, while cost/benefit issues can drive the choice for other types of sources (e.g., small hydro, fuel cells, gas turbines). In any case, EGs can feed reactive power to face loads demand, thus reducing distribution losses, improving node voltage stability, and increasing the power factor at PCC. By request of the MCU, EGs can also adjust their active power flow, at the expense of local stored energy. This can be done to meet special needs of the microgrid (e.g., voltage support, thermal limitation in feeders, intentional islanding conditions) or to respond to utility power requests. B. Islanded Operation In this case, the power balance must be ensured within the microgrid. We distinguish two situations. Case B: G L ): In this situation, the total > Ptot 1) Overgeneration (Ptot power generated by distributed sources exceeds the loads consumption. Under steady-state conditions, the extra power is stored in distributed ES devices according to their state of charge, and the EGs are driven accordingly. Under transient conditions, the dynamic power unbalance is temporarily faced at the expense of the energy stored in UI-ES, since the UI acts as voltage source and automatically fulfills every dynamic power request. However, within few line cycles, the EGs power commands are adapted to the new situation, and the load power demand is shared among DERs. The state of charge of UI-ES is promptly restored to ensure the capability to face new transients. If overgeneration lasts too long, the power generated by renewable sources is scaled down to meet the actual power consumption, according to a suitable sharing criterion. Within their kilovoltampere ratings, the EGs can also feed reactive power to meet load demand and stabilize node voltages. Case B: G < P L ): In this situation, the 2) Under Generation (Ptot tot power generated within the microgrid is not enough to fulfill the loads demand. The power balance must therefore be ensured by taking advantage of the distributed ES units, according to their state of charge. Clearly, this kind of operation can be maintained for a limited time. Then, nonpriority loads must be disconnected to prevent full discharge of ES units, and in particular, of UI-ES. In addition, in this case, EGs can feed reactive power, within their kilovoltampere capability, to stabilize node voltages and reduce distribution losses.
max PGn . (2.b)
n=1
The control master then executes a control algorithm that depends on the operating mode (grid-connected or islanded) and the relative amount of generated and absorbed power.
VI. D ISTANCE -BASED P OWER S HARING C RITERION As mentioned in the previous section, the MCU performs its control functions by assigning proper power commands to EGs, according to a suitable power sharing criterion. Here, we refer to the distance-based power sharing criterion,
1084
IEEE JOURNAL OF EMERGING AND SELECTED TOPICS IN POWER ELECTRONICS, VOL. 2, NO. 4, DECEMBER 2014
formerly presented in [24], which is simple and intuitive. It is based on the remark that distribution loss is a general index of microgrid performance. In fact, its reduction means increase of energy efficiency, limitation of thermal stress in the feeders, reduction of voltage drops across distribution lines, and increase of the power factor at PCC. Note also that the distribution loss decreases if each load is supplied by the generators nearby. A general and effective load power sharing criterion is therefore to split the power demand of each passive node among all active nodes inversely to their distance. This approach only requires to know node-to-node distances, and it was demonstrated in [24] and [25] that it leads to quasiminimum losses. Moreover, it can easily be adjusted to comply with EGs power limits. The mathematical formulation of the control algorithm is the following. Let D be the K × K matrix of node to node distances (excluding node 0) 0 d12 . . . d1K d 0 . . . d2K . (3) D = 21 . . . . . . . . . ... dK 1 dK 2 . . . 0 p
From (3), we may derive submatrix D a , which is an N × M matrix, the generic element dnm of which gives the distance between active node n and passive node m. It can be obtained by 1 d1 d12 . . . d1M d1 d2 . . . d M 2 p T 2 2 (4) Da = K a D K p = ... ... ... ... d1 d2 . . . d M N
N
N
where K a is an N × K matrix the nth row of which is zero except for the element which corresponds to the node where the nth EG is connected, which is set to one. K p is a similar M × K matrix built for passive nodes. Let S˙m = Pm + j Q m be the complex power absorbed by the load connected at the generic passive grid node m (m = 1 . . . M), and S˙n = Pn + j Q n be the complex power delivered by the EG at the generic active node n (n = 1 . . . N). In grid-connected operation, the MCU shares the power demand S˙m of each passive node among all active nodes n, including the utility at PCC, in inverse proportion to their distance dmn from node m. Accordingly, the complex power S˙mn requested from the passive node m to the active node n is N −1 1 1 . (5) S˙mn = Pmn + j Q nm = S˙m n dm dmn n=1
The total power requested to the active node n becomes S˙n =
M
S˙mn = Pn + j Q n .
(6)
m=1
In practice, this power request can exceed the actual power capability of node n. Thus, sharing criterion (5) must be adjusted to comply with the actual power capability of
active nodes. For this purpose, (5) is rewritten in the form N −1 N −1 βn P βn Q β β n P n Q n + j Qm n (7) S˙m = Pm n dm dmn dm dmn n=0
n=0
where terms βn P and βn Q (which range from nearly zero to one) account for the residual active and reactive power capacity at node n. They are updated at each control step, thus ensuring fast adaptation of the sharing criterion to any deviations from the power balance. The power sharing criterion (7) ensures dynamic saturation management, i.e., it distributes the load power demand among the various EGs according to their actual power capacity. The actual control step duration is limited by the calculation time of power terms (half line period at least) and communication delays. As a matter of fact, the above control requires communication, data processing, and measurement capability at every grid node, both active and passive. The required communication speed is, however, limited and can be met by commercial power line communication devices, e.g., based on PoweRline Intelligent Metering Evolution technology with orthogonal frequency-division multiplexing. As concerns the evaluation of node-to-node distances, they may be available by the distribution system operator. Otherwise, online estimation can be done by applying ranging techniques on the power lines, e.g., based on time-of-arrival measurements [26], [27]. As compared with other cooperative control techniques employing node-to-node communication, e.g., those described in [28]–[30], the proposed control approach has some interesting features: it only requires communication of nontimecritical data, i.e., power terms which are averaged over a line period; it does not need accurate synchronization among microgrid nodes; the control algorithm requires simple computations, for both MCU and LCUs, and does not rely on a detailed knowledge of electrical distribution network parameters. Moreover, the different levels of control (internal loops, primary control, and secondary control) do not interfere since the proposed control strategy prioritizes and decouples their domains of intervention. Finally, the saturation of the power and energy capabilities of distributed units are considered and automatically managed by the power-sharing algorithm, in both grid-connected and islanded operations. This prevents potential instabilities and ensures proper microgrid behavior even during heavy transients. Harmonic compensation techniques can also be implemented by taking advantage of the wide control bandwidth of EG inverters, e.g., by adopting the technique discussed in [31]. VII. A PPLICATION E XAMPLE The proposed master–slave control was tested for different network topologies and operating conditions, both static and dynamic, in grid-connected and islanded modes. The microgrid operation was analyzed through MATLAB/Simulink simulations, where EGs and UI were modeled in detail, including the inverter structure and the internal control loops, to test the control system in a realistic implementation scenario. The actual behavior of the control was also verified on an
CALDOGNETTO AND TENTI: MICROGRIDS OPERATION BASED ON MASTER–SLAVE COOPERATIVE CONTROL
Fig. 2.
1085
Simulated microgrid. TABLE I B US L ENGTHS
Fig. 3. Current behavior at the transition from grid-connected to islanded operation.
TABLE II P OWER R ATINGS OF UI AND EG
experimental test bed. The results are reported in the following section. A. Simulation Results For the sake of brevity, we consider the simulation of the simple microgrid in Fig. 2, which includes the UI, an EG, and a single resistive load. The parameters of distribution lines BG, B1, B2 are given in Table I, their values being those measured in the experimental test bed. In grid-connected operation, the UI compensates for the reactive power and harmonic distortion measured at PCC, and keeps control over the state of charge of its ES device (UI-ES). In islanded operation, the UI becomes the only voltage source of the microgrid, thus compensating for any power needs of the microgrid. In both cases, the MCU controls EG too, to ensure proper power management. An adequate sizing of the storage and backup units of the UI is needed to guarantee the energy balance within the microgrid. The storage unit must face temporary or light power unbalances, while the backup unit intervenes in case of persistent or heavy unbalances. The power ratings of UI and EG are given in Table II. The nominal line voltage is VG_RMS = 240 V. Initially, the microgrid operates in grid-connected mode and the load resistance is set to R L = 13.1 , corresponding to PL = 4.4 kW. The load power measurement and power-sharing algorithm (7) are updated every 180 ms. The power-sharing criterion is implemented by imposing zero active power exchange with UI. This implies that within the EG power capability, the load power is shared
Fig. 4.
Current behavior at load step change.
between the EG and the grid based on the inverse-distance criterion. At t = 5 s, the microgrid switches to islanded operation. The time behavior of the currents in the microgrid around the transition is shown in Fig. 3. Before the transition, the current fed by the EG equals the current drawn from the grid, the load being at the middle point between the grid and EG (Table I). When the mains disconnect, the UI keeps on as voltage source and feeds the power initially provided by the grid, seamlessly changing its role from grid-supporting to grid-forming unit. Thus, following the transition, the UI-ES begins discharging, but the MCU intervenes and commands the EG to inject the total load power. Further, the MCU requests the EG to feed an additional amount of active power to restore the state of charge of UI-ES. At t = 10 s, the load is suddenly changed to R L = 38.4 , corresponding to PL = 1.5 kW. This gives rise to a transient when UI temporarily absorbs the excess of generated power, until EG power references are updated to restore the power balance. Fig. 4 shows the behavior of the microgrid currents around the load step. At the first update of EG commands, the power balance in the microgrid is restored and UI-ES begins
1086
IEEE JOURNAL OF EMERGING AND SELECTED TOPICS IN POWER ELECTRONICS, VOL. 2, NO. 4, DECEMBER 2014
TABLE III S ETUP PARAMETERS
Fig. 5.
Power and energy fed by the UI along the simulation cycle.
Fig. 7. Fig. 6.
Active power measured at microgrid nodes.
Experimental setup organization.
discharging down to the initial state of charge. A portion of the load power is, therefore, diverted to UI by the MCU, which correspondingly adapts the EG power commands. Fig. 5 shows the power and energy delivered by UI along the entire simulation cycle. In case of transients, the MCU ensures the power balance using the energy stored in UI-ES, then drives the microgrid into a new steady-state condition where the load power is shared between the EG and the mains and the state of charge of UI-ES is restored. B. Experimental Results Fig. 6 shows the setup of the experimental test bed. The distribution grid topology is the same as that of Fig. 2; however, the UI is moved at PCC and is replaced by a second EG. The microgrid is, therefore, composed of two current controlled inverters (EGs) and a parallel RC load. The parameters are given in Table III. The communication between units is achieved via a dedicated Ethernet network. The data exchanges and the internal
states of the controllers are recorded for monitoring and postprocessing purposes. The execution rate of the power-sharing algorithm (7) and the power measurement updates are set very slow (5 s) to allow extensive data logging, system monitoring, and visualization. At every iteration, the MCU dispatches the updated power references to EGs, and subsequently, acquires the power data measured at load terminals and the actual power capacity of the EGs. The design and implementation of the experimental setup took advantage of hardware-in-the-loop, real-time simulation, and rapid prototyping tools as a guide to develop, test, and debug the entire microgrid prototype [32]. The experimental results reported here show the response of the islanded system to the connection and disconnection of the resistive–capacitive load specified in Table III. In particular, Figs. 7 and 8 show the power measurements at the nodes of the microgrid. The variables are shown with a time step of 5 s [points (a), . . . ,(i), in the horizontal axis]. Initially, the load is disconnected, and the two EGs exchange
CALDOGNETTO AND TENTI: MICROGRIDS OPERATION BASED ON MASTER–SLAVE COOPERATIVE CONTROL
1087
The change in load absorption is acquired at instant (d). EG2 already being saturated, the additional active power demand is charged to EG1 , which, however, can meet the request partially, since its residual power capacity amounts to 200 W. Its residual reactive power capacity is also limited to 500 VA, while the ideal power-sharing algorithm (5) would require 640 VA. This leads EG1 into saturation for both active and reactive power. Finally, the load disconnection occurs at instant (g), and the corresponding zero power references are dispatched by MCU at instant (i). Then the system returns to the initial state. The green trace in Figs. 7 and 8 represents the power contribution of the UI, which guarantees the power balance within the microgrid during transient conditions. Fig. 8.
Reactive power measured at microgrid nodes.
VIII. C ONCLUSION
Fig. 9. Test bed response to power reference update at instant (c) of Figs. 7 and 8. From top to bottom: load voltage [500 V/div], load current [10 A/div], EG1 current [8 A/div], and EG2 current [8 A/div]. Time scale: 20 ms/div.
zero power with the grid. Immediately after load connection, the total absorbed power rises up to 2900 W. This is the power acquired at instant (b) by the MCU. If the pure distance-based criterion (5) was applied, it would produce a power sharing between EG1 and EG2 approximately equal to (1.9 kW, 1 kW). This power request exceeds EG2 power capacity P2S , which is limited to 500 W (Table III). In this situation, the saturation management algorithm (7) comes into play, allowing EG2 beta coefficient (β2 P ) to decrease until P2S constraint is fulfilled. Hence, the resulting active power sharing becomes (2400 W, 500 W). On the other hand, the ideal reactive power sharing given by (5) corresponds to (−570 VA, −320 VA) and remains within the power capability of both EGs. The saturation management algorithm (7) does not intervene, and correspondingly, coefficients β Q remain stuck at one. The calculated EGs power references are dispatched by MCU at instant (c), and Fig. 9 shows the waveforms acquired immediately after the EGs execute the power commands. The power step is accomplished smoothly, and the voltage-support effect of the EGs current injection causes the load power absorption to increase to (3.11 kW, −1 kVA), due to the voltage increase across the load itself.
A scalable and flexible microgrid architecture and master–slave control strategy have been presented, which allow plug-and-play integration of DERs and ensure efficient and reliable operation of the microgrid in every operating condition. Communication among microgrid nodes is needed; however, the required bit rate and computational complexity are within the capability of state-of-the-art SMs. A control algorithm was also devised, which is capable of fully exploiting any available energy sources thanks to a suitable distance-based load power sharing technique. The proposed architecture and control strategy were implemented and validated by simulations and experimental tests in various microgrid configurations and operating modes. The master–slave control demonstrated good steady-state performance, in terms of distribution efficiency, voltage stability, and power quality, and stable operation even during severe dynamic conditions, e.g., in case of nonintentional islanding. R EFERENCES [1] M. Shahidehpour and S. Pullins, “Is the microgrid buzz real?” IEEE Electr. Mag., vol. 2, no. 1, pp. 2–5, 2014. [2] L. Tao, C. Schwaegerl, P. Mancarella, G. Strbac, N. Hatziargyriou, and B. Buchholz, “European roadmap for microgrids,” presented at the CIGRE, 2010. [3] H. Farhangi, “The path of the smart grid,” IEEE Power Energy Mag., vol. 8, no. 1, pp. 18–28, Jan./Feb. 2010. [4] D. E. Olivares et al., “Trends in microgrid control,” IEEE Trans. Smart Grid, vol. 5, no. 4, pp. 1905–1919, Jul. 2014. [5] X. Tan, Q. Li, and H. Wang, “Advances and trends of energy storage technology in microgrid,” Int. J. Elect. Power Energy Syst., vol. 44, no. 1, pp. 179–191, Jan. 2013. [6] T. Stetz, F. Marten, and M. Braun, “Improved low voltage gridintegration of photovoltaic systems in Germany,” IEEE Trans. Sustainable Energy, vol. 4, no. 2, pp. 534–542, Apr. 2013. [7] H. Sugihara, K. Yokoyama, O. Saeki, K. Tsuji, and T. Funaki, “Economic and efficient voltage management using customer-owned energy storage systems in a distribution network with high penetration of photovoltaic systems,” IEEE Trans. Power Syst., vol. 28, no. 1, pp. 102–111, Feb. 2013. [8] P. Basak, S. Chowdhury, S. Halder nee Dey, and S. P. Chowdhury, “A literature review on integration of distributed energy resources in the perspective of control, protection and stability of microgrid,” Renew. Sustain. Energy Rev., vol. 16, no. 8, pp. 5545–5556, 2012. [9] G. Parise, L. Martirano, and L. Parise, “Ecodesign of ever netload microgrids,” IEEE Trans. Ind. Appl., vol. 50, no. 1, pp. 10–16, Jan./Feb. 2014.
1088
IEEE JOURNAL OF EMERGING AND SELECTED TOPICS IN POWER ELECTRONICS, VOL. 2, NO. 4, DECEMBER 2014
[10] R. Majumder, “Aspect of voltage stability and reactive power support in active distribution,” IET Generat., Transmiss. Distrib., vol. 8, no. 3, pp. 442–450, Mar. 2014. [11] J. Rocabert, G. M. S. Azevedo, A. Luna, J. M. Guerrero, J. I. Candela, and P. Rodríguez, “Intelligent connection agent for three-phase gridconnected microgrids,” IEEE Trans. Power Electron., vol. 26, no. 10, pp. 2993–3005, Oct. 2011. [12] R. N. Anderson, A. Boulanger, W. B. Powell, and W. Scott, “Adaptive stochastic control for the smart grid,” Proc. IEEE, vol. 99, no. 6, pp. 1098–1115, Jun. 2011. [13] A. L. Dimeas and N. D. Hatziargyriou, “Operation of a multiagent system for microgrid control,” IEEE Trans. Power Syst., vol. 20, no. 3, pp. 1447–1455, Aug. 2005. [14] J.-Y. Kim et al., “Cooperative control strategy of energy storage system and microsources for stabilizing the microgrid during islanded operation,” IEEE Trans. Power Electron., vol. 25, no. 12, pp. 3037–3048, Dec. 2010. [15] A.-H. Mohsenian-Rad, V. W. S. Wong, J. Jatskevich, R. Schober, and A. Leon-Garcia, “Autonomous demand-side management based on game-theoretic energy consumption scheduling for the future smart grid,” IEEE Trans. Smart Grid, vol. 1, no. 3, pp. 320–331, Dec. 2010. [16] E. Serban and H. Serban, “A control strategy for a distributed power generation microgrid application with voltage- and current-controlled source converter,” IEEE Trans. Power Electron., vol. 25, no. 12, pp. 2981–2992, Dec. 2010. [17] J. M. Guerrero, M. Chandorkar, T. Lee, and P. C. Loh, “Advanced control architectures for intelligent microgrids—Part I: Decentralized and hierarchical control,” IEEE Trans. Ind. Electron., vol. 60, no. 4, pp. 1254–1262, Apr. 2013. [18] W. Dan, T. Fen, T. Dragicevic, J. C. Vasquez, and J. M. Guerrero, “Coordinated primary and secondary control with frequency-bussignaling for distributed generation and storage in islanded microgrids,” in Proc. 39th Annu. Conf. IEEE Ind. Electron. Soc. (IECON), Nov. 2013, pp. 7140–7145. [19] K. De Brabandere, B. Bolsens, J. Van den Keybus, A. Woyte, J. Driesen, and R. Belmans, “A voltage and frequency droop control method for parallel inverters,” IEEE Trans. Power Electron., vol. 22, no. 4, pp. 1107–1115, Jul. 2007. [20] J. M. Guerrero, J. C. Vasquez, J. Matas, L. G. de Vicuña, and M. Castilla, “Hierarchical control of droop-controlled AC and DC microgrids—A general approach toward standardization,” IEEE Trans. Ind. Electron., vol. 58, no. 1, pp. 158–172, Jan. 2011. [21] J. Vasiljevska, J. A. P. Lopes, and M. A. Matos, “Evaluating the impacts of the multi-microgrid concept using multicriteria decision aid,” Electr. Power Syst. Res., vol. 91, pp. 44–51, Oct. 2012. [22] A. Bidram and A. Davoudi, “Hierarchical structure of microgrids control system,” IEEE Trans. Smart Grid, vol. 3, no. 4, pp. 1963–1976, Dec. 2012. [23] A. Bidram, A. Davoudi, F. L. Lewis, and J. M. Guerrero, “Distributed cooperative secondary control of microgrids using feedback linearization,” IEEE Trans. Power Syst., vol. 28, no. 3, pp. 3462–3470, Aug. 2013. [24] P. Tenti, A. Costabeber, T. Caldognetto, and P. Mattavelli, “Improving microgrid performance by cooperative control of distributed energy sources,” in Proc. ECCE—IEEE Energy Convers. Congr. Exposit., Sep. 2013, pp. 1647—1654. [25] P. Tenti, A. Costabeber, F. Sichirollo, and P. Mattavelli, “Minimum loss control of low-voltage residential microgrids,” in Proc. 38th Annu. Conf. IEEE Ind. Electron. Soc. (IECON), Oct. 2012, pp. 5650–5656. [26] T. Erseghe, F. Lorenzon, S. Tomasin, A. Costabeber, and P. Tenti, “Distance measurement over PLC for dynamic grid mapping of smart micro grids,” in Proc. IEEE Int. Conf. Smart Grid Commun. (SmartGridComm), Oct. 2011, pp. 487–492.
[27] A. Costabeber, T. Erseghe, P. Tenti, S. Tomasin, and P. Mattavelli, “Optimization of micro-grid operation by dynamic grid mapping and token ring control,” in Proc. 14th Eur. Conf. Power Electron. Appl. (EPE), Aug./Sep. 2011, pp. 1–10. [28] T. L. Vandoorn, J. D. M. De Kooning, B. Meersman, and L. Vandevelde, “Review of primary control strategies for islanded microgrids with power-electronic interfaces,” Renew. Sustain. Energy Rev., vol. 19, pp. 613–628, Mar. 2013. [29] J. M. Guerrero, L. Hang, and J. Uceda, “Control of distributed uninterruptible power supply systems,” IEEE Trans. Ind. Electron., vol. 55, no. 8, pp. 2845–2859, Aug. 2008. [30] S. Bolognani and S. Zampieri, “A distributed control strategy for reactive power compensation in smart microgrids,” IEEE Trans. Autom. Control, vol. 58, no. 11, pp. 2818–2833, Nov. 2013. [31] A. Costabeber, P. Tenti, T. Caldognetto, and E. V. Liberado, “Selective compensation of reactive, unbalance, and distortion power in smart grids by synergistic control of distributed switching power interfaces,” in Proc. 15th Eur. Conf. Power Electron. Appl. (EPE), Sep. 2013, pp. 1–9. [32] T. Caldognetto, S. Buso, and P. Mattavelli, “Digital controller development methodology based on real-time simulations with LabVIEW FPGAc hardware-software toolset,” Electron. J., vol. 17, no. 2, pp. 110–117, Dec. 2013.
Tommaso Caldognetto (S’10) received the M.S. (Hons.) degree in electronic engineering from the University of Padova, Padua, Italy, in 2012, where he is currently pursuing the Ph.D. degree in information engineering with the Graduate School, Department of Information Engineering. His current research interests include real-time simulation for power electronics, design of controllers for microgrid applications, and power electronic architectures for distributed energy sources.
Paolo Tenti (M’85–SM’90–F’98) is a Professor of Electronics with the Department of Information Engineering, University of Padova, Padua, Italy. His current research interests include industrial and power electronics, electromagnetic compatibility, application of modern control methods to power electronics, EMC analysis of electronic equipment, and cooperative control of distributed electronic power processors in smart grids. Dr. Tenti was an Executive Board Member of the IEEE Industry Applications Society (IAS) and the Chair of various society committees from 1991 to 2000. He served as the IEEE IAS President, in 1997, and chaired the IEEE World Conference on Industrial Applications of Electrical Energy in Rome in 2000. From 2000 to 2001, he was the IEEE-IAS Distinguished Lecturer of electromagnetic compatibility in industrial equipment. From 2002 to 2008, he served the University of Padova in the capability of Department Director and the Chairman of the Board of Directors. He is also the President of CREIVen, an industrial consortium for research in industrial electronics with a special emphasis on electromagnetic compatibility.
