Centralized Control for DC Microgrid Using Finite State Machine 1 Mahmoud Saleh, Student Member, IEEE, Yusef Esa, Student Member, IEEE, and Ahmed Mohamed, Member, IEEE Abstract— In this paper, an autonomous communication-based centralized control for DC microgrids (MG) has been developed and implemented. The proposed controller enables smooth transition between various operating modes. Finite state machine (FSM) has been used to mathematically describe the various operating modes (states), and the events that may lead to mode changes (transitions). Therefore, the developed centralized controller aims at optimizing the performance of MG during all possible operational scenarios, while maintaining its reliability and stability. Results of selected drastic cases have been presented. These results show stable transition between modes, verifying the validity and applicability of the proposed controller. Index Terms--Centralized Control, Communication based control, DC microgrid, Finite state machine, Smart grid, State flow I. INTRODUCTION mart grids are intelligent and dynamically interactive realtime power grids, introduced with the urgent need for more resilient, and self-healing capabilities. The realization of smart grid became more feasible with the outburst of new technologies, e.g. information and communications technology (ICT), advanced metering infrastructure (AMI), microgrids, electrical vehicles, etc. Essentially it involves integration of measuring circuits, sensors and communication networks, among various technologies, that is capable of supervising the electric grid system, and control of the energy flow in real time [1-3]. Microgrid is one of the main pillar of smart grids, which prominently contribute to its level of resiliency, and could indemnify for partial grid generation loss [4-6]. Microgrid is a local distribution network comprising distributed energy resources (DER), energy storage systems (ESS) and loads, which aggregately function in a controlled fashion as a single entity, in either a grid-tied mode or an islanded mode [7]-[8]. Each of the MG components is controlled via a dedicated local controller. Local controllers must be coordinated to maintain load/generation balance, and optimize the performance of a MG. Coordination can be achieved through one of the following three architectures. Communication-less control: where there are no communication links between local controllers, and the deviation in the DC bus voltage is used to signal load/generation unbalances [9-10]. Therefore, it is also known as voltage droop control. This architecture has an advantage of being relatively reliable since it does not depend on communication systems and their inherent drawbacks (e.g. delays and packet loss). However, some of the main drawbacks of this architecture are suboptimal microgrid operation, and circulating currents. Distributed communication-based control: where communication links exist between the local controllers (LCs). LCs coordinate their decision, but final control decisions are being processed and executed locally. This architecture, if properly designed, is immune to single point failure, i.e. MG can maintain full operation even with the failure of some of the S LCs or communication links. However, it introduces complexity, and cannot achieve optimal MG operation [11]. Centralized communication-based control: where event/command signals between the LCs and a supervisory centralized controller, are being transmitted/received directly. Since the supervisory controller has global information on the various MG components, a predesigned algorithm can be implemented to achieve any desired objective, e.g. optimal operation. However, the reliability of this architecture is dependent on that of the communication system being utilized [11]. Development of an automated control system, which preserves MG stability/reliability during all possible scenarios while maintaining optimal operation is badly needed. Therefore, this challenge has attracted the attention of many researchers globally. However, the research that has been performed to address this challenge either focused on a specific case [12], or did not consider all possible scenarios/modes of operations. Moreover, it was mostly focused on designing a controller that guarantees stable MG operation [13]-[16], overlooking some other essential performance features including: energy/economic savings during grid connection, mitigating renewable energy intermittency, etc. In this paper, an autonomous centralized control, will be developed for DC MG, and implemented using finite state machine (FSM). The controller includes a forethought control logic implemented within, to guarantee MG reliability and resiliency during various modes of operation. II. MODEL ARCHITECTURE The DC microgrid topology considered in this paper comprises the following: a 1.5 kW battery system integrated to a 300V common DC bus through a bidirectional DC-DC charger; a 4 kW peak photovoltaic (PV) system that is connected to the DC bus through a DC-DC boost converter; a 6 kW controlled DC load that is connected to the DC bus; an ACDC bidirectional smart inverter tying the main grid to the common DC bus. As illustrated in Fig. 1, a microgrid Central Controller (MGCC) is used to manage and control the microgrid. The various individual converters are controlled locally through their LCs. Monitoring the AC and DC buses is being achieved by an AC agent and a DC one, respectively. The AC agent is used to detect any voltage/frequency deviations according to IEEE standards [18]. The DC agent monitors the DC bus voltage deviation [19]. The LCs, AC and DC agents, along with the solid state relays (SSRs) from the protection system, all report their statuses/events to the MGCC. The MGCC processes all inputs through a predesigned logic yielding the proper mode/action to be performed, to optimally achieve stability and reliability. The control hierarchy for the DC MG, is a communication based scheme. In the primary layer, the local controllers are state driven (i.e. controlling their respective converters by 2 S1 AC Agent S10 CInv Mode 1 Grid A V SSRPCC DC Agent A A A V V Inverter V S2 S3 S11 Solar panels CBi Mode 2 SSRBo Boost converter S1 S4 MGCC S8 , S9 Mode 3 Report AC voltage/ frequency CBo level violations S2: Report DC voltage level Batteries violations SSRBi S3 SSRPCC fault signal S4 SSRPC fault signal S5 Bi- fault signal Bidirectional converter S6 Utility control signal S7 Energy price (high/low) Load shedding CL S8 SOC (depleted/charged) S9 Batteries over discharge S10 Inverter status S11 Boost status DC S6 ,S7 S5 Utility Loads DC bus Pulses/ Commands Measurements Status/ Mode Transition Fig. 1. Microgrid architecture continuously monitoring certain state variables), which requires incessant communication, e.g. voltage/current measurements and pulse signals to the switches of the converters. In the secondary layer, the modes and set points are being assigned to each LC by the MGCC, to maintain the required voltage level within the DC MG and optimize its operation. The MGCC is event driven, i.e. it does not require continuous communication with the LCs. The devised logic within the MGCC was conceived such that the MGCC only takes an action if new event occurs, which reduces the communication requirement within the secondary layer, e.g. the required bandwidth. The tertiary controller handles the functionalities that exceed the boundaries of a single microgrid, e.g. utility control of a group of microgrids. The main focus of this paper is the secondary layer. III. SECONDARY CONTROL USING FINITE STATE MACHINE Finite state machine (FSM) provides a mathematical means to represent a finite number of states, and describe the transition between them. To represent the proposed control scheme, FSM has been used. FSM can be thought of as a machine with a finite number of operational conditions called states, or in this case, “modes.” The machine can be in only one state at any given time, and can transition to other states based on events. The FSM for the entire DC MG control could be represented by the following variables (∑, M, m00, δ), where: ∑: is a finite set of inputs to the MGCC, which are the events in Table 1. M: is a finite, non-empty set of modes, which comprises all modes sets: - M00 (Grid-tie/Energy saving mode) = {m00, m01, m02}, such that ∀m0j (m0j∈ M00 → m0j ∈ M) where j = , , . - M10 (Islanding mode) = {m10, m11, m12, m13}, such that ∀m1j (m1j∈ M10 → m1j ∈ M) where j = , , , . - M20 (Utility mode) = {m20}, such that m30 ∈ M. - M30 (Shutdown mode) = {m30}, such that m40 ∈ M. Therefore M = {M00, M01, M03, M04}, which will be more explicitly explained in section V. m00: is the initial mode of the FSM, m00 ∈ M. δ: is the mode transition function; δ: M × ∑ → M, it can be seen in the transition Table 1. The control action of the LCs is dependent upon the mode/submode of operation being triggered by the MGCC, e.g. during m00 (grid-tied sub-mode), CBo (boost converter LC) is MPPT controlled, while during m13, it is voltage controlled. The modes are based on the event signals being transmitted to the MGCC, and the logic/algorithm implemented within, which will trigger a proper transition. Transitions will be discussed further in section IV. The levels of load shedding are supposed to be dynamic within the islanding mode. The logic was prepared to have downstream load shedding during M10, i.e. no reconnection of loads happens unless normal operation is retained, to preserve the safety of the loads. The levels of load shedding will be selected based on the emergency loads, available resources at the instant of load shedding, and a margin of safety to account for solar intermittency. An average constant load within the time interval of the load shedding process was assumed. Four levels of load shedding where used within the islanding mode: (1) the first level of load shedding that is triggered at the moment of islanding. This level is function of the available resources; (2) the second level of load shedding takes place when the batteries are depleted; it is related to the battery and solar systems sizes, and the percentage of current that will be assigned to the batteries to charge with, e.g. less than 50% of I1c (rated current at rate of charge one C). The percentage of the I1c depends on the amount of the emergency loads and its priority; (3) the third level occurs when the boost converter is tripped or its sunset, and is function of the battery size; and (4) the last level of load shedding is triggered when the bidirectional converter is tripped while the boost converter is still available. At this level of load shedding, it is preferable to keep a minimal portion of the total load, and only keep the loads that are less sensitive to voltage variations, in case of intermittency. However, if voltage variations are significant, transition to M 30 will occur to preserve load safety. IV. OPERATIONAL MODES LAYERS AND TRANSITION CONDITIONS The architecture of the control scheme starts with the main layer that encompasses four modes: (1) grid-tied (energy saving) mode, which is the initial mode M00; (2) islanded mode M01; (3) utility mode M03; and (4) shutdown mode M04. M00 is assumed to be the initial mode of the entire FSM. It aims at achieving economic operation. A. Main layer Within this layer, several transitions may occur. For instance, if a grid outage takes place and gets reported by SSR PCC (S3 = 0), or the AC agent signals AC voltage/frequency exceeding permissible limits and lasting beyond the average clearing time specified in [36] to the MGCC (S1 = 0), a transition to M10 will happen. However, in case the utility sends a signal to take control of the MG (S6= 1), if all MG resources are available, a transition to M20 will occur, and if all resources are not available at any given instant, a transition to M30 will happen. Other transitions may happen depending on the events being triggered, which are summarized in Table. 1 and Fig. 2. B. Grid-tie/Energy saving Mode Transitions within that mode are triggered by S7, S5 and/or S8, e.g. starting with m00, the MGCC commands the inverter to maintain the DC bus voltage, CBo to operate as MPPT and the bidirectional converter LC (CBi) to be idle, i.e. current control 3 with Iref = 0. If the utility signals the MGCC that the energy price is low (S7 = 1), the MGCC checks the last status of CBi to confirm that the battery is not full (S8 = 0), and the SSRBi to assure no fault operation (S5 = 1), a transition to m01 takes place. In m01 the bidirectional starts charging the battery system with I1c to exploit the advantage of low energy price, while the inverter and the boost converter are still maintaining the same operation from m00. However, if the energy price is high (S7 = 0), the SSRBi last report was no fault operation (S5 = 1), and the CBi last signal that the battery has the capability to discharge (S8 = 1), a transition to m02 happens. In m02 the LCs maintain the same tasks of m00, except for the bidirectional converter, starts discharging with the maximum current I1c. The rest of transitions can be observed from Fig. 2 and Table. 1. C. Islanding Mode M10 is triggered when SSRPCC signals the circuit breaker to open and informs the MGCC with that event (S3 = 0), e.g. due to a power outage. M10 includes four sub-modes, islanding m10, contingency m11, critical m12, and extreme m13 sub-mode. A number of transitions may possibly takes place, one at a time. For example, starting from sub-mode m10, shown in Fig. 2, the MGCC trigger the first level of load shedding, CBo to operate as MPPT, and CBi to fix the DC bus voltage. If the SSRBi detects a fault where it is located, e.g. due to abnormal operation of the bidirectional converter or a fault, and report it (S5 = 0), a transition to m13 will happen. In m13, CBo maintains the DC bus voltage, and a maximum level of load shedding will be triggered. At this level, the load almost equals ~10-20% of the solar peak power that was available within the previous hour. The type of loads connected during that level of load shedding should be able to handle a wide range of voltage variation, since the battery is not available to buffer the expected power intermittency. If the DC voltage changes beyond the permissible limits of [36] during m13 (S2 = 0), a transition to m13 will take place to protect the loads as shown in Fig. 2. However, starting from m10, if CBi informs that the batteries are depleted (S8 = 0), while the boost converter is still available, through checking the last status signal from the SSRBo by the MGCC (S4 = 1), a transition T11 to m11 will occur. In this case, the second level of load shedding will take place, such that a portion of the available solar energy is used to charge the batteries with a maximum of half I1C, to guarantee continuous operation of the loads for as long as possible. The rest of energy is used to maintain supplying the remaining loads. This happens while the boost converter is MPPT controlled, and the bidirectional converter is still maintaining the DC bus voltage. On the other hand, also starting from m10, if solar intermittency occurred while the bidirectional converter is maintaining the DC bus voltage, the battery system might over discharge to keep supplying the loads. If this takes place for a time interval that is greater than the settling time of the nested PI of the bidirectional converter LC (S9 = 1), or the SSRBo signals that the boost convert is tripped (S4 = 0), then a transition T17 to m12 will take place. The reason for adding the settling time of the converter PI to the transition condition, is to guarantee that the transient oscillations of the bidirectional converter LC does not lead to false transitions. The controller oscillations may happen when the converter is trying to maintain the DC bus voltage. A condition to prevent over discharge was added, to extend the battery system availability during this mode. The rest of the transitions could be observed from Fig. 2, and Table. 1. D. Utility and shutdown Modes The goal of the utility mode M20, is for the utility to take control over the MG, according to a predefined agreement, e.g. to aggregate multiple MGs for virtual power plant operation, or enhance the voltage level by injecting reactive power. In this case, the inverter LC is performing current or P/Q control. As for the shutdown mode M30, its objective is to guarantee a shutdown in case all the resources are tripped or not available at any given moment, or in case the permissible limits of the DC voltage are violated during m13, in order not to jeopardize the loads safety. V. RESULTS AND DISCUSSIONS In order to illustrate the operation of the proposed secondary communication-based FSM-enabled controller, some selected scenarios will be presented and discussed. Each case is presented by a set of four subplots. Each subplot has five sections, each section reflects a new event or a set of events taking place. The (a) subplot of each figure represents DC currents for the inverter, boost converter, bidirectional converter and DC load. Subplots (b), (c) and (d), demonstrate the DC bus voltage, AC current from the inverter to the grid, and AC voltage, respectively. A. Case One This case shows the operation of the MGCC to control the DC MG in case of islanding, showing solar intermittency impact on the transition between modes of operation among other events. Section (1) displays the transition T 11 from m00 to m10, due to a blackout being signaled by SSRPCC (S3= 0). The first level of load shedding is triggered by the MGCC, reducing DC load current and dropping the AC current from the MG to the grid to zero as shown in Figs. 3(a) and 3(c), respectively. During m10, the MGCC commands CBi to maintain the DC bus voltage to 300 V, and CBo to maintain MPPT control, which can be noticed in Fig. 3(a). Throughout section (2), solar intermittency caused the batteries to over discharge beyond 5 A, for a period of time longer than the settling time of CBi PI (200 msec). Consequently, S8 becomes one, signaled by CBi, therefore, the MGCC checks the last status of SSRBi, if there is no fault (S5 = 1), a transition T12 to m12 happens. In m12 another level of DC load shedding is introduced, that can be seen from the DC load current in Fig 3-a; while the LCs maintain the same duties as in m10. ). During m12, the LCs maintain the same duties as in m10. During section (3), the boost converter is tripped, and in section (4), it is reconnected. Concurrently, C Bi reacts to maintain the DC bus voltage (i.e. discharging when the boost converter was tripped, and charging when it was reconnected). Section (5) represents the tripping of the bidirectional converter, where the bidirectional output current drops to zero, triggering m14. Within m14 maximum load shedding is triggered, CBo is commanded to switch to voltage control, where the DC load current drops to ~10-20% of solar energy, and becomes equal the boost converter current, which can be noticed in Fig. 3(a). It can be observed from Fig. 3(b), the DC bus voltage during all events was maintained within the acceptable limits, by using the suggested FSM logic to maneuver various critical situations. 4 Table 1. Transitions among various modes/sub-modes of DC MG Transition condition Transition Description Tripping of the Inverter triggered M01, bidirectional converter fixing DC bus voltage and boost S3 == 0 converter operating at MPPT while the first level of load shedding is triggered S3 == 1 && S1 == 1 Synchronization conditions are set for reconnection the DC MG to the grid S3==0 && S5 == 0 && S4 ==0 Tripping of the Inverter, Boost and Bi-directional converter will cause the MG to shut down S6 == 0 Utility control over the MG is disabled thus the operation is restored to grid tie mode S6 == 1 && S3 ==1 && S5 == 1&& S4 ==1 After confirming that all resources are operational the utility is granted control over the MG S3 ==0 && S5 == 0&& S4 ==0 Tripping of the Inverter, Boost and Bi-directional will cause the MG to shut down Tripping of the Boost and bidirectional converter, or the DC bus voltage deviated beyond the S5 == 0 &&(S4 == 0 || (S2 == 0 && after (ts))) limits, ts is added to guarantee unnecessary transition due to disturbance during settling time (ts) of the PI S8 == 1 || S5 == 0 Batteries are full charged or bidirectional converter is tripped S7 == 1 && S8 == 0 && S5 == 1 Energy price is low, batteries are not fully charged, and the SSR Bi reports no fault S7 == 0 && S8 == 1 && S5 == 1 Energy price is High, Batteries are not depleted, and the SSR Bi reports no fault Transition T1 T2 T3 T4 T5 T6 Main Mode Transitions T7 T01 T02 T03 Grid tie Mode Transitions Emergency Mode Transitions m00: m01: m02: m10: m11: m12: m13: m20: m30: T04 S8 == 0 || S5 == 0 Batteries are depleted or bidirectional converter is tripped T11 S8 == 0 && S4 == 1 T12 (S4 == 0 && S5 == 1) || (S2 == 0 && after (ts)) T13 T14 T15 T16 T17 S5 == 0 (S4 == 1 && S5 == 0) || S8 == 0 S5 == 1 S5 == 0 (S4 == 0 && S5 == 1) || (S9== 1 && after (ts)) Batteries are depleted and boost converter is available Boost converter is tripped and bidirectional converter is available, or DC bus voltage deviated severely Bidirectional converter is tripped Bidirectional converter is tripped and the boost converter is available, or batteries are depleted Bidirectional converter is reconnected Bidirectional converter is tripped Boost converter is tripped and bidirectional converter is available, or batteries over discharge Grid-tie saving sub-mode Batteries charging sub-mode Batteries discharging sub-mode Emergency sub-mode Contingency sub-mode Critical sub-mode Extreme sub-mode Utility mode Shut-down mode T1 M00 m02 T03 T02 T04 m01 T01 m00 T5 T2 m10 M20 T4 T7 m20 T3 T11 T16 T17 T15 m11 T6 m30 M30 T12 m12 T13 T14 M10 m13 Grid-tie/energy saving mode Islanding mode Utility mode Shut-down mode Main mode transitions Sub-mode transitions Fig. 2. Operational modes/ sub-modes of DC MG. B. Case Two This case shows the operation of the MGCC, during the gridtied/energy saving mode M00. Throughout section (1), a signal sent by the utility to the MGCC, informing it that the energy price is low (S7 = 1). The MGCC checks the last SOC of the battery system through the latest signal received from the CBi, confirming it is not full (S8 = 0). It also checks the last status of SSRBi assuring normal operation (S5 = 1), which leads the MGCC to trigger T02 to m01. Each converter LC maintains their ongoing tasks; the inverter maintains the DC bus voltage to 300 V, and CBo operates as MPPT. CBi initiates maximum charging (Iref = I1c), using current control as can be noticed in Fig. 4-a, where the bidirectional converter current becomes -5 A, i.e. charging with 5 A. Therefore, the inverter DC current increases to 5 A to maintain the dc bus voltage to 300 V as shown in Fig. 4(b). The AC current increases as well from the grid to the inverter, as shown in Fig. 4(c). For the interval of section (2), CBi signals that the battery system is fully charged (S8 = 1), then m00 is retained by the MGCC through T01, as shown in Fig 4(a), where the bidirectional converter output current drops to zero. The current of the inverter drops to zero as well. Therefore, the DC bus voltage is regulated as shown in Fig 4(b), since the DC MG can self-sustain its DC loads at that interval. During section (3), while m00 is still active, some solar power fluctuations start causing the boost converter output current to decrease gradually. The inverter DC/AC current increases in order to maintain the DC bus voltage, which can be noticed in Figs. 4(a), 4(b) and 4(c), respectively. During section (4), the irradiance goes back to its value of section (1), and the boost converter output current as well. This results in dropping the inverter DC/AC current to almost zero, maintaining the DC bus voltage, as shown in Figs. 4(a), 4(b) and 4(c), respectively. During the last segment, the utility signals high energy price to the MGCC (S7= 0). The MGCC confirms the availability of the bidirectional converter, by checking the last status report it got from the SSRBi. If there was no fault (S5 = 1), a transition T03 from m00 to m02 takes place. CBi switches to current control and starts discharging at the rated current (I1c) with 5 A. Therefore, the inverter DC current drops to -5 A to maintain the DC bus voltage to 300 V, i.e. sending 5 A to the grid, as shown in Figs. 4(a) and 4(b), respectively. VI. CONCLUSION Autonomous centralized communication based control for DC MG was developed in this paper. All possible scenarios, including severe conditions were conceived to better prepare the logic implemented within the MGCC using FSM. Reliable/ resilient operation while maintaining optimal performance, and maneuver intensive conditions was demonstrated in the results. The proposed control scheme has event driven controlled MGCC and state variables driven controlled LCs. Such hybrid combination, reduces the required communication bandwidth, compared to the regular control communication based techniques. FSM was used as a representation of the eventdriven MGCC, where the MG responds to an event by making a transition from one mode to another, and it can be in only one mode at a time and can transition to another mode based on some events. Moreover FSM was utilized due to the similarity of event driven control and discrete mathematical models, which increase the feasibility of implementing the logic within the MGCC. 5 VII. REFERENCES [1] [2] [3] [4] [5] [6] [7] [8] [9] [10] [11] [12] [13] [14] [15] Fig. 3. Case two, showing operation during islanding mode. . [16] [17] [18] G. Turner, J. Kelley, C. Storm, D. Wetz, W. Lee, “Design and Active Control of a Microgrid Testbed” IEEE Transactions on Smart Grid, VOL. 6, NO. 1, JANUARY 2015. F. Li et al., “Smart transmission grid: Vision and framework,” IEEE Trans. Smart Grid, vol. 1, no. 2, pp. 168–177, Sep. 2010. "Smart grid," in Energy.gov, 2011. [Online]. Available: http://energy.gov/oe/services/technology-development/smart-grid. M. Saleh, A. Althaibani, Y. Esa, Y. Mhandi, A. Mohamed, “Impact of Clustering Microgrids on their Stability and Resilience during Blackouts,” presented at Offenburg University of Applied Sciences. ICSGCE, Offenburg, Germany, 2015. M. Saleh, Y. Esa, A. Moahmed “Design and Implementation of CCNY DC Microgrid Testbed,” IAS, IEEE, October 2016, Portland, OR. H. Farhangi, “The path of the smart grid,” IEEE Power Energy Mag., vol. 8, no. 1, pp. 18–28, Jan./Feb. 2010. “Microgrid Activities." Microgrid Activities. Department of Energy, n.d. Web. 28 Mar. 2016. N. Hatziargyriou, Ed., Microgrids: Architectures and control. United States: John Wiley & Sons, pp.4-70, 2013. D. Chen and L. Xu, “Autonomous dc voltage control of a dc microgrid with multiple slack terminals,” IEEE Trans. Power Syst., Vol. 27, No. 4, pp. 1897-1905, Nov. 2012. R. S. Balog, “Autonomous local control in distributed DC power systems,” Ph.D. dissertation, Dept. Electr. Comput. Eng., Univ. Illinois Urbana-Champaign, Champaign, IL, USA, 2006. T. Dragicevic, X. Lu, J. C. Vasquez, and J. M. Guerrero, “DC microgrids—Part I: A review of control strategies and stabilization techniques,” IEEE Trans. Power Electron., vol. 31, no. 7, pp. 4876–4891, Jul. 2016. D. Salomonsson, L. Soder, and A. Sannino, “An adaptive control system for a DC microgrid for data centers,” IEEE Trans. Ind. Appl., vol. 44, no. 6, pp. 1910–1917, Nov./Dec. 2008. Z. Yixin, Z. Fang, X. Liansong, "Communication Platform for Energy Managem System in a Master-slave Control Structure Microgrid," IPEMC2012-ECCEAsia, China, 2012. W. Li, X. Mou, Y. Zhou, and C. Marnay, “On voltage standards for DC home microgrids energized by distributed sources,” in Proc. 7th Int. Power Electron. Motion Control Conf. (IPEMC), vol. 3. Harbin, China, Jun. 2012, pp. 2282–2286. Q. Shafiee, T. Dragicevic, J. C. Vasquez, and J. M. Guerrero, “Hierarchical control for multiple DC-microgrids clusters,” IEEE Trans. Energy Convers., vol. 29, no. 4, pp. 922–933, Dec. 2014. L. Che and M. Shahidehpour. Dc microgrids: Economic operation and enhancement of resilience by hierarchical control. IEEE Transactions Smart Grid, 5(5):2517–2526, 2014. IEEE Standard for Interconnecting Distributed Resources with Electric Power systems, IEEE Std 1547 – 2003, June 2003. J. Choi, H. Jeong, J. Choi, D. Won, S. Ahn and S.Moon, “Voltage control scheme with distributed generation and grid connected converter in a DC microgrid,” Energies, 2014, 7, 6477–6491. Mahmoud S. Saleh is a PhD student at the Smart Grid Laboratory, Department of Electrical Engineering, City collage of New York. He received his M.S. in Electrical Engineering from CCNY in 2013. His research interest includes microgrids design, control and optimization. Yusef Esa is a M.S. student at the Smart Grid Laboratory, City Collage of New York. He received the B.S. degree in electrical engineering from CCNY in 2015. His research interest includes microgrids design. Ahmed A. Mohamed (El-Tallawy) (GS’2009, M’2013) is an Assistant Professor of Electrical Engineering, City College of New York. He is the director of the Smart Grid Laboratory at CCNY. . His main research interests include microgrid design and control and electric vehicles. Fig. 4. Case one, showing operation during grid-tie mode. 6 View publication stats