Autonomous Active Power Control for Islanded AC Microgrids With Photovoltaic Generation and Energy...

882 IEEE TRANSACTIONS ON ENERGY CONVERSION, VOL. 29, NO. 4, DECEMBER 2014

Autonomous Active Power Control for IslandedAC Microgrids With Photovoltaic Generation

and Energy Storage SystemDan Wu, Fen Tang, Tomislav Dragicevic, Member, IEEE, Juan C. Vasquez, Member, IEEE,

and Josep M. Guerrero, Senior Member, IEEE

Abstract—In an islanded ac microgrid with distributed energystorage system (ESS), photovoltaic (PV) generation, and loads, acoordinated active power regulation is required to ensure efficientutilization of renewable energy, while keeping the ESS from over-charge and overdischarge conditions. In this study, an autonomousactive power control strategy is proposed for ac-islanded micro-grids in order to achieve power management in a decentralizedmanner. The proposed control algorithm is based on frequencybus-signaling of ESS and uses only local measurements for powerdistribution among microgrid elements. Moreover, this study alsopresents a hierarchical control structure for ac microgrids that isable to integrate the ESS, PV systems, and loads. Hereby, basicpower management function is realized locally in primary level,while strict frequency regulation can be achieved by using addi-tional secondary controller. Finally, real-time simulation resultsunder various state of charge (SoC) and irradiance conditions arepresented in order to prove the validity of the proposed approach.

Index Terms—Active power control, energy storage system, fre-quency bus-signaling, photovoltaic systems, secondary control.

I. INTRODUCTION

AMICROGRID can be considered as a local grid with mul-tiple distributed generators (DGs), energy storage system

(ESS), and loads, able to operate either in grid-connected orislanded mode, with possibility of seamless transitions betweenthem [1]. In islanded mode, the power exchange among DGs,ESS, and loads should be balanced inside the isolated microgridin order to keep the frequency stable.

Due to its installation cost decrease in recent years, photo-voltaic (PV) generation has emerged as one of the major DGsources and is widely used in microgrids [2]–[4]. However, theintermittent nature of power supplied from the PV generationmakes the ESS an indispensable component to keep the powerbalance between generation and consumption. In islanded mi-crogrid comprised of the ESS and PV generation, the ESS unitis usually operated as a grid forming unit that regulates ac bus,while the PV systems work as grid feeding units that inject

Manuscript received March 1, 2014; revised August 2, 2014; acceptedSeptember 4, 2014. Date of publication October 7, 2014; date of current versionNovember 20, 2014. Paper no. TEC-00144-2014.

D. Wu, T. Dragicevic, J. C. Vasquez, and J. M. Guerrero are with the De-partment of Energy Technology, Aalborg University, 9220 Aalborg, Denmark(e-mail: dwu@et.aau.dk; tdr@et.aau.dk; juq@et.aau.dk; joz@et.aau.dk).

F. Tang is with the National Active Distribution Network Technology Re-search Center, Beijing Jiaotong University, Beijing 100044, China (e-mail:ftang_nego@126.com).

Color versions of one or more of the figures in this paper are available onlineat http://ieeexplore.ieee.org.

Digital Object Identifier 10.1109/TEC.2014.2358612

all available power into the system [5]. In this sense, the ESSplays an important role for achieving the goal of power balanceand grid frequency support in a safe range of state of charge(SoC). However, this simple active power regulation strategywill lead SoC out of safe region if imbalance between con-sumption and generation lasts long enough. These situations arereferred to as overcharge and overdischarge conditions, and it iswell known that they may bring permanent damage to the ESS.On the other hand, strict active power regulation of the ESS tomaintain it within the safe SoC limits while ignoring the im-balance of power generation and consumption will deterioratethe frequency regulation function [6]. Hence, the coordinatedactive power control strategy should take into account status ofall microgrid elements such as the SoC of ESS, power availablefrom the PV systems, and demand of power consumption.

At the same time, such a coordinated power control strategyneeds to be designed with respect to specific microgrid configu-rations [7]. The most popular and well-known control paradigmin that sense is a centralized control structure [8]–[10]. The mul-tiple functions can be conveniently integrated into the microgridcentral controller which collects measurement signals from allelements and sends back commands after processing the datathrough its own algorithm. Due to limited amount of data that itcan process and inherent single point of failure, it is much moreappropriate to implement this kind of controller in concentratedsystems.

Decentralized control structure removes the central controllerand the operation of system is regulated by a number of localcontrollers which can optionally communicate with one an-other [11], [12]. The limitation of physical location can be over-come, but communication network may get complicated whenmicrogrid contains a lot of distributed units. Other availableparadigms include a hybrid control structure which groups themicrogrid elements into subdecentralized systems [13]. It com-bines the advantages of both central and decentralized controlstructures, but has complex element composition and the func-tions of each central and local controllers are not clearly de-fined. Moreover, the coordinated power control in all aforemen-tioned research studies still relies on external physical digitalcommunication links.

In order to enhance system reliability and expandability, andensure its robustness against the failure of external communi-cation system, autonomous power control strategy is needed.Droop control strategy, using frequency deviation of each unitto distribute active power, is widely accepted to fit into this

0885-8969 © 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.

WU et al.: AUTONOMOUS ACTIVE POWER CONTROL FOR ISLANDED AC MICROGRIDS WITH PHOTOVOLTAIC GENERATION AND ENERGY 883

requirement [14], [15]. However, the active power distributionis based on unified local control algorithm, which ignores theinherent power regulation difference between the ESS and therenewable energy sources. For example, the power generationof PV systems is usually based on current control mode (CCM)inner loop that follows maximum power point tracking (MPPT)algorithm, while the ESS is based on voltage control mode(VCM) inner loop aimed to regulate the bus voltage and fre-quency. Another promising approach is imposing higher orderfrequency waveforms on dc and ac power lines to pass coor-dination signals [16], [17]. The multiple resonant frequenciesof these signals should be properly designed to avoid overlap-ping, and the coordinated signals may introduce noises intothe distributed units. Bus-signaling method, by using differ-ent bus voltage/frequency thresholds to trigger mode-changingactions for the DGs and the ESS coordination is proposedin [18]–[20]. However, when the number of DGs increases, itbecomes difficult to determine the bus voltage/frequency thresh-olds. Recently, an interesting active power control law based onmodified droop method is presented in [21], which takes intoaccount both SoC, and available power in the ESS and renew-able energy sources, respectively. Nevertheless, the switchingactions of droop curves may trigger stability problems inducedby sudden bus frequency changes in microgrids.

In this study, a decentralized autonomous active power controlstrategy which is compatible with a hierarchical control schemeis proposed for islanded ac microgrids formed by the distributedESS, the PV systems, and loads. Based on the proposed bus-signaling method, the coordination performance of power regu-lation among microgrid elements is achieved in a decentralizedmanner. Moreover, an optional centralized secondary control isdefined for executing restoration of bus frequency when/if theislanded microgrid is required to reconnect with utility grid.The proposed autonomous active power control has several keyadvantages.

1) It keeps the SoC of ESS in a safe range, while efficientlyutilizing the renewable energy from the PV systems.

2) The active power control principle is simple and indepen-dent from inner control loop design, which can be easilyimplemented on top of conventional ESS and PV controlsystems without altering inner loop structure.

3) Coordinated active power distribution relies on continu-ous and smooth frequency bus-signaling to avoid suddenbus frequency changes. The proposed autonomous controlstrategy is an extension of our previous study described in[22], while a more complete description of autonomousactive power control in full range of SoC, and its imple-mentation taking into account the ESS and PV systemprime sources are addressed in this study.

II. SYSTEM DESCRIPTION

An exemplary multiple bus microgrid, shown in Fig. 1, con-sists of one ESS which acts as a grid forming unit, two PVsystems and distributed loads as grid following units. Batterybank system is used as the ESS which fixes bus frequency andcompensates power imbalance between power generation and

Fig. 1. Islanded ac microgrid configuration with ESS and PV systems.

Fig. 2. ESS system configuration with power stage and control algorithm.

consumption. Each PV system is used to supply renewable en-ergy to the microgrid.

A. ESS Grid Forming Unit

Fig. 2 shows the configuration of the ESS unit, which con-sists of a battery bank, bidirectional converter, and output filter.The modeling of the ESS with battery bank and associated SoCestimation algorithms are described in [23] in detail. Based onFig. 2, the grid forming unit control is implemented on grid-connected dc/ac inverter, which is formed by a typical doubleloop VCM control and a grid frequency control (GFC). Theinner loop VCM controller, using typical capacitor voltage con-troller (VC) and inductor current controller (CC), aims at fixingmicrogrid frequency and voltage. The frequency reference ofinner loop is given by the GFC based on ESS condition, whichis illustrated in Section III in detail. The frequency-signalingperformance is executed in the GFC by establishing a map-ping between ac bus frequency and estimated SoC of battery.In this study, SoC is estimated with a simple ampere countingalgorithm

SoC(t) = SoC0 +

t∫

0

ηbat ·ibat(t)Cbat

dt (1)

884 IEEE TRANSACTIONS ON ENERGY CONVERSION, VOL. 29, NO. 4, DECEMBER 2014

Fig. 3. PV system configuration with power stage and control algorithm.

where SoC0 is initial SoC, ηbat is charging/discharging ef-ficiency, ibat is battery current, and Cbat is battery capacity.However, it should be noted that more advanced SoC algorithmsmay be directly used instead [24].

In this study, only one ESS is used as a master unit. In caseswhen microgrid system needs to be upgraded to higher powerratings, multiple ESS units can be straightforward implementedby using ESS-coordinated control, which is described in detailin [25].

B. PV System Grid Following Unit

Fig. 3 shows the configuration of the PV systems, which con-sists of the PV array, dc/ac converter, output filter, and localcontrol system. For detailed model construction of the PV pan-els, one may refer to [26]. In the control part, the output powergenerated by the PV system is regulated by the output voltage ofPV panels VPV , which operates at the MPP when the ESS hascapability to compensate power imbalance. MPPT algorithm,shown in Fig. 3, utilizes perturb and observe method (P&O)[27], [28] to calculate the reference of VPV . Then, three phaseoutput currents are regulated by the typical CCM control struc-ture including dc VC VCR and inner loop CC CCR which isimplemented in d-q reference frame. An active power controller(APC) which is given in detail in next section, is implemented torealize power curtailment when the PV system is required to de-crease power generation and operate at OFF-MPP status. Whenexecuting power curtailment, the trigger signal EN_MPPT inFig. 3 from the APC block is set to zero to disable the MPPT.Thereby, the MPPT and power curtailment of the PV systemsare never enabled at the same time.

III. AUTONOMOUS ACTIVE POWER CONTROL

The autonomous active power control and coordinated perfor-mance among microgrid components is achieved by frequencybus-signaling method throughout the microgrid system. Thismeans that microgrid bus frequency is regulated by ESS masterunit which uses frequency to reflect its SoC condition, while thePV systems and distributed loads adjust power generation andconsumption based on measured frequency.

Fig. 4. GFC control algorithm of ESS.

A. ESS Master Control With GFC

According to different SoC scenarios, the ESS bus-signalingcontrol can be classified into high SoC control and low SoCcontrol. The high SoC control targets the coordination betweenthe ESS and PV system when microgrid system continuouslygenerates excess of renewable energy which leads to high SoClevel. When SoC becomes critically high and goes over the up-per threshold SoCu , the ESS boosts output frequency to informthe PV systems that they need to start decreasing power gener-ation. In the high SoC range (SoCu , 100%), shown as the GFCblock in Fig. 4, the ESS regulates frequency with slope of m1for bus-signaling. Similarly, when SoC is below low-thresholdSoCd , the ESS is approaching overdischarge situation. In thisrange (SoC0 , SoCd ), the ESS controls bus frequency to con-stantly decrease from nominal value with slope m2 to inducethe loads shedding procedure. When the SoC is in the saferange (SoCd , SoCu ), bus frequency is kept at nominal value f ∗.The determination of SoC thresholds SoCu and SoCd takes intoaccount the following issues.

1) The higher the SoCu selected, the more efficiently can therenewable energy from PV system be used.

2) SoC has an estimation error and should give a margin ofovercharge scenario (5% is considered in this study).

3) Based on [29], SoCd should be determined higher thanthe SoC value linked to end-voltage to protect the battery.

Besides, the thresholds SoCu and SoCd are determined basedon specific applications of battery systems, and a more detaileddescription of determining up and low SoC thresholds can befound in [30]. This bus-signaling control from ESS master unitcan be expressed as

⎧⎪⎪⎪⎪⎨⎪⎪⎪⎪⎩

f = f ∗ + m1 · (SoC − SoCu ), SoCu < SoC < 100%

f = f ∗, SoCd ≤ SoC ≤ SoCu

f = f ∗ − m2 · (SoCd − SoC), SoC0 < SoC < SoCd

(2a)where the slopes m1 and m2 are determined as

m1 =fmax − f ∗

100% − SoCu(2b)

m2 =f ∗ − fmin

SoCd − SoC0. (2c)

Finally, fmax and fmin represent the maximum and min-imum bus frequencies given by specific requirement of grid

WU et al.: AUTONOMOUS ACTIVE POWER CONTROL FOR ISLANDED AC MICROGRIDS WITH PHOTOVOLTAIC GENERATION AND ENERGY 885

Fig. 5. Coordinated operation of microgrid based on master-slave control.

standard. When bus frequency reaches fmin , it indicates thatthe SoC is below the minimum threshold SoC0 and there is notenough power stored to ensure normal system operation. At thistime, the overall system should be shut down. Fig. 5 shows thisbus-signaling performance by the ESS. In addition, when busfrequency decays sharply due to failure scenarios of ESS mas-ter unit, microgrid can take actions: 1) enabling VCM controlmode in CCM units to perform grid support [20]; 2) connectingback-up equipment like diesel generators to support local loads.Multiple ESS can also be implemented in a coordinated way toimprove system redundancy and avoid the single-point-failureof ESS master unit [25].

B. PV System Slave Control With APC

Corresponding to different frequency ranges determined bythe ESS, the PV systems adjust the output power dependingon measured bus frequency fmeas . When fmeas is not abovenominal frequency, PV systems are working at the MPP to makefull use of renewable energy. When fmeas is above f ∗, the PVsystems start to decrease generated power to limit the SoC ofthe ESS reflected by frequency. In this way, the higher the busfrequency is, the lower power PV systems generate. This powerregulation strategy can be expressed as

⎧⎨⎩

PPV = PMPP , fmeas ≤ f ∗

PPV = PMPP − n · (fmeas − f ∗), fmeas > f ∗(3a)

n =PMPP

fmax − f ∗ (3b)

where PPV is the output power of PV system, PMPP is the max-imum power extracted from MPPT algorithm, and n is the slopeof PV generation decrease with respect to frequency changes.Since MPPT algorithm is not enabled when SoC > SoCu , thevariable PMPP in (3) in high SoC case refers to the last powerstatus at MPP when SoC reached SoCu , and is used as the initialpoint of active power control to decrease power generation inorder to have a smooth transition between two conditions.

Meanwhile, microgrid bus frequency fmeas is measured bythe phase lock loop (PLL) block in PV systems. It can be ex-pressed as

fmeas =1

σs + 1f (3c)

where σ is time constant of bus frequency measurement for theAPC control. When APC is enabled, taking (3c) into (3a), theoutput power of PV system can be deduced as

PPV = PMPP − n

(1

σs + 1f − f ∗

). (4)

By replacing f ∗ in (2a) when SoC > SoCu into (4), PPV canbe expressed as

PPV = PMPP +nσs

σs + 1f − nm1(SoC − SoCu ). (5)

Based on (5), the partial derivative can be calculated as

∂PPV

∂f=

ΔPPV

Δf=

nσs

σs + 1. (6)

Then, output power inertia of the PV systems can be deducedas [31]

Gi(s) =ΔPPV

sΔf=

nσ

σs + 1. (7)

Since n is fixed by (3b), the inertia response of the PV systemscan be tuned independently of the time constant σ. Usually thetime constant of PLL used in the CCM inner loop control iswithin 1 s to ensure fast dynamic of current tracking. Comparedto the time constant of inner loop PLL, the boost of bus fre-quency determined by the SoC, as presented in (2), is changingmuch slower (σ � 1s). Therefore, when designing the APCalgorithm, the dynamics of inner loop control and active powerregulation can be considered decoupled. The rate of frequencyis much slower than inner power control loops. Thus, the sensi-tivity of power regulation against frequency change is relativelylow. The slave control of the PV system based on APC is pre-sented in Fig. 5. This frequency signaling principle for powerregulation of the PV systems can be found in [32] with its ap-plication in microgrid described in [33].

886 IEEE TRANSACTIONS ON ENERGY CONVERSION, VOL. 29, NO. 4, DECEMBER 2014

Fig. 6. PV system control with (a) V–P characteristic, (b) APC control, and(c) inertia power control.

Fig. 7. APC control algorithm in the PV systems.

The aforementioned description of the APC gives the fi-nal power regulation of the PV system using frequency bus-signaling. While implementing the APC on the PV controller inspecific, it is achieved by controlling the PV panel output volt-age VPV , with the full structure being shown in Fig. 6. Basedon V–P characteristic [see Fig. 6(a)], the PV system can operateeither at MPP or OFF-MPP by controlling VPV in the range(VMPP , VOC ) [see Fig. 6(b)], where VMPP is the PV panel volt-age corresponding to PMPP which is calculated from the MPPTblock, VOC is open-circuit voltage of PV panels. Then, insteadof using (3) to control output power, the APC control establishesrelationship between bus frequency f and VPV to indirectly con-trol PPV [see Fig. 6(c)]. The implementation of the APC blockin the PV systems is shown in Fig. 7 and is expressed as⎧⎨

⎩VPV = VMPP , fmeas ≤ f ∗

VPV = VMPP + n′ · (fmeas − f ∗), fmeas > f ∗.(8a)

Similar to (3b), the voltage boosting coefficient n’ in (8a) isderived as

n′ =VOC − VMPP

fmax − f ∗ . (8b)

In Fig. 7, the signal EN_MPPT setting to 1 and 0 represent thePV systems operate under the MPP and power curtailment con-ditions, respectively. The VMPP is therefore only being tracked

Fig. 8. Load slave control based on (a) frequency signaling of ESS and(b) relay control of loads.

when SoC ≤ SoCu . When SoC is in high scenario, the valueVMPP is selected as an initial PV voltage status to execute APCcontrol in OFF-MPP condition.

C. Load Slave Control

Besides power generation, demand side management strat-egy in microgrid can also act as an effective energy reserve tomeet supply-demand balance. According to different roles per-formed by microgrid loads, priorities are assigned to them. Forexample, critical devices are given as high priority, while non-critical loads that can be cut off when energy storage is low areassigned as low priority. Previous research are proposed usingfrequency thresholds to perform loads shedding actions basedon traditional droop control [34], [35]. These frequency thresh-olds are set corresponding to the load priorities. The loads withlower priorities are first cut off when frequency drops to presetlevels. In this way, the lower the bus frequency is, the loads withhigher priority are shed one by one. The loads recovery processadopts similar manner, where loads with higher priorities arefirst recovered when bus frequency gets restored from lower tohigher levels. This principle of ac bus frequency signaling toperform loads shedding and recovery procedure is employed inthis study. However, the main difference is that the frequencysignaling here is determined as a function of the SoC in (2a) butnot instantaneous power, as presented in load slave control ofFig. 5. In this way, bus frequency triggers the on/off actions ofload contactors depending on the SoC value of the ESS. Fig. 8(a)shows a two-step loads shedding procedure with bus-signaling,where Load2 holds higher priority than Load1 . SoCr1 , SoCr2 ,SoCs1 , and SoCs2 represent SoC thresholds for connecting andtripping Load1 and Load2 . And fL1 ON , fL2 ON , fL1 OFF , and

WU et al.: AUTONOMOUS ACTIVE POWER CONTROL FOR ISLANDED AC MICROGRIDS WITH PHOTOVOLTAIC GENERATION AND ENERGY 887

TABLE IBUS-SIGNALING OF LOAD SLAVE CONTROL

Estimated SoC Frequency thresholds Load1 status Load2 status

SoCr1 fL 1 O N ON NCSoCr2 fL 2 O N NC ONSoCs1 fL 1 O F F OFF NCSoCs2 fL 2 O F F NC OFF

Note: NC indicates not changing action.

Fig. 9. Central secondary control configuration.

fL2 OFF represent frequency thresholds to close and open loadscontactors. In order to avoid chattering phenomenon (repetitiveON–OFF actions of load contactors), a relay control is utilized,as shown in Fig. 8(b). The description of Load1 and Load2 statusaccording to bus-signaling effect is summarized in Table I.

IV. CENTRALIZED SECONDARY CONTROL

With only local autonomous active power control, steady statebus frequency deviation Δf will be produced as PPV decreasesto steady state point Pe , as shown in Fig. 5. Although the max-imum frequency deviation of Δf can be designed to stay withinthe preselected allowable range according to (1), it can also becompletely eliminated by additional centralized secondary con-trol when tight bus frequency regulation is required, as shown inFig. 9. Notice that the communication link between the centraland local level is of low bandwidth, and operates as an op-tional component since the coordination performance is alreadyachieved by local power controller. This tradeoff between theinvestment in the communication link and high quality of powersupply can be decided by customer with respect to requirementsof different applications.

Fig. 10 shows the centralized secondary control performancefor bus frequency restoration. When SoC > SoCu , both re-sponse of the ESS master control and the PV system slavecontrol shift downward by the restoration term δf sent from cen-tral controller. The central secondary control monitors the acbus frequency and utilizes PI controller for the bus frequencyrestoration that can be expressed as

δf = Gs(s) · (f ∗ − fsec) =(

kpsec +kisec

s

)· (f ∗ − fsec)

(9)

Fig. 10. Central secondary control performance. (a) ESS master control and(b) PV system slave control.

Fig. 11. Secondary control algorithm of (a) GFC block and (b) APC block.

where Gs(s) is the central secondary controller with kpsec andkisec as proportional and integral terms, and fsec is measuredbus frequency by the central controller.

With restoration term δf, the ESS and the PV system coordi-nation response based on (2) and (3) can be rewritten as

f = f ∗ + δf + m1 · (SoC − SoCu ), SoCu < SoC < 100%

(10)

PPV = PMPP − n · (fmeas − f ∗ − δf), fmeas > f ∗ + δf.

(11)

Corresponding to (11), the APC control strategy of the PVsystem is revised as

VPV = VMPP + n′ · (fmeas − f ∗ − δf), fmeas > f ∗ + δf .(12)

Based on (10) and (12), the control blocks of the GFC andAPC with secondary control are shown in Fig. 11. By combining(9)–(11), it can be deduced that the inertia response of the PVsystem is consistent with (7), which indicates that the effect

888 IEEE TRANSACTIONS ON ENERGY CONVERSION, VOL. 29, NO. 4, DECEMBER 2014

TABLE IIPOWER STAGE AND CONTROL PARAMETERS

Parameter Symbol Value Unit

Power StageNominal Bus Voltage V∗ 230 VNominal Bus Frequency f∗ 50 HzInverter Inductor of ESS L i n 1.8 mHOutput Inductor of ESS Lo 1.5 mHFilter Capacitor C 27 μFOutput Inductor of PV Lf 3.6 mHActive Power Loads R1 /R2 /R3 300 Ω ,Reactive Power Loads L1 /L2 /L3 0.4/0.4/0.5 H

Inner loop ControlESS VC kp V , ki V 0.1, 200 -, s−1

ESS Current Controller kp I , ki I 15, 50 -, s−1

PV Current Controller kp P V , ki P V 20, 50 -, s−1

Active Power ControlMaximum Bus Frequency fm a x 50.5 HzSoC Upper-threshold SoC u 95 %SoC Low-threshold SoC d 40 %Boosting Slope of Frequency m 1 0.1 Hz/%Dropping Slope of Frequency m 2 0.4 Hz/%Tripping Thresholds of Loads fL 1−O F F , fL 2−O F F , 49.5, 49.4 HzRecovery Thresholds of Loads fL 1−O N , fL 2−O N 49.8, 49.7 Hz

of secondary frequency restoration is decoupled from the localactive power regulation.

V. HARDWARE-IN-THE-LOOP RESULTS

The proposed autonomous active power control strategy isimplemented on an islanded ac microgrid consists of one ESS,two PV systems, and three distributed load, as shown in Fig. 1.The power stage and control parameters are presented in Ta-ble II. The overall microgrid model is established with MAT-LAB/Simulink toolbox and then downloaded into dSPACE 1006platform based on real-time simulation, in which the time spanof simulation equals to that in the real system. Fig. 12 showssimulation results with local active power control between theESS and PV system which is summarized as follows.

1) Scenario S1 : The SoC of the ESS [see Fig. 12(a)] is lowerthan the 95%, so that two PV systems are operating at 2and 1.3 kW, respectively. And ESS is charging power of1.72 kW [see Fig. 12(c)]. Bus frequency is kept at nominalvalue 50 Hz [see Fig. 12(b)].

2) Scenario S2 : The SoC of ESS is above 95%, and busfrequency is increasing to inform power decrease fromPV systems. Both PV systems receive that frequency andgradually decrease their generation to 0.95 and 0.62 kW,respectively. In this way, the charging power gets limitedto zero with time. In steady state, the bus frequency isstable at 50.3 Hz.

Fig. 13 shows the simulation results with both local and cen-tral control for active power regulation. Compared with Fig. 12,it can be seen that coordination performance of active powerregulation [see Fig. 13(c)] remains the same with only localcontrol shown in Fig. 12. However, with central control, the busfrequency deviation can be effectively eliminated in steady state[see Fig. 13(b)].

Fig. 12. Simulation results in high SoC scenario of ESS with only local activepower control.

Fig. 13. Simulation results in high SoC scenario of ESS with both local andcentral control.

Fig. 14 shows simulation results of coordinated performanceof the ESS, the PV systems, and distributed loads in full rangeof SoC, which is summarized as follows.

1) Scenario S1: The SoC is in low range that there is co-ordination with ESS and distributed loads. When SoCdrops to 20% and 15%, two parts of noncritical load are

WU et al.: AUTONOMOUS ACTIVE POWER CONTROL FOR ISLANDED AC MICROGRIDS WITH PHOTOVOLTAIC GENERATION AND ENERGY 889

Fig. 14. Simulation results in full SoC scenario of coordination with ESS, PVsystems, and distributed loads.

successively tripped by measuring the bus frequency dropsto 49.5 and 49.4 Hz. When the SoC rises up to 30% and35% at 49.7 and 49.8 Hz, respectively, the two parts ofloads are recovered successively.

2) Scenario S2: The SoC is in moderate range, then thePV generation are 2 and 1.3 kW. The bus frequency iscontrolled at 50 Hz in this range. In this scenario, reactivepower changes from 1.2 to 2.4 kVAr with power factorchanging from 0.8 to 0.5, and the overall microgrid systemmaintains good dynamic performance.

3) Scenario S3: The microgrid operation is based on coordi-nation between the ESS and PV systems to limit the ESScharging power by decreasing power generation in highSoC case, similarly shown in Fig. 13.

4) Scenario S4: Total active power load increases 3.2 kW,while reactive power load increases to 3.4 kVAr. It can beseen that the instantaneous increased active and reactive

power is generated by the ESS. Afterward, the PV sys-tems gradually restore generation again in an autonomousway. It can be seen after a short dynamic process that thebus frequency is regulated at 50 Hz in steady state. Sameas shown in S2 , the autonomous power regulation per-formance can be maintained and sudden reactive powerchanges have little effect.

Fig. 15 shows simulation results of microgrid system per-formance when sudden solar irradiation increases from 500 to1000 W/m2.

1) Scenario S1: The ESS is in the moderate SoC con-dition [see Fig. 15(d)], while irradiation of Fig. 15(a)is 500 W/m2. In this range, the PV system oper-ates at MPP with 2075 W [see Fig. 15(b)] by usingMPPT algorithm [enable MPPT signal is set at unityin Fig. 15(c)]. The dynamic of MPP tracking is shownin Fig. 15(b).

2) Scenario S2: The ESS is approaching to be fully charged[SoC is above 95% in Fig. 15(d)], and the APC controldominates the PV system to gradually decrease powergeneration [see Fig. 15(b)]. In this range, the MPPT algo-rithm is stalled, as Fig. 15(c) shows. In order to show fre-quency behavior followed by coordination between ESSand PV system, secondary control is not presented here[see Fig. 15(e)]. Also, it can be noticed that when the APCcontrol is enabled, a dead band is set at 0.05 Hz around thenominal bus frequency 50 Hz to decrease sensitiveness ofinstantaneous frequency ripple.

3) Scenario S3: Solar irradiation changes from 500 to1000 W/m2. The PV system tends to generate higherpower of 2375 W due to increased solar irradiation. Thisinstantaneous generated power from the PV system canbe absorbed by ESS at 1250 W in Fig. 15(f). Then result-ing from the APC control, the PV generation is able todecrease power again to limit charging power of the ESS[see Fig. 15(b) and (f)].

Fig. 16 shows simulation results of overall microgrid systemperformance when sudden solar irradiation decreases from 500to 150 W/m2 to simulate partial shading situation.

1) Scenarios S1 and S2: Solar irradiation is 500 W/m2, andmicrogrid system performance can be referred to scenar-ios S1 and S2 of Fig. 15.

2) Scenarios S3: 70% solar irradiation changes from 500to 150 W/m2, in case of partial shading. The instanta-neous power generation decrease is supported by the ESS[see Fig. 16(f)]. Then SoC of ESS starts to decrease asESS discharges power to supply power together with PVsystem [see Fig. 16(d)]. Corresponding to the SoC de-crease, the microgrid bus frequency also decreases by bussignaling effect [see Fig. 16(e)]. When the SoC drops tothreshold SoCu = 95%, MPPT algorithm is enabled again[see Fig. 16(c)] and the PV system operates at MPP of576 W [see Fig. 16(b)]. The dynamics of MPP tracking ofthe PV generation in S1 and S3 are shown in Fig. 16(b).

Fig. 17 shows the dynamics of the PV and ESS systems whensolar irradiation increases, as presented in Fig. 15. Although thePV system generates higher power at 2475 W instantaneously,

890 IEEE TRANSACTIONS ON ENERGY CONVERSION, VOL. 29, NO. 4, DECEMBER 2014

Fig. 15. Simulation results of system performance when 50% solar irradiationincreases from 500 to 1000 W/m2.

Fig. 16. Simulation results of system performance when 70% solar irradiationdecreases from 500 to 150 W/m2.

WU et al.: AUTONOMOUS ACTIVE POWER CONTROL FOR ISLANDED AC MICROGRIDS WITH PHOTOVOLTAIC GENERATION AND ENERGY 891

Fig. 17. Simulation results of system dynamic performances when solar irra-diation suddenly increases.

Fig. 18. Simulation results of system dynamic performances when solar irra-diation suddenly drops.

the power generation can be suppressed in short time due toAPC control to limit ESS charging power.

Fig. 18 shows dynamics of the PV and ESS systems whensolar irradiation decreases suddenly as presented in Fig. 16.In the dynamic process, the sudden power generation drop issupported by ESS system, while PV system gradually restorespower generation to supply loads together with the ESS.

VI. CONCLUSION

This study proposed an autonomous active power control tocoordinate distributed components of microgrid consisting ofthe ESS, the PV systems, and loads. Additionally, a centralizedsecondary control was applied to effectively eliminate steady

state deviation of the bus frequency. By the proposed activepower control, SoC of the ESS can be kept within the safe limitsby automatically adjusting the power generation from the PVsystems and load consumption. This coordination performancewas obtained by using only local controllers and does not relyon external communication links. Therefore, the risk inducedby the failure of the communication links can be avoided andthereby the reliability of the system is enhanced. Finally, theproposed control strategy is verified by the hardware-in-the-loop simulation results.

REFERENCES

[1] J. M. Guerrero, M. Chandorkar, T. Lee, and P. C. Loh, “Advanced controlarchitectures for intelligent microgrids—Part I: Decentralized and hierar-chical control,” IEEE Trans. Ind. Electron., vol. 60, no. 4, pp. 1254–1262,Apr. 2013.

[2] M. Datta, T. Senjyu, A. Yona, T. Funabashi, and K. Chul-Hwan, “Afrequency-control approach by photovoltaic generator in a PV–dieselhybrid power system,” IEEE Trans. Energy Convers., vol. 26, no. 2,pp. 559–571, Jun. 2011.

[3] S. B. Kjaer, J. K. Pedersen, and F. Blaabjerg, “A review of single-phase grid-connected inverters for photovoltaic modules,” IEEE Trans.Ind. Appl., vol. 41, no. 5, pp. 1292–1306, Sep./Oct. 2005.

[4] Q. Li and P. Wolfs, “A review of the single phase photovoltaic modulein-tegrated converter topologies with three different DC link configurations,”IEEE Trans. Power Electron., vol. 23, no. 3, pp. 1320–1333, May 2008.

[5] L. J. Ricalde, E. Ordonez, M. Gamez, and E. N. Sanchez, “Design ofa smart grid management system with renewable energy generation,” inProc. IEEE Symp. Comput. Intell. Appl. Smart Grid, 2011, pp. 1–4.

[6] F. Giraud and Z. M. Salameh, “Steady-state performance of a grid-connected rooftop hybrid wind-photovoltaic power system with batterystorage,” IEEE Trans. Energy Convers., vol. 16, no. 1, pp. 1–7, 2001.

[7] M. H. Nehrir, C. Wang, K. Strunz, H. Aki, R. Ramakumar, J. Bing, Z.Miao, and Z. Salameh, “A review of hybrid renewable/alternative energysystems for electric power generation: Configurations, control and appli-cations,” IEEE Trans. Sustain. Engergy, vol. 2, no. 4, pp. 392–403, Oct.2011.

[8] K. Jong-Yul, J. Jin-Hong, K. Seul-Ki, C. Changhee, P. June-Ho, K. Hak-Man, and N. Kee-Young, “Cooperative control strategy of energy storagesystem and microsources for stabilizing the microgrid during islandedoperation,” IEEE Trans. Power Electron., vol. 25, no. 12, pp. 3037–3048,Dec. 2010.

[9] H. Kanchev, L. Di, F. Colas, V. Lazarov, and B. Francois, “Energy man-agement and operational planning of a microgrid with a PV-based activegenerator for smart grid applications,” IEEE Trans. Ind. Electron., vol. 58,no. 10, pp. 4583–4592, Oct. 2011.

[10] F. Valenciaga and P. F. Puleston, “Supervisor control for a stand-alonehybrid generation system using wind and photovoltaic energy,” IEEETrans. Energy Convers., vol. 20, no. 2, pp. 398–405, Jun. 2005.

[11] A. L. Dimeas and N. D. Hatziargyriou, “Operation of a multiagent systemfor microgrid control,” IEEE Trans. Power Syst., vol. 20, no. 3, pp. 1447–1455, Aug. 2005.

[12] J. Lagorse, M. G. Simoes, and A. Miraoui, “A multiagent fuzzy-logic-based energy management of hybrid systems,” IEEE Trans. Ind. Appl.,vol. 45, no. 6, pp. 2123–2129, Nov./Dec. 2009.

[13] M. S. Khan and M. R. Iravani, “Supervisory hybrid control of a micro gridsystem,” in Proc. IEEE Canada Elect. Power Conf., 2007, pp. 20–24.

[14] D. Wu, F. Tang, G. J. M. Vasquez, J. C. G. Chen, and L. Sun, “Autonomousactive and reactive power distribution strategy in islanded microgrids,” inProc. Appl. Power Electron. Conf. Exp., 2014, pp. 2126–2131.

[15] E. A. A. Coelho, P. C. Cortizo, and P. F. D. Garcia, “Small-signal stabilityfor parallel-connected inverters in stand-alone AC supply systems,” IEEETrans. Ind. Appl., vol. 38, no. 2, pp. 533–542, Mar./Apr. 2002.

[16] T. Dragicevic, J. M. Guerrero, and J. C. Vasquez, “A distributed con-trol strategy for coordination of an autonomous LVDC microgrid basedon power-line signaling,” IEEE Trans. Ind. Electron., vol. 61, no. 7,pp. 3313–3326, Jul. 2014.

[17] D. J. Perreault, R. L. Selders, and J. G. Kassakian, “Frequency-basedcurrent-sharing techniques for paralleled power converters,” IEEE Trans.Power Electron., vol. 13, no. 4, pp. 626–634, Jul. 1998.

892 IEEE TRANSACTIONS ON ENERGY CONVERSION, VOL. 29, NO. 4, DECEMBER 2014

[18] D. Boroyevich, I. Cvetkovic, D. Dong, R. Burgos, F. Wang, and F. Lee,“Future electronic power distribution systems a contemplative view,” inProc. Int. Opt. Elect. Electron. Equipment Conf., 2010, pp. 1369–1380.

[19] J. Schonberger, R. Duke, and S. D. Round, “DC-bus signaling: A dis-tributed control strategy for a hybrid renewable nanogrid,” IEEE Trans.Ind. Electron., vol. 53, no. 5, pp. 1453–1460, Oct. 2006.

[20] D. Wu, J. M. Guerrero, J. C. Vasquez, T. Dragicevic, and F. Tang, “Coor-dinated power control strategy based on primary-frequency-signaling forislanded microgrids,” in Proc. IEEE Energy Convers. Congr. Expo., 2013,pp. 1033–1038.

[21] J. G. de Matos, L. A. de S. Ribeiro, and E. de Carvalho Gomes, “Powercontrol in AC autonomous and isolated microgrids with renewable energysources and energy storage systems,” in Proc. 39th Annu. Conf. IEEE Ind.Electron. Soc., 2013, pp. 1827–1832.

[22] D. Wu, F. Tang, T. Dragicevic, J. C. Vasquez, and J. M. Guerrero, “Coor-dinated primary and secondary control with frequency-bus-signaling fordistributed generation and storage in islanded microgrids,” in Proc. 39thAnnu. Conf. IEEE Ind. Electron. Soc., 2013, pp. 7140–7145.

[23] T. Dragicevic, J. M. Guerrero, J. C. Vasquez, and D. Skrlec, “Supervi-sory control of an adaptive-droop regulated DC microgrid with batterymanagement capability,” IEEE Trans. Power Electron., vol. 29, no. 2,pp. 695–706, Feb. 2014.

[24] T. Dragicevic, S. Sucic, and J. M. Guerrero, “Battery state-of-charge andparameter estimation algorithm based on Kalman filter,” in Proc. Eurocon,2013, pp. 1519–1525.

[25] X. Lu, K. Sun, J. M. Guerrero, J. C. Vasquez, and L. Huang, “State-of-charge balance using adaptive droop control for distributed energystorage systems in DC microgrid applications,” IEEE Trans. Ind. Electron.,vol. 61, no. 6, pp. 2804–2815, Jun. 2014.

[26] M. G. Villalva, J. R. Gazoli, and E. R. Filho, “Comprehensive approachto modeling and simulation of photovoltaic arrays,” IEEE Trans. PowerElectron., vol. 24, no. 5, pp. 1198–1208, May 2009.

[27] R. A. Mastromauro, M. Liserre, and A. Dell’Aquila, “Control issues insingle-stage photovoltaic systems: MPPT, current and voltage control,”IEEE Trans. Ind. Inform., vol. 8, no. 2, pp. 241–254, May 2012.

[28] T. Esram and P. L. Chapman, “Comparison of photovoltaic array maximumpower point tracking techniques,” IEEE Trans. Energy Convers., vol. 22,no. 2, pp. 439–449, Jun. 2007.

[29] D. Linden and T. B. Reddy, Handbook of Batteries. New York, NY, USA:McGraw-Hill, 2002.

[30] B. Xiao, Y. Shi, and L. He, “A universal state-of-charge algorithm forbatteries,” in Proc. 47th ACM/IEEE Design Autom. Conf., Jun. 2010,pp. 687–692.

[31] M. Datta, H. Ishikawa, H. Naitoh, and T. Senjyu, “LFC by coordinatedvirtual inertia mimicking and PEVs in power utility with MW-class dis-tributed PV generation,” in Proc. IEEE 13th Workshop Control ModelingPower Electron, 2012, pp. 1–8.

[32] Sunny Island 5048 Installation and Instruction Manual, SMA Technolo-gies AG, 2006.

[33] Y. Zhangang, C. Yanbo, and W. Chengshan, “Construction, operation andcontrol of a laboratory-scale microgrid,” in Proc. Int. Conf. Sustain. PowerGener. Supply, 2009, pp. 1–5.

[34] J. A. Pec, as Lopes, C. L. Moreira, and A. G. Madureira, “Defining controlstrategies for microgrids in islanded operation,” IEEE Trans. Power Syst.,vol. 21, no. 2, pp. 916–924, May 2006.

[35] I. J. Balaguer, Q. Lei, S. Yang, U. Supatti, and Z. Peng, “Control forgrid connected and intentional islanding operations of distributed powergeneration,” IEEE Trans. Ind. Electron., vol. 58, no. 1, pp. 147–157,Jan. 2011.

Dan Wu received the B.S. and M.S. degrees in electri-cal engineering from the Beijing Institute of Technol-ogy, Beijing, China, in 2009 and 2012, respectively.She is currently working toward the Ph.D. degree inthe Department of Energy Technology, Aalborg Uni-versity, Aalborg, Denmark.

She is a Member of the Microgrid Research Group,Aalborg University. Her research interests includemicrogrids and distributed generation systems.

Fen Tang received the B.S. degree in electrical en-gineering, and the Ph.D. degree in power electronicsand electric drives from Beijing Jiaotong University,Beijing, China, in 2006 and 2013, respectively.

She is currently a Post-Doctoral Researcher in Bei-jing Jiaotong University. From January 2013 to Jan-uary 2014, she was a Guest Post-Doctoral Researcherwith the Department of Energy Technology, AalborgUniversity, Aalborg, Denmark. Her research interestsinclude microgrid, wind power generation system,power converter for renewable generation systems,

power quality, and motor control.

Tomislav Dragicevic (S’09–M’13) received theM.E.E. degree in power systems and the Ph.D. de-gree in power electronics from the Faculty of Electri-cal Engineering, Zagreb, Croatia, in 2009 and 2013,respectively.

Since 2010, he has been actively cooperating inan industrial project related to the design of electricalpower supply for remote telecommunication stations.Since 2013, he has been a Full-time Post-DoctoralResearcher with Aalborg University, Aalborg, Den-mark. His research interests include modeling, con-

trol and energy management of intelligent electric vehicle charging stations andother types of microgrids based on renewable energy sources, and energy stor-age technologies.

Juan C. Vasquez (M’12) received the B.S. degree inelectronics engineering from the Autonomous Uni-versity of Manizales, Manizales, Colombia, in 2004.He received the Ph.D. degree in automatic control,robotics, and computer vision from the TechnicalUniversity of Catalonia, Barcelona, Spain, in 2009.

Since 2011, he has been an Assistant Professorin Microgrids with the Institute of Energy Technol-ogy, Aalborg University, Aalborg, Denmark, wherehe is co-responsible for the Microgrids Research Pro-gram. His current research interests include opera-

tion, power management, and hierarchical control and optimization applied todistributed generation in ac/dc microgrids. He is currently a Member of theTechnical Committee on Renewable Energy Systems, Aalborg Univesrity.

Josep M. Guerrero (S’01–M’04–SM’08) receivedthe B.S. degree in telecommunications engineering,the M.S. degree in electronics engineering, and thePh.D. degree in power electronics from the TechnicalUniversity of Catalonia, Barcelona, Spain, in 1997,2000, and 2003, respectively.

Since 2011, he has been a Full Professor with theDepartment of Energy Technology, Aalborg Univer-sity, Aalborg, Denmark, where he is responsible forthe Microgrid Research Program. Since 2014, he hasbeen the Chair Professor at Shandong University, Ji-

nan, China. His research interests include different microgrid aspects, includingpower electronics, distributed energy-storage systems, hierarchical and cooper-ative control, energy management systems, and optimization of microgrids andislanded minigrids.

Dr. Guerrero was the Chair of the Renewable Energy Systems TechnicalCommittee of the IEEE Industrial Electronics Society. In 2014, he was awardedby Thomson Reuters as ISI Highly Cited Researcher. He is an Associate Editorof the IEEE TRANSACTIONS ON POWER ELECTRONICS, the IEEE TRANSACTIONS

ON INDUSTRIAL ELECTRONICS, and the IEEE INDUSTRIAL ELECTRONICS MAG-AZINE, and an Editor of the IEEE TRANSACTIONS ON SMART GRID. He has beenGuest Editor of the IEEE TRANSACTIONS ON POWER ELECTRONICS, the IEEETRANSACTIONS ON INDUSTRIAL ELECTRONICS, and the IEEE TRANSACTIONS

ON SMART GRID.