06601722

230 IEEE TRANSACTIONS ON POWER SYSTEMS, VOL. 29, NO. 1, JANUARY 2014

Optimal Decentralized Voltage Control forDistribution Systems With Inverter-Based

Distributed GeneratorsVito Calderaro, Member, IEEE, Gaspare Conio, Vincenzo Galdi, Member, IEEE, Giovanni Massa, Member, IEEE,

and Antonio Piccolo, Member, IEEE

AbstractThe increasing penetration of distributed generation(DG) power plants into distribution networks (DNs) causes var-ious issues concerning, e.g., stability, protection equipment, andvoltage regulation. Thus, the necessity to develop proper controltechniques to allow power delivery to customers in compliancewithpower quality and reliability standards (PQR) has become a rele-vant issue in recent years. This paper proposes an optimized dis-tributed control approach based on DN sensitivity analysis andon decentralized reactive/active power regulation capable of main-taining voltage levels within regulatory limits and to offer ancillaryservices to the DN, such as voltage regulation. At the same time, ittries to minimize DN active power losses and the reactive powerexchanged with the DN by the DG units. The validation of theproposed control technique has been conducted through a severalnumber of simulations on a real MV Italian distribution system.

Index TermsDistributed power generation, energy resources,photovoltaic systems, power distribution, power systems, reactivepower control, voltage control, wind energy.

I. NOMENCLATURE

AI Artificial intelligence.

AR Allowed range.

CR Control range.

DG Distributed generation power plant.

DN Distribution network.

EC European community.

Down and upper threshold values for eachDG.

Maximum frequency.

Active power losses objective function.

Reactive power objective function.

DG active power variations at the PCC.

Manuscript received January 31, 2013; revised May 26, 2013; accepted Au-gust 06, 2013. Date of publication September 17, 2013; date of current versionDecember 16, 2013. Paper no. TPWRS-00132-2013.V. Calderaro, V. Galdi, G. Massa, and A. Piccolo are with the Department

of Industrial Engineering, University of Salerno, 84084 Fisciano, Italy (e-mail:[email protected]; [email protected]; [email protected]; [email protected]).G. Conio is with Italian Vento Corporation Group, Naples 80122, Italy

(e-mail: [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/TPWRS.2013.2280276

DG reactive power variations at the PCC.

DG output voltage variations due to .

DG output voltage variations due to.

DG converter output current.

DN feeder current.

MOGA Multi-objective genetic algorithm.

Total number of DN branches.

Total number of DGs.

OR Operative range.

OSAB-DC Optimized SAB-DC.

PCC Point of common coupling.

DG active power.

DG maximum available active power.

PF Power factor.

Total real power losses objective on branch.

PQR Power quality and reliability standards.

PV Photovoltaic system.

DG reactive power.

constraint due to the convertercurrent.

Total amount of reactive at the thDG PCC.

constraint due to the convertervoltage.

DG maximum reactive power capability.

RES Renewable energy sources.

Voltage versus active power sensitivitycoefficient.

Voltage versus reactive power sensitivitycoefficient.

SAB-DC Sensitivity analysis-based decentralizedcontrol.

0885-8950 2013 IEEE

CALDERARO et al.: OPTIMAL DECENTRALIZED VOLTAGE CONTROL FOR DISTRIBUTION SYSTEMS WITH INVERTER-BASED DISTRIBUTED GENERATORS 231

DG converter output voltage before .

Maximum DG converter output voltagebefore .

DG output voltage at the PCC.

Maximum allowed DG output voltage.

Minimum allowed DG output voltage.

DN feeder voltage.

WDG Wind Distributed Generation unit.

DG transformer and grid filters reactance.

DN reactance at the PCC.

II. INTRODUCTION

T HE penetration of distributed generation power plants(DGs) into distribution networks (DNs) is rapidly in-creasing across the world in recent years. Various incentiveprograms have encouraged the adoption of renewable energysources (RES)-based DGs in order to achieve the ambitiousgovernments targets related to the promotion of a more sustain-able development in the energy sector. For example, in 2009the European Community (EC) has officially recognized [1]the need to promote RES as a priority measure. Furthermore, asignificant increase of RES units within the EC and through theworld is expected in upcoming years [2], [3].Nevertheless, the connection of DGs to DNs poses

well-known technical challenges for the network manage-ment [4]. Among them, voltage rise represents one of the mainissues concerning DG integration into DNs [5], especiallyin the case of nonprogrammable RES-DGs (e.g., wind- orsolar-based ones) in weak DNs, characterized by a small valueof reactance versus resistance ratio [6]. Thus, the necessity todevelop proper control techniques to ensure power delivery tocustomers in compliance with PQR requirements and to furnishancillary services to the network is becoming a relevant issue[7].It is possible to distinguish two main categories through the

various approaches proposed to face DGs integration issues:centralized and decentralized control strategies. The first havebeen extensively investigated by the scientific community[8][10]. However, the application of centralized controlstrategies to the existing networks faces several drawbacks: inaddition to the heavy investments necessary for devices andcontrol systems, all centralized approaches require a highlyreliable communication channel through the overall DN [11].For these reasons, the second category is gaining relevant in-terest in recent years. The adoption of decentralized approachesallows the DGs to provide ancillary services to the DN, such asreserves and voltage support [12]. Moreover, it can have posi-tive effects on both loss minimization and increased generationcapacity because of their flexibility [13].Among decentralized approaches, the ones facing with DGs

capability to provide reactive power support to the DN repre-sent an emerging class of reactive dispatch technologies not yetextensively investigated. One of the most relevant features of

this type of approach is to implement the control action at thepoint of common coupling (PCC), allowing efficient, flexible,and scalable control configurations [14]. An optimization tech-nique aimed at minimizing power losses within the network andvoltage deviation with respect to a reference signal is proposedin [15] for photovoltaic systems (PVs) and applied to the singlefeeder DN presented in [14]. A two-stages voltage control archi-tecture for PV generation and Plug-in hybrid electric vehiclesactive load is proposed in [16]. The optimized control of thereactive power output from PV inverters to reduce voltage de-viations with respect to a control reference is proposed in [17].In [18] the Authors present an optimization strategy applied toa new algorithm for decentralized voltage control based on sen-sitivity analysis [19]. It is capable of maximizing active powerproduction taking into account capability curves of wind DGs(WDGs).This paper presents the extension and completion of previous

works [18] and [19]. The local voltage control strategy allowsto obtain the maximum allowable active power production foreach DG and tries to maximize active power production by con-trolling the DGs reactive/active power exchange with the DNand avoiding, as much as possible, the DG disconnection due tothe infringement of voltage regulatory limits. Its main contribu-tions and differences from the literature can be summarized asfollows: 1) a modular approach that allows to implement validlocal control actions even in absence of a widespread communi-cation channel; 2) a unified strategy for RES management thatpermits to use the same local voltage control approach at thePCC for any type of DG connected to the grid by means ofelectronic converters; 3) an optimization procedure that bringstogether the proposed control technique and the optimizationof its parameters used for voltage regulation; and 4) an artifi-cial intelligence (AI)-based optimization technique capable ofavoiding local minima trapping due to the nonconvexity natureof the problem and to face with uncertainties in a better waycompared with similar approaches presented within the litera-ture [18].The remainder of this paper is organized as follows.

Section III contains the overall description of the reactiveand active power control capabilities of inverter-based DGs,Section IV describes the sensitivity analysis based decentral-ized control (SAB-DC) method, and Section V presents theoptimization approach applied to it in order to compute theoptimized SAB-DC (OSAB-DC). Simulation results obtainedby the OSAB-DC application to a real Italian distributionsystem are illustrated in Section VI, while Section VII containsfinal remarks.

III. REACTIVE/ACTIVE POWER CAPABILITY OFINVERTER-BASED DGS

Typically, RES-DGs are connected to the DNs by means ofelectronic power converters. Among the decentralized controlapproaches, the capability of performing distributed control ac-tions taking into account the converters reactive and/or activepower controllable domains has recently been in-vestigated. Such an approach could be useful to regulate voltageprofiles and/or to offer ancillary services to the DN, maximizingactive power production at the same time [18][21]. In order


Fig. 1. Control system structure.

to define the characteristics of the overall methodology pre-sented within the paper, the structure of the proposed controlsystem applied to a generic schematic diagram of an inverterbased RES-DG is depicted in Fig. 1. It represents the core of theOSAB-DC system and can be implemented on the on-board in-telligence of each RES-DG plant connected to the DN by meansof electronic converters. and are used by theOSAB-DC algorithm to perform the local control action.Concerning the inverter-based DG, the converter voltagedepends on the dc-link voltage and the parameters of the

adopted modulation technique. The reactance representsthe total reactance of the transformers and grid filters used forthe DG connection to the DN.Starting from the maximum power supplied by the primary

source (e.g., PV array, wind turbine), the converter current andvoltage limits impose circular constraints to the DG capabilityto vary and [21].From the equations necessary to compute and it is

possible to write

(1)

Taking into account the root mean square of the convertervoltage as follows:

(2)and, working on the equations, the following can be stated:

(3)

Since in almost all power systems applications is signif-icantly greater than the DN feeder inductance , it is pos-sible to compute the first circular constraint due to the convertervoltage , as stated by the following equation [22]:

(4)

With an analogous approach, it is possible to prove that thecircular constraint imposed by the converter current is

(5)

In order to evaluate the controllable domainof the DG it is necessary to take into account the converters

and values. Thus, the reactive versus active powerconstraints can be written as [21]

(6)

For each working point, the boundary of reactive power de-viation available for the control action must be contained withinthe capability curve defined by

(7)

Fig. 2 shows the capability curves for different PF values,whereas % 1.05 p.u., 0.95p.u., and 1.01 p.u, [21]. To better explain the meaningof the curve it is possible to say that, e.g., for each value ofthe available and each value of the admitted PF, there areupper and lower limits to the capability of varying . Theupper and lower points on the curve in the figure represent theselimits.Since the controllable domain equations mainly concern the

grid-side converter model, this approach can be used for ei-ther full power back-to-back converter-connected WDGs andsingle- or double-stage inverter-connected PVs.

IV. SAB-DCThe proposed controller performs a SAB-DC strategy by

local control of reactive and/or active power exchange betweeneach DG and the DN at the DG connection bus.Sensitivity analysis can be very useful to evaluate the re-

lationships among voltage profiles changes andvalues necessary to implement a correct control action. In fact,sensitivity coefficients give information concerning the varia-tion in the output of a model can be apportioned, qualitatively


Fig. 2. DG capability curves.

and quantitatively, to different sources of variation. Thus, theyare very useful to predict or to face losses, bus voltages, andbranch flow variations due to changes in loads and/or genera-tion profiles [23], [24].Within the paper, the sensitivity coefficients are computed

only for the DN buses where DGs are connected. The evaluationof voltage versus reactive power sensitivity coefficient isimplemented by fixing all DN parameters, including the loadand the generation profiles. At each time step, only the reac-tive power of each bus where the sensitivity parameters mustbe calculated is varied in turn, as stated in [10]. The voltagesat each bus are recorded at each step, and then the sensitivitycoefficients are computed. An analogous procedure is used tocalculate the voltage versus active power sensitivity coefficients

. Thus, if time step represents the system measurementtime step, for each DG, it is possible to state that

(8)

Taking into account the relationships among voltage vari-ations and reactive/active power variations stated in (8), theproposed SAB-DC can be defined. It is focused on main-taining voltage levels within regulatory limits and avoiding,as much as possible, the DG disconnection. Its peculiarityis that the control action starts only if the voltage reacheswarning values, defined by the introduction of two thresholdlevels placed around the nominal voltage value (1p.u.). Two operational ranges are defined within the allowedrange (AR) : the operative range (OR)

, and the control range (CR), as

depicted in Fig. 3.The control operations begin only if the DG bus voltage level

enters the CR, causing a violation of the OR limits. More pre-cisely, when the voltage value enters the CR, a certain amountof reactive and/or active power is injected/absorbed proportion-ally to the difference between the voltage value within the ARand the threshold value placed between the CR and the OR. Theproportionality terms are represented by the sensitivity coeffi-cients as in (8). To better explain the operation of the SAB-DC,

Fig. 3. Allowed, operative, and control ranges for the SAB-DC.

the control algorithm flow chart in the case of upper thresholdviolation is shown in Fig. 4.At the generic time instant , the procedure begins evalu-

ating the difference between and the threshold value, as follows:

(9)

The control strategy considers related toinjection/absorption by

(10)

If reactive power injection is contained within the capabilityregion of the DG converter and the voltage value is in theupper CR, the control system tries first to compensate theentire amount of voltage variation by increasing reactive powerinjected into the DN. Thus, the amount of to be variedis computed as

(11)

The maximum amount of used to bring backvoltage levels within the OR is limited by the DG capabilitycoverage . For each working point,represents the maximum value of reactive power that the con-verter is able to absorb and/or inject into the DN. Therefore,the DG reactive power is chosen according to (7) to yield

(12)

Then, only if is not capable of maintaining thevoltage within the OR, an active power curtailment has to takeplace. The necessary amount is evaluated according to (8) and(10) as follows:

(13)

This active power curtailment allows to decrease voltagelevels and to obtain a higher reactive power injection capa-bility because the working point is moved leftmost on thecapability curve. The right part of the flowchart depicts theSAB-DC in the case of DG bus voltage contained within theOR. If a control action has been previously performed, theSAB-DC tries to recover at its maximum available


Fig. 4. SAB-DC flowchart.

value and to reduce the exchanged withthe DN. Thus, is modified according to

(14)

Furthermore, if is below the active poweramount proportional to , the residual voltage variationis used to reduce reactive power injection into the grid

(15)

The reactive power injection is decreased proportionally tothe entire amount of if no active power curtailmenthas occurred during previous time step and is shown as follows:

(16)

As depicted within this section, the proposed control systemis characterized by two internal loops. The first one acts byvarying the reactive power up to the capability limits in orderto control voltage levels, avoiding active power curtailments. Ifthe reactive control action is not sufficient to maintain voltagelevels within the AR, the second stage of the control algorithmtakes place, reducing active power injection. The DG is discon-nected only if neither reactive and active power control actionsare able to maintain voltage levels within the AR.

The algorithm acts in a similar manner in the case of bottomOR violation. The main difference is that, in this case, only thereactive loop is available to perform the control action. In fact, itis not possible to increase active power more than its maximumavailable value.

V. CR DEFINITION: OPTIMIZATION PROBLEM FORMULATIONAND SOLUTION ALGORITHM

The main issue of the SAB-DC concerns the definition of theoptimal threshold values to adopt for eachDG, where represents the DG index. For a proper approach,it is fundamental to consider that the value of the thresholds de-termines the starting point of the control action, affecting theamount of and exchanged by the DG with the DN.In fact, the SAB-DC actions take place evaluating the differ-ence between the measured voltage and the CR/ORthreshold (or ). Thus, real powerlosses through the DN are influenced by the control action aswell.An optimization approach finalized to the computation of the

optimal thresholds is defined within the paper to prove the gen-eral validity of the SAB-DC method and to allow its applica-bility to every type of DN. In order to find optimal solutions,AI-based optimization techniques can be applied. The most rel-evant advantages of these techniques are to not require differen-tiability and continuity of the objective function and to reducethe risk to be trapped in local minima compared with classicaloptimization strategies. Furthermore, these methods have been


proved to be very suitable to solve nonlinear nonconvex prob-lems [25], [26] like the one addressed within the paper. In fact,AI-based approaches have been proved to be more useful [18]of other ones in which the threshold estimation was based on un-certainties evaluations [27]. To drive an effective optimizationprocedure it is fundamental to obtain threshold values by ap-plying them to a sample composed by variable generation andload demand profiles, assuring the coverage of a wide range ofworking conditions. In particular, DGs thresholds are computedby solving a multi-objective optimization problem based on thedefinition of two objective functions: reactive power objective : it tends to control thereactive power flow through the DN. It could be useful toimprove voltage profiles, to allow high PF levels and toreduce power losses;

power losses objective : it could help to increasesystem efficiency and decrease power generation cost.

More precisely, the objective functions are computed asfollows:

(17)

(18)

The optimization problem is implemented on an overall timeinterval . It consists of optimal thresholds computation, aimedat minimizing the two objective functions. Thus, the overall op-timization problem formulation results in

(19)

(20)

where (20) defines the threshold matrix containing the thresholdvalues of each DG controller. An approach based on multi-ob-jective genetic algorithm (MOGA) interacting with a dailypower-flow routine is performed to solve the optimizationproblem (19). The power flow routine is applied to daily gener-ation and load demand profiles in order to maintain the problemformulation as much general as possible. Typical MOGAoperators like elitism, selection, crossover, and mutation aretaken into account. Several simulations with different MOGAoperators are conducted in order to prove the validity of theproposed approach. For each power-flow routine iteration,the optimization problem is subject to the following systemoperating constraints. Constraints applied to voltage levels, power factor, and re-active and active power capabilities of DGs:

(21)(22)

(23)(24)

Thresholds constraints:

(25)

Distribution line limits related to the maximum transmit-table real power between nodes and :

(26)

VI. CASE STUDY SIMULATIONS AND RESULTS

In order to prove the validity of the proposed OSAB-DC, areal Italian radial DN has been used. The single-line diagram ofthe DN is depicted in Fig. 5. It consists of a 20-kV distributionsystem fed by a 132-kV, 50-Hz subtransmission system witha short-circuit level of 750 MVA through a 150/20-kV /Ygtransformer with rated power 25 MVA, %,and [28]. The primary substation transformers tapis fixed to 1.006 p.u., as one of the two classical strategies usedin Italian distribution systems [20].The sensitivity coefficients have been evaluated according to

the procedure stated in Section IV. Due to the offline procedure,several computations have been performed for various genera-tion and loads values. Even if a very relevant change among thesensitivity coefficients over the various computations is not ob-served, those calculated for the worst case scenario have beenchosen for the SAB-DC optimization phase.At the beginning of the simulations, the power factor has been

set to 1. The maximum voltage variation of % around thenominal value has been imposed as stated in [29].The application of the optimization strategy to the network

has been simulated running the MOGA in interaction with apower-flow routine based on daily generation and load fore-casting, computed on a 10-min base in order to assure assess-ment of the state changes.Two different DGs penetration scenarios applied to the DN

under test have been supposed in order to obtain a general vali-dation of the proposed strategy. In Scenario A, fourWDGs havebeen connected to the DN. In Scenario B, four PVs have beenadded to the previously defined scenario. The connection busesare highlighted in Fig. 5 and Table I, where the rated powervalues of the DGs are specified.The same rated power has been supposed for the WDGs,

while two different configurations have been used for the PVsdue to the modular characteristic of PV plants.Fig. 6 shows the generation power profiles for WDGs and

PVs and load demand for three different load types: residential,commercial, and industrial sectors.Without loss of generality and due to the upper OR infringe-

ment shown in Fig. 7 for both scenarios, the optimizationstrategy has been run to calculate only the upper thresholds.Thus, several simulations of the optimization strategy havebeen computed varying the MOGA parameters for each sce-nario, in order to investigate the assessment of the optimalsolution.


Fig. 5. Italian radial DN under test.

TABLE IDGS RATED POWER AND PCC

Table II shows the intervals adopted for the various MOGAparameters aimed at validating the proposed methodology andthe ones capable of assuring the optimal convergence of opti-mization procedure.As shown in the table, they correspond to ranges that assure

a variability of the MOGA evolution that is not too excessivewhile preventing the premature ending of the simulation due toa minimum trap.Fig. 8 depicts the Pareto fronts of the optimal solution for both

scenarios. It represents the tradeoff between the two conflictingobjective functions. Each one of the points within the frontsrepresents a valid solution for the optimization problem.The controlled voltage values obtained using the solution that

minimizes the reactive power objective function are shown inFig. 9. It is worth noting that, in either scenario, the voltagevalues are contained within regulatory limits.Fig. 10 shows the reactive power injection into the DN by the

DGs participating in the control actions.Scenario A concerns only the reactive power support from

WDG connected at Bus 54, while Scenario B concerns the twoWDGs connected at Buses 46 and 54 and the PV at bus 47.These DGs are those that enter the CR and infringe the voltageconstraints in Fig. 7.It is worth noting a sort of sharing of the control action. In

fact, the reactive power from WDG connected at bus 46 tends

Fig. 6. Generation power profiles and load demand.

to compensate the feeder voltage boosting due to the end ofthe regulation action from WDG at bus 54 around the seventhoperation hour.Fig. 11 shows the capability coverage of the proposed

OSAB-DC for the DGs involved with the control action.The capability curve is computed from the equations in

Section II. It represents the limit to the converters capability


Fig. 7. DGs bus voltages without control action.

TABLE IIMOGA VARIED PARAMETERS AMONG THE VARIOUS TESTS

to vary with respect to the available and the allowedPF range. The crosses in Fig. 11, Scenario A, represent thecontroller working points at each time step within the dailysimulation interval.Similarly, the points in Fig. 11, Scenario B, show the daily

working points of each one of the OSAB-DC controllers thatinfringe the OR limits giving rise to the reactive/active controlaction.It can be stated that the optimized control action does not

require any curtailment of the available active power becauseof the absence of any capability curve constraint infringement.Thus, maximum available active power delivery is allowed bythe proposed control strategy at a local level.

Fig. 8. Pareto fronts of the optimal solution.

Fig. 9. Controlled voltage profiles.

Furthermore, a different wind generation power profile hasbeen applied to the WDG controllers optimized for the first case


Fig. 10. Reactive power sharing among DGs.

Fig. 11. Capability coverage of controlling DGs.

Fig. 12. Modified wind generation profile for Scenario B.

Fig. 13. DGs voltage profiles (a) Without control action enabled (b) With con-trol action enabled.

study in Scenario B in order to assess the robustness of the pro-posed solution, e.g., to unpredicted forecasting errors. Fig. 12depicts the generation power profiles.Load demands have been maintained the same as in the

previous example. Fig. 13 shows the worsening of the uncon-trolled voltage profiles and the correct action performed bythe OSAB-DC. As depicted in Fig. 14, the control action ismaintained within the capability coverage of the DGs evenin this case. Obviously, the PF drops below 0.95 (approxi-mately 0.948). This value has been considered to be acceptabledue to the PF Italian requirements for DGs connected bymeans of electronic converters (to be maintained within therange ). Taking into account more restrictive PF


Fig. 14. Capability coverage for modified wind profiles.

Fig. 15. Generation power profiles and load demand for Winter (Wt) andSummer (Smr) Weekdays.

regulations, the control action necessary to avoid DGs discon-nection could be performed by a curtailment of active powerproduction.Considering that Scenario B represents the worst case among

the two scenarios introduced previously, two further optimiza-tion procedures are presented for this DGs integration scenario.Two different sets of generation and load profiles depicted inFig. 15, concerning a Winter Weekday and a Summer Weekday,have been considered in order to prove the advantages of theproposed optimization methodology. Fig. 16 shows the uncon-trolled voltage profiles at the DGs PCC for the two case studies.It is worth noting that the optimization procedure conducts

to the calculation of thresholds capable of performing a cor-rect control action for either of the case studies, as depicted inFig. 17. Fig. 18 shows the reactive power contribution to thecontrol action by each DG.For each case study, it is worth noting that the control action is

performed by the DGs that infringe the AR, just as it did for the

Fig. 16. DGs bus voltages without control action.

Fig. 17. Controlled voltage profiles.

previous scenarios. Either the Winter and Summer Weekdaysscenarios allow to perform a control action that avoid the DGsdisconnection while maximizing active power production. It is


Fig. 18. Reactive power sharing among DGs.

important to point out that these control actions also take placewithout any active power curtailment.

VII. DISCUSSION AND CONCLUSION

This paper has presented an optimization strategy applied toan SAB-DC for RESDGs, able tomaintain voltage levels withinregulatory limits while producing the maximum available ac-tive power. It avoids the DG disconnection due to violation ofvoltage limits at the connection bus and allows the DGs to offerancillary services to the DN. A unified approach to the con-troller modeling in the case of RES DGs connected to the DN bymeans of electronic converters and the validity of the optimiza-tion strategy have been proposed and proved through severalsimulations run with different DGs penetration levels.All of the solutions presented by the Pareto fronts are valid

thresholds for the optimization problem. The choice of one ofthem represents the optimal tradeoff between active losses andreactive power minimization. It could be useful for the distribu-tion system operator (DSO) in order to determine the amount ofthe independent power producer (IPP) participation to the an-cillary service operations. Thus, it allows to face the conflictinginterests between DN losses minimization or reactive power ac-tion reduction aimed at maximum power production by the IPP.Results obtained by the OSAB-DC application to different

DGs scenarios have been shown. Its robustness with respectto unpredicted changes in generation power profiles has beenproved through the simulations as well. This property allows re-ducing the communication channel reliability requirements thatmainly concern only the optimized thresholds sharing and out-

ages communications. No continuously detailed information isrequired to implement the correct control action.The OSAB-DC is well suited for large DG penetration cases

due to its local nature that reduces the complexity of the con-trol system compared with a completely centralized approach.It is characterized by a unified and modular configuration, andits simplicity and effectiveness would allow its practical imple-mentation. In fact, its extension due to new DGs connectionsonly requires, for example, that the DSO adds the capabilityspecifications to the constraints section and two variables pereach DG to the chromosome that represents the optimal solu-tion to find. By run the MOGA on the DN model it would bepossible to evaluate the optimal thresholds. Then, the online ap-plication of the OSAB-DC would be implemented by sensingvoltage levels at the DG connection bus. These features allowthe OSAB-DC to be suitable for smart grid development sce-narios because of its intrinsic modular and scalable architecture.

APPENDIX

The MOGA used within the paper is implemented as a con-trolled elitist genetic algorithm. It is a variant of the Nondom-inated Sorting Genetic Algorithm II. Due to its elitist nature,it always tends to prefer individuals with better fitness value.It also controls the elite members of the population as the algo-rithm progresses in order to support the individuals that can helpto increase the diversity of the population necessary to convergeto an optimal Pareto front.The MOGA has been run several times on the validation

scenarios during the validation phase of the proposed controlmethodology. These computations have been conducted byvarying the MOGA parameters within the ranges indicated inthe test range column in Table II. It has been noted that thePareto front found by the MOGA tended to the same conver-gence region if the setting parameters were contained within theoptimal range presented in Table II in each one of the validationscenarios. Thus, this parameter range represents the optimal setto assure a global convergence of the optimization procedure.

REFERENCES[1] Directive 2009/28/EC of the European Parliament and of the

Council, Official J. Eur. Union, pp. 1662, 2009.[2] Communication from the Commission to the European Parliament, the

Council, the European Economic and Social Committee and the Com-mittee of the Regions, Energy roadmap 2050, in Proc. COM, 2011,pp. 120.

[3] Int. Energy Agency, World Energy Outlook 2012, Executive Sum-mary, 2012, pp. 19.

[4] R. A. Walling, R. Saint, R. C. Dugan, J. Burke, and L. A. Kojovic,Summary of distributed resources impact on power delivery systems,IEEE Trans. Power Del., vol. 23, no. 3, pp. 16361644, Jul. 2008.

[5] C.Masters, Voltage rise the big issue when connecting embedded gen-eration to long overhead 11 kV lines, Power Eng. J., vol. 16, no. 1,pp. 512, 2002.

[6] A. E. Kiprakis and A. R. Wallace, Maximising energy capture fromdistributed generators in weak networks, Proc. Inst. Elect. Eng., Gen.,Transm., Distrib., vol. 151, no. 5, pp. 611618, 2004.

[7] E. J. Coster, J. M. A.Myrzik, B. Kruimer, andW. L. Kling, Integrationissues of distributed generation in distribution grids, Proc. IEEE, vol.99, no. 1, pp. 2839, Jan. 2011.

[8] A. Casavola, G. Franz, D. Menniti, and N. Sorrentino, Voltage reg-ulation in distribution networks in the presence of distributed genera-tion: A voltage set-point reconfiguration approach,Electr. Power Syst.Res., vol. 81, pp. 2534, 2011.


[9] M. S. P. Carvalho, P. F. Correia, and Lus A. F. M. Ferreira, Dis-tributed reactive power generation control for voltage rise mitigationin distribution networks, IEEE Trans. Power Syst., vol. 23, no. 2, pp.766772, May 2008.

[10] A. Keane, L. F. Ochoa, E. Vittal, C. J. Dent, and G. P. Harrison, En-hanced utilization of voltage control resources with distributed genera-tion, IEEE Trans. Power Syst., vol. 26, no. 1, pp. 252260, Feb. 2011.

[11] H. E. Farag and E. F. El-Saadany, A novel cooperative protocol fordistributed voltage control in active distribution systems, IEEE Trans.Power Syst., vol. 28, no. 2, pp. 16451656, May 2013.

[12] A. G. Madureira and J. A. Pecas Lopes, Coordinated voltage supportin distribution networks with distributed generation and microgrids,IET Renew. Power Gener., vol. 3, no. 4, pp. 439454, 2009.

[13] L. F. Ochoa and G. P. Harrison, Minimizing energy losses: Optimalaccommodation and smart operation of renewable distributed genera-tion, IEEE Trans. Power Syst., vol. 26, no. 1, pp. 198205, Feb. 2011.

[14] K. Turitsyn, P. Sulc, S. Backhaus, and M. Chertkov, Options for con-trol of reactive power by distributed photovoltaic generators, Proc.IEEE, vol. 99, no. 6, pp. 10631073, Jun. 2011.

[15] H.-G. Yeh, D. F. Gayme, and S. H. Low, Adaptive VAR control fordistribution circuits with photovoltaic generators, IEEE Trans. PowerSyst., vol. 27, no. 3, pp. 16561663, Aug. 2012.

[16] B. A. Robbins, C. N. Hadjicostis, and A. D. Domnguez-Garca, Atwo-stage distributed architecture for voltage control in power distribu-tion systems, IEEE Trans. Power Syst., vol. 28, no. 2, pp. 14701482,May 2013.

[17] K. Tanaka, M. Oshiro, S. Toma, A. Yona, T. Senjyu, T. Funabashi, andC. H. Kim, Decentralised control of voltage in distribution systemsby distributed generators, IET Gener. Transm. Distrib., vol. 4, no. 11,pp. 12511260, 2010.

[18] V. Calderaro, V. Galdi, G. Massa, and A. Piccolo, Distributed gen-eration management: An optimal sensitivity approach for decentral-ized power control, in Proc. 3rd Conf. Exhib. Innovative Smart GridTechnol., 2012, pp. 18.

[19] V. Calderaro, V. Galdi, G. Massa, and A. Piccolo, Distributed gen-eration and local voltage regulation: An approach based on sensitivityanalysis, in Proc. 2nd Conf. Exhib. Innovative Smart Grid Technol.,2011, pp. 18.

[20] V. Calderaro, G. Conio, V. Galdi, and A. Piccolo, Reactive powercontrol for improving voltage profiles: A comparison between two de-centralized approaches, Electr. Power Syst. Res., vol. 83, no. 1, pp.247254, 2012.

[21] N. R. Ullah, K. Bhattacharya, and T. Thiringer, Wind farms as reactivepower ancillary service providersTechnical and economic issues,IEEE Trans. Energy Conv., vol. 24, no. 3, pp. 661672, Sep. 2009.

[22] M. Ohshima and E. Masada, The P,Q controllable domain of a singlephase PWM converter to preserve sinusoidal AC current waveform,IEEE Trans. Power Electron., vol. 15, no. 3, pp. 485494, .

[23] D. K. Khatod, V. Pant, and J. Sharma, A novel approach for sensitivitycalculations in the radial distribution system, IEEE Trans. Power Del.,vol. 21, no. 4, pp. 20482057, Oct. 2006.

[24] H. M. Ayres, W. Freitas, M. C. De Almeida, and L. C. P. Da Silva,Method for determining the maximum allowable penetration levelof distributed generation without steady-state voltage violations, IETGener. Transm. Distibr., vol. 4, no. 34, pp. 495508, 2010.

[25] J. G. Vlachogiannis and K. Y. Lee, A comparative study on particleswarm optimization for optimal steady-state performance of power sys-tems, IEEE Trans. Power Syst., vol. 21, no. 4, pp. 17181728, Nov.2006.

[26] G. K. Viswanadha Raju and P. R. Bijwe, Reactive power/voltage con-trol in distribution systems under uncertain environment, IET Gener.Transm. Distrib., vol. 2, no. 5, pp. 752763, 2008.

[27] F. Bignucolo, R. Caldon, and V. Prandoni, Radial MV networkvoltage regulation with distribution management system coordinatedcontroller, Electr. Power Syst. Res., vol. 78, pp. 634645, 2008.

[28] S. Conti and A. M. Greco, Innovative voltage regulation method fordistribution networks with distributed generation, in Proc. Int Conf.Electricity Distrib., 2007, pp. 14.

[29] IEEE Application Guide for IEEE Std. 1547, IEEE Standard forInterconnecting Distributed Resources With Electric Power Systems,1547.2-2008, 2009, pp. 1-207.

Vito Calderaro (M10) received the M.Sc. degree inelectronic engineering and Ph.D. degree in informa-tion engineering from the University of Salerno, Fis-ciano, Italy, in 2001 and 2006, respectively.He is currently a Research Assistant with the

Department of Industrial Engineering, University ofSalerno, Fisciano, Italy. His current research focuseson integration of distributed generation systemson electrical distribution networks, soft computingmethodologies in power system applications, andprotection and diagnostic of complex systems. He

has served as reviewer for many international journals and conferences.

Gaspare Conio received the M.Sc. degree in elec-tronic engineering from the University of Salerno,Fisciano, Italy, in 2008.Since 2009, he has worked in cooperation with

the University of Salerno on grid integrationof wind power plants. In 2010, he joined theItalian Vento Power Corporation Group, Naples,Italy, where he is involved with the development,operation, and maintenance of large wind farms.His main research activities concern the gridintegration of renewable energy power plants,

power systems control, and energy storage.

Vincenzo Galdi (M98) received the M.Sc. degreein electronic engineering from the University ofSalerno, Fisciano, Italy, in 1994, and the Ph.D. de-gree in electrical engineering from the Universityof Naples, Naples, Italy, in 1999.Since 2003, he has been an Associate Professor

with the University of Salerno, Fisciano, Italy,now with the Department of Industrial Engi-neering. His current research interests includeapplication of ICT to power systems and topower management and energy efficiency in

transportation systems, smart grids, and ITS.

Giovanni Massa (M12) received the M.Sc. degreein electronic engineering and Ph.D. degree in infor-mation engineering from the University of Salerno,Fisciano, Italy, in 2009 and 2013, respectively.His main research interests focus on smart-grid

development issues, including integration of dis-tributed generation systems on electrical distributionnetworks, soft computing methodologies in powersystem applications, energy saving in electric trans-portation, and Intelligent Transportation Systems.

Antonio Piccolo (M06) received theM.Sc. degree inelectrical engineering from the University of Naples,Naples, Italy, in 1974.Since 1990, he has been a Full Professor with the

University of Salerno, Fisciano, Italy, currently as afoundingmember of the Department of Industrial En-gineering. His main fields of interest are the appli-cation of ICT to electric energy infrastructures andtransportation systems. He has served as a reviewerand session chairman for many international confer-ences. He has authored or coauthored more than 130

papers mainly in international journals and conferences in the field of powersystems.Prof. Piccolo is a Chartered Engineer.

06601722

Documents

Transcript of 06601722