1

Centralized and distributed active and reactivepower control of a utility connected microgrid using

IEC61850Alba Colet-Subirachs, Albert Ruiz-Alvarez, Oriol Gomis-Bellmunt, Member, IEEE,Felipe Alvarez-Cuevas-Figuerola, Antoni Sudria-Andreu, Senior Member, IEEE

Abstract—This paper describes the control algorithm of a util-ity connected microgrid, based on independent control of activeand reactive power and operating in centralized and distributedoperation mode. The microgrid is composed of several (three inthe experimental tests conducted) configurable units includinggeneration, storage and load units. These units are interfacedwith the microgrid through a Voltage Source Converter (VSC)and are controlled by the nodes of the communication system bymeans of IEC 61850. The proposed control strategy is validatedby simulation in Matlab/Simulinkr and with experimental tests.

Index Terms—Microgrid, control algorithm, IEC 61850.

NOMENCLATURE

AcronymsCHP Combined Heat and PowerRTC Real Time ClockSoC State of ChargeVSC Voltage Source Converter

Subscript Superscriptsi Number of iSocket ∗ Set pointM Maximumm Minimump Active powerq Reactive power

I. INTRODUCTION

THE need for more reliable and flexible power systemsalong with the great potential of modern control and

communication systems and power electronics, has led todevelopment of the smart grid concept [1]–[3]. Modern gridswill be required to be active and to adapt to a number offault events ensuring the system optimum performance duringand after faults occur. Furthermore, modern grids will haveto integrate the increasing penetration of renewable energy of

A. Colet-Subirachs, A. Ruiz-Alvarez, O. Gomis-Bellmunt and A. Sudria-Andreu are with Catalonia Institute for Energy Research (IREC), Electri-cal Engineering Area, C Josep Pla, 2, edifici B2, Planta Baixa - 08019Barcelona, Spain (e-mail: acolet@irec.cat, aruiz@irec.cat, ogomis@irec.cat,asudria@irec.cat)

O. Gomis-Bellmunt and A. Sudria-Andreu are with Centre d’InnovacioTecnologica en Convertidors Estatics i Accionaments (CITCEA-UPC), De-partament d’Enginyeria Electrica, Universitat Politecnica de Catalunya, ETSd’Enginyeria Industrial de Barcelona, and EU d’Enginyeria Tecnica Industrialde Barcelona, Barcelona - 08028, Spain (e-mail: gomis@citcea.upc.edu,antoni.sudria@citcea.upc.edu)

F. Alvarez-Cuevas-Figuerola is with Endesa Servicios.sl, Av Paral·lel, 51 -08004 Barcelona, Spain (e-mail: felipe.alvarezcuevas@endesa.es)

intermittent nature. This can be achieved using more flexiblepower systems including power electronics, energy storagesystems, demand side management and microgrids.

A microgrid is defined as an aggregator of several microgen-eration units, storage devices and controllable loads operatingas a single system that provides electricity and thermal energy[4], [5]. Microgrid is a concept that incorporates distributedenergy resources (DER), including distributed generation (DG)and distributed storage (DS). To manage it, a network ofcommunication devices must provide the microgrid with thenecessary intelligence to allow customers and utility compa-nies to collaboratively manage the power generated, deliveredand consumed through real-time, bidirectional communica-tions. Thus, communications are essential in order to ensurethe proper operation of all the microgrid components [6]. Pro-tection devices, control commands and power flow regulation,together with real-time measurements must be integrated ina network that provides the necessary levels of quality andreliability. The architecture of this network can be centralized,distributed or hierarchical. In a hierarchical architecture thereare different types of controllers that, depending on the controllevel, can be classified into: Distribution Management System(DMS), Microgrid Central Controller (MGCC) and LocalControllers (LC) [7].

A control strategy must be devised to ensure the long-termstable operation of the microgrid under various load conditionsand different configurations [8], [9]. Depending on the timeframe this control algorithms can be separated in:• Primary, there is a momentary adjust,• Secondary, the operation range is several minutes [10]

and• Tertiary, the operation range is 15-20 minutes.

The microgrid can operate in grid-connected or stand-alonemode. Concerning the methods used for managing the powerof a microgrid two different classes can be distinguished: thecentralized operation that concentrates information in a nodeand the decentralized operation which uses local informationand provides more autonomy to the DER units [11], [12].

The present paper focuses on a simple method for con-trolling the active and reactive power of a utility connectedMicrogrid operating in grid-connected mode. It is based onpeer-to-peer and plug-and-play concepts [13] and operateseither in centralized or in distributed mode. Around the worldthere are a number of active microgrid projects involved intesting control strategies and operational modes [14]. This

2

paper presents the design, simulation and experimental resultsof a control algorithm implemented in a utility connectedmicrogrid. Communications between devices is based on thestandard IEC61850.

The paper has been structured as follows. Section II de-scribes the microgrid under investigation and the projectframework. The control algorithm to manage the microgridis presented in section III. The overall system is simulatedin Matlab/Simulinkr (section IV). Section V describes themicrogrid test platform where the control algorithms arecurrently being tested. Finally, the results are analysed insection VI.

II. MICROGRID CONCEPT

The Catalonia Institute for Energy Research (IREC) iscurrently developing a microgrid based on real and emulatedenergy resources in order to evaluate different scenarios [15].This paper describes a part of such microgrid that is alsoinvolved in the Smartcity project located in Malaga (southof Spain) and managed by Endesa, the local utility.

The Malaga Smartcity project is a demonstration projectin which it is intended to deploy and integrate the followingitems in the current grid:• A highly reliable and efficient Broadband Power Line

communications framework.• Micro generation and micro storage within the low-

voltage grid.• Mini generation and mini storage within the medium-

voltage grid.• A small fleet of bi-directional electrical vehicles.• A new and efficient street lighting system.• A few thousands of smart meters.• Improved grid self healing automation.The communication architecture of such city is composed

of a hierarchical layer system. The bottom layers are embodiedby these two elements:

iNode (Intelligent Node): Develops the global managementof microgrid tasks and connects supervising and control sys-tems (through a Gateway) to the terminal equipment (iSocket).Its functions are managing the data received from iSockets andsetting overall operation of the microgrid, developing its ownalgorithms. Its main tasks include:• Regulation: Control of energy generation and consuming

entities.• Billing: Energy measurement and real-time pricing.• Management: Asset management and condition based-

maintenance.• Metering: full system monitoring.• Security of the microgrid electrical system

The operational requests of this controller are aggregation andcoordination of iSockets and electrical safety guarantee.

iSocket (Intelligent Socket): It is an element located inthe lowest hierarchy layer of the communication system. Ithandles the device connected to it (generation, storage orloading), based on the instructions received from the iNode.The operational requests of this controller are local regulationand electrical safety guarantee.

III. CONTROL ALGORITHM

This paper proposes implementing a control algorithm withthe capability to achieve the system goals using the availableunits, interfaced to the microgrid through a voltage-sourceconverter (VSC). The algorithm uses decision laws that havecertain parameters to manage the microgrid and establishesthe equilibrium points of the system in a indirect manner.It is based on independent control of active and reactivepower in grid connected mode (PQ control) and consists onsecondary and tertiary microgrid’s control. One advantage ofthis algorithm is its simplicity in terms of the reduction ofinformation flow and data exchange between all nodes in thenetwork and the fast adaptability to configuration changes inthe microgrid, such as unit losses or reconnections.

Two operational control modes are considered:A The centralized control mode suggests that a central node,

iNode, collects the microgrid measurements (sent fromiSockets) and decides next actions according to the utilitygoals.

B The distributed control mode suggests that advancedcontrollers are installed in each node forming a dis-tributed control system. These controllers are completelyindependent from the superior nodes and only use theelectric market price for the regulation.

To analyze the control algorithm it is important to take intoconsideration the sign criteria used in this paper (1):{

P < 0→ GenerationP > 0→ Load

{Q < 0→ CapacitiveQ > 0→ Inductive

(1)

A. Centralized operation mode

This operational mode centralizes the control of all devicesin the iNode, which controls the whole microgrid to followgiven active and reactive power references.

1) iNode controller: iNode uses two independent PI con-trollers (Fig. 1), which are responsible for controlling thewhole active and reactive power flux of the microgrid.

pPk

iPk

s

Active Power PI Controller

Integrator

*P

totP

pFSaturation

Reactive Power PI Controller

Integrator

*Q

totQ

Saturation

qFpQk

iQk

s

Fig. 1. iNode PI controller

According to Fig. 2, the iNode has the following inputs:• P ∗ [W], Q∗ [VAr]: active and reactive power set points

from the utility

3

• Ptot [W], Qtot [VAr]: total active and reactive measure-ments which can be obtained as a direct value providedby the metering system or as a calculated value using thepower sent by each iSocket, i.e, the sum of all active andreactive power of the connected iSockets

e the current price of energy received from the utility[ce/kWh].

The iNode outputs are:• Fp active power control signal and• Fq reactive power control signal,

where −100 6 Fp 6 100 and −100 6 Fq 6 100.

1

I-NODECentralized operationP*, Q* FP, FQ

P1, Q1

VSC unit

Pi, Qi

P1, Q1

P1*, Q1*

P2, Q2

P2*, Q2*

P3, Q3

P3*, Q3*

PN, QN

2

Priority Load

Diesel generator3

VSC unit

VSC unit

Non-priority Load

Pi, Qi

Pi*, Qi*Battery

banki

VSC unit

PN*, QN*Wind

turbineN

VSC unit

P2, Q2

P3, Q3

PN, QN

Fig. 2. Centralized operation block description

2) iSocket controllers: iSockets receive the control signalsFp and Fq and apply equations (3) and (4) to calculatethe active and reactive power to set (P ∗i , Q∗i ) to the VSCconnected to them.

P ∗i =

Pi,M βpi < Fp (2a)

Pi,m+(Pi,M − Pi,m)Fp − αpi

βpi − αpiβpi≥Fp≥αpi(2b)

Pi,m αPi > Fp(2c)

with βpi > αpi

Q∗i =

Qi,M βqi < Fq(3a)

Qi,m+(Qi,M −Qi,m)Fq − αqi

βqi − αqiβqi≥Fq≥αqi(3b)

Qi,m αqi > Fq(3c)

with βqi > αqi

These equations are piecewise-defined functions dividedinto different sections depending on the power profile to beset to each microgrid unit. As an example, Fig. 3 shows theprofile corresponding to a load that has two states: connected(P > 0) and disconnected (P = 0). When Fp ≥ βpi the loadswitches from the disconnected to the connected mode and itdisconnects when αpi ≥ Fp. Finally, if βpi ≥ Fp ≥ αpi, itscurrent state depends on the previous situation.

However, this paper considers a power profile divided intothree sections (Fig. 4): the first equations (2a, 3a) corresponds

100

Pi*

FPi,m

βpi αpi-100

Generation area

Consumption area

i,M

Fig. 3. On/off load profile function set by the iSocket

to the maximum power to be set (Pi,M , Qi,M ), the secondsection (2b, 3b) is a ramp between maximum and minimumpower, and the third (2c, 3c) corresponds to the minimumvalue (Pi,m, Qi,m).

100

Pi*

Fp

i,m

βpi αpi-100

Generation area

Consumption area

100

Qi*

Fq

-100

i,m

αqi βqi

i,M

Capacitive area

Inductive area

i,M

Fig. 4. General power profile functions set by the iSocket

The parameters α and β, configured at each iSocket, areused to define the power limits and the participation prioritiesof the microgrid units. Its value is recalculated dynamically:as there is a change in the state of a microgrid unit, theiSocket interfaced with it perceives this variation, and modifiesα and β. Also, these parameters are recalculated according tovariables such as the energy price.

Depending on the type of microgrid unit, iSockets aredivided into three main cases:

a) Generation iSockets: A generation node i will becharacterized by:{

αpi = αpi0 − kEei,c + kp · eβpi = βpi0 − kEei,c + kp · e

{αqi = αqi0

βqi = βqi0(4)

Where

αpi0, βpi0, αqi0, βqi0 are a selectable parameters use toprioritize each power generation source,ei,c is the generation cost [ce/kWh],kE is a multiplier of the generation cost andkp is a multiplier of the energy cost.

b) Storage iSockets: A storage node i will be character-ized by: αpi = αpi0 − kwWi,a + kpe

βpi = βpi0 − kw(Wi,M −Wi,a) + kpeαqi = αqi0

βqi = βqi0

(5)

Where

Wi,M is the maximum storable energy (100%)Wi,a is the available energy (0-100%) andkw is a multiplier of the available energy.

4

c) Load iSockets:{αpi = αpi0 − krei,crc + kpe

βpi = βpi0 − krei,crc + kpe

{αqi = αqi0

βqi = βqi0(6)

Where

ei,crc is the cost of load reduction [ce/kWh] andkr is a multiplier of the load reduction cost.

For reactive power control, αqi and βqi parameters can beequal for all units as in the previous expressions or voltagedependant. If they are equal reactive power will be proportion-ally divided between units. This can imply important voltagedeviations if some nodes are located at large distances andthere are important impedances between them. In this cases,αqi and βqi parameters can be voltage dependant, so that thenodes with different voltages can eventually contribute moreor less to the total reactive power reference.

B. Distributed operation mode

In this operational mode iSockets become more independentwithin the operating area. They manage themselves and do notreport their decisions to iNode. As can be seen in Fig. 5 there isno feedback of the metered variables Pi and Qi, from the VSCunit to the iNode since the iSocket manages that information.

iSockets calculate their local active and reactive powerusing the same expressions as in centralized operation modeusing Fp = 0 and Fq = 0, therefore (choosing appropiateα and β parameters) the microgrid will respond to energyprice increases connecting more generation, disconnectingnon-critical loads and selling the energy stored in storageunits. For energy price reductions the microgrid units willrespond storing energy, disconnecting high cost generatorsand connecting loads. The resulting microgrid operating pointsobtained depend on energy prices and the availability of energystorage systems.

1

I-NODEDistributed operation

VSC unit

P1, Q1

P1*, Q1*

P2, Q2

P2*, Q2*

P3, Q3

P3*, Q3*

PN, QN

2

Priority Load

Diesel generator3

VSC unit

VSC unit

Non-priority Load

Pi, Qi

Pi*, Qi*Battery banki

VSC unit

PN*, QN*Wind

turbineN

VSC unit

Fig. 5. Distributed operation block description

IV. SIMULATION STUDY

The proposed control scheme has been simulated usingMatlab/Simulinkr. The scenario presented consists of sixnodes (Figure 6):

TABLE IMICROGRID SIMULATION PARAMETERS

Parameters Generation nodes Storage nodes Load nodesiSocket1 iSocket2 iSocket3 iSocket4 iSocket5 iSocket6

Pi,m [W] -4000 -3000 -4000 -4000 0 0Pi,M [W] 0 0 4000 4000 3000 2000αPi0 40 -60 -180 -180 -120 -80βPi0 80 -20 60 60 -60 -20kp 0 0 6 6 2 2kE 0 0 - - - -kw - - 1.2 1.2 - -kr - - - - 1 0

• Generation nodes:– iSocket1 is managing a wind turbine of a 4kW rated

power. It is a priority generation node due to therenewable nature of the source. In such case theα and β parameters are configured to operate as apriority generator and have a high value, near upperlimit of Fp.

– iSocket2 is a diesel generator of 3kW rated power.Its use has associated a cost of generation.

• Storage nodes: there are two storage nodes. Both controla battery with the same characteristics but different stateof charge. The power profile set to each of them dependson this state.

– iSocket3 controls a 20% charged battery.– iSocket4 controls a 75% charged battery.

• Load nodes– iSocket5 manages a priority load which need to be

supplied almost uninterruptedly, as a result the α andβ parameters are configured to operate as a priorityload and had a value near the lower limit of Fp.

– iSocket6 controls a conventional dispatching load.Table I shows the description of the parameters used in thesimulation.

FP

EeneP6*

iSocket6

FP

EeneP5*

iSocket5

FP

EeneP4*

iSocket4

FP

EeneP3*

iSocket3

FP

EeneP2*

iSocket2

FP

EeneP1*

iSocket1P*

P1

P2

P3

P4

P5

P6

FP

iNode P

P_sp P

con6

P_sp P

con5

P_sp P

con4

P_sp P

con3

P_sp P

con2

P_sp P

con1

t

conP6

conP5

conP4

conP3

conP2

P1

conP1Eene

P6

P5

P4

P3

P2

FP

Ptot

Clock

Fig. 6. Microgrid Matlab/Simulinkr block description

Static and the dynamic analysis have been conducted. Inboth cases the analysis is focused on active power. Reactivepower would be shared proportionally by all the involved units.The cases studied have focused on changes that occur due tothe variation of energy price.

A. Steady state analysisA case study is presented to illustrate the equilibrium points

of the proposed algorithm. These equilibrium points can be

5

calculated using the equations (??)-(6). Fig. 7 shows themicrogrid operating point when Fp = −22 and and e = 10ce/kWh, which corresponds to a generation of 2417 W. Thisvalue is derived from the sum of power set points provided bythe microgrid units. As it can be seen in the right graph, whenFp increases, the total power set to the microgrid increases too.Likewise the left graph shows the layout of the parameters αp

and βp for each iSocket. As an example it can be observed thatthe conventional load, iSocket6, has αp6 = −60 (P6,m =0 W)and βp6 = 0 (P6,M =2000 W) and at the point in the studyit has a reference power consumption of 1267W (P ∗6 =1267W).

-100 -80 -60 -40 -20 0 20 40 60 80 1000

1

2

3

4

5

6

7

-4000 W 0 W

-4000 W 1 Wind turbine

-3000 W 0 W

-150 W 2 Diesel generator

-4000 W 4000 W

-3467 W 3 Battery 75%

-4000 W 4000 W

933W 4 Battery 20%

0 W 3000 W

3000 W 5 Priority load

0 W 2000 W

1267 W 6 Conventional load

FP

iSoc

ket

-100 -50 0 50 100-1.5

-1

-0.5

0

0.5

1

1.5x 10

4

-2417 W

FP

Pow

er [

W]

Fig. 7. Microgrid steady state analysis when e = 10 ce/kWh and Fp = −22

If there is a significant increase in energy price (from10 to 15 ce/kWh), the price sensitive elements respond tosuch a change. In this case (Fig. 8), the load and storageiSockets scroll right. So the microgrid modifies its responseand at the studied point, Fp = −22, increases the generation,P ∗ =5284 W. In this new scenario, the parameters of iSocket6are αp6 = −50 and βp6 =10 and its consumption is reducedto 933 W. Therefore, when the energy price increases a 50%,the conventional load is reduced by a 26,4%.

-100 -80 -60 -40 -20 0 20 40 60 80 1000

1

2

3

4

5

6

7

-4000 W 0 W

-4000 W 1 Wind turbine

-3000 W 0 W

-150 W 2 Diesel generator

-4000 W

-4000 W 3 Battery 75%

-4000 W 4000 W

-1067 W 4 Battery 20%

0 W 3000 W

3000 W 5 Priority load

0 W 2000 W

933 W 6 Conventional load

FP

iSoc

ket

-100 -50 0 50 100-1.5

-1

-0.5

0

0.5

1

1.5x 10

4

-5284 W

FP

Pow

er [

W]

Fig. 8. Microgrid steady state analysis when e = 15 ce/kWh and Fp = −22

B. Dynamic analysis

The system behavior in centralized mode has been studiedwith step changes of the reference power P ∗ in time intervalsof 30s. The energy price is fixed at 10 ce/kWh. Fig. 9 and 10show the system response.

0 50 100 150-1

-0.5

0

0.5

1x 10

4

Time [s]

Pow

er [

W]

Ptot

P*

0 50 100 150-50

0

50

Time

FP

T0

T0

Fig. 9. Microgrid dynamic set points when e = 10 ce/kWh

As can be seen in Fig. 9 the sum of powers of themicrogrid, Ptot, fits the sequence of values entered, P ∗ ={8000, 0,−5500} W. Likewise, the control signal Fp, changesits operating point according to the PI output of the iNode.Figure 10 shows the power provided by each microgridunit, Pi. According the sign criteria established in (1), thegeneration units (iSocket 1 and 2) remain at the bottom of thegraph and the loads (iSocket 5 and 6) at the top. The windturbine (iSocket1), presents a fluctuating power curve due tothe turbulent wind flow implemented in the simulation. On theother hand, the priority load is almost supplied uninterruptedlyand remained stable at 3000W.

Comparing the results of the steady state analysis and thedynamic analysis at Fp = −22 and e =10 ce/kWh (Figure7 and 10), it can be seen that the metered values of eachiSocked match reference values, except for the wind turbine(P ∗1 = −4000W and P1 = −1647W) because its generationdepends on available wind power.

0 50 100 150

-4000

-3000

-2000

-1000

0

1000

2000

3000

4000

Time [s]

Pow

er [

W]

1 Wind turbine2 Diesel generator3 Battery 75% 4 Battery 20% 5 Priority load6 Conventional load

T0

P6=1267W

Fig. 10. Microgrid Pi response for a e = 10ce/kWh during the dynamicsimulation

V. SYSTEM DESCRIPTION

The microgrid experimental platform under investigation(Fig. 11) is composed of two main systems: the communi-cation system and the power system.

A. Power system description

The microgrid power system is composed of a three con-figurable units:

6

Wide Area Network

Microgrid comunication bus

I-SOCKET 1Generation unit

I-NODE

I-SOCKET 2Storage system

I-SOCKET 3Consumption unit

CONTROL CENTER

IEDs

IED

Ethernet

IEC 61850

CAN Bus CAN Bus CAN Bus

ACDC

DCAC

ACDC

DCAC

ACDC

DCAC

LV grid

Emulator 2Emulator 1 Emulator 3

Eolic generation group

Energy storage group

Load group

400V/400V4kVA

400V/400V4kVA

400V/400V4kVA

VSC

VSC VSC

VSC

VSC

VSC

Fig. 11. Microgrid functional block description

• Generation unit: emulates different types of generationsuch as wind and solar, reproducing the real behaviour,and in the case of renewable energy sources, reproducingthe variable nature and dependence on external climato-logical factors.

• Load unit: emulates the real behaviour of different typesof consumption based on sensitive-loads and/or non-sensitive-loads using various load profiles.

• Energy storage unit: emulates a storage system which,according to needs, can be either a battery or an electricvehicle.

The use of these configurable units (shown in Fig. 12)allows reproducing the scenarios that are deemed of interestwithout having to wait for appropriate weather conditions.The reconfigurability property of them allows emulating anysituation generating or consuming real power.

Each unit of the microgrid power system has two VSCcomposed of a three-phase inverter, AC and DC transducersand protective devices. Its design enables utilization as aconfigurable renewable generation emulator, a battery bankor a load. The converters are connected in a back to backconfiguration, so while one acts like a bidirectional boost-rectifier, the other one transfers power to the grid according tothe set point given by the iSocket. Therefore, there is a powerflow through these devices in which only converter losses areconsumed.

B. Communication system description

The communication system is composed of four nodes: oneiNode and three iSockets. Each node is implemented in a

Fig. 12. Microgrid power units configuration

Linux embedded control board with the following character-istics:• ARM9 based microcontroller, 180MHz• 32MByte flash memory• 32MByte SDRAM memory• Isolated 10/100 Base-T Ethernet• Isolated CAN• RTC• 12 isolated digital inputs• 2 isolated digital outputsDepending on the communication layer, a different commu-

nication protocol is used:• Communication between iNode and iSocket is done with

the standard IEC 61850 [16], [17]. Fig. 13 shows the setof objects defined in the microgrid using this standard[18]. The IEC 61850 series defines an abstract com-munication architecture that provides a hierarchical setof classes associated to services. These services, calledACSI (Abstract Communication Service Interface), aremapped to MMS (Manufacturing Message Specification).Furthermore, it uses a configuration language and anobject model to describe the system components, knownas IEDs (Intelligent Electronic Device).

M

LN DRCT: DER unit controller characteristics

LN DRCS: DER unit status

LN DRCC: DER unit control actions

LN FSEQ: Sequencer

LN DCCT: DER ecnonomic dispatch parameters

LN MMTR: Metering

LN CSWI: Switch controller

LN XCBR: Circuit creaker

LN ZINV: Inverter

LN ZRCT: Rectifier

LN MMDC: DC Measurements

LN MMXU: AC Measurements

LD LOAD

M

LN DRCT: DER unit controller characteristics

LN DRCS: DER unit status

LN DRCC: DER unit control actions

LN FSEQ: Sequencer

LN DCCT: DER ecnonomic dispatch parameters

LN MMET: Metereological conditions

LN MMTR: Metering

LN CSWI: Switch controller

LN XCBR: Circuit creaker

LN ZINV: Inverter

LN ZRCT: Rectifier

LN MMDC: DC Measurements

LN MMXU: AC Measurements

LD WIND TURBINE

GLN DGEN: DER unit generator

LN DRAT: DER generator ratings

M

LN DRCT: DER unit controller characteristics

LN DRCS: DER unit status

LN DRCC: DER unit control actions

LN FSEQ: Sequencer

LN DCCT: DER ecnonomic dispatch parameters

LN MMTR: Metering

LN CSWI: Switch controller

LN XCBR: Circuit creaker

LN ZINV: Inverter

LN ZRCT: Rectifier

LN MMDC: DC Measurements

LN MMXU: AC Measurements

LN ZBAT: Battery

LN ZBTC: Battery charger

LD BATTERY

Fig. 13. Set of objects defined in the microgrid test bench with IEC 61850standard

• Communication between iSocket and its appropriate VSCuses a CAN proprietary protocol. The iSocket transmitsa data frame that contains a command word, a P ∗i and

7

Q∗i to be set to the VSC. The VSC answers with threeCAN messages, which contain the status of the unit andother information such as Pi and Qi.

Data exchange between iNode and iSockets will be mon-itored in a IEC 61850 SCADA. Fig. 14 shows the graphicdisplay of the remote microgrid viewed by the operator. In thisapplication the software system is only monitoring the remotestatus and alarm conditions of the microgrid. The monitoredvalues of the microgrid units are: the active and reactivepowers, the ac and dc voltages and currents, the temperatureand the supplied energy among others. The SCADA has thenecessary data to analyse the response of the microgrid to thevarious tests performed.

Fig. 14. Microgrid IEC 61850 SCADA

VI. EXPERIMENTAL RESULTS

On the basis of the previously described experimentalsystem, experimental test were performed. To develop thesetests, the microgrid units are configured to emulate a dis-patching load (managed by iSocket3), a battery (managed byiSocket2) and a micro wind turbine (managed by iSocket1).The battery is characterized by a Ni-Cd model described inTable II (to study changes in the battery SoC during testing,its capacity has been reduced to 1 A/h). The wind power curveimplemented in the wind turbine emulator is shown in TableIII. Control parameters of iSockets are in Table IV, resultingin the control equations of Table V.

The proposed control strategy is validated by experimen-tal tests in two different operational modes: centralized anddistributed.

1) Centralized operation mode: To test the microgrid re-sponse in centralized operation mode, two case studies areconsidered: In the first the energy price is constant and in thesecond it changes.

a) Case Study 1 constant energy price: Consideringconstant energy price of e = 10 ce/kWh, different active andreactive power references has been applied according to TableVI. The microgrid response is shown in Fig. 15. The batterystate of charge plot shows the evolution of the battery state ofcharge. The power plot, shows the active power evolution ofthe three units: wind turbine, battery and load. The active andreactive total power plot shows the whole microgrid active

TABLE IICONFIGURATION CHARACTERISTICS OF THE BATTERY EMULATOR

Characteristics SoC

Capacity = 25 A/h Cut voltage = 541,6 V 0%Nominal voltage = 650 V Discharge voltage = 650 V 10%

Charge voltage = 715 V 80%Overcharge voltage = 845 V 100%

Battery state of charge

650

700

750

800

850

dc Voltage

500

550

600

650

700

750

800

850

1 11 21 31 41 51 61 71 81 91

dc Voltage

Battery  State of Charge [%]

TABLE IIICONFIGURATION CHARACTERISTICS OF THE WIND TURBINE EMULATOR

Characteristics

Sweep area= 3.8m2 ρ = 1.2kg ·m2

Micro wind turbine available power

‐3500

‐3000

‐2500

‐2000

‐1500

‐1000

‐500

0

0 100 200 300 400 500 600 700 800 900 1000Time [s]

‐4500

‐4000

‐3500

‐3000

‐2500

‐2000

‐1500

‐1000

‐500

0

0 100 200 300 400 500 600 700 800 900 1000

Power [W]

Time [s]

TABLE IVMICROGRID TEST PARAMETERS

Parameters iSocket1 iSocket2 iSocket3Pi,m [W] -4000 -4000 0Pi,M [W] 0 4000 3000Qi,m [VAr] -3000 -3000 -3000Qi,M [VAr] 3000 3000 3000αPi0 40 -180 -100βPi0 80 60 -30αQi0 -100 -100 -100βQi0 100 100 100kp 0 6 2kE 0 - -kw - 1.2 -kr - - 0

and reactive powers. The control signals plot illustrates thecontrollers outputs Fp and Fq . Finally, the Eene plot the energyprice, which is constant during the 600 seconds analysed.

It can be seen in Power plot (Fig. 15) that the power curvesshape of the wind turbine and the battery are mirrored (thebattery compensates the wind power fluctuation) to ensurethat the overall power reference is tracked. In the Active andReactive total power plot (Fig. 15), it can be checked that thevalue corresponding to the active and reactive power, Ptot andQtot, is in accordance with the reference values, P ∗ and Q∗.The control signals sent from the iNode to the iSocket areshown in the control signals plot.

In the time interval t0 7→ t1 the active power reference is 0W, the battery SoC decreases slowly because the wind turbinecan almost supply the load and few power from the batteryis needed. In the interval t1 7→ t3, the microgrid reference isP ∗ =5000 (t1 7→ t2) and 3000 W (t2 7→ t3). The controller

8

Fig. 15. Centralized operation Microgrid response (Case study 1)

charges the battery to have the demanded P ∗ resulting inincreased battery SoC . As the SoC of the battery increases, itis more reluctant (α and β parameters change) to charge. Thus,to keep P ∗, the control signal Fp is increased and the powergenerated by the wind turbine is reduced. When the battery isfully charged, t3 7→ t4, the unique consumption is the load andto maintain the P ∗ set point, the wind turbine must disconnect.In the time interval t4 7→ t5 the microgrid generates 4000 Wby discharging the battery, reducing the load consumption andreconnecting the wind turbine. The battery SoC steady dropsto a 38%. At t4 7→ t5 there is an inductive reactive powerreference. Finally, in t5 7→ t6 there is a capacitive reactivepower reference.

b) Case Study 2 non-constant energy price: To analysethe effect of energy price change, Figure 16 and Table VIImust be considered. Table VII shows that the active powerreference is maintained at 1000 W during all the test, whilethe energy price is increased in t1 7→ t2. The energy priceincrease implies a new equilibrium that makes the controllerincrease Fp in order to maintain the active power reference.

A. Distributed operation mode

Fig.17 shows the system response with respect to thechanges on energy price, during a time interval of 200 seconds.It can be seen that the battery is sensible to the changesin the energy price. As it increases (t0 7→ t1 to t1 7→ t2),the microgrid reduces the consumption (P 07→1

tot = 908W toP 17→2tot = −886W) and the battery tends to discharge. If there

is a drop on energy price the microgrid consumes more powerfrom the grid and charges the battery (Table VIII).

VII. CONCLUSIONS

This paper has presented a control algorithm of a utility con-nected microgrid. The control scheme is based on independentcontrol of active and reactive power. The microgrid operationmodes include centralized and distributed operation. The mi-crogrid is controlled by a central iNode and several iSocketsfor each unit. The controllers to be implemented in each nodehave been discussed along with the required communicationsystems. The proposed scheme has been validated by means

9

TABLE VMICROGRID TEST FUNCTIONS

iSocket1 iSocket2 iSocket3

Graphical instruction definition

-100 100

4000

-4000

P1* [W]

Fp

βP1=80αP1=40

-100 100

4000

-4000

P2* [W]

FpβP2=60

αP2=-60-100 αP3=-80

3000 W

100

4000

-4000

P3* [W]

FpβP3=-10

αpi and βpi functions

{αp1 = 40

βp1 = 80

αp2 = −180 + 6e

+1.2Waβp2 = 60 + 6e

−1.2(100−Wa)

{αp3 = −100 + 2e

βp3 = −30 + 2e

αqi and βqi functions

{αq1 = −100

βq1 = 100

{αq2 = −100

βq2 = 100

{αq3 = −100

βq3 = 100

P ∗i =

0

−4000+ 4000βp1−αp1

(Fp−αp1)

−4000

4000

−4000+ 8000βp2−αp2

(Fp−αp2)

4000

3000

3000βp3−αp3

(Fp −αp3)

0

Q∗i =

3000

−3000+ 9000βq1−αq1

(Fq−αq1)

−3000

3000

−3000+ 9000βq2−αq2

(Fq−αq2)

−3000

3000

−3000+ 9000βq3−αq3

(Fq−αq3)

−3000

TABLE VICENTRALIZED OPERATION MICROGRID RESPONSE

Interval P ∗[W] Q∗[VAr]t0 7→ t1 0 0t1 7→ t2 5000 0t2 7→ t3 3000 0t3 7→ t4 3000 0t4 7→ t5 -4000 675t5 7→ t6 -4000 -350t6 7→ t7 -1000 0t7 7→ 1000 0

TABLE VIICENTRALIZED OPERATION MICROGRID RESPONSE UNDER A ENERGY

PRICE CHANGE

Interval P ∗[W] e[ce/kWh] Fp

t0 7→ t1 1000 10 [-10, 12]t1 7→ t2 1000 15 [24, 37]t2 7→ 1000 10 [-10, 0]

TABLE VIIIMICROGRID DISTRIBUTED OPERATION RESPONSE

Interval e [ce/kWh] Ptot [W] P2 [W]

t0 7→ t1 10 908 0t1 7→ t2 15 -886 -1650t2 7→ t3 5 2551 2330t3 7→ 20 -2844 -3580

of simulations and experimentally. The experimental microgridis composed of three configurable units: a generation unit, astorage unit and a load unit. These units are interfaced with

Fig. 16. Centralized operation microgrid response under a energy pricechange (Case study 2)

2000

4000

Power [W]P3: Load [W]

P2: Battery [W]

t1 t2 t3t0

214 52

266

‐4000

‐2000

0

2000

4000

0 50 100 150 200

Power [W]P3: Load [W]

P2: Battery [W]

P1: Wind turbine [W]

t1 t2 t3

2000

6000Total Power  Ptot: Metered 

real power [W]

t0

‐4000

‐2000

0

2000

4000

0 50 100 150 200

Power [W]P3: Load [W]

P2: Battery [W]

P1: Wind turbine [W]

t1 t2 t3

10

15

20Eene [c€/kWh]

Energy price

‐6000

‐2000

2000

6000

0 50 100 150 200

Total Power 

Time [s]

Ptot: Metered real power [W]

t0

203 40,87

243,87

1,27232689

‐4000

‐2000

0

2000

4000

0 50 100 150 200

Power [W]P3: Load [W]

P2: Battery [W]

P1: Wind turbine [W]

t1 t2 t3

0

5

10

15

20

0 50 100 150 200

Eene [c€/kWh]

Time [s]

Energy price

t3t2t1t0

‐6000

‐2000

2000

6000

0 50 100 150 200

Total Power 

Time [s]

Ptot: Metered real power [W]

t0

134,1 65,4

199,5

107 49,6

156,6

1,31854839

Fig. 17. Microgrid distributed operation response

the microgrid through a Voltage Source Converter (VSC) andare controlled by the nodes of the communication system bymeans of IEC 61850. Results have shown good performancefor different scenarios.

REFERENCES

[1] H. Farhangi, “The path of the smart grid,” IEEE Power and EnergyMagazine, vol. 8, no. 1, pp. 18 –28, january-february 2010.

[2] S. Karnouskos and T. de Holanda, “Simulation of a smart grid city withsoftware agents,” in Computer Modeling and Simulation, 2009. EMS’09. Third UKSim European Symposium on, 25-27 2009, pp. 424 –429.

[3] M. Hommelberg, C. Warmer, I. Kamphuis, J. Kok, and G. Schaeffer,“Distributed control concepts using multi-agent technology and auto-matic markets: An indispensable feature of smart power grids,” in PowerEngineering Society General Meeting, 2007. IEEE, 24-28 2007, pp. 1–7.

[4] G. Venkataramanan and C. Marnay, “A larger role for microgrids,” IEEEPower and Energy Magazine, vol. 6, no. 3, pp. 78 –82, may-june 2008.

[5] J. P. Lopes, S. A. Polenz, C. Moreira, and R. Cherkaoui,“Identification of control and management strategies for lvunbalanced microgrids with plugged-in electric vehicles,” ElectricPower Systems Research, vol. 80, no. 8, pp. 898 – 906, 2010. [On-line]. Available: http://www.sciencedirect.com/science/article/B6V30-4Y718XY-3/2/b2c39b33262d943a387e12a4cbaa1e7b

10

[6] U. Abdulwahid, J. Manwell, and J. Mcgowan, “Development of adynamic control communication system for hybrid power systems,” IETRenewable Power Generation, vol. 1, no. 1, pp. 70 –80, march 2007.

[7] R. Zamora and A. K. Srivastava, “Controls for microgrids with storage:Review, challenges, and research needs,” Renewable and SustainableEnergy Reviews, vol. 14, no. 7, pp. 2009 – 2018, 2010. [On-line]. Available: http://www.sciencedirect.com/science/article/B6VMY-4YR85T0-1/2/871c4a3d9819025f9e31166e0443eda5

[8] K. De Brabandere, K. Vanthournout, J. Driesen, G. Deconinck, andR. Belmans, “Control of microgrids,” in Proc. IEEE Power EngineeringSociety General Meeting, 2007, pp. 1–7.

[9] F. Katiraei, R. Iravani, N. Hatziargyriou, and A. Dimeas, “Microgridsmanagement,” IEEE Power and Energy Magazine, vol. 6, no. 3, pp.54–65, 2008.

[10] J. Lopes, C. Moreira, and A. Madureira, “Defining control strategies formicrogrids islanded operation,” IEEE Transactions on Power Systems,vol. 21, no. 2, pp. 916 – 924, may 2006.

[11] S.-J. Ahn, J.-W. Park, I.-Y. Chung, S.-I. Moon, S.-H. Kang, and S.-R. Nam, “Power-sharing method of multiple distributed generatorsconsidering control modes and configurations of a microgrid,” IEEETransactions on Power Delivery, vol. 25, no. 3, pp. 2007–2016, 2010.

[12] S. Grillo, S. Massucco, A. Morini, A. Pitto, and F. Silvestro, “Microtur-bine control modeling to investigate the effects of distributed generationin electric energy networks,” IEEE Systems Journal, vol. 4, no. 3, pp.303–312, 2010.

[13] J. Eto, R. Lasseter, B. Schenkman, J. Stevens, D. Klapp, H. Volkommer,E. Linton, H. Hurtado, and J. Roy, “Overview of the certs microgridlaboratory test bed,” in Proc. CIGRE/IEEE PES Joint Symp. Integrationof Wide-Scale Renewable Resources Into the Power Delivery System,2009, p. 1.

[14] B. Kroposki, R. Lasseter, T. Ise, S. Morozumi, S. Papatlianassiou, andN. Hatziargyriou, “Making microgrids work,” IEEE Power and EnergyMagazine, vol. 6, no. 3, pp. 40 –53, may-june 2008.

[15] M. Roman-Barri, I. Cairo, A. Sumper, and A. Sudria, “Experience on theimplementation of a microgrid project in barcelona,” June 2010, iEEEISGT Europe 2010, Gothenburg, Sweden, October 11-13, 2010.

[16] R. E. Mackiewicz, “Overview of iec 61850 and benefits,” in Proc. /2006IEEE PES Transmission and Distribution Conf. and Exhibition, 2006,pp. 376–383.

[17] H. Frank, S. Mesentean, and F. Kupzog, “Simplified application ofthe iec 61850 for distributed energy resources,” in ComputationalIntelligence, Communication Systems and Networks, 2009. CICSYN ’09.First International Conference on, 23-25 2009, pp. 172 –177.

[18] A. Ruiz-Alvarez, A. Colet-Subirachs, O. Gomis-Bellmunt, F. Alvarez-Cuevas-Figuerola, and A. Sudria-Andreu, “Design, management andcomissioning of a utility connected microgrid based on iec 61850,” June2010, iEEE ISGT Europe 2010, Gothenburg, Sweden, October 11-13,2010.

Alba Colet-Subirachs received the Industrial Engi-neering degree from the school of Industrial Engi-neering of Barcelona (ETSEIB), Technical Univer-sity of Catalonia (UPC), Barcelona, Spain, in 2009.She worked at the Centre of Technological Inno-vation in Static Converters and Drives (CITCEA-UPC) in a cooperation project in which she gotthe experience in programming DSP applied tothe power electronics converters. She is currentlyinvolved in Smart Electrical Grid projects at theCatalonia Institute for Energy Research (IREC). Her

research interests include renewable energy integration in power systems andpower electronics.

Albert Ruiz-Alvarez received the degree in in-dustrial engineering from the School of IndustrialEngineering of Barcelona (ETSEIB), UniversitatPolitecnica de Catalunya (UPC). He obtained dis-tinction in his Electrical Final Dissertation entitled:Design and implementation of a device for powertransformers monitoring, in December 2009. In May2009, he became a member of the Catalonia Institutefor Energy Research (IREC), in the Power Electron-ics and Electrical Power Grids research area, wherehe works as a Project Engineer. In IREC he has

carried out projects related to smart grids, IEC 61850 and renewable energies.

Oriol Gomis-Bellmunt (S’05-M’07) received thedegree in industrial engineering from the Schoolof Industrial Engineering of Barcelona (ETSEIB),Technical University of Catalonia (UPC), Barcelona,Spain, in 2001 and the PhD in electrical engineeringfrom the UPC in 2007. In 1999 he joined EngitrolS.L. where he worked as project engineer in theautomation and control industry. In 2003 he de-veloped part of his PhD thesis in the DLR (Ger-man Aerospace center) in Braunschweig (Germany).Since 2004 he is with the Electrical Engineering

Department of the UPC where he is lecturer and participates in the CITCEA-UPC research group. Since 2009 he is also with the Catalonia Institute forEnergy Research (IREC). His research interests include the fields linkedwith smart actuators, electrical machines, power electronics, renewable energyintegration in power systems, industrial automation and engineering education.

Felipe Alvarez-Cuevas Figuerola is responsiblefor Telecontrol projects of Endesa network Factory.He was recently strongly involved in the processof standardization of Telecontrol protocols in theEndesa Group. Within this project, he has beenworking since the last eight years in the automationof a few hundreds of hydro power plants and HVsubstations and a few thousands of MV substations.Currently he is responsible for integration of theEndesa?s Smart City Project in Malaga.

Antoni Sudria-Andreu (M’95,SM’05) receivedM.Sc.E.E and Ph.D. degrees from UniversitatPolitecnica de Catalunya (UPC), Barcelona, Spain,in 1979 and 2005. In 1979 he joined ElectronicsEngineering Department, UPC, as Assistant Pro-fessor. He joined the Power Electronics ResearchDepartment of a railway industry in 1980. In 1985he joined the Electrical Engineering Department asAssociate Professor. In 2001 he founded the Centerof Technological Innovation in Static Convertersand Drives (CITCEA-UPC). He is the head also of

Electrical Engineering research area of the Catalonia Institute for EnergyResearch (IREC) since January 2009.