Abstract This paper provides an overview of the

MicroGrid paradigm. This includes the basic architecture,control and protection and energy management.

Index Terms MicroGrid, microsources, power electronics,voltage source inverter, microturbines, protection.

I. INTRODUCTION

The MicroGrid concept assumes a cluster of loads andmicrosources operating as a single controllable system thatprovides both power and heat to its local area. This conceptprovides a new paradigm for defining the operation ofdistributed generation. To the utility the MicroGrid can bethought of as a controlled cell of the power system. Forexample this cell could be controlled as a single dispatchableload, which can respond in seconds to meet the needs of thetransmission system. To the customer the MicroGrid can bedesigned to meet their special needs; such as, enhance localreliability, reduce feeder losses, support local voltages,provide increased efficiency through use waste heat, voltagesag correction or provide uninterruptible power supplyfunctions to name a few.

II. BASIC STRUCTURE

The microsources of special interest for MicroGrids are small(<100-kW) units with power electronic interfaces. Thesesources, (typically microturbines, PV panels, and fuel cells)are placed at customers sites. They are low cost, low voltageand have high reliable with few emissions. Power electronicsprovide the control and flexibility required by the MicroGridconcept. Correctly designed power electronics and controlsinsure that the MicroGrid can meet its customers as well asthe utilities needs. The above characteristics can be achievedusing a system architecture with three critical components;

¥ Local microsource controllers¥ System optimizer¥ Distributed protection

Figure 1 illiterates the basic MicroGrid architecture. In thisexample the electrical system is assumed to be radial withthree feeders A, B and C and a collection of loads. The radialsystem is connected to the distribution system through a

The work described in this report was coordinated by the Consortium forElectric Reliability Technology Solutions, and funded by the AssistantSecretary of Energy Efficiency and Renewable Energy, Office of PowerTechnologies of the U.S. Department of Energy under Contract No. DE-AC03-76SF00098 and by the California Energy Commission, Public InterestEnergy Research Program, under Contract No. BG 99-396 (00)

Robert H. Lasseter is Professor at University of Wisconsin-Madison, 1415Engineering Dr., Madison, WI 53706 (lasseter@engr.wisc.edu).

separation device, usually a static switch. The feeder voltagesat the loads are usually 480 volts or less. Feeder A indicatesthe presents of several microsources with one providing bothpower and heat. Each feeder has circuits breakers and powerflow controllers. Consider the power flow controller near theheat load in feeder A. This controller regulates feeder powerflow at a level prescribed by the Energy Manger. As loadsdown stream change the local microsources increase ordecreases their power output to hold the power flow constant.In this figure feeders A and C are assumed to have criticalloads and include microsources, while feed B is assumed tohave non-critical loads which can be shed when necessary. Forexample when there are power quality problems on thedistribution system the MicroGrid can island by using theseparation device shown in the figure. The non-critical feedercan also be dropped using the breaker at B.

Figure. 1. MicroGrid Architecture

III. MICROSOURCE CONTROLLER

Microsource controller is an important component of theMicroGrid infrastructure. This controller responds inmilliseconds and uses local information to control themicrosource during all events. A key element is thatcommunications among microsources are unnecessary for basicoperation. Each inverter is able to respond to load changes in apredetermined manner without communication of data fromother sources or locations, which enables plug and playcapabilities. Plug and play implies that a microsource can beadded to the MicroGrid without changes to the control andprotection of units that are already part of the system. Thebasic inputs to this controller are steady state set points foroutput power, P, and local bus voltage, V.

MicroGridsR.H.Lasseter, Fellow, IEEE

3050-7803-7322-7/02/$17.00 © 2002 IEEE

Basic Control of Real and Reactive PowerThere are two basic classes of microsources; one is a D.C.source, such as fuel cells, photovoltaic cells, and batterystorage, the other is a high frequency ac source such as themicroturbine, which needs to be rectified. In both cases theresulting D.C. voltage is converted to an acceptable ac sourceusing a voltage source inverter. The general model for amicrosource is shown in Figure 2.0. It contains three basicelements; prime mover, dc interface and a voltage sourcedinverter. Coupling to the power system is through aninductor. The voltage source inverter provides control of boththe magnitude and phase of its output voltage, V. The vectorrelationship between the inverter voltage, V, and the systemvoltage, E, along with the inductor s reactance, X, determinesthe flow of real and reactive power (P &Q) from themicrosource to the system.

X

EV+

PrimeMover Inverter

DCInterface

Figure 2.0 Interface Inverter System

The P & Q magnitudes are coupled as shown in the equationsbelow. For small changes P is predominantly dependent onthe power angle, _p, while Q is dependent on the magnitude ofthe converter’s voltage, V. These provide for a basic feedbackloops for the control of output power and bus voltage, Ethrough regulation of reactive power flow.

PVE

X p

QV

XV E p

p V E

=

= −

= −

3

2

3

2

sin

( cos )

δ

δ

δ δ δ

Voltage regulation through droopIntegration of large numbers of microsources, implied in theMicroGrid concept, is not possible with basic P-Q controls.Voltage regulation is necessary for local reliability andstability. Without local voltage control, systems with highpenetration of microsources can experience voltage and/orreactive power oscillations. Voltage control requires care toinsure that there are not large circulating reactive currentsbetween sources. The issues are identical to those encounteredin the control of large synchronous generators. In the powergrid the impedance between generators is usually large enoughto greatly reducing the possibility of circulating currents. In aMicroGrid, which is typically radial, the problem of largecirculating reactive currents is immense. With small errors involtage set points the circulating current can exceed the ratingsof the microsources. This situation requires a voltage vs.

reactive current droop controller. Basically, as the reactivecurrent generated by the microsource becomes more capacitivethe local voltage set point is reduced. Conversely as thecurrent becomes more inductive the voltage set point isincreased. The function of the basic controller is shown inFigure 3.0. The Q limit shown in the figure is a function ofthe volts-ampere (VA) rating of the inverter and the powerbeing provided by the prime mover.

QinductiveQcapactive

Vset point

QmaxQmax

Q2max= VA2- P2

Figure 3.0. Voltage set point with droop

Fast load tracking and the need for storageA system with clusters of microsources and storage could bedesigned to operate both isolated and connected to the powergrid. In isolated operation, load-tracking problems arise sincemicroturbines and fuel cells have slow response and areinertia-less. The prime movers output power time constantsrange from 10 to 200 seconds, which is too slow for mostloads. It must be remembered that the current power systemshave storage provided though the generators inertia. When anew load comes on line the initial energy balance is satisfiedby the system s inertia. This results in a slight reduction insystem frequency. A system with clusters of microsourcesdesigned to operate in an island mode must provide someform of storage to insure initial energy balance.

The necessary MicroGrid storage can come is several forms;batteries or supercapacitors on the dc bus for each microsource;direct connection of ac storage devices (batteries, flywheelsetc.); or use of traditional generation with inertia with themicrosources. If the MicroGrid is not required to operate inisland mode the energy unbalance can be met by the ac systemwithout providing storage on the MicroGrid.

Frequency droop for power sharingMicroGrids provide premium power through the ability tosmoothly move from dispatched power mode (while connectedto the utility grid) to load tracking (while in island mode). Inthe island mode such problems as slight errors in frequenciesgeneration at each converter and the need to change power-operating points to match load changes imply a need for acomplex communication system. This is not so. These issuescan be addressed using power vs. frequency droop functions ateach microsource without an explicit communication network.

When grid connected, the loads in the MicroGrid receivepower both from the grid and from the microsourcesdepending on the customer s situation. With loss of the grid

306

due to voltage drops, faults, blackouts etc. the MicroGridsmoothly transfer to island operation. With separation fromthe grid the voltage phase angles at each microsource in theMicroGrid change resulting in an apparent reduction in localfrequency. This frequency reduction coupled with a powerincrease allows for each microsource to provide it sproportional share of load without new power dispatch fromthe Energy Manager. In fact in island operation the EnergyManager is not used except for reconnection to the grid.

Consider two microsources, as suggested in Figure 4.0. Inthis example, the sources are assumed to have differentratings, P1max, and P2max. The dispatched power in grid mode( P01 and P02 ) is defined at base frequency, ω0 . The droop isdefined to insure that both systems are at rated power at thesame minimum frequency.

P

ω

ωo

wmin

w1

P1max

P01P02

P2max

Figure 4.0. Power vs. Frequency Droop Control

During a change in power demand, these two sources operateat different frequencies, which causes a change in the relativepower angles between them. As this happens, the twofrequencies tend to drift towards a lower, single value ω1 .Unit 2 was initially operating at a lower power level than Unit1. However, at the new power level Unit 2 has increased itsshare of the total power needs. Although power is adjustedwithin fractions of seconds, frequency restoration can takelonger. Because droop regulation decreases the MicroGrid, arestoration function must be included in each controller. Thedroop control design is based on each microsource having amaximum power rating. As a consequence, droop is dependenton the dispatched power level while the microsources areconnected to the grid.

MicroGrids can provide premium power through the ability tosmoothly move from dispatched power mode (while connectedto the utility grid) to load tracking (while in island mode). Inthe island mode such problems as slight errors in frequenciesgeneration at each converter and the need to change power-operating points to match load changes imply a need for acomplex communication system. This is not so. These issuescan be addressed using power vs. frequency droop functions ateach microsource without an explicit communication network.

When grid connected, the loads in the MicroGrid receivepower both from the grid and from the microsourcesdepending on the customer s situation. With loss of the griddue to voltage drops, faults, blackouts etc. the MicroGrid

smoothly transfer to island operation. With separation fromthe grid the voltage phase angles at each microsource in theMicroGrid change resulting in an apparent reduction in localfrequency. This frequency reduction coupled with a powerincrease allows for each microsource to provide it sproportional share of load without new power dispatch fromthe Energy Manager. In fact in island operation the EnergyManager is not used except for reconnection to the grid.

IV. SYSTEM OPTIMIZATION

System optimization is provided by the EnergyManager. The Energy Manager uses information onlocal electrical and heat needs, power qualityrequirements, electricity and gas costs,wholesale/retail service needs, special grid needs,demand-side management requests, congestion levels,etc. to determine the amount of power that theMicroGrid should draw from the distribution system.Some key functions of the Energy Manager are;

¥ Provide the individual power and voltage set point foreach power flow/microsource controller

¥ Insure that heat and electrical loads are met¥ Insure that the MicroGrid satisfies operational contractswith the transmission system

¥ Minimizes emissions and system losses¥ Maximize the operational efficiency of themicrosources.

¥ Provides logic and control for islanding andreconnecting the MicroGrid during events.

V. PROTECTION

Protection must respond to both system and MicroGrid faults.If the fault is on the utility grid, the desired response may beto isolate the MicroGrid from the main utility as rapidly asnecessary to protect the MicroGrid loads. The speed ofisolation is dependent on the specific customer s loads on theMicroGrid. In some cases sag compensation can be usedwithout separation from the distribution system to protectcritical loads. If the fault is within the MicroGrid, theprotection coordinator isolates the smallest possible section ofthe radial feeder to eliminate the fault. Most conventionaldistribution protection is based on short-circuit currentsensing. Power electronic based microsources can notnormally provide the levels of short circuit required.Microsources may only be capable of supplying twice loadcurrent or less to a fault. Some overcurrent sensing deviceswill not even respond to this level of overcurrent, and thosethat do respond will take many seconds to respond, rather thanthe fraction of a second that is required.

The unique nature of the MicroGrid design and operationrequires a fresh look into the fundamentals of relaying. Oneapproach that is quite powerful is to develop a real-time faultlocation technique that will identify the exact location of thefault much more accurately that the classical relaying iscapable of doing under any circumstances. These methods mayprove to costly. Low cost approach such a CT based zerosequence detection, and differential current and/or voltagemethods also show promise. These means are not in common

307

use on distribution systems but can provide the requiredfunctions. The nuances of applying these techniques todistribution systems in a variety of different configurations arenot as well understood as overcurrent sensing.

Robert H. Lasseter (F’92) received the Ph.D. in Physics atthe University of Pennsylvania, Philadelphia in 1971. He wasa Consultant Engineer at General Electric Co. until he joinedthe University of Wisconsin-Madison in 1980. Researchinterest focus on the application of power electronics toutility systems and technical issues which arise from therestructuring of the power utility system. This work includesinterfacing micro-turbines and fuel cells to the distributiongrid, MicroGrids, control of power systems through FACTScontrollers, use of power electronics in distribution systems,harmonic interactions, simulation methods, power electroniccircuits.. Professor Lasseter is a Fellow of IEEE, an expertadvisor to CIGRE SC14 and Chairman of IEEE WorkingGroup on Distributed Generation

308