42B33367d01

8/6/2019 42B33367d01

1/5

u k Z ,Damir NouoselX

s open transmission access is becoming a reality, a major concern ofelectric power utilities is to maintain the reliability of the grid. Increasedpower transfers raise concerns about steady-state overloads, increased

risks of voltage collapses, and potential stability problems. Strengthening theprotection and control strategies is what utilities must do to prevent a localproblem from spreading to other parts of the grid.

This article defir:es th e framework and motivation for development of a

multilayered protection and control scheme that starts with local measure-ment devices and integrates higher-level control schemes into an overallcontrol strategy.

Protection and Control ApproachesAlthough the complexity of the voltage collapse problem has beeG studied byvarious researchers, many questions still remain unanswered. The multitude ofapproaches have resulted in the implementation of a few protection and con-trol schemes against voltage collapse.

The symptoms often associated with collapse are low voltage profile, highconsumption, heavy transmission system loading, long distances between mostof the generation and loads, and insufficient reactivepower compensation facil-ities, The onset of voltage collapse can sometimes be precipitated by the activa-tion of limiters when some of the generators reach their reactive generationcapability limits and cannot maintain voltages with increasing demand.

Disturbances leading to potential voltage instability problems can be splitinto two categories:

Disturbances of topology, which may involve equipment outages, or faultsfollowed by equipment outagesLoad disturbances, representing the fluctuations of load which may havedynamics of their own.

They can be slow load changes (normal random load fluctuations) or fastload fluctuations (such as outages of large blocks of loads). A very typical sce-nario may involve a rapid load pickup (in some recorded cases several hundredMW per minute), corresponding deterioration of t he voltage profile in the net-

work, triggering of automatic protection and control events (such a s activationof limiters in generator excitation, tap blocking of transformers, load sheddingetc.), and final descent to a collapse, often accompanied by a cascaded actionof assorted protective relaying.

An example of voltage collapse simulated on a New England system model isshown in Figure 1. The random load is modeled as increasing in time linearly.The onse t of collapse is identified as a point whe re slopes of load voltages(shown here for 10 representative buses) suddenly change downwards, and thesystem loses its stable equilibrium.

I ABB Electric Systems Technology Institute Georgia Institute of Technology

40 IEEE Computer Applications in Power ISSN 08950156/97/$10.0001997EEE

Authorized licensed use limited to: Univ of Calif Santa Barbara Downloaded on February 24 2009 at 00:02 from IEEE Xplore Restrictions apply

8/6/2019 42B33367d01

2/5

Smart DevicesAn element of great importance in any mitigation scheme is to track the proxim-ity of the power system to a collapse. Knowing such proximity permits properutilization of control and protective equipment to steer the system from a cri-sis. At present, estimating the proximity to collapse in real-time still faces manydifficulties. Beside computational issues (e.g., speed, complexity, etc.), any cen-tralcontrol method is subject to the reliability of longdistance data communi-

cations. Although this problem is mitigated by the increasing use of redundantdata from microwave, optical fiber, and wireless systems and through sta teestimation methods, there is always concern when reliance must be placed onreceiving data from a remote location. To reduce or eliminate the need for thistransmiss ion of dat a, a local undervoltage relay (or any of its variations) is oftenused. The collapse is deemed imminent when the observed voltage falls belowa threshold. Selecting a proper value for the threshold, however, can be diffi-cult, because systems may experience a normal voltage when they are at thebrink of collapse.

Advances in microprocessor technology have made it affordable to improvethe computing capabilities of existing decentralized subsystems. Local devicescan now utilize advanced algorithms to make local decisions based on localmeasurements and possibly selected remote information. Utilities can startimproving the control and protectio n of their grids by enhancing devicesalready in use at substations. The enhanced devices form the line of defense atthe low level and offer the most advanced protection schemes that us e localinformation. In time, the communication links will gradually be built to integrateall local devices into a control network. This progressive strategy helps the util-ity to spread its investment over time.

Voltage Instability PredictionTracking how close a system operation is to a point of collapse has always beena challenging problem. Toward practical applications, the key element that dis-tinguishes one method from another is in regarding what information isrequired. Most methods in existence today require that system-wide informa-

tion be available. Fortunately, recent R&D efforts to make effective use of localmeasurements have resu lted in smart algorithms that can predict collapse from

local information. Implementation ofsuch smart algorithms is simpler, faster,

&fultilayeg-ed rotection and less costly when compared with theand control schemes Figure 2 shows a load bus and the

traditional approach.

start with L Q C ~ rest of the system treated as a Theveninequivalent. The following principle iswell known in electricci rcuit theory:

easurement deuicesand ntegrate higherlevel control schemes Maximal power t ransfer tf lZapp = I.&" I

where the apparent impedance Zcppsmerely the ratio between the voltage and current phasors measured at the bus.This maximal-power-transfer relation, holding true regardless of the load char-acteristic, separates the impedance plane into two regions, as shown in Figure 3. As the load varies, its Zappraces a path in the plane, and voltage instability

.- occurs, in the steady-state sense, when ZaDp rosses the Thevenin circle.Tracking parameters as the system app roaches voltage instability, th

fore, becomes the problem of tracking the distance of the present-time Z,the Thevenin circle. This is the principle behind the voltage instability pretor (VIP). The circle is not a fixed object, because it represents the rest ofsystem lumped together; such collection involves thousands of pieces of eq

ere-,p to!dit-the

uip-

7 41ctober 199

Authorized licensed use limited to: Univ of Calif Santa Barbara. Downloaded on February 24, 2009 at 00:02 from IEEE Xplore. Restrictions apply.

8/6/2019 42B33367d01

3/5

Figure I . Voltage collapse simul ated on a NewEngland system model

Figure 2. Local bus and the rest of the system treate das a Thevenin equivalen t

ment, any of which can change with time. Quite likelywhen the voltage is unstable, the circle expands(transmission becoming weaker) and the impedanceZu,, moves toward it (increasing load).

Tracking the Thevenin equivalent is essential fora VIP-based device to properly detect a voltage col-lapse . There a re many methods to t rack theThevenin parameters, and they can be based on therelation E =I/ + Z,,,, . I , where the unknowns are E(equivalent source in Figure 2) and Z,,,. In the idealcondition, two different measurement s et s (V and I)taken at two different times are sufficient to com-pute t he two unknowns. In th e real environment,however, measurements are not precise, and theThevenin parameters drift due to the systemschanging conditions. To suppress numerical oscil-lations in the estimated parameters, a larger datawindow must be used. Intuitively, this window mustcontain sufficient variations in th e measure ments.

Since load variations take place with a nonuniformrate, the time span of the window is not fixed. In fact,a faster variation in the data takes less time for aVIP-based device to produce an out put .

For illustration, a multinode power networkmodel is driven to maximal transfer by graduallyincreasing the load demand. The critical loading is163.4 percent of the base-case loading; beyond thisloading level, the power-flow equations admit nosolutions. A VIP is placed at each load bus and isunaware of t he changes tha t take place in the rest ofthe network (no data communications). Its inputsare the local measurements (bus voltage and load

current) and its output is a stream of Thevenin para-meters which vary with time. The plot in Figure 4 shows that the estimated Thevenin impedancemerges with the load impedance at the point of col-lapse. The load increase is evident by a decayingload-impedance profile.

VIP Interpretation and ApplicationsVIP can be viewed as an adaptive relay. Two differentinterpretations can be given. The first interpretationof VIP is direct from Figure 3: an impedance relaywith a self-tuned setting (the setting being the varying radius of th e red circle). The second interpreta-

tion is an adaptive voltage relay, and can be betterseen when the two curves in Figure 4 are multipliedby th e load current profile. The result is shown in Fig-ur e 5 ; the green curve is associated with the loadvoltage, and the red curve is associated with the volt-age drop across the Thevenin impedance. If on eviews th e red curve a s the voltage setpoint of th erelay, then the setpoint is tuned so that at the col-lapse, the load voltage is equal to t he setpoint. Recallthat the conventional voltage relay measures thevoltage only, and its decision is based on a fixed set-

42 IEEE Computer Applications in Power

I 1

Authorized licensed use limited to: Univ of Calif Santa Barbara. Downloaded on February 24, 2009 at 00:02 from IEEE Xplore. Restrictions apply.

8/6/2019 42B33367d01

4/5

Tracking the Theveninequivalent is essential for aVIP-based device to p roperly

detect a voltage collapse

point. The VIP, on the other hand, makes use of anadditional input (namely, electric current) to adaptits setpoint to the systems condition.

The algorithm can be coded and embedded inexisting microprocessor-based relays/controllers, orimplemented as an individual unit. The requiredinputs are local bus voltage and load current, whichare already available at substations for other applica-tions. The output of the VIP can be used to guidecontrol actions at the substation. For example, if thetotal load connected to the substation is found to beexcessive by the VIP (Zapp n Figure 3 is sufficientlyclose to the red circle), a partial shedding can beissued to maintain a sufficient margin. In anotherexample, tap-changing transformers are frequentlyused to regulate the voltage on the load side. Theiractions basically drain the reactive power from thesystem to support the voltage at their loads. A VIPthat processes measurements on the primary side ofa tap-changing transformer can detect when thedrain is excessive, and thus the decision to block thetapchanging action can be carr ied out.

Another application is to enhance t he perfor-mance of a static var controller (SVC) by adding volt-

age-collapse prediction. Traditionally, SVC behaviorcan mask the imminent collapse, leading to suddenand unexpected loss of reliable power supply. TheVIP can be incorporated to ensure accurate collapseprediction taking into account the SVC operation.

Integrated Control and ProtectionTodays communication and computer technologieshave created a new revolution in the power industry,especially in the field of power system control wherevertical integration is much improved. Communica-tion capability is one of t he potential benefits forcomputer relays, which communicate not only with a

center, but with each other. This in turn will facilitatethe overall system-wide protection and control phi-losophy.

The Self-Managing And Reliable Transmiss ionGrid (SMARTGrid) is seen a s the fu ture of pro tec-tion and control systems. It is an automated systemof monitoring, control, and protection devices thatimproves the reliability of the transmission grid bypreventing wide-spread break-ups. It utilizes infor-mation technology to improve grid reliability andcapability. It builds on existing components such as

Figure 3. Maximal power t ransf er s reached (voltageinstability) when the appar ent impedance of he loadlands on he Thevenin circle, Izapp= Izm,,

Figure 4. Thevenin impedance and load impedancemerge at the point of collapse

Figure 5. VIP can be viewed as a voltage re lay withan adaptive setpoint

October 199743

Authorized licensed use limited to: Univ of Calif Santa Barbara. Downloaded on February 24, 2009 at 00:02 from IEEE Xplore. Restrictions apply.

8/6/2019 42B33367d01

5/5

G f i eir gri

the local controllers together:Ontinu-

OuS1Y report their to the The t en -tral computer combines the rep orted proximities t ocollapse, performs extra calculations and issues coordi-

nating The central computer can override theload-shedding decision of individual devices.

Smart devices processing Only local measurementscan be counted upon when center-based emergency con-trois fail to mitigate an aggravating sit uation. Thes e

In a SMARTGridsmart devices such as

relays, sensors, meters, energy manage-men t sy s t ems , SCADA, SVC, e tc . , byadding better algorithms and utilizinginformation more effectively. The devicesare coordinated under several hierarchicallevels. Figure 6 shows a two-level design.

At th e subst ation level, each deviceprocesses local measurements (substationvoltage, load curren t, etc.), a ssess es th eproximity to system instability or thermalcapability, and carries out its own controldecisions. No long-distance communica-t ions a re requi red to implement theSMARTGrid concept at th e s ubs tati onlevel. At t he regional level, several regionalcomputersfrom substation-level devices and performthe coordinating role. The process is simi-

Figure 6. Smart devices such a s VIP-based devices are deployed to for mraw and processed data the low-level control, and communication links to regional centers and

h &her are needed for proper coordination

of Washington. He was a visiting faculty member at Clemson Universityduring 1991-1993, and is now with the Power Systems Center, ABB Elec-tric Systems Techno logy Institut e, in Raleigh, North Carolina. His areasof interest in power systems include stability, computer-based protec-tion and control, and power quality. He is an IEEE member.

Miroslav M. Begovic received his BSEE and MSEE from Belgrade

University, and his PhD in e lectrical engineering from Virginia Polytech-nic Institu te and S tate University. Since 1989, he has been a member ofthe faculty of the School of Electrical and Computer Engineering, Geor-gia Institute of Technology in Atlanta, Georgia, whe re he cur ren tlyholds the position of associate professor. His current interests are inthe general area of computer applications in power system monitoring,

-

lar at the global level. Long-distance communications(SCADA or dedicated links) are required for th e regionaland global levels.

There a re two major steps in realizing a SMARTGrid.The first ste p is to provide immediate solutions bymeans of enhanced products tha t will improve the sys-tem reliability and performance. For example, imple-

menting th e VIP method is done at the device level, anddoes not require communication links. This strategy isappealing to utilities, because they need not invest indeveloping new techniques, but can actually implementexisting technologies. Implementation of t hese technolo-gies form th e sub st at io n or local level of SMARTGridhierarchy. The second step is th e integration, whichrewires advanced technologies and methods to tie all

devices also formthe fall-&kprotection and control scheme when communicationschannels fail.

for any global

Estimate Voltage-Stability Margin, PICA 97: Proceedings o f the 20thIntemational Conference on Power Industry Computer Application, IEEE,May 1997.

C. W. Taylor, Power System Voltage Stability, McGraw-Hill, 1994.Proceedings o f Bulk Power System Voltage Phenomena-Ill: Voltage Sta-

bility, Security, and Control, Davos, Switzerland, August 1994.IEEE Power Systems Relaying Committee, Working Group K12,

Voltage Collapse Mitigation, 1995, available on th e World-Wide Web,http://www.rt66.com/-w5sr/psrc.htm l.

C. Barbier and J. Barret, An Analysis of Phenomena of Voltage Col-lapse on a Transmission System, Revue Genem.de de IElectricite, spe-cial CIGFE issue, pages 3-21, July 1980.

N. Grudunin and I. Roytelman, Heading Off Emergencies in LargeElectric Grids, EEE Spectrum, April 1997, pp. 4247.

protection and Control, and design and analysis of renewable energysources. He is a member of Sigma Xi, Tau Beta Pi, Eta Kappa Nu, and

Damir Novosel is manager of the Power System Center at ABB Elec-Phi Kappa Phi, and is an IEEESenior Member.

BiographiesKhoi Vu received his BSEE, MSEE, and PhD degrees from the University

I tric Systems Technology Institute. He received his PhD from Mississip-pi State University, where he was a Fulbright scholar, in 1991. H isresearch area is computer-based protection and control of power sys-terns. He is a member of Eta Kappa Nu and an IEEE Senior Member.

For Further ingK. VU, M. Begovic, D. Novosel, M. Saha, Use of Local Measuremen ts to

44 IEEE Computer Applications in Power

A h i d li d li i d U i f C lif S B b D l d d F b 24 2009 00 02 f IEEE X l R i i l