THE COLOMBIAN 2007 BLACKOUT

THE COLOMBIAN 2007 BLACK OUTc. A. Ruiz, N. J. Orrego and J. F. Gutierrez

Abstract--On April 26th, 2007, at 9:58 a.m., a fault on thecolombian power system caused by maintenance in the 230 kVTorca Bus gave origin to black out.

The consequent loss of those important branches causedcascading outages of tie lines that turned out in cascade eventsthat concluded in an electrical separation with Ecuador, causingmany economical losses for the country. The Colombian powersystem frequency suffered a mayor imbalance that leaded to adisoperation of the automatic load shedding system and thisresulted out in the Colombian blackout.

This paper provides a short description of the power systemanalysis operation and of the main events that triggered theblackout, identifying some of the main causes that could lead tothose events.

Index Terms-- Blackouts, power system security, powersystem stability, power system restoration.

I. INTRODUCTION

WHEN cascading outages happen the lights tum out andeverything stops. With the facility down and in the

dark, there is nothing to do but to sit and wait until the utilitycompany finds out the problem and solves it. This processusually takes only few minutes but sometimes it takes hoursand to complete system restoration, it can even take days.

Blackouts are by sure the most troublesome problem that autility company will have to deal with. Statistics show thatpower failures are, in general, a strange situation in mostcountries. Blackouts are also short in duration. Studies haveshown that 50 percent of blackouts last to 6 s. And 35 percentare less than 11 min long. These failure times are not a causeto concern for most commercial users, but for those industrieswhere there are computer-based operations, transportationcontrol systems, medical facilities, and communications sitesthey can be a problem.

When service continuity is critical, redundancy must becarried throughout the system. It means that some sites shouldnever depend upon one critical path for ac power. Forexample, if the facility is fed by a single step-downtransformer, a lightning flash or other catastrophic event couldresult in a transformer failure that would bring down theelectric supply.

A blackout is the total failure of the electric supply in an

This work was supported by the Universidad Nacional de Colombia andXM Los Expertos en Mercados an ISA Group enterprise.

Cesar Augusto Ruiz is from Universidad Nacional de Colombia sedeManizales (e-mail: [email protected]).

Jorge Fernando Gutierrez is from Universidad Nacional de Colombia,Serle Manizales(e-mail: [email protected])

Nolasco de Jesus Orrego is from XM - ISA, Medellin, (e-mail:[email protected])

978-1-4244-2218-0/08/$25.00 ©2008 IEEE.

area caused by the failure of some power device for example:a line short circuit or the outage of a substation braker.

In Colombia, it could be as a result of a terrorist attack tothe energy infrastructure, but in other countries, could becaused by investment delays as a result of a tight powersystem regulation that leads to devices obsolescence.

A blackout can be produced by load increments ordecrements beyond a safe level, if the protection system failsin shedding load or disconnecting generators in order to keepthe load flow in a secure value.

A blackout can be prevented if the vulnerability of thesystem is determined in the planning stages of the operation inorder to establish the actions that must be followed to assurethe safe operation of the system.

CND-XM, the Colombian system operator issues every daythe colombian generation dispatch considering the systemreliability using the VERPC (Valor Esperado deRacionamiento Programado Condicionado - Expected ValueOf Conditioned Loss Of Load) in order to reduce lost of loadsupply and providing energy in an economical way.

II. ANTECEDENTS

There are about 82 main blackouts registered in countriesas: United States, France, Canada, Brazil, Portugal, Chile,Mexico, Italy, Sweden and Colombia among others. In theFig. 1 it is presented the incidence of blackouts from 1966 topresent time. It can be seen that from 2003 the occurrence ofthose contingencies is been drastically increasing. Notice thatin this year it is been registered 8 events by may [2-10].

18 bleck outs In the world 17

16

14 I12 J

la 12

10 15

I 88

6::I•

Fig. 1: Blackouts by year

Some of the effects (duration and people affected) of the mainevents are presented in Table I.

Authorized licensed use limited to: Universidad Nacional de Colombia. Downloaded on March 13,2010 at 11:46:14 EST from IEEE Xplore. Restrictions apply.

TABLE I.CIVILIANS AFFECTED BY BLACKOUTS

time civiliansyear country

(hours) (millions)1989 CANADA 9 61999 BRASIL 9 902003 USA-CANADA 9 502003 CANADA 5 4,32003 ITALY 24 562005 JAVA ISLAND 7 1002007 COSTA RICA 1,5 52007 COLOMBIA 4,5 382008 CANADA 3 0,12008 AUSTRALIA 72 0,42

In Fig. 2, it is presented the percentage of blackouts bycountry. It can be seen that developed countries are taking thebigger share of the main blackouts.

BLACK OUTS PERCENTAGE

Fig. 2: Blackouts density by country

A list of the most important events, by their socio ­economic impacts should include the following:

1. USA Northeast blackout on November, 1965.2. USA New York blackout on July 13, 1977. During

this event there were disturbs and riots registeredalong the city.

3. Canada - Hydro Quebec on March 13, 1989. It wascaused by GMD (Geomagnetic Disturbances) andwas one of the largest ever recorded. The entireQuebec Interconnection was blacked out as a resultof this disturb 6 million people were affected duringabout 9 hours.

4. USA California on 2000 - 2001 experienced severalpower shortages as a result of its electricityderegulation process.

5. U.S.-Canada blackout on August 14,2003. This eventaffected 50 millions of people in 8 American statesand 2 Canadian provinces and 63 GW of load wascurtailed [7].

6. Italy on September 28th, 2003. As a result of twoSwiss power lines overloaded a loss of synchronismbetween the Italian system and the UCTE power

2

system that affected the whole country for about 1,5hours [4].

7. Colombia on April 26, 2007. This event affectedalmost the whole country and will be detailed in thenext section.

8. Argentina on May 17th, 2007. A fire in a Transenersubstation caused a blackout that lasted for 3 hoursaffecting a mayor area in Buenos Aires.

9. Spain Barcelona on July 23,2007. The whole city wasaffected and in some parts of it the blackout lastedfor 78 hours.

III. THE COLOMBIAN SCENARIO BEFORE THE BLACKOUT

The system operator of the Colombian power system(Centro Nacional de Despacho - CND-XM) solves everydaythe energy dispatch process observing the market regulatorreliability guidelines. In this process it is included a blackoutrisk equalizing policy called VERPC. CND must find asolution of the dispatch problem so that the expected relativeload shedding in the different sub-areas of the interconnectedsystem (SIN - Sistema Interconectado Nacional) are equalizedamong the different sub-areas and minimized. This criteria isequivalent to the worldwide well-known N-l criteria [13].

On March 15th, 2007 ISA the main transmission system

operator in Colombia, announce about some maintenanceworks on Torca substation to CND. Those works will be doneon line with the substation equipment energized. Thatsubstation is double bus and a sectioned transfer bus. Thatmaintenance was approved to be done on April 26th

• On April25th the system operator, solved the day-ahead energy dispatchconsidering the risk of failure on the 230 kV Torca - Guaviopower line during the maintenance, and approved thatmaintenance to be done by 9:20 AM.

On Fig. 3, can be seen the normal state before themaintenance. The power line flows were below the limits andthe frequency and voltages were between their normal values.

Fig. 3. SIN state before contingency

On Fig. 4, can be seen the transfer limits between areas(blue) and the real values before contingency (white). At thatmoment the system was generating 7083 MW and there were

Authorized licensed use limited to: Universidad Nacional de Colombia. Downloaded on March 13,2010 at 11:46:14 EST from IEEE Xplore. Restrictions apply.

3

Fig. 5 Guavio-Circo transmission line outage.

In Fig. 7, is presented the frequency evolution during thecontingency. Note the frequency deviations between thesystem islands. The oriental area were experience overfrequency and for rest of the country the lost of generationcaused a low frequency.

After that, Colombia and Ecuador were insulated by theopening of Jamondino-Pomasqui circuits. Finally, other plantsof the Colombian system were shut down in the sequencedescribed below:

Calderas Generator 1, at 9:58.42.885

Fig. 7 Frequency response during the contingency

~f~·..---~•.,.~--~---~--~...L-_-~" " m 1_'_'-- ..

tr,,'IQ'::-CA.·...'..:-~t • ~·.. 1lQ!-'""....:~~ ,. ..T....-;~T ......:.:hlO._I._'.-..............,.Ull"'"'*_~.·ot:

In spite of that situation, Guavio and Chivor, the biggestgeneration plants in that area were on line, but thetransmission corridor that carries out the energy was overloadso frequency and bus voltages were decreasing beyond thesafe limits causing this area insulation from the rest of thesystem. That caused that thermo plants of Paipa and Yopalwere shut down. At that moment there was a lost of generationclose to 2100 MW that corresponds to 31% of the load (6644MW). That causes separation of the network into islands andthe disconnection of:

• Guavio-Tunal transmission line• Tunal- Reforma transmission line

In Fig. 6 can be seen that the disconnection of those linescauses the islanding of Boyaca and Santander regions.Disparo de Guavio - Tunal/Reforma Boyee' y santanderes

. ~ .a:".,~".~~~

Fig. 4 Pre-contingency power flows between areas

IV. EVENT DEVELOPMENT

At 09:58 during the maintenance on one the Torca bus 1disconnect switch L141 an operation of the M240 circuitbreaker that couples the substation buses occurred. Thatoperation caused the desenergization of the entire substationand the circuit opening of Chivor 1 y 2, Bacata 1 y 2, Guavio1 y 2 and 230/115 kV Torca transformers 2, 3 y 4. Those linesand transformers are the power supply to Bogota [11].

A cascading event result as a consequence of Torcaoutage. First, Guavio-Circo power circuits were opened as canbe seen on Fig. 5. That outage caused low voltage levels insome bars (blue), overloads flow on 220kV Villavicencio ­Tunal transmission line (red) and the outage of Termo Zipagen~ration plant [11].

not any international power transfers with Ecuador. Thispower interchange with Ecuador it is been done using theJamondino - Pomasqui tie line and the political agreementbetween the countries is called TIES.

Authorized licensed use limited to: Universidad Nacional de Colombia. Downloaded on March 13,2010 at 11:46:14 EST from IEEE Xplore. Restrictions apply.

Calderas Generator 2, at 9:58.43.130Guaca Generator 3, at 9:58.43.000San Carlos Generator 8, at 9:58.43.940San Carlos Generator 6, at 9:58.44.240San Carlos Generator 5, at 9:58.44.290Tennocentro Generator 1, at 9:58.51.708Tennocentro Generator 2, at 9:58.51.951Playas Generator 1, at 9:58.50.567Playas Generator 3, at 9:58.59.393Playas Generator 2, at 9:58.50.890

That lost of generation caused the frequency responsepresented in Fig. 8.

lOi...--------..".......--------------,

.o+---------------~....·_-·--·----·------··-·;

Iss

J••

!ltoi----------"",...,.--------~lJIIr_----l

tlU+-------~IloOoI.JO\UUi..--------~---I

t60+-------...---.....--_...------,---__-~o

Fig. 8Frequency response to generation shut down

That lack of generation causes an active power imbalance thatcould not be corrected by the load shedding systemimplemented in Colombia called (EDAC - EsquemaAutomatico de Desconexi6n de Carga) because of the extremesituation presented, so the system totally collapsed.

V. SYSTEM RESTORATION

The system restoration was planned by XM whocoordinated the actions between 40 transmission controlcenters, generation companies and regional transmissionoperators.

The strategy relied in those plants with black star capacityand the tie-lines operation with Ecuador and Venezuela. Postcontingency guides developed previously were used in thesystem operation restoration in every area. Those guides aredeveloped by XM and the system agents and are continuallyactualized.

After 4, 5 hours the system was totally reestablished tononnal. The restoration process duration can be consideredappropriate in comparison with others, for example:

• 07/12/2006 Auckland, New Zeeland took 10 hours.• 08/22/2005 Iraq took 7 hours• 11/25/2005 Musterland, took six days because of the

heavy weather conditions.But those 4,5 hours of inactivity had a cost of nearly US$

130 millions to the country.

4

VI. CONCLUTIONS

The Colombian power system operator XM, after analyzingthis event, issued a document with some conclusions about thesystem behavior during this contingency and its restorationprocess [11]. Some of those conclusions are presented below:

• The load shedding system EDAC system and theprimary and secondary generation reserves hadoperated as they were intended to do.

• It is necessary to install a contingency alann device inorder to improve time responses during contingencies.

• It is necessary to reinforce communication channelsbetween the regional transmission operators and themain system operator increasing its redundancy anddeveloping new communication protocols.

• It is important to analyze the system vulnerability tothe substation failure in order to know which of themare critical

VII. ACKNOWLEDGMENT

The authors gratefully acknowledge the contributions ofCorredor P., Murcia A., et Al and XM for their work on theoriginal version of this document.

VIII. REFERENCES

[1] Corredor P., Murcia A., et AI. "Evento del 26 de Abril de 2007 en el SINColombiano", 2007

[2] US-Canada Power System Outage Task Forces. Final Report on theAugust 14, 2003, Black-Out in the US and Canada: causes andrecommendations. April 2004

[3] Taylor C. Preventing Black Outs. IEEE/PES 2004[4] Berizzi A. The Italian 2003 blackout, IEEE PES General Meeting,

Denver, CO, June 6-12, 2004.[5] "Report on the blackout in Italy on 28 September 2003," available at

http://www.energie-schweiz.ch. November 2003.[6] Dagle 1. E. Data Management Issues Associated with the August 14,

2003 Blackout Investigation, IEEE, IEEE PES General Meeting,Denver, CO, June 6-12, 2004.

[7] Hauer J. F., Bhatt N, Shah K. et AI. "Performance of 'WAMS East' inProviding Dynamic Information for the North East Blackout of August14,2003". IEEE PES General Meeting, Denver, CO, June 6-12, 2004.

[8] Corsi S. and Sabelli C. General Blackout in Italy Sunday September 28,2003. IEEE PES General Meeting, Denver, CO, June 6-12, 2004.

[9] Gomes P. New Strategies to Improve Bulk Power System Security:Lessons Learned From Large Blackouts;. IEEE PES General Meeting,Denver, CO, June 6-12, 2004.

[10] Larsson S. and Ek Svenska E. The Black-out in Southern Sweden andEastern Denmark. IEEE PES General Meeting, Denver, CO, June 6-12,2004.

[11] ISA XM. Evaluacion del Esquema de Desconexion Automatica de Cargapor Baja Frecuencia - EDAC- 2007 Gerencia Centro Nacional deDespacho Direccion Planeacion de la Operacion XM Mayo 28, 2007

[12] NERC. August 14, 2003 Blackout: NERC Actions to Prevent andMitigate the Impacts of Future Cascading Blackouts February 10, 2004

[13] Lopez L 1. M., Murillo S C. E. Gutierrez G J. F. AContingency-BasedSecurity-Constrained Optimal Power Flow Model for Revealing TheMarginal Cost of a Blackout Risk-Equalizing Policy in the ColombianElectricity Market. 2006 IEEE PES Transmission And DistributionConference and Exposition Latin America. 2006.

[14] ELECTRICITY RESTRUCTURING 2003 Blackout Identifies Crisisand Opportunity for the Electricity Sector Report to the Chairman,Senate Committee on Governmental Affairs United States GeneralAccounting Office GAO November 2003

[15] Imai S. TEPCO's Observations on August 14 Blackout andRecommendations to Prevent Future Blackouts based on TEPCO'sExperience. IEEE PES General Meeting, Denver, CO, June 6-12, 2004.

Authorized licensed use limited to: Universidad Nacional de Colombia. Downloaded on March 13,2010 at 11:46:14 EST from IEEE Xplore. Restrictions apply.

[16] Van Hulst N. Lessons Learned from the Power Outage in North Americaand Europe Director, Long-Term Co-operation and Policy AnalysisRIETI Policy Symposium Tokyo, 15 December 2004

[17] Illic M. Granger M. et AI. Electrical Blackouts: A Systemic Problem.[18] Mandozzi M., Menditto V., Moreschini G. et AI. "Recent improvements

of emergency automatic control of Enel power system in interconnectedand isolated operation", CIGRE Session, paper, Paris, 1992.

[19] Veloza O. P. and Cespedes R.H. "Vulnerability of the ColombianElectric System to Blackouts and Possible Remedial Actions", IEEE PW2006.

IX. BIOGRAPHIES

Cesar Augusto Ruiz Beltran: He received a degree inelectrical engineering from the Universidad Nacional deColombia - Manizales in 2008. He was a passant student at XMS.A E.S.P, owned by Interconexi6n Electrica S.A (ISA).Currently, his working as an engineer at CODENSA S.A. ESP.

Nolasco De Jesus Orrego: studied electrical engineering at theUniversidad de antioquia, then he studied at the University of Wisconsin­Madison, where he obtained a Master of Science in electrical power systems,he is currently working at XM S.A E.S.P, owned by Interconexi6n ElectricaS.A (ISA).

Jorge Fernando Gutierrez Gomez: studied electrical engineering at theUniversidad Industrial de Santander, where he also obtained a Master ofScience in electrical power systems; he is currently a full time professor at theUniversidad Nacional de Colombia Sede Manizales. His main areas ofinterests include power systems operation and control, transmission lines andenergy markets.

5

Authorized licensed use limited to: Universidad Nacional de Colombia. Downloaded on March 13,2010 at 11:46:14 EST from IEEE Xplore. Restrictions apply.