5
1 A Knowledge-Based System for Supervision and Control of Regional Voltage Profile and Security Alessandro B. Marques Glauco N. Taranto Djalma M. Falcão Petrobras Federal University of Rio de Janeiro Formely with FURNAS COPPE Brazil Brazil 2 Previous Reports TF 38.02.23 Report Draft Porto PowerTech, 2001 A.B. Marques, G.N. Taranto, and D. M. Falcão, A Supervisory Knowledge-Based System For Monitoring and Control of Regional Voltage Profile.IEEE Transactions on Power Systems A.B. Marques, G.N. Taranto, and D. M. Falcão, A Knowledge- Based System for Supervision and Control of Regional Voltage Profile and Security.(Submitted) TF 38.02.20 Final Report (2nd Draft) 3 Cooperative Work Federal University of Rio de Janeiro – COPPE CEPEL ONS: The Brazilian ISO Objectives: Analyze coordinated voltage schemes for possible implementation in the Brazilian System Survey implementation issues and difficulties 4 Motivation Characteristics of the Q-V Interaction Highly non-linear Discrete elements Customized Control Scheme Radial/meshed networks Short/long lines Availability of reactive support in the area 5 Knowledge Based Systems Most KBS uses production rules like If <condition>, Then <action or conclusion> Uncertainty may exists in the rules Linguistic Uncertainty (low, high, normal, etc.) Evidential Uncertainty (relationship is not certain) Linguistic uncertainty may be efficiently dealt with by Fuzzy Logic (Linguistic Variables) Example: 1.0 Very-High High Normal Very-Low Low Voltage (pu) µ (V) 0 0.85 0.90 0.95 1.00 1.05 1.10 1.15 6 Fuzzy Inference System (FIS) RULES Inference Engine Fuzzification Defuzzification x Crisp Output y = f(x) Fuzzy Input Fuzzy Output Crisp Input If x is A, Then y is B Antecedent Consequent

A Knowledge-Based System Previous Reports for Supervision ...falcao/Wm2002.pdf · Distribution: Light, Cerj, Escelsa SC Valadares Mascarenhas Vitória Campos Adrianópolis 500 345

  • Upload
    others

  • View
    1

  • Download
    0

Embed Size (px)

Citation preview

Page 1: A Knowledge-Based System Previous Reports for Supervision ...falcao/Wm2002.pdf · Distribution: Light, Cerj, Escelsa SC Valadares Mascarenhas Vitória Campos Adrianópolis 500 345

1

A Knowledge-Based Systemfor Supervision and Controlof Regional Voltage Profileand Security

Alessandro B. Marques Glauco N. Taranto Djalma M. Falcão Petrobras Federal University of Rio de Janeiro Formely with FURNAS COPPE Brazil Brazil

2

Previous Reports

TF 38.02.23 Report DraftPorto PowerTech, 2001

A.B. Marques, G.N. Taranto, and D. M. Falcão, “A SupervisoryKnowledge-Based System For Monitoring and Control ofRegional Voltage Profile.”

IEEE Transactions on Power SystemsA.B. Marques, G.N. Taranto, and D. M. Falcão, “A Knowledge-Based System for Supervision and Control of Regional VoltageProfile and Security.” (Submitted)

TF 38.02.20 Final Report (2nd Draft)

3

Cooperative Work

Federal University of Rio de Janeiro – COPPECEPELONS: The Brazilian ISOObjectives:

Analyze coordinated voltage schemes for possibleimplementation in the Brazilian SystemSurvey implementation issues and difficulties

4

Motivation

Characteristics of the Q-V InteractionHighly non-linearDiscrete elements

Customized Control SchemeRadial/meshed networksShort/long linesAvailability of reactive support in the area

5

Knowledge Based Systems

Most KBS uses production rules like If <condition>, Then <action or conclusion>

Uncertainty may exists in the rulesLinguistic Uncertainty (low, high, normal, etc.)Evidential Uncertainty (relationship is not certain)

Linguistic uncertainty may be efficiently dealt withby Fuzzy Logic (Linguistic Variables)Example:

1.0Very-HighHighNormalVery-Low Low

Voltage (pu)

µ (V )

0

0.85 0.90 0.95 1.00 1.05 1.10 1.15

6

Fuzzy Inference System (FIS)

RULES

Inference Engine

Fuzzification Defuzzification

x

Crisp

Output

y = f(x)

Fuzzy

Input

Fuzzy

Output

Crisp

Input

If x is A, Then y is BAntecedent Consequent

Page 2: A Knowledge-Based System Previous Reports for Supervision ...falcao/Wm2002.pdf · Distribution: Light, Cerj, Escelsa SC Valadares Mascarenhas Vitória Campos Adrianópolis 500 345

2

7

The Rio Area

Control Variables

Regulated Variables

Energy importing area with apeak load of 6,000 MWEquivalent of the BrazilianSouth/Southeastern subsystem730 buses, 1146 branches and104 generatorsLong transmission corridors(750 Km)Relatively weak reactive supportTransmission: FurnasDistribution: Light, Cerj, Escelsa

SC

Valadares

Mascarenhas

Vitória

Campos

Adrianópolis

500 345 kV

Marimbondo V. Grande

L.C.Barreto

Furnas

Aparecida

N. Peçanha

Santa CruzRIO AREA

Jacarepaguá

Grajaú

V. Redonda

Funil

Angra

C. Paulista

Itutinga

Tijuco Preto

CampinasPoços deCaldas

Araraquara

138 kV

138 kV

230 kV

230 kV

345 kV

500 kV

138 kV

F1

F2

F3

F4

138 kV

500 kV

345 kV500 kV

345 kV Corridor500 kV Corridor

about750 km (470 miles)

to Itaipu

to Jaguara

8

FIS-Power System Integration

POWER

SYSTEM

FISMarimbondo (kV)

Furnas (kV)

S. Cruz (kV)

Angra (kV)

Adriano (kV)

Jacare (kV)

Grajau (Mvar)

Cap./Rea. Switching

9

Linguistic Variable

Jacarepagua_138kV_Voltage

kV

10

Control Scheme Overview

Hierarchical Structure

Task Oriented ControlContinuous FIS: high-side voltage set-pointsDiscrete FIS:capacitor/reactor switchingDefensive Layer: checkon voltage limits in thetransmission corridors

Set Point ControlAVRsJVCs

ContinuousFIS

DiscreteFIS

PowerSystem

JVC/AVR

DefensiveLayer

SCADA dedicated channels

High Level(task-oriented control)

Low Level(set-point control)

11

Control Objectives

Minimize the number of capacitor/reactor bankswitching operationsMaximize (minimize) the generation of reactivepower from the line charging of the network inheavy (light) load conditionsHold the voltage at regulated buses smoothmost of the time, avoiding excessive spikescaused by the switching of discrete devices

12

Control Objectives (cont.)

Voltage Profile Control “Keep the voltage at Jacarepagua and

Adrianopolis around the desired valuesestablished by the electric utilities of theRio Area”

Security “Keep the Grajau SC reactive power output

between zero and –100 Mvar”

Page 3: A Knowledge-Based System Previous Reports for Supervision ...falcao/Wm2002.pdf · Distribution: Light, Cerj, Escelsa SC Valadares Mascarenhas Vitória Campos Adrianópolis 500 345

3

13

Continuous FIS (CFIS)

Two FIS used for differentload conditionsQualitatively both FIS usethe same rulesDifferent voltage set-pointvalues for some bussesCan be made different, ifnecessaryDecision to choose each FISis based on system demand,time of day, etc.

Heavy / MedianLoadFIS

Light / MinimumLoadFIS

SCADA

14

CFIS Rules

UpDownDown

UpRE

S.Cru

SUUpUpLILowNTL5SDREREIndHighOK4RESDRELCHighOK3UpUpUpCapLowLow2REREREOKOKOK1

AngrFurnMariGrajaJacarAdriaRule

OUTPUTSINPUTS

LI – Low Inductive

RE – Remain

SU – Small-Up

SD – Small-Down

NTL – Not too Low

Cap – Capacitive

LC – Low Capacitive

Ind – Inductive

Contain 29 rules based on experience and off-linestudies. Examples:

15

Control Surface

Typical nonlinear characteristic of the rules

16

Discrete FIS (DFIS)

First Layer: receives informationfrom the defensive layer and outputsvague information for the amount ofreactive power to be switched on/off(few, much, too much, etc).

Second Layer: decides which shuntbank should be switched on/off ineach substation.

Third Layer: deliberates the onlyone device, among the outputs of thesecond layer that should be switchedon/off.

DeliberativeLayer

Substation#1

Substation#n

Substation#1

Substation#n

1st layer

2nd layer

3rd layer

SCADA

17

DFIS RulesThe DFIS rules are specific for each layer andsubstationExample for the first layer (Jacarepagua substation)

If <Voltage is low and Adrianopolis_voltage is not too low and Grajau_output is low capacitive>, Then <switch on large amount of shunt capacitors>

Example for the second layer (Jacarepagua substation)

If <Switching on is large amount of shunt capacitors and Number of reactors switched on is zero>, Then <switch on large capacitor>

18

Simulation Setup

Santa Cruz power stationnot includedLoad variation and circuitoutageAvailable switchingelementsSimulations step 5 sFIS action at every 40 s

0 2000 4000 6000 s

Trip of one 500 kV circuitAngra-Adrianopolis (300s)

Area Load (MW)

6000

3000

Banks Available In operationCapacitors inJacarepagua

2 (100 Mvar) 1

Reactors inJacarepagua

3 (30 Mvar) 0

Capacitors inAdrianopolis

3 smalls (10 Mvar)2 medians (40 Mvar)2 larges (160 Mvar)

320

Reactors inAdrianopolis

2 smalls (50 Mvar)1 large (150 Mvar)

00

Page 4: A Knowledge-Based System Previous Reports for Supervision ...falcao/Wm2002.pdf · Distribution: Light, Cerj, Escelsa SC Valadares Mascarenhas Vitória Campos Adrianópolis 500 345

4

19

Simulation Tool

Fast Simulation Program (FastSim++) developed atCOPPE/Federal University of Rio de Janeiro, Brazil

Representation only of the mid and long term dynamics, like:

LTC actuation;

SVC;Demand curve

FISAdriano (kV)

Jacare (kV)

Grajau (Mvar)

Marimbondo (kV)

Furnas (kV)

S. Cruz (kV)

Angra (kV)

20

Simulation Results

Event Time(sec)

Switching Event

Time(sec)

Switching

1 40 Capacitor bank atJacarepagua switched on

10 1800 Small reactor bank atAdrianopolis switched on

2 720 Small capacitor bank atAdrianopolis switched off

11 3960 Small reactor bank atAdrianopolis switched off

3 800 Small capacitor bank atAdrianopolis switched off

12 4080 Small reactor bank atAdrianopolis switched off

4 840 Small capacitor bank atAdrianopolis switched off

13 4240 Small capacitor bank atAdrianopolis switched on

5 920 Median capacitor bank atAdrianopolis switched off

14 4320 Small capacitor bank atAdrianopolis switched on

6 1080 Capacitor bank atJacarepagua switched off

15 4400 Small capacitor bank atAdrianopolis switched on

7 1280 Capacitor bank atJacarepagua switched off

16 4440 Median capacitor bank atAdrianopolis switched on

8 1440 Median capacitor bank atAdrianopolis switched off

17 4480 Capacitor bank atJacarepagua switched on

9 1640 Small reactor bank atAdrianopolis switched on

18 4640 Median capacitor bank atAdrianopolis switched on

Discrete elements switching sequence

21

Marimbondo 500 kV Voltage

Initially voltage aboveupper limitLow voltage at loadbussesAntagonistic situationDefensive layer triggersDFIS that switchescapacitor banks onAt 700 sec, terminalvoltage reaches lowerlimit (not shown).DFIS orders capacitorbank switching off

0 500 1000 1500 2000 2500 3000 3500 4000 4500 5000490

500

510

520

530

540

550

560

570

580Marimbondo 500 kV

Time(sec)

Vol

tage

(kV

)

Upper Limit

Lower Limit

22

Angra 500 kV Voltage

Similar behavior toMarimbondoAt 1800 sec,underexcitation limiter isreached (not shown)

0 500 1000 1500 2000 2500 3000 3500 4000 4500 5000490

500

510

520

530

540

550

560

570

580Angra 500 kV

Time(sec)

Vol

tage

(kV

)

Upper Limit

Lower Limit

23

Jacarepagua 138 kV Voltage

Initial low voltageArrows represent time ofshunt bank switchingsAt 40 sec, 100 Mvarcapacitor bank switcheson

0 500 1000 1500 2000 2500 3000 3500 4000 4500 5000120

125

130

135

140

145

150Jacarepaguá 138 kV

Time(sec)

Vol

tage

(kV

)

Upper Limit

Lower Limit

jp-c-on

ad-mc-off

jp-c-off

ad-sr-off jp-c-on

24

Adrianopolis 138 kV VoltageInitial low voltageDesired voltage range inheavy conditions:between 143 and 145 kVSimilar behavior toJacarepaguaWorth mentioning:smoothness between3000 and 4000 sec whenload ramp is positivePriority given tocontinuous controlsAt 4000 sec, Marimbondoreaches upper limit

0 500 1000 1500 2000 2500 3000 3500 4000 4500 5000120

125

130

135

140

145

150Adrianópolis 138 kV

Time(sec)

Vol

tage

(kV

)

Upper Limit

Lower Limit

jp-c-on jp-c-off

ad-mc-off

Page 5: A Knowledge-Based System Previous Reports for Supervision ...falcao/Wm2002.pdf · Distribution: Light, Cerj, Escelsa SC Valadares Mascarenhas Vitória Campos Adrianópolis 500 345

5

25

Grajau SC Reactive Power

Output most of the timeremains within desirablerange [ -100 ; 0 Mvar]Fast response at 300 secwhen a 500 kV circuittrips out.

0 500 1000 1500 2000 2500 3000 3500 4000 4500 5000-400

-300

-200

-100

0

100

200

300

400

500Grajaú SC Mvar

Time(sec)

Rea

ctiv

e P

ower

Upper Limit

Lower Limit

26

Final Remarks

Encouraging results in the experimentconducted with a model of the Rio AreaCan be used as a decision support tool tohelp operators or as an automatic controltoolFuture work: tests with present systemconfiguration