dynamic var planning

7/30/2019 Dynamic VAR Planning

1/31

DYNAMIC VAR PLANNING IN A LARGE POWERSYSTEM USING TRAJECTORY SENSITIVITIES

1September 5, 2009

by

Bishnu Sapkota

Advisor : Vijay Vittal

Arizona State University


2/31

Outline

Background

Objective

Load modeling for voltage stability studies

Corrective measures for voltage stability

Dynamic VAr Planning

Study results

Conclusions

2


3/31

Background

Power system stability- The ability

of an electric power system,

operating at a given initial condition,to regain a state of operating

equilibrium after being subjected to

a physical disturbance, with most of

the system variables bounded, sothat practically the entire system

remains intact.

Voltage stability - ability of a power

system to maintain steady voltagesat all the buses in the system afterdisturbance.

3

Fig.1 Classification of power system stability


4/31

Background Based on actual incidents, short term voltage stability has been an

increasing concern in power industry

The short term voltage stability problem may arise in two forms. Slow voltage recovery

Fast collapse

Several events that have occurred and recent industry work mainly

deal with slow voltage recovery following short circuits with stalling

and slow tripping of residential air conditioner compressor motors

However, the fast voltage collapse has not been addressed in the

literature

VAr planning using static criteria are well established.

The suitability of dynamic reactive power source is another subject

which has drawn significant attention regarding the fast voltage

collapse4


5/31

Objectives of the study

Fast voltage collapse problem associated with the fault on the bulkpower network in presence of significant induction motor loads

Dynamic VAr planning methodology based on trajectory sensitivity

analysis

Comparison of efficacy of SVC and STATCOM to prevent fastvoltage collapse

5


6/31

Voltage stability causes and analysis

Causes of voltage instability Increase in loading

Generators, synchronous condensers, or SVCs reaching reactivepower limits

Tap-changing transformer action

Load recovery dynamics Tripping of heavily loaded lines, generators

Methods of voltage stability analysis Static analysis methods

Dynamic analysis methods

6


7/31

Load characteristics

The accuracy of analytical results depends on proper modeling of

power system components, devices, and controls

Loads are most difficult to model

Complex in behavior varying with time and location

Statistical in nature, and consists of a large number of continuous and discrete

controls and protection systems

Dynamics of loads, especially, induction motors at low terminal

voltage should be properly modeled

7


8/31

Induction motor characteristics

Impact of fault on transmission

grid

Depressed voltages at distributionfeeders and motor terminals.

Reduction of electrical torque by the

square of the voltage resulting in

slow down of motors.

The slow down depends on the

mechanical torque characteristics

and motor inertias.

With fault clearing

8

Partial voltage recovery Slowed down motors draw high reactive currents, depressing voltage magnitudes

Motor will reaccelerate to normal speed if, electrical torque>mechanical torque

Else, the motors will rundown, stall, and trip

The problem is severe in summer time period with large proportion of air conditionermotors

Speed per unit

Electric

torque

1.0

Current

Mechanical

torque

1.0

5.0

Torqu

eor

currentperuni

t

Fig. 2 Induction motor characteristics


9/31

Air conditioner motor characteristics

Characteristics

Main portion (80-87%) consumed by air compressor motor

Electromagnetic contactor drop out between (43-56%) of the nominal

voltage and reclose above drop out voltage

Stalling at (50-73%) of the nominal voltage

Thermal overload protection act if motors stall for 5-20 seconds The operation time of TOL (Thermal Over-load) protection relay is

inversely proportional to the applied voltage at the terminal

Air conditioners should be modeled to analyze the short term

voltage stability problem Quite important for utilities in the Western interconnection

9


10/31

Load modeling

Old models

Loads are represented as lumped

load at the distribution feeder The diversity in composition and

dynamic behavior of various

electrical loads is not modeled

WECC interim model 20% of the load as generic

induction motor load

80% constant current P and

constant impedance Q

Composite load model

Representation of distribution

system equivalent

Parameters of various load

components

Substation

Capacitor

Transmission Bus

Bus 1

Bus 2

Distribution Bus

OLTC

LM SM DL EL IL

Distribution FeederBus 3

Distribution

Capacitor

10Fig.3 Composite load model structure


11/31

Corrective measure- Static VAr

compensator (SVC) Variable reactive power source

Can generate as well as

absorb reactive power Maximum and minimum limits

on reactive power output

depends on limiting values of

capacitive and inductivesusceptances.

Fixed capacitor with thyristor

controlled reactor model is

used.

Droop characteristics of 2%

over the control range is used.

11

CFixedCapacitor

Transformer

TCR

XC

XL

Vref

VXSL

I

Fig. 5 V-I Characteristics of SVC

Fig. 4 Schematic diagram of an SVC

V

I


12/31

Corrective measure- Static compensator

(STATCOM)

Voltage source converter device

Alternating voltage sourcebehind a coupling reactance

Can be operated at its full output

current even at very low voltages

Depending upon manufacturer's

design, STATCOMs may have

increased transient rating both in

inductive as well as capacitive

mode of operation.

PI controller to regulate the

voltage and a droop

characteristics of 2% over thecontrol range is used. 12

VT

ILmax ILTICmaxIICT

InductiveCapacitive

Transformer

DC-AC switching

converter

I

X

System bus

Cs

Vdc

E

Fig. 6 Schematic diagram of STATCOM

Fig. 7 V-I Characteristics of STATCOM

VT


13/31

Trajectory sensitivity Why?

The influence of parameters on the non-linear, non-smooth

behavior exhibited by a disturbed power system is difficult toexplore

Normal linearization techniques involving linearization of system

model about an operating point are not very useful

What?

Based on linearizing the system around a trajectory rather than

around an equilibrium point

Change in trajectory due to small change in parameters isobserved

13


14/31

Dynamic VAr planning

The analytical description of power system is given by

The flows of x and y can be defined as,

Sensitivities of the system flow to the parameters are,

14

),,( Oyxfx

),,(0 Oyxg

),,()( OI txtxox

),,()( OI txtyoy

OOO

O''

w

w' )()()( txtxtx

OO

O

O''

w

w' )(

)()( ty

tyty

(5)

(6)

(7)

(8)

(9)

(10)


15/31


A simple numerical procedure is used to evaluate these

sensitivities

These sensitivities are calculated along the trajectory, and arecalled trajectory sensitivities

Sensitivities of the change in bus voltage magnitudes to thereactive power injection at a given bus are calculated, where Vis one of the output variable represented by y, and reactivepower Q is one of the parameters represented by O

15

O

OIOOI

OO

'

'|

w

w

),,(),,( txtxxx oxox

OOIOOI

OO

' '|ww),,(),,( txtxyy oyoy

(11)

(12)


16/31


The trajectory sensitivity index (TSI) proposed in this work isdefined as,

Wk is the weighting factor to designate the importance of the

time instant k and Wbi is the weighting factor to represent theimportance of bus ion the sensitivity calculation

If the voltage collapse is local in nature, then there will only be asmaller number of buses with non-zero W

bi. The selection of

the weights Wk

depends upon the type of voltage instabilitybased on time frame of interest

The bus with the highest trajectory sensitivity index is selectedas a candidate bus for dynamic VAr support

16

w

w

k

k

tt

n

i

j

i

bi

N

kkj

Q

VWWTSI

11(13)


17/31

17

System description and tools

Component Number

Buses 14585Plants 1938

Generators 1842

Shunts 1179

Lines 11655

Transformers 4384

Phase Shifters 59

Converters 8

Component Number

Buses 581Plants 37

Generators 32

Shunts 39

Lines 614

Transformers 74

Phase Shifters 0

Converters 0

System components Study subsystem components

Number of 12.5 kV feeders in the subsystem = 375

Following tools have been used in this study

Dynamic security analysis tools PSAT,VSAT, and TSAT PSLF, Matlab


18/31

Load modeling

Composite load model consisting of

static and motor loads is used.

The percentage of motor load (bothmegawatts and MVAr) at each

distribution substation is about 72%

of the total load at that substation.

The induction motors are classifiedas,

Small motors - Low inertia motors

Large motors - High inertia motors

Trip motors - Low inertia motors with theoption of tripping under low voltage

condition

18

Fig. 8 Detailed load modeling sample at

representative bus

Static

LoadLM STSM

69 kV Bus

DistributionTransformer

12.5 kV Bus

The percentages of small, large, and trip motors at each bus are

63, 10, and 27 respectively.


19/31

Study case

Fault on a 500 kV transmission line and the subsequent double line

outage as follows, which causes voltage collapse in many zones.

line connecting 500 kV Bus 9 and Bus 23line connecting 500 kV Bus 1 and Bus 4

19

Fig. 9 Double line outage considered most critical

The lines are the major transmission corridors for importing power to the

subsystem


20/31

Bus voltages without motor dynamics

The contingency for the

analysis is defined as follows,

A fault is applied at an artificialbus 90000 which is created at

40% of the 500 kV line 1-4

The fault is cleared after 5

cycles by removing the twolines

No serious voltage problems

are observed when motor

dynamics are not included The bus voltages restore

almost to the pre-fault values

20

Time (sec)

Busvoltagemagnitudes(pu)

Fig. 10 Bus voltages without motor dynamics


21/31

Bus voltages with motor dynamics

The trip motor (27% of the

motor load at each bus)parameter Vt is chosen as 0.55

pu and Tv is chosen as 10

milliseconds

Low voltage magnitudes areobserved at most of the buses

with few bus voltages dropping

to as low as 40%

Although the voltage drop hasbeen observed at 230 kV and

500 kV level also, the drop is

more pronounced at the 69 kV

and 12.5 kV buses21

Fig. 11 Bus voltages without motor dynamics and

without corrective action

Busvoltagem

agnitude(pu)

Time (sec)

0.00 2.00 4.00 6.00 8.00 10.000.00

0.22

0.44

0.66

0.88

1.10

BUS108 230 kVBUS18 500 kV

BUS333 69 kV BUS735 12.5 kV


22/31

Motor voltage and reactive power demand

It is observed that the reactive

power demand of the induction

motor increases during the

post contingency period

This large increase in reactive

power demand further reduces

the voltage magnitudes in thesystem, thereby resulting in

voltage collapse

At such voltage levels, even

the SM motor would stall,overload, and trip

The typical tripping time for

motors stalled at 50% voltage

is 15-20 seconds

Fig. 12 Reactive power demand of induction motorsat a representative bus

Fig. 13 Terminal voltage magnitude of induction moto

at representative bus

Inductionmotortermin

alvoltage(pu)

Time (sec)0.00 2.00 4.00 6.00 8.00 10.000.00

0.22

0.44

0.66

0.88

1.10

Time (sec)0.00 2.00 4.00 6.00 8.00 10.00

0.00

3.00

6.00

9.00

12.00

15.00

ST MOTOR

SM MOTOR

Inductionmotorreac

tivepowerdemand(MVAR)


23/31

Generator response

The outputs of the generators

especially in zone 126 are

exceeding their maximum VAr

capacity, thereby stressing the

generating units

This shows that there is a

deficiency of dynamic reactive

power in the zone. If appropriate

actions are not taken, the zone

could be subjected to severe

voltage instability

23

Fig. 14 Generator reactive power output

Generatorreact

ivepower(MVAR)

Time (sec)

0.00 2.00 4.00 6.00 8.00 10.00-70

12

94

176

258

340

GEN580

GE581

GEN579


24/31

Modal analysis

Modal Analysis is performed in

two ways

Case A: By increasing the power

transfer with the power factor same as inthe base case

The zones associated with the critical

modes do not represent the zones with

very low post fault voltage magnitudes in

time domain analysis

Case B: By increasing the powertransfer at low power factor (0.3)

Critical zones - 126,127,128, 129

Modal analysis at point ofcollapse may not accurately

pinpoint the buses or zones that

require reactive power support

when fast voltage collapse occurs24

S. No. Modes Participating

zones

1 0.014904+j0 119/124/125

2 0.048814+j0 1657

3 0.057995+j0 1843

4 0.063687+j0 1657

5 0.069565+j0 1843/1844

Case A: Critical modes at nose point

S. No. Bus No. Zone No. Bus Participation

Factor

1 397 128 0.53294

2 391 128 0.52416

3 378 126 0.507754 591 129 0.50457

5 396 128 0.50157

6 595 126 0.4993

7 379 126 0.49719

8 389 128 0.49445

9 333 126 0.49431

10 380 128 0.49396

Case B: Bus participation factors


25/31

Trajectory sensitivity and location of

dynamic VAr W

bihas been chosen to be 1 for

all the buses, since the voltage

collapse has affected all the busvoltage magnitudes

Wk

=1 for time instants t=0.1, 0.2,

0.3, 0.4, 0.5, 0.6, 0.7, 0.8, 0.9, 1.0

seconds after the fault, and zerofor all other instants

It is observed that the 230 kV

buses 108, 102, 107, and 96 have

higher sensitivities

These buses are selected primarily

as the locations for reactive power

sources

25

Fig. 15 TSI values

T j t iti it d l ti f


26/31


dynamic VAr

STATCOMs with a rating of 350 MVAr

each are placed at buses 108, 102,

107, and 96 The 230 kV voltages recover to nearly

1 pu in about 4 seconds

However, the voltages at the 69 kV

buses and the distribution feeders(12.5 kV) take much longer time (about

7 seconds) to recover to pre-fault

values

Since the voltage collapse occurs at

the low voltage level (69 kV, and 12.5

kV), it appears that the shunt

compensation must be provided at 69

kV buses for effective voltage recovery

Fig. 16 230 kV bus voltages in the system with 900 MVArSTATCOMS at three buses

Fig. 17 Bus voltages (12.5-69 kV) in the system with

900 MVAr STATCOMS at three 230 kV buses

Busvolta

gemagnitude(pu)

Time (sec)

0.00 2.00 4.00 6.00 8.00 10.000.00

0.24

0.48

0.72

0.96

1.20

Busvoltagemagn

itude(pu)

Time (sec)0.00 2.00 4.00 6.00 8.00 10.00

0.00

0.24

0.48

0.72

0.96

1.20

BUS591 69 kV

BUS817 12.5 kV


27/31


dynamic VAr

The 69 kV buses that areconnected to the 230 kV buses

with high sensitivities are

chosen as locations for

STATCOMs The STATCOMs are removed

from the 230 kV buses and then

connected to 69 kV buses

The maximum limit of the valueof shunt compensation at a 69

kV bus is chosen to be 72 MVAr

based on the physical

constraints at the substation27

Buses selected as STATCOM location

S. NO. BUS NO. SIZE (MVAR)

1 17 72

2 102 72

3 170 72

4 322 72

5 328 44

6 342 72

7 343 728 344 72

9 357 72

10 373 72

11 375 72

12 381 72

13 477 72

14 535 72

15 550 72

16 551 72

17 554 72

18 584 72

19 591 72

TOTAL 1340 (MVAR)



28/31


dynamic VAr

The STATCOMs act quickly to

increase their injected current and

thus provide the required reactive

power

Voltages at all levels are restored to

respective pre-fault values

The STATCOMs are required toprovide the reactive power to their

full capacity only for a short period

of time right after fault

Power electronic based shuntcompensation with high short term

ratings may be an effective option to

prevent voltage collapse caused by

presence of significant inductionmotor loads

Fig. 18 Bus voltages after placement of

STATCOMs at 69 kV buses

Fig. 19 STATCOM output for the fault under study

Busvoltagemagnitude(pu)

Time (sec)

10.000.00 2.00 4.00 6.00 8.000.00

0.24

0.48

0.72

0.96

1.20

BUS18 500 kV

BUS108 230 kV

BUS333 69 kV

BUS735 12.5 kV

OutputcurrentofSTA

TCOM(pu)

Time (sec)

0.00 2.00 4.00 6.00 8.00 10.00-1.00

-0.78

-0.56

-0.34

-0.12

0.10


29/31

Comparison of SVC and STATCOM

The bus voltage recovers to 1 pu in

about 2 seconds when STATCOMs

are used It takes about 4 seconds for the

voltage to recover to 1 pu when SVCs

are used

The maximum VAr output of the SVCdecreases with the square of the AC

system voltage, while that of

STATCOM decreases linearly with

the AC system voltage The STATCOM is therefore found

more effective during large system

disturbances

29

Fig. 20 Terminal voltage of Induction motor

at a bus 735

Inductionmotorterminalvoltage(pu)

Time (sec)

10.000.00 2.00 6.00 8.00

0.24

0.48

8.000.00 2.00 4.00 6.00 8.000.00

0.72

0.96

1.20

4.00 6.000.00

0.72

0.96

1.20

Without VAr

support

With SVC

With STATCOM


30/31

Conclusions

Voltage collapse for the case under study is caused by short-term

voltage instability initiated by a high reactive power demand of the

induction motors during the disturbance period

A fault on a high voltage transmission line might cause severevoltage dips at the distribution feeders, even though the voltage dips

at high voltage level is not that severe

Trajectory sensitivity analysis can be very useful in determining the

location of dynamic VAr sources

Power electronic based shunt compensation with high short term

ratings may be an effective and attractive option

STATCOMs provide a better option to improve short term voltageinstability problems than SVCs

Smaller modules of STATCOMs at 69 kV buses are found to be

more effective than larger modules at 230 kV buses in order to

obtain faster voltage recovery30


31/31

31

dynamic var planning

Documents