DYNAMIC VAR PLANNING IN A LARGE POWER SYSTEM USING TRAJECTORY SENSITIVITIES

by Bishnu Sapkota Advisor : Vijay Vittal

Arizona State University

September 5, 2009
1

Outline
Background Objective Load modeling for voltage stability studies Corrective measures for voltage stability Dynamic VAr Planning Study results Conclusions

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, so that practically the entire system remains intact. Voltage stability - ability of a power system to maintain steady voltages at all the buses in the system after disturbance.

Fig.1 Classification of power system stability

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 collapse
4

Objectives of the study

Fast voltage collapse problem associated with the fault on the bulk power 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 fast voltage collapse

Voltage stability causes and analysis

Causes of voltage instability
Increase in loading Generators, synchronous condensers, or SVCs reaching reactive power 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

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

Induction motor characteristics

Impact of fault on transmission grid
Depressed voltages at distribution feeders 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.
5.0 Torque or current per unit Current Electric torque Mechanical torque

1.0

Speed per unit

1.0

With fault clearing

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 conditioner motors 8
Fig. 2 Induction motor characteristics

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

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
Transmission Bus Bus 1

OLTC Distribution Bus Bus 2

WECC interim model

20% of the load as generic induction motor load 80% constant current P and constant impedance Q
Bus 3

Substation Capacitor Distribution Feeder

Composite load model

Representation of distribution system equivalent Parameters of various load components
LM SM DL EL IL Distribution Capacitor

Fig.3 Composite load model structure 10

Corrective measure- Static VAr compensator (SVC)

V

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 inductive susceptances. Fixed capacitor with thyristor controlled reactor model is used. Droop characteristics of 2% over the control range is used.

Transformer

Fixed Capacitor C TCR

Fig. 4 Schematic diagram of an SVC

XSL

V XL

XC

Vref

Fig. 5 V-I Characteristics of SVC

11

Corrective measure- Static compensator (STATCOM)

System bus
VT

Voltage source converter device Alternating voltage source behind 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 the control range is used.

Transformer I X E
DC-AC switching converter

Cs Vdc

Fig. 6 Schematic diagram of STATCOM

VT

ICT

ICmax Cma

ILmax

ILT

Capacitive

Inductive

Fig. 7 V-I Characteristics of STATCOM

12

Trajectory sensitivity
Why? The influence of parameters on the non-linear, non-smooth behavior exhibited by a disturbed power system is difficult to explore 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 is observed

13

Dynamic VAr planning

The analytical description of power system is given by

f ( x, y , )

(5) (6) (7) (8)

0 g ( x, y , )
The flows of x and y can be defined as,
x(t )
x

( x , t, )
o

y (t ) ( x , t, ) Sensitivities of the system flow to the parameters are,

y o

x(t ) y (t )

x(t ) y (t )

x (t ) y (t )

(9)

(10)
14

Dynamic VAr planning

A simple numerical procedure is used to evaluate these sensitivities
x
y

x
y

( x , t,
o

)
)

( x , t, )
o

(11)

( x , t,
o

( x , t, )
o

(12)

These sensitivities are calculated along the trajectory, and are called trajectory sensitivities Sensitivities of the change in bus voltage magnitudes to the reactive power injection at a given bus are calculated, where V is one of the output variable represented by y, and reactive power Q is one of the parameters represented by
15

Dynamic VAr planning

The trajectory sensitivity index (TSI) proposed in this work is defined as,
TSI
Nk j k 1

n i 1

bi

V Q

(13)
t tk

Wk is the weighting factor to designate the importance of the time instant k and Wbi is the weighting factor to represent the importance of bus i on the sensitivity calculation If the voltage collapse is local in nature, then there will only be a smaller number of buses with non-zero Wbi. The selection of the weights Wk depends upon the type of voltage instability based on time frame of interest The bus with the highest trajectory sensitivity index is selected as a candidate bus for dynamic VAr support
16

System description and tools

System components Component Buses Plants Generators Shunts Lines Transformers Phase Shifters Converters Number 14585 1938 1842 1179 11655 4384 59 8 Study subsystem components Component Buses Plants Generators Shunts Lines Transformers Phase Shifters Converters Number 581 37 32 39 614 74 0 0

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
17

Load modeling
Composite load model consisting of static and motor loads is used. s T The percentage of motor load (both megawatts and MVAr) at each distribution substation is about 72% of the total load at that substation. The induction motors are classified as,
Small motors - Low inertia motors Large motors - High inertia motors Trip motors - Low inertia motors with the option of tripping under low voltage condition

69 kV Bus
Distribution Transformer

12.5 kV Bus
Static Load
SM ST

LM

Fig. 8 Detailed load modeling sample at representative bus

The percentages of small, large, and trip motors at each bus are 63, 10, and 27 respectively.
18

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 23 line connecting 500 kV Bus 1 and Bus 4

Fig. 9 Double line outage considered most critical

The lines are the major transmission corridors for importing power to the subsystem
19

Bus voltages without motor dynamics

The contingency for the analysis is defined as follows,
Bus voltage magnitudes (pu)

A fault is applied at an artificial bus 90000 which is created at 40% of the 500 kV line 1-4 The fault is cleared after 5 cycles by removing the two lines

No serious voltage problems are observed when motor dynamics are not included The bus voltages restore almost to the pre-fault values

Time (sec)

Fig. 10 Bus voltages without motor dynamics

20

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 are observed at most of the buses with few bus voltages dropping to as low as 40% Although the voltage drop has been observed at 230 kV and 500 kV level also, the drop is more pronounced at the 69 kV and 12.5 kV buses

1.10 BUS18 500 kV Bus voltage magnitude (pu) 0.88

BUS108 230 kV

0.66

0.44

0.22 BUS333 69 kV 0.00 BUS735 12.5 kV

0.00

2.00

4.00 6.00 Time (sec)

8.00

10.00

Fig. 11 Bus voltages without motor dynamics and without corrective action 21

Motor voltage and reactive power demand

Induction motor reactive power demand (MVAR) 15.00

Induction motor terminal voltage (pu)

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 the system, 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

12.00 SM MOTOR 9.00

6.00 3.00 ST MOTOR

0.00

0.00

2.00

4.00 6.00 Time (sec)

8.00

10.00

Fig. 12 Reactive power demand of induction motors at a representative bus

1.10 0.88 0.66 0.44 0.22 0.00

0.00

2.00

4.00 6.00 Time (sec)

8.00

10.00

Fig. 13 Terminal voltage magnitude of induction motor at representative bus

Generator response
T The outputs of the generators especially in zone 126 are e 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
340

Generator reactive power (MVAR)

258

GEN579 176 GE581 94 GEN580

12

-70 0.00 2.00 4.00 6.00 8.00 10.00 Time (sec)

Fig. 14 Generator reactive power output

23

Modal analysis
Modal Analysis is performed in two ways
Case A: By increasing the power
transfer with the power factor same as in the 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 A: Critical modes at nose point S. No. Modes Participating zones 1 0.014904+j0 119/124/125 2 3 4 5 0.048814+j0 0.057995+j0 0.063687+j0 0.069565+j0 1657 1843 1657 1843/1844

Case B: By increasing the power

transfer at low power factor (0.3) Critical zones - 126,127,128, 129

S. No. 1 2 3 4 5 6 7 8 9 10

Modal analysis at point of collapse may not accurately pinpoint the buses or zones that require reactive power support when fast voltage collapse occurs

Case B: Bus participation factors Bus No. Zone No. Bus Participation Factor 397 128 0.53294 391 128 0.52416 378 126 0.50775 591 129 0.50457 396 128 0.50157 595 126 0.4993 379 126 0.49719 389 128 0.49445 333 380 126 128 0.49431 0.49396 24

Trajectory sensitivity and location of dynamic VAr

Wbi has been chosen to be 1 for all the buses, since the voltage a collapse has affected all the bus c voltage 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 zero for 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

Fig. 15 TSI values

25

Trajectory sensitivity and location of dynamic VAr

1.20

STATCOMs with a rating of 350 MVAr each are placed at buses 108, 102, e 107, and 96 1 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

Bus voltage magnitude (pu)

0.96

0.72

0.48

0.24

0.00

0.00

2.00

4.00 6.00 Time (sec)

8.00

10.00

Fig. 16 230 kV bus voltages in the system with 900 MVAr STATCOMS at three buses
1.20

Bus voltage magnitude (pu)

0.96
BUS591 69 kV

0.72

0.48
BUS817 12.5 kV

0.24

0.00

0.00

2.00

4.00 6.00 Time (sec)

8.00

10.00

Fig. 17 Bus voltages (12.5-69 kV) in the system with 900 MVAr STATCOMS at three 230 kV buses

Trajectory sensitivity and location of dynamic VAr

Buses selected as STATCOM location

T The 69 kV buses that are connected 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 value of shunt compensation at a 69 kV bus is chosen to be 72 MVAr based on the physical constraints at the substation

S. NO. 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 TOTAL

BUS NO. SIZE (MVAR) 17 72 102 72 170 72 322 72 328 44 342 72 343 72 344 72 357 72 373 72 375 72 381 72 477 72 535 72 550 72 551 72 554 72 584 72 591 72 1340 (MVAR)

27

Trajectory sensitivity and location of dynamic VAr

1.20

The STATCOMs act quickly to increase their injected current and i thus provide the required reactive t power Voltages at all levels are restored to respective pre-fault values The STATCOMs are required to provide the reactive power to their full capacity only for a short period of time right after fault Power electronic based shunt compensation with high short term ratings may be an effective option to prevent voltage collapse caused by presence of significant induction motor loads

0.96 Bus voltage magnitude (pu)

0.72 BUS735 12.5 kV 0.48 BUS333 69 kV 0.24

BUS18 500 kV

BUS108 230 kV

0.00

0.00

2.00

4.00 6.00 Time (sec)

8.00

10.00

Fig. 18 Bus voltages after placement of STATCOMs at 69 kV buses

0.10

Output current of STATCOM (pu)

-0.12

-0.34

-0.56

-0.78

-1.00

0.00

2.00

4.00 Time (sec)

6.00

8.00

10.00

Fig. 19 STATCOM output for the fault under study

Comparison of SVC and STATCOM

The bus voltage recovers to 1 pu in about 2 seconds when STATCOMs a are used a It takes about 4 seconds for the voltage to recover to 1 pu when SVCs are used The maximum VAr output of the SVC decreases 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

1.20 Induction motor terminal voltage (pu)

0.96

0.72

With STATCOM With SVC VC

0.48

0.24

Without VAr support 0.00 2.00 4.00 6.00 Time (sec) 8.00 10.00

0.00

Fig. 20 Terminal voltage of Induction motor at a bus 735

29

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 severe voltage 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 voltage instability 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 recovery
30

31