Power Sys Dynamics and Stability

Power System Dynamics and Stability

Dr. Federico Milano
E-mail: Federico.Milano@uclm.es Tel.: +34 926 295 219
Departamento de Ingeniera Electrica, Electronica, Automatica y Comunicaciones
Escuela Tecnica Superior de Ingenieros Industriales
Universidad de Castilla-La Mancha June 26, 2008 Introduction - 1
Programa de Doctorado
Technical and Economical Management of Generation, Transmission and Distribution Electric Energy Systems
Area de Ingeniera Electrica de la E.T.S. de Ingenieros Industriales de la Universidad de Castilla - La Mancha
Note
This course is partly based on the course ECE664 hold by Prof. Dr. C.
Canizares at the University of Waterloo, Ontario, Canada. I wish to sincerely thank Prof. Dr. C. Canizares for his courtesy in sharing this material.
Objectives
Understand the modeling and simulation of power systems from phasor

analysis to electromagnetic transients.
Discuss the basic denitions, concepts and tools for stability studies of
power systems.
Familiarize with basic concepts of computer modelling of electrical power

systems.
Outlines: System Modeling
Synchronous machine. Transformer. Transmission line. Cable. Loads.
Outlines: System Analysis
Basic stability concepts: Nonlinear systems. Equilibrium points. Stability regions. Power Flow: System model. Equations and solution techniques. Contingency analysis.
Outlines: Voltage Stability
Denitions. Basic concepts: Saddle-node bifurcation. Limit-induced bifurcation. Continuation Power Flow (CPF). Direct methods. Indices. Protections and controls. Real case example: August 2003 North American blackout.
Outlines: Angle Stability
Denitions. Small-disturbance: Hopf Bifurcations. Control and mitigation. Practical applications. Transient Stability (large-disturbance): Time domain. Direct Methods: Equal Area Criterion. Energy Functions. Real case example: May 1997 Chilean blackout.
Outlines: Frequency Stability
Denitions. Basic concepts. Protections and controls. Real case example: October 2003 Italian blackout. Real case example: November 2006 European blackout.
10
Outlines: Software Tools
Outlines. UWPFLOW. Matlab. PSAT.
11
References
P. Kundur, Power system stability and control, Mc Graw Hill, 1994. P. Sauer and M. Pai, Power system dynamics and stability, Prentice Hall,
1998.
A. R. Bergen and V. Vittal, Power systems analysis, Second Edition,

Prentice-Hall, 2000.
C. A. Caizares, Editor, Voltage stability assessment: concepts, practices

and tools, IEEE-PES Power System Stability Subcommittee Special Publication, SP101PSS, May 2003.
12
References
P. M. Anderson and A. A. Fouad, Power system control and stability, IEEE

Press, 1994.
J. Arrillaga and C. P. Arnold, Computer analysis of power systems, John

Wiley, 1990.
I. S. Duff, A. M. Erisman and J. K. Reid, Direct Methods for Sparse

Matrices, Oxford Science Publications, 1986.
J. Stoer and R. Bulirsch, Introduction to Numerical Analysis, Second

Edition, Springer-Verlag, 1993.
13
References
M. Ili and J. Zaborszky, Dynamics and Control of Large Electric Power c

Systems, Wiley, New York, 2000. C. A. Canizares, UWPFLOW, available at www.power.uwaterloo.ca
F. Milano, PSAT, Power System Analysis Toolbox, available at

www.power.uwaterloo.ca
Journal papers and technical reports. Course notes available on line.
14
Evaluation
Two projects are required. The projects concentrates in the various topics discussed in class. These will require the use of MATLAB, PSAT and UWPFLOW (the last
two are free software for stability studies co-developed at the University of Waterloo, Canada).
Reproducing examples presented in the slides using UWPFLOW and

MATLAB.
Stability analysis of the IEEE 14-bus test system using PSAT.
15
Evaluation
Alternatively, the students can develop a user dened model in PSAT. Interested students are invited to contact Dr. Federico Milano.
16
Contents
Introduction Generator Modeling Transmission System Modeling Load Modeling Power Flow Outlines Stability Concepts Voltage Stability Angle Stability Frequency Stability Software Tools Projects
17
Generator Modeling
Generator overview. Synchronous machine. Dynamic models of generators for stability analysis: Subtransient model Transient model Basic control models Steady-state model.
Universidad de Castilla-La Mancha June 26, 2008 Generator Modeling - 1
18
Generator Overview
Generator components:
Steam at pressure, P Fuel Boiler Enthalpy, h Turbine Torque at speed, Generator Power at voltage, V Current, I
Firing
control
Governor
Excitation
system
Power set-point
Ref
Ref V
Generator: Synchronous machine: AC stator and DC rotor. Excitation system: DC generator or static converter plus voltage
regulator and stabilizer
19
Generator Overview
Generator with DC exciter and associated controls:

Commutator Amplidyne field dc Exc slip rings Regulator transfer PTs Field breaker ac Gen CT
Exciter field rheostat (manual control)
Magnetic amplifier Transistor amplifier
Stabilizer
Reference and voltage sensing Limiter sensing
Compensator
Magnetic amplifier
Other sensing MG set PMG Regulator power M 9052
Other inputs
Station auxiliary power
20
Generator Overview
Generator with static exciter and associated controls:

Auxiliary power input start up
Voltage buildup element
FDR
Syn machine
Excitation power current transformer
CT
slip rings
Power rectifier Excitation power Excitation breaker
Excitation power potential transformer
PTs Linear reactor Trinistat power amplifier
Gate circuitry
Base adjuster manual control Regulator transfer Stabilizer Reference & voltage sensing Limiter sensing Other inputs Compensator Voltage adjuster
Signal mixing amplifier Rectifier current limit
Other sensing
21
Synchronous Machine
a
Damper windings
r dr
c
Effects of induced currents in the rotor core
b
F D D Q1 Q2
r ar
Q 2 Q 1
DC eld
c qr
22
Synchronous Machine
Electrical (inductor) equations: v d = ri + dt
d = L(r )i + dt 1 T dL(r ) Te = i i 2 dr Mechanical (Newtons) equations: J dr + Dr dt dr dt = Tm Te = r
23
Synchronous Machine
Stator equations: v0s rs vds = vqs
rs
Rotor equations: vF rF 0 = 0 0
ids r qs d ds dt qs rs iqs ds rQ1 rQ2 iF F
i0s
0s
rD
i D iQ1 iQ2
d D + dt Q1 Q2
24
Synchronous Machine
Magnetic ux equations:
0s ds qs F D Q1 Q2
L0 Ld Lq Md Md Mq Mq LF Md Md LD LQ1 Mq Mq LQ2 Md Md Mq Mq
i0s ids iqs iF iD iQ1 iQ2
25
Synchronous Machine
Transformation equations: v0s vas vbs = P T vds vqs vcs P = 2 cos r 3 sin r 1/ 2
ias
1/ 2 cos(r sin(r
2 3 ) 2 3 )
ibs = P T ids iqs ics 1/ 2 cos(r + sin(r +

2 3 ) 2 3 )
i0s
26
Synchronous Machine
Mechanical equations: 2 d 2 J r + Dr p dt p d r dt Te = Tm Te = r = p (iqs ds ids qs ) 2
27
Synchronous Machine
3-phase short circuit at generator terminals:

ia (t)
2|Ea (0)| x d
Subtransient component Transient component Steady state
2|Ea (0)| x d
2|Ea (0)| xd
Ea (0) is the open-circuit RMS phase voltage.
28
Synchronous Machine
Assuming balanced operation (null zero sequence), the detailed machine

model can be reduced to phasor models useful for stability and steady-state analysis.
Phasor models are based on the following assumptions: The rotor does not deviate much from the synchronous speed, i.e. r s = (2/p)2f0 . The rate of change in rotor speed is small, i.e. |dr /dt| 0
29
Synchronous Machine
Subtransient models capture the full machine electrical dynamics,

including the few initial cycles (ms) associated with the damper windings.
Transient models capture the machine electrical dynamics starting with

the eld and induced rotor core current transient response.
Damper windings transients are neglected. Steady-state models capture

the machine electrical response when all transients have disappeared after a few seconds.
30
Subtransient Model
External phase voltages and currents:

2 vas 2 2 = vqs + vds
vas i2 as ias
tan1
vds vqs ids iqs
= i2 + i2 qs ds = tan
1
31
Subtransient Model
Subtransient internal voltages associated with the damper windings (D and Q1 ): d eq dt d e dt d e vds d e vqs q = = 1 [eq + (x x )ids e ] q d d Td0 1 [ed (x x )iqs e ] q q d Tq0
= rs iqs x ids d
= rs ids + x iqs q
32
Subtransient Model
Transient internal voltages associated with the eld (F ) and rotor-core induced current windings (Q2 ): d eq dt d e dt d e vds d e vqs q = = 1 [ef + (xd x )ids e ] q d Td0 1 [(x x )iqs e ] q q d Tq0
= rs iqs x ids d = rs ids + x iqs q
33
Subtransient Model
Steady-state equations: ea vqs vds Mechanical equations: d r dt d dt = 1 [Pm vas ias cos(vas ias ) Dr ] M = rs iqs xd ids = rs ids + xq iqs
= r = r s
34
Subtransient Model
The subtransient reactances (x , x ) and open circuit time constants q d (Tq0 , Td0 ), as well as the transient reactances (x , x ) and open circuit q d time constants (Tq0 , Td0 ) are directly associated with the machine
resistances and inductances:
xd xq x d x q x d x q
= 0 Ld = x + xM d = 0 Lq = x + xM q xM d xLF = x + xLF + xM d xM q xLQ2 = x + xLQ2 + xM q xM d xLF xLD = x + xM d xLF + xM d xLD + xLF xLD xM q xLQ1 xLQ2 = x + xM q xLQ1 + xM q xLQ2 + xLQ1 xLQ2
Universidad de Castilla-La Mancha
June 26, 2008
Generator Modeling - 18
35
Subtransient Model
Some denitions: xF xD xQ1 xQ2 = xLF + xM d = xLD + xM d = xLQ1 + xM q = xLQ2 + xM q
36
Subtransient Model
Time constants:
Td0 Tq0 Td0 Tq0
= = = =
xF 0 rF xQ2 0 rQ2 xM d xLD 1 xLD + 0 rD xD 1 xM q xLQ1 xLQ1 + 0 rQ1 xQ1
37
Subtransient Model
Typical machine parameters:

2-pole Conventional Cooled Conductor Cooled 1.7-2.17 .264-.387 .23-.323 1.71-2.16 .245-1.12 .116-.332 4.8-5.36 .032-.059 .3-1.5 .042-.218 .27-.42 .00145-.00229 5-7 Conventional Cooled 1.21-1.55 .25-.27 .184-.197 1.17-1.52 .47-1.27 .12-.308 5.4-8.43 .031-.055 .38-1.5 .055-.152 .16-.27 .00146-00147 6-8 4-pole Conductor Cooled 1.6-2.13 .35-.467 .269-.32 1.56-2.07 .47-1.27 .12-.308 4.81-7.73 .031-.055 .36-1.5 .055-.152 .29-41 .00167-00235 6-8
xd x d x d xq x q x q Td0 Td0 Tq0 Tq0 x rs M
1.7-1.82 .18-.23 .11-.16 1.63-1.69 .245-1.12 .116-.332 7.1-9.6 .032-.059 .3-1.5 .042-.218 .118-.21 .00081-.00119 5-7
38
Subtransient Model
Typical machine parameters:

Salient-pole Dampers No dampers .6-1.5 .25-.5 .2-.5 .4-.8 Combustion Turbines 1.64-1.85 .159-.225 .102-.155 1.58-1.74 .306 .1 4.61-7.5 .054 1.5 .107 .113 .034 18-24 Synchronous Compensator 1.08-2.48 .244-.385 .141-.257 .72-1.18 .57-1.18 .17-.261 6-16 .039-.058 .15 .188-.235 .0987-.146 .0017-006 2-4
xd x d x d xq x q x q Td0 Td0 Tq0 Tq0 x rs M
.6-1.5 .25-.5 .13-.32 .4-.8
= x q
.135-.402 4-10 .029-.051
= x q
.135-.402 8-10 .029-.051
.033-.08 .17-.4 .003-.015 6-14
.033-.08 .17-.4 .003-.015 6-14
39
Subtransient Model
In practice most of these constants are determined from short-circuit

tests.
All the e voltages are internal machine voltages directly assocated with the internal phase angle . The internal eld voltage ef is directly proportional to the actual eld dc voltage vF , and is typically controlled by the voltage regulator. The mechanical power Pm is controlled through the governor.
40
Subtransient Model
A simple voltage regulator model (IEEE type 1):

Se vr max vref + vm 1 Tr s + 1 V + Ka Ta s + 1 vr min Kf s Tf s + 1 vr + 1 Te s + 1 vf
41
Subtransient Model
A static voltage regulator model:

Vc = |Vt | + (Rc + jXc )It VT 1 1+sTR IT Vuxl + + VS Vmin Vmax 1+sTC 1+sTB VC VA max 1+sTC1 1+sTB1 VA min sKF 1+sTF KA 1+sTA + Vef l vref + KIR 0 + |VT |VR max KC If Ef HV
Gate
If ref + If
|VT |VR min
42
Subtransient Model
A simple governor model (hydraulic valve plus turbine):

Torder ref + + 1/R + Tmin
Tin
Tmax Tin 1 Ts s + 1
Governor
T3 s + 1 Tc s + 1
Servo
T4 s + 1 T5 s + 1
Reheat
Tmech
43
Subtransient Model
A governor-steam turbine model:

w_ref K1 1.0 1 1+ sT_R 1 + sT_1 1 + sT_2 0.0 + 1 T_C1 1 1 + sT_C2 1 sT_C3 + P_ref
P_GV K_HP
+ +
1 1 + sT_HP + + F_IP + F_HP + P_MAX
P_mech
1 1 K_RH sT_RH
1 1+sT_IP
1 1 + st_LP F_LP
44
Example
For a 200 MVA, 13.8 kV, 60 Hz generator with the following p.u. data:
rs x xd xd xd

= = = = = = = = = = = = =
0.001096 0.15 1.7 Ld = 0.00451 Md = 0.00411 LF = 0.00438 LD = 0.00426
0.238324 0.184690 1.64 xq
xq xq xq

Lq = 0.00435 Mq = 0.00395 LQ2 = 0 LQ1 = 0.00405 rF = 0.000742 rD = 0.0131
0.185151 6.194876 0.028716 0
Td0 Td0 Tq0 Tq0 p
rQ2 = 0 rQ1 = 0.0540
0.074960
2 M = 10 D = 0
45
Example
A three-phase fault from open circuit conditions, i.e. before the fault vas = 13.8/ 3 kV and ias = 0, and after the fault vas = 0, is
simulated using the detailed machine equations:
vas , vbs ,vcs [kV]
1 0 1 2
0.02
0.04
0.06
0.08
0.1
0.12
0.14
0.16
0.18
0.2
t [s]
46
Example
10 9 8 7
iF [p.u.]
6 5 4 3 2 1
0.1
0.2
0.3
0.4
0.5
0.6
0.7
0.8
0.9
t [s]
47
Example
20
ia [p.u.]
0 20
0.1
0.2
0.3
0.4
0.5 t [s]
0.6
0.7
0.8
0.9
20
ib [p.u.]
0 20
0.1
0.2
0.3
0.4
0.5 t [s]
0.6
0.7
0.8
0.9
20
ic [p.u.]
20
0.1
0.2
0.3
0.4
0.5
0.6
0.7
0.8
0.9
t [s]
48
Transient Model
Obtained by eliminating electromagnetic differential equations and the

damper winding dynamic equations.
For this reason, the damping D in the mechanical equations, which is

typically a small value, is assumed to be large to indirectly model the signicant damping effect of these windings on r .
It is the typical model used in stability studies.
49
Transient Model
Neglecting the induced currents in the rotor (winding Q2 ):

x = xq Tq0 = 0 q
e = 0 d
This leads to the transient equations: d e dt a = = e a ea 1 [ef + (xd x )ids e ] d a Td0 1 [ef ea ] Td0 +jx ids ( + /2) + jxq iqs d = vas vas + rs ias ias +jxd ids ( + /2) + jxq iqs
= vas vas + rs ias ias
50
Transient Model
The phasor diagram in this case is: e a ea e a iqs vas jxd ids rs ias = vas + rs ias + jx ids + jxq iqs d = vas + rs ias + jxd ids + jxq iqs ea jxq iqs j(xd x )ids d
ids
ias
51
Transient Model
For faults near the generator terminals, the q axis has little effect on the system response, i.e. iqs 0. This results in the classical voltage source and transient reactance
generator model used in simple stability studies:
rs
jx d
ias ias +
e a
Pm
vas vas
52
Transient Model
A further approximation in some cases is used by neglecting the eld dynamics, i.e. Td0 = 0. In this case, e is a xed variable controlled directly through the voltage a regulator via ef . The limits in the voltage regulator are used to represent limits in the eld
and armature currents.
These limits can be soft, i.e. allowed to temporarily exceed the hard
steady-state limits, to represent under- and over-excitation.
53
Steady-state model
When all transient are neglected, the generator model becomes: ea ea ias ias = vas vas + rs ias ias +jxd ids ( + /2) + jxq iqs = ef = ids + jiqs
54
Steady-state model
For a round rotor machine (xd = x ), the steady-state model leads to d

the classical short circuit generator model:
rs
jxd
ias ias +
ea
Pm
vas vas
55
Steady-state model
Based on this simple model, the eld and armature current limits can be
used to dene the generator capability curves (for a given terminal voltage vas
= vt ):
Q ef limit Qmax
vt ea x
ia limit vt i a P Qmin Pmin

2 vt x
Pmax
56
Steady-state model
Considering the voltage regulator effect, the generator can be modeled

as a constant terminal voltage within the generator reactive power capability, delivering constant power (Pm ).
This yields the PV generator model for power ow studies:

P
where P
Qmin
change.
= Pm = constant, and V = vt = constant for Q Qmax ; otherwise, Q = Qmax,min and V is allowed to
57
Example
The generator data are:

va PG v = 10 j0.1
r (t) = 0 t + /2 + xd
Td0
= xq = 0.9, = 2s
x = 0.2 d
= large = constant = 0.5 at
The generator is operating in steady-state delivering PG
ea = 1.5. At t = 0 there is a fault and the line is disconnected. Find va (t) for t > 0.
58
Example
Steady-state conditions: ea = vas vas + (rs + jxd )ias ias = v + j(xd + xL )ias 1.5 1 = j(0.9 + 0.1) = 1.5 sin j(1.5 cos 1) = {v i } = {i } as as 1.5 sin sin1 (PG /1.5) = 19.47 0.65 39.64 = ias = =
ias PG
59
Example
iqs ids ea
= ias cos(ias ) = 0.33419.47
= ias sin(ias )( + /2) = 0.558 70.53 = vas + rs ias + jx ids + jxq iqs d = v + j(xd + xL )ias + jx ids + jxq iqs d j0.9(0.33419.47 ) = 1 + j0.1(0.65 39.64 ) + j0.2(0.558 70.53 )
d ea dt
= 1.11019.47 1 = [ea + (xd x )ids e ] d a Td0
60
Example
Transient: d 1 ea (t) = [ef + (xd x )ids e (t)] d a dt Td0 ef = ea steady-state = ids = e (0) = a d ea (t) = dt 1.5 0 open line 1.110 1 [1.5 e (t)] a 2
61
Example
Solution: d ea (t) = 0.5e (t) + 0.75 a dt e (t) = Ae0.5t + B a e (0) = A + B = 1.110 a e () = B = 0.75/0.5 = 1.5 a e (t) = 0.390e0.5t + 1.5 a
e a
va (t) = e (t) since ias = ids = iqs = 0 a

1.5 1.11 0 t
62
Transmission System Modeling
Transformers: Single phase: Detailed model Phasor model Three phase: Phase shifts Models
Universidad de Castilla-La Mancha June 26, 2008 Transmission System Modeling - 1
63
Transmission System Modeling
Transmission Lines: Single phase: Distributed parameter model Phasor lumped model Three phase: Distributed parameter model Reduced models Underground cables
64
Single-phase Transformers
The basic characteristics of this device are: Flux leakage around the transformer windings is represented by a leakage inductance L . The core is made of magnetic material and is represented by a magnetization inductance (Lm L ), but saturates.
Losses in the windings (Cu wires) and core (hysteresis and induced currents) are represented with lumped resistances (r and Gm ). Steps up or down the voltage/current depending on the turn ratio a = N1 /N2 = V1 /V2 .
65
leakage core
i1 + v1 N1 N2
i2 + v2
2 m
magnetizing
66
The equivalent circuit is: d di v = ri + = ri + L dt dt i r v 1 1 = 1 i2 r2 v2 i L + Lm Lm /a 1 d 1 + dt i2 Lm /a L2 + Lm /a2
67
Equivalent circuit:
i1 + v1 Lm r1 L1 im Gm i + e N1 : N2 a:1 ai + e/a L2 r2 i2 + v2
68
The phasor equivalent circuit is:
1 V1 = (Z1 + Zm )I1 + Zm I2 a 1 aV2 = Zm I1 + (a2 Z1 + Zm ) I2 a V Z + Zm Zm /a I 1 = 1 1 V2 Zm /a Z2 + Zm /a2 I2
69
Phasor equivalent circuit:

I1 + V1 Z1 Im Ym aV2 a:1 a2 Z2 I2 /a + I2 + V2
70
This can also be readily transformed into a ABCD input-output form based on the following approximation, since Zm Z1 (Z1 a2 Z2 ): V1 I1 = a(1 + Z Ym ) Z /a V2 I2
aYm 1/a A B V2 = C C I2
= r1 + jX1 + a2 (r2 + jX2 )
71
Phasor equivalent circuit with the approximation Zm Z1 :

I1 + V1 Z Im Ym aV2 a:1 I2 /a + I2 + V2
72
Or form: Z = Y1 Y2 = = Z a (1 a) 1 Z 1 Z
(a2 Z Ym + a2 + a)
2
1 for Ym 0 (a a) Z
73
equivalent circuit:
I1 Z I2
+ V1 Y1 Y2
+ V2
74
Neglecting Zm (Ym is small given the core magnetic properties):

1. Time domain:
v1 i2
2. Phasor domain:
di1 + av2 = ri1 + L dt = ai1
V1 I2
= Z I1 + aV2 = aI1
75
Equivalent circuit neglecting Zm :

I1 + V1 aV2 a:1 Z + I2 = I1 /a + V2
76
Certain transformers have built-in Under-Load Tap Changers (ULTC). This is either operated manually (locally or remote controlled) or
automatically with a voltage regulator; the voltage control range is limited (
10%) and on discrete steps ( 1%).
The time response is in the order of minutes, with 1-2 min. delays, due to
ULTCs being implemented using electromechanical systems.
77
These are typically used to control the load voltage side, and hence are
used at subtransmission substations.
Nowadays, power electronic switches are used, leading to Thyristor

Controller Voltage Regulators (TCVR), which are faster voltage controllers and are considered Flexible AC Transmission systems (FACTS).
These types of transformers are modelled using the same transformer models, but a may be assumed to be a discrete controlled variable
through a voltage regulator with a dead-band.
78
Transformers with special connections and under-load tap changers can

also be used for phase shift control and are known as Phase Shifters.
These control the phase shift difference between the two terminal voltages within approximately 30 , thus increasing the power capacity
of a transmission line (e.g. interconnection between Ontario and Michigan).
Phase shifters are modeled using a similar model but the tap ratio is a
phasor as opposed to a scalar:
a = a A equivalent cannot be used in this case.

79
Phasor equivalent circuit with complex tap ratio a:

I1 + V1 Z Im Ym aV2 a:1 I2 /a + I2 + V2
80
Single-phase Transformers (Example)
Single phase 8/80 kV, 30 MVA transformer with X = 10% and Xm 10X . Detailed model parameters: a = L1 = = L2 Lm = = 8/80 = 0.1 X1 X 0 20 0.1 (8 kV)2 = 0.283 mH 2 377 30 MVA L1 = 28.3 mH 2 a X 0.1 (8 kV)2 10 = 10 = 5.66 mH 0 377 30 MVA
81
Per unit parameters: Z Zm Ym = j0.1 = = 10Z = j1 1 = j1 Zm 8/8 kV =1 80/80 kV
a =
82
Per unit parameters:
A B
= =
C = D =
1(1 + j0.1(j1)) = 1.1 j0.1 = j0.1 1 1(j1) = j1 1 =1 1
83
Per unit parameters: j0.1 = j0.1 1 1 =0 (1 1) j0.1 1 = j1 (1 (j0.1)(j1) + 1 1) j0.1

2 2
Z = Y1 Y2 = =
84
Three-phase Transformers
a1
+ aVa2
T1 + Va2 a:1 T2 + Vb2 a:1 T3 + Vc2 a:1
a2
b1
+ aVb2
b2
c1
+ aVc2
c2
85
The 3 single-phase transformers form a 3-phase bank that induces a

phase shift, depending on the connection:
Vab1 Vab2 Vab1 a apu
= aVa2 330 Va2 = 3a30 Vab2 = 3a30 = 3apu 30 =
86
Y:
Y YY
a= apu =
3a30
: :
3apu 30 a = 3a 30 apu = 3apu 30 a=a a=a
87
In balanced, normal systems, the net phase shift between the

generation and load sides is zero, and hence is neglected during system analyses:
Generator side Load side
In these systems, the p.u. per-phase models of the transformers are

identical to the equivalent single-phase transformer models.
88
For smaller transformers (e.g. load transformers), integral designs are

preferred to transformers banks:
a b c
core
3 windows
vabc1 vabc2
r133 r233 L1133 L2133
iabc1
L1233
L2233
iabc2 i d abc1 dt iabc 2
89
Saturation
The magnetization inductance Lm changes with the magnetization

current due to saturation of the magnetic core.
Saturation occurs due to a reduction on the number of free magnetic

dipoles in the enriched core.
This results in the core behaving more like air than a magnet, i.e.
magnetic conductivity decreases.
It is typically represented using a piece-wise linear model.
90
Saturation
The magnetization inductance Lm changes with the magnetization

current due to saturation of the magnetic core:
m Lm1 Lm2
ims
im
Lm (im ) =
m1
for
Lm2 for im > ims
im ims
91
Single-phase Transmission Line
Alossless line can be represented using a series of lumped elements: l = D 0 ln 2 R 20 ln(D/R)
c = D R R 0 0 = =
distance between wires wire GMR wire radius
4 107 [H/m]
8.854 1012 [F/m]
92
i1
+
l [H/m]
+ +
i2
+
dv
di v
v1
c [F/m]
v2
dx
93
The equations for this line are: dv di = ldx di v i = l dt x t i v dv = c = cdx dt x t
These are DAlambert equations with solution: i1 (t) = 1 1 v1 (t) i2 (t ) + v2 (t ) Zc Zc

I2 (t )
i2 (t)
1 1 v2 (t) i1 (t ) + v1 (t ) Zc Zc
I1 (t )
June 26, 2008
Transmission System Modeling - 32
94
where: Zc = l/c chracteristic impedance 1 wave speed lc d travelling time for line length s
r/2 i2 + Zc v2
s = =
Distributed parameter equivalent circuit:

i1 + v1 Zc r/2
I2 (t )
I1 (t )
95
Example: t=0
+ +
i1
Trans. Line
i2 = 0
+
v1
v2
v1 (t) = E i2 (t) = 0
96
Example: t 0 2 3 4 5 I1 (t ) 0 2E/Zc 2E/Zc 0 0 2E/Zc I2 (t ) 0 0 2E/Zc 2E/Zc 0 0 i1 E/Zc E/Zc E/Zc E/Zc E/Zc E/Zc i2 0 2E 2E 0 0 2E
97
Example: i1 E/Zc
2 3 4 5 6
E/Zc v2 2E
98
Phasor model from the distributed line model: d V = (r + jl)I = zI dx d I = (jc)V = yI dx 0 z V d V = dx I y 0 I
99
The solution to this set of linear dynamical equations is: cosh d Zc sinh d V2 d V1 = dx I2 1/Zc sinh d cosh d I1
where
ed + ed cosh d = 2 ed ed sinh d = 2 = zy propagation constant Zc = z/y

characteristic impedance
100
This can be converted into the equivalent circuit:

I1 Z I2
+ V1
Y1 /2 Y2 /2
+ V2
101
equivalent circuit: Z Y
= =
sinh d zd d
Z
tanh d yd d
Y
for for
d < 250 km d < 100 km
Z Z
Z Z
Y Y Y 0
102
Three-phase Transmission Line

d n d n
2R a b d d c
dd 11111111111111111111 00000000000000000000 11111111111111111111 00000000000000000000 11111111111111111111 00000000000000000000 11111111111111111111 00000000000000000000

103
Typically the phase wires are bundled (e.g. 4 wires/phase) and the guard
wires are grounded at every tower, i.e. correspond to the neutral.
For a total of N wires, the per unit length equations are:

v1 6 7 6 7 6 v2 7 6 7 x 6 . 7 6 . 7 4 . 5 vN 2 3 = 2 r11 6 6r 6 21 6 6 . 6 . 4 . rN 1 2 l11 6 6l 6 21 6 6 . 6 . 4 . lN 1 r12 r22
. . .
rN 2 l12 l22
. . .
lN 2
i1 76 7 . . . r2N 7 6 i2 7 76 7 76 7 . 76 . 7 .. . 76 . 7 . . 54 . 5 . . . rN N iN 3 2 3 i1 . . . l1N 7 6 7 . . . l2N 7 6 i2 7 7 6 7 7 6 7 . 7 . .. . 7 t 6 . 7 6 . 7 . . 5 4 5 . . . lN N iN ... r1N
32
104

i1 6 7 6 7 6 i2 7 6 7 x 6 . 7 6 . 7 4 . 5 iN 2 3 = 2 g11 6 6g 6 21 6 6 . 6 . 4 . gN 1 2 p11 6 6p 6 21 6 6 . 6 . 4 . pN 1 p1 N N g12 g22
. . .
... ...
.. .
gN 2 p12 p22
. . .
... ... ...

.. .
pN 2
...
v1 76 7 g2N 7 6 v2 7 76 7 76 7 . 76 . 7 . 76 . 7 . 54 . 5 vN gN N 31 2 3 p1N v1 7 6 7 6 7 p2N 7 7 6 v2 7 7 6 7 . 7 t 6 . 7 . 7 6 . 7 . 5 4 . 5 pN N vN g1N
32
cN N
105
In phasor form, these equations are:

V1 6 7 6V 7 d 6 27 6 7 6 . 7 dx 6 . 7 4 . 5 VN 2 3 I1 6 7 6 7 d 6 I2 7 6 7 dx 6 . 7 6 . 7 4 . 5 IN 2 3 = z11 6 6z 6 21 6 . 6 . 6 . 4 zN 1 2 y11 6 6y 6 21 6 . 6 . 6 . 4 yN 1 2 z12 z22
. . .
... ...
.. .
zN 2 y12 y22
. . .
... ... ...

.. .
yN 2
...
I1 76 7 z2N 7 6 I2 7 76 7 76 7 . 76 . 7 . 76 . 7 . 54 . 5 zN N IN 32 3 y1N V1 76 7 7 6V 7 y2N 7 6 2 7 76 7 . 76 . 7 . 76 . 7 . 54 . 5 yN N vN z1N
32
d V = [z]I dx
d I = [y]V dx
106
The image method is useful for computing line parameters:

11 2Ri 00 11 00 11 00
dij
111 000 2Rj 111 000 111 000 111 000
hi
ij
Dij
hj [m]
Earth
hj hi j
image
i
image
107
The line parameters zij and yij can be computed using Carsons
formulas:
rii rij rij
= riint + rii = = rij
riint from tables
+b2 [(c2 ln a)a2 cos 2 + a2 sin 2]
4104 {/8 b1 a cos
+b6 [(c6 ln a)a6 cos 6 + a6 sin 6] a =
+b3 a3 cos 3 d4 a4 cos 4 b5 a5 cos 5 +b7 a7 cos 7 d8 a8 cos 8 b9 a9 cos 9 + . . .} 4 5104 D f /

June 26, 2008
108
Dij for i = j 2 b1 = 6 s bk = bk2 k(k + 2) c2 = 1.3659315 +1 k = 1, 2, 3, 4, 5, . . . s= 1 k = 5, 6, 7, 8, 13, . . .

June 26, 2008
D=
2h
for i
=j
1 b2 = 16 d k = bk 4
0 for i = j = ij for i = j
1 1 ck = ck2 + + k k+2
Universidad de Castilla-La Mancha Transmission System Modeling - 47
109
xii xij xij
0 2hi = ln + xiint + xii 2 Ri 0 Dij = ln + xij 2 dij =
xiint from tables( 0)
+b3 a3 cos 3 b4 [(c4 ln a)a4 cos 4 + a4 sin 4] +b5 a5 cos 5 d6 a6 cos 6 + b7 a7 cos 7 b8 [(c8 ln a)a8 cos 6 + a8 sin 8] + . . .} 1 2Dij pij = ln 20 dij
4104 {1/2(0.6159315 ln a) b1 a cos d2 a2 cos 2
gij pii
= gij 0 2hi 1 ln = 20 Ri
110

f (Hz)
2 4 6 8 10 20 40 60 80 100 200 400 600 800 1000 2000
Rac /Rdc
L /L ac dc
0.99992 0.99970 0.99932 0.99879 0.99812 0.99254 0.97125 0.93898 0.89946 0.85639 0.66232 0.47004 0.38503 0.33418 0.29924 0.21204
f (Hz)
4000 6000 8000 10000 20000 40000 60000 80000 100000 200000 400000 600000 800000 1000000 2000000 4000000
Rac /Rdc
L /L ac dc
0.15008 0.12258 0.10617 0.09497 0.06717 0.04750 0.03879 0.03359 0.03004 0.02124 0.01502 0.01227 0.01062 0.00950 0.00672 0.00475
1.0002 1.0007 1.0015 1.0026 1.0041 1.0164 1.0632 1.1347 1.2233 1.3213 1.7983 2.4554 2.9421 3.3559 3.7213 5.1561
7.1876 8.7471 10.0622 11.2209 15.7678 22.1988 27.1337 31.2942 34.9597 49.3413 69.6802 85.2870 98.4441 110.0357 155.5154 219.8336
111
The [z] parameters depend on the line frequqnecy , i.e. this is a

frequency dependent model.
These number of conductors, and hence equations, can be reduced

based on the following observations:
The voltage of all Nb conductors in a phase bundled are at the same voltage (e.g. v1 = v2 = = vN b = va ). The current in each phase is shared approximately equally by each conductor in the bundle (e.g. i1 = i2 = = iN b = ia /Nb ). The voltage in the guard wires is zero (e.g. vg = 0).
112
This reduces the matrices to: zaa zab zac yaa yab yac zab zbb zbc yab ybb ybc zac
[z]N N
[y]N N
zbc zcc yac ybc ycc
d Vabc dt d Iabc dt
= [zabc ]Iabc = [yabc ]Vabc
113
A line is transposed to balance the phases. The length of the barrel B must be much less than the wavelength (s/f 5000 km @ 60 Hz). B 50 km.
B a b c 1 2 3 3 1 2 2 3 1
B/3
B/3
B/3
zaa [[zabc ] = zab zac

June 26, 2008
zab zbb zbc
zac
zbc zcc
yaa [yabc ] = yab yac
yab ybb ybc
yac
ybc ycc
114
The phasor equations can be diagonalized using eigenvalue (modal)

analysis techniques:
T = TI [z]N N TI (diagonal matrix)
[zm ] [ym ] Vm Im
1 T = TV V = TI V 1 T = TI I = TV I
T = TV [z]N N TV
(diagonal matrix)
There are N modes of propogation, one for each eigenvalue.
115
Diagonalization of a 3-phase transposed line through sequence

transformation:
1 = TS Vabc 1 = TS Iabc
V0pn I0pn 1 1 1 3 1 1 a
2
TS
a =
1120
a a2
1 TS
1 1 = 1 3 1
1 a a2
a a
2
116
Transformation of zabc into z0pn :

1 [z0pn ] = TS [zabc ]TS zs + 2zm 0 = 0 zs zm 0 0
zp = zn
z0 3zp
0 z0 0 =0 zs zm 0
0 zp 0
0 zn
1 Similar for [y0pn ] = TS [yabc ]TS .
117
Diagonalization of a 3-phase transposed line through 0

transformation:
V0 I0
= T 1 Vabc
= T 1 Iabc 2 0 1 1 1 1/2 T = 3/2 3 1 1/ 2 3/2
T 1 = T T
118
Transformation of zabc into z0 : [z0 ] = T T [zabc ]T zs + 2zm = 0 0 0 z0 0 =0 zs zm 0
0 zs zm 0
0 z 0
z = z = zp
z0 3z
0 z
Similar for [y0 ] = T T [yabc ]T .
119
These simplications lead to the following per-phase (positive sequence),

per-unit length formulas:
r l
= =
c =
rtables Nb 0 Dm ln 2 Rb 20 ln Dm Rb
120
where: Dm is the GMD of the 3 phases: Dm =

3
dab dac dbc
Rb is the GMR of the bundled and wires: Rb =

Nb
R d12 d13 d1Nb
Rb is the GMR of the bundled: Rb =

Nb
Rd12 d13 d1Nb
121
Example 1
24.14 km transposed distributed line:

70.68 mH
Bus 1
15 miles
Bus 2
Bus 3
0.741 325/ 3 kV
1 ms
70.16 mH 219.1 mH 251.2
R0 = 0.3167 /km
R1 = 0.0243 /km
L = 3.222 mH/km 0
C0 = 0.00787 F/km
L = 0.9238 mH/km 1
C1 = 0.0126 F/km
122
Example 1
2 187.79 cos(377t) 6 7 6187.79 cos(377t 2/3)7 4 5 187.79 cos(377t + 2/3) 3 = 2 3 2 32 3 vBU S1a 0.714 0 0 i1 6 7 6 7 6 a7 6 vBU S1 7 + 6 0 0.714 0 7 6 i1b 7 4 4 54 5 b5 vBU S1c 0 0 0.714 i1c 2 32 3 0.07068 0 0 i1 7 6 a7 d 6 6 7 6 i1 7 0 0.07068 0 4 54 b5 dt 0 0 0.07068 i1c 32 3 2 vBU S1a 1 1 1 7 76 1 6 7 6 vBU S1 7 6 2 1/ 2 4 1/ 2 5 4 b5 p p 3 0 3/2 3/2 vBU S1c | {z }
T 1
+ 2 3 =
vBU S10 6 7 6vBU S1 7 5 4 vBU S1 2 3
i1 6 07 6i1 7 4 5 i1
3 2 i1 6 a7 1 6 7 T 4 i1b 5 i1c
123
Example 1
i10 (t) = i20 (t) = r0 ZC0 0 = = =
r0 1 vBU S10 i10 (t) I20 (t 0 ) Zc0 2 1 r0 vBU S20 i20 (t) I10 (t 0 ) Zc0 2 0.3167 24.14 = 7.6451 3.222 103 = 639.85 0.00787 106
24.14 3.222 103 0.00787 106 = 0.12156 ms
124
Example 1
i1 (t) = i2 (t) = r ZC = = =
r 1 vBU S1 i1 (t) I2 (t ) Zc 2 1 r vBU S2 i2 (t) I1 (t ) Zc 2 0.0243 24.14 = 0.5866 0.9238 103 = 270.77 0.0126 106
24.14 0.9238 103 0.0126 106 = 0.08236 ms
125
Example 1
i1 (t) i2 (t) r ZC
1 r vBU S1 i1 (t) I2 (t ) Zc 2 r 1 vBU S2 i2 (t) I1 (t ) = Zc 2 = r = = ZC =
126
Example 1
2 3 vBU S2a 6 7 6 vBU S2 7 4 b5 vBU S2c 2 3 i1 6 a7 6 i1 7 4 b5 i1c 3 2 1 1 6 61 34 1 | 2 2 1/ 2 1/ 2 {z
T
2 vBU S2a 6 7 6 vBU S2 7 4 b5 vBU S2c
i2 6 07 T 6i2 7 4 5 i2 2 32 3 251.2 0 0 i2 6 7 6 a7 6 0 251.2 0 7 6 i2b 7 4 54 5 0 0 251.2 i2c 2 3 32 0.28926 0 0 i2 7 6 a7 d 6 6 7 6 i2 7 0 0.28926 0 4 54 b5 dt 0 0 0.28926 i2c
vBU S20 76 7 7 6vBU S2 7 3/2 5 4 5 p vBU S2 3/2 } 0 p
32
127
Example 1
300
vaBUS2 vasource
200
100
[kV]
100
200
300
0.002 0.004 0.006 0.008
0.01
0.012 0.014 0.016 0.018
0.02
t [s]
128
Example 1
300
200
100
[kV]
100 200
300
vaBUS2 vbBUS2 vcBUS2

0 0.002 0.004 0.006 0.008 0.01 0.012 0.014 0.016 0.018 0.02
400
t [s]
129
Example 2
50 n 28 8 n
1 a b 45 110 c
11111111111111111111 00000000000000000000 11111111111111111111 00000000000000000000 11111111111111111111 00000000000000000000

40 40
130
Example 2
Line data: 4 Drake wires per phase (Nb = 4) 927 kcmil = 469.8 mm2 ACSR, 24 Al/13 steel, 3 layers R = 0.554 in = 1.407 cm R = 0.425 in = 1.080 cm rdc = 0.1032 /mile = 0.0645 /km rac@25 C = 0.1061 /mile = 0.0663 /km rac@100 C = 0.1361 /mile = 0.0851 /km d = 200 km f = 60 Hz = 377 rad/s
131
Example 2
r Dm
Rb
= = = = = =
Rb l c
rac@100 C = 0.02134 /km Nb 3 45 45 90 = 17.27 m 4 1.080 30.48 30.48 43.11 cm = 0.1442 m 4 1.407 30.48 30.48 43.11 cm = 0.1541 m 17.27 7 = 0.9573 103 H/m 2 10 ln 0.1442 28.854 1012 = 0.01179 H/m 17.27 ln 0.1541
132
Example 2
z = r + jl = y = = = Zc = = 0.361586.62 /km 4.444 106 90 S/km zy 0.0021788.31 km1 z y 285.21 1.69 = jc
133
Example 2
= D = cosh d ed + ed = 2 = 0.96860.11
= Zc sinh d = = ed ed Zc 2 71.4686.65 1 sinh d Zc 8.787 104 90.03 S
C = =
134
Example 2
Z = zd = Y = 72.3086.62 = yd 8.89 104 90 S sinh d = Z d = 71.3386.65 Z tanh d/2 = Y d/2 = 9.032 104 89.95 S Y
135
Cables
Steel pipe (lled with insulating oil) Skid wires Metallic tapes Paper/oil insulation Screen Conductor (stranded copper)
11111111111 00000000000 11111111111 00000000000 11111111111 00000000000 11111111111 00000000000 11111111111 00000000000 11111111111 00000000000 11111111111 00000000000 11111111111 00000000000 11111111111 00000000000 111 000 11111111111 00000000000 111 000 11111111111 00000000000 111 000 11111111111 00000000000 11111111111 00000000000 1111 0000 11111111111 00000000000 1111 0000 11111111111 00000000000 1111 0000 1111111 0000000 11111111111 00000000000 1111111 0000000 11111111111 00000000000 111 000 1111111 0000000 11111111111 00000000000 111 000 1111111 0000000 11111111111 00000000000 111 000 1111111 0000000 11111111111 00000000000
136
Cables
Single-phase and three-phase cables:
111111111 000000000 111111111 000000000 111111111 000000000 111111111 000000000 111111111 000000000 111111111 000000000 111 000 111111111 000000000 111 000 111111111 000000000 111 000 111111111 000000000 111 000 111111111 000000000 111 000 111111111 000000000 111111111 000000000 111111111 000000000 111111111 000000000 111111111 000000000 111111111 000000000
conductor
sheath
SF6 gas
111111111 000000000 111111111 000000000 111 000 111111111 000000000 111 000 111111111 000000000 111 000 111111111 000000000 111 000 111111111 000000000 111 000 111111111 000000000 111111111 000000000 11 11 00 00 111111111 000000000 11 11 00 00 111111111 000000000 11 11 00 00 111111111 000000000 11 11 00 00 111111111 000000000 11 11 00 00 111111111 000000000 111111111 000000000 111111111 000000000 111111111 000000000
137
Cables
Model for a single-phase cable:
VC
VS
11111111 00000000 111111 000000 11111111 00000000 111111 000000 11111111 00000000 111 000 111111 000000 11111111 00000000 111 000 111111 000000 11111111 00000000 111 000 111111 000000 11111111 00000000 111111 000000 11111111 00000000
A
1111111 0000000 1111111 0000000 1111111 0000000 1111111 0000000 11111111 00000000 1111111 0000000 11111111 00000000 1111111 0000000 11111111 00000000 1111111 0000000 1111111 0000000 11111111 00000000 11111 00000 1111111 0000000 1111111 0000000 11111111 00000000 11111 00000 1111111 0000000 11111111 00000000 1111111 0000000 11111 00000 1111111 0000000 11111111 00000000 11111 00000 1 0 1111111 0000000 11111111 00000000 11111 00000 1 0 11111111 00000000 1111111 0000000 11111 00000 11111111 00000000 1111111 0000000 11111 00000 11111111 00000000 1111111 0000000 11111 00000 11111111 00000000 1111111 0000000 11111111 00000000 1111111 0000000 11111111 00000000 1111111 0000000 V
I1 I2 I3
138
Cables
Model for a single-phase cable: V1 V2 V3 IC IS IA = Vcore Vsheath = VC VS = Vsheath Varmour = VS VA = VA = I1 = I1 I2 = I2 I3
139
Cables
Loop equations: V1 z11 z12 0 I1
d V2 = z12 z22 z23 I2 dx V3 I3 0 z23 z33 V1 y1 0 0 I1 d I2 = 0 y2 0 V2 dx V3 I3 0 0 y3
140
Cables
The elements of the impedance matrix z arecomputed as follows: z11 z12 z22 z23 z33 = zC ext + zCS + zS int = zS mut = zS ext + zSA + zAint = zAmut = zAext + zAE + zE
141
Cables
These impedances are calculated using the formulas:

11111111 00000000 11111111 00000000 11111111 00000000 11111111 00000000 q 11111111 00000000 11111111 00000000 11111111 00000000 11111111 00000000 11111111 00000000 11111111 00000000 11111111 00000000 r 11111111 00000000 11111111 00000000
zext zint zmut zins
m [I0 (mr)K1 (mq) + K0 (mr)I1 (mq)] 2rD m [I0 (mq)K1 (mr) + K0 (mq)I1 (mr)] = 2qD = 2qrD 0 r 1 = y = j 2 q =
June 26, 2008
142
Cables
= I1 (mr)K1 (mq) + K1 (mr)I1 (mq) j
m =
Where In and Kn are modied Bessel functions, as follows: In (x) Kn (x) = =

k=0
1 k!(n + k + 1)
x 2
n+2k
In (x) In (x) 2 sin(n)
143
Cables
The Gamma function (n) is dened as follows: (n) =

0
ex xn1 dx
(n + 1) = n(n) (n + 1) = n! for n = 1, 2, 3, . . . For asymmetric cables a nite element method is needed to compute
these impedances.
144
Cables
From the relations between loop voltages/currents and node

voltage/currents, the node equations are:
d VS = zCS dx VA zCA
e.g.
VC
zCC
zCS zSS zSA
zCA zSA zAA
IC
d dx
zCC = z11 + 2z12 + z22 + 2z23 + z33 IC y1 y1 0 VC = y1 y1 + y2 IS y2 VS IA 0 y2 y2 + y3 VA

IS IA
June 26, 2008
145
Cables
For a three-phase cable made of 3 single-phase cables: d [V] = [z]99 [I] dx d [I] = [y]99 [V] dx [zaa ]33 [z] = [zab ]33 [zac ]33 [yaa ]33 [y] = 0 0
June 26, 2008
[zab ]33 [zbb ]33 [zbc ]33 0 [ybb ]33 0
[zac ]33 [zbc ]33 [zcc ]33 0 0 [ycc ]33
146
Load Modeling
RLC models. Induction motors. Detailed models. Phasor models. Aggregated models: Impedance models. Power models. Induction motor power models.
Universidad de Castilla-La Mancha June 26, 2008 Load Modeling - 1
147
Load Classication
By demand level: Residential: lighting and heating (RL + controls); AC (motor + controls); appliances (small motors + controls). Commercial: similar types of devices as residential. Industrial: motor drives (induction and dc motor-based mostly); arc
furnaces; lighting; heating; others (e.g. special motor drives).
By type: RLC + controls. Drives: ac/dc motors + electronic controls. Special (e.g. arc furnace).
148
Remarks on Load Classication
Most controls are implemented using a variety of power electronic

converters.
Aggregate load models are necessary at the transmission system

modeling level.
Only large loads can be represented with their actual models.
149
RLC Loads, Resistor
Ideal, linear resistors (R), inductors (L) and capacitors (C ). Resistor: i

+
R v v = Ri
Time domain: Phasor domain: V = RI
150
RLC Loads, Inductor
Inductor: i
+
L v di v=L dt
Time domain:
Phasor domain: V = jLI = ZL I
151
RLC Loads, Inductor
The inductor time domain model can be discretized using the trapezoidal
integration method as follows:
i = ik+1
1 L
vdt
t = ik + (vk+1 + vk ) 2L t t = vk+1 + (ik + vk ) 2L 2L

hk
152
RLC Loads, Capacitor
This yields the equivalent resistive circuit:

vk+1
+
ik+1 hk
2L t
153
Capacitor: i
+
C v
Time domain: dv i=C dt Phasor domain: V 1 I C = ZC I = j
154
The capacitor time domain model can be discretized using the trapezoidal
integration method as follows:
v vk+1 ik+1
1 C
idt
t = vk + (ik+1 + ik ) 2C 2C 2C = vk+1 (ik + vk ) t t

hk
155
This yields the equivalent resistive circuit:

vk+1
+
ik+1 hk
t 2C
156
Induction Motor
Arbitrary 0dq reference:

d c s ar 0 cr bs br cs d as q r
r
b s r
a s
157
Induction Motor
Electrical (inductor) equations: d [abcs ] dt d [vabcr ] = [rabcr ][iabcr ] + [abcr ] dt i Lsr(r ) L abcs abcs = abcs Lsr (r ) Labcr abcr iabcr [vabcs ] = [rabcs ][iabcs ] +
L(r ) [i]
Te
1 T d[L(r )] [i] [i] 2 dr
158
Induction Motor
Mechanical (Newtons) equations: d J r + dr dt d r dt Stator transformation equations: vas v0s vds = Ks vbs vcs vqs = Tm Te = r
ias i0s ids = Ks ibs ics iqs
159
Induction Motor
Where the transformation matrix Ks is as follows: 1/2 1/2 1/2 2 sin sin( 2/3) sin( + 2/3) Ks = 3 cos cos( 2/3) cos( + 2/3) 1 sin cos 1 1 sin( 2/3) cos( 2/3) Ks = 1 sin( + 2/3) cos( + 2/3)
160
Induction Motor
Rotor transformation equations: var v0r vdr = Kr vbr vcr vqr
iar i0r idr = Kr ibr icr iqr
161
Induction Motor
Where the transformation matrix Kr is as follows: 1/2 1/2 1/2 2 sin sin( 2/3) sin( + 2/3) Kr = 3 cos cos( 2/3) cos( + 2/3) 1 sin cos 1 Kr = 1 sin( 2/3) cos( 2/3) 1 sin( + 2/3) cos( + 2/3) = r
162
Induction Motor
Stator equations: v0s rs vds = 0 vqs 0
0 rs 0
Rotor equations referred to the stator: 2 a rr 0 0 0 0 i0r 0r rr d i + ( r ) + 0 = 0 qr dt dr rr 0 dr 0 iqr qr dr 0 0 rr a= Ns Nr
ids + qs + d ds 0 dt qs rs iqs ds
i0s
0s
163
Induction Motor
Magnetic ux equations: Lls 0 0s 0 Lls + M ds Ls 0 0 qs = 0 0 0r dr 0 M qr 0 0
0 0 Ls 0 0 M
0 0 0 L lr 0 0
0 M 0 0 L + M lr
L r
0 i0s i ds M iqs 0 i 0r 0 i dr iqr L r
164
Induction Motor
Mechanical equations: 2 d 2 J r + Dr p dt p d r dt d dt Te = Tm Te = r = = 3p (iqs ds ids qs ) 22
165
Induction Motor
Equivalent circuit representation of these equations:

iqs + vqs rs + Lls L lr +
rr
Ns : Nr
iqr
ds M
( r ) dr vqr
166
Induction Motor
ids + vds rs + Lls L lr +
rr
Ns : Nr
idr
qs M
( r ) qr vdr
i0s + v0s
rs
Lls
L lr
rr
Ns : Nr
i0r
v0r
167
Induction Motor
Assuming a balanced, fundamental frequency ( = 0 ) system, the model can be reduced to a p.u. transient model (3rd order model): Vas Ias d ER dt d E dt I = VasR + jVasI = IasR + jIasI 1 = 0 EI [ER + (xS x )IasI ] T0 1 = 0 ER [EI + (xS x )IasR ] T0
168
Induction Motor
VasR ER
VasI EI d dt
Te TL
= rs IasI + x IasR 1 = (TL Te + D0 D) H 1 (ER IasR + EI IasR ) = 0 = f ()
= rs IasR x IasI
169
Induction Motor
where: xs x
T0
0 r slip = 0 = xls + xm stator reactance x xm = xls + lr transient reactance xlr + xm = x + xm lr 0 rr

open circuit transient time constant
170
Induction Motor
If the electromagnetic transients are neglected, the system can be

reduced to the following quasi-steady state equivalent circuit model: jx jxls rs Iar rr Ias lr
+ Vas jxm
rr 1
171
Induction Motor
Or equivalently:
Ias + Vas jxm
1 rr
rs
jxls
jx lr
Iar
Plus mechanical equations.
172
Induction Motor
This simplication can be justied from the point of view of the

torque-speed characteristic:
Te Te max
Steady state characteristic
Transient start up
r c
=1
173
Induction Motor
If the mechanical dynamics are ignored, the slip becomes a xed value,
and hence the equivalent circuit can be reduced to a simple equivalent reactive impedance, i.e. a Z load.
For loads with multiple IMs, a equivalent motor model can be used to
represent these motors.
Neglecting the motor dynamic equations in large or equivalent aggregate

motors can lead to signicant modeling errors, as the transient model and mechanical time constants can be on the same range as the generator time constants.
174
Induction Motor
Double-cage IMs can be modeled by introducing an additional inductance

on the rotor side, which leads to a subtransient model.
Since rotor cores are laminated, eddy currents do not play a signicant
role on the system dynamics.
Controls can also be modeled in detail by representing the converters

with ideal electronic switches plus their control systems.
For balanced, fundamental frequency systems, approximate equivalent

models of the IM and its controls can be readily implemented.
175
Induction Motor
For example, an IM with voltage-fed eld oriented control can be

represented by:
dr dt diqs dt
P (Te TL ) 2J
L r rs + M 2 rr L r 2 M rr iqs ids + qr Lr
M L r dr + r vqs dids dt =
L r rs + M 2 rr L r 2 M rr ids + iqs + dr Lr
M L + r qr + r vds
176
Induction Motor
d qr dt d dr dt Te dTref dt
rr rr M = ( r ) qr + iqs dr Lr Lr rr rr M = ( r )qr dr + ids Lr Lr 3pM = (iqs ids ) dr qr 4L r
= Ls L M 2 r = kq3
pTL 3p2 M iqs dr 2J 8JL r
+ kq4 (ref r )
177
Induction Motor
duqs dt
rr M 3pM kq1 rr M 2 [ iqs dr ids iqs + dr r 4Lr Lr Lr
L r rs + M 2 rr L r
iqs dr
L r dr uqs ]
+kq1 kq3
pTL 3p2 M iqs dr 2J 8JL r

aLr
+kq2 Tref 2pM iqs dr duds dt +kq1 kq4 (ref r ) = kd1

M rr rr ids dr Lr Lr
+ kd2 (ref ) dr
178
Example
A 21 MW load at 4 kV and 60 Hz is made of: An inductive impedance load with G = 0.06047, B = 0.03530. An aggregated induction motor model with rs = 0.07825, xls = 0.8320, rr = 0.1055, x = 0.8320, xm = 16.48. lr This data is all in p.u. on a 100 MVA, 4 kV base.
179
Example
The Z load model is then: PL PZ PIM 21 MW = 0.21 = 100 MVA = PZ + PIM

2 = VL G = 0.06
= 0.15
180
Example
ZIM
jxM (rr / + jx ) lr = rs + jxls + rr / + j(xM + x ) lr
= 0.07825 + j0.832 + =
13.7114 + j1.7386/ 0.1055/ + j17.312 28.652/ 0.07825 + 0.01113 2 + 299.71
0.18342/ 2 + 237.37 +j 0.832 + 0.01113/ 2 + 299.71 (0.00087/ 2 + 28.652/ + 23.452) + j(0.19628/ 2 + 486.73) = 0.01113/ 2 + 299.71
181
Example
PIM
2 = VL GIM = GIM
0.15 = ZIM YL ZL = = = =
(0.01113/ 2 + 299.71)(0.00087/ 2 + 28.652/ + 23.452) (0.00087/ 2 + 28.652/ + 23.452)2 + (0.19628/ 2 + 486.73)2 0.0191 (by trial-and-error) 4.6221 + j3.0742 1 (G + jB) + ZIM 1 = 3.3654 + j2.1597 YL
182
Impedance Models
Ignoring fast and slow transients, certain loads can be represented

using an equivalent impedance.
A Z load model is typically used for a variety of dynamic analysis of

power systems.
Using these load models, an equivalent impedance can be readily

obtained for all loads connected at a particular bus at the transmission system level.
ULTCs are used to connect distribution systems (subtransmission and LV

systems and the loads connetced to these) to the transmission system to control the steady state voltage on the load side.
183
Power Models
Hence, for slow dynamic analysis of balanced, fundamental frequency

system models:
2
PL QL VL
2 = VL GL = PL0
VL VL0 VL VL0
This is typically referred as a constant P Q model.
2 = VL BL = QL0 P L PL0 VL0 QL QL0
184
Power Models
Load recovery of certain loads (e.g. thermostatic) with respect to voltage

changes can be modeled as:
Nps Npt
dx(t) dt PL (t)
VL (t) x(t) + PL0 = Tp VL0 = VL (t) x(t) + PL0 Tp VL0
PL0
VL (t) VL0
Npt
dy(t) dt
y(t) VL (t) + QL0 = Tq VL0 y(t) VL (t) + QL0 Tq VL0
Nqs
QL0
VL (t) VL0
Nqt
Nqt
QL (t) =
185
Power Models
This results in the following time response:

P PL0
tf V VL0
tf
Universidad de Castilla-La Mancha June 26, 2008
Load Modeling - 40
186
Example
Identication of LD1 -LD4 loads at paper mill in Sweden:

Net 894 MVA, 30 kV
1 2
LD1 LD2 3 LD3 LD4 4 G1 6 G2 5
187
Example
Measurements vs. model:
188
Power Models
Certain loads have been shown to behave in steady state as follows: PL QL

= KP VL P fLP PL0
VL VL0 VL VL0
KQ VL Q fLQ
PL0
If f f0 , then the following approximation holds: PL QL PL0 PL0 VL VL0 VL VL0
189
Power Models
Load Filament lamp Fluorescent lamp Heater Induction motor (half load) Induction motor (full load) Reduction furnace Aluminum plant
P
1.6 1.2 2.0 0.2 0.1 1.9 1.8
Q
0 3.0 0 1.6 0.6 2.1 2.2
P
0 -0.1 0 1.5 2.8 -0.5 -0.3
Q
0 2.8 0 -0.3 1.8 0 0.6
190
Power Models
ZIP model: PL QL Jimmas model: PL QL = P LZ = QLZ VL VL0 VL VL0

2
= P LZ = QLZ
VL VL0 VL VL0
+ P LI
2
VL VL0 VL VL0
+ P LP + QLP
+ QLI
+ P LI
2
VL VL0 VL VL0
+ PL P + QLP dVL + KV dt
+ QLI
191
Induction Motor Power Models
Walves model: PL = Kpf f + Kpv VL + T dVL dt
QL
dVL PL0 + Kpv (VL VL0 ) + T dt = Kqf f + Kqv VL QL0 + Kqv (VL VL0 )
192
Induction Motor Power Models
Mixed model: PL
= Kpf f + Kpv VL + Tpv
dVL dt dVL dt
PL0 + Kpv (VL VL0 ) + Tpv QL

= Kqf f + Kqv VL + Tqv
dVL dt
QL0 + Kqv
dVL (VL VL0 ) + Tqv dt
193
Power Flow Outlines
Power Flow: System model. Equations. Solution techniques: Newton-Raphson. Fast decoupled.
Universidad de Castilla-La Mancha June 26, 2008 Power Flow Outlines - 1
194
Power Flow Model
The steady-state operating point of a power system is obtained by solving

the power ow equations.
Power ow system model corresponds to the steady state model. Generator: Generates and injects power P in the system while keeping the output voltage V constant within active and reactive power limits
(capability curve):
Pmin P Pmax Qmin Q Qmax

195
Power Flow Model
Thus, it is modeled as a P V bus:

P = constant Q = unknown
V = constant = unknown
When Q reaches a limit it becomes a P Q bus:

P = constant Q = constant
V = unknown = unknown
196
Power Flow Model
Slack bus: The phasor model needs a reference bus. A large generator is typically chosen as the the reference bus, as it
should be able to take the power slack:
P = unknown Q = unknown
V = constant =0
Pslack =
L
PL + Plosses
PG
G
197
Power Flow Model
Hence, if Q reaches a limit:

P = unknown Q = constant
V = unknown =0
198
Power Flow Model
Load: Loads are typically connected to the transmission system through

ULTC transformers.
Thus, most loads in steady state represent a constant power demand in the system, and hence are modeled as a P Q bus:
P = constant Q = constant
V = unknown = unknown
199
Power Flow Model
Transmission system: AC transmission lines and transformers in steady state are basically
modeled using the following model:
Si Sk Z Ik /a Ik Y1 Y2 aVk Vk
Vi
Ii
a : 1
200
Power Flow Model
Hence: Si = V i I i
= Vi (Vi aVk )
1 Z
Y
+Vi Y1
= Vi (Y + Yi ) Vi + (aY) Vk
Yii Yik
Similarly for Sk .
201
Power Flow Model
Thus, for an N bus system interconnected through an ac transmission system, an N N bus admittance matrix can be dened: I Y 1 11 I Y 2 21 . . . . . . = Ii Yi1 . . . . . . IN YN 1 I = Ybus
node injections
Y12 Y22
. . .
.. .
Y1i Y2i
. . .

.. .
.. .

.. .
Yi2
. . .
Yii
. . .
YN 2 V
YN i
V 1 Y2N V2 . . . . . . YiN Vi . . . . . . YN N VN
Y1N
node voltages
202
Power Flow Model
where: 1 Yii = N Yi k=1 Zik + = sum of all the Ys connected to node i = Yij = 1 Zij = negative of the Y between nodes i and j
Ybus
203
Power Flow Equations
In steady state, a system with n generators G and m loads L can be

modeled as:
SG1 VG1
Transmission System
SL1 VL1
Ybus (N N ) SGn VGn N =n+m SLm VLm
204
Hence the power injections at each node are dened by: Si = Pi + jQi = V i I i
N
= Vi
k=1
Yik Vk N
= Vi i
k=1
(Gij jBij )Vk k

for generator buses for load buses
Si
Gi
SL i
205
This yields two equations per node or bus: Pi (, V, Pi ) =

N
Pi
k=1
Vi Vk [Gik cos(i k ) + Bik sin(i k )] = 0
Qi (, V, Qi ) =
N
Qi
k=1
Vi Vk [Gik sin(i k ) Bik cos(i k )] = 0
And two variables per node: PQ buses Vi and i PV buses i and Qi
Slack buses Pi and Qi

206
These equations are referred to as the power mismatch equations. The equations are typically subjected to inequality constraints
representing control limits:
0.95 Qmini
Vi
1.05 for all buses
Qi Qmaxi for generator buses
207
In summary, typical power ow data are as follows:

Bus PQ PV slack Parameters Variables
P, Q P, V V,
V, Q, P, Q
208
In constrained power ow analysis, standard buses can degenerate if

some limits is reached: Bus PQ PV slack Parameters Variables
P , Vmax,min P , Qmax,min Qmax,min ,
Q, V, P, V
Continuation power ow techniques are by far more accurate and robust

than standard power ow analysis if limits are taken into account.
209
The slack bus can be single or distributed. This refers to losses. For single slack bus model, all system losses are cleared by the
slack bus.
For distributed slack bus model losses are shared (equally or

proportionally) among all or part of the generators:
Continuation power ow techniques are by far more accurate and robust

than standard power ow analysis if limits are taken into account.
210
The distributed slack bus model is based on a generalized power center

concept.
This is practically obtained by including in power ow equations a variable kG and rewriting the system active power balance as follows:
nG i nP
(1 + kG i )PGi
PLi Plosses = 0
The parameters i allow tuning the weight of the participation of each

generator to the losses.
For single slack bus model, i = 0 for all generators but one.
211
Power Flow Solution
The power ow equations can be represented as F (z) = 0 There are 2 equations per bus, with 2 known variables and 2 unknown variables per bus; the problem is of dimension 2N . Since these equations are highly nonlinear due to the sine and cosine
terms, Newton-Raphson (NR) based numerical techniques are used to solve them.
212
Power Flow Solution
A robust NR technique may be used to solve these equations:

1. Start with an initial guess, typically Vi0
0 Pslack = 0 (at start). 0 = 1, i = 0, Q0 = 0, Gi
2. At each iteration k(k matrix:
= 0, 1, 2, . . .), compute the sparse Jacobian

F1 z1 |z k F1 zN |z k
...
.. .
F z
zk
= Jk =
. . .
. . .

2N 2N
FN z1 |z k
...
FN zN |z k
213
Power Flow Solution

3. Find z k by solving the following linear set of equations (the sparse matrix Jk is factorized and not inverted to speed up the solution process):
Jk z k = F (z k )
4. Computes the new guess for the next iteration, where is a step control constant to guarantee convergence (0
< < 1):
z k+1 = z k + z k
214
Power Flow Solution

5. Stop when:
F (z k+1 ) = max |Fi (z k+1 )| This is basically the technique used in M ATLABs fsolve() routine,
based on either numerical or actual Jacobians.
215
Power Flow Solution
If only the unknown bus voltage angles and magnitudes are calculated
using NRs method (the generator reactive powers and active slack power are evaluated later):
H N
Jk z k = F (z k ) M P ( k , V k ) k = L V k /V k Q( k , V k )
216
Power Flow Solution
Where: H M N = = = P (k ,V k ) P V V (k ,V k ) Q (k ,V k ) Q V V (k ,V k )
L =
217
Power Flow Solution
Assuming: (i k ) < 10 , then cos(i k ) sin(i k ) 1 i k
The resistance in the transmission system are small, i.e. R X , then Gij Bik . The M and N matrices may be neglected, and: H M V k BV k
V k B V k
218
Power Flow Solution
Thus the linear step equations may be decoupled and reduced to: B k B V k = P ( k , V k )/V k
= Q( k , V k )/V k
where B is the imaginary part of the Ybus matrix, and B is the imaginary part of the admittance matrix obtained by ignoring the system resistances, i.e.
B = B .
219
Power Flow Solution
The fast Decoupled iterative method is then dened as follows:

0 1. Start with an initial guess, typically i
= 0, Vi0 = 1.
2. Solve for k
3. Update k+1 4. Solve for V k 5. Update
B k = P ( k , V k )/V k . = k + k . B V k = Q( k , V k )/V k .
V k+1 = V k + V k .
6. Compute unknown generator powers and check for limits. 7. Repeat process for k
= 1, 2, . . ., until convergence.
220
Power Flow Solution
This technique requires a relatively large number of iterations as

compared to the robust NR method.
It is signicantly faster, as there is no need to re-compute and re-factorize

the Jacobian matrix every iteration.
It is sensitive to initial guesses and there is no guarantee of convergence,

particularly for systems with large transmission system resistances.
221
Example
For the following system:

1 2 All lines: 200 MVA 138 kV
P1
3
P2
X = 0.1 p.u. B = 0.2 p.u.
200 MVA 0.9 p.f. lagging
V1 = 1, 1 = 0, V2 = 1, V3 = 1, and P2 = P1 /2. Determine the voltage phasor angles 2 and 3 and the shunt Q by
solving the PF equations.
222
Example
The Ybus matrix is: 2 B +2 j = j19.8 jX 2 1 Y12 = Y13 = Y23 = = j10 jX 19.8 10 10 B11 B12 B13 10 = j B21 B22 B23 Ybus = j 19.8 10 10 10 19.8 B31 B32 B33 Y11 = Y22 = Y33 =
223
Example
Mismatch equation P1 :
3
P1 0
= P1 = P1
k=1 3
V1 Vk [G1k cos(1 k ) + B1k sin(1 k )] B1k sin(k )
k=1
= P1 + B12 sin(2 ) + B13 sin(3 ) = P1 + 10 sin(2 ) + 10 sin(3 )
224
Example
Mismatch equation Q1 :
3
Q1 0
= Q1 = Q1 +
k=1 3
V1 Vk [G1k sin(1 k ) B1k cos(1 k )] B1k cos(k )
k=1
= Q1 19.8 + 10 cos(2 ) + 10 cos(3 )
225
Example
3
P2 0
= P2
k=1
V2 Vk [G2k cos(2 k ) + B2k sin(2 k )]

3
= P1 /2
k=1
B2k sin(2 k )
= 0.5P1 + 10 sin(2 ) + 10 sin(2 3 )
226
Example
3
Q2 0
= Q2 = Q2 +
k=1 3
V2 Vk [G2k sin(2 k ) B2k cos(2 k )] B2k cos(2 k )
k=1
= Q1 + 10 cos(2 ) 19.8 + 10 cos(2 3 )
227
Example
3
P3 0
= P3
k=1
V3 Vk [G3k cos(3 k ) + B3k sin(3 k )]

3
= 0.9
k=1
B3k sin(3 k )
= 0.9 + 10 sin(3 ) + 10 sin(3 2 )
228
Example
3
Q3 0
= Q3 = Q
k=1
V3 Vk [G3k sin(3 k ) B3k cos(3 k )]

3
1 0.92 +
k=1
B3k cos(3 k )
= Q 0.436 + 10 cos(3 ) + 10 cos(3 2 ) 19.8
229
Example
Thus 6 equations and 6 unknowns, i.e. 2 , 3 , P1 , Q1 , Q2 , and Q, can be solved using M ATLABs fsolve() routine:
>> global lambda >> lambda = 1; >> z0 = fsolve(@pf_eqs,[0 0 0 0 0 0], optimset(Display,iter)) First-order Trust-region Iteration Func-count f(x) optimality radius 0 7 0.945696 18 1 1 14 0.000661419 0.705901 0.0205 1 2 21 9.98637e-18 0.0257331 2.52e-09 1.76 Optimization terminated: first-order optimality is less than options.TolFun. z0 = -0.0100 -0.0500 0.6000 -0.1870 -0.1915 0.2565 Norm of step
230
Example
Where lambda is used to simulate constant power factor load changes and pf eqs.m is:
function F = pf_eqs(z) global lambda d2 d3 P1 Q1 Q2 Q = = = = = = z(1); z(2); z(3); z(4); z(5); z(6); = = = = = = P1 + 10*sin(d2) + 10*sin(d3); Q1 - 19.8 + 10*cos(d2) + 10*cos(d3); 0.5*P1 - 10*sin(d2) - 10*sin(d2-d3); Q2 + 10*cos(d2) - 19.8 + 10*cos(d2-d3); -0.9*lambda - 10*sin(d3) - 10*sin(d3-d2); Q - 0.436*lambda + 10*cos(d3) + 10*cos(d3-d2) - 19.8;
F(1,1) F(2,1) F(3,1) F(4,1) F(5,1) F(6,1)
231
Example
Observe that as the load (lambda) is increased, convergence problems

show up:
lambda = 20; z0 = fsolve(@pf_eqs,[0 0 0 0 0 0], optimset(Display,iter)) Norm of step First-order optimality 360 143 Trust-region radius 1 1
Iteration Func-count f(x) 0 7 396.67 1 14 246.008 1 . . . 9 70 2.68094e-07 0.179729 0.000311 25.9 10 77 7.43614e-16 0.00127159 1.64e-08 25.9 Optimization terminated: first-order optimality is less than options.TolFun. z0 = -0.2501 -1.2613 12.0000 7.0658 4.8031 20.1667
232
Example
Convergence problems develop until the equations fail to converge:

lambda = 22; z0 = fsolve(@pf_eqs,[0 0 0 0 0 0], optimset(Display,iter)) Norm of step First-order optimality 396 168 Trust-region radius 1 1
Iteration Func-count f(x) 0 7 480.33 1 14 310.93 1 . . . 90 595 0.00622095 0.0211144 0.0322 0.0211 91 602 0.00621755 0.0211144 0.0155 0.0211 Maximum number of function evaluations reached: increase options.MaxFunEvals. z0 = -0.3322 -1.7225 13.1600 11.8633 8.5479 29.1072
233
Stability Concepts Outlines
Basic stability concepts Nonlinear systems: Ordinary differential equations (ODE) Differential algebraic equations (DAE) Equilibrium points: Denitions. Stability: Linearization. Eigenvalue analysis. Stability regions.
Universidad de Castilla-La Mancha June 26, 2008 Stability Concepts Outlines - 1
234
ODE Systems
Nonlinear systems are represented by a nonlinear set of differential

equations:
x = s(x, p, )
where
x n state variables (e.g. generator angles) uncontrollable parameters (e.g. loads)
p k controllable parameters (e.g. compensation) s() n nonlinear functions (e.g. generator equations)
235
ODE Systems
For example, for a simple generator-system model:

Generator
PG + jQG jx G jxL
PL + jQL jxth
System
V1 1
V2 2
V 0
Innite bus (M
= )
AVR
The generator is modeled as a simple d axis transient voltage behind transient reactance.
236
ODE Systems
The generator has only the mechanical dynamics:

where
= = r 0 1 (PL PG D) = M EV sin x + xL + xth G V1 V sin 1 xL + xth
PG
= =
237
ODE Systems
If the AVR is modeled, V1 may be assumed to be kept constant by varying E , with the generators reactive power within limits: V12 V1 V QG = cos 1 xL + xth xL + xth QG min QG QG max
238
ODE Systems
If the AVR is not modeled: x = [, ]T state variables p = [E , V ]T controlled parameters = PL uncontrolled parameters Hence, assuming p = [1.5, 1]T , M = D = 0.1, and x = 0.75: x = x 1 2 s(x, p, ) = x2 = 10 = 20 sin x1 x2
These are basically the pendulum equations.
239
DAE Systems
Differential Algebraic Equation (DAE) models are dened as: x = f (x, y, p, ) 0

where:
= g(x, y, p, )
y m algebraic variables (e.g. load voltages)
f () n nonlinear differential equations (e.g. generator equations) g() m nonlinear algebraic equations (e.g. reactive power
equations)
240
DAE Systems
For example, for the generator-innite bus example with AVR, for QG min QG QG max : 0 0 0 = 1 EV = PL sin D M x V1 V EV sin 1 sin = x xL + xth V1 E V12 = QG + cos(1 ) xG xG V12 V1 V = QG + cos 1 xL + xth xL + xth
241
DAE Systems
Thus, for: x = [, ]T y = [E , 1 , QG ]T p = [V1 , V ]T = [1, 1]T = PL M = D = 0.1, x = 0.75, x = 0.25, xL + xth = 0.5 G x = x 1 2 f (x, y, p, ) = x2 = 10 13.33y1 sin x1 x2 0 = 1.333y1 sin x1 2 sin y2 g(x, y, p, ) = 0 = y3 4 + 4y1 cos(y2 x1 ) 0 = y3 2 + 2 cos y2
242
DAE Systems
If QG = QG max or QG = QG min : x = [, ]T y = [E , 1 , V1 ]T p = [QG , V ]T = [0.5, 1]T = PL M = D = 0.1, x = 0.75, x = 0.25, xL + xth = 0.5 G x = x 1 2 f (x, y, p, ) = x2 = 10 13.33y1 y3 sin x1 x2 0 = 1.333y1 sin x1 2y3 sin y2 2 g(x, y, p, ) = 0 = 0.5 4y3 + 4y1 y3 cos(y2 x1 ) 2 0 = 0.5 2y3 + 2y3 cos y2
243
DAE Systems
If the Jacobian matrix Dy g = [gi /yi ]mm is nonsingular, i.e.

invertible, along the trajectory solutions, the system can be transformed into an ODE system (Implicit Function Theorem):
= h(x, p, ) = s(x, p, )
x = f (x, h(x, p, ), p, )
In practice, this is a purely theoretical exercise that is not carry out due
to its complexity.
The system model should be revised when Dy g is singular.
244
Equilibria
For the ODE system, equilibria are dened as the solution x0 for given parameter values p0 and 0 of the set of equations s(x0 , p0 , 0 ) = 0 There are multiple solutions to this problem, i.e. multiple equilibrium
points.
245
Equilibria
The stability of these equilibria is dened by linearizing the nonlinear system around x0 , i.e. x = si (x0 , p0 , 0 ) xj x x0
x
nn
= Dx s|0 x where Dx s|0 = Dx s(x0 , p0 , 0 ) = s/x|0 is the system Jacobian

matrix evaluated at the equilibrium point.
246
Equilibria
From linear system theory, the linear system stability is dened by the eigenvalues i of the Jacobian matrix Dx s|0 , which are dened as the
solutions of the equation:
Dx s|0 v Dx s|T w 0
= v = w
right e-vector left e-vector
an n + an1 n1 + . . . + a1 + a0 = 0
det(Dx s|0 In ) = 0
247
Equilibria
There are n complex eigenvalues, left and right eigenvectors associated with the system Jacobian Dx s|0 . In practice, these eigenvalues are not computed using characteristic
polynomial but other more efcient numerical techniques, as the costs associated with computing these values is rather large in realistic power systems.
248
Equilibria
These eigenvalues dene the small perturbation stability of the ODE

system, i.e. the local stability of the nonlinear system near the equilibrium points:
Stable equilibrium point (s.e.p.): The system is locally stable about x0 if all the eigenvalues i (Dx s|0 ) are on the left-half (LH) of the
complex plane.
Unstable equilibrium point (u.e.p.): The system is locally unstable about x0 if at least one eigenvalue i (Dx s|0 ) is on the right-half
(RH) of the complex plane.
249
Equilibria
The equilibrium point x0 is a bifurcation point if at least one eigenvalue i (Dx s|0 ) is on the imaginary axis of the complex plane. Some systems have equilibria with eigenvalues on the imaginary axis
without these being bifurcation points; for example, a lossless generator-innite bus system with no damping.
250
Equilibria
For example, for the simple generator-innite bus example with no AVR: x1 x2 = x2 = 10 20 sin x1 x2
the equilibrium points can be found from the steady-state (power ow) equations:
0 0
= x20 = 10 20 sin x10 x20
251
Equilibria
which leads to the solutions: sin1 (/2) x10 = 0 x20

repetitions of these three):
This yields basically three equilibrium points (other solutions are just s.e.p /2 < x1s < /2 u.e.p.1 x1u1 = x1s + u.e.p.2 x1u2 = x1s
sin1 (PL /2) 0 = 0 0
252
Equilibria
This corresponds to the intersections of PG = E V /x sin with PL :

Stable
EV /x
Unstable
PG PL
Bifurcation Unstable
2 (x2 )
s (xs ) /2 1 (x1 )
, (x1 )
253
Equilibria
The stability of these equilibria is determined using the system Jacobian: s /x1 |0 s1 /x2 |0 1 Dx s|0 = s2 /x1 |0 s2 /x2 |0 0 1 = 20 cos x10 1
254
Equilibria
det(Dx s|0 I2 )
= 2 + + 20 cos x10 = 0 PL = 1 1,2 1 1 1 80 cos x10 = 2 2
= det 20 cos x10
1,2 (Dx s|xs ) = 0.5 j4.132 3.192 1,2 (Dx s|xu1 /u2 ) = 5.192
Universidad de Castilla-La Mancha Stability Concepts Outlines - 22
June 26, 2008
255
Equilibria
For this system, the equilibria are: stable if PG >0 unstable if PG <0 bifurcation point for PG =0 = /2
256
Equilibria
For DAE systems, the equilibria z0 = (x0 , y0 ) for parameter values p0 and 0 are dened as the solution to the nonlinear, steady state problem: f (x , y , p , ) = 0 0 0 0 0 F (x0 , y0 , p0 , 0 ) = 0 g(x0 , y0 , p0 , 0 ) = 0 In this case, the linearization about (x0 , y0 ) yields: x = Dx f |0 x + Dy f |0 y 0 = Dx g|0 x + Dy g|0 y
257
Equilibria
Hence, by eliminating y from these equations, one obtains: Dx s|0 = Dx f |0 Dy f |0 Dy g|1 Dx g|0 0 Observe that, as mentioned before, the nonsingularity of the Jacobian Dy g|0 in this case is required. The same local stability conditions apply in this case based on the eigenvalues of Dx s|0 .
258
Equilibria
For the generator-innite bus example with AVR example (within QG

limits), the steady-state solutions are obtained from solving the steady-state or power ow equations:
0 0 0 0 0
= x20 = 10 13.33y10 sin x10 x20 = 1.333y10 sin x10 2 sin y20 = y30 4 + 4y10 cos(y20 x10 ) = y30 2 + 2 cos y20
259
Equilibria
Which in M ATLAB format are:

function f = dae_eqs(z) global lambda x10 x20 y10 y20 y30 = = = = = z(1); z(2); z(3); z(4); z(5); = = = = = x20; 10*lambda - 13.33 * y10 * sin(x10) - x20; 1.333 * y10 * sin (x10) - 2 * sin(y20); -y30 - 4 + 4 * y10 * cos(y20 - x10); y30 - 2 + 2 * cos(y20);
f(1,1) f(1,2) f(1,3) f(1,4) f(1,5)
260
Equilibria
This generates the equilibrium point for = PL = 1:

>> lambda = 1; >> z0 = fsolve(@dae_eqs,[0 0 1 0 1],optimset(Display,iter)) Norm of step First-order Trust-region optimality radius 133 1 10.2 1 0.219 2.5 0.0534 2.5 4.89e-06 2.5 2.23e-13 2.5 is less than options.TolFun.
Iteration Func-count f(x) 0 6 102 1 12 1.52385 1 2 18 0.0050436 0.298687 3 24 2.72816e-05 0.0607967 4 30 2.0931e-13 0.000413599 5 36 4.98474e-28 5.13313e-08 Optimization terminated: first-order optimality z0 = 0.7539 0 1.0959 0.5236
0.2679
261
Equilibria
And the symbolic Jacobian matrices:

syms x1 x2 y1 y2 y3 real z = [x1 x2 y1 y2 y3]; F = dae_eqs(z); f = F(1:2); g = F(3:5); Dxf Dyf Dxg Dyg = = = = jacobian(f,[x1,x2]) jacobian(f,[y1,y2,y3]) jacobian(g,[x1,x2]) jacobian(g,[y1,y2,y3])
262
Equilibria
And the symbolic Jacobian matrices:

Dxf = [ 0, [ -1333/100*y1*cos(x1), Dyf = [ 0, [ -1333/100*sin(x1), Dxg = [ 1333/1000*y1*cos(x1), [ -4*y1*sin(-y2+x1), [ 0, Dyg = [ 1333/1000*sin(x1), [ 4*cos(-y2+x1), [ 0, 1] -1]
0, 0,
0] 0]
0] 0] 0]
-2*cos(y2), 4*y1*sin(-y2+x1), -2*sin(y2),
0] -1] 1]
263
Equilibria
These generate the following eigenvalues at the equilibrium point:

Dxf = [ 0, [ -1333/100*y1*cos(x1), Dyf = [ 0, [ -1333/100*sin(x1), Dxg = [ 1333/1000*y1*cos(x1), [ -4*y1*sin(-y2+x1), [ 0, Dyg = [ 1333/1000*sin(x1), [ 4*cos(-y2+x1), [ 0, 1] -1]
0, 0,
0] 0]
0] 0] 0]
-2*cos(y2), 4*y1*sin(-y2+x1), -2*sin(y2),
0] -1] 1]
264
Equilibria
These generate the following eigenvalues at the equilibrium point:

x1 = z0(1); x2 = z0(2); y1 = z0(3); y2 = z0(4); y3 = z0(5); A = vpa(subs(Dxf),5); B = vpa(subs(Dyf),5); C = vpa(subs(Dxg),5); D = vpa(subs(Dyg),5); Dxs = A - B * inv(D) * C; ev = vpa(eig(Dxs),5) ev = -.50000+3.5698*i -.50000-3.5698*i
Hence, this is a s.e.p.

265
Equilibria
A u.e.p. can be computed as well from these equations (neglecting QG

limits):
>> z0 = fsolve(@dae_eqs,[3 0 0.1 3 3],optimset(Display,iter)); z0 = 2.7465 >> >> >> >> >> >> >> 0 1.9491 2.6180 3.7321
x1 = z0(1); x2 = z0(2); y1 = z0(3); y2 = z0(4); y3 = z0(5); A = vpa(subs(Dxf),5); B = vpa(subs(Dyf),5); C = vpa(subs(Dxg),5); D = vpa(subs(Dyg),5); Dxs = A - B * inv(D) * C; ev = vpa(eig(Dxs),5)
ev = 4.2892 -5.2892
266
Equilibria
Associated with every s.e.p. xs there is a stability region A(xs ), which

basically corresponds to the region of system variables that are all attracted to xs , i.e.
x(t ) xs :
x1 (0) A(xs ) xs x1 (t)
stable
A(xs )
(stability region boundary)
x2 (0)
unstable
x2 (t)
267
Equilibria
if A(xs ) is known, the stability of a system for large perturbations can be

readily evaluated.
However, determining this region is a rather difcult task. This can realistically be accomplished only for 2- or 3-dimendional
systems using sophisticated nonlinear system analysis techniques.
268
Equilibria
= 0.2
10
x 0 x
2
u2
xs
u1
5 A(x )
s
10
15 10
0 x1
10
269
Equilibria
= 0.5
10
x 0 x
2
u2
u1
10
A(x )
s
15 10
0 x1
10
270
Equilibria
= 1.0
10
x 0 x
2
u2
u1
5 A(x )
s
10
15 10
0 x1
10
271
Equilibria
= 1.5
10
0 x
2
xs
u1
A(x ) 5
s
10
15 10
0 x1
10
272
Equilibria
= 1.8
10
x 0 x
2
u1
A(x )
s
10
15 10
0 x1
10
273
Equilibria
= 1.9
10
x 0 x
2
u1
A(x )
s
10
15 10
0 x1
10
274
Equilibria
In real systems, trial-and-error techniques are usually used: A contingency yields a given initial condition x(0). For the post-contingency system, the time trajectories x(t) can be
computed by numerical integration.
If x(t) converges to the post-contingency equilibrium point xs , the system is stable, i.e. x(0) A(xs ). If it diverges, the system is unstable.
275
Voltage Stability Outlines
Denitions. Basic concepts. Continuation Power Flow. Direct Methods. Indices. Controls and protections. Practical applications. Examples.
Universidad de Castilla-La Mancha June 26, 2008 Voltage Stability - 1
276
Voltage Stability Denitions
IEEE-CIGRE classication (IEEE/CIGRE Joint Task Force on Stability)

Terms and Denitions, Denitions and Classication of Power System Stability, IEEE Trans. Power Systems and CIGRE Technical Brochure 231, 2003:
Power System Stability
Rotor Angle Stability
Frequency Stability
Voltage Stability
Small Disturbance Angle Stability
Transient Stability
Large Disturbance Voltage Stability
Small Disturbance Voltage Stability
Short Term Short Term Long Term
Short Term
Long Term
277
Power system stability is the ability of an electric power system, for a

given initial operating condition, to regain a state of operating equilibrium after being subjected to a physical disturbance, with most system variables bounded so that practically the entire system remains intact.
Voltage stability refers to the ability of a power system to maintain steady

voltages at all buses in the system after being subjected to a disturbance from a given initial operating condition.
278
The inability of the trnasmission system to supply the demand leads to a

voltage collapse problem.
This problem is associated with the disappearance of a stable

equilibrium point due to a saddle-node or limit-induced bifurcation point, typically due to contingencies in the system (e.g. August 14, 2003 Northeast Blackout).
279
Voltage Stability Concepts
The concept of saddle-node and limit induced bifurcations can be readily

explained using the generator-load example:
PG + jQG jxL PL + jQL
V1 1
V2 2
280
Neglecting for simplicity losses, electromagnetic dynamics, and the transient impedance in the d-axis transient model, the generator can be
simulated with:
1 1
= 1 = r 0 1 = (Pm PG DG 1 ) M
281
The load can be simulated using the mixed models. For P , neglecting voltage dynamics (Tpv = 0) and voltage dependence ( = 0): PL PL 2 = Kpf f2 + Kpv [V2 + Tpv V2 ] = P d + DL 2 1 (PL Pd ) = 2 = DL
282
For Q, neglecting frequency dependence (Kqf = 0) and voltage dependence ( = 0) QL QL V2 = Kqf f2 + Kqv [V2 + Tqv V2 ] = Qd + V2 = 1 (QL Qd )
283
The transmission system yields the power ow equations: PL PG QL QG V1 V2 sin(2 1 ) XL V1 V2 sin(1 2 ) = XL V1 V2 V22 + cos(2 1 ) = XL XL V12 V1 V2 = cos(1 2 ) XL XL =
Finally, since the system is lossless: P m = Pd

284
Hence, dening:
the system equations are:
= 1 2 = 1 = 2 V1 V2 sin DG XL V1 V2 sin Pd XL
V2
1 M
Pd 1 DL
= = 1
V1 V2 V22 + cos Qd XL XL
285
Observe that these equations also represent a generator-dynamic load system with no AVR and with XL including XG , where V1 would stand for EG . The steady-state load demand may be assumed to have a constant
power factor, i.e.
Qd = kPd
If generator reactive power limits are considered, and neglecting XG ,
one has a DAE system.
286
For QGmin QG QGmax : V2 = 1 M Pd 1 DL V10 V2 sin DG XL V10 V2 sin Pd XL
= = 1
2 V10 V2 V10 + cos 0 = QG XL XL
with
x = [, , V2 ]T p = V10
y = QG = Pd
287
For QG = QGmin,max : V2 0
with
V1 V2 sin DG Pd = XL V1 V2 1 sin Pd = DL XL 1 M = 1 V22 V1 V2 + cos Qd XL XL V12 V1 V2 + cos XL XL y = V1 = Pd
= QGmin,max
x = [, , V2 ]T p = QGmin,max
288
All the equilibrium point for the system with and without limits can be
obtained solving the power ow equations:
0 0 0
V10 V20 sin 0 XL 2 V20 V10 V20 cos 0 = kPd + XL XL 2 V10 V10 V20 = QG0 + cos 0 XL XL = Pd
289
And the stability of these equilibrium points come from the state matrix: Without limits, or for QGmin QG QGmax : DG V10 V20 V10 M XL sin 0 M M XL cos 0 V10 V20 V10 1 Dx s|0 = DL XL cos 0 DL XL sin 0 10 V20 0 V XL sin 0 2 V20 + V10 cos 0 XL XL
290
For QG = QGmin,max : Dx s|0 = DG V10 V20 V10 M M XL cos 0 M XL sin 0 V10 V20 V10 1 DL XL cos 0 DL XL sin 0 10 V20 0 V XL sin 0 2 V20 + V10 cos 0 XL XL T V10 0 M XL sin 0 1 V10 V20 sin 0 V10 sin 0 2 V10 + V20 cos 0 D X XL L L XL XL V20 V10 cos 0 XL XL cos 0
291
Assume XL = 0.5, M = 1, DG = 0.01, DL = 0.1, = 0.01, k = 0.25. With the help of M ATLAB and the continuation power ow routine of PSAT, for the system without limits and V1 = V10 = 1, the power ow solutions
yield a PV or nose curve (bifurcation diagram):
292

1 0.9 0.8 0.7
s.e.p.
x3 = V2
0.6 0.5 0.4 0.3 0.2 0.1 0
SNB
u.e.p.
0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8
= Pd
293
The saddle-node bifurcation point SNB corresponds to a point where the state matrix Dx s|0 is singular (one zero eigenvalue). This is typically associated with a power ow solution with a singular PF Jacobian Dz F |0 . This is not always the case, as for more complex dynamic models, the
singularity of the state matrix does not necessarily correspond to a singularity o the PF Jacobian, and vice versa.
294
Observe that the SNB point corresponds to a maximum value max = Ps max 0.78, which is why is also referred to as the
maximum loading or loadability point.
For a load greater than Pd max , there are no PF solutions. This point is also referred to as the voltage collapse point.
295
Modeling a contingency at 0 = Pd0 = 0.7 by increasing XL = 0.5 0.6:

1 0.9 0.8 0.7
operating point
x3 = V2
0.6 0.5 0.4 0.3 0.2 0.1 0
contingency
XL = 0.5 XL = 0.6
0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8
= Pd
296
The dynamic solution yields a voltage collapse:

6
V2 V1
3 Operating point Contingency
0 Voltage collapse 1 0 0.2 0.4 0.6 0.8 1 1.2 1.4 1.6 1.8 2
t [s]
297
For the system with limits and QGmax,min = 0.5, the PV or nose
curve is:
1 0.9 0.8
s.e.p.
0.7
LIB
x3 = V2
0.6 0.5 0.4 0.3 0.2 0.1 0
u.e.p.
0.1
0.2
0.3
0.4
0.5
0.6
0.7
= Pd
298
In this case, the maximum loading or loadability point max = Pd max 0.65 corresponds to the point where the generator reaches its maximum reactive power limit QG = QG max = 0.5, and hence losses control of V1 . This is referred to a limit-induced bifurcation or LIB point. Beyond the LIB point, there are no more power ow solutions, due to
the limit recovery mechanism of the AVR.
In this case the LIB point is also a voltage collapse point.
299
Modeling a contingency at 0 = Pd0 = 0.6 by increasing XL = 0.5 0.6:

1 0.9 0.8 0.7
operating point contingency
x3 = V2
0.6 0.5 0.4 0.3 0.2 0.1 0
XL = 0.5 XL = 0.6
0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8
= Pd
300
The dynamic solution also yields a voltage collapse:

6
V2 V1
3 Operating point 2 Contingency
t [s]
301
Collapse problems can be avoided using shunt or series compensation:

V1 1
V2 2
jxC
302
In this case, the system equations are: For QGmin QG QGmax : V2 0

where
1 (Pd V10 V2 BL sin DG ) M 1 (V10 V2 BL sin Pd ) = DL 1 = [V22 (BL BC ) + V10 V2 BL cos kPd ] 2 = QG V10 BL + V10 V2 BL cos = BL = 1 XL BC = 1 XC
303
For QG = QGmin,max : V2 0 1 (Pd V10 V2 BL sin DG ) M 1 (V10 V2 BL sin Pd ) = DL 1 = [V22 (BL BC ) + V10 V2 BL cos kPd ] 2 = QGmin,max V10 BL + V10 V2 BL cos =
304
These equations generate the PV curves: Without limits:

1.5
XL = 0.6
XL = 0.5
1
V2
XL = 0.6
0.5
Bc = 0 Bc = 0 Bc = 0.5
0 0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1
= Pd
305
With limits:
1.5
XL = 0.6 XL = 0.5
1
V2
XL = 0.6
0.5
Bc = 0 Bc = 0 Bc = 0.5
0 0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1
= Pd
306
Applying compensation at t = 1.3 s without limits:
1.2 1
0.8 0.6
Apply compensation
Contingency 0.4
V2 V1
0.2 Operating point 0
0.2
10
t [s]
307
Applying compensation at t = 1.3 s with limits:
1.2 1
0.8 0.6
Apply compensation
Contingency 0.4
V2 V1
0.2 Operating point 0
0.2
10
t [s]
308
Continuation Power Flow
These PV or nose curves are obtained using a continuation power ow. This technique traces the solutions of the power equations F (z, p, ) = 0
as changes.
309
Continuation Power Flow
The algorithm steps are:

1. Predictor 2. Parametrization
(z1 , 1 ) 2 1 z1
3. Corrector
(z0 , 0 )
z2
These methods guarantee convergence.
310
CPF Predictor
Tangent vector method: Dz F |1 F dz d

t1
z1 =
k t1 t1
1
(z2 , 2 ) (z1 , 1 )
(z1 , 1 )
311
CPF Predictor
Good method to follow closely the PV curves, but relatively slow. The tangent vector t denes sensitivities at any power ow solution
point.
This vector can also be used as an index to predict proximity to a

saddle-node bifurcation, as opposed to using the smallest eigenvalue, which changes in highly non linear fashion.
312
CPF Predictor
Secant method: z1 1 = k(z1a z1b ) = k(1a 1b )

(z2 , 2 ) (z1 , 1 ) (z1b , 1b ) (z1a , 1a )
313
CPF Predictor
This method is faster but can have convergence problems with sharp
corners (e.g. limits):
(z2 , 2 ) (z1b , 1b ) (z1a , 1a )
314
CPF Parametrization
Used to avoid singularities during the predictor step. Methods: Local: interchange a zi z with , i.e. rotate the PV curve. Arc length (s): assume z1 (s) and 1 (s); thus solve for z1 and 1 : F Dz F |1 z1 + + 1
1
T z1 z1 + 2 1
= k
No real need for parametrization in practice if step cutting is needed.
315
CPF Corrector
The idea is to add an equation to the equilibrium equations, i.e. solve for (z, ) at a given point p: F (z, p, ) = 0 (z, ) = 0 These equations are nonsingular for the appropriate choice of ().
316
CPF Corrector
Perpendicular intersection technique:

(z1 , 1 ) (z1 , 1 ) z (z2 , 2 )
F (z, p, ) = 0 (z1 + z1 z)T z1 + (1 + 1 )1 = 0

317
CPF Corrector
Local parametrization technique:

(z1 , 1 ) (z1 , 1 ) z (z2 , 2 )
F (z, p, ) = 0 = 1 + 1 or zi = zi1 + zi1

318
Direct Methods
Allow to directly nd the maximum loading point (saddle-node or limit

induced bifurcation).
This problem can be set up as an optimization problem:
Max. s.t.
F (z, p, ) = 0 zmin z zmax pmin p pmax
319
Direct Methods
If limits are ignored, the solution for given values of the control paramters (p = p0 ) to the associated optimization problem, based on the
Lagrangian and KKT conditions, is given by:
F (z, p0 , )
T Dz F (z, p0 , )w T D F (z, p0 , )w
0 = 0
steady stae solution singularity condition
= 1
w = 0 condition
This yields a saddle-node bifurcation point, and corresponds to the left eigenvector (w ) saddle-node equations.
320
Direct Methods
The solution to these nonlinear equations do not converge if the

maximum loading point is a limit-induced bifurcation.
In this case, the otpimization problem

Max. s.t.
F (z, p0 , ) = 0 zmin z zmax
must be solved using standard optimization techniques (e.g. Interior Point Method).
The solution to this optimization problem may be a saddle-node or

limit-induced bifurcation.
321
Direct Methods
Observe that if the control parameters p are allowed to change, the

problem is transformed into a maximization of the maximum loading margin.
Other optimization problems can be set up to not only

maximize/guarantee loading margins but at the same time minimize costs (e.g. maximize social welfare).
322
Direct Methods
For example, for an OPF-based auction in electricity markets, the

following multi-objective optimization problem can be posed: Min. s.t.
T T G = 1 (CD PD CS PS ) 2 c
f (, V, QG , PS , PD ) = 0 cmin c cmax 0 PS PSmax 0 PD PDmax
PF equations loading margin Sup. bid blocks Dem. bid blocks
f (c , Vc , QGc , c , PS , PD ) = 0 Max load PF eqs.
323
Direct Methods
With the phyiscal and security limits: Iij (, V ) Iijmax Iji (, V ) Ijimax Iij (c , Vc ) Iijmax Iji (c , Vc ) Ijimax QGmin QG QGmax Gen. Q limits QGmin QGc QGmax Vmin V Vmax Vmin Vc Vmax
Thermal limits
V security lim.
324
Direct Methods
More information about this probem can be found in:

F. Milano, C. A. Canizares and M. Invernizzi, Multi-objective Optimization for PRicing System Security in Electricity Marktes, IEEE Trans. on Power Systems, Vol. 18, No. 2, May 2003, pp. 596-604.
325
Indices
Indices have been developed with the aim of monitoring proximity to

voltage collapse for on-line applications.
For example, the minimum real eigenvalue of the system Jacobian can be
used to measure proximity to a saddle-node bifurcation, since this matrix becomes singular at that point.
Many indices have been proposed, but the most popular/useful are: Singular value. Reactive power reserves.
326
Indices
The singular value index consists of simply monitoring the singular value of the Jacobian of the power ow equations as changes, e.g.
327
Indices
Observations: Computationally inexpensive. Highly nonlinear behavior, especially when control limits are reached. Cannot really be used to detect proximity to limit-induced bifurcation. Useful in some OPF-applications to help represent voltage stability as
a constraints.
328
Indices
The reactive-power-reserve index consists of monitoring the difference

between the reactive power generator output and its maximum limit, e.g. for a generator bus system:
329
Indices
Observations: Computationally inexpensive. Highly nonlinear behavior. Only works for the right generators, i.e. the generators associated
with the limit-induced bifurcation.
Cannot really be used to detect proximity to a saddle-node bifurcation.
330
Voltage Stability Applications
These concepts and associated techniques are applied to real power

systems through the computation of PV curves.
These curves are obtained either through a simple series of continuous

power ows or using actual CPFs.
In both cases, these nose curves are computed with respect to load
changes, which are dened as follows:
PL QL
= PL0 + PL = QL0 + QL
331
Generator power changes are then dened, with the exception of the
slack bus, as:
PG = PG0 + PG For a distributed slack bus model: PG = PG0 + ( + kG )PG

for all generators, where kG is a variable in the power ow equations replacing the variable power in the slack bus.
332
The Available Transfer Capability (ATC) of the tranmsission system is

typically obtained from (NERC denition): ATC
= TTC + ETC + TRM
The ATC can be associated to the maximum loading margin max of the
system if N-1 contingency criteria are taken into account.
333
TTC or Total Transfer Capability is the maximum loading level of the system considering N-1 contingency criteria, i.e. the max for the worst
realistic single contingency.
ETC or Existent Transmission Commitments represents the current

loading level plus any reserved transmission commitments.
TRM or Transmission Reliability Margin which is an additional margin

dened to represent other contingencies.
334
For example, for WECC, systems should be operated a minimum 5%

distance away from the maximum loadibility point when contigencies are considered:
335
Control and Protections
To improve system loadability, i.e. avoid voltage stability problems, the

most common control and protections techniques are:
Increase reactive power output from generators, especially in the

critical area (the area most sensitive to voltage problems).
Introduce shunt compesation through the use of MSC, SVC or

STATCOM (see slides 304 and 305).
Use undervoltage relays.
336
Secondary Voltage Regulation
One way of improving reactive power support is to coordinate the reactive

power outputs of generators.
The French and Italians typically referred to as secondary voltage

regulation or control.
The basic structure and controls are:
337
338
339
Observations: Control areas and associated pilot buses and controlled generators
must be properly identied.
Control hierarchy is important, i.e. PQR is slower than AVR, and

RVR is slower than PQR.
It is relatively inexpensive compared to shunt compensation solutions.
340
Undervoltage Relays
These relays are installed on sub-transmission substations (loads) to

shed during long duration voltage dips.
The idea is to reduce load demand on the system to increase the

loadability margin (see slide 334).
Operation is somewhat similarly to taps in a LTC: Discrete load shedding steps (e.g. 1-2% of total load). Activated with a time delay (e.g. 1-2 mins.) after the voltage dips
below certain values (e.g. 0.8-0.9 p.u.)
The lower the voltage, the faster and larger the load shed.
341
Example
For a 3-area sample system:

100 MW 150 MW 150 MW
60 MVAr Area 1
v3<d3
150 MW 56 MVAr
Bus 2
1.020
50 MVAr R = 0.01 p.u. X = 0.15 p.u.
V2 2
Bus 3
50 MVAr
50 MW 40 MVAr 100 MW
342
Example

Bus Name Area 1 Area 2 Area 3
PG
(p.u.) 1.5 0 0.5
PL
(p.u.) 0 1.5 0.5
QL
(p.u.) 0 0.56 0.40
Using UWPFLOW to obtain the system PV curves, for a distributed slack

bus model:
343
Example
Data le in EPRI format (3area.wsc):

HDG UWPFLOW data file, WSCC format 3-area example April 2000 BAS C C AC BUSES C C | SHUNT | C |Ow|Name |kV |Z|PL |QL |MW |Mva|PM |PG |QM |Qm |Vpu BE 1 Area 1 138 1 150 60 0 0 0 150 0 0 1.02 B 1 Area 2 138 2 150 56 0 50 0 100 0 0 1.00 B 1 Area 3 138 3 50 40 0 50 0 100 0 0 1.00
344
Example

C C AC LINES C C M CS N C |Ow|Name_1 |kV1||Name_2 |kV2|||In || R | X | G/2 | B/2 |Mil| L 1 Area 1 138 Area 2 1381 15001 .01 .15 L 1 Area 1 138 Area 3 1381 15001 .01 .15 L 1 Area 2 138 Area 3 1381 15001 .01 .15 C C SOLUTION CONTROL CARD C C |Max| |SLACK BUS | C |Itr| |Name |kV| |Angle | SOL 50 Area 1 138 0. END
345
Example
Generator and load change le (3area.k):

C C UWPFLOW load and generation "direction" file C for 3-area example C C BusNumber BusName DPg Pnl Qnl PgMax [ Smax 1 0 1.5 0.0 0.0 0 0 2 0 0.0 1.5 0.56 0 0 3 0 0.5 0.5 0.40 0 0
Vmax Vmin ] 1.05 0.95 1.05 0.95 1.05 0.95
346
Example
Batch le for UNIX (run3area):

echo -1- Run base case power flow uwpflow 3area.wsc -K3area.k echo -2- Obatin PV curves and maximum loading uwpflow 3area.wsc -K3area.k -cthreearea.m -m -ltmp.l -s echo - with bus voltage limits enforced uwpflow 3area.wsc -K3area.k -c -7 -k0.1 echo - with current limits enforced uwpflow 3area.wsc -K3area.k -c -ltmp.l -8 -k0.1
347
Example
Batch le for Windows (run3area.bat):

rem -1- Run base case power flow uwpflow 3area.wsc -K3area.k rem -2- Obatin PV curves and maximum loading uwpflow 3area.wsc -K3area.k -cthreearea.m -m -ltmp.l -s rem - with bus voltage limits enforced uwpflow 3area.wsc -K3area.k -c -7 -k0.1 rem - with current limits enforced uwpflow 3area.wsc -K3area.k -c -ltmp.l -8 -k0.1
348
Example
PV curves (threearea.m):
Profiles 150
100
50 kVArea 3 kVArea 2 kVArea 1 0 0 0.2 0.4 0.6 0.8 L.F. [p.u.] 1 1.2 1.4
138 138 138
1.6
max 1.6 p.u.

Universidad de Castilla-La Mancha Voltage Stability - 74
June 26, 2008
349
Example
Hence, if contingencies are not considered:

TTC
= = = = = =
PL0 + max 670 MW PL0 350 MW 0.05TTC 33.5 MW 286.5 MW
PL
ETC
TRM
ATC
350
Example
The singular value index obtained with UWPFLOW is as follows:

0.7
0.6
Full matrix sing. value index
0.5
0.4
0.3
0.2
0.1
0.2
0.4
0.6
0.8 L.F. [p.u.]
1.2
1.4
1.6
351
August 14 2003 Blackout
This information is extracted from the nal report of the US-Canada Joint
Task Force.
The full details of the nal report can be found on the Internet at:
https://reports.energy.gov/B-F-Web-Part1.pdf
The report is titled: U.S.-Canada Power System Outage Task Force.

Final Report on the August 14, 2003 Blackout in the United States and Canada: Causes and Recommendations. Washington DC: USGPO, April 2004.
352
3 Interconnections, 10 Regional Reliability Councils:
353
Regions and control areas of the North American Electric Reliability

Council (NERC):
354
Reliability coordinators (some are also system and market operators,

such as the IMO and PJM):
355
Reliability coordinator and system/market operators in the Ohio area,

where the system collapse started:
356
Four causes: Inadequate system understanding: FirstEnergy, ECAR Inadequate situational awareness: FirstEnergy Inadequate tree-trimming: FirstEnergy Inadequate diagnostic support: MISO, PJM
357
At 12:15 EDT, MISO began having problems with its state estimator; it did
not return to full functionality until 16:04.
Sometime after 14:14, FirstEnergy began losing its energy management

system (EMS) alarms but did not know it.
At 14:20, parts of FEs EMS began to fail rst remote sites, then core
servers but FE system operators did not know this and FE IT support staff did not tell them.
Without a functioning EMS, FE operators did not know their system was
losing lines and voltage until about 15:45.
358
FirstEnergy did not do sufcient system planning to know the

Cleveland-Akron area was seriously decient in reactive power supply needed for voltage support.
At 13:31 EDT, FE lost its Eastlake 5 unit, a critical source of real and
reactive power for the Cleveland-Akron area.
FE did not perform contingency analysis after this event.
359
Lost of the Eastlake 5 unit and beginning of reactive power problems.
360
Between 15:05 and 15:41 EDT, FE lost 3 345 kV lines in the

Cleveland-Akron area under normal loading due to contact with too-tall trees but did not know it due to EMS problems.
Line loadings and reactive power demands increased with each line loss. Between 15:39 and 16:08 EDT, FE lost 16 138kV lines in the
Cleveland-Akron area due to overloads and ground faults.
361
Line outages:
362
At 16:05.57 EDT, FE lost its Sammis-Star 345 kV line due to overload. This closed a major path for power imports into the Cleveland-Akron area
and initiated the cascade phase of the blackout.
363
The tipping point in Ohio:
364
Effect of a line trip: increasing loading on other lines.
365
Effect of a line trip: decreasing voltages leading to voltage collapse.
366
Cascade: Denition: A cascade on an electric system is a dynamic, unplanned

sequence of events that, once started, cannot be stopped by human intervention.
Power swings, voltage uctuations and frequency uctuations cause

sequential tripping of transmission lines, generators, and automatic load-shedding in a widening geographic area.
The uctuations diminish in amplitude as the cascade spreads.

Eventually equilibrium is restored and the cascade stops.
367
The system goes haywire:
368
Higher voltage lines are better able to absorb large voltage and current
swings, buffering some areas against the cascade (AEP, Pennsylvania).
Areas with high voltage proles and ample reactive power were not
swamped by the sudden voltage and power drain (PJM and New England).
After islanding began, some areas were able to balance generation with
load and reach equilibrium without collapsing (upstate New York and southern Ontario).
369
Sequence of events:
370
Sequence of events:
371
Sequence of events:
372
Sequence of events:
373
When the cascade was over at 4:13 pm, as many as 50 million people in
the northeast U.S. and the province of Ontario had no power.
This is a good example of a voltage instability problem triggered by a

series of contingencies.
374
375
Had the system properly monitored and NERC recommended operating

guidelines followed, the system might have been saved.
An operating rule regarding max. system loadability margins similar to

WECCs might have given the operators a better picture of the situation, but without proper monitoring, these would have probably not worked either.
In the absence of these, under-voltage relays with ULTC blocking might

have saved the system, as these would have automatically shed load when the voltages started to collapse in First Energys (FE) region in Ohio.
376
Voltage Stability Report
Much more information regarding the issue voltage stability can be found
in: C. A. Canizares, editor, Voltage Stability Assessment: Concepts, Practices and Tools, IEEE-PES Power System Stability Subcommittee Special Publication, SP101PSS, August 2002. IEEE-PES WG Report Award 2005.
This is a 283-page report, published after 5 years in preparation, and

coauthored by several voltage stability from around the world.
More details about the report can be found at:

http://thunderbox.uwaterloo.ca/claudio/claudio.html/#VSWG
377
Angle Stability Outlines
Denitions. Small disturbance: Hopf Bifurcations. Control and mitigation. Practical example. Transient Stability Time Domain. Direct Methods. Equal Area Criterion. Energy Functions. Practical applications.
Universidad de Castilla-La Mancha June 26, 2008 Angle Stability - 1
378
Angle Stability Denitions

Frequency Stability
Voltage Stability
Transient Stability
Short Term
Long Term
379
Angle Stability Denitions
Rotor angle stability refers to the ability of synchronous machines of an

interconnected power system to remain in synchronism after being subjected to a disturbance. It depends on the ability to maintain/restore equilibrium between electromagnetic torque and mechanical torque of each synchronous machine in the system.
In this case, the problem becomes apparent through angular/frequency

swings in some generators which may lead to their loss of synchronism with other generators.
380
Small Disturbance
Small disturbance (or small signal) rotor angle stability is concerned with
the ability of the power system to maintain synchronism under small disturbances. The disturbances are considered to be sufciently small that linearization of system equations is permissible for purposes of analysis.
This problem is usually associated with the appearance of undamped

oscillations in the system due to a lack of sufcient damping torque.
Theoretically, this phenomenon may be associated with a s.e.p.

becoming unstable through a Hopf bifurcation point, typically due to contingencies in the system (e.g. August 1996 West Coast Blackout).
381
Hopf Bifurcation I
For the generator-load example, with AVR but no QG limits:

Generator
PG + jQG jx G jxL
PL + jQL
V1 1 +
V2 2
Kv s
V10
382
Hopf Bifurcation I
The DAE model is: E E V2 sin DG Pm = X E V2 1 sin Pd = DL X 1 M = Kv (V10 V1 )
1 V22 E V2 V2 = + cos kPd X X E V2 V1 V2 sin sin 0 = XL X 1 1 E V2 V1 V2 2 0 = V2 cos + cos XL X X XL

383
Hopf Bifurcation I
Observe that the algebraic constraint can be eliminated, since: V1r V1i
Thus:
= V1 cos = V1 sin
V1i V2 E V2 E XL 0= sin V1i = sin XL X X V22 1 E V2 V1r V2 1 + cos XL X X XL E XL 1 + cos 1 XL X
and
0 V1r
= V2
384
Hopf Bifurcation I
This yields the following equations, which are better for numerical time
domain simulations:
E V2
E V2 = sin DG Pm X E V2 1 sin Pd = DL X 1 M = Kv V10 = 1

2 2 V1r + V1i
V22 E V2 + cos kPd X X
Observe that in this case, Pm = Pd , i.e. generation and load are

assumed to be balanced.
385
Hopf Bifurcation I
The PV curves for M = 0.1, DG = 0.01, DL = 0.1, = 0.01, Kv = 10, XG = 0.5, V10 = 1, k = 0.25 are:
1
OP
0.9 0.8 0.7 0.6
HB
HB
V2
0.5 0.4 0.3 0.2 0.1 0
xL = 0.6
xL = 0.5
0.1
0.2
0.3
Pd
0.4
0.5
0.6
0.7
0.8
386
Hopf Bifurcation I
The eigenvalues for the system with respect to changes in Pd for xL = 0.5:
3
Imag
3 5
1 Real
387
Hopf Bifurcation I
There is a Hopf bifurcation for Pd = 0.65, xL = 0.5:

6
HB
2
()
10
0.1
0.2
0.3
Pd
0.4
0.5
0.6
0.7
0.8
388
Hopf Bifurcation I
A Hopf bifurcation with eigenvalues = j yields a periodic

oscillation of period:
T = Hence, for the example:
j3
2s
389
Hopf Bifurcation I
The contingency xL = 0.5 0.6 yields:

2.5
E V1 V2
1.5
0.5
0.5
0.2
0.4
0.6
t [s]
0.8
1.2
1.4
1.6
390
Hopf Bifurcation I
Notice that in this case no oscillations are observed, which is a

trademark of Hopf bifurcations and small-disturbance angle instabilities.
The reason for this is that the oscillation period is 2 s (typical in practice
where these kinds of oscillations are in the 0.1-1 Hz range), but the bus voltage collapses well before the oscillations appear, which is atypical and is probably due to the chosen impedances and time constants.
This example stresses the point that angle instabilities do lead to voltage
collapse, and vice versa, voltage instabilities lead to angle/frequency oscillations, even though the reason behind each stability problem are fairly different.
391
Hopf Bifurcation II
Single-machine dynamic-load system with SVC:

PG + jQG jX PL + jQT
E V Vref jBC
V 0
392
Hopf Bifurcation II
The total reactive power absorbed by the load and the SVC is as follows: EV V2 + cos() + V 2 BC QT (V, ) = X X The SVC controller is modeled as a rst order pure integrator.
Vref + 1/sT V BC
393
Hopf Bifurcation II
The resulting differential equations of the SMDL system with SVC are as
follows:
V BC
= EV 1 [Pd sin() D] = M X 1 1 EV = [kPd + V 2 (BC ) + cos()] X X 1 (Vref V ) = T
394
Hopf Bifurcation II
BC is the equivalent susceptance of the SVC; T and Vref are the SVC
time constant and reference voltage, respectively.
In the following, it is assumed that T = 0.01 s and Vref = 1.0 p.u. Observe that also in this case it is possible to deduce the set of ODE, i.e.
the algebraic variables can be explicitly expressed as a function of the state variables and the parameters.
395
Hopf Bifurcation II
The state matrix of the system is as follows:
0 EV M X cos() A= EV sin() X 0
D | M
| |
0 0
| 0 E | M X sin() | 0 1 V2 | X [E cos() 2V + 2V BC X] | 1 | T | 0
396
Hopf Bifurcation II
Eigenvalue loci:
400
300
200
100
Imaginary
Hopf Bifurcation P =1.4143

d
100
200
300
400 0.4
0.2
0.2
0.4
0.6
0.8
Real
397
Hopf Bifurcation II
A complex conjugate pair of eigenvalues crosses the imaginary axis for Pd = 1.4143, thus leading to a Hopf bifurcation. The HB point is: (0 , 0 , V0 , BC0 , Pd0 ) = (0.7855, 0, 1, 1.2930, 1.4143)
398
Hopf Bifurcation II
Bifurcation diagram Pd - :
2.5
(rad)
P
1.5
max d
Hopf Bifurcation
1
0.5
0.2
0.4
0.6
0.8
P (p.u.)
d
1.2
1.4
1.6
1.8
399
Hopf Bifurcation II
We simulate a step change in Pd from 1.41 p.u. to 1.42 p.u. for t = 2 s. For t > 2 s the system does not present a stable equilibrium point and
shows undamped oscillations (likely an unstable limit cycle), as expected from the P - curve.
For t = 2.57 s, the load voltage collapses. Note that, in this case, the generator angle shows an unstable trajectory
only after the occurrence of the voltage collapse at the load bus.
400
Hopf Bifurcation II
Time domain simulation results:

0.85
(rad)
0.8
0.75 1.8
1.9
2.1
2.2
2.3
2.4
2.5
2.6
t (s)
1.5
V (p.u.)
1 0.5 0 1.8
1.9
2.1
2.2
2.3
2.4
2.5
2.6
t (s)
401
Hopf Bifurcation II
The use of the SVC device gives a birth to a new bifurcation, namely a
Hopf bifurcation.
This Hopf bifurcation cannot be removed by simply adjusting system

parameters.
However SVC and load dynamics can be coordinated so that the

loadability of the system can be increased.
402
Control and Mitigation
For the IEEE 14-bus test system:

Bus 13
Bus 14 Bus 10 Bus 12
Bus 11
Bus 09
Bus 07
Bus 05 Bus 01
Breaker
Bus 02
Bus 03
403
Generator speeds for the line 2-4 outage and 40% overloading:
1.001
1.0005
0.9995
0.999
0.9985
0.998
10
15 time (s)
20
25
30
404
This has been typically solved by adding Power System Stabilizers (PSS)
to the voltage controllers in certain generators, so that equilibriun point is made stable, i.e. the Hopf is removed.
vs max vSI Tw s Kw Tw s + 1 T1 s + 1 T2 s + 1 T3 s + 1 T4 s + 1 1 T s + 1 vs
vs min
FACTS can also be used to address this problem.
405
For the IEEE 14-bus test system with PSS at bus 1:

Bus 13
Bus 11
Bus 09
Bus 07
Bus 05 Bus 01
Breaker
Bus 02
Bus 03
406
Generator speeds for the line 2-4 outage and 40% overloading:
1.001
1.0005
0.9995
0.999
0.9985
0.998
10
15 time (s)
20
25
30
407
Data regarding this system are available at:

http://thunderbox.uwaterloo.ca/claudio /papers/IEEEBenchmarkTFreport.pdf
More details regarding this example can be found in:

F. Milano, An Open Source Power System Analysis Toolbox, accpeted for publication on IEEE Trans. On Power Systems, March 2004.
408
For the IEEE 145-bus, 50-machine test system:

1 93 110 43 33 5 4 3 49 2 114 34 38 40 84 35 41 37 39 47 50 51 102 55 53 56 101 52 61 54 86 127 135 138 128 133 132 120 122 121 95 67 111 8 9 69 63 64 65 68 97 125 124 80 112 11 12 72 71 21 59 98 13 70 100 103 58 14 15 17 16 81 25 108 22 73 31 26 106 105 75 82 109 27 91 29 24 28 20 19 18 83 78 30 23 10 32 94 60 89 96 77 76 123 131 134 42 48 44 46 85 117 139 118 137 129 136 49 36 99
AREA 2
87 88 45 116
AREA 1
115 145 144 143 142 141 140
113 6
57 7 104 126 66
62 119
92
90 130
107
79
74
409
For an impedance load model, the PV curves yield:

(a) 0.94 HB Voltage (p.u.) 0.92 0.9 0.88 0.86 0.84 0.82 0 0.002 0.004 0.006 0.008 0.01 0.012 Operating point HB Base Case Line 7990 Outage
Constant impedance load line
(b) L.F. (p.u.) 0.94 Voltage (p.u.) Base Case Line 7990 Outage
Operating point
0.93
HB
0.92 0.91 0.9
HB
Constant impedance load line
2.5
3.5
4.5
5.5 x 10
6
3
410
Indices based on the singular values have been proposed to predict Hopf
bifucations:
0.015 HBI2 Base Case HBI Base Case 1 HBI Line 7990 Outage 2 HBI1 Line 7990 Outage
0.01
Hopf Indices 0.005
HB
0
HB
0.002
0.004
0.006
0.008
0.01
0.012
411
More details regarding this example can be found in:

C. A. Canizares, N. Mithulananthan, F. Milano, and J. Reeve, Linear Performance Indices to Predict Oscillatory Stability Problems in Power Systems, IEEE Trans. On Power Systems, Vol. 19, No. 2, May 2004, pp. 1104-1114.
412
Small Disturbance Applications
In practice, some contingencies trigger plant or inter-area frequency

oscillations in a heavily loaded system, which may be directly associated with Hopf bifurcations.
This is a classical problem in power systems and there are many

examples of this phenomenon in practice, such as the August 10, 1996 blackout of the WSCC (now WECC) system.
413
Small Disturbance Applications
Observe that the maximum loadability of the system is reduced by the

presence of the Hopf bifurcation.
This leads to the denition of a dynamic ATC value.

1
HB
0.9
OP
HB
0.8 0.7
ETC
ATC
TRM
0.6
V2
0.5
TTC
0.4 0.3 0.2 0.1 0

Worst Contingency
0.1
0.2
0.3
Pd
0.4
0.5
0.6
0.7
0.8
414
August 10, 1996 WSCC Blackout
This information was extracted from a presentation by Dr. Prabha Kundur,

President and CEO of PowerTech Labs Inc.
The material is availbale at:

toronto.ieee.ca/events/oct0303/prabha.ppt
415
System conditions: High ambient temperatures in Northwest, and hence high power
transfers from Canada to California.
Prior to main outage, three 500 kV line sections from lower Columbia
River to loads in Oregon were out of service due to tree faults.
California-Oregon interties loaded to 4330 MW north to south. Pacic DC intertie loaded at 2680 MW north to south. 2300 MW ow from Britsh Columbia.
416
Growing 0.23 Hz oscillations caused tripping of lines:
3000 2900 2800 2700 2600 2500 2400 2300 0 3 6 12 16 19 22 25 28 31 40 43 47 50 53 56 59 62 65 68 71 74
3 4
4 1 2
Time in seconds
417
Event 1: 14:06:39 Big Eddy-Ostrander 500 kV LG fault - ashed to tree Event 2: 14:52:37 John Day-Marion 500 kV LG -ashed to tree Event 3: 15:42:03 Keeler-Alliston 500 KV - LG - ashed to tree Event 4: 15:47:36 Ross-Lexington 500 kV - ashed to tree Event 5: 15:47:36-15:48:09 8 McMary Units trip
418
As a results of the undamped oscillations, the system split into four large
islands.
419
PSS solution:
San Onofre (addition)
Palo Verde (tune existing)
420
PSS solution:
3000
With existing controls Eigenvalue = Frequency = Damping =
2800
0.0597 + j1.771 0.2818 Hz
2600 2400 2200
0.0337
15
30
45
61
72
Time in seconds 3000
With PSS modications

2800
Eigenvalue =
2600
0.0717 + j1.673 0.2664 Hz
Frequency =
2400
Damping =
2200 0 18 50 32 Time in seconds 68 75
0.0429
421
7.5 million customers experienced outages from a few minutes to nine

hours.
Total load loss: 35,500 MW.
422
Transient Stability
Large disturbance rotor angle stability or transient stability, as it is

commonly referred to, is concerned with the ability of the power system to maintain synchronism when subjected to a severe disturbance, such as a short circuit on a transmission line. The resulting system response involves large excursions of generator rotor angles and is inuenced by the nonlinear power-angle relationship.
The system nonlinearities determine the system response; hence,

linearization does not work in this case.
423
Transient Stability
For small disturbances, the problem is to determine if the resulting steady

state condition is stable or unstable (eigenvalue analysis) or a bifurcation point (e.g. Hopf bifurcation).
For large disturbances, the steady state condition after the disturbance
can exist and be stable, but it is possible that the system cannot reach that steady state condition.
424
Transient Stability
The basic idea and analysis procedures are: Pre-contingency (initial conditions): the system is operating in
normal conditions associated with a s.e.p.
Contingency (fault trajectory): a large disturbance, such as a short

circuit or a line trip forces the system to move away from its initial operating point.
Post contingency (fault clearance): the contingency usually forces

system protections to try to clear the fault; the issue is then to determine whether the resulting system is stable, i.e. whether the system remains relatively intact and the associated time trajectories converge to a reasonable operating point.
425
Transient Stability
Based on non linear theory, this analysis can be basically viewed as

determining wheter the fault trajectory at the clearance point is outside or inside of the stability region of the post-contingency s.e.p.
426
Time domain analysis
Given the complexity of power system models, the most reliable analysis
tool for these types of studies is full time domain simulations.
For example, for the generator-load example:

Generator
PG + jQG jx G jxL
PL + jQL
V1 1
V2 2
jxC
427
The ODE for the simplest generator d-axis transient model and
neglecting AVR and generator limits is:
V2
where
1 (Pd E V2 B sin DG ) M 1 = (E V2 B sin Pd ) DL 1 [V22 (B BC ) + E V2 B cos kPd ] = = B= 1 1 = X XG + XL
428
The objective is to determine how much time an operator would have to connect the capacitor bank BC after a severe contingency, simulated here as a sudden increase in the value of the reactance X , so that the
system recovers.
In this case, and as previously discussed in the voltage stability section,

the contingency is severe, as the s.e.p. disappears.
Full time domain simulations are carried out to study this problem for the parameter values M = 0.1, DG = 0.01, DL = 0.1, = 0.01, E = 1, Pd = 0.7, k = 0.25, BC = 0.5.
429
A contingency X = 0.5 0.6 at tf = 1 s, with BC connection at tc = 1.4 s yields a stable system:

1.4

1.2 V2
0.8
0.6
0.4
0.2
tf tc
0.2
0.5
1.5
2.5
3.5
4.5
t [s]
430
If BC is connected at tc = 1.5 s, the system is unstable:

6

5 V2
tc
3
tf
0.2
0.4
0.6
0.8
1.2
1.4
1.6
1.8
t [s]
431
Direct Methods
Time domain analysis is expensive, so direct stability analysis technique

have been proposed based on Lyapounovs stability theory.
The idea is to dene an energy or Lyapounov function (x, xs ) with

certain characteristics to obtain a direct measure of the stability region
A(xs ) associated with the post-contingency s.e.p. xs . A systems energy is usually a good Lyapounov function, as it yields a
stability measure.
432
Direct Methods
The rolling ball example can used to explain the basic behind these
techniques:
u.e.p.2
v m h
u.e.p.1
s.e.p.
There are 3 equilibrium points: one stable (valley bottom), two unstable
(hill tops).
433
Direct Methods
The energy of the ball is a good Lyapounov or Transient Energy Function

(TEF):
= Wkinetic + Wpotential = WK + WP 1 = mv 2 + mgh 2 = ([v, h]T , 0)
The potential energy at the s.e.p. is zero, and presents local maxima at the u.e.p.s (WP 1 and WP 2 ). The closest u.e.p. is u.e.p.1 since WP 1 < WP 2 .
434
Direct Methods
The stability of this system can then be evaluated using this energy: if W < WP 1 , the ball remains in the valley, i.e. the system is stable, and will converge to the s.e.p. as t .
depending on friction (inconclusive test).
If W > WP 1 , the ball might or might not converge to the s.e.p., When the balls potential energy WP (t) reaches a maximum with respect to time t, the system leaves the valley, i.e. unstable
condition.
435
Direct Methods
The valley would correspond to the stability region when friction is

large.
In this case, the stability boundary A(xs ) corresponds to the ridge where the u.e.p.s are located and WP has a local max. value. The smaller the friction in the system, the larger the difference between the ridge and A(xs ). For zero friction, A(xs ) is dened by WP 1 .
436
Direct Methods
The direct stability test is only a necessary but not sufcient condition: (x, xs ) < c (x, xs ) > c
potential energy function.
x A(xs )
Inconclusive!
where the value of c is usually associated with a local maximum of a
437
Direct Methods
For the simple generator-innite bus example, neglecting limits and AVR:
Generator
PG + jQG jx G jxL
PL + jQL jxth
System
V1 1
V2 2
V 0
= = r 0 1 EV = PL sin D M X
= XG + XL + Xth
438
Direct Methods
The kinetic energy in this system is dened as: WK = And the potential energy is: WP = (Tc Tm )d (Pc Pm )d in p.u. for r 0 EV ( (PG PL )d PL )d X s s

1 M 2 2
E V B(cos cos s ) PL ( s )
where s is the s.e.p. for this system.
439
Direct Methods
With WP presenting a very similar prole as the rolling ball example:

PG
unstable stable unstable
E V X
u2
u1
max
WF WF 2 WF 1
min max
u2
u1
440
Direct Methods
Hence, the system Lyapounov function of TEF is: T EF = (x, xs ) = ([, ]T , [s , 0]T ) 1 = M 2 E V B(cos cos s ) 2 PL ( s ) Thus, using similar criteria as in the case of the rolling ball: If T EF < WP 1 system is stable. If T EF > WP 1 inconclusive for D > 0 (friction). If T EF > WP 1 unstable for D = 0 (unrealistic).
441
Direct Methods
This is equivalent to compare areas in the PG vs. graph (Equal Area

Criterion or EAC):
PG
pre-contingency
post-contingency
PL
contingency (fault)
(0) = spre
(tc ) spost
u1 post
fault clearing time
442
Direct Methods
Thus, comparing the acceleration area:

(tc )
Aa
=
spre (tc )
(PL PGf ault )d EV PL Xf ault d
=
spre
versus the deceleration area:

spost
Ad
=
(tc ) spost
(PGpost PL )d EV PL d Xpost
=
(tc )
443
Direct Methods
In conclusion: If Aa < Ad system is stable at tc . If Aa > Ad inconclusive for D > 0.
If Aa > Ad unstable for D = 0 (unrealistic).
444
Direct Methods: Example 1
A 60 Hz generator with a 15% transient reactance is connected to an

innite bus of 1 p.u. voltage through two identical parallel transmission lines of 20% reactance and negligible resistance. The generator is delivering 300 MW at a 0.9 leading power factor when a 3-phase solid fault occurs in the middle of one of the lines; the fault is then cleared by opening the breakers of the faulted line.
Assuming a 100 MVA base, determine the critical clearing time for this
generator if the damping is neglected and its inertia is assumed to be
H = 5 s. Assuming D = 0.1 s, determine the actual critical clearing time.
445
Pre-contingency or initial conditions: PGpre QL EV sin spre = PL = Xpre V2 EV = + cos spre Xpre Xpre
446
Where: Xpre = 0.15 + 0.2 = 0.25 2
300 MW PL = 100 MVA E 3 = sin spre 0.25 QL = 3 tan(cos1 0.9) E 1 + cos spre 1.4530 = 0.25 0.25
447
Eipre
= E sin spre = 0.75 1.3633 E 2pre + E 2pre r i 1.5559 tan1

Eipre Erpre
Erpre
= E cos spre = = =
spre
= =
28.82 = 0.5030 rad
448
Fault conditions: PGf ault = = EV sin Xf ault 1.5559 sin Xf ault
where, using a Y- circuit transformation due to the fault being in the middle of one of the parallel lines:
jXf ault j0.15 j0.2 E V 0 j0.1 j0.1
449
Xf ault PGf ault Aa
= = =
0.15 0.2 + 0.1 0.2 + 0.15 0.1 0.1 2.394 sin

(tcc ) spre (tcc )
(PL PGf ault )d (3 2.394 sin )d
=
0.503
= =
3((tcc ) 0.503) + 2.394(cos (tcc ) cos(0.503)) 3(tcc ) + 2.394 cos (tcc ) 3.6065
450
Post contingency conditions: Xpost PGpost = = = 3 = = = spost 0.15 + 0.2 = 0.35 EV sin Xpost 4.446 sin 4.446 sin spost 42.44 0.7407 rad
451
spost
Ad
=
(tcc ) 2.4
(PGpost PL )d (4.446 sin 3)d
=
(tcc )
= 4.446(cos 2.4 cos (tcc )) 3(2.4 (tcc )) = 3(tcc ) + 4.446 cos (tcc ) 3.9215
452
Aa
= Ad = 3(tcc ) + 2.394 cos (tcc ) 3.6065 = 3(tcc ) + 4.446 cos (tcc ) 3.9215
(tcc )
= 81.17 = 1.4167 rad
453
During the fault: M = 1 EV = sin PL M Xf ault H = f 5s = 60 Hz = 0.0265 s2 = = 37.70(3 2.394 sin )
454
Integrating these equations numerically for (0) = spre = 28.82 :

220 200 180 160 140
[deg]
120 100 80 60 40 20
0.05
0.1
0.15
t [s]
0.2
0.25
0.3
0.35
455
For D = 0.1 and a clearing time of tc = 0.27 s, the system is stable:

150
[deg]
100
50
0.2
0.4
0.6
0.8
t [s]
1.2
1.4
1.6
1.8
10
[deg]
10
0.2
0.4
0.6
0.8
t [s]
1.2
1.4
1.6
1.8
456
For a clearing time of tc = 0.28 s, the system is unstable; hence tcc 0.275 s:
2500 2000
[deg]
1500 1000 500 0
0.2
0.4
0.6
0.8
t [s]
1.2
1.4
1.6
1.8
40
30
[deg]
20
10
0.2
0.4
0.6
0.8
t [s]
1.2
1.4
1.6
1.8
457
Generator-motor, i.e. system-system, cases may also be studied using

the EAC method based on an equivalent inertia
M = M1 M2 /(M1 M2 ), and damping D = M D1 /M1 = M D2 /M2 . For the generator-load example neglecting the internal generator
impedance and assuming an instantaneous AVR:
V1 1
V2 2
458
The energy functions, with or without generator limits, can be shown to

be:
WK WP
1 M 2 2 = B(V1 V2 cos V10 V20 cos 0 ) 1 1 2 2 2 + B(V2 V20 ) + B(V12 V10 ) 2 2 V2 QG ln Pd ( 0 ) + Qd ln V20 =
V1 V10
The stability of this system can then be studied using the same energy evaluation previously explained for T EF = (x, x0 ) = WK + WP .
459
Thus for V1 = 1, XL = 0.5, Pd = 0.1, and Qd = 0.25Pd , the potential energy WP (, V2 ) that denes the stability region withr espect
to the s.e.p. is:
8 7 6 5
WP
4 3 2 1
0 400 200
u.e.p. saddle
0 200
s.e.p. node
1 0.5 400 0
2 1.5
V2
460
Simulating the critical contingency XL = 0.5 0.6 for Pd = 0.7 and

neglecting limits, the energy proles are:
0.2
Wp Wk+Wp
0.1
T EF
0.1
0.2
0.3
0.4 0.9
1.1
1.2
1.3
1.4
1.5
1.6
t [s]
461
The exit point on A(xs ) is approximately at the maximum potential

energy point.
Thus, the critical clearing time is: tcc 1.42 s A similar value can be obtained through trial-and-error.
462
Direct Methods: Conclusions
The advantages of using Lyapounov functions are: Allows reduced stability analysis. Can be used as an stability index. The problems are: Lyapounov functions are model dependent; in practice, only
approximate energy functions can be found.
Inconclusive if test fails. The post-perturbation system state must be known ahead of time, as
the energy function is dened with respect to the corresponding s.e.p.
Can only be used as an approximate stability analysis tool.

463
Transient Stability Applications
In practice, transient stability studies are carried out using time-domain

trial-and-error techniques.
These types of studies can now be done on-line even for large systems. The idea is to determine whether a set of realistic contingencies make
the system unstable or not (contingency ranking), and thus determine maximum transfer limits or ATC in certain transmission corridors for given operating conditions.
464
Thus, the maximum loadability of the system may be affected by the

size of the stability region, leading to the denition of a true ATC value.
1
HB
0.9
OP
HB
0.8 0.7
ETC
ATC
TRM
0.6
V2
0.5
TTC
0.4 0.3 0.2 0.1 0

Worst Contingency
0.1
0.2
0.3
Pd
0.4
0.5
0.6
0.7
0.8
465
Critical clearing times are not really an issue with current fast acting
protections.
Simplied direct methods such as the Extended Equal Area Criterion (Y.
Xue et al., Extended Equal Area Criterion Revisited, IEEE Transaction on Power Systems, Vol. 7, No. 3, Aug. 1992, pp. 1012-1022) have been proposed and tested for on-line contingency pre-ranking, and are being implemented for practical applications through an E.U. project.
466
Chilean Blackout (11/07/2003)
An example of an application of transient analysis techniques can be

found in L.S. Vargas and C. A. Canizares, Time Dependence of Controls to Avoid Voltage Collapse, IEEE Transaction son Power Systems, Vol. 15, No. 4, November 2000, pp. 1367-1375.
This paper discusses the May 1997 voltage collapse event of the main
power system in Chile.
467
Most of Chile lost power in a major blackout on Friday evening, snarling

rush hour trafc in the capital. The blackout began at about 7:20 pm, and power was back in about a third of the affected areas at 9:00 pm.
At 9:00 pm lights were gradually coming on in parts of the capital, home

to 5 million people or one third of the countrys population. Television reports said power went out as far as Puerto Montt, a city some 600 miles south of Santiago, and in areas the same distance to the north.
Source Reuters
468
Main system characteristics: Extension: 756626 km2 . Inhabitants: 14.5 mil. National consumptions: 33531 GWh. National peak load: 5800 MW. Installed capacity: 8000 MW. Frequency: 50 Hz. Trans. Level: 66/110/154/220/500 kV. Four interconnected systems: SING, SIC, AISEN, MAGALLANES
469
470
Initial state of SIC system: 2500 MW load. Power ow south-north near 1000 MW (900 MW through 500 kV lines
and 100 MW through 154 kV lines).
Events: Line 154 kV trips. Major generator in the south hits reactive power limits and losses
voltage control.
Operator tries to recover falling voltages by connecting a capacitor

bank near Santiago.
471
472
The line trip and generator limits yield a voltage collapse associated with
a limit-induced bifurcation problem:
473
PV curves:
474
The connection of the capacitor bank after the generator limits are
reached did not save the system, as the faulted system trajectories had left the stability region of the post contingency operating point.
475
If the capacitor bank is connected before the generator limits are reached,
the system would have been saved, as the faulted system trajectories were still within the stability of the post-contingency operating point.
476
Frequency Stability Outlines
Denitions. Basic Concepts. Practical applications, controls and protections. Italian Blackout (28/10/2003) European Blackout (4/11/2006)
Universidad de Castilla-La Mancha June 26, 2008 Frequency Stability - 1
477
Frequency Stability Denitions

Frequency Stability
Voltage Stability
Transient Stability
Short Term
Long Term
478
Frequency Stability Denitions
Frequency Stability refers to the ability of a power system to maintain

steady frequency following a severe system upset resulting in a signicant imbalance between generation and load.
Thus, frequency stability analysis concentrates on studying the overall

system stability for sudden changes in the generation-load balance.
479
Frequency Stability Concepts
For the generator-load example, with AVR but no QG limits:

Generator
PG + jQG jx G jxL
PL + jQL
V1 1 +
V2 2
Kv s
V10
480
Neglecting losses and electromagnetic dynamics, the generator with a very simple AVR and no limits can be modeled using a d-axis transient
model:
G G E
= G = r 0 1 (Pm PG DG G ) = M = Kv (V10 V1 )
481
The load can be simulated using simplied mixed models and constant
power factor:
2 V2
= 2 = =
1 (PL Pd ) DL
1 (QL kPD )
Qd
482
The transmission system yields the power ow equations (X = XL = XG ): PG = = QL = = QG = E V2 PL = sin(G 2 ) X V1 V2 sin(1 2 ) XL V22 E v2 + cos(2 G ) X X V22 V1 V2 + cos(2 1 ) XL XL V12 V1 V2 cos(1 2 ) XL XL
483
Dene: = G 2 = 1 2 = G = 2
484
This yields the DAE model: E E V2 Pm sin DG = X E V2 1 sin Pd = DL X 1 M = Kv (V10 V1 )
1 V22 E V2 V2 = + cos kPd X X E V2 V1 V2 sin sin 0 = XL X 1 1 E V2 V1 V2 2 0 = V2 cos + cos XL X X XL

485
And the equilibrium equations: 0 0 0 0 0 0 = = = = = = E V2 Pm sin DG X E V2 sin Pd DL X V10 V1 V22 E V2 + cos kPd X X E V2 V1 V2 sin sin XL X 1 E V2 V1 V2 1 2 V2 cos + cos XL X X XL
486
Hence: x = [, , E , v2 ]T y p = [V1 , ]T = [Pm , V10 ] = Pd
Observe that in this analysis, Pm = Pd to study the effect of

generator-load imbalances in the system.
487
In normal operating conditions Pm = Pd = 0, from the rst 2

equilibrium equations.
Hence, these equations can be replaced by the following 4 power ow equations, with 4 unknowns (E , V2 , , ): 0 0 0 0 = = = = E V2 sin Pd X V22 E V2 kPd + cos X X V10 V2 sin Pd XL V22 V10 V2 kPd + cos XL XL
488
Simulating a 50% generation and load reduction, respectively, for M = 0.1, DG = 0.01, DL = 0.1, = 0.01, Kv = 10, XL = 0.5, XG = 0.5, V10 = 1, Pd0 = 0.7, k = 0.25. Initial solution: E V2 V1 = = = = = = 0 0.7266 1.3463 0.7826 1.0000 0.4636
489
Time domain simulation:

1.5 1 0.5 0 0.5
50% generator drop
1 1.5 2
50% load drop
2.5 3 3.5
E V2 V1
6 7 8 9 10
t [s]
490
The generation reduction yields a 0.5 Hz frequency drop and a load

voltage increase.
The load reduction yields a 0.5 Hz frequency increase, and an even

larger load voltage increase, as the reactive power demand drops.
The system reaches new s.e.p.s in both cases, as expected. Observe that the AVR keeps the generator terminal fairly stable and close to its set value V10 .
491
These frequency excursions due to generation-load imbalances are

typical.
It might lead to unstable conditions due to device protections such as

frequency relays in generators and loads.
Frequency problems may be solved manually by operators or

automatically through controls and/or protections.
Generator governors automatically regulate local frequency excursions.
492
Centralized frequency regulators, such as automatic Area Control Error

(ACE) regulators, may be used to regulate power exchanges among control areas by controlling the frequency deviations on the interties.
Examples of frequency instabilities are: The Italian blackout of Tuesday, October 28, 2003 (Material courtesy
of Prof. Alberto Berizzi, Politecnico di Milano, Italy).
The European blackout of Saturday, November 4, 2006 (Material

courtesy of Prof. Edmund Handschin, University of Technology, Dortmund, Germany).
493
Italian Blackout (28/10/2003)
494
495

3:10.47 ETrans (operator of interties between Switzerland and rest of Europe) lets the GRTN (Italian operator) know of the Mettlen-Lavorgo line trip (1320 MW) and the overloading of the Sils-Soazza line (1650 MW), and requests a 300 MW demand reduction to relief the overload. 3:18.40 ETrans contacts EGL (Switzerland operator) requesting the tripping of a transformer in Soazza. 3:21.00 GRTN reduces the power imports by 300 MW.
496

3:22.03 ATEL (Swiss energy company) changes the connection of the transformer at Lavorgo. 3:25.22 Protections trip the Sils-Soazza line (1783 MW). This is basically the beginning of the cascading events that follow, severing Italy from the rest of Europe and leading to the collapse of the Italian system.
497
498
Imports from France:
499
Frequency in Italy:
500
Frequency and voltages in Europe:
501
502
503
504
European Interconnected Systems
505

System Capacity GW Nordel UPS/IPS UKTSOA Nordel 94 337 85 600 Peak Load GW 66 215 66 390 Energy TWh 405 1285 400 2530 Population Mio. 24 280 65 450
506
450 million people served 2530 TWh used 600 GW installed capacity at 500 e/kW = 300 Ge 230.000 km HV network at 400 000 e/km = 90 Ge Approx. 5.000.000 km MV+LV network 1500 e investment per EU citizen Largest man-made system
507
European Blackout (4/11/2006)
508
509
Area 1: The frequency drops to 49 Hz, which causes an automatic load

schedding.
Area 2: Real power surplus of 6000 MW.
510

50.1 f [Hz] 50.0 49.9 49.8 49.7 49.6 49.5 49.4 49.3 49.2 49.1 49.0 48.9
22:10:00 22:10:10 22:10:20 22:10:30 22:10:40 22:10:50 22:11:00
frequency
511
512
Contents
Overview UWPFLOW Matlab PSAT
Universidad de Castilla-La Mancha June 26, 2008 Software Tools - 1
513
Overview
Software packages for power system analysis can be basically divided

into two classes of tools:
Commercial softwares. Educational/research-aimed softwares.
514
Overview
Commercial softwares: PSS/E EuroStag Simpow CYME PowerWorld Neplan
515
Overview
Commercial software packages follows an all-in-one philosophy and are

typically well-tested and computationally efcient.
Despite their completeness, these softwares can result cumbersome for

educational and research purposes.
commercial softwares are closed, i.e. do not allow changing the source
code or adding new algorithms.
516
Overview
For research purposes, the exibility and the ability of easy prototyping
are often more crucial aspects than computational efciency.
At this aim, there is a variety of open source research tools, which are
typically aimed to a specic aspect of power system analysis.
An example is UWPFLOW which provides an extremely robust algorithm

for continuation power ow analysis.
517
Overview
C anf FORTRAN are very fast but requires keen programming skills and
are not suitable for fast prototyping.
Several high level scientic languages, such as Matlab, Mathematica and

Modelica, have become more and more popular for both research and educational purposes.
At this aim, there is a variety of open source research tools, which are
typically aimed to a specic aspect of power system analysis.
Matlab proved to be the best user choice.
518
Overview
Matlab-based power system analysis tools: Power System Toolbox (PST) MatPower Voltage Stability Toolbox (VST) Power Analysis Toolbox (PAT) Educational Simulation Tool (EST) Power system Analysis Toolbox (PSAT)
519
Overview
Comparison of Matlab-based power system analysis softwares:

Package EST MatEMTP MatPower PAT PSAT PST SPS VST PF CPF OPF SSA TD EMT GUI GNE
520
Overview
The features illustrated in the table are: power ow (PF) continuation power ow and/or voltage stability analysis (CPF-VS) optimal power ow (OPF) small signal stability analysis (SSA) time domain simulation (TD) graphical user interface (GUI) graphical network editor (GNE).
521
Overview
An important but often missed issue is that the Matlab environment is a

commercial and closed product, thus Matlab kernel and libraries cannot be modied nor freely distributed.
To allow exchanging ideas and effectively improving scientic research,

both the toolbox and the platform on which the toolbox runs should be free (Richard Stallman).
An alternative to Matlab is the free GNU/Octave project.
522
UWPFLOW
UWPFLOW is a research tool that has been designed to calculate local

bifurcations related to system limits or singularities in the system Jacobian.
The program also generates a series of output les that allow further
analyses, such as tangent vectors, left and right eigenvectors at a singular bifurcation point, Jacobians, power ow solutions at different loading levels, voltage stability indices, etc.
523
UWPFLOW Features
Adequate handling of generators limits, with generators being able to

recover from a variety of limits, including S limits.
Steady state models of generators and their control limits (AVR and
Primemover limits) are included.
Voltage dependent load models for voltage stability analysis are also
included.
524
UWPFLOW Features
Either BPA/WSCC ac-dc (HVDC systems) input data formats (and

variations) or IEEE common format may be used.
Detailed and reliable steady state models of SVC, TCSC and STATCOM
models, and their controls with the corresponding limits are included.
Secondary voltage control, as dened by ENEL (elecetricity company of

Italy), can be modeled and simulated in the program.
525
UWPFLOW Features
The program is able to compute the minimum real eigenvalue and the
related right and left eigenvectors and several voltage stability indices.
The program generates a wide variety of output ASCII and MATLAB (.m)
les as well as IEEE common format data les.
The program has being designed to automatically run script les.
526
UWPFLOW Usage
Like any other UNIX program, i.e., command-line options (-option) with redirection of output (>) from screen into les:
uwpflow [-options] input_file [[>]output_file]
For example, to generate the program help:

uwpflow -h
527
UWPFLOW Example
3-area sample system:

100 MW 150 MW 150 MW
60 MVAr Area 1
v3<d3
150 MW 56 MVAr
Bus 2
1.020
50 MVAr R = 0.01 p.u. X = 0.15 p.u.
V2 2
Bus 3
50 MVAr
50 MW 40 MVAr 100 MW
528
UWPFLOW Example

Bus Name Area 1 Area 2 Area 3
PG
(p.u.) 1.5 0 0.5
PL
(p.u.) 0 1.5 0.5
QL
(p.u.) 0 0.56 0.40
Using UWPFLOW to obtain the system PV curves, for a distributed slack

bus model:
529
UWPFLOW Example

HDG UWPFLOW data file, WSCC format 3-area example April 2000 BAS C C AC BUSES C C | SHUNT | C |Ow|Name |kV |Z|PL |QL |MW |Mva|PM |PG |QM |Qm |Vpu BE 1 Area 1 138 1 150 60 0 0 0 150 0 0 1.02 B 1 Area 2 138 2 150 56 0 50 0 100 0 0 1.00 B 1 Area 3 138 3 50 40 0 50 0 100 0 0 1.00
530
UWPFLOW Example

C C AC LINES C C M CS N C |Ow|Name_1 |kV1||Name_2 |kV2|||In || R | X | G/2 | B/2 |Mil| L 1 Area 1 138 Area 2 1381 15001 .01 .15 L 1 Area 1 138 Area 3 1381 15001 .01 .15 L 1 Area 2 138 Area 3 1381 15001 .01 .15 C C SOLUTION CONTROL CARD C C |Max| |SLACK BUS | C |Itr| |Name |kV| |Angle | SOL 50 Area 1 138 0. END
531
UWPFLOW Example
Generator and load change le (3area.k):

C C UWPFLOW load and generation "direction" file C for 3-area example C C BusNumber BusName DPg Pnl Qnl PgMax [ Smax 1 0 1.5 0.0 0.0 0 0 2 0 0.0 1.5 0.56 0 0 3 0 0.5 0.5 0.40 0 0
Vmax Vmin ] 1.05 0.95 1.05 0.95 1.05 0.95
532
UWPFLOW Example
Batch le for UNIX (run3area):

echo -1- Run base case power flow uwpflow 3area.wsc -K3area.k echo -2- Obatin PV curves and maximum loading uwpflow 3area.wsc -K3area.k -cthreearea.m -m -ltmp.l -s echo - with bus voltage limits enforced uwpflow 3area.wsc -K3area.k -c -7 -k0.1 echo - with current limits enforced uwpflow 3area.wsc -K3area.k -c -ltmp.l -8 -k0.1
533
UWPFLOW Example
Batch le for Windows (run3area.bat):

rem -1- Run base case power flow uwpflow 3area.wsc -K3area.k rem -2- Obatin PV curves and maximum loading uwpflow 3area.wsc -K3area.k -cthreearea.m -m -ltmp.l -s rem - with bus voltage limits enforced uwpflow 3area.wsc -K3area.k -c -7 -k0.1 rem - with current limits enforced uwpflow 3area.wsc -K3area.k -c -ltmp.l -8 -k0.1
534
UWPFLOW Example
PV curves (threearea.m):
Profiles 150
100
50 kVArea 3 kVArea 2 kVArea 1 0 0 0.2 0.4 0.6 0.8 L.F. [p.u.] 1 1.2 1.4
138 138 138
1.6
535
UWPFLOW Example
The singular value index obtained with UWPFLOW is as follows (-0

option):
0.7
0.6
Full matrix sing. value index
0.5
0.4
0.3
0.2
0.1
0.2
0.4
0.6
0.8 L.F. [p.u.]
1.2
1.4
1.6
536
Matlab Overview
Matlab is a general purpose environment for mathematical and

engineering analysis.
Is vector/matrix based. A variable is by default a matrix. Is an interpreted language, thus can be slow for heavy applications. Is not open source. The GNU-Octave project provides a good alternative:
http://www.octave.org
537
Matlab Example
Generator-load example (see the introductory example in the Voltage

Stability section starting from slide 279): :
V1 1
V2 2
538
Matlab Example
For QGmin QG QGmax : V2 = 1 M Pd 1 DL V10 V2 sin DG XL V10 V2 sin Pd XL
= = 1
2 V10 V10 V2 0 = QG + cos XL XL
with
x = [, , V2 ]T p = V10
y = QG = Pd
539
Matlab Example
For QG = QGmin,max : V2 0
with
V1 V2 = sin DG Pd XL V1 V2 1 sin Pd = DL XL = 1 V1 V2 V22 + cos Qd XL XL V1 V2 V12 + cos XL XL y = V1 = Pd
1 M
= QGmin,max
x = [, , V2 ]T p = QGmin,max
540
Matlab Example
Assume XL = 0.5, M = 1, DG = 0.01, DL = 0.1, = 0.01, k = 0.25. The time domain integration can be solved with the help of M ATLAB.
541
Matlab Example
Differential equations without limits:

function dx = example(t,x) global M DL DG tau k Pd V10 if t <= 1 XL = 0.5; else XL = 0.6; end delta = x(1); omega = x(2); V2 = max(x(3),0); dx(1,1) = omega - (V10*V2*sin(delta)/XL-Pd)/DL; dx(2,1) = (Pd-V10*V2*sin(delta)/XL-DG*omega)/M; dx(3,1) = (-V2*V2/XL+V10*V2*cos(delta)/XL-k*Pd)/tau;
542
Matlab Example
Initialization and main routine:

clear all global M DL DG tau k Pd V10 XL = 0.5; M = 0.1; DL = 0.1; DG = 0.01; tau = 0.01; k = 0.25; Pd = 0.7; V10 = 1; x0 = [0.4636; 0.0000; 0.7826]; tmax = 2; [t,x] = ode23(example,[0 tmax],x0);
543
Matlab Example
Graphical commands:
figure plot(t,x(:,1),b-) hold on plot(t,x(:,2),g--) plot(t,max(x(:,3),0),c-.) plot([0 tmax],[V10 V10],r:) legend(delta,omega,V2,V1) xlabel(t [s]) ylim([-1 6])
544
Matlab Example
The dynamic solution without limits:

6
V2 V1
3 Operating point Contingency
t [s]
545
Matlab Example
Differential equations without limits:

function dx = example(t,x) global M DL DG tau k Pd V10 if t <= 1, XL = 0.5; else, XL = 0.6; end delta = x(1); omega = x(2); V2 = max(x(3),0); V10 = 1; Q = V10*V10/XL - V10*V2*cos(delta)/XL; if Q > 0.5 a = 1/XL; b = -V2*cos(delta)/XL; c = -0.5; V10 = (-b + sqrt(b*b - 4*a*c))/2/a; end dx(1,1) = omega - (V10*V2*sin(delta)/XL-Pd)/DL; dx(2,1) = (Pd-V10*V2*sin(delta)/XL-DG*omega)/M; dx(3,1) = (-V2*V2/XL+V10*V2*cos(delta)/XL-k*Pd)/tau;
546
Matlab Example
The dynamic solution with limits:

6
V2 V1
3 Operating point 2 Contingency
t [s]
547
PSAT Features
PSAT has been thought to be portable and open source. PSAT runs on the commonest operating systems PSAT can perform several power system analysis:
1. Continuation Power Flow (CPF); 2. Optimal Power Flow (OPF); 3. Small signal stability analysis; 4. Time domain simulations.
548
PSAT Features
PSAT deeply exploits Matlab vectorized computations and sparse matrix

functions in order to optimize performances.
PSAT also contains interfaces to UWPFLOW and GAMS which highly

extend PSAT ability to solve CPF and OPF problems, respectively.
549
Synoptic Scheme
Input
Simulink Models Other Data Format Saved Results
Data Files Simulink Library
Simulink Model Conversion
Power Flow & State Variable Initialization
Conversion Utilities
Settings
User Defined Models
Interfaces
GAMS Static Analysis Optimal Power Flow UWpflow Continuation Power Flow Dynamic Analysis Small Signal Stability Time Domain Simulation
PSAT
Command History
Plotting Utilities
Output
Text Output
Save Results
Graphic Output
550
PSAT Features
In order to perform accurate and complete power system analyses, PSAT

supports a variety of static and dynamic models.
Dynamic models include non conventional loads, synchronous machines

and controls, regulating transformers, FACTS, wind turbines, and fuel cells.
551
PSAT Features
Besides mathematical algorithms and models, PSAT includes a variety of

additional tools, as follows: 1. User-friendly graphical user interfaces; 2. Simulink library for one-line network diagrams; 3. Data le conversion to and from other formats; 4. User dened model editor and installer; 5. Command line usage.
552
PSAT Features
Not all features are available on GNU-Octave:

Function Continuation power ow Optimal power ow Small signal stability analysis Time domain simulation GUIs and Simulink library Data format conversion User dened models Command line usage Matlab yes yes yes yes yes yes yes yes GNU/Octave yes yes yes yes no yes no yes
553
Getting Started
PSAT is launched by typing at the Matlab prompt:

>> psat
which will create all structures required by the toolbox and open the main GUI.
All procedures implemented in PSAT can be launched from this window

by means of menus, buttons and/or short cuts.
554
Getting Started
Main PSAT GUI:
555
Simulink Library
PSAT allows drawing electrical schemes by means of pictorial blocks. The PSAT computational engine is purely Matlab-based and the Simulink
environment is used only as graphical tool.
A byproduct of this approach is that PSAT can run on GNU/Octave, which

is currently not providing a Simulink clone.
556
Simulink Library
PSAT-Simulink Library:
557
Other Features
To ensure portability and promote contributions, PSAT is provided with a

variety of tools, such as a set of Data Format Conversion (DFC) functions and the capability of dening User Dened Models (UDMs).
The set of DFC functions allows converting data les to and from formats
commonly in use in power system analysis. These include: IEEE, EPRI, PTI, PSAP, PSS/E, CYME, MatPower and PST formats. On Matlab platforms, an easy-to-use GUI handles the DFC.
558
Data Format Conversion
GUI for data format conversion:
559
User Dened Models
The UDM tools allow extending the capabilities of PSAT and help
end-users to quickly set up their own models.
Once the user has introduced the variables and dened the DAE of the
new model in the UDM GUI, PSAT automatically compiles equations, computes symbolic expression of Jacobians matrices and writes a Matlab function of the new component.
Then the user can save the model denition and/or install the model in
PSAT.
If the component is not needed any longer it can be uninstalled using the
UDM installer as well.
560
User Dened Models
GUI for user dened models:
561
Command Line Usage
PSAT is provided with a command line version. This feature allows using
PSAT in the following conditions: 1) If it is not possible or very slow to visualize the graphical environment (e.g. Matlab is running on a remote server). 2) If one wants to write scripting of computations or include calls to PSAT functions within user dened programs. 3) If PSAT runs on the GNU/Octave platform, which currently neither provides GUI tools nor a Simulink-like environment.
562
Power System Model
The standard power system model is basically a set of nonlinear

differential algebraic equations, as follows:
x = f (x, y, p) 0 = g(x, y, p)
Rn ; y are the algebraic variables y Rm ; p are the independent variables p R ; f are the differential equations f : Rn Rm R Rn ; and g are the algebraic equations g : Rm Rm R Rm .
where x are the state variables x
563
Power System Model
PSAT uses these equations in all algorithms, namely power ow, CPF,
OPF, small signal stability analysis and time domain simulation.
The algebraic equations g are obtained as the sum of all active and
reactive power injections at buses:
where gpm and gqm are the power ows in transmission lines, M is the
T T set of network buses, Cm and [gpc , gqc ]T are the set and the power
g g g p = pm pc m M g(x, y, p) = gq gqm cCm gqc
injections of components connected at bus m, respectively.
564
Component Models
PSAT is component-oriented, i.e. any component is dened

independently of the rest of the program as a set of nonlinear differential-algebraic equations, as follows:
xc Pc Qc
= fc (xc , yc , pc ) = gpc (xc , yc , pc ) = gqc (xc , yc , pc )
where xc are the component state variables, yc the algebraic variables (i.e.
V and at the buses to which the component is connected) and pc are independent variables. Then differential equations f are built concatenating fc of all components.
June 26, 2008
Software Tools - 53
565
Component Models
These equations along with Jacobians matrices are dened in a function

which is used for both static and dynamic analyses.
In addition to this function, a component is dened by means of a

structure, which contains data, parameters and the interconnection to the grid.
566
Component Models: Example
Lets consider the exponential recovery load (ERL). The set of differential-algebraic equations are as follows: xc1 xc2 Pc Qc = xc1 /TP + P0 (V /V0 )s P0 (V /V0 )t = xc1 /TP + P0 (V /V0 )t
= xc2 /TQ + Q0 (V /V0 )s Q0 (V /V0 )t = xc2 /TQ + Q0 (V /V0 )t
where and P0 , Q0 and V0 are initial powers and voltages, respectively, as given by the power ow solution.
Observe that a constant PQ load must be connected at the same bus as the ERL to determine the values of P0 , Q0 and V0 .
567
Component Data:
Column 1 2 3 4 5 6 7 8 9 10 Variable Description Bus number Power rating Active power voltage coefcient Active power frequency coefcient Real power time constant Reactive power time constant Static real power exponent Dynamic real power exponent Static reactive power exponent Dynamic reactive power exponent Unit int MVA kV Hz s s -
Sn Vn fn TP TQ s t s t
568
Exponential recovery loads are dened in the structure Erload, whose

elds are as follows: 1. 2. 3. 4. 5. 6.
con: exponential recovery load data. bus: Indexes of buses to which the ERLs are connected. dat: Initial powers and voltages (P0 , Q0 and V0 ). n: Total number of ERLs. xp: Indexes of the state variable xc1 . xq: Indexes of the state variable xc2 .
569
PSAT Example
This section illustrates some PSAT features for static and dynamic
stability analysis by means of the IEEE 14-bus test system.
All data can be retrieved from the PSAT web site:

http://www.power.uwaterloo.ca/fmilano/
570
PSAT Example
IEEE 14-bus test system:
Bus 13
|V| = 1.047 p.u. <V = 0.2671 rad
Bus 14 Bus 10
|V| = 1.0318 p.u. <V = 0.2622 rad |V| = 1.0207 p.u. <V = 0.2801 rad
Bus 12
|V| = 1.0534 p.u. <V = 0.2664 rad
Bus 11
|V| = 1.0471 p.u. <V = 0.2589 rad
Bus 09
|V| = 1.0328 p.u. <V = 0.2585 rad
Bus 07
|V| = 1.0493 p.u. <V = 0.2309 rad
Bus 06
|V| = 1.07 p.u. <V = 0.2516 rad
Bus 04
Bus 08
|V| = 1.09 p.u. |V| = 1.012 p.u. <V = 0.2309 rad <V = 0.1785 rad
Bus 05 Bus 01
|V| = 1.06 p.u. <V = 0 rad |V| = 1.016 p.u. <V = 0.1527 rad
Breaker
Bus 02
|V| = 1.045 p.u. <V = 0.0871 rad
Breaker Bus 03
|V| = 1.01 p.u. <V = 0.2226 rad
571
PSAT Example
Power ow report:
572
PSAT Example
Continuation power ow analysis (GUI):
573
PSAT Example
Continuation power ow analysis (plots):
574
PSAT Example
Nose curves at bus 14 for different contingencies for the IEEE 14-bus test
system:
Base Case
1 0.8
Voltage [p.u.]
Line 2-4 Outage Line 2-3 Outage
0.6 0.4 0.2 0
1.1
1.2
1.3
1.4 c
1.5
1.6
1.7
1.8
575
PSAT Example
Optimal power ow analysis (GUI):
576
PSAT Example
Comparison between OPF and CPF analysis:

Contingency
BCP
[MW]
[p.u.] 0.7211 0.5427 0.2852
M LC
[MW] 445.8 399.5 332.8
ALC
[MW] 186.8 148.6 73.85
None Line 2-4 Outage Line 2-3 Outage
259 259 259
Because of the denitions of generator and load powers PG and PL , one has c = + 1.
577
PSAT Example
Time domain simulation: It has been used a 40% load increase with respect to the base case
loading, and no PSS at bus 1. A Hopf bifurcation occurs for the line 2-4 outage resulting in undamped oscillations of generator angles.
A similar analysis can be carried on the same system with a 40% load
increase but considering the PSS of the generator connected at bus 1. In this case the system is stable.
578
PSAT Example
Time domain simulation (without PSS):

1.002 1.0015 1.001
Generator Speeds [p.u.]
1 - Bus 1 2 - Bus 2 3 - Bus 3 4 - Bus 6 5 - Bus 8
1.0005 1 0.9995 0.999 0.9985 0.998
10
15
Time [s]
20
25
30
579
PSAT Example
Eigenvalue analysis (with PSS):

10 8 6 4 2 Imag 0 2 4 6 8 10
0.8
0.6 Real
0.4
0.2
0.2
580
Project 1
Reproduce the examples illustrated in the following slides: Voltage Stability: Slides 341-350 (using UWPFLOW) Angle Stability: Slides 444-456 (using Matlab) Frequency Stability: 479-489 (using Matlab) The software UWPFLOW is freely available at:
http://thunderbox.uwaterloo.ca/claudio/software/pow.html
Universidad de Castilla-La Mancha June 26, 2008 Projects - 1
581
Project 1
Voltage stability (slides 341-350): Write the 3area.wsc data le in the EPRI data format (the format
is fully described in the UWPFLOW documetation).
Write the generator load change le 3area.k using the format

described in the UWPFLOW documentation.
Write a batch le for running the simulations as described in the

slides.
582
Project 1
Voltage stability (slides 341-350): Run a base case power ow, a continuation power ow with and
without enforcing voltage and current limits. Compute also the singular value index.
Report power ow results as given by UWPFLOW and plots PV

curves and the singular value index using Matlab.
583
Project 1
Angle stability (slides 444-456): Write a Matlab function with the system differential equations. Use a Matlab script le to initialize and solve the time domain simulation (function ode23). Using a trial-and-error technique nd the clearing time tc of the system for D = 0.1, D = 0.05 and D = 0.2. For each value of the damping D, provide plots of the rotor angle and the rotor speed .
584
Project 1
Frequency stability (slides 479-489): Write a Matlab function with the system differential equations. Use a Matlab script le to initialize and solve the time domain simulation (function ode23). Run the time domain integration for a 25%, 50% and 60% generation drop at t = 1 followed by a 25%, 50% and 60% load drop at t = 5,
respectively.
For each value of the generation and load drop, provide plots of , , E , V2 and V1 .
585
Project 2
Reproduce the results for the IEEE 14-bus tests system illustrated in the
paper: F. Milano, An Open Source Power System Analysis Toolbox, accepted for publication on the IEEE Transactions on Power Systems, March 2005, 8 pages.
The full paper as well as the software PSAT is available at:

http://www.power.uwaterloo.ca/fmilano/
586
Project 2
The IEEE 14-bus test system is provided wintin the PSAT main distribution (folder tests). For the base case power ow, the continuation power ow and the
optimal power ow routines, use the le:
d 014 dyn l10.mdl
For the time domain simulations without PSS, use the le:
d 014 dyn l14.mdl
For the time domain simulations with PSS, use the le:
d 014 pss l14.mdl
587
Project 2
Hints: For static analyses (PF, CPF, OPF), disable loading dynamic
components by checking the box Discard dynamic components in the GUI Settings (within the menu Edit in the main window).
To simulate a line outage in static analyses, uncheck the box initially

close box (within the Simulink block mask) of the breaker of the line that one wants to keep out.
588
Project 2
Hints: For OPF analysis, disable the base case powers in the GUI for OPF
Settings and set the weighting factor to 1 (maximization of the ditance to voltage collapse).
For time domain simulations, remember to uncheck Discard dynamic

components box and do not allow the conversion to PQ buses when the program asks for.

Power Sys Dynamics and Stability

Uploaded by

Document Information

Original Description:

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

Power Sys Dynamics and Stability

Uploaded by

Copyright:

Available Formats

Power System Dynamics and Stability

Power System Dynamics and Stability

Departamento de Ingeniera Electrica, Electronica, Automatica y Comunicaciones

Escuela Tecnica Superior de Ingenieros Industriales

Universidad de Castilla-La Mancha June 26, 2008 Introduction - 1

Power System Dynamics and Stability

Universidad de Castilla-La Mancha June 26, 2008 Introduction - 2

Power System Dynamics and Stability

Universidad de Castilla-La Mancha June 26, 2008 Introduction - 3

Power System Dynamics and Stability

Understand the modeling and simulation of power systems from phasor

Familiarize with basic concepts of computer modelling of electrical power

Universidad de Castilla-La Mancha June 26, 2008 Introduction - 4

Power System Dynamics and Stability

Outlines: System Modeling

Synchronous machine. Transformer. Transmission line. Cable. Loads.

Universidad de Castilla-La Mancha June 26, 2008 Introduction - 5

Power System Dynamics and Stability

Outlines: System Analysis

Universidad de Castilla-La Mancha June 26, 2008 Introduction - 6

Power System Dynamics and Stability

Outlines: Voltage Stability

Power System Dynamics and Stability

Outlines: Angle Stability

Power System Dynamics and Stability

Outlines: Frequency Stability

Universidad de Castilla-La Mancha June 26, 2008 Introduction - 9

Power System Dynamics and Stability

Outlines: Software Tools

Outlines. UWPFLOW. Matlab. PSAT.

Universidad de Castilla-La Mancha June 26, 2008 Introduction - 10

Power System Dynamics and Stability

A. R. Bergen and V. Vittal, Power systems analysis, Second Edition,

C. A. Caizares, Editor, Voltage stability assessment: concepts, practices

Universidad de Castilla-La Mancha June 26, 2008 Introduction - 11

Power System Dynamics and Stability

P. M. Anderson and A. A. Fouad, Power system control and stability, IEEE

J. Arrillaga and C. P. Arnold, Computer analysis of power systems, John

I. S. Duff, A. M. Erisman and J. K. Reid, Direct Methods for Sparse

J. Stoer and R. Bulirsch, Introduction to Numerical Analysis, Second

Universidad de Castilla-La Mancha June 26, 2008 Introduction - 12

Power System Dynamics and Stability

M. Ili and J. Zaborszky, Dynamics and Control of Large Electric Power c

F. Milano, PSAT, Power System Analysis Toolbox, available at

Journal papers and technical reports. Course notes available on line.

Universidad de Castilla-La Mancha June 26, 2008 Introduction - 13

Power System Dynamics and Stability

Reproducing examples presented in the slides using UWPFLOW and

Stability analysis of the IEEE 14-bus test system using PSAT.

Universidad de Castilla-La Mancha June 26, 2008 Introduction - 14

Power System Dynamics and Stability

Universidad de Castilla-La Mancha June 26, 2008 Introduction - 15

Power System Dynamics and Stability

Power System Dynamics and Stability

Universidad de Castilla-La Mancha June 26, 2008 Generator Modeling - 1

Power System Dynamics and Stability

Universidad de Castilla-La Mancha June 26, 2008 Generator Modeling - 2

Power System Dynamics and Stability

Generator with DC exciter and associated controls:

Exciter field rheostat (manual control)

Magnetic amplifier Transistor amplifier

Reference and voltage sensing Limiter sensing

Other sensing MG set PMG Regulator power M 9052