Power System Stability

Lectures prepared by

Prof. S. Shahnawaz Ahmed

•Definition of stability: steady state and transient stability

• Modeling for transient stability : classical model, swing

equation

•Evaluation of a two-machine system stability through

equal area criterion

•Evaluation of a multi-machine system stability through

step by step solution of swing equations

•Factors affecting stability: design , operation and control

related factors

•Frequency stability

•Voltage stability
For subsequent materials assistance taken mainly from :

[1]. John J. Grainger, William D. Stevenson, Jr. "Power System Analysis",

McGraw-Hill, 1994. Ch.16.

[2] LEONARD L. GRIGSBY, editor: POWER SYSTEM STABILITY AND CONTROL, CRC
Press, 3rd Edition, 2012

[3] Jan Machowski, Zbigniew Lubosny, Janusz W. Bialek and James R. Bumby:
Power System Dynamics Stability and Control, John Wiley & Sons, Ltd, UK, 2020.
Stability: This is the ability of a power system (having
multiple synchronous machines i.e. generators and
motors interconnected through a transmission
network) to attain the original or a new steady state
after it experiences respectively a small disturbance or
a large disturbance.

•This ability mainly stems from the adjustment

capability of synchronous machines by adjusting their
rotor angles in the event of any disturbance.
•The stability arising from small disturbance (e.g.
change in AVR gain parameter, gradual or slow
variation in load or generation etc.) is termed
steady state stability

•The stability arising from a large disturbance (e.g.

fault, line switching, sudden large load or
generation change) is termed transient stability.
•If the stability is lost depending upon the intensity of the
disturbance (e.g. a large change in load in a short time) then
it manifests in any or all of the forms of oscillating rotor
angle, rotor speed, line current and terminal voltage and
fluctuating or monotonically varying frequency eventually
leading to the loss of synchronism between the machines
and ultimately a blackout in the worst case.
•Whatever the disturbance may be an interaction between the
mechanical systems (i.e. turbine and its control elements) and
the electrical system (i.e. generator stator and rotor circuit,
damper winding and exciter) is involved.

•However, determination of stability requires modeling all or

some of the components depending upon the period of interest
and response times of the components. As for instance, for the
first swing or about a half cycle following a fault the stability can
be decided modeling only the generator with a constant emf
and a transient reactance while its exciter and turbine-governor
are not considered in the model.
•Steady state stability (also termed small signal stability)
analysis can be done by linearizing the model equations
around an initial operating point.

•The model equations termed swing equations are second

order differential equations each relating rotor
acceleration of a machine to the balance between its
mechanical power input and electrical power output.

•Transient stability analysis requires time-domain solution

of the swing equations step by step i.e. every Δt for
multiple machines. However, for only two-machine
system a criterion known as equal area criterion can be
developed to determine transient stability without
requiring to solve swing equation.
Swing equation’s different variants

(Multiplying both sides by ωm and substituting P = T ωm )

M = Jωm is inertia constant or angular momentum

of rotor at sync. speed

when H is another inertia constant as follows

or, H in pu-second on the machine’s rating as base

Since δ is the angle with respect to a

synchronously rotating reference axis, dδ/dt
is the difference from the synchronous speed
 If swing eqn. needs to be referred to
system base in a multi-machine system then
all powers are in pu of system base MVA (Ssystem )
on respective rating as base
and H needs to be converted as follows

implied to be on system base.

For two coherently swinging machines at the same power plant

If these are on same base for both machines

Example:
 For two non-coherent machines:

Power angle equation for a two-machine

system or one machine to infinite bus (OMIB) system
is termed power-angle eqn for machine 1
when armature loss neglected

The swing equation then becomes

-Pmax sinδ

Slope of power angle curve at an initial operating angle

is called synchronizing power coefficient (Sp) . If Sp is
positive that angle represents a valid point and a point of
stable equilibrium.

The angular frequency of oscillation of rotor angle

around an equilibrium point is
 prefault
During fault network
Post fault

During fault

Before fault the system was in equilibrium

condition when the machine was
delivering 1 pu power (equal to
mechanical input neglecting loss) to the
infinite bus and its rotor was at an angle
of 28.44° with respect to the
synchronously rotating reference frame.
However, for this small
Prefault: system a Δ-Y and then
another Y-Δ
transformation will also
provide the same
transfer admittance

Reminder: δ0 is the initial position of

the rotor relative to the synchronously
rotating axis before any disturbance.
When the system faces a disturbance
the rotor speed drifts from the
synchronous speed and the angle δ
also changes.
Reduction of Y matrix using Kron’s technique [Ref. P.M. Anderson and A.A. Fouad, Power System Control
and Stability, 2nd Edition, October 2002, Wiley-IEEE Press].
Example 16.4
Post fault network
 Equal Area Criterion (EAC)
This is a technique to examine stability of a two-machine or one machine to an infinite bus (OMIB) system without
numerical solution of the swing equation in the time domain i.e. δ vs. t.

Development of the criterion using a particular example:

•Before fault , the short line is ineffective and so electrical power

out put of the generator goes to other end and is equal to the
mechanical input from the turbine. (Pm = Pe at δ0 and no acceleration)

•During a three phase fault at point P (at t=0) on the short line
no power is transferred to the infinite bus and hence
electrical power output from the generator is zero. (Pe=0 so that Pa ≠0 i.e. the rotor
will oscillate)

•After the fault is cleared (t = tc), power same as before fault is transferred (and the
already developed acceleration becomes negative) .
Pre fault;
The rotor stays at equilibrium or steady state angleδ0

During fault at t=0, Pe drops sharply to zero

and Pm causes the rotor to accelerate and
move with more than sysnc. speed from b to c
(angle δc).

At t=tc when fault is cleared Pe rises sharply to d i.e. above Pm so

that acceleration becomes negative but due to inertia already the
rotor moved up to peak e and angle δx when its speed becomes
again the sync. speed . The rotor then starts slowing down and
descends from e to f (angle δy at which the speed again becomes
sync. speed) through a. But in its motion from a to f Pm exceeds Pe so
that again it accelerates and ascends up to e.
Equal area criterion says that if following a fault any one of the two pairs of areas (A1, A2)
or (A3,A4)can be obtained in the power angle curve of a synchronous machine such that
A1=A2 or A3=A4 then the machine will remain stable i.e. regain the synchronous speed .

This is more precisely stated as “the kinetic energy (KE) added to the rotor during a fault
must be removed after the fault is cleared in order to restore the rotor speed to sync.
speed”.
In fact the criterion for stability is A1 ≤A2 or A4 ≤A3.

The areas below Pm line represent addition of KE while those above Pm represent removal
of KE.

However, A1 =A2 or A3= A4 criterion covers up to a critically stable condition in which the
rotor relative angle can swing up to a maximum value δmax and still regain the synchronous
speed. The corresponding clearing angle and time are termed critical clearing angle and
critical clearing time.
Critical clearing angle: δcr

This is such an angle that when the rotor position is at or before this angle, the
fault must be cleared else the rotor will swing beyond a point δmax where Pm will
permanently be more than Pe and hence the machine will loose stability.

The time corresponding to δcr is called tcr (critical clearing time).

Analytical derivation of EAC:
Basic to this is finding two points (during and after clearing a fault)
where the rotor speed regains synchronous speed.

infinite bus is

Since δ is the angle with respect to a

synchronously rotating reference axis,
dδ/dt is the difference from the
synchronous speed
 Notes:

i.e. P-δ curve area below Pm line representing

kinetic energy added during fault should be equal
to the area above Pm line representing kinetic
energy removed after clearing fault in order for the
machine to be stable.
Analytical expression of δcr and tcr (i.e. when δ = δcr and t=tc=tcr )
(Finding tcr using power –angle curve is possible for the very
special case when electrical power transferred is zero (Pe=0) during fault)
During fault if Pe=0 then the swing equation becomes

Integrating it twice gives δ as in 16.53

Now δ at t=tc

Now, with respect to the power-angle curve showing critical

clearing angle (CCA)
Note: In practice relay and circuit breaker times are set to be
much lower than the critical clearing time.
Application of EAC to a more generalized case and finding CCA when output or transferred
electrical power during a fault is not zero.

.
P

Note: If r1=0 i.e. a fault at the sending end bus and r2=1 i.e. after the fault is cleared the original power
curve is followed then eqn 16.73 reduces to 16.67.
 Classical model

•Usually for transient stability analysis the

following fundamental assumptions are used.

Note: In practice computer models use detailed machine and

turbine models and solve a set of differential equations for
each machine employing 4th order Runge-Kutta method .
•Swing Equation is solved step-by-step based on classical
model to obtain the swing curve (δ vs. t)
 Notes:
1. By forming the pre fault bus admittance matrix and adding at the diagonal positions
the admittances obtained by converting loads (as follows) , the augmented admittance
matrix is formed.

2. Then the during fault condition is reflected appropriately in this augmented Ybus
and it is reduced to nxn matrix eliminating non generator buses when n is the number
of generators.

3. Similarly the after fault condition is reflected appropriately in this augmented Ybus
and it is reduced to nxn matrix eliminating non generator buses when n is the number
of generators.

4. The Pei in the swing equation of a generator ‘i’ is evaluated using an equation where
the admittances between generator buses i and j are taken from the reduced matrix
corresponding to during or after fault condition whichever is needed. e.g., in a 3 generator
system this eqn. is
Note: Conductance G has not been
neglected in the line admittances
 Discontinuity in Pa at the
Solution of swing equation using an algorithm suitable for “hand calculation”. If the step size Δt is middle of an interval
small then its accuracy increases. However, Δt =0.05 sec is often used for hand calculation purpose.

when δ is in electrical degrees

•In the methods for “hand calculation” the change in the
angular position of the rotor during a short interval of time
is computed by making the following assumptions:

1. The accelerating power Pa computed at the beginning of

an interval is constant from the middle of the preceding
interval to the middle of the interval considered.

2. Throughout any interval the relative angular velocity is

constant at the value computed for the middle of the
interval.

Δ
By an analysis of the load flow results:
Pm2 Pm1

δ10

δ20
For the system and given fault condition the power transferred and swing equations
during and after fault are as follows. (Y values are taken from corresponding reduced matrix
whose formation shown in latter slides)
During fault After fault
Pe1 Pe1 =0

Swing eqn for machine 1

Pe2

Swing eqn for machine 2

Notes:
Since the method assumes discontinuity in

can give more insight into the trend of frequency change

meaningful rather when averaged over several intervals
ωn = ω0 + ωr, n-1/2 ; ω should preferably be in radians/sec
Pa at the middle of an interval so that if an

fn= ωn /2π. However, frequency at every step is not

actual discontinuity occurs at the
beginning of an interval it must be
considered by taking average of the two
values. (δ0 )

(increase or decrease)
Δδn /Δt =ωr, n-1/2
δn = δn -1+ Δδn (in any step δn is updated
first )
Δδn = Δδn -1+ kPa,n-1 (in any step Δδn is
updated last for use in the next step)

≈0.5 sec =T
 Average frequency of
oscillation of δ2 is fr=
I/T ≈ 1/0.5 = 2 Hz

The system as a whole is unstable because machine 1

swing angle is varying indefinitely
Using “hand-calculation” algorithm Using 4th order Runge-kutta method and a computer program

Unstable Stable

Both Tables are outcome of classical model

based stability analysis
Note on Ybus formation and reduction:

Note: In load flow analysis the Ybus is formed without

considering machine nodes. But for stability analysis since the
Ybus matrix needs to be reduced among machines’ nodes, the
machine terminal bus is ‘moved’ to internal node so that machine
reactance is also added with transformer reactance in Ybus. e.g.
Bus 1 is now ‘behind’ Xd1’ instead of XT1.

Prefault Ybus: Xd1’, XT1 Admittance of the

load at a bus is
added with the bus
self admittance

Similarly other elements are computed

 Prefault Ybus need not be reduced as
Pe (before fault) is calculated by load flow.

During fault Ybus:

A three-phase fault at
bus 4 so it is same as
reference bus and
hence reflected by
simply deleting 4th row
and 4th column

Then reduce it to 3x3 matrix by

Kron’s formula
noting n= 3 (buses 1,2,3)
and r=1 (bus 5)
Post fault Ybus formation and reduction:

The fault is cleared by removing line 4-5

so that in prefault Ybus the elements Y44, Y55
need to be recomputed by subtracting from
them y45 (=1/(0.018+j0.11)
and half of corresponding line charging
susceptance (j0.226/2).

Also Y45= Y54 =0.0+j0.0

Then reduce it to a 3x3 matrix by Kron’s formula

noting that n =3 (buses 1,2,3) and r=2 (buses 4,5)

As explained physically , Y12 and Y21=0 in reduced matrix

implying machines 1nd 2 become uncoupled after the fault is cleared in this case.
FACTORS AFFECTING TRANSIENT STABILITY Means to improve stability:
a)Design strategies:
Swing equation: i)Choice of machines with high H and low Xd’
where possible.
so that ii) Reduction of transfer reactance X by parallel
lines, series compensation, decreasing
transformer reactance etc. so that Pe increases
when during a fault.
iii)Use of HVDC line to interconnect the areas
within a system or to interconnect two systems so
that oscillation of machines in one area /system
though this equation is valid for a special case will not spread to other area/system.

b) Operational strategies:
An analysis of the above set of equations shows that i) Increase machine excitation so that E and hence Pe
•A) acceleration (due to a fault) or deceleration (due to increases during a fault; this is performed by what is
known as PSS (Power System Stabilizer)
increase in electrical load) of a synchronous machine is ii) Fast valve control of turbines i.e. during a fault the
less and hence stability is more if turbine valve will close faster while during a sudden
1.H is higher i.e. machine rotor size is higher. increase in electrical load demand the valve will open
faster so that the difference between Pm and Pe and
2. The difference between mechanical input power (Pm) hence acceleration or deceleration reduces.
and electrical power transfer (Pe) is less. Pe can be iii) Faster fault clearing (much less than critical clearing
increased by increasing E i.e. excitation and decreasing time) using high speed relays and CBs.
iv) Selective pole operation of CBs to clear a fault i.e.
reactance X. healthy phases remain intact so that Pe after fault
increases and hence rotor acceleration becomes less.
•B) The critical clearing time increases i.e. a machine can v) In case generation i.e. mechanical power output from
turbine cannot be increased when needed, the last resort
retain stability for a longer time if is to shed electrical demand (load) using under frequency
1. H is higher auto load shed relays installed at substations, to reduce
2. Pre-fault power (equal to Pm neglecting loss) is lower. the difference between Pm and Pe .
 Rotor angle vs. frequency vs. voltgae stability:
A brief overview of rotor angle stability.
Rotor angle stability is concerned with the ability of interconnected synchronous machines of a power system
to remain in synchronism under normal operating conditions and after being subjected to a small scale or large
(i.e. transient stability) disturbance. It depends on the machines’ ability to maintain/restore equilibrium
between electromagnetic (i.e. electrical) torque and mechanical torque of each synchronous machine in the
system. A fundamental factor influencing this category of system stability is the manner in which the power
outputs of synchronous machines vary as their rotor angles change. The time frame of interest in transient
stability studies is usually limited to 3–5 s following the disturbance. It may extend to 10 s for very large systems
with many machines leading to several modes of oscillations.

Frequency stability:
A subtle difference between frequency and rotor angle stability:
Though change in rotor angle causes change in frequency, the frequency stability is concerned
with the ability of a power system to maintain steady frequency within a nominal range
following a severe system upset resulting in a significant imbalance between the overall system
generation and load. It depends on not only machines’ ability but also on the ability of the system
to restore balance between system generation and load, with minimum loss of load. Obviously a
frequency stability study needs modeling fast processes viz. primary control i.e. generators’
governor response and secondary control i.e. AGC actions, modeling voltage and frequency
sensitivity of loads, under frequency relay actions etc. and also slow processes such as boiler and
turbine dynamics. Usually the devices used for frequency control are not modeled in rotor angle
or voltage stability studies. The overall time frame of interest for frequency stability extends to
several minutes. Therefore, it is categorized as a long-term phenomenon.
Voltage stability:

Voltage stability is concerned with the ability of a power system to maintain steady voltages at all buses in the
system under normal operating conditions, and after being subjected to a small or a large disturbance. Voltage
instability may occur in both forms i.e. (i) progressive fall or rise and (ii) sustained fall at some or all the buses.

The possible outcome of voltage instability is loss of load in the area where voltages reach unacceptably low values, or
a loss of integrity of the power system.

The main factor contributing to voltage instability is usually the higher reactive power losses and the resulting voltage
drops that occur when high levels of active power and reactive power flow through inductive reactances associated
with the transmission network; this limits the capability of transmission network for power transfer. The power
transfer and voltage support are further limited when some of the generators hit their field current limits. The driving
force for voltage instability is usually the loads; in response to a disturbance resulting in network-wide voltage
reduction, power consumed by the loads tends to be restored by the actions of fast acting devices such as
electronically controlled motor loads, HVDC converters, under voltage load shed relays or slow acting devices such as
distribution voltage regulators, tap-changing transformers and thermostats . So the time frame of interest for voltage
stability problems may vary from a few seconds (short term) to tens of minutes (long term).

However, stability is usually determined by the resulting outage of equipment rather than the severity of the initial disturbance. In many
cases, static analysis techniques can be used to estimate stability margins.

A criterion for small disturbance voltage stability is that, at a given operating condition for every bus in the system, the bus voltage
magnitude increases as the reactive power injection at the same bus is increased. A system is voltage unstable if, for at least one bus in
the system, the bus voltage magnitude (V) decreases as the reactive power injection (Q) at the same bus is increased. In other words, a
system is voltage stable if V–Q sensitivity is positive for every bus and unstable if V–Q sensitivity is negative for at least one bus.

The loadability margins ( i.e. limits of increasing load until voltage collapse occurs) is one of the indices of voltage instability and this
can be computed using power flow based methods.
 All solutions to these power flow
equations, as the system load level
PL is increased, can be plotted to
yield PV curves (bus voltage vs.
active power load levels) or QV
curves (bus voltage vs. reactive
power load levels) for this system.

PV curve showing loadability limit for generator-load

example considering generator reactive power limits