ELEC0047 - Power system dynamics, control and stability

Transient stability analysis and improvement

Thierry Van Cutsem

[email protected] www.montefiore.ulg.ac.be/~vct

November 2017

1 / 34
 Transient stability analysis and improvement Introduction

Classification of power system instabilities

2 / 34
 Transient stability analysis and improvement Introduction

(Rotor) angle stability

most of the electrical energy is generated by synchronous machines

in normal system operation:
all synchronous machines rotate at the same electrical speed 2πf
the mechanical and electromagnetic torques acting on the rotating masses of
each generator balance each other
the phase angle differences between the internal e.m.f.’s of the various
machines are constant = synchronism
following a disturbance, there is an imbalance between the two torques and
the rotor speed varies
rotor angle stability deals with the ability to keep/regain synchronism after
being subject to a disturbance

3 / 34
 Transient stability analysis and improvement Introduction

Small-disturbance angle stability

Small-signal (or small-disturbance) angle stability deals with the ability of the
system to keep synchronism after being subject to a “small disturbance”
“small disturbances” are those for which the system equations can be
linearized (around an equilibrium point)
→ tools from linear system theory can be used (in particular eigenvalue and
eigenvector analysis)
following a small disturbance, the variation in electromagnetic torque Te can
be decomposed into:
∆Te = Ks ∆δ + Kd ∆ω
Ks ∆δ: synchronizing torque Kd ∆ω: damping torque
a decrease in synchronizing torque will eventually lead to aperiodic instability
(machine “going out of step”);
a decrease in damping torque will eventually lead to oscillatory instability
(growing oscillations).

4 / 34
 Transient stability analysis and improvement Introduction

Transient (angle) stability

Transient (angle) stability deals with the ability of the system to keep
synchronism after being subject to a large disturbance
typical “large” disturbances:
short-circuit cleared by opening of circuit breakers
more complex sequences: backup protections, line autoreclosing, etc.
the nonlinear variation of the electromagnetic torque with the phase angle of
the machine’s internal e.m.f. must be taken into account
→ numerical integration of the differential-algebraic equations is used to assess
the system response
unacceptable consequences of transient instability:
generators losing synchronism are tripped by protections (to avoid equipment
damages)
large angle swings create long-lasting voltage dips that disturb customers.

5 / 34
 Transient stability analysis and improvement Introduction

Remarks

Small-disturbance angle stability:

depends on operating point and system parameters
does not depend on the disturbance (assumed infinitesimal and arbitrary)
is a necessary condition for operating a power system (small disturbances are
always present !)

Transient stability:
depends on operating point and system parameters
depends on the disturbance also
- the system may be stable wrt disturbance D1 but not disturbance D2
- if so, the system is insecure wrt D2, but as long as D2 does not happen, it can
operate. . .

6 / 34
 Transient stability analysis and improvement Introduction

Objectives of this lecture

We focus on a simple system: one machine and one infinite bus

allows a complete analytical treatment
and, hence, a good understanding of behaviour
is a good introduction to more complex systems

central derivation: equal-area criterion

analogy with pendulum motion in Mechanics
large-disturbance stability analyzed through energy considerations
certainly the most classical power system stability development
(can be found even in the simplest textbooks on power system analysis!).

7 / 34
 Transient stability analysis and improvement “Classical model” of synchronous machine

“Classical model” of synchronous machine

1 damper windings d1 et q1 ignored
2 magnetic flux in f and q2 windings assumed constant
⇒ model valid over no more than ' 1 second
3 same transient reactances in both axes: Xd0 = Xq0 = X 0
4 stator resistance neglected.

V̄ + jX 0 I¯ = Ē 0 = E 0 ∠δ
Assumption # 2 ⇒ ϕ constant

Using a synchronous reference (angular speed ωN ):

d d 1 d
θ = (ϕ + δ) + ωN ⇒ δ = ωpu − 1
dt dt ωN dt
d
2H ωpu = Pmpu − Ppu
dt
8 / 34
 Transient stability analysis and improvement One-machine infinite-bus system under classical model

One-machine infinite-bus system under classical model

System with:
one synchronous machine represented by classical model with constant
mechanical power Pm
one infinite bus with constant voltage
some loads, represented as constant impedances.

System reduction:
replace network, loads and infinite bus by a Thévenin equivalent
merge the machine reactance with the Thévenin impedances into Zeq
take the Thévenin emf as synchronous phase angle reference.
9 / 34
 Transient stability analysis and improvement One-machine infinite-bus system under classical model

1
Yeq = = Geq +j Beq = |Yeq | e jη
Zeq

in practice: − π/2 ≤ η ≤ 0

Active power produced by the machine:

P = real(Ē 0 I¯∗ ) = real[E 0 e jδ Yeq
∗
(E 0 e −jδ − Eeq )] = Geq E 02 − |Yeq |E 0 Eeq cos (δ − η)

P(0) = Geq E 0 (E 0 − Eeq )

Geq E 0
ν = η + arccos
|Yeq |Eeq

10 / 34
 Transient stability analysis and improvement Network configurations and equilibria

Network configurations and equilibria

Typical large disturbance:

short-circuit at t = 0, cleared by protections opening the faulted line at t = te

configuration symbol time interval stable equilibrium

pre-fault v t<0 (δvo , ωvo = 1)
fault-on d t ∈ [0 te ] -
post-fault p t > te (δpo , ωpo = 1)

a simpler sequence of events: “self-cleared” fault (no line opening).

Fictitious scenario, used in quick stability tests
a more complex sequence of events: line reclosing
the fault has been cleared −→ final configuration = pre-fault
the fault is permanent −→ new line tripping (usually stays open for quite
some time)
what follows applies to any sequence
in some cases, there may be no post-fault equilibrium (e.g. if many lines are
tripped to clear the fault). In this case, the system is unstable.
11 / 34
 Transient stability analysis and improvement Network configurations and equilibria

One Thévenin per configuration −→ one P(δ) function per configuration

The relative positions of the curves is realistic:

during fault: capability of evacuating power on the network much decreased
due to low voltage
post-fault: system weaker owing to the line tripping.

Pm and E 0 are supposed to remain constant throughout all configurations.

12 / 34
 Transient stability analysis and improvement The equal-area criterion

The equal-area criterion

Swing equation:
M δ̈ = Pm − P(δ)
2H
where M = and P(δ) changes from one configuration to another.
ωN

Multiplying by δ̇:
M δ̈ δ̇ = [Pm − P(δ)] δ̇
Integrating from 0 to t:
Z t
1
M δ̇ 2 − [Pm − P(δ)] δ̇ dt = C
2 0

Changing variable in the integral:

Z δ
1
M δ̇ 2 + [P(u) − Pm ]du = C
2 δv0

“kinetic” energy + “potential” energy = constant

13 / 34
 Transient stability analysis and improvement The equal-area criterion

Fault-on period (t ∈ [0 te ]): P(δ) = Pd (δ)

At t = 0 : δ̇ = 0 and δ = δv0 . Hence, C = 0 and

Z δ
1
M δ̇ 2 + [Pd (u) − Pm ]du = 0
2 δv0

At t = te : δ = δe and δ̇ = δ̇e . Hence

Z δe
1
M δ̇e2 = − [Pd (u) − Pm ]du (1)
2 δv0

14 / 34
 Transient stability analysis and improvement The equal-area criterion

Post-fault period (t ∈ [te ∞[): P(δ) = Pp (δ)

Z δe Z δ
1
M δ̇ 2 + [Pd (u) − Pm ]du + [Pp (u) − Pm ]du = 0
2 δv0 δe

Z δe Z δ Z δp0
1
M δ̇ 2 + [Pd (u) − Pm ]du + [Pp (u) − Pm ]du + [Pp (u) − Pm ]du = 0
2 δv0 δp0 δe

The 2nd and 4th terms being constant:

Z δ
1
M δ̇ 2 + [Pp (u) − Pm ]du = K
2
| {z } δp0
Vk
| {z }
Vp

Vk : kinetic energy

Vp : potential energy in final configuration with reference in δp0

15 / 34
Transient stability analysis and improvement The equal-area criterion

16 / 34
 Transient stability analysis and improvement The equal-area criterion

for small enough te values, the system trajectory in the (δ, δ̇) state space, is a
closed curve along which:

∀t ≥ te : Vk (t) + Vp (t) = K = Vk (te ) + Vp (te )

at points C and C’: potential energy is maximum, kinetic energy is zero

at points D and D’: potential energy is zero, kinetic energy is maximum
there is a te value for which the trajectory passes through the unstable
equilibrium point (δpi , 1). For this and for larger te values, stability is lost
the domain of attraction of the post-fault stable equilibrium (δp0 , 1) is the
part of the (δ, δ̇) state space bounded by that trajectory; instability results
from a lack of attraction towards this equilibrium
if the system model was dissipative (as the real system is, e.g. to due
damping) the stable trajectories would spiral up to (δp0 , 1) and Vk (t) + Vp (t)
would decrease with time

17 / 34
 Transient stability analysis and improvement The equal-area criterion

Stability condition

System stable if (δe , δ̇e ) belongs to the domain of attraction of (δp0 , 0), i.e. if
Z δe Z δpi
1
M δ̇e2 + [Pp (u) − Pm ]du < [Pp (u) − Pm ]du
2 δp0 δp0

or, using (1):

Z δe Z δe Z δpi
− [Pd (u) − Pm ] + [Pp (u) − Pm ]du < [Pp (u) − Pm ]du
δv0 δp0 δp0

Z δe Z δpi
⇔ [Pm − Pd (u)]du + [Pm − Pp (u)]du < 0
δv0 δe

⇔ Aacc − Adec < 0

Aacc = “accelerating area”, corresponding to P(δ) < Pm

Adec = “decelerating area”, corresponding to P(δ) > Pm .
18 / 34
 Transient stability analysis and improvement Critical clearing time

Critical clearing time

Maximum duration tc of the fault-on period after which the system still
returns to (or remains in the neighbourhood of) its post-fault equilibrium

Let us denote by δc the rotor angle at t = tc , i.e. δc = δ(tc ).

The system is at the stability limit when δe = δc and Aacc − Adec = 0, or

Z δc Z δpi
[Pm − Pd (u)]du + [Pm − Pp (u)]du = 0
δv0 δc

solve the above equation with respect to δc

integrate the system trajectory in the fault-on configuration, to find the time
tc such that δc = δ(tc ).
except in some simple configurations (e.g. Pd = 0), this integration has to be
performed numerically.

19 / 34
 Transient stability analysis and improvement Extensions of the equal-area criterion

Extensions of the equal-area criterion

Accounting for damping

stability becomes asymptotic

equal-area criterion pessimistic in terms of δc and tc
the first angle deviation is little decreased by damping. Damping is effective
in subsequent oscillations, for which a more detailed model is required.

20 / 34
 Transient stability analysis and improvement Extensions of the equal-area criterion

Two-machine system
M1 δ̈1 = Pm1 − P1 (δ1 − δ2 ) M2 δ̈2 = Pm2 − P2 (δ1 − δ2 )
can be combined into:
 
Pm1 Pm2 P1 (δ1 − δ2 ) P2 (δ1 − δ2 )
δ̈1 − δ̈2 = − − − (2)
M1 M2 M1 M2

Defining:
M1 M2
M12 = δ12 = δ1 − δ2
M1 + M2
Pm1 M2 − Pm2 M1 P1 (δ12 ) M2 − P2 (δ12 ) M1
Pm12 = P12 =
M1 + M2 M1 + M2
(2) can be rewritten as:
M12 δ̈12 = Pm12 − P12 (δ12 )
which is the swing equation of an equivalent one-machine infinite-bus system.

Remark. If M1 = M2 = M, then M12 = M/2 : two machines of inertia M oscillate

against each other as one machine of inertia M/2 against an infinite bus.
21 / 34
 Transient stability analysis and improvement Extensions of the equal-area criterion

Extensions to multi-machine systems ?

rigorously speaking, the equal-area criterion does not apply to systems with
more than two machines
but the underlying energy concept inspired much research into “direct
methods” for transient stability analysis
it also inspired “hybrid” methods:
detailed time simulation complemented with stability assessment inspired of
equal-area criterion
relying on a two-machine equivalent: one machine corresponds to the
machine(s) loosing synchronism, the other machine to the rest of the system

the concept of critical clearing time tc applies, whatever the complexity of

the model
“critical group”: the set of machines which loose synchronism with respect to
the remaining of the system, for a clearing time a little larger than tc .

22 / 34
 Transient stability analysis and improvement Transient stability improvement

Transient stability improvement

Decrease the accelerating area and/or increase the decelerating area

23 / 34
 Transient stability analysis and improvement Transient stability improvement

Modifying the pre-disturbance operating point:

reducing the active power generation
operating with higher excitation

Automatic emergency controls:

actions on network:
line auto-reclosing
fast series capacitor reinsertion
fast fault clearing - single pole breaker operation
actions in generators:
(turbine) fast valving
generation shedding
action on “load”: dynamic braking

Other means:
equip generators with fast excitation system
control voltage at intermediate points in a long corridor: through
synchronous condensers or static var compensators.
24 / 34
 Transient stability analysis and improvement Modifying the pre-disturbance operating point

Modifying the pre-disturbance operating point

Active power generation

Decreasing Pm :
decreases the accelerating area
increases the decelerating area
(side effect: pre- and post-fault equilibria also modified)

Reactive power generation

r
0 + 0 − Q 2 P
E (0 ) = E (0 ) = (V + X 0 ) + (X 0 )2
V V

For given values of V and P, increasing the reactive power production Q

increases the emf E 0 which prevails in the post-fault period
this, in turn, somewhat increases the magnitude of the P(δ) curve
and, hence, the decelerating area and the stability margin.

25 / 34
 Transient stability analysis and improvement Line auto-reclosing

Line auto-reclosing

coming back to pre-fault configuration increases the decelerating area

reclosing possible only after air has recovered its insulating properties
(delay ' 0.5 s).

26 / 34
 Transient stability analysis and improvement Fault clearing

Fault clearing

Fault clearing delay:

should be as short as possible
typical values : 5 cycles (0.1 s at 50 Hz)
stability must be checked with respect to scenarios where primary protection
fails clearing the fault (due to protection or breaker malfunction), which is
eliminated by the slower backup protection.

Single-phase tripping-reclosing:
most of the faults (' 75 %) are of the phase-ground type
for such faults, it is of interest to open the faulted phase only and keep the
other two in service ⇒ protect each phase separately
in case of 3-phase fault with malfunction of one breaker, the other two
operate and the fault changes into phase-ground (less severe, cleared by
backup protection)
at EHV level, the three poles of the breaker are usually separate (for insulation
reasons); it does not cost much to add a separate control on each phase.
27 / 34
 Transient stability analysis and improvement Turbine fast valving

Turbine fast valving

Principle

Decrease the mechanical torque as fast as possible

Pm (δ) curves obtained from the Pm (t) and δ(t) evolutions (eliminating t)

fast valving not fast enough to act during fault

main effect = increase of decelerating area.
28 / 34
 Transient stability analysis and improvement Turbine fast valving

Speed of action
mechanical torque must be decreased rapidly: typically less than 0.5 s
gates of hydro turbines cannot be moved so quickly ⇒ applies to steam
turbines
delays:
to take the decision from measurements (selectivity !)
to close the valves. They are closed much faster than in normal operating
conditions by emptying the servomotor of its oil.

Decision criterion
cannot rely on rotor speed only: due to inertia it takes time to reach an
emergency value
additional signal: rotor acceleration, drop of electrical power, difference
between electrical power and an image of mechanical power.

29 / 34
 Transient stability analysis and improvement Turbine fast valving

Implementation

Action:
on Control Valves (CV), normally controlled by Speed Governor (SG)
on Intercept Valves (IV): fully open in normal conditions, closed to cancel the
torque developed in Medium Pressure (MP) and Low Pressure (LP) bodies

30 / 34
 Transient stability analysis and improvement Turbine fast valving

Temporary valving:
Pm returns to its pre-disturbance value
typical sequence:
intercept valves are closed for a short time, then re-opened
control valves are left unchanged

Sustained valving:
Pm remains at a lower than pre-disturbance value
e.g. because post-disturbance network configuration expected to be too weak
even if it produces less, the generator remains synchronized with the network
typical sequence:
intercept valves are quickly closed
control valves are closed partially and at a lower speed
intercept valves are re-opened
control valves remain partially closed.

Valve re-opening:
after some delay due to servomotors
with a limited speed to avoid wearing the turbine, but this is not an issue if
stability has been preserved.
31 / 34
 Transient stability analysis and improvement Generation shedding

Generation shedding

Principle

Trip one or several generators in order to preserve synchronous operation of the

remaining generators
applies mainly to hydro plants
those including multiple generators
combinations of 1, 2, 3, . . . generators can be dropped
may be also used with thermal plants:
tripped generator not stopped, used to feed its own auxiliaries
(called tripping to houseload)
plant remains in operation and can be re-synchronized with shorter delay
usually applies to large power plants evacuating power through long corridors.

32 / 34
 Transient stability analysis and improvement Generation shedding

Simple illustrative example

Assumptions:
two identical machines oscillating in phase (“coherent” machines)
each machine: inertia M, transient reactance X 0 , mechanical power Pm
both machines = machine of inertia 2M, reactance X 0 /2, mech. power 2 Pm

Equal-area criterion:
area ABCD > area DEF ⇒ system unstable
ABCD area = energy acquired by both machines at the time the fault is
cleared.
33 / 34
 Transient stability analysis and improvement Generation shedding

Assume that one machine is tripped intentionally.

Assume for simplicity that this takes place at the time the fault is cleared.

The right figure relates to the remaining generator. Compared to the left figure:
the P(δ) curve has lower magnitude, since the reactance between the emf
0
and the infinite bus is Xe + X 0 instead of Xe + X2
the mechanical power is Pm instead of 2Pm .
The system is stable if area GHK is larger than half of area ABCD .
34 / 34