PAPER TITLE

Ajay Kumar Gudla and Saikat Chakrabarti

Department

of Electrical Engineering
Indian Institute of Technology Kanpur
Kanpur, Uttar Pradesh,India Tel.07753058469
Email: ajaykg@iitk.ac.in
Department of Electrical Engineering
Indian Institute of Technology Kanpur
Kanpur, Uttar Pradesh,India
Email: saikatc@iitk.ac.in

AbstractEstimating Rotor angle stability, on-line is possible

now with the advent of PMUs. In this paper, rotor angle
of synchronous generator is estimated using damper current
observers [11] and kalman filter. Here, measurements are taken
from hardware PMU. This method is tested on single machine
system connected to load. 7th order model of synchronous
machine is used. After disturbance occurs, in sub-transient and
transient state, damper winding observer along with RungeKutta method estimates rotor angle. In steady-state this method
switches to Kalman filter method. This helps us estimate the
rotor angle better.

I. INTRODUCTION
Modern Power Systems have to ensure desired level of
stability and security. As the loads keep on increasing and
generators are limited, modern day power systems are forced
to operate near the steady state stability limit. In case of
any change in operating condition and/or disturbance, the
system state changes from steady-state to dynamic state. The
rotor angle and speed of the synchronous generator are the
most important reference quantities for power system dynamic
assessment and control [1].
Phasor Measurement Units (PMUs) are used to monitor
the system state in real time. The potential applications of
PMU include determination of fast transient stability swings
[2], predicting a developing transient and initiating important
relays or other control actions [3][6]. In this paper, hardwired
PMU is used for measurement purpose.
The research done so far relies on classical and lower
order generator models to infer rotor angles from phasor
measurements and numerical computation of the rotor speeds
[7], [8]. The lower order models dont include effect of
damper windings, making the estimate of rotor position less
accurate during transient and sub-transient time interval. This
paper utilizes 7th order model of synchronous machine for
estimation of rotor angle.
II. DESCRIPTION OF MACHINE MODEL
In the present paper, 7th order Synchronous machine model
is used. The stator circuit consists of 3- armature windings,
rotor circuit consists of one field winding and 2 damper
windings, each in d-axis and q-axis. This model is similar
to model used in Matlab/Simulink environment. Here, q-axis

leads the d-axis. is the angle between phase-a and q-axis.

The following transformation matrix(T ) is used to transform
3 voltages and currents into qd0 form [16].

2
cos cos( 2
3 ) cos( + 3 )
2
2
sin sin( 2
(1)
T =
3 ) sin( + 3 )
3
0.5
0.5
0.5
All the voltages(both stator and rotor), can be written in the
following form [10].
di
(2)
dt
Here, the convention used for polarities of currents, is
as follows. Stator current is positive, when its direction
is out of terminals.Positive direction of field and damper
currents, when going into the machine. In the above eq.(2),
v = [vd vq v0 vf d 0 0]T , i = [id iq i0 if d ikd ikq ]T . The
Resistance matrix(R) is,

Ra Lq 0
0
0
MKQ

Ld
Ra
0 MF MKD
0

0
0
R
0
0
0
a
(3)

0
0
0
R
0
0
fd

0
0
0
0
Rkd
0
0
0
0
0
0
Rkq
v = Ri L

The inductance matrix(L) is,

Ld
0
0
MF
0
L
0
0
q

0
0
L
0
0

MF
0
0
L
F

MKD
0
0 MKR
0
MKQ 0
0

MKD
0
0
0
LKD
0

0
MKQ
0
MKR
0
LKQ

(4)

In above matrices, all parameters are in p.u. A nomenclature

list for above parameters is given in Table II. The Swing
equation of the generator can be written as,
2H d
.
= Tm Te = Ta
s dt

(5)

Where, Tm , Te , Ta are mechanical, electrical and acceleration

torque of Synchronous machine, in N-m, respectively. , s
are actual and Synchronous speed of machine, in rad/sec. If

we convert speed, time in above equation into p.u. and write

it in the form of state-space equations, then we get
d
Ta
=
(6)
dt
2Hs
d
=1
(7)
dt
In the above eq.(7), is rotor angle. From eq.(2), (6), (7) we
can derive the state space form, with state vector [iT ]T and
input vector [v T Tm 1]T . It is to be noted that, in this state
space form, the obtained state matrix is time varying because
of the presence of speed, stator inductance terms. They change
during transients.
III. DAMPER WINDING OBSERVER DESIGN
An observer is a system, that can give an estimate of
the internal state of a system, by using output and input
measurements of that system. When a state of a system
is non-observable or observed, but noise is included in the
measurement, then we can design special observers for those
states. In our present system, state vector includes damper
winding currents ikd , ikq , which can not be measured by any
physical methods. So, we need to design an observer, to
observe these damper currents. Here, the observer is designed
from state equations [11], [15]. From eq.(2), take the equations
with damper winding voltages(= 0).
vkd = 0 = Rkd ikd

MKR dif d
LKD dikd
MKD did
+
+
R dt
R dt
R dt
(8)

LKQ dikq
MKQ diq
+
(9)
R dt
R dt
Now, write Euler approximation, for first order derivative
of current. Then, we get
vkq = 0 = Rkq ikq

1 Rkd R t
MKD
)+
(id (n + 1)
LKD
LKD
MKR
(if d (n + 1) if d (n))
id (n))
LKD
(10)
1 Rkq R t
MKQ
ikq (n + 1) = ikq (n)(
)+
(iq (n + 1)
LKQ
LKQ
iq (n))
(11)

ikd (n + 1) = ikd (n)(

This observer is able to estimate the damper winding

currents in transient condition also. It uses information of other
states and previous knowledge about damper currents.
IV. ESTIMATION OF STATOR INDUCTANCES
In Synchronous machine, the stator mutual and self inductances vary with rotor position. These variations are because
of variations in the permeance of magnetic flux path, which is
due to non-uniform air-gap. If we neglect variation in air-gap
due to stator slots, rotor circuit doesnt see any variation in
permeance. It is because of cylindrical structure of stator. So,
rotor position doesnt effect rotor circuit inductances. [9]

After qd0 transformation, the inductances become constant.

So, they are independent of rotor position. Here, saturation
effect is not considered. This is the case in steady-state. But,
when a sudden change occurs, we can see a considerable
change in Ld , Lq during transient period. The reason is, as
follows. When disturbance comes, sudden change in armature current takes place. This causes increse in flux linkage
due to current. But, flux that links the field doesnt change
immediately. So, this disturbance flux causes changes in airgap flux. Now, current is induced in damper windings. The
flux produced by it, will not allow disturbance flux into rotor.
Here, the air-gap flux path changed, which causes change in
reluctance. And, this changes reactance. [11]
Since, damper winding current decays quickly, we dont
consider that effect in transient period. Still, some opposition
will be there for disturbance flux. So, small changes in original
inductance. In steady-state, disturbance flux lines are super
imposed on these field flux lines. So, air-gap flux lines comes
back to normal path. So, variation in inductance or reactance
vanishes.
To estimate change in Ld , Lq , an estimator is designed in
this section, based on d-axis and q-axis voltage equations in
eq.().
MF dif d
base dt
MKD dikd
Ld did
+

(12)
base dt
base dt
vq = Ra iq Ld id + MKD ikd + MF if d
MKQ dikq
Lq diq
+

(13)
base dt
base dt
From above eq.(), (), we can write
#
"

did
1
.
i
Ld
q
base dt
=
diq
1
Lq
id
base . dt

id

 iq

Ra
0
0
0
.MKQ
if d

0
Ra MF MKD
0
ikd
ikq

id
# i 
"

MKD
F
q
0
0 0 M
vd
b ase
b ase

MKQ
if d + vq
0 0
0
0
ikd
b ase

ikq
(14)
vd = Ra id + Lq iq MKQ ikq +

In the above eq.(14), except Ld , Lq , all other parameters

are known, whether they are machine parameters or taken as
measurements or estimated. So, we can estimate Ld , Lq from
above esimator.
V. KALMAN FILTER METHOD
Kalman filter is used to estimate the state of a controlled
process. Here, we are using discrete Kalman filter. The state
equations of the system are

We can write above equation (22), in the form

xk+1 = Axk + Buk + vk
zk = Hxk + wk

k + Bu
k
xk+1 = Ax

Where xk is system state vector, uk is input vector, zk is

noisy measurement or output vector. Variables vk , wk represent
the process and measurement noise respectively. Here, we
assume that both noises are independent. They have known
co-variance matrices Qk , Rk respectively. By using feedback
control, Kalman filter estimates the state. First, Kalman filter
estimates the state at some time. Then, it gets the feedback, in
the form of noisy measurements. So, we can divide Kalman
filter equation into 2 parts. (i) time update equation or prediction step (ii) measurement update equation or correction
step.
We can write Kalman filter algorithm as follows.
1) give initial values for x
k , Pk
2) Prediction step:
a) Predict the current state.
xk + Buk
x
k+1 = A

(16)

b) Predict the error co-variance

P k+1 = APk AT + Q

(17)

These are used to obtain Priori estimate for next

prediction step.
3) Corrector step:
This step incorporates new measurement into Prediction
step Priori estimate to get an improved Posteriori estimate. These equations are responsible for feedback.
a) Calculate Kalman gain Kk+1
Kk+1 = P k+1 H T (HP k+1 H T + R)1

(18)

Here, Kalman gain is taken such that, the Posteriori

co-variance is minimized.
b) Correct the state estimate with measurement zk
x
k+1 = x
k+1 + Kk+1 (zk+1 H x
k+1 )

(19)

= B.t, I is identity matrix.

where A = (I + A.t), B
Above equation (23) is in the form of eq.(15)
In the present problem, we are taking damper currents
ikd , ikq and as Kalman filter state variables. And we
consider measurement from PMU, with noise. is
change in frequency or speed in p.u.. Other state equations in
eq.() are used in the method, to calculate stator inductances
Ld , Lq . Now, we get matrices A, B as follows.

A=

B=

Rkd .base
LKD

0
MKD .iq
2H

0
Rkq .base
LKQ
MKQ .id
2H

(20)

xk+1 xk
= Axk + Buk
t
= (I + A.t)xk + (B.t)uk

xk+1

(21)
(22)

(24)

Now, the measurement or output equation is,


 ikd
zk = = 0 0 1 ikq

(25)

(26)

VI. ESTIMATION OF ROTOR ANGLE

This section describes, how the rotor angle is estimated
using 2 methods, mentioned before. In the first method,
first well estimate damper winding curents using observers.
Then, we estimate stator inductances Ld , Lq using the method,
mentioned in section V. Then, we find acceleration torque,
using below formula.
Te = (Lq Ld )id iq + MF if d iq + MKD ikd iq
MKQ ikq id

4) goto step-2.
We can continue to run this, for the duration of interest.
k are priori and posteriori state
In the above algorithm, x
k , x
estimates of k th iteration. P k , Pk are Priori and Posteriori
estimate error co-variances of k th iteration.
In the current work, for programming purpose, we need
to convert continuous time equations into discrete time
equations(15). From the definition of time derivative of variable x and after using Euler theorem, we obtain
x k =

0
0

MKR .if d
MKD .id
LKD
LKD
MKQ .iq
LKQ
((Ld Lq )iq id )+Tm (MF if d iq )
2H

c) Correct the error co-variance

Pk+1 = (1 Kk+1 H)P k+1

(23)

(15)

Ta = Tm Te

(27)
(28)

Then, by solving swing equation mentioned in eq.(5) using

some numerical method, we can find speed and rotor angle.
This speed is taken as initial speed for starting next iteration
and at the end of the iteration, well estimate actual speed.
In the second method, we use Kalman filter to estimate
damper winding currents and speed. Then, using same method
in section V, we find Ld , Lq . These values are used to estimate
ikd , ikq and speed in next iteration. From speed, we can obtain
rotor angle, using same method used before. Both methods can
be understood better, by the flow chart in Fig. 1. In that figure,
t is time in sec. and T is sum of tansient and sub-transient
time.

ikd (p.u.)

0.5

0.5
0

10

15

20

25

30

Time (seconds)

Fig. 3. d-axis damper winding current ikd observed

1

ikq (p.u.)

0.5
0
0.5
1
1.5
2
0

10

15

20

25

30

Time (seconds)

Fig. 4. q-axis damper winding current ikq observed

1.04

estimated
actual

1.03
Speed (p.u.)

Fig. 1. Flow chart for rotor angle estimation

1.02
1.01
1
0.99
0.98
0

Fig. 2. Single line diagram of the test system

VII. RESULTS
By using the estimation algorithm described in section VI,
rotor angle of synchronous generator is estimated. The required measurements are taken from PMU, placed at generator
bus and from synchronous machine. Generator data is given
ib Table I. The single line diagram of the test system is
shown in fig.2. The model is simulated in OPAL-RT [] and
measurements are taken from a hardware PMU interfaced with
the OPAL-RT. PMUs data reporting rate is 30 times per
second. A 3 fault is applied at the end of transmission
line after the system reached steady-state. After 0.1 sec, that
fault is cleared.
Fig.3 and 4, shows damper winding currents ikd , ikq
observed by the observer in the section III. Before the fault,
damper currents are zero. After transient and sub-transient
period, they are settling to zero.
Speed of the machine, estimated using the methods mentioned in section VI is shown in fig.5 and fig.6. It is given in
p.u. And from speed, we estimated rotor angle. It is shown in
fig.7. and its manified view is shown in fig.8, which contains
transient and sub-transient parts.

10

15
Time (seconds)

20

25

30

Fig. 5. Estimated and actual Speed

VIII. CONCLUSION AND FUTURE WORK

In this paper, Kalman Filter method is applied to estimate
rotor angle of synchronous generator during transient and
TABLE I
T EST MACHINE DATA :
Quantities
Rated MVA
Rated KV
d-axis subtransient reactance(xd 00 )
d-axis transient reactance(xd 0 )
d-axis synchronous reactance(xd )
q-axis subtransient reactance(xq 00 )
q-axis transient reactance(xq 0 )
q-axis synchronous reactance(xq )
leakage reactance(xl )
Stator resistance(Ra )
d-axis subtransient open-circuit time constant (d0 00 )
d-axis transient open-circuit time constant
(d0 0 )
q-axis subtransient short-circuit time constant (q0 00 )
q-axis transient short-circuit time constant
(q0 0 )
Inertia co-efficient (H)

Unit

pu
pu
pu
pu
pu
pu
pu
pu
sec

Values
100
13.8
0.14
0.1813
1.3125
0.14
0.1813
1.2578
0.0742
0.0025
0.03

sec

5.89

sec

0.03

sec

0.6

sec

3.01

single machine connected to load through a transmission line.

Generally, PMU measurements contain noise. Kalman filter is
used to filter that noise in the measurement. So, this method
produces accurate results even in the presence of noise also.
Results shown in the section VII are based on measurement
from hardware PMU.
If saturation of parameters considered, it gives results with
more accuracy. If we can find the rotor angles of generators
in a power system on-line, we can reduce the network based
on synchrony.

TABLE II
D EFINITIONS OF VARIABLES AND CONSTANTS :
Parameters
MF
MKD
MKQ
MKR
LKD
LKQ
Rkd , Rkq
Rf d
Ld , Lq , L0
vq , vd , v0
vf d
iq , i d , i 0
if d
ikd , ikq

Parameter names
Stator to field mutual inductance
Stator to d-axis damper winding mutual inductance
Stator to q-axis damper winding mutual inductance
Rotor mutual inductance
d-axis damper winding self-inductance
q-axis damper winding self-inductance
d-axis, q-axis damper winding resistance
field resistance
d-axis, q-axis and zero sequence stator reactances
Stator voltages in qd0 axis
field voltage
Stator currents in qd0 axis
field current
d-axis, q-axis damper winding currents

estimated
actual

1.02
Speed (p.u.)

R EFERENCES

1.01
1
0.99
10

10.5

11

11.5
Time (seconds)

12

12.5

13

Fig. 6. Estimated and actual Speed : magnified view

55

Rotor angle (deg)

50
45
40
35
estimated
actual

30
25
0

10

15
Time (seconds)

20

25

30

Fig. 7. Estimated and actual Rotor angles

50

Rotor angle (deg)

45
40
35
30

estimated
actual

25
9.8

9.9

10

10.1
10.2
Time (seconds)

10.3

10.4

10.5

Fig. 8. Estimated and actual rotor angle: magnified view

steady-state. Present method uses field voltage and current

from the generator and the terminal measurements obtained
from the PMU. This method is tested on system in which a

[1] Angel, A.D., Glavic, M. and Wehenkel, L. Using Artificial Neural

Networks to Estimate Rotor Angles and Speeds from Phasor Measurements Proceedings of Intelligent System Applications in Power ISAP2003,
Lemnos, Greece, Paper ISAP03/017, 2003.
[2] C. W. Liu, M. C. Su, S. S. Tsay, and J. Y. Wang, Application of a
novel fuzzy neural network to real-time transient stability swings prediction
based on synchronized phasor measurements, IEEE Transactions on
Power Systems, vol. 14, No. 2, pp. 685692, May 1999.
[3] A. Diu, L. Wehenkel, EXaMINE Experimentation of a Monitoring and
Control System for Managing Vulnerabilities of the European Infrastructure for Electric Power Exchange, IEEE/PES Summer Meeting, Chicago,
USA, July 2002.
[4] M. Pavella, D. Ernst and D. Ruiz-Vega, Transient Stability of Power Systems: A Unified Approach to Assessment and Control, Kluwer Academic
Publishers, 2000.
[5] E. De Tuglio, M. Dicorato, M. La Scala, P. Scarpellini, A Corrective
Control for Angle and Voltage Stability Enhancement on the Transient
Time-Scale, IEEE Trans. on Power Systems, Vol. 15, No. 4, pp. 13451353, 2000.
[6] C. Taylor (Convener) CIGRE Task Force 38.02.17, Advanced Angle
Stability Controls, CIGRE Technical Brochure, No. 155, April, 2000.
[7] C.W. Liu and J. Thorp, Application of synchronized phasor measurements to real-time transient stability prediction, IEE Proc. Gener. Transm.
Distrib., Vol. 142, No.4, pp. 355-360, July 1995.
[8] S. Rovnyak, C-W. Liu, J. Lu, W. Ma and J, Thorp, Predicting future
behavior of transient Events Rapidly Enough to Evaluate remedial Control
Options in Real time, IEEE Transactions on Power Systems, Vol. 10, No.
3, pp. 1195-1203, 1995.
[9] Kundur,P. Power System Stability and Control Mc, Graw Hill, New
York,1994.
[10] Anderson,P.M., Fouad,A.A. Power System Control and Stability IEEE
Press, Piscataway, NJ,USA, 1993.
[11] Manish Jha Estimation of the Rotor Angle of Synchronous Generator
Using PMU Measurements, M.Tech. thesis, Dept. Elect. Eng., IIT Kanpur,
Kanpur, India, 2014.
[12] OPAL-RT real-time power system simulator (OPAL-RT). Available:
http://www.opal-rt.com/
[13] Welch, G., and Bishop, G. An Introduction to the Kalman filter Notes
of ACM SIGGRAPH tutorial on the Kalman lter, 2001.
[14] Simon,D. Optimal State Estimation John Wiley & Sons, 2006.
[15] E. Kyriakides and G. T. Heydt, On-line Estimation of Synchronous
Generator Parameters using a Damper Current Observer and a Graphic
User Interface IEEE Trans. Energy Conv., vol. 19, No.3, pp.499-507,
Sept. 2004.
[16] Bimal K. Bose, Modern Power Electronics and AC Drives PrenticeHall,Inc., NJ, 2002.
[17] IEEE Standard 115-1995: Test procedures for synchronous machines.
Part I: Acceptance and performance testing, Part II: Test procedures and
parameter determination for dynamic analysis.
[18] Esmaeil Ghahremani, Innocent Kamwa, Dynamic State Estimation in
Power System by Applying the Extended Kalman Filter With Unknown
Inputs to Phasor Measurements IEEE Trans. Power Sys., vol.26, No.4,
pp.2556-2566, Nov 2011.