10.

2478/v10048-010-0018-2

MEASUREMENT SCIENCE REVIEW, Volume 10, No. 3, 2010

Determination of Load Angle for Salient-pole

Synchronous Machine
D. Sumina1, A. ala2, R. Malari3
Faculty of electrical engineering and computing, University of Zagreb, Unska 3, 10000, Zagreb, Croatia
1
[email protected], 2 [email protected], 3 [email protected]
This paper presents two methods for load angle determination for salient-pole synchronous generator. The first method uses
optical encoder to detect the rotor position. In some cases the end of the rotor shaft is not free to be used and mounting of an
encoder is impossible. Therefore, the second method proposes estimation of the load angle based on the measured electrical values
that have been already used in excitation control system of the synchronous generator. Estimation method uses corresponding
voltage-current vector diagram and parameters of the synchronous generator, transformer and transmission lines. Both methods
were experimentally verified on the digital control system and synchronous generator connected to power system. The estimation
and measured results were compared. The accuracy of load angle estimation method depends on voltage and current
measurement accuracy as well as generator, transformer and transmission line parameter accuracy. The estimation method gives
satisfactory accuracy for load angles less than 120 el. Thus, it can be applied in excitation control system to provide stable work of
synchronous generator in under-excitation operating area.
Keywords: Load angle measurement, load angle estimation, synchronous generator

1. INTRODUCTION

angle can be used as a measure of how close the generator is

to pole slipping and can be also used to limit the excitation
current during its operation in capacitive mode [3]-[6]. The
application can be also possible in the implementation of the
power system stabilizer, where change of the load angle can
be used as a stabilizer input signal [7].
The load angle of a synchronous generator is the essential
variable for the transient stability studies of power systems
[1]. In recent years, an application of the load angle as an
input variable in the excitation control and protection
systems of synchronous generators and power systems has
been investigated [8], [9], [10]. Therefore, real-time
measurement of the load angle is required for these
applications.

SYNCHRONOUS GENERATOR connected to a power

system needs to remain synchronized in operating
conditions even in the case of extreme situations. The
exceeding of allowed loading would cause the activation of
the generator protection and disconnection of the
synchronous generator from the power system. Automatic
voltage regulators of synchronous generators have the
excitation current limitations dictated by P-Q diagram
(Fig.1). This enables optimal utilization of generator loading
and safer work of a generator operating in parallel with the
power system [1], [2]. If the generator excitation becomes
too weak for a particular active power, the magnetic flux in
the rotor will not develop the required torque to maintain the
generator in synchronism, and the rotor will start to pole
slip. Under-excitation (capacitive) operating mode of a
synchronous generator appears in systems with underloaded long power transmission lines, by connecting long
power transmission lines to voltage and by asynchronous
work of regulating transformer regulators.
The load angle is the angular displacement of the
generator's rotor from the no-load position referred to as
terminal voltage. Load angle provides information about the
generator operating point position in relation to the stability
limit. The load angle increases with increasing load and
under-excitation. Power system is characterized by
significant changes in synchronous generator performance,
so that the transition from one to another steady operating
condition is often accompanied by significant changes in the
dynamic load angle. This results in changes of the load
angle, in a relatively short period of time, from the
maximum to the minimum value. The load angle value is
important information especially in the case when generator
operates in under-excitation state, because the load angle
then assumes the maximum allowed value in relation to the
stability limit of synchronous generator operation. The load

Fig.1. Limitations in the P-Q diagram of the synchronous

generator

For the measurement of the load angle, the main problem

is the rotor position detection. The load angle measurement
system described in [11] uses an optical encoder to detect
the rotor position. The load angle was determined by a
procedure based on measured terminal voltage and rotor

89

MEASUREMENT SCIENCE REVIEW, Volume 10, No. 3, 2010

position signals. The load angle measurement for detecting

the rotor position based on a photoelectric sensor mounted
on the stator is proposed in [12]. Air gap sensor for
detection of rotor position was proposed in [13].
Disadvantage of these two methods is the mounting of
encoder or sensor on the generator. Several other papers
have described the procedures for indirect determination of
load angle [14]-[17]. Indirect determination is based on
available measured values.
This paper presents a load angle measurement using the
optical encoder. It also presents the load angle estimation
method for synchronous generator connected over the
transformer and transmission line to the power system
(Fig.2). The proposed estimation method uses the available
measured values that have already been introduced for the
purpose of the control system excitation. This method has
been implemented in the digital control system and requires
knowledge of generator and the equivalent network
parameters.

(channel A) has a frequency of 5000 pulses per revolution

and the second pulse train (null mark) has a frequency of 1
pulse per revolution.
After the generator is synchronized on the power system
with no load and with synchronous speed, calibration
procedure is performed. Digital control system detects the
time difference between the occurrence of the null mark
pulse and the occurrence of the next positive zero crossing
of the power system voltage. This time difference is the
offset or calibration time, shown as Tcal in Fig.3. As the
synchronous generator is loaded, the rotor position pulse
advances with respect to the zero crossing of the power
system voltage. Thus, the time difference between the
occurrence of the null mark pulse and the occurrence of the
next positive zero crossing of the power system voltage
changes to T in Fig. 3. In this way, the current load angle is
given as the time difference:
T = T Tcal

(1)

The load angle is determined by the following expression:

T
360
Tv

(2)

where Tv is the time period of the power system voltage.

If the time is expressed as the number of pulses, then the
load angle is determined by:

Fig.2. Synchronous generator connected to power system

2. LOAD ANGLE DETERMINATION

1
N
f0
=
360 = 360
1
Nv
Nv
f0
N

A. Load angle measurement

The load angle was measured using optical encoder and
the digital control system. The load angle was measured at
the exact moment of the networks phase voltage passing
through zero, which means that for the power systems
frequency of 50 Hz the sampling frequency is 100 Hz. The
angle is measured based on the difference between the
momentary position of the rotor and the network phase
voltage.

(3)

where N is the number of pulses between the occurrence of

the null mark pulse and the occurrence of the next positive
zero crossing of the power system voltage, and Nv is the
number of pulses in one power system voltage time period.
The resolution of load angle measurement is in this case
0.36 el.
B. Load angle estimation based on the quadrature-axis
synchronous reactance and the equivalent reactance of the
transformer and transmission line
For the load angle estimation voltage-current vector
diagram is used (Fig.4.), where 1 is the angle between
induced voltage E0 and generator voltage U, is the load
angle between induced voltage E0 and AC network voltage
Um, I is the armature current, R is the equivalent resistance
of stator, transformer and transmission line, Xe is the
equivalent reactance of the transformer and transmission
line and Xq and Xd are synchronous generator reactances.
In [17] the estimation method is proposed based only on
the quadrature-axis synchronous reactance Xq and equivalent
resistance R, whereas the method proposed in this paper
includes, in addition, the equivalent reactance of the
transformer and transmission line Xe.
The load angle is obtained from the diagram in Fig.4:

Fig.3. Load angle measurement principle

Fig.3 shows the principle of operation for measuring the

load angle. From the optical encoder, two pulse trains are
connected to the digital control system. The first pulse train

90

MEASUREMENT SCIENCE REVIEW, Volume 10, No. 3, 2010

= arctg

I Xq P I R Q + X e I P
Um S + I R P + Xq I Q + X e I Q

The standstill DC decay test consists in cutting off the

constant DC supply, allowing the current from the coils to
reach zero and keeping the rotor in a fix position. The dc
flux decay needs a dc source connected to two machine
stator phases and the field winding in short-circuit as it is
shown in Fig.5.
In order to perform the test, the synchronous generator
must be at standstill and positioned either on the d- or qaxis. Voltages and currents are measured with Hall-effect
sensors.
Initially, the switch Sa is closed and the stator windings are
excited with a constant DC current, which will set up a
magnetic flux density along the machine. A small DC
excitation current is usually applied; hence, the machine is
in the unsaturated condition. Once the stator excitation
current is at steady state, the switch Sb is closed to shortcircuit the generator stator terminals. The machine voltage
and winding currents are acquired at a rate of 5 kHz
throughout the transient. The decrement of the stator current
during short-circuit is used to determine reactances, time
constants and fundamental parameters of the synchronous
generator. Derived parameters can be readily obtained. The
test duration is considered to be finished when the stator
current has reached its steady state on the d- and q- axis; in
case of testing the direct axis, the field current must also be
at steady state. It is important to employ the field current in
the estimation process because the model will be able to
predict the field winding dynamic behavior with high
accuracy. The limiting resistor R will limit the current
applied to the machine windings. The value of the resistance
must guarantee a negligible current variation before and
after the test because the DC source is always connected.
The internal resistance of switch Sb must have a much lower
value than the stator winding resistance; with this
consideration, the decrement in the winding currents is not
affected. In addition, it is recommended that the current
sensor has a low internal resistance when compared to the
winding resistance.
The identification procedure starts from a mathematical
model, which consists of the machine voltage equations for
the d- and q-axis. The d- and q- axis current time-variation
is considered as being given by a sum of exponential
functions, which means, in a per-unit (p.u.):

(4)

where is the load angle, I is the armature current, P is the

active power, Um is the power system voltage, Q is the
reactive power and S is the apparent power. The values of
the quadrature-axis synchronous reactance Xq, equivalent
resistance R and reactance Xe must be known for this
estimation method. Measurement of the quadrature-axis
synchronous reactance Xq is presented in the next
subsection. Reactance Xe is calculated and in p.u. system is
equal to 0.1 p.u. and equivalent resistance is equal to 0.01
p.u. In section III it was determined that estimation method
in dynamic operations is not significantly dependent on the
change of reactance Xe.

Fig.4 Vector diagram of the synchronous generator connected to

AC network

C. Measurement of the quadrature-axis synchronous

reactance
Usually, the characteristic values of a synchronous
machine are obtained from a short-circuit test at no load
generator operation. This test implies risks and it cannot be
performed every time, so the DC decay test is an alternative
method. It determines the characteristic values of a
synchronous machine for two axes in function of the
saturation state of the leakage paths. In general, the DC
decay test is done for positions of the rotor in d- and q- axis
[18].

id (t )
= Ad e1d t + Bd e 2 d t + Cd e3 d t
id 0
iq (t )
iq 0

= Aq e

1q t

+ Bq e

2 qt

(5)

(6)

where the coefficients A, B, C, and are functions of

time-constants and reactances for each of the axes [18].
The developed identification program is based on the
Gauss-Newton curve fitting algorithm (Fig.6) [19]. Starting
from the expressions of the currents, the identification is
applied on parts of the curves. Estimated data are used as
initial values.

Fig.5. Experimental setup for DC decay test

91

MEASUREMENT SCIENCE REVIEW, Volume 10, No. 3, 2010

Fig.6. Structure of identification procedure

The identification method gives following results: Xd=0.8

p.u and Xq=0.59 p.u.
3. EXPERIMENTAL VERIFICATION OF LOAD ANGLE
DETERMINATION METHODS

Digital control system (Fig.7) based on NI cRIO-9012 [13]

was used for excitation control of synchronous generator, as
well as to implement the load angle determination methods.
The cRIO-9012 controller operates the National Instruments
LabVIEW Real-Time Module, while the connectivity to
sensors and actuators was obtained with four I/O modules,
two NI-9215 (16-bit analog input, 100KS/s), NI-9401
(digital I/O, 100 ns) and NI-9263 (16-bit analog output,
100KS/s) [20]. NI LabVIEW graphical programming tools
can be used for easier modeling of control algorithms. For
testing purposes a software monitoring tool was also
developed. It enables regulator parameters optimization as
well as displaying and recording of the testing results via
PC. Experimental setup is presented in Fig.8.

Fig. 8 Experimental setup

A. Static comparison of the measured and estimated load

angle
The static comparison of the measured and estimated load
angle was performed by three appropriate experiments.
Active power was changed from 0 p.u. to 1.0 p.u. with the
step change of 0.2 p.u. In the first experiment (Fig.9),
reactive power of the generator is zero.
30

Measured value
Estimated value
Difference

20

15

10

Difference / el

Load angle / el

25

0
0
0,0

0,2

0,4

0,6

0,8

1,0

Active power / p.u.

Fig.9. Measured and estimated load angle, and estimation error

with reactive power of 0 p.u. during active power change from 0
p.u. to 1.0 p.u.

In the second experiment (Fig.10) the reactive power is

inductive (1.0 p.u.) and is kept constant during the whole
range of active power change. In the third experiment
(Fig.11) reactive power is capacitive and is kept constant
(1.0 p.u.). In the whole range of active power change, the
difference between estimated and measured load angle is
less than 4 el.

Fig.7. Digital control system

92

MEASUREMENT SCIENCE REVIEW, Volume 10, No. 3, 2010

12

5
20
30

10
2

20
6
10
3
0

-1
0,0

0,2

0,4

0,6

0,8

1,0

Measured value
Estimated value
Difference

-10

Active power / p.u.

t/s

Difference / el

Load angle / el

15

Difference / el

Load angle / el

Measured value
Estimated value
Difference

-3
6

Fig.12. Measured and estimated load angle for active power step
change and reactive power of 0 p.u.

Fig.10. Measured and estimated load angle, and estimation error

with inductive reactive power of 1.0 p.u. during active power
change from 0 p.u. to 1.0 p.u.

25
6
20

30

-1

20

0,4

0,6

Measured value
Estimated value
Difference

-3
0

-3
0,2

-2

0
0,0

3
10

-5

Measured value
Estimated value
Difference

10

Difference / el

Load angle / el

40

15

Difference / el

Load angle / el

50

0,8

t/s

Fig.13. Measured and estimated load angle for active power step
change and inductive reactive power of 1.0 p.u.

1,0

Active power / p.u.

Fig.11. Measured and estimated load angle and estimation error

with capacitive reactive power of 1.0 p.u. during active power
change from 0 p.u. to 1.0 p.u.

200
200
150

100

Load angle / el

100

The estimated and measured results were also compared

with three different experiments. In these experiments there
was active power step change (at the moment of t 1s) from
0 p.u. to 1.0 p.u. and again to 0 p.u. after four seconds. In all
experiments the reactive power was kept constant using the
reactive power controller. In the first experiment (Fig.12)
the reactive power was set to 0 p.u. In the second
experiment the generator was in inductive operating mode
and reactive power was set to 1.0 p.u. (Fig.13). In the third
experiment (Fig.14) the generator was in capacitive
operating mode and has lost the synchronism.

50
50
0
0
-50
-50

-100

Measured value
Estimated value
Difference

-100

Difference / el

B. Dynamic comparison of the measured and estimated

load angle

150

-150
-200

-150
0

t/s

Fig 14. Measured and estimated load angle for active power step
change and capacitive reactive power of 1.0 p.u.

C. Sensitivity of estimation method due to error of

estimator parameters

Measured and estimated results demonstrate good

matching except near the moment when the generator loses
synchronism. In that moment the voltage-current vector
diagram used as a base in the presented estimation method is
not valid. For algorithms that need to keep the generator in
stable operating mode, operating range with load angles less
than 90 el. is relevant. So, if excitation control algorithm
ensures stable operating mode of a generator, load angle will
never exceed 90 el.

Parameters needed for the load angle estimation are

quadrature-axis synchronous reactance Xq, equivalent
resistance R and reactance Xe. An estimation method
dependence of parameters determining accuracy is analyzed
in capacitive operating mode with constant reactive power
(1 p.u.). There is active power step change (at the moment
t = 1 s) from 0 p.u. to 1.0 p.u. and again to 0 p.u. after four

93

MEASUREMENT SCIENCE REVIEW, Volume 10, No. 3, 2010

seconds. Stator resistance R, quadrature-axis reactance Xq

and equivalent reactance Xe in estimator are changed from
50 % to 150 % of nominal value with step of 50 %. In the
first set of experiments, value of stator resistance in
estimator is changed (Fig.15). Two different validation
criterions are used to determine sensitivity of estimation
method to parameter change. The first criterion is the mean
absolute error (MAE) between the measured and estimated
load angle. Second criterion is static error (STAT). This
error is defined as the absolute error between the measured
and estimated load angle 3 seconds after step change of
active power. MAE and STAT criterions for stator
resistance change are shown in Fig.16. In the second
experiment, value of reactance Xq in estimator is changed
(Fig.17). MAE and STAT criterions for this set of
experiments are shown in Fig.18. In the third experiment,
value of reactance Xe in estimator is changed (Fig.19). MAE
and STAT criterions for this set of experiments are shown in
Fig.20.

12

Difference / el

10

MAE
STAT

2
0,4

0,6

0,8

1,0

1,2

1,4

1,6

R / Rn

Fig.16. MAE and STAT criterions due to different stator

resistance R in estimator

70

60

60
50

40

40

Load angle / el

Load angle / el

50

30
20

R(estimator) = 0,5 Rn
Measured value
Estimated value

10

30

20

Xq(estimator) = 0,5 Xqn

Measured value
Estimated value

10

0
0

-10
-20

-10

t/s

70

60
50

50
40

Load angle / el

Load angle / el

t/s

60

30

R(estimator) = Rn
Measured value
Estimated value

20
10
0

40
30

Xq(estimator) = Xqn
Measured value
Estimated value

20
10
0

-10

-10

t/s

70

t/s
80

60

60

Load angle / el

50

Load angle / el

40
30

R(estimator) = 1.5 Rn
Measured value
Estimated value

20

40

20

Xq(estimator) = 1.5 Xqn

Measured value
Estimated value

10
-20

0
-40

-10
0

t/s

t/s

Fig.15. The measured and estimated load angle with different

stator resistance R in load angle estimator

Fig.17. The measured and estimated load angle with different

quadrature-axis reactance Xq in load angle estimator

94

MEASUREMENT SCIENCE REVIEW, Volume 10, No. 3, 2010

12
30
10

Difference / el

Difference / el

25

MAE
STAT

20

15

10

MAE
STAT

5
2
0
0,4

0,6

0,8

1,0

1,2

1,4

1,6

0,4

Xq / Xqn

Load angle / el

40

30

Xe(estimator) = 0,5 Xen

Measured value
Estimated value

-10
1

t/s

Load angle / el

70

1,4

1,6

60

4. CONCLUSIONS

50

This paper presents and compares methods for the load

angle determination for salient-pole synchronous generator.
First method presents measurement using an optical encoder
to detect rotor position. The load angle is measured against
the power system voltage. Second method presents the load
angle estimation method based on the measured values
(voltages and currents) of synchronous generator. An
accuracy of the presented load angle estimation method
depends on voltage and current measurement accuracy as
well as on parameters determining accuracy (quadratureaxis reactance Xq, equivalent resistance R and reactance Xe).
In static conditions the difference between estimated and
measured load angle is less than 4 el. and in dynamic
conditions measured and estimated results demonstrate good
matching for load angles less than 90 el.
It was also concluded that errors in generator parameter
determination have influence on the static accuracy of the
presented estimation method. The average dynamic absolute
error significantly depends on generator quadrature-axis
reactance change.
Excitation control system with additional signal and
estimated load angle can improve the stability of a
synchronous generator in capacitive operating mode. Future
research will be based on development of excitation control
algorithms based on the estimated load angle of a
synchronous generator.

40
30

Xe(estimator) = Xen
Measured value
Estimated value

20
10
0
-10
0

t/s
70
60
50

Load angle / el

1,2

The results of sensitivity analysis show that static accuracy

of the presented estimation method is practically
independent of errors in determining stator resistance. The
average absolute dynamic error is less than 1 % in
conditions of stator resistance change. So the determining
accuracy of stator resistance value does not have great
influence on the accuracy of the presented load angle
estimation method.
However, errors in determining the reactances Xq and Xe
have influence on the static accuracy of the presented
estimation method. The average dynamic absolute error
significantly depends on quadrature-axis reactance Xq
change.

50

1,0

Fig.20. MAE and STAT error of load angle estimation due to

different equivalent reactance Xe in estimator

60

10

0,8

Xe / Xen

Fig.18. MAE and STAT criterions due to different quadrature-axis

reactance Xq in estimator

20

0,6

40
30
20

Xe(estimator) = 1,5 Xen

Measured value
Estimated value

10
0
-10
-20
0

t/s

Fig.19. The measured and estimated load angle with different

equivalent reactance Xe in load angle estimator

95

MEASUREMENT SCIENCE REVIEW, Volume 10, No. 3, 2010

APPENDIX A

[10]

The synchronous generator's rated parameters are:

Voltage
Current
Power
Frequency
Speed
Power factor
Excitation voltage
Excitation current

400 V
120 A
83 kVA
50 Hz
600 min-1
0.8
100 V
11.8 A

[11]

[12]

[13]

REFERENCES
[1]

[2]

[3]

[4]

[5]

[6]

[7]

[8]

[9]

Kundur, P. (1993). Power System Stability and

Control. Power System Engineering Series. McGrawHill.
Anderson, P.M., Fouad, A.A. (1994). Power System
Control and Stability. IEEE Power System
Engineering Series. IEEE Press.
Girgis, G.K., Vu, H.D. (1995). Verification of limiter
performance in modern excitation control systems.
IEEE Trans. on Energy Conversion, 10 (3), 538-542.
Eberly,
T.W.,
Schaefer,
R.C.
(1995).
Minimum/maximum excitation limiter performance
goals for small generation. IEEE Trans. on Energy
Conversion, 10 (4), 714-721.
Ramos, A.J.P., Lins, L.R., Fittipaldi, E.H.D.,
Monteath, L. (1997). Performance of under excitation
limiter of synchronous machines for system critical
disturbances. IEEE Trans. on Power Systems, 12 (4),
1702-1707.
Working group of the excitation systems
subcommittee. (1995). Underexcitation limiter models
for power system stability studies. IEEE Trans. on
Energy Conversion, 10 (3), 524-531.
Sumina, D., Buli, N., Erceg, I. (2010). Threedimensional power system stabilizer. Electric Power
Systems Research, 80 (7), 886-892.
Chapman, J.W., Ilic, M.D. King, C.A. (1993).
Stabilizing a multimachine power system via
decentralized feedback linearizing excitation control.
IEEE Trans. on Power Systems, 8 (3), 830-839.
Lu, Q., Sun, Y., Xu, Z., Mochizuki, T. (1996).
Decentralized nonlinear optimal excitation control.
IEEE Trans. on Power Systems, 11 (4), 1957-1962.

[14]

[15]

[16]

[17]

[18]

[19]

[20]

Redfern, M.A., Checksfield, M.J. (1995). A new pole

slipping protection algorithm for dispersed storage and
generation using the equal area criterion. IEEE Trans.
on Power Delivery, 10, 194-202.
Barrera-Cardiel, E., Pastor-Gomez, N. (1999).
Microcontroller-based power-angle instrument for a
power systems laboratory. In Proceedings of the IEEE
Power Engineering Society: Summer Meeting. IEEE,
1008-1012.
Chen, Y., Zhang, C., Hu, Z., Wang, X. (2000). A new
approach to real time measurement of power angles of
generators at different locations for stability control. In
Proceedings of the IEEE Power Engineering Society:
Winter Meeting. Vol. 2, IEEE, 1237-1242.
Despalatovi, M., Jadri, M., Terzi, B. (2008). Realtime power angle determination of salient-pole
synchronous machine based on air gap measurements.
Electric Power System Research, 78 (11), 1873-1880.
Cari, E.P.T., Alberto, L.F.C., Bretas, N.G. (2000). A
novel methodology for power angle estimation of
synchronous generator based on trajectory sensitivity
analysis. In Proceedings of the IEEE Power
Engineering Society: Winter Meeting. Vol. 2, IEEE,
1237-1242.
Del Angel, A., Geurts, P., Ernst, D., Glavic, M.,
Wehenkel, L. (2007). Estimation of rotor angles of
synchronous machines using artificial neural networks
and local PMU-based quantities. Neurocomputing, 70
(16-18), 2668-2678.
Ghahremani, E., Karrari, M., Malik, O.P. (2008).
Synchronous generator third-order model parameter
estimation using online experimental data. IET
Generation, Transmission & Distribution, 2 (5),
708-719.
Idzotic, T., Erceg, G., Sumina, D. (2004). Load angle
estimation of a synchronous generator. In IEEE
Mediterranean Electrotechnical Conference. Vol. 3,
IEEE, 893-896.
Vicol, L., Tu Xuan, M., Wetter, R., Simond, J.-J.,
Viorel, I.A. (2006). On the identification of the
synchronous machine parameters using standstill DC
decay test. In ICEM 2006: International Conference
on Electrical Machines.
Malari, K., Malari, R., Hegedu, H. (2008). Gauss
Newton curve fitting.
http://sourceforge.net/projects/gauss-newton/
National Instruments: CompactRIO.
http://www.ni.com/compactrio/about.htm
Received January 26, 2010.
Accepted June 7, 2010.

96