Induction Motors Thermal Monitoring by Means of Rotor Resistance Identification

Rachid Beguenane and Mohamed El Hachemi Benbouzid

Laboratoire des Systimes Automatiques
UniversitC de Picardie - Jules Verne
7, Rue du Moulin Neuf - 80000 Amiens, FRANCE

Abstract-Rotor and stator temperature is of concern in both

short-term machine protection and in longer term condition
monitoring for large induction machines. Especially, when oper-
ating with overload cycles, it is necessary to monitor rotor bars
and stator winding temperatures to make sure that the tempera-
ture remains below prescribed limits. The purpose of this paper
is to present a thermal monitoring technique, for induction mo- where
tors, based on the rotor resistance identification. Experimental d-axis (q-axis) stator voltage
results are shown for a 4-kW four-pole squirrel cage induction d-axis (q-axis) stator current
motor. d-axis (q-axis) rotor current
per phase stator (rotor) resistance
per phase stator (rotor) inductance
per phase mutual inductance
I. INTRODUCTION supply fiequency
slip frequency, wsi= ws- w,
T he use of comprehensive monitoring schemes for con-
tinuous assessment of the electrical and mechanical
condition of rotating electrical machines is becoming increas-
rotor fiequency
time derivative operator.
ingly important. A successful condition monitoring scheme on
a machine would be capable of providing adequate warning of Equation (1) can be transformed as follows: we make the
imminent failure of its critical components, both electrical and d-axis coincide and synchronously rotate with the stator cur-
mechanical, and would enable preventive maintenance and rent vector I,. Thus it can simplified as
repair work to be scheduled. Parameters presently monitored
on induction machines include terminal voltage, current, rotor
speed, etc. The parameters so monitored have been found
capable of predicting and identifying stator ground and inter-
turn faults, phase failure, stall, overload, bearing failure, rotor
defects, shaft eccentricity but with varying degrees of success Since the time constant for the variation of the rotor resis-
as summarized in [ 11-[2]. tance is much larger than the time constant of the induction
Rotor temperature is given considerable importance in the motor, the steady-state model can be adopted for the rotor
design and protection of machines but has been left out of the resistance identification. Therefore, the following steady-state
conventional as well as the recently developed monitoring equation could be derived fiom (2).
schemes. The function of a temperature monitor is taken over
by other secondary monitoring systems. One obstacle appears
to be the difficulties in accessing information from the rotor
while the machine is in normal operation. The availability of a
continuous temperature profile across the rotor can contribute
much to a comprehensive condition monitor.
This paper describes an electrical model of the induction Considering the last three rows of (3) and solving the result-
motor and the process by which the rotor resistance in that ing equations system, we get
model is estimated fi-om measurements of stator voltage, sta-
tor current, stator excitation frequency, and rotor speed. This
is the basis for thermal monitoring of the induction motor.

11. ROTORRESISTANCE IDENTIFICATION

The method to be proposed is based on the proper selec-
tion of coordinate axes, namely the d-axis of the rotating
fiame is set to be coincident with the stator current.vector [3].
It is well known that an induction motor can be repre-
sented by the following differential equations in d-q axes V", 1,
fixed in the stator. Fig. 1. Phasor diagram of stator voltage and current.

0-7803-39U6-0/97/$10.00 0 1997 IEEE. TD2-4.l

 I r 1 IV. SIMULATION
RESULTS
The proposed identification technique has been simulated
on a 4-kW four-pole squirrel cage induction motor which
ratings are summarized in the appendix. The motor model is
represented by (1) as well as the following mechanical equa-
tion.
To avoid time-consumed computations related to d-q Park
transformation, electrical quantities are expressed in the a-p D 3 M
=--.com+-.p.-.(i,,.ira-i,.i,p)-- T/ (10)
axes fixed in the stator (Fig. 1). In this case, we get J 2 J J
where D (0.001 N m s ) is the viscosity resistance, J (0.024
kg m2)the total inertia, p the pole pairs number and T, the load
torque.
The validity of the proposed identification technique is
P = vSa.is, + vsp . i P well verified by simulations as illustrated by Fig. 2 and 3.
where (6) These results are obtained for a light load condition at 1482
Q= vsa.isp -vsp.isa rpm, where the identified resistance is R, = 1.45 R comparing
Using the simplified Park matrix to combine (5) and (6) into to the real value R, = 1.44 R and for a load torque at 1460
(4), we obtain a more suitable expression of the rotor resis- rpm, where the identified resistance is R, = 1.44 R. Moreover,
tance. the identification method has been tested for a specific case
where the real rotor resistance is given an exponential profile.
I 1 R,(t)= R,, +(l-e-'.2')
where Rd = 1.44 R. Figure 4 shows then a good identification
of the rotor resistance profile.

V. EXPERIMENTAL
RESULTS
We can now summarize the identification procedure of the
rotor resistance as: A. Experimental Benchmark
Acquisition of the stator voltage, current, and motor The proposed thermal monitoring technique has been
speed: vis, vL, v3s;i],, &, ij, and wm= orlp. tested on 4-kW squirrel cage induction motor with dc genera-
Computation of measwed voltages and currents park tor coupled on its shaft. Figure 5 shows a view of the experi-
components: v,,, vsb and i,, isp. mental benchmark during test.
Computation of the reactive power Q by (6).
Identification of R, by using formula (7).

111. INDUCTION MOTORTHERMAL MONITORING i . :

It is easy to monitor rotor temperature from its resistance 2

identification and then its temperature dependence, which is
0' 05 1 15 2 25 3
given by Tm(r)
Fig 2 Rotor resistance identification for a load corresponding to 1482 rpm
R= R, [l+a.AT] (8)
where Ro = resistance at reference temperature To= 25C
CL = resistance temperature coefficient
AT = temperature increase.

The proposed identification method is unfortunately un-

able to identify the stator resistance. To overcome this prob-
lem, two approaches could be used. The first one correspond 0' 05 1 15
Tma(*l
2 25 3

to EDF (French Electricity Board) experience which states Fig 3 Rotorresistance sponding to 1460 rpm
that the rotor temperature is 10C greater than the stator one
[4]. The second approach is based on Kubota work in [ 5 ] . In
this case, the stator resistance is deduced from
RA"
R,=k.R,, k=- (9)
Rm
where R, and R, are the nominal values of stator and rotor 0 05 1 15
Tmclr)
2 25 3

resistance as given by the machine manufacturer. Fig 4 Exponential profile rotor resistance identification

TD2-4.2
 Fig. 5. View of the experimental benchmark.

B. Experimental Results
Figure 6 shows experimental results of the rotor resistance
identification for a load corresponding to 1482 rpm. Then,
30
0
- 20 40
keh"
60 I

Fig. 9. Estimated rotor temperature for a load corresponding to 1482 rpm.

Fig. 7, as for Fig. 6, illustrates specific identification proc-

esses of the rotor resistance for 1460 rpm, where each curve
corresponds to a resistance identification done 5 mn after the
previous one: fiom the induction motor starting until one hour
operation. These figures illustrate clearly that the identifica-
tion process is quite satisfactory. More, the convergence is
experimentally well verified since the identification process I
O 10 20 30 40 50 60
converges quickly and independently of the initial values as it limelmn)

is clearly shown by Fig. 8.
The results of rotor temperature estimation, using (S), are
illustrated by Fig. 9 and 10. Estimated rotor temperature is, in
this case, in good agreement with induction machine classical
theory. In fact, one should note that the magnitude of the
temperature differential increases with the load level. More,
for the thermal monitoring of the induction motor (rotor and
stator), we have used (9) to estimate the stator temperature:
this is in a good concordance with [4].
Fig. 10. Estimated rotor temperature for a load corresponding to 1460 rpm.
VI. CONCLUSION
The viability of induction motor thermal monitoring using
only rotor resistance identification has been investigated.
Apart fiom its benefits as a rotor condition monitor for induc-
tion machines on a long-term basis, such a technique can be
an effective tool for machine designers and protection system
designers, who are concerned about induction machine tem-
perature monitoring.

APPENDIX
PARAMETERS OF THE MACHINE UNDER TEST
~~~

Rated values Power 4 kW

Frequency 50 Hz
Voltage ( A N ) 220/380 V
Current (NY) 1518.6 A
2 , . . : . , Speed 1440 rpm
Pole pair (p) 2
Constants JL 1150
R, 1.440 R
1 0.013 H
M 0.143 n
a (Cu or Al) 0.004 1/"C

0 0.2 0.4 0.6 0.8 1

nM(d
REFERENCES
Fig. 7. Rotor resistance identification for a speed of 1460 rpm.
[ l ] A.H. Bonnen et al., "Cause and analysis of stator and rotor failures in
three-phase squirrel-cage induction motors," IEEE Trans. Indus. Appl.,
Vol. 28, N04,pp. 921-937, July-August 1992.
[2] P.J. Tavner et al., "Monitoring generators and large motors," in Proc.
- f -5-
IEE, Vol. 133, Pt. B, pp. 169-180, May 1986.
0 0.005 0.01 0 0.005 0.01 [3] C.C. Chan et al., "An effective method for rotor resistance identification
for high-performance induction motor vector control," IEEE Trans.
Yvl, 37,N"6, pp. 477482,Dsssmbar 1990,
2 n d .leern,
~
[4] M. Dessoude, Electricit6 de France (EDF), R&D Division, Electrical
Equipment Department (France), personal communication, June 1994.
[5] H. Kubota et aL, "DSP-based speed adaptive flux observer of induction
motor," IEEE Trans. Indus. Appl., Vol. 29, N02, pp. 344-348, March-
Fig. 8. Convergence of the identification process April 1993.

TD24.3