Energy Conversion and Management 42 (2001) 519528

www.elsevier.com/locate/enconman

Loadless full load temperature rise test for three phase

induction motors
Hamid M.B. Metwally *
Department of Electrical Power Engineering, Faculty of Engineering, Zagazig University, Zagazig, Egypt
Received 9 December 1999; accepted 24 August 2000

Abstract
This paper introduces a new method for measuring the full load temperature rise of three phase induction
motors. The method is very simple, cheap and reliable. It does not require a mechanical load to be coupled
to the motor shaft, as in the conventional method. Also, it does not need any auxiliary machines or special
power electronics to generate the two frequency supplies, as needed by the dual frequency method. The
method is simply to run the motor at no load without any thing coupled to the shaft but with the supply
voltage slightly higher than the rated voltage of the motor. This causes the motor to draw from the supply a
no load power higher than that drawn at rated voltage. The value of this power can be controlled by
changing the applied voltage to simulate the full load conditions on the motor. The results obtained in this
study showed that an input voltage of about 120% of the rated voltage of the motor is a suitable value to
cause the full load losses to occur in the motor body, and hence, the full load temperature rise is ob-
tained. 2000 Elsevier Science Ltd. All rights reserved.

Keywords: Induction motor performance; Induction motor testing; Induction motor temperature rise

1. Introduction

To test three phase induction motors for temperature rise, the full load losses must be dissi-
pated in the motor body while the motor is running at rated speed. This means that a mechanical
load of at least the same power rating of the motor must be coupled to the motor shaft. Then, the
system must be left to run at full load for about three hours till the steady state temperature rise is
reached. The initial cost of building this test rig is quite high. This also necessitates a dissipation of
energy equal to the power rating of the motor multiplied by the time of the test. So, the running

*
Tel.: +20-55-622-411; fax: +20-55-324-987.

0196-8904/01/$ - see front matter 2000 Elsevier Science Ltd. All rights reserved.
PII: S 0 1 9 6 - 8 9 0 4 ( 0 0 ) 0 0 0 7 9 - 0
520 H.M.B. Metwally / Energy Conversion and Management 42 (2001) 519528

cost of these tests is also high, especially for large size motors. In order to avoid the above dif-
culties, it is necessary to nd an equivalent load test in which the full load losses are dissipated in
the motor body without actually loading the motor. Several authors have recognized this fact and
a number of loading schemes have been developed. The dual frequency method was rst proposed
by Ytterberg [1] as early as 1921. In this method, two voltage supplies of dierent frequencies are,
connected in series with the induction motor. Recently, several authors [28] have introduced
dierent forms of this dual frequency method. All these forms of the method require some facility
to generate the two distinct frequencies. Some authors have used rotating electrical machines,
while others have used electronic converters. Ghoneem [9] suggested a method of synthetic
loading in which three diodes and a variable resistance are used. The diodes and the variable
resistance are connected across the motor such that a controllable dc current circulates through
the motor windings. This increases the currents drawn from the supply by the stator windings and,
hence, increases the motor losses. The variable resistance is used to control the losses to be equal
to the full load losses. Fong [10] suggested the phantom loading method in which two equally
rated machines are required. Further improvement of the method is given in Refs. [11,12]. The
variable inertia test is proposed by Garvey et al. [13]. The test is purely mechanical in nature. It
does not require any electrical connection of the test machine to any other machine, but it requires
mechanical coupling of the test machine to an inertia load. The load is very simple and compact
compared with the electrical load.
In this work, a new method for equivalent loading of three phase induction motors is intro-
duced. It is very simple. It does not require any mechanical load on the motor shaft. Also, there is
no need for any power electronics between the motor and the supply.

2. Basic idea of the method

The per phase equivalent circuit of the induction motor is shown in Fig. 1. Under any operating
condition, the power drawn from the supply is:
Pin 3V1 I1 cos /1 1
This power supplies:
1. The stator copper losses 3I12 R1 ,

Fig. 1. Equivalent circuit of the motor.

 H.M.B. Metwally / Energy Conversion and Management 42 (2001) 519528 521

Fig. 2. Measured and calculated no load losses.

2. The core losses 3E12 =R0 ,

3. The rotor copper losses 3I22 R2 ,
4. The load power 3I22 R2 1 s=s.
If the motor is running under no load conditions, item 4 is not present. The total power drawn
from the supply is then consumed as losses. The no load losses (P0 ) are then given by:
P0 3I12 R1 3E12 =R0 3I22 R2 2
and the voltage E1 is given by:
E1 V1 I1 R1 jX1 3
From Eqs. (2) and (3), it can be seen that the no load losses of the motor depend on the applied
voltage (V1 ). As the applied voltage is increased, each of E1 , I1 and I2 is increased, and hence, the
losses are increased. This increase in current and power losses takes place even if the parameters of
the motor equivalent circuit per phase are considered constants and independent of the applied
voltage. Taking a 2 hp motor as an example, the parameters of the equivalent circuit are calcu-
lated from the conventional no load and blocked rotor tests. Fig. 2 shows how the no load losses,
both measured and calculated, change with the applied voltage. The large discrepancy depicted in
the gure between measured and calculated no load powers is due to the fact that the calculations
are made based on the assumption that the motor parameters are constants. In fact, these pa-
rameters change widely with the applied voltage, especially the no load loss resistance (R0 ) and the
magnetizing reactance (X0 ). A computer program, based on the trial and error method, is de-
veloped to calculate R0 and X0 at dierent values of the applied voltage. Assuming that the other
four parameters are constants, the program searches for the values of R0 and X0 which achieve
close agreement between the calculated and measured characteristics (power, current and power
522 H.M.B. Metwally / Energy Conversion and Management 42 (2001) 519528

Fig. 3. Variation of R0 and X0 of the 2 hp motor.

Fig. 4. Measured and calculated no load power.

factor). It has been found that R0 and X0 vary with the applied voltage as shown in Fig. 3. Taking
this variation into account, complete agreement between the measured and calculated current and
power is obtained, as in Figs. 4 and 5. As the voltage increases above the rated voltage, each of R0
 H.M.B. Metwally / Energy Conversion and Management 42 (2001) 519528 523

Fig. 5. Measured and calculated no load current.

and X0 drops quickly, and hence, the no load current and power are increased very fast. For the
above motor, the full load losses are about 500 watts. The gure shows that this amount of power
can be drawn from the supply while the motor is running unloaded simply by increasing the
applied voltage to about 120% of the rated voltage. This voltage ratio is conrmed experimentally
by measuring the no load power and the temperature rise at dierent voltages for 53 motors, as
will be explained in the following sections. This fact is exploited to simulate the full load con-
ditions on the motor by operating it at no load with higher voltage.

3. Experimental work

Tests have been performed on three motors to establish the relationship between the applied
voltage at no load and the temperature rise of the motor. These motors are rated at 0.75, 2 and 2
hp, respectively. The two 2 hp motors are from dierent manufacturers. Each motor is left to run
at no load under a given supply voltage for three hours till the steady state temperature rise is
reached. The dc resistance of the phase winding is measured each 15 min during the test. The
temperature rise is then calculated using the following formula:
T2 RT2 RT1  235 T1 =RT1 ; 4
where T2 is the temperature rise, T1 is the ambient temperature, RT1 is the phase resistance at
temperature T1 , RT2 is the phase resistance at temperature T2 . The test is then repeated many
times at dierent values of supply voltage. A family of curves, for each motor, representing the
relationship between temperature rise and time at dierent values of supply voltage is obtained.
Fig. 6 shows one of these families of curves for the above 2 hp motor. The applied voltage is
524 H.M.B. Metwally / Energy Conversion and Management 42 (2001) 519528

Fig. 6. Temperature rise characteristics at dierent percent voltages.

Fig. 7. Temperature rise/voltage characteristics.

changed from 100% to 125% of the rated voltage in steps of 5%. This covers the expected range of
temperature rise of the motors. The results showed that the temperature rise increases as the
applied voltage is increased. This can be seen in Fig. 7. A temperature rise of 80C is reached by
applying a voltage of 122%, 120% and 115% of the rated voltage of the three motors. This means
that if the 2 hp motor, for example, is left to run under no load with an applied voltage of 120% of
its rated voltage, the full load temperature rise of 80C is reached, which is exactly equivalent to a
complete load test on the motor. This 80C temperature rise is the allowable upper limit for class
B insulation under full load.
 H.M.B. Metwally / Energy Conversion and Management 42 (2001) 519528 525

4. Extracted data of motors designed by Siemens

Shoubra Co. for engineering industries under a license from Siemens Co. is producing three
phase squirrel cage induction motors. These motors are totally enclosed, fan cooled and are of
dierent sizes and speeds. The technical specications of these motors are available. Also, the no
load characteristics, namely the variation of no load current and power with applied voltage, are
available for applied voltages from zero to 125% of the rated voltage. These experimental data are
exploited to calculate the voltage required to produce the full load losses for each motor. From
the rated output and eciency, the full load losses are calculated. Then, the required voltage and
the corresponding current are obtained from the no load characteristics. The above procedure is
repeated for 50 motors of dierent sizes and speeds. The results obtained are given in Tables 14

Table 1
Full load losses with required percent voltage and percent current for two pole motors
Output HP Eciency (%) Losses (W) Voltage (%) Current (%)
0.25 62 110 119.5 121.5
0.5 69 165 120.2 121.1
0.75 71 225 119.5 123.3
1.0 74 264 120.5 117.5
1.5 77 329 122.1 111.4
2.0 78 423 119.3 108.5
3.0 82 483 116.8 105.2
4.0 83 615 120.0 113.3
5.5 85 706 122.1 88.4
7.5 85 971 115.8 91.4
10.0 87 1121 117.4 83.8
15.0 88 1500 123.7 80.0
20.0 90 1667 122.2 78.5
25.0 91 1830 127.4 75.5

Table 2
Full load losses with required percent voltage and percent current for four pole motors
Output HP Eciency (%) Losses (W) Voltage (%) Current (%)
0.16 58 87 119.1 121.5
0.25 60 120 120.2 122.2
0.5 67 182 118.4 123.6
0.75 72 214 118.6 118.7
1.0 74 264 117.9 114.9
1.5 75 367 113.2 116.1
2.0 75 500 118.8 114.9
3.0 78 620 117.5 105.6
4.0 79 797 116.8 100.8
5.5 83 819 115.8 90.9
7.5 84 1048 117.6 91.5
10.0 86 1221 116.8 89.7
15.0 88 1500 122.1 77.7
20.0 89 1854 123.7 80.0
526 H.M.B. Metwally / Energy Conversion and Management 42 (2001) 519528

Table 3
Full load losses with required percent voltage and percnet current for six pole motors
Output HP Eciency (%) Losses (W) Voltage (%) Current (%)
0.33 62 153 116.8 120.3
0.5 66 191 111.8 120.0
0.75 70 236 115.6 119.9
1.0 71 306 114.5 119.2
1.5 72 428 117.8 118.7
2.0 76 474 116.8 118.8
3.0 78 620 118.2 112.6
4.0 80 750 113.7 98.7
5.5 83 819 116.8 100.0
7.5 84 1048 117.1 97.0
10.0 84 1429 117.1 96.1
15.0 88 1500 118.4 93.8

Table 4
Full load losses with required percent voltage and percent current for eight pole motors
Output HP Eciency (%) Losses (W) Voltage (%) Current (%)
0.5 63 220 113.2 123.3
0.75 68 259 114.5 112.7
1.0 68 350 115.8 117.2
1.5 72 430 115.8 122.3
2.0 73 555 115.8 110.8
3.0 75 733 119.0 104.4
4.0 80 750 117.4 96.7
5.5 80 1000 113.7 99.0
7.5 83 1127 113.6 98.1
10.0 84 1430 121.6 97.3

Fig. 8. Variation of voltage required to produce full load losses at no load (as percent of rated voltage).
 H.M.B. Metwally / Energy Conversion and Management 42 (2001) 519528 527

Fig. 9. Variation of current required to produce full load losses at no load (as percent of rated current).

and are shown in Figs. 8 and 9. Fig. 8 shows that the voltage required to produce full load losses
at no load is about 120% of the rated voltage. This voltage circulates a current varying from 120%
of the rated value for motors of small sizes to a current of 80% of the rated value for large size
motors. The current variations are shown in Fig. 9.

5. Conclusions

This paper presents a new method for testing the full load temperature rise of three phase
induction motors. The method is simply to connect the motor under test while it is unloaded to a
supply of voltage higher than the rated voltage of the motor by about 20%. It has been found that
a voltage of 120% of the rated voltage of the motor under test is a suitable value for motors of
dierent sizes and speeds. This value of voltage circulates a current which exceeds the full load
current for small size motors and stays less than the full load current for large size motors. The
method is very simple to implement. It does not require any mechanical load to be coupled to the
motor shaft. The motor draws only the full load losses from the electrical supply. Hence, there is
no need for some arrangement to either dump an electrical power equal to the rated power of the
motor under test or to return it to the supply. The method is cost eective and is suitable for use
by manufacturers.

References

[1] Ytterberg A. A new method for fully loading induction machines not having a driving motor or a load machine, the
so-called Jolting test, Teknesk Tidsykvijt, 1921.
[2] Kron AW. Testing induction motors by means of a two frequency supply. ETZ-A (Germany) 1973;94:7782.
528 H.M.B. Metwally / Energy Conversion and Management 42 (2001) 519528

[3] Meyer A, Lorenzen HW. Two-frequency heat run: a method of examination for three phase induction motors.
IEEE Trans 1979;PAS-98(6):233847.
[4] Grantham C, Rahman F. A novel machineless dynamometer for load testing three phase induction motors, IEE
Sixth International Conference on Electrical Machines and Drives, 810 September, 1993, p. 2027.
[5] Colak I, Garvey SD, Wright MT. Mixed-frequency testing of induction machines using inverters. Fifth European
Conference on Power Electronics and Applications, vol. 5, 1993. p. 31722.
[6] Garvey SD, Colak I, Wright MT. Aspects of mixed frequency testing of induction machines. ICEM94
International Conference on Electrical Machines. Paris, France, 1994. p. 6238.
[7] Schwenk HR. Equivalent loading of induction machines for temperature tests. IEEE Trans PAS 1977;96(4):
112631.
[8] Jordan HE, Cook JH, Smith RL. Synthetic load testing of induction machines. IEEE Trans PAS 1977;96(4):
11014.
[9] Ghoneem G. Temperature tests for three phase induction motors by a new synthetic loading system. Port Said Sci
Engng Bull 1995;7:26775.
[10] Fong W. New temperature test for polyphase induction motors by phantom loading. Proc IEEE 1972;1A(7):8837.
[11] Fong W. Temperature testing of skin eect rotor induction motors by synthetic loading. Proc IEE 1976;123(6):
5468.
[12] Ho SL. Further development of phantom loading in induction motors. International Conference on Electrical
Machines, vol. 2, 1992. p. 298302.
[13] Garvey SD, Colak I, Wright MT. Variable inertia test for full load temperature rise testing of induction machines.
Proc IEE Electric Power Appl 1995;142(3).