IEEE TRANSACTIONS ON ENERGY CONVERSION, VOL. 22, NO.

2, JUNE 2007 383

Analysis of Repairs on Three-Phase Squirrel-Cage

Induction Motors Performance
Edson da Costa Bortoni, Senior Member, IEEE, Jamil Haddad, Afonso Henriques Moreira Santos,
Erick Menezes de Azevedo, and Roberto Akira Yamachita

Abstract—This paper presentsthe results of research conducted decision making process of whether a motor should be repaired
to assess the effects of preventive maintenance and repairs on three- or not.
phase squirrel-cage induction motor performance regarding en- Among the factors that shall be considered when conducting
ergy efficiency. The results brought interesting insights on how
simple tasks can improve the energy efficiency, both in industry the analysis, efficiency of the repaired motor is the most impor-
and in repair shop procedures. tant variable and, unfortunately, also the most unpredictable one.
Although a small efficiency reduction in a rewound motor may
Index Terms—Maintenance, motor efficiency, motor repair.
be expected, this is acceptable, since it should be compensated
for by a lower repair cost when compared to the investment in
NOMENCLATURE purchasing a new motor.
I0 No load current (A). A set of “rules of thumb” is often used, for instance, repair
I100 Full load current (A). cost should not exceed a percentage of the new motor price, or,
n100 Full load speed (rpm). to discard motors with a lower rated power and/or older than a
P Rated power (hp). given age. Recent papers give systematic approaches to reduce
P0 No load loss (kW). the empirical aspects involved in this decision making [7], [8].
PC Full load core loss (kW). The former aspect is that there is a common sense about the
PJ1 Full load stator loss (kW). importance and effectiveness of periodic cleaning and lubrica-
PJ2 Full load rotor loss (kW). tion to maintain motor systems operating in the best conditions.
PS Full load stray loss (kW). These procedures have an unknown energetic component that
PWF Full load windage and friction losses (kW). shall be quantified.
R75 Stator winding resistance referred to 75 ◦ C (Ω). A great effort and even a lot of money are spent to seek even
∆θ Full load temperature rise (◦ C). a half percent efficiency gain in energy systems. The energetic
η50 1/2 load efficiency (%). benefits of preventive maintenance when applying simple and
η75 3/4 load efficiency (%). low cost tasks cannot be neglected and, thus, is treated and
η100 Full load efficiency (%). quantified as the second point of the paper.
ρxy Correlation coefficient.
σ Standard deviation.
II. ADOPTED APPROACH
I. INTRODUCTION The research was conducted in two phases. The first
HREE-PHASE induction motors are the most significant phase was focused only on the corrective maintenance by study-
T industrial loads almost everywhere. This statement is also
noticed in Brazil as these machines are responsible for about
ing the repair shops overall quality and the motor performance
before and after rewinding due to an imposed burnout. Both
25% of the gross energy consumption. The great number of preventive maintenance and repair are addressed in the second
installed units magnifies the number of failures and also the phase, where the effect of simple tasks such as cleaning and
cases of misapplications. lubrication on the motor’s performance is studied. Tests before
Maintenance is related to motor energy efficiency in those two and after the burnout, were also applied.
respects. The first, a subject exhaustively treated in international
technical literature, is related to the motor efficiency behavior
after repairs [1]–[6]. The conclusions of these studies support the A. Analysis of the Repair Shops
In order to know the motors’ performance after the repair,
Manuscript received April 20, 2005; revised July 10, 2005. This work was the first step was to check the ability of the maintenance shops
supported in part by PROCEL, the Brazilian National Electrical Energy Con- to supply their services. Twelve repair shops were studied. The
servation Program. Paper no. TEC-00152-2005. checked points were mainly aspects that could influence the
E. C. Bortoni, J. Haddad, and A. H. M. Santos are with Itajubá Fed-
eral Univeristy, Itajubá MG 37500-903, Brazil (e-mail: [email protected]; quality of the service, which are shown in Table I, classifying
[email protected]; [email protected]). each qualitative item as good (•), regular (◦·), or poor (◦).
E. M. Azevedo and R. A. Yamachita are with the Energy Studies Group, In general, the services delivered by these shops are of good
Itajubá MG 37500-903, Brazil (e-mail: [email protected]; rakira@
unifei.edu.br). quality. The main negative point occurs when the shops do not
Digital Object Identifier 10.1109/TEC.2006.874213 have the correct tools to perform their job.

0885-8969/$25.00 © 2006 IEEE

384 IEEE TRANSACTIONS ON ENERGY CONVERSION, VOL. 22, NO. 2, JUNE 2007

TABLE I TABLE II
SHOP CHARACTERISTICS REGARDING THE SERVICE QUALITY SHOP CHARACTERISTICS REGARDING THE EMPLOYED TECHNOLOGY

One example is the use of a torch for stator winding removal

instead of using a burnout furnace with temperature control.
This procedure, used in some small or rural shops in Brazil, can
cause severe consequences in the stator core integrity, including
changes in magnetic characteristics and interlaminar insulation, TABLE III
resulting in greater iron loss. SHOP CHARACTERISTICS REGARDING THE EMPLOYED TOOLS
Once motor maintenance becomes a usual practice—there
are shops in the region that repair up to 400 motors a month—it
was realized that the trust-based relationship between customers
and repair shops is the most important factor when choosing a
shop. Table II shows the used technology repair survey, while
Table III shows the survey of tools employed in the shop.

B. Analysis of the Motors

As a first step, an experiment was conducted with a set of eight
low-voltage motors with ratings from 3 to 15 (hp), to determine
the influence of rewinding on motor efficiency. This range was
chosen because it represents 65.2% of Brazilian installed mo-
tors. While motors within this range are virtually disposable
items in large industries, this is not true in the small ones, which
represent 68.7% of the cases.
In order to reflect actual conditions, these motors were care-
fully chosen and taken directly from different industries and
field applications to the laboratory. Newer motors were offered
to the partner companies on exchange for the old ones.
The motors were tested before and after an imposed failure.
Insulation resistance and stator winding resistances were also
measured. Efficiency at 50%, 75%, and 100% of rated load were
initially obtained applying the direct method [10]. To strengthen TABLE IV
QUANTITIES BEFORE REPAIR
the analyses and to obtain more information, free and locked
rotor, temperature rise and loss segregation tests were applied
[11].
The method of induced overload burnout damage caused by
rotor locking during starting was chosen because overheating
and insulation breakdown are the major initiators of stator wind-
ing failures in industry [12], making this analysis more realistic.
Tables IV and V present motor performance before the im-
posed failure. Table VI and Table VII present the results after
BORTONI et al.: ANALYSIS OF REPAIRS ON INDUCTION MOTORS PERFORMANCE 385

TABLE V
QUANTITIES BEFORE REPAIR

Fig. 2. Percentage variation of PWF .

TABLE VI
QUANTITIES AFTER REPAIR

Fig. 3. Percentage variation of PC .

TABLE VII
QUANTITIES AFTER REPAIR

Fig. 4. Percentage variation of PS .

Fig. 5. Percentage variation of PJ1 .

increase of another (y) while a negative coefficient shows that

an increase in a variable (x) is associated to a decrease of
Fig. 1. Percentage variation of R75 . another (y). Absolute correlation coefficient values less than
0.4, between 0.4 and 0.6, and greater than 0.8 indicates weak,
the repairs. Rated current, rated speed, and no-load current were medium, and strong correlation between variables (x, y), re-
registered to estimate rotor losses behavior, iron losses, and rated spectively. The following expression can be used to determine
load losses. the correlation coefficient [13].
The percentage variations of these values are now presented

1
(x − x̄)(y − ȳ)
graphically in Figs. 1––12, where the effects of maintenance ρxy = n−1 . (1)
σx σy
upon motor performance can be observed.
Fig. 1 shows the stator resistance variation. Figs. 2––8 show Table VIII shows a matrix of correlation coefficients between
important full load characteristics of motor losses variation, the quantities under analysis.
temperature rise, and speed. When analyzing the results, some expected high correlations
Figs. 9 and 10 show the behavior of full load and no-load were detected, for example, a positive relationship between sta-
currents, while Figs. 11 and 12 present the variation of the total tor losses with stator resistance (0.991) and with stator current
no-load loss and full load efficiency. (0.715), or the negative relationship between rotor speed and
Correlation coefficient is a normalized measurement of the rotor losses (−0.971).
linear relationship strength between variables (x, y) and ranges Nevertheless, some intermediate and less intuitive correla-
between −1 and +1. A positive correlation coefficient means tions can also be found, as the dependence of no-load current
that a positive growth in a variable (x) is associated with the with the core losses (0.701). This relationship may be suitable
386 IEEE TRANSACTIONS ON ENERGY CONVERSION, VOL. 22, NO. 2, JUNE 2007

Fig. 6. Percentage variation of PJ2 . Fig. 10. Percentage variation of I0 .

Fig. 7. Absolute variation of ∆θ. Fig. 11. Percentage variation of P0 .

Fig. 8. Absolute variation of n100 . Fig. 12. Absolute variation of η100 .

Another often and improper procedure is the use of a torch for the
removal of damaged coils, which may change core proprieties
and deteriorate interlaminar insulation.
Considering the general classification of the surveyed shops,
a clear relationship between the characteristics of losses-
efficiency, the quality of repair procedures, the available tools,
Fig. 9. Percentage variation of I100 . and the overall quality of work, was observed. Table XI shows
motors and related repair shop.
The effects of these observations can be seen when analyz-
to estimate the core loss behavior from simple current mea-
ing the windage and friction losses, where the greater increase
surements, avoiding the need of application of specific tests.
(35.6%) comes from one of the worst shops, while the best-
The same analysis can be done considering and quantifying the
ranked shops provided a reduction of 23%–47% in these losses.
qualitative aspects of repair shops showed from Tables I. to III.
The same occurs with the Hysteresis and Foucault losses, where
The influence of simple actions such as cleaning or lubricating
the greater reduction (−10.5)% is achieved in shops of better
is verified. Another set of five used motors was selected with
ranking.
ratings varying from 3 to 10 (HP). Tables IX and X show the
Due to the importance and the need for reliable information,
results.
reference documents and recommended practices for owners
The motors taken from the field were tested (first table line
and users of establishments that offer motor repair services have
for each motor) and passed through a cleaning and lubrication
been published in order to ensure higher quality, better service
process, bearings replaced and tested again (second line). After
specifications, cost effectiveness, and timely repairs [14]–[17].
the imposed failure, corrective maintenance was applied and the
In general, an average full-load efficiency increase of 0.34%,
motors were tested one more time (third line).
varying from −0.9 to +2.0 percent points, was noticed. This
result differs slightly from that obtained by other researchers,
III. ANALYSIS OF RESULTS
which found an average efficiency reduction. The reason for
Regarding the motor repair shops, it may be accepted that the this discrepancy is that the motors employed in these tests were
majority of the contacted shops followed adequate procedures already in use and taken from the field, while those used in the
for induction motors maintenance. published literature were new ones.
In some cases it was noticed that the use of inadequate tools An average reduction of stator winding resistance of about
and insulation materials resulted in a motor life-time reduction. −1.15%, accompanied by a stator copper loss reduction of
BORTONI et al.: ANALYSIS OF REPAIRS ON INDUCTION MOTORS PERFORMANCE 387

TABLE VIII
MATRIX REPRESENTATION OF CORRELATION COEFFICIENT

TABLE IX TABLE X
QUANTITIES BEFORE AND AFTER PREDICTIVE MAINTENANCE AND AFTER QUANTITIES BEFORE AND AFTER PREDICTIVE MAINTENANCE AND AFTER
CORRECTIVE MAINTENANCE CORRECTIVE MAINTENANCE

TABLE XI
MOTORS AND RELATED REPAIR SHOP
0.94%, contributes to this efficiency increase. Even a small
percentage reduction in stator copper loss is significant as it
accounts for about 40% of full load losses in the induction mo-
tors.
Considering that the repairer used a wire of the same original
gauge, this reduction of stator resistance may be explained by phenomenon as it affects core magnetic characteristics and in-
the use of better raw materials, fewer coils or by changing the terlaminar insulation.
original winding pattern. In general, the original windings were The full load speed has a reversal characteristic with the rotor
made in winding machines while in the shops they were hand- losses, as it decreases as the losses increase, and the other way
made. It is very difficult to copy the original winding with an around. The high and negative correlation coefficient (−0.971)
equal number of coils inside the slots. reinforces this statement.
The greatest increase of stator resistance (23.5%) occurred in In addition to an increase in motor efficiency, full load speed
one shop that has a coil winder, while the higher reductions in reduction decreases heat dissipation, thus increasing the tem-
stator resistance (both −12.7%) happened in shops with hand- perature of operation. The efficiency of this heat transfer also
wound coils. Hand made coiling may allow a smaller curve depends on the environmental temperature. This is noticed when
shape diameter in the coil heads, resulting in a smaller wire comparing rotor losses, full load, and temperature.
length required for a same number of loops. The stator resistance variation resulting from rewinding,
Due to the severe nature of the induced breakdown (overload windage and friction losses, replacement of bearings, cleaning
burning) and the repair procedure used in some shops (use of and lubrication are important factors regarding the efficiency
torch), a large variation in core loss was expected. An aver- differences after repair were completed.
age full-load core loss increase of 0.34% was observed, with The average reduction in windage and friction losses
final values differing from about −10 to +10 percent from was the most expressive variation observed (−19%) ranging
initial value. Temperature increase is the main cause of this from +35.6% to −47.3%. Cleaning, lubrication, and bearing
388 IEEE TRANSACTIONS ON ENERGY CONVERSION, VOL. 22, NO. 2, JUNE 2007

replacement are the main reasons for this change. This loss re- [15] The Effect of Repair/Rewinding on Motor Efficiency, “EASA/AEMT
duction represents an increase of about 0.5% point increase in Rewind Study and Good Practice Guide Best Practices to Maintain Motor
Efficiency,” http://www.easa.org, 2004.
motor efficiency. [16] T. H. Bishop, “Motor Repair Industry Response to the Energy Policy Act
of 1992: A Status Report”.
[17] R. L. Nailen, “A user’s view of motor repair standards and specifications,”
IV. CONCLUSION IEEE Trans. Ind. Appl., vol. 24, no. 6, Nov.–Dec. 1988, pp. 1131–1137.
This paper presented an evaluation of the performance of
three-phase induction motors after preventive and repair main-
tenance. Regarding motors already in use, the results show that
motor efficiency has a small increase after the repair, making Edson da Costa Bortoni (S’94–M’96–SM’05) was born in Maringá, Brazil,
motor repair an attractive practice. It was also noticed that the on December 1, 1966. He received the degree from Itajubá Federal University
overall quality of the repair shop related to its adopted main- (UNIFEI), Itajubá, Brazil, in 1990, the M.Sc. degree in energy systems plan-
ning from University of (UNICAMP), Campinas, Brazil, in 1993, and the D.Sc.
tenance procedures, employed materials, equipment used, and degree in energy and electrical automation from the University of São Paulo
personnel qualifications and capabilities, are of great importance (USP), São Paulo, Brazil, in 1998.
to have a motor successfully repaired and provide an efficient He was a Professor at São Paulo State University (FEG-UNESP) and Uni-
versity of Amazonas, Brazil. Presently, he is a Professor at UNIFEI. His current
application. As windage and friction have a considerable influ- research interests include instrumentation, synchronous machines, and energy
ence in motor performance, this work also showed how low cost systems.
and simple practices of preventive maintenance, such as clean- Dr. Bortoni is a member of ISA and SPIE.
ing and lubricating the motor, can result in better performance
and increases the overall efficiency of motor systems.

ACKNOWLEDGMENT
Jamil Haddad was born in Itajubá, Brazil, on June 5, 1959. He received the
The authors gratefully acknowledge the contributions of degree from Itajubá Federal University (UNIFEI), Itajubá, Brazil, in 1982, the
Dr. R. P. Tabosa, and Dr. G. A. Soares from PROCEL—the M.Sc. degree in power systems from UNIFEI, in 1983, and the D.Sc. degree in
Brazilian National Electrical Energy Conservation Program— energy systems planning from University of Campinas (UNICAMP), Campinas,
Brazil, in 1992.
for their valuable comments and support during the development He is a Professor at UNIFEI and an Advisor of Brazilian Energy Ministry.
of this work. E. C. Bortoni would like to acknowledge the con-
tribution of Dr. E. Jacobson, from Quadlogic Controls, for the
technical review of the paper.

REFERENCES
[1] Selected Bibliography on Electric Motor Repair. U.S. Department of Afonso Henriques Moreira Santos was born in Itanhandu, Brazil, on April 1,
Energy, Washington, DC, Nov. 1999. 1957. He received the degree from Itajubá Federal University (UNIFEI), Itajubá,
[2] M. Zeller, “Rewound high-efficiency motor performance.” Guides to En- Brazil, in 1978, the M.Sc. degree in mechanical engineering from UNIFEI,
ergy Management. BC hydro. Canada, Aug. 1992. in 1981, and the D.Sc. degree in automation from University of Campinas
[3] D. H. Dederer, “Rewound motor efficiency.” Ontario Hydro Technology (UNICAMP), Campinas, Brazil, in 1987.
Profile, Canada, Nov. 1991. From 1997 to 2000, he was a Director of ANEEL, The Brazilian Electrical
[4] A. Charette and P. Anger, “Impact de la ráparation des moteurs álectriques Energy Agency. From 2001 to 2004, he was the National Secretary of Energy.
sur leur rendement.” Laboratoire des technologies álectrochimiques et des Presently, he is a Professor at UNIFEI and is an Advisor of Brazilian Energy
álectrotechnologies d’Hydro-Quábec (LTEE), 1995. Ministry, of Power and Energy Utilities, and of Engineering Companies.
[5] J. S. Allen, “Repair procedures to maintain energy efficiency” in IEE
Half-Day Colloquium on Refurbishment of Machines, May 1998.
[6] D. G. Walters, “The repair of low tension induction motors—Best practices
to maintain energy efficiency” in IEE Half-Day Colloquium on Refurbish-
ment of Machines, May 1998.
[7] C. Yung and A. H. Bonnett, “Repair or replace?,” IEEE Ind. Appl. Mag.,
Sep.–Oct. 2004, pp. 48–52.
[8] S. Nadel, M. Shepard, S. Greeberg, Z. Katz, and A. Almeida, Energy Erick Menezes de Azevedo was born in Rio de Janeiro, Brazil, on May 5,
Efficient Motor Systems, ACEEE, 1992. 1974. He graduated from Itajubá Federal University (UNIFEI), Itajubá, Brazil,
[9] C. Yung and A. H. Bonnett, “A repair-replace decision model for petro- in 1998, received the M.Sc. degree in power systems from UNIFEI, in 2000,
chemical industry electric motors,” in Petrol. Chem. Ind. Conf. 2002. and the D.Sc. degree in energy systems planning from UNICAMP, in 2003.
Industry Applications Society 49th Annual, Sep. 23–25, 2002, pp. 55–66. Presently, he is a Researcher at UNIFEI on energy markets.
[10] Methods for Determining Losses and Efficiency of Rotating Electrical
Machinery from Tests, IEC Standard 34-2.
[11] IEEE Standard Test Procedure for Poly-Phase Induction Motors and Gen-
erators, IEEE Standard 112, 1997.
[12] O. V. Thorsen and M. Dalva, “A survey of faults on induction motors
in offshore oil industry, petrochemical industry, gas terminals, and oil
refineries,” IEEE Trans. Ind. Appl., vol. 31, no. 5, Sep.–Oct. 1995, pp.
1186–1196. Roberto Akira Yamachita was born in Guarulhos, Brazil, on August 30, 1966.
[13] J. W. Barnes, Statistical Analysis for Engineers and Scientists—A Com- He received the degree from Itajubá Federal University (UNIFEI), Itajubá,
puter Based Approach. New York: McGraw-Hill, 1994. Brazil, in 1992, the M.Sc. degree in power systems from UNIFEI, in 1998.
[14] IEEE-PCIC Recommended Practice for Repair and Rewinding of Motors He is currently researching energy efficient systems and renewable energy
for the Petroleum and Chemical Industry, IEEE Standard 1068, 1996. systems at Energy Studies Group, Itajubá, Brazil.