A NEW METHOD FOR STATOR CORE REPAIR

REDUCES CORE LOSSES AND REPAIR TIME

Copyright Material IEEE
Paper No. PCIC-2006-14

Chuck Yung
Senior Member, IEEE
Electrical Apparatus Service Association, Inc. (EASA)
1331 Baur Blvd.
St. Louis, MO 63132
USA
cyung@easa.com

Abstract — The steel laminations used in the stator core of

electric motors and generators are electrically insulated from one
another to reduce eddy-current losses. When a bearing failure
results in friction between the rotor and stator, or a ground fault
occurs, laminations are often fused together. Since traditional
methods of core repair are labor-intensive, even moderate core
damage may preclude economical repair in the few percent of
motor failures with significant core damage. There is a new
process that can restore the coreplate without dismantling the
core, saving labor and eliminating the chance of unsatisfactory
restack results. This chemical conversion process simultaneously
etches interlaminar shorts and forms a durable coreplate with
excellent electrical insulation. The potential benefits to industry,
in terms of labor saved and reduced mean time to repair, are
substantial. The cost to restack the stator core of a 5,000 kW
machine can exceed $30,000 US, and require two weeks time.
This new process can restore the core in less than one day.

Index Terms – Coreplate, Lamination, Core loss, Core test.

I. INTRODUCTION

When, during the course of a repair, a motor or generator

stator is found to have unacceptable core losses, the repair-
replace decision becomes one of economics. Is the motor large
enough, or special enough, to justify the labor-intensive process
required to dismantle, clean, re-insulate and restack the core? If
a replacement motor is not available within a reasonable time,
traditional methods of restack make for a costly repair.
When the time required to repair a damaged core exceeds Fig. 1. A 9.5 MW stator core being prepared for the new
allowable downtime, a desperate motor user might direct the treatment.
service center to rewind a motor with a marginal core. The motor
thus repaired is returned to service with reduced efficiency and Tests were carried out in 5 different service centers in the North
higher operating costs. America. Because service centers do not all use the same equip-
The new process has reduced static core losses to approxi- ment, and larger electrical machines require the use of a loop test
mately 4 watts/kg (2 watts/lb) in nearly every repair, even in cases [1][2], it was not always possible to obtain power factor
where the initial static core loss approached 12 watts/lb (26 watts/ readings.Core testing in the repair industry is done by static core
kg) . Note that only static core loss tests were done. Static core test, resulting in comparative numbers for watts loss. The values
tests are performed using a commercial core loss tester or a loop are considerably different than those obtained during a dynamic [3]
test [1][2], as compared to dynamic core loss measurement core test. A core test, whether by loop test or commercial core
described in IEEE 112 (Method B.) Cores up to 9.5 MW have been tester, only qualifies the eddy-current losses and identifies hot
successfully repaired with the new process (Fig. 1). spots due to localized shorting of laminations. The watts loss

1
values presented in this paper would be alarming to those familiar B. Coreplate
with electrical steels, without this vital bit of information.
The expensive nature of the traditional restack procedure has Due to the labor-intensive nature of traditional restack meth-
driven the use of alternative procedures. This paper also ex- ods, a lot of alternatives have been tried and, in some places,
plores some of those options. adopted. One of the better coreplate coatings used by manufac-
Historically, when core damage occurred, there were several turers of rotating electrical equipment is a particular form of iron
repair options. In descending order of cost these are: oxide: Fe3O4. The fact that Fe3O4 is durable, has excellent
• Replace the entire laminated core. electrical insulating properties and good thermal conductivity
• Dismantle, clean, re-insulate and re-assemble the original has given rise to one of the more enduring core repair myths: that
laminations. a damaged stator core can be soaked with water and allowed to
rust, thus forming a coating of iron oxide. Red rust is FeO2 – which
• Spread the damaged areas, clear the shorted laminations,
is fragile, a poor electrical insulation, and inhibits heat transfer.
and re-insulate them.
The properties of various iron oxides vary widely. FeO can be
• Use one of a variety of methods to repair the damaged
explosive, and when mixed with aluminum powder becomes
areas, then separate and re-insulate the laminations.
thermite, a useful product for burning holes through steel but
Strictly speaking, coreplate is the coating applied to electrical hardly a desirable material in a petro-chemical plant. Fe3O2 is
steel as interlaminar insulation. Interlaminar insulation can also hematite, also called iron ore.
be formed on semi-processed steel during the annealing pro-
Because the formation of Fe3O4 requires exacting control of
cess, by controlling the humidity and oxygen in the annealing
temperature, humidity, oxygen content and oven exposure time,
oven. This is done subsequent to the annealing process. There
the manufacturers’ cost — for energy to operate the oven and
are three distinctly separate factors that contribute to core
additional handling required to load and unload the oven — is a
losses:
significant portion of the total cost of laminations.
• Hysteresis losses
• Notching stresses
• Eddy-current losses
The first two are addressed by the manufacturer when the
motor is produced. Hysteresis losses are attacked by the anneal-
ing process, by controlling grain size. Notching stresses, induced
when the laminations are stamped, contribute to the hysterisis
losses and are relieved by the annealing process. Motor opera-
tion or subsequent failure should not alter these. Eddy-current
losses in individual laminations vary as the square of the thick-
ness of the lamination. When two or more adjacent laminations
are shorted together, they behave as a single, thicker lamination.
Burs left by the punching process are a likely place for the
interlaminar insulation to be compromised. Likewise, any wind-
ing failure that includes a ground fault is likely to fuse laminations,
creating localized eddy currents. The interlaminar insulation
used to control eddy-current losses can be fragile, depending on Fig. 2. The relatively large backiron of a 2-pole lamination
the method the manufacturer selected to form it. requires more exposure time than smaller/low-speed ma-
Fully-processed laminations are already coated by the manu- chines.
facturer of the electrical steel. They do not require annealing by
the motor manufacturer. While fully processed steel has a higher When numerous sizes and designs of laminations (Fig. 2) are
initial cost, the indirect costs of production may be lower since the processed simultaneously in large batches, it is difficult to opti-
laminations do not have to be coated or annealed. Larger mize the exposure time for all. Deviation from the correct mix of
laminations, when annealed, are more difficult to handle without time, oxygen, humidity and temperature can result in formation of
bending or deforming them. This results in increased handling iron oxides other than the desirable Fe3O4.
time and higher rejection rates. For NEMA 400-frame motors and Aside from those critical factors, other variables which affect
larger, most manufacturers use fully processed steel. quality control include the processing of large 2-pole laminations
(more backiron area) with smaller and/or low speed laminations,
A. Traditional Restack stacking too many laminations, and surface rust present before
oven processing. The lamination in Fig. 3 illustrates a common
The damaged core is completely dismantled [4], separated problem when a lamination passes through the oven too quickly.
into individual laminations. Those laminations are then cleaned, The circumferential band of color variation clearly indicates that
sanded and coated with an insulating material – usually an the entire lamination did not reach the desired temperature.
organic varnish (C3) or inorganic material (C5) developed for use Some manufacturers use a less well-controlled process, which
as coreplate.The pieces are reassembled to match the original results in a “steam blued” coating – a thin blue coating as in Fig. 4.
configuration of the core, compressed at 5-8.8 kg/cm2 (75- Other core repair alternatives also have drawbacks. Sodium
125 psi), and tested to confirm that the watts loss is acceptable. silicate (Na4SiO4), colloquially called “waterglass” when suffi-

2
 Fig. 3. The differences in color and coreplate formed on the
backiron of this lamination illustrate what happens when cycle
time is too short.

Fig. 5. Segmented laminations are typical for cores with bore

diameters greater than 89 cm (35”).

translates to delays in returning a motor to service, and consid-

erable expense. Obtaining the correct stacking pressure [typically
5-8.8 kg/cm2 (75-125 psi) of lamination area], and maintaining
the stack geometry are but two potential problems. Cores with
bore diameters 89 cm (35”) and larger are often constructed of
segmented laminations. These can best be described as a series
of short arc-shaped pieces of lamination (Fig. 5) stacked end-to-
end in a continuous spiral until the entire laminated core is
formed. Orientation of the segments is critical. Inattention to
Fig. 4. An inferior oxide coating with blue coloration. detail may result in a core that is eccentric, tilted or otherwise
departs from proper core geometry.
A stator core measuring 122 cm (48”) long with a 137 cm (54”)
ciently diluted with water, is used to coat laminations; or in inside diameter, of segmented construction, requires approxi-
attempts to soak between laminations after fused laminations mately 15,400 individual pieces of lamination. Laboriously
have been separated. Sodium silicate is water soluble, conse- cleaning, sanding and recoating each piece, then restacking
quently it does not provide a permanent coating. them and pressing with the correct pressure, all the while
To dispel other repair myths: Etching with hydrogen peroxide keeping the core concentric and round, is a slow process.
does not work. Muriatic acid or other mild acids may etch the
surface including the shorted areas when prepared at the correct
II. A BETTER ALTERNATIVE, BUT FIRST SOME
strength. When too weak a concentration is used, the damage is
not cleared, and if not adequately diluted, acids may cause
HISTORY
further damage to the core. Acids do mothing to restore the
interlaminar insulation. In the late 1800s, the growing firearms industry had three
options to improve the rust-resistant characteristics of their
When a damaged core is dismantled and repaired, and the
product:
damaged laminations are cleaned and recoated, the material
used for coreplate is yet another variable in the long-term • Various bluing techniques
success of the repair. One product called lamination coating is • A stabilized flash-rust process
actually a class B insulation intended to coat the stator bore and • A “new” process called Parkerizing
prevent rust. When used to coat the repaired laminations, it will The durability of the latter process led to its adoption by
not withstand a normal burnout oven cycle, so the next rewind will manufacturers of military weapons. Those who served as mem-
likely require another expensive restack. bers of the armed forces are well familiar with the greenish gray
The labor required to dismantle and repair a stator core coating of the M1 Garand, the M1 carbine and the M14 rifle. One

3
of the author’s hobbies, the restoration of historical firearms, led
to the belief that this finish could be adapted to the coating of
laminations for electrical equipment, specifically rotating electric
machines.

III. THE BETTER ALTERNATIVE

The process as adapted involves the use of an aqueous

solution of phosphoric acid with certain elements (including zinc,
manganese and iron) in suspension.
Phosphoric acid has been used for decades to etch damaged
laminations of large (i.e., several MW) machines. After the
etching process, insulating material had to be applied to restore
the coreplate. This requires separating individual laminations in
the damaged areas, applying epoxy and trying to soak it in
between the laminations. The use of phosphoric acid for this
purpose has a long history of success. The new process includes
certain additives which complete the process. Fig. 6. A bearing failure results in friction between rotor and
Proper application of the new process clears the interlaminar stator, with subsequent increase in core losses.
shorts between laminations in stator cores. The chemical solu-
tion etches the shorted laminations, then converts the steel
surface to form a durable coreplate. It is worth noting that two key
ingredients – zinc and manganese – are sometimes added to
electrical steel to improve its electrical resistance. The treatment
reduced static core losses by 32-38%, and improved stator core
power factor by 10-12%.
The final core loss after treatment ranged from just under, to
just over, 4.3 watts/kg (2 watts/lb) as measured using a core-los
ltester. There is not universal agreement about “acceptable”
watts/pound limits, but most commercially available core testers
suggest a static core loss limit of approximately 13 watts/kg
(6 watts/lb).
The static core test is used to compare before and after
burnout core condition. Manufacturers of commercial core loss
testers use algorithms consistent with references 1 and 2,
although not all use the same acceptance criteria. While a
commercial core loss tester might consider 13 watts/kg (6 watts/lb) Fig. 7. Core repair by separating fused laminations and
an acceptable static core loss, the author suggests 8.5 watts/kg reinsulating them is labor intensive.
(4 watts/lb) as a conservative limit.
Direct comparison of static and dynamic core losses confirm greater convenience.
that the core loss values do not match. However, the results do The results of treating those two cores are condensed in
correlate[5]. That is, if a static core loss increases, dynamic core Tables B-I and B-II. The watts loss of each core was reduced
loss determination[3] will confirm this. from a pre-treatment value of 17-26 watts/kg (8-12 watts/lb), to
Several individual laminations were treated, and resistance less than 4.3 watts/kg (2 watts/lb). A third core, repaired at a
tests of the insulation qualities indicated improvement. To judge different service center, yielded similar results.
whether this process might be a cost-effective repair, an entire
laminated core was treated next. The core losses of several B. The Treament Process
small cores treated by immersion yielded improved results with
no negative side-effects.
Successful treatment depends on several easily-controlled
steps. First, the solution and the damaged core must be pre-
A. Tests of Motors heated to between 82°-100° C (180°- 210 ° F) . Second, the stator
bore is scrubbed with a coarse abrasive or bead-blasted. Best
Two identical damaged 6-pole cores, in NEMA 405 frames results were obtained with a drum sander, with the drum abrad-
were treated (Figs. 6 and 7). Prior tests involved total immersion ing the surface circumferentially with respect to the bore of the
of the core in the solution. For these larger cores, several stator. The heated core must be continuously wetted with the
alternatives to immersion were tried. Applying rags soaked with solution for 15 minutes, after which the core should be rinsed
solution worked, but the use of commercially available spray using cold water and then baked dry. A final core test confirms the
bottles to keep the core surface wetted yielded good results with results.

4
 The chemical solution results in a physical micro-coating on
the surface of the steel, but must be able to reach the damaged 15
areas. While the steps seem simple and the description brief,
each step is important to the process. Preheating the core is
critical to the ability of the solution to penetrate between the
12
tightly compressed lamination stack. This can be explained by
Poiseuille’s Law:

P1 - P2 π × Pressure difference × Radius 4 9

Volume flowrate = =

Watts/kg
R 8 × Viscosity × Length

8nl
Where R =
pi × r 4 6

An example familiar to most readers is soldering. The techni-

cian does not have to understand Poiseuille’s Law to realize that
the solder flows towards the heat source. The higher the solder 3
temperature, the lower the viscosity (the solder is thinner). The

Before

Before

Before

Before

Before
After

After

After

After

After
lower the viscosity of the liquid, the faster the flow.
To apply this to the preheated machine core: Comparing the
teeth to the backiron, the teeth have more surface area per unit 0
Motor 1 Motor 2 Motor 3 Motor 4 Motor 5
volume and the backiron has more mass to retain heat. The teeth
cool at a faster rate, and Poiseulle’s Law explains that the liquid Fig. 8. Improvement in watts loss. Also see Appendix B.
solution flow towards the region of higher temperature (lower
pressure). Preheating the core before applying the solution, then
removing it from the heat source, is a critical step in the process. 100° C
Scuffing of the surface assures a clean steel surface for the
chemical reaction. A polished surface is not desirable. The use
of a drum sander results in the sanding action parallel to the 80° C
laminations, which avoids smearing of adjacent laminations. A
relatively rough surface provides more surface area for the zinc
Hot Spot Temperature

phosphate to form. The micro-porosity of the surface retains

resin when the rewound core is VPI processed. 60° C
Preheating the solution is done to maximize the amount of
solids in suspension. At room temperature, some of the solids
percolate out, and must be reabsorbed by heating and the
solution before application. 40° C
The exposure time is the one variable that can be adjusted.
The longer the contact time between the solution and the steel
being treated, the more aggressive the coating formed. Zinc salts
20° C
leave a white residue on the core surface, hence the need to rinse
Before

Before

Before

Before

Before
the core. Note that zinc salts do not promote rusting.
After

After

After

After

After
The following insulating materials were soaked for several
hours in the solution, dried and hipot tested: 0° C
• Nomex Motor 1 Motor 2 Motor 3 Motor 4 Motor 5
• DMD Fig. 9. Reduction of hot spot tempearture. Also see
• Mylar Appendix B.
• Several Dacron-based products used as tie cord or lacing.
No adverse effects were noted; the insulation did not delami-
nate or exhibit other harmful effects, and each of the insulating treatment process, they were able to further reduce the core
materials passed a 1-minute 2 kV AC hipot test. losses to 5 watts/kg (2.33 watts/lb).
After success with several cores up to 100 kW (125 hp) rating, The largest of several cores treated to date (Figs. 8 and 9) is
a 1550 kW (2100 hp) core was selected for treatment. The stator a 9.5 MW core, shown in Figs. 1 and 10.
had core losses of 25.9 watts/kg (11.98 watts/lb) before any The 9.5 MW core weighed 17,250 kg (38,000 pounds), with a
repairs were attempted. The service center tried physically core length of 203 cm (80”) and a bore diameter of 216 cm (85”).
manipulating the core, rusting the core, and using sodium Widespread shorting of the core included the surface and numer-
silicate. The core losses were reduced to 13.2 watts/kg ous slots, as evidenced by the thermal images in Appendix C.
(6.13 watts/lb) before contacting the author. Using the new Visual examination of the stator core revealed that the core is

5
 to use plastic or glass beads. Unfortunately, it was later discov-
ered that the outside contractor had not completely removed the
aluminum oxide from the hopper before cleaning this core. This
resulted in the localized surface heating visible in the thermal
image in Fig. 11.
A drum sander was used to scrub the surface in those areas,
during the loop test, which cleared surface shorts and decreased
the temperature to that of the rest of the core.
Damaged areas of the core were carefully marked during the
core loss test, and thermal images taken to document the
condition of the core. The core was baked for 24 hours at 99° C
(210° F) prior to treatment. The solution was heated and applied
by two technicians wearing full personal protective equipment
(PPE) (Fig. 10). Keeping the damaged areas continuously wet-
ted was challenging given the size of the core. (A pressurized
applicator would be more practical for a core of this size.)
Following treatment, the core was pressure-washed with cold
water, and allowed to dry.
A second core test was performed using the same loop-turns
of cable, placed in the same position on the core, with the same
applied voltage. The watts/loss improvement (Table V) was not
as dramatic as for previous cores treated, reflecting the large
Fig. 10. The 9.5 MW stator core being treated with the new core size and localized nature of the damage. Thermal images
process. of the core, pre- and post-treatment, are more dramatic than the
core loss figures.
comprised of approximately 34,400 individual pieces. The time to treat the core, rinse it with cold water, and conduct
The original winding had been treated by global VPI. Because a loop test was less than 8 hours. The stator core was dried,
they had experienced similar problems with an identical machine allowed to cool to ambient, and thermal images (Appendix C)
repaired under supervision of an OEM engineer, the machine were taken during a final loop test. The ampere-turns, ambient
owner’s representative directed that the coils be removed using temperature and test time duplicated the before-treatment test.
mechanical force after cutting off one entire coil extension. At his Three stator cores have been processed twice each through
insistence, open flame had been used to “warm” the insulation to normal burnout oven cycles (370° C) with no apparent degrada-
facilitate removal. tion of the interlaminar insulation.
The stripped core was then cleaned by an outside contractor The question must be asked: “What can the new treatment
specialized in abrasive-blasting. The contractor was instructed process not do?”
• It cannot restore missing iron.
• The etching action is not aggressive enough to remove/
repair large volumes of shorted laminations.
• The solution works best on bare steel, so it is doubtful
whether it would be helpful in treating a shorted core with the
windings intact, except when the damage is superficial
surface shorting. In those cases, if the surface is properly
prepared and the core preheated, the treatment method
should be helpful.

IV. CONCLUSIONS

The new process for restoring interlaminar insulation reduces

the labor required to repair damaged cores by as much as 90%.
By reducing the repair time without compromising the end result,
machines which might otherwise be scrapped can be economi-
cally repaired. This helps the end user by reducing repair costs,
and by avoiding long lead times when a replacement machine is
not immediately available.
Treatment of brand new cores, donated by a manufacturer,
Fig. 11. With the exception of small areas of surface damage indicate a modest reduction in core losses. This process should
caused by aggressive blast media, the temperature of the be useful to manufacturers, in reducing the reject rate of new
entire core was reduced by ~40° C (50° F). cores. The process is applicable to DC armatures (Fig. 12), and

6
 Fig. 12. DC armature cores can also be repaired using the
new process.

should be equally effective in reducing rotor losses.

When damage occurs during a motor failure, increased core
losses adversely affect insulation life as well as operating cost
unless corrective measures are taken. The new process greatly
decreases both MTTR and repair cost, as compared to traditional
restack methods. The reparability of motors previously consid-
ered uneconomical to repair will benefit motor users.

V. REFERENCES

[1] P. Christensen, EASA Tech Note 17: Stator Core Testing,

Electrical Apparatus Service Association, Inc. (EASA), St.
Louis, MO, Oct. 1992 (Updated Oct. 2003).
[2] IEEE Std. 432-1992, IEEE Guide for Insulation Mainte-
nance for Rotating Electric Machinery (5 hp to less than
10,000 hp), Institute of Electrical and Electronics Engineers,
Inc. (IEEE), New York, NY, Sept. 1992.
[3] IEEE Std. 112-2004, IEEE Standard Test Procedure for
Polyphase Induction Motors and Generators, Institute of
Electrical and Electronics Engineers, Inc. (IEEE), New York,
NY, May 1997.
[4] C. Yung, “Motor Restack Procedures”, Currents, Electrical
Apparatus Service Association, Inc., July 1999.
[5] The Effect of Repair/Rewinding on Motor Efficiency, Electri-
cal Apparatus Service Association, Inc. (EASA) and
Association of Electrical and Mechanical Trades (AEMT),
St. Louis, MO, 2003.

VI. VITA

Chuck Yung is a technical support specialist for the Electrical

Apparatus Service Association (EASA). Mr. Yung worked in the
electrical machinery repair industry from 1974 until joining the
staff of EASA in 1998. He received his AS from NHCC in 1988
and his BBA from the University of Houston in 1990. He holds one
provisional patent and is the author of more than 100 technical
articles and papers, including one published in IEEE Industry
Applications.

7
 APPENDIX A

TERMS
Burnout process. The most commonly used method for
winding removal is to heat the stator to a controlled temperature
of 360° C (700° F) to pyrolize the windings. Breaking down the
volatile products makes it easier to remove the windings, which
in turn prevents mechanical damage to the laminated core.

Coreplate is an electrical insulation on electrical steel lamina-

tions. Coreplate is usually comprised of organic varnishes (C3),
or inorganic material (C5). It varies widely in quality, depending
on motor size and manufacturer. Some European manufacturers
use a “steam blue” process, which results in a fragile insulation,
blue in color – and susceptible to damage from the burnout
process.

Laminations are thin sheets of steel, stacked together to

comprise the stator core of an electric motor. They must be
electrically insulated from one another, to control eddy-current
losses when the motor operates. Typical thickness of an indi-
vidual lamination is .5 mm-.7 mm (0.022” to 0.028”), so a typical
core has approximately 40 laminations per inch of stack length.
Given a typical 300 hp core is over 61 cm (2 feet) long, the labor
required to manually clean a thousand laminations (24 / 0.024 =
1,000) is evident.

Power factor is not reported by all commercial core loss te

lters. It s an indication of the stacking “tightness” and can indicate
how widespread the damage. PF below 0.70 is considered
“good”, while 0.7 – 0.9 is “marginal”. Values of PF in excess of 0.9
are considered “bad”.

Watts loss/pound(kg) is a measure of the quality of

interlaminar insulation (coreplate). Watts/pound limits vary with
type of steel (carbon or silicon) and the core loss tester used but
the lower the watts loss/pound, the better. Commercial core
testers generally use 13-17 watts/kg (6-8 watts/lb) as an upper
limit. Some service centers use a more conservative value of
8.6 watts/kg (4 watts/lb).

8
 APPENDIX B

TREATMENT RESULTS
TABLE B-I TABLE B-II
TREATMENT OF CORE #1 TREATMENT OF CORE #2

Core loss Power Hot spot Core loss Power Hot spot
watts/kg (watts/lb) factor temperature watts/kg (watts/lb) factor temperature

Before 8.10 (3.75) 0.70 43° C (109° F) Before 9.09 (4.21) 0.67 53° C (128° F)

After 4.23 (1.96) 0.50 23° C (74° F) After 5.05 (2.34) 0.58 28° C (83° F)

This NEMA 405 frame stator (Fig. 7) was fanned (laminations This NEMA 405 frame stator (Fig. 6) was not fanned (laminations
spread individually) and processed. Ambient temperature of the spread individually), watt-knocked or otherwise prepared prior to
core was 22° C (71° F). the etching treatment. Ambient temperature was 22° C (71° F).

TABLE B-III TABLE B-IV

TREATMENT OF CORE #3 TREATMENT OF CORE #4

Core loss watts/kg Power Hot spot Core loss Power Hot spot
(watts/lb) factor temperature watts/kg (watts/lb) factor temperature

Before 13.60 (6.30) 0.74 84° C (182° F) Before 13.20 (6.13) — 43° C (109° F)

After 6.09 (2.82) 0.74 27° C (81° F) After 5.00 (2.33) — 18° C (65° F)

This 100 hp, 8-pole stator was not mechanically manipulated A 2100 hp, 2-pole stator treated after other techniques failed to
prior to etching treatment. reduce core losses to an acceptable level.

TABLE B-V
TREATMENT OF CORE #5

Core loss watts/kg Power Hot spot

(watts/lb) factor temperature

Before 4.97 (2.30) — 53° C (128° F)

After 4.10 (1.90) — 28° C (83° F)

9.5 MW stator

9
 APPENDIX C

THERMAL IMAGES OF CORE BEFORE AND AFTER TREATMENT

Fig. C-1. Thermal images taken before the 9.5 MW stator core was treated. Note the widespread heating.

Fig. C-2. The same 9.5 MW stator core after treatment. Average temperature was reduced approximately 50° F. The small
localized hot spots on the surface are believed to be the result of abrasive blasting by an outside contractor.

10