Você está na página 1de 5

Detecting Load Unbalance and Shaft Misalignment

Using Stator Current in Inverter-Driven Induction Motors


Ramzy R Obaid Thomas G. Hahetler Rangarajan M.Tallam
School ofElechical and Computer Engineering
Georgia Institute of Technology
Atlanta, GA 30332-0250 USA
Phone :(404) 894 9829 Fax :(404) 894 4641
Email :thabetler63ee.gatecb.edu

Abstract- This paper examines the effect of changing the input stator current are present. the effects of supplying the
frequency of the induction motor with an adjustable speed drive induction motor from an adjustable speed drive on the
on the detection of mechanical fault conditions, such as load detection of load unbalance and shaft misalignment are
unbalance and shafi misalignment The mechanical force investigated. Mechanical fault conditions are detected using
resulting from these fault conditions is highly dependent on the
rotational speed of the motor. When the motor is run at low a simple and low-cost algorithm that utilizes only a single
speeds, the mechanical fault produces less force, resulting in phase of the stator current. The employed algorithm could be
smaller frequency harmonics in the stator current easily incoprated into the drive at little or no additional
Nevertheless, results show that the fault signatures are evident cost. Most drives use DSPs, and usually have enough room
in the stator current spectrum even when the drive input to include a simple algorithm such as the one implemented in
frequency is very low. In addition, the drive frequency this paper.
harmonics do not affect the fault detection, since the frequency
components of interest are much lower than the drive Mechanical faults of interest are tested for the case where
frequencies. It is shown that the tested fault couditious are the induction motor is operated at different drive input
detected in an induction motor "Iing at a rotational speed of . frequencies (in an open-loop control). However, this paper
as low as 150 rpm. Mechanical fault conditions are detected does not consider the case where the supply frequency is
using a simple, low-cost algorithm that utilizes a single phase of changing while the induction machine is in operation. Since
the stator current The employed algorithm could be easily the fault detection scheme utilized in this work depends on
incorporated into the drive at no additional cost,
the supply kequency, baseline measurements would be
complicated if the frequency is changing during machine
I. INTRODUCTION operation.

Load (or rotor) unbalances and shaft misalignments are


common mechanical fault conditions in induction machine n. STATOR CURRENT AND MECHANICAL FAULTS
applications. It is often desirable, as with all other motor
faults, to detect the presence of such conditions early enough Load unbalances and shaft misalignments in induction
in order to avoid sudden downtimes and effectively plan machines can be detected using only a single phase of the
maintenance [l]. Adjustable speed drives control the stator current. The basic concept is that these fault conditions
majority of induction motors in industry applications. These create air-gap eccentricities, which change the frequency
drives introduce voltage harmonics into the machines supply, spectrum of the line current from. that of a non-faulted
which may affect the vibration levels of induction machines machine. The air-gap eccentricity affects the inductances of
[2, 31. Considerable research has been focused on the the motor resulting in harmonics (L) at rotating kquency
detection of air-gap eccentricities caused by mechanical fault sidebands of the supply frequency predicted by [SI
conditions in line fed induction machines [4, 5 , 6,71. Early
detection of these faults was shown to be possible by
monitoring the frequency spectrum of a single phase of the
motor stator current for the existence of specific harmonic
frequencies associated with the presence of these faults. In
[7], mechanical fault detection was focused on line fed
induction motnrs only. In this paper, the work is extended to
examine the ability to detect meclianical fault conditions in where f. is the input kquency, m is an integer, s is the per-
the case where other sources of frequency harmonics in the unit slip, and p is the number of poles. This technique for

0-7803-7817-2/03/$17.00 02003 IEEE 1454


detecting air-gap eccentricities has the advantage of not TABLE I
requiring any knowledge of the construction of the motor. L m R O U E N f f SDEBAhDS FOR THE TESTE0 INDUCTION
For a four-pole machine and m=l, equation (I) reduces to M A C H M E AT DIFFERENT DRIVE INPUT FREQUENClES

(2) LTh% Input Frequency FrequencySdebands with -


1

which is equivalent to @fr, where f, (Hz) is the rotating fmk W890M


mechanical speed of the motor. The slip in equations ( I ) and
(2) indicates load dependency of these rotating frequency
sidebands. Therefore, it is important to consider the case
when the motor is run at no-load condition as well as when
loaded.
The experiments presented in this paper examine the
ability to detect these sideband kequencies when a rotor
unbalance or a shaft misalignment is present in an inverter
fed induction machine operated at different drive input
kequencies. The drive control scheme is open loop using
constant volts per hertz control. It is important to note that
both the inverter output voltage and frequency are assumed
constant for a particular speed command for all the
experiments in this work. Since the predicted frequency
sidebands depend on the supply frequency, reference
measurements would be complicated if the line frequency is
changing during machine operation. This assumption also
makes the operation similar to running the machine &om the
utility supply line. The experiments examine the fault
detection schemes when the machine is operated at a drive I, !
input frequency of as low as 5Hz. In addition, the
experiments are conducted with the motor run at different I!
I/
,
load conditions to examine the effect of loading the machine i,l !i , ,
on detecting these mechanical fault conditions. 5 10 15 20 25 30 35 40 45 50 55
F-MW (W
Fig. 1. Current Pequsncy spechum (m) of the tesl machine
ID. EXPERIMENTS (four-pole) with B rotor unbalance, driven at 5fi and no-load.

Two types of mechanical faults were evaluated in the lab to


examine their affects on the lime current frequency spechum
while driving the motor with an adjustable speed drive: rotor, I /
or load unbalance and radial shaft misalignment. Each of
these types was induced mechanically to different degrees of
severity and the stator current was acquired and analyzed for
each severity level. An ABB ACSSOO drive was used to
drive the four-pole, S HF', 230V three-phase induction motor
under test. The test motor was run at 40Hz,ZOHz, 10Hq and
SHz, respectively. Table 1 lists the locations of these
predicted frequency sidebands for the test motor when run, at
different drive input-frequenciesat no-load.
For each of these frequencies, the stator current was
acquired at each intensity level of unbalance and
misalignment, then processed by a variable frequency notch
filter developed by Hurst [9]in order to remove the dominant 1 2 3 4 5 6 7 8
drive ftequency and the higher order drive harmonics. Frequency (Hz)
Figures 1 and 2 depict the stator current spechum of the
Fig. 2. Sidebands of the input Pqumcy (SHz) aosociatcd 4 t
h
motor with rotor unbalance when driven at 5Hz. diffmnt severity levels of rotor unbalance a1 no-load

1455
In Fig. I , the range 0 -
ll.f, shows the frequency The computed rms values are then plotted against the
severity levels for each fault type. This is repeated for each
sidebands of 5Hz at 2.5Hz and 7.5Hz are evident. In Fig. 2,
the critical sidebands of the drive input frequency are shown of the considered drive input frequencies. Figures 4 through
for different severity levels of rotor unbalance when the 11 show the results for detecting load unbalances and radial
motor is operated at 5 H z and no load. shaft misalignments, respectively.
Load unbalances were created by adding a steel bolt and The results presented in Figures. 4 through 11 show that
nut at different radial distances from the rotor shaft on a the two different types of faults impose similar affects on the
balanced metal disk. At a given rotational speed of the frequency spectrum of the current. They both result in the
motor, the centrifugal force that acts on the shaft increases same induced sidebands, but with different amplitudes. The
with the increase of the radial distance between the bolt and magnitude of these harmonics is highly dependent on the
the center of the shaft. The exact force could be calculated machine installation and is affected by the complex
using mteractions of the mechanical and magnetic systems.
F = m , r . w2 , (2)

where F is in Newton, m is in kg, r is the radial distance in


; 280.00, I
meter, and o is the angular speed in radidsec. Placing the
bolt at farther holes on the disk creates a higher centrifugal
force, which causes a, more severe load unbalance. Three
severity levels of load unbalance were tested in this
experiment at each drive frequency.

Radial, or vertical, shaft misalignments were created by 0.00 I


installing additional shims of specific thickness under the Bda;ed Leal Lea2 W 3

base of the machine to lift it upward with respect to the shaft ~

Severityof Load lhbalince


of its coupled load. Increasing the thichess of the shim I
installed would increase the intensity of the shaft
misalignment. Three intensity levels of radial shaft
misalignment were tested in this experiment at each drive
frequency.
+Wth~oLoad i
i

-
-m- Wih Load
lV. ANALYSIS AND OBSERVATTONS I
For each experimental run, the filtered current signal is
analyzed as follows. First, the rms value of the critical inpul
-
Fig. 4. Ststor anrcnt fiesquency sidebands, versus levell of load
unbalance a1 & 40B2with no load and wiih load wnditiom.

frequency sidebands, which are induced by the load


unbalance or shaft misalignment is computed and recorded.
200 00
This is done for the well-balanced and well-aligned cases as
well as for each fault severity level. The steps are
summarized in the flowchart in Figure 3.
E

I' jE
Hunt
nolchbmd bosrd and 50.00
signal anti-aliasing camputn
-filter
I

Compute
i I
Bda;ed M l M Z Lea3

Severlty of Load lhbalance


FFT of I

I
i .+WthNoLoad ~

~ +WthLoad 1
Fig 5 Smlor c m 0 1 hequcncy ndcbands. rms vmus lcvcls of load
Fig. 3. Flowchart of the fault detection steps unbalance 81 t = ZOHz with no load and wlb load condinans

1456
It is also important to note that the baseline of the noise
levels do not change significantly with the presence of any of ao.00

these mechanical faults or any level of them Moreover, the


drive fkequencies do not affect the frequency sidebands
associated with these mechanical conditions, as they are
mainly multiples of the input frequency. This means that E 80.00

detecting load unbalances and shail misalignments in an


inverter-driven motor may be possible by utiling a simple 0E' 30.00

scheme such as the one presented in 171. The presence of a


mechanical fault could be detected without the need of
ono
Alignad MI W 2 m.
3
excessive filtering or using spectral analysis.
Severltyof Radial Mlsallgnment ~

'60.00

-
5
8090

Fig. 8. Stalorcvrrml fgequmcy sidebands, rms vcrrus levels ofradial shaft


0 40.00
misalignment a1 & = 4OHz with no load aad with load conditions.
000 I 1
B d d MI W Z b d 3

Severltyof Load &balance 1 00.00 , 1

1 -4-WthbbLOad
i1
1 +WlhLoad

Fig. 6. Stator current fcequmcy aidebands, llos v m levels of load


Allonad Wl W 2 W 3
unbalance a1 f. = 10- with no load and with load conditions.
SevertIy of Radial Mlsallgnment

,
+WfihbbLoad
-m-WfihLoad j

I
I
0.00 UFig. 9.misalignment
Stator current frqucncy sidebands, rmsv c m levels ofradial shaft
B d d Wl at f. 20th with no load and with load conditions.
W 2 W 3 -
Sverlty of Load Unbalance

V. CONCLUSIONS

1
i
-+-WfihNoLoad
t W ~ L o a d I This paper has examined the effects of running the induction
motor with an adjustable speed drive at different input
frequencies on the detection of load unbalance and shaft
misalignment. The magnitudes of the harmonics associated
Fig. 7. Stator cumnt fceqequcncy sidebands, rms V-E levels of load with these mechanical fault conditions were shown to be
unbalance at 2 = 5 t h with no load and with load conditionr. sufficiently large in the current spectrum, making the fault

I 457
detection possible by means of previously proposed simple REFERENCES
techniques. Results have shown that load unbalances and Williams, J.H., Davisr, A., and Draks, P.R, “Conditim-Bared
shaft misalignments in an induction motor driven by an Maintenance and M a c h e Diagaoslics, 1‘ ed. London, Chopman &
adjustable speed drive, could be detected using simple and H d
low-cost techniques that utilize a single phase of the stator Riley, C.M.; Habetla, T.G.; Obaid, RR, “A method for the sensorless
dete-ation of the vibration level in invmer-drivcn induction
current. The presence of these mechanical faults was motors”, Electric Machines and Drives, 1999. Intemnationol
detected even when the drive input frequency was ConferencelEMD‘99, pp.165 -167.
considerably low. Moreover, the mechanical fault detection Yacamini, R, Chaog, S.C., “Noise and vibration from induction
techniques employed in this paper could be easily machines fed from harmonic S O U ~ C ~ SIEEE
” Trmsaction on Energy
Conversion, ,Volume: 10 Issue: 2 , lune 1995. pp. 286 -292.
incorporated into the drive at little or no additional cost. T h o m n , W.T.; Barbour, A, ‘The on-line prediction of airgap
eccentricity levels in large (MW m g e ) ?-phase induction moton”,
Elecnie Machines and Drivm, 1999. hlemotionol Conference IEMD .
‘99,pp.383-385.
Schosn, R R, L i s B.K.,Habetla, T.G., Schlag, I. H.,and Farag, S.,
“‘An Unsupcrviped, &-Line S m m for Induction Motor Fault
.I ! Detection Uskg Stator I k m I Monitoring,” IEEE Trmacfiom on
60.00 Indway Appliealiom, vol. 31,no.6,pp. 1274-l279,Nov/Dec 1995.
Dorrsll, D.G.,Thomon, W.T., and Roacb S . ”Aoalyris ofairgap flus
current, and vibration signals BI a function of the combination of static
aod dynamic airgap ecmtricity in 3-phase induction mtd’
Trmwtionr on Indushy Applic.tions. B E E , Volume: 33 Issue: 1 ,
1997pp. 24 -34.
Obaid, RR, Habetla, T.G., and Gritter, D.1. “A Silified Technique
!
000 J for Detecting Mechanical Faults using Stator Curnnt in S d
Aligral -1 -2 M 3 Induction Motors”, Conference R-rd of 2000 IEEK 35” U S , VoLl
i pp.479-483.
Severity of Radial Mlsallgnment
.%’ i K(iman, G.B. and Stien, 1. “Induction Motor Fault Detection via
Passive Cumnt Monitoring”. Pmccedinp of the Intemorionol
Gnferrenuon Eiecnic Machiner,pp. 13-17. August, 1990.
H w l , LD. ‘ T d u c c r l e s a Control and Monitoring of Induction
+WihthLoad ’ Machines by Daktion ofMagnctic Saliency Hamonid’, PhD Thcsb,
ii Georgia lnstihae of Technology. Atlanta, GA,1996.
~

~ tWdhLoad

Fig 10 Stator currmi fmcqumcy rtdchands, nns verws levels of d a l shaft


mraltpmmt at I = 1OAz with no Inad and wth load m d ”

0.00 I
Aligral llrm I W 2 W s .
i
Severttyof Radlal Misalignment

-
Fig I I Slator c u m s frequcnc) ridehands. nns YCMS levels ofradml shaft
msallgnmml at t SH2 wth no load and wth load rondtlioos

1458

Você também pode gostar