You are on page 1of 10

INFLUENCING FACTORS ON MOTOR VIBRATION & ROTOR CRITICAL

SPEED IN DESIGN, TEST AND FIELD APPLICATIONS


Copyright Material IEEE
Paper No. PCIC-(do not insert number)

Raj Mistry Bill Finley. Scott Kreitzer Ryan Queen


Member, IEEE Senior Member, IEEE Member, IEEE Member, IEEE
Siemens Industry, Inc. Siemens Industry, Inc. Siemens Industry, Inc. Siemens Industry, Inc.
4620 Forest Avenue 4620 Forest Avenue 4620 Forest Avenue 4620 Forest Avenue
Norwood, OH 45212 Norwood, OH 45212 Norwood, OH 45212 Norwood, OH 45212
USA USA USA USA
rajendra.mistry@siemens.com bill.finley@siemens.com scott.kreitzer@siemens.com ryan.queen@siemens.com

Abstract – This paper will present various case studies of how variation in performance can also happen in the field due to
the rotor or system natural frequencies can be strongly wear or environmental temperature changes. The reader will
influenced by its external and internal factors and how small learn how to avoid these types of situations, which can cost the
variations in these factors can influence the motor vibration at user millions of dollars in down time.
the manufacturer and in the field. Motors constructed to API The second example will demonstrate an actual situation
541 standards are required to have a rotordynamic lateral seen in the field and includes a motor driving a reciprocating
natural frequency that is removed from the operating speed by compressor powered from an ASD. This less common
at least 15%. The location of this natural frequency can depend circumstance introduces many new concerns that are not
on many factors such as bearing clearance, bearing type, always obvious. For example, the changes in speed make it
residual unbalance, oil temperature, oil viscosity, and bearing virtually impossible to avoid all the possible speeds at which a
housing stiffness. Depending on the design, some motors are torsional resonance could exist. The solution may be as simple
more sensitive to these parameters than others, and small as blocking out the speed ranges around the torsional
changes in these factors may cause large variances in the resonances. However, this potential solution can be
motor natural frequency. As a result, small variations in test complicated by the fact that there can be significant variations
setup, manufacturing tolerances, or field installations within in the torsional stiffness calculations of the rotating
critical components can cause noticeable differences between components. In some cases, this situation can force the
the calculated and measured natural frequencies. Variation in necessity of performing field tests to verify the exact location of
motor vibration may also be seen between the motor operating the resonances and then make field adjustments as necessary.
on the manufacturer’s test stand and the motor operating in the The third example will demonstrate a condition where there
field. In the field some apparently minor changes on ambient was considerable variation in ambient temperature, and the
conditions or set up can significantly change the motor motor and driven equipment did not grow thermally at the same
vibration. Additionally this paper will propose a worst case rate. In this example, the system integrator did not have a good
calculation method for motor natural frequencies that will understanding of the temperature change effects on all of the
provide greater confidence to the end user that the motor will components in the system. In this particular case, if the correct
operate successfully in the field before the motor is installed. coupling was used, this problem could have been avoided. This
paper will also demonstrate how the system performed with the
Index Terms — Induction Motor, Vibration, Lateral Critical wrong coupling and how the problem was fixed with the proper
Speed, Torsional Resonance, Coupling Misalignment. coupling. The basic question in solving this field problem was
this: how much coupling misalignment was acceptable and
I. INTRODUCTION why did this create vibration? This paper will discuss how to
predict the level of vibration with varying degrees of
This paper will review several real world case studies that misalignment. Although the solution to this problem seems
created considerable havoc for the end user until the root cause simple, it took considerable time and effort to get to the root
was discovered. These situations could have been avoided if cause. Unfortunately in some instances, by the time the
the basic knowledge demonstrated in this paper was known. problem is resolved, there may already be hundreds of similar
The paper will also demonstrate how even a minor variation in applications set up in the field.
motor dimensions or tolerances can have a significant effect on
the motor performance. This paper will also show the motor’s II. INFLUENCING FACTORS ON LATERAL
sensitivity through both calculations and models and CRITICAL SPEED
demonstrate the actual tested performance changes as a result
of these minor variations. A. Description of Influencing Factors
The first example will illustrate the influence of various motor
design and manufacturing parameters to a rotordynamic critical There have been many cases where minor changes in motor
speed. For example, a 0.001 inch (1 mil) change in bearing parameters can cause significant performance changes. For
diameter or a minor change in oil viscosity can move a motor example, small changes in bearing clearances can change the
rotor resonance much closer to the operating speed. This bearing stiffness enough to move the critical as much as 200 to

1
th
400 RPM. According to API 541 4 edition [1], the separation
margin (SM) between the critical speed and operating speed
should be greater than 15%. In this example, this seemingly
minor change in clearance can take a design which was above
the 15% separation margin (SM) to below the SM limits of API.
As only a few mils of clearance can influence rotordynamic
performance, it is easy to understand that this change in critical
speed could result from wear on the bearing journals over time,
which can be exacerbated by excessive stopping and starting.
In some cases, it is also possible to see significant changes in
bearing stiffness and critical speed as a result of oil film
viscosity. This will be demonstrated with actual test results and
with calculations utilizing finite element simulations. In addition,
it will be demonstrated how calibration of the model boundary
conditions is also necessary to achieve accurate simulations.
For 2 pole and 4 pole sleeve bearing machines above 5000
horsepower, achieving the appropriate 15% separation margin
can be challenging. In the absence of test data, the accuracy of
the critical speed calculation becomes extremely important. For
this reason, worst case parameters should be used in the
vibration analysis, and the motor should be designed to meet
the 15% separation based on the most conservative estimation
of critical speed.
Calculated critical speeds are typically depicted with a Bode Fig. 1 Bode Plot Example, Shaft Vibration and Phase
plot, which is the plot of shaft vibration in mils (peak-to-peak)
versus shaft rotational speed. The shaft vibration is typically 1. Bending Modes: The sensitivity of the critical speed to
plotted at the shaft location for each of the bearing proximity these parameters depends largely on the mode shape of the
probes. In addition, the phase versus rpm is also included. An natural frequency. All mode shapes have a combination of shaft
example of these plots is shown in Fig. 1. This case represents bending and shaft displacement at the bearing supports. Mode
a typical configuration of two proximeter probes per journal shapes with a large amount of shaft deflection and relatively
bearing mounted at 45 degree angles from the vertical. In this small amounts of shaft displacement at the bearing supports
example it is apparent that a rotordynamic resonance exists are called bending modes. An example of this type of mode is
around 2250 RPM. A point of inflection also occurs in the phase shown pictorially in Fig. 2. In this case, the shaft has a large
angle around this rotational speed which is also a strong deflection at the center and a relatively small amount of shaft
indicator of a critical speed at this location. These calculated displacement at the bearing. The degree of influence of each
plots can be directly compared to motor coast down data taken parameter on critical speed strongly depends on the ratio of
with the corresponding shaft probes. At this point the accuracy shaft bending to shaft displacement at the bearing. In general,
of the critical speed prediction can be determined. however, the critical speeds of bending modes are heavily
There are multiple influencing factors that affect the influenced by shaft and rotor stiffness, moderately influenced by
calculation of critical speed, some of which are affected by support & foundation stiffness, and only lightly affected by
manufacturing and some of which must be determined bearing properties.
experimentally. Most of these factors can be separated into one 2. Rigid Body Modes: In contrast to bending modes, rigid
of four categories: Bearing stiffness and damping, support body or cylindrical modes have a small amount of shaft bending
stiffness, rotor stiffness, and rotor weight. This list is by no and a relatively large shaft displacement at the bearings, as
means comprehensive but represents the ones that are most shown in Fig. 3. In general, rigid body modes are strongly
influential to the rotordynamic calculations: influenced by bearing and support stiffness and damping,
moderately influenced by rotor weight, and lightly influenced by
Bearing Stiffness and Damping shaft and rotor stiffness. The influencing factors and their
• Bearing clearance degree of influence on critical speed for these two modes are
• Bearing diameter summarized in Table 1.
• Oil viscosity
B. Effects of Bearing Stiffness and Damping on Critical Speed
Support Stiffness
• Stiffness of bearing housings and motor frame Figs. 4 and 5 show how differences in bearing clearances
• Foundation stiffness can affect the critical speeds. This particular example was
taken from a 4 pole induction motor with a shaft height of 630
Rotor Stiffness mm. The bearing was a 200 mm flange mounted plain
• Shaft stiffness cylindrical journal bearing. Many sleeve bearing manufacturers
• Rotor core stiffness offer several standard clearances ranging from large to small. In
• Bearing span from DE to NDE this example the standard clearances are shown in Fig. 4 using
the following notation: Auto +1 (large), Auto (middle), and
Rotor Weight

2
Auto -1 (small). A non-standard, tighter clearance is also listed 0.0160

DIA CLEARANCE (INCHES)


here as “Auto -2. “ 0.0150
The calculated critical speeds for the various bearing 0.0140
clearances are shown in Fig. 5. For a plain cylindrical bearing, 0.0130
the critical speed drops almost 200 RPM for each 2 mil 0.0120
increase in bearing clearance. It is worth noting that this bearing 0.0110
mode was mostly a rigid body mode which is strongly 0.0100
influenced by bearing stiffness. A true bending mode would 0.0090
show much less sensitivity to these bearing properties. 0.0080
Additionally, a four lobe bearing was analyzed, and the results 0.0070
in Fig. 5 also show that this bearing type is much less sensitive 0.0060
to bearing properties, even in a rigid body mode. "AUTO -2" AUTO -1 AUTO AUTO +1

Fig. 4 Diametric Clearances for a 200mm Plain Cylindrical


Journal Bearing [2]

3000

2500

2000

RPM
1500
AUTO+1
1000 AUTO
AUTO-1
500 AUTO-2
Relatively small shaft
movement at the bearing
0
Large shaft Plain Plain Four Lobe,
deflection at the rotor
Cylindrical, Cylindrical, 200mm
Fig. 2 Bending Mode 200mm 180mm Journal
Journal Journal

Fig. 5 Effects of Bearing Type, Bearing Journal, and Bearing


Clearance on Critical Speed

C. Effects of Rotor Stiffness and Weight on Critical Speed

The lateral stiffness of an induction rotor core can be difficult


to calculate due to its non-homogenous, complex geometry and
variety of different materials. Although core stiffness is small
compared to a similar sized ring made of solid steel, this
stiffness can influence the critical speed by as much as several
hundred RPM. The rotor core stiffness is influenced by the
Relatively large shaft amount of shrink fit to the shaft, the type of rotor construction,
movement at the bearing and the axial pressure on the laminated core. A process for
Large shaft measuring this stiffness is outlined in the example below. This
deflection at the rotor
particular rotor was part of the same 4 pole induction motor
mentioned previously.
Fig. 3 Rigid Body Mode
In order to ensure that the rotor model was accurate for the
case in question, two calibration tests were performed. The first
Bearing Support Rotor Rotor test was a rotor bump test with the rotor hanging vertically to
Stiffness Stiffness Stiffness Weight simulate a zero stiffness support. This test was performed with
Bending a set of six radial accelerometers spaced along the length of the
Light Moderate Heavy Moderate
Mode rotor and second set of six additional accelerometers placed 90
Rigid degrees from the first set. The rotor was then excited by striking
Body Heavy Heavy Light Moderate it with a calibrated force impact hammer. The vibration output
Mode was then normalized per unit of input through the transfer
function. The results showed the first mode at 142 - 143 Hz and
Table 1: Degree of Influence on Critical Speed for Various the second mode at 270 – 276 Hz. This simple arrangement
Mode Shapes was then modeled under a free-free boundary condition. The

3
rotor core was modeled as a cylinder and then the elastic
modulus was adjusted until the calculated mode matched the
experimental. In this case, the rotor core elastic modulus was
lowered to 5% of steel in order to match the measurements
above.
The second rotor calibration test was done with the rotor
horizontally on v-blocks in order to simulate a near infinite
support stiffness. The rotor was positioned with the center of
the journal in the center of the v-blocks. A bump test as
described above was then performed with thirty five
accelerometer positions along the length of the rotor. The
response showed a natural resonance at 45.5 Hz as illustrated
in the frequency response in Fig. 6. Each of the accelerometer
readings were broken down to their position on the shaft at 45.5
Hz in order to see the mode shape of the v-block test (Fig. 7).
Fig. 7 Experimental Mode Shape of Rotor on V-Blocks

The rotor model in Fig. 8 utilized a pinned boundary condition


at the center of the bearings and a rotor modulus of 5% of steel.
With these boundary conditions, the calculated first natural
frequency was 55.74 Hz. The two tests conducted and the
model simulations show that the rotor model is very accurate.

Fig. 6 Frequency Response of Rotor Bump Test on V-Blocks


Fig. 8 Rotor Model with Pinned Boundary Condition at the
Fig. 7 shows that the v-block support stiffness was not infinite Bearing Midpoint
as seen by the bearing centerline not being maintained on the
zero axis. A simple calculation was performed using the ratio of D. Effects of Support Stiffness and Foundation Rotor Stiffness
frequencies as a function of the ratio of compliance to provide on Critical Speed
the frequency if infinite stiffness were present. The natural
frequency at infinite stiffness was determined from the 1. Calibrations of the Structure: The calibration of the
measurement values as follows: supporting structure is also critical in obtaining an accurate
rotordynamic calculation. In an induction motor, the support
stiffness includes the bearing housing, end shield, and motor
frame. In this example, the support stiffness was measured by
bump testing the assembled motor in the horizontal, vertical,
fsys measured natural frequency of the rotor in v-blocks and axial directions on each end of the machine. Each test was
(hz) completed using 3 accelerometers to ensure accurate
fk=∞ calculated natural frequency of the rotor with infinite measurements. The impact force and vibration were captured
support stiffness (hz) for each bump test and the transfer function of the dynamic
stiffness was plotted. Fig. 9 shows one example of the bump
Applying these equations, the measured 45.5 Hz was test frequency plots, taken from the NDE horizontal probe. A
adjusted to an infinite stiffness at the bearings yielding the second order best fit polynomial was applied to the data up to
result of 56 Hz below. In this case the shaft displacement at the the natural frequency, which in this case was 111 Hz for the
bearing was 0.00081 inches and the maximum shaft NDE horizontal. Then the best fit curves for all 6 plots were
displacement at the middle of the core was 0.0024 inches. evaluated at the operating speed of 34.6 Hz (2076 RPM).
After the completion of the dynamic stiffness testing, the
model was updated to use the calibrated stiffness values
listed in Table 2. The calculated versus measured vibration
results are shown in Fig. 10.

4
Commercially available software was used for the finite
element calculations. In each of the cases the rabbet fit was
held as fixed and a vertical load of 17793N (approximately half
the rotor weight) was placed at the bearing centerline locations
(111mm from the end-shield front face on the flange mount and
20mm from the end-shield front face on the center mount).
Static deflections in the axial, horizontal and vertical directions
were calculated due to the resulting bending moment caused
by this vertical load. Figs. 11 and 12 show the exaggerated
deflection shapes and Table 3 shows the static deflections and
corresponding cross coupling stiffness values.

Fig. 9 Support Stiffness Bump Test Results, NDE Horizontal

DE Axial 2.592E+06 lb / in
Horizontal 11.898E+06 lb / in
Vertical 8.406E+06 lb / in
NDE Axial 2.679E+06 lb / in
Horizontal 11.490E+06 lb / in
Vertical 11.661E+06 lb / in
Fig. 11 End Shield Deflection, Flange Mounted Design
Table 2 Measured Dynamic Stiffness Values at 34.6 Hz

Fig. 12 End Shield Deflection, Flange Mounted Design with


Gussets (Left); Centrally Mounted Design (Right)
Fig. 10 Calibrated Model on Motor Manufacturers Test Base,
Calculated (Left) Versus Measured (Right) A simulation was also done with the center mount endshield
having a load located at the flange mount location. The results
2. The effects of bearing and end shield design on show that the geometry change from center to flange was not
support stiffness: A stiffness analysis was performed on the that significant. The 15 times increase in stiffness from center to
original end shield as well as a few additional variations of the flange was a result of the change of the load location. The
end shield design: deflection shape in all cases gave an “S” shape due to the
bending moment. From Fig. 11 it can be seen that the section
Design A) Original design - Flange mounted bearing with no from the OD of the endshield to the ID of the end shield is very
gussets linear. The deflections occurred more on each end of this
Design B) Design A with inboard gussets: 20mm thick, region near the OD and ID. As a result, the gusset design had
320mm height with a 150 x 250 mm chamfer very little impact on the original design and was not a
Design C) Design A with no gussets but 25mm thicker plate recommended method to increase overall stiffness in this case.
Design D) Center Mount with end shield thickness of 50mm Due to this deflection shape, it was very difficult to retrofit the
original flange mount design to affect the deflection at the OD
and ID locations.

5
Static Deflection (mm) be taken to avoid any system structural resonances that may
Design A Design B Design C Design D be lowered into the motor operating range.
Axial 0.04892 0.04605 0.01121 0.00312
Horizontal 0.00005 0.00004 0.00003 0.00002 E. The Effects of Flanged Mounted vs. Centrally Mounted
Vertical 0.00380 0.00360 0.00090 0.00034 Bearings on Critical Speed
Total Deflection 0.04907 0.04619 0.01125 0.00313
Static Stiffness (N/m) Bearing type, whether centrally mounted or flange mounted,
Design A Design B Design C Design D can also have a significant effect on bearing critical speed.
Axial 3.64E+08 3.86E+08 1.59E+09 5.71E+09 Centrally mounted bearings typically have a bearing midpoint
Horizontal 3.56E+11 4.45E+11 5.93E+11 8.90E+11 that is more inboard than a flange mounted bearing of the
Vertical 4.68E+09 4.94E+09 1.98E+10 5.23E+10 same size. A typical flange mounted and centrally mounted
bearing is shown in Fig. 14. The original design of the 4 pole
Total Deflection 3.63E+08 3.85E+08 1.58E+09 5.68E+09
Stiffness Multiple of Design A 1.06 4.36 15.66
example described above used a flange mounted style
bearing. In order to increase the rotordynamic critical, the
design was replaced with a centrally mounted bearing. The
Table 3 Static Deflections and Static Stiffnesses for the Four
End Shield Designs centrally mounted bearing reduced the overall bearing span
by 180mm and raised the critical speed from 2406 to 2571
3. End Shield Effect on Resonance: The end shield RPM, an increase of 165 RPM.
stiffness, which was calculated using finite element analysis,
was compared to the calibrated equivalent stiffness values
which were used to match test results. It was found that the
increase in end shield stiffness had little effect on the
resonance location. By calculation, the heavier end shield
increased the response by 25 RPM to an approximate value of
2431 RPM.
4. Test Results on Rubber: In order to simulate the
same motor on a flexible foundation, the motor was placed on
low stiffness hard rubber mats and a coast down test was
performed. The measured results are shown in Fig. 13. By
comparing this data with Fig. 10, it is observed that there is a
low frequency response at approximately 18.3 Hz on the
rubber mounts and that the critical appears at or above 2500
RPM.
Fig. 14 Flange Mounted Bearing (Left) Versus Centrally
Mounted Bearing (Right)

III. OTHER VARIATIONS THAT CAN CAUSE


VIBRATION

A. Motor on an ASD Driving a Reciprocating Compressor

Rotors in ASD applications are subject to significantly


different service conditions than when used on a fixed power
supply. There is a much greater risk of running on a lateral or
torsional resonance when the system can potentially run at any
speed (from zero to full speed) versus one fixed speed. On the
right hand side of Fig. 15, one can see the relative order of the
pulsating torque as a function of 1-12 times the rotational
speed. This is dependent on the number of torsional pulsations
per revolution the driven equipment produces. For example, a
4-piston or 4-lobe compressor will produce a pulsating torque of
Fig. 13 Motor on Rubber Mats 4 times the rotational speed. In this example there are two
fixed torsional resonant frequencies of the drive train. Both have
The stiffness and damping of the rubber mats is unknown, upper and lower speed ranges between which torsional
but the test was performed to confirm the de-coupling effect vibratory forces can be excessive if the forcing frequency
between the rotordynamic critical and the structure critical. In happens to be in this range.
this case, the rotordynamic critical stayed the same or even What is critical to the user is that the torsional stiffness and
increased slightly, which is counterintuitive. This data suggests resonance have a significant tolerance due to manufacturing
that the relative shaft vibration of the motor could be the same variations and may change over time. As a result, special
or better on a flexible base, such as a motor mounting plate precautions need to be taken. For example, it may be
placed on anti-vibration mounts (AVM’s). This improvement necessary to enlarge the unsafe speed avoidance range. To
may not always be seen in all motor designs, and care should help understand this we will explain the possible variations.

6
Key along one Arm
Range to avoid
for 7th order
torsional force Without key,
torque could be
distributed on
each arm

Fig. 16 Typical Spider Shaft

Fig. 15 System Response Spectrum of a Typical Motor

1) Spider Shaft Stiffness: Components such as the motor


shaft stiffness are greatly influenced by the shrink onto the shaft
and how each spider arm is bonded to the core. On a typical
spider shaft as shown in Fig. 16, the question is: do all spider
arms contribute to the shaft stiffness or just the spider arm with
the keyway? If the shrink fit is enough to fix all the bars to the
core, then the key would have little value but may still be
provided for safety. If all arms contribute to the stiffness, then
the torsional critical would be raised. If only one arm
contributes, (for example the one with the key) this would result Fig 17 Spider Shaft Deflection Modes: Bending (Top) and
in a much lower stiffness and resulting lower torsional critical Twisting (Bottom)
speed. Another consideration is whether or not the rotor core is
rigid along the length so that the force applied from the core to In conclusion, to avoid problems it is critical to understand all
the shaft is equal along the length of the mating surface. Or of the possible variations and tolerances, set a reasonable
would the laminated core slip between laminations causing an speed avoidance range around the resonances, and field test to
uneven distribution of force and a bending of the spider arms verify the actual resonant frequencies. It is also important to
along the length. See Fig. 17 for the different bending modes of monitor vibration over time, retest and then make adjustments
the spider bar. as necessary.
2) Bond between shaft and rotor: Another consideration
is whether the bond between the rotor and shaft changes over B. How Ambient Temperature Changes Can Affect Vibration
time, either becoming loose or maybe rusted together, thereby
increasing the bond. All of these possibilities must be This next example will illustrate how vibration can change
considered in the analysis to determine the possible effects on from acceptable levels to unacceptable levels as a result of
the range of the torsional resonances and range of speeds that completely normal ambient temperature changes. It is not
must be avoided. If a smaller avoidance range or block out unusual for motors and driven equipment to be used outdoors.
range is required, then a field test would be highly No matter what geographic location the product is located, it will
recommended. At the same time, however, it is not always experience ambient temperature swings. It is easier to
recommended to set the range so tight that minor changes understand that the motor shaft height will grow from cold to hot
(which can be affected by temperature and climate changes) do due to heat from inside the motor, and that the thermal growth
not later create a problem. Once set, it is important to monitor of the motor and driven equipment may not be the same.
the vibration and re-evaluate if vibration begins to increase over However, it may be more difficult to understand how ambient
time. temperature swings can change these results. The following
For example, in the diagram shown in Fig. 15, a 7th order example will demonstrate.
vibratory force would need to avoid running at speeds in the The motor shaft height on a 580 frame is 14.5 inches from
range between 610-650 RPM and 1020-1100 RPM. If a second the foundation to the center of the shaft (shown in Fig. 18). As
resonance unrelated to the torsional resonance exists at the the changes in shaft height are very small (0.001 inches of
same frequency, the magnitude of vibration could be greatly displacement), it is important to first understand that the shaft
increased. An example of this would be an air compression height has a significant tolerance. This tolerance is typically
chamber resonance (acoustical resonance). compensated during installation by placing shims under the

7
motor feet. This is unimportant for the purposes of this
discussion, but it is important to note that the relative shaft
height can easily be adjusted during installation using shims.
Once installed, the shaft height will be fixed at approximately
14.5 inches, plus or minus a small shift in height depending on
Gearbox Motor
temperature.

Fig. 19 Motor and Driven Equipment with Different Shaft


Heights

Fig. 18 Motor with 580 Frame Shaft Height

When the equipment gets hot, the shaft height will grow a
certain number of mils. A rule of thumb commonly used in the
industry for steel structures is: 1 mil per inch for every 100
degrees centigrade of temperature change. The stator winding
inside of the motor will normally have a temperature rise of 80 * “Soft foot” describes the condition where the four mounting feet are
to 105 degrees Celsius. On a totally enclosed fin cooled not all in the same plane. Measured in mils (1 mil = .001 inches)
machine, the bearing housings could rise as much as the stator ** To find angular misalignment in mils/inch of coupling diameter,
winding rise. However, in this example an open machine was measure widest opening in mils, then subtract narrowest opening in
used, and it was determined that the housing rise would be only mils, and divide by diameter of coupling in inches.
50% of the stator rise. For a class B rise (80 C), the housing Note: Up and down motion of driving and driven shafts with
would then be 40 degrees Celsius. Using the above rule of temperature may be in either direction.
thumb, the shaft height rise would be:
Fig. 20 Suggested Alignment Tolerances [3]
1 mil/in per 100 deg C x 14.5in x 40/100 = 5.8 mils
If a spacer is required after the product is installed, it may be
If the driven equipment grows to the same center line as the difficult or often impossible to move the motor away from the
motor, there is no problem. If they grow at different rates, driven equipment. However, special couplings are available
misalignment and excessive vibration could occur. The normal today that can be flexible in very short distances between shaft
variation between shaft centerlines can not exceed 1.0 - 2.5 ends. An example of this coupling will be discussed later in this
mils TIR (total indicator run-out) depending on the rotational example.
speed, unless a special coupling is used which can allow for Once the change from cold to hot for both the load and driven
this excessive offset. It is important to note that the offset of equipment is known, adjustments can be made to the cold
shaft centerlines is equal to ½ the TIR. alignment to compensate for the hot alignment. Keep in mind
If the driven equipment grows at a much different rate, that this compensation only applies to the vertical shaft height
significant vibration could occur. The driven equipment may change, and that either the motor or load shaft centers could
even be located on a different foundation and have a totally also move horizontally. This is common on gear boxes where
different shaft height, which can affect its sensitivity to thermal the distance between gears will tend to grow with increasing
change. See the example in Fig. 19. To calculate the vertical temperature. Torque applied to the shaft can also cause the
growth from cold to hot, one would need to know the thermal equipment centers to move apart. This movement is harder to
change, the growth in the driven equipment and foundations, measure and normally not done. However, it is common to
and any other mounting medium that could affect the total shaft measure the hot alignment at standstill to ensure the
height. If the coupling contains a spacer or spool piece calculations are accurate.
extension or if a flexible coupling is used (which many times The analysis so far has been fairly straight forward, but it has
also includes a spacer), the misalignment allowance can be not yet considered the effects of large ambient temperature
increased. Fig. 20 provides an estimation on the length of swings. In this example, the hot alignment was perfect under a
spacer required per inch of shaft offset. full-load condition, but the ambient changed a total of 50
degrees Celsius from -10 degrees C to + 40 degrees C. The

8
driven load was a water cooled gear box and the motor was air bearing was 1200 lbs based on a total rotor weight of 2400 lbs,
cooled. The cooling water temperature remained constant with it would lighten the load to a point where the vibration would
ambient temperature. And for the sake of calculation, it can be become excessive. The result of unloading the bearing and the
assumed that the gear box temperature also remained corresponding vibration was modeled, and the results of this
constant. At the same time, the air cooled motor was directly analysis are shown in Fig. 22. This phenomenon was also
affected by the ambient air temperature, and as a result, the verified in the field under test. The vibration started at less than
motor temperature rose 50 degrees C with a calculated offset 1 mil at 1800 RPM, increased to 2 mils vibration at 6 mils
as follows: misalignment, and then climbed to 3-4 mils vibration as the
misalignment approached 12 mils. The problem was totally
1 mil/in x 14.5in x 50/100 = 7.25. mils off-set eliminated by switching to a coupling shown in Fig. 23.

The installed coupling, which is fairly common and good for


many applications, is shown in Fig. 21. The coupling’s literature
also stated that it was capable of handling 10 mils of
misalignment. Although this coupling can take the force of
misalignment, it was not known how this force would affect the
performance of the drive train; therefore, the drive train had to
be evaluated.

Fig. 21 Coupling [4]

The force per mil of misalignment was measured to


understand the effect on vibration, and it was discovered to be Fig. 22 Vibration as a Result of Unloading the Bearing
relatively linear at approximately 100 lbs per mil of
misalignment. In this case, the 7.25 mils of misalignment
resulted in over 700 lbs of radial force on the motor drive end
bearing. When the force was in the same vertical direction as
the rotor weight (vertical down), the force started to approach
the bearing loading limit. The vibration on the drive end,
however, did not significantly change. The vibration on the non-
drive end increased slightly due to the bearing being off loaded,
but the vibration was still not excessive.
Now assume the misalignment was in the opposite direction.
In this case the hot alignment was perfect at the maximum
ambient temperature of 40 degree C and then the ambient
temperature dropped 50 deg C. Since the gear box temperature
did not change, the bearing was now being unloaded by 100 lbs
for every mil of misalignment. If the bearings were antifriction
bearings instead of sleeve bearings, the unloading would
eventually lead to a bearing loading below the minimum level
Fig. 23 Flexible Coupling [4]
required to minimize skidding. Skidding would eventually lead
to premature bearing failure. Sleeve bearings, however, require
a certain level of loading to remain stable and maintain the To confuse the issue even further, there were more than 50
stiffness required to minimize vibration. In an extreme case the of these systems in the field, and they all performed differently.
The quality of the alignment was inconsistent between units,
shaft could be lifted off its bearing and become totally
and the coupling unbalance was sometimes in phase with
dependent on the coupling and load bearing for any resistance
to movement. Since this set up would normally have minimal motor unbalance and sometimes out of phase, where it would
stiffness or resistance to movement, any unbalance of the actually decrease the overall vibration. At times, the magnetic
unbalance inside the motor would lift the rotor and off-load the
motor rotor or coupling would go virtually unchecked and could
bearing even further. Other applications would pull the rotor
vibrate out of control. There are other factors, such as magnetic
pull, that can also add or subtract from the force being applied down, loading the bearings even further. As a result, none of
to the bearing. In this case, where the rotor weight on each the sets were performing in the same way and only a few were
drastically different.

9
Fig. 23 shows a coupling that can handle significant V. REFERENCES
misalignment. Normally the flange for the disc pack would be at
the end of the shaft. In this case, however, there was no room [1] API 541, June 2004, American Petroleum Institute,
for a spool piece between the flanges which was needed for Washington D.C.: API.
flexibility. In the picture the design has a flexible disc pack on [2] RENK, Manual for the Application of RENK Sliding
the left and a flexible disc pack on the right which can be Bearings, RENK Corporation, Duncan, S.C. 2001.
located against the motor housing and the driven equipment [3] EASA, Mechanical Reference Handbook, Electrical
housing. Housings are not shown in Fig. 23. Since this Apparatus Association, Inc 1993.
application was already installed in the field and the shaft ends [4] Siemens, Flender Standard Couplings, Siemens AG,
were very close together, a special coupling was required. In Bocholt, Germany.
this case, the flanges were able to be turned away from the end
of the shaft and placed closer to the motor and the driven VI. VITA
equipment. The spool piece in the center was split in order to
separate the components for removal of the motor. This William R. Finley received his B.S. Degree in Electrical
provided all the flexibility that was needed. This coupling now Engineering from the University of Cincinnati, Cincinnati, OH.
applied less than 10 lbs per mil of offset as compared to 100 lbs He is presently Sr. Director of Technology for Siemens Industry,
per mil on the original coupling. Vibration remained unchanged Inc. Drive Technologies He has worked in various engineering
throughout all future misalignment conditions. Now all 50 design and management positions, including Electrical and
applications in the field performed virtually the same and could Mechanical Design, Product Development, Quotation,
meet the vibration requirement of 1.5 mils as defined in API 541 Computer Systems, and Business Unit and Operations
[1]. Management. He is a senior member of IEEE and has
In conclusion, these types of applications will have minimal previously published more than 25 technical papers. He is
problems when the radial loading on one bearing does not currently active in over 10 NEMA, IEEE and IEC working
exceed 10% of the rotor weight. In the original drive train with groups and sub-committees. He is Chairman of the
the original rigid coupling, 10% would have amounted to only International Standardization Group & Vice Chair of the
120 lbs or 1 mil offset maximum. Within that limit the vibration Technical committee of NEMA, Chair of CANENA THC 2 and is
increase and bearing life reduction would not be significant. Any the Chief Delegate to IEC TC2.
greater force would require the use of a flexible coupling to
guarantee a long life. Scott Kreitzer graduated with a BSME degree from Wright
State University in 1993 and received a Master of Science
IV. CONCLUSIONS degree in Aerospace Engineering from the University of
Cincinnati in 1995. Scott worked for Reuland Electric in 1994 as
As shown in these examples, even a small change in a Design Engineer developing high-speed AC induction motors.
manufacturing tolerances or dimensions can have a dramatic He is currently the Manager of Engineering in the Above NEMA
effect on performance. These effects can also be seen in the motor development group at Siemens Industry, Inc. Scott is an
field due to wear or a changing environment. When applying a associate member of IEEE.
product in any application, specifically in critical applications, it
is important to ensure that enough margin is allowed so that Rajendra Mistry, PE received his B.E. degree in Mechanical
these variations (which may not always be small) do not Engineering in India and a Bachelor of Technology in Electrical
adversely affect the performance of the machine. The 15% Engineering in the U.K. He is currently a Principal Engineer at
margin on the critical speed may be enough to compensate for Siemens Industry, Inc. (Norwood) in the engineering
wear, viscosity or thermal changes, but it may not be enough to development department responsible for developing Above
compensate for a base with varying flexibility, whether by NEMA induction motors. In addition to his industry role, he has
design or otherwise. In some cases it may be possible to attended several courses in vibrations, design for
intentionally design a weak motor foundation to drop a motor manufacturing, concurrent engineering, and digital signal
lateral resonance, which was once above operating speed on a processing. He is also responsible for certifying induction
massive foundation, to below operating speed. However, great motors for Hazardous location. He is a member of American
care should be taken in this case, and close coordination with Society of Mechanical Engineers (ASME) American Material
all manufacturers is recommended. In areas such as torsional Science International (ASM) and American Foundry Society
resonances, with many influencing factors in the drive train, it (AFS). He holds four patents for components in hydraulic
may be necessary to allow for additional tolerances resulting elevators and on Induction motors.
from manufacturing variations. Regardless, when assembling a
system with multiple components, consider utilizing an expert Ryan Queen received his B.S. degree in Mechanical
with experience in designing the integration of multiple systems. Engineering in 2006 from The Ohio State University,
Columbus, Ohio. He is currently a Sr. Product Engineer in the
Advanced Technologies group of Siemens Industry, Inc. He is
responsible for advanced mechanical calculation, design, and
simulation of AC induction motors and components. He also
participates in Siemens Network of Competency for both
rotordynamics and noise and machine vibration.

10

You might also like