Balancing, and Its Effects on Vibration Response

By Victor Wowk, P.E.

If I ever came across a perfectly balanced rotor, I would:

1. Think that I died and went to heaven, and

2. Not know it, because perfect balance is unmeasurable since it does not create any
vibration.

Unbalance is a defect on the rotating part that creates an unbalanced centrifugal force that
transmits into the stationary structure as an oscillating motion. We cannot directly measure the
unbalance on the rotor, nor the centrifugal force that it creates. What we can measure is oscillating
motion on the stationary structure.

Forces, especially the oscillating ones, are what does damage to mechanical parts. Since we can
only measure oscillating motion, we tend to make judgments about damage potential based on the
measured motion alone. The vibration motion measured at a point on the stationary structure depends
not only on the centrifugal force itself, but on the machine size and configuration, the bearings, the
stiffness of the support structure, and whether resonance is an active player. This article will discuss all
of these factors. The vibration analyst is not only a data gatherer and a messenger, but also a judge.
He/she must interpret the evidence, giving due consideration to all the factors that influence the
measured data, then assess the damage risk.

Forces and Motion

A pure unbalance creates a centrifugal force that is radial, and is well defined by the formula:

(1) F c = mr 2 F c is the peak force

m is the imaginary mass
r is the radius of the imaginary mass
ω is the rotating speed in radians per second

We use the concept of an imaginary mass for unbalance because we do not always know the
exact cause for each specific rotor. It could be caused by nonuniform material density (porosity), or
eccentricity from poor fits. When we place the correction weight, then the mass and radius of the
correction weight becomes 180 ° opposite from this imaginary mass.

Consistent units should be used in equation (1) with the appropriate conversion factors. In
English units, the formula is―

(2) rpm
2
F  Lb  = 1.77oz∗in 
f
1,000

This formula can also be presented graphically as in Figure 1.

 Figure 1

This centrifugal force from the imaginary heavy-spot is only one of many, since every small
piece of the rotor creates its own centrifugal force. The majority of these forces are countered by equal
masses and forces on the other side of the rotating center. These forces are constant and balanced.
They stretch the rotor outward and cause a circumferential tangential stress at the inner diameter. If the
rotor is accelerated in speed, the stress will grow until it causes cracking at the inner diameter that
propagates outward and the rotor flies apart. This is called the burst speed, and there are test facilities
that do just that in a containment structure that captures these exploding parts. Engineers use a healthy
safety factor of 4x to 10x and define that as the maximum safe speed. These constant, and balanced,
centrifugal forces are mostly undetectable (except for strain gages mounted on the rotor and transmitted
out), so we generally are unaware of them until the rotor separates. The unbalance centrifugal force
from the imaginary heavy spot, however, is another story (Figure 2).

The unbalanced centrifugal force has two effects:

1. It will distort the rotor.

2. It will couple into the stationary structure as an oscillation at the frequency of rotation.
 Figure 2

The unbalanced force is also a constant force on the rotating frame of reference. It will grow
once during spin-up to its maximum, remain constant during rotation at constant speed, then decline
during coast-down. But since it is unbalanced, it rotates around 360°. The bearing is the coupling
device between the rotating and stationary parts. The bearing transmits the rotating heavy spot force to
the stationary structure, where it becomes an oscillating force. The bearing gets beat up in the process.
The oscillating forces on the stationary structure causes an oscillating motion that depends on the
stiffness of the structure. Since it is motion, now we can measure something. The centrifugal forces
are created on the rotating part where the defect is, but the damaging forces are felt at the bearings and
the stationary structure.

We could measure the heavy spot force with a weighing scale, if the scale was fast enough in
response, or the rotor turned slow enough. The scale would display a sinusoidal alternating reading
riding on top of a constant weight. The vibration transducer functions like a fast electronic scale
reading of the oscillating force alone, being unresponsive to the constant force. The vibration
transducer does not actually measure force, but rather motion. With the measured motion, I could
estimate the force causing this motion if I know something about the mass of the rotor, or the stiffness
of the supports.

If I measure motion with a proximity probe, then I measure eccentricity of the shaft
centerline directly, assuming low runout or proper compensation for runout with a slow-roll check.
With this eccentricity measure and knowledge of the rotor weight, then I can calculate the unbalance
using the fundamental unbalance equation:

(3) U = Me U = Unbalance
M = Mass of rotor
e = Measured eccentricity
 The calculated unbalance can then be plugged into equation (1) or (2), along with a speed
measurement, to calculate the peak centrifugal force. Forces on machine elements are what causes
damage, not motion.

If I measured motion with an accelerometer, which is the most common motion sensor for
vibration today, then I can calculate the force from Newton's Second Law:

(4) F = ma F = force
m = mass
a = acceleration

I would need to estimate the mass in motion, which is no easy task for a support structure bolted
to a concrete foundation since it also has significant stiffness. For a spring-isolated machine, the
estimated mass in motion is all the mass above the springs. If desired, the unbalance itself can be
calculated by combining equations (1) and (4)

Ma
(5) U = U is the unbalance
2
M is the mass in motion
a is the acceleration
ω is the circular speed in radians per second

Fortunately, there is an easier way to estimate damage risk with acceleration measurements, that
will be presented later.

Machine Size and Configuration

A large and heavy rotor is going to be relatively insensitive to unbalance just by virtue of its
inertia. Eccentricity, e, is defined in ANSI S2.41 to be numerically identical to specific unbalance,
lb-in of unbalance per pound of rotor weight. This is an interesting philosophical concept. It says
explicitly than unbalance and manufacturing variability are inseparable, because eccentricity is
determined by manufacturing processes of drilling on center, turning round and concentric, and
assembly tolerances. For our discussion here, this equality states that a heavy rotor can accumulate
more heavy-spot mass before the rotor is pulled off the center of rotation a specified amount. This has
implications for the machine designer who must accommodate a certain amount of material buildup,
such as in a centrifuge, and not over vibrate. A heavier rotor would preform better in that application,
because it would be less sensitive to material buildup.

This also explains why we spend so much effort balancing fans. They are light, high-speed
rotors, many times mounted on soft supports. This makes them sensitive to small unbalances, which
create large motion.

The centrifugal force due to unbalance is supposed to be a purely radial force. It is mostly
radial on center suspended and symmetrical narrow rotors. Overhung rotors will couple this radial
force into the axial direction depending on the amount of overhang, the position of the center of gravity
of the rotating system, and the spacing between bearings. In other words, we can get significant axial
motion from unbalance depending on the machine configuration. Bearings do not tolerate axial forces
well. High axial motion at synchronous speed will always get my attention because it kills bearings.
Bearings

Unbalance does not pose serious consequences for rotors themselves, but rather the components
surrounding the rotor. Rolling-element bearings are the most chronic casualty. Rolling-element
bearings have small radial and axial clearance, and when this clearance is exhausted by the unbalance
eccentricity, then the rollers press through the hydraulic lift provided by grease or oil, and encounter a
hard limit. The local contact forces in the bearing translate into very high stresses because of the small
contact area, especially in ball bearings. The surfaces in contact suffer fatigue damage, and cracking,
or spalling, ensue. The rolling-element bearings, being rigid, also transmit these unbalance forces well
to the surrounding structure.

Plain bearings, on the other hand, tolerate unbalance better. The shaft journal has a clearance to
wiggle around inside the bearing and is also cushioned by a thicker oil film. Plain bearing machines
are smother and quieter for the same unbalance forces. The general rule for concern is that when the
shaft motion, as measured with a proximity probe, approaches ½ of the radial clearance, then there is a
risk of rubbing.

There is a very good design solution to obviate the damage to bearings from unbalance,
especially rolling element bearings. The solution is compliant bearings. The obvious example is wheel
bearings on vehicles. These are rigid, rolling-element bearings that are compliant mounted, with
springs on one side and rubber on the other. They are exposed to gross unbalances, in addition to
severe road shocks, and still survive to a normal old age.

Support Structure

The structure surrounding a machine is a major factor defining both vibratory motion and
damage. The two are related, sometimes in a reciprocal manner. Figure 3 will illustrate the concept.
There are two identical motors with the same internal unbalance. One is mounted on springs in (a) and
the other is hard bolted to a concrete floor in (b). The soft supported motor will have more motion but
the bearings will have smaller contact forces at the rollers. The hard bolted motor will have less
vibratory motion but high contact forces in the bearings. So here is a contradictory situation to
common wisdom where more motion is better, and less motions leads to early failure. This is why
vibration analysis remains a human function with a high level of judgment.
 Figure 3

Another aspect of the surrounding structure is that it both transmits and absorbs the vibratory
energy. Vibration energy is channeled along the structural paths. A rigid structure will pass through
any vibratory energy while absorbing very little. A flexible structure will not transmit as much, but
rather absorb it by converting it to kinetic energy of motion. The net result is that vibration energy
settles on the flexible components that it encounters as it races along the structural paths at the speed of
sound, which is 16,000 ft/sec in steel. It should be no surprise then that the handrail vibrates more than
the floor it is attached to.

The final aspect of the support structure is that it can become frequency creative. It can
generate harmonies in the vibration spectrum when the structure goes non-linear in elasticity. A
structure has two loads—a constant static load from gravity or other steady force, and a dynamic
oscillating load from vibration (Figure 4). When the oscillating load input is high due to a gross
unbalance, then the waveform can be distorted, or clipped, on the positive high side. This, we know,
generates multiples of the fundamental synchronous frequency, or harmonics. They are artifacts of the
signal processing instrumentation, but surprisingly, they are real enough to excite resonances. The
bottom line is that gross unbalance will generate harmonics.
Figure 4: Clipping of the oscillating load due to non-linear elasticity creates harmonics in the vibration
spectrum.
Resonance

The source of all machine vibration is some repetitive event within the cycle of operation,
unbalance force being an obvious contributer. All psychical components, including the shaft itself,
transform this force into motion, and the motion can be modified up or down. If it attenuates it, then
we call it damping. If it amplifies it, then it is resonance. The key factor on which way the structure
modifies is the natural frequency of the components in relation to the operating speed. Knowing the
natural frequency of at least the component that my vibration sensor is attached to, is the mark of a
good analyst. I would not attempt to judge unbalance, nor balance anything without some knowledge
of the natural frequencies. This is such an easy and quick test with a rubber hammer or fist blow, that it
amazes me why others don't do it more often. I would like to see all vibration analyst do it routinely,
but I am also just as happy charging time doing it myself.
 When the structure is actively amplifying the unbalance force, the resulting vibration will be
very large and the structure will suffer fatigue damage. But resonance is not always bad. We have
resonance machines in the form of watches, musical instruments, and sensors. Resonance on a panel
that makes a lot of noise may be irritating, but of little consequence. A machine on springs resonating
at the natural frequency of the springs can do this until the 23rd century and I don't care. This is like
your car shaking on its springs on a windy day. The key questions to ask are:

1. Where is the source of energy?

2. Where is the vibration energy dissipated?
3. What is the force transmission path?

This type of interpretation of the measured motion takes us to the highest level of vibration analysis.

On a flexible rotor, where the shaft resonates at its critical speed, the shaft suffers fatigue
damage because there is reversed bending of the shaft.

Damage Risk

All machines vibrate during operation as a normal consequence of being alive and doing work.
The question that everyone likes answered is how much is normal, and at what level do we call it bad.
There are industry guidelines on balance levels that are based on experience, and most of these are
valid and good criteria to start from. The basic premise of these guidelines are that there is a balance
level that is economically achievable and will not cause premature wear. So they are fair to both
sides—the machine builder and the customer. The machine builder only has control of the balance
level, the bearings, and the fits if the machine is fully assembled at the factory. The machine builder
has no control of the site conditions, which are the support structure. The support structure, we know,
can modify the motion measured from small unbalances. So what is a vibration analyst to do with
higher than normal amplitude at rotating speeds? Here are some suggestions:

1. Look at the overall pattern at each bearing in all three orthogonal directions. This can tell us
something about the flexibility of the structure, and whether resonance is active, because
resonance is strongly directional.

2. Stop the machine and measure the natural frequencies of the major components and structural
parts.

3. Don't get too excited if there is a large unbalance. Unbalance is not serious enough for
immediate repair. Machines have operated for many years with unbalance levels above industry
guidelines, even 2 to 5x the acceptable levels. What unbalance will do is beat-up the bearings
and transmit elsewhere to do collateral damage.

4. Do get excited if:

a) There is high axial motion.
b) The bearings are generating large shock pulses.
c) There are harmonics in the spectrum.
d) There is structural cracking.
e) Controls malfunction.
f) The operator complains of numbness in the hands.
 Generally, we would like the dynamic oscillating forces to be less than 10-percent of the rotor
weight. This is the old 10-percent rule, and there is even a formula to calculate the test weight for
balancing that will generate a centrifugal force of 10 percent. It is:

W
(6) Ur = 56,347 W = journal weight, Lb
rpm2
Ur = residual unbalance, oz·in

If I were measuring unbalance in acceleration units, then 0.1 g is very close to the 10-percent
rule. So I would like to see everything balanced to 0.1 g or less, but not zero. Zero unbalance is not
achievable. There will always be some residual unbalance. There is no such thing as perfect balance
on this side of the galaxy, just like there is no perfect woman nor perfect man.

Victor Wowk, P.E., is the president of Machine Dynamics, Inc., based in Albuquerque. He is the author
of Machinery Vibration: Balancing, published in 1995 by McGraw-Hill. He teaches a two-day
balancing seminar. Schedules are posted at www.machinedyn.com.