APPLICATIONS

The color of runout

17-4 PH stainless steel as seen under polarizing light at
50X magnification. The colors show martensite matrix
formations and delta-ferrite with copper precipitates –
all of which contribute to the material’s tendency
toward high levels of electrical runout.

4 O R B I T [ Vo l . 2 5 N o. 3 2 0 0 5 ]
 APPLICATIONS

Understanding and
Mitigating Shaft Runout
Introduction
This article explains what runout is, why it is guarantee that runout can be mitigated or pre-
important, and the root causes of runout in vented in every application, it can be managed
machinery shafts. It also outlines common meth- effectively and kept to within allowable levels in
ods for reducing runout to allowable levels and the vast majority of applications. This is evi-
suggests best practices to observe during fabri- denced by the millions of successful proximity
cation and machining to help avoid runout probe applications for turbomachinery around
difficulties in the first place. While there is no the world over the past forty years.

Nathan Littrell
Senior Engineer, Bently Nevada™ Asset Condition Monitoring
GE Energy
nate.littrell@ge.com

[ Vo l . 2 5 N o. 3 2 0 0 5 ] O R B I T 5
APPLICATIONS

(elliptically
exagerated
for clarity)

Probe Output
Max
0˚ 0˚

270˚ 90˚ 270˚ 90˚ Min

180˚ 90 180 270 360

180˚
Degrees of Shaft Rotation

a) Out-of-round shaft with b) Perfectly round shaft with

perfectly uniform electrical non-uniform electrical
properties. properties.

Figure 1 – “Apparent” probe gap for a) an out-of-round shaft with uniform electrical properties and b) a perfectly
round shaft with non-uniform electrical properties.

What is Runout?
The signal from an eddy current proximity probe is a Notice also that these signals have nothing to do with
function of the gap between the probe tip and the target the dynamic motion or vibration of the shaft. They are
material. However, it is also a function of the electrical inherent properties of the shaft that will be observed
conductivity and magnetic permeability of the target regardless of whether it is stationary or rotating at high
material. Thus, two different materials (for example, speed. These signals are known as runout. For conven-
4140 type steel and aluminum) can be positioned with ience, we divide runout into two primary categories as
the same physical gap from a proximity probe, yet will follows:
give different outputs due to their dissimilar electrical
Mechanical Runout is a measure of the shaft’s
properties.
deviation from a perfectly uniform radius as its
For a rotating shaft, physical out-of-roundness results in circumference is traversed. This type of runout
a change in gap. This is shown in Figure 1a. However, a can be measured by a dial indicator.
perfectly round shaft with non-uniform electro-mag-
Electrical Runout is a measure of a shaft’s electrical
netic properties will also result in a change in probe
property variations as its circumference is traversed.
output, even though the physical gap is uniform. This is
This type of runout cannot be measured by a dial
shown in Figure 1b. In this example, both shafts give
indicator.
identical probe outputs even though they have different
physical shapes. In practice, mechanical runout can Because a proximity probe senses both types of runout,
indeed be somewhat sinusoidal as shown in the exam- it is customary to speak of Total Indicated Runout (TIR)
ple. However, electrical runout is rarely – if ever – which is simply the sum of mechanical runout and elec-
sinusoidal and is generally characterized by a noisy trical runout. In most cases, when runout is discussed in
waveform with numerous spikes. It is represented as a conjunction with proximity probes, it is understood to
sinusoid in this example merely for illustrative purposes. mean TIR.

6 O R B I T [ Vo l . 2 5 N o. 3 2 0 0 5 ]
 APPLICATIONS

BECAUSE THE RUNOUT SIGNAL IS NOT RELATED TO ACTUAL SHAFT

VIBRATION, IT CAN LEAD TO ERRONEOUS VIBRATION READINGS
AND MACHINERY DIAGNOSTIC CONCLUSIONS.

Why Be Concerned About Runout? Sources of Mechanical Runout

Even if a target material has non-uniform physical or • Machining processes
electrical properties, it does not create a problem for — So-called “lobing” of the shaft (see Figures 2 and 3).
probes observing the same location on the target at all This is particularly problematic when centerless
times – such as an axial position measurement using the grinding machines are used because variations in
end of a shaft. However, proximity probes are often used shaft hardness can result in a non-circular geome-
for radial vibration measurements where the “track” try. Grinding on centers provides a reference for the
observed by the probe is constantly changing (repeating wheel to work against and is less prone to runout.
itself every 360 degrees) as the shaft rotates. This results
— Tool chatter. Selection of the correct tool and holder,
in a proximity probe signal composed of both actual
as well as adjustment, is critical for all machining
vibration and runout. Because the runout signal is not
processes. Make certain tools are not dull.
related to actual shaft vibration, it can lead to erroneous
vibration readings and machinery diagnostic conclu- — Improper feed rate and speed of cutting tools.
sions. To avoid this problem, the amount of TIR must be Surface finish is strongly affected by cutting tool
kept to allowable levels, generally 25% or less of feeds and speeds.
expected vibration amplitudes. • Dents from handling

Many customers specify the amount of allowable runout • Rust patches

for new or refurbished rotors as part of their purchasing • Rotor bow due to thermal effects, gravity, or other
documentation to their vendors. American Petroleum influences/loads
Institute (API) Standard 612 is one such frequently cited
• Defective or worn bearings in the machine or lathe
specification. It pertains specifically to mechanical drive
supports
steam turbines and requires the TIR to be 0.25 mil pp or
25% of allowable vibration, whichever is greater. API 617 Sources of Electrical Runout
has identical runout requirements and pertains to
process centrifugal and axial compressors as well as • Metallurgy
turbo-expanders. Similar API standards exist for other The material chemical composition is fundamental to its
machine types. electrical and magnetic properties. As well, the material’s
purity can affect runout. In general, non-ferrous materi-
Failure to meet runout specifications can cause expen-
als such as copper and aluminum exhibit the fewest
sive delays and re-work, impacting both the customer
electrical runout problems, since they are devoid of any
and their machinery supplier. For this reason, discovering
significant magnetic effects.
and correcting electrical runout issues early in the manu-
facturing process can save a great deal of cost. It is Conversely, the worst materials in terms of electrical
much easier to treat the problem when the shaft is on runout are precipitation-hardened steels. Precipitation
blocks or mounted on a lathe than when installed in hardening is a process where clumps of different crystal
the machine. states are formed in the matrix of the parent metal. The

[ Vo l . 2 5 N o. 3 2 0 0 5 ] O R B I T 7
APPLICATIONS

probe observes these clumps as they pass by while the • Heat Treatment
shaft rotates, producing the unwanted runout signal.
The purpose of heat treatment is to modify the crystal
17-4 PH can be particularly troublesome in this regard
structure of a material to tailor the material mechanical
(see photo on page 4).
properties (toughness, ductility, etc.) to the application.
™ ®
Bently Nevada Proximitor sensors are calibrated to AISI The magnetic properties of ferrous materials are a
4140 type steel. However, this material is available in sev- function of the crystal structure, so it follows that heat
eral grades, and variations in probe system response will treatment is a factor in the resulting electrical uniformity
vary among these grades. In general, the vacuum arc of the shaft.
remelt (VAR) or double vacuum arc remelt (DVAR) materi-
Many large shafts are quenched as part of the heat
als possess the best homogeneity and exhibit the fewest
treatment process by lowering into tanks of salt water
number of problems with electrical runout.
or other liquids. Most commonly, the shaft is horizontal
It is recognized that the choice of shaft materials is rarely when lowered into the quench tank, which results in an
as simple as merely considering the material’s runout asymmetrical quench profile. It is recommended to lower
properties. Instead, designers are faced with multiple cri- the shaft vertically into the quench tank if possible to
teria and inevitable tradeoffs. Pumps are a good example improve the radial homogeneity of the quench.
of machines that must often employ more exotic materi-
• Grinding
als due to the corrosive nature of the process fluid that
Grinding the bearing journals to final dimensions and fin-
will be handled, whether seawater, liquefied sulfur, acids,
ish is generally the last step in the shaft manufacturing
or others. Motors are another machine type that com-
process. The grinding process generates significant heat
monly use materials other than 4140 type steels. As will
that is localized at the point where the grinding wheel
be discussed later, when a designer requires certain shaft
touches the shaft. It is important to have maximum
metallurgies, yet the material exhibits intractable runout
coolant flow on the work piece and to start and stop the
characteristics, one approach is to attach a collar or
grinding process slowly. Avoid sudden increases or
coating of a different material to the shaft.
decreases in feed rate when grinding. It is also recom-
• Forging mended to keep the wheel freshly dressed to limit heat
The forging process involves forming an ingot into the build up. Because grinding results are highly dependent
rough shape of the shaft using enormous hydraulic on the operator, it is recommended that this step be
hammers and presses. During the forging process, the closely monitored if runout problems are occurring.
material flows into the shape of the shaft and gains a
• Magnetism
grain structure that is present throughout the cross
Residual magnetic fields in the shaft can cause signifi-
section of the material. This grain structure defines a set
cant variation in the output of the proximity probe
of large scale boundaries that contain the smaller scale
system. Degaussing (discussed later in this article) is the
crystal boundaries. A non-uniform grain structure can
recommended remedy.
result in electrical runout.

8 O R B I T [ Vo l . 2 5 N o. 3 2 0 0 5 ]
 APPLICATIONS

• Stress Effects Measuring Mechanical Runout

Stress affects the crystal structure and magnetic proper- The first step in dealing with runout is to make an accu-
ties of materials. Occasionally, a shaft with runout rate measurement of the physical profile of the shaft to
problems can be traced back to an event that caused determine the mechanical runout. Once the mechanical
the probe area to undergo significant mechanical stress. profile has been determined, electrical runout can then
It is best practice to support shafts in slings in such a be inferred – generally by simply subtracting the
way that they are not subject to significant bending mechanical runout from the TIR measurement made
stresses during installation and handling. Bead blasting with a proximity probe.
or other impact-based cleaning processes create com-
When assessing mechanical runout, accuracy is para-
pressive stress in the surface of the shaft and can induce
mount. Special care must be taken due to the extremely
runout.
small dimensions being measured. This requires a meas-
• Handling uring instrument capable of resolving increments finer
In addition to the stress effects mentioned above, it is than 0.1mil (.0025 mm). While there are several choices
possible to ‘bruise’ metal by hitting or dropping the shaft for such instruments, some are more practical, accurate,
on the probe tracks during intermediate steps of manu- and convenient than others.
facture. The external damage is erased by subsequent • LVDTs (Linear Variable Differential Transformers)
steps such as machining, but the damage to the crystal
LVDTs operate on the principle of a transformer with
structure may go quite deep into the material. Thus, it is
a movable core. As the core moves, the gain of the
important to handle the shaft carefully at all steps in the
transformer changes and the displacement is
process.
inferred from that signal. Units are available with a
• Plating resolution in the 0.01 mil (.0003 mm) range and are

Occasionally, a rotor is refurbished by plating the bear- particularly well-suited for highly accurate mechan-

ing journal area to replace worn material. Sometimes, ical runout determination.

this surface will be intentionally “roughed up” to allow • Dial indicators

the chrome plating to adhere better. However, the probe
While inexpensive, reliable, and found in most every
will “see through” the plating to the rough surface under-
machinist’s tool box, mechanical dial indicators are
neath, resulting in runout. Also, chrome plating has very
generally limited to increments no finer than a tenth
different electrical properties than typical shaft materials
of a mil. Analog versions use a conventional needle-
and strongly affects proximity probe output. In general,
type indicator that can be very difficult to read with
plating in the area of proximity probes is not recom-
the required resolution, and for this reason are
mended. However, when this is not an option, plating
unsuitable for runout measurements. In addition,
thickness should be at least 20 mils to prevent the “see
they do not allow for automated data acquisition.
through” effect mentioned above, and the Proximitor®
sensor should be calibrated to the plating material rather Dial indicators with a digital display are also avail-
than the substrate shaft metallurgy. able, and are generally capable of providing the

[ Vo l . 2 5 N o. 3 2 0 0 5 ] O R B I T 9
APPLICATIONS

necessary resolution. In addition, some of the more runout once it is already present. These methods will be
advanced versions feature an electrical output in discussed later in this article. However, before attempting
addition to the display, making them suitable for to reduce the amount of electrical runout, it is generally
automatic data acquisition. recommended that mechanical runout be addressed
first, to try and bring the TIR within tolerances. This
• Form measuring equipment
These are specialized devices that evaluate compo-
nents in terms of geometric dimensioning and
tolerancing definitions. The machine typically holds
the component vertically on a turntable and meas-
ures the form of the surface using a stylus. Output is
the radial deviation from absolute roundness along
with values describing the concentricity, eccentric-
ity, and roundness as defined by Geometrical
Dimensioning and Tolerancing (GDT) standards.
Unfortunately, such equipment is of only academic
interest for most rotating machinery because it can-
not handle components larger than 60 kg.

Based on the foregoing discussion, only two practical

choices exist for measuring mechanical runout in most
instances: LVDTs and electronic dial indicators.

Measuring Electrical Runout

Since electrical and mechanical runout are remedied in
different ways, it is important to obtain separate profiles
of the mechanical and electrical runout. Often, it is
easiest to measure TIR and mechanical runout simulta-
neously. The electrical runout is then found by simply
subtracting the mechanical runout from the TIR, as previ-
ously mentioned. Measurements should be made at
suitably small intervals (typically every 10-20 degrees)
to provide sufficient detail in the profiles.
Excerpt from a typical runout report for both coupling
While we have already discussed several practices that
and non-coupling ends of a pump rotor. The top trace of
can help prevent electrical runout at the manufacturing each plot shows the runout waveform while the bottom
stage, there are also methods (such as diamond burnish- trace shows the Keyphasor® pulse, indicating one
ing and degaussing) that can be used to reduce electrical complete shaft revolution between pulses.

1 0 O R B I T [ Vo l . 2 5 N o. 3 2 0 0 5 ]
 APPLICATIONS

serves two purposes. First, the processes of grinding and • Lobing Effects
machining to further reduce mechanical runout can As mentioned earlier, lobing is a common artifact of
themselves introduce additional electrical runout. Thus, centerless grinding operations. When measuring the
there is little point in proceeding to address excessive shaft profile in a lathe or other device where the shaft
electrical runout until mechanical runout has been is rotated about its axial centerline, the user is mea-
addressed. Second, because it is typically more difficult suring radial (rather than diametral) variations. As a
to address electrical runout, mitigation is generally only result, there is no ambiguity in the profile measure-
appropriate when mechanical runout reduction alone ment. In contrast, v-blocks cause the user to measure
cannot bring the TIR to within the specified limits. diametral variation, and can result in ambiguity
regarding the shaft profile. This is easiest to visualize
Sources of Error and Non-Repeatability by way of examples, as shown in Figures 2 and 3.
API 687 (Repair of Special Purpose Rotors) provides a Notice that the dial indicator in Figure 2 gives
very detailed description of how to measure runout. API exactly the same output shape (dark blue line) for
specifications, in general, require that: both shafts and that it reflects the change in diame-
1. The shaft be supported in v-blocks; ter (not radius) as the shafts are rotated. The user
may incorrectly conclude that the one-lobed shaft
2. The probe be perpendicular to one face of the v-block;
had a two-lobed profile, and efforts to correct this
3. Runout be measured in terms of peak-to-peak probe through grinding would only exacerbate the prob-
output. lem. In Figure 3, notice that the three-lobed shaft
One of the primary reasons that v-blocks are recom- provides a dial indicator output suggesting perfect
mended is that the runout measurement should be made roundness, when, in fact, it has three lobes. Only by
in apparatus separate from that in which the machining examining the motion of the center of the inscribed
was actually performed. For example, if a lathe has bear- circle for all three shafts does the user obtain the
ing wear that produces an elliptical shaft cross-section, true profile. This ambiguity can be removed by mak-
the shaft will appear perfectly round as long as it is in that ing the mechanical runout measurement with a
particular lathe. By moving the shaft to a separate meas- fixture that rotates the shaft about its centerline
uring environment (i.e., v-blocks or a balancing machine), (such as a balancing machine). As previously
the error introduced by the lathe will not be masked. mentioned, it is not recommended that the meas-
urement be made on the same lathe in which the
However, while the use of v-blocks represents recognized shaft is being machined, as the runout measured
good practice, it is not immune from its own sources of becomes the combined effect of the shaft and the
errors as detailed below. lathe bearings, and the two can offset one another.
• Failure to mount the probe perpendicular to one Oil wedge – It is known that a film of oil builds up
face of the v-block between two surfaces moving relative to each other.
This is a common error made in the field and results This oil film becomes part of the runout measure-
in incorrect mechanical runout readings. It affects ment and is unpredictable. A moving measurement
only the mechanical runout measurement (not the may exhibit a dependency on shaft speed, even at
electrical). The maximum mechanical runout error slow roll. Thus, when documenting runout under
introduced is the sine of the probe’s angular devia- slow roll conditions, it is important to record the
tion from block face perpendicularity. actual shaft rotative speed.

[ Vo l . 2 5 N o. 3 2 0 0 5 ] O R B I T 1 1
APPLICATIONS

Figure 2 – When mounted in v-blocks and measured with a dial indicator, these one- and two-lobe shafts give identical
dial indicator profiles. The only way to ascertain the true mechanical profile is to make the dial indicator measurements
using apparatus that keeps the shaft fixed about its centerline – such as a lathe. This allows radial, rather than diametral,
variation to be observed.

Figure 3 – This 3-lobed shaft appears to be perfectly round when mounted in v-blocks and measured with a dial indicator.

1 2 O R B I T [ Vo l . 2 5 N o. 3 2 0 0 5 ]
 APPLICATIONS

WHEN ASSESSING MECHANICAL RUNOUT,

ACCURACY IS PARAMOUNT.

Stick slip – V-block measurements will sometimes Methods of Mitigating Electrical Runout
use an apparatus (such as a drive belt) that slowly
The old adage “an ounce or prevention is worth a pound
rotates the shaft. However, manual barring of the
of cure” is particularly true for electrical runout. In the
shaft is most common. While API specifications
section “Sources of Electrical Runout,” we noted a variety
recommend rotation intervals of no more than 20
of things that can lead to electrical runout and offered
degrees, this is a relatively large gap between data
advice on how to carefully choose and handle materials
points. As the rotor is moved, it may not settle into
to minimize the potential for electrical runout. However, if
position repeatably, leading to significant error. To
excessive levels of electrical runout exist, there are steps
help counteract this “stick slip” effect, smaller
that can be taken to reduce them.
measurement intervals (10 degrees or less) are
recommended.
Degaussing (Demagnetizing)
Bow/sag – If a shaft has a bow or sag from gravity
One method of checking residual magnetic field strength
(and all shafts exhibit some level of this), it is possible
is by using a small, hand-held field indicator, available in
that this will show up as runout as the shaft flexes
digital and analog versions from manufacturers such as
during rotation. Anisotropic stiffness (unequal with
Magnaflux®. Even a relatively small amount of localized
respect to direction) will definitely cause irregularity
residual magnetism can contribute to runout. For exam-
in the runout reading due to rotor sag. To minimize
ple, a localized concentration of 5 gauss on a rotating
this effect, make the runout measurements as close
shaft can give electrical runout on the order of 0.5 mil.
as possible to the shaft supports (e.g., v-blocks).
Therefore, it is always good practice to check the shaft
Inconsistent Transducer Models – When measur- with a field indicator and, if required, degauss in the area
ing TIR, it is not necessary to use the same probe in of the probe tracks.
the shop as the installed probes in the field, as this
A degausser emits an AC pulse of decreasing strength.
is rarely practical. However, it is strongly recom-
The magnetic field generated “scrambles” the domains
mended that the same probe series be used to
in the material to reduce the residual magnetism. While
eliminate possible sources of inconsistencies. For
special degaussing apparatus is available, a very com-
example, if the machine will be permanently moni-
mon field practice is to use an arc welder set to AC with
tored with Bently Nevada 3300 XL 8mm proximity
the cables shorted together. The cables are waved over
probes, it is advisable to use this type of transducer
the area to be degaussed, or sometimes wrapped
system for the bench runout measurements as well.
around the shaft. The current in the cables sets up a
While the differences between transducer series are
large enough magnetic field to effectively degauss the
generally small, runout measurements are typically
shaft.
trying to resolve dimensions of 0.25 mils or less. As
such, even the smallest sources of variation can
affect the results.

[ Vo l . 2 5 N o. 3 2 0 0 5 ] O R B I T 1 3
APPLICATIONS

are strongly encouraged to enlist the assistance of a

qualified GE services professional when burnishing is
required. In addition to performing the burnishing work,
these individuals can provide the necessary hands-on
training for those customers that prefer in-house
competencies in the use of burnishing tools.

Alternate probe track material

In some instances, electrical runout can prove quite
intractable. In such instances, the most expedient solu-
A diamond-tipped burnishing tool. The tool is
tion is generally to use an alternate target material for
mounted in a lathe and can provide surface finishes
of 10 microinches or less. When used by a skilled
the probe to observe. The two most common methods
practitioner, diamond-tip burnishing can be effective for this are collars and coatings.
in reducing electrical runout because it alters the
• Collars
shaft’s crystal structure.
Collars can be very effective, provided they are
attached to the shaft in such a way that they can-
Probe Gap
not come loose or induce additional loads or
A simple first attempt at reducing electrical runout is to stresses on the machine. Additionally, the collar
gap the probes closer to the shaft. Sometimes this can must be ground after it is shrunk onto the shaft to
change the runout measurement. However, be certain ensure that it is suitably concentric. However, some
that the probes are not gapped so close that they take shaft geometries cannot accommodate a collar. In
the probes outside their linear region or allow the probes other situations, the shaft geometry may allow for
to contact the shaft during periods of high vibration. a collar, but significant thermal gradients due to
differential expansion problems may make use of
Burnishing a collar unwise.
Burnishing is a technique of smoothing the surface of the • Coatings
shaft using a rounded diamond tip mounted on a lathe.
Depositing a layer of less runout-prone material
The burnisher tip is pushed against the shaft surface by
onto a shaft can be employed successfully, and
a spring loaded tool holder. This process mechanically
there are several technologies for this. The idea is
alters the crystal structure on the surface of the shaft by
similar to that of plating (already discussed) and
plastic deformation, allowing surface finishes of less than
many of the same considerations apply. Primary
10 microinches to be realized. While burnishing can be
concerns are to choose a material that is non-
an effective method of reducing electrical runout, it is
ferrous and applies with sufficient density that
more of an art than a science. Unless applied by a skilled
inclusions do not generate a runout signal of their
practitioner, burnishing can actually worsen electrical
own. The material must also be applied in a thick
runout. Further, if some burnishing is good, more is not
enough layer to prevent the probe from seeing
necessarily better. Once burnishing has minimized the
through to the substrate.
electrical runout, additional burnishing may increase
rather than decrease the runout. For these reasons, we Obviously, both approaches require a proximity
have deliberately chosen not to include step-by-step probe system calibrated to the target material, not
burnishing instructions in this article. Instead, customers the shaft material.

1 4 O R B I T [ Vo l . 2 5 N o. 3 2 0 0 5 ]
 APPLICATIONS

A Word About Compensation Occasionally, users will request that we provide compen-
sation features in our permanent monitoring hardware.
When performing machinery diagnostics, a common
Both vector and waveform compensation are valuable
practice is to subtract a known runout signal from the
features when performing machinery diagnostics, and
overall vibration waveform to obtain a “runout-free”
the runout signal can generally be validated and
waveform. This is known as compensation and is a way
of dealing with both mechanical and electrical runout. updated as needed as part of the diagnostic process.
®
Many diagnostic products (such as the ADRE System This is not the case for permanent monitoring and we
and System 1® software) allow such compensation. In strongly advise against the use of compensation for
addition to waveform compensation for unfiltered plots machinery protection applications. Runout signals can
such as timebases and orbits, the signal can also be fil- change over time due to factors such as surface
tered to a specific frequency, such as shaft rotative scratches incurred during operation or maintenance,
speed (1X). This allows it to be characterized as a vector and/or changes in the amount and distribution of shaft
and used to compensate filtered plots (such as Bodé magnetism. When compensation is embedded in a per-
and Polar – see Figures 4 and 5). manent monitor, the runout profile stored in the monitor

Figure 4 – ADRE ®
Sxp software is one example of a diagnostic system that provides runout compensation features as
shown by the compensated (green) and uncompensated (red) data in this polar plot.The red arrow denotes the 1X runout
vector. For a polar plot, compensation has the effect of shifting the data to the origin at slow-roll speeds.

[ Vo l . 2 5 N o. 3 2 0 0 5 ] O R B I T 1 5
APPLICATIONS

remains fixed even though the actual runout may be or they may mask legitimate high vibration problems,
changing over time. phase changes, or other conditions indicative of an
emerging machinery malfunction.
The nature of combining waveforms and vectors is that
they reinforce in some places and counteract in others – Back when Bently Nevada™ monitoring hardware utilized
unlike simple scalar addition. Figure 5 illustrates this con- analog meter movements, users would sometimes want
cept, showing how the runout signal increases the to “compensate” for runout by using the offset adjust-
observed vibration signal in some places (i.e., below 6200 ment potentiometer in the meter. For example, if the
rpm) and decreases it in others (i.e., above 6200 rpm). peak-to-peak amplitude of the runout signal was 0.5
In this case, the vibration is changing while the runout mils, they reasoned that they could simply offset their
remains fixed. However, the same effect can occur when meter by 0.5 mils. Thus, a meter that would normally
the runout is changing, regardless of whether the vibra- indicate 3.2 mils of vibration would indicate only 2.7 mils.
tion levels are stable or changing. Changes in runout This approach was particularly faulty because it not only
may make the vibration look worse than it actually is, failed to recognize that runout can change over time, but

Figure 5 – Bodé plot of same data as in Figure 4. Notice how the compensated data (green) has lower amplitude than
uncompensated data (red) below 6200 rpm, but this is reversed for speeds above 6200 rpm. The complex nature in which
vectors and waveforms combine can cause runout to either increase or decrease the actual vibration amplitude.

1 6 O R B I T [ Vo l . 2 5 N o. 3 2 0 0 5 ]
 APPLICATIONS

that runout is a complex waveform and does not follow

Slow Roll Runout the rules of scalar subtraction.

as a Diagnostic Tool The inherent problems in using embedded compensation

in machinery protection systems was addressed a num-
As pointed out in this article, runout can change,
ber of years ago in American Petroleum Institute
and for this reason, embedded compensation in
Standard 670. It specifically prohibits the use of compen-
a permanent monitoring system is not recom-
mended. However, runout can vary for more sation in permanent monitoring systems. Consistent
reasons than just shaft scratches or changes in with API 670, it has long been our practice to provide
shaft magnetism. One of the most serious mal- compensation in systems used for diagnostic purposes,
functions that can result in a runout change is a but not machinery protection purposes.
shaft crack. While this is not the only symptom of
a crack, any time the slow roll (typically less than Summary
400 rpm) runout amplitude/phase vectors change,
it is imperative to understand why. As we have shown, many factors can influence the
amount of runout present in a shaft. The best approach
Our first rule of shaft cracks states that, “If a shaft
is to prevent runout – rather than mitigate it after the
is cracked, it is almost certainly bowed.” This bow
fact – through appropriate diligence at all stages of the
can change the 1X slow roll runout vectors once
manufacturing process. However, mitigation will still
crack propagation takes place. Thus, if the shaft is
bowed, it may simply be gravity sag, or it may be sometimes be necessary and this article has discussed
more serious. The 2X slow roll runout vector should several methods that can be employed with good suc-
also be checked. Cracks can cause stiffness asym- cess, ranging from degaussing and burnishing to the
metry if they propagate in an uneven pattern, use of alternate materials for the probe tracks. In all
creating a characteristic twice-per-revolution situations, runout can be effectively managed and
flexing. Certain rotor designs are inherently asym- should not preclude users from using proximity probes
metric – such as 2-pole generator rotors, – where on machinery with fluid-film bearings, as these trans-
“normal” shaft asymmetry will yield a noticeable ducers afford the most sensitive and reliable machinery
2X component, but its amplitude and phase condition measurements available.
should not change over time.
For those experiencing runout-related problems or
The moral of this story? Always remember to treat
runout data as a valuable source of diagnostic desiring to prevent such problems from occurring in the
information – not merely “noise” that interferes first place, an excellent approach is to enlist the service
with the true vibration. of GE Energy’s field professionals. They can develop a
runout mitigation plan specific to your operations as well
You can read more about the topic of shaft crack in
the January 1986 issue of ORBIT. as provide the necessary training for your personnel.

[ Vo l . 2 5 N o. 3 2 0 0 5 ] O R B I T 1 7