tive maintenance program must be developed with clear goals and objectives that
permit maximum utilization of the technologies. The program must be able to cross
organizational boundaries and not be limited to the maintenance function. Every func-
tion within the plant affects equipment reliability and performance, and the predictive
maintenance program must address all of these influences.
Vibration monitoring and analysis is the most common of the predictive maintenance
technologies. It is also the most underutilized of these tools. Most vibration-based
predictive maintenance programs use less than 1 percent of the power this technology
provides. The primary deficiencies of traditional predictive maintenance are:
• Technology limitations
• Limitation to maintenance issues
• Influence of process variables
• Training limitations
• Interpreting operating dynamics
13.1.1 Technology Limitations
Most predictive maintenance programs are severely restricted to a small population
of plant equipment and systems. For example, vibration-based programs are generally
restricted to simple, rotating machinery, such as fans, pumps, or compressors. Ther-
mography is typically restricted to electrical switchgear and related electrical equip-
ment. These restrictions are thought to be physical limitations of the predictive
technologies. In truth, they are not.
Predictive instrumentation has the ability to effectively acquire accurate data
from almost any manufacturing or process system. Restrictions, such as low speed,
are purely artificial. Not only can many of the vibration meters record data
at low speeds, but they can also be used to acquire most process variables, such as
temperature, pressure, or flow. Because most have the ability to convert any propor-
tional electrical signal into user-selected engineering units, they are in fact multime-
ters that can be used as part of a comprehensive process performance analysis
program.
13.1.2 Limitation to Maintenance Issues
From its inception, predictive maintenance has been perceived as a maintenance

that quantify the operating condition of a machine-train or process system and eval-
uates them using a design benchmark that defines normal for the system.
13.1.3 Influence of Process Variables
In many cases, the vibration-monitoring program isolates each machine-train or a
component of a machine-train and ignores its system. This approach results in two
major limitations: it ignores (1) the efficiency or effectiveness of the machine-train
and (2) the influence of variations in the process.
When the diagnostic logic is limited to common failure modes, such as imbalance,
misalignment, and so on, the benefits derived from vibration analyses are severely
restricted. Diagnostic logic should include the total operating effectiveness and effi-
ciency of each machine-train as a part of its total system. For example, a centrifugal
pump is installed as part of a larger system. Its function is to reliably deliver, with the
lowest operating costs, a specific volume of liquid and a specific pressure to the larger
system. Few programs consider this fundamental requirement of the pump. Instead,
their total focus is on the mechanical condition of the pump and its driver.
The second limitation to many vibration programs is that the analyst ignores the
influence of the system on a machine-train’s vibration profile. All machine-trains are
Operating Dynamics Analysis 269
affected by system variations, no matter how simple or complex. For example, a com-
parison of vibration profiles acquired from a centrifugal compressor operating at 100
percent load and at 50 percent load will clearly be different. The amplitude of all rota-
tional frequency components will increase by as much as four times at 50 percent
load. Why? Simply because more freedom of movement occurs at the lower load.
As part of the compressor design, load was used to stabilize the rotor. The designer
balanced the centrifugal and centripetal forces within the compressor based on the
design load (100 percent). When the compressor is operated at reduced or excessive
loads, the rotor becomes unbalanced because the internal forces are no longer equal.
In addition, the spring constant of the rotor-bearing support structure also changes
with load: It becomes weaker as load is reduced and stronger as it is increased.
In more complex systems, such as paper mills other continuous process lines, the

To be effective, predictive analysts must have a thorough knowledge of machine/
system design and machine dynamics. This knowledge provides the minimum base
required to effectively use predictive maintenance technologies. Typically, a graduate
mechanical engineer can master this basic knowledge of machine design, machine
dynamics, and proper use of predictive tools in about 13 weeks of classroom training.
Nonengineers, with good mechanical aptitude, will need 26 or more weeks of formal
training.
13.1.5 Understanding Machine Dynamics
It Starts with the Design
Every machine or process system is designed to perform a specific function or range
of functions. To use operating dynamics analysis, one must first fully understand how
machines and process systems perform their work. This understanding must start with
a thorough design review that identifies the criteria that were used to design a machine
and its installed system. In addition, the analyst must also understand the inherent
weaknesses and potential failure modes of these systems. For example, consider the
centrifugal pump.
Centrifugal pumps are highly susceptible to variations in process parameters, such as
suction pressure, specific gravity of the pumped liquid, back-pressure induced by
control valves, and changes in demand volume. Therefore, the dominant reasons for
centrifugal pump failures are usually process related.
Several factors dominate pump performance and reliability: internal configuration,
suction condition, total dynamic pressure or head, hydraulic curve, brake horsepower,
installation, and operating methods. These factors must be understood and used to
evaluate any centrifugal pump-related problem or event.
All centrifugal pumps are not alike. Variations in the internal configuration occur in
the impeller type and orientation. These variations have a direct impact on a pump’s
stability, useful life, and performance characteristics.
There are a variety of impeller types used in centrifugal pumps. They range from
simple radial-flow, open designs to complex variable-pitch, high-volume enclosed
designs. Each of these types is designed to perform a specific function and should be

OPPOSED CONFIGURATION
Figure 13–1 Impeller orientation.
Multistage pumps that use opposed impellers are much more stable and can tolerate
a broader range of process variables than those with an inline configuration. In the
opposed-impeller design, sets of impellers are mounted back-to-back on the shaft. As
a result, the other cancels the thrust or axial force generated by one of the pairs. This
design approach virtually eliminates axial forces. As a result, the pump does not
require a massive thrust-bearing or balancing piston to fix the axial position of the
shaft and rotating element.
Because the axial forces are balanced, this type of pump is much more tolerant of
changes in flow and differential pressure than the inline design; however, it is not
immune to process instability or to the transient forces caused by frequent radical
changes in the operating envelope.
Factors that Determine Performance
Centrifugal pump performance is primarily controlled by two variables: suction con-
ditions and total system pressure or head requirement. Total system pressure consist
of the total vertical lift or elevation change, friction losses in the piping, and flow
restrictions caused by the process. Other variables affecting performance include the
pump’s hydraulic curve and brake horsepower.
Suction Conditions. Factors affecting suction conditions are the net positive suction
head, suction volume, and entrained air or gas. Suction pressure, called net positive
suction head (NPSH), is one of the major factors governing pump performance. The
variables affecting suction head are shown in Figure 13–2.
Centrifugal pumps must have a minimum amount of consistent and constant positive
pressure at the eye of the impeller. If this suction pressure is not available, the pump
will be unable to transfer liquid. The suction supply can be open and below the pump’s
centerline, but the atmospheric pressure must be greater than the pressure required to
lift the liquid to the impeller eye and to provide the minimum NPSH required for
proper pump operation.
At sea level, atmospheric pressure generates a pressure of 14.7 pounds per square inch

chronic failure problems.
Total System Head. Centrifugal pump performance is controlled by the total system
head (TSH) requirement, unlike positive-displacement pumps. TSH is defined as the
274 An Introduction to Predictive Maintenance
(H
vp
) VAPOR PRESSURE
(Hf) FRICTION LOSS IN SUCTION
VELOCITY HEAD LOSS
AT IMPELLER
USEFUL PRESSURE
AVAILABLE N.P.S.H.
LOSS DUE TO
USEFUL PRESSURE AT SURFACE OF LIQUID
ATMOSPHERIC PRESSURE AT SURFACE OF LIQUID
STATIC LIFT
Figure 13–2 Net positive suction head requirements.
total pressure required to overcome all resistance at a given flow. This value includes
all vertical lift, friction loss, and back-pressure generated by the entire system. It deter-
mines the efficiency, discharge volume, and stability of the pump.
Total Dynamic Head. Total dynamic head (TDH) is the difference between the dis-
charge and suction pressure of a centrifugal pump. Pump manufacturers that generate
hydraulic curves, such as those shown in Figures 13–3, 13–4, and 13–5, use this value.
These curves represent the performance that can be expected for a particular pump
Operating Dynamics Analysis 275
200
150
50
100
100 200 300 400 500 600 700 800 1000

of 100psig and a positive pressure of 10psig at the suction will have a TDH of
90psig.
Most pump hydraulic curves define pressure to be TDH rather than actual discharge
pressure. This consideration is important when evaluating pump problems. For
example, a variation in suction pressure has a measurable impact on both discharge
pressure and volume. Figure 13–3 is a simplified hydraulic curve for a single-stage
centrifugal pump. The vertical axis is TDH, and the horizontal axis is discharge
volume or flow.
The best operating point for any centrifugal pump is called the best efficiency point
(BEP). This is the point on the curve where the pump delivers the best combination
of pressure and flow. In addition, the BEP defines the point that provides the most
stable pump operation with the lowest power consumption and longest maintenance-
free service life.
In any installation, the pump will always operate at the point where its TDH equals
the TSH. When selecting a pump, it is hoped that the BEP is near the required flow
where the TDH equals TSH on the curve. If it is not, some operating-cost penalty will
result from the pump’s inefficiency. This is often unavoidable because pump selection
is determined by choosing from what is available commercially as opposed to select-
ing one that would provide the best theoretical performance.
276 An Introduction to Predictive Maintenance
200
100
100 200 300 400 500 600 700 800 1000
150
50
65%
70% 75%
80%
80%
75%

its performance (i.e., flow and head) at various BHPs. Figure 13–5 is an example of
a simplified hydraulic curve that includes the BHP parameter.
Note the diagonal lines that indicate the BHP required for various process conditions.
For example, the pump illustrated in Figure 13–2 requires 22.3 horsepower at its BEP.
If the TSH required by the application increases from 150 feet to 175 feet, the horse-
power required by the pump increases to 24.6. Conversely, when the TSH decreases,
the required horsepower also decreases.
The brake horsepower required by a centrifugal pump can be easily calculated by:
With two exceptions, the certified hydraulic curve for any centrifugal pump provides
the data required by calculating the actual brake horsepower. Those exceptions are
specific gravity and TDH.
Specific gravity must be determined for the specific liquid being pumped. For
example, water has a specific gravity of 1.0. Most other clear liquids have a specific
gravity of less than 1.0. Slurries and other liquids that contain solids or are highly
Brake Horsepower
Flow GPM Specific Gravity Total Dynamic Head Feet
3960 Efficiency
=
()
¥¥
()
¥
Operating Dynamics Analysis 277
viscous materials generally have a higher specific gravity. Reference books, like Inger-
soll Rand’s Cameron’s Hydraulics Databook, provide these values for many liquids.
The TDH can be directly measured for any application using two calibrated pressure
gauges. Install one gauge in the suction inlet of the pump and the other on the dis-
charge. The difference between these two readings is TDH.
With the actual TDH, flow can be determined directly from the hydraulic curve.
Simply locate the measured pressure on the hydraulic curve by drawing a horizontal

278 An Introduction to Predictive Maintenance
deviations from good engineering practices result in turbulent suction flow and cause
hydraulic instability that severely restricts pump performance.
The restrictions on discharge piping are not as critical as for suction piping, but using
good engineering practices ensures longer life and trouble-free operation of the pump.
The primary considerations that govern discharge piping design are friction losses and
total vertical lift or elevation change. The combination of these two factors is called
TSH, which represents the total force that the pump must overcome to perform prop-
erly. If the system is designed properly, the discharge pressure of the pump will be
slightly higher than the TSH at the desired flowrate.
In most applications, it is relatively straightforward to confirm the total elevation
change of the pumped liquid. Measure all vertical rises and drops in the discharge
piping, then calculate the total difference between the pump’s centerline and the final
delivery point.
Determining the total friction loss, however, is not as simple. Friction loss is caused
by several factors, all of which depend on the flow velocity generated by the pump.
The major sources of friction loss include:
• Friction between the pumped liquid and the sidewalls of the pipe
• Valves, elbows, and other mechanical flow restrictions
• Other flow restrictions, such as back-pressure created by the weight of liquid
in the delivery storage tank or resistance within the system component that
uses the pumped liquid
Several reference books, like Ingersoll-Rand’s Cameron’s Hydraulics Databook,
provide the pipe-friction losses for common pipes under various flow conditions.
Generally, data tables define the approximate losses in terms of specific pipe lengths
or runs. Friction loss can be approximated by measuring the total run length of each
pipe size used in the discharge system, dividing the total by the equivalent length used
in the table, and multiplying the result by the friction loss given in the table.
Each time the flow is interrupted by a change of direction, a restriction caused by
valving, or a change in pipe diameter, the flow resistance of the piping increases sub-

Bypass Operation. Many pump applications include a bypass loop intended to prevent
deadheading (i.e., pumping against a closed discharge). Most bypass loops consist of
a metered orifice inserted into the bypass piping to permit a minimal flow of liquid.
In many cases, the flow permitted by these metered orifices is not sufficient to dissi-
pate the heat generated by the pump or to permit stable pump operation.
If a bypass loop is used, it must provide sufficient flow to ensure reliable pump oper-
ation. The bypass should provide sufficient volume to permit the pump to operate
within its designed operating envelope. This envelope is bound by the efficiency
curves that are included on the pump’s hydraulic curve, which provides the minimum
flow needed to meet this requirement.
Stable Operating Conditions. Centrifugal pumps cannot absorb constant, rapid
changes in operating environment. For example, frequent cycling between full-flow
and no-flow ensures premature failure of any centrifugal pump. The radical surge of
back-pressure generated by rapidly closing a discharge valve, referred to as hydraulic
hammer, generates an instantaneous shock load that can literally tear the pump from
its piping and foundation.
In applications where frequent changes in flow demand are required, the pump system
must be protected from such transients. Two methods can be used to protect the
system.
280 An Introduction to Predictive Maintenance
• Slow down the transient. Instead of instant valve closing, throttle the system
over a longer interval. This will reduce the potential for hydraulic hammer
and prolong pump life.
• Install proportioning valves. For applications where frequent radical flow
swings are necessary, the best protection is to install a pair of proportioning
valves that have inverse logic. The primary valve controls flow to the
process. The second controls flow to a full-flow bypass. Because of their
inverse logic, the second valve will open in direct proportion as the primary
valve closes, keeping the flow from the pump nearly constant.
Design Limitations. Centrifugal pumps can be divided into two basic types: end-

resolution vibration data, incoming product characteristics, all pertinent process data,
and actual operating control parameters.
Vibration Data
For steady-state operation, high-resolution, single-channel vibration data can be used
to evaluate a system’s operating dynamics. If the system is subject to variables, such
as incoming production, operator control inputs, or changes in speed or load, multi-
channel, real-time data may be required to properly evaluate the system. In addition,
for systems that rely on timing or have components where response time or response
characteristics are critical to the process, these data should be augmented with time-
domain vibration data.
Data Normalization
In all cases, vibration data must be normalized to ensure proper interpretation. Without
a clear understanding of the actual operating envelope that was present when the vibra-
tion data were acquired, it is nearly impossible to interpret the data. Normalization is
required to eliminate the effects of process changes in the vibration profiles. At a
minimum, each data set must be normalized for speed, load, and the other standard
process variables. Normalization allows the use of trending techniques or the com-
parison of a series of profiles generated over time.
Regardless of the machine’s operating conditions, the frequency components should
occur at the same location when comparing normalized data for a machine. Normal-
ization allows the location of frequency components to be expressed as an integer
multiple of shaft running speed, although fractions sometimes result. For example,
gear-mesh frequency locations are generally integer multiples (e.g., 5¥, 10¥), and
bearing-frequency locations are generally noninteger multiples (e.g., 0.5¥, 1.5¥). Plot-
ting the vibration signature in multiples of running speed quickly differentiates the
unique frequencies that are generated by bearings from those generated by gears,
blades, and other components that are integers of running speed. At a minimum, the
vibration data must be normalized to correct for changes in speed, load, and other
process variables.
Speed. When normalizing data for speed, all machines should be considered to be

speeds.
Load. Data also must be normalized for variations in load. Where speed variations
result in a right or left shift of the frequency components, variations in load change
the amplitude. For example, the vibration amplitude of a centrifugal compressor taken
at 100 percent load is substantially lower than the vibration amplitude in the same
compressor operating at 50 percent load.
In addition, the effect of load variation is not linear. In other words, the change in
overall vibration energy does not change by 50 percent with a corresponding 50
percent load variation. Instead, it tends to follow more of a quadratic relationship.
A 50 percent load variation can create a 200 percent, or a factor of four, change in
vibration energy.
None of the comparative trending or diagnostic techniques used by traditional vibra-
tion analysis can be used on variable-load machine-trains without first normalizing
the data. Again, since even machines classified as constant-load operate in a variable-
load condition, it is good practice to normalize all data to compensate for load varia-
tions using the proper relationship for the application.
Other Process Variables. Other variations in a process or system have a direct effect
on the operating dynamics and vibration profile of the machinery. In addition to
changes in speed and load, other process variables affect the stability of the rotating
elements, induce abnormal distribution of loads, and cause a variety of other abnor-
malities that directly impact diagnostics. Therefore, each acquired data set should
Operating Dynamics Analysis 283
include a full description of the machine-train and process system parameters. For
example, abnormal strip tension or traction in a continuous-process line changes
the load distribution on the process rolls that transport a strip through the line. This
abnormal loading induces a form of misalignment that is visible in the roll and its
drive-train’s vibration profile.
Analysis of shaft deflection is a fundamental diagnostic tool. If the analyst can estab-
lish the specific direction and approximate severity of shaft displacement, it is much
easier to isolate the forcing function. For example, when the discharge valve on an

lems exist. An example is the following description of the imbalance failure mode,
which was obtained from a failure-mode chart: Single-plane imbalance generates a
dominant fundamental (1¥) frequency component with no harmonics (2¥, 3¥, etc.).
Note, however, that the failure-mode charts are simplistic because many other
machine-train problems also excite, or increase the amplitude of, the fundamental (1¥)
frequency component. In a normal vibration signature, 60 to 70 percent of the total
overall, or broadband, energy is contained in the 1¥ frequency component. Any devia-
tion from a state of equilibrium increases the energy level at this fundamental shaft
speed.
14
FAILURE-MODE ANALYSIS
285
14.1 COMMON GENERAL FAILURE MODES
Many of the common causes of failure in machinery components can be identified by
understanding their relationship to the true running speed of the shaft within the
machine-train.
Table 14–1 is a vibration troubleshooting chart that identifies some of the common
failure modes. This table provides general guidelines for interpreting the most
common abnormal vibration profiles. These guidelines, however, do not provide
positive verification or identification of machine-train problems. Verification requires
an understanding of the failure mode and how it appears in the vibration signature.
The sections to follow describe the most common machine-train failure modes:
critical speeds, imbalance, mechanical looseness, misalignment, modulations, process
instability, and resonance.
14.1.1 Critical Speeds
All machine-trains have one or more critical speeds that can cause severe vibration
and damage to the machine. Critical speeds result from the phenomenon known as
dynamic resonance.
Critical speed is a function of the natural frequency of dynamic components such as
a rotor assembly, bearings, and so on. All dynamic components have one or more

Damaged Rolling Impact rates for Radial Uneven vibration levels, often with
Element Bearings the individual & shocks. °Impact-Rates:
(Ball, Roller, etc.) bearing components° Axial
Also vibrations at
very high frequencies
(20 to 60kHz)
Journal Bearings Sub-harmonics of Primarily Looseness may only develop at operating
Loose in Housings shaft rpm, exactly Radial speed and temperature (e.g.,
1/2 or 1/3 ¥ rpm turbomachines)
Oil Film Whirl or Slightly less than Primarily Applicable to high-speed (e.g., turbo)
Whip in Journal half shaft speed Radial machines
Bearings (42% to 48%)
Hysteresis Whirl Shaft critical speed Primarily Vibrations excited when passing through
Radial critical shaft speed are maintained at
higher shaft speeds. Can sometimes be
cured by checking tightness of rotor
components
Damaged or Worn Tooth meshing Radial Sidebands around tooth meshing
Gears frequencies (shaft rpm & frequencies indicate modulation (e.g.,
¥ number of teeth) Axial eccentricity) at frequency corresponding to
and harmonics sideband spacings. Normally only
detectable with very narrow-band analysis
Mechanical 2 ¥ rpm
Looseness
Faulty Belt Drive 1, 2, 3 & 4 ¥ rpm Radial
of belt
Unbalanced 1 ¥ rpm and/or Primarily
Reciprocating multiples for higher Radial
Forces order unbalance
and Couples

DD
PD
11 –
•
For Ball Detect 1(Hz) =
Con l
2
n1
2
DD
PD
(1 – )
2
•
repsor
The best way to confirm a critical-speed problem is to change the operating speed of
the machine-train. If the machine is operating at a critical speed, the amplitude of the
vibration components (1¥, 2¥, or 3¥) will immediately drop when the speed is
changed. If the amplitude remains relatively constant when the speed is changed, the
problem is not critical speed.
14.1.2 Imbalance
The term balance means that all forces generated by, or acting on, the rotating element
of a machine-train are in a state of equilibrium. Any change in this state of equilib-
rium creates an imbalance. In the global sense, imbalance is one of the most common
abnormal vibration profiles exhibited by all process machinery.
Theoretically, a perfectly balanced machine that has no friction in the bearings would
experience no vibration and would have a perfect vibration profile—a perfectly flat,
horizontal line—however, no perfectly balanced machines exist. All machine-trains
exhibit some level of imbalance, which has a dominant frequency component at the
fundamental running speed (1¥) of each shaft.

always be higher than any subsequent harmonics.
Lift/Gravity Differential
Lift, which is designed into a machine-train’s rotating elements to compensate for the
effects of gravity acting on the rotor, is another source of imbalance. Because lift does
not always equal gravity, some imbalance always exists in machine-trains. The vibra-
tion component caused by the lift/gravity differential effect appears at the fundamen-
tal or 1¥ frequency.
Other
All failure modes create some form of imbalance in a machine, as do aerodynamic
instability, hydraulic instability, and process loading. The process loading of most
Failure-Mode Analysis 289
Figure 14–1 Single-plane imbalance.
machine-trains varies, at least slightly, during normal operations. These vibration com-
ponents appear at the 1¥ frequency.
14.1.3 Mechanical Looseness
Looseness, which can be present in both the vertical and horizontal planes, can create
a variety of patterns in a vibration signature. In some cases, the fundamental (1¥) fre-
quency is excited. In others, a frequency component at one-half multiples of the shaft’s
running speed (e.g., 0.5¥, 1.5¥, 2.5¥) is present. In almost all cases, there are multi-
ple harmonics, both full and half.
Vertical
Mechanical looseness in the vertical plane generates a series of harmonic and half-
harmonic frequency components. Figure 14–3 is a simple example of a vertical
mechanical looseness signature.
In most cases, the half-harmonic components are about one-half of the amplitude of
the harmonic components. They result from the machine-train lifting until stopped by
the bolts. The impact as the machine reaches the upper limit of travel generates a fre-
290 An Introduction to Predictive Maintenance
Figure 14–2 Multiplane imbalance generates multiple harmonics.
quency component at one-half multiples (i.e., orders) of running speed. As the machine

ances do not generate multiple harmonics. In these cases, the vibration profile con-
tains unique frequencies that indicate looseness, but the profile varies depending on
the nature and severity of the problem.
With sleeve or Babbitt bearings, looseness is displayed as an increase in subharmonic
frequencies (i.e., less than the actual shaft speed, such as 0.5¥). Rolling-element bear-
ings display elevated frequencies at one or more of their rotational frequencies. Exces-
sive gear clearance increases the amplitude at the gear-mesh frequency and its
sidebands.
Other forms of mechanical looseness increase the noise floor across the entire band-
width of the vibration signature. Although the signature does not contain a distinct
292 An Introduction to Predictive Maintenance
Figure 14–4 Horizontal looseness creates first and second harmonics.