Condition Monitoring based on Acoustic Emission

by Trevor Holroyd
Posted 7-13-09

When we talk about machinery condition monitoring, we’re talking about condition in the sense of a machines ability to
continue to perform its intended function in an efficient manner. Unfortunately there is no way to directly measure
machine condition and so it is necessary to infer it from indirect measurements. For Condition Monitoring (CM) purposes a
range of technologies are available, each having its own strengths and weaknesses, and it is usual to consider each of
them as a tool in the CM toolkit.

The Acoustic Emission (AE) technique has a 40 year history of use for machinery condition monitoring and although it got
off to a slow start, in recent years it has gained very widespread acceptance across industry. The particular strength of AE
is its ability to directly detect the processes associated with wear and degradation (including friction, impacts, crushing,
cracking, turbulence, etc,). It does this by detecting the surface component of stress waves that these processes invariably
generate. These stress waves travel all over the machines surface, which means that sensor positioning is not critical.

Another convenient feature of AE signals is that they generally have a high Signal to Noise Ratio (SNR), which means that
the signals related to machine condition can be clearly seen and are not buried in other inconsequential signals. This is a
direct consequence of the high frequency and resonant nature of an AE sensors response. An example of an application
that makes full use of this high SNR is the monitoring of blowers in a blower house. This application environment involves
very high noise and vibration levels with adjacent identical blowers cutting in and out in response to process demands
without warning. Despite this, clear trends of processed AE signals have successfully identified a number of faults including
defective motor coupling (see Figure below showing good and bad examples) and build up of lime on the rotors enabling
wash initiation and avoidance of rotor collision. The high and variable background noise together with the problems of
cross-talk and sympathetic resonance from adjacent blowers prevents the successful use of Vibration techniques in this
application.

AE sensors, signals and technology have little in

common with those of the Vibration technique and this
has many positive consequences in practical terms. For
example, with AE it isn’t a pre-requisite to perform a
frequency analysis (or FFT) and interpret the signal
levels at all possible defect frequencies before
determining whether or not there is a fault. In fact,
with AE the usual measurement sequence is the logical
and time saving one of:

(a) instantly alert to the presence of a fault on a

machine without knowledge of machine details (such as
bearing type or number),
(b) only look at the trend of readings to reveal the rate
of degradation on those machines with a fault flagged,
(c) only carry out an FFT if diagnostic information is
required of a fault condition.
Of course, such a radical departure from the norm was
initially met with a fair degree of scepticism along the
lines of ‘It can’t be that easy otherwise why would people use Vibration?’ A fair point in its time! The only answer was to
relentlessly demonstrate instant fault detection on the shop floors of those who were interested but doubtful. A not
untypical example of readings from such a demonstration is shown below for one-off measurements taken with a portable
AE based CM instrument (MHC type) on the various white metal bearings of a generating set in a power station. (Note: for
this instrument the common interpretation for all types of machines, bearings and rotational speeds down to 35 rpm is that
an item is suspect if Distress® is 10 or more.)

Prior to these measurements being taken, the client believed that the main exciter was as good as new! However upon
seeing the readings the client explained that the exciter had previously shown problems of arcing in the bearings and the
resulting surface damage had been blended out prior to the bearing being returned to service.

The high SNR which AE signals enjoy has other consequence too, as will now be described:

Measurement Point Distress®

HP Rotor DE 03

HP Rotor NDE 04
 IP Rotor DE 05

IP Rotor NDE 03

LP1 Rotor DE 04

LP1 Rotor NDE 03

LP2 Rotor DE 03

LP2 Rotor NDE 03

Generator Rotor DE 04

Generator Rotor NDE 03

Exciter DE 19

Exciter NDE 16

Pilot Exciter 07

Very slowly rotating machinery

At slow rotational speeds the rate at which AE signals are generated reduces, even becoming infrequent at the lowest
speeds. However, this does not necessarily pose a problem provided signals are processed accordingly since AE sensors
are insensitive to the low frequency sounds and vibrations which are ever present on the shop floor. Commercially
available AE instrumentation is able to easily monitor down to 0.25 rpm without requiring special expertise.
Non-repetitive or random fault types
Because AE signals can be simply analysed directly in the time domain, without recourse to frequency analysis, they are
equally sensitive to non-repetitive signals. One example of this has been the detection of carbonisation of oil in a high
temperature white metal bearing where the small particles of carbon were crushed in the bearing as they randomly
formed. Another example has been the detection of sporadic air bubbles in the oil flowing through a plain bearing.
Reciprocating machinery
It is normal to get load reversals, valve openings/closures, sliding actions and gas transfers during the operation of
reciprocating machinery and each of these has an associated AE signature. A full interpretation of the AE signal requires
knowledge of the timing sequence of all these actions but it can also be useful to simply observe occasional or persistent
variations from the norm. This is illustrated in the Figure below which shows the maximum, minimum and mean values of
the AE envelope signal as a function of
time within the cycle of a diesel engine
over 12 successive engine cycles.
Comparison of the minimum and mean
waveforms clearly indicates an occasional
drop out of part of the operating process
(thought to be due to a sticking valve)
which is only detectable because of the
high SNR of AE signals - a low SNR would
have required averaging to reveal the
waveform in the first place thus masking
exceptional occurrences.
Intermittent actions

Similar comments to those for reciprocating machinery apply to intermittent actions. For a single action the high SNR of AE
signals is particularly relevant in allowing the signals to be directly analysed without the need for averaging of multiple
actions.

The high SNR and the resulting suitability of time domain signal processing has had a positive impact on simplifying AE
instrumentation. In particular for rotating machinery it has allowed generic time domain signal processing algorithms to be
developed which are quick and easy to use. Initially this led to the creation of very simple yet effective portable
instruments but, more recently, these algorithms have been incorporated into the AE sensor itself (see example shown
below). The importance of this development is that it eases the way for the utilisation of AE sensors with other technology
sensors (such as rms vibration and temperature) in shop floor installations as it removes the need for a separate AE
instrument.

With increased understanding of the fundamentals underpinning the use of AE for CM and the overwhelming evidence of its
benefits to industry, it is not surprising that AE has gradually moved centre stage over the years. In addition to having its
own standard ISO 22096 (Condition monitoring and diagnostics of machines – Acoustic Emission) it is specifically listed in
the CM standard ISO 18436 (Condition monitoring and diagnostics of machines – Requirements for qualification and
assessment of personnel) as one of the four main CM techniques, together with Infra-red Thermography, Lubrication
Management & Analysis and Vibration Analysis. Alongside these developments the British Institute of Non-Destructive
Testing as a UKAS accredited, third party certifying body has recently created a certification programme of training and
examinations that conforms to the ISO 18436 international standard.

Understanding The Basics Of Balancing & Measuring Techniques

Gary K. Grim & Bruce J. Mitchell VibrAlign

Why Balance? All rotating components experience significant quality and performance improvements if balanced. Balancing
is the process of minimizing vibration, noise and bearing wear of rotating bodies. It is accomplished by reducing the
centrifugal forces by aligning the principal inertia axis with the geometric axis of rotation through the adding or removing
of material. In order to understand the basics of balancing it is necessary to define the following fundamental terms.

FUNDAMENTAL TERMS

CENTER OF GRAVITY (C.G.):

When a is the acceleration due to gravity, the resultant force is the weight of the body. For this reason the term center of
gravity can be thought of as being the same as the center of mass. Their alignment would differ only in large bodies where
the earth's gravitational pull is not the same for all components of the body. The fact that these points are the same for
most bodies, is the reason why static (non-rotating) balancers, which can only measure the center of gravity, can be used
to locate the center of mass. Additional information on static balancers will be reviewed in the following pages.

CENTER OF MASS:
The center of mass is the point in a body where if all the mass was concentrated at one point, the body would act the
same for any direction of linear acceleration. If a force vector passes through this point the body will move in a straight
line, with no rotation. Newton's second law of motion describes this motion as F = ma. Where the sum of forces, F, acting
on a body is equal to its mass, m, times its acceleration, a.

F=ma

GEOMETRIC AXIS:

The geometric axis is also referred to as the shaft axis or the engineered axis of rotation. This axis of rotation is
determined either by the rotational bearing surface, which exists on the work piece, or by the mounting surface. An
adequate mounting surface establishes the center of rotation at the center of mass plane (the plane in which the center of
mass is located).

PRINCIPAL INERTIA AXIS:

When a part is not disc shaped and has length along the axis of rotation, it spins in free space about a line. This line is
called the principal inertia axis. The center of mass is a point on this line. It takes energy to disturb a part and cause it to
wobble or spin on another inertia axis. Examples of this would be a correctly thrown football or a bullet shot from a rifle.
When the principal inertia axis coincides with the axis of rotation the part will spin with no unbalance forces. In this case
the static as well as the couple unbalance are equal to zero.

In summary, a state of balance is a physical condition that exits when there is uniform total mass distribution. Static
balance exists when the center of mass is on the axis of rotation. Whereas, both static and couple balance exist when the
principal inertia axis coincides with the axis of rotation.

TYPES OF UNBALANCE

The location of the center of mass and the principal inertia axis is determined by the counter balancing effect from every
element of the part. However, any condition of unbalance can be corrected by applying or removing weight at a particular
radius and angle. In fact the amount of unbalance, U, can be correctly stated as a weight, w, at radius, r.

U=wt

Static unbalance can also be determined if you know the weight of the part and the displacement of the mass center from
the geometric axis. In this case, U, is equal to the weight, w, of the work piece times the displacement, e.

U=we

STATIC UNBALANCE:

Is a condition that exists when the center of mass is not on the axis of rotation. It can also be explained as the condition
when the principal axis of inertia is parallel to the axis of rotation. Static unbalance by itself is typically measured and
corrected on narrow disc-shaped parts, such as a Frisbee. To correct for static unbalance requires only one correction. The
amount of unbalance is the product of the weight and radius. This type of unbalance is a vector, and therefore, must be
corrected with a known weight at a particular angle. Force unbalance is another name for static unbalance.

As discussed earlier, a workpiece is in static balance when the center of mass is on the axis of rotation. When this
condition exists, the part can spin on this axis without creating inertial force on the center of mass. Parts intended for
static applications, such as speedometer pointers or analog meter movements, benefit from being in static balance in that
the force of gravity will not create a moment greater at one angle than at another which causes them to be non-linear. The
following drawing represents an example of static unbalance.
COUPLE UNBALANCE:

Is a specific condition that exists when the principal inertia axis is not parallel with the axis of rotation. To correct couple
unbalance, two equal weights must be added to the workpiece at angles 180° apart in two correction planes. The distance
between these planes is called the couple arm. Couple unbalance is a vector that describes the correction. It is common for
balancers to display the left unbalance vector of a couple correction to be applied in both the left and right planes.

Couple unbalance is expressed as U = wrd where the unbalance amount, U, is the product of a weight, w, times the radius,
r, times the distance, d, of the couple arm. Couple unbalance is stated as a mass times a length squared. Common units of
couple unbalance would be g-mm2 or oz-in2. The angle is the angle of the correction in the left plane. (Please note: In
mechanics, the angle is perpendicular to the plane of the radius vector and the couple arm vector. This is an angle 900
from the weight location.) Couple unbalance can be corrected in any two planes, but first the amount must be divided by
the distance between the chosen planes. Whereas static unbalance can be measured with a non-rotational balancer, couple
unbalance can only be measured by spinning the workpiece.

A combination of force and couple unbalance fully specifies all the unbalance which exists in a part. Specifying unbalance in
this manner requires three individual correction weights. The following drawing represents an example of couple
unbalance.

TWO PLANE UNBALANCE:

Is also referred to as dynamic unbalance. It is the vectorial summation of force and couple unbalance. To correct for two
plane unbalance requires two unrelated correction weights in two different planes at two unrelated angles. The
specification of unbalance is only complete if the axial location of the correction planes is known. Dynamic unbalance or
two plane unbalance specifies all the unbalance which exists in a workpiece. This type of unbalance can only be measured
on a spinning balancer which senses centrifugal force due to the couple component of unbalance.

DYNAMIC BALANCING:

Is a term which specifies a balancer that spins and measures centrifugal force. It is necessary to use this type of balancer
when measuring couple or two plane unbalance. Typically it can also be used to provide greater sensitivity to measure
static or force unbalance. The following drawing represents an example of dynamic unbalance.

UNITS OF UNBALANCE

Unbalance can be specified as the weight of mass to be added or removed at a correction radius. The weight units can be
any convenient units of measure which take into account the weighting equipment available and the size of the whole unit
of measure. Grams (g), ounces (oz), and kilograms (kg) are the most common units. Occasionally Newton's (N) are
specified, but for practical use must be converted to available weight scale units. Length units usually correspond to the
manufacturers standard drawing length units. Most commonly these are inches (in), millimeters (mm), centimeters (cm),
and meters (m). The most common combinations used to specify unbalance are ounce-inches (oz-in), gram-inches (g-in),
gram-millimeters (g-mm), gram-centimeters (g-cm), and kilogram-meters (kg-m).

MOTION OF UNBALANCED PARTS

What is the effect of unbalance on a rotating part? At one extreme, if mounted in a rigid suspension, a damaging force
must exist at support bearings or mounting surface to constrain the part. If the mount is flexible, the part and mount will
exhibit significant vibrations. In a normal application, there is a combination of both.
Consider an unbalanced thin disc mounted on a simple spring suspension. The spring will respond differently depending on
the speed at which the disc rotates. At very low speeds (less than one half the resonant frequency of the spring mass) the
unbalance of the disc generates very little centrifugal force, causing a small defection of the spring and a small motion of
the mass.

With rigid bodies the unbalance remains the same although an increase in speed causes an increase in force and motion.
Force increases exponentially as the square of the change in speed. Twice the speed equates to four times the force and
four times the motion. In other words, force is proportional to the square of the rotating speed. An equation for estimating
force is:

F=1.77U(rpm/1000)2

CENTRIFUGAL FORCE

Centrifugal force caused by 0.001 oz-in of unbalance at various speeds.

The centrifugal force of the unbalance is outward from the center of the part, at the location of the weight. In a hard
suspension balancer the force bends a rigid spring causing the high spot of vibration to occur at the location of the weight.

At speeds twice or greater than the resonant frequency of the spring-mass, the unbalance force is much greater than the
spring force. The motion of the unbalanced part is limited by its own inertia. The part rotates about the present center of
mass at any running speed in this range. Displacement peak is equal to the center of mass eccentricity, e, and therefore
Xp = e. The formula for displacement peak, Xp, is Unbalance, U, divided by the part weight. (Note: the weight units of
unbalance must be the same as part weight units.) In a balancer this would be termed a soft suspension.

Xp=U / weight of the part

At remaining speeds near the resonant frequency, the amplitude of motion can get much larger than at higher speeds even
if the unbalance force is less. The resonance exists when the resisting force of the part inertia is equal to and opposed to
the resisting spring force. The only resisting force is due to mechanical damping. When the damping is low, the amplitude
of vibration may be fifty times greater at resonance. In the past some balancing companies ran their balancers at this
speed to gain sensitivity. However, with the great improvements of present day electronics, this range of speed is
considered unpredictable and is therefore typically avoided.

A part other than a thin disc, which has length along the rotating axis, has a similar response when rotated supported in a
suspension system at each end. With speeds below resonance (in a hard suspension), the force generated by centrifugal
force divides between the two suspension points just as a simple static load divides between two fulcrum points. With
speeds above resonance (in a soft suspension), the part spins, not only about the center of mass, but also about the
principal inertia axis. The peak displacement at any point along the part equals the distance between the principal inertia
axis and the geometric axis. It should be noted that there may be several resonance speeds. Resonance of the total mass
on a spring system will cause the part to translate. At a different speed, the part rotational inertia and spring system will
cause it to rotate about a vertical axis. This is another reason to avoid this range of running speed.
BALANCING EQUIPMENT

STATIC BALANCERS:

Static balancers do not rotate the part in order to measure unbalance. Instead, their operation is based on gravity
generating a downward force at the center of gravity. An example of and older form of static balancer is a set of level
ways. Although extremely time consuming, this old method is still effective at minimizing static unbalance. The force
downward on the center of gravity will cause the part to rotate until the C.G. is directly below the running surface, which
identifies the location of the heavy spot. Typically with level way balancing the unbalance amount is not known and the
part is corrected by trial and error until the part no longer rotates. However, it is possible to measure unbalance amount
on a level way balancer. This is accomplished by rotating the heavy spot up 90°, and then measuring the moment of
torque. Historically, this was often achieved by using a hook scale to determine force at a known radius.

Modern static balancers measure parts with the parts rotational axis in a vertical orientation, directly over a pivot point.
This type of gage can quickly sense both amount and angle of unbalance. Gravity acting on an offset center of mass
creates a moment on the part which tilts the gage.

Static balancers can be divided into two types depending upon how they react to this unbalance moment: those with a free
pivot where the amount of tilt is measured as a direct indication of the amount of unbalance, and those that restrict
amount of tilt and measure the moment of unbalance.

Static balancers which have a free pivot offer no resistance to the downward force of gravity on the C.G. It is necessary
that the C.G. of the workpiece and tooling together be a proper distance below the pivot point. The distance the C.G. is
below the pivot point determines the sensitivity of the balancer. This distance is often set up by an adjustable
counterweight connected to the tooling below the pivot. With no part on a leveled set of tooling, the C.G. initially is directly
below the pivot point. When an unbalanced part is placed on the tooling it causes the C.G. to raise and shift away from the
center in the direction of the unbalance. Moment caused by the gravity on the new C.G. causes the tooling to tilt, until the
new C.G. is directly below the pivot. As it tilts the moment arm and, consequently, the moment, are reduced to zero. The
amount of tilt is determined by measuring the distance between an arm extending from the tooling and the machine base.
The amount of tilt is proportional to the amount of part unbalance.

Measuring unbalance on a static balancer is most often achieved with two LVDT's oriented at 90° to each other. A typical
pivot consists of points in a socket, ball on an anvil, a small diameter flexure in tension, hydraulic sphere bearings, and air
sphere bearings. Each have problems associated with keeping the pivot free. The mechanical point contact system must be
mechanically protected to prevent flat spots on the ball, or a point of indentation in the anvil. The wire flexure can be bent
or broke if not protected. The sphere bearings must be kept perfectly clean to prevent drag. Two additional concerns are
that the sensitivity is dependent upon the weight of the part and the pivot must be well protected to prevent damage that
can effect balancer performance.

There is however a better alternative that overcomes these problems, it is called the stiff pivot balancer. With this type of
balancer the pivot is a post which acts as a stiff spring. The moment due to unbalance bends the post a small amount and
the tilt is measured to determine the amount of unbalance. With a stiff pivot balancer the calibration is not effected by part
weight and the balancer is accurate, simple, and extremely rugged.

DYNAMIC BALANCERS:

The previously described static balancers depend totally upon the force of gravity at the C.G. As a result, with a static
balancer, it is not possible to sense the couple component of unbalance. To sense couple unbalance the part must be spun.
Such a balancer is termed a centrifugal or dynamic balancer. Dynamic balancers consist of two types: soft suspension and
hard suspension.

The most common dynamic balancers fixture the workpiece with the shaft axis horizontal. There are, however, both soft
and hard bearing vertical balancers too.

SOFT SUSPENSION DYNAMIC BALANCERS:

Are also referred to as a soft bearing balancers. The soft suspension balancer operates above the resonant frequency of
the balancer suspension. With this type of balancer the part is force free in the horizontal plane and rotates on the
principal inertia axis. The amplitude of vibration is measured at the bearing points to determine the amount of unbalance.
There are problems in using the measured information to correct the balance of the part. Each individual part has its own
calibration factor and crosstalk of correction information. Stated in a different way, if a balanced part has one unbalance
weight added in one correction plane, the information necessary to predict the new line of the principle inertia axis is not
available. One weight causes vibration at both suspensions and the amplitude and ratio of these two vibrations is not
known. When the influence of a weight in a second plane is added, it is not possible to separate the information on the two
weights.

To determine the calibration and crosstalk factors, trial weights must be added individually in each plane, and the reaction
measured. When using an unbalanced part the effect of initial unbalance must be removed from the trial weight
measurements. When these factors have been determined, each channel reads out only the unbalance in the
corresponding correction plane. These two channels then have what is called plane separation. The main disadvantage of
soft suspension balancers is the requirement of extra setup spins for the calibration of different size and weight
workpieces.

DYNAMIC HARD SUSPENSION BALANCERS:

Are also referred to as a hard bearing balancers. The hard suspension balancer operates at speeds below the suspensions
resonant frequency. The amplitude of vibration is small, and the centrifugal force generated by the unbalance is measured
at the support bearings. With a hard suspension balancer it is only necessary to calibrate the force measurement once.
This one time calibration is typically performed by the balancer manufacturer at their own facility.

Using the force measurement and an accurate speed measurement, the balancer electronics can calculate the corrections
which are required at the support bearing planes. However, since corrections cannot be made at the bearing planes, the
unbalance information must be translated to the two correction planes. For the calculation, the location of the correction
planes relative to the bearing planes are entered by the operator when the balancer is set up for a particular part.

In addition to the advantage of being inherently calibrated, hard suspension balancers are: easier to use, safer to use, and
provide rigid work supports. With hard suspension balancers it is possible to provide hold-down bearings to handle the
negative load which can be generated when a part is run outboard of the two support bearings.

All of the balancers described are implemented with analog electronics. However, the basic calculations required for plane
separation and plane translation require complicated circuits, which in turn require trimming and setup. Computer
electronics are ideally suited to these applications. In addition computer electronics can memorize part setups for easy
recall, collect unbalance data, provide statistical information, and output the data to a printer or disk drive.

SUMMARY:

Virtually all rotating components experience significant quality improvements if balanced. In today's global market
consumers look for the best products available for their money. They demand maximum performance, minimum size, and
lower cost. In addition everything must be smaller, more efficient, more powerful, weigh less, run quieter, smoother and
last longer.

As consumer demands continue to increase, balanced components will remain an essential ingredient. Balancing will
always be one of the most cost effective means of providing quality products to consumers.