THE ARTIFICIAL INTELLIGENCE SYSTEMS FOR MACHINE CONDITION MONITORING AND DIAGNOSTICS BY VIBRATION

Alexej V. Barkov, Natalia A. Barkova

Published in the Proceedings of the Saint Petersburg Post-graduate Institute of the Russian Federation Power Industry and Vibration Institute, USA, Volume 9, Saint Petersburg, 1999.

One of the leaders in vibration diagnostics of rotating machinery is the Association of companies "VibroAcoustical Systems and Technologies" (VAST, Inc.) from Saint Petersburg, Russia. VAST together with foreign colleagues and companies that represent it in different countries throughout the world conducts research and development of methods and means for machine condition monitoring and diagnostics. The main products of VAST are the artificial intelligence systems that are delivered to many countries of the world.
This article discusses the main problems in machine vibration diagnostics and the capabilities of VAST in this area of activities.

1. Introduction

In recent years, when one refuses to maintain and repair the equipment by schedule, in practice it is repaired by three main methods: But only the third method is economically efficient. Its successful use enables to: For the last method, a detailed diagnostics is necessary and it is desirable to detect all the defects that influence the service life long time before failure so that to have time to prepare for maintenance. Practice has shown that in mechanical and electromechanical equipment, efficient diagnostics is possible mainly by vibration signal as: Such conventional methods as checking the temperature, analysis of lubrication and others are not necessary when your approach is correct - they are substituted by vibration analysis.

2. The History of Machine Condition Vibration Diagnostics.

In any industry, the time to introduce an idea in reality takes about 20-30 years. The same period of time was needed for the development of efficient machine condition diagnostics. The main methods of diagnostics appeared in sixties-eighties when the means of vibration analysis that were better than the human's hearing capabilities were introduced. But for practical realization of vibration diagnostics methods, 2-3 decades were necessary and only last years these methods were introduced practically in all industries.

In Russia, almost half of the efficient machine diagnostic methods were born in the laboratories of shipbuilding industry where the leaders of VAST, Inc. worked from the later sixties. The confirmation of the fact that they were among the originators of machine vibration diagnostics can be found in many Russian and Western publications and magazines. It is well known that practically all the diagnostics has appeared in the Navy of Russia, USA and Great Britain, that means where this problem was real and where there were no problem with financing the investigations and where the best scientists and engineers were working.

But the measuring means were not so common that time to make the vibrational diagnostics so popular as it is nowadays. In practice, these means became real in the beginning of nineties when the instruments became to be designed on the basis of microcomputers. Only then the complicated methods of vibration analysis became accessible to a wide class of users. Now each 2-3 years there appears a new generation of measuring and analyzing techniques. One can say that modern instrument comprises of a transducer and microcomputer. Nowadays the so called virtual instruments when the instrument is a combination of a transducer and a personal computer are wide spread. One of its manufacturers with which VAST, Inc. cooperates is the company Data Physics, Inc. (USA).

Naturally five years later, the manufacturers of such instruments appeared in Russia too. Now there are about a dozen such companies. VAST, Inc. produces DC-11 and besides Russia its production has begun now also in Canada.

But detailed machine diagnostics by vibration means not only the methods of diagnostics and measuring and process analyzing instruments. There are two more mandatory components:

The leaders in developing the data bases always were the companies of the US that for many years made efficient computerized systems for machine monitoring, i.e. for watching the processes including the machine vibration. The software included the data base with the machine specification data and the results of measurements, friendly user's interface with different alarm levels, capability of graphical analysis with trend drawing and other means of information display for a specialist - diagnostician.

The problems of the second component were solved by one method - each time an expert in diagnostics of a certain type of equipment was invited. In many countries and especially in the US there is a special system of preparing and upgrading the qualification of such experts. But in some countries including the countries of CIS (former USSR) such system does not exist.

But there are two more methods to solve the problems of this second component - the development of artificial intellect. One method is to develop the learning artificial intellect system where on the first stage the teaching of the system is conducted by the software architect and then the user supplements the system by the rules that he needs. The second method is to design a self learning (adaptive) system with strict learning algorithms written by the designers.

Almost all the companies that design the diagnostic systems use the first method. VAST, Inc uses the second method. Only last years there appear its followers. It is the most complicated method but such systems give a real diagnosis and condition prediction immediately after there purchase. The artificial intellect designed by VAST, Inc. is used also by a number of the world manufacturers of monitoring and diagnostic systems. Among them the leading manufacturers of Europe - Bruel & Kjaer (nowadays it has united with the company "Schenk") (Denmark - Germany), "Diagnostics Instruments" (Great Britain), and also manufacturers in the USA and Canada - DPL (Canada), VibroTek, Inc. (former Inteltech Enterprises, Inc.), USA. To realize it VAST, Inc. conducted serious projects with each of these companies.

The main reason why the VAST, Inc. projects are so widely used is that such systems really substitute the experts and the productivity of such diagnostic systems on its base increases in many times.

3. Vibration, its Measurement and Analysis

Vibration and noise are natural processes in the machines and equipment and they are excited by the same dynamic forces that are the reasons of the wear and different types of defects.

Naturally, the vibration and noise transform from one into another on the border of gas and solid states, but the human being directly perceives only noise and only in limited low frequency range the vibration.

The oscillation velocity is responsible for the transformation of the vibration into noise. The inflationary pressure in air is proportional to the velocity of the oscillation surface. That is why the standards as, a rule, limit the oscillation velocity of the machines and equipment.

But the vibration control and vibration diagnostics are different practical problems. In diagnostics the defect is defined by the oscillation force that is applied to the defective zone and the force is linear connected with the oscillation acceleration but not with its velocity. So for diagnostics more often the vibration acceleration is measured and the vibration velocity is measured additionally for the machine vibration control and in restricted low frequency range.

For vibration measurements as a rule are use the sensors of vibration acceleration that work using the pjezoelectric effect. In such sensors the output electric charge is proportional to the force applied to the sensor. Only for some on-line vibration control systems for large machines with fluid film bearings the sensors of oscillation displacement mounted in the bearings are used (usually two sensors for each bearing). These sensors enable to measure the trace of the rotor center (its orbit), i.e. to measure directly the wear of the bearing's shells.

To measure the noise the microphones with different methods of converting the sound pressure into an electric signal are used. For machine diagnostics sometimes are used directional microphones that enable to define the direction to the point of noise emission. In practice with a microphone one can conduct remote measurement of the object vibration, i.e. the value of its vibration velocity.

In VAST, Inc. the noise measurements are used very seldom for diagnostic purposes as in the air the noise from different sources are summarized practically without losses from many sources and it is very difficult to analyze the noise from a certain source in the presence of other noise sources. Besides this the problem of extracting certain noise components, excited by a defect, is meshed by its possible refraction, multiple reflection and other effects during its propagation.

There is one more reason because of which the noise cannot be recommended for machine diagnostics. In close proximity of the place where the defect appears it is necessary to take into account the form of the object oscillation. Here a significant contribution both in the vibration and noise give the pseudo-components of complicated forms, i.e. those, that while the distance from the object is increased, transform into simple (wave) form. When you measure the noise on some distance from the object the information, that is contained in the pseudo-noise or pseudo-vibration, disappears.

After converting the vibration (noise) signal into an electric signal it is necessary to analyze thoroughly the latter receiving but not loosing the diagnostic information. There are very strict requirements to the analyzing instruments for diagnostics. The typical operations that the instruments analyzing the vibration must do are the following:

  1. To measure the overall vibration level in a frequency range required by the machine vibration control standards and in units required by these standards.
  2. Spectral analysis of the vibration, i.e. to divide the vibration into components of different frequencies, defined by the nature of the oscillation forces.
  3. To analyze the power oscillations of separate vibration components extracted preliminary from the vibration signal. This is as a rule the analysis of the envelope spectra of random high frequency vibration signal.
  4. To analyze the vibration signal form, i.e. the analysis of the time signal (the oscillograph mode of operation).
It should be noted that not all of the manufactured comparatively simple analyzing instruments can do all the mentioned types of analysis, at least with the quality necessary for the diagnostics.

4. The Quality and Units of the Vibration Measurements

Diagnostics means the search of weak components on the background of strong ones. The weak and strong components usually differ also by the frequencies. By the power these components can differ in 106 times, that is why not their power is measured but their amplitudes. In this case the difference between the weak and strong components decreases up to 103 times. But not only it is necessary to detect a weak component but also to define its features. So the signal analyzer must be capable without any switching to ensure the dynamic range about 104 times. It should be added also that the machines, because of for example different rotating speeds, can have different maximum vibration amplitude that can differ in 100 times. Hence it is evident that a good instrument must have a dynamic range of measurements without changing the sensor about 106 times.

For convenience the vibration components that so strongly differ from one another are compared in acoustics in logarithmic scale.

Two components, whose power differ from each other in 10 times, in logarithmic scale are considered to differ by 10 dB. The difference of their amplitudes are different. In acoustics the difference in amplitudes in 10 times is defined as the difference in logarithmic scale by 20 dB. So now it is necessary to correspond the basic reference points of certain units of vibration acceleration, velocity, displacement, sound pressure and decibels (dB). According to the IEC standards:

The vibration displacement, velocity and acceleration measured in one point and expressed in dB coincide by their value only for one frequency - 1000 rad/s or 159 Hz.

5. The Instrument Requirements

One of the requirements applied to the instruments is its good linearity. This linearity is necessary to enable the analysis of small components obviating the nonlinear distortion from the strong components. Naturally, that the limitation of this linearity is the dynamic range of 80 dB, viz. not worse than 0.01%. In practice in best cases it is possible to achieve the value of 0.03%. It means that the distortion appears on the level –70 dB. It is quite sufficient for diagnostic measurements.

The next requirement is applied to the frequency range. The typical requirement is to have a frequency range from 2 Hz up to 20 kHz. But sometimes a more wide frequency range is needed. In some cases the frequency range has to be widened up to 40 kHz, in some industries the frequency range has to be from 0.3 Hz or even from zero. There are no criteria here and VAST, Inc. and VibroTek, Inc. provides all the diagnostics in the frequency range from 2 Hz up to 25 kHz.

And the last requirement concerns the resolving capacity of the instruments in the frequency range, viz. the number of lines in the signal spectrum. The typical requirement is to have from 100 to 800 lines, but in some instruments there are up to 6400 lines and even more. VAST and VibroTek limit themselves by real problems when to have 1600 lines is usually sufficient.

The instruments for vibration spectrum and enveloped spectrum analysis that VAST and VibroTek uses most often are shown on fig.1.

Fig.1. Some instruments with which the software designed by VAST and VibroTek work (from left to right and downward):

It should be noted that the list of necessary types of analysis and instruments' requirements is realized now in the portable instruments only of VAST and DPL production.

6. The Sources of Vibration in the Rotating Machines.

Dynamic forces are acting in each machine. These forces are the sources not only of the noise and vibration but the defects as well. These defects change the features of the forces and hence the characteristics of the noise and vibration. One can say that the machine functional diagnostics without changing the machine operation mode is the investigation of dynamic forces but not the vibration and noise. The latter only contain the information about the dynamic forces, but in the process of the transformation of dynamic forces into the vibration and noise some information is lost. Even more information is lost during transformation of the forces and their work into heat energy. That is why when there is a choice between two types of signal (temperature or vibration) in diagnostics vibration should be preferred.

The main dynamic forces that act in a rotating machine and excite its vibration and noise are shown in table 1.

Table 1.
The types of oscillating forces that act in rotating machines

FORCESOURCE
1. Mechanical Nature
CentrifugalRotor imbalance
KinematicSurface roughness
ParametricThe shaft, bearing and other units' rigidity fluctuation
Friction forcesThe rolling and slide friction units
ShocksFaulted friction surfaces
2. Electric and Magnetic Nature
MagneticThe air gap volume fluctuation in the magnetic conductor
ElectrodynamicVariable components of current and magnetic flux
MagnetostrictiveThe magnetostrictive effect in the magnetic conductor
3. Flow Dynamics
Lifting flowThe movement of a blade in an irregular flow or a group of nonuniform blades in uniform flow
Friction forcesThe boundary between the flow and fixed parts
Pressure pulsationsThe flow turbulence, burble, cavitation

The main forces of the mechanical nature are the following:

The main forces of the electrical and magnetic nature are the following:

The main forces of the flow dynamics nature are the following:

The forces of hydrodynamic origin are mainly of the same nature as in the gas environment, but the pressure pulsations can be added because of cavitation, that can appear in some certain condition in the fluid flow.

The dynamic forces in the machines excite the vibration either directly or the forces excite the noise and the noise excites the vibration of the machine case.

The vibration, depending on the nature of the forces that excites it, can be deterministic (more often periodical) or random.

One of the simplest examples of a deterministic vibration signal is a harmonic oscillation (fig.2).

Fig.2. A simplest harmonic oscillation

It is characterized by an amplitude (peak value Xpeak, root mean square value XRMS or mean value Xmean) of the detected signal, by the frequency f=1/T and initial phase.

A random signal (fig.3) can have any value in a certain range, so it is characterized not by an amplitude, frequency and phase, but by its peak value, root mean square value, mean value of the detected signal and peak-to-peak value.

Fig.3. Random vibration signal

A periodic vibration can be presented as a spectrum. It can contain one component (a harmonic signal) (fig.4a, or fig.4b) or many multiple components (fig.5 right)

Fig.4. The time domain signals and their spectra

If the signal represents a combination of two simple harmonic components (fig.4c) with different frequencies and amplitudes then in its spectrum will be clearly seen just these two harmonic components with different frequencies and amplitudes (fig.4c, right).

Time dimain vibration signalSpectrum
Fig.5. A complex periodical vibration signal and its spectrum

As the periodic components are represented by spectra, the random components should be represented by the spectra too but the spectra have to be continuous (fig.6).

Fig.6. A spectrum of random vibration components

The spectrum is convenient as it divides the vibration into components with different features and often with different nature.

A typical spectrum (fig.7) has as a rule many harmonic components in the low frequency range. They become rare while the frequency increases and practically they are absent in the high frequency range.

Fig.7. Vibration signal spectrum

When choosing the frequency range of vibration for machine and equipment condition diagnostics one must take into account the features of vibration of different frequency.

For example in the infralow frequency band the vibration can be excited even not by the tested machine but by the nearby operating machines including the transport running on far distance from the equipment that is measured.

The peculiarity of low frequency vibration is in slight damping with distance and hence in the point where the transducer is fixed the vibration comes from all the units of the tested machine, from the machines with which it is connected, from the nearby equipment. That is why during vibration analysis using the low frequencies there appears a difficulty of faulted unit localization and the problem of interference immunity. On these frequencies (in the frequency range up to the third - fifth harmonics of the rotation frequency) the machine is oscillating as a whole, so to swing all the machine large forces and severe defects are needed.

In the middle frequency band the vibration is excited mainly by the oscillating forces applied to the points nearest to it. In the vibration spectrum there are seen a large number of harmonic components of different frequencies but because of many resonances the ratio of amplitudes of these components differs significantly from the ratio of oscillating forces that excite them. As a consequence there is a distortion of information about the defects, the sources of the oscillating forces, and no reproducibility of results when the rotation speed varies even slightly.

In the high frequency band the vibration becomes of a wave nature. In the spectrum there is little number of lines, lack of information (for the first sight), but even small forces are enough to excite the vibration.

The vibration in ultrahigh frequency band is excited mainly by the microshocks but it propogates only in a homogeneous medium (metall without bolts and weld joints). Usually it is very difficult or even impossible to reach the optimal points for measurements if it is not a vessel or a pipeline.

7. The Forces That Excite the Vibration in the Rotating Machines' Units

7.1. Rotors and Couplings

The main force applied to the rotor is the centrifugal force. It has the frequency that equals the rotation frequency RPM/60 and a circular form. If there is any fault, especially a misalignment of rotors, the form will deviate from a circular one - in the spectrum appear superharmonics (multiple harmonics) kRPM/60.

The second force is the parametric one. It appears when the element has a splineway or such fault as a crack, viz. when the rigidity of the shaft depends on the rotation angle and changes twice per revolution. In this case in the vibration spectra even harmonics of the rotation frequency appear 2kRPM/60.

The third force is the shock force. It appears when the joint coupling has a defect. The joint coupling strips off from the stable position several times during the revolution and the shaft hits the bearings. As a result the jumps of the high frequency bearing vibration power appear several times per revolution (kRPM/60 in the envelope spectrum).

Each component of these forces can coincide with the resonance of certain machine elements. It can lead to sharp excursion of the vibration amplitude on this frequency.

7.2. Rolling Element Bearings.

In both the rolling element and slide bearings act two main types of forces - kinematic and friction forces. In faulted rolling element bearings sometimes appear the third type of forces - shock pulse forces. The vibration created by rolling element bearings besides the rotation frequency fr is characterized by the following main frequencies:

Usually even harmonics of this frequency appear in the vibration.

We have discussed the forces of the kinematic nature. If the bearing is a new one and all the surfaces are "circular" then only the vibration on the frequencies multiple to BPFO (кfout) should be seen ("irregular road surface"). If there are any severe defects then the shaft will "jump" with the frequencies that correspond to all appropriate defects. If these "jumps" are high enough the lubrication can be forced through and "dry" shocks appear that excite high frequency vibration. Just the same shocks can appear if the lubrication is not so good and its layer can be easily "torn".

In the bearing act also the friction forces. They excite the high frequency vibration and if there are any defects that lead to even partial discontinuity of the lubrication layer then the value of the friction forces and the power of vibration will fluctuate. To detect the faults by this symptom it is necessary to measure the envelope spectrum of the high frequency bearing's vibration.

7.3. Journal Bearings

Besides the forces of kinematic origin and friction forces in the journal bearings, the forces act hat are the results of nonlinear interaction of the static load with the friction forces. These forces accompany the rotor self-oscillations in the bearings.

The rotor self-oscillations in the journal bearings are very much alike the pendulum oscillations of the rotor in relation to the equilibrium position in the lowest point of the bearing. The rotor is shifted from the equilibrium position by the friction forces and is returned in it by the gravity force. The reason of this unstable equilibrium is the nonlinear dependence of the friction forces from the thickness of the lubrication layer that grows while the rotor position deviates from the equilibrium position. The self-oscillation frequency is the lesser the larger is the gap in the bearing, i.e. the more is the bearing's wear.

As a rule the rotor self-oscillation frequency changes abruptly from the RPM to 1/2 RPM but sometimes, with increasing the wear, to 1/3 RPM. The reason of the rotor self-oscillation can be not only its wear, but also the decreased quality of lubricant or failure in feed lubrication. The self-oscillation can appear also in the rolling element bearings but only with large wear. The frequency of the rotor self-oscillation in the rolling element bearings as a rule coincides with the second order of the rotating frequency of the cage.

The shock forces that act in the journal bearings can be of two types. A "dry" shock with the disruption of the lubrication layer is very dangerous but it appears very seldom and is accompanied with significant growth of high frequency vibration. "Hydraulic" shock does not disrupt the lubrication layer, but because of uneven wear of the bearing in the loaded zone, where the thickness and the velocity of the lubrication flow jump, the turbulent breakaway of the flow occurs. The moment of the breakaway of the flow is sensed by the measuring system as a shock, accompanied by an impulse increase of the high frequency vibration. Such shock does not lead to fast destruction of the bearing but it is a cause of fast uneven wear.

The friction forces in the journal bearing are rather stronger than in the rolling element bearings but as the high frequency bearing vibration, when there is no turbulence of the lubrication flow, is activated only by the boundary friction, the random vibration of the journal bearing is significantly lower than in the rolling element bearings.

7.4. Gearing

The oscillating forces in the gearing are generated in the gearing zone and can be of kinematic, parametric and shock origin. Unlike the rotor bearings, in the gearing the static load is defined in the gearing not by the force of gravity of the gearing wheel, but by the transferred torque that often has a dynamic component because of a defect in gearing wheel that has no contact with the wheel under control.

In a nondefective gearing there are some small oscillating forces of kinematic, parametric and very often of shock nature when each tooth enters mesh. The frequency of oscillating forces is defined by the number of teeth and rotating frequency of the gear.

FT = T1*RPM1/60 = T2*RPM2/60,

where

The vibration on the other frequencies is not connected with the construction peculiarities but is defined by the technological deviations during their manufacturing or by the wheels' defects.

First of all these are the defects of separate tooth. In this case once a revolution of the faulted gear wheel the forces that can be of different nature occure, including kinematic (when the loaded surface of the tooth has a smooth unevenness), parametric (when in the zone of gearing there is a change of rigidity) or shock (when loaded surface of the tooth has an abrupt change). With these defects the low and high frequency gearing vibration increases, but the later does not reach the measuring points on the gearing box case.

One more peculiarity of oscillating forces in gearing is the appearance of low frequency forces when the defects are on both wheels. These are the oscillating forces with the frequency FSH that is a subharmonic of the rotating frequencies of both gear wheels. On this frequency appear the oscillating forces of kinematic, parametric or shock origin, but the gear box case reach also only the low frequency components, that are excited by these forces

FSH = RPM1/(60 k1) = RPM2/(60 k2)

where k1 and k2 are integers.

7.5. Magnetic system of electric motor

The formation of electromagnetic and electrodynamic forces in electric motors have their own peculiarities.

7.5.1. Induction motors

The electromagnetic forces acting in the alternating current motors (induction motors and synchronous motors) have their own peculiarities - their frequency is twice the frequency of the magnetic field because it is proportional to the magnitude of the magnetic field ignoring its direction. So the main electromagnetic forces in an alternating current machine are acting with the frequency 2fm, where fm is the frequency of the supply voltage (a.c. mains).

The second by its magnitude oscillating force has the slot frequency. The vibration with the slot frequency is sometimes traditionally called "magnetic noise". This vibration not always shows up visibly on the background of other components with nearby frequencies. There is a peculiarity of forming the oscillating forces defined by the slots of the rotor and stator. It consists in the fact that the slots of the rotor enter the stator field with the frequency fz=Z*RPM/60. But the field itself is a pulsating one and can be resolved into two different components that rotate in opposite directions with the power-line (mains) frequency fm. As there are two poles under which are the slots and the magnetic field has maximum magnitude the forces act on three slot frequencies:

fz1 = Z*RPM/60,

fz2 = Z*RPM/60 - 2fm,

fz3 = Z*RPM/60 + 2fm.

But the form of the stator oscillation has not so complicated form as the field has and respectfully the oscillation amplitude has an increased value only on one or two frequencies. It is defined by the machine construction, viz. by the number of poles and slots.

Some words about the electrodynamic forces.

When the rotor winding (squirrel cage) is absolutely symmetrical the electrodynamic forces have no alternating components. They generate only a constant (operating) torque. If the winding, i.e. the currents induced in it, is not symmetrical than a low frequency pulsating torque with a double slip frequency appears:

fs = 2Sfm,

where S is the slip of the rotor that is equal S=(fm - p*RPM/60)/fm. Here p is the number of pole pairs.

If the field of the stator is unsymmetrical, i.e. besides the field that rotates in the main direction with the frequency fm a badly compensated field, that rotates in the opposite direction, is present, then there appears an alternating electrodynamic force and correspondingly a torque of forces with a frequency 2fm. This situation appears both when there are unsymmetrical stator windings or when the power-line is unsymmetrical.

But even this is not all. If the form of the supply voltage is distorted and if in the machine windings the higher harmonics are present, then the additional components of the rotating electromagnetic field in the air gap with the frequencies f=3kfm and electrodynamic forces and torques with the frequencies f = 6kfm appear.

Additional electromagnetic forces, that appear with such defects as the volume deviation of certain parts of the air gap, should be mentioned.

The appearance of the air gap static eccentricity leads to the increase of radial vibration with the frequency 2fm and increase of the number of slot harmonics with frequencies .

When there is a dynamic eccentricity of the air gap then the radial vibration with the rotating frequency RPM/60 is increased.

The magnetic saturation of the iron in the slot zone of the rotor is characterized by the increase of the radial machine vibration with the frequencies kfm, and sometimes 6kfm.

When the rotor is displaced in the axial direction then the vibration in this axial direction will be increased on the frequency 2fm.

In the magnetic circuit of the alternating current machines act magnetostrictive forces with the frequency 2fm. They are manifested most of all on the end surfaces of the rotor iron core. When the laminated core became loose then besides the magnetostrictive forces the shocks of the sheets one on another appear with the same frequency and the vibration and noise increases on the frequencies 2kfm.

7.5.2. Synchronous Machines

The main differences between the synchronous machines and induction motors are the following:

  1. The absence of vibration on the frequencies multiple to the slip frequency as there is no slip in the synchronous machines.
  2. The defects of excitation windings manifest themselves as in the induction motors the dynamic air gap eccentricity manifest itself.
  3. The slot vibration harmonics (magnetic noise) are defined by the stator, but not the rotor.
  4. In salient pole machines a radial vibration with the frequencies kp*RPM/60, where р is the number of pole pairs, appears.
7.5.3. Direct Current Electric Machines.

The peculiarity of the direct current motors is the absence of rotating electromagnetic field in them. So there are no oscillating forces on the twice line frequency. By the same reason the magnetic noise in these motors is the only one component with the slot frequency.

The electromagnetic field in the air gap of the direct current motor has much more complicated form than in the air gap of the alternating current motors with distributed winding of rotor and stator. This field is formed by the main poles and when the motor is loaded a significant contribution give the additional poles. The main vibration component, i.e. the slot vibration harmonic, is defined by the fluctuations of certain parts of the air gap volume during the rotation of the slotted armature. It is formed mainly on the edges of the main and additional poles and significantly changes with any variation of each air gap form and hence with the appearance of any defect in the magnetic system. As the slot vibration has a significantly high frequency and increases first of all near the pole where the air gap has changed there appears a possibility to trace the faulted pole in large direct current motors.

Electrodynamic oscillating forces in defect-free direct current motors are absent and are generated only when appear the defects first of all such as poor contact in the brush and commutator unit. The forces, largest by their magnitude, appear when one or the commutator plates became open. In this case the frequency of the pulsating electordynamic forces is defined by the number of opens in the armature per revolution, i.e. the number of brush rows and is equal 2kр*RPM/60. The second force by its magnitude appears when the condition of commutation the current in the brush and collector unit is distorted and is defined by the number of plates in the collector that can differ in some machines from the number of slots in the armature. In this case the collector vibration harmonics have the frequency kzk*RPM/60. If the motor is supplied by a rectifier then in the exciting circuit or in the armature circuit can appear besides the constant current also the alternating components of the current with the frequencies kf*m. Most often these frequencies are kf*m = k6fm. In this case the electordynamic forces and vibration with such frequencies appear in the machine and they can be significantly large. Such vibration is especially seen when the motor is supplied by electronic rectifiers (SCR's - Silicon Controlled Rectifiers). The output current of such rectifiers have a large number of alternating voltage components with different frequencies.

7.6. Impellers

In the pumps, hydraulic turbines, gas and steam turbines, compressors, blowers, fans and other similar bladed machines the impellers that are the sources of noise and vibration define mainly the operation life.

There are no other forces of mechanical nature generated by an impeller besides the cetrifugal forces, defined by the imbalance. All the main oscillating forces are defined by the interaction of the flow with the blades of the impellers and with the inner surface of the case. In the pumps and hydrolic turbines these forces have the hydrodynamic nuture, in the gas and steam turbines, compressors, blowers, fans and so on these forces are of aerodynamic origin.

The main force applied to each blade of the impeller is the lifting force that is a result of the interaction of the flow with the blade that has a position with some angle against the flow. The sum of the lifting forces that act on an impeller without defects in a uniform flow is directed along the axis of its rotation and has no variable component. When the mean value or the direction of the lifting force of one of the blades differs form the others the impeller will be exposed to first of all a radial to the rotation axis force with the rotating frequency and secondly, to a moment of forces with the same frequency, the vector of which will be also directed perpendicular to the rotation axis.

If the flow in the operating zone of the impeller is nonuniform then a radial force and a moment of forces of the same nature is applied to the impeller but the frequency of these forces is defined by the number of blades on the impeller fb=Zb*RPM/60. This frequency is called the blade-passing frequency (number of blades times shaft-rotating frequency).

The described forces and moments with frequencies k*RPM/60 and kfb are transferred via the rotation supports from the impeller to the case of the machine and simultaneously the pulsation of flow pressure impact the same case via the liquid or gas. However the case is subject not only to pressure pulsation of the flow on these frequencies. First of all these are the result of pulsation of the flow on the boundary layer between different mediums (air or hydrodynamic friction), and then the pulsation of pressure resulting from flow turbulence, the flow burble from the blade tips, and finally, the pulsation from cavitation in liquid in the pumps where this cavitation is present.

The case vibration, excited by the variable pulsation of pressure in the flow, has a random nature and hence has no fixed frequency.

8. Main Methods of Diagnostics by Vibration

Two main concepts concerning the machine condition estimation by vibration are known in the worldwide practice.

  1. Vibration monitoring defined as watching the changes in machine vibration condition and the analysis of the reasons of these changes.
  2. Vibration diagnostics when the defects in the controlled objects are detected and identified (the type and severity of the defects).

The main difference between monitoring and diagnostics is in that monitoring has no object to detect the defects in their incipient stage of development. Its function is to detect in time severe defects in assumption that at least not long enough before failure any defect is only a unit in the chain of defects and at least one defect in this chain influence significantly the machine vibration that can be detected by comparatively simple methods of analysis of vibration signal that is measured in one or several machine control points.

According to this aim the vibration monitoring requires the measurement with little time intervals between them not to miss quickly developing defects. For this it is natural that the specialists want to have on-line monitoring systems that conduct measurements with time intervals of parts of seconds or several seconds.

The second natural user's desire is to decrease the number of measurement channels that define the cost of the system. Hence, the absence of vibration transducers on many units disable the detection of defects in their incipient stage of development in these units.

The objective of diagnostics is to detect the incipient defects, watch and predict their development, plan the maintenance of machines. But if the problem is to organize the maintenance according to real machine condition then the task becomes very complicated. In this case it is necessary to detect all the defects on their incipient stage of development. The fact that there is no defect that develops suddenly (except hidden defects of manufacturing and mounting) is already proven, at least for the rotating machinery.

So the main peculiarities of vibration diagnostics are the following:

To demonstrate the existing diagnostic methods for different units of rotating machines it is necessary to fulfill the main rule of the quantitative diagnostics - the machine condition must be defined by the deviations of the diagnostic parameters from their standard values.

Two main interconnected problems of diagnostics derive from this rule - how to find the optimal diagnostic parameters and how to conduct a standard for each parameter. Let us begin with the second problem that is common for all the branches of technical diagnostics.

A standard for a machine without defects can be conducted by three methods:

  1. by measuring each diagnostic parameter for a group of machines without defects, defining its mean value and boundaries of their permissible variations;
  2. by watching the deviations of the diagnostic parameter on the beginning stage of a certain object's operation, defining the trend of this parameter and its permissible variation, after this you can use these values as a standard for this parameter for the certain object for its further operation (standard by the history);
  3. making all this job in the beginning stage of a certain measurement of a diagnostic parameter use it for the second part of this one measurement (immediate standard).

The best diagnostic parameters for machine diagnostics are those that enable to build immediate standards. But unfortunately, in vibration diagnostics there are only few such parameters that really manifest the machine condition. In particular such standards can be used for modulated signals. For example, for the vibration components the power of which is constant in machines without defects and pulsates in faulted machines. Or for such oscillating forces for which the frequencies are constant in the machines without defects and vary in defective machines.

An example of using the immediate standards in practice is the diagnostics of friction units by the oscillation of the friction forces and, hence, by the oscillation of the power of the high frequency random vibration excited by them. An other example is the diagnostics of a rotor of an induction motor by its pulsating moments and hence by the rotating frequency variations. These methods were primarily offered by the specialist of VAST, Inc. in late seventies.

8.1. The Peculiarities of Rolling Element Bearing Diagnostics

For many years the machine diagnostics by high frequency vibration could not be developed because of the absence of effective methods of its analysis. But in 1968 the Swedish specialists offered their method of high frequency vibration analysis that was sensitive to the appearance of micro shocks in the contacts of friction elements in the rolling element bearings. Further it was called "Shock Pulse Method". To conduct such type of high frequency vibration analysis special measuring instruments were designed that even nowadays are very widely used because of their low price.

The kernel of the shock pulse method is in the fact that the presence of even incipient defect in the bearing leads to the appearance of high frequency pulses and hence to the increase of peak levels in the high frequency vibration. The root mean square (RMS) values in this case can stay even stable without changes. So the ratio of peak and RMS values, that is called the peak factor, is a diagnostic symptom. If there are no shock pulses then the peak factor value of the rolling element bearing high frequency vibration is less than five, but when the shock pulses are present, then this symptom can exceed the value ten. The principle of the shock pulse method can be explained by an example of high frequency vibration time signal of a bearing (Fig.8) without faults a) and a bearing with a cavity on the rolling surface b).

Fig.8. Time domain high frequency rolling element bearing vibration
a) bearing without fault, b) bearing with a cavity on the rolling surface

The shock pulse method and comparatively simple instruments, that realize it, measure the peak factor (or some of its modifications, for example the crest factor) of the vibration in the frequency range higher than 25 kHz, is widely used to control the rolling element bearings' condition. The method is very sensitive even to the incipient defects but not all types of defects in the rolling element bearings lead to shock pulses appearance. Such defects that are not followed by the displacement of lubrication layer, for example mounting defects, cannot be detected by shock pulse method. But it is well known that such defects significantly influence the life time of the bearing. As the shock pulses can appear as a result of different defect types and deterioration of grease or lubrication it does not enable to identify the type of defect. Besides this the shock pulse method does not enable to make a long term condition prediction as it can not define the defect type, and as it is well known that different defects have different rate of defect development. And one more significant peculiarity of shock pulse method - the presence of shock pulses in high frequency vibration of the non-faulted low speed bearings with the rotating speed 50-100 RPM, which makes the use of this method in principle difficult and decreases its reliability.

So the shock pulse method enables to control the rolling element bearing condition but not to diagnose it.

Not sufficient efficiency of defect development prediction by shock pulse method and the breakdown of diagnostic reliability for low speed machines was the reason to search for more efficient methods of incipient defects detection and prediction of their development.

In 1978 by the specialists from Saint Petersburg was proposed a method of diagnosing that was named similar to existing methods "Enveloping Method". This method, where not the high frequency vibration itself is analyzed, but the low frequency oscillation of its power, enables to overcome all the restrictions of the shock pulse method and significantly widens its area of application in machine diagnostics by high frequency vibration, increases the reliability of diagnostic results, and, what is especially important, increases the quality of long term condition prediction of the diagnosed equipment.

The burden of the enveloping method consists in the following. The friction forces that excite the high frequency random vibration are stationary only when there are no defects. In non-defective friction units the random high frequency vibration is stationary also . Its power is constant in time. When appears a defect (see fig.9), that leads even to a partial displacement of the lubrication, then the friction forces periodically vary in time and shocks appear. They excite high frequency vibration. The shocks can also appear if the lubrication or grease are not in a very good condition and its layer is easily torn.

Fig.9. High frequency random amplitude-modulated signal

So in the presence of defects the magnitude of the friction forces and the power of vibration varies in time, i.e. a modulation of the high frequency vibration power appears (fig.9).

The depth of modulation m of a random amplitude-modulated vibration signal X(t) can be determined in percentage using the mean value of the envelope ,

,

where , are the maximum and minimum values of the enveloped signal, respectfully.

When a defect type changes then the modulation frequency changes too. The more is the development of the defect, the larger is the depth of the modulation. Hence the frequency of modulation defines the type of the defect and the depth of modulation - its development. Fig. 10 presents as an example on the left side the time domain vibration signals of bearing without faults a), with wear b) and with a cavity on the friction surface c). So the most full information consists in the envelope spectrum of the high frequency signal. The envelope spectra of the bearing's vibration for a bearing without defects a), with wear b) and with a cavity on a friction surface c) are shown on the right side of fig. 10.

Time domain signal of high frequency vibration Envelope spectra of high frequency vibration
Fig.10. Time domain signals of high frequency bearing vibration and their envelope spectra
a) a bearing without faults, b) a bearing with a wear of friction surface, c) a bearing with a cavity on the surface of its race

The figure shows that if a bearing has no defects the envelope spectra of its random vibration does not contain any harmonic component. The envelope spectra of vibration of a worn bearing contains a strong harmonic component that indicates a smooth and periodical variation of vibration signal power. In a bearing with shock pulses the power of high frequency vibration changes abruptly and in its envelope spectrum appears already a number of multiple harmonic components.

In the envelope spectrum of high frequency vibration it is possible to watch simultaneously the development of all defects by the values of the excess of harmonic components on certain frequencies against the background value of the random components. So the possibility to define the partial depth of modulation appears, i.e. the depth of modulation for each existing defect. It enables to define the development of each defect and to identify their types. And this enables to give a prognosis of diagnosed unit condition as each type of defect has its own rate of development.

The depth of modulation m is interconnected with the difference of the harmonic and random components levels of the envelope spectrum (see fig.10) by the following expression:

,

where:

One of the main problem for VAST, Inc. specialists for 30 years was the analysis of high frequency vibration, search and extraction the diagnostic information from this part of the vibration signal. The main patents in this field belong to VAST's specialists and just in this aspect of high frequency vibration application for rotating machine and equipment diagnostics VAST, Inc. is one of the world leaders.

The significant increase of the quality of the diagnostic results gained is defined not only by the use of a new vibration analysis method but also by the increased volume of the collected information. In particular the shock pulse method requires measurements during time of 2-3 revolutions of the diagnosed unit while the use of the envelope method requires the measurements during 50-100 revolutions. It means that the time of measurement for low speed machines can be of several minutes.

A quick implementation of the diagnostic methods on the base of the envelope method several years ago was not possible because there were no comparatively cheap instruments, but in 1991 there appeared first Russian instruments that could conduct spectral analysis of the enveloped spectra of machine vibration random components. Nowadays not only in Russia but also in other countries are manufactured conventional measuring and analyzing instruments that have in their typical functions an option for spectral analysis of the enveloped random vibration with preliminary extraction of the random components from the measured signal.

The results of rolling element bearing diagnostics by enveloped vibration spectra are significantly influenced by the quality of lubrication. For example, because of poor lubrication the disruption of lubrication film can occur and the symptoms of these disruptions are very similar to the symptoms of cavities on the races or rolling elements. That is why in the cases when the diagnostics of the bearings are made by a single measurement it is expedient to control also the spectrum of the bearing unit using the standard calculated by a group of similar machines. In this case by the complex of results of vibration spectra and enveloped spectra measurements it is possible by simple methods exclude the mistakes in defect identification by single vibration measurements of any rolling element bearing in any stage of its service life.

8.2. The Peculiarities of Fluid Film Bearings Diagnostics

The fluid film bearings are practically always diagnosed by joint measurements of low frequency vibration and enveloped spectra bearing vibration excited by the friction forces.

The defects of the fluid film bearings are detected by three main symptoms.

1. By the fluctuations of the thickness and location of the lubricant wedge that modulate the random vibration by the low frequency random process. It results in the raise of the beginning part of the enveloped spectrum (in the low frequency range).

Fig.11. Enveloped spectrum of high frequency vibration of a fluid film bearing with a wear
(detected by the instability of the lubricant wedge)

2. By appearance of shaft's auto-oscillations in the bearings. These are the sources of vibration with the subharmonics of the rotating frequency (1/2 or 1/3 of RPM).

Fig.12. Enveloped spectrum of fluid film bearing vibration with auto-oscillation of the shaft

3. By appearance of shocks (dry and hydrodynamic). In this case in the envelope spectrum of the high frequency bearing vibration a series of weakly damped components multiple to RPM appears.

Fig.13. Enveloped spectrum of vibration of fluid film bearing
without defects (a) and with shocks in the bearing (b).

8.3. The Peculiarities of Gearing Diagnostics

The diagnostics of gears and gearing is conducted also by the enveloped spectrum of the high frequency vibration that is measured on the bearings of these gears and by the spectra of the low frequency and middle frequency vibration where the features of the gear mesh vibration is analyzed.

The main diagnostic symptom of the defects is the appearance of dynamic load applied to the bearings. This symptom works much more efficient than the symptom, used by many specialists, - the appearance of shocks in the gearing. The point is that when there is a defect of teeth this shock is often "negative", it means that the load in the gearing is not growing but declines. In such cases the method, based on detecting the shocks in the gearing, does not work.

In the enveloped spectrum of the gearbox bearing vibration (see fig.14) with a defect of a gearing wheel, for example with an uneven wear of teeth, both the harmonic components with the frequencies multiple to the rotating frequency and the teeth harmonics with side components that differ by the rotating frequency appear.

Fig.14. Vibration spectrum of a gearbox bearing and its envelope spectrum when the teeth is worn

8.4. The Peculiarities of Diagnostics of the Electric Motors' Electromagnetic Systems

The AC electric motors' diagnostics is based on the methods detecting the components of the vibration of electromagnetic and electrodynamic origin that were discussed previously. The envelope vibration spectra for the electromagnetic systems' diagnostics are not used.

To detect one part of the defects, that are accompanied by appearance of alternating electrodynamic forces and pulsating moments, the case vibration is analyzed in two directions - radial and tangent (see fig.15).

To detect the other part of defects, that are accompanied by deviation of radial electromagnetic forces, the radial vibration of the case is investigated.

Fig.15. An example how to chose the measurement points to control the vibration of electric machine

Many defects in the electric machine rotors lead to the modulation of vibration and are easily detected by single vibration measurements. These are the defects of the squirrel cage and the dynamical eccentricity of the air gap in the induction motors.

For example to control the squirrel cage condition in an induction motor by pulsating moments it is necessary to measure the spectrum of low frequency vibration with high frequency resolution (see fig.16). The measurement time in this case can be comparatively large as some induction motors have the motor slip frequency sfm, where fm is the supply main's frequency, less than 1% of the rotating frequency, especially if the motor is working with little load.

Fig.16. Vibration spectra of an induction motor
a) without defects and b) with a broken bar of the squirrel cage

The diagnostic symptom of such defect is the appearance of pulsating moments at two times slip frequency 2sfm and, hence, the appearance in the vibration spectrum at the frequencies k*RPM/60 side bands that differ from the multiple rotating frequencies by 2k1sfm.

The diagnostic symptom of an air gap dynamic eccentricity (uneven rotating air gap) in the induction motors is the presence of a nonsymmetrical magnetic field, rotating together with the rotor, and, hence, the appearance of amplitude modulation of slot harmonics of the vibration at frequencies K1*RPM/60. It means that side bands appear at the multiple slot harmonics at frequencies . This symptom is the same for detecting dynamic eccentricity of the air gap in induction motors, the defects in the excitation system in synchronous machines and defects in the direct current motors' armature windings.

Fig.17 shows the vibration spectra of synchronous machine without defects and with a defect in the exciting winding with a short-circuited section of one of the windings. The widening of the lines is the result of sideband components' presence.

Fig.17. The vibration spectrum of a synchronous machine
a) without defects and b) with the defect of the exciting winding

The diagnostic symptom of static eccentricity of rotor (uneven stationary air gap) in the AC electric motor is the raise of vibration components at frequencies that differ from the slot passing frequency by and also the raise of radial vibration component at the double frequency of the supply mains. The main attention is paid to the raise of the components at the frequencies as the raise of the vibration at the frequency is usually less and besides this it can be caused by other reasons.

An example of a radial vibration spectrum of an AC motor without a defect and with a static ecentricity of rotor is given on fig.18.

Fig.18. The spectra of radial vibration of induction motor
a) without defects and b) with a static eccentricity of the air gap.

Some defects of the stator windings lead to appearance of pulsating moments in the machine. They are detected by the ratio of tangential against radial vibration at the frequency 2fm in the induction and synchronous machines.

At higher frequencies the difference between the radial and tangential vibration decreases. Then the defects are detected by the increase of corresponding vibration components.

A similar method is used to detect the changes in the vibration condition of the AC machines because of the unsymmetry of the supply voltage or the distortion of its form. To differentiate the unsymmetry of the supply voltage and the defects in the stator windings is difficult. It is necessary to compare the parameters of several machines and if all the machines have the pulsating moments at frequencies 2fm, then the reason of it is the unsymmetry of the supply voltage.

And the last point. Unfortunately, some electrical machine defects are detected only by the growth of corresponding vibration components. So a large number of measurements are needed to construct a standard of a machine without defects both by group or by the history of measurements. Such defects include the static eccentricity of the air gap in the AC electrical machines, mechanical unbalance of the rotor and some other defects.

The diagnostic capabilities for the DC machines are a little bit less than for the AC machines because the number of components of the electromagnetic origin in such machines is less.

By pulsating moments in the DC machine it is possible to detect the circuit openings in the commutator circuit. In this case appear new vibration components at the frequencies 2kp*RPM/60.

Besides this if in the supply voltage the alternative components at frequency of the supply voltage fm are present, then the pulsating moments appear with this frequency and its multiple frequencies kfm.

The armature windings defects and the dynamic eccentricity of the air gap are detected by the modulation of the slot passing vibration component.

The distortion of the commutation condition leads to the appearance or increase of the vibration at the commutation frequencies and their harmonics.

The most complicated problem is to detect the defects of the excitation system that lead to the increase in the slot passing frequency vibration in the DC machine. To do this it is necessary to have a standard of slot vibration. This standard it is necessary to derive by a group of machines, working with the same load, because for many machines even without load there is an individual dependence of the slot vibration on the value of the load.

8.5. The Rotors and Couplings Diagnostic Peculiarities

All the peculiarities of rotor's diagnostics are connected with the necessity to create a standard or template of the vibration spectra components at the frequencies multiple to the rotor's rotating speed. The availability of such standard enables to detect three main defects of the rotor and shaft. They include:

The rotor's unbalance is detected mainly by the increased machine's vibration at the rotor's rotating frequency, the unsymmetry of rigidity - by the growth of the vibration at the twice rotating frequency harmonic, the shaft's wobbling is detected by several multiple harmonics of the rotating frequency with the multiplication coefficient up to fifth-seventh order.

8.6. The Peculiarities of Diagnostics of Impellers

Such units in rotating machines as impellers in the pumps, turbines, compressors and fans require detailed diagnostics. As it was mentioned earlier their diagnostics is based on the analysis of pressure pulsation in liquid (gas) or machine's case vibration excited by these pulsation.

The main method of impellers' diagnostics is the analysis of envelope spectra of high frequency random vibration and noise. By this analysis are detected such defects as the wobbling of impeller, the defects of separate blades, increased turbulence of the flow and the cavitation in the liquid. But to do this it is necessary to measure the envelope spectrum of the case in the nearest adjacency of the blades of the diagnosed impeller. This problem can be solved easily in the pumps and compressors, but for the turbines it is a complicated problem as usually the case of a turbine has a thermoinsulating casing which is very difficult to take off. That is why the place where to fix the transducer must be prepared during a maintenance when the casing is dismounted.

The certainty of the impeller's diagnostics can be significantly increased if besides the envelope spectrum of the impeller the spectrum of low frequency vibration of the machine case or bearing shield is measured. In this case the severity of the defect can be defined much more accurate by the analysis of blade vibration components and the vibration at the rotating frequency and its multiple harmonics.

The diagnostic symptom of faulted blades is the growth of the low frequency vibration at the harmonic frequencies k*RPM/60 (that is one of the symptoms of the impeller wobbling), accomponied by the growth of harmonics at the same frequencies in the envelope spectrum, and the levels of these harmonics can exceed the levels of the blade harmonics. In the spectrum and the envelope spectrum an appearence of side band components for the blade harmonics are possible. The main difference between the symptoms of defects on the blades and the impeller wobbling is considerred to be the mainly growth of the harmonics on the frequency k*RPM/60 in the envelope spectrum of the case vibration, and if the machine has several impellers, then there has to be mainly the growth of the harmonics k*RPM/60 in the envelope spectrum of the machine's case vibration near the faulted impeller, but not all the impellers mounted on one shaft. As an example of capabilities to detect and separate the impellers wobbling from the defect of blades, fig. 19 shows the vibration spectra of cases of two multistage compressors one of which has the wobbling of the impeller and the second has faulted blades on one of the wheels of the impeller. In both spectra the symptoms of wobbling and defected wheels can be detected, namely, the side band harmonics near the blade harmonics (the components with the frequencies ). But when there is a shaft's wobbling the side band components are detected simultaneously around the harmonics of both stages, but when the blades have defects, then only the harmonics of the stage with the defective blades have side band components.

Fig.19. The vibration spectra of a multistage compressor
(a) with the shaft wobbling and (b) with a defect of a blade.

If in the impeller there is only one wobbling wheel then the identification of a defect will be much more complicated. In this case the decision must be made using the measurement of envelope spectrum of high frequency random vibration. Fig.20 shows the envelope vibration spectra of the same stages of the two compressors the vibration spectra of which are shown on fig.19. In the envelope spectra of vibration, when there are defects of blades, the number of components at frequencies k*RPM/60 and their levels are much higher than for the impeller wobbling.

Fig.20. Envelope spectra of vibration of a multistage compressor
a) with impeller wheel wobbling and b) with a defect of a blade.

9. Vibration Diagnostic Systems

Nowadays in the world practice three main types of machine condition monitoring and diagnostic systems by vibration are used:

Machine condition monitoring system of any type must have four main subsystems: If machine condition monitoring system must fulfill also the diagnostic functions then it should include a subsystem that generates the diagnostic decisions. It can be: The main feature of a portable monitoring and diagnostic system is the work of an operator that includes the connection of the primary vibration transducers to the portable instruments that measures and analyzes the signal. This work is made manually and before each measurement.

When you use a portable system usually it is necessary to fix the transducers on the object for the time of measurement. But for the places where there is no access the transducers can be mounted stationary and the communication lines wiring can be made to the place where the operator has an access. The advantage of a portable system is in the possibility to enlarge the number of points where the vibration is controlled up to the necessary number (on each machine unit) by enlarging the time intervals between the measurements. These intervals are defined by the time period of the long term condition prediction and usually includes several days or weeks.

A typical portable monitoring and diagnostic system is show on the left part of Fig.21.

Fig.21. A portable monitoring and diagnostic system on the basis of a digital instrument for vibration signal measurement and analysis with a possibility of its expansion up to a stand or a simplest on-line system.

If the portable systems fulfill only monitoring functions they are not efficient and they have to be substituted by on-line systems capable to measure the vibration practically constantly.

The distinction of a stand system for monitoring and diagnostics is a group of transducers that are mounted in all necessary points to control the machine during the follow-up or concurrent measurements and analysis of diagnostic signals by means of algorithms and commands selected by the operator.

As a rule the stand systems are used during machine testing, for example, for product inspection or for periodical testing of machines that are mounted for this on special stands. In the last case the machine condition monitoring and diagnostic system is an integrated part of a stand and the diagnostics is conducted by modeling a standard for a group of similar machines.

A typical stand monitoring and diagnostic system that can be made on the basis of a portable system is shown on the right part of Fig.21. By the hardware components it does not differ from an on-line monitoring and diagnostic system.

What are the peculiarities of on-line monitoring systems? (See Fig.22)

Fig.22. A structure of an on-line monitoring and diagnostic system consisting of a central diagnostic station - a computer with circuit boards to convert the signal into digital form, and application programs to control, analyze the signals, monitor and diagnose, to display the machine condition, with external block of amplification and commutation, vibration transducers and rotating speed meter.

The first peculiarity is the automatic operation of all the system, including the planning of measurements. All the measurements, signal analizing and all further operations of monitoring are made automatically.

The second peculiarity - in the system there can be a lack of transducers to measure the vibration of each unit. In this case the on-line system can have no subsystem for detailed diagnostics.

The third peculiarity - in the configuration of an on-line system can be included a portable instrument to make additional (to the mandatory monitoring vibration measurements) measurements of other signals. In this case the system can be capable to provide detailed machine condition diagnostics.

What from is it expedient to begin the implementation of diagnostic systems on a plant? Best of all to divide this job into three-four stages.

On the first stage it is better to try to achieve economic efficiency with minimum costs. To do this it is necessary to purchase a portable system for automatic diagnostics with high efficiency (two-three dozens machines a day) and begin the diagnostics of ancillary equipment. In two-three months such system enables to transfer the maintenance of two-three hundred machines according to its real condition, decreasing the accident faults in dozens of times. It fully pays off all the costs and during this time the operators gain experience in working with the system.

On the second stage it is necessary to solve two main problems - to prepare the specialists and to mount the vibration transducers in the points of the diagnosed machines that have no access for the operator.

On the next stage it is expedient to increase the number of portable systems up to optimal number (one system for 100-400 machines with the number of measurement points 2000-5000), and only after this to begin implementation of on-line condition monitoring and diagnostic systems. The latter can be explained by the fact that if the cost of a portable system does not exceed 1-2% of the cost of machines diagnosed and about the same sum is payed for the installation of the transducers in the inaccessible points, the cost of on-line monitoring system can reach 10-20% of the cost of the machine. Just for this reason the on-line monitoring systems are most frequently installed on the most vital equipment.

Summary

For operator's efficient work with a monitoring and diagnostic system he must be appropriately prepared and be able to address the user's support group of the enterprise that designed the system at any moment. How is this job organized by VAST, Inc.?
  1. While purchasing the system the user is delivered by the primary 2-3 days training and receives a set of operation documentation including the methodological manual concerning the monitoring and diagnostics. Besides this the application software includes the tutors that give the operator all the necessary data about the symptoms and parameters used for conducting each certain diagnose.
  2. After a little work with the system on his enterprise the user can take part in the seminars, organized by VAST, Inc. in cooperation with different educational institutions 2-3 a year, where the users can exchange their experience. In the frame of such seminars a week course is organized where the participant can receive a certificate. These courses provide the participants with special methodical and educational literature.
  3. The monitoring and diagnostic systems supplied are provided with means of electronic communication (option) with the support group that can help the user in complicated cases to verify the diagnosis and condition prediction made automatically.
  4. And finally a Club and Forum is organized by the support group in the Internet where each user can address his question and get a detailed answer.
  5. Both the hardware and the software for machine condition monitoring are continuously updated. During our supervision (two years for the software) the user is provided with free modified versions of his software by the electronic communication lines. To upgrade the hardware free of charge during our supervision (one year for the hardware) the user must deliver it in the support group at VAST, Inc.
  6. After the warranty the user can receive the necessary hardware and software to expand the capability of his system with large discounts.

Bibliography

See Russian Pages

  1. Вибрация и вибродиагностика судового электрооборудования./ Александров А.А., Барков А.В., Баркова Н.А., Шафранский В.А.,/ Л., Издательство Судостроение, 1986.
  2. Барков А.В., Баркова Н.А., Азовцев А.Ю. Мониторинг и диагностика роторных машин по вибрации. Рекомендации для пользователей систем диагностики. Издательство СПбГМТУ, Санкт-Петербург, 2000.

Back to articles/notes page

Back to home page VAST and VibroTek, Inc.