|
|
|
|
|
AN OVERVIEW OF CONDITION MONITORING WITH EMPHASIS TOWARDSONE OF THE MOST POWERFUL OF APPLICATIONS:VIBRATION MONITORINGCondition MonitoringWhat is condition monitoring ? An incremental measure of a machines ability to safely and efficiently perform its predefined duty, enabling fault and failure prediction, to assist formulation of a planned maintenance schedule. A variety of techniques may be employed to achieve the desired results, including; Oil Debris Analysis (Tribology), Temperature Measurement (Thermography), Machine Performance \ Efficiency, Acoustic Emissions However the technique noted for its overall ability is that of Vibration Monitoring. This being one of the methods employed, by Proviso Systems Limited, to provide Engineers with a comprehensive report of a machine's condition. Condition based maintenance objectives
Vibration MonitoringPreface. With today's highly mechanised industries comes the greater need for machine maintenance. Three common methods of approach to achieve this are as follows:- 1. Breakdown Maintenance Simply, if it breaks, fix it. A method commonly employed for duplicate machines or non critical plant. 2. Planned Preventative Maintenance Periodically carry out pre-scheduled tests, examinations and replacements to reduce the possibility of unforeseen defects and downtime. 3. Predictive Maintenance Employs any of a number of methods of testing a machines operating condition, enabling trending of levels and hence rates of deterioration. This predictive technique may then be used to govern any further maintenance action. Ideally, a combination of these methods would be employed to ensure both cost effective machine operation and maximum production capabilities. Vibration Monitoring is often used as a versatile maintenance tool for both planned and predictive maintenance. Its implementation may even be useful on a defective machine to assist in fault identification. With the influence of technology, vibration monitoring systems are readily available in integrated packages designed for all types of application, from a multiplexed fixed installation to the simpler hand held units. The utilisation of any system is dependent upon such factors as; location; practicality; effectiveness; and cost. One factor has remained consistent throughout the technological development of systems; namely, the vibration signal generated as a machines' components rotate. Whether an engineer uses a screwdriver, to his ear, to audibly detect the impacts, or an accelerometer to electronically measure their frequency and amplitude, the signals are actually identical. Technology now enables this signal to be registered, stored, and downloaded to a p.c. for analysis. To assist in the understanding of the vibration signal it is essential to understand the machinery on test and the parameters by which its performance can be monitored. Principles Behind the Science Vibrations may be defined as the oscillatory, or cyclic motion of a solid body, about some equilibrium point when excited by an appropriate force. In simple terms, take the example of Fig.1 A thin beam clamped to the edge of a bench illustrates what might occur when such a body is excited by an impulse. The practical example may be a ruler clamped to the edge of a table. Fig.1
In vibration of this nature, the magnitude of oscillation will tend to die down until the beam is at rest. This being termed as "damped oscillation". When first excited the vibrations are characterised by two specific features. These are: i) the amplitude of vibration D ii) the frequency of vibration where f = 1/T We can observe these features during the vibrations. The amplitude D, is the maximum distance moved during one vibration cycle, i.e. the Peak to Peak movement. This peak to peak movement is known as "displacement" and can be expressed as a linear measurement. Depending on the amplitude its magnitude may be expressed directly in mm or microns. The level is dependant on the excitation. In this case the vibrations die down quickly, owing to the natural damping of the beam. The second characteristic we can observe is the frequency of the oscillations. That is the number of oscillations per second (Hz). The frequency of oscillations is a function of the size, weight, and stiffness of the beam. Hence vibrations may be specified in terms of both frequency and amplitude. The frequency tells us what is vibrating. The amplitude tells us how bad it is vibrating. The vibration amplitude, in this example, has been specified as a distance moved, i.e. "displacement". The vibration can, however, be expressed in other ways; namely, Velocity of vibration, and Acceleration of vibration. These terms are often used in practice, and it is important to understand these concepts of velocity and acceleration of vibration. In our example, the 'displacement' may be observed by the eye, and consequently is easy to comprehend. However, any moving object has a velocity. It will therefore be appreciated that the cyclic motion of the beam must have a velocity of vibration (measured in mm per second). Velocity of vibration is a function of displacement, and frequency. In a similar fashion any moving object which changes direction (as described by the cyclic motion of the beam), is said to have acceleration of vibration ( measured in m\sec\sec, or 'g') where 'g' is the unit of gravity. The acceleration level reaches a maximum at the two extremes of the beams movement. So it can be seen that for any single vibrating source, the vibration may be specified in terms of either its displacement, velocity or acceleration levels, together with its frequency. Referring to the case of the beam, the equivalent displacement, velocity and acceleration levels might be as shown in Fig.2
With reference to Fig.2, a vibration would be expressed in terms of its most predominant feature, depending upon its frequency of vibration. In physical terms, it is convenient to think of the following as: Displacement : the peak to peak movement Velocity : the energy content of the vibration Acceleration : the force of the vibration. To summarise then; Low frequency vibrations give high displacement levels (0 - 50Hz). Low to medium frequency vibrations give high velocity levels (10 - 1kHz). Medium to high frequency vibrations give high acceleration levels (300 - 15kHz). In practice the is some overlap to the general frequency bands. The Vibration Transducer (ACCELEROMETER). Data capture regarding the vibration emitted by a machine, or other body, begins with the sensor. Simply, this may well consist of a piezoelectric crystal which has a mass attached to one of its surfaces. When the mass is subjected to a vibration signal, the mass converts the vibration (acceleration) to a force, this then being converted to an electrical signal representative of the incoming vibration signal. This is the basis of the ACCELEROMETER. The accelerometer output may then be processed to provide the instantaneous velocity and displacement signals. Figure 3.
Time domain frequency signals The vibration pattern, or signature, produced by a typical machine is normally complex. A typical wave form might be as shown in Figure 4.. This complex wave form is not easy to analyse, being a mixture of all the vibration components generated by the machine. Namely gears, bearings and other moving parts in addition to it supporting structure. Further analysis of this time domain wave form, indicates the running condition of the machine. The complex time domain wave form is a composition of all the numerous individual frequency components, and by extracting the individual frequency components from the wave form, it is possible to identify what part of the machine is generating what frequency. Then, by measuring the amplitude of the individual frequencies, assess the condition of the machine. This process of extracting the individual frequency components is complicated, it is, however, the basis of vibration analysis and can be illustrated by the following example. Consider the following complex time wave form, simply comprised of three fundamental component frequencies. Figure 4.
The wave form is represented in the 'time domain' and is often referred to as a 'time trace'. If we separate the individual frequencies present in the time domain and plot them on a third axis, the result would be as shown in Figure 5.
An observer viewing the two wave forms on the amplitude \ frequency plane would see the separate frequencies as shown The data has now been translated into what is known as a vibration spectrum with the vibration amplitude being shown in the frequency domain. This form of data representation is one of the most useful tools utilised to analyse complex signals. The mathematics involved in conversion of time domain to frequency domain data is complex, however, today's processors handle these mathematics with ease. The complex conversion involves the use of F.F.Ts. or Fast Fourier Transforms.
Figure 6.
Getting Started Choice of software and hardware Various software packages are available, several used by Proviso Systems, in addition to a variety of data collectors. In essence the software must provide an easy to use system for both database set-up, subsequent analysis and further management of the collected data. In addition to a comprehensive self reporting module. The hardware or data collector should also be easy to use and capable of communicating with the chosen software without the loss of any of the collectors facilities. The full specification of the systems available are published by the manufacturer and should be obtainable from the supplier. Which items should be monitored.? The criticality of the machine will help to identify its priority for monitoring. By answering the following questions, each of the machines can be allotted a "critical value"; 1. If this machine were to fail unexpectedly, how detrimental would this be with regard to, lost production, time and costing.? 2. Is this a single line component.? 3. Is there a replacement or spares readily available.? 4. Does this machine have a history of problems.? 5. What is the age of the machine.? 6. Is there any current form of condition monitoring being used and has it been effective.? Having decided upon the items of equipment to monitor, several questions should be answered with regard to each item, these are as follows; 1 Machine Name - Plant No. 2 General Arrangement Diagram 3 Component details - i.e. bearing Nos., gear types and teeth Nos. 4 Speeds, reductions and ratios 5 Lubrication method and type. Database Compilation. When the machine information has been correlated, a comprehensive database may be compiled in order to establish a structure (database) within which all data is stored, see the following screen shot.
The aim of the database is to electronically reproduce a model of the machine within the software, from which possible fault conditions may be calculated. From this database the data acquisition parameters may be defined. (These being the key areas of data which should be most closely monitored as areas of concern.) The resultant database will hold information, relevant to every test point on the machine, with regard to the frequency spectrum bandwidth, spectrum definition (lines of resolution), analysis parameters (areas of concern), number of averages to be taken and any relevant alarm levels. From within the database a LIST is created. This enables communications between the data collector and the p.c. for both uploading machine test points and down loading point data. When the data collector has been loaded with the route, data can then be collected from the machines to be tested. Figure 7. Test Repeatability and Accelerometer Interface Node Fixing.
Once the machine has been selected for monitoring, the machine is assessed for practical measurement locations. Measurement position is critical in all vibration monitoring systems, data being acquired from principle bearing housings in a vertical, horizontal and axial plane, where practical. The accelerometer or interface node should be fitted in the loaded region of the bearing to ensure maximum signal strength. Each node position is individually identified, for easy location and reference, e.g. R1 = motor none dive end (NDE) bearing, R2 = motor drive end (DE) bearing. This is shown in Figure 7..
Figure 8. Repeatability is also a critical point. On fixed installations the accelerometer can be fitted direct to the machine surface using a threaded stud mount. On the portable systems, this is achieved by fixing accelerometer interface nodes to the selected measurement position. The interface node ensures that each measurement is repeated with regards to position, pressure and surface area when an accelerometer, with a bayonet or screw fixing, is attached. See figure 8. Steps should also be taken where possible to ensure each test is made under similar conditions with regard to temperature, load, speed and lubricant levels. Faults and Fault Calculation. The key to fault frequency calculations is simply the number of tones generated in a one second intervals (frequency), which would be generated due to a known effect or defect on that particular machine. Since the tones are calculated during one second samples the units are measured in HERTZ (Hz), cycles per second, commonly known as frequency. i.e. each fault will appear at a specific frequency. The fault will be exhibited as a peak, of an amplitude "y" at a frequency "x". The amplitude of this peak may be expressed in several units, these being; MICRONS - Displacement. MM\SEC - Velocity. Gs - Acceleration. All of which are recognised units of vibration quantification. All faults are in some way related to the rotational speed of the shaft on which that fault would be generated and since the faults are presented in Hz, cycles per second, it is generally considered that the shaft speed is converted from revolutions per minute (rpm) to revolutions per second (rps). This is achieved simply by dividing the shaft rpm by 60 seconds. For example a motor rotating at 1500rpm will have a rotational frequency of; 1500rpm = 25 Hz (cycles per second). 60seconds The rotational frequency may now be related to a common base, known as ORDERS, one order being the speed of the specific shaft in Hz. Two orders would then equate to 2x the rotational frequency. In the case of the 1500 rpm motor shaft, this would be; 2 x 25Hz = 50Hz, etc., etc. Several vibration fault characteristics are found at specific frequencies related to the shaft speed and hence are easily targeted in orders of that speed. These more commonly being; IMBALANCE at 1 Order. INSECURITY at 2 Orders MISALIGNMENT at 2, 3, and 4 Orders. NOTE - Reference to the included typical spectrum, Figure 9, will assist in the understanding of these and the following fault frequencies.
Gear Defects Assume a single reduction gearbox with a reduction of 27 teeth on the input shaft pinion meshing with 54 teeth on the output shaft gear, a ratio of 2:1. The gearbox is driven by the 1500rpm motor, mentioned previously. The fundamental gearmesh frequency is calculated by ; Shaft Rotational Frequency x Number Of Teeth On Gear. 25 Hz x 27 teeth = 675 Hz This is also equivalent to 27 ORDERS of the fundamental shaft speed and may be classed as 1 GEARMESH ORDER. The gearmesh frequency may also exhibit upper and lower sidebands, these being at plus and minus 1 Order of the shaft rotational frequency from the gearmesh frequency. i.e. 675 Hz + 25 Hz and 675 Hz - 25 Hz 650 Hz and 700 Hz. Variation in either fundamental, sidebands and \ or harmonics of the gearmesh frequency will highlight various gear defects. Figure 10
Bearing Defects. A rolling element bearing will generate four fundamental fault frequencies. These may be calculated from details of the bearing dimensions, number of elements, contact angle and the speed at which it rotates. Again each of these four fundamental fault frequencies may also exhibit harmonics (multiples of the fundamental) and sideband activity, dependant upon the severity of the fault. These defect frequencies are related to ;
Figure 11 To assist in frequency calculation, bearing defect frequencies are held in a parts list, within the software, along with a gear frequency calculator and many other fault finding aids to assist in database compilation. In addition to the previously highlighted defects, a bearing suffering from initial deterioration and/or initial lubrication deficiency, as opposed to a distinct defect, would be seen to generate an increase in high frequency activity. This being indicative of an increase in metal to metal contact between the rotating elements. This would be indicated on the spectrum by an evident increase in baseline activity as the frequencies of the impacts becomes broadband. This is generally identified by spectral baseline lift. To assist in the monitoring of this area of the frequency spectrum the PL31 / PL33 and Dataline data collectors used, utilise a number of high bandpass filters, namely :-
The PL31/33 and Dataline Data Collectors. Proviso Systems have wide experience of both types of meter in many industrial applications including Water Authorities, Mineral Extraction, Confectionery, Manufacturing, Chemical and Coking Plants and Quarry applications. The latest addition to the meter range is the Dataline DSP (Digital Signal Processing), this has been designed to improve the data collection time for high resolution data collection and slow speed applications. Once the relevant component fault frequencies have been identified and the database compiled, the data collector may be loaded with a list in preparation for data acquisition at the remote site. Each measurement point on the meter having been pre programmed with the relevant information will save the spectrum and trend data and then proceed to the next list location. In addition to the pre programmed parameters, an OVERALL vibration level is always recorded with each spectra. Since the data collector has been pre programmed, any of the parameters measured which are considered to be in an alarm status, are identified by an "Alarm" message on the data collector screen. The dataline data collector will also alarm on any possible cable faults with the message "Accel not connected". On returning to the host computer the data is downloaded and all measurement points, shown to be in alarm on the data collector, are further identified for full analysis. Trend Analysis. NOTE - Reference to the included trend plot, Figure 12, will assist in the understanding of trend analysis and its associated alarm levels. Trend data or single value data is an excellent method of monitoring specific frequencies without spending too much time analysing complicated spectral information. Since the trend can be programmed to look at the same bandwidths as the spectral data, it is easy to use the trend data as a filter for your fault finding analysis, i.e., if the trend data is not in alarm there is unlikely to be anything to analyse on the spectra this is assuming your trends are set to mirror your spectra. Each of the targeted trend parameters are individually updated by the software following each data download. This data is presented in the form of a plot showing parameter amplitude against time. Figure 12 A High Frequency Bearing Fault Trend (G's)
Before
After
A Trend of the 30 Order Band (mm/sec)
Before
After
Fig 13.
|
|
Send mail to
dave@proviso-systems.co.uk with
questions or comments about this web site.
|
