Wear Force on Metal Surfaces - Essay Example

THE DIFFERENCES BETWEEN ENVELOPING AND PEAKVUE By: Seyed Mirhadizadeh Presented 4th January Introduction: The forces of wear and tear act slowly on the metal surfaces and it is sometimes impossible to detect the point of the fault or even know whether a fault really exists. Physical observation and sound detection are the two physical means of detecting the existence of faults in bearing systems. However, other technological ways of detecting the existence of faults on bearings have been developed over time. Some of the methods include: airborne ultrasound (acoustic emission), shock pulse method, shock energy, enveloping and demodulation (which describe essentially the same process), and peakvue. This paper seeks to concentrate on the working differences between enveloping and peakvue, and further explain their functionality. Peakvue Vibration signals from working machinery can be described as complex signals made up of a combination of sinusoidal waveform of uneven frequency, amplitudes and phase differences which can all be traced to the fundamental rotational speed. Peakvue as a technique captures the peak values of the stress waves that are produced and then remits the repetition frequency of the impacts via a spectral analysis. For a professional using peakvue, the most important factors are the peak values. These values are captured as resonation of stress waves. Due to its high precision, the method is able to detect the emission of stress waves caused by direct contact of adjacent metallic surfaces at an early stage. Generally, through spectral analysis the frequency of repetition of the stress waves is availed. By peakvue, the resonance zones are isolated through use of filters. Capturing the peak values for particular intervals of the selected sampling time is done through the application of high frequency clustering of signals with over one hundred kilo Heartz. The method is the trade mark of Computer System Incorporation. The method takes the analyst through four stages The initial stage in which the low frequency signals are eliminated by taking the entire signal through a high pass filter. The amplitudes are run through the accelerometer which detects the measurements and is able to classify them according to the initially specified cut-off frequency level. All the measurements reading below the designated cut-off level are classified together. They are effectively eliminated which implies that only the high frequency readings- those above the cut-off level, proceed to the second phase of the analysis procedure. By elimination it means that such values are truncated from the recorded measurements such that their consideration ends at that phase. The second phase that involves the digital conversion of frequency. The high frequency signal is changed from its initial analog form to digital form in order to begin the analysis phase. Normally, high frequency values will be recorded for particular sections along the surface of the bearing. Therefore whenever the metallic surfaces come into contact during the rotation, the frequencies hit a peak. If special readings were availed at phase two for any specific time duration, it will be analyzed here. If for specified time duration the amplitude levels of the converted signals read beyond a predefined threshold it is then matched to a digital value. The essence of this matching is that it is not always possible to produce signals with equal frequencies, even when the components are running on a fairly flat or even ground. This could be the result of instantaneous change in the position of the load, continued distribution of the lubricant along the colliding surfaces, and varying positioning of the emerging fault. Therefore, for a analyst, the simplest way to conclude that the frequencies were derived from a ‘certain’ spot along the surface of the bearing is to cluster all measurements within a certain range. These measurements clustered together are then assigned a specific digital value, which identifies them together. The display is rendered once the digital value corresponding to the particular time interval is processed using the Fast Fourier Transform (FFT) algorithm. The FFT algorithm works out the Discreet Fourier Transform and the corresponding inverses. To obtain the Discrete Fourier Transform, a sequence of amplitude values is decomposed to form components of varying frequencies. The frequencies so formed are categorized according to their closeness, that is, according to pre-determined intervals. The classification is done with reference to how often a specific range acquires numerical frequency. For example if the outlier frequency is too scarce, the measurements thus classified can be overlooked and ignored. If there is higher consistency in the numerical frequency of measurements within a certain range, they will be passed as valid and analyzed. The level of the considered frequencies may determine the level of damage, whether visible or not. The second most important part of the analysis is to establish the part of the bearing that bears the fault. Frequencies of similar classification earlier assigned the same digital value are matched to a section of the bearing as they were recorded during rotation of the components. This means that if multiple faults occur on the same bearing, they could easily be classified according to the differences of their assigned digital values. However, it may be difficult to label different parts of the bearing as faulty when their respective levels of damage are relatively equal, since there will hardly exist any distinguishing factor in their recorded frequencies- meaning they occupy the same digital value. The peakvue method samples the unconverted analog signal at a relatively high frequency so as to capture the short term stress waves. The wave form and the resultant spectrum are viewed by the analyst and using a peak hold algorithm, the wave form retains the peak levels making it trend-able. Peakvue requires the use of arbitrary values for the cut–off frequency which may not be necessarily accurate. If for instance the cut-off level is set too high, the analyst may fail to detect any shortcomings in the performance of the bearings. This could in effect lead to an emergency failure that the system is otherwise aimed at guarding against. On the other hand, if the cut-off frequency is set too low, it may end up sounding alarm unnecessarily. This may cause the department or the user significant investment in replacing the still safely functioning bearings as well as variable production time. Enveloping This process is also called enveloping and demodulation. Two functions are performed using this process. The two are: Extraction of the high amplitude. For the high amplitude to be extracted the frequency runs through the accelerometer and the low and high frequencies are separated. The mechanism works much similar to the peakvue, except that a standard cut-off value is not as much fixed as in peakvue. This means that the analyst can set the cut-off value at own standard and continue to run the system so as to detect any excessive readings as alerts for possible bearing faults. Delimitation to low frequency signals of the frequency that earlier had high projections; the mathematical amplitude of the frequency. This again involves the classification of the frequencies according to the signal amplitude, which in effect sees fairly similarly peaked amplitude values being clustered together. Clustering together of the waves means that those that have close values are classified and recorded together, while those that occurred so rarely so as to fail pairing with any other readings are truncated and treated as erroneous measurements. In the envelope method, spectrum will feature specially average frequencies whenever there do not exist faults. On the other hand if a fault for instance an internal crack exists, peaks or sharp rise in readings will be seen to read on the accelerometer. With subsequent development of the fault the amplitude of the previously seen peaks will increase again and the normal frequency readings will rise to the same levels as the existing peaks. The use of enveloping widely finds ground in fault diagnosis of rolling elements. This method makes use of the high frequency section of the spectrum. When an object comes into contact with a bearing with a spall kind of defect the amplitude rises and is recorded on the accelerometer as a peak over the duration that the object passes through the defective section of the bearing. In order to understand the differences between peakvue and enveloping methods of bearing defect detection it is important to understand the stages of a bearing failure. Stage One At this stage, the bearing will mostly have suffered internal damage which may be difficult to see even when conducting a physical examination of its condition. Since when using the advanced fault detection methods including peakvue and enveloping it is possible to detect a fault at this stage, it is advised that a constant eye is kept over the bearing as well as constantly lubricating it, and checking on its stability and alignment. Since while using enveloping it is possible to lower the frequency cut-off reading to as low as the analyst intends to determine the accuracy of their readings, it is advisable that it is employed for this initial stage. The most obvious distinguishing factor between a damaged and a bearing in good working condition is lack of harmonics. For a damaged bearing, frequencies of the harmonics begin to show considerably. When using enveloping, it is necessary to set the cut-off mark, usually written Fmax, so as to record any harmonics and teir numerical frequency. Stage Two At this stage it is recommended that the bearing is watched more closely than when at the first stage. The internal crack will perhaps be showing on the surface and such signs as flakes or cracks will be evident. Both peakvue and enveloping will still be effective tools of checking the faults. Stage Three At this phase it is only recommended that you continue running the machine on the faulty bearing if the failure costs are effectively offset by the need to continue running the same. Otherwise it is time to remove the already visibly damaged bearing. The peakvue readings start to even out at this stage, which makes it relatively impossible to track the exact part that is faulty. Due to its ability to pick out individual harmonics, enveloping remains a strong tool of measurement. For peakvue, the harmonics will be seen as a cluster of emergent peaks that crowd a section of the frequency domain. Stage Four The bearing has already suffered substantial damage at this level. Continued use of the bearings could result in a catastrophic failure. Peakvue takes a break when the measurements to be taken are at this advanced level since it becomes no longer tenable to establish the real position of the fault using the recorded frequencies. However, enveloping is still a strong tool because it is able to isolate individual harmonics, which enables the analyst to distinguish the damaged and undamaged parts of the bearing. Bibliography: Estupinan, E., Saavedra, P. (n.d.). Diagnostic Techniques for the Vibration Analysis of Bearings. Web. Liu, F. (2012). Recorded Webinar: Signal Analysis and Measurement Techniques in MATLAB. Web, 4th January 2013. MECH7350; Rotating Machinery. (n.d.). Analysis Techniques. Analysis of vibration Signal. Web. Mobius Institute. (n.d.). Detecting Rolling Element Faults with Vibration Analysis. Vibration Analysis Training and Certification. ISO 18436 parts 1,2 & 3 Web, 3rd January 2013. Tranter, J. (2011). Detecting Bearing Faults. Condition Monitoring: Vibration. Web. Read More

If special readings were availed at phase two for any specific time duration, it will be analyzed here. If for specified time duration the amplitude levels of the converted signals read beyond a predefined threshold it is then matched to a digital value. The essence of this matching is that it is not always possible to produce signals with equal frequencies, even when the components are running on a fairly flat or even ground. This could be the result of instantaneous change in the position of the load, continued distribution of the lubricant along the colliding surfaces, and varying positioning of the emerging fault.

Therefore, for a analyst, the simplest way to conclude that the frequencies were derived from a ‘certain’ spot along the surface of the bearing is to cluster all measurements within a certain range. These measurements clustered together are then assigned a specific digital value, which identifies them together. The display is rendered once the digital value corresponding to the particular time interval is processed using the Fast Fourier Transform (FFT) algorithm. The FFT algorithm works out the Discreet Fourier Transform and the corresponding inverses.

To obtain the Discrete Fourier Transform, a sequence of amplitude values is decomposed to form components of varying frequencies. The frequencies so formed are categorized according to their closeness, that is, according to pre-determined intervals. The classification is done with reference to how often a specific range acquires numerical frequency. For example if the outlier frequency is too scarce, the measurements thus classified can be overlooked and ignored. If there is higher consistency in the numerical frequency of measurements within a certain range, they will be passed as valid and analyzed.

The level of the considered frequencies may determine the level of damage, whether visible or not. The second most important part of the analysis is to establish the part of the bearing that bears the fault. Frequencies of similar classification earlier assigned the same digital value are matched to a section of the bearing as they were recorded during rotation of the components. This means that if multiple faults occur on the same bearing, they could easily be classified according to the differences of their assigned digital values.

However, it may be difficult to label different parts of the bearing as faulty when their respective levels of damage are relatively equal, since there will hardly exist any distinguishing factor in their recorded frequencies- meaning they occupy the same digital value. The peakvue method samples the unconverted analog signal at a relatively high frequency so as to capture the short term stress waves. The wave form and the resultant spectrum are viewed by the analyst and using a peak hold algorithm, the wave form retains the peak levels making it trend-able.

Peakvue requires the use of arbitrary values for the cut–off frequency which may not be necessarily accurate. If for instance the cut-off level is set too high, the analyst may fail to detect any shortcomings in the performance of the bearings. This could in effect lead to an emergency failure that the system is otherwise aimed at guarding against. On the other hand, if the cut-off frequency is set too low, it may end up sounding alarm unnecessarily. This may cause the department or the user significant investment in replacing the still safely functioning bearings as well as variable production time.

Enveloping This process is also called enveloping and demodulation. Two functions are performed using this process. The two are: Extraction of the high amplitude. For the high amplitude to be extracted the frequency runs through the accelerometer and the low and high frequencies are separated. The mechanism works much similar to the peakvue, except that a standard cut-off value is not as much fixed as in peakvue. This means that the analyst can set the cut-off value at own standard and continue to run the system so as to detect any excessive readings as alerts for possible bearing faults.

Read More