Practical Solutions to Machinery and Maintenance Vibration Problems http://updateinternational. http://updateint ernational.com/Book/Vibrati com/Book/VibrationBook8a.ht onBook8a.htm m
Table of Contents Page No. Section 1, Vibration Due to Rolling Element Bearings
2
Section 2, High Frequency/Ultraso Frequency/Ultrasonic nic Measurements
3
Section 3, Considerations for Evaluating Amplitudes Due to Bearing Defects
5
Section 4, Importance of Pickup Mounting and Location
7
Section 5, Use of of IBF Units to Accurately Determine Bearing Bearing Lubrication 8 Section 5, Use of of IBF Units to Accurately Accurately Determine Bearing Bearing Lubrication Lubrication 9 Section 7, Calculations for Bearing Defect Frequencies
12
Section 8, Further Considerations for Bearing Defect Vibration Amplitude and Frequency Symptoms
15
Section 9, Estimating Remaining Bearing Life
16
Section 10, Vibration Symptoms of a Loose Bearing
17
Section 11, Further Considerations Regarding Calculated Bearing Defect Frequencies
19
Section 12, Vibration Due to Oil Whirl (in Plain or Sleeve Bearings)
21
1
Section 1, Vibration Due to Rolling Element Bearings
One of the first points to make regarding vibration due to rolling element bearings is that generally speaking, the bearing is not a source of mechanical trouble but merely a result of other problems. Most often, when symptoms of bearing defects arise, they are accompanied by other symptoms, such as for misalignment, unbalance or lubrication. When dealing with rolling element bearing analysis, a good rule is to not only analyze the symptoms of bearing defects but to also determine why the bearing is defective. Look for symptoms of such sources as unbalance, misalignment, poor assembly, etc. There are many different symptoms of distressed rolling element bearings, and much good information has been published on the subject. It is, therefore, not the purpose of this this section section to supersede supersede information information that that is likely likely already known, but instead to highlight some of its more practical applications and strengthen certain portions.
Figure 1: Major parts of a rolling element bearing
2
Section 2, High Frequency/Ultrasonic Measurements
One of the most difficult Bearing problems problems to identify identify is incipient failure failure (failure that is just about to happen). Vibration velocity readings help us to identify bearing defects in the latter stages of bearing life. However, in order to get the earliest warning of impending trouble, it is necessary to monitor the ultrasonic vibration levels. Most instruments now have the capability to measure very high frequency vibration in the 5 kHz - 50 kHz range, (300 Kcpm - 3000 Kcpm). The terminology for the particular unit of measurement varies from manufacturer manufacturer to manufacturer, manufacturer, such as "spike energy," shock pulse," "bearing defect energy," "high frequency detection," etc., etc. However, they all measure pretty much the same thing. The instrument manufacturers will be able to furnish a detailed definition and description description of their specific specific units of measurement. For the purposes of this text, general terms will be used. These high frequency measurements sense the low energy, repetitive, metal-tometal impacts that occur in the earliest stages of an incipient bearing failure. The rate, amplitude and frequency are all combined to give a single "numerical" output. To maintain impartiality, this text will identify all these units with the generic term IBF (Incipient Bearing Failure) number. The problem with all these units is that it is almost impossible to accurately determine bearing condition based on a single measurement. Measurements are very sensitive to a host of external influences such as differences in installation, load, lubrication, pickup mounting, location and number of interfaces. Other problems, such such as cavitation, cavitation, gear gear mesh and and steam leaks, leaks, also have have significant significant effects in IBF numbers (i.e. it is possible to install two identical, brand new bearings in two identical identical machines machines and get two different IBF readings). readings). The use of IBF readings excels when trended over a period of time. If used correctly, IBF units will provide the earliest warning system for impending failure. In this manner, change becomes more significant than absolute values. If this technique technique is to be successful, accurate repeatable data has to be acquired. acquired. Of all the vibration parameters measured, IBF units are the most sensitive to transducer location and mounting. Small changes in location and/or mounting techniques can produce large changes in IBF amplitudes, making it very difficult to identify true changes in bearing condition.
3
Fig. 1 shows a typical IBF trend. Notice that the IBF number actually drops either before or as the the bearing bearing fails. If a single single IBF reading reading of, say, 0.5 is obtained, obtained, will will the bearing last another 5 years, 5 months or 5 days? This is not known unless enough plots have been taken to show a trend. It is not recommended that a diagnosis be made based on IBF units alone. Instead, always verify the diagnosis with other vibration units such as velocity. Remember, trending IBF units provides only an early warning system.
4
Section 3, Considerations for Evaluating Amplitudes Due to Bearing Defects
Whether plotted by hand or by computer, trends of IBF units or vibration units (usually velocity) vs calendar dates are certainly valuable for monitoring how fast a bearing may be deteriorating. However, such trends do not always accurately predict the approximate approximate time for for failure. Usually Usually trend plots of IBF units show show increases as the bearing deteriorates, but as actual failure occurs (such as cracking), cracking), the IBF units may show a sharp decrease. When IBF units are still increasing, most careful analysts monitor trends of velocity amplitudes as well. But again, amplitude increase is not always present as a bearing deteriorates. The difficulty is due to the entirely different nature of vibration amplitude increases due to bearing defects, as compared to amplitude increases due to other vibration sources such as unbalance, misalignment, bent shaft and so on. For example, as fly ash builds up on a fan, its unbalance may progressively increase. Increased unbalance shows increased amplitude at 1 x rpm. Or, for example, greater shaft-to-shaft misalignment will give higher amplitudes for its frequencies. Therefore, the analyst is tempted to think that greater defects at the source always produce larger larger amplitudes. amplitudes. True for for almost all all vibration vibration sources. However, not necessarily true for bearing defects. Sometimes bearing defect amplitudes amplitudes increase as the defect gets worse -- but not always. Amplitudes at bearing defect frequencies may not increase much at all and, in some situations, may actually decrease as the bearing gets worse. Instead of the bearing defect defect amplitudes amplitudes increasing, increasing, they they may instead instead increase increase the number number of sidebands. What may have started out as a relatively sharp peak may appear to be spreading out to cover a wider frequency range. As with trends of IBF units, just before or right right at bearing bearing failure, failure, the vibration vibration amplitude amplitude suddenly suddenly decreases. Therefore, trend plots based on amplitude, may or may not give an accurate picture of what is happening in the bearing. Vibration amplitudes measured on the machine's case that are considered as acceptable or even smooth if originated from, for example, unbalance, are not acceptable for amplitudes due to bearing defects. For example, a pump's case reading of 0.1 in/sec or 3 mm/sec is considered acceptable, but if that amplitude originates from a bearing defect, that bearing is usually so bad that the defect is visible to the eye (conversions between English and metric units are approximate). Vibration amplitudes amplitudes at bearing defect frequencies are usually very low. For a "regular speed" machine, such as 1800 rpm or 1500 rpm, a bearing is considered mildly bad at only 0.04 in/sec or 1 mm/sec. Although this amplitude appears mild, the bearing can still immediately fail. For very slow speed machines, such as papermachine papermachine dryer rolls rolls (under 100 to 200 200 rpm), a bearing bearing defect defect velocity velocity amplitude of only 0.02 in/sec or 0.5 mm/sec requires a bearing change.
5
(In Update's course Practical Solutions to Machinery and Maintenance Vibration Problems Part II this subject is covered in considerably more detail and actually takes a full day. However, this textbook is for Practical Solutions to Machinery and Maintenance Vibration Problems Part I and the information provided is sufficient for analysis of bearing defects in about 90 percent of all situations.)
6
Section 4, Importance of Pickup Mounting and Location
In order to accurately monitor and trend IBF units over time, it is necessary to acquire accurate readings. Of all the various vibration units available, IBF units are the most sensitive to pickup location and mounting techniques. Small changes in location and/or mounting techniques can give rise to large changes in IBF amplitude readings, making it very difficult to identify the genuine changes in IBF number. The preferred location is as physically close to the bearing as possible and preferably (but (but not essentially) essentially) in in the "load "load zone." zone." For the usual usual vibration vibration units, such as velocity or displacement, having one to a few interfaces between the pickup location location and and the bearing bearing itself itself is not very very important. important. However, However, for IBF measurements, each interface decreases the accuracy of the reading. Therefore, try to choose a location with the least amount of interfaces between pickup location and the bearing. As with all maintenance work, compromises may be necessary. If so, IBF absolute numbers may have to become secondary to IBF changes from one time to another. For initial maintenance setup on large machines, it may be necessary to review the engineering drawings in order to find an appropriate pickup location that will provide a good transmission transmission path. path. The location location must must be clearly clearly marked on the machine so that subsequent readings will be acquired in the same location. Stud mounting is the best, but obviously more expensive. A practical alternative would be the magnetic holder. Use of an extension probe "stinger" or stem usually does not produce as good results unless of short length and adequately tightened to the pickup. Update also suggests small drill point holes on the case's measuring points.
7
Section 5, Use of IBF Units to Accurately Determine Bearing Lubrication Lubrication
One of the challenges of maintenance departments worldwide is to develop effective lubrication programs for the many grease lubricated rolling element bearings. In In an effort to to ensure that that a bearing bearing is not lost due to lack of of lubrication, lubrication, there is a strong tendency to over-lubricate rolling element bearings. Overlubrication is probably responsible for more bearing failures than underlubrication. There have been several approaches over the years to solve lubrication problems of rolling element element bearings. bearings. Some special special instruments instruments have have built-in computers. Strong claims are made for their effectiveness. Others use certain features of their regular vibration instruments, such as measuring IBF units. However, none of these approaches address the question of how much grease the bearing needs. needs. A fully fully sealed bearing bearing does does not require require regreasing regreasing through through the course of its life, which if correctly installed in a machine that is well balanced and well aligned, can run in excess of 10 years. Why is it, therefore, that relubricating bearings can can be called called for every few months? months? The usual usual answer is, is, "To flush the the bearing of contamina contaminants." nts." However, However, if the bearing bearing lip seals are in in good conditio condition, n, there should be no contaminants and, therefore, no requirement for regular lubrication. This is not a section on bearing lubrication. Its focus is only on how some specialists use IBF units to accurately determine the quality and quantity of bearing lubrication lubrication in rolling element bearings. IBF measurements are made on a regular basis, watching watching for significant significant change. When significant significant change change occurs, occurs, the technician verifies the initial data. Then, while monitoring the IBF reading, grease is applied. After applying two or three pumps of the grease gun, the technician pauses for a few seconds seconds while observing observing the the IBF reading. reading. Typically, Typically, the reading reading decreases. Grease is applied until no further reduction in the IBF reading can be obtained. At that point, the bearing is considered to be properly lubricated. The bearing should should be monitored monitored for for 10 to 15 15 minutes minutes after the process process to ensure ensure that the IBF reading does not increase. If it does, the process is repeated. Sometimes overgreasing overgreasing will show an IBF increase. The bearing should then be checked 24 to 48 hours later to ensure no further increase has occurred. A return to previous levels probably means the bearing is starting to fail. It is possible to temporarily reduce the IBF readings on severely defective bearings by applying grease.
8
Section 6, Use of Velocity Spectra to Determine Bearing Condition
One of the keys to the successful analysis of distressed bearings is pattern recognition. recognition. Since there are many different configurations, configurations, types and sizes of rolling element bearings, it is very difficult to accurately determine one specific frequency and amplitude that will be generated by a bearing defect. Therefore, it is necessary to observe the familiar "patterns" developed by about 80 percent of distressed bearings rather than their absolute amplitudes and frequencies. Stage One: Fig. 2 represents the velocity spectrum for the first stage of a bearing defect. An increase in IBF units has occurred. Note that there is no change on the vibration velocity spectrum. At this stage, it should be noted that there is no discernible change in noise or temperature and no visual indication that the bearing is distressed.
Stage Two: Fig. 3 represents the velocity spectrum for the second stage of failure. IBF continues to increase. The first indication of a problem on the velocity spectrum is what is often referred to as the bearing component's natural frequency. This low amplitude vibration usually occurs somewhere between 30 Kcpm and 120 Kcpm depending on the bearing. These frequencies are independent of operating speed and non-synchronous. As the problem continues to deteriorate, 1 x rpm sidebands begin to appear. appear. As deterioration deterioration continues, continues, the other component component natural natural frequencies appear, again developing 1 x rpm sidebands until the familiar "haystack shape" has developed. developed. The difference frequencies between sidebands usually indicate the rpm of the defective bearing.
9
10
Section 7, Calculations for Bearing Defect Frequencies
Fig. 4 shows various formulas which can be used to calculate these frequencies. Unfortunately, much of the information required is not always readily available, so the use of the approximations given in Fig. 6 are recommended. The only information required to calculate these approximations is the machine's rpm and the number of balls. This information is usually available. available. Some bearing manufacturers now offer databases and computer programs to determine bearing frequencies. Another fact that assists in recognizing fundamental bearing frequencies is that they almost occur at non-integer multiples of operating speed (non-synchronous) such as 4.3 x rpm, 5.6 x rpm, 6.8 x rpm, etc. They do not occur at full integer (synchronous) (synchronous) multiples such as 2 x rpm, 6 x rpm, 8 x rpm and so on.
11
All bearing frequency calculations are made with the assumption that pure rolling contact is occurring. In practice, however, it is unlikely that this type of contact is occurring perfectly, which can lead to small frequency errors. Also, errors in accurately determining rpm, errors caused by FFT bandwidth, etc., require the acceptance of a certain amount of approximation. As with the component natural frequencies, the amplitudes of these peaks are relatively small. As the bearing deteriorates further, 1 x rpm sidebands develop, especially around the inner race frequencies. This is due to amplitude modulation as the defect(s) passes in and out of the "load zone." They can occur at +/ - 1 x rpm, +/- cage frequency, or +/- ball spin frequency, depending depending on the situation. Stage Three: Fig. 5 represents the velocity spectrum spectrum for the third stage of failure. IBF has reached a maximum. As the problem develops further, bearing defect frequencies that can be calculated appear. Multiples or harmonics of these frequencies are also common. The more harmonics of a bearing defect frequency, the greater the deterioration. However, rpm must also be considered. Low speed machines show considerably lower amplitudes, as well as less bearing defect harmonics, for the same deterioration as in higher speed machines.
12
Stage Four: Fig. 6 represents the velocity spectrum for the fourth stage of failure. This is the final 13
detectable stage, and the IBF units have dropped. At this time, bearing defects start to become less distinct as the noise floor rises. Multiple sidebands occur around harmonics of fundamental bearing defect frequencies. Eventually the spectrum becomes erratic, erratic, and broadband noise occurs. occurs. Notice that the amplitude amplitudess of individual peaks have decreased. Due to the spread over a very large frequency range, and much broadband noise, this stage is often confused with the symptoms of cavitation. It is recommended, however, that when a significant increase in IBF units occurs or "haystack" activity is present on the velocity spectrum, then data collection time intervals should be considerably reduced.
14
Section 8, Further Considerations for Bearing Defect Vibration Amplitude and Frequency Symptoms
As vibration amplitudes due to bearing problems are often relatively small when compared to other vibration sources, "threshold-type" alarms that measure "overall" vibration do not usually give adequate warning for impending bearing failure. Instead, "power band-type" alarms, which measure "root mean square" (rms) values, are most often more appropriate as they give a better idea of "vibration defect energy."
15
Section 9, Estimating Remaining Bearing Life
The writer views all of the symptoms only as warnings. It is analogous to measuring just how much thickness of automobile tire rubber is left above the cords. Assuming a "reading" that indicates that the tire's cords are about to be exposed, the driver might ask if the car could be driven for three hours at the maximum speed without experiencing a blowout. The answer can only be that there is just so much thickness of rubber left, that the condition can cause a blowout within the first five minutes but may not result in a blowout for several weeks or even several months. Obviously, the rubber thickness measurement is only a warning of what can probably happen. It may even predict what what will more likely likely happen, happen, but it is not a sure indication of what will happen. The same is true for bearings showing more and more symptoms via the spectrum, as failure approaches. For bearings with symptoms of imminent failure, it is suggested that the analyst indicate to the operator: "From the best vibration and IBF knowledge we have today (which is incomplete), the symptoms indicate the machine should be shut down as quickly as possible. However, if you have reasons to want to take your chances and continue running, there is always a possibility that this particular failing bearing may run much longer than symptoms indicate. But at this level, the risk is yours. I can only report that I recommend recommend changing the bearing." For bearings that that are deteriorating deteriorating but not as as close to imminent failure, a rough rough estimate may be given, but in the same spirit as the analogy of the tire. Operating under the same operating conditions as before, it is reasonably possible to predict how long a bearing will last if the symptoms and data are compared with that bearing's previous symptoms of failure. Most of the time, however, this experience is not available, so it is not recommended that specific bearing life predictions predictions be made. Many factors factors other than than bearing bearing conditions, conditions, such such as speed, also affect decisions as to whether a bearing should be changed or allowed to run. Try to avoid specific time predictions, since errors in time prediction often lead to a loss of credibility even though the defective bearing has been correctly identified.
16
Section 10, Vibration Symptoms of a Loose Bearing
A bearing can be installed in such a way that "hand feel" indicates a proper fit. However, when the machine is running, the bearing fit might not be tight enough for operating conditions. While running, the fit is loose and the race that is not supposed to rotate, rotates. This, of course, causes excessive rubbing and most often produces symptoms similar to that of a rub or looseness. If no bearing defects are present and the only problem is bearing looseness, the usual symptoms are more harmonics of 1 x rpm than usual. For example, unbalance alone produces almost no larger than usual harmonics, such as a 2, 3 or 4 x rpm. Shaft-to-shaft misalignment usually increases the amplitudes of the harmonics in this low end of the harmonic range (see section "Evaluating Harmonics for Complete Analysis to Determine Vibration Source"). Harmonic amplitudes caused by misalignment usually diminish considerably after 3 or 4 x rpm (unless magnified by resonance or produced by the segments of coupling jaws). However, However, higher higher than usual amplitudes amplitudes of harmonics harmonics due to to bearing looseness can occur in the same low end of the harmonic range (but most often do not show until at least 3 or 4 x rpm) and continue on to still higher harmonic frequencies. These harmonic frequencies are usually still in the range that Update calls "lower harmonics" (approximately 10 x rpm and below). Although more harmonics of higher than usual amplitude are present, they are not absolutely sequential. For example, the harmonics created by looseness at 2 and 3 x rpm may be small and appear higher than usual at 4, 5 and 6 x rpm. For greater looseness or harder rubs, the harmonics can continue, for example, on up to say 8 x rpm; but occasionally intermediate harmonics may be missing, such as the 4th and 7th harmonics. For extreme cases, these harmonics can continue on up to very high frequencies. Synchronous time averaging will not eliminate the true harmonics of 1 x rpm. As this condition is not an actual bearing defect, such as a spall, pit or crack, the harmonics created are not considered to be "bearing defect frequencies." True bearing defect frequencies, such as spalls, cracks and pits, emit frequencies that are non-synchronous with running speed. So when there is question as to whether a peak is due to a bearing defect frequency or due to some other source, it's a good idea to determine whether it is synchronous. Spectra showing rubs or looseness always vary from machine to machine. Only a general idea can be formed by pattern similarity. similarity. Conclusions: 1. IBF units are an early warning system only. 2. IBF units should be trended over time where change is more significant than absolute values.
17
3. IBF units are sensitive to outside influences such as cavitation, transducer mounting, lubrication and so on. 4. When observing velocity spectra, pattern recognition rather than absolute amplitude values are most important. 5. All bearing defects (other than too much clearance or looseness) create peaks at frequencies non-synchronous with running speed. 6. As defects progress, look for increased sideband activity. 7. Difference frequencies between sidebands usually indicate the rpm of the defective bearing. 8. Don't attempt to be too specific in predicting time to failure.
18
Section 11, Further Considerations Regarding Calculated Bearing Defect Frequencies
A common error is to suppose that most calculated bearing defect frequencies will be present almost almost always always in the the higher frequency frequency portion portion of the vibration vibration spectrum. spectrum. This, of course, can occur; but most bearing defect frequencies that can readily be calculated, are revealed at frequencies that are relatively low and can easily be mixed up with those from other vibration sources. For example, sometimes bearings have defect frequencies and harmonics of those frequencies in the range that normally shows the lower harmonics due to shaft/coupling misalignment. If the instrument resolution is not high, accuracy is sacrificed and sometimes bearing defect frequencies are confused with frequencies such as those originating from electrical hum, misalignment harmonics of 2 x and 3 x rpm, vanepass frequencies, and so on. The calculated frequencies for a defective cage or retainer are almost always in the range of less than 50 percent of rpm. Typically, the frequencies are from a little over 30 percent to slightly under 50 percent of rpm. A surprising number of bearing cage cage defect frequencies frequencies are are in the range of 41 41 to 48 percent percent of rpm, causing causing the symptom to look like that for oil whirl. Yet, rolling-type bearings are not subject to oil whirl! Another characteristic of a cage defect frequency is that the calculated amount (less than 1 x rpm) often does not appear on the spectrum. Instead, a harmonic of that number does (such as 2 x the bearing cage defect frequency). This can come surprisingly close to the 1 x rpm frequency and be revealed only with proper resolution. For example in Fig. 7, consider the spectrum's first peak at 28.0 Hz and its third peak at 96.2 96.2 Hz. At first glance, the first peak looks looks as if it's surely surely a ½ x rpm subharmonic. Through a quick calculation, it is shown to be a non-synchronous 0.47 x rpm. The third peak appears to be its third harmonic but calculates to be 3.44 x 28 and, therefore, not its full integer multiple (therefore non-synchronous). Calculation indicates it is also close enough to appear to be 1 ½ x rpm.
19
Also notice that the frequency range goes up to 800 Hz. If 400 lines were used, the range for each line would be 2 Hz (120 cpm). Reducing the top frequency to 200 Hz gives 0.5 Hz (30 cpm) per line. Increasing the number of lines to 1600 gives a resolution of 0.125 Hz or 7.5 cpm. Obviously, data collecting for routine predictive maintenance requires fast procedures, but when a bearing or some other defect is suspected, a further set of very accurate data is required for proper analysis.
20
Section 12, Vibration Due to Oil Whirl (in Plain or Sleeve Bearings)
Oil whirl (sometimes called "oil whip") is one of the easiest vibrations to recognize as it is one of those rare vibrations with a frequency of well below 1 x rpm running speed. Its frequency has been reported to be anywhere from approximately 45 to almost 50 percent of rpm. Simply view oil whirl vibration frequency as "slightly less than ½ x rpm. Referring to the diagram showing the normal running conditions between a shaft and a plain bearing, notice that the shaft is very rarely operating and running with its own centerline coincident with the bearing's centerline. Therefore, there is a section wherein the clearance for lubricant would be less than in other areas. This could be the normal result of the weight of the rotor, the always present partial coupling and shaft misalignment, misalignment, and so on. The narrow gap area acts act s to build up the pressure of the rotating oil, forming a higher pressure point between the shaft and bearing. As the speeds increase, pressure increases, thereby pushing the shaft and rotor further upward (in most cases), and/or toward one side. This allows the gap to widen at the pressure point and move the new pressure point to a position closer to the shaft's bottom. All described so far apply to rotors without oil whirl and are considered normal.
However, if the load is too light (rare), the clearance too great (also rare), or if there is any other reason why the pressure point proceeds to a place whereby the higher pressure can lift the shaft high enough so that the higher pressure section can "escape," then oil whirl results. The shaft that is lifted high enough to allow the escape of the higher pressure section is now no longer supported in this position. The shaft suddenly drops -- the gap narrows again -- and the higher pressure again develops. The process repeats itself cyclically; cyclically; pressure lifting the shaft -- escape of pressure -- dropping of shaft -- pressure buildup buildup -- lifting of shaft -- escape of pressure -- dropping dropping of shaft -- pressure buildup buildup -- lifting lifting of shaft shaft -- and so on. on. The cyclical frequency of all this is the average oil velocity. The oil's velocity right on the shaft's surface is equal to that of the shaft's surface speed. The oil's velocity right on the bearing's surface is zero. The average velocity is 50 percent of the rotor's rpm. With a little slippage, the actual velocity is slightly less than ½ x rpm. 21
Do not confuse it with vibration at exactly ½ x rpm, which does not originate with oil whirl but instead is most often associated with bearing looseness or a rub. The most common published reason for oil whirl is that the bearing loads are too light, relative to the oil pressures built up by higher speeds. Yet in maintenance work of already built machinery with more established design, it has been found that this is the least probable source. (Yet, it is a common source for newly designed machinery that hasn't been de-bugged.) Another reason given is bearings that have too much clearance. Again, with established designs, it is doubted this is very common. An approach used by Update is to first remember that when rotors vibrate for any reason, the centerline of the shaft will be tracing an orbit around the actual axis of rotation. Visualizing the shaft's surface relative to the bearing's inside diameter, the distance between the two will be changing at a rate equal to the vibration frequency. When shaft vibration reaches several mils, the opening and closing of the oil gap for the film will also be equal to several mils. The orbit that seems to create oil whirl most often is from large coupling or shaft misalignment. The misalignment, for example, can position the shaft at a location so as to more easily lift the shaft and thereby allow escape . Misalignment can cause bearings and shafts that do not normally result in oil whirl to be in a threshold position position for it. The solution is not to change the bearing, decrease the clearance, or increase the load (although each of these could work), but to first analyze the vibration at the lower end of the frequency range, such as at 1 x and 2 x rpm and so on, so as to determine if there is a large unbalance or large misalignment. If there is, the unbalance or misalignment can usually be corrected more easily than making bearing alternation alternations. s. For most process plant machinery that can't readily be shut down for bearing changes, rebalancing or realignment, there are some relatively successful methods for curing oil whirl temporarily until permanent changes can be made. The most common successful method reported is to increase the oil's temperature, usually by approximately 10 percent. Although this seems to work more frequently than cooling the oil, there are cases reported whereby the oil whirl problem worsened with increased oil temperature, instead cooling the oil eliminated the oil whirl. Some have accomplished the desired results by decreasing the oil's viscosity (most common success), whereas certain situations required raising the viscosity to eliminate the oil whirl. For altering the bearing itself, several methods have been reported. One is to make the bearing ID egg-shaped, usually scraping material away in the 9:00 and 3:00 22
o'clock positions. Another way is to form a dam or step on the top of the bearing's ID so that the oil will produce a back pressure -- and therefore higher pressure on the top of the shaft -- preventing it from lifting enough to cause the previously described oil pressure section escape. Another Another has been to reduce the surface area with grooves, especially at the bottom half, so as to increase the bearing load. This section is not complete. Bearing and rotating machinery manufacturers have much more experience and knowledgeable details on the subject. Considering the high cost of shutdown and the high cost of a mistake, it is suggested that such sources be consulted in the event bearings are to be altered as a solution.
23
