|
Introduction The predictive maintenance philosophy of using vibration information to lower operating costs and increase machinery availability is gaining acceptance throughout the industry. Since most of the machinery in a predictive maintenance program contains rolling element bearings, it is imperative to understand how to monitor and diagnose problems associated with them. Bently Nevada has adopted a two-part philosophy with regard to rolling element bearing monitoring and diagnostics: (1) the monitor system will provide adequate warning to avert catastrophic machine failures and (2) diagnostic data will be available so that when warning is given, the bearings will have visible damage. This philosophy should be kept in mind during the following discussion. Rolling Element Bearing Characteristics Any discussion of monitoring and diagnostics for rolling element bearings would not be complete without a comparison with the techniques used for fluid film bearings. The construction of a fluid film bearing is such that a fluid film supports the shaft during operation. By design, the shaft can experience motion relative to the bearing. Because of this freedom of motion, the industry-accepted vibration measurement for a fluid film-bearing machine is a shaft relative measurement, i.e., proximity probe. A rolling element bearing, by design, has extremely small clearances, which do not allow a significant amount of shaft motion relative to the bearing (Figure 1). Forces from the shaft are transferred through the rolling elements to the bearing outer race and then ultimately to the bearing housing. Because of this transmission, a casing (bearing housing) measurement is normally acceptable for monitoring machines with rolling element bearings. However, as explained later in this discussion, a method called REBAMŽ is available from Bently Nevada Corporation that allows vibration measurements directly at the bearing outer ring, which contains the outer race. This direct measurement greatly enhances bearing vibration data, and in some cases, this is the only measurement that can provide adequate vibration information.(Reference Hansen, J. Seven and Harker, Roger G., "A New Method for Rolling Element Bearing Monitoring in the Petrochemical Industry," Presented at the Vibration Institute Seminar, New Orleans, Lousiana, June 1984.) Shaft relative vibration measurements (i.e., proximity probe) are also useful when clearances increase during failure and for observation of rotor problems that are not related to bearings. A classical characteristic of rolling element bearings is the generation of specific vibration frequencies based on the bearing geometry, number of rolling elements and the speed at which the bearing is rotating (Appendix A). (Reference Foiles, Bill, "Rolling Element Bearing Frequencies," Edited by Bently Nevada Corporation.) |
|
|
The most prominent of these characteristic-bearing frequencies are the Outer Race Element Pass frequency, Inner Race Element Pass frequency, Element Spin frequency, and the Fundamental Train frequency or Cage frequency. These vibrational components are generated even in a new bearing, but the amplitudes are small. Bently Nevada defines the frequency range from the Outer Race Element Pass frequency (1EPx) to seven times this value (7EPx) as the Prime Spike frequency region. This range contains the Inner and Outer Race Element Pass and Element Spin frequencies, and is therefore a valuable region to monitor bearing condition. The Cage frequency lies below 1/2 rotor speed (for a stationary outer ring) and cage damage would therefore show up in the Rotor frequency region, defined below. Distinguishing between these two regions enhances the ability to determine if a vibration increase is caused by a failing bearing or a rotor-related malfunction (imbalance, misalignment, fluid induced instability, etc.). It should be kept in mind that, from a plant manager's point of view, it is much more important to determine when a bearing needs to be replaced to avert a machine failure and unnecessary downtime, than it is to determine what components within the bearing are being damaged. The primary goal of a rolling element bearing monitoring system is to satisfy this need. The secondary goal is to provide data that is appropriate for diagnosing the failure of the bearing with the purpose of determining the root cause (improper mounting, lubrication, loading, etc.) so that similar failures can be avoided in the future. Vibration Characteristics of Rolling Element Bearings The vibrations produced by machines with rolling element bearings can be separated into three frequency regions. (See example in Appendix A for calculation of monitor filter ranges for rolling element-bearing applications). 1. Rotor Vibration Region Rotor-related vibrations normally occur in the range of 1/4 to 3 times shaft rotative speed (1/4X - 3X) and are best measured in terms of velocity or displacement. Many rolling element-bearing failures are the direct result of a rotor-related malfunction (e.g., unbalance, misalignment, or rotor instability). Rotor-related malfunctions must be corrected to eliminate bearing overloading and subsequent failure. Most general-purpose equipment with speeds from 1200 to 3600 rpm generates rotor-related vibration signals between 10 and 500 Hz (600 cpm to 30 kcpm). It is, therefore, imperative for diagnostics to monitor this frequency range in order to determine when/if a bearing failure is caused by a rotor-related malfunction. Without this data, the rotor-related malfunction would remain undetected, and bearings will continue to fail and need periodic replacement. It should be mentioned here that the Rotor Vibration region is not exclusive to rotor-related vibration components. Bearing-related frequencies can also occur in this region. A damaged cage can produce vibrational components below 1/2 shaft rotative speed. In addition, studies have shown that with the REBAMŽ monitoring system, spalling on the inner race can produce signal components in the Rotor Vibration region. 2. Prime Spike Region The second vibration frequency region to monitor for machines with rolling element bearings is the Prime Spike (element passage) region. As previously mentioned, a rolling element bearing generates characteristic frequencies based on its geometry and speed. Prime Spike is a term used by Bently Nevada to describe a vibration frequency range, which includes those bearing frequencies that are generated by the rolling elements traversing either an inner or outer race flaw. This frequency range is normally 1 to 7 times the Outer Race Element Passage frequency (1-7 EPx). Vibrations in this range can be measured effectively in terms of acceleration, velocity or displacement. Field studies indicate that approximately 90% of all bearing failures are related to either an inner or outer race flaw. The other 10% are related to either a rolling element flaw, which produces vibrations in the Prime Spike region in acceleration and velocity signals, and the Rotor region in REBAM signals, or a cage flaw which produces vibration in the Rotor region. By establishing a frequency band around the predominant bearing failure frequencies and filtering out the rotor-related vibration frequencies, it is possible to gain improved monitoring of bearing condition. 3. High Frequency (Spike Energy) Region The third region is the high frequency (Spike Energy) region. This region covers frequencies from 5 kHz to approximately 25 kHz and is measured in terms of acceleration. If high frequency region measurements are used for bearing failure detection, they should be used as a supplement to measurements made in the Rotor-Vibration and Prime Spike regions. High frequency measurements have two primary uses:
Causes of Failure in Rolling Element Bearings A rolling element bearing has a finite life and will fail due to fatigue, even if operated under ideal design conditions. Rolling element bearing manufacturers realize this fact and have developed design life limits (L10/B10) to let users know how long a bearing should last when installed and operated within design limits. L10/B10 is defined as the rating life of a group of apparently identical rolling element bearings operating under identical loads and speeds with a 90% reliability before the first evidence of fatigue develops. Unfortunately, most "real world" installations are not under ideal conditions and the bearings prematurely fail well before reaching their design life. Most premature bearing failures can be attributed to one or more of the following causes:
A rolling element bearing progresses through three failure stages: 1. Prefailure 2. Failure 3. Near Catastrophic/Catastrophic Note: Each of these different failure stages exhibits specific vibration characteristics, which require specific diagnostic/monitoring techniques. Prefailure Stage -- During the prefailure stage, the bearing is in the earliest stages of failure. It develops hairline cracks or microscopic spalls that are not normally visible to the human eye, since most of the early damage occurs below the surface of the race. During this stage there may be an increase in the high frequency (7 EPx) vibration produced by the bearing. If temperature or Prime Spike vibration measure-ments are taken during this stage, the levels will be normal. At this stage, the bearing usually has a significant amount of safe operating life left, and it is not economical to replace it at this time. Failure Stage During the failure stage, the bearing develops flaws that are visible to the human eye. At this stage, the bearing usually produces audible sound, and the temperature of the bearing will rise. Vibration amplitudes in the "bearing-related" range (Prime Spike) reach easily detectable levels. Once the failure stage is reached, it is necessary to either change the bearing or increase the frequency of monitoring to estimate how long the bearing will safely operate before causing a catastrophic machine failure. This stage is considered the economical time at which to replace the bearing. If the bearing is not removed during the failure stage, it will eventually enter the final progression of failure, the near catastrophic/catastrophic stage. Near Catastrophic/Catastrophic Stage When the bearing enters this stage, rapid failure of the bearing is imminent. Audible noise produced by the bearing increases significantly, and the bearing temperature increases until the bearing overheats. Rapid wear causes the bearing clearance to increase, which then allows significant shaft motion relative to the bearing. Since a rolling element bearing is designed to restrict shaft motion, it can be very dangerous to allow the bearing to reach this stage due to the probability of creating a rub within the machine. Bearing-related (Prime Spike) vibration amplitude levels will show significant increases in this stage. High frequency vibration data may be unreliable in this stage and caution should be used in its interpretation. Due to "self-peening" of the bearing flaws, high frequency amplitude levels often decrease during this stage, and it can appear that the bearing is in an earlier stage of failure. The occurrence of this "self-peening" phenomenon is especially true for low speed machines. Transducers and Instrumentation for Vibration Measurement and Monitoring of Rolling Element Bearings 1. REBAMŽ Instrumentation System Rolling Element Bearing Activity Monitor. The REBAM system uses a high-gain, low-noise eddy current proximity transducer that is installed in the bearing housing observing the bearing outer ring (Figure 2). The bearing outer ring contains the outer race. The REBAM transducer measures the very small (microinch/micrometre) deflection of the outer ring as the rolling elements pass the area observed by the transducer. These deflections are measured in terms of displacement. The operating frequency range for the REBAM transducer system is from 0 Hz (dc) to 10 kHz (0 to 600 kcpm). The REBAM system is a direct and very sensitive method of rolling element bearing measurement. It offers a very high signal-to-noise ratio, as compared to casing-mounted acceleration or velocity measurements. Through the use of electronic filters, the REBAM vibration signal is separated into Rotor Vibration and Prime Spike regions as previously discussed. Typical Prime Spike amplitudes are 10 to 50 microinches (0.25 to 1.3 micrometres) for a good bearing and 2 to 5 times that for a damaged bear-ing. However, the amplitude of the REBAM signal is highly dependent upon the amount of loading on the elements as they pass the location of the probe, and it is, therefore, not possible to give broad guidelines for a healthy or a damaged bearing. A common practice is to take readings on what is known to be a healthy bearing and set the monitor Alert and Danger alarm levels at 1.5 and 2 times the baseline level. Field and lab tests confirm that using such alarm levels provides adequate failure protection. 2. Casing Vibration Instrument Systems Rolling element-bearing condition can be monitored by using casing measurements. Overall velocity or displacement, Prime Spike velocity, and the high frequency acceleration regions can be used. Bently Nevada can provide accelerometer or velocity transducer-based systems to monitor rolling element bearing condition. Overall casing velocity or displacement provides a means for determining the general mechanical condition of rolling element-bearing machinery. For a velocity transducer-based system, the frequency range used is from 10 Hz to 1 kHz (600 cpm to 60 kcpm). For an accelerometer-based system, the frequency range used is from 10 Hz to 20 kHz (600 cpm to 1.2 million cpm). Depending on the machine speed, the velocity system frequency range is likely to span the Rotor frequency region and the lower end of the Prime Spike frequency region. The acceleration system will cover the Rotor frequency region, Prime Spike region, and into the high frequency region. As stated previously, the Prime Spike region is used by Bently Nevada to monitor the rolling element bearing-related frequencies (inner/outer race defects). By filtering out the rotor-related vibration signals (i.e., 1X, 2X, etc.), it is possible to get better signals related to the rolling element bearing condition. The Prime Spike frequency region includes the fundamental element passage frequency (EPx) and harmonics up to 7 EPx. If an accelerometer-based system is used, Prime Spike and high frequency measurements are available. However, due to the noise susceptibility problems previously mentioned, the high frequency measurements should be used only as a possible early indicator of impending rolling element bearing problems. The high frequency range used by Bently Nevada is from 5 kHz to 25 kHz (300 kcpm to 1500 kcpm). High frequency measurements are useful when diagnosing certain machine malfunctions mentioned previously. The following alarm threshold guidelines were established for the three diagnostic techniques based on field tests conducted by Bently Nevada: 1. Overall velocity: 0.3 in./sec. (7.7 mm/s) pk. 2. Prime Spike region: 0.1 to 0.15 in./sec. (2.5 to 3.8 mm/s) pk. 3. High frequency region: 3 to 4 g pk. Note: These alarm threshold levels for balance-of-plant equipment should be considered as an initial guide only. The levels may need adjusting up or down, depending on how an individual monitored machine behaves. When using casing measurement systems, two key factors should be considered: (1) signal amplitude versus transmission distance and (2) the measurement's susceptibility to noise. The farther away a vibration measurement is made from the vibration source, the more the signal will be attenuated. Most rolling element bearing machines have joints between machine parts, which further attenuates the signals. Figure 3 shows that there is sharp vibration signal attenuation at each joint. The closer the measurement is made to the machine vibration source, the better the measurement. For a rolling element bearing, it is suggested to be within 1 to 2 inches (25 to 50 mm) from the bearing. Another fact that must be considered is a comparison of the various transducer systems' signal-to-noise ratios. System signal-to-noise ratio is defined as the ratio of the amplitude of a desired signal at any point to the amplitude of noise signals at that same point. A comparison of the signal-to-noise ratios for the various transducer systems is shown in Figure 4. If not carefully considered, these two factors can have a large negative impact on the success of using casing measurements for monitoring. Summary Based on the Bently Nevada two part philosophy of (1) providing adequate warning to avert machine failures and (2) removing bearings only when they are likely to have visible evidence of an impending failure, the following conclusions are drawn:
Hansen, J. Steven and Harker, Roger G., " A New Method for Rolling Element Bearing Monitoring in the Petrochemical Industry," Presented at the Vibration Institute Seminar, New Orleans, Louisiana, June 1984. Foiles, Bill, "Rolling Element Bearing Frequencies," Edited by Bently Nevada Corporation. Figure 1 Return to article
Note: It has been observed that traversing an interface (water jackets, double cases, etc.) can attenuate the signal by as much as a factor of ten (20 dB). Return to article
|
|
