Page 1 of 15 DoctorKnow Application Paper Title: Characterizing Shaft Misalignment Effects Using Dynamic Measurements Source/Author:Dan Nower & Curt Thomas Product: Corrective Technology: Corrective Classification: Biographies CHARACTERIZING SHAFT MISALIGNMENT EFFECTS USING DYNAMIC MEASUREMENTS By Dan Nower, P.E. & Curt Thomas Computational Systems, Incorporated Dan Nower is an Applied Development Engineer in the Engineering Division at Computational Systems, Incorporated. He is responsible for the development of the alignment product line. Dan has over ten years of alignment and vibration analysis experience. He received a BS in Mechanical Engineering from the University of Missouri-Rolla He is a registered engineer in the state of Tennessee, an associate member of ASME, has published several papers and is the holder of two alignment related patents. Curt Thomas is a Consultant working with Computational Systems, Incorporated. He is responsible for much of the testing of all alignment products. Curt has two years experience in the alignment area. His prior experience includes steam turbine-generator repair work and a construction engineer in the U.S. Army. He has a BS in Mechanical Engineering from the Georgia Institute of Technology Abstract A test system has been developed for measuring the effects of shaft misalignment under a variety of common field conditions. This system allows five common types of couplings to be tested under loaded conditions for both offset and parallel misalignment in the horizontal and vertical directions. The dynamic measurements that were collected include vibration spectral and waveform data, axial and radial phase data at 1x, 2x and 3x RPM, motor current data, and temperature data. This information should be extremely useful when troubleshooting possible misalignment problems Introduction We all know the importance of proper shaft alignment. Misalignment can lead to excessive vibration and stresses, premature component failures, and eventually shortened life spans of machines, Finding misalignment while a machine is shutdown is not often difficult. Just measure the misalignment with a reliable alignment system. But what about the step prior to this, when the machine is operating? Most of us are familiar with the traditional vibration analysis rules: 1. 2x running speed amplitude -Radial direction for offset misalignment -Axial direction for angular misalignment -Most misalignments are a combination of the two, so 2x in any direction
Page 2 of 15 2. 0 and 180 Phase Shifts Radial direction for offset misalignment Axial direction for angular misalignment How accurate are these rules? Is the 2x RPM peak always a good indicator?. How does the coupling design affect the shape of the spectral information? Do the phase rules work? If so, how much misalignment does it take to cause the phases to shift?
Page 3 of 15 Beyond the traditional rules, those that are involved in the alignment area have noticed interesting phenomena with respect to motor current, and temperature. Little has been documented on these parameters. This undertaking monitored these parameters, but this paper deals mainly with the accuracy of the traditional rules on a limited number of couplings and touches briefly upon motor current and temperature results. Previous Work Dewell, David L., "Detection of a Misaligned Metallic Disc Flexible Coupling Using Real Time Spectrum Analysis." Master Thesis, Virginia Polytechnic Institute and State University, March 1980. Lorenc, Jerome A., "Changes in Pump Vibration Levels Caused by The Misalignment of Different Style Couplings." Proceedings, Eight International Pump Users Symposium, Houston Texas, March 1991. Nower Daniel L., "Preliminary Report on Characterizing Shaft Misalignment Effects Using Dynamic Measurements", Presented at the Vibration Institute Meeting, Wilmington, Virginia, June 1992. Piotrowski, John D., "How varying Degrees of Misalignment Affect Rotating Machinery - A Case Study." Proceedings, Machinery Vibration Monitoring and Analysis Meeting, New Orleans Louisiana, June 1984. Test Machine & Couplings The configuration of the test machine is shown in Figure 3. The motor drove the generator which provided power to the heating element located below the motor. Much care was taken to correct for soft foot. The couplings tested are listed in Table 1 along with their manufacturer's recommended alignment tolerances. MOTOR: Dayton 1 Hp 3450 RPM Loaded, 3550 Rpm Unloaded 15 amps, full load 60 Hz, Single-Phase 115 Volts 28 Rotor Bars 24 Stator Slots Anti-Friction Bearings Loaded by heating element located underneath
Page 4 of 15 GENERATOR: Dayton 360O RPM 16.7 amps, maximum 60 Hz, Single-Phase 120 Volts 2 Kilowatt Anti-Friction Bearings Provides power to heating element Parameters Monitored Misalignment Conditions Besides the aligned condition, the machine was increasingly misaligned in the vertical and horizontal directions. The pure offset misalignments were typically 10, 25, and 50 mils, and pure angle misalignments were typically 1,5, and 15 mils/inch (17.45 mils/inch = 1 ) Vibration & Motor Current Vibration spectra and time waveforms were gathered using normal averaging and high resolution time synchronous averaging. Time synchronous averaging was utilized because the high level of electrical signals generated by the single-phase motor.
Page 5 of 15 Motor current was gathered in different frequency spans and resolutions to monitor the line and rotor bar frequencies. Phases Phase at 1x, 2x, and 3x RPM were monitored in the radial and axial directions. The outboard axial readings were taken around the clock face, at 12, 3, 6, and 9 o'clock Inboard axial readings were also taken. If phases did not stabilize after one minute, phase at that point and speed was labeled "unstable". Temperature Temperature was measured at the end of each misalignment route, which took about 2% hours to complete. Temperature was measured with an infrared temperature probe at all bearings and at the coupling hubs. Amplitude & Spectral Results General The amplitude data was analyzed to locate those multiples of running speed that provided the highest levels and those that trended upward as the misalignment increased. Harmonics up to 8x were reviewed, but only up to 6x was included in the analysis since amplitudes at 7x and 8x were so small. The amplitudes were sorted into blocks by increasing amounts of offset and angular misalignment. If the amplitudes in each block trended upward it was flagged as good indicator. The number of good blocks and the highest amplitudes determined which harmonic of running speed was the best indicator for each coupling. Table 3 shows the results for each coupling. The Grid coupling had the best misalignment response with 63% of the amplitude blocks trending upward. The Shim coupling was the worst with 12.5%.
Page 6 of 15 Rotor bar frequencies (24x and 28x) were also reviewed. They did not show an upward trend as the misalignment increased in any direction. Outboard measurement points trended misalignment as often as inboard measurement points, however the highest amplitudes were found at the inboard bearings. Also, no correlation was found with respect to: a Angular misalignment more prone to revealed itself in the axial direction and offset misalignment in radial direction. b. Vertical misalignment affects the horizontal measurement points and visa versa. Bun Coupling The spectra shown are from point Generator Inboard Horizontal (GIH) as the horizontal offset misalignment increased from the aligned condition to 75 mils. This point provided the most reliable spectral data and also provided the best upward trend. The spectral content shown in Figure 5 is typical of this coupling. As shown in the data, 2x times running speed proved to be the most consistent misalignment indicator for this coupling. The highest amplitudes (0.292 ips-pk) were generated at GIH as the machine was offset in the vertical direction 75 mils (12.5 mils above tolerance). Based upon the data the following was concluded for this coupling: 1. 2x RPM was the best amplitude to track to find misalignment as a whole. 2. Horizontal misalignment was difficult to find. 3. 1x RPM was not much help. Jaw Coupling The spectra shown in Figure 6 are from measurement points Generator Inboard Axial (GIA) as the horizontal angular misalignment increased. The offset misalignment was increased to 25 mils (10 mils above tolerance) and the angular was increased to 15 mils/inch (0.86 ). These spectra show the typical
Page 7 of 15 high harmonic content that was generated while the jaw coupling was in service. Amplitudes at 2)( RPM were much lower, but the overall energy generated was high. 3x and 5x prove to be the best harmonics to track a misaligned condition. Amplitudes at 3x reached 0.09 ips, pk at GIA and GIH, caused by both angular and offset misalignment Amplitudes at 5x reached 0.12 and 0.15 ips, pk at GIV and GIH, caused by both angular and offset misalignment. Based upon the data the following was concluded for this coupling: 1. 3x RPM was the best amplitude to track to find misalignment as a whole. 2. 5x RPM was a good indicator in this case because of the frequency response function (FRF). 3. Horizontal and vertical misalignment was just as easy to find when 3x is tracked. 4. 1x RPM was not much help. Shim Coupling The spectra shown in Figure 7 are from the measurement point Motor Outboard Vertical (MOV) as the horizontal angular misalignment increased. The offset misalignment was increased to 50 mils and the angular was increased to 15 mils/inch (0.86 ). This was the quietest and smoothest running coupling tested. All amplitudes were low. The highest amplitude was 0.0654 ips-pk at 2x on measurement point GIH in the aligned condition. Typically the amplitudes were highest at 2x running speed, but were typically down around 0.01 ips-pk. Consistent upward trends were difficult to find, with 6x the best indicator. Amplitude trends typically decreased as the misalignment increased and started to increase at the maximum misalignment.
Page 8 of 15 Based upon the data the following was concluded for this coupling: 1. 6x RPM was the best amplitude to track to find misalignment as a whole However its track record was not good, correct only 30% of time. 2x RPM had the highest amplitudes, but did not typically (only 10%) trend upward. 2. Any misalignment was difficult to find. Therefore, caution should be used when using this coupling. The vibration maybe low, however stresses are still working on the critical components. 3. 1x RPM was not much help. Another machine with this type of coupling was tested at a customer's facility. The machine was a 700 Hp induction motor driving a multi-stage pump. The response was measured from several misaligned conditions. The amplitude and spectral response generated similar information. No amplitudes trended upward as the misalignment increased. A typical set of spectra are shown in Figure 8.
Page 9 of 15 Grid Coupling The spectra shown in Figure 9 are from measurement points Generator Inboard Vertical (GIV) as the horizontal offset misalignment increased. The offset misalignment was increased to 50 mils and the angular was increased to 15 mils/inch (0.86 ). These spectra show that the 4x RPM peak is a good indicator of the amount of misalignment, with the maximum amplitude of 0.2056 ips-pk showing up at GIV with 50 mils of offset in the horizontal direction. Not only did the 4x harmonic peak generate the highest amplitudes, but also showed an upward trend as the misalignment increased 80% of the time. Accuracy (trended upward) of the 1x and 2x RPM peaks were 35% and 60%, respectively.
Page 10 of 15 Based upon the data the following was concluded for this coupling: 1. 4x RPM was the best amplitude to track to find misalignment as a whole. Also producing the highest amplitudes. 2. Misalignment was not difficult to find. 3. 1x RPM was not much help. Rubber Internal Gear Coupling This was also a quiet and smooth operating coupling. Again, all amplitudes were low. The highest amplitude was 0.0204 ips-pk at 2)( RPM at measurement point GIH which was misaligned by 15 mils/inch in the horizontal direction. Typically the amplitudes were highest at 2)( running speed, but were typically down around 0.01 ips-pk. As shown in Figure 10, consistent upward trends were difficult to find, with 6)( being the best indicator. Amplitude trends typically increased as the misalignment increased but upward trends ware minimal, typically starting at 0.004 ips-pk and increasing to 0.010 ips-pk. Based upon the data the following was concluded for this coupling: 1. 6x RPM was the best amplitude to track to find misalignment as a whole. However its track record was not excellent, correct 53% of time. 2x RPM had the highest amplitudes and trended upward 43% of the time. 2. Any misalignment was difficult to find since the synchronous amplitudes ware so low. Therefore, caution should be used when using this coupling. The vibration maybe low, however stresses are still working on the critical components. 3. 1x RPM was not much help. Another machine with this type of coupling was tested at a customer's facility. The machine was a 100 Hp induction motor driving a single stage centrifugal water pump. We measured the response from several misaligned conditions. The amplitude and spectral response generated similar information. This being small upward trends with increased misalignment and 2x being the best indicator. A typical set of spectra are shown in Figure 11.
Page 11 of 15 Phase Results It was difficult to decide how to reduce all the phase data down into something practical, Phase differences were calculated across the coupling for the axial and radial measurement points, Phase differences were also calculated from end-to-end of the motor and the generator. These differences sorted by the type of coupling, type of misalignment, and misalignment direction. Each group was considered a phase block. A phase shift was flagged if it was a multiple of 180, ±15 Unstable phases were not considered in phase shift calculations. Total phase accuracy was calculated by taking all 180 phase shifts and dividing by the total number of times a phase shifts could have occurred. The phase block accuracy was determined by dividing the number of phase blocks that shifted by the total number of applicable phase blocks. If just the raw percentages are reviewed (see Table 4), phase analysis does not appear to be a good indicator of misalignment. The numbers typically ran in the low to mid 20% range. Phase indicated misalignment the best on the Jaw Coupling and the worst on the Rubber Gear Coupling.
Page 12 of 15 However, the percent accuracy is not how the phase is typically used in the field. It is utilized by determining the number of 180' phase shifts measured during operation. The greater the number of 180' phase shifts the more likely that a misaligned condition does exist. The results of this are shown in the final column of Table 4. The number of phase shifts increased in some of the couplings as the misalignment increased to a certain point. A YES is logged in the above table for those that did. It is interesting to note that for the maximum misalignments (50 mils offset and 15 mils/inch angle) the number of phase shifts per condition decreased. Also there appears to be a relationship between the accuracy percentages and the phase shifts, with the cut off point being around 15% Tests on the 100 Hp and 700 Hp machines produced similar results. The results from the 700 Hp machine (which used a shim type coupling) were: Total Accuracy = 20%, Block Accuracy = 40%, and the number of phase shifts did increase with the amount of misalignment. The 100 Hp machine (which used a shim type coupling) results were: Total Accuracy = 7%, Block Accuracy = 0%, and the number of phase shifts did not increase with the amount of misalignment. Motor Current Results We believe that the motor current portion of the experiment actually raised more questions than it answered. It was expected that by measuring the motor current it could be determined that proper alignment can save money by reducing the amount of wasted energy consumed to overcome forces induced by misalignment. This held true for the Bun coupling. With higher degrees of misalignment, the motor current increased small amounts, ranging from 2.1% to 4.9%. The savings typically did not increase as the misalignment increased This raises a question, Are the saving percentages the same for larger machines if they are properly aligned? Jaw coupling motor current results did differ from the bun coupling. All motor currents at line frequency decreased as the misalignment increased, opposite of what was expected. This was true regardless of the type or direction of misalignment. However, "two times slip" sidebands around line frequency and all harmonics of line frequency increased in amplitude. The increase in amplitudes occurred during the initial misalignment and remained at that those levels. They did not increase as the misalignment Increased. The remaining couplings produced wild results. The motor currents varied up and down as the misalignment increased. Further investigation revealed that the electrical system under test did not have a stiff voltage. When the system was loaded, the voltage was pulled down significantly. In order to compensate for this, voltage and input power factor should have been measured, but was not. Also, the '"two times slip" sidebands did not react as they did with the Jaw coupling. All delta amplitudes (between
Page 13 of 15 the line amplitude and the lower sideband amplitude) remained stable as the misalignment increased. The Shim Coupling remained around 49 db, the Grid Coupling around 47 db, and the Rubber Gear around 51 db. Temperature Results As with the other measured parameters (vibration, motor current, and phase) the temperature response to misalignment varied with the type of coupling. Some indicated misalignment well, especially when looking at the coupling temperature. Table 5 below shows how the couplings ranked using temperature as a misalignment indicator. The All Points Block Accuracy is the number of temperature blocks that showed an upward trend divided by the total number of chances. This includes the temperatures at the motor bearings, generator bearings and the coupling hubs. Considering all couplings, the coupling measurement point showed the most consistency, therefore the Coupling Temperature Block Accuracy column shows how often this particular location showed an upward trend. The table shows that the Grid coupling was the best of the couplings tested and the shim coupling was the worst. Results from the larger machines (700 Hp and 100 Hp) showed similar results. Coupling block results on the shim coupling (700 Hp) and the rubber gear coupling (100 Hp) were 0% These two couplings also performed poorly as an indicator in the above data. Conclusions Amplitude and Spectral Probably the most significant item learned from this experiment is that the spectral information generated by a machine is very dependent upon the type of coupling. Figure 12 shows a set of spectra from the same measurement point (GIH) with similar misaligned conditions. Troubleshooting personnel should take this into consideration while doing their job. Also beware of the type of coupling being used prior to performing running alignments or running soft foot checks. CSI does not condone these type of checks. Phase Analysis When looking at the number of phase shifts over the complete machine, phase analysis is a good indicator for several couplings and not for others. So, before performing this type of analysis know what type of flexible coupling is used on the machine being tested. See Table 4 for results on each coupling. Motor Current Hopefully you can learn from our mistakes. Much care has to be taken when determining the effects of misalignment on the efficiency of the total machine. Just measuring motor current is not enough. To determine the effects the machine, it should be operated at the same load before and after alignment. For
Page 14 of 15 example, a motor-pump combination should be operated at the same flow rate before and after. Also, if the machine is not on a stiff (with respect to voltage) system, measure voltage and input power factor. The proper amount of savings can be calculated from this information. Temperature If using temperature as a misalignment indicator you should be aware of what type of coupling is installed on the machine, see Table 5 to see those couplings tested. Also, the best measurement point location is the coupling itself, preferably one of the coupling hubs. Be sure to use a non-contact method, such as infrared, to measure since the coupling is rotating. Some of the upward trends found were small. The increase is small enough that a variation in ambient temperature or the process could affect the temperature more than the misaligned condition. These parameters were held constant during the test, but the process may have affected the results from the large machines. General It is evident from the data that the troubleshooter in the field should always be aware of the type of flexible coupling being used before any final decisions are made. Different couplings affect the spectral shape (Figure 12), phase (Table 4), motor current, and temperature (Table 5) During the test, responses were measured within three operating hours of the initial startup for each condition. Misalignment maybe like heart disease, it maybe required to operate in a poor condition awhile before symptoms start showing up. In the case of misalignment, time to allow the misalignment stresses to open up clearances Most of the results generated from this experiment were from a machine smaller than the typical machine in the field Further testing should be performed on larger machines using the same type couplings. Some tests have been performed with the shim and rubber gear designs, which confirmed most of the results obtained from the small machine.
Page 15 of 15 All contents copyright 1998-2006, Computational Systems, Inc. All Rights Reserved.