Shaft alignment——misalignment detection

Misaligned rotating machinery causes high cost to the industry as it causes premature damages to the machinery, loss in production and excessive energy consumption. Misalignment is the most common cause of machinery malfunction. A poorly aligned machine could cost 20% to 30% of machine down time, replacement parts, inventory and energy costs. Large returns are usually seen by regularly aligning the machine. The total operation life is extended and process conditions are optimized for efficiency. Hence it becomes extremely important for the maintenance and engineering professionals to understand machine malfunctions caused by misalignment.

Vibration signatures are widely promoted for studying machine malfunctions. However, most literature will not be able to provide a clear picture of signature characteristics uniquely and directly attributable to misalignment. Every author will report different signatures. There are still no reports of systematic, controlled experiments with varying parameters. However we can carry out various experiments to elucidate any consistent features of vibration signatures for misaligned machinery.

To begin with lets consider a simulator which is fault free and is able generate controlled faults, it should possess three machine operating parameters, coupling types, amount of misalignment and the motor speed which will be systematically varied while all the other parameters will be held constant. The machine should be fault free with the exception of deliberate misalignment which is varied systematically. Therefore, baseline vibration data is recorded for each of the test conditions. Vibrations should be monitored via sensors which should be placed in strategic locations to get accurate data. The X, Y, Z coordinate system is used to show direction. For this experiment we can use three different stiffness couplings, four levels of offset should be used on the left bearing housing to simulate a combination of angular and parallel misalignment. Equivalent offset on the right bearing housing to get parallel misalignment. The experiment must contain four speeds of rotation and the goal here must be to determine the effects of coupling stiffness, level and type of misalignment and finally the speed of rotation on vibration spectra.

Data can be acquired from the specially designed hardware and software for the simulator. The purpose of this experiment should be to examine the spectra due to misalignment between the motor and the rotor shafts. Spectral comparisons should be made across coupling measurement points on left bearing housing and the motor. The data should be compared in both vertical and axial direction. If the results at 960 and 2100 RPM do not show significant vibration, the study can be limited to a higher speeds of 2900 and 3800 RPM. A correlation between misalignment and vibration signature could not be discerned. The data for all cases contained several harmonics. Both axial and lateral vibration was present in all cases. The dominant harmonic varied from condition to condition. As a general rule, as expected, increased misalignment yielded increased vibration peaks. As another general rule, peak vibrations in the misaligned machine were in the axial direction. For predictive maintenance applications where the goal is machinery health monitoring, it is sufficient to realize that the problem is complex. One can routinely trend the vibration spectra until it becomes severe. But for root cause analysis, one must exercise caution and perform a detailed analysis. Obviously, the rules provided in training courses and wall charts are doubtful at best. The observed changes that occurred with shifts of speed and misalignment do not show a typical signature for misalignment vibration spectra. Hence, it can be concluded that misalignment vibration is a strong function of machine speed and coupling stiffness. A single point vibration spectrum does not provide good and reliable indication of machinery misalignment. Observations of spectra in axial and radial directions at various speeds and several points are needed to diagnose misalignment effects. Orbital plots of vertical and horizontal measurements in the time domain might also be needed. Non-linear dynamic modelling must be performed to gain full understanding of misalignment effects. Finally, more work in this field is simply needed in order to develop simple rules for diagnosing machinery shaft misalignment.