DIAGNOSIS OF ROLLING ELEMENT BEARING USING VIBRATION AND ACOUSTIC EMISSION SIGNALS

Abstract

Transmitting load and mitigating friction are the primary objectives of deploying a rolling element bearing in machinery. The bearings are often expected to perform for a more extended time with moderate maintenance requirements leading to most machinery breakdown. As any unplanned maintenance may lead to a significant production loss, monitoring the condition of bearings has become standard industry practice. Bearings used in critical machines are equipped with dedicated condition monitoring systems to gather data in the form of signals. A substantial amount of work has been devoted to utilizing the information concealed in these signals for better fault diagnosis of bearings. This study is dedicated to developing methods for fault diagnosis of bearings by identifying the defective raceways, estimation of cage slip, and defect size. Fault diagnosis of rolling element bearings begins with identifying defective components characterized by particular fault frequency in the recorded signal. Classically, the fault diagnosis is done by converting a time-domain signal to a frequency domain signal and manually locating the fault frequency. Modern-day fault diagnosis methods use intelligent algorithms for fault frequency identification. Though these smart algorithms show promising results, they need extensive data engineering and usually do not consider common machinery problems such as unbalance, misalignment, and looseness. In this study, an autonomous algorithm is developed for fault frequency identification. Compared to sophisticated, intelligent algorithms, it has a simple implementation and considers common machinery issues during fault diagnosis. The algorithm uses the harmonic product spectrum method for fault frequency identification, which takes a pre-processed input vibration signal. The pre-processing stage of the algorithm first removes effects of speed fluctuation, unbalance, misalignment and looseness, and then obtains an envelope which is fed to the harmonic product spectrum method. When used in conjunction with the harmonic product spectrum method, two methods of obtaining envelopes are analysed for their efficacies in fault diagnosis. The numerical model presented in the study provides a physics-based understanding of the bearing dynamics and asserts the utility of deploying the envelope demodulation technique in fault diagnosis. Once the defective raceway is identified, the next step is to estimate the damage severity. Any prediction about damage severity helps in effective maintenance planning. One of the ways to assess damage severity is to estimate the size of the defect. The study aims to develop a methodology to estimate the size of a defect present on the bearing raceway by identifying the entry and exit events concealed in the vibration signal recorded in the vicinity of bearing having raceway defect. The method developed in the study blends mechanics-based and signal processing-based approaches together. The mechanics-based approach is based on the Hertzian contact theory and considers bearing operational parameters along with the effect of the load zone. It encompasses a novel approach to correlate peak acceleration with the location of a rolling element in the load zone. The signal processing approach for locating entry and exit events encompasses concepts of cross-correlation, Variational mode decomposition (VMD), curve fitting, and root location. The proposed method is used to estimate the size of the defect present in the form of an artificially generated spall. Experiments are performed at various speeds and sizes of spall on outer as well as inner raceways. The findings of the study demonstrate the enhancement in the efficacy of estimating the artificially generated spalls size. In an ideal scenario, the rolling element bearings are assumed to operate under pure rolling conditions but seldom do they. Owing to variable operating conditions, assembly imperfections, or lubrication irregularities, bearings tend to have some slip during motion. Persistent slip may lead to an elevated degradation rate of the bearing; hence, an estimation of slip is the next task undertaken in the study. A time-domain analysis method is developed to estimate slip in the form of percentage cage slip for bearings with raceway defects. The method extracts impulses evident in the signal's time domain, and the time span between two impulses is used as a baseline for estimation of slip in the bearing. The methodology takes care of challenges emerging out of speed variation, which affects the time span between impulses. Unlike other cage slip estimation methods, the method proposed in the study needs no extra measurement system. Methods of estimating defect width and cage slip are demonstrated using experimentally recorded signals. Moreover, to support the findings, a mechanics-based dynamic model of rolling-element bearing including cage slip is developed. The simulation results form a base for cage slip estimation methodology. Additionally, the dynamic model correlates the force on the rolling element with peak acceleration recorded during rolling element-defect interaction. Experimental verification of the correlation advocates using the underlying empirical method for obtaining force from peak acceleration. The numerical models describe the physics behind the methods developed in the study. Condition monitoring systems installed industry-wide depend extensively on vibration signals. However, in recent years, acoustic emission measurement and analysis have found significant applications in the field of in-situ and off-site fault diagnosis and condition monitoring of machines. To study the vibration response of rolling element bearings, several models have been developed. Contrary to this, very few attempts can be located in literature, aiming to model acoustic emission of rolling element bearings. Hence, this study explores that avenue by developing a model for acoustic emission. The proposed model is a blend of a multi-body dynamic model and an asperity-based acoustic emission model of a rolling element bearing. The dynamic model of bearing is solved to compute the contact deformations and thus contact forces. From these contact forces and lubricant film thickness, asperity forces and their deformations are computed, which are fused into the acoustic emission model to compute the acoustic emission generates in the bearing. The model is used to analyze the effect of load, speed, and radial clearance on acoustic emission of healthy bearing. Further, the simulated acoustic emission signals generated by the model are analyzed for their usefulness in fault diagnosis of the bearing.