Haba, Usama A. M. (2019) Information Extraction Based on the Analysis of Motor Supply and Structural Vibration for Machinery Condition Monitoring. Doctoral thesis, University of Huddersfield.

Reciprocating compressors (RC) are one of the most widely used industrial machines due to their flexibility and reliability. Commonly exposed to harsh working environments, compressors experience various faults that affect their operational performance and functionality. Unfortunately, conventional condition monitoring methods such as vibration monitoring shows inefficiency sometimes in detecting multi-faults that occur in RCs even thoigh it needs a high-cost system for monitoring equipment.

The AC motor driving the RC undertakes an oscillating torque which induces additional components in the measured current signals and changes with the presence of faults in either compressor or motor with consequent. Extensive investigations have shown that motor current signature analysis (MCSA) has the potential to be an accurate and cost-effective technique for the detection and diagnosis of common RC faults (i.e discharge valve leakage, intercooler leakage, broken rotor bars, stator winding asymmetry, and discharge valve leakage combined with stator winding asymmetry). However, a full study has not been found to consolidate this deduction. This research has adopted a model based simulation and experimental evaluation to verify and detail the use of MSCA for monitoring compressors under a wide range of load and fault conditions.

The study firstly develops an extended and comprehensive dynamic model of the compressor, which links the various effects including mechanical dynamics, electrodynamics and fluid dynamics and, thus allows the current signature variations, particularly the sideband patterns to be simulated, and studied under common faults and their combinations at different severity. The model was validated against measured current signatures. Moreover, an advanced and effective technique, modulation signal bispectrum (MSB) has been identified to be the accurate tool as it can accurately extract the effect of sidebands and suppresses inevitable noise influences.

Subsequently, a number of experiments were carried out based on a general purpose two-cylinder RC. Experimental results have shown that the seeded compressor faults caused observable changes in the motor current signature, inducing non-linear and modulation characteristics into the measured signal. The above-cited compressor faults generate different patterns and varying load on the motor, thereby inducing changes in modulation features in the current signal. Especially for the combined fault, an increase in sidebands can be observed by MSB analysis due to its high performance of noise reduction and nonlinear feature extraction.

The MSB results in primary diagnostic features; sideband peaks at the first and second orders of compressor operating frequency has the ability to differentiate RC fault cases over a wide range of operating conditions, which allowed fault diagnosis without the need for additional measurements such as pressure, speed, or temperature. The successful detection and diagnosis of common compressor and motor faults through experiments and model developments confirm the capability of MSB based MCSA method.

With the successful diagnosis of RC and motor faults, this research study is progressed to validate the capability of MSB based methods to diagnose different common compressor faults relating to compressor leakages, motor faults, and combined fault. The results show that MSB signatures allow accurate differentiation between normal condition and abnormal change induced by these seeded faults compared to conventional analysis, confirming the effeciancy of the signal processing technique proposed in this thesis.

FINAL THESIS - Haba.pdf - Submitted Version
Available under License Creative Commons Attribution Non-commercial No Derivatives.

Download (7MB) | Preview


Downloads per month over past year

Add to AnyAdd to TwitterAdd to FacebookAdd to LinkedinAdd to PinterestAdd to Email