Non-Intrusive Fault Detection in Reciprocating Compressors

This research presents techniques developed for non-intrusive sensing and fault detection in reciprocating compressors driven by induction motors. These procedures are “non-intrusive” because they rely only on voltage and current signals measured on the compressor power cable. The electrical sensor based method allows for easy and non-intrusive determination of many fault sensitive signals that usually require complicated, expensive, and time consuming operations to measure. The electric signals are processed and used with the inverted dynamic motor model equations and motor parameters (which are also determined non-intrusively) to recover the instantaneous angular speed of the compressor shaft, as well as the torque of electromagnetic origin provided by the motor. These two intermediate signals and compressor parameters such as crank shaft inertia are then used to solve for the compressor load torque. This load torque signal has high fault diagnostic value because it is composed of pressure and friction torques, and these signals are close to mechanical phenomena of diagnostic interest in the compressor. All these signals are recovered at a fine resolution giving high level of detail on a sub-shaft revolution basis. The use of the load torque signal in determining faults and additional diagnostic information is also given. A procedure for determining the cylinder suction and discharge pressure from the load torque signal and knowable cylinder parameters such as cylinder volume, crank arm length, and gas coefficients is discussed. The load torque signal is also useful directly: for the two piston machine used in the research, symmetry of the twice per rotation peak of the load torque is a valuable diagnostic measure. Reed valve leakage faults were investigated by drilling small holes of varying sizes in one cylinder’s suction reed. The asymmetry in each cylinder’s pressure torque peak increases with increasing leak size, providing both an indication and measure of leakage severity.