Model Based Analysis of Compressor Valve Leakage and its Effects on the Efficiency of the Motor-Compressor

In the development of algorithms to detect and diagnose component malfunctions impacting system performance and efficiency, engineers are often faced with the following questions: what is the relevant information required, what sensor or set of sensors is needed to provide this information, how should the information from the sensor signals be extracted, which sensor provides the earliest signal of system efficiency deterioration, and how do these signals change as the severity of the fault increases? A typical approach used to answer these questions is to experiment empirically with the actual system; however, this approach can be expensive and time consuming because it involves a variety of complex instrumentation and requires a fair amount of time to run the tests and analyze the data. An alternate approach, which is considerably less expensive is to use a model of the system and run computer simulations to get insight and understanding into its behavior. Once the computer model is constructed, simulations of the model can be readily obtained and system response to specific component failure studied.

In this paper we show how the key questions related to system performance can be answered through system modeling and simulation. The paper shows how a motor-compressor system can be modeled to yield insight and understanding into system behavior and the model is used to assess and compare the performance of three techniques to detect and diagnose incipient faults of the reciprocating compressor valves.