Research On Fault Mechanisms Of Reciprocal Compressor And Its Fault Advance Warning

The technology of condition monitoring and fault diagnosis has been altering the traditional post-maintenance and periodically planned maintenance to predictive maintenance and intelligent maintenance in the process industry. The key factors to achieve this attention are the development of techniques of advance fault warning and fault features extraction, as well as the remote monitoring and diagnosis system.The reciprocal compressor (RC) is a kind of important equipment for transporting gaseous materials in processing industry, whose operating conditions are usually the controlling factor for a production process. The most common fault of reciprocal compressor, accounting for over60%of all faults, occurs in its valves. This fault will lower its compression efficiency, and even cause plant shutdown. However, due to the complex structure of RC and the existance of numerous inner oscillating impact sources in RC during operation, it becomes very difficult to extract fault signals from its valves and to achieve its advance fault warning. Another serious fault of RC is piston rod fracture. The fault tends to result in severe mechanical accidents, even cause explosions and casualties. However, at present, only an accident analysis can be made after the accident, but has no method to avoid it in advance. The remote monitoring and diagnosis system is a platform for implementing condition monitoring and diagnosis, and its development till not good enough in its universality, expansibility, and real-time execution, thus, it is till unable to provide an uniform platform for different kinds of rotary and reciprocal equipment commonly used in enterprises. Moreover, it is inconvenient for upgrading the system itself. This dissertation focuses on the study of valve faults and piston rod fracture, and the study on remote monitoring and diagnosis system. The research works down in this dissertation are summarized as follows:1) Research on techniques of advance warning and feature extraction dealing with different faults of valves:The RC’s force characteristics are discovered through the force analysis for RC carried out by the author. With the thermodynamic and dynamic analyses on the exhaustion-and-suction process of valves, a mathematical model is established with which the valve’s motion, and the valve’s fault can by simulated and the valve’s working can be described.Based on the theory of lifting wavelet transform and wavelet de-noising principle, and maximum likelihood estimate method, the characteristics of early stage vibration signals of valves are studied, and then two methods, the adaptive construction wavelet function based on signal features and the principle of wavelet de-noising threshold based on prior distribution knowledge of extracted signals, are proposed by the author. By combining the above two methods to extract characteristics of early stage fault, and by confirmation of simulated signal and testing data, the advance fault warning of pneumatic valves can be realized effectively.Based on the modulation character of valve’s vibration signal, the technique of energy operator (EO) envelop demodulation is discussed and compared with the common technique of Hilbert transform envelop demodulation. It is discovered that the EO method is much better both in computing speed and in envelop accuracy, especially for signals with impulse characteristic. In order to solve the problem that the carried wave band needs to be pre-determined before the envelop demodulation is analyzed, an analytical method for envelop demodulation based on the combination of wavelet packet transform and EO is proposed, and its advantages are verified by analyzing the simulated signal. The method has been used in extracting features from differing faults of pneumatic valve and achieved good effects.2) Research on advance warning technique of the piston-rod fracture accident:Based on the analyses of forces on piston rod, fracture mechanisms of crack generation, and propagation, and fracture of piston rod, the author points out that the advance warning of fracture for piston rod can be realized by detecting whether the piston rod has a crack and its status.The technique of acoustic emission (AE) and its common analytical methods of signals are introduced, and the relationship between the AE and the crack fatigue propagation are analysed. After simulating the working condition of piston rod by repeated tension-compression test using a multifunctional fatigue testing machine, the relationship between AE signals and stress and strain during fatigue propagation of a crack on the piston rod specimen is studied and it is found out that it is feasible to evaluate propagation degree of a piston rod crack and predict its fatigue life by AE signal’s counts. By analyzing the AE signals of those piston rod specimens with and without cracks, it is found that it is feasible to realize the advance warning of the piston-rod fracture with the AE technique. After discussing sensitive degree of the different characteristic parameters of AE signals on piston rod with and without crack, it is proposed that the amplitude of AE signals can be selected as the characteristic index to judge whether the piston rod has a crack. The potential problems during application of acoustic emission to detect piston rod crack and installation way of AE sensor are also discussed, and the effect of these problems and their solution, as well as designs installation way of AE sensor studied.3) Research on the distributed remote monitoring and diagnosis system (DRMDS) based on middleware technique:To solve the problems that are encounted by the remote monitoring and diagnosis system at present, the author discusses the middleware technique and the advantage and disadvantage of different network structure are discussed. Based on a kind of DRMDS with a three-level network system structure, and middleware technique as its core, and a technique to realize the system are present. Based on the monitoring and diagnostis characteristics of various faults of reciprocal and rotary machinery, kinds of functions of DRMDS are presented that DRMDS should have to meet with the above characteristics. And finally a DRMDS system is successfully developed which has been used in several domestic petrochemical corporations.At the end of this dissertation, the author summarizes the whole research work and gives suggestions of the direction for further research.