Research On Air Separation Unit Load Change Control And Diagnose

Cryogenic air separation unit (ASU) has been used to produce oxygen, nitrogen, and argon, and other gases, as desired. Recently, in order to decrease the oxygen venting capacity, load change operation has become one of the most important requirements for an air separation unit. It poses a problem that the process variables showed a large amount of interaction, especially in load change situations. Hence, there are various technologies such as process simulation, real-time monitoring and fault diagnosis need to be combined in automatic load change system.Based on the developed understanding of load change of air separation unit, its characteristics and problems, the thesis designed a novel type of automatic load change system used in air separation unit. Taking use of the knowledge of expert system and multivariable statistics, it realized an expert system which is capable of real-time diagnosis and online abnormal condition monitoring system. Finally, a software platform was designed in order to support integrated diagnoses. Research work mainly focuses on the following parts:In order to improve load change operate of air separation unit, a novel automatic load change system was developed in this study. The whole system was based on process simulation. It could change the working load by gradually applying a sequence of steady-state optimal points. Unit interfaces and general calculation driver module were also designed. Finally, human machine interface were developed base on the perspective of improving the reliability of automatic load change.Expert system is one of the important techniques in the domain of fault diagnosis. Many applications of expert system occurred over the last decade. This thesis developed a real-time fault diagnosis expert system which used in air separation unit, established the rules of expert system and developed the mixed inference engine which is able to decide its work direction. Finally, a knowledge database was designed for the early detection of nitrogen block which is a typical abnormal condition.Due to the high dimensionality of the complicated process as like as air separation process, principal component analysis (PCA) was introduced to reduce the dimension of the process data. After successful combined PCA methods and real process, the thesis developed an online condition monitoring system used in statistical models. Finally the thesis verified the design and analyzed the performance of the system.At last, a novel concept of integrated diagnosis platform was proposed. The thesis described the details of the platform by defining module interfaces and data flows.