The First Sets Of Atmospheric And Vacuum Distillation Unit Of The Karamay Petrochemical Plant Operation Optimization

This paper aims at the project of operation optimazation on the atmospheric and vacuum towers in Kelamayi petrochemical factory. Mainly it discusses two problems: software instruments of viscosity and flash point using neural networks and operation optimization on the atmospheric and vacuum towers using NLJ method.On the basis of comprehension of process flow , process mechanism and principle of crude oil distillation, we analyze various factors which affect viscosity and flash point. Sequentially, we establish software instruments of viscosity and flash point about No.3 side line of the atmospheric tower, the No.l, No.2, No.3, No.4 side line of the vacuum tower.In this paper, we adopt two methods: RBF neural network and Multiple neural network. The tenique of software instrument is through some optimization criterion, chosing a batch of secondary variables which are related closely to primary variables and can be easily measured, using software to implement prediction about primary variables. The content of software instrument are the type choice of secondary variables; the number choice of secondary variables; the choice of testing point; the manipulation of prcess data and the confirmation of the method of software instrument. From the training and testing curve, software instruments obtain certain accuracy, satisfying the requirement of project.In order to implement operation optimization of the atmospheric and vacuum towers, we establish the yield model of No.3 side line of the atmospheric tower and No.3 side line of the vacuum tower. On the basis of these , we adopt NLJ method to implement optimization. The characteristic of NLJ method is high convergence rate and can economize calculating time. The content of process optimization include four parts: the choice of decision variable; the determination of object function; the establishment of constraint condition and the choice of optimization algorithm. From the result of optimization, the rate of production is added and energy consumption is decreased.