Data Processing And Optimal Scheduling Of Intermediate Separation System

M-Phenylenediamine,which is one of the most important intermediate in the dye industry,brings great value to the fine chemical plant.For an intermediate separation system,if the raw material supply-side exhibits an irregular pattern for the reason of maintenance or unstable production,scheduling must be performed via some means of production technique,mode switch or load change techniques for instance.Thereby it offers much more freedom to the scheduling work but also brings some challenge to the operating activities.Without the precise knowledge of each unit,manual scheduling is not reliable which is based on personal experience.Thus,the information technology and the scheduling optimization method are of great importance to the balance of supply and demand,improving energy efficiency and reducing energy cost.The problem of scheduling changing feeds on parallel units is addressed in this paper.The main contributions and research work are listed as following:1)Data collection and data processing.Make full use of the historical data and statistic data to improve the performance of each unit.Data processing is very vital to the accuracy of modeling and optimization.Steady state detection,data reconciliation based on several operating modes and identification of gross errors,all efforts to improve the quality of the data-driven model to overcome the problem of blind scheduling without knowing the actual production rate and energy consumption information.2)Learn from the real-world scheduling problems and analyze the scheduling method by the distillation separation system.For the nature of different operating modes in the distillation system,we proposed the mathematical model to illustrate the system.Using the Generalized Disjunctive Programming(GDP)method,the logic constraints about mode-switch and load-change can be fully described.Finally,we formulate a mixed integer nonlinear programming model.3)The proposed modeling framework is applied to a real case study from a fine chemical plant in which the detailed scheduling information of each units can be acquired.The main goal of this part is to demonstrate how the constructed model can be designed to improve a decision-making methodology for a scheduling problem.Comparative results between the optimal scheduling and the manual scheduling demonstrate that the optimal results have a better energy performance.Then a collaborative production strategy is built for the analysis of the uncertainty in supply side.The model is decomposed into scenarios and solution of each scenario will give feasible solution for that scenario.