(610c) Optimal Integration of Production Planning and Process Operation in Petrochemical Industry

Production planning optimization can improve production efficiency and increase benefits. For a petrochemical industry which has a large number of units, a variety of products and diversiform processes, there are many production plans for option, so optimal one is essential to insure it running safely and bring maximal economic benefits. However, whether the optimized production plan in mathematical meaning is feasible and how to adjust the plan depend on process operation. In this paper, production planning and process operation were optimized and integrated as two hierarchies, to find a feasible production plan and corresponding process operation condition.

Production planning (Hierarchy I) and process operation (Hierarchy II), which was an emphasis of this paper, could be regarded as two hierarchies. The Production plan from Hierarchy I was assigned to units, which was a down procedure. Units of Hierarchy II submitted some feedback information to Hierarchy I, which was an up procedure. Thus, a closed loop system was formed. The system included three parts: modeling and solution of Hierarchy I with commercial production planning software to determine a production task, process simulation and optimization to find operation condition which could fulfill the plan and satisfy quality indexes and revision of model of Hierarchy I. A strategy of optimal integration of two hierarchies was presented. Firstly, production planning model was established by GIOPIMS (Graphic I/O Petro-chemical Industry Modeling System), which had interactive, visual and friendly user interface, and solved with LP solver embedded in GIOPIMS. The plan and correlative quality indexes from Hierarchy I were transferred to simulation model of Hierarchy II, which was executed by process simulation software, such as Aspen. To find optimal operation condition which could fulfill the plan and satisfy quality indexes, perhaps there were many algorithms. For example, secant search was appropriate for finding optimal operation condition of preflash tower of a refinery distillation unit. However, for the whole unit which was multivariable, stochastic search was needed. If process operation condition could be obtained, the plan is feasible; otherwise, the plan was infeasible. Some constraints of the plan vectors must be calculated in Hierarchy II, to revise LP model of Hierarchy I. Then a new solution was obtained from revised model. Optimal integration of production planning and operation process of units was realized undergoing solution procedure of determination, examination, revision and determination.

There was a case study of optimal integration in this paper. In the case, there were three towers: preflash tower, atmospheric tower and vacuum tower in a refinery. Preflash tower had two process operation parameters: top condensed temperature and bottom stripper steam flowrate. Atmospheric and vacuum tower had nine process operation parameters respectively, including top condensed temperature, bottom stripper steam flowrate, temperature of furnace, side stripper steam flowrate, flowrate and return temperature of pumparounds, etc. All parameters were given a value range based on practical production data or experiences. For preflash tower, top condensed temperature had little effect on the preflash products and could be regarded as a constant determined from practical production data. Then, bottom stripper steam flowrate could be obtained by secant method after 9 iteratives. However, for atmospheric and vacuum tower, all the nine parameters must be determined simultaneously and could be obtained by stochastic search method after 20 iteratives. At last, a feasible optimized production plan and corresponding process operation condition were determined to guide practical production.