280037 Feedstock Optimization and Scheduling of a Polymer Plant

Wednesday, October 31, 2012: 1:50 PM
326 (Convention Center )
Pablo A. Marchetti1, Ignacio E. Grossmann2, Wiley A. Bucey3 and Rita A. Majewski3, (1)Chemical Engineering Department, Carnegie Mellon University, Pittsburgh, PA, (2)Department of Chemical Engineering, Carnegie Mellon University, Pittsburgh, PA, (3)Braskem America, Pittsburgh, PA

Polymer manufacturing is a continuous process where several campaigns of different product families are scheduled in a cyclic manner. The raw materials, usually referred as ‘feedstock’, are by-products of petroleum or natural gas such as ethylene and propylene, and their prices are inextricably linked to the prices of such commodities. The polymer market is extremely competitive and the quality expectation of customers is very high. Polymer plants usually operate at full capacity, seeking to improve the return on assets and to reduce supply chain costs [1]. In the last years, significant progress has been made in the development of new computational tools and models to solve problems in the context of Enterprise-wide Optimization [2]. Nonlinear programming formulations as the ones required to model the manufacturing processes of polymer production can now be accurately solved with reasonable computational effort. As an example, a multiperiod nonlinear programming optimization model for planning of production and operation of a real world multi-plant polymer facility as been developed by Jackson et al. [3]. They include detailed nonlinear empirical process models obtained from experimental results and actual plant operation data.

We propose in this work a nonlinear programming model for the selection of the optimal balance of feedstocks for manufacturing multiple grades of polymer product in a polypropylene plant. The feedstocks have different prices and propylene purities, and the process includes a reactor and a distillation column with various types of recycles. The model constraints consist of material balances around the system, lower and upper bounds on flow rates, constraints on compositions, and limits on the catalyst yield and flow. The main decision variables are the production rates of polypropylene for each product type, the flowrates and compositions of each stream, and the catalyst flow. The optimal operation balances the production rate with the costs of feedstocks, maximizing the plant throughput. The different products differ on how much catalyst is needed to produce a given production rate, and the objective function is driven by the economic tradeoff of selling prices versus feedstock costs.

The representation of the distillation process has been one of the major challenges. Approximate solutions for the performance of the distillation column are obtained using the aggregated group-based method of Kamath et al. [4]. This method models a counter-current cascade of trays, without providing tray-by-tray details, and was chosen in order to maintain an adequate balance between solution accuracy and model size. Comparisons of the model with real plant data are presented.

Based on simplified and aggregated distillation models, single and multiple-product nonlinear programming feedstock optimization models have been developed. In the latter, a fixed time horizon is considered to produce a given schedule of polymer products. The optimal solution slows down the production rate for certain products to obtain a better separation on the distillation unit, thus reducing the production cost by using more refinery grade as feed to the column. Different time horizons were tested to measure the tradeoff of decreasing feedstock costs versus increasing the production rate. In addition, a “slack” product has been included to assess the benefits of having some extra production when the schedule finishes earlier. Overall, the results consistently showed the convenience of running the plant at its maximum flowrate whenever possible. A user-friendly computational tool that implements the proposed models to assist on purchase decisions by practicing engineers is also described.

References:

[1]   Kadipasaoglu S., Captain J., James M. Polymer supply chain management. Int. J. Logistics Systems and Management. 2008; 4:233-253.

[2]   Grossmann I. E. Enterprise-wide Optimization: A new frontier in Process Systems Engineering. AIChE Journal. 2005; 51:1846-1857.

[3]   Jackson J. R., Hofmann J., Wassick J., Grossmann I. E. A nonlinear multiperiod process optimization model for production planning in multi-plant facilities. Proceedings FOCAPO 2003 (Eds. I. E. Grossmann and C. M. McDonald), pp. 281-284 (2003).

[4]   Kamath R. S., Grossmann I. E., Biegler L. T. Aggregate models based on improved group methods for simulation and optimization of distillation systems. Computers and Chemical Engineering. 2010; 34:1312-1319.


Extended Abstract: File Uploaded
See more of this Session: Industrial Applications In Design and Operations
See more of this Group/Topical: Computing and Systems Technology Division