(558a) Simultaneous Optimization Approach for Feedstock Selection and Operating Conditions for an Olefins-Aromatics Planning Problem

Feedstock selection has become the key factor for improving competitiveness in the actual market environment of all ethylene producers around the world. Demands and prices variations, not only of ethylene, but of propylene, benzene and other by-products have a direct impact in the feedstock selection and the operating conditions used to process them. Furthermore, depending on the feedstock, different alignments and degrees of co-crack are required in order to adapt to the market conditions and at the same time satisfy demands, respecting/observing all plant constraints. This poses challenges in ensuring the most profitable business operation execution strategy.

To solve this complex problem, an advanced optimization application that represents the real business problem is needed. Optience SCMart® has been used since last year by COPESUL as the application platform to satisfy their planning and scheduling needs. The main module for the feedstock selection is a multiperiod nonlinear planning model which includes all plant constraints and uses a low order nonlinear polynomial correlation yield model based on SPYRO® results. The planning model is supported by a scheduling and a feedstock blending model to achieve optimal results at the daily level. The feedstock selection uses a feedstock database library that tracks the different available feedstock qualities and properties.

The main differences between Optience SCMart and other approaches are the use of the nonlinear polynomial correlation for the yield estimation in directly optimizing the feedstock selection. Other LP based applications use a simple ‘look-up' table or first order Taylor's series expansion that has been generated a priori. Optience SCMartxs approach also allows a simple way of considering the feedstock that is currently in inventory and will be mixed or not with the new purchases.

Here, we discuss briefly the planning module that is used to support the naphtha and condensate purchases at Copesul. The nonlinear planning model, inside the module, consists of about 4,000 variables per period including flows, inventories, properties and other process variables. It covers two ethylene plants, one C4s unit, and two aromatics plants, in which the main products are ethylene, propylene, butadiene, and the different aromatics.

We present different case studies or scenarios for which different feedstock are selected according to the market conditions. Accordingly, different optimal operating conditions are selected by the tool. These process variables include COTs, possible co-cracks, furnace utilization, ethane recycling flow and loads to other downstream units.