PITA Fiscal Year 2013 Projects

Computational Models and Algorithms for Enterprise-wide Optimization of Process Industries

Lead University: Carnegie Mellon University
PI: Ignacio E. Grossman, Chemical Engineering Dept.
Co-PI: Lorenz T. Biegler, Chemical Engineering Dept.
PA Industry: Air Products and Chemicals, PPG Glass R&D Center

Enterprise-wide optimization (EWO), which has become a major goal in the process industry, involves optimizing the supply, manufacturing and distribution operations to reduce costs and inventories. We have established a program on EWO at the Center for Advanced Process Decision-making (CAPD), which currently involves twelve companies (ABB, Air Liquide, Air Products, Braskem, Cognizant, Dow, Ecopetrol, ExxonMobil, Petrobras, Praxair, Sasol, Unilever). This proposal seeks partial support for EWO projects with two Pennsylvania based companies, and two companies that have operations in the state: ABB, Air Products, Braskem and Dow Chemical. With ABB the goal is to develop a model to optimize the design and operation of the supply chain for manufacture and distribution of electric motors. The challenge is accounting for various types of motors and their repair, which introduces uncertainties in the demands and reverse flows. With Air Products the challenge is to develop multiperiod decision-making and optimization tools that include market forces and uncertainties. With Braskem the research involves developing a model for optimizing the manufacture of different grades of propylene. The challenge is accounting for the distillation column and reactor for the scheduling decisions of multiple grades of polyproylene. With Dow we are involved in three different projects: Planning and Scheduling of Multiproduct Batch Plants; Design of Supply Chains with Risk of Disruptions in their Facilities; Scheduling and Dynamic Optimization of Multiproduct Batch Plants.

The first gives rise to a multiscale problem where changeovers and blending operations must be accounted for. The second one gives rise to a stochastic programming model that considers safety stocks and rerouting to satisfy customer demands. Finally, the third involves a mixed-integer dynamic optimization model that simultaneously can optimize the production recipe as well as the assignment and sequencing decisions in the batch plant.