The goal of the economic optimization of a bio-fuel manufacturing plant is to determine the most profitable way to operate the plant based on the current market and plant conditions. This optimization must take into account the complex interactions among the economic factors, the process interconnections, and the operational constraints. In order to achieve the economically optimal operation, the economic optimization must be integrated with the the underlying control system. Despite attracting significant interest from both academia and industry, the implementation of a truly integrated economic optimization and automated control solution for the production plant has proven a major challenge. The fundamental challenge lies in the development of consistent, accurate models for both the economic optimization and the automatic control layers in order to ensure that the optimal operational settings determined by the economic optimization can be realized by the controller.

This work describes the development of an integrated plant-wide economic optimization and Model Predictive Control (MPC) strategy for the economic optimization of an ethanol manufacturing plant. The plant-wide optimization takes into account the various sections of the ethanol plant: milling and slurry, fermentation, distillation, dehydration, separation, and drying. The goals of the economic optimization are to determine operational conditions that balance production (faster rates) with yield (slower rates) to maximize profit, utilize resources to minimize costs, and maintain feasible operation in all sections of the plant. The control layer uses nonlinear MPC to optimize the dynamic operation of the individual units given the operational directives provided by the economic optimization.

The approach uses a flexible, parametric hybrid modeling framework that enables the transparent inclusion of individual models developed for the various components of the process into a complex, modular plant-wide model that is suitable for the economic optimization and can be easily managed. This approach maintains the consistency between the plant-wide model used for economic optimization and the models developed for the MPC of the individual units while allowing additional process information to be included in the economic optimization model.

The solution is used for both online operation as well as for off-line scenario analysis and is implemented in the Pavilion8 software package. This tool allows direct control and provides tools for analyzing the behavior of the economic optimization and MPC systems.