7-09: Advanced sensors and control systems for biofuels and biochemicals production

Production of lignocellulosic bioproducts will grow significantly over the next decade. In order to be competitive with fossil-based fuels and chemicals, maintaining cost-effectiveness will be critical.  Majority of research in reducing operating costs of biofuels and biochemical has focused on finding cheaper feedstocks, developing efficient and robust microorganisms, process integration, and co-product utilization. There has been relatively little emphasis on process control and optimization methods, which have significantly reduced operating costs in similar industries, such as petrochemicals and pulp and paper. It is often cited that the minimal emphasis on control and optimization is due to the inherent nature of biochemical systems, which are notoriously difficult to predict and control due to process nonlinearity, slow dynamics, and lack of suitable non-invasive on-line measurement techniques.

In this research we present a strategy that utilizes an advanced control scheme, called model predictive control (MPC), in conjunction with on-line sensors. Raman spectroscopy sensors are used to track carbohydrate and metabolite concentrations during continuous fermentation. These state measurements are used to make predictions of future system trajectories using an unstructured, unsegregated mechanistic system model. Solution to a nonlinear optimization problem at each sample instance provides an optimal control input scheme during operation. The strategy is tested using various control configurations, as well as controller robustness against model mismatch. The overall goal is to reduce operating costs of the plant, by improving productivity, product yield and titer.