(81b) Advanced Design of Experiment Methodologies for Enhanced Process Understanding

Motivated by recent robotic advances, a host of automated, high-throughput devises are enabling a data-rich experimental environment. This talk will address two recently introduced methodological advances that enable the design of information-reach experiments and model the time-resolved data that are collected. They both contribute significantly to the optimization of a multitude of processes and enhance our understanding of them. In such data-rich environment, the traditional Design of Experiments (DoE) and the Response Surface Methodologies are showing their limitations. We introduce the Design of Dynamic Experiments (DoDE) methodology, a generalization of DoE, and the Dynamic Response Surface Methodology (DRSM) to model time-resolved data. The first, DoDE, enables the design of experiments with time-varying inputs. The second one, DRSM, enables the estimation of a single model that incorporates all time-resolved data of a process output. Exploring the interrelationship among DRSM models of several measured reacting species, one can be led to the discovery of the stoichiometry of the reactions taking place. This enhances our understanding of complex kinetics and the discovery of optimal operating conditions that minimize the generation of unwanted byproducts. Examples from a series of industrial collaborations will be sited.