Chapter 13: Statistical Design of Experiments for Reaction Modelling and Optimisation

Statistical design of experiments, often known as DOE, is a powerful method for understanding how to control industrial processes to ensure consistent quality. In this chapter, we will see how DOE enables efficient experimentation to maximise our learning on multi-factor systems. An example case study is used in which sequential DOE yielded a comprehensive model of a yeast fermentation process for the production of a biosurfactant. The DOE approach will be compared with the limited understanding that could be gained from traditional trial-and-error and one-factor-at-a-time approaches. This is not intended as a guide on how to design experiments for other multifactor processes or systems. Rather, the aim is to provide an introduction to what is possible with DOE and to illustrate the value of carefully designing experiments with the intention of building useful multifactor empirical models. The objectives of the case study are introduced before a brief discussion of the broader reasons for experimenting on industrial processes. The value of the final model is then presented, followed by an explanation of how this model was incrementally built and the additional information that each step in the sequence of experiments contributed to the overall solution.