Chemisthrill

Optimal Design: Adapting Experiments to Real-World Constraints

This episode explores how optimal designs handle practical limitations. It covers constrained factor spaces, mixture experiments, mixture-process designs, covariates, and time trends—showing how to design useful experiments when factors cannot be freely controlled or changed independently.

Optimal Design: Building Efficient Experiments

This episode explains how optimal designs help researchers choose informative experimental runs while saving time, cost, and resources. It covers the coordinate-exchange algorithm, optimisation criteria, blocking, random blocking, mixed models, and split-plot designs for hard-to-change factors.

Episode 2 Description

From Experimental Design to Process Optimisation: Response Surface Methodology

In this episode, we move from identifying important factors to understanding how those factors work together and how to find the best operating conditions.

We begin by introducing Response Surface Methodology, or RSM, and explain why continuous factors often require more than a simple straight-line model. You will learn how quadratic terms describe curvature, how response surfaces and contour plots help us visualise process behaviour, and how designs such as Central Composite Designs, Box–Behnken designs, Definitive Screening Designs, and I-optimal designs support optimisation.

The episode also explores how to optimise several responses at the same time using desirability functions. We discuss the difference between finding a mathematical optimum and finding a practical, robust operating region that can tolerate normal process variation.

Finally, we look at sequential experimentation—starting with screening, moving towards improvement, and then building a detailed response surface model around the promising region.

By the end of this episode, you should understand how design of experiments develops from:

Finding important factors → modelling the response → predicting behaviour → optimising the process.

This episode is useful for anyone working in process development, chemistry, engineering, manufacturing, formulation, or quality improvement.

Hello! This is my first episode after a long hiatus. I am currently pursuing my second degree at Imperial College London in Digital Chemistry with AI and Automation. This first episode is inspired by a recent module assignment on DoE (Design of Experiments). I wanted to review the material here just in case, and hopefully, others will find it useful too! I’m planning to release more episodes soon.

About this episode: We dive into the DoE Foundation, Randomization, Blocking, Independence, Replication, and Factorial Experiments