Peter Goos presents four case studies from the Optimal Design of Experiments book he co-authored with Bradley Jones. He introduces each case by describing the type of experiment and when and why it is useful. Next, he demonstrates how to set up the experiment in the JMP software and how to analyze the resulting data.
Presentation material taken from Optimal Design of Experiments: A Case Study Approach by P. Goos and B. Jones.
The presenter designs an experiment for investigating a chemical reaction using a full cubic model in two factors. Because many factor-level combinations are known to be infeasible in advance, the two factors cannot be varied completely independently of each other.
The presenter creates a blocking design to determine what factors impact the stability of a vitamin. The process mean shifts randomly from day to day and the investigator wants to explicitly account for these shifts in the statistical model.
The presenter designs an experiment run in a wind tunnel where it is possible to vary certain factors without shutting down the tunnel and other factors can only be changed by shutting down the tunnel.
The presenter creates a definitive screening design to determine optimal settings for a laser etching process.