A Two Week Course in Design of Experiments for Engineers

This course has been taught over 15 times, with revisions and improvements each time, audiences ranging from 20 - 28 engineers.

Assumptions: Each engineer either has their own computer with either Statistica or JMP usable from that computer, or the engineers work in pairs at such computers.

Note: Some technicians have taken this course and benefitted greatly. Judgment should be exercised in making sure the people in the class have some responsibility for developing or improving a process, and some experience in use of SPC.

The course works best when the two weeks are separated by 3 - 4 weeks to allow participants to work using the ideas of Week 1 before beginning Week 2. It is possible and in fact desirable to break the course up further into four 2.5 day portions to generate a greater amount of project worki.

Topics covered in detail in the course:

Week 1:

Introduction and overview
    Why experiment?
    Basic ideas of experimentation
    Case studies of successful experimentation
 

Simple comparative designs
    Simple split lot experiments
    Simple comparative experiments
 

Running the experiment and checking assumptions
    Randomization
    Residuals
    Model checks
 

Factorial experiments
    Multiple factors
    Key concepts - interaction, replication, blocking, randomization
    Design methods
    Analysis methods

Week 2:

Fractional factorial designs
    Introductory comments
    Nonreplicated designs
    True fractional factorials
    Analysis of fractional factorials
    Robust design
 

Variance components
    Design to assess variance components (Passive Data Collection)
    Estimation of variance components
    Use of variance component estimates in management decision making

Integrated wrap-up
    Simulation experience with Design of Experiments
    Concluding discussion