The statistical techniques taught in "Strategies for Process Investigations" are used to answer the following questions:

• How are the quantities that describe a process related? Is reactor yield influenced by temperature? Is mean particle size influenced by power input to the grinding mill? How can I tell from the process variables that I have observed?
  
• How can we estimate models from data?
  
• How well do we know the relationships from the data? Is there a clear picture of the relationships, or is it obscured by noisy measurements?
  
• What is the best way to plan an experimental program so that we get the most information to answer specific questions about a process?
  
• How can we improve our process operation using a systematic combination of experimentation and modeling?

CHEE418/801 is a follow-up course to an introductory course in probability and statistics, and is designed to give you a more comprehensive understanding of how models are estimated from data, and how experimental programs can be designed to make the resulting data as informative as possible. The focus of the course is largely on empirical models - models that are estimated from data - that are sometimes called "data-driven models". However, the techniques for estimating parameters, making decisions about parameters, and planning experiments also apply equally to fundamental or first-principles models.

The objectives of the course are:

1. to provide you with a strong background for developing empirical models between process variables through model building, including multiple linear regression with emphasis on evaluation and interpretation of the resulting model;
   
   
2. to provide you with basic techniques for the initial screening of process variables including 2-level, complete and fractional, factorial designs, and higher-order experimental designs;
   
   
3. to provide you with an understanding of Response Surface Methodology, which is a systematic procedure in which experimental design and modeling steps are used to optimize the performance of a process;
   
   
4. to give you an idea of how to go further - estimating parameters that appear nonlinearly in models, and estimating models to describe time-varying behaviour.
   

I hope that you enjoy the course, and find it useful.

Martin Guay

Last UpdatedSeptember 9, 2011