By Dr. Douglas Montgomery
The last 25 years have seen many important and fundamental changes in how experiments can be designed. These changes have come about because of increases in computing power and the development of efficient algorithms for implementing optimal design methodology. This enables engineers, scientists and other that use designed experiments to create designs that are customized to the specific characteristics of their problem and not have to rely on tables of libraries of standard design from textbooks or software.
This webinar gives a basic overview of optimal design methodology and demonstrates how it is implemented in modern computer software. Examples of how this approach can be used to create custom designs are shown for a variety of situations, including cases where there are constraints on the region of experimentation, restrictions on the number of runs that can be made, or non-standard models that need to be fit to the results. Some specific new developments in DOX that have origins in optimal design methodology that will be discuss include no-confounding fractional factorial designs, designs that include one-step screening and response surface modeling, and designs for computer experiments.