Statistical methods and experiment designs for bulk factor screening in manufacturing - in the style of Evolutionary Operations

A common engineering problem in manufacturing is identifying which of many possible factors influences a response of interest, or in other words “What is causing a performance, reliability, quality or cost issue?” Evolutionary Operations (commonly called EVOP) harnesses the muscle of manufacturing operations to generate large data sets with minimal disruption in the factory, and that capture the natural variance of the materials and processes and materials used in the manufacture of a product. Supersaturated Experiments methods guide experimental plans and supports statistical analysis of the data upon which sound conclusions and interpretations can be drawn. This paper presents these methods as they can be applied together to the problem of screening of many factors to find those that influence a particular response of interest. A case study from the Virtual Cell Factory is presented.