Uncertainty Research Lab Summer 2022: Data Science for Engineers
Before the 1950’s, the Air Force believed in the ‘average man’ . The ‘average man’ was believed to be the average in human body data as a way to oversimplify the variability in a population of pilot bodies. This was to simplify the design for a cockpit to fit everyone since it would fit the ‘average man.’ It was only until further investigation by Gilbert Daniels revealed that out of the thousands of Air Force pilots (over 4,000 pilots), no one was average in all dimensions of a pilot. Daniels quoted that “‘the average man’ is a misleading and illusory concept as a basis for design criteria,” illustrating the importance of handling variability in the real world and the consequences of only using a single value (the mean) to represent a distribution . However, this is not the only example of variability negatively affecting the effectiveness of designs, and furthermore, this is not a widely known study. Despite the results of this study, there is a lack of research regarding how engineers handle variability. Many industries rely on single values (like the mean) when doing engineering design work. This flattens real world variability, and opens up the potential to design shortfalls and probability of failure. For example, if an engineer were building a structural system out of an aluminum alloy, they might use a mean strength value of that alloy when doing calculations for the design. Because real world materials have variability, meaning the samples will exhibit a range of different strength value, this design approach could lead to mechanical failure if the sample used in the construction falls below the mean strength used in the design. This is why it is important that engineers are given adequate tools and knowledge on how to approach real world variability. In this study, we conducted interviews with engineers to understand how they approach real world variability. The goal of this study is to better understand how engineers reason under uncertainty, and to provide more baseline research that will help us to better educate engineers on how to approach variability.