MAE PhD Defense – Christopher Elliott

Title: Methods for Streamlined Firefly Optimization and Interpretation: Application to Engineering Design

(Advisor: Dr. Buckner)

Biography:

Chris was raised in Asheville NC, and completed his undergraduate and graduate coursework at North Carolina State University.  He earned a master’s degree in mechanical engineering in 1998, with a minor in mathematics, and has worked at Caterpillar since then.  His responsibilities at Caterpillar have focused in four areas: machine noise control, engine systems development and research, hydrostatic transmission design optimization, and control system development. For the past five years, Chris has been pursuing his doctoral degree, working under the direction of Dr. Gregory Buckner. His doctoral research includes electromechanical system design, control system development, and numerical methods for engineering design optimization.

Abstract:

This research introduces a computationally efficient firefly algorithm (FA) tuned to find multimodal objective function minima and a new method for programmatically and visually identifying locally optimal solutions. A novel plot of distances between cost-sorted designs (the cost-sorted distance or CSD plot) is shown to reveal clusters of designs at minima; the best design associated with each cluster is readily apparent.  An investigation is made of the methods’ capability to address uncertainty in the design process.  Correlation between the percent of fireflies in clusters and robustness measures for associated designs is considered.  Finally, A novel magnetorheological fluid device type is described and its design optimized using FA and CSD.