Adam Baker
(Advisor: Prof. Dimitri Mavris)

will defend a doctoral thesis entitled,

A Bayesian Technology Maturation Framework Applied to Test Evaluation for Structural Technologies

On

Monday, March 17 at 8:00 a.m. 
Collaborative Visualization Environment (CoVE)
Weber Space Science and Technology Building (SST II)

Teams Meeting: Link

Abstract
One of the biggest challenges in the development of complex systems is the selection of technologies that will be incorporated into the system’s design. Technology selection must occur early in the design process since the included technologies will enable many of the program or system goals. The development programs for these systems can span multiple decades, so less mature technologies are often explored that will be developed alongside the system and reach full maturity by product deployment. This can maximize the system’s designed performance advantage over the state of the art once the system reaches production, but it also creates risk in the program if the technology does not meet the projected performance or schedule. Appropriately assessing the maturity of technologies and adequately developing maturation plans to address their shortcomings is therefore a critical challenge in complex system design. 

The current technology development process is characterized by a holistic combination of modeling, expert judgment, and benchmark testing. Technologies are characterized through the assignment of an ordinal technology readiness level and then progress through pre-defined benchmark tests based on that assigned level. This is potentially problematic as the prescribed tests may or may not suitably address the underlying areas of uncertainty in the technology. Restricting the use of data from tests to deterministic passing or failing grades for benchmark performance levels is reductive and both fails to fully exploit the information that is generated by these tests and is potentially problematic as it falsely assumes the result will always be repeatable and reproducible.

The purpose of this research is to develop a framework that will reframe the technology development process to be defined based on the underlying uncertainty sources in a technology. This requires creating a baseline representation of uncertainty in a technology, identifying how these uncertainties impact the system of interest, and finally showing how these uncertainties can be addressed through Bayesian inference using test results. This inference can both leverage test data to reduce the uncertainty in phenomenological properties of the technology and challenge the input assumptions or computational models for the technology. The common thread of uncertainty throughout the technology development process creates traceability within the process and generates actionable data at every step. It is then demonstrated how this resulting framework can be exploited to enable the design of physical tests that can optimally address the underlying sources of uncertainty in a technology. This in turn enables enhanced decision-making for testing as a given test’s efficacy at reducing uncertainty can be traded against the cost of that test. The results demonstrate that a Bayesian framework has the capacity to adequately capture the various forms of information generated in the current technology maturation process and will create traceability that can enable the exploitation of test data and the design of physical tests for an individual technology.

Committee

• Prof. Dimitri N. Mavris – School of Aerospace Engineering (Advisor)
• Prof. Graeme J. Kennedy – School of Aerospace Engineering
• Prof. Christos E. Athanasiou – School of Aerospace Engineering
• Prof. David Goldsman – School of ISYE
• Dr. Burak Bagdatli – School of Aerospace Engineering