Rhines Hall Room 125
549 Gale Lemerand Drive
Gainesville, FL 32611
Tom Gruenwald, Ph.D.
Executive V.P., Business Development
Blue Wave AI Labs
Dr. Tom Gruenwald co-founded Blue Wave AI Labs in 2016. Blue Wave solves technologically complex and economically impactful problems through the application of AI techniques. His team works with Constellation, Southern Nuclear Power, and Cooper Nuclear plant, as well as the US Department of Energy and the US Department of Defense. Constellation credits Blue Wave with saving them over $35M during the last two years by using its AI-based algorithms, which predict plant conditions two years into the future. Blue Wave also works closely with the defense department providing AI-based algorithms to speed up hypersonic design and other problems important to national security.
He holds a Ph.D. and an M.S. in theoretical physics from Purdue University, where he was awarded the George Tautfest Award for outstanding graduate research. Tom holds a B.S. in Physics from the University of Cincinnati. He was the Purdue School of Science Outstanding Alumnus in 2008. He continues to be active at Purdue, where he participates in research in nuclear physics and guest lectures on careers in the business world for science majors. Tom has authored and published numerous scientific papers.
Historically, several problems have persisted that impact the ability to further improve the economics of reload fuel planning. These problems can limit reductions in the amount of fresh fuel required to be loaded into the core (known as the reload batch size), resulting in excess direct fuel costs. Additionally, they can have a generation impact by less reactivity occurring throughout the fuel cycle than expected or by the potential need to derate power if conditions require it.
Inability to predict Moisture Carryover (MCO) – The amount of moisture mixed with steam leaving the reactor’s moisture separators, referred to as MCO, has been near impossible to predict by conventional methods. There are design specifications limiting how much MCO is permissible before the operator will take remedial action (of which one costly option is a power derate).
Unpredictability of Eigenvalue in BWRs – The hot reactivity parameter of the core, k_effective, is one of the most fundamental parameters in nuclear engineering and has been notoriously difficult to predict accurately in boiling water reactors (WRs).
We will discuss the machine learning predictive models for both MCO and Eigenvalue. This will include feature selection, modeling techniques, their use in the reload planning process and the economic benefits of their application. These new capabilities extend beyond just core design. The same concepts apply to cycle management strategy evaluation. If unforeseen changes occur relative to the planned operating strategy, such as a fuel failure, unplanned downtime, or startup delay, this predictive suite can be utilized to analyze alternate operating scenarios and provide user-friendly comparisons.
Department of Materials Science & Engineering