Kayla Clements
Email me: clemekay@oregonstate.edu
I’m a third year graduate student at Oregon State University under Dr. Todd Palmer. I work with the Center for Exascale Monte Carlo Neutron Transport (CEMeNT), a PSAAP-III focused investigatory center focused on advancing time-dependent Monte Carlo neutron transport simulation capabilities on exascale computing systems. I focus on uncertainty quantification methods for Monte Carlo radiation transport solvers, as well as using dynamic-mode decompostion to solve for alpha-eigenvalues in time-dependent radiation transport problems.
I got my bachelor’s degree in Nuclear Engineering with a minor in French and Francophone studies from the University of Florida in 2019. Go Gators.
Current Work
Uncertainty Quantification
Specifically, uncertainty quantification methods for problems solved using Monte Carlo radiation transport (MCRT) solvers. In general, the goal of uncertainty quantification is to characterize and/or reduce how sources of uncertainty propogate through some numerical solver and affect your quantity of interest. This can be done by evaluating the statistics of your quantity of interest with respect to the uncertainty source. If you had an exact analytic solution method for your problem, you could simply evaluate the statistics of your quantity of interest to understand how an uncertainty source is affecting your solution, e.g. perform a sensitivity study. However, MCRTs introduce additional uncertainty due to the use of finite numbers of particle histories. If you wanted to use an MCRT solver to perform UQ, you could over-resolve the problem to drive down that solver uncertainty. Instead, while interning at Sandia National Laboratories, I’ve been working to develop a method that can separate the uncertainty contributions of the solver from those of the uncertainty source.
Previous Work
Reactor Physics, Idaho National Lab
- Experiment Analysis Intern, Jan - June 2020. Optimized critical configurations of INL’s Transient Reactor Test Facility (TREAT) by integrating upgrade fuel models into a standard TREAT model using MCNP. Parameters such as excess reactivity, power peaking factor, and power coupling factor were calculated to optimize placement of the upgrade fuel in the core. Critical models were made for various experiment configurations in TREAT.
- Reactor Physics Intern, June - Aug 2019. Modeled upgrade fuel in MCNP for TREAT using fabrication and technical specification documents for the upgrade’s previous design work in the 1970s and ’80s. The MCNP model of the upgrade fuel was then implemented in an existing model of TREAT’s current design to find critical configurations and possible uses of the upgrade fuel.
Nuclear Data, Brookhaven National Lab
- National Nuclear Data Center Intern, June - Aug 2018. Under Gustavo Nobre, wrote scripts to automate runs of EMPIRE, a nuclear reaction code, and analyze the generated data. Some of the data generated include reliable evaluated files across the whole nuclide chart, including nuclei off-stability. An adiabatic model previously developed by Nobre et. al was used to describe statically-deformed nuclei in the rare-earth region and applied to all isotopes of Gadolinium and Tungsten.
Extracurricular Interests
- Hiking, cooking and baking, crochet and painting, film photography, my cat.