Masters Theses

Date of Award

8-2020

Degree Type

Thesis

Degree Name

Master of Science

Major

Nuclear Engineering

Major Professor

Steve E. Skutnik

Committee Members

Steve E. Skutnik, Jamie B. Coble, Ondrej Chvala

Abstract

Cyclic Voltammetry (CV) has been under recent consideration for assessing the bulk uranium content in the electrorefiner used in pyroprocessing. Many prior studies have focused on either measuring the cell’s electrochemical characteristics, or were limited to a small number of different electrorefiner contents. This study however focuses on assessing CV for predicting the cell’s uranium content with a wide array of cell contents. To do this, a simplified CV model was used on electrorefiner content data from a pyroprocessing flowsheet model and on data covering a uranium mass fraction range of 0.25% to 10.0%. Linear trends were evaluated in the CV output from the model (irreversible peak current difference) under these two ranges. A 7% maximum relative error was observed between the model’s CV output and the predicted output from the trend for the data covering uranium weight percents from 0.25 wt% to 10.0 wt%, and 1.97% for the data from the flowsheet model. These trends were then used to reverse-correlate back to uranium weight percent in the electrorefiner. A maximum error of 1.90% between the reverse-correlated and actual uranium weight percent was reported for the flowsheet data. Further analysis was performed to measure the uncertainty in the evaluated linear trends in the CV output, including measuring the impact of data size and data noise. Increasing the data size decreased uncertainty in the linear regressions’ slope and intercept values found via the CV model, and increasing the CV output noise significantly increased these regression uncertainties. Data noise had a roughly linear contribution to regression uncertainty (e.g. 15% data noise resulted in 3 times the regression uncertainty of 5% noise, assuming the same data size), while data size had a somewhat lesser effect. When these trends were used to reverse-correlate from the CV output back to uranium mass percent, data noise proved to have the largest impact on accuracy, with maximum prediction error increasing to 40-50% at the extreme error values. These results illustrate that cyclic vvoltammetry can produce viable results for monitoring the electrorefiner uranium content, so long as any potential biases within the system are mitigated.

Download
Files over 3MB may be slow to open. For best results, right-click and select "save as..."

Share

COinS