Uncertainty Quantification Of Plutonium Inventory Calculations Through Xenon Gas Sampling Of Underground Waste Tanks Containing High-level Radioactive Waste From Uranium Reprocessing

An experimental method has been developed by the Pacific Northwest NationalLaboratory (PNNL) for estimating the plutonium mass contained in tanks storing high-levelradioactive waste produced from nuclear fuel reprocessing at the Hanford site in the state ofWashington, USA. These tanks contain various waste products from chemical separationprocesses for plutonium and uranium, such as the PUREX process. The method developed byPNNL for plutonium mass quantification employs a mathematical model that correlates themeasured concentration of xenon isotopes to the spontaneous fission of plutonium-240 and itsmass. Coupled with known isotopic ratios of plutonium isotopes in the tank, this model allowsfor the estimation of the plutonium inventory of the tank without the need to directly samplethe solid and liquid waste material. A sensitivity analysis was performed to identify the highestcontributing factors to the uncertainty in the final plutonium calculation and is presented in thepaper. This work was performed with uncertainty quantification (UQ) toolboxes written forMATLAB and Python programming environments, UQLab and Chaospy respectively. Themathematical model described by PNNL was transcribed into MATLAB and Python scripts alongwith input values for each parameter of the model. Each input value was randomly sampledwith a Gaussian distribution created by an associated uncertainty and used with the PNNLmodel to create a polynomial chaos expansion. The resulting analysis produced an uncertaintyin the final calculation of plutonium as well as a normalized sensitivity of each input parameter.The input parameters of greatest concern for driving higher uncertainty in the waste tank werethe spontaneous-fission yields of the xenon isotopes. For the benchtop experiments, highsensitivity parameters included the spontaneous fission yields, the activity measurements, andthe uncertainty in the Pu-240 branching ratio for spontaneous fission. We found that betteraccuracy of Pu-240 spontaneous fission yield values of xenon isotopes can increase theaccuracy of PNNL’s experimental method.