A fundamental approach to Nuclear Waste Repository research involves the collection of experimental data in a laboratory setting, development of empirical and/or mechanistic numerical models representing those observations, and application of these models (or Reduced Order Models) into reactive transport and performance assessment models as predictive tools for informing society of impacts and risks associated with nuclear waste repository scenarios. Both the assimilation and interpretation of experimental data must take advantage of ML opportunities to improve predictive tools, particularly from the standpoint of uncertainty quantification. However, a number of challenges need to be met in the coming years to successfully integrate these new data science approaches. A recent effort at Lawrence Livermore National Laboratory (LLNL) has led to the developement of a data digitization pipeline to simulate radionuclide sorption that is amenable to both ML as well as traditional surface complexation modeling. With the LLNL Surface Complexation/Ion Exchange (L-SCIE) database that is already available, we have begun developing mechanistic surface complexation and ML models of radionuclide retardation specific to the nuclear waste disposal problem. In this presentation, we will introduce this novel approach to mining community data to develop robust models of radionuclide reactive transport.

Machine learning and data mining applications in nuclear waste repository reactive transport modeling and performance assessment

ACS Spring 2022

The official archive and repository for all ACS Meetings & Expositions abstracts and select posters or presentations.