Salt Chemistry Optimization and Prediction Engine (SCOPE) for Nuclear Waste Treatment Planning
This project aims to integrate Super Intelligence (SI) into the Department of Energy’s nuclear waste treatment system through the Salt Chemistry and Prediction Engine (SCOPE). SCOPE is a hybrid, first-principles and data-driven scientific machine-learning framework designed to predict salt chemistry and support more informed decisions throughout waste treatment.
Led by Savannah River National Laboratory, the project team will develop an SI-ready salt database drawing on relevant storage and processing information from the Savannah River Site (SRS) and Hanford. The database will bring together knowledge currently distributed across reports, spreadsheets, operator logs, and other records. Organizing these sources into a consistent resource will provide the foundation for training and evaluating models that can help researchers understand complex salt chemistry.
The team will use a chemistry and physics-informed SCOPE workflow to train supervised surrogate models. These models will be developed to represent important chemical behaviors and make predictions that can be evaluated against measured sample data. Comparing model predictions with experimental measurements will help the researchers assess accuracy and determine whether the approach is ready for practical use. The workflow will also support iterative refinement as additional data and validation results become available.
If training and validation are successful, researchers will apply SCOPE to flowsheet prediction, scenario testing, and early warning. Flowsheet prediction could help assess how changes in treatment conditions affect expected process outcomes. Scenario testing could allow teams to explore potential operating conditions and compare possible responses before making decisions in active treatment settings. Early-warning capabilities could help identify chemical conditions or process trends that merit closer attention.
By combining historical operational knowledge, measured samples, and scientific machine learning, the project seeks to strengthen the evidence available to waste-treatment teams. Its results could provide a foundation for more consistent predictions, better-informed planning, and a clearer understanding of salt chemistry across DOE treatment activities.
Partners
- Savannah River National Laboratory
