CX-020913: GUARDIAN: General Active Sensing for Condition Assessment – Duke University

Duke University, in collaboration with Sandia National Laboratory (SNL) and Westinghouse, proposes to develop a robust active sensing framework (GUARDIAN) through the integration of model-based inference and mobile actuating/sensing robots. The tasks associated with this project are (1) Application description – develop a set of robotic responses to changes in a set of reactor vessel internal (RVI) components; (2) Computational modeling – steady-state approximations of the elastodynamics; (3) GUARDIAN framework – consisting of two main components, the estimator (a physics-based inverse problem module used to determine the location of an anomaly given a set of observations) and the controller (optical measurement system to determine the next set of measurement locations based on the data gathered by the estimator); and (4) Simulated and physical experiments – test the developed GUARDIAN system using both simulated data from finite element models and experiments conducted on a fabricated cylindrical pressure vessel with both static and moving components to simulate behavior modes potentially leading toward failure. Existing equipment and laboratory facilities will be used. Duke University will construct and deploy the robotic system. SNL will supply access to SIERRA-SD, a massively parallel structural acoustics code. Westinghouse will be responsible for deployment of technology, data collection, and providing physical insight of reactor internals and other passive structures in a nuclear power plant.