Isogeometric analysis for nuclear engineering

C56-40a-273805The Office of Nuclear Energy has developed powerful computer modeling and simulation tools for nuclear reactor design and testing. Their flagship multiphysics platform is ideal for simulating both light-water and non-light water reactors, but it cannot run on the graphics processing units (GPUs) that are increasingly used in high performance computers. Also, non-light-water reactors continue to grow more complex, making it difficult to approximate them into the representation needed for simulation. This makes the platform difficult for new, inexperienced, and occasional users. This project will address both limitations. It will integrate next generation isogeometric analysis technology, which allows users to skip the approximation step and saves up to 80 percent of the total development time. Second, it will integrate a new simulation library that will allow the multiphysics platform to run on GPUs. These improvements will make the flagship multiphysics code more attractive and accessible to new and inexperienced users, who will no longer need special skills to set up a useful simulation in a reasonable time frame. In Phase I, proof of concept will be established by demonstrating that isogeometric analysis can be run on GPUs using a subset of functionality of the flagship code. Once this is successful, the Phase II project will expand the proof-of-concept, so it works with all the functionality of the flagship code. This proof of concept will be shared back to the open-source code, where developers will be able to add the high-performance computing capability to other tools developed by the same office. The research and development done in this project will benefit other commercial projects aimed at nuclear energy and other industries including nuclear fusion. Eliminating the defeaturing and approximation step through isogeometric analysis not only makes simulation easier for new users, but it also means that users can test multiple variations of a design concurrently. This leads to better, safer designs, and faster progress from idea to development - all benefits that are particularly crucial in nuclear reactor development.