Enhancing the SAPHIRE Solve Engine: Initial Progress and Efforts

Improving methods and tools used for reactor safety, whether deterministic or probabilistic, is crucial. This research aims to address the current status of a tool vital for probabilistic safety approaches, showcasing initial progress alongside a roadmap for future development. This investigation focuses on the SAPHIRE Solve Engine (SAPHSOLVE), which plays a pivotal role in transitioning from legacy systems to state-of-the-art excellence. Key objectives include efficiently handling large models, enabling seamless model exchange with other tools, and ensuring model tracking capabilities. robust The ongoing efforts involve extracting and enhancing processing engines within the SAPHIRE API, with a specific emphasis on secure cloud solving in addition to existing desktop capabilities. The current initiative concentrates on enhancing SAPHSOLVE, requiring a detailed analysis. This involves a benchmarking approach comprising performance assessments and source code analysis to identify bottlenecks and optimize quantification performance. Synthetically generated models facilitate the model generation process, a crucial step in these endeavors. The benchmarking results underscore SAPHSOLVE's proficiency, with the Windows version demonstrating superior performance compared to the Linux version. Notably, the truncation size of cut sets significantly affects quantification time. Addressing bottlenecks in the source code has led to an average 29% increase in speed. Ongoing efforts are focused on further enhancements and exploring parallel processing capabilities. The pursuit of enhanced performance extends to two distinct approaches: shared data parallelization for low-level implementations and distributed data parallelization for high-level implementations within SAPHSOLVE.