Advanced Solver Development for Large-scale Dynamic Building System Simulation

US National Institute of Standards and Technology (NIST)

Developed a new numerical solver based on a preconditioned Newton-Krylov method that significantly improved the computational efficiency of large-scale dynamic building system simulations using HVACSIM+, saving more than 90% of the computational time compared to the original solver while maintaining the same level of robustness.
Created AHU-VAV system models of various scales using HVACSIM+ to test the performance of the new solver, demonstrating a deep understanding of dynamic HVAC system modeling.
Developed a generic smoothing technique to improve the efficiency and robustness of building system simulations with discontinuities.
Investigated multiple strategies for constructing physics-based preconditioners to further improve the efficiency of dynamic building system simulations, particularly for systems with frequent on/off controls and setpoint reset strategies.