GPU-accelerated Updated Lagrangian Particle Hydrodynamics(ULPH) modeling of Heat Conduction and Convection Problems
Please login to view abstract download link
In this paper, an Updated Lagrangian Particle Hydrodynamics (ULPH) method incorporating high-order nonlocal differential operators is employed to simulate fluid heat conduction and natural convection processes. The introduction of high-order matrix inversion significantly increases computational demand. Therefore, GPU parallel acceleration was implemented within the ULPH framework. We have conducted several simulations including natural convection within a cubic cavity and between concentric cylinders. The results show that the ULPH method aligns well with existing literature, and the GPU acceleration efficiency improves with the increase in computational scale. This method exhibits inherent accuracy and convergence, making it promising for simulating complex fluid dynamics scenarios.
