ACCELERATING TURBULENCE MODEL SIMULATIONS WITH A HIGHEFFICIENCY GPU-OPTIMIZED ALGORITHM

Resumo

© 2025, Begell House Inc. All rights reserved.This study presents an in-depth comparative analysis with a widely used turbulence model in Computational Fluid Dynamics (CFD): the standard k-ε model. The research focuses on turbulent flow over a backward-facing step (BFS), a classical problem known for its complex recirculation and reattachment phenomena. Simulations were conducted using both an innovative Graphics Processing Unit (GPU) based parallel processing algorithm developed on the Nvidia Compute Unified Device Architecture (CUDA) platform, and a Central Processing Unit (CPU) based commercial software. The numerical simulation analysis spans a broad range of Reynolds numbers, representing different levels of turbulence intensity, and compares the performance of these two approaches. The primary objective of this study is to evaluate the predictive capabilities of the standard k-ε model in terms of reattachment length, a critical parameter for accurately capturing the dynamics of separated flows. The simulation results obtained from both software platforms are rigorously compared with classical experimental data at ReH = 36,000 to assess the accuracy and reliability of each approach. The GPU simulations were performed on an Nvidia GeForce RTX™ 3090Ti with 24 GB of video memory, while the commercial simulations were run on an Intel®Core™ i7-12700H CPU, featuring a 2.3 GHz base clock and 14 cores. The results indicate that GPUs offer a more optimal architecture for CFD problem-solving, leveraging large-scale computational parallelization.

Palavras-chave