FlowBench
收藏资源简介:
FlowBench是由爱荷华州立大学创建的一个大规模流体模拟数据集,包含超过10,000个样本,旨在评估神经PDE求解器在复杂几何形状上的性能。数据集涵盖了多种复杂几何形状(参数化和非参数化)和流体条件(雷诺数和格拉晓夫数),捕捉了从稳态到瞬态的各种流体现象。每个样本都包含速度、压力和温度场的数据,以及升力、阻力和努塞尔数等工程特征。FlowBench的创建过程包括使用高保真模拟器进行直接数值模拟,确保数据的准确性和可靠性。该数据集主要应用于工程领域,如航空航天、汽车设计和生物流体学,旨在解决复杂几何形状上的流体动力学问题。
FlowBench is a large-scale fluid simulation dataset created by Iowa State University, containing over 10,000 samples, designed to evaluate the performance of neural PDE solvers on complex geometries. The dataset encompasses a variety of complex geometries (parametric and non-parametric) and fluid conditions (Reynolds number and Grashof number), capturing diverse fluid phenomena ranging from steady-state to transient flows. Each sample includes data of velocity, pressure and temperature fields, as well as engineering features such as lift force, drag force and Nusselt number. The creation of FlowBench involved direct numerical simulations using high-fidelity simulators, ensuring the accuracy and reliability of the dataset. This dataset is mainly applied in engineering fields such as aerospace, automotive design and biofluid dynamics, aiming to solve fluid dynamics problems on complex geometries.

- 1FlowBench: A Large Scale Benchmark for Flow Simulation over Complex Geometries爱荷华州立大学 · 2024年



