A new CFD method enhanced by machine learning for violent free surfaces

通过机器学习增强的新 CFD 方法，适用于暴力自由表面

• 批准号: 22K14332
• 负责人: カムラ モハメド・ムスタフア・ザキ・アハメド
• 金额: $ 2.66万
• 依托单位: Kyushu University
• 依托单位国家: 日本
• 项目类别: Grant-in-Aid for Early-Career Scientists
• 财政年份: 2022
• 资助国家: 日本
• 起止时间: 2022-04-01 至 2023-03-31
• 项目状态: 已结题
• 来源: https://kaken.nii.ac.jp/en/grant/KAKENHI-PROJECT-22K14332/
• 关键词: THINC; multiphase; volume of fluid; HPC; Machine learning; multiphase flow

Preliminary results for a new formula that will replace the currently employed Tangent Hyperbolic function in the THINC VOF scheme utilized for interface capture treatment in multiphase flow have been achieved. The proposed function, unlike the Tangent Hyperbolic function, can be integrated twice, saving computing costs associated with numerical integration. In a subset of test scenarios, the proposed scheme gives results that are comparable to state-of-the-art THINC techniques at around half the computational cost. However, the current results are still limited to Cartesian (Hexahedral mesh) grids, and there are some difficulties in extending the new approach to different types of unstructured grids.

一个新的公式，将取代目前采用的切线双曲函数在THINC VOF计划用于界面捕获处理多相流的初步结果已经实现。与切线双曲函数不同，所提出的函数可以积分两次，节省了与数值积分相关的计算成本。在测试场景的一个子集，所提出的计划给出的结果是相媲美的最先进的THINC技术在大约一半的计算成本。然而，目前的结果仍然局限于笛卡尔（六面体网格）网格，并有一些困难，扩展到不同类型的非结构化网格的新方法。