波浪环境下水上飞机喷溅性能预测中的无网格并行方法研究

The spray performance prediction of water-based aircraft in the wave environment is a typical fluid-solid coupling problem, for which the meshfree parallel computing is an important method. Finite Particle Method (FPM) is a generalized SPH meshless particle method, which has global two order accuracy. However, in face of the complex moving interface and large-scale data calculations, the shortcomings of FPM are obvious, including computational instability and long computing time. This project will focus on the key interface algorithm and GPU parallel computing in the spray simulation of water-based aircraft by FPM. First, the FPM-DSPH interface coupling equation will be derived based on the matrix theory, and further an interface double-coupling algorithm will be proposed to improve the interface calculation accuracy and stability by introducing the contact force function into FDM-DSPH equation. Second, based on the CUDA platform, the GPU-FPM parallel codes will be developed, and the memory management and data exchange details will be optimized to improve the parallel efficiency according to the calculation characteristics of FPM. Final, the large-scale meshfree parallel wave water area will be constructed and the spray performance of the full-scale aircraft will be calculated. Meanwhile, a fluid-solid coupling simulation software platform will be built. The research results of this project could also provide some theoretical and technical support for the anti-wave capacity prediction and structural strength design of ship and offshore platform.

波浪环境下水上飞机喷溅性能预测是典型的流固耦合问题，无网格并行计算是解决此类问题的重要方法。有限粒子法是一种具有全域二阶精度的广义SPH方法，但在面对飞机喷溅中的复杂运动界面和大规模数据计算时，其数值不稳定和计算效率低的缺点十分明显。基于此，本项目拟针对有限粒子法在飞机喷溅性能预测中的关键界面算法及其GPU并行计算展开研究。首先，利用矩阵论相关方法推导FPM-DSPH界面耦合方程，同时引入接触力函数，建立固液界面双重耦合算法，提高界面计算精度和稳定性；其次，基于CUDA平台开发GPU-FPM并行程序，结合FPM计算特点提出内存管理和数据交换优化策略，提高并行效率；最后，基于上述研究工作实现大规模无网格并行化波浪水域构建及全尺寸实机喷溅性能的数值预测，搭建流固耦合动力学仿真软件平台。项目研究成果可同时为船舶、海洋工程平台的抗浪能力预测、结构强度设计等提供一定的理论和技术支撑。

水上飞机水面滑行喷溅性能预测是一个典型的流固耦合问题，涉及复杂多相界面、数值波浪水池构建及大规模数值计算等多项关键问题亟待解决。本项目从算法理论研究、并行程序开发和滑水仿真计算三个方面层层递进开展研究。具体来说，首先基于矩阵论的相关理论和方法，开展了无网格FPM修正算法的理论研究，并进一步推导FPM-DSPH界面耦合方程，建立了多种方法自适应计算的无网格计算框架。其次，突破了流固耦合仿真计算时涉及的多项关键技术，基于CUDA平台完成了无网格流固耦合仿真计算的单GPU和多GPU程序开发与优化。最后，基于无网格框架建立了数值波浪水池构建的数学模型，基于前述算法理论成果和GPU并行程序实现了水上飞机在平静水面和波浪水面上拖曳滑行数值仿真与喷溅性能预测。本项目的研究成果可为结构物入水、滑行溅水等涉及自由液面大变形问题的仿真计算提供有力的理论和技术支撑。