EAGER: Fast High-Accuracy Navier-Stokes Solvers for Reacting Flow Simulations

对反应流的科学理解，比如在火灾和汽车发动机中遇到的流动，通常依赖于模拟来深入了解流动的行为以及与化学反应的相互作用。这种模拟使预测能力能够增强隧道和防火建筑等安全基础设施的设计，以及燃煤锅炉等清洁燃烧反应堆的设计，并且具有国家重要性。不幸的是，传统的反应流模拟方法在计算和经济上都是昂贵的。例如，即使在最快的超级计算机上，单个高保真室内火灾计算也可能需要两周的计算时间。提出的研究旨在挑战传统的仿真方法，在不牺牲精度和保真度的情况下，消除仿真中昂贵的元素并用合适的近似值代替它们。所提出的方法有望将模拟成本降低一半，并且易于在现有软件中实现。这项工作将以更低的成本加快模拟的周转时间，从而增强反应流系统的设计和分析工作流程。本提案旨在通过用不影响精度或稳定性的合适近似代替昂贵的压力场计算，开发低马赫湍流反应流的快速高精度求解器。这些方法将使模拟复杂流（如火灾）的成本减半，从而加快周转时间，实现更高分辨率的计算。对于不可压缩流体的初步结果表明，与传统方法相比，该方法的加速可达60%。提出的方法将在Python中原型化，并使用制造解决方案的方法进行验证。然后，它们将在大规模的多物理场代码中实现，以研究和演示新算法的性能和可扩展性。所提出的算法不受反应模型、燃烧化学、辐射和其他多物理场现象的影响，这使得它们非常灵活，适用于非常广泛的变密度和反应流问题。该奖项反映了美国国家科学基金会的法定使命，并通过使用基金会的知识价值和更广泛的影响审查标准进行评估，被认为值得支持。

The scientific understanding of reacting flows, such as those encountered in fires and car engines, often relies on simulations to gain insight into how the flow behaves and interacts with chemical reactions. Such simulations enable predictive capabilities that enhance the design of safe infrastructure such as tunnels and fire-safe buildings, cleaner burning reactors such as coal boilers, and are of national importance. Unfortunately, conventional methods for reacting flow simulations are often expensive, both computationally and economically. For example, a single high-fidelity indoor fire calculation could take up two weeks of calculation even on the fastest supercomputers. The proposed research aims at challenging conventional simulation methods by eliminating the costly elements in a simulation and replacing them with suitable approximations, without sacrificing accuracy and fidelity. The proposed methods are expected to halve the cost simulations and are easy to implement in existing software. The work will enable faster turnaround times for simulations at a lower cost thus enhancing design and analysis workflows of reacting flow systems.This proposal aims to develop fast high-accuracy solvers for low-Mach turbulent reacting flows by replacing expensive calculations of the pressure field with suitable approximations that do not affect accuracy or stability. These methods will halve the cost of simulations of complex flows such as fires leading to faster turnaround times and enabling higher-resolution calculations. Preliminary results for incompressible flows showed speedups of up to 60% compared to conventional methods. The proposed methodologies will be prototyped in Python and verified using the method of manufactured solutions. They will then be implemented in a large-scale multiphysics code to study and demonstrate performance and scalability of the new algorithms. The proposed algorithms are agnostic to reaction models and combustion chemistry, radiation, and other multiphysics phenomena making them extremely flexible and applicable to a very wide range of variable density and reacting flow problems.This award reflects NSF's statutory mission and has been deemed worthy of support through evaluation using the Foundation's intellectual merit and broader impacts review criteria.