Collaborative Research: CDS&E: Leveraging hardware acceleration for accurate particle dynamics in turbulent flows

纳米粒子存在于日常生活的许多方面，从食品、药品、化妆品、纺织品、木材防腐剂到轮胎、电子产品和发动机。无论纳米粒子的类型如何，它的形成、生长和最终的破坏都涉及到导致许多不同形状和大小的粒子分布的物理过程。在大多数情况下，工业界需要更好的方法来控制颗粒分布；在其他情况下，目标是防止颗粒的形成。无论哪种方式，人们都需要了解粒子如何随着时间的推移而表现和进化，以实现这些目标。该项目旨在提高模拟复杂流场（例如，内燃机中的纳米颗粒）中纳米颗粒群的传输和进化的软件的预测能力。该项目开发的改进工具将通过更好地预测烟灰颗粒的形成、生长和破坏，使燃烧领域受益。这将支持更清洁的内燃机、燃气涡轮发动机和熔炉的设计。除了燃烧之外，通过该项目开发的工具和获得的理解可以通过确定可以利用哪些环境特性来增强/触发某些颗粒，增加某些类型颗粒的形成，或完全抑制它们来推进制造。最后，要开发的软件并不局限于固体氧化物纳米颗粒和煤烟。事实上，任何纳米到微尺寸材料的分散阶段都可以用这里开发的工具来描述，包括气溶胶、灰尘、等离子体中的带电粒子的分散，甚至是天体物理学中代表大型物体的“粒子”。除了该项目的科学目标外，pi还将参与公共宣传和教育工作，包括制作向公众解释计算流体动力学的视频，举办向研究人员教授软件技能的讲习班，并让来自不同背景和代表性不足的本科生参与研究。该项目将提高模拟纳米粒子在复杂流场中迁移和演化的数值框架的预测能力。研究目标包括：(1)在不影响物理精度的前提下，求解纳米粒子数密度函数的种群平衡方程（PBE）。该方法不仅提供了粒子的完整分布，而且还提供了它们的所有相关性质；(2)评估耦合流动求解器和PBE求解器的计算效率数值实现；(3)通过与其他研究小组共享开发的软件和合作，使人们能够对纳米颗粒在广泛领域的分布有新的物理和化学见解。这些目标将通过利用底层计算机体系结构的优势来实现：使用传统的中央处理单元（cpu）来解决流场和粒子传输，同时使用蒙特卡罗求解器在图形处理单元（gpu）上执行粒子种群的时间演化。pi计划将NGA软件作为开源发布，并围绕NGA建立一个用户社区，确保感兴趣的研究人员能够为代码库做出贡献。这将使该项目得到更广泛的发展。高级网络基础设施办公室的软件集群对这方面特别感兴趣，该办公室为该奖项提供了共同资金。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。

Nanoparticles are present in many aspects of everyday life, from food, drugs, cosmetics, textiles, and wood preservatives to tires, electronics, and engines. Regardless of a nanoparticle's type, its formation, growth, and eventual destruction involves physical processes that lead to distributions of particles of many different shapes and sizes. In most cases, industry wants better means to control the particle distributions; in other cases, the objective is to prevent the formation of particles. Either way, one needs to understand how particles behave and evolve over time to accomplish those goals. This project aims to improve the predictive capabilities of software for simulating the transport and evolution of populations of nanoparticles in complex flow fields (e.g., nanoparticles in a combustion engine). The improved tools developed in this project will benefit the combustion field by enabling better predictions of the formation, growth, and destruction of soot particles. This will support the design of cleaner internal combustion engines, gas turbine engines, and furnaces. Beyond combustion, the tools developed and understanding gained through this project could advance manufacturing by identifying which environmental properties can be leveraged to enhance/trigger certain particles, increase the formation of certain types of particles, or suppress them altogether. Finally, the software to be developed is not limited to solid-oxide nanoparticles and soot. In fact, any dispersed phase of nano- to micro-size materials can be described by the tools developed here, including dispersion of aerosols, dust, charged particles in plasmas, and even "particles" representing large objects in astrophysics. In addition to the scientific objectives of the project, the PIs will engage in public outreach and education efforts, including producing videos explaining computational fluid dynamics for the public, running workshops teaching software skills to researchers, and involving undergraduate students from diverse and underrepresented backgrounds in research. This project will improve the predictive capabilities of numerical frameworks for simulating the transport and evolution of populations of nanoparticles in complex flow fields. The research objectives include: (1) Solving the population balance equation (PBE) for nanoparticle number density functions with no compromise in physical accuracy. The approach provides not only the full distribution of particles but also all their relevant properties; (2) Evaluate a computationally efficient numerical implementation of a coupled flow solver and PBE solver; and (3) Enable new physical and chemical insights into nanoparticle distributions across a wide range of fields by sharing the software developed and working with other research groups. These objectives will be achieved by leveraging the strengths of the underlying computer architectures: using traditional central processing units (CPUs) to solve the flow fields and particle transport, while performing the temporal evolution of the population of particles on graphics processing units (GPUs) using a Monte Carlo solver. The PIs plan to release the NGA software as open source and build a user community around NGA by ensuring that interested researchers are able to contribute to the codebase. This will allow a wider growth of the project. This aspect is of special interest to the software cluster in the Office of Advanced Cyberinfrastructure, which has provided co-funding for this award.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.