CAREER: Using Stochastic Techniques to Understand and Predict the Flow of Non-spherical Particles

气体中含有密集悬浮粒子的流动被称为颗粒流，在自然界和工业中都很常见。在许多具有实际重要性的情况下，悬浮颗粒具有不规则形状（即非球形），这使得用现有方法预测悬浮颗粒的流动行为尤其具有挑战性。因此，许多微粒过程的设计往往依赖于昂贵的经验试错测试。这个CAREER项目将开发一个基于物理的随机模型，该模型可以解释不规则的粒子形状，从而更准确地预测大规模系统中的粒子动力学。该项目的结果将有助于将理想球形颗粒的颗粒流理论扩展到更现实的颗粒介质，并为颗粒技术中出现的技术挑战提供新的解决方案。该项目将包括研究生和本科生的研究培训，并将为他们可能从事的涉及粒子技术的职业做好准备。研究小组将与普渡大学工程拓展俱乐部合作，为当地学校的K-12学生展示颗粒物流动。CAREER项目的目标是使用随机方法开发一个基于物理的模型，用于预测包含数十亿粒子的系统中的粒子流动。目前最先进的非球形粒子离散元方法仅限于少于一百万个粒子。相比之下，一杯沙子含有大约1亿个颗粒。为了实现这一目标，该项目将开发高保真模拟，捕捉碰撞粒子的动力学，以构建大型系统的随机模型。将进行离散元模拟，以确定非球形粒子在碰撞过程中如何散射并重新分配旋转和平移能量。然后将使用机器学习工具来构建概率分布函数，将粒子的碰撞前状态与碰撞后状态联系起来。概率分布函数将被整合到直接模拟蒙特卡罗求解器中，该求解器可以模拟包含数十亿粒子的系统的动力学。为了验证随机模型，将首先与相对小规模系统的确定性离散元素模拟进行比较。然后将随机模型的准确性与文献中可用的实验数据和内部测试的数据进行比较，以评估更大规模系统的准确性。除了捕捉由于粒子形状效应而产生的复杂物理现象外，该项目还将创建一个框架，以改进对其他复杂现象（如粒子磨损和聚集）的预测。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。

Flows involving dense suspensions of particles in gas, which are known as granular flows, are common in nature and industry. In many cases of practical importance, the suspended particles have an irregular shape (i.e., non-spherical), which makes predicting the flow behavior of the suspension especially challenging with currently available methods. As a result, the design of many particulate processes often relies on costly empirical trial-and-error testing. This CAREER project will develop a physics-based stochastic model that accounts for irregular particle shapes to predict particle dynamics more accurately in large-scale systems. Results of the project will be useful in extending granular flow theory for idealized spherical particles to more realistic granular media and in providing new solutions to technical challenges that occur in particle technology. The project will involve research training for graduate and undergraduate students and will prepare them for possible careers involving particle technology. The research team will participate with the Purdue Engineering Outreach club to bring demonstrations of particulate flows for K-12 students in local schools.The goal of this CAREER project is to use stochastic methods to develop a physics-based model for predicting particle flows in systems containing billions of particles. Current state-of-the-art discrete element methods for non-spherical particles are limited to fewer than one million particles. By comparison, a single cup of sand contains approximately 100 million particles. To achieve this goal, the project will develop high-fidelity simulations that capture the dynamics of colliding particles to construct a stochastic model for large-scale systems. Discrete element simulations will be performed to determine how non-spherical particles scatter during collisions and redistribute rotational and translational energies. Machine learning tools will then be employed to build probability distribution functions that relate the pre-collision to the post-collision states of particles. The probability distribution functions will then be incorporated into a direct simulation Monte Carlo solver that can simulate the dynamics of systems containing billions of particles. To validate the stochastic model, comparisons will first be made to deterministic discrete element simulations for relatively small-scale systems. The accuracy of the stochastic model will then be assessed for larger scale systems by comparing results with available experimental data in the literature and with data from in-house tests. In addition to capturing the complex physics that arise due to particle shape effects, the project will create a framework for improving predictions of other complex phenomena such as particle attrition and agglomeration.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.