GOALI: Computational Multibody Dynamics: Addressing Modeling and Simulation Limitations in Problems with Friction and Contact

这个GOALI项目汇集了威斯康星大学麦迪逊分校的一个研究小组和卡特彼勒的工程师团队，旨在调查和实验验证能够通过计算机模拟预测大型动态系统时间演化的建模和数值解方法。拟议的研究、开发和验证工作将增加计算机模拟在研究过程中所发挥的作用，如运输、混合和压实离散材料(粉末、颗粒、砾石、岩石)。尽管这些过程与广泛的行业相关：例如，制造业、建筑业、农业、食品和制药业，但除非问题的规模和/或分析的持续时间大大缩短，往往低于实际有用的限度，否则它们通过计算机建模和模拟进行的调查在今天提供的好处有限。更好地理解大型多体系统的动力学是很重要的，因为在水之后，离散介质在工业中被操纵得最多。此外，全球约50%的贸易产品是颗粒状的，它们的制造、运输、加工、包装和混合构成了具有挑战性的问题，预计本研究的结果将提供有价值的见解。要研究的颗粒系统有数百万到数十亿个成分，它们的集体行为主要是由单个元素之间的摩擦接触相互作用所主导的。研究工作将集中在(1)研究两类求解方法，即Krylov子空间方法和原始-对偶内点方法，以解决当前由于实际应用的多尺度方面而导致的数值解的局限性；(2)研究增广拉格朗日公式，以了解它们是否可以增强互补方法在多体动力学中所起的建模作用；以及(3)从精度、稳健性和可扩展性方面比较两类求解方法，即基于惩罚的和基于互补的方法，以汇编事实证据，总结每种方法的优点和缺点。

This Grant Opportunity for Academic Liaison with Industry (GOALI) Program project brings together a research group from the University of Wisconsin-Madison and a team of engineers from Caterpillar in a joint effort that aims at investigating and experimentally validating modeling and numerical solution methods capable of predicting through computer simulation the time evolution of large dynamic systems. The proposed research, development, and validation effort will increase the role that computer simulation plays in investigating processes such as transporting, mixing, and compacting discrete material (powders, grains, gravel, rocks). Although these processes are relevant in a broad spectrum of industries: e.g., manufacturing, construction, farming, food and drug, their investigation through computer modeling and simulation provides limited benefit today unless the problem size and/or duration of the analysis are drastically curtailed, often below practically useful limits. Better understanding the dynamics of large multibody systems is important since after water, discrete media are manipulated in industry the most. Moreover, about 50% of all traded products worldwide are in granular form and their manufacturing, transport, processing, packing and mixing pose challenging problems in which the outcomes of this research is anticipated to provide valuable insights. The granular systems to be investigated have millions to billions of components whose collective behavior is dominated by frictional contact interactions between individual elements. The research effort will concentrate on (1) investigating two classes of solutions methods; i.e., Krylov-subspace and Primal-Dual Interior Point methods, to address current numerical solution limitations that stem from the multiscale aspect of practical applications; (2) investigate augmented Lagrangian formulations to understand whether they can enhance the modeling role that complementarity approaches play in multibody dynamics; and (3) compare two classes of solution methods; i.e., penalty-based and complementarity-based, in terms of accuracy, robustness and scalability in order to compile a factual body of evidence that summarizes the advantages and disadvantages of each approach.