Collaborative Research: Uncertainty Quantification and Model Validation in Thin-Walled Structures: A Probabilistic Paradigm for Advancing Analysis-Based Design

本项目将研究薄壁结构非线性稳定性分析的先进框架，同时考虑与几何缺陷相关的不确定性。薄壁结构最常见的失效模式是屈曲失效。这种屈曲是良性的还是灾难性的，在很大程度上取决于结构的细节，最重要的是它的几何缺陷。 研究的目的是提供一个飞跃的不确定性量化和模型验证的薄壁结构的分析，通过设计一个概率范式，使用先进的不确定性量化技术沿着与新的结构分析方法。几何不确定性的表征和表示方法以及概率缺陷灵敏度分析框架的建议，有可能铺平道路的薄壁结构的完全随机非线性分析。所提出的技术是由高精度的缺陷实验测量和不完美的members.If成功的测试补充，这项研究的结果将改变如何在薄壁结构的先进的基于分析的设计中处理几何缺陷。由于该项目中采用的随机方法具有非常普遍的性质和几何学驱动的性质，所开发的方法将适用于各种各样的问题，并将对所有科学和工程学科产生潜在的影响。除了研究生指导，本科生研究将是马萨诸塞州达特茅斯大学（UMD）和约翰霍普金斯大学（JHU）努力的重点。参与的本科生将被要求与当地高中和社区大学的学生以及参加UMD大一暑期学院的学生分享他们的经验，希望他们的成功能被其他人效仿。此外，与专门针对代表性不足的群体的计划合作，邀请合格的学生参加这两所大学的研究。除了通过会议和论文进行学术传播，并将所有开发的工作作为开源提供外，PI还将利用其参与工程委员会（例如AISC，AISI等）的机会。在学术界之外传播研究结果。

This project will pursue research towards an advanced framework for nonlinear stability analysis of thin-walled structures taking into account the uncertainty associated with geometric imperfections. Thin-walled structures most commonly fail in buckling failure mode. Whether this buckling is benign or catastrophic depends, to a great extent, on the details of the structure, most importantly its geometric imperfections. The research aims at providing a leap towards uncertainty quantification and model validation in the analysis of thin-walled structures by devising a probabilistic paradigm that uses advanced uncertainty quantification techniques along with novel structural analysis methods. Methodologies for the characterization and representation of geometric uncertainties as well as probabilistic imperfection sensitivity analysis frameworks are proposed that have the potential to pave the way for a fully stochastic nonlinear analysis of thin-walled structures. The proposed techniques are complemented by experimental measurements of imperfections with high accuracy, and testing of imperfect members.If successful, the results of this research will transform how geometric imperfections are handled in advanced analysis-based design of thin-walled structures. Because of the very general and mathematics-driven nature of the stochastic approaches adopted in this project, the methodologies developed will be applicable to a wide variety of problems and will have potential impact across all science and engineering disciplines. Alongside graduate student mentoring, undergraduate research will be a focus of the efforts at both University of Massachusetts Dartmouth (UMD) and Johns Hopkins University (JHU). The involved undergraduates will be asked to share their experience with students at local high schools and a community college as well as students attending Freshman Summer Institute at UMD, in hopes that their success will be emulated by others. Also, in collaboration with programs that specifically target underrepresented groups, qualified students are invited to participate in research at both universities. In addition to academic dissemination through conferences and papers and providing all developed work as open source, the PIs will capitalize on their involvement in engineering committees (e.g. AISC, AISI etc.) to disseminate the findings outside of the academic community.