Development of Key Design Strategies for Hot Form Quench (HFQ)

热成形淬火（HFQ）是伦敦帝国理工学院发明的一种新的热成形技术，能够为汽车和航空航天工业提供轻量化解决方案。HFQ能够生产具有成本效益和复杂形状的结构，具有理想的几何特征，包括通过高强度，轻质铝合金的紧密角半径和尖锐的强调造型线。然而，目前对其应用和设计能力的了解不足，这意味着结构设计师经常会忽视HFQ，而不是充分利用它的潜力。这项工作旨在开发先进的设计方法，以指导首次使用HFQ技术的制造优化设计。这项工作分为三个子项目，旨在解决HFQ的尖端设计挑战。简而言之，这些是：生产具有紧密角的深拉部件的设计方法，具有a级表面光洁度的成型部件的设计方法，以及应用新兴的机器学习领域的先进技术来辅助HFQ工艺设计的方法。迄今为止，该项目承担的主要任务包括：-进行文献综述，以确定目前使用HFQ技术成形部件的复杂性和挑战，并了解与成形a类表面相关的关键要求挑战-回顾应用于固体力学应用的尖端机器学习方法；有可能应用于这项工作-初步研究，包括几个设计变量对各种成形响应的影响，以及在这项工作中使用机器学习方法的工作流程。进一步探讨了影响角形成形厚度分布的因素。关键子问题涉及（HFQ条件下的平均冷工具和热毛坯）：总结表A.2在等温条件下模拟运行中材料、加工和几何因素对形成紧角盒形的影响。探讨上述因素的综合影响，建立等温条件下弯角成形响应的趋势和敏感性。c.考虑HFQ条件，考察不同工艺参数下弯角成形时HFQ工艺的非等温特性。研究了高热量条件下传热和接触压力对紧半径的影响。从文献中探索新的工具策略，用于HFQ成形角，例如宏观纹理工具表面（Zheng等人，2017）和半径轮廓（Wang & Masood, 2011）。影响因素对表面缺陷影响的研究a.材料、加工和几何因素对刀坯界面接触压力的影响（从附录B开始），比较等温条件和HFQ条件B。探讨高通量条件下上述因素对接触压力的综合影响。通过实验研究（在印象技术有限公司），分析和关联表面质量和缺陷与接触压力和有限元成形应变在HFQ条件下。开发a类表面的设计方法并进行HFQ成形试验3。通过进一步研究设计支持工具的机器学习方法，开发HFQ的最佳设计策略和平台；使用结果1。和2。划分复杂的大型设计域，并研究基于可制造性约束（如细化约束）元模型的深度神经网络。研究基于图像的复杂几何图形（点1.e.）的泛化方法，并使用cnn预测全场有限元结果的图像（Zimmerling等人，2019b）。将基于可制造性的约束集成到HFQ优化平台中

Hot Form Quench - HFQ - is a new hot forming technology invented at Imperial College London capable of providing lightweighting solutions to the automotive and aerospace industries. HFQ enables the production of cost-effective and complex-shaped structures, with desirable geometric features including tight corner radii and sharp accent styling lines through high strength, lightweight aluminium alloys. However, there is currently insufficient knowledge on its application and design capability, meaning that HFQ can often be overlooked by structural designers and not utilised to its full potential. This work aims to develop advanced design methodologies to guide optimal design for manufacturing using HFQ technology for the first time. The work is divided into three sub-projects, aimed at tackling cutting edge design challenges for HFQ. Briefly, these are: design methodologies for producing deep drawn components with tight corners, design methodologies for forming components with a Class-A surface finish and methods to apply advanced technologies from the newly emerging field of Machine Learning to aid in the design for HFQ process.Major tasks undertaken in the project to date include:- a literature review to identify the intricacies and challenges currently associated with forming components using HFQ technology, and to understand key requirements challenges associated with forming Class-A surfaces- a review of Cutting edge machine learning methods applied to solid mechanics applications, which have the potential to be applied to this work- preliminary research including the influence of several design variables on various forming responses and a workflow for using machine learning methods for this workFuture work includes the following:1. Further investigation into the effects of influential factors on the thickness distribution for forming corner shapes. Key sub-problems involve (HFQ conditions below mean cold tools and hot blank):a. Summarise the effect of material, processing and geometry factors in forming box shapes with tight corners from the simulation runs in Table A.2, under isothermal conditionsb. Explore combined effects of the above factors and establish trends and sensitivities with corner forming responses, under isothermal conditions c. Investigate the non-isothermal nature of the HFQ process when forming tight corners under various processing parameters, by considering HFQ conditions.d. Investigate heat transfer and contact pressure effects on tight radii under HFQ conditionse. Explore novel tooling strategies from the literature, for HFQ forming corners, such as macro textured tool surfaces (Zheng et al., 2017) and radii profiles (Wang & Masood, 2011)2. Investigation on the effect of influential factors on the surface defects a. Effect of material, processing and geometry factors on contact pressures at the tool-blank interface (started in Appendix B ), comparing isothermal conditions with HFQ conditionsb. Explore combined effects of the above factors on contact pressures under HFQ conditionsc. Through experimental studies (at Impression Technologies Ltd), analyse and correlate surface quality and defects with contact pressures and FE forming strains under HFQ conditionsd. Development of design methods for Class-A surfaces and conduct HFQ forming trials3. Development of an optimal design strategy and platform for HFQ by further investigating machine learning methods for design support tools:a. Use outcomes of 1. and 2. to divide complex, large design domains and investigate deep neural networks for metamodels of manufacturability based constraints (e.g. thinning constraints)b. Investigation into image-based methods for complex geometries (point 1.e.) for generalisation and using CNNs to predict images of full field FE results (Zimmerling et al., 2019b)c. Integration of the manufacturability based constraints into an optimisation platform for HFQ