Generative Design for a New Generation of Resilient Steel Structures

在过去的20年里，由于故意攻击、事故、设计或维护错误以及极端环境行为引发的结构性倒塌事件不断增加。因此，进行了深入的研究，以更好地了解建筑物在未明确设计的荷载下的结构行为，也许这是无法合理预见的。重点一直放在理解和减少所谓的“不成比例崩溃”的可能性上，这种崩溃是指崩溃的程度与其引发事件的规模完全不成比例的事件的展开，往往发生在从一个结构要素或系统到另一个结构要素或系统的渐进“多米诺骨牌”效应中。尽管在罕见的，但极端的，加载情况下的结构行为的理解的进步，设计建筑物的方法仍然在很大程度上是传统的，没有范式转变的方式结构的设计和建造的明确目的，提高结构的弹性。相反，弹性被视为一个单独的设计后评估，而不是从一开始就融入设计过程。随着最近在计算能力和数据科学方面的突破性创新，已经创造了一个丰富的环境，其中生成设计技术不仅有可能创建先进的分析解决方案，而且还可以将我们陈旧的结构系统进化为以前从未设想过的先进系统。生成式设计方法涉及基于高级设计目标和参数化约束的多个解决方案的生成。这一过程本质上摆脱了过去经验中存在的偏见，这些偏见往往会限制创造力，并使用数据随着时间的推移不断改进。进化原理的使用利用了计算和数据的巨大力量，同时将所有最终设计决策留给工程（人类）专家的神圣领域。总之，拟议的研究，通过使用生成设计原理和算法，将证明，不仅可以大大提高结构的弹性，但它可以在一个适度的成本溢价。该研究项目的成果将为结构工程师提供必要的工具，用于设计能够阻止倒塌的建筑物，将结构弹性作为从概念设计阶段到完工的基本设计目标。该研究预计将对加拿大钢结构建筑设计和施工的标准实践产生重大影响。

Over the past 20 years, there has been a perceived proliferation of structural collapses triggered by intentional attacks, accidents, errors in design or maintenance, and extreme environmental actions. As a result, intensive research has been carried out to acquire a better understanding of structural behaviour under loading for which a building was not explicitly designed, and perhaps that could not have reasonably been foreseen. The emphasis has been squarely on understanding and diminishing the possibility of what has become known as a "disproportionate collapse", an unfolding of events wherein the extent of collapse becomes completely disproportionate to the magnitude of its initiating event, and often taking place in a progressive "domino" effect from one structural element or system to another. Despite advances in the understanding of structural behaviour under rare, but extreme, loading scenarios, the approach to designing buildings has remained largely conventional, with no paradigmatic shift in the way structures are designed and built for the express purpose of enhancing structural resilience. Instead, resilience is treated as a separate post-design assessment, rather than woven into the design process from the outset. With recent ground-breaking innovations in both computing power and data science, a rich environment has been created wherein generative design techniques have the potential not only to create advanced analytical solutions, but also to evolve our well-worn structural systems into advanced systems never before conceived. The generative design method involves the generation of multiple solutions based on high-level design goals and parameterised constraints. This process inherently sheds pre-existing biases from past experience that tend to limit creativity, and uses data to engender continual improvements over time. The use of evolutionary principles leverages the enormous power of computing and data, while leaving all final design decisions in the sacrosanct domain of the engineering (human) expert. In summary, the proposed research, through the use of generative design principles and algorithms, will demonstrate that not only can greatly enhanced structural resilience be achieved, but it can be done at a modest cost premium. The research program outcomes will provide structural engineers with essential tools for designing buildings capable of arresting collapse by regarding structural resilience as a fundamental design objective from the conceptual design stage through to completion. The research is expected to have a significant impact on standard practice for the design and construction of steel buildings in Canada.