Resource-efficient, Adaptable and Practical Digitally Designed (RAPID) Structures

The built environment is the dominant user of highly carbon-intensive materials such as concrete and steel, leading to vast carbon emissions as well as global habitat loss and pollution. The Institution of Structural Engineers has neatly summarised the most effective strategy for combatting these harms – ‘Use less stuff’. Structural optimization is a powerful digital approach which finds out how much ‘less stuff’ it is physically possible to use whilst ensuring safety and functionality in a given scenario, along with the structural form which obtains this optimal material-efficiency. Applying such techniques at the early stages of the design process, before the initial structural concept has been fixed, provides the greatest possible material savings, with case studies suggesting reductions in material usage (and hence embodied carbon) of up to 50%. To achieve these levels of material savings requires the highest levels of design freedom to be permitted, i.e. requiring so-called topology or layout optimization. However, the current state-of-the-art methods available for this are not yet ready for widespread implementation within the built environment. Thus, the aim here is to develop a framework suitable for optimization-driven design at the concept stage, by addressing the key blockers below. Any real structure must resist a huge variation in loadings during its lifetime, yet current methods become impractical with more than a handful of loading cases. Furthermore, within an optimization-driven framework, it is typically unclear which cases may be critical for inclusion in a simplified initial phase. Incorrect decisions can lead to late changes at huge financial and environmental cost, or in the worst case even unsafe structures. The first objective of this project is therefore to develop effective techniques which will allow all loading cases to be considered from inception. A further limitation is that the structural forms which obtain the minimum material usage are complex and impractical to construct. However, previous studies have shown that significant rationalisation can be obtained with only a small increase in material usage – but suitable methods are needed to be able to identify these rationalised solutions in a timely manner. Developing such methods is the second goal of this project. The final objective is to facilitate the holistic design of structures, to permit these benefits to be realised for common scenarios which involve loads distributed over an area such as a floor or a façade. This will require combined optimization of span or panelling layouts and structural form. Sustainability of the methods themselves is also a key part of this project, attained by using rigorous and computationally efficient mathematical programming approaches. This avoids the need for power-hungry supercomputing time, or additional model training processes. Furthermore, these approaches provide exact, globally optimal benchmark values, which can be used to inform decision-making on more general questions, such as whether new construction is the right answer, or on the dichotomy between lean design vs design for future flexibility. The computational efficiency of these methods means they can be run interactively on standard laptop/desktops and provide solutions quickly to engage with fast-moving design workflows. To facilitate this in real-world settings, an open-source Python framework will be developed, accompanied by interfaces to industry-standard software and easy to use web-apps. Through this, it is envisioned that real-world adoption of these approaches will be accelerated, providing a crucial leap in the march towards net-zero.