The built environment generated 37% of global CO2 emissions in 2021, reported by UN environment programme. It is estimated that more than 170 billion m2 housing and office floor area will be added globally by 2050, meaning that floor area equivalent to the size of Paris (> 100 km2), will need to be created on a weekly basis. To reduce carbon emissions in the buildings sector, it is crucial that the amount of material used to create building structures is kept to a minimum. So it is vitally important to design buildings that are structurally efficient.
Structural layout/topology optimization provides a powerful means of designing highly efficient building structures. Since the method is based on mathematical derivations (e.g., gradient), it requires clearly defined design inputs to function correctly. This often makes the method appear ‘rigid’, lacking flexibility to deal with a large spectrum of problems found in practice. Furthermore, its computational efficiency is likely to be significantly influenced by problem complexity, so only limited types of problems can be solved. On the other hand, versatile optimization methods are also available, though mostly using heuristic algorithms that are often developed via human intuition. The effectiveness of these methods is likely to be heavily influenced by user’s experience, and the results are often found to gain small incremental improvements that may be insufficient to properly address the urgency of the climate crisis.
To improve the flexibility of the structural layout optimization method, this PhD project will develop a novel AI-in-the-loop optimization framework, permitting mathematical and heuristic optimization methods to work seamlessly to tackle problems within their domains of proficiency. The use of the standard layout optimization method will provide rigorous mathematical foundations to generate highly efficient structural designs (e.g., potentially significant savings of building materials), while the adoption of AI will enable experience-based heuristic methods to tackle a wide range of design problems. The PhD research project involves several elements: first, challenging design problems will be identified and then simplified via, e.g., supervised learning, to provide approximated models that can be incorporated in mathematical optimization algorithms. Second, subjective design factors, such as aesthetics or unforeseen requirements, will be addressed via heuristic algorithms such as fuzzy logic to generate design galleries instead of a unique solution. Finally, an interactive AI-in-the-loop optimization framework will be developed to solve relatively challenging problems that cannot be tackled effectively via standard layout optimization methods.
This is an ambitious PhD project which offers many fantastic opportunities to highly motivated students:
• Participate in world-leading research in structural optimization, for designing highly efficient next-generation building structures.
• Contribute towards the NetZero target, one of the most urgent and challenging missions in our generation.
• Learn and exploit a wide range of powerful tools such as layout optimization, parametric modelling and AI.
• Engage with the construction industry.
• Develop software tools to deliver impact.
Academic requirements: A Master's degree in relevant fields such as Structural Engineering, Computer Science, Mechanical Engineering or related disciplines.
The start date is 01 October 2024.
Interested candidates are strongly encouraged to contact the project supervisors to discuss your interest in and suitability for the project prior to submitting your application. Please refer to the EPSRC DTP webpage: https://www.sheffield.ac.uk/postgraduate/phd/scholarships/science-engineering for detailed information about the EPSRC DTP and how to apply.
Continue with Facebook
