Enhancing Explanation to Improve Building-Design Support
PhD candidate | Andre Yu Li, M.Arch. |
Supervisor 1 | Prof. Ian F.C. Smith (computer science) TUM Georg Nemetschek Institute |
Supervisor 2 | Prof. Frank Petzold (architecture) Chair of Architectural Informatics |
Motivation and Goals
The success of building projects depends on effective communication and collaboration among multiple disciplines in the AEC fields. The growing number of design criteria and new strategies for design support have increased complexity, leading to greater risks of project delays and resource waste. Addressing such challenges through explicitly represented sequences has potential for ensuring transparent, accountable, and understandable decisions. This project involves exploring the potential of increasing explainability in building design, its importance, challenges, and emerging solutions. A framework is developed for incorporating explainability and design sequence into building-design support and for validating it on full-scale building-design cases.
Recent Results
A preliminary investigation was conducted to examine the impact of representing design sequence on outcomes in the context of generative design (Figure 2). As another aspect of the project, explanation strategies were compared and evaluated across various design methods, and these findings were presented at the recent GNI Symposium. A proposal has been submitted to CAADRIA 2025 to host a workshop titled “Dimensions of Explainability”, which aims to introduce the concept of explainability to researchers in this field and to gather insights on current trends in explainability, both generally and within the AEC sector. Searches are underway to find good design cases for assessing design sequences in full-scale projects.
Conference Articles
Yu-Li, A., Petzold, F., & Smith, I. F. C. (2024). Impact of Design Sequence on Building Design with Generative Methods. The 31st EG-ICE: International Conference on Intelligent Computing in Engineering, 559-568.
Li, A., & Smith, I. F. C. (2023). Explainability for Support of Open-Sequence Building Design. The 30th EG-ICE: International Conference on Intelligent Computing in Engineering, 52–61.https://www.ucl.ac.uk/bartlett/construction/sites/bartlett_construction/files/explainability_for_support_of_open-sequence_building_design.pdf