Architecture Engineers Award

Introduction Ice-shell architecture in cold regions faces significant challenges related to snow accumulation, radiation exposure, and long-term weatherability. While snow distribution data is critical for optimizing architectural performance, current design practices lack effective tools for predicting snow drift during early design stages. This study addresses this gap by proposing a computational snow drift simulation method tailored to ice-shell architecture, enabling architects to integrate environmental performance considerations into early-stage design decision-making. Limitations of Existing Snow Drift Prediction Methods Traditional snow drift simulation techniques are often complex, computationally demanding, and inaccessible to architects during conceptual design phases. These limitations restrict their application primarily to engineering validation rather than design exploration. This section discusses the technological and methodological gaps in existing approaches, hig...

With the growing integration of artificial intelligence in architectural practice, machine learning (ML) has emerged as a promising tool for enhancing design efficiency and decision-making. While ML techniques are well known for identifying complex patterns within large datasets, their application in predicting architectural design parameters—particularly for form generation—remains relatively underexplored. This study investigates the feasibility of a machine learning–based framework capable of predicting numerical design parameters to support the generation of architectural form in a controlled, human-centered design context. Complexity of Architectural Form Generation Architectural form generation is influenced by multiple interdependent factors, including spatial logic, functional requirements, and human-centered considerations. These factors introduce a high level of complexity that challenges conventional rule-based or deterministic design approaches. This section discusses ho...