Yakim Milev
Defining Architectural Typologies Through Structural Topologies and Machine Learning
Summary
This practice-led research explores the problem of stadia design in relation to design parameters and drivers, morphological classification, and machine learning. Stadia are one of the largest and most challenging genres to construct, but little has been done to understand how they are designed as an interconnected system of design drivers in contrast to the traditional compartmentation of the design process into separate disciplines. This research examines the benefits and limitations machine learning brings to the complexity of stadia design and operation in relation to efficiency of use and a structured approach to understanding stadia morphology, form generating design drivers and adaptability to change.
Research questions
The research asks: How can an integrated structural and architectural design process delineate a new understanding of stadia as a genre and a morphological typology? What are the specific parameters and design drivers that define the design of stadia? What are the needs of the building industry and society as a whole towards this genre, how do they change and what useful design responses does a machine learning supplemented design process provide?
Methodology
The project methodology uses approaches from both architectural and structural engineering fields. Due to the specific requirement for an input dataset for machine learning analysis the first stage of the research involved the construction of a typological framework through Grounded theory and the selection of an architectural category for further analysis. The initial typological analysis of the case studies explored the larger architectural genre of assembly buildings and through machine learning classification of architectural and structural plans and 3d models identified the most influential design drivers. These design parameters served for the construction of a hierarchy three for the classification of the case studies and the subsequent selection of research focus. In terms of organisational and morphological classification stadia presented distinct characteristics that are not present in the other examples of assembly buildings. In order to validate the design drivers the research also focused on the historical development of stadium design parameters and relevant legislation. The second phase of the research takes practical approaches from the field of structural engineering like algorithmic design and uses empirical study in order to validate the construction of an analytical machine learning model. This model is based on a parametric 3d model that generates simplified digital representations of the different morphological typologies. The parametric model is controlled with machine learning and evaluates the design efficiency through travel distances, distances to programmatic features, and sightline quality. The third phase of the research brings functional change to the machine learning model and explores the different morphological and organisational consequences of adaptability to changing requirements.
Machine Learning Based Typological Analysis with Quantified Spatial Data
