Parametric Archetype: A Synthetic Digital Method of Building Material Stock Representation Based on Distance Measurement


Loading...

Date

2023-10-21

Publication Type

Conference Paper

ETH Bibliography

yes

Citations

Altmetric

Data

Abstract

Building material stock (BMS) is a crucial inventory of secondary resources which contain comprehensive information for analyzing the potential of material reuse and urban harvesting. Due to the complexity of urban building systems and the large number of buildings, obtaining building information one by one is impractical. Existing methods for stock representation mainly start from data collection, and utilize techniques such as clustering, machine learning, computer vision, et cetera, to process and analyze large and complete datasets. However, it is noticed that data on urban buildings, especially for building materials, is very limited or rather inaccessible. Existing methods cannot be applied in data-scarce cities and are also challenging to update over time. Therefore, this study proposes a synthetic approach named parametric archetype for the digital representation of BMS. This approach combines distance measurement, which is a distance within dimensions describing building features, to match instance buildings dynamically to a parametric archetype with the highest similarity. The weight and types of different building features, which may influence building material (composition and properties) in distance measurement, can be determined by supervised, semi-supervised, or unsupervised learning, whether relying on ample available data or domain rules/expert knowledge when data is scarce. This way, the parametric archetype model can use data more efficiently to form a synthetic and extensible representation for urban-level BMS (Figure 1). The parametric archetype is anticipated to offer an approach for describing, quantifying, and modeling the real building material stock system incrementally and transparently.

Publication status

published

External links

Book title

Proceedings of the 43rd Annual Conference of the Association for Computer Aided Design in Architecture

Journal / series

Volume

2

Pages / Article No.

44 - 52

Publisher

Association for Computer Aided Design in Architecture

Event

43rd Annual Conference of the Association for Computer Aided Design in Architecture (ACADIA 2023)

Edition / version

Methods

Software

Geographic location

Date collected

Date created

Subject

Organisational unit

08058 - Singapore-ETH Centre (SEC) / Singapore-ETH Centre (SEC)
08060 - FCL / FCL

Notes

Conference lecture held on October 27, 2023

Funding

Related publications and datasets