Value Diversity as a Driver for Renovation Design Support: A Clustering-Based Approach to Accelerate the Exploration of Design Space

Publikation: Bidrag til bog/antologi/rapport/proceedingKonferencebidrag i proceedingsForskningpeer review

Sustainable building renovation design is a highly challenging and ever more pressing issue in the construction sector. Architects must search through a large number of potential designs while aiming to optimize a range of Key Performance Indicators (KPIs) that measure, for example, construction cost, energy efficiency, an occupant’s sense of privacy, and so on.
Software tools for supporting this task often seek to automatically discover renovation solutions (or scenarios) through search procedures that optimize certain criteria, and deliver a small number of similar, optimal designs with no intermediate layers of abstraction, and no support in analyzing and exploring the diversity of the design space.
In contrast, we present a new approach that emphasizes the diversity of KPI outcome first, that combines both physics-based KPIs in addition to human-centered KPIs such as privacy, thermal comfort, and so on. The architect is presented with clusters of designs that have relatively similar KPI evaluations, and can then subsequently explore (i.e. fine-tune) their designs within each cluster via a semantic domain model of renovation, called NovaDM that we have previously developed. Thus, each cluster represents a qualitatively distinct combination of near-optimal KPI values, i.e. all lying close to the Pareto front, and exploring the design space within a cluster ensures that the design solution (or scenario) obtains the desired combination of values (e.g. medium construction cost, high privacy, etc.).
OriginalsprogEngelsk
TitelProceedings of the 18th International Conference on Computing in Civil and Building Engineering : ICCCBE 2020
RedaktørerEduardo Toledo Santos, Sergio Scheer
Antal sider20
UdgivelsesstedCham
ForlagSpringer
Udgivelsesår2021
Sider207-227
Kapitel2
ISBN (trykt)978-3-030-51294-1
ISBN (Elektronisk)978-3-030-51295-8
DOI
StatusUdgivet - 2021
Begivenhed18th International Conference on Computing in Civil and Building Engineering: ICCCBE 2020 - São Paulo, Brasilien
Varighed: 18 aug. 202020 aug. 2020
Konferencens nummer: 18

Konference

Konference18th International Conference on Computing in Civil and Building Engineering
Nummer18
LandBrasilien
BySão Paulo
Periode18/08/202020/08/2020
Serietitel Lecture Notes in Civil Engineering
Vol/bind98

Se relationer på Aarhus Universitet Citationsformater

ID: 192426148