Data mining of building performance simulations comprising occupant behaviour modelling. Issue 2 (3rd July 2019)

• Record Type: Journal Article
• Authors: Darakdjian, Quentin; Billé, Sylvain; Inard, Christian
• Is Part Of: Advances in building energy research. Volume 13:Issue 2(2019)
• Page Start: 157
• Page End: 173
• Publication Date: 2019-07-03
• Subjects: Occupant behaviour modelling -- multi-agent system -- principal component analysis -- energy modeling -- building simulation; Sustainable buildings -- Design and construction -- Periodicals; Architecture and energy conservation -- Periodicals; Buildings -- Energy conservation -- Periodicals; 696
• DOI: 10.1080/17512549.2017.1421099
• Languages: English
• ISSNs: 1751-2549

Abstract

ABSTRACT: Occupant behaviour is now widely recognized as a major factor in the disparity between predicted and measured building performance. Stochastic models are a convenient way to model the rational, diverse and complex nature of occupant behaviour, including presence and adaptive behaviour. The FMI standard was used to co-simulate the building energy modelling program EnergyPlus and a multi-agent platform that contains stochastic models in an integrated environment. Using an office building as a case study, we show that data mining, through a correlation matrix and a principal component analysis, was an efficient way of investigating the cumulated influence of occupant behaviour on energy performance. The organisation of simulations was achieved using design of experiments in order to take into consideration multiple building configurations. This paper demonstrates how data mining of stochastic simulations can be used to identify the determinants that have the greatest influence on building energy needs.