An approach for building occupancy modelling considering the urban context

https://doi.org/10.1016/j.buildenv.2020.107126Get rights and content

Highlights

  • •Model of interactions between occupancy behaviour and urban systems is proposed.

  • •Competing risk hazard model is used to operationalise the model using Wi-Fi logs.

  • •The hazard model flexibly captures dependence of transition probability on duration.

  • •The model helps gain insights into exogenous effects on occupancy behaviour.

  • •Compared to MCM, hazard model better predicts occupancy with flexible work agendas.

Abstract

Building occupancy, which reflects occupant presence, movements and activities within the building space, is a key factor to consider in building energy modelling and simulation. Characterising complex occupant behaviours and their determinants poses challenges from the sensing, modelling, interpretation and prediction perspectives. Past studies typically applied time-dependent models to predict regular occupancy patterns for commercial buildings. However, this prevalent reliance on purely time-of-day effects is typically not sufficient to accurately characterise the complex occupancy patterns as they may vary with building's surrounding conditions, i.e. the urban environment. Therefore, this paper proposes a conceptual framework to incorporate the interactions between urban systems and building occupancy. Under the framework, we propose a novel modelling methodology relying on competing risk hazard formulation to analyse the occupancy of a case study building in London, UK. The occupancy profiles were inferred from the Wi-Fi connection logs extracted from the existing Wi-Fi infrastructure. When compared with the conventional discrete-time Markov Chain Model (MCM), the hazard-based modelling approach was able to better capture the duration dependent nature of the transition probabilities as well as incorporate and quantify the influence of the local environment on occupancy transitions. The work has demonstrated that this approach enables a convenient and flexible incorporation of urban dependencies leading to accurate occupancy predictions whilst providing the ability to interpret the impacts of urban systems on building occupancy.

Introduction

Building occupancy is a key parameter of building energy modelling and a basic factor of modelling occupant behaviour [1]. Bahaj and James [2] discovered that different occupancy patterns lead to different energy consumptions in identical buildings. Therefore, accurately modelling and comprehensively understanding how people use building can help improve energy efficiency [3]. Although the IEA Annex 66 on Definition and Simulation of Occupant Behaviour in Buildings has given guidance on the current state of occupancy modelling, there are still unanswered questions relating to the enhancement of modelling techniques, from the sensing, interpretation and prediction perspectives [4].

Driven by multidisciplinary factors, occupancy has three features that pose challenges to occupancy modelling: stochasticity, diversity and complexity [4]. Stochasticity has been the research focus over a past decade and has enriched the methodological approaches for modelling random occupancy chain. In fact, occupancy is not purely random but could also be characterised by deterministic factors in surrounding systems. These factors could be the internal conditions of building environment and/or the conditions external to the building system. Such conditions can be natural (weather and local topology) as well as anthropogenic, related to urban systems (conditions of transport network and presence of event). Incorporating both of those conditions as potential driving factors of building occupancy into a modelling framework could be a promising way of understanding and enhancing the way we model the complexity and diversity of occupancy patterns.

Considering spatial and temporal dependencies between building occupancy and conditions on surrounding urban systems, this paper proposes a modelling framework that comprehensively and flexibly accounts for such interactions. The framework is operationalised by the means of a competing risk hazard model using implicit occupancy data inferred from wireless (Wi-Fi) networks in the case study area. This approach has apparently not been applied in this discipline so far and has capabilities exceeding those of the more conventional approaches, e.g. MCMs. In particular, it can more easily incorporate multiple exogenous variables simultaneously and allow insights into exogenous impacts on occupancy and thus energy consumption. As an additional contribution, this paper also presents and applies a methodology of processing customarily available Wi-Fi data for occupancy modelling.

Section snippets

Literature review

The review is structured to correspond to the three contributions of this paper. Section 2.1 summarises studies related to interactions between human behaviour and factors originating in urban systems. Section 2.2 comparatively discusses the applications of MCMs and hazard-based models in occupancy modelling to date. Section 2.3 reviews different approaches to implicit occupancy detection, including discussion of the challenges associated with the use of Wi-Fi data for occupancy modelling in

Methodology

The following sections describe the urban-system level modelling framework, the HBM approach, the development of the conventional discrete-time MCM, the case study description, Wi-Fi data pre-processing and the performance metrics used to evaluate the two methods (HBM and MCM) from the estimation, interpretation and prediction perspectives.

Comparison of transition probability distributions

Comparison between the MCM and HBM with respect to the distribution of the transition probabilities requires plotting of the underlying PDFs. However, a PDF underlying a HBM parameterised using covariates will also depend on those covariates, themselves being particular to the circumstances they reflect. Hence for the purpose of demonstrating how the HBM captures more complex duration-dependencies in the transition distributions when compared to the MCM, we use the example of a baseline HBM

Conclusions

In this paper, we propose an urban-system level framework conceptualising the interactions between building occupancy and urban systems. We demonstrate how the framework can be operationalised using a competing risk hazard model, incorporating Wi-Fi connection logs, to analyse the occupancy of Imperial College Faculty Building in April–May 2017. A time-inhomogeneous discrete-time MCM was also implemented as the reference method. Comparing the performance of the two models, we identify three key

Declaration of competing interest

The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper.

Acknowledgements

This research was financially supported by the China Scholarship Council and Imperial College London. The authors would like to appreciate Mr. Matthew Balyuzi (Networks Project Team Leader, Imperial College London) for providing Wi-Fi connection data and discussing associated protocol issues. The authors acknowledge the support from the Integrated Development of Low-Carbon Energy Systems (IDLES) research programme at Imperial College London funded by the Engineering and Physical Sciences

References (65)

  • D. Wang et al.

    Modeling occupancy in single person offices

    Energy Build.

    (2005)
  • J.A. Oliveira-Lima et al.

    Load forecast on intelligent buildings based on temporary occupancy monitoring

    Energy Build.

    (2016)
  • A. Roy

    Correlates of mall visit frequency

    J. Retailing

    (1994)
  • J. Stephenson et al.

    Energy cultures: a framework for understanding energy behaviours

    Energy Pol.

    (2010)
  • S. D'Oca et al.

    The human dimensions of energy use in buildings: a review

    Renew. Sustain. Energy Rev.

    (2018)
  • F. Haldi et al.

    Interactions with window openings by office occupants

    Build. Environ.

    (2009)
  • A.G. Parsons

    The association between daily weather and daily shopping patterns, australas

    Market. J.

    (2001)
  • I. Eliasson et al.

    Climate and behaviour in a Nordic city

    Landsc. Urban Plann.

    (2007)
  • V. Fabi et al.

    Occupants' window opening behaviour: a literature review of factors influencing occupant behaviour and models

    Build. Environ.

    (2012)
  • S. D'Oca et al.

    A data-mining approach to discover patterns of window opening and closing behavior in offices

    Build. Environ.

    (2014)
  • V. Motuziene et al.

    Modelling the effect of the domestic occupancy profiles on predicted energy demand of the energy efficient house

    Procedia Eng

    (2013)
  • C. Zimring et al.

    Influences of building design and site design on physical activity: research and intervention opportunities

    Am. J. Prev. Med.

    (2005)
  • E. Delzendeh et al.

    The impact of occupants' behaviours on building energy analysis: a research review

    Renew. Sustain. Energy Rev.

    (2017)
  • G. Happle et al.

    A review on occupant behavior in urban building energy models

    Energy Build.

    (2018)
  • P.D. Andersen et al.

    Dynamic modeling of presence of occupants using inhomogeneous Markov chains

    Energy Build.

    (2014)
  • Y. Chen et al.

    An agent-based stochastic Occupancy Simulator

    Build. Simul.

    (2018)
  • S. D'Oca et al.

    Occupancy schedules learning process through a data mining framework

    Energy Build.

    (2015)
  • C.R. Bhat

    A hazard-based duration model of shopping activity with nonparametric baseline specification and nonparametric control for unobserved heterogeneity

    Transp. Res. Part B Methodol.

    (1996)
  • J. Pawlak et al.

    A framework for joint modelling of activity choice, duration, and productivity while travelling

    Transp. Res. Part B Methodol.

    (2017)
  • C. Jiang et al.

    Indoor occupancy estimation from carbon dioxide concentration

    Energy Build.

    (2016)
  • S.H. Ryu et al.

    Development of an occupancy prediction model using indoor environmental data based on machine learning techniques

    Build. Environ.

    (2016)
  • W. Wang et al.

    Modeling occupancy distribution in large spaces with multi-feature classification algorithm

    Build. Environ.

    (2018)
  • Cited by (0)

    View full text