Tracking the human-building interaction: A longitudinal field study of occupant behavior in air-conditioned offices

https://doi.org/10.1016/j.jenvp.2015.01.007Get rights and content

Highlights

  • A one-year case study of office occupant thermal comfort and behavior is conducted.

  • Behaviors are examined over time against environmental and personal variables.

  • Certain behaviors show clear thermal ties; others likely relate to non-thermal factors.

  • Personal thermal acceptability ranges explain inter-individual variations in comfort.

  • Morning clothing and metabolic rates affect daily comfort and behavior trajectories.

Abstract

This paper presents findings from a one-year longitudinal case study of occupant thermal comfort and related behavioral adaptations in an air-conditioned office building. Long-term data were collected via online daily surveys and datalogger measurements of the local thermal environment and behavior. Behavioral outcomes are examined against both environmental and personal thermal comfort variables. Key personal variables include one's currently acceptable range of thermal sensations, which significantly explains inter-individual variations in thermal comfort responses. Results also show substantial between-day clothing adjustments and elevated metabolic rates upon office arrival, which may affect subsequent thermal comfort and behavior trajectories. Behavior sequencing appears complex, with multiple behaviors sometimes observed within a short time period and certain behaviors subject to contextual constraints. By elucidating the nature of the human-building interaction, the paper's findings may inform the improved measurement, modeling, and anticipation of occupant behavior as part of future sustainable building design and operation practices.

Introduction

Building occupants interact with their surrounding environments in deliberate and meaningful ways that contribute to both energy consumption and Indoor Environmental Quality (IEQ), and thus warrant significant attention in the building design and operation processes. For example, occupants' thermally adaptive behaviors (i.e., turning on fans/heaters, opening windows) are strongly tied to space heating and cooling loads, which make up 37% and 54% of total site energy consumed in commercial and residential buildings in the United States, respectively (U.S Department of Energy, 2011). These behaviors also modify key thermal comfort determinants like air temperature, air velocity and clothing insulation level (Baker & Standeven, 1997). Recent studies have begun to quantify the magnitude of occupant behavior's influence on energy use and comfort, reporting significant impacts that have intensified the focus on behavior as a key topic of built environment research (e.g. Bourgeois et al., 2006, Hong and Lin, 2013).

If the general importance of the human-building interaction is well established, however, the mechanisms behind this interaction are still being explored. Increasingly, this effort has involved the collection of longitudinal data, which allow one to observe occupant comfort and adaptive behavior as they evolve together across the day and season.

Nevertheless, longitudinal studies are time-consuming and expensive to carry out, and existing comfort and behavior data are accordingly limited in their coverage of certain adaptive actions, building types and climates. In particular, few existing studies examine thermal behaviors in air-conditioned buildings, or in buildings in climates with large seasonal variations. Moreover, existing studies do not generally examine action hierarchies across several possible thermal behaviors, and have not fully characterized the relationship between behavior and occupants' personal thermal preferences. Going forward, new longitudinal studies that address such shortcomings are needed to improve the understanding of interactions between building occupants and their interior environments.

This paper presents findings from a one-year longitudinal case study of occupant thermal comfort and several related behavioral adaptations in an air-conditioned office setting in Philadelphia, USA. Offices were chosen as the context for the research because of their significant contribution to energy use in the United States – representing the most prevalent type of floor space in the commercial sector, which is currently responsible for 19% of U.S. energy consumption (U.S Department of Energy, 2011). Moreover, multiple longitudinal studies of thermal comfort and behavior have been published for the office setting. The current case study builds a novel longitudinal protocol from the data collection and analysis approaches of these existing studies and from a theoretical framework in the psychology literature, which yields new types of survey response items and behavior data that contribute to a more comprehensive understanding of how and why office occupants interact with a common U.S. office context.

Like many recent studies of thermal comfort and behavior in offices, this paper approaches behavioral action through the general lens of Humphrey's adaptive principle, which states: “If a change in the thermal environment occurs, such as to produce discomfort, people react in ways which tend to restore their comfort” (Humphreys, 1997). However, while previous studies mostly focus on the environmental (external) determinants of discomfort and related behavior, the current study also seeks to explore personal (internal) determinants. Social psychologists have long suggested the need to include such internal variables in theories of behavioral action, particularly as part of research on pro-environmental behavior (Clarke et al., 2003, Guagnano et al., 1995, Wilkie, 1990). Within this context, internal variables are broadly defined to include one's motivation, environmental knowledge, locus of control, and attitudes, amongst many other concepts.

Kolmuss and Agyeman (2002) highlight motivation as the strong internal stimulus around which behavior is organized, and hypothesize that primary motives (i.e., altruistic, social values) are superseded by more immediate motives related to one's needs (i.e., being comfortable). This hypothesis aligns well with the adaptive principle above, and is considered a good starting point for exploring the key drivers of office occupant behavior. To further frame the current study, we adopt a theoretical formulation of comfort-driven behavior based on perceptual control theory (PCT) (Powers, 1973). Under PCT, behavior is the by-product of a negative feedback loop in which an organism attempts to control the current perception of its environment around some reference level. In the context of this paper, PCT suggests thermal comfort and adaptive behavior may be understood as part of the interplay between one's thermal sensation (current perception) and reference range of acceptable sensations (reference perception). The latter is thus focused on as a personal variable of potentially large significance to observed behavior.

Previous long-term field studies of thermal comfort and behavior in offices generally follow from the data collection and analysis approach of the European Smart Controls and Thermal Comfort (SCATs) project (see Humphreys et al., 2007, McCartney and Nicol, 2002). The SCATs project tracked thermal comfort, preference, and related behavioral adaptations (clothing, windows, doors, fans, and heating) from 1997 to 2000 in offices from twenty-five buildings located across Europe (nine air-conditioned; nine naturally ventilated; seven mixed-mode/other). The field monitoring combined longitudinal and cross-sectional field surveys with concurrent measurements of the local environment, establishing a set of environmental and personal variables that have been recorded in many subsequent field studies on comfort and behavior. Analysis of the SCATs data, together with similar data collected in the UK and Pakistan (see McCartney et al., 1998, Nicol et al., 1999) also first introduced the concept of simulating occupant behavior stochastically using generalized linear models; here, the probability of a given behavior occurring is modeled on a zero to one scale in terms of relevant predictor variables such as indoor or outdoor temperature.

In the decade following the SCATs project, a number of similar longitudinal studies have been reported in the literature. Rijal et al. (2007), for example, used over one year of longitudinal survey data on comfort and building control use in the UK to develop a simulation algorithm for window opening behavior. The algorithm calculates the probability of a window opening once a ± 2K deadband2 around comfort temperature has been breached, in terms of operative indoor and outdoor air temperatures. In the corresponding data analysis, the authors observed both seasonal and diurnal changes in the proportion of windows open, with the greatest observed proportions occurring in the afternoon in summer. The authors also suggest the consideration of “active” and “passive” window users, as previously suggested in Reinhart (2004) in the context of lighting.

Yun and Steemers (2008) conducted a field study in the summer of 2006 in UK private and shared private offices in naturally ventilated buildings. Indoor and outdoor temperatures were monitored along with window state and, for the first week of the study, occupancy (through observation). Data analysis showed significant correlations between window opening and indoor temperature, as well as time of day effects, where openings were far more frequent upon office arrival than during the day. Sub-models of window opening probability were developed for occupant arrival, intermediate, and departure periods, with indoor temperature and previous window state as predictor variables. In a subsequent paper (Yun, Tuohy, & Steemers, 2009), the authors incorporated “active,” “medium,” and “passive” window users into their modeling algorithm to represent inter-individual behavioral variation.

Herkel, Knapp, and Pfafferott (2008) monitored large and small window states alongside indoor/outdoor temperatures and occupancy in 21 naturally ventilated offices in Germany for 13 months. They observed strong seasonal changes in the percentage of open windows, with a consistently large percentage of windows open in the summer, sudden increases/decreases in the percentage of windows open in spring and fall, respectively, and a low percentage open in the winter. Outdoor temperature was more strongly correlated with window open percentage than indoor temperature in their study. Time of day was also found to be a significant factor, with most window openings and closings occurring upon arrival. The authors developed a series of quadratic functions to predict window opening probability for five segments of the day in terms of outdoor temperature.

Finally, Haldi and Robinson (2008) conducted a longitudinal study in eight Swiss office buildings across the summer of 2006. Several occupant adaptations were surveyed multiple times per day alongside indoor and outdoor temperature recordings. Logistic regression revealed that the probability of occupants interacting with personal and environmental characteristics is better described by internal than external temperature, with the exception of clothing adjustment, which is more strongly related to day-to-day changes in outdoor conditions. A later paper by the authors (Haldi & Robinson, 2009) examined longitudinal data on window opening in naturally ventilated cellular offices in terms of indoor/outdoor temperature as well as humidity, rainfall, wind speed, and occupancy, developing sub-models for window opening probability for arrival, intermediate, and departure times. The authors suggested that individual behavior differences could be modeled through the classification of “low,” “average,” and “high” activity occupants.

This paper contributes findings on the following research items that have been examined by at least some of the studies reviewed above:

  • The general frequency of thermally adaptive behaviors in a field office environment.

  • Temporal dynamics in thermal comfort and behavior, within a day and across all four seasons.

  • The relative effectiveness of indoor and outdoor temperature in describing the probability of a given behavioral state.

In addition, the paper seeks to examine items that are not substantively covered by previous studies:

  • Long-term comfort and behavior in an air-conditioned U.S. office building located in a dense urban context and subtropical climate with hot, humid summers and cold winters.

  • Direct consideration of personal thermal acceptability ranges as they relate to inter-individual variation in thermal comfort and behavior.

  • The sequencing/hierarchy of available behavioral actions.

  • Social restrictions on behavior and other reasons that occupants may not take available adaptive actions.

  • Analysis of high-resolution datalogger information on personal fan/heater use (frequent in HVAC buildings when available, see Fig. 1) as well as on windows.

Section snippets

Data collection

Longitudinal comfort and behavior data were collected between July 2012 and August 2013 at the Friends Center office building in Center City Philadelphia, USA (Fig. 2). Data collection proceeded in three stages, described below: 1) Semi-structured interviews; 2) Site selection and subject recruitment for the longitudinal study; and 3) Longitudinal survey and datalogger measurements.

Semi-structured interview results

The initial round of semi-structured interviews with 32 office occupants helped shape the development of the longitudinal survey instruments and guide the interpretation of longitudinal data that were collected. Findings of particular importance to the longitudinal study include: the tendency for occupants to prefer “Slightly Cool” thermal sensations in summer and “Neutral” sensations in winter (Fig. 7); the common perception that cold discomfort adaptations are more effective than warm

Summary and interpretation of findings

  • 1.

    Certain behaviors reveal clear thermal drivers and directions, while others are less strongly related to the thermal environment. In Table 10, Table 11, indoor and outdoor temperature show significantly positive associations with the probabilities of low clothing level, fan use, and window use, and significantly negative associations with the probabilities of high clothing level and heater use.

The strength and directions of these relationships make intuitive sense: behaviors that manage warm

Conclusion

This paper has presented findings from a one-year longitudinal case study of a medium-sized air-conditioned office building in a sub-tropical climate. The study paired multiple daily surveys of occupant comfort and behavior with datalogger measurements of behavior and the local thermal environment near each occupant. Results show that while certain behaviors have a clear relation to changes in the thermal environment, others may be better described by non-thermal factors. One's personally

References (49)

  • J. Langevin et al.

    Reducing energy consumption in low-income housing: Interviewing residents about energy behaviors

    Applied Energy

    (2013)
  • J. Langevin et al.

    Modeling thermal comfort holistically: Bayesian estimation of thermal sensation, acceptability, and preference distributions for office building occupants

    Building and Environment

    (2013)
  • K.J. McCartney et al.

    Developing an adaptive control algorithm for Europe: Results of the SCATs project

    Energy and Buildings

    (2002)
  • J. Nakano et al.

    Differences in perception of indoor environment between Japanese and non-Japanese workers

    Energy and Buildings

    (2002)
  • G.R. Newsham

    Clothing as a thermal comfort moderator and the effect on energy consumption

    Energy and Buildings

    (1997)
  • J.F. Nicol et al.

    Climatic variations in comfort temperatures: The Pakistan projects

    Energy and Buildings

    (1999)
  • K.C. Parsons

    The effects of gender, acclimation state, the opportunity to adjust clothing and physical disability on requirements for thermal comfort

    Energy and Buildings

    (2002)
  • C.F. Reinhart

    Lightswitch 2002: A model for manual control of electric lighting and blinds

    Solar Energy

    (2004)
  • H.B. Rijal et al.

    Using results from field surveys to predict the effect of open windows on thermal comfort and energy use in buildings

    Energy and Buildings

    (2007)
  • S. Schiavon et al.

    Dynamic predictive clothing insulation models based on outdoor air and indoor operative temperatures

    Building and Environment

    (2013)
  • G.Y. Yun et al.

    Time-dependent occupant behaviour models of window control in summer

    Energy and Buildings

    (2008)
  • G.Y. Yun et al.

    Thermal performance of a naturally ventilated building using a combined algorithm of probabilistic occupant behaviour and deterministic heat and mass balance models

    Energy and Buildings

    (2009)
  • N. Baker et al.

    A behavioural approach to thermal comfort assessment

    International Journal of Solar Energy

    (1997)
  • A.C. Boerstra et al.

    Personal control over temperature in winter in Dutch office buildings

    HVAC&R Research

    (2013)
  • Cited by (0)

    1

    Tel: +1 215 895 2341.

    View full text