You are on page 1of 14

Energy Efficiency (2016) 9:1257–1270

DOI 10.1007/s12053-015-9422-7

ORIGINAL ARTICLE

Should we design buildings that are less sensitive to occupant


behaviour? A simulation study of effects of behaviour
and design on office energy consumption
Sami Karjalainen

Received: 24 March 2015 / Accepted: 29 December 2015 / Published online: 14 January 2016
# Springer Science+Business Media Dordrecht 2016

Abstract In an ideal world, occupants have an under- Keywords Behaviour change . Building design . Robust
standing of how building systems work and are moti- design . Energy conservation . Energy efficiency .
vated to use the systems as they were designed to be. Control system . Usability . Personal control
There is considerable evidence, however, that occupants
do not understand the principles of how buildings func-
tion and that they use the systems non-optimally. The
Introduction
purpose of the paper is to quantify the effect of occupant
behaviour on energy consumption and show how it is
Buildings are not used as the designers intended
affected by design strategies. Numerical simulations of
an office were performed with the dynamic thermal
Ideally, occupants understand how their building’s sys-
simulation software TRNSYS. Three types of behaviour
tems work and are motivated to use those systems as
(‘careless’, ‘normal’, and ‘conscious’) and two types of
was intended, thereby ensuring high energy-efficiency
design (‘ordinary’ and ‘robust’) were considered. The
and providing a safe, healthy, and productive work
results show that the effect of occupant behaviour on
environment. For several reasons, however, the ideal is
energy consumption is greatly diminished with robust
not achieved in practice and buildings under-perform.
design solutions, solutions that make buildings less
Buildings are complicated systems made up of diverse
sensitive to occupant behaviour. The careless user con-
independent systems. There is a great deal of evidence that
sumes 75–79 % less energy if the robust design solu-
occupants do not understand the principles of how build-
tions are applied rather than the ordinary design solu-
ings operate and that they use the systems non-optimally
tions. It is argued that a realistic view of occupant
or even inappropriately from the energy point of view.
behaviour is advantageous in the creation of energy-
Numerous examples are documented in the literature.
efficient buildings (that is, leaving less need to learn
Kempton (1987) analysed the common-sense beliefs
how buildings work, to be motivated to save energy,
about home heating control and found two theories as to
or to perform specific energy-saving actions). However,
how a thermostat works: a feedback-based theory and a
the possibility of personal control should not be
valve theory. In the feedback theory, a thermostat senses
eliminated.
room temperature. In the valve theory, a thermostat dial
is like a car’s accelerator pedal and controls the amount
of heat. The latter misconception affects the use of
thermostats. In a study of use of room air-conditioning
S. Karjalainen (*)
VTT Technical Research Centre of Finland Ltd, P.O. Box 1000, units, Kempton et al. (1992) found that how people
FI-02044 Espoo, VTT, Finland operated them was governed by multiple, overlapping
e-mail: sami.karjalainen@vtt.fi systems of belief and preferences related to health,
1258 Energy Efficiency (2016) 9:1257–1270

thermal comfort, common-sense beliefs about how Lighting is often on for no good reason, with 60 % of
air-conditioners function, etc., in addition to eco- European offices using lights even in the hottest weather
nomic factors. (Nicol 2001) (when more natural light is typically avail-
Furthermore, Drake et al. (1986) found that occu- able). Occupants of offices and schools commonly have
pants are active in creating theories on how the systems either all or none of the lighting on at any given time
work and use the systems in accordance with their (Hunt 1979). Occupants in private offices with occu-
beliefs even when this may involve circumvention. pancy sensors are less likely to turn off the lights when
Examples given of user actions include blocking sen- leaving the room; i.e., people modify their behaviour in
sors, taping over air diffusers, and bringing in fans and the presence of occupancy sensors (Pigg et al. 1996).
heaters. These actions often interfere with the intended Masoso and Grobler (2010) studied energy use in
performance and increase energy consumption. commercial buildings in Botswana and South Africa
Widespread misconceptions exist with regard to and found that more energy is used during non-
comfortable room temperatures. An interview study working hours (56 %) than the work day (44 %). The
(Karjalainen and Vastamäki 2007) reported that 41 % main reason behind this pattern is that occupants leave
of interviewees believed that comfortable room temper- lights and other equipment on at the end of the day. It is
atures are lower in summer than in winter, which may well known, for example, that computers are commonly
lead to increased energy use. left on when workers leave the office. A field study
Karjalainen and Koistinen (2007) studied the use of performed in the USA found that only 44 % of com-
thermostats in offices and found that usability problems puters and 32 % of monitors were switched off at night
were fundamental and varied. Occupants did not always (Webber et al. 2001). A field study in the UK found that
know even that they were able to control the temperature 60 % of occupants never switch off their computers
room-specifically, because the device was not when leaving the office (Zhang et al. 2011). Murtagh
recognised at all or its purpose remained unclear. et al. (2013) went further, arranging focus-group discus-
Although the room thermostats in offices were simple sions in which they collected reasons for not switching
to operate, the symbols seen in user interfaces were things off. These include the following: ‘I failed to
often not understood correctly. An important reason suspend/hibernate it’, ‘The computer will not boot up
for many of the usability problems is that the systems correctly in the morning’, lack of incentive to shut down
are produced without a realistic view of their users. In the computer, the car industry consuming more energy
other words, occupants are expected to have knowledge than computers do, and that turning the unit off will
that they do not actually possess. shorten the service life of the computer. Clearly, there
The occupants themselves do recognise that their are multitudinous reasons for not shutting down a
knowledge is limited. When Karjalainen (2007) asked computer.
occupants to rate their knowledge of the heating and Mulville et al. (2014) see potential for reducing desk-
ventilation systems in Finland, reported knowledge of level energy consumption by up to 20 % through be-
the systems in office buildings was low, with a mean haviour changes in UK offices. Also, Junnila (2007)
value of 2.03 (very bad = 1, …, very good = 5). calculated the energy savings potential in four banking
Passiveness too is a factor. Foster and Oreszczyn organisations in the Nordic region, finding that 20 %
(2001) studied the use of Venetian blinds in offices in savings on total electricity use can be gained with no or
London and found that the majority of blinds tend to be very low investment costs. The calculated savings were
kept down most of the time. A review of literature on attributed to the reduced energy consumption of office
occupant interaction with window blinds (Van Den equipment and lighting.
Wymelenberg 2012) reveals that manual blinds have a Leaman (1999) has made observations on occupant
very low rate of daily change: 15 % are adjusted every behaviours in his studies of buildings in use. The obser-
day or so, one third are not adjusted more often than vations show that occupants often choose the easiest and
every 4 months, and 30 % are reported by users as never quickest option, rather than the best one, for their im-
being adjusted. Neither are window-shading devices mediate benefit. They spring into action in response to
(including blinds and roller shades) operated actively: arbitrary, external events. They are likely to make the
most office occupants do not adjust the shades more decision to use a switch or control only after being
often than weekly or monthly (O’Brien et al. 2013). prompted by an event, rather than in advance of it.
Energy Efficiency (2016) 9:1257–1270 1259

They often wait a fairly long time before taking action: study performed in Finland with 1000 respondents
until they have reached a ‘crisis of discomfort’. They showed that people make significantly less effort to save
may overcompensate in their reactions to relatively mi- energy in offices than in homes (Karjalainen 2007).
nor annoyances. They tend to operate the control that is Occupants have less motivation to save energy in offices
closest to hand or otherwise more convenient, rather because they do not pay for the energy themselves and
than the one that would be logically the most appropri- because they consider their energy use to be negligible
ate. In addition, having made the decision to use a (Karjalainen 2007).
switch, occupants often leave systems in the new state Secondly, energy costs in offices are low when com-
rather than changing their state back again later, at least pared to the personnel costs of office work, e.g.,
until another crisis of discomfort is reached. Wargocki and Seppänen (2006). This means that from
O’Brien and Gunay (2014) conclude their review on the economic perspective, it is of fundamental impor-
adaptive comfort behaviours by arguing that ‘occupants tance to provide productive work environments and less
are not illogical and irrational but rather they attempt to important to save on energy costs.
restore their comfort in the easiest way possible’. Factors other than those of economy may be more
Occupant behaviour is not easily modelled, since it is effective in encouraging energy conservation in offices.
affected by many contextual factors. O’Brien and Handgraaf et al. (2013) gave office occupants feedback
Gunay organised the contextual factors affecting occu- on their energy consumption and compared the effect of
pant behaviour into nine categories: (1) availability of monetary and social rewards on energy conservation.
personal control, (2) accessibility of personal control, The results show that public social rewards were more
(3) complexity and transparency of automation systems, effective than monetary rewards for stimulating energy
(4) presence of mechanical/electrical systems providing conservation.
alternative means of comfort, (5) view and connection to Leygue (2014) classified the types of motivation to
the outdoors, (6) interior design, (7) experiences and save energy at work as self-directed (involving self-
foreseeable future conditions, (8) visibility of energy image and a ‘warm glow’) and altruistic (oriented to-
use, and (9) occupancy patterns and social constraints. ward the planet and toward the organisation, including
the company’s image and finances). She performed a
Can we modify occupant behaviour? survey in UK workplaces and found that people are
motivated mostly by environmental considerations.
The fact that occupant behaviour has a large effect on Delmas and Lessem (2014) compared private feedback
buildings’ energy consumption is well accepted; see, for and publicly posted information on conservation behav-
example, the review by Lopes et al. (2012). We also know iour in experiments at halls of residence at the University
that it is possible to reduce energy consumption by modi- of California. They found no effect of private information
fying occupant behaviour. Several studies have shown that provision on conservation behaviour. Public information
household energy consumption can be reduced by means (posters with red or green stickers as indicators of the level
of feedback on one’s energy consumption, see reviews by of energy use) in combination with private feedback mo-
Darby (2001); Abrahamse et al. 2005; Fischer (2008); tivated 20 % savings in electricity consumption. The
Ehrhardt-Martinez et al. (2010) and Faruqui et al. (2010). authors concluded that public information can encourage
It should be noted here that research into energy-related conservation, partially by ‘allowing people to communi-
behaviours has focused largely on the residential sector cate their greenness to the world’.
(Lopes et al. 2012); little research has dealt with energy Normative social influence may affect energy use in
feedback in office environments, and we cannot generalise offices, although the studies described next were carried
the results for households to offices, since energy-related out in other contexts. Nolan et al. (2008) studied energy
behaviours are hugely complex and shaped by contextual conservation in California households and found that
factors (Lopes et al. 2012; Steg and Vlek 2009; Coleman et normative social influence affects energy behaviour;
al. 2013; Littleford and Ryley 2014). that is, ‘witnessing the actions of other people has a
It is clear that the economic environment of offices is powerful effect on behaviour’. While the participants
very different from that of households. Occupants are believed that their neighbours’ behaviour had a very
not responsible for the energy costs in office spaces as limited influence on their own behaviour, this actually
they typically are in households. An interview-based was the factor with the strongest effect on the
1260 Energy Efficiency (2016) 9:1257–1270

participants’ energy-conservation behaviour. People focused on work, accomplishing the tasks for which
seem to hold incorrect beliefs as to what motivates them: they were hired, instead of energy-saving activities.
the behaviour of others motivated them to conserve
energy, although they did not recognise this themselves. Should we design buildings that are less sensitive
Siero et al. (1996) examined energy-related behaviour in to occupant behaviour?
a metallurgy company and found that employees who
received comparative feedback saved more energy than One strategy for conservation of energy in offices is to
those who received only information about their own target modification of occupant behaviour in such a way
performance. Hardly any changes in attitudes or inten- that occupants reduce their energy use. There are chal-
tions were recorded. lenges in modifying occupant behaviour, especially in
Greater concern for the environment is associated with offices (see Subsection BCan we modify occupant
pro-environment behaviour, although this correlation is behaviour?^); i.e., we cannot assume that it is easy to
generally not strong (Steg and Vlek 2009). Spence et al. modify behaviours, let alone achieve a significant and
(2014) see highlighting the relationship between climate permanent decrease in energy use.
change and energy savings as of possible benefit for Another strategy is to design solutions that make
promotion of pro-environment behaviour. buildings less sensitive to occupant behaviour—that is,
Recent studies involving energy conservation in of- make design choices that diminish the effect of occupant
fices include the following. In their field study of indi- behaviour on energy consumption.
viduals’ energy use and feedback in UK offices, Murtagh O’Brien (2013) has argued for occupant-proof build-
et al. (2013) found that people’s reports of their pro- ings. He uses the term ‘robust’ (earlier used in this
environment behaviour was not correlated with actual context by Palme et al. 2006 and Hoes et al. 2009) for
energy-linked behaviour. Although a reduction in energy building-design strategies that are less sensitive to occu-
consumption was achieved, the authors found consider- pant behaviour. A few examples of robust design are
able lack of motivation to conserve energy among the occupant detection for the control of lighting, an over-
office occupants. Though occupants consumed much hang above windows to limit the solar energy entering
more energy at their desks than they really needed, they the room through the windows, and thermochromic win-
were not very prepared to change their energy-use be- dows to adapt to changing levels of sunshine and limit
haviour, and 41 % of the office occupants did not access the solar energy entering the room (O’Brien 2013). Also,
their worker-specific energy feedback even once. solutions that consume less energy, such as employing
Carrico and Riemer (2011) carried out two interven- LED lighting instead of fluorescent tubes and opting for a
tions in an office environment to study their effect on laptop computer instead of a desktop PC, can be deemed
energy consumption. They distributed group-level feed- robust, as the effect of occupant behaviour on the energy
back via e-mail and used educators to disseminate in- consumption is lower than with the alternatives.
formation and encourage colleagues to reduce their Table 1 gives on overview of these two strategies and
energy use. The feedback and peer education resulted shows their significant differences from the occupant
in a 7 and 4 % reduction in energy use, respectively. point of view.
Orland et al. (2014) educated users and prompted Now that background information has been laid out,
behaviour change through serious game-style interven- the paper can proceed toward its goal of quantifying the
tions in an office environment. They created an ‘energy effect of occupant behaviour on energy consumption
chicken’ game in which chickens gain or decline in and showing how this is affected by design strategies.
health in line with the energy consumption of a specific
device. The reported results are that average plug-load
energy consumption was reduced by 13 %: 7 % on work Method
days and 23 % on non-work days.
Coleman et al. (2013) found barriers to utilisation of Basic cases
energy-feedback technologies in offices. Occupants of-
ten feel that the only energy consumption under their Numerical simulations were performed to study the
direct control is that of their workstation equipment effect of behaviour and that of design on office energy
(e.g., computer, monitor, desk lamp). Occupants are consumption. The simulations were performed with the
Energy Efficiency (2016) 9:1257–1270 1261

Table 1 Overview of the two strategies for energy savings from Table 2 The simulated office
the occupant perspective
Area 10.4 m2
Saving energy Saving energy Volume 31.2 m3
by modifying by designing
Window 1.8 m2, U = 0.68 W/m2K, g = 0.41,
occupant solutions that
orientation: north-facing
behaviour are less sensitive
to occupant External wall 6.3 m2 (without window), concrete,
behaviour U = 0.19 W/m2K, orientation:
north-facing
Occupants have to be Yes No Other walls No heat transfer between this and
motivated to save other offices—i.e., adjacent
energy rooms are assumed to have
Occupants have to Yes No identical thermal conditions
perform energy-saving Ventilation 1.5 (dm3/s)/m2 (1.8 1/h), heat
actions in order to recovery efficiency 70 %,
save energy supply air temperature 20 °C
Occupants need to More Less The air-handling unit includes a
possess knowledge heat-recovery unit and a heating
of how the building coil but no cooling or
works humidification. The air-flow
rate is chosen in line with
regulations ( D2 National
Building Code of Finland 2003).
dynamic thermal simulation software TRNSYS 17.1 The air-flow rate during the
(Klein et al. 2012) and its building model Type 56. night and on weekends is 25 %
A single-person office modelled on the basis of a of the daytime air-flow rate.
room in Finland’s METOP low-energy building (Laine Infiltration 0.1 1/h
and Saari 1994) was simulated. Weather data from Location Helsinki, Finland
Helsinki were utilised. Detailed information on the of- Room air temperature Heating set point: winter season
fice is presented in Table 2. set point 21 °C, summer season
(June–August) 23 °C
The simulation model calculates heating and cooling Cooling set point: winter season
demand—the energy needed to keep the room air tem- 24 °C, summer season
perature between the set points. The room temperature (June–August) 26 °C
set points utilised in the simulations are shown in Table No heating or cooling energy is
used when the room air
2. Adjacent rooms are assumed to have identical thermal temperature is between
conditions—i.e., no heat is transferred between rooms. the heating and cooling set points.
The amount of energy needed by the air-handling unit to Occupancy One person is working in the
heat the air flow is calculated and added to the heating- office between 8 a.m. and
demand value. The electricity consumption considered 4 p.m. on Monday to Friday,
except for 1 h at lunchtime,
includes the energy use of lighting and the user’s com- between 12 p.m. and 1 p.m.
puter. Simulations spanning a year were performed with
1-h time step.
Six cases were simulated to study the effects of below). These behaviours were chosen for simulation
behaviour and design. Three types of behaviour were because of their large effect on energy use.
considered: ‘careless’, ‘normal’, and ‘conscious’ behav- In basic cases, it is supposed that the window is not
iour. The careless user does not care about consumption operable because this is a common situation in me-
and leaves the lights and computer on when leaving the chanically ventilated office buildings; see e.g., Rijal et
office, while normal and conscious users turn them off. al. (2007). If the windows are operable in mechanical-
The conscious user turns the lighting and computer off ly ventilated offices, their use is often discouraged
also for the lunch hour. Table 3 provides a summary of (Rijal et al. 2007; Wang and Greenberg 2015).
the behaviours simulated. No other behavioural ele- Additional simulation cases were performed in which
ments than lighting use and computer use are considered the user opens the window (BThe effect of window-
(except in the additional simulation cases, addressed opening behaviour^).
1262 Energy Efficiency (2016) 9:1257–1270

Table 3 The behaviours simulated

Use of lighting Computer use

Careless Leaves the lighting on when leaving the office. Lighting is on continuously. Leaves the computer on when leaving the
office. The computer is on continuously.
Normal Turns lights on when the lighting level is beyond a certain threshold (total Has the computer on during the working day.
horizontal radiation outdoors < 120 W/m2). Turns the lighting off when
leaving the office after work or when the lighting level is above a certain
threshold (total horizontal radiation outdoors > 200 W/m2).
Conscious As above but lighting is turned off for the 1-h lunch period every day. As above but the computer is turned off for the
1-h lunch period every day.

Two types of design were considered: ‘ordinary’ and similar to those in the basic cases described in Section
‘robust’ design. The latter refers to the following design BBasic cases^.
solutions: occupant detection for the control of lighting,
an overhang above the window to limit the solar energy
entering the room through the window, LED lighting, The effect of window-opening behaviour
and a laptop as the computer. The focus of the study is
not on the building structures and other elements that are Additional simulation cases were examined to assess the
difficult to change without a great cost. Table 4 presents effect of window-opening behaviour on energy con-
a summary of the simulated designs. sumption. Occupants’ window-opening behaviour is
influenced by many interconnected factors, and no con-
sensus has been reached about how this behaviour in
Additional cases offices should be modelled (Rijal et al. 2007; Herkel et
al. 2008; Fabi et al. 2012; Zhang and Barrett 2012). The
Additional simulation cases were performed to study the driving forces are multidisciplinary and were divided by
effects of climate, window-opening behaviour and room Fabi et al. (2012) into five categories: physical environ-
temperature set points. mental, contextual, psychological, physiological and
social. Most field studies have been performed in natu-
rally ventilated buildings in which the proportion of
The effect of climate open windows is increased in line with higher outdoor
and indoor temperatures (Rijal et al. 2007; Fabi et al.
Simulations were performed in three climates to study 2012; Zhang and Barrett 2012). In these simulations, it
the sensitivity of the results to the weather. In addition to is supposed that the window is opened at 1 p.m. after
the northern European location (Helsinki, Finland), ad- returning to the office after lunch if the outdoor temper-
ditional cases were simulated with weather data from ature is 10 °C or higher. The air change rate through an
central Europe (Frankfurt, Germany) and southern opened window is dependent on a number of factors,
Europe (Napoli, Italy). All the other parameters were including window type, window width, window-

Table 4 The designs simulated

Occupant detection Overhang above the window Power use for lighting Power use of a computer

Ordinary No No 156 W (~15 W/m2) 230 Wa (PC and monitor)


2
Robust Yes Yes 42 W (~4 W/m ) (LED lighting) 60 W (laptop and external
Lighting controlled by occupant Projection: 1 m, installed monitor)
detection—i.e., automatically 0.2 m above the window
turned off if the room is
unoccupied
a
Default values in the TRNSYS simulation program were used. The power use values are highly variable between different PC systems, as
shown in ASHRAE (2009). The default values in TRNSYS are among the highest
Energy Efficiency (2016) 9:1257–1270 1263

opening angle, temperature difference and wind speed. Table 6 Room air temperature set points in additional simulation
cases studying the effects of differences between the cooling and
The average air change rate due to the occupants has
heating set points
been measured to be from 0.25 to 0.8 in the studies
reviewed by Fabi et al. (2012). In these simulations, it is Winter season Summer season
supposed that the air change rate through the opening is (June–August)
0.5 1/h. Heating Cooling Heating Cooling
set point set point set point set point
(°C) (°C) (°C) (°C)
The effect of room air temperature set points
Difference of 3 °C 21 24 23 26
Because there are individual differences in the room Difference of 1 °C 22 23 24 25
temperature preferences and condition tolerance, addi- Difference of 5 °C 20 25 22 27
tional simulation cases were examined to assess the
effect of the room temperature set points on energy
consumption. The set points used were dependent on The electricity consumption totals 3.4 MWh with the
season, as shown in Table 2, and the difference between careless-behaviour case and ordinary design while it is
the cooling set point and heating set point was 3 °C in between 0.1 and 0.9 MWh in the other cases.
both seasons. Additional simulations were performed Energy consumption for heating (including energy
with lower and higher set points (Table 5) and with a used by the heating coil for heating the inlet air) is
difference of 1 and 5 °C between the cooling and between 0.2 and 0.7 MWh in all the cases considered.
heating set point (Table 6). No heating or cooling energy The need for heating energy is lowest with careless
is used when the room air temperature is between the behaviour, because the electrical energy used is convert-
heating and cooling set points. ed into thermal energy in the room.
The energy consumed for cooling comes to 2.6 MWh
in the case of careless behaviour and ordinary design,
Results while it is 0.2 MWh or less in the other five cases. In two
of the cases, no cooling is needed (that is, there is no
Basic cases need for a cooling system).
The room temperatures experienced in 1 day (13
The total energy consumption, including cooling, March) are presented in Fig. 2 for all the cases. With
heating and electricity, is presented in Fig. 1 for all six careless behaviour and ordinary design temperature, the
cases. One of the cases showed much higher energy room air temperature is 24 °C throughout the day. This
consumption than the others. With careless behaviour is a result of heat gains from lighting and computer use.
and ordinary design, the energy consumption for the The cooling system keeps the temperature from rising
single-person office over the span of 1 year is above 24 °C. No cooling energy is used in the other five
6.1 MWh. For the other cases, it is between 0.8 and cases, since the temperature does not increase to 24 °C.
1.3 MWh. The room temperatures are constantly between 21 and
22 °C in all cases with robust design. In most of the
Table 5 Room air temperature set points in additional simulation cases, heating energy is needed during certain hours to
cases studying the effects of low and high set points keep the room temperature above 21 °C.
Winter season Summer season Figure 3 highlights the effect of behaviour on total
(June–August) energy consumption with the two design options simu-
lated. The effect of behaviour on consumption is very
Heating set Cooling set Heating set Cooling set large with the ordinary design but is considerably di-
point (°C) point (°C) point (°C) point (°C)
minished when the robust solutions are applied. A care-
Normal (N) 21 24 23 26 less user consumes 78 % less energy with the robust
Low (L) 19 22 21 24 design than with the ordinary design.
High (H) 23 26 25 28 The total energy costs are low relative to the person-
nel costs of the office work. If an average energy price of
1264 Energy Efficiency (2016) 9:1257–1270

Fig. 1 Yearly energy


consumption of the office,
depending on behaviour (careless,
normal, or conscious) and design
(ordinary or robust)

€130/MWh is assumed, the total cost of energy for a savings in Italy and Germany than in Finland (Fig. 4,
single-person office is €109–796 per year, depending on Table 7).
the behaviour style and design.
The effect of window-opening behaviour

Additional cases The simulations showed that window-opening behaviour


has a minor effect on yearly energy consumption. The
The effect of climate small increase in heating demand and the small decrease
in cooling demand are not visible in Fig. 5. The reasons
Weather data from three European locations were used behind the minor effect are as follows: the relatively
to study the effect of climate. Figure 4 shows that the small difference between the outdoor and indoor temper-
robust design solutions led to lower total energy con- atures during the time when the window was open; and
sumption compared with the ordinary design in all three the small proportion of time when the window was open.
climates. With normal and conscious behaviour, the The window was opened for a total of 311 h over the 1-
robust design solutions provide larger proportional year simulation period (3.6 % of the time).

Fig. 2 Room air temperatures


through a full day of the
simulation (13 March)
Energy Efficiency (2016) 9:1257–1270 1265

the ordinary ones with both low and high room temper-
ature set points.
The difference between the cooling and heating set
point has an effect on energy consumption. The total
energy consumption is higher when there is a small
difference (1 °C) between the cooling and heating set
point and lower when the difference is large (5 °C). This
is shown in both ordinary and robust design cases. The
total energy consumption is lower in the robust case
with a set point difference of 1 °C than in the ordinary
case with a set point difference of 5 °C (Fig. 7).

Summary

Table 7 provides an overview of the total energy savings


with the robust design compared with the ordinary de-
sign in all the simulation cases. In all cases, the careless
Fig. 3 Maximum and minimum energy consumption, depending user consumes 75–79 % less energy if the robust design
on the design. The ‘max.’ values refer to careless behaviour and solutions, rather than the ordinary design solutions, are
‘min.’ values to conscious behaviour applied. Savings of between 9 and 60 % were achieved
with the robust design in the simulation cases displaying
The effect of room air temperature set points normal and conscious occupant behaviour.

The room temperature set points affect cooling- and


heating-energy consumption. Low set points generally Discussion
create more demand for cooling, while high set points
create more demand for heating. The effect on total Designing with people in mind
energy consumption depends on the case; i.e., an in-
crease in the set points may increase or decrease the total Buildings’ energy systems are often designed such that
energy consumption; see Fig. 6. The robust design their use is demanding from the user perspective.
solutions lead to lower total energy consumption than Occupants must have an understanding of how the

Fig. 4 Yearly energy


consumption in different climates.
A northern European location
(Helsinki, Finland) is compared to
central (Frankfurt, Germany) and
southern (Napoli, Italy) European
locations
1266 Energy Efficiency (2016) 9:1257–1270

Table 7 Total energy savings with the robust design compared separately. In the basic case the building is located in Finland,
with the ordinary design in different cases. All three types of the window is non-operable and the room air temperature set
behaviour (careless, normal and conscious) are considered points are as given in Table 2

Careless Normal Conscious

Basic case (Fig. 1) 78 % 26 % 23 %


From 6.12 to 1.32 MWh From 1.14 to 0.84 MWh From 1.08 to 0.84 MWh
German climate (Fig. 4) 79 % 45 % 41 %
From 6.3 to 1.3 MWh From 1.1 to 0.6 MWh From 1.0 to 0.6 MWh
Italian climate (Fig. 4) 76 % 60 % 57 %
From 6.59 to 1.55 MWh From 1.05 to 0.42 MWh From 0.95 to 0.41 MWh
Operable window (Fig. 5) 79 % 25 % 21 %
From 6.11 to 1.30 MWh From 1.12 to 0.85 MWh From 1.07 to 0.85 MWh
Low room air temperature 75 % 42 % 38 %
set points (Table 5; Fig. 6) From 6.12 to 1.50 MWh From 1.29 to 0.75 MWh From 1.19 to 0.74 MWh
High room air temperature 79 % 10 % 9%
set points (Table 5, Fig. 6) From 5.80 to 1.24 MWh From 1.09 to 0.98 MWh From 1.07 to 0.98 MWh
Set point difference of 1 °C 77 % 42 % 39 %
between the cooling and From 6.29 to 1.42 MWh From 1.58 to 0.91 MWh From 1.49 to 0.92 MWh
heating set points
(Table 6, Fig. 7)
Set point difference of 5 °C 79 % 22 % 19 %
between the cooling and From 5.96 to 1.24 MWh From 1.01 to 0.78 MWh From 0.96 to 0.78 MWh
heating set points
(Table 6, Fig. 7)

systems work in order to use them properly, and they solve the problems. What we can do is develop systems
have to be motivated to save energy and active in and devices that have greater usability and are easier to use.
performing energy-saving actions. Designers often make optimistic assumptions (e.g.,
We should not blame office occupants for their behav- Foster and Oreszczyn 2001; Lutzenhiser 1992; Karlsson
iour, nor is it realistic to suppose that users would devote et al. 2007); however, a realistic view of occupants’
much of their valuable time to learning the way in which actions would be advantageous for creation of energy-
the building works. It is best instead to find other ways to efficient buildings.

Fig. 5 Yearly energy


consumption with non-operable
and operable windows. In the lat-
ter case, the window is opened for
3 h in the afternoon if the outdoor
temperature is 10 °C or higher
Energy Efficiency (2016) 9:1257–1270 1267

Fig. 6 Yearly energy


consumption plotted against
different room air temperature set
points. N refers to normal set
points, L to low set points and H
to high set points (see Table 5)

The following strategies should be considered It is advantageous to accept the fact that office occu-
when one is designing energy systems with people pants do not behave optimally. This does not mean that
in mind: the aim of modifying occupant behaviour is not impor-
tant. A combination of strategies is necessary for reduc-
& We cannot assume that people have the knowledge ing energy use, and, of course, it is important to find
required for the most appropriate use of the systems. ways to change occupant behaviour such that people use
While they may have the knowledge, we should not less energy.
base our designs on that supposition.
& We cannot assume that people are motivated to save Personal control and energy use
energy in the workplace. Some are, but we should
not make this assumption in our designs. Should we eliminate the possibility of personal control
& We cannot assume that people actively perform when we design solutions that are less sensitive to
energy-saving actions. While some of them display occupant behaviour?
this behaviour, we should not base our designs on It is easy to assume that energy consumption is
the assumption that they do. greater if people have possibilities for personal control,

Fig. 7 The yearly energy


consumption of the office plotted
against the difference between the
cooling and the heating set point.
Simulations with the set point
difference of 1, 3 and 5 °C were
calculated
1268 Energy Efficiency (2016) 9:1257–1270

including the opportunity to adjust the room temperature Limitations of the simulations and suggestions for future
with a thermostat. However, when experts (of HVAC work
systems and thermal comfort) were presented with the
statement ‘Allowing occupant control leads to higher The purpose of the simulations was to quantify the
energy use’ at a workshop and asked whether they effects of occupant behaviour and design strategies on
agreed or disagreed (Pfafferott and Boerstra 2014), they energy consumption. The simulations were performed
had divergent views. It was evident that many factors in certain sets of conditions only, and no results are
affect the influence of personal control on energy con- available for other types of office building designs.
sumption, and no general conclusion can be stated. The occupant-behaviour models were greatly simpli-
The simulations performed in the study described fied. The office simulated was mechanically ventilated
here show—not surprisingly—that energy consumption with a mechanical cooling system. Adjacent rooms were
is higher when the difference between the cooling and assumed to have identical thermal conditions, i.e., no
the heating set point is small and lower when the set heat transfer between rooms was considered.
point difference is large (as is shown in Fig. 7). This Simultaneous heating and cooling may occur in real-
means that energy consumption is lower if people can world buildings, but the energy-demand calculations
tolerate a wider range of room temperatures. Those performed assume that no heating or cooling energy is
occupants who show dissatisfaction with a small devia- used when the temperature is between the cooling and
tion from the ‘optimal’ temperature and adjust the set heating set point.
points accordingly use more energy than their counter- Future work should study a wider range of robust
parts who are tolerant of wider range of temperatures. In technologies in different climates and other conditions.
short, less tolerant occupants use more energy that the It would be valuable to know which robust solutions to
more tolerant ones. choose for specific types of building (either new or
Several studies have showed that people are more renovated). The current research does not evaluate dif-
condition-tolerant if they have control over their ther- ferences in initial costs of robust and ordinary building
mal environment (Humphreys and Nicol 1998; techniques. Future studies could include economic cal-
Leaman and Bordass 2000; Brager et al. 2004; Luo culations for robust solutions compared with ordinary
et al. 2014): occupants accept a broader range of room solutions. These calculations should take into account
temperatures if they have effective control of the tem- both the costs of running the building under the various
perature. There is a psychological effect that is impor- solutions and the capital costs of providing those
tant to recognise here: thermal comfort is improved if solutions.
occupants perceive themselves as having real control
over the temperature. They often do not use the con-
trols frequently, but they are more satisfied with the Conclusions
thermal environment when they know that they could
adjust the temperature if ‘a crisis of discomfort’ is Occupant behaviour has a large effect on energy con-
reached. If the occupants have no opportunities for sumption, but energy systems are often designed with-
personal control, a small deviation from the tempera- out a realistic view of occupants and their behaviour.
ture they find optimal may lead to thermal The actual behaviour of occupants is often not antici-
dissatisfaction. pated at the design stage. Energy systems in buildings
Personal control does not only improve comfort: are frequently designed in such a way that their use is
office occupants who feel more control over tempera- demanding from the user perspective. Occupants must
ture have less building-related symptoms (Boerstra et al. have an understanding of how the systems work in
2013). order to use them properly, and they have to be
Accordingly, creating buildings that are less sensitive motivated to save energy and actively perform energy-
to occupant behaviour should not mean eliminating the saving tasks.
possibility of personal control, since personal control The effect of occupant behaviour can be diminished
not only improves occupant satisfaction but also in- greatly via robust design solutions—solutions that make
creases toleration of existing conditions and thereby buildings less sensitive to occupant behaviour. This
supports conservation of energy. should be taken to refer not to removing opportunities
Energy Efficiency (2016) 9:1257–1270 1269

for personal control but to making buildings less de- Fabi, Andersen, R. V., Corgnati, S., & Olesen, B. W. (2012).
Occupants’ window opening behaviour: a literature review
manding from the occupant perspective (that is, leaving
of factors influencing occupant behaviour and models.
less need to learn how buildings work, to be motivated Building and Environment, 58, 188–198.
to save energy, or to perform specific energy-saving Faruqui, A., Sergici, S., & Sharif, A. (2010). The impact of
actions). informational feedback on energy consumption—a survey
of the experimental evidence. Energy, 35, 1598–1608.
Fischer, C. (2008). Feedback on household electricity consump-
Acknowledgments The work reported upon here has been sup-
tion: a tool for saving energy? Energy Efficiency, 1, 79–104.
ported by the Indoor Environment Program of RYM Ltd and the
Foster, M., & Oreszczyn, T. (2001). Occupant control of passive
Evidence project funded by the Academy of Finland.
systems: the use of Venetian blinds. Building and
Environment, 36, 149–155.
Handgraaf, M. J. J., Van Lidth de Jeude, M. A., & Appelt, K. C.
(2013). Public praise vs. private pay: effects of rewards on
References energy conservation in the workplace. Ecological
Economics, 86, 86–92.
Herkel, S., Knapp, U., & Pfafferott, J. (2008). Towards a model of
Abrahamse, W., Steg, L., Vlek, C., & Rothengatter, T. (2005). A user behaviour regarding the manual control of windows in
review of intervention studies aimed at household energy office buildings. Building and Environment, 43, 588–600.
conservation. Journal of Environmental Psychology, 25, Hoes, P., Hensen, J. L. M., Loomans, M. G. L. C., de Vries, B., &
273–291. Bourgeois, D. (2009). User behavior in whole building sim-
ASHRAE. (2009). ASHRAE Handbook of Fundamentals, SI ed. ulation. Energy and Buildings, 41, 295–302.
Atlanta, GA: American Society of Heating, Refrigerating and Humphreys, M., & Nicol, F. (1998). Understanding the adaptive
Air-Conditioning Engineers. approach to thermal comfort. ASHRAE Transactions, 104,
Boerstra, A., Beuker, T., Loomans, M., & Hensen, J. (2013). 991–1004.
Impact of available and perceived control on comfort and Hunt, D. R. G. (1979). The use of artificial lighting in relation to
health in European offices. Architectural Science Review, 56, daylight levels and occupancy. Building and Environment,
30–41. 14, 21–33.
Brager, G. S., Paliaga, G., & de Dear, R. (2004). Operable win- Junnila, S. (2007). The potential effect of end-users on energy
dows, personal control and occupant control. Ashrae conservation in office buildings. Facilities, 25, 329–339.
Transactions, 110, 17–35. Karjalainen, S. (2007). The characteristics of usable room tem-
Carrico, A. R., & Riemer, M. (2011). Motivating energy conser- perature. VTT Publications 662. Espoo: VTT Technical
vation in the workplace: an evaluation of the use of group- Research Centre of Finland.
level feedback and peer education. Journal of Environmental Karjalainen, S., & Koistinen, O. (2007). User problems with
Psychology, 31, 1–13. individual temperature control in offices. Building and
Coleman, M. J., Irvine, K. N., Lemon, M., & Shao, L. (2013). Environment, 42, 2880–2887.
Promoting behaviour change through personalized energy Karjalainen, S., & Vastamäki, R. (2007) Occupants have a false
feedback in offices. Building Research and Information, 41, idea of comfortable summer season temperature. In
637–651. Proceedings of Clima 2007 WellBeing Indoors. Helsinki,
D2 National Building Code of Finland. (2003). Indoor climate and Finland.
ventilation of buildings, regulations and guidelines. Helsinki: Karlsson, F., Rohdin, P., & Persson, M.-L. (2007). Measured and
Housing and Building Department, Ministry of the predicted energy demand of a low energy building: important
Environment. aspects when using Building Energy Simulation. Building
Darby, S. (2001). Making it obvious: designing feedback into Services Engineering Research and Technology, 28, 223–
energy consumption. In P. Bertoldi, A. Ricci, & A. de 235.
Almeida (Eds.), Energy efficiency in household appliances Kempton, W. (1987). Two theories of home heat control. In N.
and lighting (pp. 685–696). Berlin Heidelberg: Springer. Quinn & D. C. Holland (Eds.), Cultural models in language
Delmas, M. A., & Lessem, N. (2014). Saving power to conserve and thought (pp. 222–242). Cambridge: Cambridge
your reputation? The effectiveness of private versus public University Press.
information. Journal of Environmental Economics and Kempton, W., Feuermann, D., & McGarity, A. E. (1992). BI
Management, 67, 353–370. always turn it on super^: user decisions about when and
Drake, P., Welch, P., & Zeisel, J. (1986). The role of occupancy how to operate room air conditioners. Energy and
analysis in diagnosing total building performance. In G. Buildings, 18, 177–191.
Davis (Ed.), Building performance: function, preservation, Klein, S. A., Beckman, A., Mitchell, W., & Duffie, A. (2012).
and rehabilitation. Philadelphia: ASTM American Society TRNSYS 17, a TRaNsient SYstem Simulation program.
for Testing and Materials. Wisconsin: Solar Energy Laboratory, University of
Ehrhardt-Martinez, K., Donnelly, K. A., & Laitner, J. A. (2010). Wisconsin.
Advanced metering initiatives and residential feedback pro- Laine, J., & Saari, M. (1994). METOP, CFC-aineeton
grams: a meta-review for household electricity saving op- matalaenergiatoimistotalo [METOP, CFC-free low-energy
portunities. Washington: American Council for an Energy- office building]. Espoo: VTT Valtion teknillinen
efficient Economy. tutkimuskeskus.
1270 Energy Efficiency (2016) 9:1257–1270

Leaman, A. (1999) Window seat or aisle? Studies of buildings in O’Brien, W., Kapsis, K., & Athienitis, A. K. (2013). Manually-
use are showing the importance of personal control of one’s operated window shade patterns in office buildings: a critical
environment, which includes windows. Architects’ Journal, 3 review. Building and Environment, 60, 319–338.
March 1999. Orland, B., Ram, N., Lang, D., Houser, K., Kling, N., & Coccia,
Leaman, A., & Bordass, B. (2000). Productivity in buildings: the M. (2014). Saving energy in an office environment: a serious
‘killer’ variables. In D. Clements-Croome (Ed.), Creating the game intervention. Energy and Buildings, 74, 43–52.
productive workplace (pp. 167–191). London: E & FN Palme, M., Isalgué, A., Coch, H., & Serra, R. (2006) Robust
SPON. design: a way to control energy use from human behavior
Leygue, C. (2014). Saving energy in the workplace; why, and for in architectural spaces. In Proceedings of the PLEA
whom? In Proceeding of Behave Conference. Oxford: Saïd Conference, the 23rd Conference on Passive and Low
Business School. Energy Architecture, Geneve, Switzerland.
Littleford, T. J., & Ryley, S. K. (2014). Firth, context, control and Pfafferott, J., & Boerstra, A. (2014). Comfort, user behavior and
the spillover of energy use behaviours between office and energy efficiency—summary of a workshop at Windsor
home settings. Journal of Environmental Psychology, 40, Conference 2014. The REHVA European HVAC Journal,
157–166. 51, 25–27.
Lopes, M. A. R., Antunes, C. H., & Martins, N. (2012). Energy Pigg, S., Ejiers, M., & Reed, J. (1996) Behavioral aspects of
behaviours as promoters of energy efficiency: a 21st century lighting and occupancy sensors in private offices: a case
review. Renewable and Sustainable Energy Reviews, 16, study of a University Office Building. In: Proceedings of
4095–4104. the 1996 ACEEE Summer Study on Energy Efficiency in
Luo, M., Cao, B., Zhou, X., Li, M., Zhang, J., Ouyang, Q., & Zhu, Buildings (pp. 8.161–8.171). Pacific Grove, CA.
Y. (2014). Can personal control influence human thermal Rijal, Tuohy, P., Humphreys, M. A., Nicol, J. F., Samuel, A., &
comfort? A field study in residential buildings in China in Clarke, J. (2007). Using results from field surveys to predict
winter. Energy and Buildings, 72, 411–418. the effect of open windows on thermal comfort and energy
use in buildings. Energy and Buildings, 39, 823–836.
Lutzenhiser, L. (1992). A question of control: alternative patterns
Siero, F. W., Bakker, A. B., Dekker, G. B., & Van den Burg, M. T.
of room air-conditioner use. Energy and Buildings, 18, 193–
C. (1996). Changing organizational energy consumption be-
200.
haviour through comparative feedback. Journal of
Masoso, O. T., & Grobler, L. J. (2010). The dark side of occu-
Environmental Psychology, 16, 235–246.
pants’ behaviour on building energy use. Energy and
Spence, A., Leygue, C., Bedwell, B., & O’Malley, C. (2014).
Buildings, 42, 173–177.
Engaging with energy reduction: does a climate change
Mulville, M., Jones, K., & Huebner, G. (2014). The potential for frame have the potential for achieving broader sustainable
energy reduction in UK commercial offices through effective behaviour? Journal of Environmental Psychology, 38, 17–
management and behaviour change. Architectural 28.
Engineering and Design Management, 10, 79–90. Steg, L., & Vlek, C. (2009). Encouraging pro-environmental
Murtagh, N., Nati, M., Headley, W. R., Gatersleben, B., Gluhak, behaviour: an integrative review and research agenda.
A., Imran, M. A., & Uzzell, D. (2013). Individual energy use Journal of Environmental Psychology, 29, 309–317.
and feedback in an office setting: a field trial. Energy Policy, Van Den Wymelenberg, K. (2012). Patterns of occupant interac-
62, 717–728. tion with window blinds: a literature review. Energy and
Nicol, J.F. (2001) Characterising occupant behaviour in buildings: Buildings, 51, 165–176.
towards a stochastic model of occupant use of windows, Wang, L., & Greenberg, S. (2015). Window operation and impacts
lights, blinds heaters and fans. In Proceedings of Building on building energy consumption. Energy and Buildings, 92,
Simulation, Seventh International IBPSA Conference (pp. 313–321.
1073–1078). Rio de Janeiro, Brazil. Wargocki, & Seppänen, O. (Eds.). (2006). Indoor climate and
Nolan, J. M., Schultz, P. W., Cialdini, R. B., Goldstein, N. J., & productivity in offices. Rehva guidebook no. 6. Finland:
Griskevicius, V. (2008). Normative social influence is Rehva.
underdetected. Personality and Social Psychology Bulletin, Webber, C. A., Roberson, J. A., Brown, R. E., Payne, C. T.,
34, 913–923. Nordman, B., & Koomey, J. G. (2001). Field surveys of office
O’Brien, W. (2013) Occupant-proof buildings: can we design equipment operating patterns, LBNL-46930. Berkeley:
buildings that are robust against occupant behaviour? In Lawrence Berkeley National Laboratory.
13th Conference of International Building Performance Zhang, Y., & Barrett, P. (2012). Factors influencing the occupants’
Simulation Association (pp. 1746–1754). Chambéry, France. window opening behaviour in a naturally ventilated office
O’Brien, W., & Gunay, H. B. (2014). The contextual factors building. Building and Environment, 50, 125–134.
contributing to occupants’ adaptive comfort behaviors in Zhang, T., Siebers, P., & Aickelin, U. (2011). Modelling electricity
offices—a review and proposed modeling framework. consumption in office buildings: an agent based approach.
Building and Environment, 77, 77–87. Energy and Buildings, 43, 2882–2892.

You might also like