Energy & Buildings 244 (2021) 111009

Contents lists available at ScienceDirect

Energy & Buildings

journal homepage: www.elsevier.com/locate/enb

Smart versus conventional lighting in apartments - Electric lighting

energy consumption simulation for three different households
Nima Hafezparast Moadab, Thomas Olsson, Géza Fischl, Myriam Aries ⇑
Department of Construction Engineering and Lighting Science, School of Engineering, Jönköping University, Jönköping 551 11, Sweden

a r t i c l e i n f o a b s t r a c t

Article history: For residential environments, energy-reducing strategies to cope with user activities and behaviour are
Received 24 November 2020 currently limited mainly to the implementation of improved lighting technology. Non-residential envi-
Revised 2 March 2021 ronments have already been operating smart lighting systems for many years. These systems use
Accepted 8 April 2021
advanced and integrated lighting technology, including an internet-based network for data communica-
Available online 13 April 2021
tion. As user activities and behaviour in the residential sector significantly differ from the non-residential
one; thus, a non-residential energy optimisation potential may not necessarily be achievable and directly
Keywords:
translatable to the residential sector. Also, the architectural typological variety amongst residences may
Indoor lighting
Energy efficiency
be larger than non-residences. In residential buildings, the effect of the composition of a household (do-
Intelligent systems mestic establishment) and user activities on smart lighting systems’ consumption and efficiency are not
Living environment explicitly investigated before. In this light simulation study in Sweden, the electric lighting energy con-
Residential sumption for a two-room apartment was modelled for three different household scenarios using DIALux
Evo and DIVA-for Rhino. The household scenarios were composed based on input by 12 existing Swedish
households and incorporate residential occupancy variety. The study’s findings suggested that the appro-
priate use of smart lighting solutions, including optimised sensor applications, has the potential to save
more than 50% of electric lighting energy consumption compared to non-smart systems. The study
demonstrated promising simulation results specifically focussing on (smart) lighting application alterna-
tives in the residential sector.
Ó 2021 The Author(s). Published by Elsevier B.V. This is an open access article under the CC BY license
(http://creativecommons.org/licenses/by/4.0/).

1. Introduction well-being, and productivity (e.g., [3,4]). Since indoor environ-

ments might not have daylight access all the time, specially
Currently, more than 19% of the total global electrical energy designed electric lighting technologies can be a way of providing/-
consumption is used for lighting, and even though the absolute substituting lighting, potentially affecting users’ health and well-
number will continue to increase [1], relatively, the percentage being [5]. Conversely, these novel lighting technologies influence
by lighting may strongly decrease due to the electrification of cars the ELEC as they include constantly switched-on data communica-
and heating systems. Current electricity production is mainly tion components despite the use of energy-efficient light-emitting
dependent on finite fossil fuels, and electricity demand is not con- diode (LED) sources. The type of protocol used for (wireless) com-
stant throughout the day. Until a complete transition is made to munication in smart systems (including lighting) has varying
fully sustainable electricity production, demand must be mini- power use. The ZigBee protocol is frequently used for smart com-
mized, and peak use needs to be flattened. Electric light, daylight, munication and consumes less power than other protocols like
or the combination is utilized for illumination of indoor environ- solely Wi-Fi-based [6].
ments. In Scandinavia, electric lighting represents 15–30% of the A literature scan (Scopus, ScienceDirect, and Google Scholar)
total electricity consumption [2] and optimising this can not only showed an apparent inequality in the numbers of studies dealing
contribute to a noticeable reduction of annually produced green- with smart or intelligent lighting in indoor environments that fun-
house gasses, but can also assist in regulating peak loads. Daylight damentally investigate non-residential buildings like offices or
use is both important for its effect on the electric lighting energy schools rather than the residential sector. In most countries, the
consumption (ELEC) as well as for its influence on people’s health, residential sector consumes significantly more energy (often but
not always in the form of electricity) than the commercial sector
⇑ Corresponding author. [7]. The difference in (electric) energy consumption is primarily
E-mail address: [email protected] (M. Aries). due to differences in user activities and behaviour (e.g., [8]). In

https://doi.org/10.1016/j.enbuild.2021.111009
0378-7788/Ó 2021 The Author(s). Published by Elsevier B.V.
This is an open access article under the CC BY license (http://creativecommons.org/licenses/by/4.0/).
N. Hafezparast Moadab, T. Olsson, Géza Fischl et al. Energy & Buildings 244 (2021) 111009

non-residential buildings, various strategies to cope with the ferent companies provided embedded software to control LED
results of user activities and behaviour are mentioned and tested lamps to create so-called Smart LEDs, which offers lighting cus-
for ELEC reduction, ranging from improvements of the lighting tomisation, particularly in the home environment connected to
technology to the introduction of occupancy sensors (e.g., [9,10]). wireless network technology (i.e., Wi-Fi, ZigBee, Bluetooth). Reli-
For (daylit) non-residential buildings, systems are installed that able results regarding implementation and their effects are not
enable communication between various system inputs and outputs available yet. Dimming strategies to positively influence user expe-
to control luminaires to offer lighting customisation related to rience are tested in non-residential environments (e.g., [29–31]),
occupancy and daylight presence. Lighting control systems are but only limited in a residential setting (e.g., [32]).
classified as smart lighting systems (SLS) in case the systems The presence and behaviour of building occupants have a signif-
involve the installation and use of advanced and integrated build- icant impact on the building’s energy performance. Even though
ing technology systems, including an internet-based network for understanding building occupants’ behavioural and presence pat-
data communication. The energy demand of these SLS contains terns has risen, occupant behaviour in building performance simu-
two main parts: power consumption for functional illumination, lations is often oversimplified due to its stochastic, diverse,
comparable with traditional lighting control systems, and standby complex, and interdisciplinary nature [33,34]. This simplification
power consumption, as a prerequisite for data communication. may contribute to performance gaps between simulated models
Principi and Fioretti [11] executed a comparative life cycle assess- and actual building energy consumption [33]. Increased computa-
ment study of office lighting using CFL and LED sources and tional power and availability of technical solutions have enabled
showed that LEDs have a lower environmental impact than CFLs, the use of more sophisticated methods [35,36]. Typical occupancy
mainly due to a lower impact during the use stage. The cumulative schedules can be defined for building energy (saving) modelling
energy demand during the use phase reduces 43–52% by going and assessing perceived comfort using occupancy sensing technol-
from CFL to LED. Daylight-linked control systems adjust the elec- ogy and occupancy-dependent controls. According to recent stud-
tric light level by providing light switching and dimming when ies (e.g., [35,37]), occupancy patterns and occupants’ actions
daylight is present. The amount of ELEC saving reported is over (window, solar shading, electric lighting, thermostat adjustment,
40% and as high as 80% for integrated daylight systems (e.g., clothing adjustment and appliance use) can be gathered and
[12,13]). Sensor-based optimising solutions enable intuitive user detected in different ways. Existing data and databases can be used
interaction by recognising the location and activities of a user to develop schedules using probabilistic data-driven and machine/
and creating an appropriate lighting environment [14]. The deep learning techniques (Markov chains). Additionally, new data
amount of ELEC saving for non-residential buildings varies can be used involving sensor data and/or time use survey data
between 3% and 60% for occupancy detection systems depending (e.g., [38]), hence inquiring qualitative and quantitative data.
on the type of system, occupant usage pattern, sensor location, Multiple studies demonstrate the importance of applying (near)
field of view, quality, calibration, and time delay (e.g., actual user behaviour data in energy simulation analyses (e.g.,
[13,15,16]). Delay time, the period or amount of time during which [37,39,40]). Mitra, Steinmetz [37] used data collected over 12 years
the light turn-off is delayed, is one of the factors affecting the per- to develop typical residential occupancy schedules for various
formance of occupancy-based and daylight-linked lighting control household types and occupant age ranges. They compared the data
systems (e.g., [17–19]). Researchers have found that reducing with residential occupancy schedules used for building perfor-
switch-off time delays from 20 to 5 min may increase energy sav- mance simulation. The results varied from exhibiting similar pat-
ings from 26% to 33% in regularly occupied spaces [20] or cause a terns to differences of more than 40%. Muroni et al. (2019)
26% improvement in the energy efficiency of smart systems by extracted occupancy, equipment, and light use profiles from data
decreasing the turn-off time delay from 20 to 1-min [18]. The effi- of 12 Dutch dwellings and compared electricity consumption as
ciency of occupancy-based lighting controls in non-residential predicted by simulation tools with on-site measurements. Integra-
buildings varied between 20% and 93%, highly depend on turn-off tion of occupancy profiles decreased the deviation between simu-
delay time [21]. Reducing the time delay from 20 min to 1 min lation and reality from 22.9% to 1.7%. The review by Delzendeh, Wu
led to 26% more energy-saving [18]. Gentile [22] argued that delay [41] identified that analysis of occupants’ behaviour impact should
time could cause a 26% up to 66% difference in ELEC. Adoptive be investigated both at a more extensive (e.g., urban analysis) as
time-delay depends on individual activities, providing an addi- well as a smaller scale (e.g., space layout, type of light sources).
tional 5% ELEC savings compared to fixed time-delay [21]. Research Recently, Wolf, Calì [42] have simulated occupancy at room level
in commercial buildings showed that standby power consumption and took weekday, time of day, occupant age, and family type into
could be as high as 25% to 30% of the energy required for lighting account. The simulations were compared to measurements with a
[23,24]. The lack of proper coordination and configuration of SLS (preliminary) positive result.
can lead to more energy consumption. For example, in low occu- As user activities and behaviour in the residential sector signif-
pancy areas or places with high daylight penetration, standby con- icantly differ from the non-residential sector, a non-residential
sumption can reach 55% of the energy use [22]. The energy energy optimisation potential may not necessarily be achievable
efficiency of SLS can be maximised by the combination of smart and directly translatable to the residential sector. Additionally,
LEDs with sensor-based optimising solutions [25]. user activities and behaviour may be much more diverse amongst
For residential environments, energy-reducing strategies to cope different residents than office employees or students. The effect of
with user activities and behaviour are mainly limited to imple- user activities and the composition of a household (domestic
menting improved lighting technology. LEDs are currently the establishment) in residential buildings on the consumption and
most efficient light sources with a luminous efficacy of 120– efficiency of SLS are not specifically investigated before. Therefore,
150 lm/W as compared to Compact Fluorescent Lamps (CFLs) of this study aimed to simulate the effect of (smart) lighting applica-
around 40–80 lm/W [26]. In a simulation study executed for Oslo tion using an occupancy protocol for a selection of household sce-
(Norway), all light sources were replaced by LEDs as part of their narios and compare their ELEC. Even though lighting simulation
energy upgrading holistic retrofit method for residential building still has restrictions in representing full reality, it can provide
blocks, and a comparable electricity consumption reduction of researchers with faster and improved ways to compare complex
40% was found [27]. Mills and Schleich [28] reported lighting solutions and results that would have taken long and costly time
energy savings by replacing CFLs with LEDs of circa 23% for a data- spans to obtain with laboratory or field tests [43]. Lighting simula-
set of German households. Additionally, over the last decade, dif- tion enables relatively quick comparisons of lighting solutions
2
N. Hafezparast Moadab, T. Olsson, Géza Fischl et al. Energy & Buildings 244 (2021) 111009

provided by varying combinations of electric and natural lighting. 2.2. Household scenario and averaged occupancy patterns
In a light simulation study, the ELEC in a two-room apartment
in Sweden for three different hypothetical household scenarios In their study, Johansson, Bagge [45] studied occupancy level
was investigated. Based on questionnaire results from 12 Swed- in 32 apartments across Sweden based on the CO2 concentration
ish families, occupancy pattern detection was used to define of the indoor air for variations during the year, week, and day. As
three different households and their characteristics. Simulated these results did not provide detailed daily occupancy profiles at
data were compared for the use of three different light sources room level, an online questionnaire was designed and distributed
(CFL, LED, and Smart-LED) and the application of sensors (occu- to 12 Swedish households living in apartments (convenience
pancy with varying times of delay and daylight harvesting), day- sampling) in Jönköping. The questionnaire investigated daily
lighting, and switch on/off occupancy-based lighting control for a activities of a household occupancy pattern and was sent out
full year. on September 11th, and the last one was returned on October
12th, 2019 (fall season to avoid days within the Swedish summer
with very long photoperiods and winter with very short photope-
2. Materials and methods riods). People were inquired about the household composition
(amount of people and age), work schedule(s), as well as the
2.1. Residential setting dimensions of their apartment, living room, kitchen, and bed-
room(s). Additionally, they were asked to fill out a table related
A 75 m2 two-room apartment was selected as the investigated to which activities the household members would perform and
residential setting, see Fig. 1. It represents one of the most common what place in the apartment (per 30 min). There was a table
typologies of residences in Sweden (36.6% of all apartment types for weekdays and weekends, and people were asked to fill them
[44]). The study took into consideration the occupancy of the out for typical days.
kitchen (12 m2), living room (24 m2), and the bedroom (12 m2). The outcome of the convenience sampling was N = 5 single,
The bathroom, entrance, and balcony (total area 27 m2) were not N = 5 two-member, and N = 2 three-member households. Their
included as the SLS application is not recommended for spaces information was used to create averaged occupancy patterns for
with too low/short occupancy [22]. The site location was set to three household types (a single-, two-, and three-member). Note
Gothenburg, Sweden (57°420 N 11°580 E) since annual files from that a two-member household can, for example, consist of two
Gothenburg’s Airport (Landvetter) were available for the required adults or one adult and a child, but for this study, only the input
weather data. For layout modelling, Autodesk Revit and Diva for for assuming two adults was included. A 24-h occupancy pattern
Rhino 6 were used, using the default material settings in Rhino was generated for each household scenario for a week and week-
(floor q = 0.20, walls/furniture q = 0.5, ceiling q = 0.90, outdoor end day for three apartment areas (living room, kitchen, bedroom).
plane q = 0.20). The thermal and optical properties of applied glaz- The occupancy pattern per household and per room is an average
ing as well as the calculation settings for direct and diffuse light of the questionnaire’s patterns, hence the partial values (between
calculation are shown in Table 1. No shades or external obstruc- 0% and 100%) in Fig. 2. Note that the higher the bar in the figure,
tions were considered in the light calculation. the more residents were at the same time at the same place. If a
person was not in any of these household areas, it was assumed
that the person was in another room or outside. The reference year
included 261 week and 104 weekend days.
The delay time energy consumption was calculated (and aver-
aged) based on the questionnaires mentioning their occupancy in
30 min intervals. If it was mentioned that an area was occupied
for 30 min, no delay was considered during this period. A delay
time was considered after the 30 min if the room was not occupied
for the next 30-minute period. See Table 2 shows three examples.
There are dedicated lighting scenes associated with activities in
each investigated area, with three levels: 100% to reach the
required light level, 0% when turned off, and 50% to accommodate
dimmed background levels, see Table 3. For example, there are two
scenes in the kitchen: a ’Food preparation and cleaning’ and an
’Eating’ scene. General light and two dimmed task lights are acti-
vated in the affected rooms (kitchen and living room) for the for-
mer. Meanwhile, for the latter, a full task light is applied in the
kitchen with dimmed general light and a dimmed task light in
the living room. Note that multiple scenes were activated with
multiple people present in the apartment at the same time (i.e.,
‘Food preparation/Cleaning’ in the kitchen and ‘Relaxing (TV)’ in
the living room). It is worth noting that by raising the number of
people living in an apartment, the accumulated occupancy dura-
tion in different areas does not necessarily increase (e.g., persons
eat together in one area). The light scenes were a simplified
assumption of how occupants may set the smart light sources
within their house. However, they are based on recommendations
for suitable scenarios in residential rooms from an energy-saving
Fig. 1. Floor plan with light source positions: General Lighting (GL) is integrated perspective [26] as well as on definitions for manual and auto-
into the ceiling, while Task Lighting (TL) is linked to furniture. Windows Win1,
Win2, and Win5 all have a windowsill height (0.90 m), while windows Win3 and
mated dimming used as a method to improve electric lighting
Win4 are full wall-sized. Only light sources in the kitchen, bedroom, and living energy consumption in other (non-residential) studies (e.g.,
room are included in calculations. [15,46,47]).
3
N. Hafezparast Moadab, T. Olsson, Géza Fischl et al. Energy & Buildings 244 (2021) 111009

Table 1
Thermal and optical properties of applied glazing as well as light calculation settings.

DIVA-for-Rhino glass material Visual transmissivity Visible transmittance sv Solar factor g [%] U-value Ug [W/m2K]
Glazing_DoublePane_Clear* 87 80 72 2.71
Light calculation settings (Radiance comment ‘rtrace’)
Direct -h -I + -oc -aa 0.100000 -ab 2 -ad 1000 -ar 300 -as 20 -dj 0.000000 -dr 2 -ds 0.200000 -lr 6 -lw 0.004000 -sj 1.000000 -st
0.150000 -L 1000.000000 -N 65 -Dm -dt 0.000000
Diffuse -h -I + -oc -aa 0.100000 -ab 2 -ad 1000 -ar 300 -as 20 -dj 0.000000 -dr 2 -ds 0.200000 -lr 6 -lw 0.004000 -sj 1.000000 -st
0.150000 -L 1000.000000 -N 148 -Dd
*
Standard material in DIVA-for-Rhino

Fig. 2. Occupancy pattern for a 24-h week (left) and weekend (right) day in the living room area for the three household types showing the average portion of time per hour
spent the space.

Table 2
Delay time energy calculation examples.

From Till What? Where? Delay?

Example 1
11:30 12:00 Eating Kitchen No
12:00 12:30 Out Out Yes
Example 2
11:30 12:00 Eating Kitchen No
12:00 12:30 Eating Kitchen No
12:30 13:00 Out Out Yes
Example 3
11:30 12:00 Eating Kitchen No
12:00 12:30 Out Out Yes
12:30 13:00 Cleaning Kitchen No
13:00 13:30 Out Out Yes

2.3. Lighting design and sensor application The TL for this study is personalised for defined characters and
designed to provide up to 400 lx for tasks like cooking or reading.
Both daylight and electric lighting were considered for compu- All household scenarios are assumed to have the same number
tational modelling. Daylight was supplied by openings located on of light sources (N = 12) at the same apartment locations. The GL
the North-East side of the apartment (reduced daylight contribu- was designed to provide a minimum amount of light (Ehor at
tion) while electric lighting systems, separated into general light- h = 0.8 m; for a kitchen: 180 lx, a living room: 150 lx, and a bed-
ing (GL) and task lighting (TL), are positioned as shown in Fig. 1. room: 100 lx) required by standards [26]. In DIALux DIALux Evo -
An ample amount of light (at floor level) is needed to provide a safe version 8.1, a calculation plane at h = 0.8 m (0.5 m from the walls)
and general environment, while TL is usually designed for provid- was used to check if the light levels (virtual sensor points devoted
ing an additional amount of light required for specific activities. for useful daylight illuminance (UDI) calculations, see Fig. 3)

4
N. Hafezparast Moadab, T. Olsson, Géza Fischl et al. Energy & Buildings 244 (2021) 111009

Table 3
The dedicated lighting scenes for each activity.

Area Scenes per activity Kitchen Lighting (%) Living room lighting (%) Bedroom Lighting (%)
General Task General Task General Task
Kitchen Food preparation/Cleaning 100 100 50 50 0 0
Eating 100 100 50 50 0 0
Living room Relaxing (TV) 0 0 100 100 0 0
Relaxing (reading) 0 0 100 100 100 100
Working 100 100 100 100 0 0
Socialising/Playing 50 50 100 100 50 50
Cleaning/Else 0 0 50 50 0 0
Bedroom Playing 0 0 100 100 100 100
Relaxing (reading) 0 0 50 50 100 100
Sleeping 0 0 0 0 0 0
Cleaning/Else 50 50 0 0 100 100

conventional setting has no occupancy detection and no daylight

harvesting capability, while the smart setting is equipped with
both occupancy detection and daylight harvesting options. The
ELEC was calculated for eight lighting combinations, see Table 5.
For the effect of daylight on the total ELEC in the conventional set-
ting, it was assumed that users turn on electric light sources when
the illuminance level is lower than standard limits and turn them
off right after when the standard level reached. It is a very opti-
mistic and probably unrealistic scenario due to the large sensitivity
range of the human eyes.
In the smart setting, the Smart-LEDs turn on whenever and
wherever occupants are present and when the daylight is not
enough to provide the required light levels. The occupant detection
system in the smart setting turns on lights when users enter the
place and turns it off immediately after the place is empty (delay
time = 0 min). The impact of a 0, 5, 10, and 15-min turn-off delay
on ELEC was investigated. This calculation required an occupancy
frequency per area, and the required information was extracted
and averaged from the questionnaires, see Fig. 2. For periods that
an area was used continuously, no turn-off delay time was
considered.
Standby energy consumption is the consumption of an SLS
when in ’turned-off’ mode. The standby power consumption was
considered for the Smart-LEDs and was extracted from smart light-
ing bulbs (Philips Hue, 0.4 W per lamp) and a bridge (1.5 W) as
Fig. 3. The apartment’s floorplan in DIALux with the virtual sensor points devoted measured by others [50,51]. The energy required for occupancy
to useful daylight illuminance calculations in the kitchen, living room, and detection and daylight harvesting sensors (battery-powered) was
bedroom. not considered in the calculation of ELEC.

agreed on the minimum levels when all light sources were on at 2.4. Procedures and analysis
100% in the three main rooms. Tables or desks for eating and work-
ing generally have this height of h = 0.8 m, whereas kitchen coun- Parameters calculated in this study were limited to horizontal
ters are slightly higher (h = 0.85–0.9 m), and bedside tables and illuminance (Ehor), see Section 2.3, and electric lighting energy
cabinets are slightly lower (h = 0.6–0.7 m) [48]. consumption (ELEC). Regarding the calculation of ELEC (in
The TL mainly provided the amount of light required for specific kWh), first, the power consumption of each light source (CFL or
tasks, and the chosen light levels are based on recommendations LED) was calculated and multiplied by the in-use time without
by the Energy Saving Trust [26] since the European/Swedish stan- the presence of daylight. The in-use time was calculated per
dard [49] does not provide values for private spaces. The lumi- room, including the lighting scenes per activity. For example,
naires’ mounting height is 2.5 m above the floor (ceiling), except the ELEC required for ’preparing food’ in the kitchen for 20 min
the TL in the bedroom, which are installed at 1.5 m. In this study, was determined. Annual-hourly room daylight illuminance,
three light source types were investigated and compared: CFLs, extracted from the software (DIVA-for-Rhino) for the reference
LEDs, and Smart-LEDs. The Smart-LEDs have the same technical year in Gothenburg, was turned into a 24-hour daily plan. Next,
characteristics as the LEDs but are additionally equipped with a to show the amount of energy that daylight can save, the amount
smart control system. The power (in watt) and beam angle per of required electric lighting was calculated by subtracting the
light source and room is shown in Table 4. minimum illuminance level as specified in the standard from
The apartment was equipped with daylight harvesting sensors the available daylight illuminance. Finally, the optimisation with
in the kitchen and bedroom as well as with an occupancy sensor sensors was calculated using the Smart-LED for four different
in each investigated room. The ELEC of the lighting systems was delay times and including standby energy consumption. See also
calculated employing two different lighting control settings. The the workflow in Fig. 4.
5
N. Hafezparast Moadab, T. Olsson, Géza Fischl et al. Energy & Buildings 244 (2021) 111009

Table 4
Power (in watt) and beam angle (in degrees) per room and light source (CFL and LED) for general and task lighting.

Room Light source Light source Light source

general lighting task lighting 1 task lighting 2
Type Power [W] Beam angle [°] Type Power [W] Beam angle [°] Type Power [W] Beam angle [°]
Living room CFL 23 90 CFL 19 70 CFL 14 45
Bedroom CFL 19 90 CFL 11 45
Kitchen CFL 27 90 CFL 14 –
Living room LED 16 90 LED 14 70 LED 10 45
Bedroom LED 14 90 LED 5 45
Kitchen LED 18 90 LED 5 –

Table 5 46–50%. Differences between the household scenarios (including

Overview of the applied settings with the light source and sensor specifications. occupancy pattern nesting) are only 1–4%.
Setting Light source Occupancy detection Daylight harvesting As can be seen in Fig. 5, substituting all CFL light sources in the
Conventional CFL No No
apartment with LED sources (with manual daylight control) means
CFL No Yes* an average saving of 34%, and changing from LED light sources
LED No No (with manual daylight control) to Smart-LEDs means an average
LED No Yes* saving of an additional 14%. Substituting all CFL light sources in
Smart Smart-LED Yes – 0 min delay Yes
the apartment with Smart-LED sources can lead to 47% saving on
Yes – 5 min delay Yes
Yes – 10 min delay Yes average. For the conventional setting with CFLs and LEDs, the situ-
Yes – 15 min delay Yes ation in which users consequently turn on/off lights when the day-
*
light level goes above/below standard values was considered. For
No sensor, manual switching.
CFL sources, the use of manual daylight control lowers the ELEC
by 13–18% on a yearly basis; for the more efficient LED sources,
3. Results this reduction is 10–14% (Fig. 6).

In Table 6, the yearly electric lighting energy consumption 3.2. Smart lighting systems
(ELEC) in kWh is presented per household and different lighting/
sensor solutions. Comparisons are made to investigate the influ- The introduction of an SLS at home involves both the change
ence of household, type of light source, light setting, including sen- from regular LED light sources to Smart-LEDs and the inclusion
sors’ application (daylight, occupancy). Additionally, the effect of of an internet-based network for data communication. Instead of
being home/away and the use of one/multiple spaces (apartment a dedicated user trying to switch on or off the lights related to a
occupancy) is investigated. subjectively assessed daylight situation, the system responds to a
sensor output. The introduced automation means the takeover of
user behaviour leading to an average saving of 10% in the case of
3.1. Household scenario daylight harvesting (from 97 to 88 kWh/year for a single-
member, 154 to 139 kWh/year for a two-member, and 119 to
Fig. 5 shows the ELEC for the three different households and the 107 kWh/year for a three-member household).
use of different light sources. The two-member household gener- Sensor input can also be related to the occupancy of a space. The
ally has the highest ELEC as their occupancy pattern is the most energy consumption for four different time delay intervals of 0, 5,
diverse between the members. The largest difference was demon- 10, and 15 min was calculated separately for weekdays and week-
strated for the single-member household’s annual ELEC compared ends in the apartment’s three spaces per household scenario. On a
to the two-member household with an increase of 60–63%. The yearly base, the occupancy-related turn-off delay time has the lar-
ELEC difference between the single and the three-member house- gest impact for the single-member household. Going from 5 to
hold was 21–33%, which means differences between the household 0 min reduces the ELEC by 10%, and from 15 to 0 min by 30%.
scenarios, including their specific averaged occupancy patterns of For the larger households, the effect by occupancy-related turn-
3–12%. Fig. 5 compares the ELEC results relative to CFL sources, off delay time from 15 to 0 min on ELEC is about 10% lower: 19%
which corrects for the specific occupancy patterns. Upgrading from for two-member and 22% for three-member household scenario
a conventional setting with CFLs to a smart setting with Smart-LEDs (Fig. 7).
showed comparable differences between the three household sce- However, the energy demand of an SLS contains, next to power
narios. Changing all light sources from CFL to LED means an ELEC consumption for functional illumination, a standby power con-
reduction of 33–35%, and from CFL to Smart-LED, a reduction of sumption, as a prerequisite for the data communication. This

Fig. 4. Overview of the workflow and studied alternatives (per household and light source).

6
N. Hafezparast Moadab, T. Olsson, Géza Fischl et al. Energy & Buildings 244 (2021) 111009

Table 6
Overview of the electric lighting energy consumption (ELEC) in kWh per year for three different household and different lighting/sensor solutions.

Single member Setting Light source Standby Occupancy & delay Daylight harvesting ELEC (kWh/yr]
Conventional CFL No No No 162
CFL No No Yes* 141
LED No No No 108
LED No No Yes* 97
Smart Smart-LED Yes Yes – 0 min Yes 88
Yes Yes – 5 min Yes 97
Yes Yes – 10 min Yes 105
Yes Yes – 15 min Yes 114
Two-member Setting Light source Standby Occupancy & delay Daylight harvesting ELEC (kWh/yr]
Conventional CFL No No No 264
CFL No No Yes* 225
LED No No No 175
LED No No Yes* 154
Smart Smart-LED Yes Yes – 0 min Yes 139
Yes Yes – 5 min Yes 148
Yes Yes – 10 min Yes 156
Yes Yes – 15 min Yes 165
Three-member Setting Light source Standby Occupancy & delay Daylight harvesting ELEC (kWh/yr]
Conventional CFL No No No 215
CFL No No Yes* 177
LED No No No 139
LED No No Yes* 119
Smart Smart-LED Yes Yes – 0 min Yes 107
Yes Yes – 5 min Yes 115
Yes Yes – 10 min Yes 123
Yes Yes – 15 min Yes 131
*
No sensor, manual switching

Fig. 7. Relative electric lighting energy consumption per household scenario for the
Fig. 5. Relative electric lighting energy consumption per light household scenario smart light setting using Smart-LED light sources with occupancy sensing and turn-
relative to CFL light source application. off delay times of 0, 5, 10, and 15 min (results are relative to the 0-min delay).

Table 7
Annual and percentual ELEC for data communication per household scenario for the
smart setting with daylight and occupancy sensors. Minimal ELEC = occupancy-
related turn-off delay of 0 min and maximal ELEC = delay of 15 min.

ELEC single ELEC two- ELEC three-

member member member
[kWh/year] [kWh/year] [kWh/year]
ELEC Smart-LEDs (Min – Max) 88–114 139–165 107–131
Electric energy consumed - 13 (11–15%) 13 (8–9%) 13 (10–12%)
data transfer system (% of
total ELEC)
Electric energy consumed – 45 (39–51%) 42 (25– 43 (33–40%)
standby use (% of total ELEC) 30%)

Fig. 6. Relative electric lighting energy consumption per household scenario. On

the left, results for a conventional light setting using CFL light sources with manual means that the data transfer system (hub/bridge) and light sources
daylight harvesting implementation and, on the right, the LED light sources with
are always in standby mode, even when the lighting system is
manual daylight harvesting implementation (results are relative to the situation
with no daylight control). in ’turned-off’ mode. Table 7 shows the annual and percentual

7
N. Hafezparast Moadab, T. Olsson, Géza Fischl et al. Energy & Buildings 244 (2021) 111009

electric energy consumed by the internal chipsets of the Smart-LED ing setting using LED light sources (without data communication
light sources in the three main areas of the apartment and one data chip) with daylight-related switching behaviour to a setting using
transfer system (hub/bridge energy consumption: 13 kWh/year) Smart-LEDs including daylight harvesting and occupancy detection
for the investigated households. The light sources’ standby energy results in a saving of 10–11% with an ELEC as low as 88 kWh per
consumption is between 42 and 45 kWh/year, which is between 25 year (single-household). However, the simulations assumed a
and 51% of the total ELEC. Data transfer counts for 8–15% of the well-installed, never failing system, which may not always be the
overall energy consumption. case. Other studies showed that poorly and inadequately imple-
mented, calibrated, commissioned, or too complex control systems
will lead to user discomfort, which can result in a lower ELEC sav-
4. Discussion
ing and, worst case, in a complete deactivation of the control sys-
tem [10]. Therefore, the main goal of SLS implementation must be
The study simulated and investigated the impact of conven-
achieving ELEC saving without sacrificing user comfort [52]. This
tional versus smart lighting with and without sensor application
calls for SLS implementation and validation studies, including user
in a two-room apartment in Sweden occupied by three different
experience and assessment.
households, including user activity patterns, on electric lighting
Daylight-linked control systems adjust the electric light level by
energy consumption. As user activities and behaviour in the resi-
providing light switching and dimming when daylight is present.
dential sector significantly differ from the non-residential sector,
The amount of ELEC saving reported is over 40% and as high as
a non-residential energy optimisation potential may not necessar-
80% for integrated daylight systems in non-residential buildings
ily be achievable and directly translatable to the residential sector.
(e.g., [12,13]). Looking at the results for a three-member house-
The largest difference between residential and non-residential
hold, the use of daylight switching for the LED conventional setting
building use may be the large variety in occupation and space
decreases the annual ELEC from 139 kWh to 119 kWh (17% saving).
use of comparable areas.
For the other two household scenarios, this is lower (11–14%).
However, the simulated apartment floorplan in Fig. 1 shows that
4.1. Household scenarios and use of smart lighting the daylight admittance is not optimal and primarily limited to
the kitchen and bedroom. The entrance to the living room is
Reviewing the three household scenarios shows a difference in shielded by the balcony. As seen in other studies (e.g., [53,54]), a
yearly ELEC. The two-member household has the highest ELEC as carefully positioned and designed daylight opening size (40–60%
their occupancy pattern is the most diverse between the members. window-to-wall ratio) can significantly influence the ELEC. For
In the two and three-member households, the probability of simul- example, a window with the highest performance, a U-value in
taneous activities in two or more areas is higher than in the single- thermal performance studies, may not always be the most optimal
member household. The most substantial annual ELEC difference energy or economical solution [55]. In certain situations, optimiz-
was demonstrated using CFLs under the conventional setting (39% ing the daylight situation may be energetically and/or economi-
between single- and two-members). Smart-LEDs using a smart set- cally more valuable than the introduction of SLS. Regulations
ting were more efficient for two- and three-member households related to daylight access are minimal for the residential sector
than single-member, most likely because of the more extended compared to the non-residential sector; hence, access to daylight
and more consistent occupancy of different areas as well as the in i.e., offices is often much higher, reflecting in the much higher
higher efficiency of smart solutions. However, the results showed saving potential. If working from home will happen more fre-
that ELEC does not necessarily go up by increasing the number of quently as an outcome of/after the COVID-19 pandemic, attention
occupants and is more affected by control strategy, occupancy, needs to be paid to the workplace’s location to benefit from day-
activities, and user behaviour. If the effects of occupancy pattern light and occupancy-controlled systems optimally.
are nested in the data by comparing the households with them- Sensor-based optimising solutions enable intuitive user interac-
selves, differences between the household scenarios are only 2% tion by recognising the location and activities of a user and creat-
to 4%. The effect of occupancy, activities, and user behaviour on ing an appropriate lighting environment [14]. The amount of ELEC
ELEC can be studied when investigating the effect of space use, saving for non-residential buildings varies between 3% and 60% for
comparing scenarios where there are people at home during week- occupancy detection systems depending on the type of system,
days. With only one person at home (single household), this person occupant usage pattern, sensor location, field of view, quality, cal-
is likely to stay in one room and use lighting in just that room. Peo- ibration, and turn-off delay time (e.g., [13,15,16,21]). Adding a
ple can either share a room or be in separate rooms (three-member delay time for systems in residential buildings increases the ELEC
household). The effect of using multiple space use for the smart set- of SLS. Delaying from 0 to 5 min increases the ELEC of the SLS by
ting with Smart-LEDs showed 88 kWh/year ELEC for one person 10% for a single-member household and from 0 to 15 min by
and 107 kWh/year for multiple people (person 2 and 3); an 30%. For the three investigated household scenarios, the increase
increase of 18%. This calls for SLS implementation and validation in energy is 6–30% (0 to 15 min delay), which is in the low range
studies, including user activities and behaviour. compared to non-residential buildings. A recent study, executed
in corridors of a high-rise apartment building, reported 14% less
4.2. Residential versus non-residential electric lighting energy saving energy use for the short delay time (5 min) setting compared to
the long delay time (15 min) setting [32]. Despite the savings, add-
As said, for residential environments, energy-reducing strate- ing a system turn-off delay time may affect user comfort. Delay
gies to cope with user activities and behaviour are currently time is known as one of the techniques to provide user comfort
limited mainly to the implementation of improved lighting tech- [13] but could result in the opposite. If the lighting turns off as
nology. For commercial buildings, replacing CFLs with LEDs shows soon as a person leaves the room, it may leave certain parts of
a reduction of energy demand by 43–52% [11]. The substitution of the apartment dark and visually uncomfortable.
CFLs with LEDs in this simulated Swedish apartment of 75 m2 The standby energy consumption counted for 25 to 51% of total
resulted in 33–35% electric energy saving per year, which is in line annual ELEC when using Smart-LEDs for the three household sce-
with comparable studies reported [27,28]. Application of Smart- narios, which is generally higher than the 25 to 30% mentioned
LEDs will only be efficient when the smartness built into the light in studies performed in commercial buildings [23,24]. The lack of
source is utilized. On average, changing from a conventional light- proper coordination and configuration of SLS can lead to more
8
N. Hafezparast Moadab, T. Olsson, Géza Fischl et al. Energy & Buildings 244 (2021) 111009

energy consumption. For example, in low occupancy areas or more than calculated in this study and empowering these systems
places with high daylight penetration, standby consumption can with smart solutions could gain more benefits.
reach 55% of the energy use in non-residential buildings [22]. The study has used an existing but randomly selected apart-
Spaces with a low occupancy rate (hallway, bathroom, balcony) ment layout for simulation, even though it is shown that several
were excluded from the study as it was expected that in these architectural variables do influence residential lighting use during
spaces, the ‘smartness’ of an SLS would not be utilised, increasing daytime [5]. The considered layout of the study case provided no
the standby consumption percentage even more. The highest direct daylight in the living room. However, the comparison, made
standby contributions are found in single-member household based on the occupancy time (day and night), window positioning,
and are likely related to the reduced space use (low occupancy). and area of each room, showed that users spend more time in the
Data transfer counted for 8–15% of the overall energy use, indicat- living room than in the kitchen or bedroom (see Fig. 2). In the lay-
ing that neglect of standby and data transfer consumption in smart out of the chosen apartment, these rooms are the spaces with the
systems can lead to unrealistically high savings expectations [21]. highest daylight admittance, which may mean that ELEC saving
The calculated ELEC saving ranges for each of the investigated may be higher if daylight availability is optimized. Moreover,
energy-reducing strategies involving (smart) lighting for the apart- reviewing previous research shows the importance of daylight in
ment were compared to a non-residential estimated saving range providing user comfort [61]. Therefore, a good floor plan should
from literature sources as discussed in this paragraph; hence pos- provide more daylight access for higher occupancy areas during
sibly incomplete. Nevertheless, as illustrated in Fig. 8, the first the daytime, like a living room.
comparison shows that the potential ELEC savings between the Currently, optimization of user comfort was not included in the
two sectors, in general, broadly vary and are not in the same order study, as the focus was on calculating energy consumption. This
of magnitude. includes the absence of luminance or solar shading, which can
influence the amount of admitted daylight as people would close
4.3. Study limitations blinds or shades in cases with high visual and/or thermal comfort.
However, the findings stress the importance of a well-considered
The validity and reliability of the results are affected by several selection of different lighting solutions and the suitable inclusion
factors. The comparison made between the three households indi- of daylight. If an evaluation of room occupancy patterns, user activ-
cated no direct relationship between the number of people in the ities, as well as daylight availability (presence and level) is
household and the ELEC. Effects of space use and occupancy hours available before the selection and implementation of lighting solu-
were shown, emphasizing, once again, the importance of including tions, an appropriate combination of LEDs, Smart-LEDs, and sen-
behavioural parameters in energy measuring/predicting residen- sors can be composed. A careless selection could result in a
tial studies (e.g., [8,56–58]). This study’s occupancy patterns were substantial misuse of energy and dissatisfied users.
extracted from subjective questionnaire results instead of basing
them on assumptions as is usually done. This is, however, still
not as accurate as on-site occupancy detection techniques or the 5. Conclusions and further research
use of validated residential occupancy profiles at room level that
are currently under development (e.g., [59]). Besides, the apart- Substitution of CFL with LED light sources leads to a minimum
ment size was set at 75 m2, and no comparisons are made with nei- of 30% ELEC saving regardless of any smartness in the system.
ther 75 m2 non-residential areas nor with larger apartments or Therefore, the increased use of LEDs should be considered the main
other types of residences. primary change that needs to and can already be implemented in
Behavioural parameters do also include light switching actions. the residential sector. An upgrade to Smart-LED sources may add
Users in the real environment do not necessarily turn on/off lights 10%. Smart-LEDs with a smart setting are recommended in combi-
immediately when the light level is above/under the limitations nation with sensor-based optimization techniques (occupancy and
mentioned in this study. Other studies indicated that there is no daylight harvesting).
significant relationship between external illuminance and lighting Comparable with non-residential environments with low occu-
use patterns, meaning that when the users turn on lights, it is less pancy, SLS for apartments (areas) with a low occupancy rate may
likely to switch them off according to daylight illuminance [60]. not be the most optimal solution. Smart lighting applications are
Therefore, the ELEC of conventional lighting systems could be much not directly connected with the number of people living in an
apartment, and it is more affected by a combination of space use,
floor plan and façade organization, occupancy patterns, and user
(light use) behaviour. Two-member household required a 20–40%
higher ELEC compared to single- or three-member households.
Daylighting has a high impact on the energy-saving potential of
Smart-LEDs with a smart setting. Therefore, including the availabil-
ity and use of daylight is of the utmost importance when imple-
menting SLS in residential environments. However, the savings
may be lower compared to non-residential environments. Consid-
eration of both delay time and standby energy consumption affects
the SLS efficiency, and neglect of standby and data transfer con-
sumption in smart systems can lead to unrealistically high savings
expectations. Standby consumption has a similar range for residen-
tial and non-residential environments, whereas the effect of sys-
tem turn-off delay time is much less widespread for the
residential sector.
As humans spend much of their time at home, a residential
Fig. 8. Calculated electric lighting energy consumption (ELEC) saving range for the
environment must be pleasant and comfortable, demanding
apartment compared with a literature-based saving range for the non-residential research that includes behavioural, comfort, and sustainability
sector per investigated energy-reducing strategies involving (smart) lighting. parameters. Research may focus on the optimal organization of
9
N. Hafezparast Moadab, T. Olsson, Géza Fischl et al. Energy & Buildings 244 (2021) 111009

spaces and design of daylight opening in relation to the required [11] P. Principi, R. Fioretti, A comparative life cycle assessment of luminaires for
general lighting for the office – compact fluorescent (CFL) vs Light Emitting
use and occupancy by residents to maximize the interaction
Diode (LED) – a case study, J. Clean. Prod. 83 (2014) 96–107.
between daylight, smart electric lighting systems, and user com- [12] A. Williams, B. Atkinson, K. Garbesi, E. Page, F. Rubinstein, Lighting controls in
fort. It can even be further expanded by the inclusion of shading commercial buildings, LEUKOS. 8 (3) (2012) 161–180.
and temperature controlling systems. All control systems can ben- [13] I. Chew, D. Karunatilaka, C.P. Tan, V. Kalavally, Smart lighting: the way
forward? Reviewing the past to shape the future, Energy Build. 149 (2017)
efit from the appropriate implementation of sensor systems 180–191.
detecting presence and activity of both users and lighting (radia- [14] S.-Y. Kwon, K.-M. Im, J.-H. Lim, LED context lighting system in residential
tion) while balancing (return of) investment, energy appropriate- areas, The Scientific World J. 2014 (2014) 16.
[15] G.R. Newsham, M.B.C. Aries, S. Mancini, G. Faye, Individual control of electric
ness, and potential user comfort and well-being opportunities. lighting in a daylit space, Light. Res. Technol. 40 (1) (2008) 25–41.
[16] N. Aste, M. Manfren, G. Marenzi, Building Automation and Control Systems
and performance optimization: a framework for analysis, Renew. Sustain.
Funding Energy Rev. 75 (2017) 313–330.
[17] N. van de Meugheuvel, A. Pandharipande, D. Caicedo, P.P.J. van den Hof,
Distributed lighting control with daylight and occupancy adaptation, Energy
This research did not receive any specific grant from funding
Build. 75 (2014) 321–329.
agencies in the public, commercial, or not-for-profit sectors. This [18] E.E. Dikel, G.R. Newsham, H. Xue, J.J. Valdés, Potential energy savings from
project was inspired by activities defined within the SMart Illumi- high-resolution sensor controls for LED lighting, Energy Build. 158 (2018) 43–53.
nation in Living Environment (SMILE) project (P45185-1), financed [19] A. Pandharipande, G.R. Newsham, Lighting controls: evolution and revolution,
Light. Res. Technol. 50 (1) (2018) 115–128.
by the Swedish Energy Agency and Smart Housing Småland. [20] T.M. Chung, J. Burnett, On the prediction of lighting energy savings achieved by
occupancy sensors, Energy Eng. 98 (4) (2001) 6–23.
[21] A. Pellegrino, V.R.M. Lo Verso, L. Blaso, A. Acquaviva, E. Patti, A. Osello, Lighting
CRediT authorship contribution statement control and monitoring for energy efficiency: a case study focused on the
interoperability of building management systems, IEEE Trans. Ind. Appl. 52 (3)
(2016) 2627–2637.
Nima Hafezparast Moadab: Conceptualization, Data curation,
[22] N. Gentile, Lighting Control Systems for Energy Saving and User Acceptance:
Formal analysis, Investigation, Methodology, Software, Writing - State-of-the-art and future directions, Lund University, Faculty of Engineering,
original draft, Writing - review & editing, Visualization. Thomas Lund, 2015.
Olsson: Writing - review & editing. Géza Fischl: Writing - review [23] C. Aghemo, L. Blaso, A. Pellegrino, Building automation and control systems: a
case study to evaluate the energy and environmental performances of a
& editing. Myriam Aries: Conceptualization, Data curation, Formal lighting control system in offices, Autom. Constr. 43 (2014) 10–22.
analysis, Methodology, Project administration, Writing - original [24] B. Roisin, M. Bodart, A. Deneyer, P. D’Herdt, Lighting energy savings in offices
draft, Writing - review & editing, Supervision. using different control systems and their real consumption, Energy Build. 40
(4) (2008) 514–523.
[25] X. Guo, D.K. Tiller, G.P. Henze, C.E. Waters, The performance of occupancy-
Declaration of Competing Interest based lighting control systems: a review, Light. Res. Technol. 42 (4) (2010)
415–431.
[26] Energy Saving Trust. The right light - selecting low energy lighting 2016, 11
The authors declare that they have no known competing finan- November 2020, p. 17. Available from: https://www.google.com/url?sa=t&rct=
cial interests or personal relationships that could have appeared j&q=&esrc=s&source=web&cd=1&ved=
2ahUKEwicl7ignqPlAhUD36QKHdLmANAQFjAAegQIARAC&url=https%3A%2F%
to influence the work reported in this paper.
2Fwww.energysavingtrust.org.uk%2Fsites%2Fdefault%2Ffiles%2Freports%
2FEST%2520Lighting%2520Guide%2520-%2520the%2520Right%2520Light.
Acknowledgements pdf&usg=AOvVaw0N13dWICjiHIcQdG4oLIZU.
[27] K. Qu, X. Chen, A. Ekambaram, Y. Cui, G. Gan, A. Økland, et al., A novel holistic
EPC related retrofit approach for residential apartment building renovation in
The authors would like to thank Mr. Moe Soheilian and Dr. Norway, Sustain. Cities Soc. 54 (2020) 101975.
Ulrika Wänström-Lindh for their input and feedback during educa- [28] B. Mills, J. Schleich, Household transitions to energy efficient lighting, Energy
Econ. 46 (2014) 151–160.
tional projects related to this study. We also would like to thank [29] S. Chraibi, P. Creemers, C. Rosenkötter, E.J. van Loenen, M.B.C. Aries, A.L.P.
the 12 Swedish families for their valuable input regarding occu- Rosemann, Dimming strategies for open office lighting: user experience and
pancy patterns. acceptance, Light. Res. Technol. 51 (4) (2019) 513–529.
[30] Y. Akashi, J. Neches, Potential recommendations for illuminance reductions by
load-shedding, Light. Res. Technol. 37 (2) (2005) 133–150.
References [31] G.R. Newsham, S. Mancini, The potential for demand-responsive lighting in
non-daylit offices, LEUKOS. 3 (2) (2006) 105–120.
[32] J. Brons, Sensor-controlled corridor lighting in a high-rise residential tower:
[1] J. De Boer, Daylighting of non-residential buildings, International Energy
occupancy patterns, dimming energy savings, and occupant acceptance,
Agency (IEA) (2019).
LEUKOS. 15 (4) (2019) 293–307.
[2] O.K. Larsen, R.L. Jensen, T. Antonsen, I. Strømberg, Estimation methodology for
[33] T. Hong, D. Yan, S. D’Oca, C.-F. Chen, Ten questions concerning occupant
the electricity consumption with daylight- and occupancy-controlled artificial
behavior in buildings: the big picture, Build. Environ. 114 (2017) 518–530.
lighting, Energy Procedia 122 (2017) 733–738.
[34] A. Wagner, W. O’Brien, B. Dong, Exploring Occupant Behavior in Buildings:
[3] M. Münch, A. Wirz-Justice, S.A. Brown, T. Kantermann, K. Martiny, O. Stefani,
Methods and Challenges, Springer International Publishing, 2017.
et al., The role of daylight for humans: gaps in current knowledge, Clocks and
[35] S. Carlucci, M. De Simone, S.K. Firth, M.B. Kjærgaard, R. Markovic, M.S.
Sleep. 2 (1) (2020) 61–85.
Rahaman, et al., Modeling occupant behavior in buildings, Build. Environ. 174
[4] M.-C. Dubois, M. Alenius, I. Bournas, N. Gentile, Daylighting and Lighting Under
(2020) 106768.
a Nordic Sky, Studentlitteratur AB, Lund, Sweden, 2019, p. 242.
[36] L.M. Candanedo, V. Feldheim, D. Deramaix, A methodology based on Hidden
[5] I. Bournas, M.-C. Dubois, Residential electric lighting use during daytime: a
Markov Models for occupancy detection and a case study in a low energy
field study in Swedish multi-dwelling buildings. Build. Environ., (2020)
residential building, Energy Build. 148 (2017) 327–341.
106977.
[37] D. Mitra, N. Steinmetz, Y. Chu, K.S. Cetin, Typical occupancy profiles and
[6] J.S. Alobaidy HAHM, R. Nordin, N.F. Abdullah, A review on ZigBee based WSNs:
behaviors in residential buildings in the United States, Energy Build. 210
concepts, infrastructure, applications, and challenges, Int. J. Elect. Electron.
(2020) 109713.
Eng. Telecommun. 9 (3) (2020) 10.
[38] V.M. Barthelmes, R. Li, R.K. Andersen, W. Bahnfleth, S.P. Corgnati, C. Rode,
[7] T.A. Nguyen, M. Aiello, Energy intelligent buildings based on user activity: A
Profiling occupant behaviour in Danish dwellings using time use survey data,
survey, Energy Build. 56 (2013) 244–257.
Energy Build. 177 (2018) 329–340.
[8] D. Cvetković, A. Nešović, I. Terzić, Impact of people’s behavior on the energy
[39] F. Yousefi, Y. Gholipour, W. Yan, A study of the impact of occupant behaviors
sustainability of the residential sector in emergency situations caused by
on energy performance of building envelopes using occupant data, Energy
COVID-19, Energy Build. 230 (2021) 110532.
Build. 148 (2017) 182–198.
[9] M.-C. Dubois, Å. Blomsterberg, Energy saving potential and strategies for
[40] A. Muroni, I. Gaetani, P.-J. Hoes, J.L.M. Hensen, Occupant behavior in identical
electric lighting in future North European, low energy office buildings: a
residential buildings: a case study for occupancy profiles extraction and
literature review, Energy Build. 43 (10) (2011) 2572–2582.
application to building performance simulation, Build. Simul. 12 (6) (2019)
[10] M. Bonomolo, M. Beccali, V. Lo Brano, G. Zizzo, A set of indices to assess the
1047–1061.
real performance of daylight-linked control systems, Energy Build. 149 (2017)
235–245.

10
N. Hafezparast Moadab, T. Olsson, Géza Fischl et al. Energy & Buildings 244 (2021) 111009

[41] E. Delzendeh, S. Wu, A. Lee, Y. Zhou, The impact of occupants’ behaviours on [52] A. Kumar, P. Kar, R. Warrier, A. Kajale, S.K. Panda, Implementation of smart LED
building energy analysis: a research review, Renew. Sustain. Energy Rev. 80 lighting and efficient data management system for buildings, Energy Procedia
(2017) 1061–1071. 143 (2017) 173–178.
[42] S. Wolf, D. Calì, M.J. Alonso, R. Li, R.K. Andersen, J. Krogstie, et al., Room-level [53] C.E. Ochoa, M.B.C. Aries, E.J. van Loenen, J.L.M. Hensen, Considerations on
occupancy simulation model for private households, J. Phys. Conf. Ser. 1343 design optimization criteria for windows providing low energy consumption
(2019) 012126. and high visual comfort, Appl. Energy 95 (2012) 238–245.
[43] C.E. Ochoa, M.B.C. Aries, J.L.M. Hensen, State of the art in lighting simulation [54] E. Ghisi, J.A. Tinker, An Ideal Window Area concept for energy efficient
for building science: a literature review, J. Build. Perform. Simul. 5 (4) (2012) integration of daylight and artificial light in buildings, Build. Environ. 40 (1)
209–233. (2005) 51–61.
[44] Statistics Sweden. Housing, construction and building (2020) [09-11-2020]. [55] G. Nair, K. Mahapatra, L. Gustavsson, Implementation of energy-efficient
Available from: https://www.scb.se/en/finding-statistics/statistics-by-subject- windows in Swedish single-family houses, Appl. Energy 89 (1) (2012) 329–
area/housing-construction-and-building/. 338.
[45] D. Johansson, H. Bagge, L. Lindstrii, Measurements of occupancy levels in [56] M.A.U. Haq, M.Y. Hassan, H. Abdullah, H.A. Rahman, M.P. Abdullah, F. Hussin,
multi-family dwellings—application to demand controlled ventilation, et al., A review on lighting control technologies in commercial buildings, their
International builidng performance simulation association, Sydney, 2011. performance and affecting factors, Renew. Sustain. Energy Rev. 33 (2014) 268–
[46] D.H.W. Li, T.N.T. Lam, S.L. Wong, Lighting and energy performance for an office 279.
using high frequency dimming controls, Energy Convers. Manage. 47 (9) [57] A. Ioannou, L.C.M. Itard, Energy performance and comfort in residential
(2006) 1133–1145. buildings: Sensitivity for building parameters and occupancy, Energy Build. 92
[47] N. Shishegar, M. Boubekri, Quantifying electrical energy savings in offices (2015) 216–233.
through installing daylight responsive control systems in hot climates, Energy [58] T. Hong, S.C. Taylor-Lange, S. D’Oca, D. Yan, S.P. Corgnati, Advances in research
Build. 153 (2017) 87–98. and applications of energy-related occupant behavior in buildings, Energy
[48] E. Neufert, P. Neufert, Architects’ Data, Wiley, 2012. Build. 116 (2016) 694–702.
[49] SS-EN15193-1. Energy performance of buildings - Energy requirements for [59] D. Wolf SC, M.J. Alonso, R. Li, R.K. Andersen, J. Krogstie, H. Madsen, Room-level
lighting - Part 1: Specifications, Module M9. SS-EN 15193-1:2017: Swedish occupancy simulation model for private households. J. Phys.: Conf. Ser.
Standards Insitute; (2017), p. 112. (Online) - CISBAT 2019. 1343 (012126) (2019) 8.
[50] Swedish Energy Agency. Smarta LED-lampor (2016) [updated October 2016. [60] G.Y. Yun, H. Kim, J.T. Kim, Effects of occupancy and lighting use patterns on
Available from: https://www.energimyndigheten.se/tester/tester-a-o/smarta- lighting energy consumption, Energy Build. 46 (2012) 152–158.
led-lampor/. [61] J.Y. Park, T. Dougherty, H. Fritz, Z. Nagy, LightLearn: an adaptive and occupant
[51] E.E. Dikel, Y.E. Li, M. Vuotari, S. Mancini, Evaluating the standby power centered controller for lighting based on reinforcement learning, Build.
consumption of smart LED bulbs, Energy Build. 186 (2019) 71–79. Environ. 147 (2019) 397–414.

11