Content uploaded by Thomas Bjørner
Author content
All content in this area was uploaded by Thomas Bjørner on Feb 23, 2021
Content may be subject to copyright.
Original Paper
I
Indoor
and
and B
uilt
uilt
Environment
An experiment of double dynamic
lighting in an office responding
to sky and daylight: Perceived
effects on comfort, atmosphere
and work engagement
Ellen Kathrine Hansen , Thomas Bjørner ,
Emmanouil Xylakis and Mihkel Pajuste
Abstract
The experiment was targeted to develop design strategies and methods by testing the complex inter-
play between the dynamics of daylight and electrical lighting in an office. The double dynamic lighting
design concept is based on the idea of adding task lighting, with a directionality referring to the daylight
inflow and a variation on direct/diffuse lighting and respective changes in colour temperature respond
to sky conditions and daylight levels. The experiment was conducted in an office space at Aalborg
University in Copenhagen from September to December 2019. Four participants moved in and
worked in the office with four-week periods of respective standard static lighting as a baseline, and
dynamic lighting. In a parallel mixed method approach with interviews and questionnaires, the dynamic
lighting was compared to the baseline and to a control group. The results indicate that the dynamic
lighting periods had a positive effect on visual comfort, perceived atmosphere and work engagement.
The studies helped to develop the definition of five dynamic light settings. Seasonal changes, time of
day, dynamic sunscreens and individual needs for task lighting can be implemented in future field
experiments as additional dynamic parameters to meet individual needs and circadian potentials for
double dynamic light.
Keywords
Dynamic lighting, Responsive lighting, Office lighting, Lighting design, Double dynamic lighting
Accepted: 10 January 2021
Introduction
Human perception and vision have evolved in response
to the natural variations of daylight patterns created by
the reflected diffuse light from changing sky conditions
and the direct light from the sun, depending on the
altitude and orientation.
1–3
This combination of diffuse
and direct light, with respective spectral distributions
and directionality, creates the perceived qualities of
dynamic daylight that we appreciate.
4
This study inves-
tigated how new sensor and lighting technologies,
responding to the sky and daylight level, can meet the
human need for variation and appreciation of a more
natural atmosphere in an office environment. Our
research question was the following: Can dynamic elec-
trical lighting, complementing and responding to the
natural dynamics and qualities of daylight, improve
Department of Architecture, Design and Media Technology,
Aalborg University, Copenhagen, Denmark
Corresponding author:
Ellen Kathrine Hansen, Department of Architecture, Design and
Media Technology, Aalborg Universitet, A. C. Meyers Vaenge
15 Copenhagen S 2460, Denmark.
Email: ekh@create.aau.dk
Indoor and Built Environment
0(0) 1–20
!The Author(s) 2021
Article reuse guidelines:
sagepub.com/journals-
permissions
DOI: 10.1177/1420326X21991198
journals.sagepub.com/home/ibe
the perceived atmosphere and visual comfort in an
office environment and thereby improve work
engagement?
Mental health disorders, such as depression, stress
and anxiety, are some of the main challenges in our
work environments,
5,6
with an estimated cost to the
global economy of 1 trillion US dollars per year in
lost productivity.
6
Research on the psychological
effects of light, including window access, has revealed
effects of visual qualities on objects and the distribu-
tion of light in a space can impact work satisfaction
and engagement.
7–14
Previous studies reported that
office workers who perceive lighting positively also rec-
ognize visual comfort and perceive the atmosphere as
more attractive.
14,15
The positive lighting also led to
better mood, higher work satisfaction, and greater
overall engagement in their work.
14,15
The development of lighting technology, sensors and
control systems enables the implementation of dynamic
lighting and has increased the awareness of positive
effects of dynamic light to support human health and
well-being. A previous literature study demonstrated
that there have been a lot of studies reflecting interest
in the non-image-forming effects of light on health and
well-being.
16
Studies of these physiological processes
often focus on the effect of a preset protocol, in
which time settings adjust only the light level and spec-
tral distribution to synchronize with biological
rhythms.
17,18
These studies, often carried out in labs,
indicated that a high light level and cooler colour tem-
perature can increase alertness and work performance.
However, other studies found no significant difference
in mental and physical health, but a higher satisfaction
with dynamic lighting compared to static lighting in an
office environment.
19
Beute and de Kort
20
investigated
photographic outdoor scenes manipulated across nat-
uralness, brightness and weather type (sunny and over-
cast) and found explicit preferences for natural, bright
and sunny photographic scenes. Bakker et al.
21
studied
luminance distribution preference in relation to the
time of day and subjective alertness, and reported
that participants preferred varying luminance distribu-
tions and they did not always prefer the same lighting,
defined according to non-visual effects on increasing
alertness and performance.
21
Studies of dynamic light-
ing have demonstrated a preference for dynamic light-
ing responding to unpredictable natural variations in
weather and light levels.
8,22,23
These studies investigat-
ed the relationship of the distribution of the light in a
space and a combination of daylight inflow from a
window and electrical lighting in real space.
Stokkermans et al.
11
investigated the perceived effect
of diffuse daylight in a space without a view. They
found very little effect on the perceived atmosphere
from this diffuse daylight and argued that this may
be due to the lack of a direct sun component and a
view. They defined this as an interesting topic to inves-
tigate in future research. Fleischer
23
pointed out that
preferences for different electrical lighting in an office
depend on weather-type, intensity and colours of the
daylight inflow.
Houser et al.
24
stated that the core of good human-
centric lighting is the outcome of good design, stressing
that light is still for vision, visibility and visual comfort.
They encouraged integrative lighting, comparing the
visual and non-visual factors in lighting design. These
factors were defined by Houser et al.
24
as follows:
temporal pattern, light spectrum, light level and spatial
pattern. In particular, the temporal and spatial patterns
have potential for understanding the complex interplay
between the dynamics of daylight and electrical light-
ing. This experiment was aimed at integrating all four
factors within the visual effect with reference to the
daylight dynamics, and thereby developing a lighting
design concept to inspire future implementation of the
non-visual effects in an integrative design approach.
The qualities of natural light have previously been
explored,
16
to provide a better understanding of how
these qualities can be translated into indoor lighting
criteria for dynamic electrical lighting, complementing
the daylight inflow in a space.
16
The qualities of com-
bined diffuse and direct light can be defined through
the light’s interaction with three-dimensional objects.
In lighting design and architecture, this has been inves-
tigated as the qualities of light modelling, which refers
to visual characteristics, such as contour, shape and
detail distinction, of objects and space.
25,26
In a spatial
context, especially with side-lit windows, the inflow of
daylight can create a light modelling effect, which is
also characterized as the flow of light,
7,27
creating a
light hierarchy and light zones. In lighting research,
the combination of direct and diffuse lighting in an
office environment has been investigated using uplights
and downlights. These studies found that a combina-
tion of direct (downlight) and diffuse (uplight) lighting
is preferred, compared to only diffuse or only direct
illumination.
12,15,23,28,29
The dynamic variation in colour appearance of day-
light also refers to the components of daylight compo-
nents, the sky and the sun. The daylight reflected from
a clear blue sky is perceived as cool white in appear-
ance, whereas an overcast sky reflects more neutral
white daylight.
1,7
During transition hours, the light
has a warmer colour appearance, as the sun rays are
more scattered, due to the low sun elevation angle.
These variations can be perceived as having a warm
colour temperature between 2000 and 3000 Kelvin
(K), neutral between 3000 and 5000 K, and cool as
5000 K and above.
7
Stage lighting designer Stanley
McCandless
30
incorporated the ‘sun and sky lighting
2Indoor and Built Environment 0(0)
effect’ into theater, a combination of direct warm and
diffuse cold lighting as a combination of colour appear-
ance and distribution. This understanding of the per-
ceived spatial character of dynamic daylight as a flow
of light, referring to the ratio of direct sunlight and
diffuse skylight to their respective CCTs (correlated
colour temperatures), was used as a reference for the
double dynamic lighting concept developed for this
experiment.
31
The novelty in this proposed design concept is that
the daylight and electrical lighting are integrated as one
lighting component, by letting the variation of diffuse/
direct ratio and the CCT of dynamic electrical lighting
complement the dynamics in the sky and daylight level.
This integrated lighting concept creates light zones in
the workplace through directional electrical light sour-
ces. This direct light is not in the form of downlight
(with a 90tilt angle), as is usually used for a combi-
nation of direct/diffuse lighting, but a spotlight with a
directionality similar to that of the daylight inflow
from the window. It thereby complements the light
modelling qualities and flow of daylight in the office
space. The direct task light has warm and neutral
colour temperatures with reference to the direct sun-
light, whereas the diffuse ambient light has a neutral or
cool colour appearance referring to the skylight. Prior
to this experiment, two pilot studies were carried out
testing ratio of flow of light and diffuse light and
respective warm, neutral and cool colours. Based on
these findings, seven light settings were developed to
respond to variations of sky type and daylight intensi-
ty. The double dynamic lighting environment created
by the integrated daylighting and seven responsive light
settings was tested in this study. This experiment was
targeted to develop integrated lighting design strategies
and methods by testing perceptions of the complex
interplay of these dynamic lighting parameters on
office workers.
Pilot studies
The location for the pilot studies was the same as for
this experiment: the office environment at Aalborg
University. The aim of the pilot studies was to evaluate
the perceived qualities of dynamic lighting through var-
iations of direct/diffuse ratios and colour temperatures,
while still meeting the standards for illumination, pro-
viding minimum 500 lx on the working area and 300 lx
in the space.
32
The traditional diffuse ceiling panels
were supplemented with ceiling-mounted spotlights
with a beam angle of 36and tilt angle of 32, provid-
ing a light flow complementing the directionality of the
daylight inflow from side windows in the given space.
All light sources were dimmable, tunable white LED
fixtures. The ambient light from the ceiling panels
provided up to 300 lx, whereas the spotlights provided
additional task lighting on the work plane of 200 lx,
resulting in a total of 500 lx.
In the first pilot study, the ratio of direct/diffuse
lighting was evaluated by 30 participants, testing the
following ratios: 0/100, 15/85, 30/70 and 45/55. The
findings indicated that direct light, provided by the
spotlight, should be more than 15% to create the qual-
ity of visual appearance of light modelling and less than
45% to avoid uncomfortable contrasts for visual tasks
and glare.
In the second pilot study, a direct/diffuse light ratio
of 30/70 was evaluated with 15 participants. Different
combinations of colour temperatures such as cold/cold,
cold/warm, neutral/neutral, warm/cold and warm/
warm were tested under two contrasting daylight con-
ditions: overcast and clear sky. The outcome suggested
that the combination of direct warm task lighting and
diffuse cool ambient lighting was perceived as the most
natural. In relation to the different daylight situations,
the outcome suggested a warmer light when the sky was
overcast and for cooler ambient light (looked more
natural) when the sky was clear, and the daylight
inflow was cooler. A full description of the pilot tests
can be found in the designated paper.
31
The outcome of the two pilot studies suggested the
following design criteria for the experimental study:
1. Direct flow of light on task area from ceiling-
mounted spotlights (with approximately 36beam
angle and approximately 32tilt angle as the
inflow of daylight) combined with diffuse light
from ceiling panels, complementing the inflow of
daylight and light modelling qualities.
2. Direct/diffuse ratio between 20/80 and 40/60 to
create visual appearance of light modelling effect
and to avoid uncomfortable contrasts.
3. Neutral ambient lighting applied (less than 5000 K)
with overcast sky condition to compensate for the
cool overcast atmosphere. Cooler ambient lighting
applied (more than 5000 K) with clear sky condition
to complement the cool, natural luminous
atmosphere.
4. Direct warm light combined with diffuse cool or
neutral CCT to create a natural luminous condition.
5. Light levels meet the standard of minimum of 500 lx
on the task area and minimum of 300 lx in the imme-
diate surroundings.
Light settings
Seven light settings for the experimental study were
designed to correspond to two parameters: first, the
daylight inflow, measured as the daylight intensity on
the work plane by a sensor and second, the sky
Hansen et al. 3
condition, defined as overcast or clear sky/partly
cloudy, determined by sky scanner data. The clear
sky/partly cloudy sky condition is referred to as clear
sky condition in this paper. The seven light settings
were defined, with four possible settings for clear and
overcast sky condition, created in steps of daylight
inflow of 50, 200, 350 and 500 lx. 750 lx was defined
to mark the amount of light for which electrical light-
ing is not needed and would be automatically switched
off. The total amount of light on the task area varies
based on the four steps, between 500 lx and 750 lx. This
affects the direct lighting component with a variation
between 26% and 40% of the total luminous condition
when electrical lighting is turned on. The colour tem-
peratures of the seven light settings were based on pre-
vious findings from the pilot studies and re-evaluated
by six lighting professionals in four study sessions in
the laboratory under the two sky conditions, which
took place at the end of August 2019 (see the defined
light settings in Figure 1(a) and (b)).
The 1OC light setting was designed identically for
both sky conditions, because this is a luminous condi-
tion, where there is only up to 50 lx on a working plane.
This setting was designed for working hours, after
sunset, and for transition hours in the morning and
evening. The direct light component was defined with
a warm colour temperature (2700 K) and the diffuse
light component with a neutral temperature (4000 K).
The light settings for the clear sky condition (2 C
and 3 C) were designed with cool diffuse, ambient light-
ing (6000 K), referring to the cool skylight on a sunny
day, and with a warm direct lighting component, a
spotlight (3000 K). The 4 C light setting was designed
with only a direct lighting component, a spotlight illu-
minating the task area with cool lighting (5000 K). The
aim of this light setting was to complement the cool
daylight inflow. This design is also supported by the
findings from Fleisher,
23
who suggests that maximum
value for pleasure can be received by the combination
of warm white and daylight colour, with a large indi-
rect component.
The light settings for overcast sky conditions (2 O
and 3 O) were designed with a neutral ambient lighting
(4000 K), supported by a warm direct spotlight
(3000 K). According to findings from previous pilot
studies, neutral/warm lighting combinations were pre-
ferred under overcast sky conditions. The 4 O light set-
ting was designed with a direct lighting component
only, providing task lighting with a neutral colour tem-
perature (4000 K).
Experimental setup
The location for the experimental study was an estab-
lished office on the Copenhagen campus of Aalborg
University in Denmark. The space dimensions were
4.4 by 5 m and height of 2.6 m. The walls were painted
white. The office was equipped with four tables, chairs
and some cabinets. The participants were asked to dec-
orate and bring their computers and work equipment
to establish their office space as they preferred it (see
Figures 2 and 3(a)). The walls had a measured
Figure 1. (a) (left) Illustration of the experimental set-up with daylight opening of side windows (uncovered), creating a
natural flow of light; electrical lighting components diffuse ambient ceiling panels; and directional task lighting, with similar
directionality as the daylight inflow. (b) (right) Seven light settings shown in relation to daylight character (‘O’ overcast or ‘C’
clear sky/partly cloudy) and intake with corresponding CCTs and intensities of electrical lighting components. All settings meet
minimum 500 lx on task area.
4Indoor and Built Environment 0(0)
reflectance of 77% and tables of 18% measured with a
Hagner EC1 Luxmeter. The two southwest-facing win-
dows were equipped with manually controllable electric
blinds. The glass area of the windows was 0.8 by 1.07
m, equal to 7.7% of the floor area. The office was
located on the third floor, and there were no obstruct-
ing buildings affecting the daylight inflow. The view
from the window can be seen in Figure 2. The building
was next to a six-lane road with a heavy traffic. The air
temperature was controlled by a central heat and air
condition system.
Four dimmable and tunable ceiling panels (2700–
6500 K, Fagerhult Multilume Flat Delta) and four
spotlights (2700–6500 K, Zumtobel’s Arcos 3) with a
Figure 2. The experimental study space with participants.
Figure 3. Flowchart showing scheduling process of input data readings to trigger lighting settings.
Hansen et al. 5
36beam angle and 32tilt angle were installed, see
Figure 2.
Daylight and sky sensors
An LM-TLM daylight sensor from Zumtobel
33
was
utilized to define the outdoor sky condition. The LM-
TLM contains eight sensors, four measuring horizontal
and four vertical illuminance, placed relating to the
four cardinal directions. The specific sensor placement
informs the amount of illuminance at each of the four
cardinal directions at any given moment horizontally
and vertically. The sky scanner was installed on the
roof of Aalborg University Copenhagen, with no
obstructing elements. A detailed description of a
study designed to determine how the illuminance
values were received by the sky scanner change regard-
ing the sky condition is given in Appendix 1.
To measure the indoor illuminance, the ESP8266
microcontroller
34
and TCS34725 light sensor
35
were
used to transmit the light data over Wi-Fi. The light
sensor provided a digital return of red, green, blue
(RGB) values that were converted to illuminance
(lux). The accuracy of the light sensor was evaluated
by comparing the measured sample value with meas-
ures from Hagner EC1 Luxmeter, under both artificial
and skylight.
The lighting control system ran on DALI. The light-
ing management system (LMS) that was responsible
for triggering light settings ran on the central control
unit (CCD) via restful API. To establish communica-
tion with the CCD, and thus exchange light data,
HTTP (Hypertext Transfer Protocol) methods had to
be established. The utilization of the REQUESTS
Python library
36
made it easy to formulate HTTP
requests as Python commands. The JSON response
content, in case of successful requests, includes all the
connected DALI devices states, from switches (True/
False regarding their state) to luminaires (intensity: 0–
100 and colour temperature for tunable white: 2700–
6500 K). These JSON data were stored and accessed via
the use of the Pandas Python library.
37–39
The Raspberry Pi was utilized as a low-cost solution
server for scheduling the execution of the Python
scripts. Seven light settings with their corresponding
illuminances and colour temperatures were coded to
be triggered at different indoor illuminance levels and
sky conditions. The scripts were set to execute with a 5-
min interval. The flowchart illustrates the sequence of
execution of the scripts. Figure 3 presents the flowchart
of the scheduling process initiated by receiving the
latest reading from the outdoor daylight scanner trans-
lated into sky condition. Once this was achieved, the
indoor horizontal illuminance was measured, and the
threshold of the latest reading was determined. The
lighting was chosen to change over a duration of
change of 10 s, without further investigations of this
variable. This whole process was set to repeat every
fifth minute.
Participants
This experimental study was performed by applying an
experimental design that occurred in a natural office
setting with four male participants, defined as the
experimental group (with mean age 25.5 years):
employees of a three-man start-up company and one
master’s student, working with media production. The
participants were recruited through an open call distri-
bution via the campus intranet. The criteria for enroll-
ment were having a desk as a primary work-place, and
the ability to work as they normally did without any
extraordinary period of absence from the office. The
participants were not paid, but they were given two
dinners during the test period, as well as a small gift
as appreciation for their participation. Prior to partic-
ipation, all participants completed informed consent,
including the right to withdraw at any time, the right
to refuse to answer the questions, and a guarantee of
participants’ anonymity. All test participants were pro-
vided with anonymized ID numbers. The participants’
personal information was kept encrypted in a database
that was separated from the other information used in
the study. We applied special ethical considerations for
this study. Legal access and permission to run the test
were provided by the university.
All participants reported on the questionnaires that
they had no visual impairments other than myopia or
hyperopia, which were corrected by wearing contact
lenses or glasses (three out of four participants), and
none of the participants reported any colour vision
deficiencies. One participant mentioned being light sen-
sitive, and two participants mentioned sometimes get-
ting eye strain when working for long periods of time in
front of a screen. The work tasks of the participants
were mainly software development, graphic work,
video production, 3 D renderings and marketing. The
participants in the game company enrolled in this
study, did team-based work, with different tasks, and
due to that, there may have been individual differences
in the effect of the dynamic lightning. Within the con-
trol group, the tasks also included group work, and
work on computers; instead of games, this group
worked on a semester project within the field service
system. However, all participants self-reported that
they spent more than 70% of their time in front of
their computer.
Participants were aware that they were participating
in a study related to the potential effects of dynamic
6Indoor and Built Environment 0(0)
lighting, but were unaware of the exact light settings,
and in which period there was static or dynamic light.
A control group was established and consisted of
four participants, like the experimental group. The con-
trol group were students from the Service System
Design master’s program at Aalborg University. They
were working in a similar office space on the same floor
as the experimental participants, with the same window
orientation. The electrical lighting condition was iden-
tical to the experimental group’s static lighting: ceiling
panels with the same light levels and CCT. The control
group consisted of three female participants and one
male participant (mean age 21.5 years). Two out of
three used glasses or lenses to work. None of them
had any colour deficiencies. The primary working
tasks of the participants were reading articles, research-
ing, brainstorming, writing and visualizations. The
data from the control group were collected for
three weeks in October 2019 and three weeks in
December during the same approximate working
hours as the data from the experimental group. The
data used for analysis were from weeks 49 and 50,
due to alignment of the experimental period.
The experimental period
The experimental period was four months in total, from
September to December 2019, covering the Autumn
and Winter season. The data were collected for
16 weeks in total, including eight weeks of static and
eight weeks of dynamic lighting periods. Each month
had a period of static and double dynamic lighting. The
monthly time spans were established due to the large
seasonal changes in daylight conditions in the northern
latitude of 56, including change in both daylight hours
and sunlight hours. All data were collected from
Monday to Friday, from 10 am to 6 pm, which was
defined according to participants’ show-up time and
approximate working hours.
Subjective assessments
Methods to evaluate the subjective response to lighting
in an office space are complex and require several cri-
teria. Our criteria for the evaluation were inspired by
previous effect evaluation of light on impression,
behaviour, perceived atmosphere and work engage-
ment,
8,13,14,40,41
as well as engagement evaluations.
42,43
The questionnaires and interviews implemented formed
a combination of both qualitative and quantitative
methods to produce transdisciplinary insights within
lighting research methods exploring how different
data sources can increase the validity and reliability.
Thereby this study was based on a convergent parallel
mixed method,
53
meaning that qualitative interview
data and quantitative questionnaire data were collected
and analysed separately but at the same time (in par-
allel). In this experiment, three focus areas were defined
to evaluate how the light and work environment were
perceived and affected the users:
1. Visual comfort, validating the way the distribution
and intensity of the light affect the perception of the
light, the light modeling of objects, and the task per-
formance at a workplace.
2. Perceived atmosphere, defined as the subjective
experience of being in the space, also considering
the peripheral light and spaciousness in the appear-
ance of the room and the relation to the outside.
3. Work engagement, the long-term affective effect of
the light, relating to perceived well-being and
motivation.
Questionnaires
To evaluate the visual comfort, perceived atmosphere
and work engagement, data were collected bi-weekly in
the form of electronic questionnaires during the first
six weeks of the experiment. The frequency of question-
naires was defined to avoid disturbing participants too
much and to ensure a high rate of answers and accu-
racy, so participants would be motivated to take time
and to be as precise as possible. However, it was
changed to weekly distribution for the last 10 weeks
to increase the validity and accuracy. The question-
naires were made in SurveyExact
44
and distributed on
Fridays, to be filled out at the end of the working week.
The questions included general questions gathering
background information (e.g. attendance and health)
and specific questions covering the following themes:
visual comfort, perceived atmosphere and work
engagement. The questions can be found in Table 3.
In the visual comfort theme, three questions were
designated for the following purposes: first, to evaluate
how frequently the lighting has been good for carrying
out work tasks, and second, to gain knowledge of
whether visual discomfort has occurred and been
caused by electrical lighting or daylight.
In the perceived atmosphere section, four questions
were designated to evaluate the following aspects: pos-
itive impact on the work atmosphere; naturalness of
combination of daylight and electrical lighting; fre-
quency of electrical lighting being stimulating for
work; and frequency of feeling comfortable being in
the room.
To gain insight on the work engagement evaluated
by participants, six questions were designated. These
addressed the following: frequency of good work per-
formance; feeling motivated; producing innovative
ideas/novelty; feeling concentrated at work; having a
Hansen et al. 7
good workflow; and willingness to take risks in work
tasks.
To analyse the participant data from the Likert
scales numerically, the ordinal data were transformed
into arithmetic data. The process for computing the
mean for the ordinal scale was as follows. First, each
of the five Likert answers was given a weight from 1 to
5 (never ¼1, rarely ¼2, occasionally ¼3,
frequently ¼4, very frequently ¼5). Then, for each of
the questions, the frequency of the answer chosen was
calculated. Lastly, to compute the mean (M), the sum
of each of the weights was multiplied by its frequency
and divided by the total amount of responses for the
present question. Standard deviation (SD) was also
calculated.
Two weeks (week 49 and 50) were analysed through
a Mann–Whitney U test during the experiment, when
the frequency of distributed questionnaires was weekly
for both the control and experimental group. The
Mann–Whitney U test, a nonparametric rank order
test, was used to analyse differences in medians for
these two weeks for the experimental group (with
double dynamic lightning) and the control group
(static lightning), respectively. The hypotheses for the
test were as follows:
H0: There is no difference between the ranks of the
test and control group.
Halt: There is a difference between the ranks of the
test and control group.
Interviews
During the study, three sessions of individual in-depth
interviews
53
with four test participants were conducted.
All interviews took place after four continuous weeks
of the same lighting condition (see Table 1): two times
with dynamic lighting conditions and once with static
lighting conditions. The interviews varied in duration
between 25 and 45 min. The interviews consisted of
sections of questions divided under themes: general
lighting experience, visual comfort, work engagement
and perceived atmosphere, which was elaborated by
participants with the usage of reaction cards,
45
also
known as card sorting.
46
The reaction card method
was modified from its original, using only 30 adjectives
divided into two categories: visual appearance and per-
ceived atmosphere (see Table 2 below). All interviews
were recorded and transcribed using Otter.ai.
47
The
AI-powered live transcriptions were later corrected in
Nvivo12,
48
where the transcripts were read and the
nodes’ visual comfort, perceived atmosphere and
work engagement were manually detected. The data
were analysed through content analysis with defined
nodes, negative or positive. In addition, a group inter-
view following the friendship-pair method
46
was con-
ducted in the end of the last static period. The group
interview was conducted after a two weeks period and
had also the purpose to provide debriefing and final
remarks from the participants.
Results
The results of the analysis of the questionnaires and
interviews are presented within each of the three-
Table 1. Overview of fourth-month experimental field study period, showing weeks with dynamic and static lighting, weeks
when questionnaires were distributed, and when interviews were conducted.
Light Settings 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52
Dynamic x X x x xxxx
Static x x xxxxx xx
Questionnaires x x xxxxx xxxxx
Interviews x x x
Control group xx
Note: The control group period is only marked with the two weeks when the data were used for analysis.
Table 2. Overview of reaction cards, in categories of visual
appearance and perceived atmosphere.
Reaction cards
In relation to
visual appearance:
In relation to
perceived atmosphere:
Comfortable Cozy
Sufficient Motivating
Task-focused Personal
Pleasant Intimate
Natural Formal
Contrasting Stimulating
Dim Relaxed
Bright Lively
Glary Detached
Uncomfortable Boring
Insufficient Lifeless
Disturbing Depressing
Unpleasant Clinical
Unnatural Demotivating
Tense Dull
8Indoor and Built Environment 0(0)
focus area: visual comfort, perceived atmosphere and
work engagement.
Visual comfort
The weekly questionnaires (distributed 12 times in
total, see Table 1) comparing static and dynamic light-
ing during the four-month period were within mean
values of 3.25 and 4 (see Table 3). This demonstrates
that the lighting was satisfactory occasionally and fre-
quently throughout the experimental period and regard-
less of light settings (see Q 1.1 in Table 3). Discomfort
caused by the electrical lighting (Q 1.2) was also found
never and rarely by the occupants, with numeric values
of 1.25 and 2, respectively. Discomfort caused by the
daylight (Q 1.3) was found to produce values of 1.25
and 2.75. The most negative value was registered in
December, which could be due to the low sun angle
and a southwest-oriented daylight opening or due to
the exposure of the 1OC light setting as a constant,
more than 80% of the time. There were no significant
differences or problems detected in the analysis of the
questionnaires (see Table 3).
In the questionnaire, three questions were desig-
nated to evaluate visual comfort (see Table 5). All of
them returned differences between the control and
experimental group with a significance of p<0.05. In
Q 1.1, which asked about the suitability of the lighting
to carry out work tasks, the test group ranked it with a
higher mean value (M ¼11.44) than the control group
(M ¼5.56). They also noted less discomfort from
artificial lighting (Q 1.2) compared to the control
group (experimental group M ¼5.7, control group
M¼11.25). In Q 1.3, where the groups noted their dis-
comfort from daylight, the test group evaluated it as
significantly higher (M ¼11.5) compared to the control
group (M ¼5.5). For all three questions, the null
hypothesis, that there is no difference between the
ranks of the experimental and control group, could be
rejected. Because the electrical lighting was less dynamic
during the comparison period in December than in pre-
vious autumn months, the dynamism of lighting could
not be evaluated in this section. The design concept
components of diffuse and direct lighting still reveal pos-
itive results over static lighting for visual comfort.
From the interviews and the content analysis, it can be
deducted that the visual comfort during dynamic lighting
ranked with 19 out of 23 positive statements (nodes) in
the first interview, when the interviewees had been
exposed to dynamic lighting over four continuous
weeks. In the second period of dynamic lighting, visual
comfort ranked 12 out of 21 positive nodes. One partic-
ipant stated the following during the first interview:
The lighting has been sufficient, no matter how the
lighting has changed. The lighting has been very pleas-
ant. It hasn’t annoyed me, it hasn’t distracted me, it
hasn’t cost me headaches or anything like that. [ID3,
1st int., Dynamic]
The participants were not aware of on which weeks the
static and dynamic lighting settings were executed. One
of the sub-questions for the researchers, if the partic-
ipants were able to notice the dynamic changes, was
confirmed by the following statement:
The electrical lighting is not too bright. And when it
gets lower... I’ve seen the adjusting a couple of times.
Not noticeably, but sometimes you can see that it
changes a bit, but the changes from the lowest to the
highest have been within a range that still feels com-
fortable and natural, even though there’s a noticeable
range. [ID3, 1st int., Dynamic]
The increase of negative nodes under visual comfort
during the second interviews in December could be
due to two factors: the low sun angle and problems
with the accuracy of the lighting system adaption.
For instance, one participant said the following in
December:
And then we also had like extreme sunlight for a very
brief moment, and then all the lights turned off and
then they didn’t turn on again for a long period of
time. And then I would have liked to turn on the
lights again. [ID4, 3rd int., Dynamic]
The visual comfort was also evaluated after four con-
tinuous weeks of static lighting for comparison to
dynamic lighting. There were 15 negative nodes out
of 23 detected, demonstrating numerous unsatisfactory
or negative effects of this period (see Table 4). One of
the participants stated the following: ‘I feel like the
visual comfort could be better, if it adjusts dynamically,
somehow’ [ID2, 2nd int., Static]. Another participant
said, ‘So it didn’t feel like it did anything smart. And
sometimes, we actually had the urge to turn them off’
[ID1, 2nd int., Static]. This indicates that the dynamic
lighting, which was previously applied, was favoured
by the participants. The interviews also revealed that
the standard light intensity level could be problematic,
especially during evening hours. For instance, one par-
ticipant stated:
I had been sitting in the evening. It was very bright. It
hurt my eyes. So during the day, the lighting was fine,
but when it got night and it was very dark outside. It
was just very intense. [ID4, 2nd int., Static]
Hansen et al. 9
Table 3. Questions, divided into three categories: visual comfort, perceived atmosphere and work engagement.
Month September October November December
Static/Dynamic Static Dynamic Static Dynamic Static Dynamic Static Dynamic Control
Week nr. 36–37 38–39 42–43 40–41 45–46 47–48 51–52 49–50 49–50
Mean/Standard Deviation M SD M SD M SD M SD M SD M SD M SD M SD M SD
Visual Comfort
1.1 During this week, the
lightning has been good
for carrying out my work
tasks.
4.00 0.00 3.25 0.83 3.00 0.00 4.00 0.71 3.75 0.43 3.75 0.43 3.50 0.35 3.63 0.48 2.50 0.71
1.2 During this week, I have
experienced discomfort
caused by the electrical
lighting.
1.50 0.50 1.25 0.43 2.00 0.71 2.00 1.00 1.50 0.50 1.88 0.78 1.25 0.31 1.63 0.70 3.00 1.00
1.3 During this week, I have
experienced discomfort
caused by daylight.
2.00 0.71 2.00 0.71 2.25 0.83 2.25 0.83 1.75 1.83 1.75 0.43 1.75 0.31 2.75 0.83 1.13 0.54
Perceived atmosphere
2.1 During this week, the
electrical lighting in the
space has had a positive
impact on the work
atmosphere.
2.25 0.83 2.25 0.83 2.75 0.43 2.75 0.43 3.00 0.71 3.00 0.00 3.00 0.50 2.88 0.60 2.13 0.60
2.2 During this week, the
combination of daylight
and electrical lighting has
felt natural for me.
3.25 0.83 3.75 1.09 2.75 1.09 3.25 0.83 3.00 1.00 2.63 0.86 2.75 0.31 2.75 0.43 2.50 1.12
2.3 During this week, the
electrical lighting has
been stimulating for my
work.
2.25 0.83 2.50 0.87 2.50 0.87 2.75 0.43 3.00 1.22 2.63 0.70 3.00 0.50 2.88 0.60 2.00 0.71
2.4 During this week, I have
been feeling comfortable
being in the room.
3.75 0.43 3.50 1.50 3.00 1.22 3.75 0.83 3.50 0.87 3.25 0.66 3.25 0.59 3.50 0.71 2.75 0.66
Work engagement
3.1 During this week, my
work performance has
been good.
3.75 0.83 3.25 0.83 3.00 0.71 3.75 0.83 3.00 0.71 3.50 0.50 3.75 0.31 3.50 0.87 3.13 0.78
(continued)
10 Indoor and Built Environment 0(0)
Table 3. Continued.
Month September October November December
Static/Dynamic Static Dynamic Static Dynamic Static Dynamic Static Dynamic Control
Week nr. 36–37 38–39 42–43 40–41 45–46 47–48 51–52 49–50 49–50
Mean/Standard Deviation M SD M SD M SD M SD M SD M SD M SD M SD M SD
3.2 During this week, I have
been feeling motivated to
work.
3.75 0.43 3.50 0.50 3.00 0.71 3.75 0.83 3.50 0.87 3.25 0.43 3.75 0.92 3.38 0.48 3.13 0.93
3.3 During this week, I have
produced some novelty/
innovative ideas.
2.25 0.83 2.50 1.12 2.00 0.71 2.75 0.83 2.75 0.83 2.38 0.86 2.50 0.79 1.75 0.83 2.88 0.60
3.4 During this week, I have
been feeling concentrated
on my work.
3.75 0.83 3.50 0.50 2.75 0.43 3.50 0.87 3.50 0.87 3.50 0.50 4.00 0.50 3.50 0.50 3.25 0.97
3.5 During this week, I have
had a good workflow.
3.50 0.50 3.25 0.43 3.00 0.71 3.50 0.50 3.50 0.50 3.25 0.43 4.25 0.59 3.38 0.70 3.00 1.22
3.6 During this week, I have
been willing to take risks
in my work tasks.
3.25 0.83 3.50 0.87 3.00 0.71 2.50 0.50 3.00 1.22 2.63 0.48 3.00 0.50 2.63 0.86 3.00 1.00
Note: For all months dynamic and static periods are shown with mean values and standard deviations.
Hansen et al. 11
The analysis of visual comfort, through the ques-
tionnaires and the interviews, revealed several findings.
First, there was a general favouritism of dynamic light-
ing. For instance, after changing from dynamic lighting
to static lighting, participants preferred the dynamic
lighting and referred to it as smart and adjusting,
while describing the static lighting period as lacking
those qualities. Second, during the static lighting
period, visual comfort, especially referring to evening
hours, was found to have negative nodes. Several par-
ticipants described it as too bright and intense. This
indicates that working with light zones and ambient
and task lighting components can provide a higher
visual comfort for participants, still meeting the stan-
dard of 500 lx on the work task area. Lastly, it was
detected that double dynamic lighting, composed of
diffuse and direct lighting components, provided
higher visual comfort for participants, compared to
the experimental group.
Perceived atmosphere
Four questions were designated for evaluating the per-
ceived atmosphere within the impact on work atmo-
sphere, naturalness of lighting, stimulation and
feeling comfortable (see Table 3). During September,
November and December, differences between ranked
Table 4. Overview of positive and negative nodes from
content analysis.
þ– Total
Dynamic (Oct.)
Visual comfort 19 4 23
Perceived atmosphere 25 3 28
Work engagement 14 3 17
Static (Nov.)
Visual comfort 8 15 23
Perceived atmosphere 10 4 14
Work engagement 11 6 17
Dynamic (Dec.)
Visual comfort 12 9 21
Perceived atmosphere 13 3 16
Work engagement 10 2 12
Note: Aggregated from three interviews.
Table 5. Mann-Whitney test results of weeks 49 and 50, comparing experimental group with control group.
Questions Mean rank Sum of rank U test Pvalue
Visual comfort
1.1 Exp. group 11.44 91.5 8.5 0.009
Control group 5.56 44.5
1.2 Exp. group 5.75 46 10 0.017
Control group 11.25 90
1.3 Exp. group 11.5 92 8 0.008
Control group 5.5 44
Perceived atmosphere
2.1 Exp. group 10.75 86 14 0.039
Control group 6.25 50
2.2 Exp. group 9 72 28 0.65
Control group 864
2.3 Exp. group 10.88 87 13 0.032
Control group 6.13 49
2.4 Exp. group 10.63 85 15 0.057
Control group 6.38 51
Work engagement
3.1 Exp. group 9.38 75 25 0.436
Control group 7.63 61
3.2 Exp. group 8.94 71.5 28.5 0.692
Control group 8.06 64.5
3.3 Exp. group 5.88 47 11 0.019
Control group 11.13 89
3.4 Exp. group 8.75 70 30 0.815
Control group 8.25 66
3.5 Exp. group 9.25 74 26 0.51
Control group 7.75 62
3.6 Exp. group 7.5 60 24 0.377
Control group 9.5 76
Note: The underlined values are enhanced for significance.
12 Indoor and Built Environment 0(0)
values were insignificant, between 0 and 0.37 for 15
questions out of 16. In October, the naturalness of the
combination of daylight and electrical lighting as well as
feeling comfortable being in the room were evaluated 0.5
higher during dynamic lighting than during the static
period in the same month. This demonstrates little evi-
dence, although it indicates that October should be
considered for further analysis.
There were four questions related to the perceived
atmosphere in the questionnaire (see Table 5). Three
out of four were found significant in their ranking dif-
ferences (p<0.05). In Q 2.1, the test group noted a
higher positive impact (M ¼10.75) of the lighting con-
tributing to the working atmosphere than the control
group (M ¼6.25). In Q 2.2, where participants were
asked to evaluate the combination of natural and elec-
trical lighting, no significant ranking differences were
discovered with a pvalue over 0.05 and close mean
rank values, M ¼8 for the experimental group and
M¼9 for the control group. In Q 2.3, the experimental
group found the dynamic lighting more stimulating
(M ¼10.88) than the control group that found static
lighting (M ¼6.813), with notable significance and
mean rank difference. In Q 2.4 about evaluating the
atmosphere as comfortable, the test group ranked it
higher (M ¼10.63) than the control group (M ¼6.38).
For three out of four questions, the null hypothesis was
rejected.
In the qualitative content analysis, the ratio of pos-
itive to negative nodes from the three interviews
showed similar results (see Table 4). In October, eval-
uating dynamic lighting, 25 out of 28 positive nodes for
perceived atmosphere were detected, and 13 out of 16
in December. In November, after four weeks of static
lighting, 10 out of 14 positive nodes were found.
The perceived atmosphere was characterized with
these following positive statements in October regard-
ing dynamic lighting:
I feel like when there’s like overcast and the sun isn’t
that much out, the lighting is kind of more comforting
in the room, because it fills more. [ID2, 1st int.,
Dynamic]
I noticed the light in the room, but it feels like the
daylight and the light in the room has blended in
together in a way. So, the weather would be like
casted into the room, projected. [ID3, 1st int.,
Dynamic]
And then, at some point during the daytime, I don’t
notice the lighting at all and it feels very natural. It’s
just there complementing the space and task. [ID4, 1st
int., Dynamic]
These statements support the aim of using daylight
components as inspiration for dynamic electrical light-
ing, creating a more natural luminous atmosphere in
the office environment, combining warm and cool
lighting.
In comparison, under static lighting, participants
gave contrasting comments about the perceived atmo-
sphere in the space. For instance, one participant said,
‘When it is bright, I feel awake. I feel focused, anything
missing in the horizon. I can do my job. I feel a strong
presence in the whole room’ [ID3, 2nd int., Static].
Another participant described the lighting with the fol-
lowing words:
The lighting has been static, but the experience has
been varying. During the daytime it’s been a natural
feeling. So, in that sense, it’s been a natural feeling, but
not really in the evenings. It’s been negative. [ID4, 2nd
int., Static]
Several participants described the same negative con-
trasting experience of lighting in relation to surround-
ings: ‘When the lights in the hallway were off and you
entered the room. That sensation...I don’t know what
it was, but it was uncomfortable and clinical.’ [ID1,
2nd int., Static]
The analysis of perceived atmosphere revealed the
following findings. First, the double dynamic lighting,
consisting of warm, neutral and cool colour tempera-
tures, was described positively, with several references
to the naturalness of the lighting and daylight compo-
nents, supporting the design concept. Second, on the
contrary, the static lighting period, especially during
evenings, triggered negative sensations among partici-
pants, resulting in clinical and uncomfortable atmos-
pheres. Lastly, the perceived atmosphere, composed of
diffuse and direct lighting, was evaluated as more com-
fortable and stimulating by the experimental group in
comparison to the static lighting of the test group. All
these findings confirm that the double dynamic lighting
performs better than static lighting in evaluation of
perceived atmosphere.
Work engagement
Six questions were designated to detect patterns in
motivation, concentration, workflow, willingness to
take risks, innovative ideas and perceived work perfor-
mance (see Table 3). The mean results show more sig-
nificant differences than in the two previous criteria,
visual comfort and perceived atmosphere. In October,
five out of six questions were answered with positive
rankings, and four of them (Q 3.1–3.4) were ranked
0.75 higher during the dynamic lighting period com-
pared to the static. October is also registered with the
Hansen et al. 13
most diverse light setting distribution (Shannon’s index
of 2) and the highest amount of clear sky conditions
(44%; see Table 3).
This correlation for the six questions within work
engagement showed the least amount of significance
in ranking differences between the control and experi-
mental group, with only one out of six with p<0.05
(see Table 5). In Q 3.3, evaluating novelty and innova-
tive ideas, the results were M ¼11.13 for the experi-
mental group and M ¼5.88 for the control group.
Other questions in this section were all found to have
insignificant pvalues. Only for Q 3.3 could the null
hypothesis be rejected.
In the qualitative content analysis, work engagement
in relation to lighting showed similar results in all
three periods, with 14 out of 17 positive nodes during
dynamic lighting in October and 10 out of 12 in
December (see Table 4). During static lighting, 11 out
of 17 positive nodes were detected in November, per-
forming with a less positive ratio compared to dynamic
lighting.
With regard to the dynamic lighting and work
engagement in October, one participant described it
as follows:
I wouldn’t necessarily say concentration or focus.
I mean, you do have that, but you’re generally
awake, I could say energized, but not like jumping
out on my own chest, just energized. I mean I don’t
get tired; we can stay there for a long time. We are very
motivated for what we are working with, but lighting
has supported that quite well. [ID3, 1st int., Dynamic]
The work engagement during the static lighting period
came with varied experiences. One of the participants
described the lighting in a positive way: ‘I’ve always
been focused and energized, until I have to take
myself home’ [ID3, 2nd int., Static]. Another partici-
pant would like to have personal control over the light-
ing: ‘I think, if we could play around with it, maybe I
could get more motivated to do my work’ [ID1, 2nd
int., Static]. The lighting was also described as having
negative effects, especially during work hours after
sunset:
I just stopped working and went home. Because I knew
it was too bright. I wasn’t too motivated to continue
finishing up the tasks I had said I had to finish that
day. I think maybe some warmer lights during the night
will probably motivate me a bit further, during the end
of the day. [ID3, 2nd int., Static]
The statement from ID3 also stresses the need for dif-
ferent lighting during the daytime, transition and night.
The analysis of work engagement revealed the fol-
lowing findings. First, the questionnaires show that
double dynamic lighting in October was found to
have the most positive effect for work engagement,
hereunder motivation, concentration and workflow. It
is implied that larger diversity of exposure to different
light settings contributes positively to work engage-
ment. Second, the period with static lighting, on the
other hand, shows that subjective experience and per-
sonal preferences can vary. Therefore, personalized
task lighting could offer individual control over inten-
sity and colour to meet the personal needs of the users
and to reinforce work engagement, keeping users moti-
vated and focused on work.
Discussion
This study was targeted to develop design strategies
and methods by testing the complex interplay between
the dynamics of daylight and electrical lighting. The
approach of setting up pre-experiments to inform the
lighting design in a complex long-term experiment was
shown to be valuable. These responsive design concepts
based on the perceived qualities of dynamic light, can
now be applied in other lighting designs and developed
further within the research methodologies and dynamic
parameters, as well as the recommendation for light
settings.
Future potential regarding the method
The effect of dynamic light on visual comfort, per-
ceived atmosphere and work engagement is determined
by many variables, including differences between par-
ticipants (e.g., age, gender, mental state, work experi-
ence, working hours, attitudes, presence and different
task types), context (e.g., social presence, location,
exposure duration, time of day and year), and param-
eters of light (e.g., illuminance level, ratio and distribu-
tion, as well as spectral distribution and the weather).
This multidimensionality of studying the effect of
dynamic light is the reason that many studies prefer
to set up tests in controlled labs and to work with
only daylight or variables in electrical lighting.
However, this experiment with four participants work-
ing for four months in a specific space and under spe-
cific lighting conditions created specific knowledge:
knowledge of the potential and complexity of designing
and testing the effect of double dynamic lighting, the
interplay of the dynamics of daylight and electrical
lighting with the light we all perceive in the built envi-
ronment. The knowledge gained from this experiment
must be considered in relation to other contexts and
geographic locations, where weather, seasons, daylight
inflow, participants, tasks and space may be different.
14 Indoor and Built Environment 0(0)
For future research, the concept is recommended for
implementation and evaluation with a greater number
of participants, and ideally, with participants who rep-
resent a higher average age and are assigned more
diverse tasks, as well as in a larger office space.
Regarding the methods for testing, we found, in par-
ticular, that the qualitative interviews were valuable to
evaluate the qualities and impact of the dynamic light-
ing over time in relation to work engagement. Further,
it was revealed that participants sometimes found it
difficult to respond to questions regarding their percep-
tion of the atmosphere and how it affects their motiva-
tion and engagement; this difficulty was reflected in
long pauses but resolved with clarifying questions to
the interviewer. The method of combining question-
naires with the designed interview guide—structured
according to the triangulation of focus areas of visual
comfort, perceived atmosphere and work engage-
ment—is recommended for future field experiments.
However, the criteria visual appearance and perceived
atmosphere may be considered as based on immediate
responses, whereas the effect on work engagement is
affected over longer periods. Future field investigations
can investigate the visual appearances and perceived
atmosphere as an immediate response to the space
more frequently, and by that, decrease participants’
recall bias effects. In this study, interviews were con-
ducted in a different office space than the experiment
office. Tests regarding the visual appearance and per-
ceived atmosphere are recommended and should be
conducted in the actual experimental office to have
the test-person experience this onsite while answering.
Another focus point for improvement is the tasks, as
these have an important influence on work engage-
ment. The control group in this study did not work
with the exact same subject areas as the experimental
group. Future experiments aim at selecting a control
group that works in the same field and with same type
of project are recommended, to minimize the influence
of different tasks and processes on the outcome, par-
ticularly of the long-term effects, such as work
engagement.
Points of improvement regarding
light settings
The results suggest points of improvement for light
settings tested in this double dynamic lighting concept.
The current seven light settings is suggested to reduce
to five settings, consisting of one setting for transition
hours and work hours after dark, two settings for over-
cast sky, and two for clear sky conditions. This is sug-
gested due to the data regarding frequency of
triggering. During the dark periods, setting 3 O, 4 O,
and 5 O were never triggered due to the daylight
inflow being lower than 200 lx under overcast sky.
The 4 C and 4 O light settings were the least triggered,
with means of only 2.5% and 0.5%, respectively. These
light settings were not found to be feasible for future
implementations. Second, the minimum threshold of 50
should be lowered (20–30 lx) for triggering scenarios
2 O and 2 C, the daytime settings, to ensure that day-
time light settings are triggered more often.
The results show that, especially during transitions
hours and dark hours, there is a potential for defining
specific settings, with warmer colour temperatures and
lower intensities for the ambient lighting.
Investigations may be carried out into whether the
ambient light during the dark hours can reduce the
level of illuminance (300 lx) as defined in current stand-
ards. These studies can be combined with an approach
supporting circadian stimuli providing low melanotic
content during evening.
Potential for developing the lighting
design concept
A potential for future development of the double
dynamic concept is adjustment to the time of the day
by adding a time factor to the data triggering the light
settings, and thereby meeting needs for circadian stim-
uli, which is especially relevant in a Nordic context.
Seasonal changes in light intensity and cloud coverage
also had a strong impact on which electrical light settings
the occupants were exposed to. During September and
October, the distribution between triggered light settings
was highest for the 2 O setting, at 33%. During winter
months, November and December, the light setting 1OC
was triggered 83% and 88% of the time, respectively,
which demonstrates potential for designing new dynamic
light settings in which the daylight inflow illuminance is
defined in relation to seasonal changes, with a summer,
autumn and winter setting. It was also possible to vali-
date the time factor, temporal pattern, in relation to non-
visual effects, with reference to timing of exposure in
relation to daily rhythm, duration of exposure and
photic history.
24
The results of this study also point toward the
potential of implementing individual settings for the
task lighting while the diffuse ambient lighting follows
the dynamics of skylight. This opens possibilities to
meet individual visual needs for different tasks, as
well as needs for higher light intensities, boosting the
lighting to stimulate individual circadian entrainment;
impacts of light on alertness and sleep have the poten-
tial to be integrated in this design concept.
3,15,17–19
The experiment also exposed the potential for a new
additional setting with the time factor as a pre-
programmed script of a built-in dynamic lighting set-
ting, changing automatically in colour temperature and
Hansen et al. 15
distributions as a pattern when the daylight level and sky
are static over a longer period of time. This is especially
relevant for the winter period in such locations as
Scandinavia with predominantly overcast sky conditions.
It was possible to detect dynamic lighting, for instance,
when one overcast light setting was present, unchanged
for more than an hour, during daylight hours. Thereby,
the dynamic direct and diffuse lighting can compensate
for the static daylight hours through dynamic light set-
tings, adding visual dynamics, full of change, instability
and unpredictability. This could introduce a new
approach to the understanding of how transition time
between lighting scenarios could be seen as visually inter-
esting and stimulating for an environment.
The switches between the light settings were pro-
grammedtohappeneverytenthminute,whena
change in sky-type and/or daylight levels was detected.
The switch duration time from one light setting to anoth-
er was set as 10 s. The interviews confirmed that the
change of switches was rarely noticed. This parameter
was not a focus area in this study. Future studies can
elaborate on this parameter and investigate the balance
of possible negative effects of dynamic lighting and the
stimulating effects of perceived variations in lighting with
reference to the perceived variations of natural light.
The concept also has potential in responding with
dynamic sunscreens. This study determined that, when
the sunscreen is drawn manually, due to direct sunlight
entering a space, it is often not removed when it is not
needed anymore. There is a potential to define settings
to meet the daylight levels in the space when the sun-
screen is drawn. It is possible to trigger the sunscreen
and the definition of sky type through the same or a
similar illuminance sky scanner to the one used in this
project, because this sky scanner registers illuminance
from four cardinal directions.
This approach of complementing the qualities of
daylight and human perception to create appropriate
light level and distribution in the individual workplace
and in the space, can be investigated in relation to
energy-saving potential. A future study can measure
and calculate the electricity used for lighting in larger
office spaces to validate the savings associated with
reducing lighting to what is actually needed for indi-
vidual task lighting and ambient lighting, whereas
today the standard is an even distribution of a fixed
level of illumination in office spaces.
Based on these findings, the double dynamic design
concept is recommended for further development in
future designs.
Conclusion
This investigation was targeted to develop an integrat-
ed lighting design strategy by testing a new dynamic
and responsive lighting design concept in an experi-
ment in an office. We examined whether dynamic light-
ing, complementing and responding to the natural
dynamics of daylight, could improve the perceived
atmosphere and visual comfort in an office environ-
ment and thereby improve work engagement. Despite
the low number of participants, the findings from this
experiment can inform the definition of future respon-
sive lighting design strategies and validate the potential
of the integrated lighting, responding to the dynamics
of daylight through a combination of direct and diffuse
lighting. We conclude the following:
1. The dynamic light settings, responding to the
dynamics of daylight through a combination of
direct task lighting and diffuse ambient lighting,
have a positive impact on visual comfort, perceived
atmosphere and work engagement compared to
static lighting. The double dynamic lighting provid-
ed a more suitable visual luminous condition and
was also found to be more pleasant, with a lower
visual discomfort factor. The double dynamic
lighting was described with several references to nat-
uralness of lighting and daylight components.
2. Qualitative interviews can contribute to in-depth
insights on how light affects the perceived atmo-
sphere and work engagement.
3. It is recommended to develop and validate the
design concept responding to sky conditions and
daylight levels. Seasonal changes, time of day,
dynamic sunscreens and individual needs for task
lighting can be implemented in future field experi-
ments as additional dynamic parameters. Visual and
non-visual needs can be considered in a single inte-
grative design approach. To implement this on a
large scale, an automated system and a user-
friendly interface must be developed, as well as a
strategy for how this corresponds to an integrated
control system of the total building operation.
This design strategy demonstrates a potential in
using new sensor and lighting technologies to meet
human needs and satisfy a desire for variations and
more natural atmospheres in indoor environments.
The findings from this experiment advance the under-
standing of the qualities of integrating the dynamics of
daylight and electrical lighting, referring to human sen-
sation of unpredictability, naturalness, flow of light,
light modeling effects and light zones. The findings
can be adapted to other contexts and facilities, such
as educational, housing and the health sector. The
core of this approach is to help to form future lighting
design guidelines and research experiments exploring
how a responsive lighting technology, reacting to and
16 Indoor and Built Environment 0(0)
complementing the daylight dynamics, can reconnect
man and nature.
Authors’ contribution
Ellen Kathrine Hansen was responsible for the design con-
cept, research process and project. Mihkel Pajuste and
Emmanouil Xylakis collected and analysed all, respectively,
qualitative and quantitative data and wrote the paper togeth-
er with Ellen Kathrine Hansen. Thomas Bjørner supervised
the data collection and analysis and was involved in writing.
Acknowledgements
Thanks to our industrial partners Tridonic, Fagerhult,
iGuzzini and Zumtobel as well as the test participants and
not least our colleagues from the Lighting Design Research
Group at AAU.
Declaration of conflicting interests
The author(s) declared no potential conflicts of interest with
respect to the research, authorship, and/or publication of this
article.
Funding
The author(s) disclosed receipt of the following financial sup-
port forthe research, authorship, and/or publication of this
article:This paper represents the final findings from the proj-
ect “Double Dynamic Lighting” co financed by Aalborg
University, Tridonic, Fagerhult, iGuzzini and Zumtobel.
ORCID iDs
Ellen Kathrine Hansen https://orcid.org/0000-0001-7858-
9511
Thomas Bjørner https://orcid.org/0000-0001-9071-7168
References
1. Knoop M, Stefani O, Bueno B, Matusiak B, Hobday R,
Wirz-Justice A, Marling K, Kantermann T, Aarts MPJ
and Zemmouri N. Daylight: what makes the difference?
Ligh Res Technol 2019; 49: 423–442.
2. Ulrich RS. Aesthetic and affective response to natural
environment. In: Altman I, Wohlwill JF (eds) Behavior
and the natural environment. Human behavior and environ-
ment (advances in theory and research). vol 6. Boston,
MA: Springer, 1983.
3. Ulrich RS. Biophilic theory and research for healthcare
design. In: Kellert S, Heerwagen J and Mador M.
Biophilic design: the theory, science and practice of bring-
ing buildings to life. Hoboken, NJ: Wiley, 2008,
pp.87–106.
4. Shrum G. Sun and earth. Light Illuminat Architect 2017;
49: 19–35.
5. Sander EJ, Caza A and Jordan PJ. The physical work
environment and its relationship to stress. In: Ayoko
OB and Ashkanasy N (eds) organizational behaviour and
the physical environment. Abingdon: Routledge, 2019,
pp.268–284.
6. World Health Organization. Mental health in the work-
place, https://www.who.int/mental_health/in_the_work
place/en/ (2019, accessed 13 March 2020).
7. Cuttle C. Lighting design: a perception-based approach.
New York, NY: Routledge, 2015.
8. De Kort YA and Smolders KC. Effects of dynamic light-
ing on office workers: First results of a field study with
monthly alternating settings. Light Res Technol 2010; 42:
345–360.
9. Leder S, Newsham GR, Veitch JA, Mancini S and
Charles KE. Effects of office environment on employee
satisfaction: a new analysis. Build Res Inform 2016; 44:
34–50.
10. Rashid M and Zimring C. A review of the empirical lit-
erature on the relationships between indoor environment
and stress in health care and office settings: problems and
prospects of sharing evidence. Environ Behav 2008; 40:
151–190.
11. Stokkermans M, Vogels I, Kort Yaw D and Heynderickx
I. A comparison of methodologies to investigate the
influence of light on the atmosphere of a space. Leukos.
J Illumin Eng Soc 2018; 14: 167–191.
12. Veitch JA and Newsham GR. Preferred luminous con-
ditions in open-plan offices: research and practice rec-
ommendations. Int J Light Res Technol 2000; 32:
199–212.
13. Veitch JA, Newsham GR, Boyce PR and Jones CC.
Lighting appraisal, well-being and performance in
open-plan offices: a linked mechanisms approach. Light
Res Technol 2008; 40: 133–151.
14. Veitch JA, Stokkermans MG and Newsham GR. Linking
lighting appraisals to work behaviors. Environ Behav
2013; 45: 198–214.
15. Boyce PR, Veitch JA, Newsham GR, Jones CC,
Heerwagen J, Myer M and Hunter CM. Lighting quality
and office work: two field simulation experiments. Light
Res Technol 2006; 38: 191–223.
16. Hansen EK and Mathiasen N. Dynamic lighting balanc-
ing diffuse and direct light. In: Arch19 – 4th conference on
architecture research care & health with the title,
Trondheim, Norway, 12–14 June 2019.
17. Morita T and Tokura H. The influence of different wave-
lengths of light on human biological rhythms. Appl
Human Sci 1998; 17: 91–96.
18. Ruger M, Gordijn MC, Beersma DG, de Vries B and
Daan S. Time-of-day-dependent effects of bright
light exposure on human psychophysiology:
comparison of daytime and nighttime exposure. Am J
Physiol-Regulat Integrat Comparative Physiol 2006; 290:
R1413-R1420.
19. Ru T, de Kort YA, Smolders KC, Chen Q and Zhou G.
Non-image forming effects of illuminance and correlated
colour temperature of office light on alertness, mood, and
performance across cognitive domains. Build Environ
2019; 149: 253–263.
20. Beute F and de Kort YA. Let the sun shine! Measuring
explicit and implicit preference for environments differing
in naturalness, weather type and brightness. J Environ
Psychol 2013; 36: 162–178.
Hansen et al. 17
21. de Bakker C, Aarts M, Kort H, van Loenen E and
Rosemann A. Preferred luminance distributions in
open-plan offices in relation to time of day and subjective
alertness. Leukos J Illum Eng Soc 2021; 17: 3–20.
22. Begemann SH, Van den Beld GJ and Tenner AD.
Daylight, artificial light and people in an office environ-
ment, overview of visual and biological responses. Int J
Industrial Ergonom 1997; 20: 231–239.
23. Fleischer SE. The psychological effect of changeable arti-
ficial lighting situations on humans. Zurich (Switzerland):
ETH Zurich Research Collection, 2001.
24. Houser KW, Boyce PR, Zeitzer JM and Herf M. Human-
centric lighting: myth, magic or metaphor? Light Res
Technol 1973; 3: 87–94.
25. Frandsen. S. The scale of light. Int Light Rev 1987; 3:
108–112.
26. Zaikina V. Light modelling in architectural spaces:
luminance-based metrics of contour, shape and retail dis-
tinctness of day-lit 3D objects. PhD Thesis, NTNU,
Trondheim, 2016.
27. Tregenza P and Daylighting WM. Architecture and light-
ing design. New York, NY: Routledge, 2013.
28. Canazei M, Dehoff P, Staggl S and Pohl W. Effects of
dynamic ambient lighting on female permanent morning
shift workers. Light Res Technol 2014; 46: 140–156.
29. Houser KW, Tiller DK, Bernecker CA and Mistrick RG.
The subjective response to linear fluorescent direct/indi-
rect lighting systems. Light Res Technol 2002; 34:
243–260.
30. McCandless SA. Method of lighting the stage, revised.
New York, NY: Theatre Arts, 1958.
31. Hansen EK, Pajuste M and Xylakis E. Flow of light:
balancing directionality and CCT in the office environ-
ment. Leukos, J Illum Eng Soc 2020. DOI: 10.1080/
15502724.2020.1808014.
32. BS EN 124643. 2011. Light and lighting—lighting of work
places—Part 1: indoor work places. London: BSI
Standard Publication, 2011.
33. Zumtobel Group. LM-TLM, https://www.zumtobel.
com/com-en/products/product_popup.html?20724105
(accessed 2 July 2020).
34. Espressif Systems. ESP8266, https://www.espressif.com/
en/products/hardware/esp8266ex/overview (accessed 13
March 2020).
35. Arduino and TCS34725 Color Sensor, http://arduino
learning.com/code/arduino-tcs34725-color-sensor.php
(accessed 13 March 2020).
36. Reitz K. Requests: http for humans, https://requests.
readthedocs.io/en/master/ (accessed 13 March 2020).
37. The Pandas Development Team. Pandas documentation,
https://pandas.pydata.org/ (accessed 13 March 2020).
38. Joe K. Leading the way through changing times.
Research World [Internet] Wiley. 2016; 2016: 56–56.
39. Beals M, Gross L and Harrell S. Diversity indices:
Shannon’s H and E, http://www.tiem.utk.edu/gross/
bioed/bealsmodules/shannonDI.html (accessed March
13, 2020).
40. Flynn JE, Spencer TJ, Martyniuk O and Hendrick C.
Interim study of procedures for investigating the effect
of light on impression and behavior. J Illumin Eng Soc
1973; 3: 87–94.
41. Vogels I. Atmosphere metrics. In: Probing experience.
Dordrecht: Springer, 2008, pp.25–41
42. O’Brien HL, Cairns P and Hall M. A practical approach
to measuring user engagement with the refined user
engagement scale (UES) and new UES short form. Int
J Human-Comput Studies 2018; 112: 28–39.
43. Creswell JW. Research design: Qualitative, quantitative
and mixed methods approaches. 4th ed. London: Sage
Publications, 2014.
44. Ramboll. SurveyXact by Rambolll, https://www.surveyx
act.com/ (2012, accessed 13 March 2020).
45. Benedek J and Miner T. Measuring desirability: new
methods for evaluating desirability in a usability lab set-
ting. In: Proceedings of UPA 2002 Conference, Orlando,
FL, 8-12 July 2002. p. 57.
46. Bjørner T. Data collection. In: Bjørner T (ed.) Qualitative
methods for user and consumer research: the value of the
qualitative approach in theory and practice. Copenhagen:
Hans Reitzel, 2015.
47. Liang S and Otter.Ai FY. https://otter.ai/ (2018, accessed
13 March 2020).
48. Richards LN. https://www.qsrinternational.com/nvivo/
nvivo-products/nvivo-12-plus (1995, accessed 13 March
2020).
49. Lothon M, Barn
eoud P, Gabella O, Lohou F, Derrien S,
Rondi S, Chiriaco M, Bastin S, Dupont JC, Haeffelin M
and Badosa J. ELIFAN, an algorithm for the estimation
of cloud cover from sky imagers. Atmos Meas Technol
2019; 13: 5519–5534.
50. Piderit MB, Cauwerts C and Diaz M. Definition of the
CIE standard skies and application of high dynamic
range imaging technique to characterize the spatial dis-
tribution of daylight in Chile. Revista de la Construcci
on.
J Construct 2014;13: 22–30.
51. Richardson W, Krishnaswami H, Vega R and Cervantes
M. A low cost, edge computing, all-sky imager for cloud
tracking and intra-hour irradiance forecasting.
Sustainability 2017; 9: 482.
52. Sun S, Ernst J, Sapkota A, Ritzhaupt-Kleissl E, Wiles J,
Bamberger J and Chen T. Short-term cloud coverage pre-
diction using ground based all sky imager. In: 2014 IEEE
international conference on smart grid communications
(SmartGridComm), Venice, Italy, 3–6 November 2014,
IEEE, pp.121–126.
53. Hansen EK, Louring S, Georgieva D, Schledermann K
M. The Impact of Dynamic Lighting in Classrooms. A
review on Methods. Institute for Computer Sciences,
Social Informatics, Telecommunications Engineering
2018. ArtsIT 2017/DLI 2017, LNICST 229, pp. 314-323.
Appendix 1. Sky- and daylight
sensors
A short study was designed to determine how the illu-
minance values received by the sky scanner change
regarding the sky condition.
49–52
Over two weeks in
May 2019, sky photographs were taken every fifth
18 Indoor and Built Environment 0(0)
minute, resulting in 2000 pictures, accompanied by
timestamps. Two research assistants manually labelled
each of the pictures taken, and for each labelled
instance, the latest sky scanner reading was matched.
There was no alternative option or software available
to be applied in this new innovative usage of the equip-
ment to be able to detect sky-types. The result of the
aforementioned process was a table of 2000 entries,
each containing sky scanner readings and the sky con-
dition labels.
The histograms of illuminance levels varied signifi-
cantly in their rate of dispersion. To determine whether
the current sky condition was overcast or clear, the
relative standard deviation (RSD) was calculated to
measure that dispersion. The experimental threshold
of 40 RSD was set for determining sky condition,
based on the data. If the reading were above the afore-
mentioned value, the sky condition would be registered
as clear, and if not, it would be registered as overcast.
See the flowchart in Figure 3 to see how this was
embedded as part of the light setting scheduling
process.
The microcontroller was used primarily due to its
ability to transfer sensor data through Wi-Fi. Data
storage could occur in the cloud, avoiding the need
for local storage or frequent maintenance checks.
For the present study, one sensor measuring hori-
zontal illuminance in the task area was placed on the
table farthest from the window and door. The
shadowing from the monitor and additional objects
on the table was taken into consideration. The imple-
mentation and placement of the compact Wi-Fi sensor
serves a twofold purpose: first, to record and store light
data related to the task zone during the experiment
period for analysis, and second, to inform the lighting
system of the current lighting condition in the room,
measured in lux, to trigger light settings accordingly.
Light data acquired from the ESP were stored in the
cloud utilizing Google Web APIs (application pro-
gramming interfaces). A Google spreadsheet was
made and configured as a Web app allowing incoming
data in.json format that were then parsed in its fields.
Appendix 2. Frequency of light
settings
Tables 6 and 7 illustrate the frequency and duration of
the seven light settings to which participants were
exposed. There were two main phases with regard to
their diversity in dynamics. September and October are
defined as more diverse with a Shannon’s index
39
of 1.8
and 2, respectively, with a larger change in settings and
an average of 12–15 switches per day in contrast to 4–7
switches during November and December (see
Table 6). The dynamic periods of November and
December are identified as less diverse in their light
setting occurrences, with a 1 and 1.2 Shannon’s
index, respectively. Note that these indexes are in rela-
tion to the designed light settings and their thresholds,
regardless the seasonal change of daylight intensity.
The seasonal patterns in daylight are visible in
Table 7. In September and October, the daylight
hours are longer, and intensities are higher. For
instance, during these months, the absence of electrical
lighting was 17% and 22% of all the working hours.
Additionally, 10 C light settings, providing 300 lx of
diffuse ambient lighting with 4000 K and direct lighting
Table 7. Frequency of occurrence of all light settings during dynamic periods and Shannon’s diversity indexes.
Light Settings September October November December
1OC for clear sky/partly cloudy 1%5%20%22%
2C 5%6%4%6%
3C 5%10%1%3%
4C 3%6%0%1%
Absence for clear sky/partly cloudy 20%17%3%2%
Total for clear sky/partly cloudy 34 44 28 33
1OC for overcast 24%24%68%61%
2O 33%24%3%4%
3O 6%8%0%0%
4O 1%1%0%0%
Absence for overcast 2%0%0%0%
Total for overcast 66 56 72 67
Shannon’s diversity index 1.8 2 1 1.2
Table 6. Frequency of lighting setting switches during
dynamic periods.
Month Switches Switches per day
September (dynamic) 174 12
October (dynamic) 208 15
November (dynamic) 52 4
4December (dynamic) 92 7
Hansen et al. 19
from a spotlight providing additional 200 lx with
2700 K, were rarely triggered with clear sky conditions,
1% and 5%, respectively.
During November and December, with less daylight,
3 O, 4 O, or 5 O (absence for overcast) light settings
were never triggered due to the daylight inflow being
lower than 200 lx under overcast sky, measured on the
table. The 4 C and 4 O light settings were the least trig-
gered, with a mean of only 2.5% and 0.5%.
Additionally, the 1OC light setting was triggered
88% and 83% of the time in November and
December, respectively, for overcast and clear sky com-
bined. This demonstrates that this light setting, which
was designed mainly to be triggered during transition
hours and after dark, was in fact triggered most of the
time during these months. This could be due to a large
change in mean daylight illuminance. The readings
from the sky scanner showed the following mean
values for direct/diffuse daylight: September
M¼17.679 lx, October M ¼15.858 lx, November
M¼2.155 lx, and December M ¼2.562 lx. Note that
there is a five-week gap between the October and
November dynamic periods (see Table 1). In addition,
the participants reported frequent usage of blinds,
which could also be a factor resulting in triggering
1OC, with lowered daylight intake.
20 Indoor and Built Environment 0(0)
