Journal of Building Engineering 45 (2022) 103407

Contents lists available at ScienceDirect

Journal of Building Engineering

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

An artificial skylight compared with daylighting and LED: Subjective and

objective performance measures☆
A. Bellazzi *, L. Danza, A. Devitofrancesco, M. Ghellere, F. Salamone
ITC-CNR, Construction Technologies Institute of National Research Council of Italy, Via Lombardia 49, San Giuliano Milanese, 20098, Milano, Italy

A R T I C L E I N F O A B S T R A C T

Keywords: In the last decade, innovative systems such as Artificial Skylights (AS) have been developed to reproduce the
Artificial skylight blue-sky effect and white sunlight, giving the impression that interiors are illuminated by natural light, even
Monitoring campaign when this is not available because of construction- or climate-related reasons. Given the hybrid characteristics
User perception
between natural and artificial lighting, the aim of this research is to compare an AS with daylight and with
Visual well-being
artificial lighting to identify similarities and differences in performances, pointing out the most suitable in­
dicators to describe it and to provide useful feedback for the design and integration of these systems.
A monitoring campaign was conducted involving nine participants who spent entire work sessions in a full-
scale living lab equipped as an office, with different lighting systems (AS, daylight and LED) and different
furniture configurations, completing a total of more than 360 questionnaires and analysing the visual and non-
visual effects.
AS and daylight share many similarities, which may be positive, such as pleasantness, circadian stimuli, colour
perception, as well as, in some scenarios, also negative, such as glare perception. As a conventional artificial
lighting system, even with AS, Unified Glare Rating correlates more strongly with users’ perception of glare than
Daylight Glare Probability, although both underestimate it. Overall, the workstation receiving diffuse light was
found to be the most comfortable and the higher installed lighting power density of AS with respect to LED is well
balanced by a better lighting quality.
Further research about AS is needed to define design guidelines and to support a holistic approach, funda­
mental to high-performing buildings.

been implemented that recreate the view outside, with poor results in
terms of reproducing daylight [9]. One example is the Virtual Natural
1. Introduction
Lighting Solutions (VNLS), but the first and second generation pro­
totypes tested ([10–14]) have not provided clear results [5]. Another
Many studies have investigated and demonstrated the numerous
solution is the artificial skylight, able to simulate a ceiling opening and
benefits of daylight (DL) in buildings such as the positive effects on vi­
daylighting, but there is little previous research on it [5], with different
sual comfort, psychological benefits, health, workplace productivity and
settings and goals. To our knowledge, two systems are studied in the
energy savings [1–3]. As it is well known, people spend a lot of time
literature. The former is a prototype made of LED (Light Emitting Diode)
indoors [4]. Since it is not always possible to provide daylight due to
and covered by a blue optical structure. For clarity, it will be hereafter
constructional, safety or hygiene reason [5], negative effects such as a
referred to as Artificial Skylight Prototype (ASp). The latter system
reduction in alertness, drowsiness, psychological stress and lower job
consists of an artificial light source, spectrally similar to the visible part
satisfaction can occur [6]. In addition, if proper lighting conditions are
of sunlight (CCT – Correlated Colour Temperature 5770 K), and a
not regularly achieved for circadian stimulus effectiveness, zones can be
nanostructured material that mimics the Rayleigh scattering process
labelled as “biologically dark” with the risk of disrupting the circadian
that occurs in the atmosphere. It is called Artificial Skylight (AS) and it is
system [7]. Natural environments have positive effects on people [8]
the system investigated in this research. Stokkermans [15] investigated
but, in an attempt to connect with nature and reproduce the positive
an Asp capable of reproducing the blue sky effect and white sunlight,
effect of natural daylight when it is not available, some solutions have

☆
All authors have seen and agree with each of the changes made to this manuscript in the revision.
* Corresponding author.
E-mail address: [email protected] (A. Bellazzi).

https://doi.org/10.1016/j.jobe.2021.103407
Received 7 June 2021; Received in revised form 28 September 2021; Accepted 1 October 2021
Available online 11 October 2021
2352-7102/© 2021 Elsevier Ltd. All rights reserved.
A. Bellazzi et al. Journal of Building Engineering 45 (2022) 103407

Nomenclature Lsi glare source luminance [cd/m2]

MCCA Main Colour Confusion Area
AS Artificial Skylight MPOD Macular Pigment Optical Density
ASp Artificial Skylight prototype OW Desk orientation toward wall
CCT Correlated Colour Temperature OF Desk orientation toward fenestrations
CR Contrast Ratio OD Desk orientation toward door
CS Circadian Stimulus p: Guth position index
DGP Daylight Glare Probability P position index
DL Daylighting PD workstation near the door
E Illuminance level (lx) PM workstation in the middle of the room
Emin minimum illuminance level (lx) PW workstation near the wall
Em mean illuminance level (lx) Q1 General user data questionnaire
Em,t Mean illuminance of the task area (lx) Q2 User initial conditions questionnaire
Ep Punctual illuminance in the middle of the desk (lx) Q3 Hourly questionnaire
Ev vertical illuminance at the eye of the observer [lx] Q4 Ambient quality questionnaire
IAQ Indoor Air Quality TES Total Error Score
IEQ Indoor Environmental Quality UGR Unified Glare Rating
FoV Field of View U0 Uniformity factor
HDR High Dynamic Range U0,t Uniformity of the task area
LCD Liquid Crystal Display U0,s Uniformity in the immediate surrounding area
LED Light Emitting Diode U0,b Uniformity of the background area
Lb background luminance [cd/m2] VNLS Virtual Natural Lighting Solutions
Ls glare source luminance in the direction of the observer’s ω glare source surface –axis of sight solid angle [sr]
eye [cd/m2] ωsi solid angle subtended by the glare source I [sr]
Lref reference luminance value

and described two tests that measured both visual and non-visual pa­ questionnaires. The test duration was approximately 60 min for each
rameters. The former test investigated whether people could appreciate skylight system. 57% of participants felt that the ASp provided the
an ASp in an office space by measuring ratings of lighting (attractive­ strongest skylight experience, even though it was not compared to a real
ness, suitability for work, glare), room conditions (attractiveness, at­ skylight. Canazei conducted two different experiments in 2016 [18] and
mosphere, visual clarity, uniformity, spaciousness), affect as well as 2017 [5], in which an AS was able to simulate a ceiling opening with a
perceived naturalness (daylight experience, naturalness of colours). cloudless sky and sun and it was compared with a fluorescent lamp. In
Twenty-nine participants took part for 30 min in a test aimed at inves­ Ref. [18] one hundred people filled out 3 different questionnaires about
tigating the difference between a room with ASp and a room with room lighting, room atmosphere and connection to nature when
standard office fluorescent lighting. The results of this first test showed exposed to an artificial skylight for 60 min. He found that room lighting
that individuals preferred the standard room for an office activity due to of the AS is more pleasant, attractive and natural and glare perception is
greater uniformity of light, visual clarity and appropriate lighting. On the same for fluorescent lamp and skylight even when the tester is sitting
further investigation through a funnel debriefing, it was found that in­ in the “sun” position (directly hit by artificial sun). Only the in­
dividuals who had a better daylighting experience had previously homogeneity is perceived as a negative aspect in an office. In the second
recognized the ASp in a better way. Pooling the results of the latter test, study [5] attention was focused on non-visual aspects, with one hundred
which focused on comparing the sun pattern preferences of thirty people testers divided into two test rooms for 72 min. The main finding was that
in an office and home environment, revealed that more realism was users felt more connected to nature with AS and perceived the office as
needed, especially in offices. Seuntiens [16] conducted an experiment to more lively, less tense and more distant than with fluorescent lighting.
analyse the preference of 20 testers regarding daylight impression in a Yasukouchi [6] tested the non-visual effects (brain arousal level, auto­
room, comparing images of different artificial skylight configurations nomic nervous activity, work performance, subjective response during
for 70 side by side comparisons in total. Subjects preferred the larger day-time exposure and night-time release of melatonin) of an AS
rectangular skylights over the square skylights in terms of daylight compared to a conventional fluorescent light in an office, involving ten
impression, and a sky with visible sun was perceived more realistically healthy male adults. Table 1 summarizes the most important aspects
than a partly cloudy one with no visible sun. Wang [17] studied the analysed by reference literature on the subject of AS considered above.
effect on alertness, mood, self-control and cognitive performance of Lit environments are very complex when it comes to analysing the
twenty-five subjects, in an office setting during daytime with ASp. He correlation between photometric measurements and visual performance
compared four different lighting settings that varied the illuminance (E) and comfort, especially when more than one visual task is present for the
and CCT at eye level, as well as the proportion of light in the blue same position and daylighting is available [19]: visual quality can be
spectral range. During the day, participants felt less tense and happier, determined by individual aspects, but with many difficulties due to the
had higher levels of self-control, and showed a faster response in various low correlation with subjective responses and limited validation [20].
task activities under the effect of 400 lx and 9300 K light. The Carlucci affirmed that visual comfort has been commonly assessed by
blue-enriched light has no relevant effects on alertness during the day measuring the amount and uniformity of light, the quality of light in
(see the effects during the night found in other researches). Meerbeek colour rendering, and the prediction of the risk of glare for occupants,
[9] studied an ASp using a commercial led with an optical structure but all these indicators are measured separately and there is no possi­
adapted to produce a blue-sky effect in combination with white sunlight. bility to summarise a global visual comfort with a single index [21].
The ASp was compared with a Standard Panel without an optical During the monitoring campaign several parameters and aspects are
structure and the same model with a Blue Filter. Thirty participants measured, and this paper aims at addressing the following main
tested the 3 systems by performing various tasks and typing objectives:

2
A. Bellazzi et al. Journal of Building Engineering 45 (2022) 103407

Table 1
Artificial skylight references. Research goal: E, illuminance; G, glare; CS, circadian stimuli; NV, non-visual effects.
References Test room Systems tested No. Of Testers Test duration Research goal

E G CS NV

Stokkermans [15] 2 Offices ASp VS fluorescent light 29 30 min divided in 3 sessions x x x

(2011) I test
Stokkermans [15] Home and office Comparison of different sun patterns 15 at home and 5 min x
(2011) II test produced by an ASp 15 in the office
Seuntiens [16] Office Simulation with sketchup of different 20 x
(2012) skylight dimension and different sky
Wang [17] (2012) Office 3.75 × 6.10 ASp 25 4 different 1-h lighting treatments x x
×3m
Meerbeek [9] (2014) Three Offices 3 × 3 ASp compared with a Standard Panel 30 About 60 min in each room x x
× 2,40 m (SP) and blue filtered panel (BF)
Canazei [18] (2016) 2 offices (3 × 4x2.4 AS VS 3 louvre luminaries with 100 tested both 60 min in each room x x x
m) fluorescent lamp systems
Canazei [5] (2017) 2 offices (3 × 4x2.4 AS VS 3 louvre luminaries with 50 per office = 72 min x
m) fluorescent lamp 100 tot
Yasukouchi [6] Office (1.96 × 2.7 × AS VS fluorescent lights 10 36 min repeated twice during the x x
(2019) 1.94 m) morning and twice during the afternoon
This research Office 6.4 × 3.7 × AS 9 3–7 h a day for 3 lighting systems in 3 x x x x
2.95 m (l x w x h) configurations each

• characterize the performance of AS when monitored data and user spent whole working days testing 3 different lighting systems in 3
preferences are compared with daylight and LED lighting to provide different furnishes configurations. All participants come from South
useful feedback for design and integration of these systems, testing Europe and have been selected in order to have a heterogeneous group
the lighting pattern in different working positions. The most considering age (average = 42 y ± 8.8), gender (5 males and 4 females),
important factors of visual comfort and users feeling are evaluated: colour of eyes (5 light and 4 dark) and absence of eye pathologies.
the amount and uniformity of light, glare, colour rendering, circa­ Choosing a small number of participants is a compromise to satisfy
dian rhythm and the user feedback; two requirements. The former is an analysis based on within-subjects
• find out whether standard indicators can correctly represent an study design. The same participant tested all the conditions. The latter
artificial skylight. is the period chosen to concentrate the experiments in the period near
the spring equinox, in order to conduct the entire campaign at the same
The monitoring campaign described in this paper involved nine times maintaining a solar path as constant as possible and similar nat­
participants who spent entire work sessions in a full-scale living lab ural light and solar radiation conditions.
equipped as an office, with different lighting systems (DL, AS, and LED All participants previously provided their informed consent for in­
lighting) and different furniture configurations, completing a total of clusion and the experiment was approved by the Ethics Review Com­
more than 360 questionnaires and analysing visual and non-visual mittee of CNR (National Research Council of Italy) with registration
effects. number: 9397/2020.

2. Method 2.2. Tested systems

2.1. Experimental campaign Three lighting systems (Fig. 2) are installed in the test room and
tested during different days: Daylighting (DL), Artificial Skylight (AS)
The test experience was performed in a full-scale living lab room of and Artificial Lighting (LED). When the AS and LED are tested, the
about 6.4 × 3.7 × 2.95 m (l x w x h) equipped as an office (Fig. 1) located windows are darkened by 100% opaque rolling curtains (measured
near Milan (45◦ 23′ N, 9◦ 15′ E), with white plaster and matte finish sur­ maximum illuminance below 2 lx). Each tested system is designed to
faces and furniture. From 20th of March to April 18, 2019, nine users represent a peculiar indoor lighting condition emphasizing specific

Fig. 1. Experimental campaing: a)Test room and equipment during DL session; b) detail of the AS.

3
A. Bellazzi et al. Journal of Building Engineering 45 (2022) 103407

Fig. 2. The three lighting systems tested: a) DL; b) AS; c) LED.

lighting aspects. power. Participants tested this lighting system over a 3-h session.
This scenario represents the common artificial lighting condition
1. Daylighting (DL): the DL lighting system consists of two 1.1 × 1.6 m with static lighting intensity and distribution over the time and
low-emitting windows (τv = 54.3, g = 30.5, Ug = 0.6 W/m2K) on the within the room, potentially minimizing the discomfort risk related
South-East facade. The total window to floor area ratio is 12.5% and to illuminance and glare.
represents the minimum value allowed by Italian technical regula­
tions. To use the daylight hours to best advantage, experiments and
2.3. Configurations and participants
participants involving were planned, for the same day, in the
morning (4 h) and in the afternoon (3 h). This scenario represents the
The three working desks are positioned inside the room in 3 different
common natural lighting with a high variability over the time. It
configurations (Fig. 3):
represents a situation with a potential high glare risk and an optimal
colour rendering and visual appearance of objects.
• Ow: workstations toward the South-West wall (shoulders toward the
2. Artificial Skylight (AS): static artificial light coming from AS (2
door);
Coelux® LS modules 0,6 × 1,2 m with a 45◦ narrow angle). Partic­
• OF: workstations toward the South-West facade (gaze on the
ipants tested this lighting system over a 3-h session. The AS tested is
fenestrations);
composed by artificial light source modules, spectrally very similar
• OD: workstations toward the South-West facade (gaze on the door).
to the visible part of sunlight (CCT 5770 K), and of a nanostructured
material which recreates the Rayleigh scattering process (that
During the different test configurations, the nine participants divided
reproduce diffuse light) that occurs in the atmosphere. Each module
in groups of three always occupy the same workstation: PD (near the
provides a luminous flux of 3300 lm with a maximum electric energy
north-east façade and entrance door), PM (in the middle of the room) and
consumption of 90 W. The modules are dimmable with a DALI-based
PW (near the South-West wall).
protocol and a smartphone app but according to the scope of the
The position of the participants and the three configurations were
research this function was not used. The system produces a uniform
chosen in order to consider peculiar lighting conditions for DL and AS
45◦ tilted beam imitating the outdoor condition typical of a cloudless
lighting systems:
day. An intense yellowish light component (CCT 5000 K) mimics the
direct sunlight and creates the shadows of illuminated object. At the
• For DL, configuration OW and OD ensure vertical lateral lighting
same time, the diffuse skylight component creates blue-toned
while configuration OF guarantees vertical front lighting with direct
shadows. During the test, modules are placed above one working
sun beam;
desk (see 2.3) and sized to ensure on it at least 300 lx (minimum
• For AS, PW workstation ensures diffuse zenithal light, PM direct “Sun”
value for office tasks according EN 12464-1) with diffuse light. Direct
light and PD diffuse ambient light. Specifically, the rotation in PM
light beams are oriented towards the centre of the room with
allows to assess the effects on performances and perception in rela­
orientation SW-NO. This scenario represents a hybrid lighting con­
tion to direct light geometry changes.
dition between natural and artificial lighting with static light in­
tensity over the time but high variability and discomfort risk related
On the contrary, the regular distribution of LED lighting elements
to illuminance, glare and colour rendering within the room.
keeps the lighting condition quite constant among workstations and
3. Artificial lighting (LED): for the artificial lighting system 8 LED
configurations.
headlamps (14W power, φ = 1300 lm, CCT 4000K and CRI>90)
Participants carry out activities such as reading on the screen or
distributed over a grid with dimension 1.2 × 1.2 m have been
paper, typing, writing and all the white coloured desks are equipped
selected. The LED lighting system was designed with Dialux® soft­
with the same furniture and hardware model: mouse, keyboards and 24′′
ware to ensure 300 lx at the desk level (+0.80 m above floor) with
black monitors set on all three workstations with the same settings of
homogeneous distribution over the room, operating at the maximum
luminosity, contrast and light temperature.

Fig. 3. Workstation configurations: a) OW – gaze on the wall; b) OF - gaze on the fenestrations; c) OD – gaze on the door.

4
A. Bellazzi et al. Journal of Building Engineering 45 (2022) 103407

2.4. Monitoring setup Table 2

Technical data – Luxmeters, Spectroradiometer and Videophotometer.
The monitoring setup is designed and applied to monitor the overall Sensors Num. Measuring Relative Signal Output
environmental data (IEQ, Indoor Environmental Quality) with a focus range spectral
on the lighting data. Fig. 4 highlights the lighting monitoring setup. A response
fixed grid with 9 luxmeters (dotted line) is installed: 3 measuring points Luxmeters low 6 20 ÷ 2′ 000lx <8% of the 4 ÷ 20 mA
are located in the middle of each working plane and 6 other luxmeters range CIE spectral
are fixed on dedicated supports with the same height of the working luminous
efficiency V
plane (+0.75 m). These luxmeters have different ranges depending on (ƛ)
whether they are directly illuminated by natural or artificial solar ra­ Luxmeters high 3 200 ÷ <8% of the 4 ÷ 20 mA
diation (Table 2); measures are taken every 10 s and averaged every range 20′ 000lx CIE spectral
minute. Other 12 measures are recorded every hour with a Konica luminous
efficiency V
Minolta T-10 [22] portable luxmeter (Fig. 4 with “Ο”). The portable
(ƛ)
luxmeter is used also in vertical mode paired with a LMK Mobile Air Portable luxmeter 1 0.01 ÷ <6% of the LCD (Liquid
Videophotometer [23] with 180◦ fisheye lens (4.5 mm focal length) at 29′ 900lx CIE spectral Crystal Display)
eyes height (1.20 m above floor) oriented towards the user’s point of luminous displayed value
view. On days representative of clear, cloud and variable sky, indoor efficiency V
(ƛ)
light spectrum measures are collected with a DeltaOhm HD30.1 spec­ Spectroradiometer 1 380–780 nm 5% for
troradiometer. Finally, a meteorological station is installed on the spectrum
rooftop near the test room with two luxmeters to measure beam and Videophotometer 1 1:30.000 14 Bit RAW image
diffuse external illuminance. HDR image - data as
(High uncompressed
Dynamic Bayer structure
2.5. Experimental analysis Range)
External luxmeters 2 0-150′ 000 lx <4% of the 4 ÷ 20 mA
The experimental campaign is aimed at assessing the lighting satis­ CIE spectral
faction and perception of users by combining the DL, AS and LED luminous
efficiency V
lighting systems with the OW, OF and OD workstation configurations.
(ƛ)
The evaluation is carried out based on the following analysis:
1
CIE: Commission Internationale de l’Eclairage – International Commission on
Illumination.

Fig. 4. Visual Monitoring asset.

5
A. Bellazzi et al. Journal of Building Engineering 45 (2022) 103407

• calculation, based on measured data, of the visual comfort indicators Table 4

regarding the level of illuminance, the glare and the circadian Considered glare indicators details.
rhythm. Indicator Formula Parameters description Glare indicator/
• analysis of the user’s perception of lighting through the colour perception scale
perception test and questionnaires; UGR Lb = background luminance 10 - Imperceptible
• correlation analysis between measured data and subjective [cd/m2]; 13 - Just perceptible
perception; Ls = glare source luminance in 16 – Perceptible
• visual performances comparison of lighting systems. the direction of the observer’s 19 - Just acceptable
eye [cd/m2]; 22 - Unacceptable
ω = glare source surface – axis 25 - Just uncomfortable
of sight solid angle [sr]; p = 28 - Uncomfortable
2.5.1. Illuminance Guth position index.
Illuminance (E) is the physical quantity that describes the amount of DGP Ev = vertical illuminance at the ≤0.35 - mostly not-
eye of the observer [lx]; perceived
light in a specific point of a given surface; a good level of visibility may
Lsi = glare source luminance 0.35 ≤ 0.40 - perceived
be ensured by an adequate value of illuminance to accomplish a task [cd/m2]; but mostly not-
[21]. The illuminance level and its distribution on the task and sur­ P = position index; disturbing
rounding area were monitored in accordance to EN 12464 [24] to verify ωsi = solid angle subtended by 0.40 ≤ 0.45 - perceived
the visual comfort and the possibility to carry out the visual task. In the glare source I [sr]; c1 = and often disturbing
5.87 ⋅ 10− 5; >0.45 - perceived and
particular, illuminance target and uniformity are verified. The illumi­ c2 = 9.18 ⋅10− 2; often intolerable
nance uniformity factor, U0, expresses the ratio between the minimum, a1 = 1.87;
Emin, and mean, Em, illuminance values of a surface. The measurements c3 = 0.16.
in Table 3 are carried out and the limits for office activity are verified for CR Ls = luminance of the 0.33 ≤ CR ≤ 3.00 -
investigated area [cd/m2]; comfortable
each task area.
Lb = background luminance
[cd/m2].
2.5.2. Glare aspects
The previous glare indicators are calculated on the basis of the HDR images
Glare is the visual discomfort produced by an unsuitable range or
collected by the video photometer, then processed with the LMK Labsoft soft­
distribution of high luminance or to extreme contrast in luminance.
ware [30]. Ev is measured with a portable Illuminance meter.
According to the European technical Standards ([24,25]), UGR – Unified
Glare source pixels are identified when their luminance value exceeds 4 times
Glare Rating is the reference indicator for the assessment of glare in the reference luminance value (Lref). Lref alternatively refers to the average
spaces with artificial lighting sources, while, for daylighting, the DGP – fisheye field of view luminance and to the average task area luminance.
Daylighting Glare Probability [26] is the reference indicator for glare
assessment due to the light entering from the windows and skylight. In 2.5.3. Circadian rhythm
accordance with EN 12665 [27], glare can be caused by extreme con­ Natural and artificial lighting (depending on spectrum output) has
trasts and it can be calculated with the Contrast Ratio (CR) between the other effects on human physiology and behaviour: for example, blue
luminance of an object and the background. light affects the circadian rhythm ([31,32]) and, as a consequence, the
The considered glare indicators are detailed in Table 4, reporting nocturnal secretion of melatonin [17]. To evaluate the impact of light
also the glare indicator/performance scale according to UNI 11665 for sources on melatonin production, the Circadian Stimulus (CS) parameter
UGR [25], to EN 17037 for DGP [26] and to Refs. [28,29] for CR. is used to measure the efficiency of the spectral irradiance to the cornea.

Table 3
Indicators and limits verified and graphical explanation (Example of the measure area in PM in Configuration OW).
Indicator Definition Limit
Ep Punctual illuminance in the 500
middle of the desk lx
Em,t Mean illuminance of the task 500
area calculated as the mean of lx
3 values: Em,t = (Ep + E1+E2)/
3
U0,t Uniformity of the task area, 0.60
calculated as the ratio between
the minimum and the mean
illuminance of the task area.
U0,t = Emin,t/Em,t,
U0,s Uniformity in the immediate 0.40
surrounding area, considering
a band of at least 0.5 m and
calculated as the ratio between
the minimum illuminance of
the task area and the mean
illuminance of the surrounding
area. U0,s = Emin,t/Em,s
U0,b Uniformity of the background 0.10
area, considering a band of at
least 3 m and within the limit
of the space and calculated as
the ratio between the
minimum illuminance of the
task area and the mean
illuminance of the room. U0,b
= Emin,t/Em,b

6
A. Bellazzi et al. Journal of Building Engineering 45 (2022) 103407

It varies between 0.1 and 0.7, where the former value is the minimum Table 6
circadian stimulus threshold, while the latter is the saturation level. Q3 Hourly questionnaire – Lighting questions.
Conventionally, a value equal to or greater than 0.3 is considered During the last hour, the light was:
optimal [33]. The offline version of the Lighting Research Center’s CS
Too bright Bright Slightly Neutral Slightly Dark Too dark
Calculator Web tool [34] was used to evaluate this indicator, using the bright dark
measured vertical illuminance to the eye of the observer and the spectral
− 3 − 2 − 1 0 1 2 3
profile of the light sources. The default value of 0.5 of the Macular During the last working hour, how did you perceive the glare from frontal/
Pigment Optical Density (MPOD) was used. right/left/top/bottom direction?
During the last working hour, how did you perceive the glare from desk/screen/
2.5.4. Questionnaires ceiling/floor/windows elements?
Perfectly Slightly Fairy difficult to Very Intolerable
During their stay in the testing facility participants have completed 4
tolerable difficult tolerate difficult
different questionnaires, as explained in Table 5. to to
Questions related to Q3 and Q4 are formulated in accordance with tolerate tolerate
EN ISO 10551 Annex C [35], with a bipolar or unipolar scale. About Q3 1 2 3 4 5
(Table 6), in question 1, the lighting level is defined by a perception During the last hour, your working activity was:
Perfectly Slightly Fairy difficult to Very Intolerable
scale (7-point bipolar scale). Question 2 and 3 are about the glare based tolerable difficult tolerate difficult
on a tolerance scale (5-point one-pole scale). Question 4 is built on the to to
working appraisal based on a tolerance scale (5-point one-pole scale). In tolerate tolerate
question 5, the colour perception of the working station is defined by a 1 2 3 4 5
During the last hour, the colour of your working station was:
perception scale (7-point bipolar scale). Finally, question 6 is an
Too cold Cold Slightly Neutral Slightly Warm Too warm
open-ended question about the mood. cold warm
Q4 has the same questions of Q3 plus some others. In particular, in − 3 − 2 − 1 0 1 2 3
question 7 the visual well-being is defined by a perception scale (5-point During the last working hour, how did you feel
one-pole scale). In question 8, the compilation of the colour test is During the last working hour, how did you feel?

defined by a tolerance scale (5-point one-pole scale), as in Table 7.

Questions 9 and 10 were useful to understand the main cause of (Bezold–Brucke (B–B) hue-shift) [36]. The perception of colours is, in
discomfort and the most important feature the users would like to parallel with illuminance and glare, an important element in deter­
change. mining visual comfort. For the purposes of visual comfort, light sources,
particularly artificial light sources, must illuminate objects and surfaces
2.5.5. Colour perception in such a way as to alter their original colours as little as possible.
An interrelation between luminance and colour perception can be During this experimentation, the Farnsworth-Munsell 100 Hue
observed: Bezold and Brücke independently discovered that a variation Colour Vision Test - FM 100 test [37] was carried out to verify the users
in luminance can alter the tone, thus changing its colour appearance colour perception with different lighting sources. Users had to execute
the test at the end of every testing session, during the morning and the
afternoon, in 5 min. At the same time, in the questionnaire, the users had
Table 5 to express a preference about the colour perception.
Questionnaire type and schedule.
Type of Data collected Timing 3. Results
questionnaire

Q1 General user data age, gender, colour of the Once at the 3.1. Monitoring-based data analysis
eyes, any visual problems, beginning of the
information about working monitoring
habits (such as the kind of campaign
3.1.1. Illuminance level distribution
light they prefer to work with, Analysing the recorded DL values of illuminance, DL-OW and DL-OD
natural, artificial or a show a similar trend with a very high maximum level of Ep (around 1600
combination of them) lx) during the morning, specifically in PD and PW that are in front of the
Q2 User initial quality and number of hours At the beginning of
windows (Fig. 5-a). Em,t and Em remain above the limit threshold of 500
conditions slept the previous night, type each daily session
of clothing worn lx (Fig. 5-b) but a low uniformity of U0,t and U0,s is still highlighted,
Q3 Hourly 14 questions about IEQ At the end of each specifically in PD and PW, because the direct solar radiation entering the
questionnaire (acoustic, lighting, IAQ – working hour room causes very sunny zones and darker ones on the desks (Fig. 5-c-d).
Indoor Air Quality, Regarding DL-OF (Fig. 6), once again, there is no uniformity on the desk
temperature etc.) of which 4
about visual quality:
area due to the screen shadow: in some days U0,t,PD and U0,t,PW are al­
brightness, glare and colour ways below the threshold value (0.6).
perception, and 1 about the With AS, in all the three configurations there is a very low uniformity
mood in the room because PD and PW are the darkest positions both in terms of
Q4 Ambient quality 12 questions about the global At the end of each
Ep and in terms of Em (about 200 lx), while PM is the brightest one, with
questionnaire well-being during the whole daily working
session: 4 questions about the session monitored illuminance values of one order of magnitude higher. The
light quality (brightness, difference between the configurations is that in AS-OF and AS-OD, the
colour, global visual well- uniformity on the desk area is equal to 1 (uniformly bright in PM and
being and difficulty to uniformly dark in PD and PW) while in AS-OW there is a low uniformity in
complete the “Farnsworth –
PM because of the screen shading. The trend of the illuminance value
Munsell colourtest”) and the
remaining questions about the recorded by the different sensors is shown in Fig. 7.
appraisal of working
conditions, the appraisal of 3.1.2. Glare and contrast
the environment and the main
Glare is assessed using DGP for DL, UGR for LED and both of them for
cause of discomfort if any and,
in this case, the change needed
AS, as illustrated in Table 8. Maximum (max), minimum (min) and

7
A. Bellazzi et al. Journal of Building Engineering 45 (2022) 103407

Table 7
Q4 Ambient quality questionnaire – Lighting questions.
How do you evaluate the visual well-being as a whole?

Comfortable Slightly uncomfortable Uncomfortable Very uncomfortable Much uncomfortable

1 2 3 4 5
How do you consider the compilation of the colour test?
Perfectly tolerable Slightly difficult to tolerate Fairy difficult to tolerate Very difficult to tolerate Intolerable
1 2 3 4 5
What was the major cause of discomfort?
Which of the following features would you have liked to change?

Fig. 5. DL- OW illuminance and uniformity trend – 22nd of March.

Fig. 6. DL-OF illuminance and uniformity trend – 18th of April.

averaged values of each position in the 3 different configurations are quite below the previous threshold, in all the configurations. Only DL-PD
listed for DL. For AS and LED, there are not the maximum, minimum and and DL-PW show glare effects in the early period of the monitoring
average columns cause they are artificial systems and the output is a session (9.30–10.30 a.m.), with direct sunlight entering the room. The
constant value that varies only for each position and configuration. highest values are recorded in DL–OF–PW and PD, when light is frontal.
As reported in Table 8, with DL, in all the three configurations, the As a result, only 3% of cases shows glare effects. For AS, DGP values are
DGP values are most of the time below the glare perception threshold very low, always below 0.2, therefore very far from the threshold value;
(0.35). In particular, DL-PM generally shows the lowest DGP values, the UGR values exceed the threshold only in AS-OW with one value

8
A. Bellazzi et al. Journal of Building Engineering 45 (2022) 103407

Fig. 7. AS: Illuminance distribution With LED-OW (Fig. 8), the illuminance distribution is uniform with a range of 400–500 lx over all the reference surfaces.

Fig. 8. LED: Illuminance distribution.

higher than “13-Just perceptible” (13.64 in AS-OW-PD) and one value perceived. So with AS, glare is perceived for front light. LEDs respect the
higher than “16-Perceptible” (16.13, the highest one, in AS-OW-PM): all limits with the highest value equal to 9.35, so neither the 10-impercep­
the others values are below 13 and in AS-OF and AS-OD no glare is tible threshold is recorded.

9
A. Bellazzi et al. Journal of Building Engineering 45 (2022) 103407

Table 8 lighting, glare is perceived from the windows or from the desk in DL. For
DGP and UGR values for the considered lighting scenarios – In bold type the AS, the Zenithal light (Fig. 9 d)) is critical too, with 44% of the cases with
values higher than the threshold value. CR values too high with respect to the threshold value. Zenithal lighting
DGP UGR gives always optimal CR values for LED. Side light incidence (Fig. 9 b))
DL AS AS LED
always ensures good CR values.

max min avg

3.1.3. Circadian rhythm
Ow In relation to the characteristics of the lighting system, during DL
PD 0.364 0.175 0.229 0.189 13.64 7.50 monitoring, the CS value is variable (Fig. 11) while AS and LED provide
PM 0.228 0.184 0.207 0.190 16.13 6.08
PW 0.339 0.175 0.230 0.174 6.36 3.35
a constant illuminance level and CS value (Table 9).
OF Analyzing DL, the maximum CS value is always near the saturation
PD 0.405 0.222 0.279 0.181 7.16 2.39 value (0.7) with a peak around 10:30 a.m. and then decrease but keeping
PM 0.263 0.195 0.239 0.189 11.01 1.26 a value above 0.3 in 95% of the cases. The lowest values are recorded in
PW 0.503 0.216 0.287 0.172 6.23 9.27
DL-OW and DL-OD in PM and PD: observing the graphs in Fig. 11, it can be
OD
PD 0.292 0.174 0.200 0.190 11.97 1.37 observed that in DL-OW the values are below 0.3 only after 14:30, while
PM 0.203 0.173 0.186 0.189 10.52 9.35 in DL-OD the trend is different due the presence of a partially cloudy
PW 0.292 0.177 0.205 0.170 6.80 6.19 weather instead of a clear sunny weather recorded in the DL-OW and DL-
OF configurations. On the contrary, with LED, the 0.3 threshold is never
reached. AS stands in a middle way of previous scenarios: in the AS-OW
CR assessment results are illustrated in Fig. 9. Lighting pattern cases
and AS-OF, CS is higher than 0.3 in Pw and PM, while it is always lower in
are grouped in relation to the geometry of the incident light with respect
PD. In AS-OD, that is with AS behind the field of view, the CS is never
to the position of the user (in relation to DL, CR is calculated for the
reached at any workstation.
highest DGP value recorded for the considered configuration and
workstation):
3.2. Questionnaire-based data analysis
a. Front lighting: light rays come principally from in front of the user
(cases DL-OF-PD/PW and AS-OW-PM); 3.2.1. Hourly-based user feedbacks (Q3 questionnaire)
b. Side lighting: light rays come principally from the left or from the During DL tests, the three configurations confirm the data moni­
right of the user (DL-OW-PD/PW and AS-OF-PM); tored: the users perceive the desk to be properly illuminated for more
c. Rear lighting: light rays come principally from behind the user (DL- than 50% of the time in all the workstations. The users in DL-PD and DL-
OD-PW/PD and AS-OD-PM) PW have an excessive bright perception during the morning
d. Zenithal lighting: light rays come principally from above (LED-OW- (9:30–11:30), especially in DL-OF when the gaze of the user is towards
PW/PM/PD and AS-OW/OF/OD-PW). the windows. On the contrary, in DL-OF-PM there is a higher perception
of right illuminance with respect to DL-OW-PM. For each lighting system,
The Field of View (FoV) is subdivided in five sector: frontal, right, the graphs of Figs. 12–14 show the correlation between the lighting
left, top and bottom, based on the setting of the beginning of the pe­ perception vote and the mean measured hourly lighting value. The score
ripheral vision zone considering a virtual cone with a 60◦ aperture angle is averaged across the three configurations (OW, OF and OD).
aligned with the fisheye image centre considered as the central zone With DL, neutral perception increases from almost 50% in DL-PD to
[38] (Fig. 10). almost 80% in DL-PW. Perception is neutral even if the illuminance level
The most critical CR values for DL and AS are due to front or rear is below or higher than 300–500 lx. With the artificial systems, the
lighting (Fig. 9 a) and c)): with AS 30% of CR values exceed the illuminance level is obviously constant but there is a difference between
threshold, with DL 55% of CR values are too high or too low. In the front AS and LED. Within AS, a too dark lighting level is confirmed in AS-PD, a

Fig. 9. CR assessment for different light incidence on workplaces: a) – Front, b) – Side, c) – Rear, d) – zenithal. Figure a), b) and c) do not consider LED, figure d) does
not consider DL.

10
A. Bellazzi et al. Journal of Building Engineering 45 (2022) 103407

Fig. 10. Overlay of FoV subdivision on a false colour luminance image captured with the 180◦ fisheye lens. (For interpretation of the references to colour in this
figure legend, the reader is referred to the Web version of this article.)

Fig. 11. Hourly mean CS values for DL for the three different configurations. Values are reported in blue for PD, in red for PM, and in green for PW. (For interpretation
of the references to colour in this figure legend, the reader is referred to the Web version of this article.)

too high lighting level is confirmed in AS-PM and an almost acceptable

Table 9
level is illustrated in AS-PW. It is highlighted that the trend is toward 1
Circadian Stimulus value for AS and LED (in bold type values below the refer­
(“slightly bright”) in AS-PD, even if the illuminance is equal to 0–199 lx,
ence CS value of 0.3).
while in AS-PM and AS-PW the perception corresponds to the illuminance
DL AS LED recorded: bright in AS-PM and neutral in AS-PW.
max min avg Within LED, illuminance levels are evenly spread across the room
OW and the reference class is always 300–499 lx: except for LED-PW with a
PD 0.645 0.251 0.500 0.135 0.169 neutral perception, in LED- PM and PD the trend is towards bright or too
PM 0.544 0.251 0.461 0.303 0.126 bright.
PW 0.640 0.343 0.518 0.407 0.167 In relation to user feedback, a global glare perception score for each
OF
PD 0.659 0.421 0.565 0.178 0.206
lighting system is considered. The vote is calculated by averaging, for
PM 0.595 0.427 0.543 0.415 0.155 each hour and for each user, the score assigned to each viewing direction
PW 0.674 0.431 0.575 0.326 0.199 (Fig. 15).
OD Comparing the lighting systems, the mean highest vote is in AS-OW-
PD 0.639 0.276 0.431 0.150 0.213
PM. Analysing the different lighting systems separately, the highest glare
PM 0.558 0.221 0.372 0.194 0.165
PW 0.636 0.338 0.484 0.280 0.187 perceptions are reached in DL-OF-PD, AS-OW-PM and LED-OW-PD, all

11
A. Bellazzi et al. Journal of Building Engineering 45 (2022) 103407

Fig. 12. Correlation between user perception of brightness and data monitored with DL.

Fig. 13. Correlation between user perception of brightness and data monitored with AS.

Fig. 14. Correlation between user perception of brightness (votes) and data monitored with LED.

12
A. Bellazzi et al. Journal of Building Engineering 45 (2022) 103407

Fig. 15. Global glare perception vote, from 1 = perfectly tolerable to 5 = Intolerable.

Fig. 16. “Frontal” field of view area glare perception score, from 1 = Perfectly tolerable to 5 = Intolerable.

situations where light is coming frontally with respect to the user. OF-PD but the highest standard deviation is recorded in DL- OW-PD (dv.st.
Analysing the “frontal” sector score, the highest values are recorded 1.29): due to the large variation of the luminous values during the day,
(Fig. 16): AS-PM reaches a value higher than 3.5 and, on the whole, the working conditions are perceived as very hard to tolerate or intolerable
PD working desk results to be the most unfavourable with all the lighting during the morning but tolerable during the afternoon. With AS, the
systems showing values higher than 3 in DL-OF, AS-OW, AS-OD and LED. difference is perceived for (the) workstation and AS-PW is more
About the hourly work hardship (Fig. 17), with DL, for all the 3 comfortable in all the configurations. With LED, the perception is similar
configurations, the mean vote is between 2 in DL-OD-PW and 3.2 in DL- in all the workstations and the votes are concentrated between 2 and 3.5

Fig. 17. Hourly work hardship, from 1 = Perfectly tolerable to 5 = Intolerable.

13
A. Bellazzi et al. Journal of Building Engineering 45 (2022) 103407

in the most of cases. within AS for each of the considered aspects (the result was predictable
The analysis of hourly responses about mood does not give peculiar since the workstation, the farthest from AS modules, receive only the
results in relation to the lighting scenario. Indeed, most of the answers reflected light from room surfaces) and confirm that AS mean perfor­
are “relaxed” for each considered scenario. mances of “useable” workstations (PM and PW) are clearly better than
The analysis of hourly colour perception feedback (Fig. 18) is per­ LED.
formed considering the peculiar colour features of lighting scenarios. DL The second session-based user response regards the testing of colour
colour changes over the day so morning and afternoon feedbacks are perception, provided by the FM-100 test. There are two relevant test
treated separately. LED and DL light sources are chromatically and results: the absolute error and the Main Colour Confusion Area (MCCA).
evenly spread across the test room (black and white coloured) so the The absolute error is defined by the mean Total Error Score (TES):
workstation position is not considered as a variable. On the basis of the the higher the TES value, the higher the colour sequencing error.
above, DL colour is mostly perceived as neutral both in the morning and Generally, a TES value equal to or below 40 means an absence of colour
in the afternoon (45%), while the second frequent perception is warmer vision deficiency.
in the morning and colder in the afternoon. LED colour perception is Scenarios mean measured TES are 43.5 (±23.5) for DL, 42.4 (±19.8)
essentially slightly warm (70%), confirming the nominal CCT values of for AS and 30.5 (±12.6) for LED, respectively. Despite the low values,
the lamps (4000 K). On the contrary, AS colour perception is more they denote how LED allows a slightly better colour perception with
sensitive to the workstation position because the device emits light both respect to DL and AS. In particular, while in LED the 71% of TES values
at 3800 K (direct beam light) and at 30,000 K (diffuse “sky” light). The are below 40, in DL and AS, the ratio drops to 57% and to 52%,
subjective feedback analysis confirms the previous concept. In the AS-PM respectively.
workstation, characterized by direct lighting, the most frequent The MCCA identifies where, throughout the colour range, the user
perception is “slightly warm” (48%) and no “cold” perception is detec­ have the greatest difficulties in distinguishing colours: in this case,
ted, while in the AS-PW workstation, under diffuse light the “slightly MCCA is the hue colour sector of the Colour Vision Deficiency Type
cold” perception is the most frequent (52%) and neither “slightly warm” diagram showing the highest error. The reference colour hue sectors are:
and “warm” feedbacks are provided. The AS-PD workstation perception R-RP (red to red/purple), RP-P (red/purple to purple), P-PB (purple-
wavers halfway between the previous ones. purple/blue), PB-B (purple/blue-blue), B-BG (blue-blue/green), BG-G
(blue/green-green), G-GY (green-green/yellow), GY-Y (green/yellow-
3.2.2. Session-based users’ feedback (Q4 questionnaire) yellow), Y-YR (yellow-yellow/red) and YR-R (yellow/red-red).
The main results of the “Q4-Evaluation of the ambient quality” filled Fig. 19 shows the MCCA for each lighting system. DL results are
at the end of each session are summarized in Table 10. grouped in “Morning” and “Afternoon” in order to take into account the
According to the analysis of the work hardship, which confirms the sky colour temperature variation during the day.
trend of the hourly analysis, DL-PD is perceived as “fairly difficult to
tolerate” and the trend in DL-PM and PW is toward “slightly difficult to 4. Discussion
tolerate”. AS-PW is perceived as “slightly difficult to tolerate” while AS-
PD and PM exceed “fairly difficult to tolerate”. LED-PD is perceived as The first main goal of this research is the characterization of the AS
“very difficult to tolerate” while LED-PM and PW are between “slightly with respect to DL and LED, by analysing and comparing objective in­
and fairly difficult to tolerate”. The visual well-being is perceived as dicators and subjective analyses. About the illuminance level, the
uncomfortable with all the lighting systems: LED-PD shows the highest analysis of the monitored data shows non-uniform light zones in AS, like
value, almost “very uncomfortable”, while AS-PW and LED-PW shows the DL. With DL, the low uniformity depends on the hour of the day and
lowest one. AS-PD and PM values are greater than 4. The most uniform varies during the day in the same workstation; with AS, it depends on the
perception is recorded with DL, with values around 3, in all working orientation of the lighting system and it is fixed for specific workstations.
stations. With LED, light is uniformly distributed.
In Table 10 the major cause of discomfort is “None” with DL, fol­ Analysing the correlation between monitored data and users’ per­
lowed by noise, “noise” with AS and “glare” with LED. A more uniform ceptions with DL, the room is perceived well illuminated, which corre­
light, as LED, permits to identify other discomfort causes: the major sponds to the illuminance level recorded that nearly exceeds 300 lx; too
cause of discomfort “None” is 12% with LED while it is equal or more high illuminance levels monitored during the morning are confirmed by
than 30% with DL and AS. It is important to highlight that the partici­ the − 1 and − 2 users’ votes. With AS, users’ preferences confirm the non-
pants knew that it was an experiment about lighting quality and this uniformity of the lighting distribution, with a AS-PD perceived as too
could have influenced their answers. dark, AS-PM as too bright and AS-PW as the most neutral working station.
AS mean values gives better results than LED about lighting levels, With LED, there is a discrepancy between monitored data and user
colour perception and visual well-being. On the contrary, LED gives perception: while the monitored illuminance levels are in the optimal
better results than AS in relation to work hardship. However it must be range of 300–500 lx, users perceived the room as “too bright”.
underline while LED lighting conditions are uniform within the room, Focusing on glare aspects, with DL and AS, nearly all the DGP in­
with AS this is not true. In particular the PD performances are the worst dicators measured are below the threshold value (0.35) and the main

Fig. 18. Colour perception feedback analysis: DL and LED scenario (left), AS scenario (right). (For interpretation of the references to colour in this figure legend, the
reader is referred to the Web version of this article.)

14
A. Bellazzi et al. Journal of Building Engineering 45 (2022) 103407

Table 10
Q4-Evaluation of the ambient quality questionnaire results.
Room lighting level (− 3 = too Colour (− 3 = too cold, Work hardship (1 = perfectly Visual well-being (1 = comfortable, 5 Major cause of
bright, 3 = too dark) 3 = too warm) tolerable, 5 = intolerable) = much uncomfortable) discomfort

DL Mean |0.31| |0.14| 2.75 2.92 33% None

PD − 0.71 − 0.06 3 3.29 22% Noise
PM − 0.24 − 0.06 2.88 2.94 20% Glare
PW 0.00 0.29 2.35 2.53 12% Air quality
10% Air Temp.
2% Relative
Humidity
AS Mean |0.71| |0.65| 3.04 3.57 35% Noise
PD 1.00 − 0.75 3.50 4.38 30% Glare
PM − 1.00 0.57 3.71 4.14 30% None
PW 0.13 − 0.63 2.00 2.25 4% Air quality
0% Air Temp.
0% Relative
Humidity
LED Mean |0.89| |0.78| 2.89 3.78 35% Glare
PD − 1.67 0.67 4.00 4.67 24% Noise
PM − 0.67 0.67 2.67 3.67 12% Air Temp.
PW − 0.33 1.00 2.00 3.00 12% None
12% Relative
Humidity
5% Air quality

Analysing the room lighting level and colour, DL is perceived as the most neutral lighting system in all the working stations, with values between − 1 and 0. With AS,
AS-PW is the more neutral working station, while AS-PD and PM are perceived as “slightly dark/slightly warm” and “slightly bright/slightly cold”, respectively. With the
LED system, all the workstations show a cold light perception with PD showing the coolest light perception.

Fig. 19. MCCA results distribution.

Observing the graph, B-BG and BG-G are the most
critical colour hue sectors. Focusing on specific sce­
narios, DL-Afternoon shows errors related to warmer
hues (Y-YR, YR-R) with respect to DL-Morning. With
LED, most of the errors are green-related (B-BG, BG-
G, G-GY), while AS shows an error distribution
similar to DL-Morning.

glare scores expressed by the users in the questionnaires are “1-Perfectly • for DL the new “Perfectly tolerable” score is expressed by 85% of
tolerable” or “2-Slightly difficult to tolerate”. Glare is measured with users and “Not perceptible” DGP indicator is recorded in 96% of
front light source as to the field of view, both for DL and AS; for DL, cases;
however, glare is perceived also in PM, therefore not only in worksta­ • for AS the new “Perfectly tolerable” score is expressed by 77% of
tions located in front of the windows. In Table 11 and Table 12 for DL users and “Not perceptible” DGP indicator is recorded in 100% of
and AS scenarios, respectively, the “2-Slightly difficult to tolerate” cases;
scores are grouped within “1-Perfectly tolerable” in order to have
comparable 4-step glare preferences as for DGP indicators. The results In relation to the previous data, the personal perception score and
are now more comparable: the DGP classification provide similar results making DGP classification
feasible to predict subjective glare sensation. However, DGP values tend
to underestimate real glare perception, both in DL and in AS scenarios:

Table 11
User glare score/DGP rating matrix with aggregation of “perfectly tolerable” and “Slightly difficult to tolerate” users perception ratings for DL scenario.
DGP

Mostly not perceptible Perceptible/Moderate Disturbing Intolerable %

cases <0.35 0.35-0.40 0.40-0.45 >0.45

Glare score Perfect and slightly difficult to tolerate <2.5 132 3 0 1 84.5%
Fairly difficult to tolerate 2.6–3.5 23 0 2 0 15.5%
Very difficult to tolerate 3.6–4.5 0 0 0 0 0.0%
Intolerable >4.5 0 0 0 0 0.0%
% 96.3% 1.9% 1.2% 0.6%

15
A. Bellazzi et al. Journal of Building Engineering 45 (2022) 103407

Table 12
User glare score/DGP rating matrix with aggregation of “Perfectly tolerable” and “Slightly difficult to tolerate” users’ perception ratings for AS scenario.
DGP

Mostly not perceptible Perceptible/Moderate Disturbing Intolerable %

cases <0.35 0.35-0.40 0.40-0.45 >0.45

Glare score Perfectly and slightly difficult to tolerate <2.5 62 0 0 0 76.5%

Fairly difficult to tolerate 2.5–3.5 19 0 0 0 23.5%
Very difficult to tolerate 3.5–4.5 0 0 0 0 0.0%
Intolerable >4.5 0 0 0 0 0.0%
% 100.0% 0.0% 0.0% 0.0%

in particular, the lower the DGP value, the more pronounced the dif­ perception score (− 0.78) is more neutral than the slightly cool percep­
ference between measurement and subjective glare perception. This tion value (− 1). In this sense, objective indicators and subjective re­
result is consistent with the previous study of Van Den Wymelenberg sponses provide comparable results.
and Inanici [39]. Focusing on the second main objective of this research, it emerges
The comparison of user mean glare score and UGR values for AS and that a single indicator is not always a proper representation of the visual
LED (Fig. 20) shows an overall correlation of 0.70. For AS, while DGP quality.
was always low with respect to the glare detection threshold (0.35), the In detail, the illuminance level for DL is respected, the color
UGR was higher than 13 (glare detection threshold) in two situations: perception is neutral, the DGP is less than 0.35 and the performance on
AS-OW-PM and AS-OW-PD. Comparing the output of the monitored data the circadian rhythm is excellent. On the other hand, work hardship is
with the questionnaires and considering that the DGP indicator and perceived by the users due to the extreme variability of the lighting
related scale are validated for higher Ev values than AS, UGR values are conditions and the glare problems in the first hours of the morning. It
more appropriate in describing the glare performance of this lighting was chosen to leave out the shielding systems for this experiment, with a
system. view to comparing the daylight beam with AS beam, however, the
Circadian stimulus effects could influence working productivity and, importance of a solar radiation control system has become evident. AS
indirectly, work hardship. The comparison of CS values with work has, in general, a good performance recorded in terms of circadian
hardship responses carried out in this experiment provides consistent stimulus, colour rendering and glare, but the output varies considerably
results only for DL, which, in the specific case, is the best performing with the workstation position and orientation. The main criticality is
scenario. With AS, the output depends on the working position, with related to the uniformity of illuminance, essentially due to the tight
good performance in AS-PW and PM and poor performance in AS-PD. beam of the light emitted. The performance of objective indicators was
With LED, the threshold value of 0.3 is never reached. However, the partly, though not in all cases, confirmed by the questionnaires: for
circadian stimulus exposure provided by lighting affects the sleep/wake example, glare is often perceived by the eye of the participants also from
cycle with medium/long term effects that cannot be clearly highlighted a position from where glare is not measured, especially under direct
based on a single day’s subjective analysis. In addition, work hardship lighting conditions. This suggests that the causes of the discomfort
perception is affected by a combination of different variables that are experienced by the participant are mostly due to the non-uniformity of
not only related to lighting (temperature, air quality, noise, etc.). illuminance and luminance of the surfaces that is detected by moving
The results of the analysis of the data relating to light colour are the gaze direction during the activity (while the glare indexes are
partially contradictory. In fact, while the best FM-100 TES values are detected in a static way by directing the fisheye lens of the video­
related to the LED lighting scenario, the best “neutrality” in colour photometer towards the front direction). Therefore, the performance of
perception is reached in DL. The same comparison demonstrates that AS the AS systems turns out to be very sensitive to the position of the system
is the second preferable solution for both analyses. In any case, even the with respect to the workstation, as demonstrated by Canazei [18].
worst FM-100 values (43.4) are very close to the reference value of 40 LED results analysis confirms the discrepancy between data and
(“None” colour vision deficiency threshold) and the worst Colour perception. Even if the best illuminance level (500 lux) and maximum

Fig. 20. UGR-Score scatterplot.

16
A. Bellazzi et al. Journal of Building Engineering 45 (2022) 103407

uniformity targets are reached, LEDs are perceived by users as the worst order to point out optimized dimming control strategies preserving
system in terms of visual comfort: it is resulted the system with the colour rendering and lighting quality as well. Further studies and tests
lowest circadian stimulus (≤0.2, Table 9). This can be related to the CCT on Artificial Skylight in different configurations and compared with
equal to 4000 K that could be too low for office activities, as explained in alternative technologies as sunlight tubes should be interesting to
other researches ([40,41]). investigate and to define design guidelines.
As reported in other studies, like [20,42], the need is highlighted for
a monitoring protocol for lighting, especially for the assessment of glare CRediT authorship contribution statement
[43].
A. Bellazzi: Conceptualization, Methodology, Investigation, Writing
5. Conclusions – original draft, Writing – review & editing. L. Danza: Supervision,
Writing – review & editing. A. Devitofrancesco: Conceptualization,
This paper describes a monitoring campaign with the aim to compare Investigation, Writing – review & editing. M. Ghellere: Methodology,
the lighting performance of an Artificial Skylight respect to daylight and Investigation, Data curation, Writing – review & editing. F. Salamone:
LED and users perception. Different configurations were chosen in order Data acquisition and Management, Hardware Setup, Writing – review &
to consider peculiar lighting conditions, testing diffuse or direct beam or editing.
frontal or lateral light. Several indicators have been measured and
compared with users perception to move a first step toward a definition Declaration of competing interest
of a global indicator.
This specific test highlighted that the performances of the artificial The authors declare that they have no known competing financial
skylight share some similarities with daylight solar radiation, such as the interests or personal relationships that could have appeared to influence
good performance of the CS and in colour rendering. In relation to other the work reported in this paper.
metrics (illuminance level, glare, contrast, user’s perception) the per­
formances are in line with those provided by other considered systems.
Acknowledgements
Given the impossibility of using lighting shielding, it is extremely
important to properly orient it and correctly install the right number of
This study was part of I-ZEB Project (2016–2019) which was funded
systems to optimize lighting and working well-being. AS could be used
by Lombardy Region and CNR (grant number no. 19366/RCC) with the
in an office but, due to its peculiar lighting features, such as tight
collaboration of 16 companies of building industry. We gratefully
lighting beam and fixed inclination, the workstation orientation and
acknowledge Paolo Cardillo for revising the English text of the
position have a great impact on visual performance (in particular in
manuscript.
glare indicators and subjective perception). PW, receiving diffuse zenith
light, ensures the best lighting performance. As far as measurements are
References
concerned, the most critical issue regards AS glare assessment. Although
DGP is designed for natural lighting and AS simulate it, its application on [1] S.G. Yong, J.H. Kim, Y. Gim, J. Kim, J. Cho, H. Hong, Y.J. Baik, J. Koo, Impacts of
AS shows a lower correlation with subjective glare perception than UGR building envelope design factors upon energy loads and their optimization in US
indicator, so the latter still remains preferable for glare evaluations. On standard climate zones using experimental design, Energy Build. 141 (2017) 1–15,
https://doi.org/10.1016/j.enbuild.2017.02.032.
the contrary, using HDR images captured with a fisheye lens allows to [2] P. Boyce, C. Hunter, O. Howlett, The benefits of daylight through windows.
better analyse the human field of view and capture the luminance range Literature Review Report Sponsored by: Capturing the Daylight Dividend Program,
perceived by the human eye; furthermore, the aggregation of the sub­ 2003.
[3] G. Vandewalle, P. Maquet, D.J. Dijk, Light as a modulator of cognitive brain
jective glare ratings expressed for different FoV sector is a possible way function, Trends Cognit. Sci. 13 (2009) 429–438, https://doi.org/10.1016/j.
for a coarse prediction of the glare indicator. Another point to investi­ tics.2009.07.004.
gate in further detail is the study of the main viewing direction and the [4] N.E. Klepeis, W.C. Nelson, W.R. Ott, J.P. Robinson, A.M. Tsang, P. Switzer, J.
V. Behar, S.C. Hern, W.H. Engelmann, The National Human Activity Pattern Survey
eye pupil size that could help to make the subjective glare more (NHAPS): a resource for assessing exposure to environmental pollutants, J. Expo.
consistent and better comparable to standard glare rating indicators. Anal. Environ. Epidemiol. 11 (2001) 231–252, https://doi.org/10.1038/sj.
The comparison between questionnaires and monitored data highlights jea.7500165.
[5] M. Canazei, W. Pohl, H.R. Bliem, M. Martini, E.M. Weiss, Artificial skylight effects
that compliance with the regulatory requirements does not always mean
in a windowless office environment, Build. Environ. 124 (2017) 69–77, https://doi.
visual quality if different indicators are considered separately. A holistic org/10.1016/j.buildenv.2017.07.045.
analysis is needed and a dynamic evaluation will better represents the [6] A. Yasukouchi, T. Maeda, K. Hara, H. Furuune, Non-visual effects of diurnal
performance of a dynamic system. In order to expand this experience, exposure to an artificial skylight, including nocturnal melatonin suppression,
J. Physiol. Anthropol. 38 (2019), https://doi.org/10.1186/s40101-019-0203-4.
the phase II of this experiment involves one hundred participants of [7] K. Konis, A novel circadian daylight metric for building design and evaluation,
different age, gender and training who were monitored in PM and PW Build. Environ. 113 (2017) 22–38, https://doi.org/10.1016/j.
workstations with OW configuration and artificial skylight as lighting buildenv.2016.11.025.
[8] V. Chulvi, M.J. Agost, F. Felip, J. Gual, Natural elements in the designer’s work
source. The focus was to investigate in depth the lighting (E and UGR/ environment influence the creativity of their results, J. Build. Eng. 28 (2020)
DGP) and colour perception during a shorter period of stay. 101033, https://doi.org/10.1016/J.JOBE.2019.101033.
There are several indoor environments where people work in [9] B. Meerbeek, P. Seuntiens, Evaluating the experience of daylight through a virtual
skylight, in: Proceedings of Experiencing Light 2014 : International Conference on
absence or with poor daylight and AS could be a valid solution [44] or as the Effects of Light on Wellbeing, 10-11 November 2014, Eindhoven, The
integrated system with natural and artificial lighting. In high efficiency Netherlands, Technische Universiteit Eindhoven. Document, Eindhoven, 2014,
buildings and in particular in Zero Energy Buildings the optimum pp. 58–61.
[10] H.J.A. de Vries, M.P.J. Aarts, M. Knoop, H. Cornelissen, Beneficial Non-visual
cooling, heating and lighting energy balance is the result of an holistic Effects of Daylight, Research into the Influential Parameters, in: Turkish National
approach combining lighting and thermal aspects and the correct design Committee on Illumination, 2009, pp. 155–161. https://research.tue.nl/en/publi
of the transparent building envelope [45]. However, the higher installed cations/beneficial-non-visual-effects-of-daylight-research-into-the-influ. (Accessed
13 November 2020). accessed.
lighting power density of AS respect to LED (7.5 W/m2 vs 4.6 W/m2) is
[11] K.C.H.J. Smolders, Y.A.W. de Kort, P.J.M. Cluitmans, A higher illuminance induces
well balanced by a better lighting quality, as demonstrated by the con­ alertness even during office hours: findings on subjective measures, task
ducted analysis. performance and heart rate measures, Physiol. Behav. 107 (2012) 7–16, https://
Focusing on the future work, the compliance potential of AS with doi.org/10.1016/j.physbeh.2012.04.028.
[12] H. Stefani, Oliver, Matthias Bues, Achim Pross, Sandra Mebben, Philipp Westner,
ZEB requirements should be investigated. In particular a monitoring H. Dudel, Rudolph, Moving clouds on a virtual sky affect well-being and subjective
campaign about the integration of AS with DL could be performed in tiredness positively, in: Light. Qual. Energy Effic. Conf. <2012, Hangzhou> Int.

17
A. Bellazzi et al. Journal of Building Engineering 45 (2022) 103407

Comm. Illum. -CIE-, Wien CIE 2012 "Lighting Qual. Energy Effic. Proc. 19 - 21 Sept. [30] LMK LabSoft, TechnoTeam bildverarbeitung GmbH (n.d.), https://www.tec
2012, China Vienna CIE, Hangzhou, 2012, pp. 113–122. hnoteam.de/product_overview/photometer_colorimeter/software/lmk_labsoft
[13] R. Rodriquez, A. Pattini, Tolerance of discomfort glare from a large area source for /index_eng.html (accessed June 1, 2021).
work on a visual display, Light. Res. Technol. 46 (2014) 157–170, https://doi.org/ [31] G.C. Brainard, J.R. Hanifin, J.M. Greeson, B. Byrne, G. Glickman, E. Gerner, M.
10.1177/1477153512470386. D. Rollag, Action spectrum for melatonin regulation in humans: evidence for a
[14] R.A. Mangkuto, M.B.C. Aries, E.J. van Loenen, J.L.M. Hensen, Analysis of various novel circadian photoreceptor, J. Neurosci. 21 (2001) 6405–6412, https://doi.org/
opening configurations of a second-generation virtual natural lighting solutions 10.1523/jneurosci.21-16-06405.2001.
prototype, Leukos 10 (2014) 223–236, https://doi.org/10.1080/ [32] K. Thapan, J. Arendt, D.J. Skene, An action spectrum for melatonin suppression:
15502724.2014.948185. evidence for a novel non-rod, non-cone photoreceptor system in humans,
[15] M. Stokkermans, Artificial Skylights: the Benefits of and Preference for Artificial J. Physiol. 535 (2001) 261–267, https://doi.org/10.1111/j.1469-7793.2001.t01-1-
Skylights and Sun Patterns - Master Thesis, 2011. 00261.x.
[16] P. Seuntiens, M. Van Boven, D. Sekulovski, Effect of skylight configuration and sky [33] M.G. Figueiro, K. Gonzales, D.R. Pedler, Designing with Circadian Stimulus, LD+A
type on the daylight impression of a room, WIT Trans. Ecol. Environ. 165 (2012), The magazine of the Illuminating Engineering Society, 2016.
https://doi.org/10.2495/ARC120051. [34] CSCalculator. http://www.lrc.rpi.edu/resources/CSCalculator_2017_10_03_Mac.
[17] D. Wang, Acute Effects of Blue-Enriched Light from an Artificial Skylight on xlsm, 2017. (Accessed 1 June 2021) accessed.
Alertness and Cognitive Performance during Daytime - Master Thesis, 2012. [35] EN ISO 10551:2019, Ergonomics of the Physical Environment - Subjective
[18] M. Canazei, M. Laner, S. Staggl, W. Pohl, P. Ragazzi, D. Magatti, E. Martinelli, P. Di Judgement Scales for Assessing Physical Environments, 2019.
Trapani, Room- and illumination-related effects of an artificial skylight, Light. Res. [36] W. von Bezold, Ueber das Gesetz der Farbenmischung und die physiologischen
Technol. 48 (2016) 539–558, https://doi.org/10.1177/1477153515577852. Grundfarben, Ann. Der Phys. Und Chemie. 226 (1873) 71–93, https://doi.org/
[19] L. Bellia, M. Musto, G. Spada, Illuminance measurements through HDR imaging 10.1002/andp.18732260904.
photometry in scholastic environment, Energy Build. 43 (2011) 2843–2849, [37] Farnsworth-munsell 100 HueColor vision test (n.d.), https://www.color-blindness.
https://doi.org/10.1016/J.ENBUILD.2011.07.006. com/farnsworth-munsell-100-hue-color-vision-test/ (accessed June 1, 2021).
[20] T. Kruisselbrink, R. Dangol, A. Rosemann, Photometric measurements of lighting [38] H. Strasburger, I. Rentschler, M. Jüttner, Peripheral vision and pattern recognition:
quality: an overview, Build. Environ. 138 (2018) 42–52, https://doi.org/10.1016/ a review, J. Vis. 11 (Issue 5) (2011), https://doi.org/10.1167/11.5.13.
j.buildenv.2018.04.028. [39] K. Van Den Wymelenberg, M. Inanici, A Critical Investigation of Common Lighting
[21] S. Carlucci, F. Causone, F. De Rosa, L. Pagliano, A review of indices for assessing Design Metrics for Predicting Human Visual Comfort in Offices with Daylight,
visual comfort with a view to their use in optimization processes to support LEUKOS - Journal of Illuminating Engineering Society of North America, 2014,
building integrated design, Renew. Sustain. Energy Rev. 47 (2015) 1016–1033, https://doi.org/10.1080/15502724.2014.881720.
https://doi.org/10.1016/j.rser.2015.03.062. [40] Q. Wang, H. Xu, F. Zhang, Z. Wang, Influence of color temperature on comfort and
[22] T-10A and T-10MA Illuminance Meters, Konica Minolta sensing [Internet], (n.d.), preference for LED indoor lighting, Optik 129 (2017) 21–29, https://doi.org/
https://sensing.konicaminolta.us/products/t-10a-t-10ma-illuminance-meters/ 10.1016/J.IJLEO.2016.10.049.
(accessed April 15, 2021). [41] M.G. Figueiro, B.C. Steverson, J. Heerwagen Phd, M.S. Rea, Circadian light and its
[23] LMK mobile air - TechnoTeam bildverarbeitung GmbH (n.d.), https://www.techno impact on alertness in office workers: a field study (n.d.), www.lrc.rpi.edu.
team.de/product_overview/photometer_colorimeter/products/lmk_mobile_air/ (Accessed 6 September 2021).
index_eng.html (accessed June 1, 2021). [42] N. Gentile, M.-C.C. Dubois, W. Osterhaus, S. Stoffer, C.N.D. Amorim, D. Geisler-
[24] EN 12464 -1 Light and Lighting - Lighting of Work Places - Part 1: Indoor Work Moroder, R. Jakobiak, A toolbox to evaluate non-residential lighting and
Places, 2011. daylighting retrofit in practice, Energy Build. 123 (2016) 151–161, https://doi.
[25] U.N.I. UNI, 11165 Light and Lighting - Indoor Lighting - Assessment of Glare with org/10.1016/J.ENBUILD.2016.04.026.
the UGR Method, 2005. [43] N.S. Shafavi, Z.S. Zomorodian, M. Tahsildoost, M. Javadi, Occupants visual
[26] EN 17037:2018 Daylight in Buildings, 2018. comfort assessments: a review of field studies and lab experiments, Sol. Energy 208
[27] EN 12665 - Light and Lighting - Basic Terms and Criteria for Specifying Lighting (2020) 249–274, https://doi.org/10.1016/j.solener.2020.07.058.
Requirements, 2018. [44] C.C. Gomes, S. Preto, Should an artificial window substitute a natural one? Adv.
[28] W. Osterhaus, Victoria Geelong, Recommended luminance ratios and their Intell. Syst. Comput. 607 (2018) 247–258, https://doi.org/10.1007/978-3-319-
application in the design of daylighting systems for offices, in: 36th Int. ANZAScA 60492-3_24.
Conf. Deakin Univ., Aust., 2002. [45] L. Belussi, B. Barozzi, A. Bellazzi, L. Danza, A. Devitofrancesco, C. Fanciulli,
[29] K. Parpairi, N.V. Baker, K.A. Steemers, R. Compagnon, The Luminance Differences M. Ghellere, G. Guazzi, I. Meroni, F. Salamone, F. Scamoni, C. Scrosati, A review of
index: a new indicator of user preferences in daylit spaces, Light. Res. Technol. performance of zero energy buildings and energy efficiency solutions, J. Build. Eng.
(2002), https://doi.org/10.1191/1365782802li030oa. 25 (2019), https://doi.org/10.1016/J.JOBE.2019.100772.

18