Solar Energy 207 (2020) 1434–1444

Contents lists available at ScienceDirect

Solar Energy
journal homepage: www.elsevier.com/locate/solener

Improving the daylighting performance of residential light wells by

reflecting and redirecting approaches
Antoine Bugeat a, *, Benoit Beckers a, Eduardo Fernández b
a
Urban Physics Joint Laboratory, Université de Pau et des Pays de l’Adour, E2S UPPA, Campus Montaury, 64600 Anglet, France
b
Centro de Cálculo, Universidad de la República, J. H. y Reissig 565, 11300 Montevideo, Uruguay

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

Keywords: The light well is a well-known architectural resource used in dense cities to provide daylight to the core of
Light well residential buildings. Due to its elongated shape, the propagation of light through it highly depends on the
Dense city properties of its surfaces. This study focuses on the improvements in daylighting performance achieved through
Daylighting
minor interventions in a conventional light well, namely, changing the surface optical properties and installing
Climate based daylight modelling
Radiosity method
mirrors to redirect the sunrays. The performance is evaluated using Climate-Based Daylight Modelling (CBDM).
The illuminance threshold is set to 100 lx for residential areas, where a small amount of daylight represents a
significant benefit for the inhabitants. The proposed radiosity algorithm allows a precise and fast calculation of
the interreflections, which play a crucial role. This study shows that small changes in diffuse reflectance have
significant effects on lighting conditions, and that the use of specular surfaces is also improving the results. Solar
redirecting devices are particularly efficient for periods and latitudes with low solar paths. They are providing
satisfying lighting conditions during the whole year.

1. Introduction the bottom depends on two main factors: the geometry of the light well,
which directly affects the amount of incoming light, and the optical
One of the major problems of urban densification is the reduction of properties of the inner surfaces, which influences the spread of light by
access to solar radiation and daylight (Compagnon, 2004; Mardaljevic inter-reflection. The size of the light well, and more specifically its
and Janes, 2013). Sufficient daylight reduces the electricity consump­ height/width ratio, namely, the “Well Index” (Serres and Murdoch,
tion for lighting (Mardaljevic et al., 2009), and is essential for human 1990; Calcagni and Paroncini, 2004; Acosta et al., 2013), as well as the
health (Brown and Jacobs, 2011; Aries et al., 2015). inclination of the walls (Kristl and Krainer, 1999; Freewan et al., 2014;
The light well is a solution to bring light into spaces located in the Ahadi et al., 2017), influence strongly the amount of light. However, in
core of buildings in dense neighborhoods. It is widely used in dense an existing environment, it is difficult to modify the geometry of the
historic city centers (Panerai et al., 2004), as well as in recent projects light well. The possibilities for improvement are therefore limited to the
(Kotani et al., 2003). A light well is a space open to the sky over several modification of the coating. Improvements produced by simple retrofits,
floors inside the volume of a building. It is designed to provide light and such as the use of white reflective materials, mirrors, and redirecting
ventilation to adjacent interior spaces (Harris, 2006). Daylight is devices, are analyzed from simulations on a geometric model of a
received at the top of the building, spreads in the well, and finally rea­ representative light well like those found in most historic city areas, as
ches the adjacent areas through the openings. Even if the amount of light Barcelona (Generalitat de Catalunya Departament de Territori i Soste­
is small, its presence is valuable because it provides information about nibilitat, 2012) and Bilbao (Departamento de Vivienda y Asuntos
the daily dynamics of natural light and thus regulates our circadian Sociales, 2002). As the study is focused on these minor modifications,
rhythm (Aries et al., 2015; Rea and Figueiro, 2018). the geometry of the model is assumed constant.
Lower floor rooms receive a limited quantity of light due to the Several studies have already highlighted the importance of the
specific elongated shape of the light wells. The amount of light reaching composition of the walls. The distribution of elements with different

Abbreviations: BD, Blind Days; CIE, Commission Internationale de l’Eclairage; DA, Daylight Autonomy; DAmax, Maximum Daylight Autonomy; DNI, Direct Normal
Illuminances; ECS, European Committee for Standardization; WI, Well Index.
* Corresponding author.
E-mail addresses: antoine.bugeat@univ-pau.fr (A. Bugeat), benoit.beckers@univ-pau.fr (B. Beckers), eduardof@fing.edu.uy (E. Fernández).

https://doi.org/10.1016/j.solener.2020.07.099
Received 9 May 2020; Received in revised form 28 July 2020; Accepted 31 July 2020
Available online 21 August 2020
0038-092X/© 2020 International Solar Energy Society. Published by Elsevier Ltd. All rights reserved.
A. Bugeat et al. Solar Energy 207 (2020) 1434–1444

reflectances (Du and Sharples, 2011) as well as the size and position of days and seasons (Reinhart et al., 2006). Therefore, the performance of
windows (Cole, 1990; Matusiak et al., 1999) have an impact on lighting the solutions investigated are compared under different climates. To
conditions. For instance, increasing the size and quantity of openings ensure a minimum daily provision of daylight throughout the year for
means decreasing the opaque surfaces that can reflect light into the these dark living spaces, we propose a method considering this low
spaces at the bottom of the light well (Littlefair, 2002). Moreover, the daylight availability.
use of specular reflective materials is efficient to bring more light to the
lower part (Kristl and Krainer, 1999; Lau and Duan, 2008). With the 2. Method
same purpose of providing light deep in the light well, the redirecting
devices prove to be effective solutions (Kotani et al., 2003; Freewan 2.1. The light well model and its variants
et al., 2014).
The light well represents a special context for light simulations. The The investigated light well represents most of those found in Euro­
inter-reflections play a major role in the total illuminance perceived and pean historic city areas (Fig. 1). With a section of 3 × 3 m, it rises over
must be precisely calculated. With standard Monte-Carlo path-tracing four floors for a total height of 12 m (WI = 4). Eight adjacent rooms
software, it is necessary to set the threshold of the number of reflections distributed on two opposite sides have access to the light well via 1 ×
to a large value to avoid underestimating the reflected part. Increasing 1.5 m openings through a 0.25 m thick wall. Only the four rooms on one
the number of reflections leads to longer calculation times and affects side are studied. An interesting result from Ahadi et al. (2017) is the very
the fluidity of the design process. To overcome this issue, we developed low, or even non-existent, variation in the performance in the rooms
an alternative method based on the radiosity method. The extension according to their orientation. As the light reaches the lower part after
proposed by Sillion and Puech (1989) is used to handle specular re­ several reflections, it enters the rooms from a multitude of directions and
flections and direct transmissions. It allows a fast and efficient calcula­ the orientation of the adjacent spaces no longer has any influence. These
tion of the hourly illuminances needed for Climate Based Daylight rooms are 4 m deep, 3 m width, and 2.75 m high. The sensors used for
Modelling (CBDM) metrics (Reinhart et al., 2006) with a precise the calculation of the metrics are distributed at the height of 80 cm from
consideration of the exchanges by diffuse and specular reflections. the floor and are spaced 25 cm for a total of 140 sensors per room.
Among the authors cited above, Ahadi et al. (2017) are the only ones The optical properties of the interiors follow the usual standards
having used CBDM metrics to assess the daylight performance. The other (Table 1): diffuse reflectance of 80% for the ceiling, 70% for the walls,
studies are based on mock-up experiments (Kristl and Krainer, 1999; Lau and 30% for the floor. The windows are composed of basic single glazing
and Duan, 2008; Cole, 1990; Matusiak et al., 1999), predictive models or with a specular reflectance of 8% and a direct transmittance of 88%. The
simulations under overcast sky conditions (Serres and Murdoch, 1990; composition of the adjacent surfaces is identical for all the presented
Calcagni and Paroncini, 2004; Acosta et al., 2013; Du and Sharples, cases. Conversely, we assessed several scenarios regarding the proper­
2011; Littlefair 2002), point-in-time simulations (Freewan et al., 2014), ties of the light well surfaces. Two variants of the light well grounds are
or questionnaire surveys (Kotani et al., 2003). Designing light wells with studied: a dark ground with a diffuse reflectance of 20% and a light
mirrors and redirecting devices requires to analyze in detail the solar ground with a diffuse reflectance of 60%. For the walls, dark and light
path according to the period of the year and the latitude. CBDM metrics scenarios are also investigated: the dark variant corresponds to a diffuse
such as Daylight Autonomy (DA) provide an accurate assessment of reflectance of 50%, equivalent to a grey or colored wall, and the light
lighting conditions according to the latitude and their variations over variant to a diffuse reflectance of 85%, equivalent to a standard white

Fig. 1. Top: geometry of the light well and the 4 adjacent rooms with their sensor grid; bottom: the 4 studied variants and the 2 others with the additional redirecting
device. Mirror surfaces are in sky blue. (For interpretation of the references to colour in this figure legend, the reader is referred to the web version of this article.)

1435
A. Bugeat et al. Solar Energy 207 (2020) 1434–1444

Table 1
Material characteristics of the interior surfaces (fixed) and of the different light well variants.
Interior materials Light well materials (4 variants *+ 2 with additional redirecting device)

(fixed) Dark surfaces Light surfaces* Mirror Top & Light Surfaces Mirror walls*

Ceiling ρd = 80% Walls ρd = 50% Walls ρd = 85% Walls ρr = 90% (top) Walls ρr = 90%
ρd = 85%
Walls ρd = 70% Ground ρd = 20% Ground ρd = 60% Ground ρd = 60% Ground ρd = 60%
Floor ρd = 30% Redirect. Device ρr = 90% Redirect. Device ρr = 90%
Windows τr = 88%
ρr = 8%

wall. Configurations with the presence of mirrors on the walls or the paths.
light well top are also studied. Mirrors have a specular reflectance of
90%. These optical properties comply with usual measured values (Di
Laura et al, 2011; Jakubiec, 2016). The CIE (2016) notation is employed 2.2. Performance metrics
to name the three surface properties: ρd for diffuse reflectance, ρr for
specular (regular) reflectance, and τr for direct (regular) transmittance. CBDM are today reference methods to quantify the daylight perfor­
In addition to diffuse and specular reflective walls, we propose a mance of interior spaces, and they are present in building directives and
simple passive device (Fig. 2) to redirect the sunrays to the bottom standards (ECS, 2018; IESNA, 2012). They allow assessment of daylight
during the darkest part of the year. The objective is to achieve accept­ over the whole year, taking into account the dynamic nature of lighting
able illumination conditions for a minimum amount of daily time conditions as functions of time, season, and climate for any location
throughout the whole year. A mirror is placed at the top to redirect the (Reinhart et al., 2006; Mardaljevic et al., 2009). In this article, we
sunrays to the bottom. It is a planar surface, and its dimensions are propose a performance assessment based on the following metrics:
identical to the section of the light well. It is designed to redirect as much
light as possible around solar noon during the winter half year. There­ • DA100 (and DA100 winter) [h/day] - Annual daylight autonomy with
fore, it is facing South (in the case of the Northern Hemisphere), and its an illuminance threshold of 100 lx. DA100 is the ratio between the
inclination is determined to redirect the sunrays in a perfectly vertical number of hours with illuminance above 100 lx and the total number
direction on the most representative day of the solar trajectories of the of hours per year. It is calculated for each sensor. Usually expressed
winter half year, i.e., around November 10. This day is halfway between as a percentage of time, this quantity is reported in average number
the autumnal equinox and the winter solstice. This proposed optimiza­ of hours per day to allow a better representation of its daily distri­
tion period can nevertheless be modified according to the needs and bution. DA100 winter is also used. It represents the DA100 but only for
requirements of each project. More complex designs, such as parabolic the winter half year between the equinoxes of September 21 and
shapes for light concentration, which have already proven their effi­ March 21.
ciency for photovoltaic and daylighting applications (Xuan et al., 2019; • BD [day / year] – BD is the number of days per year when the median
Li et al., 2020) or solar trackers, can also be considered for process illuminance in a room does not reach 100 lx during the day. In other
optimization but are not studied in this paper. The sensitivity of the words, a blind day is a day when natural light is not able to provide
device performance is investigated according to its two main orientation suitable lighting conditions inside a room. With this indicator, the
parameters, the azimuth, ideally oriented towards the South, and the days are counted, not the hours as for DA. This distinction allows us
elevation. to have information on the annual distribution of days with unsat­
The model is initially studied for Barcelona’s (41◦ ) climate. Other isfactory lighting conditions.
locations are later surveyed: Stockholm (60◦ ), Bilbao (43◦ ), Mexico City • Eavg (and Eavg winter) [lux] - It is the average illuminance over the
(19◦ ), and Quito (0◦ ). Barcelona and Bilbao have quite the same latitude, year (or over the winter half year). As for DA100, it is calculated for
but Bilbao presents much cloudier sky conditions. As an indication, the each sensor and the median of the sensor grid is used to state the
annual direct normal illuminance is 50% higher in Barcelona than in performance of one room. Unlike the two other metrics, Eavg is giving
Bilbao (Fig. 3). From Quito to Stockholm, an interval of about an evaluation based on absolute amounts of received light. It
20◦ separates the latitudes of the cities, thus offering very different solar removes the biases caused by the notion of the illuminance
threshold. For instance, a configuration that brings more light into a
room does not affect the DA100 if the room is already sufficiently
illuminated; in such a case, the average illuminance records more
accurately the contribution.
• DAmax,2000 [%] - With the use of reflective materials and mirrors, the
question of glare due to excessive amounts of light arises. This
additional indicator allows estimating the changes generated by the
modifications. The DAmax,2000 is a variant of the DA proposed here to
quantify the percentage of time when the illuminance can be
considered too high and cause visual discomfort (Reinhart et al.,
2006). The threshold of disturbing illuminance is chosen at 2000 lx
(Nabil and Mardaljevic, 2005).

The ground floor room is the least lit room and therefore the one we
are most concerned about. So, the majority of the metric values in this
article refers to it. We chose to use the median value of the metrics
obtained on the sensor grid to describe the performance of one room.
The median on the work plane makes it possible to give spatial infor­
mation: this means that the metric is bigger in 50% of the space and
Fig. 2. Scheme of the sun redirecting device designed for the winter half year. lower in the other half.

1436
A. Bugeat et al. Solar Energy 207 (2020) 1434–1444

Fig. 3. Annual distribution of Direct Normal Illuminances (DNI) for Barcelona and Bilbao.

A horizontal illuminance threshold of 100 lx is used for this study. discretized into surface elements, and a system of linear equations
We choose to use a lower level of illuminance compared to the usual linking the illuminances of each of these elements is established. The
lighting recommendations and standards that are around 300 lx or more resolution of this system directly gives the illuminances for the whole
(ECS, 2018; IESNA, 2012; Reinhart and Wienold, 2011). As a reminder, scene. In the case of light wells, the great advantage of this method is
the aim is to improve the lighting conditions of dark living spaces where that it takes into account the infinity of diffuse reflections in a short time.
the lighting needs are less demanding. For instance, light levels of less It gives users the possibility to assess comfortably a multitude of con­
than 100 lx, far below the usual target illuminance level used for figurations. The weak point of this method, in its original formulation, is
workspaces, can be sufficient in living and dining rooms (Di Laura et al, that it is unable to take into account other than perfectly diffuse
2011). In the definition of the Useful Daylight Illuminance, a CBDM metric reflection or transmission models. However, the proposal made by Sil­
proposed by Nabil and Mardaljevic (2005), 100 lx is also used as the lion and Puech (1989) makes it possible to extend its utilization to en­
limit above which natural light is effective, either as the only source of vironments composed of specular surfaces such as mirrors and
illumination or in conjunction with artificial lighting. The other reason transmitting surfaces such as glazing. This method, combined with the
for this choice is that people tend to tolerate and appreciate much lower concept of daylight coefficients of Tregenza and Waters (1983), has
illuminance levels in natural light than in artificial light (Paule et al., already demonstrated its effectiveness for daylight studies in dense
2015; Jakubiec et al., 2018). For these reasons, we believe that urban environments (Bugeat et al., 2019).
perceiving even low levels of natural light in these dark living spaces
represents a significant benefit to the comfort and quality of life of the 2.3.2. Built environment and light source modelling
inhabitants, and this must be considered during the design process. In the radiosity method, it is necessary to discretize the environment
into a mesh of surface elements called patches. In this study, the surfaces
2.3. Simulation methodology of the built environment are meshed with 25 × 25 cm patches and then,
a total of 6512 patches.
2.3.1. Radiosity light transport algorithm Unlike the built environment, the sky vault is discretized into a finite
The objective is to calculate the hourly illuminances on each sensor number of elements that are solid angles. It is meshed into 5000 tiles
during a whole year, and to deduce the performance metrics. Instead of according to the method proposed by Beckers and Beckers (2012),
using the widely used methods of ray tracing, implemented in the ensuring a partition with tiles of equal solid angles and the best possible
Radiance calculation engine (Ward, 1994) and the majority of com­ aspect ratio. The luminance of each sky tile is calculated using the model
mercial software (Ayoub, 2020), we chose to develop a program based proposed by Perez et al. (1993). This model requires the direct normal
on the radiosity method. Ray tracing and radiosity are the two main and diffuse horizontal irradiances obtained from IWEC and SWEC data
families of radiation simulation methods and each one offers different files, available in the Energy Plus format (.epw) at energyplus.net/
possibilities. In the ray tracing approach, a high number of rays are weather, as well as the calculation of the sun’s position for each time
traced from each sensor to find the path leading to the light sources step (Beckers and Beckers, 2010). The sun is modeled by the sky tile in
directly or after reflections. From these paths, we obtain the impact of which it is located. To switch from the normal direct irradiation, pro­
the light sources on the illuminance of the sensors or, more specifically, vided by the weather files, to the luminance of the sun tile (1), two steps
the daylight coefficients in the case of an annual calculation of are necessary: the calculation of the direct luminous efficiency accord­
daylighting (Tregenza and Waters, 1983; Mardaljevic, 1999; Reinhart ing to Perez (1990), then the calculation of the corresponding luminance
and Walkenhorst, 2001). This robust and flexible method has the of the sun tile from its solid angle:
advantage of being able to take into account any type of reflection or Lv,dir = ηEe,ndir /ω (1)
transmission. However, it shows some weaknesses for specific applica­
tions, such as light wells, where the access of light to the areas of interest where Lv,dir is the direct luminance of the sun, Ee,ndir the direct normal
is difficult. In this case, a large number of reflections must be taken into irradiance, η the direct luminous efficacy, and ω the solid angle of the
account to achieve a correct final result (Fig. 4). Moreover, the rays sky tile. In the radiosity method, the luminous exitance is the quantity
emitted for the calculation of the illuminance of a sensor are useful only that defines the light source elements. So, the luminance of a patch is
for this sensor and cannot be reused for other ones (view-dependent). considered independent of direction. In this particular case, luminous
Radiosity, the alternative method here developed and used, treats exitance is equal to the luminance multiplied by π. This is possible
the problem in a “Finite Element Method” approach: the environment is because sky tiles do not correspond to surface elements but solid angles.

1437
A. Bugeat et al. Solar Energy 207 (2020) 1434–1444

2.3.3. Extended view factor computation environment and the sensors Ec (4) are then expressed as the product
The next step consists of calculating the extended view factor be­ between the matrix of the contributions of the sky tiles on the illumi­
tween each elements of the scene. The conventional view factors allow nances of the built environment Dck, and the vector of the luminous
determining the amount of light received directly on each element from exitances of the sky and the sun Mk. As only the sky tiles and the sun emit
all the other elements that emit or reflect light diffusely. The method light, and only the sensors and the built environment receive and reflect
proposed by (Sillion and Puech, 1989) makes the radiosity method light, Eq. (3) is rewritten:
applicable in environments composed of glazing or mirrors. It consists of
Ec = Dck Mk (4)
replacing conventional view factors by extended view factors. They
represent the fraction of light sent by tile j received on tile i directly and with Dck = (I – Fextcc Rcc)-1 Fextck
after one or more specular reflections or direct transmissions. Thus, where Ec is the vector of illuminances of the built environment and
exchanges by specular reflection and direct transmission are directly sensors, Mk the vector of luminous exitances of the sky and sun tiles, Dck
included in the extended view factors. For their calculation, we used the the contributions matrix of sky and sun tiles on the illuminances of the
powerful open-source ray-tracing programs of the Embree library built environment, Fextcc the square matrix of extended view factors
working on CPU hardware (Wald et al., 2014). between the built environment patches and sensors, Fextck the matrix of
To compute view factors from a patch, rays are emitted from its extended view factors from the built environment patches and sensors to
surface to the scene. The origin points are randomly generated over the the sky tiles, Rcc the diagonal matrix of the diffuse reflectances of the
patch surface, and the directions are produced in the deterministic built environment, and I the identity matrix. The calculation by direct
approach proposed by Beckers and Beckers (2016). To determine these inversion, allowing the infinity of diffuse reflections to be taken into
directions, a hemisphere is meshed into tiles of equal view factors with account, is thus performed only once for Dck. Then, it is easy to obtain
the best possible aspect ratio. The directions formed by the hemisphere’s the hourly illuminances needed for the CBDM metrics after a quick
center and the center of each tile constitute the set of ray directions. In multiplication between this matrix and the different Mk vectors of sky
this way, each ray represents the same view factor ratio, or weight, equal and sun exitance of the year. This simulation step takes only 5 s with
to the inverse to the number of rays emitted. The view factor of a patch is MATLAB©.
the sum of the impacting rays’ weights. To obtain the extended view
factor, the rays’ paths are prolonged if they hit a non-diffuse patch. Their 2.3.5. Importance of a correct consideration of inter-reflections
weights decrease according to the characteristics of the intersected The great advantage of radiosity is its ability to take into account the
surfaces, namely ρr and τr, and their trajectories are extended in the infinity of diffuse reflections in a very fast way. Fig. 4 shows the influ­
corresponding directions. This process is repeated until reaching a ence of the number of diffuse reflections on the accuracy of DA100 and
diffuse patch or a sky tile. The extended view factor of a patch is the sum Eavg calculations in a light well composed of light diffuse surfaces during
of the weights of the rays impacting directly, and after specular re­ the winter half year. We used the Jacobi iterative method (Saad, 2003)
flections and direct transmissions. The angular dependence of specular for the solution of the system only for this figure. With this method, each
reflectance and direct transmittance, according to Fresnel relations, is iteration corresponds to an additional diffuse reflection. The number of
simplified by following the model proposed by Schlick (1994). reflections required is higher during the winter half year. In fact, as the
sunrays reach higher positions in the light well, the path to the bottom is
2.3.4. Solving the systems of illuminance longer and more reflections are needed. Twenty-four and twenty-eight
Once the extended view factors calculated, the last step is to solve the reflections are required to bring the percent error (Eq. (5)) below 1%
system of linear equations, linking the illuminance of all the patches of of the Eavg and the DA100 respectively. This huge amount leads to long
the scene. The radiosity method was mainly developed for image computation times with path tracing techniques and highlights the in­
rendering (Goral et al., 1984). So, the final output quantity, namely the terest to use a radiosity algorithm in this context.
radiosity, corresponds to the light emitted and reflected by a patch. However, it is not possible to consider the infinity of specular re­
However, for daylighting studies, light received by a patch is required. flections with the concept of extended view factors. Fortunately, light
The illuminance equation (2) is an adaptation of the radiosity equation propagates more quickly deep in the light wells with specular surfaces.
(Beckers, 2013). The number of specular reflections required is then smaller. Fig. 5 is
∑
∀i, Ei = (Fijext Mj + Fijext ρd,j Ei ) (2) similar to Fig. 4, showing the influence of the number of specular re­
j flections in a light well with mirror walls. Since the calculation of the
infinity of specular reflections is not possible, DA100 and Eavg obtained
where Ei is the illuminance received by patch i [lux], Fext
ij the extended with 30 specular reflections replaces DA100, ∞ refl. and Eavg, ∞ refl. in Eq.
view factor of patch i to patch j, Mj the luminous exitance of patch j (5). We chose to limit the number of specular reflections to fifteen. With
[lm⋅m− 2], and ρd,j the diffuse reflectance of patch j. The system can be this number, the error of the DA100 and Eavg falls below 1% in the ground
written in matrix form (3). floor room (Fig. 5).
The total computing time ranges from 1 min, in a light well without
(3)
− 1
E = Fext M + Fext R E = (I − Fext R) Fext M
mirrors, to 3 min in the extreme case with all mirror walls. The hardware
is a laptop equipped with an Intel i7 CPU. Since the solution time of the
where E is the vector of illuminances, M the vector of luminous exi­
radiosity system under MATLAB© takes only 5 s, the calculation of the
tances, Fext the square matrix of extended view factors, R the diagonal
extended view factors is the most time-consuming part.
matrix of diffuse reflectances, and I the identity matrix.
The calculation of CBDM metrics requires the hourly illuminances on Xn − X∞
(5)
refl. refl.
Percent Error = × 100
the sensors for a whole year, which means a multitude of different sky X∞ refl.
conditions. Instead of making a complete radiative calculation for each
time step, requiring a significant amount of computing time, a solution 3. Results and discussions
independent of the lighting conditions is calculated only once: this is the
concept of Daylight Coefficients (Tregenza and Waters, 1983). The 3.1. Influence of surface properties
method consists of determining the contributions of each sky tile on the
illuminance of each sensor and patch of the built environment. This 3.1.1. Propagation of daylight by diffuse reflections
method, involving the discretization of the sky dome, is perfectly suited Fig. 6 shows the distribution of DA100 in four light well configura­
to radiosity algorithms. The illuminances received on the built tions with different wall compositions. The first two cases are only

1438
A. Bugeat et al. Solar Energy 207 (2020) 1434–1444

Fig. 4. Convergence of DA100 and Eavg during the winter half year according to the number of diffuse reflections in the 4 floors within a light well composed of light
diffuse surfaces in Barcelona.

Fig. 5. Convergence of DA100 and Eavg during the winter half year according to the number of specular reflections in the 4 floors within a light well composed of
mirror walls in Barcelona.

composed of diffuse surfaces. The first one consists of dark walls with a around solar noon and the summer solstice. There is a substantial dif­
50% diffuse reflectance and a floor with a 20% diffuse reflectance. The ference between the winter and summer periods.
second one involves light diffuse surfaces with a diffuse reflectance of In Fig. 7, each curve represents the DA100 and Eavg reached on each
85% for the walls and 60% for the floor. floor as a function of the diffuse reflectance of the walls and with a
In the light well with dark walls, there is a significant disparity be­ constant diffuse reflectance of the ground equal to 60%. The DA100 on
tween the different floors. The top floor receives a sufficient amount of the top floor are insensitive to wall reflectance as most of the light comes
light with an average duration of 10 h 35 per day above 100 lx. From the directly from the sky or the sun. This light is sufficient to produce an
lower floor, this duration drops to 5 h 04 and then to 0 h 55 on the first illumination of more than 100 lx in almost every moment of the day. As
floor and finally to a near-zero value on the ground floor. By switching to mentioned above, the sky view factor decreases drastically in the lower
light diffuse surfaces, the performance of the lower floors is considerably floors, where reflected light is predominant. The shape of the curves of
improved. On the ground floor, the light reaches the interior of the room the lower floors illustrates the strong influence of the reflectance of the
through inter-reflections, which allows satisfactory lighting conditions light well walls. On the ground floor, the DA100 is zero for reflectance
to be achieved for an average of 2 h 53 per day. from 0 to 50% and then grows exponentially. Increasing the reflectance
The Blind Day metric and the calendar of illuminance levels above by 10%, from 75% to 85%, triplicates the DA100 on the ground floor
100 lx on the ground floor provide information on the annual distribu­ (from 1 h to 2 h 53 per day). The same observations are valid for the
tion of good lighting conditions. In the case of dark surfaces, the few average illuminance curves. The average annual illuminance increases
hours with illuminance above 100 lx are exactly around solar noon and from 42 lx to 75 lx, with the same slight increase of 10% on the reflec­
the summer solstice. In the case of bright surfaces, the hours with illu­ tance of the walls. These results also underline the need to use the most
minance above 100 lx are more frequent, but remain also centered accurate reflectance values possible from builders’ data or precise

1439
A. Bugeat et al. Solar Energy 207 (2020) 1434–1444

Fig. 6. Distribution of DA100 and ground floor performance for different configurations.

Fig. 7. Impact of diffuse reflectance of light well walls on DA100 (left) and Eavg (right). The reflectance of the light well floor is 60%.

measurements as input data. With mirrors and reflective materials, the issue of glare due to
excessive amounts of light arises. In Fig. 8, the risks of glare according to
3.1.2. Propagation of daylight by specular reflections the DAmax,2000 metric are compared in three cases: the light well with
The results obtained by placing mirrors in the light well once again diffuse dark surfaces, with diffuse light surfaces, and with mirror walls.
highlight the great potential of improvement by changing the properties On the last floor, which is the most affected since it is the most exposed,
of the surfaces. The use of mirrors considerably enhances the illumina­ the DAmax,2000 values are very low in the 3 cases studied. Switching to
tion levels on the darker floors. Mirrors offer a clear advantage over light mirrors or white reflective walls only slightly increases the risk of glare
diffuse surfaces. Their installation is not always possible because of compared to dark walls. On the lower floors, switching to mirrors has
installation costs, privacy issues, or architectural constraints. Even if only a small impact in the area very close to the openings. On these
mirrors enable illuminances higher than 100 lx during the whole year, floors, the DAmax,2000 is not even half of the level of the last floor without
there is a clear disparity between winter and summer. mirrors.

1440
A. Bugeat et al. Solar Energy 207 (2020) 1434–1444

Fig. 8. Distribution of the DAmax,2000 in three configurations: left, diffuse dark surfaces; middle, diffuse light surfaces; right, mirror walls.

3.2. Redirecting the sunrays are obtained during the whole year and also during the winter half
(September 21 to March 21). During the winter period, the redirection
3.2.1. Performance improvement in winter device plays a decisive role in the ground floor room. It allows a change
By modifying the characteristics of the walls of the light well, its from 0 h 40 to 3 h 30 per day with an illuminance of more than 100 lx.
performance can be considerably improved, but significant seasonal The map representing the distribution of hours with illuminance levels
disparities remain. Due to the solar trajectories and sky conditions in the above 100 lx shows that almost all days of the year now have suitable
annual cycle, the distribution of satisfying lighting conditions is lighting conditions. This more uniform annual distribution of lighting
centered on the summer solstice while the winter period remains the conditions is also illustrated by the decrease in the number of blind days
least well served. Illuminance may not reach 100 lx for several consec­ from 154 to 42 days.
utive days during this period. It is important to note that the device does not increase the amount of
Fig. 9 illustrates the effect on instantaneous illuminances thanks to light received during the summer half year, because it acts then as a
the device on a sunny day close to November 10 at solar noon. The mask for the sunrays. Between April 21 and August 21, the altitude of
sunspot that is redirected towards the ground of the light well is the sun is higher than 61◦ (angle of elevation of the device). It thus
perfectly visible at this time. The addition of the device drastically en­ makes it possible to limit the undesirable light and heat contributions
hances the amount of light perceived in the lower areas. Illuminance during the summer period. This shadow lowers the amount of received
levels of around 500 lx and above are reached on the ground floor. The light, but the light from the sky is already sufficient at this period to
difference between the lower and upper parts is reduced. Lighting achieve suitable levels of illumination. The simulations showed very
conditions at the top floor are very slightly affected. small variations in DAmax,2000 on all floors with the addition of the
Fig. 10 gives information on the ability of the device to provide good redirecting device. The extra light provided by the device is too small to
lighting conditions throughout the period. The DA100 and Eavg in Fig. 10 exceed the 2000 lx threshold in the lower floors. The occasional varia­
tions in the DAmax,2000 mainly concern the space very close to the win­
dows and the top floor where the sunrays are redirected into the room
only a few hours in the year.

3.2.2. Orientation sensitivity

Tables 2 and 3 show the device performances in function of the two
orientation parameters: the azimuth, ideally oriented towards the South,
and the elevation, ideally considered at 61◦ according to the proposed
design system. The evaluation is based on the winter DA100 and Eavg, and
the annual number of blind days.
Table 2 relates to the azimuthal shift. It presents the performance of
six configurations with an azimuth difference of +10◦ between each of
them, ranging from a perfect South orientation (+0◦ ) to a South-East
orientation (+50◦ ). Differences in the performance of the device are
weak up to 40◦ angle and then start to decrease significantly from 50◦ .
The results show that the azimuth does not play a significant role in the
performance of the device. Indeed, the altitude of the sun in winter re­
mains most of the day very close to the daily maximum. However,
another orientation of the device changes the time around which the
Fig. 9. Illumination received at solar noon on a sunny day of November. On the
left: light diffuse walls. On the right: light diffuse walls plus redirecting device.
device takes effect.

1441
A. Bugeat et al. Solar Energy 207 (2020) 1434–1444

Fig. 10. Distribution of DA100 and details of the performance on the ground floor during the year and during the winter half of the year. Both cases have light diffuse
surfaces. On the right, the redirecting device is included.

This is because a larger part of the sky is hidden in a more horizontal

Table 2
position. The sun is the light source that we want to benefit from, but it is
Influence of azimuthal orientation relative to the South on the performance of
not the only one. The sky also plays a valuable role in the general illu­
the device in a light well with diffuse light walls.
mination of the light well. A too big tilt of the device limits the sky view
0◦ 10◦ 20◦ 30◦ 40◦ 50◦
factor from the light well surfaces and results in lower illuminance
levels.

3.3. Variations due to climate

DA100 winter [h/day] 3 h30 3 h30 3 h30 3 h23 3 h16 2 h45
September 21 – march 21 Two identical light wells under different climatic conditions may
Eavg. winter [lux] 89 lx 89 lx 88 lx 85 lx 81 lx 72 lx require different interventions to achieve an equivalent level of perfor­
September 21 – march 21
mance. In Table 4, the selected configuration of light wells previously
Blind Days [days/year] 42 d 42 d 42 d 44 d 45 d 51 d
analyzed under Barcelona climatic conditions is now studied in different
cities: Stockholm, Bilbao, Mexico City, and Quito.
At equivalent latitude, a more pronounced cloud cover reduces all
Table 3
the performance metrics uniformly in any light well configuration. For
Influence of the elevation angle in a light well with diffuse light walls.
all the configurations, the average annual illuminance is 20% higher in
− 10◦ − 5◦ 0◦ 5◦ 10◦ Barcelona than in Bilbao. It is interesting to see that the redirecting
device remains very efficient in Bilbao despite the cloudier conditions.
For the light diffuse walls, during winter half year, Eavg is more than
doubled thanks to the device, both in Bilbao and in Barcelona. However,
if we set a target of 3 h per day of DA100 during the winter period and a
DA100 winter [h/day] 2 h50 3 h25 3 h30 3 h29 3 h29
maximum of 1/6th of blind days per year, the recommendations are
September 21 – march 21
Eavg. winter [lux] 57 lx 72 lx 89 lx 90 lx 86 lx different for these two cities. In Barcelona, both objectives are achieved
September 21 – march 21 with diffuse light surfaces and the installation of a redirecting device. In
Blind Days [days/year] 59 d 44 d 42 d 42 d 39 d Bilbao, it is not sufficient, and the use of mirrors is required.
Regarding the impact caused by latitude, there are significant dif­
ferences between the performance of the same light well located in
Table 3 concerns the elevation angle. It presents the performance of
Stockholm or Quito, both on the absolute quantities of received light and
configurations with 5◦ and 10◦ offsets in both directions. Little changes
on daily and annual times with satisfying illuminance. In Mexico City
in performance are recorded with a device in a more vertical position.
and Quito, the simple application of diffuse light surfaces easily achieves
This is because, although the sunlight is not optimally redirected to­
the double objective proposed before. In Stockholm, the use of mirror
wards the bottom of the light well, it is nevertheless redirected towards
walls and redirecting devices is mandatory. The main cause is the dif­
the walls, which subsequently scatter the light. However, tilting the
ference in solar paths. In tropical latitudes (Quito and Mexico City), the
device to a more horizontal position negatively affects performance.
sunrays reach the lower parts of the walls of the light well throughout

1442
A. Bugeat et al. Solar Energy 207 (2020) 1434–1444

Table 4 4. Conclusion
Performances in different light well’s configurations in Stockholm, Bilbao,
Barcelona, Mexico, and Quito. Changing the surface characteristics or adding redirecting devices
are efficient measures to improve the annual performance of a light well.
This study has shown that CBDM metrics computed using our radiosity
method give a quick and precise evaluation of the quantities of light
received and their annual and daily variation for the tested
configurations.
Dark Light +Device Mirror +Device The results show a very high sensitivity of the performance con­
Walls Walls Walls
cerning diffuse reflectance of the light well surfaces. Because the adja­
DA100 [h/day] ground floor cent rooms have a restricted direct view to the sky, the reflected part of
Stockholm 59◦ N 0 h00 1 h22 2 h49 6 h45 7 h04
the light is getting a major importance. Variations of 10% on the diffuse
Bilbao 43◦ N 0 h00 1 h51 3 h12 7 h56 7 h58
Barcelona 41◦ N 0 h02 2 h53 4 h24 9 h12 9 h11 reflectance of the walls can produce differences in average annual illu­
Mexico 19◦ N 0 h45 5 h03 5 h54 10 h01 10 h00 minance of up to twice in ground floor rooms. Whatever the climate and
Quito 0 ◦
0 h22 4 h54 – 9 h52 – the configuration, choosing diffuse light walls is the strict minimum for
DA100 winter [h/day] ground floor
good performance. Considerable increases in illuminance levels are also
Stockholm 59◦ N 0 h00 0 h00 1 h54 2 h04 3 h10
Bilbao 43◦ N 0 h00 0 h11 2 h21 5 h35 6 h06
obtained with mirror walls. Specular reflections allow a higher quantity
Barcelona 41◦ N 0 h00 0 h40 3 h30 7 h17 7 h29 of light to penetrate deeper into the light well. So, the use of mirror walls
Mexico 19◦ N 0 h02 3 h54 5 h35 9 h21 9 h22 is mandatory in narrow light wells. The orders of magnitude of the
Quito 0 ◦
0 h17 4 h43 – 9 h47 – benefits obtained in the case studied in the different climates are:
Eavg [lux] ground floor
switching from dark walls to diffuse white walls uniformly increases the
Stockholm 59◦ N 7 lx 42 lx 62 lx 245 lx 268 lx
Bilbao 43◦ N 9 lx 61 lx 86 lx 291 lx 306 lx average illuminance level by a ratio of 5, and switching to mirror walls
Barcelona 41◦ N 12 lx 75 lx 104 lx 343 lx 356 lx by 30.
Mexico 19◦ N 26 lx 134 lx 148 lx 449 lx 453 lx A key factor that emerged from this study is the height of the sun.
Quito 0◦ 21 lx 119 lx – 432 lx – With a low sun, direct rays only reach the upper part of the light well,
Eavg winter [lux] ground floor
Stockholm 59 N
◦
3 lx 17 lx 49 lx 67 lx 113 lx
heavily penalizing the propagation of light towards the lower floors.
Bilbao 43◦ N 5 lx 31 lx 69 lx 139 lx 199 lx This leads to important seasonal differences in illuminance levels,
Barcelona 41◦ N 7 lx 43 lx 101 lx 194 lx 278 lx resulting in a very dark winter period in mid and high latitudes. Using
Mexico 19◦ N 12 lx 82 lx 124 lx 364 lx 371 lx the redirecting device produces a localized enhancement of light which
Quito 0◦ 20 lx 113 lx 404 lx
ensures in the lowest spaces satisfying lighting conditions throughout
– –
Blind Days [days/year] ground floor
Stockholm 59 N
◦
365 d 253 d 113 d 104 d 59 d the year, in critical latitudes as Stockholm and even in cloudy climates.
Bilbao 43◦ N 365 d 204 d 92 d 22 d 21 d The performance of the device is not sensitive to changes in position
Barcelona 41◦ N 360 d 154 d 42 d 0d 1d (inclination and azimuth), and is, therefore suitable for different urban
Mexico 19◦ N 236 d 2d 0d 0d 0d fabrics. At different latitudes, the same light well can produce signifi­
Quito 0◦ 297 d 5d – 0d –
DAmax,2000 [%] top floor
cantly unequal lighting conditions. In tropical locations, such as Mexico
Stockholm 59◦ N 2% 3% 3% 4% 3% City or Quito, white walls with a high diffuse reflectance are fully suf­
Bilbao 43◦ N 2% 3% 4% 4% 3% ficient to achieve good lighting conditions throughout the year. At
Barcelona 41 N
◦
2% 3% 5% 5% 4% higher latitudes, such as in Barcelona, Bilbao, or Stockholm, due to
Mexico 19◦ N 2% 4% 5% 5% 4%
lower sun elevations, high reflective walls need to be combined with
Quito 0◦ 2% 3% – 5% –
redirecting devices to achieve suitable lighting conditions all year
round.
the year thanks to very high solar trajectories. At higher latitudes This study shows the importance of inter-reflection or redirection of
(Stockholm), the propagation of light by inter-reflection between the daylight in light wells during architectural projects and urban planning.
walls is much longer and more difficult, since the sunrays reach the Good performance levels can be achieved without changing the geom­
interior of the light well at higher positions and with very horizontal etry and shape of the buildings, thus avoiding heavy work and reduction
trajectories. of living space. Even if the performances are known to be sensitive to
Concerning the question of unwanted excess of light, the frequency climatic and geometric aspects, the potential for improvement offered
of illuminance above 2000 lx on the top floor remains low and shows by the modification of the characteristics of surfaces and redirecting
only slight differences between the 5 climates and between all the devices is always considerable. The article proposes both performant
studied configurations. From these results, it can be concluded that the assessment and effective simulation methods for the design of these
use of reflective materials has only little influence on the risk of visual specific spaces.
and thermal discomfort. On the other hand, more noticeable differences
are observed between the different cities based on the average annual Declaration of Competing Interest
illuminance, even though it very rarely reaches critical levels. In the case
of the light well with diffuse light surfaces, the average annual illumi­ The authors declare that they have no known competing financial
nance in Mexico City is more than twice that of Bilbao, and the switch to interests or personal relationships that could have appeared to influence
mirror walls increases illuminance levels by a factor of 4 for all climates. the work reported in this paper.
In these cities located in tropical latitudes, climatic conditions are more
constraining in maintaining thermal comfort and the additional solar Acknowledgments
gain created by the use of reflective materials can become penalizing. To
identify these possible overheating risks in hot climates, a complete This work was supported by the Communauté Pays Basque, Bayonne,
thermal study is required implying a precise calculation of solar gain and France, the Région Nouvelle-Aquitaine, Bordeaux, France and the
natural ventilation by CFD simulations, which is essential in this type of project FSE_1_2017_1_144731 from Agencia Nacional de Investigación e
architectural environment (Muhsin et al., 2017). Innovación (ANII, Uruguay).

1443
A. Bugeat et al. Solar Energy 207 (2020) 1434–1444

References Jakubiec, J.A., Quek, G., Srisamranrungruang, T., 2018. Towards Subjectivity in Annual
Climate-Based Daylight Metrics. In: Proceedings of the Building Simulation and
Optimization Conference (BSO 2016). Cambridge, pp. 24–31.
Acosta, I., Navarro, J., Sendra, J.J., 2013. Predictive method of the sky component in a
Kotani, H., Narasaki, M., Sato, R., Yamanaka, T., 2003. Environmental assessment of
courtyard under overcast sky conditions. Sol. Energy 89, 89–99. https://doi.org/
light well in high-rise apartment building. Build. Environ. 38, 283–289. https://doi.
10.1016/j.solener.2012.12.009.
org/10.1016/S0360-1323(02)00033-1.
Ahadi, A.A., Saghafi, M.R., Tahbaz, M., 2017. The study of effective factors in daylight
Kristl, Ž., Krainer, A., 1999. Light wells in residential building as a complementary
performance of light-wells with dynamic daylight metrics in residential buildings.
daylight source. Sol. Energy 65, 197–206. https://doi.org/10.1016/S0038-092X(98)
Sol. Energy 155, 679–697. https://doi.org/10.1016/j.solener.2017.07.005.
00127-3.
Aries, M.B.C., Aarts, M.P.J., van Hoof, J., 2015. Daylight and health: a review of the
Lau, B., Duan, Z., 2008. The daylight benefit conferred upon adjoining rooms by specular
evidence and consequences for the built environment. Light. Res. Technol. 47, 6–27.
surfaces in Top-Lit Atria. Archit. Sci. Rev. 51, 204–211. https://doi.org/10.3763/
https://doi.org/10.1177/1477153513509258.
asre.2008.5125.
Ayoub, M., 2020. A review on light transport algorithms and simulation tools to model
Li, G., Xuan, Q., Akram, M.W., Akhlaghi, Y.G., Liu, H., Shittu, S., 2020. Building
daylighting inside buildings. Sol. Energy 198, 623–642. https://doi.org/10.1016/j.
integrated solar concentrating systems: a review. Appl. Energy 260, 114288. https://
solener.2020.02.018.
doi.org/10.1016/j.apenergy.2019.114288.
Beckers, B., Beckers, P., 2010. Les anomalies, l’équation de Kepler, la position du soleil.
Littlefair, P., 2002. Daylight prediction in atrium buildings. Sol. Energy 73, 105–109.
Report Helio_006_fr, http://heliodon.net/downloads/Beckers_2010_Helio_006_fr_2.
https://doi.org/10.1016/S0038-092X(02)00038-5.
pdf.
Mardaljevic, J., 1999. Daylight simulation: validation, sky models and daylight
Beckers, B., Beckers, P., 2012. A general rule for disk and hemisphere partition into
coefficients. PhD Thesis. De Montfort University, Leicester.
equal-area cells. Comput. Geom. Theory Appl. 45, 275–283. https://doi.org/
Mardaljevic, J., Heschong, L., Lee, E., 2009. Daylight metrics and energy savings. Light.
10.1016/j.comgeo.2012.01.011.
Res. Technol. 41, 261–283. https://doi.org/10.1177/1477153509339703.
Beckers, B., 2013. Taking Advantage of Low Radiative Coupling in 3D Urban Models In:
Mardaljevic, J., Janes, G.M., 2013. Multiscale Daylight Modeling for Urban
Proceedings of the Eurographics Workshop on Urban Data Modelling and
Environments, in: Solar Energy at Urban Scale. John Wiley & Sons, Ltd, pp. 159–190.
Visualization (UDMV 2013). Girona, pp. 17-20. https://doi.org/10.2312/UDMV/
https://doi.org/10.1002/9781118562062.ch8.
UDMV13/017-020.
Matusiak, B., Aschehoug, Ø., Littlefair, P., 1999. Daylighting strategies for an infinitely
Beckers, B., Beckers, P., 2016. Fast and Accurate View Factor Generation In: Proceedings
long atrium: an experimental evaluation. Light. Res. Technol. 31, 23–34. https://doi.
of the First International Conference on Urban Physics (FICUP 2016). Quito, pp. 22-
org/10.1177/096032719903100105.
31.
Muhsin, F., Yusoff, W.F.M., Mohamed, M.F., Sapian, A.R., 2017. CFD modeling of natural
Brown, M.J., Jacobs, D.E., 2011. Residential light and risk for depression and falls:
ventilation in a void connected to the living units of multi-store housing for thermal
results from the LARES Study of Eight European Cities. Public Health Rep. 126,
comfort. Energy Build. 144, 1–16. https://doi.org/10.1016/j.enbuild.2017.03.035.
131–140. https://doi.org/10.1177/00333549111260S117.
Nabil, A., Mardaljevic, J., 2005. Useful daylight illuminance: a new paradigm for
Bugeat, A., Fernández, E., Beckers, B., Aguerre, J., 2019. A Multi-Scale Consideration of
assessing daylight in buildings. Light. Res. Technol. 37 (1), 41–57. https://doi.org/
Daylight in a Real Urban Context. In: Proceedings of the 16th Conference of
10.1191/1365782805li128oa.
International Building Performance Simulation Association (IBPSA 2019). Rome, pp.
Panerai, P., Castex, J., Depaule, J.-C., Samuels, I., 2004. Urban forms: the death and life
3460-3467. https://doi.org/10.26868/25222708.2019.210650.
of the urban block. Routledge.
Calcagni, B., Paroncini, M., 2004. Daylight factor prediction in atria building designs.
Paule, B., Boutillier, J., Pantet, S., 2015. Shading device control: Effective impact on
Sol. Energy 76, 669–682. https://doi.org/10.1016/j.solener.2004.01.009.
daylight contribution. In: Proceedings of International Conference Future Buildings
CIE (Commission Internationale de l’Eclairage), 2016. International Lighting
and Districts Sustainability from Nano to Urban Scale (CISBAT 2015). Lausanne, pp.
Vocabulary. CIE DIS 017/E:2016, Commission Internationale de l’Eclairage, Vienna.
241-246.
Cole, R.J., 1990. The effect of the surfaces enclosing atria on the daylight in adjacent
Perez, R., Seals, R., Michalsky, J., 1993. All-weather model for sky luminance
spaces. Build. Environ. 25, 37–42. https://doi.org/10.1016/0360-1323(90)90039-T.
distribution—Preliminary configuration and validation. Sol. Energy 50, 235–245.
Compagnon, R., 2004. Solar and daylight availability in the urban fabric. Energy Build.
https://doi.org/10.1016/0038-092X(93)90017-I.
36, 321–328. https://doi.org/10.1016/j.enbuild.2004.01.009.
Rea, M.S., Figueiro, M.G., 2018. Light as a circadian stimulus for architectural lighting.
Departamento de Vivienda y Asuntos Sociales, 2002. Decreto 317/2002, de 30 de
Light. Res. Technol. 50, 497–510. https://doi.org/10.1177/1477153516682368.
diciembre, sobre actuaciones protegidas de rehabilitación del patrimonio urbanizado
Reinhart, C.F., Walkenhorst, O., 2001. Validation of dynamic RADIANCE-based daylight
y edificado. Comunidad Autónoma del País Vasco Departamento de Vivienda y
simulations for a test office with external blinds. Energy Build. 33, 683–697. https://
Asuntos Sociales.
doi.org/10.1016/S0378-7788(01)00058-5.
Di Laura, D.L., Houser, K.W., Mistrick, R.G., Steffy, G.R., 2011. The lighting handbook
Reinhart, C.F., Mardaljevic, J., Rogers, Z., 2006. Dynamic daylight performance metrics
reference and application. Illuminating Engineering Society of North America, New
for sustainable building design. LEUKOS 3, 7–31. https://doi.org/10.1582/
York.
LEUKOS.2006.03.01.001.
Du, J., Sharples, S., 2011. The variation of daylight levels across atrium walls:
Reinhart, C.F., Wienold, J., 2011. The daylighting dashboard – a simulation-based design
Reflectance distribution and well geometry effects under overcast sky conditions.
analysis for daylit spaces. Build. Environ. 46, 386–396. https://doi.org/10.1016/j.
Sol. Energy 85, 2085–2100. https://doi.org/10.1016/j.solener.2011.05.015.
buildenv.2010.08.001.
ECS (European Committee for Standardization), 2018. Daylight in buildings. EN 17037:
Saad, Y., 2003. Iterative methods for sparse linear systems. SIAM.
2018. European Committee for Standardization.
Serres, A.W., Murdoch, J.B., 1990. On the Efficiency of Skylight Wells. J. Illum. Eng. Soc.
Freewan, A.A.Y., Gharaibeh, A.A., Jamhawi, M.M., 2014. Improving daylight
19, 73–86. https://doi.org/10.1080/00994480.1990.10747943.
performance of light wells in residential buildings: nourishing compact sustainable
Schlick, C., 1994. An inexpensive BRDF model for physically-based rendering. Comput.
urban form. Sustain. Cities Soc. 13, 32–40. https://doi.org/10.1016/j.
Graph. Forum 13, 233–246. https://doi.org/10.1111/1467-8659.1330233.
scs.2014.04.001.
Sillion, F.X., Puech, C., 1989. A general two-pass method integrating specular and diffuse
Generalitat de Catalunya Departament de Territori i Sostenibilitat, 2012. Decret 141/
reflection. SIGGRAPH Comput. Graph. 23, 335–344. https://doi.org/10.1145/
2012 sobre condicions mínimes d’habitabilitat dels habitatges i la cèdula
74334.74368.
d’habitabilitat. Generalitat de Catalunya Departament de Territori i Sostenibilitat
Tregenza, P.R., Waters, I.M., 1983. Daylight coefficients. Light. Res. Technol. 15, 65–71.
Secretaria d’Habitatge i Millora Urbana, Agència de l’Habitatge de Catalunya.
https://doi.org/10.1177/096032718301500201.
Goral, C.M., Torrance, K.E., Greenberg, D.P., Battaile, B., 1984. Modeling the Interaction
Wald, I., Woop, S., Benthin, C., Johnson, G.S., Ernst, M., 2014. Embree: a Kernel
of Light between Diffuse Surfaces. SIGGRAPH Comput. Graph. 18, 213–222. https://
framework for efficient CPU ray tracing. ACM Trans. Graph. 33 https://doi.org/
doi.org/10.1145/964965.808601.
10.1145/2601097.2601199.
Harris, C.M., 2006. Dictionary of architecture and construction. McGraw-Hill.
Ward, G.J., 1994. The RADIANCE Lighting Simulation and Rendering System, in:
IESNA (Illuminating Engineering Society of North America), 2012. LM-83-12 IES Spatial
Proceedings of the 21st Annual Conference on Computer Graphics and Interactive
Daylight Autonomy (sDA) and Annual Sunlight Exposure (ASE). Illuminating
Techniques, SIGGRAPH 1994. pp. 459–472. https://doi.org/10.1145/192161
Engineering Society.
.192286.
Jakubiec, J.A., 2016. Building a Database of Opaque Materials for Lighting Simulation.
Xuan, Q., Li, G., Lu, Y., Zhao, B., Zhao, X., Su, Y., Ji, J., Pei, G., 2019. Overall detail
In: Proceedings of the 36th International Conference on Passive and Low Energy
comparison for a building integrated concentrating photovoltaic/daylighting system.
(PLEA 2016). Los Angeles.
Energy Build. 199, 415–426. https://doi.org/10.1016/j.enbuild.2019.07.018.

1444