Content uploaded by Terri Peters
Author content
All content in this area was uploaded by Terri Peters on Sep 10, 2019
Content may be subject to copyright.
Generative Design Approaches to Daylight in MURBs
Terri Peters1, Jake Wolf2, Brady Peters2 Ted Kesik
1Department of Architectural Science, Ryerson, Toronto, Canada
2John H Daniels Faculty of Architecture, Landscape and Design, University of Toronto, Toronto,
Canada
Abstract
Access to acceptable levels of daylight are important for
people’s quality of life. Multi-unit residential buildings
(MURBs) are known to perform poorly in terms of
daylight compared to other residential building types.
There are neither appropriate agreed upon metrics, nor
effective methods for designing for daylight in MURBs.
This paper presents results of a speculative design study
that utilizes generative design to explore alternative
geometries for the MURB tower typology. The
experiment combines a genetic algorithm for spatial
variation with climate-based daylight modeling (CBDM)
to test the new forms against variations of the point tower
MURB floorplan are commonly used. This paper
identifies the poor performance of typical MURBs for
daylight, and proposed new techniques for form
generation. A new workflow has been developed and
tested and a number of challenging issues have been
identified.
Introduction
As the trend toward urban intensification continues in
Canadian cities and around the world, a large number of
existing and new housing are multi-unit residential
buildings (MURBs) According to Canadian Census data,
one third of new housing are MURBs and 1.9 million
households live in this kind of housing (2016 Census). It
is imperative that quality of life and inhabitant wellbeing
be prioritized in this building type, as the compact
footprint and repeated floorplate typical of MURBs pose
challenges to attaining satisfactory levels of daylight and
ventilation. MURB unit aspect ratios of 1:2 are not
uncommon in current practice, which leads to suites that
are deep with unacceptable levels of daylight (Peters and
Kesik 2018). Numerous studies show daylight has a great
impact on people’s health and wellbeing (Veitch and
Galasiu 2012) and daylight is central to architectural
quality and desirability. Further studies are needed into
performance based design exploration of formal and
geometric options that maximize daylight. Many
considerations drive MURBs design, but these are
suspended in this study to isolate daylight as a design
parameter.
MURBs design are not optimized for daylight, either in
the building form or in the arrangement of individual
units. MURB design tends to prioritize efficiency of
repeating floorplates and developers seek financial gains
for maximizing number of units and ease of construction.
MURB research tends to focus on quantitative building
performance aspects such as energy efficiency,
ventilation and to a lesser extent daylight (Kesik and
O’Brien 2017). This study is part of a larger ongoing
study into daylight metrics, tools and design workflows in
MURBs, and this sub-study offers a more speculative
consideration of how generative design can facilitate
daylight can be a design driver for new geometries and
building forms in MURBs. This paper uses generative
design to develop new geometries that privilege daylight,
and climate-based daylight modeling (CBDM) to test the
new forms against variations of the point tower MURB
floorplan are commonly used. The aim is to propose and
test new ways of working to compare options of MURB,
and new design options for better daylighting (Figure 1).
While daylighting was the primary consideration in the
research underlying this paper, future studies will take a
more robust generative design approach to consider
additional factors to housing quality such as visual
privacy, amenity such as terraces, green roofs and other
considerations.
Generative design tools offer designers exciting
possibilities in performance based design. There are few
MURBs experiments event though new workflows are
urgently needed to offer better quality of life. Some
recent MURB examples were examined as inspiration for
this paper, especially those with highly articulated facades
that are designed for daylight. For example, 56 Leonard
Street by Herzog and de Meuron is an irregular residential
tower design in New York with a non-repeating floorplan
and highly articulated and expressive façade. It serves as
a formal inspiration to this research project offering an
alternative tower aesthetic, although it is unclear if the
façade design is performance driven. According to the
architects, it was designed using formal strategies of
staggering, setback and pixilation and it has been an
architectural and commercial success (Herzog and de
Meuron website). MVRDV’s three story ‘Valley’ project
in the Netherlands was a relevant precedent for this
research in terms of design workflow and computational
design. Designers at MVRDV were able to articulate and
parametricize the project’s design objectives and use a
genetic algorithm to generate options in terms of daylight,
views, structural design and code compliance
(Christodoulou et al 2018). While the MVRDV project
used multi-parameter optimization, this paper focuses on
daylight.
Figure 1: Generative Design Workflow
Methods
This research uses two methods of performance-based
design evaluation, first a genetic algorithm (Rutten 2013)
and then climate based daylight modeling (CBDM) using
DIVA-for-Rhino (Solemma 2016). Computational design
was used and in particular a Genetic algorithm (GA) was
developed because the design issues of form and daylight
are complex, with potentially competing variables. The
goal was to find solution “sets” of optimal design options
rather that a singular solution. This allowed the setting of
specific performance targets, and enabled the options to
be optimized for very specific targets.
Figure 2: Isometric view of Point Tower ‘base case’
geometry and CBDM analysis of typical floor. Shown
are the percentage of occupied hours that these nodes
receive the target illuminance of 100 lux.
Genetic Algorithm
The initial design approach was to break up the building’s
floor-plates into modular ‘tiles’ for analysis and use a
genetic algorithm to test a variety of combinations of
these tiles based on daylight performance. With each
arrangement of tiles, a new set of floor-plates is created
and the entire building is re-tested for daylighting quality.
Because the computational load of a full DIVA-for-Rhino
for every design options was deemed not feasible, a proxy
for daylighting quality was developed based on
daylighting rules of thumb (Figure 3). These rules state
that typically adequate daylight will penetrate 1.5 times
the height of the window head (Reinhart, 2005). This
basic rule of thumb does not consider the orientation of
the building and so the algorithm does not either, the best
performing options from this set of iterations are then
tested in DIVA-for-Rhino.
Figure 3: Daylight Rule of Thumb is typically measured
as 1.5-2 times the head height of the window.
The study experimented with: the building footprint size
and shape, number of storeys, floor-to-floor height and
depth of the floor/ceiling structure. For this study, new
options are compared to a ‘base case’, the point tower
typology which is a typical MURB high-rise with a 27m
x 27m footprint, that is 26 storeys tall, with 3m external
floor-to-floor height and .5m of structure between floors
(figure 2). The relevant geometric parameters became
inputs into the genetic algorithm, the floor area on each
floor relate to a 3m x 3m grid. Using Galapagos’s
Simulated Annealing algorithm, these grid tiles are either
turned ‘on’ or ‘off’ (Rutten 2013). For each combination,
all ‘on’ tiles are treated as a new floorplate. To determine
the ‘fitness’ of each floor-plate a combination three
factors are combined to provide an overall fitness value.
In using the parametric tool, the floorplate was arranged
into zones for analysis. The 27mx27m floorplate was
considered in three 9mx9m square zones which were then
divided into smaller ‘tiles’ of 3mx3m. The smallest unit
of measure in the geometry studies was 3m tiles are
approximately the size of a room. The building cores for
the stairs, elevators, and unit entry areas are simplified as
the middle 9mx9m zone of the building. In this study, the
cores are considered a low priority for daylight to
prioritize daylight in the principle rooms of the dwellings.
A genetic algorithm was used three ways to evaluate the
performance of the design options. A light ray factor is
achieved by arraying each floor-plate with points, and
from each point rays are cast at an angle of 33.69 degrees
(which corresponds to the daylight rule of thumb) toward
the exterior of the building. The ratio of the rays cast to
FITNESS
split contours into grid of tiles
(cores are excluded
from gene pool)
is a given tile
“on” or “o”
draw line from center of tile to closest
point on floorplate edge positive factor
do any rays escape with-
out intersecting a floor-
plates above?
o
on
shoot rays in this direction at angles
between 33 and 89 .
no
yes
negative factor
“o” tiles are ignored
all adjacent tiles
form a region
are there more regions
than floors?
yes
no
gross floor area
of all regions
x
ray
factor
region
factor
positive factor
negative factor
building
footprint
building
height get contours
floor-to-floor
heights
GENES
INPUTS
x
the rays which escape (i.e. those that do not intersect a
ceiling above) provides a percentage of rays that hit the
sky. A perfect score would be 1, meaning all tiles have
light access. However, if this were the only driving factor
the algorithm would find a solution in which most tiles are
‘off’, leaving the few remaining tiles with lots of access
to light but creating a building with almost no occupiable
space. The second evaluation metric, the GFA factor is
indexed to the floor area. This is the ratio of the ground
floor area (GFA) of the generated floorplates to the
possible GFA (i.e. if all tiles were ‘on’). Lastly, as a way
to encourage the algorithm to seek tile combinations that
make continuous floor-plates (i.e. not an archipelago of
floor regions) the region factor acts as a penalty for each
region beyond the minimum (in this case 26 regions, one
for each floor).
Daylight Modeling
DIVA-for-Rhino was used to evaluate the best versions of
four studied options generated by the GA. In order to test
the best case scenarios, the window to wall ratio was set
at 100% WWR to test all possible daylight availability,
the building is positioned on a 45 degree angle facing
South-East/South-West and internal partition walls are
not considered. Vancouver was used as the building
location and surrounding buildings and site context were
not considered. This study assumed the buildings is
occupied 8am-6pm with Daylight Savings Time (DSTI)
invoked, analyzed in 60 minute increments. This is a
typical occupancy schedule for daylighting, although
some studies are challenging these assumptions in
residential applications (Dogan and Park 2017). For the
analysis grid, the node height and spacing assumptions are
0.5 m high, the node spacing is 0.5 m apart. This gives a
relatively human scale proportion for the unit of measure
when compared to the dwelling-scale measure used in the
geometry study. The target illuminance threshold for
measuring two similar dynamic daylight metrics: spatial
daylight autonomy (sDA) was tested and the mean spatial
daylight autonomy was also calculated (Illuminating
Engineering Society of North America, 2012).
There are a number of daylighting performance metrics
(DPM) used to quantity natural light and its availability
including glare, aesthetics and non-visual effects of
daylight. These metrics are not developed for use in
MURB, and normally focus on performance issues facing
workplace and office buildings (Dogan and Park 2018)
which are not suitable for MURB. This study uses Spatial
Daylight Autonomy (sDA) a commonly used annual
daylighting metric that is more relevant than other
options, such as Daylight Factor, for analysis of MURB.
DA refers how much of the interior area (in this case the
sensors were set at 0.5 high) receives a target daylight
illuminance (often set to 300 lux in office settings, in this
case 100 lux was used) at least 50% of the occupied hours.
Mean Spatial Daylight Autonomy (Mean sDA) is the
percentage of annual daytime hours that a floorplate’s
daylight levels are above a specified illumination level
(IESNA 2012). Mean DA was found to be more of a
relevant dynamic daylight metric to compare and interpret
the performance of the different floorplans.
Table 1: Surface reflectance and glass transmittance
used in the DIVA simulations.
Material
Reflectance/
Transmittance
Ceiling
70%
Floor
20%
Internal Walls
70%
Façade Spandrel
Panel
35%
Façade Glazing
65%
Results
Four series of design experiments were carried out and
compared to an analysis of the base case MURB (figure
1). The point tower base case MURB has low levels of
spatial daylight autonomy (DA100lux [50%]=2%) and
low mean spatial daylight autonomy (15%) on all floors.
Point Tower- Articulated Floorplates
Option A1 generated options that added articulation to the
floorplates by creating unique floorplates that removes
tiles that do not receive daylight, and relates to
surrounding floors. When a floor tile is generated, a
ceiling tile is added so each floor remains only one storey
tall. The ceiling tiles block daylight rays so options are
generated that optimize for the best performance for the
overall building form. In some cases, the floorplates have
large balconies, and often there are internal courtyards.
The highest performing floor is the top level, floor 26. The
spatial daylight autonomy is 9% (compared to base case
of 2%) and the mean spatial daylight autonomy is 28%,
(base case is 15%). This floorplan arrangement of offers
a number of unusual spatial relationships for MURB.
There are large (3m x 3m minimum) outdoor balconies
with views down multiple levels into the building next to
the core, bringing daylight into the core and heightening
the sense of being at the top of the building. As with 76
Leonard Street, parts of the floorplan and in some cases
the units could be geometrically distinct from one other,
there could be a sense of being in a separate house, with
different visual and acoustic relationships from the
neighbours.
Floor 9 is also one of the highest performing floorplates
(Figure 4). The algorithm generated this arrangement to
allow light to come in from the sides, adding articulation
to the floorplate and the overall building form. The
arrangement of spaces is highly unusual for MURB, with
spatial complexity despite the unit being arranged over
one level. There are balconies with views to the sky as the
next floor is five stories above, and internal courtyards
where there can be indoor-outdoor views into the rooms.
The daylight is improved and there would be cross
ventilation, large outdoor areas and increased views to the
sky inside the principle rooms. The spatial daylight
autonomy is 8% (compared to base case of 2%) and the
mean spatial daylight autonomy is 27%, (base case is
15%).
Figure 3: Option A1 Series. The four best optimized
versions of this series.
Figure 4: Detail of the building geometry of Option A1-1
highlighting Floor 9, and the CBDM unit analysis of
Floor 16 using DIVA-for-Rhino. The analysis shows
percentage of occupied hours that these nodes receive the
target illuminance of 100 lux.
In terms of spatial daylight autonomy six of the
floorplates equal the spatial daylight autonomy target of
2% as in the base case, six perform less well in terms of
DA but better in mean DA, and fourteen exceed the base
case with the highest being 9%. Reflecting on results of
Option A1 and the testing of version 1, the limitation of
keeping each unit a single floor height was determined to
be restrictive, and Option A2 explores the potentials of
generating double height spaces.
Table 2: Option A1 Version 1 DIVA results
Point Tower- Double Height Spaces
Option A2 improves upon Option A1 by improving
daylight by generating double height spaces. The top four
versions of a number of generate versions in this Option
study are shown in image X. The algorithm generated
options based on rules that dictated that when a floor tile
does not have a corresponding tile above, a double-height
space is created and a ceiling tile is added two floors
above. Option A2 produced a wider range of performance
than A1. In terms of spatial daylight autonomy, two of
the floorplates equal the spatial daylight autonomy target
of 2% as in the base case, eleven perform less well in
terms of DA but better in mean DA, and thirteen exceed
the base case with the highest at 16%.
Figure 5: Option A2 Series. The four best optimized
versions.
Figure 6: Detail of the building geometry of Option A2-1
highlighting Floor 3, and the CBDM unit analysis of
Floor 9 using DIVA-for-Rhino. The analysis shows
percentage of occupied hours that these nodes receive the
target illuminance of 100 lux.
Some of the results are impacted by the different floor
areas generated per floor and the corresponding daylight
performance. For example, floor 3 is the highest
performing at 16% DA. This floorplate is small as it is
generated by accommodating many double height spaces
on the first floor. The entire floorplate of Level 3 is only
198m2 (2131 sqft) including the core zone, so there will
be fewer units. These units would arguably be very high
quality and higher value, as their views into neighbouring
suites on the floor would be at the double height part of
that suite, so there would be increased privacy between
neighbours. The dwellings on level 3 would enjoy the
highest level of spatial daylight autonomy, double height
spaces and terraces with four story high views to the sky.
Level 3 is also clearly distinguishable as an architectural
feature on the building’s façade from street level, making
it unique and potentially more desirable.
Table 3: Option A2 Version 1 DIVA results
Reflecting on results of Option A2 and the testing of
version 1, the double height spaces do as expected bring
in more daylight and the decision to limit the height to two
levels seems appropriate. The wide variety of
performance was unexpected, and many floorplates
performed better, but many also much worse than
expected. This Option could use further refining, and a
follow up study will add in the parameter of minimum
floor area per floor in order to create more usable results.
However, care would need to be taken to test different
weightings of the floor area parameter because if a
specific minimum or maximum floor area rule was
incorporated then it would significantly lower the number
of generated options. This could potentially weaken the
overall findings of generating options designed for
achieving daylit interior spaces, as it could shift the focus
to floorplates that maximize quantity of interior floor area
Point Tower- Complexity
The starting point for the Option B1 series was the 27mx
27m point tower, but the goal was to create more
articulated and high-performance floorplate geometries,
with a repeated single level floorplan. Option-B1
generates options that are based on the rule that when a
tile on one floorplate does not have a corresponding tile
above, a ‘ceiling’ tile is added so each floor is only one
storey tall. Option B1-1 was the best performing version
and it has large balconies and internal courtyards. The
higher floors are slightly better performing than the lower
levels. All of the floors perform better than the point
tower base case examined in Option A1 and A2. The very
articulated forms look similar to another typical MURB
typology, the ‘barbell’ shape, or H-shape tower. For
comparison, a typical H-shape tower was analyzed (see
figure X). The H-shape tower is a relevant ‘base case’ for
Options B1 and Options B2. The H-shape base case
MURB has higher levels of spatial daylight autonomy
than the point tower base case, (DA100lux [50%]=3% and
4% on level 26) and higher mean spatial daylight
autonomy (17% and 19% on level 26). Option B1-1 did
not perform significantly higher than the H-shape base
case see Table 4 below. It is surprising that the B1
Options did not greatly improve in terms of daylight.
Given the nature of the GA, it was expected that the GA
should have found options that were more close to the H-
shape if it is in fact higher performing. There could be a
number of reasons for this, including limitations in the
GA fitness functions.
!!!! !!!!!!!!!!!!!!!!!!!!
Figure 7: The geometry and typical unit analysis of the
commonly used MURB geometry, the H-Shape Building
using DIVA-for-Rhino. The analysis shows percentage of
occupied hours that these nodes receive the target
illuminance of 100 lux.
Figure 8: Option B1 Series. The four best optimized
versions and a unit analysis of Floor 24 using DIVA-for-
Rhino. The analysis shows percentage of occupied hours
that these nodes receive the target illuminance of 100 lux.
Table 4: Option B1 Version 1 DIVA results
Point Tower- Complexity
Option-B2 generated sets of two floorplates, and these
pairs are repeated 13 times. When a tile on the bottom
floor-plate does not have a corresponding tile above, a
double-height space is created and a ceiling tile is added
two floors above. In the most optimized option, B2-1, the
double height space has only one double height space, a
single tile adjacent to the core. B2-2, B2-3 and B2-4 have
more double height spaces. The set of alternating large
and small floorplates could be imagined as potentially an
open and daylit garden level and a more enclosed area.
The ‘floating’ terraces on odd floors are disconnected
from the floorplate, but these could be connected and be
useful and unusual outdoor space. In this option, the floor
plate has the core exposed in areas so it is daylit in areas.
The forms generated are interesting form a privacy
perspective, as the B2-1 option is broken into four distinct
stacked ‘buildings’. With so much outdoor space shown,
it is clear that the north facing decks would be less useful.
A future study could then maximize the large outdoor
decks based on daylight to gain better performance on
smaller floorplates
Figure 9: Four Options B2 Series. The four best optimized
versions. B2-1 Floor 8 and Floor 9 South-East/South-
West facing unit analysis using DIVA-for-Rhino. The
analysis shows percentage of occupied hours that these
nodes receive the target illuminance of 100 lux.
Figure 10: B2-1 Floor 8 and Floor South-East/South-
West facing unit analysis using DIVA-for-Rhino. The
analysis shows percentage of occupied hours that these
nodes receive the target illuminance of 100 lux.
Table 5: Option B2 Version 1 DIVA results
Discussion
As noted in the introduction, this speculative design study
had a number of limitations because the complex
variables of MURB design were not all taken into
account. Building energy performance, structural
efficiency, fire security, and economic viability are
acknowledged as highly relevant issues in developing
optimized design solutions for MURBs but in the interest
of focusing on daylight, these aspects were not
parametricized or evaluated in this study. Some of these
variables could be taken into account in a further study. A
more detailed study could take into account surrounding
buildings, which would naturally impact the daylight in
MURBs. Additionally, only one climate zone (Vancouver
BC) and orientation (South-East/South-West) was tested
and this could be expanded to compare to other building
types and orientations. Larger questions of ‘what is good
daylight in MURBs’ will be explored in future studies
involving on site measurements and occupancy surveys to
better explore what metrics, illuminance thresholds and
environmental considerations impact daylight in MURBs.
Conclusions
This study focuses on daylight, yet there are a number of
building performance aspects that need to be critically
examined in order to design for quality of life in MURB.
With so many variables, there are difficulties
ininterpreting results due to conflicting findings. New
studies investigate the need for new metrics specific to
MURB, such as visual privacy (Alkalali et al, 2018),
which examines the often conflicting desires for privacy
inside the home and windows for views and daylight.
Ventilation autonomy (Ko et al, 2018) is a metric that
allows for the simultaneous assessment and visualization
of ventilation, thermal and lighting comfort data in order
to make it easier for designers to understand trade offs and
benefits of design options.
Computational design tools offer great potentials for
designing MURBs with better daylight. In particular, the
use of a genetic algorithm to better understand complex
variables and parametricize design aspects has been
useful. While a GA should not guide the design of
MURBs, GAs can uncover design opportunities and
moments that can be further explored by the designer. In
this paper, the intention was to explore generative design
and climate-based daylight modelling as means of
developing performance based and aesthetically
interesting building forms that offer enhanced daylighting
in MURBs. This paper identified the poor performance of
typical MURBs for daylight, and proposed new
techniques for form generation. A new workflow was
developed and tested and a number of challenging issues
have been identified. This section summarizes the interim
conclusions stemming from the study underlying this
paper.
1.!Building Scale: The design of the building’s
geometry, in particular making the perimeter form
more articulated, is shown to have a large impact on
daylight in individual units. There is a need for more
studies that experiment with the building’s geometry
for improved daylight in MURB units.
2.!Metrics for Measuring Daylight: There is a need for
better metrics for measuring daylight quality in
MURBs. Spatial Daylight Autonomy was found to
be useful when considered in relation to results from
other metrics such as Mean Spatial Daylight
Autonomy in order to enable comparison and
interpretation of results. This conclusion supports
earlier research by Peters and Kesik 2018.
3.!Iterative Workflow: The generative design
workflow developed and tested here of using a GA in
Grasshopper and CBDM in Diva for Rhino tests these
workflows for applicability as of an iterative design
process. The fact that these tools are within the
architect’s design environments (Grasshopper and
Rhino are increasily used by designers) shows
promise that this workflow could be practical.
4.!Unusual Spatial Qualities for MURBs: The
generative design tool found spatial arrangements not
normally seen in MURBs design. Double height
spaces, courtyard outdoor spaces within the
floorplate, and large outdoor terraces were all highly
performing options for daylight. The outdoor
qualities more closely match environmental
characteristics found in other residential building
types, giving the MURBs patios and front yards that
may make them more desirable. The typical point
tower base case MURB has low levels of spatial
daylight autonomy (DA100lux [50%]=2%) and low
mean spatial daylight autonomy (15%) on all floors.
All of the studies presented in this paper exceeded
these levels in either DA or Mean DA.
5.!Daylight Proxy: The daylight proxy used in the GA
based on Daylight Rule of Thumb worked well, and
when compared to the DIVA analysis the best
performing GA options were also highly performing
for Mean Spatial Daylight Autonomy. The proposed
generalized methodology could work with different
proxys including climate and orientation.
6.!Localized Optimization vs Building Optimization:
A significant challenge in setting the parameters and
interpreting the results of this study has been in the
balancing design moves that contribute to localized
optimization and those which contribute to overall
building performance. For example, to make a suite
daylit internal courtyards and outdoor terraces shape
the floorplate but impact via overshading and
preventing double height spaces on the floor below.
could be tested in a future study. A follow up study
will test where partition walls would go based on
daylight as a spatial organizing feature; could
consider outdoor spaces – interior and exterior
balconies; and could test optimizing for other climate
zones.
Acknowledgements
The authors gratefully acknowledge the BC Housing
Research Excellence Program for funding this project.
Special thanks to our colleagues for sharing their
knowledge and providing feedback on this paper and the
underlying study.
References
Alkhalili, N., Kesik, T., O’Brien, W.O., Peters, T. (2018)
Developing and testing visual privacy metrics. 7th
International Building Physics Conference IBPC 2018
Ithica (USA). September 23-26, 703-708.
Christodoulou, A., Vola, M., Rikken, G. Case Study for
the Application of Multidisciplinary Computational
Design Assessment and Constructability Optimisation
Tools. SimAUD 2018, Delft, (Netherlands) 51-58.
DIVA. (2016). Version 4. Solemma LLC. Retrieved from
http://solemma.net
Dogan, T and Park, Y.C. (2017). A new framework for
residential daylight performance evaluation. Proceedings
of Building Simulation 2017: The 15th IBPSA Conference,
San Francisco, (USA). August 7-9.
Dogan, T and Park, Y.C (2018) A critical review of
daylighting metrics for residential architecture and a new
metric for cold and temperate climates. Lighting Res.
Technol. 2018; 0: 1-25.
Galasiu, A. and Reinhart, C. (2008). Current daylighting
design practice: a survey. Building Research &
Information 36(2), 159–174.
Government of Canada, Canadian Census Data, (2016).
Herzog and de Meuron, “56 Leonard Street”
https://www.herzogdemeuron.com/index/projects/compl
ete-works/301-325/305-56-leonard-street.html Accessed
5 May 2018
Illuminating Engineering Society of North America,
IESNA. (2012). LM-83-12 IES Spatial Daylight
Autonomy (sDA) and Annual Sunlight Exposure (ASE).
New York, NY, USA: IESNA Lighting Measurement.
Kesik T., & O’Brien, L. Enhancing the livability and
resilience of Multi-unit residential buildings (MURBs):
MURB Design Guide, Version 1.0 January 2017.
https://www.daniels.utoronto.ca/faculty/kesik_t/MURB-
Design-Guide/MURB-Design-Guide-v1.0-Jan2017.pdf
Ko, W.H., Schiavon, S., Brager, G., Levitt, B. Ventilation,
thermal and luminous autonomy metrics for an intergrated
design process. Building and Environment 145(2018)
153-165.
Peters T., & Kesik T. (2018). Daylight Simulation for
Multi-Unit Residentail Buildings: Occupant Centered
Approaches to Assessment. eSim 2018. Montreal
(Canada) May 9-10.
Reinhart, C.F., 2005. A simulation-based review of the
ubiquitous window-head-height to daylit zone depth rule-
of-thumb. Proceedings of Building Simulation 2005: The
9th International IBPSA Conference, Montreal (Canada)
August 15-18, 1011–1018.
Rutten, D. (2013). Galapagos on the Logic and
Limitations of Generic Solvers. Architectural Design,
March/April (222) 132-135.
Veitch, J., & Galasiu A. (2012). The Physiological and
Psychological Effects of Windows, Daylight, and View at
Home: Review and Research Agenda. National Research
Council of Canada Institute for Research in Construction
NRC-IRC Research Report RR-325.