You are on page 1of 25

Lighting Res. Technol.

2010; 42: 7–31

A rules of thumb-based design sequence for


diffuse daylight
CF Reinhart Dr Inga,b and VRM LoVerso PhD Archa,c
a
National Research Council Canada, Institute for Research in Construction
1200 Montreal Road, K1A 0R6 Ottawa, Canada
b
Harvard University, Graduate School of Design 48 Quincy Street, Cambridge MA 02138, USA
c
TEBE Research Group, Politecnico di Torino, corso Duca degli Abruzzi 24, 10129 Turin, Italy

Received 20 December 2007; Revised 20 February 2009; Accepted 27 February 2009

This paper proposes and validates a daylighting design sequence for sidelit
spaces. Since the design sequence uses the daylight factor as a performance
metric, it is aimed towards spaces that primarily receive diffuse daylight. It should
be complemented by a design analysis that looks at direct sunlight for glare and
energy considerations. The sequence interconnects and refines earlier proposed
rules of thumb and is intended to be used during the earliest design stages when
concepts regarding programming, floor plans, massing and window areas are
initially explored. All steps within the sequence were ‘validated’ using Radiance
simulations of over 2300 sidelit spaces. During step one of the sequence the
effective sky angles are calculated and target daylight factors are defined for all
potential daylit zones within a building. In step two a refined version of the
‘daylight feasibility study’ is used to help the design team to identify building
zones with high daylighting potential based on a target mean daylight factor
criterion. During step three suitable interior room dimensions and surface
reflectances are determined using a combination of the Lynes’ limiting depth,
‘no sky line’, and window-head-height rules of thumb. Step four provides a more
accurate estimate of the required glazing area for each zone based on the Lynes
daylight factor formula which is also validated as part of this work. The effect of
external obstructions is considered throughout the process. The paper closes with
a discussion of the merits of the design sequence compared to the glazing factor
spreadsheet calculation method promoted by LEED-NC 2.2.

List of symbols DF average daylight factor on the


work plane (%)
AEA adjusted effective aperture (–)
DFF daylight feasibility factor
Aglazing net glazing area (m2)
DLynes limiting room depth (m)
Atotal total area of interior surfaces
hwindow-head-height window-head-height (m)
(including windows) (m2)
OF obstruction factor (–)
Awall gross window þ wall area (m2)
Rmean area-weighted mean surface
 obstruction angle (8)
reflectance (–)
 vis visible glazing transmittance (–)
y sky angle ¼ vertical angle of sky,
calculated from the centre of the
Address for correspondence: CF Reinhart, Harvard glazing (8)
University, Graduate School of Design, 48 Quincy Street,
Cambridge MA 02138, USA. w interior room width (m)
E-mail: reinhart@gsd.harvard.edu WWR window-to-wall ratio (%)

ß The Chartered Institution of Building Services Engineers 2010 10.1177/1477153509104765

Downloaded from lrt.sagepub.com at SETON HALL UNIV on April 4, 2015


8 CF Reinhart and VRM LoVerso

1. Introduction thumb, i.e. ‘a useful principle having wide


application but not intended to be strictly
Daylighting rules of thumb are simple, numer- accurate or reliable in every situation’4 In the
ical expressions that relate a design quantity of context of daylighting another explanation
interest, e.g. how far daylight will penetrate might be that rules of thumb tend to be purely
into a building, to one or several design empirical, i.e. neither derived from scientific
parameters, e.g. the window-head-height. principles nor experimentally validated.
According to two recent surveys of close to Notable exceptions are two equations derived
400 design practitioners ‘experience from pre- by Lynes, which approximate the average
vious work’ and ‘rules of thumb’ are the two daylight factor and limiting depth of a
daylight prediction tools on which design sidelit space5 as well as a previous study by
practitioners most commonly rely during the one of the authors that compared the window-
schematic design phase.1,2 The main appeal of to-head height rule of thumb to annual day-
rules of thumb is that they are easy to learn light simulations.3 Both studies are further
and offer quick advice regarding key design described below.
parameters without slowing down the design The objective of the current paper is to
process. Daylight simulations and scale critically review several previously proposed
models offer an effective means to go beyond daylighting rules of thumb (Section 2), to test
rules of thumb and refine a daylighting con- their validity using Radiance simulations of
cept later in the design process. But, since over 2300 sidelit spaces (Sections 3 and 4),
many design teams still lack the know-how, and to reformulate and combine them into a
time, or resources for such detailed design four-step design sequence for diffuse day-
investigations, the daylighting analysis of lighting (Section 5). The limits and merits of
many buildings begins and ends with the use the sequence as well as the US Green Building
of rules of thumb. Given the resulting impor- Council’s Leadership in Energy and
tance of rules of thumb for daylighting design Environmental Design – New Construction
practice today, one might wonder which rules (LEED-NC) glazing factor formulae are
are actually being used in the ‘trenches’. The discussed in Section 6.
participants of the two above-mentioned
surveys listed a series of different daylighting
rules-of-thumb. The rules were concerned with 2. A review of three rules of thumb
different design aspects such as space propor-
tions, glazing area, building orientation, etc. This section reviews three previously pub-
While most of the rules cited evolved around lished daylighting rules of thumb that can be
common themes and ideas, their exact wording used to influence key daylighting-related
substantially differed from participant to par- design decisions.
ticipant. As an example, the popular rule of
thumb that links window-head-height and 2.1 Daylight feasibility study
daylight penetration depth was quoted by The ‘daylight feasibility study’ offers a
many participants with recommended ratios quick test to identify which zones within a
ranging from 1.5 to 2.5. A review of various building can potentially be daylit or not.
popular design guides showed that these In this context a zone consists of a series of
guides also differ in their exact formulation spaces in a building with similar daylighting
of this rule.3 Why is the formulation of rules of characteristics. Key parameters that influence
thumb so ‘non-standardised’? Part of the these daylighting characteristics are façade
answer might be that they are ‘just’ rules of orientation and external obstructions such as
Lighting Res. Technol. 2010; 42: 7–31

Downloaded from lrt.sagepub.com at SETON HALL UNIV on April 4, 2015


A design sequence for diffuse daylight 9

surrounding landscape or buildings. The zero (no obstructions) and 100% (view out
daylight feasibility study suggests that if a fully obstructed).
zone’s adjusted effective aperture, AEA, is The daylight feasibility study was originally
larger than a certain threshold level called the introduced in the 1989 Public Works and
daylight feasibility factor, DFF, then the zone Government Services Canada (PWGSC)
has a high potential for daylighting and ‘PWC Daylighting Manual’.6 The ‘Tips for
merits a more comprehensive daylighting Daylighting with Windows’ document from
analysis. The adjusted effective aperture is Lawrence Berkeley National Laboratory
defined as the product of a zone’s effective (LBNL) later cited a slightly modified version
aperture and its obstruction factor, OF. The of the rule.7 Both design guides presented the
effective aperture is in turn the product of a rule ‘as is’ without any further justification.
zone’s window-to-wall ratio, WWR, and the The DFFs given by PWGSC and LBNL were
mean visible transmittance of all façade 0.22 and 0.25, respectively.
openings (Equation (1)). While the daylight feasibility test does not
seem to have been systematically validated in
Adjusted effective aperture the past, its appeal for designers is that it can
easily be applied from the earliest design
¼ WWR  vis  OF4DFF ð1Þ
stages onwards. As a simple pass/fail criterion
the test can effectively inform fundamental
The WWR is here defined as the overall programming decisions such as where to
area of all translucent or transparent façade locate zones with a high need for daylight
openings excluding mullions and frames within a building. The physical basis of the
divided by the zone’s exterior wall area, rule is that the adjusted effective aperture is a
Awall. Awall is the part of the window wall direct measure of the relative amount of the
that borders the space, i.e. it is the product of daylight flux that can enter a particular space
the space width times the floor-to-ceiling compared to the same space with an unob-
height. The obstruction factor approximates structed, fully opened façade. External
the effect of external obstructions and is obstructions, reduced window opening sizes
determined as a function of the ‘percentage and glazing transmittances all reduce the
of the view obstructed as seen from a typical incoming daylight to a point below which its
task location’.6 As shown in Figure 1, the contribution to lighting the space ceases to be
obstruction factor is defined for four ‘per- meaningful. This minimum level corresponds
centage of obstructed view’ ranges between to the DFF (Equation (1)). The important

If view <50% obstructed, If view >50% and <70% obstructed, If view >70% and <90% obstructed, If view >90% obstructed,
OF = 1.00 OF = 0.85 OF = 0.65 OF = 0.40

Figure 1 Relationship between the percentage of view obstructed by surrounding landscape and the obstruction
factor

Lighting Res. Technol. 2010; 42: 7–31

Downloaded from lrt.sagepub.com at SETON HALL UNIV on April 4, 2015


10 CF Reinhart and VRM LoVerso

question is what this minimum level should be lighted. When the average daylight factor is
for different space types, i.e. what does a less than 2%, the interior space will seem
daylight feasibility factor of 0.22 (PWGSC) or dimly lighted’.9 One of the compliance paths
0.25 (LBNL) imply for the amount of day- for the daylighting credit 8.1 within the
light available within a space? This question is LEED for New Construction version 2.2
addressed below. green building rating system requires a glazing
factor above 2% in 75% of all regularly
occupied areas. The LEED glazing factor
method is examined in detail in Section 6.
2.2 Daylight quantity – average daylight While these few examples show that the
factor daylight factor plays a pivotal role in current
The daylight feasibility study evaluates norms and guidelines, the authors would like
whether the amount of daylight penetrating to stress that a number of research efforts
through a façade can – in principle – lead to currently aim towards developing new day-
good daylighting in a space bordering the lighting metrics in order to complement/
façade or not. The rule essentially constitutes replace the daylight factor. There is a partic-
a ‘necessary condition’ for daylighting. How ularly strong push towards climate-based
satisfactory the daylight within a space actu- metrics which take local climate, specific
ally ends up being is further dependent on the lighting requirements and space use into
specific lighting requirements of the space account.10
(quantity) as well as the uniformity of the The latter caveat notwithstanding, the
daylight throughout the space. Daylight daylight factor remains an important day-
quantity and uniformity both depend on lighting design factor and rules of thumb to
interior space dimensions and surface reflec- estimate the average daylight factor for a
tances. This section deals with the quantity space are hence useful for design practitioners
aspect. Uniformity is addressed in Section 2.3. using this metric. In a seminal paper in 1979
One still commonly used metric to char- Lynes derived the following expression
acterise the overall quantity of diffuse day- (Equation (2)) of the average daylight factor
light within a space is the average daylight (DF) in percentage for a rectangular, sidelit
factor on the work plane. A number of design space5:
guides and standards recommend different
average daylight factor values for different Aglazing vis y
space types. For example, the British DF ¼ ð2Þ
Atotal 2ð1  Rmean Þ
Standards Institution, BS 8206 Part 2 pre-
scribes that: ‘if electric lighting is not nor-
mally to be used during daytime, the average where Aglazing ¼ net glazing area,  vis ¼ visual
daylight factor should not be less than glazing transmittance, y ¼ sky angle ¼ vertical
5% . . . [whereas] . . . if electric lighting is to be angle in degrees subtended at the centre of
used throughout daytime, the average day- the glazing area by direct view of the sky
light factor should be not less than 2% if a (Figure 2(a)), Atotal ¼ total area of all interior
predominantly daylit appearance is wanted’.8 surfaces including windows and Rmean ¼ area-
Along the same lines, the Illuminating weighted mean surface reflectance of all
Engineering Society of North America interior surfaces. While Lynes derived DF
(IESNA) Lighting Handbook suggests that as the mean daylight factor over all inte-
‘when an average daylight factor is 5% or rior room surfaces, DF is nowadays usually
greater an interior space will appear to be well interpreted to be the mean daylight on the

Lighting Res. Technol. 2010; 42: 7–31

Downloaded from lrt.sagepub.com at SETON HALL UNIV on April 4, 2015


A design sequence for diffuse daylight 11

(a) (b)

α
θ γ

Work plane height No sky line depth

Figure 2 (a) Geometrical relationship between the sky angle y and the obstruction angle, ; (b) definition of the no
sky line depth

work plane. Lynes derived Equation (2) by formula (Equation (3)) which is identical to
equating the light flux entering a sidelit space the Lynes formula except that the
to the flux absorbed by all interior surfaces. 1/2(1Rmean) is changed to 1/(1Rmean 2), i.e.:
Based on empirical findings he further
expressed the effect of surrounding buildings Aglazing vis y
on the incoming flux as a function of the sky DF ¼ ð3Þ
Atotal ð1  R2mean Þ
angle, y. The rule is well established and
different versions of it appear in several design
guides.6,7,11
The only study that the authors are aware For the scale model measurements used by
Crisp and Littlefair the ‘derivate’ of the Lynes
of that ever empirically tested the Lynes
formula generated a better fit to the results
formula (Equation (2)) was presented by
than the original formula (Equation (2)). The
Crisp and Littlefair in 198412. In that paper
derivate is currently recommended in the
the authors compared predictions from
IESNA Lighting Handbook9 as well as in
Equation (2) to daylight factor measurements other guides and standards6,13. Both formulae
in a model classroom under an artificial sky. will be systematically tested further below.
From what the authors could gather from the
Crisp and Littlefair paper, measurements
2.3 Daylight uniformity – limiting room depth
were taken in a scale model of a classroom
In a sidelit space where an oversupply of
that was supposed to be 7.8 m deep, 8.8 m
daylight near the window may push the
wide and had a variable height of 2.7–4.1 m.
average daylight factor unduly up, the con-
Another parameter that was varied was the
cept of a maximum space depth for day-
wall reflectance. The effect of neighbouring
lighting can help designers to avoid too high
buildings was not taken into account. The illuminance contrast ratios. In the above-
classroom was sidelit through a window that mentioned paper Lynes also proposed a
spanned the whole width of the façade. The limiting room depth for daylighting, DLynes:
study found that the Lynes formula tended to
underestimate the mean daylight factor in the
2 . 1 1

investigated spaces with a standard deviation DLyne ¼ þ
of about 10%.The authors accordingly sug- 1  Rmean w hwindow-head-height
gested a modified version of the Lynes ð4Þ

Lighting Res. Technol. 2010; 42: 7–31

Downloaded from lrt.sagepub.com at SETON HALL UNIV on April 4, 2015


12 CF Reinhart and VRM LoVerso
where w ¼ interior room width, hwindow-head-
use of a shading device, the ratio range can
height ¼ distance between the floor and the
window head. In Lynes’ original paper the increase up to 2.5’.3 In short,
Rmean in Equation (4) was the mean area- 2  hwindow-head-height

f
weighted reflectance of the back half of a if a shading device is required
space. But, for ease of use the rule has been
generally simplified to the use of Rmean as the Ddaylight 5
mean reflectance of all interior room surfaces penetration
2:5  hwindow-head-height
including glazings. Equation (4) is based on
if no shading device is required
the assumption that if the ‘ratio of the average
daylight factor in the front half of a room ð6Þ
[which includes the window] to the average
daylight factor in the back half of the room Equation (6) was derived based on an series
exceeds three, the diversity is likely to prove of annual daylight simulations of the daylight
unacceptable’5. penetration depth in sidelit spaces for five
In the presence of external obstructions North American cities. In that study the ‘daylit
Lynes further suggested for Equation (4) that area’ was defined as ‘the area in which a task-
the limiting depth should correspond to the specific minimum illuminance is maintained
smaller measure of either DLynes or the ‘no through daylight for a significant proportion
sky line’, i.e. the line on the working plane of the year when the area is occupied. This
beyond which no direct sky can be seen proportion is also called daylight autonomy.
(Figure 2(b)). The angle a from Figure 2(b) The boundary/depth of the daylit area corre-
can usually be approximated to be equal to y. sponds to points at which the daylight auton-
Thus, the no sky line depth becomes: omy falls to half of its maximum value’.3
Dno sky line ¼ 2.4 A four-step design sequence for daylighting
ðhwindow-head-height  work plane heightÞ  tanðyÞ The basic assumption of the ‘new’ day-
ð5Þ lighting design sequence proposed in this paper
is that – once validated – Equations (1)–(6)
can be effectively combined to initially iden-
Please note that this approximation only tify potential daylit zones within a building,
holds if the height difference between the to choose appropriate zone dimensions, and
obstructing building and the centre of the to determine the required glazing area and
glazing is substantially larger than the dis- type based on desired daylight factor levels.
tance between the window head and the The resulting design sequence should have a
centre of the glazing. strong appeal for designers as it allows them
Finally, an earlier mentioned study by one to quickly establish a daylighting concept for
of the authors validated the following rela- a whole building. At this point the authors
tionship between the depth of the daylight would already like to stress that the design
area, Ddaylit area, and the window-head-height: sequence only considers diffuse daylight, i.e.
‘In a sidelit space with a standard window and it helps to design minimum façade openings
venetian blinds, the depth of the daylit area and space dimensions required to reach ade-
usually lies between 1 and 2 times the quate lighting levels throughout a space
window-head-height. The exact number for under overcast sky conditions. The effect of
a particular space is largely influenced by the direct sunlight is ignored by the sequence.
glazing type and the target illuminance level This limitation is further explored in the
in the space. If a space does not require the discussion section.
Lighting Res. Technol. 2010; 42: 7–31

Downloaded from lrt.sagepub.com at SETON HALL UNIV on April 4, 2015


A design sequence for diffuse daylight 13

Several issues have to be resolved before Equating this obstruction correction factor
the design sequence can be recommended to with the obstruction factor, OF, in the day-
practitioners: light feasibility study changes Equation (1) to:
 Suitable daylight feasibility factors, DFF,
have to be determined for different space y
types. WWR  vis  4DFF ð8Þ
908
 Equations (2)–(4) have to be tested in a
series of sidelit spaces.
 The effect of external obstructions has to be A limitation of Equation (7) for practical
expressed through a single, convenient design purposes is that the sky angle accord-
quantity in Equations (1)–(3). ing to Lynes is only defined for linear
 External obstructions must be considered obstructions, i.e. a space that is obstructed
when determining the limiting room depth. by a structure of uniform height that is
located parallel to the façade plane of the
2.5 Effective obstruction angle space (e.g. Figure 3(b)). How can the sky
This paragraph proposes a common angle be used for a discontinuous obstruction
obstruction correction factor for Equations situation such as the one in Figure 3(a)? In
(1)–(3). As explained above, Equation (2) order to answer this question one should note
uses the sky angle, y, to characterise the that the underlying notion of Equation (7) is
degree of obstruction of a space (Figure 2a). that each solid angle of the celestial hemi-
Intuitively, the effect of external obstruction sphere that is ‘seen’ from an obstructed space
on the daylight available in a building should equally contributes to the daylight within the
vanish for a sky angle of 908 (no obstructions) space. Using this simplification it is possible
and reach its maximum for a sky angle of 08 to theoretically replace any complex external
(fully obstructed sky). Following Equation (2), obstruction situation with a linear obstruction
which postulates a simple linear relationship that shades the same amount of sky. The
between y and the obstruction correction ‘effective’ sky angle resulting from this linear
factor, one gets (Equation (7)): obstruction will be used during the validation
Obstruction correction factor of Equation (2), (3) and (7). The concept of
the effective sky angle is largely identical to
y that of the ‘sky solid angle’ proposed by
¼ for 0  y  908 ð7Þ
908 Capeluto.14 Using Radiance simulations of

(a) Target space (b)

Target space

Figure 3 Two examples of external obstructions – (a) discontinuous obstruction, sky angle 608, single story target
building; (b) continuous obstruction, sky angle 608, target building same height as the neighbouring structures

Lighting Res. Technol. 2010; 42: 7–31

Downloaded from lrt.sagepub.com at SETON HALL UNIV on April 4, 2015


14 CF Reinhart and VRM LoVerso

sidelit spaces, Capeluto showed that a correc- Tregenza already showed in 1980 that mea-
tion factor of the solid angle of sky ‘seen’ by sured internal to external illuminance ratios in
an obstructed façade (measured in steradians) the same space can vary by ‘as much as twice
divided by p correctly approximates the effect or as little as half the real value’.17
of external obstructions on the daylight factor
in sidelit spaces. The ‘solid angle over p 3.1 Space descriptions
correction term effectively corresponds to the All spaces investigated in this study were
y/908 term in Equation (7). In practice, there rectangular, sidelit rooms with a single
are a number of ways to estimate this correc- window of varying dimension and position
tion factor. As shown by Capeluto, a simple within the façade. Table 1 shows all design
raytracer combined with a three-dimensional variables that were changed resulting in a total
CAD model of a building and its surround- of 2304 different spaces. All spaces had the
ings can be used to quickly model the amount same window-head-height of 2.74 m (9 ft).
of sky ‘seen’ by any particular building Spaces that had a window header with a
surface. Alternatively, a Waldram diagram height of 0.9 m had an increased floor-to-
can be used to approximate the same quantity ceiling height of 3.65 m to maintain the same
manually. window-head-height. The visual glazing trans-
mittance of the window was either 35% or
75%. The width of the window with respect to
3. Methodology the façade varied between full width and 50%
of the space width. The balustrade below the
In order to test the validity of Equations (1)–(4) main window had a height of 0.75 m and was
for sidelit spaces a series of Radiance simula- either glazed with the same visual transmit-
tions was carried out.15 Radiance is a tance as the main window or opaque. The
validated backward raytracer that can reli- overall space width varied from 2.74 to 9.86 m
ably model the daylight factor distribution in (3.6 times the window-head-height). All walls
the standard sidelit spaces investigated in this and window frames had a diffuse reflection of
study.16 Simulations were used as opposed to 50%, floors had a diffuse reflectance of 20%,
measurements since the daylight factor is and ceiling reflectances were varied between
notoriously difficult to measure reliably in 50% and 80%. External ground and neigh-
real spaces. The reason for this is that actual bouring building reflectances were set to 30%
overcast sky conditions hardly ever corre- and the thickness of all external walls was
spond to the idealised CIE overcast sky. 250 mm. The depth of all target spaces was

Table 1 Design variables that were modified for the Radiance parameter study

Design variable Range #

Glazing transmittance,  vis (%) 35 75 2


Ceiling reflectance (%) 80 50 2
Window width (–) Full façade width 50% of façade 2
width, centred
Balustrade (–) Yes No 2
Window header (–) Yes No 2
Space width (–) 1.0 * window- 1.8 * window- 3.6 * window- 3
head-height head-height head-height
Sky angle (8) 90 75 60 45 30 15 6
Neighbouring construction (–) Continuous Discontinuous 2
Target building (–) Same height as neighboring structure Two storey 2

Lighting Res. Technol. 2010; 42: 7–31

Downloaded from lrt.sagepub.com at SETON HALL UNIV on April 4, 2015


A design sequence for diffuse daylight 15

varied to match that of the Lynes uniformity floor were simulated in all 2304 spaces using
depth from Equation (4). For example, for a Radiance. Table 2 shows the Radiance sim-
9.86 m wide and 2.74 m high space with an ulation parameters used. For all spaces the
area-weighted mean surface reflectance of following quantities were calculated: mean
48% the space depth was set to 8.25 m. These daylight factor for the front half, back half
particular depths were chosen to combine and full space; window-to-wall-ratio, total
Equation (1)–(4) into a self-consistent day- room surface area, area-weighted mean sur-
lighting design sequence. The combination of face reflectance and glazing area.
all design variables in Table 1 resulted in 24
room geometries with window-to-wall ratios
(WWR) ranging from 21% to 84%. Note that 4. Results
a WWR of 84% corresponds to a ‘fully glazed
façade’ i.e. a rough façade opening of 100%. 4.1 Lynes daylight factor formula
In order to quantify the effect of neighbour- Figure 4 shows a plot of the mean daylight
ing buildings all spaces were located on the factor calculated with Radiance versus the
second floor of a dense urban surrounding mean daylight factor according to the Lynes
(Figure 3)34. As shown in Table 1 the formula (Equation (2)) for all 2304 spaces.
surrounding landscape was defined through Upon first inspection the two prediction
three parameters. The height of all neighbour- methods agree pretty well with an R2 of 0.90
ing buildings, whether these buildings were and a slope of 0.81. The mean bias error
forming a continuous obstruction or not, and (MBE) and root mean square error (RMSE)
the height of the target building in which the for all spaces are 4% and 50%, respectively.
investigated space was located. The height of Most of the larger relative errors appear for
all neighbouring buildings was varied between daylight factors under 1%. Excluding those
six different sky angles with respect to the
target space from 908 (no obstructions) to 158
Mean daylight factor according to lynes formula (%)

5 y = 0.813x
(severe obstructions). The height or the target R2 = 0.8961
building, i.e. the building in which the
4
investigated space was located, was either
two stories – meaning that the investigated 3
building was located directly under the roof –
(Figure 3(a)) or corresponded to the height of 2
the neighbouring buildings (Figure 3(b)). The
neighbouring buildings either consisted of 1
eight ‘continuous’ blocks (Figure 3(b)) or 64
‘discontinuous’ blocks (Figure 3(a)). 0
0 1 2 3 4 5 6
Mean daylight factor according to Radiance (%)

3.2 Radiance simulations


Figure 4 Mean daylight factor calculated with Radiance
Daylight factor levels on a 0.305 m  0.305 versus the mean daylight factor according to the Lynes
m (1 ft  1 ft) grid located 0.85 m above the formula (Equation (2)) for all 2304 spaces

Table 2 Utilised Radiance simulation parameters

Ambient Ambient Ambient Ambient Ambient Direct


bounces division sampling accuracy resolution threshold

6 1000 20 0.05 300 0

Lighting Res. Technol. 2010; 42: 7–31

Downloaded from lrt.sagepub.com at SETON HALL UNIV on April 4, 2015


16 CF Reinhart and VRM LoVerso

low values (which anyhow lie below typical (Equation (3)) to Radiance-based mean day-
daylight factor target levels) from the statis- light factors. The correlation between the two
tical analysis leads to MBE and RMSE values data series is equally good as for the Lynes
of 18% and 18%, respectively. formula with an R2 value of 0.90. The main
How significant are these errors when it difference is that the slope of the best fit curve
comes to using the Lynes formula for practical in Figure 5 is 1.13, i.e. the rule tends to yield
design purposes? First, in order to judge the slightly higher values than Radiance. This is
magnitude of these errors one should remem- not surprising given that:
ber that the Radiance simulations themselves
also include errors of around 10% (MBE) and 1 1
25% (RMSE).18 Second, the key issue for a 2
¼
ð1  Rmean Þ ð1 þ Rmean Þð1  Rmean Þ
user of these tools is not necessarily the size of ð9Þ
the relative errors of both methods but rather 1

how reliably each method predicts a particular 2ð1  Rmean Þ
performance metric. Using a target mean
daylight factor level of 2%, the Lynes formula
and Radiance disagree in 304 of all 2304 spaces
Using the same analysis as for the Lynes
(13%), i.e. there are 304 spaces for which
daylight formula for daylight factor target
Radiance predicts a mean daylight factor of levels of 2% yields that Lynes modified and
2% and Lynes does not or vice versa. For these Radiance diverge in 208 spaces with the rule of
304 spaces the Lynes predictions tend to be thumb predicting for 158 spaces that the space
more conservative: there are only 18 spaces meets the 2% requirement when Radiance
(less than 1%) for which Lynes predicts a predicts the opposite. In other words the
daylight factor above 2% while Radiance does modified Lynes formula agrees more often
not. These numbers suggest that the Lynes with Radiance than the Lynes formula but
formula is a surprisingly reliable calculation when both methods diverge the rule of thumb
method to estimate the mean daylight factor in is more optimistic than Radiance. While the
sidelit spaces. overall performances of the modified and
Figure 5 compares the mean daylight factor the original Lynes formula are satisfactory,
according to the modified Lynes formula the authors tend more towards using the
original Lynes formula (Equation (2)) as it
Mean daylight factor according to lynes formula (%)

includes a ‘safety margin’. The topic is


5
revisited in the discussion section.
4

3
4.2 Daylight feasibility study
y = 1.1323x Figure 6 compares the mean daylight factor
R2 = 0.8965
2
calculated with Radiance versus the daylight
feasibility factor from Equation (8). A strong
1 correlation between the two quantities is
visible with an R2 of 0.87. The main signif-
0
0 1 2 3 4 5 6
icance of the figure is that it directly links the
Mean daylight factor according to Radiance (%) daylight feasibility factor to a daylight factor
requirement for a space. For example, a mean
Figure 5 Mean daylight factor calculated with Radiance
versus the mean daylight factor according to the daylight factor distribution of 2% roughly
modified Lynes formula (Equation 3) for all 2304 spaces corresponds to a daylight feasibility factor
Lighting Res. Technol. 2010; 42: 7–31

Downloaded from lrt.sagepub.com at SETON HALL UNIV on April 4, 2015


A design sequence for diffuse daylight 17

0.60 DFF = 0.0883 DFmean


In order to gain more insight into Equation
R2 = 0.87 (10) one can also ‘derive’ it analytically from
0.50 Equations (2), (4) and (8). Using:
Daylight feasibility factor

0.40
Ag
0.30 WWR ¼ ;
Awall
0.20

Atotal ¼ 2  Awall þ DLynes hfloor-to-ceiling
0.10 
0.00
þ w  DLynes ;
0.00 1.00 2.00 3.00 4.00 5.00 6.00
Mean daylight factor according to Radiance (%) hfloor-to-ceiling
c¼ ð11Þ
Figure 6 Mean daylight factor calculated with Radiance hwindow-head-height
versus the daylight feasibility factor (Equation (8)) for all
2304 spaces

Solving for DDF the following relationship


of 0.18. In other words, if for a given is established:
combination of external obstructions,
window-to-wall ratio, and glazing type the 3  Rmean
daylight feasibility factor lies above 0.18, then DFF4  DF ð12Þ
22:5  c
a space proportioned according to Equation
(3) should have a mean daylight factor of
around 2%. Similarly, a daylight factor crite- This is an analytically accurate equation
rion of 3% would translate into a daylight which links DDF to a minimum DF, also
feasibility factor of 0.088*3.0%  0.26. Using accounting for the mean interior surface
the best-fit line in Figure 6, Equation (8) can reflectance and the relative window-header
be rewritten as: geometry. Applying Equation (12) to the sidelit
spaces investigated in this study which have an
DFF 908 0:088  DF 908 average area-weighted reflectance of 0.44 and a
WWR4    ð10Þ ‘c’ value of 1.16 reduces Equation (12) to
vis y vis y
DFF  0.097*DF. This result is reasonably
close to the best fit slope of 0.088 from Figure
6. While Equation (12) yields more insight in
Equation (10) establishes a minimum the relationship between DDF and DF than a
WWR criterion that can be used from the slope of 0.088, the authors still recommend the
earliest design stages onward. Since the WWR use of Equation (10) ‘as is’, because neither
defined in Equation (7) excludes mullions and Rmean nor ‘c’ are typically known at the time
window frames from the window area the the daylight feasibility study is to be applied in
rough facade opening resulting from any the design sequence (Section 5).
given WWR is about 20–25% larger then
the WWR, i.e. only building zones for which 4.3 Limiting Depth
Equation (10) yields a minimum WWR below This section investigates the limiting room
80% can be ‘easily’ daylit without the use of depth criterion from Equation (4). As
any advanced daylight techniques such a light described in the methodology section, the
redirecting façade elements, high-reflective depth of all investigated spaces was chosen in
surfaces, etc. accordance with Equation (4). If Lynes’
Lighting Res. Technol. 2010; 42: 7–31

Downloaded from lrt.sagepub.com at SETON HALL UNIV on April 4, 2015


18 CF Reinhart and VRM LoVerso

original derivation of the formula is valid, ‘the as high as 20. It falls below three for a few of
ratio of the average daylight factors in the the most heavily obstructed spaces.
front and back half of . . . [all spaces Since the Lynes limiting depth formula
should] . . . lie below three’.5 Figure 7 shows does not result in spaces in which a 3:1
that this is actually not the case. The figure daylight factor uniformity criterion is met, the
shows the ratios of simulated daylight factors question arises whether its use should be
in the front and in the back half of the spaces recommended at all? To answer this question
for different sky angles. The dotted line one should first reexamine the 3:1 uniformity
corresponds to a ratio of three as predicted ratio criterion itself, i.e. do daylit spaces
by Lynes. In the absence of external obstruc- with a daylight factor uniformity ratio
tions (sky angle of 908) the ratio lies between above three exhibit ‘inadequate’ daylighting
4.8 and 6.0. With obstructions the ratio goes conditions? Probably not: Figure 8 shows
that spaces that do meet the 3:1 daylight
20.0
Continuous obstruction
factor ratio criterion have uncommonly low
18.0
Discontinuous obstruction
room depths. The solid line in Figure 8(a)
16.0 shows the daylight factor distribution in one
of the study spaces with a WWR of 28%
Ratio DFFront/DFBack

14.0

12.0 and no external obstruction. The dotted line


10.0 shows the front-to-back daylight factor
8.0 ratio for varying room depths. For the
6.0 investigated space the ratio reaches three at
4.0 Theoretical lynes uniformity ratio 2.74 m which corresponds to the space’s
2.0 window-head-height. Figure 8(b) shows the
0.0
0 15 30 45 60 75 90
corresponding 3:1 room depths for all 2304
More obstruction Sky angle (°) Less spaces. In the absence of external obstructions
the allowable room depth lies between 0.9
Figure 7 Ratio of the mean daylight factor in the front
half to the back half of all spaces plotted against the sky and 1.3 times the window-head-height.
angle. According to Lynes the ratio should be around The depth only rises to 2.5 times the
three for all spaces (dotted line)
window-head-height for a few very heavily

(a) 10 5 (b) 3
Limiting depth (multiple of window-head-height)

8 4
Daylight factor ratio
Daylight factor (%)

2
Lynes uniformity ratio
6 3

Daylight factor
4 Daylight factor ratio 2
1

2 1

0 0
0 1 2 3 4 5 0 15 20 45 60 75 90
Distance to facade (m)/Limiting room depth (M) Sky angle (°)

Figure 8 (a) Plot of the daylight factor distribution with distance to the facade in one of the database spaces with a
WWR of 28% and without external obstructions. The dotted line shows the front-to-back daylight factor ratio for
various space depths; (b) Limiting depth for all spaces leading to a front-to-back daylight factor ratio of 3:1

Lighting Res. Technol. 2010; 42: 7–31

Downloaded from lrt.sagepub.com at SETON HALL UNIV on April 4, 2015


A design sequence for diffuse daylight 19

obstructed spaces with average daylight fac- sky line rule becomes more conservative than
tors of less than one. the other two methods for sky angles below
Based on the above argument the authors 458. In fact, the daylit area for the no sky line
conclude that a front-to-back daylight factor depth becomes a narrow band that is less than
ratio below 3 is not a suitable requirement for one times the window-head-height wide for a
good daylighting. The question remains 158 sky angle. This prediction reflects that
whether one should use the Lynes limiting daylighting becomes meaningless for heavily
depth formula? As explained above, Lynes obstructed facades. Based on Figure 9 the
actually suggested in the presence of external authors recommend interpreting the minimum
obstructions to use the smaller of the limiting of the three depths as the ‘depth of the daylit
depth and the ‘no sky line depth’.5 Figure 9 zone’ for a sidelit room. The implications of
compares the Lynes limiting depth of all this interpretation for practical considerations
spaces with their corresponding no-sky-line are further discussed below.
depth. The figure also shows a limiting depth
of 2.5 times the window-head-height which
corresponds to the recommended room depth 5. A new design sequence for diffuse
for a sidelit space without a shading device daylighting
(Equation (6)).
Figure 9 shows that the Lynes limiting Based on the results from the proceeding
depths vary between 1.7 and 3.0 times the section Table 3 proposes a new daylighting
window-head-height for all investigated design sequence for sidelit spaces. The design
spaces, i.e. they sometimes go beyond the sequence is meant to serve two purposes:
upper limit of 2.5 predicted by the window- 1) It allows design teams to quickly verify a
head-height rule. The predictions of both building zone’s potential for diffuse day-
rules of thumb do not change for different lighting based on its obstruction situation
obstruction situations. The no sky line limit- and intended usage.
ing depth lies above those of the two other 2) It can be used throughout all design stages
rules for sky angles above 458. At around 458 to check whether daylighting remains fea-
all three methods more or less agree. The no sible as the building design evolves.
In the following an example application of
2 1 1
the design sequence from Table 3 is given for
Limiting depth (multiple of window-head-height)

1–R w + hwindow-head-height
4
No sky line
2.5*hwindow-head-height a heavily obstructed urban site that is
bordered by a 20 m wide street and obstructed
3
by a 40 m high existing building (Figure 10).
The program for a new building foresees
2 retail spaces at street level and private offices
starting on the eighth floor (assuming a floor-
1
to-floor height of 4m). The design sequence is
applied in the following to determine whether
both building zones can be daylit.
0
90° 75° 60° 45° 30° 15° Step 1: The effective sky angles from the
Sky angle (°)
centre of the façade for zones 1 and 2 are
Figure 9 Comparison of the Lynes limiting depth, the no y1 ¼ arctan(20m/38m)288 and y2 ¼ arctan
sky line depth, and window-head-height rule of thumb (20m/6m)738, respectively. The desired day-
for all 2304 spaces
light factor level for both zones is 2%.

Lighting Res. Technol. 2010; 42: 7–31

Downloaded from lrt.sagepub.com at SETON HALL UNIV on April 4, 2015


20 CF Reinhart and VRM LoVerso
Table 3 Design sequence for diffuse daylight

Step (1) Site conditions and programming needs:


Mentally divide your design into daylit zones, i.e. areas within your building with comparable programmatic needs for
daylight and similar exposure to the celestial hemisphere. Calculate the effective sky angles, y (in degree), and define
target daylight factors, DF (in percent), for all potential daylit zones.
Step (2) Daylight feasibility test:
Pick a glazing transmittance,  vis, and determine the minimum required window-to-wall ratio, WWR, for each daylit
zone from step (1) according to:
0:088  DF 90o
WWR > 
vis y
Given that this WWR excludes mullions and frames, only zones with a minimum WWR below 80% can be realistically
daylit. Remember that increasing the visual transmittance of the selected glazing can reduce the required minimum
WWR.
! If zones that should be daylit have a minimum WWR above 80% (Aglazing ¼ 0.8 * Aopening) you should reconsider
your design scheme or space usage for these zones.
! Continue to steps (3) and (4) for zones that pass the daylight feasibility test.
Step (3) Room proportions:
Working with the formulae below select mean surface reflectances, space widths and window-head-heights for all
zones that passed step (2). The depth of the daylit area within each zone is going to be:
0   1
2 1 1
B = þ C
B 1  Rmean w hwindow-head-height C
B C
B
Depth of daylit area < minimum B ðh window-head-height  work plane heightÞ  tanð y Þ C
C
B 2:0  h if a shading device is required C
@ window-head-height A
2:5  hwindow-head-height if no shading device is required
Note that working with these formulae should be an iterative process until you have satisfactory daylighting depths
for all zones.
Step (4) Required glazing area:
In order to calculate the minimum required glazing area, Aglazing, required for each daylit zone you may assume that
the room depth of the zone corresponds to the depth of the daylight zone that you calculated in step (3). Calculate the
total interior surface area, Atotal, according to this ‘virtual room depth’ and derive Aglazing using:
Aglazing ¼ DF  2Atotal ð1Rmean Þ
vis  y

! Verify that the resulting Aglazing and  vis values are compatible with competing design requirements such as solar
gain control and glare.

Step 2: Choosing a typical double glazing


with a visual transmittance of 0.8 results
in minimum WWRs for zones 1 and 2
θ2 of (0.088*2*908)/(0.8*288) ¼ 71% and
Zone 2 32 m above street level
(0.088*2*908)/(0.8*738) ¼ 27%, respectively.
Both zones show some daylight potential
even though the daylighting potential for the
θ1 retail space (zone 1) is borderline. It is
Zone 1 Street level
therefore worthwhile to continue the analysis
Existing building Street width (20 m)
(40 m high)
New building
for both zones in steps 3 and 4.
Step 3: Assuming a window-head-
Figure 10 Section of a hypothetical urban site. The site height and a floor-to-ceiling-height for the
is bordered by an existing building and the design
sequence is applied to determine the daylighting poten- retail space of 3.5m, a space width of 10m, a
tial of zone1 (retail) and zone 2 (office) mean surface reflectance of 50%, no shading
Lighting Res. Technol. 2010; 42: 7–31

Downloaded from lrt.sagepub.com at SETON HALL UNIV on April 4, 2015


A design sequence for diffuse daylight 21

device and a workplane height of zero The analysis for the retail space shows that
results in: there will only be a 1.9 m deep daylit area
directly bordering the façade which could
Depth of daylit area ðretail Þ5minimum
0  1  1 serve as a niche application such as a seating
2 1
10:5= 10m þ 3:5m  10:3m area for a cafe. The required glazing area for
B C the retail space fell from 71% (step 2) to 31%
@ ð3:5m  0mÞ  tanð288Þ  1:9m A ¼ 1:9m
because of the depth of the daylit area is so
2:5  3:5m  8:8m low. The resulting design assumes that the
ð13Þ designer only intends to daylit a narrow band
adjacent to the façade. This is an unlikely
This shows that – due to the high obstruc- scenario given that even partial daylight is a
tion – the retail space is only going to have a sought after amenity in retail spaces. If the
very narrow daylit area very close to the designer instead ‘demanded’ a daylit zone
window. For the offices the window-head- depth of 6 m for the retail space – essentially
height is set equal to 2.8m, the space width is ignoring the result of step (3) – then the
3m, the mean surface reflectance is 50%, and required WWR would double to 59%. The
a shading device is required. This results in: resulting expanded daylight area would have a
strong daylight illuminance gradient which
Depth of daylit area ðofficeÞ5minimum would have be balanced with an electric
0 1  1
2 1 lighting system that is mostly switched on in
10:5= 3m þ 2:8m  5:8m
B C the back of the space. The designer would
@ ð2:8m  0:85mÞ  tanð738Þ  6:4m A ¼ 5:6m obviously be well advised to analyse the energy
2:0  2:8m ¼ 5:6m and glare implications of this dramatic design
ð14Þ change.

Step 4: The minimum required glazing 6. Discussion


areas for daylighting a 1.9m (5.6) deep zone
for the retail (office) space are: This section discusses the merits and limita-
Aglazing ðretailÞ tions of the proposed design sequence. Predic-
tions by the sequence are compared to those of
2  2  2ð10m  3:5m þ 10m  1:9m the LEED-NC 2.2 glazing factor method.
þ1:9m  3:5mÞð1  0:5Þ
¼
0:8  288 6.1 A sequence for diffuse daylighting
 11m2 ) WWR ¼ 31% The design sequence proposed in this paper
provides an effective and reliable means for
Aglazing ðofficeÞ designing sidelit spaces that meet a given
average daylight factor requirement.
2  2  2ð3m  2:8m þ 3m  5:6m Predictions of the sequence tend to be slightly
þ5:6m  2:8mÞð1  0:5Þ ð15Þ lower than Radiance simulations allowing for
¼ some safety margin within the analysis. Since
0:8  738
2 the sequence should be easy to learn and
 2:8m ) WWR ¼ 33%
apply it has the potential to become popular
among design practitioners. In fact, the first
The two resulting minimum WWRs of author has already used the design sequence
around 30% follow good daylighting practice. in a homework assignment in two classes for
Lighting Res. Technol. 2010; 42: 7–31

Downloaded from lrt.sagepub.com at SETON HALL UNIV on April 4, 2015


22 CF Reinhart and VRM LoVerso

architectural students at the Harvard Design mainstream, the authors maintain that, if no
School during the fall 2008 term. These initial daylight simulations are carried out because
teaching experiments confirmed that architec- of time, know-how and/or financial con-
tural students can learn how to use the design straints, then applying the validated daylight-
sequence within a 90-min lecture. The results ing design sequence is preferable to using no
for the homework assignments further sug- quantitative information or (even worse)
gested that – if used in combination with relying on rules of thumb that yield wrong
other daylighting analysis methods – the results (see below).
sequence can help designers to come up with In any instance, the design sequence should
a solid first daylighting scheme for their be interpreted as a tool for designing spaces in
projects. On the other hand, it was noted which a certain amount of daylight is avail-
that – if used in isolation – the sequence able under diffuse sky conditions. In predo-
tempts designers to oversize their windows. minantly overcast, temperate climates the
This is not surprising given that the sequence sequence should yield meaningful results by
is based on a daylight factor approach which itself. In sunnier climates the sequence
favours a ‘the more the better’ attitude obviously has to be combined with an addi-
towards daylight.10 Given these known lim- tional solar gains and glare analysis. In the
itations of the daylight factor metric one extreme case of a sunny climate and a South-
might therefore wonder whether a widespread or West-facing, unobstructed façade, the
adoption of the design sequence is actually results of the design sequence become largely
desirable? This question is addressed in the irrelevant as overheating and glare concerns
following. will (or should) determine the overall facade
As mentioned in the introduction, the design. In denser urban settings in which
daylight factor is currently under scrutiny as façades are subject to moderate or high
a daylight performance metric as it does not external obstructions the effect of direct
take direct sunlight, local climate, façade sunlight, orientation and local climate is
orientation and movable shading devices somewhat tempered and the design sequence
into account.10 Climate-based daylight per- again becomes more relevant even in sunny
formance metrics, that are based on annual climates.
daylight simulations combined with an occu-
pant behaviour model, can be used to over- 6.2 Limitations
come these limitations. Integrated As shown above, the design sequence can
daylighting/energy design tools such as be confidently applied to standard sidelit
Daylight1-2-319 and ComFen20 have been spaces with mostly Lambertian surfaces. The
specifically developed to allow simulation effect of certain design measures such as
novices to calculate more advanced metrics highly reflective ceilings can only be partly
in offices and classrooms at the earliest design considered through the sequence through an
stages. The use of such tools is therefore increased mean room reflectance. The method
recommended over the use of rules of thumb. cannot distinguish between specular versus
But, these tools follow a somewhat ‘micro- matte reflectances or between highly reflective
scopic’, space-by-space approach to daylight- ceilings versus highly reflective walls or floors.
ing design whereas the design sequence fosters Likewise, the effect of light-redirecting façade
a more holistic initial approach to daylighting elements cannot be predicted using the
that encompasses the whole building. Also, sequence. If such ‘advanced’ daylighting mea-
given that it will still take some time before sures are of interest the use of daylight
the use of simulations becomes more simulations is recommended.
Lighting Res. Technol. 2010; 42: 7–31

Downloaded from lrt.sagepub.com at SETON HALL UNIV on April 4, 2015


A design sequence for diffuse daylight 23

The sequence has been validated in ‘empty’ different scene assumptions, i.e. Crisp and
spaces. Accordingly the results cease to be Littlefair worked with a ground reflectance of
reliable if internal partitions and other furni- about 20% whereas this study used a 30%
ture are introduced. As shown in an earlier ground reflectance. This value was chosen
parametric study of open plan offices the based on a sample of reflectance measure-
presence of obstructions with a height of only ments of typical urban sidewalks. Other
1.1 m can significantly reduce the annual differences might have been caused by the
amount of daylight available beyond the different type of benchmark data that was
partition21, i.e. the sequence is only applicable used in the two studies: Artificial sky mea-
within the area between the façade and any surements in the Crisp and Littlefair study
objects higher than about 1.1 m. Likewise, and Radiance simulations in this study. While
objects such as desks which are located near it is hard to determine now which method is
the facade block all daylight incoming more reliable, one should remember that the
through glazed areas below desk height. In daylight factor is an idealised mathematical
the presence of such objects any blocked construct that is hard/impossible to measure
glazing area should not be considered in the correctly in practice17 and that scale model
calculation. Note, however, that in the measurements – same as simulations – come
absence of such objects low-lying façade with a significant error margin.22 For a self-
areas do contribute to the overall daylight shading scene parallax errors become increas-
within a space. This is especially true in the ingly important for artificial sky measure-
presence of significant external obstructions ments.23 For a daylight factor evaluation
when most daylight is reflected at least once involving self-shading one could therefore
outside before entering a space, i.e. the overall argue that simulations are more reliable
directionality of incoming daylight changes than artificial sky measurements.
from ‘downwards’ to ‘sideways’.

6.3 Lynes formula and its derivates 6.4 LEED-NC 2.2 Glazing Factor
Since its initial publication in 1979 the A spreadsheet method that bears some
Lynes daylight factor formula5 and its deri- resemblance to the Lynes daylight factor
vate (Equation (4)) have been quoted by a formula is the earlier mentioned glazing
number of design guides. Its popularity factor method, a compliance path for
underlines the attraction of easy-to-use rules ‘Daylighting Credit 8.1’ within LEED–NC
of thumb for design practitioners. As 2.224. Within LEED a compliance path is a
explained above, the comparison of predic- method to demonstrate that a building design
tions using either formula lead to good meets the intent of a given credit. The glazing
approximations of daylight factor simulations factor compliance path requires a ‘minimum
in Radiance. The authors slightly favour the glazing factor of 2% in a minimum of 75% of
original version since it provides reliable, all regularly occupied spaces’.24 The ‘glazing
slightly more conservative results than the factor’ is a somewhat confusing term that was
derivate formula. This conclusion diverges introduced in LEED-NC version 2.2. Its
from those of the Crisp and Littlefair12 study definition is identical to that of the daylight
in which the derivate was developed and factor except that it rather vaguely refers to a
which showed significantly better results for ‘known overcast sky condition’ as the refer-
the derivate formula. What might be the ence sky as opposed to the CIE overcast sky
reasons for these opposing conclusions? Part that is used by the daylight factor.25 The
of the discrepancy might have been caused by LEED reference guide provides the following
Lighting Res. Technol. 2010; 42: 7–31

Downloaded from lrt.sagepub.com at SETON HALL UNIV on April 4, 2015


24 CF Reinhart and VRM LoVerso

formula to calculate the mean glazing factor slope ¼ 0.8) the glazing factor method fails
in a space: in the presence of external obstructions
(crosses). This finding is not surprising as
Glazing Aglazing Window geometry this limitation is clearly stated in the LEED
¼  reference guide. Nevertheless, Figure 11
factor Afloor factor
clearly shows how severe this shortcoming
 vis actual Window height of the glazing factor method is given that
 
 vis reference factor ‘green’ building design generally promotes the
(re)development of urban sites. As a ‘quick
ð16Þ
fix’ Figure 12 compares the predictions of the
glazing factor method enhanced with the
Some of the elements of the Lynes formula obstruction correction factor from Equation
can be found in Equation (16) even though (7) with the Radiance simulations for the 2304
the area in the denominator is the floor area spaces:
instead of the sum of all internal surfaces. The
geometry and height factors are correction Glazing Aglazing Window geometry
factors to account for where within the ¼ 
building envelope a particular glazing is factor Afloor factor
located. The LEED Reference Guide is  vis actual Window height y
explicit in that the glazing factor calculation   
 vis reference factor 908
method ‘does not take into account light
shelves, partitions, significant exterior ð17Þ
obstructions or exterior reflective surfaces’.24
Figure 11 compares the predictions of the
glazing factor method with the Radiance Figure 12 shows that adding the obstruc-
simulations for the 2304 spaces. tion correction factor introduces a dramatic
While the two prediction methods agree improvement over Figure 11. In fact, the least
reasonably well for spaces without external square fit for all 2304 spaces now corresponds
obstructions (black dots; R2 ¼ 0.9; to R2 ¼ 0.90 with a slope of 0.79, i.e. there is
some safety margin between the prediction of
6.00
the enhanced glazing factor method and
Glazing factor according to LEED 2.2 (%)

Glazing factor with obstruction correction factor (%)

5.00 6

4.00 5

3.00 4

2.00 3

y = 0.7854x
1.00 Spaces without obstructions
2 R2 = 0.9017
Spaces with obstructions
0.00 1
0 1 2 3 4 5 6
Mean daylight according to Radiance (%)
0
0 1 2 3 4 5 6
Figure 11 Mean daylight factor calculated with Radiance Mean daylight factor according to Radiance (%)
versus the mean glazing factor according to LEED-NC 2.2
for all 2304 spaces. Spaces without external obstructions Figure 12 Mean daylight factor calculated with Radiance
are marked as black dots; all other spaces are marked as versus the enhanced glazing factor method with an
crosses obstruction correction factor of y/908 for all 2304 spaces

Lighting Res. Technol. 2010; 42: 7–31

Downloaded from lrt.sagepub.com at SETON HALL UNIV on April 4, 2015


A design sequence for diffuse daylight 25

Radiance. Comparing once more the results to reasonably accurate results compared to
from Radiance versus those of the glazing simulations, it does not lead to spaces that
factor for a target mean daylight factor level meet the formal requirement of the compli-
of 2% leads to diverging results in 329 spaces ance path, i.e. a glazing factor/daylight factor
and only 16 spaces in which the glazing factor above 2% in 75% of the space area. This is
prediction is over 2% when Radiance is not. important since several simulation programs
This excellent agreement suggests that (a) the now automatically calculate the daylight
glazing factor should be renamed again into factor on a point-by-point basis in a space
the daylight factor to avoid further confusion in order to demonstrate compliance with the
and (b) an obstruction correction factor of y/ LEED daylighting credit. Figure 13 shows
908 should be added to Equation (16). that getting the credit via individual point-by-
Finally, before closing this section, it is point calculations is harder than just meeting
instructive to reconsider once more the actual the 2% mean glazing factor criterion. This
requirement of the glazing factor method in dilemma provokes the question as to whether
LEED namely a ‘minimum glazing factor/ the whole glazing factor/daylight factor con-
daylight factor of 2% in at least 75% of all cept in LEED should be discarded altogether?
regularly occupied spaces’. How far are the The authors’ opinion in this matter is ‘yes’ but
mean daylight factor and the overall daylight not because the 75% of floor area require-
factor distribution within a space correlated? ment is not met. Field work by Roche and
Figure 13 compares the mean daylight factor Littlefair in the UK suggests that in primarily
according to Radiance to the percentage of overcast climates ‘satisfaction with daylight
floor area with a daylight factor over 2%. As can be maximised for average daylight factors
one would expect the correlation between the between 2% and 5%’.26 So, a mean glazing
two quantities is reasonably high. What might factor/daylight factor requirements of 2%
be surprising is that the percentage of floor seems reasonable. The daylight factor should
be replaced with climate-based metrics
area with a daylight factor over 2% never
because of its already mentioned intrinsic
reaches 75% for all 2304 spaces even though a
limitations as a design metric.10 Yet, this
large proportion of these spaces have a mean
change should only take place once suitable
daylight factor over 2%. This reveals that
minimum levels for one or several climate-
while the glazing factor rule-of-thumb leads
based metrics have been established and
validated. Rules of thumb for the daylight
Percentage of floor area with a DF over 2% (%)

factor will remain a useful design tool, at least


60 until comparable rules are available for
climate-based metrics.
45
y = 14.342x
R2 = 0.8222
7. Conclusion and outlook
30

This paper introduces and validates a new


15
design sequence for diffuse daylighting and
reviews several other rules of thumb. The
0
0.00 1.00 2.00 3.00 4.00 5.00 6.00
daylight feasibility study is expanded and
Mean daylight factor according to Radiance(%) linked to a minimum daylight factor criterion.
Figure 13 Mean daylight factor calculated with Radiance
Results from the Lynes mean daylight factor
versus the percentage of floor area with a daylight factor formula are compared to Radiance simula-
over 2% for all 2304 spaces tions revealing that there are only 18 out of
Lighting Res. Technol. 2010; 42: 7–31

Downloaded from lrt.sagepub.com at SETON HALL UNIV on April 4, 2015


26 CF Reinhart and VRM LoVerso

2304 spaces for which Lynes predicts a mean References


daylight factor above 2% when Radiance did
1 Galasiu A, Reinhart CF. Current daylighting
not. At the same time it was suggested that
design practice: A survey. Building Research
the Lynes limiting depth formula should be and Information 2007; 36(2): 159–174.
used in combination with the validated 2 Reinhart CF, Fitz A. Findings from a survey
window-head-height rule of thumb and the on the current use of daylight simulations in
‘no sky line’ method. Combined, the three building design. Energy and Buildings 2006;
rules of thumb form a powerful and easy-to- 38(7): 824–835.
use design sequence for diffuse daylighting 3 Reinhart CF. A simulation-based review of the
(Table 3). ubiquitous window-head-height to daylit zone
While the paper shows that the rules of depth rule of thumb: Proceedings of the
thumb used by the design sequence yield Buildings Simulation. Montreal, Canada:
IBPSA, 2005.
surprisingly reliable estimates, the LEED 4 Houghton-Mifflin. The American Heritage
glazing factor formula is less reliable for the Dictionary of the English Language,
standard sidelit space investigated. A closer 4th Edition. New York: Houghton Mifflin
investigation of the glazing factor approach in Company & Bartleby.com, 2004.
LEED-NC 2.2 suggests that an obstruction 5 Lynes JA. A sequence for daylighting design.
correction factor of y/908 should be added to Lighting Research & Technology 1979; 11(2):
the glazing factor formula to include the 102–106.
effect of external obstructions. The glazing 6 Public Works and Government Services
factor should also be renamed again into the Canada (PWGSC). PWC Daylighting Manual.
Ottawa, Canada: PWGSC, 1989.
daylight factor to avoid confusion.
7 O’Connor J, Lee E, Rubinstein F, Selkowitz S.
While the design sequence is based on the Tips for Daylighting with Windows Report
daylight factor, a design metric with several LBNL-39945. Berkeley CA, USA: Lawrence
recognised shortcomings, its use is recom- Berkeley National Laboratory, 1997.
mended by the authors with all the caveats 8 British Standards Institution. British Standard
that come with a method that only considers BS8206 Lighting for buildings – Part 2: Code of
diffuse daylight, at least until a comparable Practice for Daylighting. London: BSI, 1992.
design sequence based on climate-based 9 Illuminating Engineering Society of North
metrics becomes available. America. IESNA Lighting Handbook, Ninth
This paper has the larger goal of bridging Edition, New York: IESNA, 2000.
10 Reinhart CF, Mardaljevic J, Rogers Z.
the gap between design practice and advanced Dynamic daylight performance metrics for
daylight simulation research by clarifying sustainable building design. Leukos 2006;
within what boundaries the use of a particular 3(1): 1–20.
rule of thumb is justifiable and when design 11 Public Works and Government Services
practitioners should ‘move on’ to simulations. Canada. Canadian Tips for Daylighting.
At the same time the paper clearly demon- Ottawa, Canada: PWGSC, 2002.
strates the need to validate rules of thumb 12 Crisp VHC, Littlefair PJ. Average daylight
showing that some popular rules are less factor prediction: Proceedings of the CIBSE
reliable than others. As far as future work is National Lighting Conference, Robinson
College, Cambridge, UK: Chartered
concerned, the design sequence has thus far
Institution of Building Services, 1984.
only been validated for sidelit spaces and clear 13 Chartered Institution of Building Services
glazings. An expansion to toplit spaces and a Engineers. Lighting Guide 10: Daylighting and
larger variety of façade elements therefore Window Design. London, UK: CIBSE, 1999.
seems to be opportune at some point in 14 Capeluto IG. The influence of the urban
the future. environment on the availability of daylighting

Lighting Res. Technol. 2010; 42: 7–31

Downloaded from lrt.sagepub.com at SETON HALL UNIV on April 4, 2015


A design sequence for diffuse daylight 27

in office buildings in Israel. Building and Discussion


Environment 2003; 38: 745–752.
15 Ward G, Shakespeare R. Rendering with
RADIANCE. The Art and Science of Lighting Comment 1:
Visualization. San Francisco, USA: Morgan J-L Scartezzini (Professor, Laboratoire
Kaufmann Publishers, 1998. d’Energie Solaire et de Physique du
16 Mardaljevic J. Validation of a lighting simu- Batiment, Ecole Polytechnique Federale de
lation program under real sky conditions. Lausanne, Batiment LE, CH-1015 Lausanne,
Lighting Research & Technology 1995; 27(4): Switzerland)
181–188.
17 Tregenza PR, Waters IM. The daylight factor The quest for reliable and ‘easy-to-use’ day-
and actual illuminance ratios. Lighting lighting design methods has been a concern for
Research & Technology 1980; 12(2): 64–68. academics and practitioners in the building
18 Reinhart CF, Walkenhorst O. Dynamic domain for several decades. Their application
RADIANCE-based daylight simulations for a
at a very early stage of the building design
full-scale test office with outer venetian blinds.
Energy & Buildings 2001; 33(7): 683–697. process is crucial and should lead to appropri-
19 Reinhart CF, Bourgeois D, Dubrous F, ate decisions regarding daylighting strategies
Laouadi A, Lopez P, Stelescu O. Daylight and energy saving measures. Rules-of-thumb
1-2-3 – A state-of-the-art daylighting design as well as nomograms and protractors such as
software for initial design investigations: those that were set-up in the early 1960s by the
Proceedings of the Buildings Simulation 2007. former Building Research Station (BRS) in the
Beijing, China: IBPSA, 2007. UK, were certainly among the first available
20 Hitchcock RJ, Yazdanian M, Lee E, Huizenga design methods of that kind. Through their
C. ComFen - A commercial fenestration/facade detailed and rigorous analysis of the Lynes’
design tool: Proceedings of SimBuild 2008. daylighting design sequence, the authors have
Berkeley. CA: IBPSA USA, 2008. identified the capabilities and limits of such an
21 Reinhart CF. Effects of interior design on the
‘easy-to-use’ daylighting design method.
daylight availability in open plan offices:
Proceedings of the American Commission for an
However, some concerns must be raised
Energy-Efficient Environment 2002 Summer about the pertinence of such a rules-of-thumb
Study on Energy Efficiency in Buildings. Pacific method with respect to what simplified com-
Grove, USA: ACEEE, 2002. puter design methods can offer today. This is
22 Cannon-Brookes SWA. Simple scale models particularly true if one considers some of the
for daylighting design: Analysis of error existing daylighting design and decision tools –
illuminance prediction. Lighting Research such as the one developed by Paule in
& Technology 1997; 29(3): 135–142. Switzerland for instance (DIAL-Europe soft-
23 Mardaljevic J. Quantification of parallax ware, www.estia.ch) – which can provide
errors in sky simulator domes for clear sky almost instant access to daylight factor and
conditions. Lighting Research & Technology daylighting autonomy calculations for con-
2002; 34(4): 313–332. ventional room spaces together with an
24 US Green Building Council. LEED-NC
Version 2.2 (2006). www.usgbc.org/LEED/,
‘‘expert system’’ based appraisal (achieved
last accessed February 2009. through fuzzy logic rules) of the daylighting
25 Moon P, Spencer DE. Illumination from a design. Rules-of-thumb cannot take all aspects
non-uniform sky. Illuminating Engineering of the imprecise nature of the daylighting
1942; 37: 707–726. design process into account (wall colours and
26 Roche L, Dewey E, Littlefair P. Occupant reflectances are probably unknown by the
reactions to daylight in offices. Lighting lighting designer at the early stage). Being able
Research & Technology 2000; 32(3): 119–126. to use linguistic variables (such as ‘clear’ or
Lighting Res. Technol. 2010; 42: 7–31

Downloaded from lrt.sagepub.com at SETON HALL UNIV on April 4, 2015


28 CF Reinhart and VRM LoVerso

‘dark’ to express wall surface reflectance for finding that gives a better fit to experimental
instance) instead of fixed numbers (such as the data, but while I have been giving this
area-weighted mean surface reflectance of explanation it has always sounded to me like
interior surfaces in the Lyne’s formula for an apology. I am, therefore, well pleased to
instance) is one of the main advantage of such note that the authors do not recommend this
modern daylighting design and decision tools. substitution, and I will enjoy expunging it
It must be emphasised, moreover, that other from my teaching material.
important aspects of building design (such as It is with these thoughts in mind that I look
visual and thermal comfort, as well as sun- at the new validated sequence proposed by
lighting and solar gains for instance) cannot the authors. Of course, any design method or
easily be accounted for by the suggested rules- calculation procedure that we promote should
of-thumb-based design sequence: a more be reliable, but if it is to further the design
holistic design approach of buildings, which process it must do more than predict numer-
was not possible at the time most of these ical quantities. The components of the expres-
‘easy-to-use’ daylighting design methods were sions need to have meaning, so that designers
created, is recommended today. are not only alerted to problems, but also to
the likely causes. For the daylight feasibility
Comment 2: test, the formula for required window to wall
ratio is outstandingly simple, but where does
C Cuttle (16, McHardy Street, Havelock the 0.088 constant come from? Once again,
North 4130, New Zealand) there is the need to explain an empirical/best-
fit intrusion, but actually, I prefer the derived
It has been pleasantly reassuring to read that procedure given in the text. It gives a value of
‘the Lynes formula is a surprisingly reliable 0.097, which is considered close enough to be
calculation method to estimate the mean accepted as confirmation of the 0.088 value,
daylight factor in sidelit spaces’. During the but if that is so, cannot the value be rounded
three decades since Lynes published his off to 0.1? Or instead, a multiplier of 10 be
‘sequence’, I have taught it, albeit with some inserted in the bottom line?
spins of my own, to well over a thousand As the depth from window wall of all of the
students. spaces examined in the Radiance simulations
It is a distinct pedagogical strength of the had been determined by Lynes’ limiting depth
‘sequence’ that, by following Lynes’ deriva- formula, it was expected that in all cases the
tion of his expressions, a teacher can give front half/back half ratio would not exceed a
students insight into the process of daylight- value of 3, but it was found that ‘this is
ing and the factors, which affect its perfor- actually not the case’. That might be seen as a
mance and suitability for a given situation. failure of the formula, but it should be noted
The components of the expressions emerge in that Lynes’ procedure included checking the
logical order and take their respective posi- ‘no-sky line’, and the uniformity criterion
tions on the right-hand side of each formula, requires that both of these limits are satisfied.
and for teaching architecture students, the The authors acknowledge this in their propo-
benefit of this introduction to an aspect of sals, and proceed to add a third check: the
architectural science can hardly be overstated. window-head-height rule. My examination of
At least, that is the procedure until we reach Figure 9 does not convince me that this is a
the Crisp and Littlefair ‘derivative’ formula, useful addition, and I prefer Lynes’ Step 4 of
which substitutes (1Rmean2) for 2(1Rmean). his ‘sequence’ in which he discusses how a
This has to be explained as an empirical designer who has determined the total
Lighting Res. Technol. 2010; 42: 7–31

Downloaded from lrt.sagepub.com at SETON HALL UNIV on April 4, 2015


A design sequence for diffuse daylight 29

window area in the previous step might design tools such as Dial-Europe. In fact, the
proceed to determine the shapes and positions corresponding author lead the development
of the windows. I think this is an aspect of the of an integrated daylighting/energy design
design process where architects are more tool called Daylight1-2-3 at the same time as
likely to accept advice than rules. we developed this design sequence.1
I need more time to work with these In our opinion the use of rules-of-thumb
proposals and to think through their implica- and computer-based design tools perfectly
tions before I am willing to move on from the complement each other: Given the way build-
familiar Lynes’ sequence. Nonetheless, I ing design is taught (and practiced) design
applaud the authors for bringing this key practitioners tend to initially follow an ‘out-
component of daylighting design into such side in’ approach not only towards day-
sharp focus. I am sure that they have noted lighting but towards all building design
that Lynes claimed that application of his aspects. Building form is developed in
average daylight factor and window area response to aesthetics, urban considerations,
formulae ‘are not restricted to side windows’. programmatic requirements and, increasingly,
Can we hope that they will apply similar environmental concerns. Once a basic build-
scrutiny to clerestories and skylights? ing form has been conceived different façade
variants can be explored. At this point within
Reply to comments: the design process, tools such as Dial-Europe
CF Reinhart and VRM LoVerso and Daylight1-2-3 can in fact support better
building design. But – since both tools only
We thank Kit Cuttle and Jean-Louis examine a single room at a time – they make
Scartezzini for their thoughtful comments, the designer take on an occupancy-centred,
which reveal how dramatically attitudes ‘inside out’ approach towards daylighting.
towards rules-of-thumb vary among daylight- One could argue that design advice provided
ing experts. Dr Scartezzini doubts their by these tools, although useful, is already ‘too
remaining pertinence for today’s design prac- late’ since the tools can help make a West
tice and instead recommends the use of facing façade better for daylighting but – at
computer-based design decision tools such this point in the design process – they cannot
as Dial-Europe. His main criticism of the easily lead the designer towards avoiding this
proposed design sequence is that it does not façade orientation in the first place. Rules-of-
consider visual and thermal comfort as well as thumb on the other hand can act as form-
solar gain control. Mr Cuttle on the other givers for buildings from the earliest design
hand has already been an avid promoter and phases onwards. Figure 1 shows an example
user of Lynes’ original design sequence for of two partly shaded façades whose window
architectural education and suggests that a design has been fully derived from the
rules-of-thumb should alert designers ‘not daylight feasibility study: The window-to-
only of [design] problems but also of likely wall-ratio falls for higher levels with less
[physical] causes [for these problems]’. He obstruction by neighbouring buildings. As
further proposes a simplification of the day- the required window-to-wall ratio reaches the
light feasibility study replacing the empirically upper feasibility limit for the lower levels the
derived 0.088 with a factor of one tenth. windows disappear since these levels ‘cannot
We agree with Mr Cuttle’s observation be daylit anyhow’. The example is ‘radical’ in
regarding the usefulness of rules-of-thumb for that it leads to a design that is most likely not
daylighting design. At the same time we are acceptable to the building occupants. But, it
also promoting the use of computer-based carries a powerful educational message and
Lighting Res. Technol. 2010; 42: 7–31

Downloaded from lrt.sagepub.com at SETON HALL UNIV on April 4, 2015


30 CF Reinhart and VRM LoVerso

Figure 1 Visualisation of a window design for a partly shaded façade derived by the daylight feasibility study
(visualisation and design by Jeff Niemasz)

highlights how effectively rules-of-thumb may changing the daylight feasibility test to
act as formgivers for design. The example Equation (1):
further shows that – as already mentioned in
the paper – the proposed design sequence
does not take overall energy use and/or glare DF 90
WWR4 ð1Þ
into account and has been derived for facades 10  vis y
that are mostly receiving diffuse daylight.
There is certainly a need to further expand the
sequence and the authors invite others to join Teaching the design sequence, the first
in this effort. author also found many students taking
Since we wrote the paper the first author exception to the no-sky-line limiting rule
has taught the design sequence to over one depth since it frequently leads to seriously
hundred architectural students and practi- limited daylit areas. In this sprit we now stress
tioners and, based on this experience, we the fact that the decision of what constitutes a
follow Cuttle’s recommendation to simplify ‘well daylit’ space rests with the designer
the daylight feasibility formula since (a) many including the ‘decision’ whether a space
students were confused as to where the 0.088 without visible access to the sky can be
comes from, (b) the number suggests a called ‘daylit’.
precision that is not there and (c) many As Dr Scartezzini suggests, interior room
students simply mistyped the formula repla- reflectances are typically unknown at the
cing 0.088 e.g. with 0.88. We are therefore early stages of design and qualitative terms
Lighting Res. Technol. 2010; 42: 7–31

Downloaded from lrt.sagepub.com at SETON HALL UNIV on April 4, 2015


A design sequence for diffuse daylight 31

such as dark, medium or bright are therefore simulation program is being used.3 At this
more adequate than absolute levels. For the point the modeler should carefully review the
2304 investigated spaces the mean interior model before proceeding to more advanced
room reflectance varied between 37% and simulations of e.g. climate-based metrics and/
49%. In reference to the CIBSE ‘Lighting or glare. This use of the Lynes formula as a
Guide 10: Daylighting and Window Design’2 simulation ‘crosscheck’ was successfully used
we are therefore suggesting values of 40%, during two practitioner workshops that the
50% and 60% for spaces with light, medium first author gave during the summer of 2009.
and dark surface reflectances.
We agree with Mr Cuttle that the Lynes
formula still needs to be systematically tested References
for non-sidelit spaces and we are currently
working on a study to this effect. 1 Reinhart C, Bourgeois D, Dubrous F, Laouadi
In closing, we are leaving the reader with a A, Lopez P, Stelescu O. Daylight 1-2-3 – A state-
final thought. Even if the reader decides not to of-the-art daylighting design software for initial
use a design sequence that is based on the design investigations: Proceedings of the
daylight factor, the fact that the Lynes Buildings Simulation 2007. Beijing, China:
formula leads to such highly correlated results International Building Performance Simulation
compared to Radiance simulations makes it a Association, 2007.
2 Chartered Institution of Building Services
powerful quality control test for computer Engineers. Lighting Guide 10: Daylighting and
models. If a computer model of a typical Window Design. London, UK: CIBSE, 1999.
sidelit space predicts a dramatically different 3 Ibarra D, Reinhart CF. Daylight factor
mean daylight factor than the Lynes formula, simulations – How close do simulation beginners
it is probably corrupt, e.g. wall thicknesses are ‘really’ get? in Building Simulation 2009.
not taken into account, material properties Glasgow, Scotland: International Building
are not set properly and/or an unreliable Performance Simulation Association, 2009.

Lighting Res. Technol. 2010; 42: 7–31

Downloaded from lrt.sagepub.com at SETON HALL UNIV on April 4, 2015

You might also like