Basic HTML Version
NORDIC LIGHT & COLOUR
89
Firstly, since the estimate is derived from daylight factors, it
requires only a modest enhancement to existing software tools
that predict DFs. Next, it provides some ‘connectivity’ to the
prevailing climate. A target that has been proposed is that a
side-lit design should achieve 300 lux across half of the work
plane for half of the year when the sun is above the horizon.
To achieve this for say, Stockholm, half of the sensor points
must have a DF of 2.5% or greater, whereas for the Madrid
the ‘target’ DF would be 1.8%. Note, the target is based on the
same criterion for internal daylight provision, it is of course the
greater prevailing diffuse illuminance for Madrid compared to
Stockholm that results in the lower ‘target’ daylight factor for
the Spanish capital. There are other advantages – a median ap-
proach informs on the spatial distribution of daylight whereas,
as noted earlier, the average daylight factor value does not
(Mardaljevic and Christoffersen, 2013).
Artificial skies
An artificial sky provides a controlled means of illuminating a
scale model for the purpose of taking measurements and also
for qualitative appraisal (Hopkinson, 1963). The most common
artificial sky is the ‘mirror box’ design. This has a horizontal
sheet of white diffusing material forming the top of the box.
The sheet is evenly lit from behind (i.e. from above) by lamps.
The four (or more) vertical sides of the box are mirrors. These
create a sky vault that extends seemingly to infinity on all sides
due to multiple reflections between the mirrors. Measure-
ments have shown that the luminance pattern in mirror box
skies can approximate that of the CIE overcast sky, and so
these can provide a controlled luminous environment for the
determination of daylight factors. Many of the larger schools
of architecture had artificial skies at one time or another, but
they tend to be less used since the computer-based methods
became more common.
Before carrying out any ‘high precision’ scale model studies
in an artificial sky, the user should question what the goal is,
and also if it justifies the effort involved. The user learns many
skills in constructing the models and carrying out systematic
measurements, and the value of that should not be overlooked.
However, we know from the work of Cannon-Brookes and
others that absolute accuracy of even the most carefully con-
structed models can be much less than is generally assumed
(Cannon-Brookes, 1997). Students should be made aware of
these limitations and not strive to achieve an ‘accuracy’ that
may indeed be illusory.
Considerable capital investment has been made in the con-
struction and operation of full-hemisphere sky simulator
domes for the evaluation of daylight in physical models. As far
as the author is aware, full-hemisphere domes cannot repro-
duce absolute illumination values. Nor can the the illumination
effect of the sun be modelled simultaneously with that of the
sky, at least not without reducing the absolute illumination
from the sky lamps to minuscule levels to maintain the correct
relative level with that from the sun lamp. I suspect that most
of us would struggle to notice the difference in visual appear-
ance between a model illuminated by a full-hemisphere sky
simulator and the same model under an improvised ‘sky’ com-
prised of some thin cloth and a few lamps. In consequence, the
author remains skeptical that viewing a model illuminated by a
sky simulator dome can offer any meaningful insight. Indeed,
the perceived benefits of model viewing under seemingly ‘con-
trolled’ conditions are, I believe, largely illusory, and if it must
be done, then it may as well be under a real sky with a real
sun. Furthermore, because of the limitations regarding abso-
lute levels, one practice occasionally seen is a designer viewing
a model under a clear sky distribution
without
sun. This is of
course an illumination condition that cannot occur in nature
and the value of using this approach must be questioned.
Climate-based daylight modelling
The accurate prediction of daylight in spaces under realistic
sun and sky conditions, and for many instances e.g. hourly for a
full year, was first demonstrated in the late 1990s (Mardaljevic,
2000)(Reinhart and Herkel, 2000). Climate-based daylight
modelling (CBDM) is over a decade old and has been used ef-
fectively on a number of projects large and small, e.g. from the
New York Times Building to residential dwellings. CBDM tools
are however still largely the preserve of lighting simulation ex-
perts and researchers. For CBDM to become mainstream the
software to do it needs to be taken up and supported by one or
more major software houses. Here lies a classic ‘chicken and
egg’ conundrum. On one hand, those who draft guidelines are
reluctant to recommend metrics founded on CBDM because
tools to predict the metrics are generally not available, at least
not as software supported by one of the major vendors. On the
other hand, the software vendors are understandably loathe
to dedicate the resources to develop and maintain CBDM tools
because – inasmuch as climate-based metrics are not in the
guidelines – there will be no real market for these new tools.
This presents something of an impasse to all those who strive
to advance daylighting standards beyond the current guide-
lines. Furthermore there is the risk of a “skills gap” developing
in daylight modelling between a small core of CBDM experts
and the rest who see little prospect for developing those expert
skills inhouse. In other words, there is a risk that the ‘head’
(i.e. those with CBDM skills) could separate from the ‘tail’ (i.e.
those without CBDM skills), leading to fragmentation in the
practitioner user base and barriers to knowledge transfer.