Download as pdf or txt
Download as pdf or txt
You are on page 1of 18

Lighting Res. Technol. 2014; Vol.

46: 619–636

LRT symposium ‘Better metrics for


better lighting’ – a summary
PR Boyce PhDa and KAG Smet PhDb
a
Canterbury, UK
b
ESAT, Light and Lighting Laboratory, KU Leuven, Ghent, Belgium

Received 29 August 2014; Revised 3 October 2014; Accepted 14 October 2014

A metric is a well-defined measure that a designer can use to evaluate a design.


For many years, the most widely used metrics in general lighting practice have
been the average illuminance, the associated illuminance uniformity, the daylight
factor, the correlated colour temperature and the CIE General Colour Rendering
Index of the light source, the Unified Glare Rating and the power density of the
installation. All these metrics have limitations with the result that there are now
many proposals for replacing them with metrics more suited to delivering better
lighting. This paper, which is based on the presentations given at the LRT
symposium held in London on the 15 July 2014, considers the desirable attributes
of a metric, summarises the proposed new metrics and considers how likely they
are to be adopted and, if they are, whether or not they are likely to lead to better
lighting.

1. Introduction generated and published to promote better


lighting. Following such guidance will cer-
The design of lighting installations is con- tainly do enough to avoid bad lighting, but it
strained by laws and regulations. Laws and may not be enough to ensure good lighting.6
regulations are produced by governments in Strictly, such guidance does not have to be
pursuit of specific objectives. For lighting, followed, but often it can be considered to
these objectives can range from ensuring the have legal force because it is taken to repre-
health and safety of the public1,2 to limiting sent best practice. Any designer who ignores
the use of electricity.3 Laws are typically best practice will have a difficult case to make
framed to express general principles. if faced with litigation.
Regulations are the means by which laws A common element of both regulations
are put into effect because regulations, such and a lot of guidance documents are metrics.
as the UK Building Regulations, can be A metric is a well-defined measure that the
prescriptive and quantitative and can more designer can use to evaluate a design. The
easily be changed when desired. Laws and the essential characteristics of a metric are that it
associated regulations have to be obeyed. should be quantitative, have application cri-
The designer of lighting installations will teria attached to it, be capable of being
also be aware of another set of constraints, calculated during design and measured after
the guidance issued by professional organisa- installation as well as having some relation to
tions in the form of lighting recommenda- the desired objective. For many years, the
tions.4,5 These recommendations are most common metric for lighting design has
been the illuminance on the task. Indeed,
Address for correspondence: PR Boyce, 60 Riverside Close, design criteria in the form of tables of
Bridge, Canterbury, Kent CT4 5TN, UK.
E-mail: prb.lrt@btinternet.com illuminances recommended for different

ß The Chartered Institution of Building Services Engineers 2014 10.1177/1477153514558161

Downloaded from lrt.sagepub.com at Harvard Libraries on June 22, 2015


620 PR Boyce and KAG Smet

tasks have been published for almost 100 used that can be calculated but not measured.
years,4,5 although the values have oscillated One relates to the ability of a light source to
somewhat over that time depending on the render colours accurately relative to a refer-
available technology, the cost of electricity ence light source, the CIE General Colour
and peoples’ expectations. Illuminance is a Rendering Index (CRI).8 Another indicates
good example of a metric because it is well the colour appearance of the light produced
defined;7 it has criterion levels for different by a light source, the correlated colour
applications; there are several different meth- temperature (CCT).7 The third measures the
ods by which it can be calculated and inex- effectiveness of a light source in converting
pensive meters by which it can be measured. electrical power to light, luminous efficacy
Its weakness is that it is only approximately measured in lumens/watt. The fourth
related to how visible a task will be. Task attempts to predict the level of discomfort
visibility will actually depend on several other glare produced by an electric lighting instal-
factors, among them being the visual size, lation, the unified glare rating (UGR).9 A
luminance contrast and colour difference of search of any light source manufacturer’s
the task seen against its immediate back- catalogue will reveal how widely the first three
ground as well as the luminance to which the of these metrics are used.
visual system is adapted.6 Nonetheless, illu- All of these metrics have limitations. The
minance is what a lighting installation delivers objective of this paper is to review what these
and, given a fixed diffuse reflectance, illumin- limitations are and to consider the merits and
ance determines what the luminance of the potential of some of the alternative metrics
surface will be and luminance is what deter- that have been proposed. The order in which
mines the state of visual adaptation. Further, the alternative metrics will be discussed is
it is rare for the designer to have an idea of determined by how close they are to being
exactly what tasks will be done under a given adopted. Some are very close while others are
installation, where they will occur and how far over the horizon.
they might change over time. Thus, the best
that can often be done is to provide enough 2. Lighting efficiency
illuminance to allow the visual system to deal
easily with whatever confronts it. This The ultimate objective of metrics associated
explains why illuminance has been such a with lighting efficiency is to limit the amount
long-lived metric. of electricity used for lighting without exces-
But there are several other metrics that are sively restricting lighting design. There are
widely used in design.4,5 For daylighting, the several metrics that have been used in
most common metric is daylight factor. This attempts to do this. Some are concerned
is the ratio of the illuminance received at a simply with components of a lighting instal-
specified point in an interior to the unob- lation such as the light source and luminaire
structed exterior horizontal illuminance (i.e. used. Others focus on limiting power density,
from where a real or simulated photocell has but only one actually attempts to quantify
a complete ‘view’ of the sky). For electric energy consumption. They are:
lighting, there is a metric for illuminance Light source luminous efficacy: This metric
uniformity over an area, usually expressed as quantifies the efficiency with which a light
the ratio of the minimum illuminance to the source converts electrical power to light and is
average illuminance. measured in lumens/watt. If the light source
Both these metrics can be calculated and requires control gear to operate, the power
measured, but there are four others widely demand of the control gear should be
Lighting Res. Technol. 2014; 46: 619–636

Downloaded from lrt.sagepub.com at Harvard Libraries on June 22, 2015


LRT symposium summary 621

included. Setting a minimum allowed lumi- density for different building types, the
nous efficacy ensures that only energy effi- number of inefficient lighting installations
cient light sources are used or developed in should be reduced, but it imposes some
the future. restrictions on lighting design and operation.
Luminaire efficiency: Luminaires differ in Lighting energy numeric indicator (LENI):
their ability to collect and emit the luminous The LENI is different from all the above
flux produced by a light source. The metric metrics in that it actually quantifies energy
that partly quantifies this ability is the light rather than power demand. LENI is defined
output ratio (LOR). This is simply the ratio of as the annual energy consumed by a lighting
the luminous flux emitted from the luminaire installation per square metre of lit area.4,10 It
to the luminous flux emitted by the light is measured in units of kWh/m2/year. LENI
source in the luminaire. However, the shape can be calculated or measured. For the
of the light distribution is very important in calculation, information is required on the
lighting design, but this is not reflected in a installed power demand for illumination and
simple LOR measure. Setting a minimum for standby purposes, the operating periods
LOR ensures that only energy efficient lumin- by day and night, and the effectiveness of any
aires are used, but this can handicap lighting controls relating to occupancy, daylight avail-
design. ability and maintaining constant illuminance.
Lighting efficacy: The efficacy of a lighting To do the calculations, either actual or
installation is measured as the initial lumi- default values can be used. LENI can also
nous flux of the entire installation divided by be measured with metering designed to spe-
the power demand of the entire installation cifically monitor the use of the lighting. If
and is measured in lumens/watt. By setting a default values are used in the calculation of
minimum value for this metric, inefficient LENI, they should be verified after construc-
installations can be eliminated. Allowance tion by metering.
can be made for the use of controls by It should be appreciated that these metrics
lowering the minimum lighting efficacy of lighting efficiency are not mutually exclu-
required. sive. Indeed, it is very likely that an installa-
Installed power density: The installed power tion that meets any LENI limit will have to
density of an installation is the power demand use a light source with a high luminous
of the whole installation divided by the area efficacy and a luminaire with a high LOR
to be lit and is measured in watts/square which means these metrics will continue to be
metre. By setting a maximum allowed value useful. LENI can be applied to any type of
for installed power density for different interior lighting installation, task/ambient
building types, the number of inefficient lighting as well as uniform lighting. LENI is
lighting installations should be reduced. included in the European Standard EN 15193
However, such restrictions inevitably limit Energy Performance of Buildings.10 It has
lighting design. been adopted by a number of European
Working plane power density: Installed countries, a trend that seems likely to con-
power density obviously increases as the tinue. It should also be appreciated that
illuminance provided by the installation LENI and all the other metrics discussed
increases. The working plane power density above have nothing to say about the visual
avoids this problem by expressing the quality of the lighting provided. A good
installed power density per 100 lx falling on lighting installation has to enable people to
the working plane. Again, by setting a max- see what needs to be seen, in comfort, as well
imum allowed value for working plane power as being energy efficient.
Lighting Res. Technol. 2014; 46: 619–636

Downloaded from lrt.sagepub.com at Harvard Libraries on June 22, 2015


622 PR Boyce and KAG Smet

3. Daylighting illuminances delivered is limited, and hence


daylight’s interaction with electric lighting
Unlike the situation for the efficiency of control systems is uncertain. Daylight factor
electric lighting, the metrics associated with is best considered as a rule of thumb. This is
daylighting can have three objectives. One is most apparent in the very limited amount of
to characterise what constitutes good day- quantitative advice given about daylighting
lighting. Another is to provide an estimate of which usually consists of little more than
the effect of daylight on the energy consump- some minimum average daylight factors.4
tion of the building. Yet another is to indicate Such advice tends to be interpreted as the
the likelihood of visual or thermal discomfort higher is the average daylight factor the better
occurring. The metric most commonly used to is the lighting, Sadly, this is not true. If it was,
assess daylighting is the daylight factor. This we should all be living in greenhouses. The
is the ratio of the illuminance received at a reason why a higher daylight factor is not a
given point on the horizontal working plane recipe for better lighting is that the windows
in an interior to the unobstructed exterior or rooflights used to deliver daylight can also
horizontal illuminance, usually expressed as a deliver solar heat gain causing thermal dis-
percentage. Daylight factor assumes a CIE comfort and glare causing visual discomfort.
overcast sky which is rotationally symmetric Thus, daylighting design is a matter of
about the vertical axis and for which the balancing positive and negative effects.
luminance at the zenith is three times greater Daylight factor is too simple a metric to do
than the luminance at the horizon. Daylight this. Despite this, daylight factor has been
factor does not take direct sunlight into used in some accreditation schemes with some
account, although it does deal with externally unintended consequences
and internally reflected light as well as light Despite its limitations, in the hands of an
from the sky. The extent to which daylight experienced designer daylight factor can be a
factor meets the three objectives of a day- useful rule of thumb to make design deci-
lighting metric is very limited.11 This is sions,12 but are there any alternative metrics
because it is a relative static measure that is for daylighting? The answer to this question is
determined by the design and setting of the positive, but it requires a very different
building and that ignores the contribution of approach. This approach is known as cli-
sunlight. However, daylight is dynamic in that mate-based daylight modeling (CBDM).11
it varies over the hour, day and year depend- The key to this approach is the existence of
ing on the climate and, even for the same data files of the irradiances or illuminances
climate, varies with the orientation of the provided by the sky and sun for a given
building. In terms of the three objectives, location over a whole year. Such climate
daylight factor does little to identify good records are now available for many locations
daylighting, partly because what constitutes throughout the world. From such records, it
good daylighting is ill-defined, but mainly is possible to calculate the illuminances that
because it ignores the effect of sunlight. will be produced at a given point in a space
Ignoring sunlight also means any attempt to depending on the prevailing climate, the
use it to predict the extent to which visual or window or skylight arrangement, and the
thermal discomfort will occur is futile. As for orientation of the building, every hour, for a
the energy consumption of a building, the fact whole year. These illuminances are absolute
that daylight factor is a relative measure that measures and can be calculated for an
ignores changes in sky type throughout the assumed horizontal working plane at desk
year means its ability to predict the height or on the ceiling, or even on a vertical
Lighting Res. Technol. 2014; 46: 619–636

Downloaded from lrt.sagepub.com at Harvard Libraries on June 22, 2015


LRT symposium summary 623

plane so that the illuminance at the eye can be space will have the recommended illuminance
calculated. These measures give a better delivered by daylight.14 By making this cal-
indication than daylight factor of how sky- culation for multiple positions, a contour
light and sunlight will be distributed across a defining an area over which a specific min-
space, where and how often visual discomfort imum DA occurs can be developed. The
caused by glare is likely to occur, and what larger the area enclosed by the contour, the
the stimulus to any ceiling-mounted sensors more likely it is that the space will be
for an electric lighting or blind control system considered to be daylit. DA is useful as a
will be. This is clearly a much richer picture of design tool for those who want to make the
how a daylighting installation will perform most use of daylight, but it has no relevance
than is possible with daylight factor alone. for assessing visual discomfort caused by
There are two metrics associated with glare and limited value for estimating the
CBDM that use the calculated illuminances. energy consequences of daylighting.
One is the useful daylight illuminance (UDI). Any metrics using CBDM require com-
Based on experience, a four-part illuminance puter simulations, and these can be expensive
scale has been developed. When daylight to create depending on the complexity of the
delivers less than 100 lx at a point, it is largely building. This may be the reason why, despite
ignored. When daylight delivers more than their advantages, UDI and DA have not
3000 lx, blinds are likely to drawn down so replaced daylight factor as the basis of day-
such high illuminances are not useful. The lighting design. The problem is that designers
intermediate range is subdivided into two are unwilling to move on from daylight factor
parts. From 100 lx to 300 lx, daylight will be until software exists that make UDI or DA
noticed but will usually be considered as easy to use, and software companies are
supplementary to electric lighting. From unwilling to develop the necessary software
300 lx to 3000 lx, daylight will be considered until there is evidence of a demand. But this
primary and, given a responsive control logjam may be about to break. The Priority
system, electric lighting is likely to be Schools Building Program in the UK has
dimmed or turned off. This means that only recently specified that proposed daylighting
illuminances in the range 100 lx to 3000 lx are designs for schools will be assessed on the
useful. By examining the distribution of basis of UDI thereby forcing people who wish
illuminances delivered over a space, over a to gain access to the program to use UDI, a
year, it is possible to estimate the extent to requirement that will generate the demand
which the design delivers useful daylight and necessary to stimulate the software compa-
hence the energy consequences. It is import- nies. This appears to be happening – during
ant to appreciate that UDI does not really the drafting of this paper two software
deal with discomfort glare apart from limiting companies announced the addition of
the illuminances received to less than 3000 lx. CBDM to their building modeling tools.
There are more comprehensive methods for If the use of UDI leads to schools that
dealing with discomfort glare from windows, show reduced energy consumption and less
and those should be used to supplement overheating in summer than those designed
CBDM.13 using daylight factor, the future for CBDM is
The other metric associated with CBDM is bright. If it does not, then daylight factor and
called daylight autonomy (DA). This is a other rules of thumb are likely to prevail. In
measure of daylight availability. DA is either case, there is still much to learn about
defined as the percentage of the occupied metrics for daylighting. While moving from
time over the year when a given position in a only considering an overcast sky through
Lighting Res. Technol. 2014; 46: 619–636

Downloaded from lrt.sagepub.com at Harvard Libraries on June 22, 2015


624 PR Boyce and KAG Smet

daylight factor to assessing the real climate by CCT is a very convenient and easily
CBDM is a step forward, until what visual understandable metric of light source colour
conditions constitute good daylighting are appearance, applicable to nominally white
defined it will be difficult to develop suitable light sources. As a rough guide, such light
metrics. sources have CCTs ranging from 2700 K to
17,000 K. A 2700 K light source, such as an
incandescent lamp, will have a yellowish
4. Colour colour appearance and be described as
‘warm’, while a 17,000 K light source, such
Colour is a perception. This means that the as some types of fluorescent lamp, will have a
colour of a surface is not solely a function of bluish appearance and be described as ‘cool’.
the light spectrum incident on it and the The CCTs of the most commonly used light
spectral reflectance of the surface. While these sources lie in the range 2700–5000 K. It is
factors are undoubtedly important, the colour worth noting that two light sources with the
perceived will also depend on its immediate same CCT can have different colour appear-
surroundings and the capabilities of the ances. This is because the two light sources
observer’s visual system.15 Despite this, the can lie at different points along the specific
metrics most commonly used to characterise iso-temperature line and hence have different
colour utilise only the spectrum of the inci- chromaticity coordinates. It is also important
dent light and the spectral reflectance in their to be aware that light sources with the same
calculation. There are two common metrics CCT can have different light spectra so how
used; the CCT and the CIE General CRI. well they render surface colours can be
different.
4.1. Correlated colour temperature Another aspect of CCT that requires
The CCT is used as a metric to describe the attention is how far the chromaticity coord-
colour appearance of the light emitted by a inates of a light source can be away from the
light source. CCT is based on the fact that the Plankian locus if the light emitted is to be
spectral emission of a black body is defined considered white. A metric, called Duv, is the
by Planck’s radiation law and hence is a distance between the chromaticity coordin-
function of its temperature only. This pro- ates of the light source and the nearest point
duces a curved line on the CIE 1931 chroma- on the Planckian locus, measured on the CIE
ticity diagram, called the Plankian locus or 1976 UCS diagram. The maximum allowed
black body line.6 When the chromaticity Duv values for white light are 0.006, posi-
coordinates of a light source lie directly on tive values being for chromaticity coordinates
the Planckian locus, the colour appearance of above, and negative for below, the Planckian
that light source is expressed by the colour locus.16 Both CCT and Duv are necessary to
temperature, i.e. the temperature of the black define the colour appearance of a light source.
body that has the same chromaticity coord- While setting a limit to how far away from
inates. For light sources that have chromati- the Planckian locus the chromaticity coord-
city coordinates close to the Planckian locus inates of a light source can be while still
but not on it, their colour appearances are claiming to produce white light is necessary,
quantified in CCT, i.e. the temperature of an there may still be some work to do in this
iso-temperature line that is closest to the area. This is because there is some evidence
actual chromaticity coordinates of the light that below 4000 K, the chromaticity coordin-
source. The temperatures are usually given in ates considered to represent white light depart
Kelvin (K). from the Planckian locus.17 This somewhat
Lighting Res. Technol. 2014; 46: 619–636

Downloaded from lrt.sagepub.com at Harvard Libraries on June 22, 2015


LRT symposium summary 625

undermines the use of the Plankian locus as different because the differences that define
the basis of the light source colour appear- the Special CRI values only take the magni-
ance and suggests the need for a different tude of the difference into account, not the
function. direction. Second, different light sources are
being compared with different reference light
4.2. CIE General Colour Rendering Index sources. This makes the meaning of compari-
The CIE General CRI measures how well a sons between different light sources uncertain.
given light source renders a set of standard Third, chromatic adaptation is dealt with by
test colours relative to their rendering under a using the von Kries transform which has been
reference light source of the same CCT as the found to be inadequate.19 Fourth, none of the
light source of interest.8 The reference light test colours used in the calculation of General
source used is an incandescent light source for CRI are saturated. Fifth, the CIE test samples
light sources with a CCT of 5000 K and less do not sample wavelength space uniformly,
and some form of daylight for light sources i.e. some areas will be underrepresented.20
with a CCT above 5000 K. The actual calcu- The non-uniform spectral sampling of wave-
lation involves obtaining the positions of a length space enables manufacturers to take
surface colour in the CIE 1964, U*, V*, W* advantage of these differences to game their
colour space under the reference light source spectra, i.e. to artificially inflate the General
and under the light source of interest after CRI value.20 Sixth, there has to be some
allowing for any chromatic adaptation by a doubt about whether the reference light
von Kries transform. The difference between sources, either incandescent or daylight, rep-
the two positions is then expressed on a scale resent perfect colour rendering.
that gives perfect agreement between the two
positions a value of 100. The CIE has 14 4.3. New colour metrics
standard test colours. The first eight form a While the CIE General CRI remains the
set of pastel colours arranged around the hue most widely used metric for describing the
circle. Test colours 9–14 represent colours of ability of a light source to render colours,
special significance, such as more saturated the arrival on the market of solid state light
primary colours, skin tones and vegetation. sources producing white light from a com-
The result of the calculation for any single test bination of three or more narrow band LEDs
colour is called the CIE Special CRI, for that has caused some consternation. The problem
test colour. The average of the Special CRIs is that such solid state light sources produce
for the first eight test colours is called the CIE values of CIE General CRI that do not match
General CRI. It is this latter index that is how people evaluate their colour rendering.21
usually presented in light source manufac- The result has been the same as occurred
turers’ catalogues. when fluorescent lamps first became wide-
The CIE General CRI has its limitations.18 spread, a growth of interest in colour metrics.
First, it should be appreciated that just The outcome of this interest has been an
because two light sources have the same outburst of new metrics. These can be divided
General CRI, it does not mean that they into two types.
render colours the same way. The General The first type is focused on measuring
CRI is an average, and there are many colour fidelity. To measure colour fidelity, it
combinations of Special CRI values that can is essential to have a reference illuminant. The
give the same average. Even when all the fidelity in question then becomes how well the
Special CRI values for two light sources are light source in question renders colours rela-
the same, their colour rendering may still be tive to how the reference illuminant does. This
Lighting Res. Technol. 2014; 46: 619–636

Downloaded from lrt.sagepub.com at Harvard Libraries on June 22, 2015


626 PR Boyce and KAG Smet

is essentially what the CIE General CRI does implications of the shape and position of the
so the proposed changes seek to address the gamut. This requirement for thought has led
limitations discussed above. The Colour to suggestions that the colour gamut be
Quality Scale (CQS) method22 uses the same reduced to a single number by calculating
reference light sources as the CIE CRI. Where the area enclosed by the gamut as a metric of
it differs is in having 15 test colours, most of colour rendering. Gamut area suffers from
them being saturated, using the CIELAB the defects of any single number index, in that
colour space, using a better transform to light sources with the same gamut area can
allow for colour adaptation, using the latest render colours differently. It is even possible
colour difference formulation and taking the to introduce a reference light source and then
root mean square of the differences between to express the gamut area relative to an ideal
the chromaticity coordinates of each test light source. The gamut area scale suggested
colour under the test and the reference light by Davis and Ohno22 relates the gamut area
sources. The result is a single number scaled created by calculating the positions of the 15
between 0 and 100. The CQS produces test colours used in the CQS on the a*, b*
general CQS indices for fluorescent lamps plane of the CIELAB colour space to the area
that are very similar to those produced by the enclosed by the same test colours illuminated
CIE General CRI but produces rather differ- by the CIE Standard Illuminant D65 repre-
ent values for white, narrow band LED light senting daylight. In this metric, differences
sources. that reduce the saturation of a test colour are
Two other refinements of colour fidelity penalised because it has been found that
metrics have been suggested.23 One, CRI2012, people prefer colours to be more rather than
suggests the use of a set of 17 theoretical less saturated.24,25
spectral reflectance curves evenly spaced Another approach to exploring the colour
across the visible wavelength range as a set properties of light sources involves the use of
of test colours. The other is to adjust the the colour appearance models. Pointer26 used
spectrum of the reference illuminant so that the Hunt27 model to generate 15 different
the discontinuity at a CCT of 5000 K is measures relating to hue, chroma and light-
avoided. These proposals are now being ness. Szabo et al.28 used the CIECAM02
considered in CIE Technical Committee 1–90. model29 to generate metrics relating to the
The second type of colour metric is called a harmony of colour pairs and triads. However,
colour preference metric. For colour prefer- it seems perverse to use a very sophisticated
ence metrics, no reference illuminant is model of colour appearance to generate a
needed. Probably the most advanced of comprehensive picture of how a light source
these metrics is the colour gamut. The will influence surface colours only to throw
colour gamut is obtained by calculating the almost all of that information away by
chromaticity coordinates of a set of test reducing the description to a single number
colours under the light source of interest metric.
and plotting them on a plane in a uniform By now, it should be apparent that quan-
colour space. When the plotted positions are tifying colour is complicated and using simple
joined together, a colour gamut is formed. metrics requires some recognition that infor-
Figure 1 shows the colour gamuts for a mation has been lost. Rea and Freyssinier30
number of light sources derived from the have suggested that at least two metrics are
same eight CIE test colours used to calculate required to give a meaningful picture of the
the CIE General CRI. A colour gamut colour rendering of any light source, one
requires the designer to appreciate the colour fidelity metric and one gamut
Lighting Res. Technol. 2014; 46: 619–636

Downloaded from lrt.sagepub.com at Harvard Libraries on June 22, 2015


LRT symposium summary 627

0.58

0.56

0.54

0.52

v’
0.50

0.48

= Spectrum Locus
0.46 = Planckian Locus
= Metal Halide
= High Pressure Sodium
0.44 = Fluorescent
= Incandescent
= Daylight

0.42

0.14 0.16 0.18 0.20 0.22 0.24 0.26 0.28 0.30 0.32 0.34
u’

Figure 1 The colour gamuts of high-pressure sodium, incandescent, fluorescent and metal halide light sources, and
for the CIE Standard Illuminant D65 simulating daylight, all plotted on the CIE 1976 Uniform Chromaticity Space (UCS)
diagram. The dashed curve is the Planckian locus

area-based metric. This view is supported by the CIE General CRI needs to be modified
the results of Smet et al.31 and Dangol et al.32 or replaced but exactly which of the many
Others have pointed out the possibility of possible metrics should be adopted remains
developing metrics based on colour differ- to be determined. One thing that is appar-
ences,33 colour harmony,28 feelings of con- ent is that more than one metric is needed
trast34 and memory colours,35 i.e. the colours to give a clear picture of the ability of a
of objects where colour carries a meaning light source to render colours, at least one
such as fruit. Yet others have developed for fidelity and one for preference. No
detailed models of colour appearance.36 The doubt the CIE will eventually make a
relative merits of these colour preference choice between the competing metrics, but
metrics are being considered by CIE before that can happen, it will be necessary
Technical Committee 1–91. to determine if there are serious differences
From the above, it should be clear that between them or if the arguments between
the situation with regard to colour metrics is their proponents are more a matter of
an excess of choice. There is no doubt that theology than reality.
Lighting Res. Technol. 2014; 46: 619–636

Downloaded from lrt.sagepub.com at Harvard Libraries on June 22, 2015


628 PR Boyce and KAG Smet

5. Discomfort glare task plane but without changes to the design


process so that designers know where the
The level of discomfort glare produced by a tasks will be and what type of lighting they
lighting installation is usually predicted by require these changes seem unlikely to be
calculating the UGR.9 UGR calculations effective. The fact is the design of lighting for
involve the luminance and size of the glare workplaces is often an exercise in prediction
source, the luminance of the background based on assumptions. Fortunately, providing
against which it is seen and the position of the recommended average illuminance uni-
the glare source relative to the line of sight. formly across a horizontal plane in a room
The higher is the luminance of the glare with high reflectance walls and ceiling will
source and the larger its size, the greater will usually provide adequate if indifferent light-
be the level of discomfort glare. Conversely, ing for many applications, provided care is
the higher the luminance of the background taken to avoid glare and to have a modicum
and the further the glare source is from the of upward light.38 This is probably why the
line of sight, the lower will be the level of horizontal plane has survived as the focus of
discomfort glare. UGR is the result of many routine lighting practice despite many years of
years of study by the CIE and represents the exhortation for its abolition by experienced
state of the art in glare prediction methods. designers seeking to make lighting more
However, there is doubt about how well it interesting.
deals with LED luminaires which contain Now, however, a much more radical
multiple small points of high luminous inten- approach to light distribution is being sug-
sity. At the moment, there are no compre- gested. This starts by undermining the funda-
hensive proposals for another metric to mental purpose claimed as the basis for
replace UGR. This is probably because, illuminance recommendations, to ensure ade-
despite all the research, there remains much quate visibility. Cuttle39 argues that while
uncertainty about the phenomenon.37 task visibility was a reasonable basis for
lighting recommendations when difficult
6. Light distribution tasks were common, over the last 30 years
many visually difficult tasks, e.g. reading a
For many years, the most widely used metric fifth carbon copy, have disappeared and,
in the design of lighting installations has been where they do occur, technology often pro-
the average illuminance on the horizontal vides a better way of either doing the task or
working plane.4,5 The horizontal working making it more visible than simply increasing
plane is usually assumed to be at desk the illuminance. Further, more and more
height and to cover the whole occupied information is being viewed on self-luminous
space. The average illuminance given in devices such as smart phones and computer
most lighting recommendations is backed up screens which higher illuminances make more
by an illuminance uniformity metric in the difficult to see. He concludes that current
form of the ratio of minimum to average lighting recommendations based on providing
illuminance. Together, these two metrics tend enough light to ensure task visibility on a
to produce a uniform distribution of light horizontal working plane cannot be justified.
across a space which means that work can be As a replacement, he suggests the basis of
positioned anywhere in the space. There have lighting recommendation should be changed
been attempts to move attention away from to providing for a criterion he calls ‘Perceived
the horizontal working plane by claiming that adequacy of illumination’, which simply
illuminance recommendations apply to the means that the space does not appear underlit
Lighting Res. Technol. 2014; 46: 619–636

Downloaded from lrt.sagepub.com at Harvard Libraries on June 22, 2015


LRT symposium summary 629

or gloomy. The metric he associates with this expected that recommended mean room sur-
perception is mean room surface exitance, face exitance levels to satisfy the perceived
where exitance is the product of the luminous adequacy of illumination criterion will vary
flux falling on a surface and the reflectance of according to people’s expectations for the
that surface. This means that mean room location, so that the level for a doctor’s
surface exitance is the ambient density of surgery will be higher than that for the
diffusely interreflected luminous flux waiting room.
(excluding direct light from the luminaires) For task/ambient illumination ratio, there
within the volume of an enclosed space, in are questions about what ratio is required to
lm/m2. Adopting mean room surface exitance give different levels of emphasis. There is also
as a basis for lighting recommendations a question about the integration with daylight
would require some basic rethinking of gen- from windows. The mean room surface
eral lighting practice, because light distribu- exitance and task/ambient illumination ratio
tions that illuminate the walls and ceiling then metrics have both been developed for electric
become much more effective than those that lighting only. Daylight from windows inevit-
concentrate their output onto the relatively ably produces a very non-uniform light
low-reflectance horizontal working plane.39 distribution. How this might distort or con-
Cuttle40 has recently gone further by sug- strain the use of these metrics is something
gesting an additional metric called the target/ that needs to be addressed. If all these
ambient illumination ratio that is used to give questions can be answered positively, then it
emphasis to specified objects within the space. should be possible to develop lighting criteria
Together, mean room surface exitance and based on these metrics and, hence, reorient
target/ambient illumination ratio can be used lighting recommendations away from lighting
to form a design procedure for first lighting a task on the horizontal working plane
the space and then any significant objects in towards the lighting of the space and the
it. This procedure is all-encompassing in that objects in it. The result should be better
it allows both art galleries and speculative lighting.
office space to be designed by the same
process, although the former will result in 7. Light spectra
very different lighting than the latter.
Interestingly, uniform illumination of a hori- At first, it may seem odd to include light
zontal working plane can occur, but now it spectra in a review of lighting metrics. After
will be the result of a considered opinion all, the spectral power distribution of a light
rather than unthinking obedience to a sched- source is measured using well-established
ule of illuminance recommendations. procedures and quantified in Système
At the moment, the worth of mean room Internationale (SI) units. The reason for
surface exitance and target/ambient illumin- including it is simply that to express the
ation ratio as metrics for determining desir- effect of a spectral power distribution on
able light distributions are matters of belief human vision, the CIE spectral luminous
rather than proof. What is required is some efficiency function for photopic vision, the
experimental evidence that mean room sur- V() function, is used. The V() function was
face exitance is related to peoples’ perception established in 1924 and, although there are
of the amount of light in a space and that other luminous efficiency functions for pho-
that, in turn, is more closely linked to their topic conditions that are recognised by the
satisfaction with the lighting than illuminance CIE,41,42 none have been widely used. Every
on the horizontal working plane. It is to be luminous flux, luminous intensity,
Lighting Res. Technol. 2014; 46: 619–636

Downloaded from lrt.sagepub.com at Harvard Libraries on June 22, 2015


630 PR Boyce and KAG Smet

illuminance and luminance found in lighting safety for pedestrians, and this is known to
recommendations and manufacturer’s cata- be related to their perception of brightness.
logues is based on the V() function. Yet, the Figure 2 shows the proposed luminous effi-
V() function has its limitations.6 Specifically, ciency functions for brightness perception at
the V() function represents the brightness three different photopic illuminances. By
response of the central 28 of the retina, i.e. the applying one of these to the spectral power
fovea, in photopic conditions. In 1924, this distribution of a light source instead of the
was reasonable because the fovea is physio- V() function, light sources for the lighting of
logically designed for the discrimination of streets where pedestrians are likely to be
detail and the main purpose of lighting in found can be compared on the basis of a
1924 was to make detail easier to see when meaningful perception.
daylight was limited. But since 1924, know- It is important to appreciate that lighting
ledge of how the visual system operates has can often have multiple objectives. For exam-
grown, and it has become apparent that the ple, when driving at night drivers are using
five different photoreceptors in the retina and different parts of their visual system for
the different signal channels to the brain different purposes. For controlling the pos-
combine in different ways to produce differ- ition of the vehicle and for detecting activity
ent spectral sensitivities for different types of at the edge of the road, peripheral vision is
performance and perception. This under- important, but for reading road signs and
standing has been largely ignored by the examining the movements of other vehicles,
world of lighting which is a pity because the foveal vision is relevant. This means the
V() function is based on the response of only designer has to check the effectiveness of a
two photoreceptor types, the medium and light source using a number of different
long wavelength-sensitive photoreceptors, luminous efficiency functions before deciding
and ignores the colour channels of the visual on which one to use. It may even be necessary
system entirely. This explains why there are to use different light sources for different
well-known spectral effects associated with objectives such as one that is efficient for
the detection of stimuli off-axis,43 and with peripheral vision for lighting the road and one
brightness perception in mesopic condi- that is efficient for foveal vision for lighting
tions.44 These spectral effects are such that road signs. Whatever the outcome the concept
to achieve the same level of performance or of benefit metrics is valuable because it places
perception, light sources with different spec- responsibility on the designer to identify what
tra have to produce different light levels. the lighting is supposed to achieve, something
Given that these spectral effects are well that requires much more thought than simply
known, it should be possible to incorporate following a table of illuminances based on the
them into lighting recommendations, but this V() function.
is rarely done. Rea45 has proposed a system Rea46 has now taken this process a step
that will allow these spectral effects to be further. His argument is that in order to
taken into account in the design process. The recognise the complexity of the visual sys-
fundamental idea is to develop a series of tem’s operations, it is necessary to separate
what are called benefit metrics to be used in the concepts of light and lighting. At the
lighting recommendations. These benefit met- moment, light is a fundamental SI quantity,
rics require the designer to first identify the like mass, time and length. However, it is an
objectives of the lighting. For example, when anomaly in the SI system because it is the only
lighting a street at night, an important quantity that depends on the response of
objective is to generate a perception of humans. Other human senses, such as
Lighting Res. Technol. 2014; 46: 619–636

Downloaded from lrt.sagepub.com at Harvard Libraries on June 22, 2015


LRT symposium summary 631

1.0

0.8

Efficiency
0.6
VB3(λ): above 25 lx
0.4

0.2

0
400 450 500 550 600 650 700
Wavelength (nm)

1.0

Efficiency 0.8

0.6
VB2(λ): between 1 and 25 lx
0.4

0.2

0
400 450 500 550 600 650 700
Wavelength (nm)

1.0

0.8
Efficiency

0.6
V’(λ): below 1 lx
0.4

0.2

0
400 450 500 550 600 650 700
Wavelength (nm)

Figure 2 Proposed luminous efficiency functions for brightness over different photopic illuminance ranges.45 The
luminous efficiency functions vary because different photoreceptors are dominant in different ranges

hearing, are not blessed with their own SI starting from scratch that would be the
quantity. Logically, it could be argued that approach adopted, but we are not starting
light should not be included in the SI system. from scratch. The world has become accus-
Only optical radiation measured in watts tomed to having a single definition of light
should be included. Optical radiation could that can be used to compare products any-
be converted into light by applying the where in the world. But, as discussed above,
appropriate luminous efficiency function for the V() function is just one among several
a given application. Maybe, if we were possible luminous efficiency functions
Lighting Res. Technol. 2014; 46: 619–636

Downloaded from lrt.sagepub.com at Harvard Libraries on June 22, 2015


632 PR Boyce and KAG Smet

depending on what aspect of visual system appropriate for different design objectives as
performance or perception is of interest. discussed above.
Therefore, what Rea46 suggests is replacing This is a revolutionary proposal, but it
the V() function with something called the does recognise the complexity of the human
universal luminous efficiency function (U(). visual system. It remains to be seen how far
Figure 3 shows the proposed universal lumi- the lighting profession will be willing to move
nous efficiency function and the luminous in this direction.
efficiency functions of all five photoreceptors
in the human retina. It is clear that the
luminous efficiency functions of all five
photoreceptors lie within the boundary of 8. Implications
the universal lumen. Light is then defined as
the optical radiation weighted by the univer- From the above, it should be clear that there
sal luminous efficiency function. Light is no shortage of proposed new metrics for
defined in this way can be used to compare many different aspects of lighting. Three
products across the world but not to design questions now need to be addressed. What is
lighting. Lighting design requires the use of necessary for these metrics to be adopted?
different luminous efficiency functions as What is the likelihood of any of these metrics

1.0
V(λ)
0.9 U(λ)

0.8

2° L-cones
0.7
2° M-cones
0.6
Efficiency

2° S-cones
0.5 10° S-cones

0.4 Rods

ipRGCs
0.3

0.2

0.1

0
400 450 500 550 600 650 700
Wavelength (nm)

Figure 3 The universal luminous efficiency function U(), the CIE photopic luminous efficiency function V() and the
luminous efficiency functions of the long wavelength-sensitive photoreceptors (L-cones), the medium wavelength-
sensitive photoreceptors (M-cones), two short wavelength-sensitive photoreceptors (S-cones), the rod photoreceptors
and the intrinsically photosensitive retinal ganglions cells (ipRGCs) in the retina.46 There are two luminous efficiency
functions for the short wavelength-sensitive photoreceptors (S-cones) because of the effect of the macula on the
S-cones at 28 eccentricity

Lighting Res. Technol. 2014; 46: 619–636

Downloaded from lrt.sagepub.com at Harvard Libraries on June 22, 2015


LRT symposium summary 633

being adopted? If they are adopted, will the So, what is the likelihood of the metrics
end result be better lighting? discussed above being adopted? The answer
The answer to the first question can take to this question has to be qualified by who is
two forms, technical and emotional. The likely to be doing the adopting. Metrics linked
technical answers express the features that a to legislation or to global trade need to be
metric needs to have to be adopted. There are adopted by recognised and authoritative
six such features. First, the proposed metric bodies. Metrics linked to design processes
has to address a recognised problem, either a can be adopted by individual designers.
lighting problem or a policy issue related to LENI is a metric linked to legislation. It is
lighting. Second, the metric has to be shown already contained in European Standard
to be effective in solving the problem. Third, EN 1519310 and is also included in the draft
the metric has to be easily understood. revision of this standard so widespread adop-
Fourth, the metric should make a significant tion is guaranteed. Some form of replacement
difference to design decisions. Fifth, the for the CIE General CRI also looks likely,
metric has to be simple to implement in although exactly which metric or metrics will
design and in construction. Sixth, there be adopted by the CIE is uncertain. A
should be criteria available to identify appro- replacement for CCT is unlikely because the
priate values of the metric. deviation of white light from the Planckian
The emotional answers reflect the fact that locus below about 4000 K is not yet widely
human inertia makes existing metrics difficult recognised as a problem. As for introducing
to change or replace. Simply meeting all the the universal luminous efficiency function,
technical requirements listed above may not this would require changing the SI, an act that
be enough to get a new metric adopted. What calls for international cooperation and agree-
is required to ensure success are advocates, ment, a process that can take many years if it
allies and perseverance. Advocates are happens at all.
required to raise awareness of the proposed Adopting CBDM as the basis of daylight-
metric and to enthuse people about it. Allies ing design is currently problematic. The
are required because a groundswell of support limitations of daylight factor are widely
for a metric will convince authoritative bodies acknowledged, but CBDM and its associated
to give it some consideration. Perseverance is metrics have to be shown to produce better
necessary because most metrics have a defen- daylighting designs before the complexity it
sive screen of vested interests around them. introduces can be justified. The process of
A light source manufacturer who has evaluating CBDM in practice is just begin-
designed a product range around existing ning. However, following its inclusion as a
colour metrics will not take kindly to a new mandatory requirement in the Priority
metric particularly if it means some existing Schools Building Program, its wider adoption
products will be downgraded. Similarly, light- does seem more likely.
ing consultants who are familiar with daylight As for mean room surface exitance and
factor may not view having to learn a whole target/ambient illumination ratio, these met-
new method of daylight evaluation with rics form a coherent process for determining
much enthusiasm. And bodies that issue light distribution, but it is not yet evident that
authoritative guidance may have difficulty this process leads to more satisfactory lighting
changing the basis of that guidance if it means installations, nor what criterion levels are
admitting that they have been wrong for appropriate for each metric. Research is
many years – or, at least, not as ‘right’ as they underway to investigate the merits of these
might wish. metrics and their use in the design process.
Lighting Res. Technol. 2014; 46: 619–636

Downloaded from lrt.sagepub.com at Harvard Libraries on June 22, 2015


634 PR Boyce and KAG Smet

If this research comes to a positive conclu- the moment, the outcome of applying these
sion, mean room surface exitance and target/ metrics to real world applications is unclear.
ambient illumination ratio should become In fact, a combination of the light distribution
more widely used, particularly if lighting metrics of Cuttle and the light spectra metrics
practitioners recognise them as a route to of Rea probably offer the best prospect for a
enhanced lighting quality for general lighting quantum leap in lighting quality. But to make
practice. that leap, it will first be necessary for authori-
The concept of benefit metrics based on the tative bodies to admit the limitations of
spectral sensitivities of different combinations current metrics and the standards that use
of photoreceptors and visual channels is them. This is because standards have a
certainly attractive but requires that much stranglehold on general lighting practice
more attention be given to identifying the through the threat of litigation against those
objectives of the lighting rather than its who fail to conform. This threat only loses its
photometric characteristics. Benefit metrics power if the limitations of the standards are
could be adopted at different times for widely recognised and the benefits of the
different applications. Outdoor lighting alternative metrics are promoted. This will
looks likely to be the first application for only happen if professional organisations
which the benefit metrics approach might be show some leadership by developing lighting
adopted. In general, the neuroscience behind recommendations based on the new metrics
benefit metrics is well established. What is and influential individuals and companies
missing is experience in trying to apply them advocate their adoption, by word and deed.
in the real world. Until this experience is In many ways, these metrics represent an
gained, it is unlikely that professional organ- opportunity for individuals, companies and
isations will be willing to abandon their organisations to gain professional credibility
reliance on recommended illuminances. through demonstrating an expertise founded
Finally, it is worth considering if adopting on a deeper understanding of their subject
any of these metrics will lead to better lighting. than is possible from simply following a table
LENI will only deliver better lighting in the of recommended illuminances, an opportunity
sense of encouraging more energy efficient that should not be missed.
lighting. Changes in the colour metrics are
important for light source manufacturers but Funding
will probably have little effect on the percep-
tion of lighting by the general public. The This paper received no specific grant from
public can certainly tell the difference between any funding agency in the public, commercial
the extremes of colour rendering but fine or not-for-profit sectors.
differences elude them. CBDM is another
matter. By exposing window or skylight
designs that can cause visual or thermal Acknowledgements
discomfort and providing better estimates of
energy consumption, CBDM should deliver This paper is based on the presentations given
better lighting. Having said that, it is the mean at the Better Metrics for Better Lighting
room surface exitance and target/ambient symposium held at University College
illumination ratio process for light distribu- London on the 15 July 2014. It is a pleasure
tion that has the most potential to deliver to acknowledge the contributions of Lou
better lighting. The same may be true of the Bedocs, Kit Cuttle, John Mardaljevic, Peter
light spectra metrics proposed by Rea but at Raynham, Mark Rea and Kevin Smet to the
Lighting Res. Technol. 2014; 46: 619–636

Downloaded from lrt.sagepub.com at Harvard Libraries on June 22, 2015


LRT symposium summary 635

symposium. All have contributed to the 15 Purves D, Beau-Lotto R. Why We See What
thoughts that lie behind this paper. We Do. Sunderland, MA: Sinauer Associates,
2003.
16 American National Standards Institute. ANSI
Standard for Electric Lamps: Specifications for
References the Chromaticity of Solid State Lighting
Products. ANSI C78.377. Rosslyn VA:
1 Fire Precautions (Workplace) (Amendment) National Electrical Manufacturers
Regulations, Norwich, UK: The Stationery Association, 2008.
Office, 1999. 17 Rea MS, Freyssinier JP. White lighting for
2 Health and Safety (Safety Signs and Signals) residential applications. Lighting Research and
Regulations, Norwich, UK: The Stationery Technology 2013; 45: 331–344.
Office, 1996. 18 Guo X, Houser KW. A review of colour
3 Building Regulations Part L (Conservation of rendering indices and their applications to
Fuel and Power), Planning Portal: UK commercial light sources. Lighting Research
Department for Communities and Local and Technology 2004; 36: 183–199.
Government, 2010. 19 Commission Internationale de l’Eclairage.
4 Society of Light and Lighting. The SLL Code A Review of Chromatic Adaptation Models.
for Lighting. London: CIBSE, 2012. CIE Publication 160:2004. Vienna: CIE, 2004.
5 The Illuminating Engineering Society of 20 Smet KAG, Schanda J, Whitehead L, Luo RM.
North America. The Lighting Handbook. CRI2012: A proposal for updating the CIE
10th Edition. New York: IESNA, 2011. Colour Rendering Index. Lighting Research and
6 Boyce PR. Human Factors in Lighting. Technology 2013; 45: 689–709.
3rd Edition. Boca Raton, FL: CRC Press, 2014. 21 Commission Internationale de l’Eclairage.
7 Commission Internationale de l’Eclairage. Colour Rendering of White LED Light
International Lighting Vocabulary. CIE Sources. CIE Publication 177:2007. Vienna:
Standard S 017/E 2011. Vienna: CIE, 2011. CIE, 2007.
8 Commission Internationale de l’Eclairage. 22 Davis W, Ohno Y. Colour quality scale.
Method of Measuring and Specifying Colour Optical Engineering 2010; 49: 033602.
Rendering Properties of Light Sources. CIE 23 Smet KAG, Schanda J, Whitehead L, Luo MR.
Publication 13.3. Vienna: CIE, 1995. CIE2012: A proposal for updating the CIE
9 Commission Internationale de l’Eclairage. CIE Colour Rendering Index. Lighting Research and
Collection on Glare. CIE Publication 146-2002. Technology 2013; 45: 689–709.
Vienna: CIE, 2002. 24 Judd DB. A flattery index for artificial illumin-
10 British Standards Institution. BS EN 15193: ants. Illuminating Engineering 1967; 62: 593–598.
2007 Energy Performance of Buildings. Energy 25 Thornton WA. Color-discrimination index.
Requirements for Lighting. London: BSI, 2007. Journal of the Optical Society of America 1972;
11 Mardaljevic J, Heschong L, Lee E. Daylight 62: 191–194.
metrics and energy savings. Lighting Research 26 Pointer MR. Measuring colour rendering –
and Technology 2009; 41: 261–283. A new approach. Lighting Research and
12 Reinhart CF, LoVerso VRM. A rules of thumb- Technology 1986; 18: 175–184.
based design sequence for diffuse daylight. 27 Hunt RWG. A model of color vision for
Lighting Research and Technology 2010; 42: 7–31. predicting color appearance. Color Research
13 Wienold J, Christoffersen J. Evaluation meth- and Application 1982; 7: 95–112.
ods and development of a new glare prediction 28 Szabo F, Bodrogi P, Schanda J. A colour
model for daylight environments with the use harmony rendering index based on predictions
of CCD cameras. Energy and Buildings 2006; of colour harmony impression. Lighting
38: 742–757. Research and Technology 2009; 41: 165–182.
14 Reinhart C. Daylighting Handbook 1. 29 Commission Internationale de l’Eclairage.
Cambridge, MA: Christoph Reinhart, 2014. A Colour Appearance Model for Colour

Lighting Res. Technol. 2014; 46: 619–636

Downloaded from lrt.sagepub.com at Harvard Libraries on June 22, 2015


636 PR Boyce and KAG Smet

Management Systems: CIECAM02. CIE 38 Jay PA. Inter-relationship of the design criteria
Publication 159:2004. Vienna: CIE, 2004. for lighting installations. Transactions of the
30 Rea MS, Freyssinier JP. Color rendering: Illuminating Engineering Society (London)
Beyond pride and prejudice. Color Research 1968; 32: 47–71.
and Application 2010; 35: 401–409. 39 Cuttle C. Towards the third stage of the
31 Smet KAG, Ryckaert WR, Pointer MR, lighting profession. Lighting Research and
Deconinck G, Hanselaer P. Memory colours Technology 2010; 42: 73–90.
and colour quality evaluation of conventional 40 Cuttle C. A new direction for general lighting
and solid-state lamps. Optics Express 2010; 18: practice. Lighting Research and Technology
26229–26244. 2013; 45: 22–39.
32 Dangol R, Islam M, Hyvarinen M, Bhusal P, 41 Commission Internationale de l’Eclairage. CIE
Puolakka M, Halonen L. Subjective prefer- 1988 28 Spectral Luminous Efficiency Function
ences and colour quality metrics of LED light for Photopic Vision. CIE Publication 86,
sources. Lighting Research and Technology Vienna: CIE, 1990.
2013; 45: 666–688. 42 Commission Internationale de l’Eclairage.
33 Sandor N, Scanda J. Visual colour rendering CIE 10 Degree Photopic Photometric
based on colour difference evaluations. Observer. CIE Publication 165:2005. Vienna:
Lighting Research and Technology 2006; 38: CIE, 2005.
225–239. 43 Akashi Y, Rea MS, Bullough JD. Driver
34 Hashimoto K, Yano T, Shimizu M, Nayatani decision making in response to peripheral
Y. New method for specifying color-rendering moving target under mesopic light levels.
properties of light sources based on feeling of Lighting Research and Technology 2007; 39:
contrast. Color Research and Application 2007; 53–67.
32: 361–371. 44 Rea MS, Bullough JD, Akashi Y. Several
35 Smet KAG, Ryckaert WR, Pointer MR, views of metal halide and high pressure sodium
Deconinck G, Hanselaer P. Correlation lighting for outdoor applications. Lighting
between color quality metric predictions and Research and Technology 2009; 41: 297–320.
visual appreciation of light sources. Optics 45 Rea MS. Value Metrics for Better Lighting.
Express 2011; 19: 8151–8166. Bellingham WA: SPIE Press, 2013.
36 Fairchild MD. Colour Appearance Models. 46 Rea MS. The lumen seen in a new light:
Chichester, UK: John Wiley and Sons, 2005. Making distinctions between light, lighting and
37 Clear RD. Discomfort glare: What do we neuroscience. Lighting Research and
really know?. Lighting Research and Technology. First published 31 March 2014.
Technology 2013; 45: 141–158. DOI 10.1177/1477153514527599.

Lighting Res. Technol. 2014; 46: 619–636

Downloaded from lrt.sagepub.com at Harvard Libraries on June 22, 2015

You might also like