Steve Fotios on ways to reduce lighting load

Reducing the energy consumed by interior lighting improves the sustainability of buildings. The best way to do this is to make good use of daylight and ensure artificial lighting is not used when daylight is adequate. Achieving this is a function of lighting controls and occupant motivation. This article discusses an approach to reducing the energy consumed by artificial lighting.

Interior lighting in buildings such as offices and education facilities is often designed to meet a target mean horizontal illuminance, this being 500 lux in many spaces. The energy consumed by lighting could be reduced by adopting a lower average illuminance, but this may lead to complaints of gloominess from occupants, and adversely affect motivation and productivity. There may also be more direct effects on visual performance.

One way to combat gloominess resulting from reduced illuminance is to consider effects of lamp spectral power distribution (SPD). Lamps are available in a wide range of SPD and this can be seen in the variation of colour rendering and colour appearance characteristics.

Akashi & Boyce [2006] surveyed the occupants of an office building in which they were able to modify the lighting. The existing lighting used three fluorescent tubes in each luminaire, these lamps being of correlated colour temperature (CCT) 3500K. In one office, the illuminance was reduced by one third from about 550 lux by the removal of one lamp from every luminaire, and in a second office the reduction in illuminance was accompanied by a change in lamp type, increasing the CCT to 6500K. In the office where only the illuminance was reduced, there was a significant increase in judgements that the office appeared gloomy; in the office where the reduction in illuminance was accompanied by the increase in CCT, there was a significant reduction in judgements that the lighting was too dim.

The Akashi & Boyce study suggests a strategy for reducing energy consumption while maintaining the same visual conditions. Can this study be relied upon? Yes, for two reasons. First, two methods of data gathering were used, a category rating task and a questionnaire seeking yes/no responses. Second, there were control conditions – a room in which only lamp type was changed and a room in which no changes were made.

There is, however, need to validate the results by replication to ensure they are not just due to chance. Any experimental procedure contains bias which may exaggerate or understate the subjective response and so data from a range of procedures and contexts are beneficial. Most studies tend to use a unique set of variables and procedures and thus set limits for the data; for example, the Akashi & Boyce study used only two lamps of different CCT and it may be found that it is a different attribute of lamp SPD that guides the brightness effect.

There have been many studies of lamp SPD and perception of the environment [Fotios, 2001]. Well known work carried out in the UK includes the visual clarity studies of Bellchambers [Aston & Bellchambers, 1969; Bellchambers & Godby, 1972] and Boyce [Boyce, 1977]. A first step towards considering a study to be reliable is that the results are published in a peer-reviewed journal – a key source for lighting studies is Lighting Research & Technology.

I am carrying out a meta-analysis of these studies with Kevin Houser of Pennsylvania State University in which we are closely inspecting experimental methodologies. One method is side-by-side matching, where two stimuli, such as illuminated rooms, are lit simultaneously by separate lamps and test participants adjust the illuminance in one room until they appear as-near-as-possible equally bright. We have shown that three steps of counterbalancing are needed to avoid biasing the results [Fotios, Houser & Cheal, 2008]; results of studies which did not do so, such as those of Bellchambers, give an unreliable estimate of lamp SPD effects on brightness and must therefore be omitted from the analysis.

In the brightness discrimination method the task is to say which of two environments is brighter, in a room lit sequentially by two different types of lamp. This can be an easier task for test participants but needs more care from the experimenter when choosing which levels of SPD and illuminance to compare.

An alternative experimental method is category rating. Typically only one light source is used at a time, and participants rate the brightness (and other attributes) of the scene using a scale of 1=dim to 8=bright. Again, precautions are needed if the work is to provide reliable evidence of lamp SPD effect [Fotios & Houser, 2009]. While side-by-side matching tasks tend to exaggerate the differences of brightness between lamps, category rating tends to understate them; the appropriate interpolation lies within this range and depends partly on the end application of the research and the physiology of human vision.

A current focus of research at the University of Sheffield is how experimental conditions affect the outcome. Laboratory studies often ask test participants to judge lighting from one or more different types of lamp; what is the difference between judgements made of two lamps seen simultaneously, two lamps seen in rapid succession or of judgements given to the lamps when seen separately? What is the difference between evaluations of a uniform field (eg a grey wall) and a furnished space? Do tests carried out using small visual fields adequately represent the visual response in full-field lighting?

An understanding of these effects and their magnitude is needed in order to be able to properly analyse and categorise the findings from previous work. There is a strong interest in this work in the US. The Illuminating Engineering Society of North America (IESNA) has established the Effects of Lamp Spectral Distribution committee to report on lamp SPD effects on brightness and visual effort. I am chairman of the brightness sub-committee. This committee has two objectives. The first is to review the body of evidence to identify reliable and appropriate evidence of lamp SPD effects on brightness. The second is to identify a tool for predicting the effect, which may be a model of human vision or a colour characteristic of the lighting. Research is under way at the University of Sheffield and Pennsylvania State University.

While the meta-analysis is yet to be completed, some previous studies do appear to provide reliable and appropriate data [Fotios & Houser, 2008]. These tend to support the Akashi & Boyce field study in that lamp SPD does have a significant effect on judgements of brightness. Imagine that you had chosen to light a room to 500 lux with lamp A. The data suggest it could be lit using lamp B to a lower illuminance while maintaining the same perceived visual environment. What is missing is a tool for predicting the relationship between lamp characteristics and the permissible illuminance reduction. Whether this saves energy depends on the efficacies of the two types of lamp; the research should also be able to provide targets for manufacturers as to tuning the spectral power distribution of lamps or LED arrays to match human vision.

A similar situation exists with lighting on residential roads, where one target of lighting is to improve visual perceptions of brightness and safety. There is evidence from trial installations and laboratory studies that lamp SPD affects perception of the visual environment at night and that consideration of lamp type enables design illuminance to be reduced, which can reduce energy consumption [ILE, 2008]. British Standard BS5489-1:2003 accounts for the effect of lamp SPD by permitting the average illuminance to be reduced by one lighting class when using lamps of colour rendering index equal to, or greater than, 60. We know that colour rendering index is not an infallible tool for comparing the colour rendering properties of lamps [Guo & Houser, 2004], let alone for comparing the effect on brightness or perceived safety in night-time roads, but it does give a simple method of discriminating between those lamps currently in common usage.

I have received funding from the Engineering and Physical Sciences Research Council to identify a better method of prescription. Three simplified examples [ILE, 2008] demonstrated that consideration of lamp SPD can reduce the energy consumed by street lighting by 34%, but also that this saving is highly context related and should not be assumed without carrying out a cost-effectiveness analysis.