In choosing which light source to use for a particular installation, at least ten criteria should be considered. These are as follows:

1. Amount of Control Desired

Light distribution from a small lamp can be controlled more easily than can light emitted from a large lamp. This is simply because it is easier to shape a reflector around a small lamp than a large one. Some lamps are as small as 10mm in length, others as large as 1m long. The pattern of light from the former can be controlled very precisely whereas the best we can hope to do with the latter is to throw light in one general direction or another. We idealise a small light source with the concept of a 'point source'. The smaller the light-emitting element of a lamp, the more closely it resembles the ideal of a mathematical point source.

2. Amount of Diffusion Desired

It is sometimes desirable to have diffuse light rather than highly directional light since the latter may cause harsh shadows. An area source or a linear source (such as a fluorescent lamp) generates more diffuse light and softer shadows than a point source.

3. Consistency and Reliability of Supply Voltage

High pressure gaseous discharge lamps are more sensitive to voltage variation than low pressure lamps. If the arc is extinguished due to a dip in voltage, the high pressure lamp may require up to 15 minutes to return to nearly full light output.

4. Ambient Temperature and Humidity

Some lamps, notably fluorescents, are very sensitive to temperature and humidity. These lamps are difficult to start when the ambient temperature is low, and once started may not produce full light output.

5. Air Conditioning Load

All artificial lighting adds an additional heat load to a building. Lamps with a higher efficacy will put less heat into a space for a given amount of light output. The most thermally efficient form of lighting turns out to be diffuse daylight, followed by direct sunlight, low pressure then high pressure gaseous discharge, with the worst of all being incandescent lamps.

6. Colour Rendition

One of the key areas of concern for the architect, interior designer and lighting engineer is the lighted appearance of the space. Most sources are available in a variety of spectral power distributions, yielding a variety of colour rendering indices (CRI) and colour temperatures. The lighting designer must be thoroughly familiar with these concepts in order to properly achieve the desired effect with the chosen lamps.

7. Cost

Some lamps are quite cheap to purchase initially. However, these tend to have low efficacy and relatively short lives. One must consider not only the initial cost, but the cost to operate the system over its entire life, including energy costs and the requirement to pay a worker to change the burned out lamps frequently.

8. Efficacy, Life, Lumen Depreciation

All of these have an effect on life cycle cost. If a lamp's lumen output declines rapidly during its life, the prudent designer initially provides more lumens than is required so that as the lamp declines with age, a sufficient amount of light is still available. In other words, if seven luminaires are required to provide the right amount of light for a space initially, we might put eight luminaires in our design so that the space is over lit at first, but the installation will still provide enough light later on as the lamps provide less than their rated lumen output due to ageing. Comparing the rate of lumen depreciation from one type of lamp to the next thus becomes an important part of the cost analysis.

9. Control

Some lamps are more easily dimmed than others. One must consider if simple on/off control is acceptable, if inexpensive dimming is desired, or if it is reasonable to incur the larger expenditures to get higher quality dimming.

10. Energy-Saving Devices

As with dimming, lamps vary in their adaptability to certain energy-saving devices and strategies. The lighting designer must be aware of which sources are suitable for use with which energy-saving methods.

Selecting a Lamp

Lamp Life

The life rating of a light source is given in burning hours and is based on measurement of a large representative group of lamps under laboratory conditions. The rating is based on lamp industry standards which specify the burning cycle per start, a 50% mortality, and approximately a minimum of 70% rated lumens. Thus if a manufacturer rates the life of a certain lamp as 1000 hours, this means that after burning for 1000 hours we can reasonably expect half of a large group of lamps to be operative and half to have expired. Rated life is only a median value and says nothing about an individual lamp. Any individual lamp may last much longer or may burn out much earlier than the rated life.

The life of a lamp is usually affected by the number of starts. For this reason the life rating only applies to a specific number of hours per start. If lamps are tested at 11 hours per start (i.e. an 11 hour burning cycle), then an installation in which the lamps are operated for only three hours at a time may experience significantly shorter lamp life.

Operating Hours per Year

The lamp life alone has little practical meaning for the lighting designer. The important consideration is how long the lamp will be in service for the particular application, usually given in annual operation hours. From this we can determine how often we anticipate lamp replacement will be required. If a lamp has a rated life of 10,000 hours and is operated 11 hours a day, 6 days a week, then on average the re-lamping cycle will be:

Time between re-lamping = 10,000 / 3432 = 2.9 yrs.

Colour Temperature and Correlated Colour Temperature

Since incandescent lamps have radiating characteristic "similar" to the blackbody radiator, they can be assigned a 'colour temperature'. However, gaseous discharge lamps do not have radiating characteristic that match the blackbody radiator. Therefore, gaseous discharge lamps may be assigned a 'correlated colour temperature'.