Less is less, less is more
Will LEDs, a technology that creates more light per unit of electricity than any commercial technology before it, increase demand for electricity? Will an efficiency technology increase overall consumption? This is the odd question posed not so much by participants in the LED marketplace but commentators around it, who refer to a concept known as the “rebound effect” of efficiency. In brief, if lighting becomes less expensive to use, the world will use more light - so much more, potentially, that the net consumption of energy by lighting will actually increase.
The New York Times published a somewhat bizarre op-ed to this effect last October, and GreenTech Media published a response, authored by investor Rob Day. As an organization that implements then owns and operates LEDs on behalf of commercial clients, we thought we’d weigh in with our experiences of the extent to which LEDs drive rebound.
The rebound that matters is in measured power consumption. Let’s review what would be required to increase wattage drawn when changing an old lamp to an LED.
The typical incandescent light bulb emits 15 lumens per watt, where a modern LED bulb emits between 80 and 120 lumens per watt. So for incandescent applications, a consumer would need to increase brightness (number of lumens) by 600-1,000% in order to increase lighting system wattage. Try to imagine your office or classroom being 6-10x brighter. Sunglasses all around.
In fluorescent and industrial (high intensity discharge) lighting, the efficiency difference is not as great, but LEDs still require considerably less wattage (the wattage saved per usable lumen is 50-75% in fluorescent and 40-60% in industrial/HID). A facility would need to double the light output to generate net positive "rebound" where those technologies are in place. Again, we’ve never come across a hotel, apartment complex, or warehouse that wanted to anywhere near double their brightness.
What we see - retrofit projects
Instead, what we have seen - in hotel, retail, office, and educational environments - indicates that users do not increase light output during the LED implementation process, let alone increase consumption nearly enough to offset energy saved. In fact, as measured by lumen output, brightness almost always decreases. We have previously theorized that this is related to lumen quality.
In certain cases, especially retail, we have seen “re-lamping”, where a business will take the opportunity to brighten the space, and in those instances the 6% number that Rob Day references seems about right. Retailers generally have incandescent/halogen lighting, which due to its extreme inefficiency creates a powerful disincentive to brighten. The resulting lighting deficiency helps explain why retail as a group seems most likely to re-lamp. The increase is 6%, though, not 600% - meaning the wattage drop over the original lighting system is still about 75%.
What we see - new construction
One of the keys to the Times op-ed is the idea that for new installations, where no lighting existed before, builders and tenants will see cheaper LED lighting as a way to drastically increase consumption. Again, though, when we’ve seen users design commercial environments from greenfield, we are not seeing an increase. Offices, actually, now that computer ubiquity means every desk emits light, specify less light from overhead than designs did 20 years ago. Where lighting standards recommend illuminance of about 100 footcandles, today’s office employees tend to prefer between 40 and 60.
We see no evidence of rebound in commercial interiors, even when LEDs are introduced during design phase.
Rebound in emerging economies
The argument in the Times op-ed that may have some energy-efficiency credibility is that the world’s emerging economies will view LEDs as an opportunity to greatly brighten their countries in an affordable way. Those countries, which per the op-ed’s image are the ones that currently show up as dark from outer space, would see such a massive increase in light output that LEDs can be reasonably accused of increasing power demand.
This seems true, or at least possibly true. The rural areas in the developing world, for example, are not well-lit electrically. It is generally cited that about 1 billion people do not have electric lighting, and many more would use more light if they could. For those populations, LEDs will accelerate the availability of lighting and the technology can be accused of a net increase in power consumption.
This line of thinking first omits the qualitative value of artificial light, and second borders on prejudicial against those not living in the already-well-lit confines of the rich world. To conclude that in “China, India and Nigeria...the sheer volume of demand will likely trump efficiency gains” is incorrect primarily because in those regions people are not looking for "efficiency gains." They would like lighting, and have every right to it that we do in the already-bright world. Where there is no lighting, introducing LEDs will increase consumption - but as mentioned above, LEDs are still a major improvement over introducing fluorescent or incandescent light.
Lighting is not about energy
LEDs are an efficient innovation, and have brought lighting into the cleantech and resource efficiency realms. Frankly, that is why Wavelength Lighting exists. An effort to address climate change is the driving force behind what we do. It is important to us that each of our lighting installations lowers carbon footprints.
However, lighting both precedes and supersedes efficiency. Lighting - unlike solar power, wind power, or an efficient HVAC unit - is an end in itself. You cannot have a discussion about LEDs and energy demand rebound without acknowledging that illumination, the primary goal of lighting, is not by nature quantitative.
Even quantitatively, though, our experience in implementing LED lighting in commercial facilities suggests that net energy rebound is nearly impossible. LEDs are having a major positive effect on energy reduction, and as the efficiency of LED technology continues to improve, as with 100% likelihood it will, this impact only increases.