Resources | LpR Article | Environment | Oct 31, 2017

A World of LED Lights - The Cost of Waiting

Today, LEDs dominate the lighting domain. They are efficient and manufacturers understand how to implement them for different lighting applications. While energy efficiency is rather well understood, the impact of the adoption speed is rarely discussed is. Benoit Bataillou, guest author from Pi Lighting, will have a look at the environmental cost of LED production versus the significant energy savings that a full transition to LEDs would bring. In a fictive scenario, he assumes that a transition takes place in a heartbeat today and compares that with the usually supposed scenario of a full transition to LED lights in 2025.

This article will be different from a typical technical article, as it will not go deep into technical details, but rather, give an areal view of a popular subject. The article is based on a hypothetical scenario: What if we changed all lighting to LEDs with a snap of the fingers?

It’s common knowledge that LEDs help to save energy and there has also been talk about the impact on manufacturing. Circular Economy questions are getting more and more urgent, and debates are ongoing about the industry footprint, depending on who or what entity makes the study.

But the question is: What would it take to replace all the light fixtures in the world with today’s LEDs? How would we produce all those LEDs and what would the result be? And if there is something that can be done within the lighting industry - does it matter?

After a first short analysis in an article last year [1], this article goes a little further on this subject using a similar approach. By gathering information and some hypotheses from different available sources a first draft of an answer can be made. The goal is to grasp the idea of the opportunity ahead. In many cases the Fermi estimation and the “Spherical Cow” [2] approximation are used to get an order of magnitude [3].

About the Goal

The goal is to estimate the impact of an arbitrary scenario: If we switched the entire world to today’s efficient LED lighting, would that make a difference? Do we have an idea?

What is the challenge ahead? What does it take to produce an LED fixture? What should we do with everything that is replaced? Do it and finally see the impact by comparing it with the regular increase of the LED population that is currently going on.

Note that there is a large quantity of public data available for the US, so the “model world” here is a world in which all volumes, consumptions and costs are the same ratio of the GDP contribution of the USA (25.89 % source [4]). The world is roughly four times larger than the USA data. In all cases the data was double-checked with EU and China data (where available).

World Statistics

This sections shows the estimate of the total number of lights that need to be replaced and also tries to give some interesting numbers.

Careful estimations provided, lighting has only a share of 14% of the electricity consumption in homes, 22% in commercial environments, and 7% in manufacturing facilities. Given the relative total consumption we end up with 14% average of energy consumed for lighting in 2016.

Different sources [5&6] show that 22,000 TWh of energy were consumed for electricity in the world in 2016. With 14%, lighting has a total electrical consumption of 3080 TWh for 2016, which is roughly the electricity production of the entire European Union [7].

Calculating the energy in Joule for 2016 from 3080 TWh using EQ 1 results in 11 Exajoules (1M TJ). Transformed into watts using EQ2, for 2016, the average power dedicated to lighting is 300 GW.

Screen Shot 2017-10-31 at 11.18.55.pngFigure 2: Breakdown of world installed based by technology (2016 data, forecasted to 2017 using source [8])

Screen Shot 2017-10-31 at 11.18.59.png

About today's efficiency of our lights
As of today [8], the total installed light sources of all sectors together can be broken down by lighting technology.

Based on the measured efficiency data, one can calculate the weighted average efficacy of the mix of all these technologies in lm/W. Gathering data from Caliper [9] for LED fixtures, and various manufacturer datasheets, we end up with the following weighted average: 38 lm/W as our world average, mostly raised by LEDs, HID and fluorescent technologies, and taken down by incandescent and halogens.

The same report gives a rough estimate of 30B fixtures. By multiplying the number of lamps with the efficiency using the calculation of EQ3 gives the amazing number of 11 Terra-lumens.

These calculations show that the average fixture consumes 10 W and produces 380 lm. It seems low for the power, but the order of magnitude is huge and the light output seems correct, given the large majority of halogen/ CFL/incandescent lights.

LED Fixture Production

Given our world statistics of the installed fixture base of 30B fixtures, the question here is: if we were to replace them all, what would be the impact?

NEMA in 2013 published a report in which the carbon footprint of several fixtures was analyzed [10]. An interesting finding of this study was that the carbon impact comes mainly from four components: Ballast (driver), Heatsink (Al), potting materials in the lamp base, and LEDs themselves.

The semiconductor and electronics industries use a large amount of energy and water, and as the NEMA report shows, the overall carbon footprint and energy cost of manufacturing an LED product is approximately twice the conventional counterpart.

Derived from 30B fixtures into the relative energy cost, and now having to spend quite a bit of energy to produce the 86% missing “non-LED” fixtures.

How much energy does it take to produce an LED fixture? - Unfortunately no exact data is available. Therefore a theoretical calculation based on some assumptions and published data from laptop computers was made, but fixture producers are welcomed to enlighten us with more accurate numbers.

An LED fixture in general does not have a battery or a screen, etc., and the electronics is simpler, but there is more and more PCB material involved for sensors, intelligence, communication and new features.

In the research paper “Economic-balance hybrid LCA extended with uncertainty analysis: case study of a laptop computer” [11], the authors found “Results show that manufacturing the computer requires 3010-4340 MJ of primary energy.” While LED fixtures have in general no screen, no battery, quite simpler electronics, the assumption was that it takes 1/10 of this energy consumption to produce a luminaire. In addition, the lower value from the cited research for energy consumption was used for the calculation, resulting in 300 MJ of energy.

To put things in relation, this 300 MJ equals 83 KWh or the equivalent of $10 of electricity in the US. To replacing all the fixtures, 172 GW of energy are necessary; that’s a lot for one year!

Circular Economy - Fixture Disposal

To see the whole picture, it is important to also have a look at the old fixtures. In the chosen scenario, the entire world is illuminated with LEDs. So apart from LED fixtures with the ability to replace the LED, everything must be replaced. In LED Lighting, replacing part of a fixture is an important debate, but very few solutions exist. This simplifies the calculation because the lifetime of any installed structure can be ignored.

Keeping the same “spherical cow” approach, there are currently about 30B “light points” in the world.

Data on large scale recycling varies quite a bit when reading through the literature. Our hypothesis is simply to add a “recycling factor”, a percentage of our new installed base that comes from recycling older materials. A study suggests that currently, between 20% and 40% of e-waste is recycled [12].

Even when assuming that proper recycling takes place with a recycling factor of 40%, this leaves a significant amount of waste. While the question of the waste raised many debates between the authors of this article, since the comparison concerns the scenario “change everything now” or “just keep going as it is”, in both cases, the waste will be the same, except there will new ways to deal with the waste that will be produced in the future. Then, maybe the second scenario might change.

Energy Savings

The latest generation of fixtures in 2017 easily exceeds 100 lm/W for the complete range of applications; and efficiency is still increasing. It is worth it to mention that this is a very careful estimate. Figure 3 is a comparison of the two scenarios “change everything now,” and the scenario “just keep going as it is” with the assumption of changing +10% a year, leading to a complete change in 2025 for the latter.


As can be seen in Figure 3, the initial production cost makes a “change everything now” worse in energy balance in the first year. However, this changes very quickly.

The most striking value is the annual difference. With “change everything now” the first year with the effort to replace all translates in a negative energy benefit. But it’s more than covered the following year when 200 GW of energy can be saved. From then on, the “just keep going as it is” scenario catches up. Anyway, the savings increase dramatically over the years.

Screen Shot 2017-10-31 at 11.20.25.pngFigure 3: Hypothetical “change everything now” scenario and “just keep going as it is” comparisons. Note that numbers are large, but the blue curve plateau is at approximately 13 GW while the lifetime is ignored

How does the benefit of saving 200 GW compare?
200 GW equals forty times the energy production of the largest coal plant in the world [13] and roughly equals the amount from 200 nuclear reactors. This should give a small impression of the usefulness of such a step. The earlier one implements an energy efficient source, even given the important production costs, it leads to a benefit.

The cost of waiting
The “cost of waiting” is even more important, since it’s the cumulated variation between “change everything now” and “just keep going as it is”. Reality will certainly lie somewhere between these two scenarios. However, this extreme theoretical case would lead to a cumulative saving of close to a Terawatt which is about a fourth of the total power consumption - not just electrical power consumption - of the US, including cars and coal. Waiting is just pure waste of energy on a very large scale.

The way to getting closer to the impossible
The largest contributor to improvements is the cost of production. Taking the very rough hypothesis above, there is close to 60% of an LED fixture energy saving which is spent in producing it. New processes will have a direct impact on that. Recycling value is obvious, recycling old parts lead to direct change in the return, as it’s equivalent to fewer fixtures to “pay” in energy.

About the hypothesis
DOE states in an interesting report published last year [14]:
“Decreasing lighting energy consumption by 75% in 2035 represents an even greater opportunity when the cumulative savings are considered. From 2015 to 2035, a total cumulative energy savings of 62 quads is possible if the DOE SSL Program Goal for LED efficacy and connected lighting are achieved - equivalent to nearly $630 billion in avoided energy costs.“

So it seems that the discussed hypothesis was actually quite conservative.

Thoughts about the LED lifetime
Lifetime calculation for the given time between 2017 and 2025 with 12h/24 running time results in 35,000 hours operation time. There are already fixtures with longer lifetimes of 50K hours and more on all the examples available, therefore it’s long enough not to worry about it. Anyway, in the “change everything now” scenario (blue curve), the calculations take a 1% “accidents” every year into account. Adding these lifetime considerations to the full picture makes the results even clearer.


Would aggressively replacing everything to LEDs positively contribute to “the big picture”? - Yes, but what else? If emerging countries still are building up their power grid, they probably should be the first to heavily invest in LEDs. More generally spoken, if the industry provides a solution, would it not be best to apply it as soon as possible? Very likely also yes.

I would like to thank Dr. Richard from CNRS for his help on this exercise, and V. Laganier from Light Zoom Lumiere for bringing the idea in the first place, last year.

[8] Energy Savings Forecast of Solid-State Lighting in General Illumination
Applications: energysavingsforecast16_0.pdf
[10] of Carbon Footprint of Electrical Products-June 2013.pdf

(c) Luger Research e.U. - 2017