Decentralized Renewables Won’t Fuel Modern Cities

Why We Can’t Ignore Fundamentals of Power Density


There are fundamental physical limits to how much energy we can extract from renewable resources for a given area of land, particularly for an increasingly urban human society. Yet, Western environmentalists calling for distributed energy as a viable solution routinely ignore extremely high population density. For example, of the world's 200 largest urban areas, 17 are in India. Covering any of the cities entirely in solar panels will generate less than half of their energy needs. If the citizens of these cities are to attain a high quality of life they will require large centralized energy generation. This is not a matter of ideological preference, but of engineering reality.

August 19, 2013 | Robert Wilson,

The 21st century will almost certainly witness a transition to an overwhelmingly urban human population, and – hopefully – a low-carbon energy system. The former scenario, however, will have a significant impact on the latter because a fundamentally urban species cannot be powered locally.

The continued, and essentially unabated, accumulation of carbon dioxide in the atmosphere may at times render considerations of the requirements of a decarbonized energy system appear somewhat self indulgent, but I must ask the reader to indulge me, and at a little length.

What would a low-carbon energy system look like? (And let's avoid such fanciful ideas as "zero carbon" because that would be truly self indulgent.) In essence, we would get as much electricity as possible from some combination of renewable and nuclear energy, and electrify as many aspects of our energy systems as is feasible. Predicting the relative composition of such a system is a largely fruitless exercise. However, we can say something about the extent to which a low-carbon energy system will be distributed and "local." This confidence comes from the difference between the high physical concentrate of energy use in cities, and the relatively low physical concentration of renewable energy resources.

Power density

There are fundamental physical limits to how much energy we can extract from renewable resources for a given area of land. If we want to rigorously quantify this we calculate an energy source's power density in watts per square meter (W/m2 ).

To get an understanding of this concept, consider the recently opened London Array wind farm to the south of England. This is the world's largest offshore wind farm and according to its owners will generate "enough energy to power nearly half a million homes." Its total capacity is 630 MW, covering a total of 100 km2, and is expected to have a capacity factor of 39 percent. In other words, the power density of the London Array will be 2.5 W/m2.

This number is also very similar to the average calculated by David MacKay for existing UK wind farms. The United Kingdom is windier than a lot of the world, and some research suggests that large extraction wind farms will reduce average power density closer to 1 W/m2, so 2-3 W/m2 can be viewed as an upper limit on the power density of large scale wind power. This power density reflects average output – peak power density of wind farms will be perhaps three times higher – and minimum power density will be close to zero. It should be noted that this calculation excludes the requirements for manufacturing steel needed for turbine towers and the extraction of fossil fuels for conversion to plastics for wind turbine blades. However, inclusion of these factors is not likely to result in a significant reduction to power density estimates.

Globally, solar radiation available for conversion to electricity averages 170 W/m2, and in sunnier locations can reach above 200 W/m2. This solar energy, however, is currently not converted at anywhere close to 100 percent efficiency. Commercial solar photovoltaic panels typically average between 10 and 15 percent efficiency. Power density of solar installations must also account for space between panels, either for servicing in solar farms or for spacing between houses in rooftop solar installation. As a result the highest power density achieved is around 20 W/m2 in desert solar PV farms, whereas solar farms in Germany generally achieve 5 W/m2. Future improvements in panel production will hopefully see significant improvements in panel efficiency. However there will remain a firm physical upper limit of 200 W/m2, which will be significantly lower when considering large-scale deployment of residential rooftop solar due to obvious physical restrictions on panel placement.

At their best biofuels might be able to produce close to 2 W/m2. However power densities of 0.5 W/m2 and below are more typical, with prominent examples of this being corn ethanol for transport and the burning wood for electricity. We will see later that this is a very important consideration for the scalability and sustainability of biofuels.

In contrast, typical generation of fossil fuel and nuclear electricity has a power density of at least an order of magnitude greater than that of renewable energy. Power densities are comfortably above 100 W/m2 after accounting for mining, and conventional power plants often have power densities in excess of 1000 W/m2. A simple example of this higher power density is this small propane powered generator, providing in excess of 1000 W/m2. This power density far exceeds any conceivable new method of generating renewable energy.

Why power density matters

A simple thought experiment demonstrates why power density needs to be a fundamental consideration when evaluating renewable energy: imagine a world where all energy comes from bio-energy. What would be the requirements?

Currently the planet consumes energy at a rate of over 16 trillion watts (TW). If we include non-commercial biomass energy used in Africa and Asia (an uncertain figure), this number would increase. For simplicity I will ignore non-commercial sources and will round our figure down to 15 TW. If we got all our energy from corn ethanol we would need to convert a total of 75 million km2 to corn ethanol plantations. This is roughly half of the land surface of the entire planet, land that is somewhat scarce. This simple thought experiment shows there are very real limits on how much energy we can, and should, get from biofuels. If we want large-scale biofuels to become truly sustainable, a questionable prospect, we will need to see significant improvements to their power density, perhaps improvements of at least an order of magnitude.

Physical concentration of energy consumption

How much energy do we consume per unit of land? For ease of comparison, this figure again can be calculated in W/m2. On a global level, if we only consider land surface area, the total is 0.1 W/m2. Global averages, however, are not very instructive; power density averaged at the scale of countries and cities is much more important. David MacKay has visualized this much better than I can in his "Map of the World." The following graph shows the average rate at which countries consume energy in W/m2, compared with the power density of different renewables:


Ideally a country wants to have lots of available land for renewable energy. (They want to be in the bottom left of this graph.) Being in the top right may lead to some problems.

Consider the United Kingdom and Germany. Both use energy at a rate of just over 1 W/m2. A back-of-the-envelope calculation will tell you that getting all of their energy needs from onshore wind will require covering half of the UK or Germany in wind turbines. If you have ever been confused by why these countries are building wind farms in the North Sea, instead of on land where it is much cheaper, now you know why. Wind energy's low power density means you need to put turbines in a lot of back yards.

Things are even worse in Japan and South Korea. If you covered all of South Korea in wind turbines, they would generate less energy than is consumed there. Japan has a similar problem. And this ignores another difficulty: trees. Both Japan (68%) and South Korea (63%) have very high forest cover. If we ignore forested land – which should be out-of-bounds for large-scale renewable energy generation, unless large-scale biomass plantations are deemed acceptable – energy is used with a power density of almost 6 W/m2 in Japan and 7.5 W/m2 in South Korea. This calculation makes it clear that these countries can only be predominantly powered by renewable energy through the large-scale utilization of power dense solar energy. Social and political constraints may mean this can only happen if the efficiency of typical solar panels increases significantly from their current 10-15 percent.

Local Energy Is Not A Solution

Some environmentalists and renewable energy advocates have an ideological preference for small and community-scale renewable energy. But what if your community looks like this?

Since 2008, the majority of humanity lives in cities. And by 2050 it is probable that we will see 70 or 80 percent of the human population living in cities. The key energy challenge this century will be providing energy for these cities, and quite clearly local distributed energy is not a solution. To see why this is the case requires untangling some issues.

Here are some considerations. An average North American has an annual energy consumption of just over 7 tonnes of oil equivalent (toe), which is the equivalent of a rate of 9,000 watts. However, this is almost double what it is in countries such as Germany, France, and Japan. A comparison of these countries in terms of key well-being measures makes one thing clear: there is no evidence that North Americans have greater well-being as a result of their excessive energy use. Americans don't live longer, aren't healthier, or better educated than countries that consume half as much energy per capita. That this high per capita energy consumption comes with a very significant environmental cost – global carbon dioxide emissions would drop by almost 10 percent if North Americans consumed like Europeans –suggests that is not desirable for other countries to emulate North American consumption patterns.

Further evidence for the desirability to limit, and probably reduce, per capita energy consumption in modernized countries is given by its evolution in recent decades. Instead of increasing in the long-term, per capita energy consumption now appears to have peaked in almost all modernized countries. Here are some examples:

Per capita consumption has decline steadily in the United Kingdom for the last decade and is now at its lowest point in over four decades:

The United States saw peak per capita consumption in the 1970s, with consumption now seeing an apparent decline. And the fact that per capita consumption did not rise in the age of the Hummer suggests significant room for movement.

Germany is also now seeing declines in per capita consumption.

In Japan per capita consumption appears to have peaked in the late ‘90s and is now in decline:

Many modernized countries are now seeing reductions in per capita energy consumption, and this is not being accompanied by a reduction in quality of living. Any sensible long-term energy and climate policy should include a strong desire to continue this trend. The belief that the world can transition to both American levels of energy consumption and to a low-carbon energy system by the middle of this century ignores the vital lessons of previous energy transitions, and given the current position of renewable and nuclear energy it appears delusional. The world therefore must be much more like Japan than America.

And cities must play a key role in reducing energy consumption. The most important and effective way to do this is simple: make them dense. For a full elucidation of why, I recommend books by Edward Glaeser and David Owen. But the key reasons are easy to understand: a dense city lets you walk or take public transport instead of drive and lets you live in a more energy efficient apartment building instead of a large inefficient house. Packing people more tightly together in cities may not be to the taste of everyone, but it appears to be one of the most achievable and practical ways to reduce how much energy people consume.

Let us now move forward to 2050 to the world as I hope it will be. Global population will have peaked below 9 billion as a result of the spread of the demographic transition to modernizing countries, and the success in reducing infant mortality and widespread availability of contraception. Perhaps 7 billion of us will live in cities, and they will consume much more like modern day Japanese than Americans.

How will we provide energy for these cities? The answer appears to be large, centralized power plants, whether they are wind, solar, or nuclear. Here I assume, wishfully, that we have managed to get rid of fossil fuels, an unlikely prospect. The answer however is almost certainly not local distributed energy, and for simple reasons.

Consider Manhattan, not where many would consider the green ideal. Yet here you will find significantly lower per capita energy consumption than in almost every American city. You will also find energy consumption far greater than can conceivably be provided by local renewables. A recent study managed to map energy consumption in the city that never sleeps right down to the individual city block. This is what it looks like:


A typical block in Manhattan consumes energy at a rate of over 1,000 kWh per square meter each year – a power density of over 100 W/m2. This is almost two orders of magnitude greater than the power density of wind power, and obviously you could not plaster Manhattan in wind turbines. Solar power is not much better. Imagine that we could cover 20 percent of Manhattan in solar panels. This would give us no better than 5 W/m2. Clearly Manhattan is not getting its energy locally. And as you can see from the above map the other boroughs of New York are not going to fare much better.

How about the rest of North America? If we reduced per capita energy consumption to Japanese levels, a sensible but unpopular idea, could many American cities run largely on local renewables? The graph below shows population density versus energy use density in this lower consumption America:

Low density Phoenix perhaps has a shot at getting most of its energy from solar power. Covering 25 percent of Phoenix in solar panels would produce as much energy as is consumed in Phoenix. The practicality of this is rather questionable, and getting more than 50 percent of Phoenix's energy from local solar will require something that currently does not exist: a cheap way to store energy on a large scale. A system involving more than 50 percent of energy coming from solar will thus inevitably require the accounting of land requirements for large-scale storage, an uncertain prospect, and significant losses of solar panel output due to efficiency losses during storage and curtailment of excess.

The prospects of American cities running largely on local renewables thus seems unlikely, and 83 percent of Americans live in cities.

A global appraisal

The world's 200 largest urban areas are home to over 1.2 billion people, and a quarter of these areas are more densely populated than New York (10,000 people per square kilometer). This is shown below:

Before asking if these cities can run on local renewables I must first mention the too real disparities in global energy consumption. Below is a comparison of the population of countries with their per capita energy consumption, with population plotted on a logarithmic scale due to China and India being much larger than other countries. I include lines showing typical European and North American per capita energy consumption.

While there are about 350 million North Americans who can, and should, reduce their energy consumption to European levels, there are even more at the bottom who must increase their energy consumption significantly to improve their quality of life. Quantitative comparisons are sobering. Over 35 countries have per capita consumption at less than 10 percent of North American levels, with populations totaling over 2 billion. Despite the apparent desires of some environmental NGOs (see page 11 of this WWF report) it is therefore undesirable to propose reductions in global energy consumption. The modernized world may consume excessive energy, but energy consumption is much too low in modernizing countries to let us decrease global energy consumption without negative humanitarian impacts.

We therefore should have a desire to both reduce excessive consumption in modernized countries and increase energy consumption in modernizing countries. I am not going to suggest a prescriptive end point. Instead I will assume that consumption levels in modern day Japan can provide a very good quality of life, and exceeding these levels is unnecessary.

If the populations of the world's 200 largest cities consumed energy like modern day Japanese energy use density would look like this:


In total, 10 cities would have power density greater than 100 W/m2, 56 would have power density greater than 50 W/m2, while 181 would have power density of over 10 W/m2. Ninety percent of the planet's 200 largest cities almost certainly cannot be powered predominantly by local renewable energy. The population densities of these cities are not significantly different than the rest of the world's cities, so we can conclude that the vast majority of cities cannot be powered by local renewables. And this suggests very serious limits to the role of local distributed energy in a world where more than 70 percent of us will probably live in cities.

The prospects are even worse in individual countries. Of the world's 200 largest urban areas, 17 are in India. Below I have isolated these cities.

One hundred twenty million people live in these cities. Covering any of them entirely in 10 percent efficient solar panels will generate less than half of their energy needs. And look at that dot in the top right that is Bombay. For Bombay to get all of its energy needs from solar in my hypothetical future it would need to harness almost 100 percent of the solar radiation that strikes it, a remote prospect. This extremely high population density is routinely ignored by Western environmentalists calling for distributed energy to be the solution to India's energy problems. It quite clearly is not.

In this century, the bulk of humanity will live in large, densely populated cities. If the citizens of these cities are to attain a high quality of life they will require large centralized energy generation. This is not a matter of ideological preference, but of engineering reality.


Robert Wilson is a PhD Student in Mathematical Ecology at the University of Strathclyde. This post is reprinted with permission from The Energy Collective.

Photo Credit: Wikipedia Commons


  • 1. Highly centralized generation is too vulnerable to terrorist attacks.
    2. Transmitting electricity efficiently is difficult. Solar Roadways is probably an important future option of solar generation and transmission.
    3. Using renewables to generate electricity and store surplus as natural gas or hydrogen warrants more investment in working prototypes.
    4. Pyrolysis of biomass (into natural gas or hydrogen) warrants more investment in working prototypes.
    5. Wave energy and OTEC warrant more investment in working prototypes.

    By John Casper on 2013 08 19

    Reply to this comment / Quote and reply

  • John

    Thanks for shaping your bullet points according to the content of my piece.


    By Robert Wilson on 2013 08 20

    Reply to this comment / Quote and reply

  • Some initial thoughts in reaction to your article:
    1)  Does your energy consumption data cover every means, including transportation?  If so, perhaps a North American goal to match Europe or Asia is overly aggressive.  A much lower target is perhaps reasonable still.
    2)  The discussion might also benefit from separating renewable energy from locally generated/distributed energy.  Especially in urban areas, combined heat and power, or better yet, combined cooling, heat, and power presents huge upsides.  And from my exposure to various researchers and their reporting, even a pure renewable future requires a good deal of centralized production, if one is being honest.
    3)  Producing locally in urban areas also requires localized or micro grids.  There are upsides to his as well, especially with respect to vulnerabilities whether environmental, cyber, etc.  After all, we have proven at least twice in the most populated region of the U.S. (the northeast) that the responsibility to adequately manage a large and growing macro grid is more challenging than perhaps all the parties can manage.  This leads to famous blackouts and brownouts.
    4)  Perhaps the question that is unanswered goes something like this:  For any given urban area, what is or how do you set the reasonable target which balances distributed and centralized energy portions?

    As is the case whenever politics (or socio-politics) and technology are combined, too many people take the extreme position and argue the case.  But the best future for almost any region on the globe is likely NOT found in those arguments.  While European population densities and their international grid networks may lead to one solution, what may work better for North America will likely be different.  Surely in Europe people will see the folly of forcing renewable sources to work in regions that aren’t conducive (such as solar PV in Germany) and limit them to regions that are better suited (such as wind, onshore and offshore, in the UK).  And surely in North America ultimately people will see the folly of building an ever larger and more complex grid in order to satisfy “sustainable” or “green” sociopolitical needs.  Regions in the Northeast, such as NYC, already depend on regions in Quebec and New Brunswick, Canada, for power that isn’t fossil derived.  And projects are already underway to transmit power 2,000 miles or greater to provide wind farm output from one region to another distant one.

    Just as the industrialized nations have thrived throughout a significantly variable and evolving ‘power density’ history due to the nature of the fuel sources and evolution of efficiency/technology, so can those same nations find their own balance between centralized and distributed, and fossil and renewable, while meeting their economic and reliability targets.  Taking advantage of existing urban real estate with PV installations (free fuel!!) is hard to resist, for example in Phoenix.  Providing the least expensive energy from large fossil plants (and nuclear) is also hard to resist.  But targeting only big, remote centralized plants is not a perfect solution.  Likewise, some hypothetically perfectly distributed, storage intense, non-fossil future with mandatory reduced demand (through demand response, etc.) is equally imperfect.

    By John Sullivan on 2013 08 27

    Reply to this comment / Quote and reply

  • While I agree we cannot completely decentralize power generation in large cities, we can at the very least create increased competition which is better for everyone in the long run.  In cases where the power company is not providing the best prices, the consumer then has the option to install solar panels and become an instant power company “competitor”.  While this is not possible for everyone, due to various reasons, the increased competition is better for everyone.

    -Mike of Solar Well Pumps

    By Mike on 2013 09 02

    Reply to this comment / Quote and reply

  • Food for thought:

    China are aiming to take grid-installed wind capacity to 75GW by the end of this year, according to the China Electricity Council (CEC).

    Wind would then be the 3rd largest power source, 860GW thermal and 280GW hydro.

    Three largest power sources are renewables.


    With respect to cost, suprisingly, probably for some of you, is that in Austalia, wind power currently beats fossil fuel generation on cost. New wind farms are capable of providing $80/MWhr whereas coal or gas are at $143 and $116/MWhr, once the price of carbon emmissions is factored in. This tax was introduced in July 2012, but even without the carbon price, wind is still cheaper than coal & gas by 14-18%. As the costs of fossil fuels have risen, there has been a 10 percent decrease in the cost of wind power since 20-11 and a 29 percent decrease in the cost of solar.

    By Benjamin on 2013 10 15

    Reply to this comment / Quote and reply

  • I have been doing some research on solar panel and wind tnibrue systems. The pricing is crazy. For a 10Kw solar grid tie kit system is between $35,000 and $40,000. If you install batteries add $3,000-$5,000. 10Kw will only power a small to medium home without electric heat. Using my electric usage and info from my local electric company payback is 60 years for solar and 361 years for wind. This is unexceptionable. The manufactures of the solar and wind are gouging the consumers. They are using the excuse of supply and demand for the high cost. When demand goes up so will mass production and the price will go down. This is true, but they are making a huge profit on what they are making now. Some items are in the 3000% markup range. Are the power companies keeping the price up so their profits don’t suffer? My local electric supplier has an ENERGY PARK that you can go see solar and wind power in action and check out real time data online. They have a 4Kw solar array installed and claim it cost $30,378 installed and a 2.5Kw wind tnibrue that cost about $20,000 installed. I think they are giving false numbers to consumers so you won’t go green. I was able to find a 3.5Kw kit online for $9,000. This doesn’t include any type of mounting materials. You can add $800-$1,200 for these materials. An installer will will charge you $2,000-$4,000 to install it. For the sake of argument lets say this will cost you $15,000 to install but still way out of line for you and me to install. That is a far cry from the $30,000 the power company claims. The power companies don’t want you to make your own electric. They are keeping the cost high so they will still make money off you. A 210w panel sells for around $600. They probably manufacture it for under $50. The cost needs to be $100-$150 to the consumer to make it affordable to the average home owner. 50 210w panels make a 10Kw system. That would be $5,000 not $40,000. The manufactures are raping us and the power companies are helping them to keep prices high. When will we get the technology at a proper cost? Let me know what you think.In response to the first 5 answers, Yes, if demand increases so will mass production and thus supply will increase and this will drive the price down. I looked into panels from China. From what I was able to find panels sell for $0.17-$0.45 per watt. In the U.S. they sell for $3.00-$5.00 per watt. Manufactures in the U.S. have lobbied for a higher tariff on solar products that makes them impractical to import on mass. These companies want to keep the price high so their profits stay high, and the power companies do not want you to produce all your own power. That would put them out of business.In response to Steve R, You need to do more research before you post another retarded comment like that. Air DOES have mass (just not very dense). When air is in motion it is called WIND . This motion creates energy. Have you ever heard of a tornado ? Get your facts together next time.

    By Negar on 2014 09 29

    Reply to this comment / Quote and reply

  • For architectures, this revolution had allowed them to get a better perspective of forthcoming endeavors including
    , commercial and residential components. Regardless of architectures, in addition, there are other professionals who use the techniques. Some instances of the professionals are multimedia suppliers and engineers, interior designers. Oftentimes, the professionals use techniques and designing jobs.

    By celinarpinkston on 2017 02 19

    Reply to this comment / Quote and reply

Submit a comment