The Azimuth Project
Stabilization wedges (Rev #29, changes)

Showing changes from revision #28 to #29: Added | Removed | Changed



Stephen Pacala and Robert Socolow have sketched a flexible plan for tackling the global warming problem for the next 50 years using only present technologies:

As they write:

The debate in the current literature about stabilizing atmospheric CO2 at less than a doubling of the preindustrial concentration has led to needless confusion about current options for mitigation. On one side, the Intergovernmental Panel on Climate Change (IPCC) has claimed that “technologies that exist in operation or pilot stage today” are sufficient to follow a less-than-doubling trajectory “over the next hundred years or more”. On the other side, a recent review in Science asserts that the IPCC claim demonstrates “misperceptions of technological readiness” and calls for “revolutionary changes” in mitigation technology, such as fusion, space-based solar electricity, and artificial photosynthesis. We agree that fundamental research is vital to develop the revolutionary mitigation strategies needed in the second half of this century and beyond. But it is important not to become beguiled by the possibility of revolutionary technology. Humanity can solve the carbon and climate problem in the first half of this century simply by scaling up what we already know how to do.

They list 15 measures, each of which could reduce carbon emissions by 1 billion tons per year by 2057. They believe global warming would be manageable, though still a serious problem, if 12 of these measures were carried out by that time. They call these measures stabilization wedges, thanks to their appearance in a chart that illustrates their effects.


The problem to be solved

Pacala and Socolow wrote this now-famous paper in 2004. Back then we were emitting about 6.2 gigatons of carbon per year, there were 375 ppm of carbon dioxide in the atmosphere, and many proposals to limit global warming urged that we keep the concentration below 500 ppm. Their paper outlined some strategies for keeping it below 500 ppm.

They estimated that to do this, it would be enough to hold emissions flat at 7 gigatons of carbon per year for 50 years, and then lower it to nothing. On the other hand, in a “business as usual” scenario, they estimate the emissions would ramp up to 14 gigatons per year by 2054. That’s 7 too many.

So, to keep emissions flat it would be enough to find 7 ways to reduce carbon emissions, each one of which ramps up linearly to the point of reducing carbon emissions by 1 gigaton/year in 2054. They called these stabilization wedges, because if you graph them, they look like wedges:

Despite the above graph, they don’t really think carbon emissions will increase linearly in a business-as-usual scenario. This is just a deliberate simplification on their part. They also show this supposedly more accurate graph:

They say the top curve is “a representative business as usual emissions path” for global carbon emissions in the form of CO2 from fossil fuel combustion and cement manufacture, assuming 1.5% per year growth starting from 7.0 GtC/year in 2004. Note this ignores carbon emissions from deforestation, other greenhouse gases, etc. This curve is growing exponentially, not linearly.

Similarly, the bottom curve isn’t flat: it slopes down. They say the bottom curve is a “CO2 emissions path consistent with atmospheric CO2 stabilization at 500 ppm by 2125 akin to the Wigley, Richels, and Edmonds (WRE) family of stabilization curves described in [11], modified as described in Section 1 of the SOM text.”

Here reference [11] is:

  • T. M. L. Wigley, in The Carbon Cycle, eds. T. M. L. Wigley and D. S. Schimel, Cambridge U. Press, Cambridge, 2000, pp. 258–276.

and the “SOM text” is the supporting online material for their paper, which unfortunately doesn’t seem to be available for free.

The proposed solutions

Their paper listed 15 possible stabilization wedges, each one with the potential to reduce carbon emissions by 1 gigaton/year by 2054. This is a nice way to start thinking about a very big problem, so many people have adopted it and modified it and criticized it in various ways, which I hope to discuss later.

Before listing their stabilization wedges, we should emphasize: stabilizing emissions at 7 gigatons is not enough to stay below 500 ppm forever! Carbon dioxide stays in the atmosphere a very long time. So, as Pacala and Socolow note:

Stabilization at any level requires that net emissions do not simply remain constant, but eventually drop to zero. For example, in one simple model that begins with the stabilization triangle but looks beyond 2054, 500-ppm stabilization is achieved by 50 years of flat emissions, followed by a linear decline of about two-thirds in the following 50 years, and a very slow decline thereafter that matches the declining ocean sink. To develop the revolutionary technologies required for such large emissions reductions in the second half of the century, enhanced research and development would have to begin immediately.

What’s the “declining ocean sink”? Right now the ocean is absorbing a lot of CO2, temporarily saving us from the full brunt of our carbon emissions — while coral reefs, shellfish and certain forms of plankton suffer from increased acidity. But this won’t go on forever; the ocean has limited capacity.

Pacala and Socolow consider several categories of stabilization wedges:

  • efficiency and conservation
  • shifting from coal to gas
  • carbon capture and storage
  • nuclear fission
  • renewable energy sources
  • forests and agriculture

Here are all 15 stabilization wedges:


We shall describe and analyze these in turn.

Wedges 1-4: efficiency and conservation

1) Efficient vehicles: increase the fuel economy for 2 billion cars from 30 to 60 miles per gallon. Or, for those of you who don’t have the incredible good luck of living in the USA: increasing it from 13 to 26 kilometers per liter. When they wrote their paper, there were 500 million cars on the planet. They expected that by 2054 this number would quadruple. When they wrote their paper, average fuel efficiency was 13 kilometers/liter. To achieve this wedge, we’d need that to double.

2) Reduced use of vehicles: decrease car travel for 2 billion 30-mpg cars from 10,000 to 5000 miles per year. In other words: decreasing the average travel from 16,000 to 8000 kilometers per year. (Clearly this wedge and the previous one are not additive: if we do them both, we don’t save 2 gigatons of carbon per year.)

3) Efficient buildings: cut carbon emissions by one-fourth in buildings and appliances. This could be done by following “known and established approaches” to energy efficient space heating and cooling, water heating, lighting, and refrigeration. Half the potential savings are in the buildings in developing countries.

4) Efficient coal plants: raise the efficiency of coal power plants to 60%. In 2004, when they wrote their paper, “coal plants, operating on average at 32% efficiency, produced about one fourth of all carbon emissions: 1.7 GtC/year out of 6.2 GtC/year.” They expected coal power plants to double their output by 2054. To achieve this wedge, we’d need their average efficiency to reach 60%.

Wedge 5: shifting from coal to gas

5) Shifting from coal to natural gas: replace 1400 gigawatts coal-burning power plants with gas-burning plants. Natural gas puts out half as much CO2 as coal does when you burn them to make a given amount of electricity. After all, it’s mainly methane, which is made from hydrogen as well as carbon. Suppose by 2054 we have coal power plants working at 90% of capacity with an efficiency of 50%. 700 gigawatts worth of coal plants like this emit 1 gigaton of carbon per year. So, we can reduce carbon emissions by one ‘wedge’ if we replace 1400 gigawatts of such plants with gas-burning plants. That’s four times the 2004 worldwide total of gas-burning plants.

Wedges 6-8: carbon capture and storage

6) Capturing CO2 at power plants: 800 GW of coal plants or 1600 GW of gas plants . Carbon capture and storage at power plants can stop about 90% of the carbon from reaching the atmosphere, so we can get a wedge by doing this for 800 GW of baseload coal plants or 1600 GW of baseload gas plants by 2054. One way to do carbon capture and storage is to make hydrogen and CO2, burn the hydrogen in a power plant, and inject the CO2 into the ground. So, from one viewpoint, building a wedge’s worth of carbon capture and storage would resemble a tenfold expansion of the plants that were manufacturing hydrogen in 2004. But it would also require multiplying by 100 the existing amount of CO2 injected into the ground.

7) Capturing CO2 at plants that make hydrogen for fuel: 250 megatons/year made from coal, or twice as much made from gas. You’ve probably heard people dream of a hydrogen economy?. But it takes energy to make hydrogen. One way is to copy wedge 6, but then ship the hydrogen off for use as fuel instead of burning it to make electricity at power plants. To capture a wedge’s worth of carbon this way, we’d have to make 250 megatons of hydrogen per year from coal, or 500 megatons per year from natural gas. This would require a substantial scale-up from the 2004 total of 40 megatons of hydrogen manufactured by all methods. There would also be the task of building the infrastructure for a hydrogen economy. The challenge of injecting CO2 into the ground would be the same as in wedge 6.

8) Capturing CO2 at plants that turn coal into synthetic fuels: 1.8 teraliters of synfuels per year. As the world starts running out of oil, people may start turning coal into synfuels, via a process called coal liquefaction. Of course burning these synfuels will release carbon. But suppose only half of the carbon entering a synfuels plant leaves as fuel, while the other half can be captured as CO2 and injected underground. Then we can capture a wedge’s worth of CO2 from plants that produce 1.8 teraliters of synfuels per year. For comparison, total yearly world oil production in 2004 was 4.7 teraliters

Wedge 9: [[nuclear power]]

9) Replacing 700 gigawatts of coal-fired power plans with nuclear power. As Pacala and Socolow already argued in wedge 5), replacing 700 gigawatts of efficient coal-fired power plants with some carbon-neutral form of power would keep us from burning one gigaton of carbon per year. To do this with nuclear power would require 700 gigawatts of nuclear power plants running at 90% capacity (just as assumed for the coal plants). The means doubling the world production of nuclear power. The global pace of nuclear power plant construction from 1975 to 1990 could do this! But of course, there’s still a downside: we can only substantially boost the use of nuclear power if people become confident about all aspects of its safety.

Wedges 10-13: renewable energy

10) Replacing 700 gigawatts of coal-fired power plants by 2000 gigawatts of peak wind power. Wind power is intermittent: Pacala and Socolow estimate that the ‘peak’ capacity (the amount you get under ideal circumstances) is about 3 times the ‘baseload’ capacity (the amount you can count on). So, to save a gigaton of carbon per year by replacing 700 gigawatts of coal-fired power plants, we need roughly 2000 gigawatts of peak wind power. Wind power was growing at about 30% per year when they wrote their paper, and it had reached a world total of 40 gigawatts. So, getting to 2000 gigawatts would mean multiplying the world production of wind power by a factor of 50. The wind turbines would ‘occupy’ about 30 million hectares, or about 30-45 square meters per person — some on land and some offshore. But because windmills are widely spaced, land with windmills can have multiple uses.

11) Replacing 700 gigawatts of coal-fired power plants by 2000 gigawatts of peak photovoltaic solar power. Solar power is also intermittent. Pacala and Socolow estimate that to save a gigaton of carbon per year, we need 2000 gigawatts of peak photovoltaic solar power to replace coal. Like wind, photovoltaic solar was growing at 30% per year when Pacala and Socolow wrote their paper. However, only 3 gigawatts had been installed worldwide. So, getting to 2000 gigawatts would require multiplying the world production of photovoltaic solar power by a factor of 700. In terms of land, this would take about 2 million hectares, or 2-3 square meters per person.

12) Using 4000 gigawatts of peak wind power to generate hydrogen for powering automobiles. Renewable energy can be used to produce hydrogen for vehicle fuel. 4000 gigawatts of peak wind power, for example, used in high-efficiency fuel-cell cars, could keep us from burning a gigaton of carbon each year in the form of gasoline or diesel fuel. Unfortunately, this is twice as much wind power as we’d need in wedge 10, where we use wind to eliminate the need for burning some coal. Why? Gasoline and diesel have less carbon per unit of energy than coal does.

13) Making 5.4 gigaliters of bioethanol per day to replace gasoline. Fossil fuels can also be replaced by biofuels such as ethanol. To save a gigaton per year of carbon, we could make 5.4 gigaliters per day of ethanol as a replacement for gasoline — provided the process of making this ethanol didn’t burn any fossil fuels! Doing this would require multiplying the world production of bioethanol by a factor of 50. It would require 250 million hectares committed to high-yield plantations, or 250-375 square meters per person. That’s an area equal to about one-sixth of the world’s cropland. An even larger area would be required to the extent that the biofuels require fossil-fuel inputs. Clearly this could cut into the land used for growing food.

Wedge 14 and 15

14) Stop deforestation, start reforestation. They say we could stop half a gigaton of carbon emissions per if we completely stopped clear-cutting tropical forests over 50 years, instead of just halving the rate at which they’re getting cut down. For another half gigaton, plant 250 million hectares of new forests in the tropics, or 400 million hectares in the temperate zone!

To get a sense of the magnitude here, note that current areas of tropical and temperate forests are 1500 and 700 million hectares, respectively.

Pacala and Socolow also say that another half gigaton of carbon emissions could be prevented by created by establishing approximately 300 million hectares of plantations on nonforested land.

15) Soil management. When forest or grassland is converted to cropland, up to one-half of the soil carbon gets converted to CO2, mainly because tilling increases the rate of decomposition by aerating undecomposed organic matter. Over the course of history, they claim, 55 gigatons of carbon has gone into the atmosphere this way. That’s the equivalent of two wedges. (Note that one wedge, ramping up linearly to 1 gigaton/year for 50 years, adds up to 25 gigatons of carbon by 2054.)

However, good agricultural practices like no-till farming can reverse these losses — also reduce erosion! By 1995, these practices had been adopted on 110 million of the world’s 1600 million hectares of cropland. If this could be extended to all cropland, accompanied by a verification program that enforces practices that actually work as advertised, somewhere between half and one gigaton of carbon per year could be stored in this way. So: maybe half a wedge, maybe a whole wedge!


As a general remark, it seems a little weird only to be listing technologies that are currently viable on large scales when we’re talking about 50 years out. But most likely the point that Pacala and Socolow want to make is that keeping emissions flat is possible with current technologies only.

Wedges 1 and 2: vehicles

Number of cars

The emissions model assumes a four-fold increase in the number of cars in 50 year: is this reasonable? It seems a plausible back-of-the-envelope estimate. Firstly, Worldometers suggests that China and India have lots of “capacity” if they reach a sufficiently high income:

China became the world’s third-largest car market in 2006, as car sales in China soared by nearly 40% to 4.1 million units. China should become the world’s second-largest car market by 2010, as low vehicle penetration, rising incomes, greater credit availability and falling car prices lift sales past those of Japan. Furthermore, vehicle penetration in China stands at only 24 vehicles per 1,000 people, compared with 749 vehicles per 1,000 people in the mature markets of the G7.

Secondly, a four-fold increase in 50 years is about 2.8 percent net increase (i.e. new cars being added minus old cars being removed) per year. Is this too high, too low or about right?

(According to Worldometers, world car production rose 6.45% in 2006, and in China it rose 40% that year. But in 2008 they report a decline in world car production, and they projected a further decline in 2009. The average world increase in car production from 1999 to 2009, based on their projection for 2009, was 3.2% per year. However, this quantity is only indirectly related to the growth in number of cars on the road.)

Higher efficiency of cars

More fuel efficient vehicles already exist. See, e.g. Most and least fuel efficient cars. The main question is how to urge the consumer to use these cars? One possible solution is higher taxation: both a tax on gasoline, and one on inefficient cars. Related to taxation is the “natural” rise of oil product prices when oil becomes more scarce.

Nevertheless, there is a problem of attitude, that is that large cars are often a status symbol and drivers may be willing to pay more for driving these cars, to promote their social status.

In addition, the rebound effect may come into play: that is, simply making cars more energy efficient might lead to more cars purchased or more distance traveled. However, because cars are a status symbol, people voluntarily buying energy efficient cars may also voluntarily try to reduce, or keep equal, their distance traveled. On the other hand, people buying energy efficient cars only to reduce taxation, may use the savings in their budget to travel more. For some empirical data, see:

Abstract: We estimate the rebound effect for motor vehicles, by which improved fuel efficiency causes additional travel, using a pooled cross section of US states for 1966-2001. Our model accounts for endogenous changes in fuel efficiency, distinguishes between autocorrelation and lagged effects, includes a measure of the stringency of fuel-economy standards, and allows the rebound effect to vary with income, urbanization, and the fuel cost of driving. At sample averages of variables, our simultaneous-equations estimates of the short- and long-run rebound effect are 4.5% and 22.2%. But rising real income caused it to diminish substantially over the period, aided by falling fuel prices. With variables at 1997-2001 levels, our estimates are only 2.2% and 10.7%, considerably smaller than values typically assumed for policy analysis. With income at the 1997 – 2001 level and fuel prices at the sample average, the estimates are 3.1% and 15.3%, respectively.

Another danger of taxation may be that when the tax becomes an income for the government, the government may become less determined to use the tax for what its purpose is for: to reduce the number of inefficient cars.

Finally, it would be reassuring to check that people do respond to taxation in the way the economic model predicts. The evidence in the UK suggests that people don’t respond to tax “incentives” in the smooth way. The already high (by global standards) tax on petrol increased by 8.7% between 2000 and 2010, at a time when incomes and inflation in the UK weren’t increasing much. Over the same time period private car miles driven (admittedly extrapolated from sampling, so possible issues) increased by 4.5 percent per driver (i.e., this isn’t due to there being more drivers) in the same time. Maybe the taxation has suppressed even large rises in miles driven, but at first glance this is not what theoretical economists argue should happen in the presence of rising taxes. Sources:

Another option is to impose stronger emission norms for cars. In this case, energy inefficient cars can neither be produced nor bought.

Less distance traveled

Taxation could also be useful for reducing the distance traveled. In addition, for large distances (> 2020 km) other means of transportation are more energy-efficient than cars. For small distances, transportation by foot and bicycle is also more environmentally friendly.

We need references here.

As trend indicators, calculated risk has a graph of US vehicle miles travelled shows US miles driven has begun increasing again.

Wedge 4: efficiency of coal plants

The Rankine cycle is the cycle used in steam turbine power plants. The overwhelming majority of the world’s electric power is produced with this cycle. Since the cycle’s working fluid, water, changes from liquid to vapor and back during the cycle, their efficiencies depend on the thermodynamic properties of water.

The thermal efficiency of modern steam turbine plants with reheat cycles can reach 47%, and in combined cycle plants it can approach 60%. An example of the latter is given by the GE H system. There are programs that attempts to develop plants with these and even higher efficiences by 2015, e.g. Vision 21.

Can we find more information on similar programs?

See also:

Wedge 5 and 6: shifting from coal to gas and carbon capture

Wedge 5 and 6 would certainly lower the amount of carbon released into the atmosphere, but there are some problems.

Wedge 5: how can a shift from coal to gas be enforced? Some Western countries are currently burning less coal voluntarily, but instead they are shipping it China ( International Herald Tribune , 23/11/10). An idea could be to pay countries to temporarily stop mining coal.

Wedge 6: there is some discussion about the feasibility of carbon capture and storage.

These issues are discussed into more detail on carbon capture and storage.

Wedge 7: hydrogen economy

So far, there have been no comments on wedge 7, but there are also ideas for a methanol economy.

Some thoughts on hydrogen as an energy carrier: outlook for 2010, 2030 and 2050 from the year that Pacala and Sokolow published their paper on the stabilization wedges:

Wedge 8: coal into synfuels

Pacala and Socolow say 1 gigaton carbon/year is the flow of carbon in 24 million barrels/day, or 1.4 teraliters/year. They assume the same value for synfuels and allow for imperfect capture, which leads them to conclude that carbon capture at synfuels plants producing 1.8 teraliters/year of synfuel can catch 1 GtC/year. Does this calculation make sense? If we’re catching just half the carbon, and 1 GtC/year = 1.4 teraliters oil/year, don’t we need to generate at least twice that — 2.8 teraliters synfuel/year — to catch wedge’s worth of carbon?

It’s also unclear what percentage of the carbon you can actually capture while turning coal into synfuels. Can we really capture half of it?

There’s also another funny feature of this last wedge. If we assume people are already committed to making synfuels from coal, then it may be true they will emit less carbon if they use carbon capture and storage as part of the manufacturing process. But compared to making electricity or hydrogen as in wedges 6 and 7, turning coal into synfuels seems bound to emit more carbon, even with the help of carbon capture and storage.

In general, it only makes sense to talk about how much carbon emission some action prevents when we compare it to some alternative action. That’s pretty obvious, but it gets a bit confusing when some of Pacala and Socolow’s wedges look like plausible alternatives to other ones.

Wedge 9: nuclear energy

From a technological point of view this wedge could be achieved rather easily, but there is (at least in some Western countries) a strong opposition against nuclear energy (a “culture war”). Nevertheless, as a replacement of coal-fired power plants, the external costs of nuclear energy should be compared to those of burning coal.

Wedge 10-13: renewable energy

To use wind power to reduce our coal burning by 1 gigaton of carbon per year (wedge 10) and also reduce vehicle carbon emissions by 1 gigaton of carbon per year (wedge 12), we would need to increase the amount of wind power produced worldwide by a factor of 150, starting from its 2004 level (this would accomplish about 30% of the carbon reductions we need by 2054 to keep CO2 levels below 500 ppm). To grow by a factor of 150 over 50 years, wind power would only need to grow at an annual rate of 10,5%.

Most of the discussion on the blog about solar panels was focused on their average power output and the amount of land needed, etc., instead of whether the necessary growth rate for solar power is feasible. For more information and specific critiques of technologies, see e.g. wind power, solar power and biofuels. About Pacala and Socolow’s calculation for wind power: it does not suffice to account for intermittency by equating baseload capacity to one third of nominal peak capacity, one needs to supplement wind farms with e.g. pumped storage systems to account for variabilities like lulls. See Capacity factor of wind power for a more information.

A problematic issue with wedges 10-13 is that if burning coal will keep on providing the cheapest form of energy (neglecting externalities) it will be difficult to replace it with other, more expensive, forms of energy.

Wedge 12: Using 4000 gigawatts of peak wind power to generate hydrogen for powering automobiles

It is remarkable that Pacala and Socolow opt for hydrogen vehicles instead of EVs/PHEVs as the former are less efficient and more expensive. The following argument leads to the estimate that a hydrogen powered vehicle fleet would require ~1,5-3 times as much renewable generation as an electric version.

For hydrogen from renewables, the NREL assumes 75% efficiency via electrolysis. Once produced there will likely be losses associated with transporting and refueling, and apparently there are significant losses due to leakage, which David MacKay expected to be quite severe in a production ready version. If we use it in an internal combustion engine we’ll see the same rough efficiency limits of ~40% (~200g/kWh gasoline equivalent) give or take, plus the time it takes the engine to warm up and light off the emissions system. Realistically the best we could hope for is ~35% efficiency in that arena. Even a Prius is only ~30-35% efficient on the unadjusted EPA tests IIRC. The wind to wheel efficiency in that case would be ~25%, compared to an EV with a wind to wheel efficiency of ~70+%. Fuel cells are possibly more efficient, but aren’t available at reasonable costs yet.

Perhaps this argument should be written out in more detail and moved to a separate page about Energy efficiency of vehicles?

Wedge 14: Stop deforestation, start reforestation

The UN’s REDD+ program is a plan to reduce emissions from deforestation and forest degradation. There is some controversy, see e.g. the Wikipedia article. There’s also a lot of argument about just how much long-term impact on atmospheric CO2 a standing forest has.

In any case, forest monitoring is possible by remote sensing, see e.g.remote sensing, see e.g.

Pacala’s Critique

Stephen Pacala critiques his own plan in light of new evidence here:

The discussion of recent evidence begins at 10:58.

At 33:34 he brings up a slide summarizing the reasons he is more pessimistic. Summarized in these points:

  • Fossil fuel emissions have been growing at 3% instead of the assumed 1.5%.

  • The Ocean CO2 sink has been growing less than anticipated.

  • CO2 fertilization (the increased growth rate in plants because of the presence of more CO2 in the atmosphere) which is a very important sink in all climate models has not been growing as expected.

  • If CO2 fertilization fails that is equivalent to requiring another 26 wedges in addition to the 8 in the original plan.

category: action