You know that one of my favorite sports is bagging on forecasts of all types, so when I come across one that’s pretty decent, I think it’s worth highlighting. Here, we see that the Department of Energy’s 1996 forecast (drawn from here) does pretty well. They overprojected the price declines up until 2000, but as you can see in the bottom graph (from Ryan Wiser’s 2007 Berkeley Lab report), the wind industry quickly caught up as the price of wind electricity dropped from 6 cents a kilowatt hour to 4 from just 1999 to 2002.

(The scales of the charts above are different: One is cents per kilowatt hour, the other in $ per megawatt hour. Basic conversion: $10/MWh = $0.01/KWh. My apologies. I’d love to play Remake the Chart today, but I’ve got a few thousand words to write.)

Note, too, that the DOE projection shows the cost reduction curve flattening out around the 3-4 cent range, which is exactly what’s happened. What they did miss is that the price spread for different projects is wide. They anticipated that wind power projects would only vary half a cent up or down from the average. In reality, the variance is 1.5 cents either way, so the range extends from some projects making power at 2 cents a kilowatt hour to other projects that make electricity at 5 cents a kilowatt hour.

It’s worth noting that all of the costs cited here for wind are easily competitive in the wholesale electricity market. For example, in 2005, a good year for wind and bad one for natural gas prices, wind was off-the-charts cheap:

It’s a pretty conventional argument: 1) “no single technology will provide all of the answers,” which is obviously true, and 2) large-scale storage options are necessary for grid-integration.

“[R]emember that wind and solar are intermittent energy sources. The sun isn’t always shining, and the wind isn’t always blowing,” Chu wrote. “Without technological breakthroughs in efficient, large scale energy storage, it will be difficult to rely on intermittent renewables for much more than 20-30 percent of our electricity.”

Which is true enough. I have a lot of hope for compressed air storage, but that’s going to take time to understand and scale up.

There was one part of Chu’s argument that I don’t like, though. He relies on a bogus forecast from the Energy Information Administration about projected future demand to make the case for nuclear.

“While we are working at hard as we can to promote energy efficiency in every part sector of America, it is likely that our energy demand will continue to rise. In fact, the Energy Information Administration projects an almost 20 percent increase in overall energy demand and over 30 percent increase in electricity demand over the next 25 years under current laws.”

As I’ve reviewed here over and over, long-run energy forecasts have been terrifically bad. With some notable exceptions, they have tended to project way too much energy usage over the last 50 years. Yet Chu, like so many others, treats them with far more respect than they deserve.

Using EIA forecasts to justify policy is a convenient way to sidestep having to make a real argument about whether it makes sense to let our energy usage grow or not. Instead, American energy usage growth is seen as inevitable, so that key point doesn’t need to be argued. This is an old and kind of dirty trick. Just check out Philip Sporn using just about the same one in 1971.

And the thing is: you don’t need EIA forecasts to make the argument that Chu does. The need to cut our carbon emissions is a good enough reason to look at nuclear power again, particularly new types of reactors. Compared with burning coal, which is how we generate just under 50% of American electricity, nuclear power starts to look pretty good, even given the problems that people have noted for years.

But no one really knows how much a new plant will cost or how long it will take to build one. And that’s using light-water reactors, which pro-nuclear guys like MIT’s Ernie Moniz say will be the only option for the coming couple of decades. If it takes the high-end estimate of 100 months to build a new nuclear plant — more than eight years — than it may not be possible to build enough nukes to, as Chu puts it, “make a serious dent in carbon dioxide emissions.”

This visual representation of bad forecasting is the from 2002 paper, “WHAT CAN HISTORY TEACH US? A Retrospective Examination of Long-Term Energy Forecasts for the United States” [pdf]. As you can see, every estimate can in too high with the exception Amory Lovins, who is not on this chart.

Charts like this are important because they remind us that supposed foresight should not foreclose debate about energy choices. Forecasts can be a particularly effective political cudgel and it’s the memory of these projections that should keep us from taking anyone seriously who claims too much certainty about the future.

Which is great, but where does that leave us?

I wrote the first three-quarters of a chapter happily blasting away at the energy forecasting mistakes of the past. But where does that leave us? We still need effective ways of thinking about the future. I’d like to see more descriptive scenarios, what a group of 70s researchers said lay “somewhere between a forecast and a fantasy.” And I’d like to see more normative scenarios that say, “This is the future I’d like to see more America, and this is the energy system that can help us get there.”

In the quest to find the limits that matter, be they geophysical, political, financial, or otherwise, we need more people willing to come out from behind their models’ baked-in assumptions and tell Americans why we need more X or less Y.

BUT, solar and wind technologies have been chasing moving regulatory and cost targets while dealing with very inconsistent government research, development, and deployment support. Up through 2000, when this study was conducted, renewable energy had actually hit the cost targets used to justify governmental support. (Government support, however, has rarely kept up its side of the bargain.)

“The most important measure of success would seem to us to be the cost of electricity generated from renewable technologies compared with the expectations that served as the justification for public-sector support,” the study’s authors conclude. “According to this measure, renewable technologies have met the goals set for them, and could be considered an important component of an ongoing movement toward sound energy policy.”

This could mean that the technology itself has actually performed better than people expected because the cost has come down without the benefits that come with scale. (And forecasters were undoubtedly banking on scale to help.)

Here’s the abstract.

This study provides an evaluation of the performance of five renewable energy technologies used to generate electricity: biomass, geothermal, solar photovoltaics, solar thermal, and wind.  We compared the actual performance of these technologies against stated projections that helped shape public policy goals over the last three decades.  Our findings document a significant difference between the success of renewable technologies in penetrating the U.S. electricity generation market and in meeting cost-related goals, when compared with historic projections.  In general, renewable technologies have failed to meet expectations with respect to market penetration.  They have succeeded, however, in meeting or exceeding expectations with respect to their cost.  To a significant degree, the difference in performance in meeting projections of penetration and cost stem from the declining price of conventional generation, which constitutes a moving baseline against which renewable technologies have had to compete.

Joel Darmstader¹, an RFF analyst, had prepared a packet of information on energy consumption trends and patterns in which he predicted energy usage of 190 quads in the year 2000. Sporn distanced himself from that number.

“In my opinion, the projections of U.S. energy consumption for the end of this century cited by Joel Darmstadter overstate the likely rate of growth,” he said. “I believe the requirements are likely to be 20-25 percent less—152,000 trillion BTU of total energy input…”

At the time, the United States was using 68 quadrillion BTU, or in Wonk, 68 quads (the term denoting quadrillion BTUs and beloved by energy nerds).

But, Sporn noted, “my lower projections will not satisfy those who believe with David Inglis” that American energy consumption increases could be held to “60 percent instead of 100 percent.” A sixty percent increase would yield year 2000 energy consumption of about 109 quads.

“I doubt that it is in the best interests of the American people to adopt Professor Inglis’s program for limiting energy growth,” Sporn warned. “Since every projection of population, GNP, and industrial production indicates major increases in total energy use, electric energy use, and per capita use, I shall discuss the impacts of environmental costs on the program of expansion and growth as I visualize that growth.”

In other words, Sporn dismissed Inglis. He sideswiped him, played the it’ll ruin the economy card, and told the audience he was going to talk about something else entirely. Inglis wacky ideas about energy weren’t even worth discussing. Who could blame him? Every projection called for major increases because from 1900 through 1970, average annual energy growth was over 3%.

So, who was right? The energy analyst, the utility executive, or the physicist-turned-professor activist?

As it turns out, Inglis was. By a mile. And they all — even Inglis the Bold — missed high. American energy usage is 2000 was actually a mere 99 quads. Last I checked, no one was complaining about how the 90s were a period of hairshirted energy restrictions (Ford Explorer, McMansion) or one of economic destruction. Indeed, the time, particularly the latter half of the decade, is notable precisely for the nation’s strong economic growth.

What’s the moral of this story?

We’re terrifically bad at imagining our energy futures. And we’re terrifically good at imagining the knock-on effects of projections made decades out. This or that is going to happen with supply or demand, ERGO, what I want is the path that must be taken.

It’s great for getting what you want in the present, but it’s a terrible way to peer, however murkily, into the future.

Inglis got much closer to reality because he didn’t just believe what the curves — every projection! — told him. He considered the social circumstances underlying those realities and came up with a nuanced and forward-looking picture of what was attainable in the world. Demand could be altered and shaved and shaped. Of course, he also got lucky as the sociopolitically precipitated energy crises helped reset demand expectations. But the American people proactively helped, too, by preventing utilities from executing the grow-and-build strategies that required energy demand growth.

1. I don’t point out Darmstadter’s forecasting miss here to impugn his work generally. He strikes me as a careful and thorough observer of energy issues. RFF, as an organization, has often made interesting and groundbreaking analyses, too. Lee Erickson, in reviewing energy forecasting methodologies in the 60s for the book Energy and Human Welfare noted that RFF analysts were the first to “explicitly consider factors such as the impact of insulation improvement on space heating requirements and of the rates of appliance saturation on residential electricity demand.” Which seems very sensible.

Here’s former Texas A&M geologist, Earl Cook,  and his full quote from a 1971 article in Scientific American.

The automobile engine and its present fuel simply cannot be cleaned up sufficiently to make it an acceptable urban citizen. It seems clear that the internal-combustion engine will be banned from the central city by the year 2000; it should probably be banned right now. Because our cities are shaped for automobiles, not for mass transit, we shall have to develop battery-powered or flywheel-powered cars and taxis for inner-city transport.

Well, we’re not quite there yet, but how about the kindler, gentler, 21st-century version: congestion taxes?

(This is part of my on-going forecasts project, which details just how bad we are at predicting what the world is going to look like.)

Photo: Tom Baker.

How much would technical learning drive down manufacturing costs? How much would economies of scale help? The list of questions was long and the list of answers was short.

So, people working on wind had to make some assumptions. One struck me as particularly interesting. It’s perhaps the simplest heuristic that I’ve ever seen. In a publication called “Wind Energy Developments in the 20th Century,” NASA chose to estimate the ultimate cost of wind power by weight:

To estimate what the cost of a wind turbine might eventually be in production, it was assumed that wind turbines could be fabricated, assembled, and installed for a cost of $2 to $3 per pound. This is a rather simplistic way to estimate the cost of a mature product, but available data show that many machinery items such as large tractors, power shovels, and steam turbines are fabricated, assembled, and distributed for $2 to $3 per pound.

I haven’t found many references to this method of cost estimation. Anyon know if NASA was right about this? Does anyone still estimate ultimate cost this way?

The top graph shows constant change in the energy supply over the last one hundred years. The bottom graph shows no change in energy supply over the next 20+ years.

“While the Nation’s energy history is one of large-scale change as new forms of energy were developed,” the DOE writes, “the outlook for the next couple of decades (assuming current laws, regulations, and policies) is for continued reliance on fossil fuels (with coal growing faster than liquid fuels and natural gas), modest growth in hydroelectric power and nuclear electric power; and a doubling of non-hydroelectric renewable energy by 2030.”

Let’s paraphrase: even though the history of the nation’s energy usage shows constant change, we’re predicting no further change. I guess when you’ve been burned by past forecasts, you get a little gunshy. Still, is it really the most credible scenario that almost nothing would change in our nation’s energy mix?

In a 1951 report for the Atomic Energy Commission, Palmer C. Putnam, inventor of the one of the world’s largest wind turbines and the DUKW amphibious transport vehicle, made some seemingly reasonable predictions about the the world energy system.

He noted the world’s “illiterate subsistence-farming populations” were “in demographic transition” to urbanized, higher-energy lifestyles. Meanwhile, “everything we know suggests that world population will double and may treble,” he wrote. (Perhaps you’ve heard similar sentiment now about a certain country in the Far East?)

More people with more more money meant that humanity would be using more energy, he reasoned. A lot more energy. The outcome of his thinking was the series of curves at the top of the post. Using different population estimates and energy demand growth rates, he came up with what he called the maximum plausible energy usage for from 1950 to 2000. The growth rates for energy demand he used don’t sound big — 3 to 5 percent per year — but what comes out of these accelerating scenarios is just astonishing.

Palmer Putnam predicted that in the year 2000, the maximum plausible energy usage for the world was 45,000 exajoules (or 43 x 10¹⁸ BTU). The curve predicting the lowest energy usage in 2000 showed that people would be using 32,000 exajoules (or 30 x 10¹⁸ BTU) of energy per year.

In reality, human energy consumption in the year 2000 totaled a mere 500 exajoules.

That is to say, to meet Putnam’s estimates, we’d have to rebuild every bit of our energy system 60 times over to meet his “minimum maximum.” Not only that, he underpredicted world population substantially, so his per capita energy demand was incredibly overblown.

What’s fascinating about energy demand scenarios like Putnam’s is that they seemed quite reasonable and buttoned up even. They seemed downright hard-headedly realistic. Except that they were totally crazy.

No fossil or renewable sources of energy could ever meet such a demand. Even taking the sunniest estimates, Putnam’s imaginary world energy system would burn through America’s coal reserves every few months! To make any kind of sense, Putnam needed not only massive amounts of nuclear power plants, but breeder reactors or fusion plants, which wouldn’t need limited uranium stocks. Never mind that finding enough metal and other materials to build such a fleet of plants would probably be impossible. Even assuming all materials could be found, the sheer scale of energy deployment required by Putnam’s energy estimates is astonishing.

“Perhaps it would be less costly, for example, to modify the pattern of the energy system so that nuclear fuels might bear half to three-fourths of the load,” he wrote. “Blanket electrification would be one way. Could we electrify most of the railroads and much of industry? Could we run overhead power lines along main toll roads and redsign trucks, buses, and cars for electric operation, relying on batteries for off-highway travel?”

Luckily, what Putnam saw coming never came to pass, but it’s important to remember that he and his estimates helped created the context in which energy planners and engineers worked. If this is where they thought the world was going, they would naturally favor huge power sources, not efficiency and renewable energy.

Steve Cohn has called this tendency the “technological aesthetics of mega-builders.” And Putnam was one of the great artists, his curves a masterpiece of the form.

One of the most optimistic predictions came from Thornton Bradshaw, president of Atlantic Richfield, who thought that the U.S. could reduce its dependence on foreign oil from 18% of total energy consumption now “to perhaps as low as 15% by 1980 and possibly 10% to 13% by 1985.” Most other speakers, including Sawhill, guessed that the U.S. would be importing 25% of its oil eleven years from now, v. about one third early this year.

In a TIME article entitled, “Project Realism,” a play on Nixon’s half-hearted Project Independence, we find these forecast gems. Thornton Bradshaw is admirably close to the money about U.S. oil imports as a percentage of U.S. energy consumption. In fact, in 1985, oil imports represented 14% of total consumption which is close to his range.

As for the other guys, 11 years after they made their prediction, the U.S. was dependent on imports for 27% of its petroleum needs, so they were pretty close, too. It’s hard to know whether they banked on the brief Alaskan oil boom, which kept down U.S. oil imports until the late 80s, or were just lucky.

By 1990, we were importing 42% of our oil.  Now, that number is 60%.

[Source: Time Magazine]