Does a Big Economy Need Big Power Plants? A Guest Post


Amory B. Lovins is the energy maven’s energy maven, viewed variously as a visionary or a heretic in his assessments of how the U.S. and the world should be generating and using energy. More specifically, he is the chairman and chief scientist at the Rocky Mountain Institute, a man who has won many awards, written many books, and, as if that weren’t enough, was a fan favorite for Energy Secretary when we asked blog readers a few months ago to give incoming President Obama some advice.

Lovins has written a guest post for us today, which I am guessing that everyone who cares about energy will find instructive in one way or another. It is especially interesting in light of forward-looking projects like this one about battery-exchange stations for electric cars — for as eager as we may be to wean ourselves from oil, it’s worth remembering that all that newly-demanded electricity doesn’t grow on trees.

Does a Big Economy Need Big Power Plants?
By Amory B. Lovins
A Guest Post

If I told you, “Many people need computing services, so we’d better build more mainframe computer centers where you can come run your computing task,” you’d probably reply, “We did that in the 1960’s, but now we use networked PC’s.” Or if I said, “Many people make phone calls, so we’d better build more big telephone exchanges full of relays and copper wires,” you’d exclaim, “Where have you been? We use distributed packet-switching.”

Yet if I said, “Many people need to run lights and motors, Wii’s, and air conditioners, so we’d better build more giant power plants,” you’d probably say, “Of course! That’s the only way to power America.”

Thermal power stations burn fuel or fission atoms to boil water to turn turbines that spin generators, making 92 percent of U.S. electricity. Over a century, local combined-heat-and-power plants serving neighborhoods evolved into huge, remote, electricity-only generators serving whole regions. Electrons were dispatched hundreds of miles from central stations to dispersed users through a grid that the National Academy of Engineering ranked as its profession’s greatest achievement of the 20th century.

This evolution made sense at first, because power stations were costlier and less reliable than the grid, so by backing each other up through the grid and melding customers’ diverse loads, they could save capacity and achieve reliability. But these assumptions have reversed: central thermal power plants now cost less than the grid, and are so reliable that about 98 percent to 99 percent of all power failures originate in the grid. Thus the original architecture is raising, not lowering, costs and failure rates: cheap and reliable power must now be made at or near customers.

“Central thermal stations have become like Victorian steam locomotives: magnificent technological achievements that served us well until something better came along.”

Power plants also got irrationally big, upwards of a million kilowatts. Buildings use about 70 percent of U.S. electricity, but three-fourths of residential and commercial customers use no more than 1.5 and 12 average kilowatts respectively. Resources better matched to the kilowatt scale of most customers’ needs, or to the tens-of-thousands-of-kilowatts scale of typical distribution substations, or to an intermediate “microgrid” scale, actually offer 207 hidden economic advantages over the giant plants. These “distributed benefits” often boost economic value by about tenfold. The biggest come from financial economics: for example, small, fast, modular units are less risky to build than big, slow, lumpy ones, and renewable energy sources avoid the risks of volatile fuel prices. Moreover, a diversified portfolio of many small, distributed units can be more reliable than a few big units.

Bigger power plants’ hoped-for economies of scale were overwhelmed by diseconomies of scale. Central thermal power plants stopped getting more efficient in the 1960’s, bigger in the 1970’s, cheaper in the 1980’s, and bought in the 1990’s. Smaller units offered greater economies from mass production than big ones could gain through unit size. In the 1990’s, the cost differences between giant nuclear plants — gigantism’s last gasp — and railcar-deliverable, combined-cycle, gas-fired plants derived from mass-produced aircraft engines, created political stresses that drove the restructuring of the utility industry.

Meanwhile, generators thousands or tens of thousands of times smaller — microturbines, solar cells, fuel cells, wind turbines — started to become serious competitors, often enabled by IT and telecoms. The restructured industry exposed previously sheltered power-plant builders to brutal market discipline. Competition from a swarm of smaller electrical sources and savings created financial risks far beyond the capital markets’ appetite. Moreover, the 2008 Defense Science Board report “More Fight, Less Fuel” advised U.S. military bases to make their own power onsite, preferably from renewables, because the grid is vulnerable to long and vast disruptions.

Big thermal plants’ disappointing cost, efficiency, risk, and reliability were leading their orders to collapse even before restructuring began to create new market entrants, unbundled prices, and increased opportunities for competition at all scales. By now, the world is shifting decisively to “micropower” — The Economist‘s term for cogeneration (making electricity and useful heat together in factories or buildings) plus renewables (except big hydroelectric dams).

The U.S. lags with only about 6 percent micropower: its special rules favor incumbents and gigantism. Yet micropower provides from one-sixth to more than half of all electricity in a dozen other industrial countries. Micropower in 2006 (the last full data available) delivered a sixth of the world’s total electricity (more than nuclear power) and a third of the world’s new electricity. Micropower plus “negawatts” — electricity saved by more efficient or timely use — now provide upwards of half the world’s new electrical services. The supposedly indispensable central thermal plants provide only the minority, because they cost too much and bear too much financial risk to win much private investment, whereas distributed renewables got $91 billion of new private capital in 2007 alone. Collapsed capital markets now make giant projects even more unfinanceable, favoring lower-financial-risk granular projects even more.

In short, many, even most, new generating units in competitive market economies have already shifted from the million-kilowatt scale of the 1980’s to the hundredfold-smaller scale that prevailed in the 1940’s. Even more radical decentralization, all the way to customers’ kilowatt scale (prevalent in and before the 1920’s), is rapidly emerging and may prove even more beneficial, especially if its control intelligence becomes distributed too.

Global competition between big and small plants is turning into a rout. In 2006, nuclear power worldwide added 1.44 billion watts (about one big reactor’s worth) of capacity — more than all of it from uprating old units, since retirements exceeded additions. But that was less capacity than photovoltaics (solar cells) added in 2006, or a tenth what windpower added, or 2.5 percent to 3 percent of what micropower added. China’s nuclear program, the world’s most ambitious, achieved one-seventh the capacity of its distributed renewable capacity and grew one-seventh as fast. In 2007, the U.S., Spain, and China each added more wind capacity than the world added nuclear capacity, and the U.S. added more wind capacity than it added coal-fired capacity during 2003 to 2007 inclusive.

What part of this story does anyone who takes markets seriously not understand? Central thermal stations have become like Victorian steam locomotives: magnificent technological achievements that served us well until something better came along. When today’s billion-watt, multi-billion-dollar plants retire, we won’t replace them with more of the same. I’m already experiencing a whiff of prenostalgia.


And if I told you "Many people need healthcare, so we better create a centralized full-service national network, so that when people get sick they can go apply to a federally funded doctor database for immediate care" you'd tell me.....

Eric M. Jones

Distributed power is a great idea. Assuming we have a "smart grid" to share it, so much the better.

The big problem is that coupling a large number of small power "stations" leads to a much higher cost per kilowatt hour.

This fact drives engineering decisions.


"Central thermal power plants stopped getting more efficient in the 1960's, bigger in the 1970's, cheaper in the 1980's, and bought in the 1990's."

This is, perhaps, the best written sentence ever to appear on Freakonomics. Bravo.


the lags in nuclear development make your comparisons of recent installed capacity disingenuous.

plus let's make sure to add all the costs of intermittancy to those solar and wind facilities.

Chris Dudley

There are some very good points here. On reliability, there is a core reliability to distributed generation in the face of natural disasters that we don't always appreciate because we don't have enough of it. In a disaster that disrupts grid power for several weeks, if there is a home in walking distance with solar power, you could perhaps keep your medicine refrigerated.

That kind of reliability is not the same thing as having the lights come on when you turn on the switch 99% of the time but it can be a very important kind of reliability nevertheless.

The most expensive kind of central generation is nuclear power. Lovins and Sheikh have looked at the issue here:

Because nuclear power is the least competitive of a set of uncompetitive technologies, we can be pretty darn sure that the tax payers will be left holding the bag as companies like Constellation Energy default in federally guaranteed construction loans.

One thing especially concerning new nuclear power is the long planning horizon for a plant. Fifteen years to build is added to sixty years to run which is added to 20 years to cool off enough for decommissioning followed by another 20 years to decommission. We must plan well into the next century with this kind of plant. Yet, the places we are considering for new nuclear power plants are often a sea level or very close: Clavert Cliffs of South Texas for example. We know that these places will face substantial sea level rise on a century time scale which may well force a shortening of the operating lifetime of the plants, making them even more expensive. It would seem that we need to consider climate in nuclear licensing decisions where we would not for other technologies. What, for example are the projected cooling water resources over the rest of the century? It would be foolish to build where drought might be expected to make power production impossible as happened recently in Alabama.

Co-gen is not the exclusive domain of combustion. Solar shows a great deal of promise to both generate power an usable heat together:

And, there may be a time when we start to produce fuels from wind power in a manner that generates heat that can be used for buildings. This is not co-gen but similar to it. A small scale is described here:



Dear Eric;

great idea is not enough- "an idea must be correct" for the purpose of accomplishing worthwhile matter. that;s where the buck stops. so that's why am happy to share `mine' windmill idea with friends. The path has become clear and obvious.

So as far as I can tell- no higher cost- From a scientifically sound-- engineering standpoint mutually `shared' benefit..

There is still -- as far applications-almost wide open field and almost untouched toppings (with a few exceptions including fairchild). if you are having difficulty- just listen to the sound, the physical and other sightings, the news- .

So as far as certain q's and a's- it's your score.

Jason Goodman

In the early 20th century, "micropower" (the Edison DC system) didn't lose to "macropower" (the Tesla/Westinghouse AC system) because of reliability or construction costs: it lost on the grounds of efficiency. Electrical transmission losses with Edison's DC system were so high that he had no *choice* but to use local micropower: by using high-voltage AC and transformers, Westinghouse was able to make transmission losses negligble.

Once your transmission system isn't leaking watts, the battle is fought between different power plants on a pure "megawatts out per dollar" basis. I would argue that the basic economics of this haven't changed for most of the century: big plants can run at higher temperatures, making them thermodynamically more efficient, and lose less energy from parasitic losses. It's more than economics, it's physics.

The author argues that improvements in micropower have made it cost-competitive to macropower, but doesn't present any good statistics to confirm this. It's certainly true that private companies (especially telecommunications, government, and hospitals) have been adopting micropower in a big way in the last couple of years, and electrical utilities are also installing lots of small gas plants. But they're not doing it because it's cheaper: they're doing it because these micropower gas plants switch on in a few seconds, allowing them to be used in emergency and unexpected-demand situations. Heat energy from coal costs 1/3 as much as heat energy from natural gas, so gas plants are not a good idea for long-term everyday use.

Now, renewable sources are another matter, but if the only consideration is whether smaller is cheaper, I'm not buying it unless I see a clearer cost analysis between micropower and macropower.


Jon S

You are assuming away all of the government grants for renewables and even more egregiously ignoring all of the bogus negative political pressure nuclear receives. Then you conclude by saying the market supports your argument?

Sorry, not convinced.

David Roberts

Bravo, Amory. Again.

What's particularly ironic about this is that the allegedly "pro-market" political party in the U.S. is absolutely and monomaniacly *obsessed* with nuclear power, which is perhaps the largest and most needy welfare recipient ever to bestride the nation's budget. Don't they value hard work and self-sufficiency? Or is that just for poor people?

Still, though it's futile to say so, this comment thread should not devolve into yet another quasi-religious argument over nukes. The Lovins point is broader:

Commodity prices are rising, almost across the board. Meanwhile, the cost of processor power -- intelligence -- is falling steeply.

The rational economic response is to *substitute intelligence for material* -- that is to say, substitute smart design and use for size and brute power. And indeed, that's what markets are doing, even against the many legal and regulatory barriers referred to in the piece.

In 10-15 years, this debate will seem quaint.



Cogeneration offers good savings. There are even hom e scaled natural gas cogenerators that that scavenge the waste heat for air conditioning and water heating. they use the prime mover mainly to turn the compressor and secondarily to generate electricity that is sold back to the grid.

But, distributed maintenance is an issue.

And with anything but the smallst units, so is NIMBYism and zoning.

I beleive Amory is fundamentally right, but glossing over many problems yet to be resolved, not he least of which is Jevon's Paradox.


Here's the problem with Nuclear Power: It's a PERFECT example of a Taleb Distribution.

One terrorist attack or serious accident WIPES OUT ALL Previous Profit.

We've just seen what happens with you ignore Black Swan's in the Stock Market. I'm voting against researching this using nuclear power.

Global Warming is another example. Water sources dry up, land becomes barren, forest fires, acidic oceans, high frequency storm are all Heavy-Loss, Catastrophic, Events.

William Tucker

Quite briefly, Lovins is drawing a false analogy between the miniaturization and distribution of computing and telecom instruments and the production of energy. Computers and telephones can be miniaturized and distributed according to Moore's Law because they involve information. You can use less and less energy to store each bit. For that reason you can have as much computing power on your desktop today as Univac had in an entire room in the 1960s. Computers can be distributed because they have become so powerful.

But things don't work that way with energy. A kilowatt is a kilowatt, whether it's generated in your backyard or at a power station. You can “distribute” generation anywhere you want but you still have use the same amount of fuel or wind or whatever. We could replace central thermal stations with gas turbines on every street corner, but the fuel is going to be expensive and produce a lot more carbon emissions, which is something Lovins conveniently overlooks.

The real irony, however, is his suggestion that wind fits this small-is-beautiful scenario. Sure wind is “distributed.” After all, you need 125 square miles of 45-story windmills to generate the same 1000 megawatts that can be generated in one square mile at a central thermal station. You've got to put them somewhere! And that's just their nameplate capacity. To produce 1000 MW of base load electricity, you'd need at least three or four 125-square-mile wind farms scattered at diverse locations around the country.

That's the reason Lovins himself has suggested covering all of North and South Dakota with wind farms. Al Gore matches him by asking for 1/5 of New Mexico, the fifth largest state, for solar collectors. On top of this, they want to rebuild the entire national grid to 765 kilovolts in order to ferry all this electricity from the remote areas where it's best generated to population centers. And Lovins calls 1000-MW power plants operating on the current transmission system “irrationally big!”

What Lovins never wants to acknowledge is the energy density of nuclear power. With nuclear, the energy produced from 500 square miles of windmills can be generated with a fuel assembly that would fit in the average living room. Why “distribute” all this generating capacity into big, ugly structures that litter the landscape and only work when the wind blows? Why not concentrate it all in one place? Then once every 18 months a single tractor-trailer can come in with a new set of fuel rods.

In one respect, though, Lovins may be right. Maybe we shouldn't be building nuclear reactors to 1500 MW. Hyperion, a New Mexico company, has invented an 80-MW mini-reactor the size of a gazebo that can power a town of 20,000. You could put it in someone's basement and no one would ever notice. While “alternate energy” has gotten more and more gigantic, nuclear is getting smaller and smaller.

Who would have thought it would be nuclear that is small and beautiful?


William Tucker

Can I add one word? In the third from last paragraph:

What Lovins never wants to acknowledge is the INCREDIBLE energy density of nuclear power.




The cost of building a solar burner is less than the cost of building a new coal burner. That makes solar steam less expensive than coal steam even when coal is delivered at no cost -- solar is cheaper than free lumps of coal.

Chris Dudley

Mike99 (#11),

Right now I think that we are looking at a possibly horrible product of your Taleb distribution and Murphy's Law: if we were to face a Price Anderson Act payout right now for a nuclear accident, I'm not sure that the federal government could keep up with its debt service obligations. One nuclear accident could make us Iceland.


Those who would like to know Lovin's (and the RMI's) current views of nuclear energy should be interested in a new report
Nuclear Power: Climate Fix or Folly?

"This 15-page January 2009 update and expansion of "Forget nuclear" in RMI's Spring 2008 Solutions Newsletter adds the latest data, expands the discussion of capital-cost escalation, and includes June 2008 cost comparisons by preeminent financial advisors Lazard. It summarizes why nuclear power cannot in principle deliver the climate-protection or energy-security and reliability benefits claimed for it. (January 2009)"

Rick Answer Analyst

Bigger is not better. Governments don't understand that yet eitiher. Interesting that during Enron's rule our local paper company brought in some generators and parked them outside the plant. It was cheaper than getting electricity through the grid. The other question comes up as to pollution from multiple sources versus handling it from one main location where it is easier to monitor.

The future will be different.

Sourendu Gupta

This makes for good reading, but links to original sources. would have been nice. They could, for example, allow me to check which of the statements about price advantages and competetiveness carry over to other countries.


Whenever solar and wind power comes up in discussions like this, and is contrasted with coal-burning, gas-burning, or nuclear power generation, I'm also amazed at the people who attempt to compare renewables with resource-depleting. I realize that solar power, for example, isn't terribly efficient just yet. But it's getting better every year, and eventually it'll be just fine in that regard... because the solar energy is already raining down on us, every day, just waiting to be used. Imagine if coal fell from the sky and had no negative effect on our atmosphere when burned? Wouldn't we be crazy not to use it? Instead, it comes out of the ground, takes millennia to form and kills us in all kinds of ways when we burn it. Sunlight? Not so much! Solar and wind power are just waiting to be used, and no matter how much we use today, there'll be just as much waiting for us tomorrow... and next year... and next century... and long after we've died off as a species.

Even if solar power costs more in the short term, the source of that power lasts until our sun burns out (by which time, if we're still around, we'll have much bigger issues to deal with than the loss of some solar power).

Are people really that stupid that they can't see how fundamentally different those two categories of energy sources are?


Carl Christopher

As always, Amory Lovins has very insightful and interesting ideas. I like to read his writings and listen to his speeches.

But Lovins has a terrible track record in predicting the future. He's been wrong for decades. I think he is wrong here as well.

Micropower makes a lot of sense in a lot of cases. But it's not taking over the world. Not now, and (I think) not in the future either.

Of course, I'm only a few years younger than the Prophet of Snowmass, and I've done no better at predicting the future. I've invested a fair amount of money in wind energy over the past two decades. I've lost it all.