This is a guest essay from Amory B. Lovins and Imran Sheikh of the Rocky Mountain Institute. It is part two of a series; see part one here.


Part two of David Bradish’s critical look at “The Nuclear Illusion” (PDF) raises two additional issues to which we respond here. As in his first critique, it appears that, unable to rebut and hence unwilling to address our paper’s data and logic, Mr. Bradish must content himself with trying to manufacture an illusion of confusion.

Does RMI’s data fit their definition [of micropower]?

Grist thanks its sponsors. Become one.

Yes, precisely; it just doesn’t fit various other definitions that Mr. Bradish has invented on his own. We clearly defines micropower (an Economist magazine term) thus at pp. 11-12:

1. onsite generation of electricity (at the customer, not at a remote utility plant) — usually cogeneration of electricity plus recovered waste heat (outside the U.S. this is usually called CHP — combined-heat-and-power): this is about half gas-fired, and saves at least half the carbon and much of the cost of the separate power plants and boilers it displaces; [and] 2. distributed renewables — all renewable power sources except big hydro plants, which are defined here as dams larger than 10 megawatts (MW).

Mr. Bradish arbitrarily and wrongly assumes “that the size of ‘micropower’ plants is 10 MW or less,” then claims this is our definition and contradicts our data. It’s not and it doesn’t. Our 10 MW limit applies only to small hydro, distinguishing it from big hydro using the most conservative criterion. Any power source except small hydro can be larger than 10 MW but still meet our micropower definition: WADE’s onsite-fueled-generator definition, which we’ve adopted, includes onsite units up to somewhat over 180 MWe for gas turbines (though few actual units are over 120 MWe) and up to 60 MWe for engines, as well as onsite (nearly always cogenerating) steam turbines of any size if they’re in China and India; however, WADE’s database excludes steam turbines elsewhere, and all units below 1 MWe.

Mr. Bradish complains that we and WADE don’t specify onsite generator units’ size distribution. The size distribution for 2006 additions can be found by looking at the Diesel and Gas Turbine Worldwide: Power Generation Order Survey (PDF). WADE details its assumptions, which we adopted, about what fraction of these units provide onsite generation and hence fit our micropower definition. To give a rough idea of the size distribution of new non-biomass decentralized generation capacity additions in 2006, after separating out peaking and standby units, about 88 percent of new onsite diesel capacity came from units less than 10 MW in size. For onsite continuously-operating gas turbines, 10 percent of new capacity came from units less than 10 MW, 74 percent came from units between 10 and 60 MW, and 17 percent came from units over 60 MW.

Mr. Bradish adds further confusion by injecting his personal opinion that “micropower” shouldn’t include what he calls “centralized renewables,” like windfarms. But we define micropower to include all renewables except big hydro, consistent with Economist usage and the terminology long established in the field. For reasons explained in our 2002 Economist book of the year Small Is Profitable: The Hidden Economic Benefits of Making Electrical Resources the Right Size, we think the key distinctions between micropower and central stations turn on their ability to capture many of the 207 “distributed benefits” catalogued in that definitive work. A windfarm totaling hundreds of megawatts still captures the deployment speed, modularity, low financial risk, and economies of scale that come from mass production of its ~1-3 MW turbines. The windfarm lacks the large unit scale, long lead times, and high financial risks of a GW-scale thermal station. Mr. Bradish thinks we should count only the ~5 percent of wind machines that are relatively small and onsite, but we clearly counted all windpower, as he notes from our posted database. Had we meant to count only rooftop solar cells and other small-scale generators resources sited at the user, we would have said so.

Grist thanks its sponsors. Become one.

In short, we chose and defined our terms carefully, presented data consistent with our definitions, and cannot be responsible for Mr. Bradish’s pretense that we meant something different and should have said so. He’s welcome to compile his own data using his own idiosyncratic definitions, but he shouldn’t blame us for not adopting them.

Mr. Bradish offers a graph from a proprietary Ventyx/Global Energy Decisions database to which we don’t have access. We therefore can’t tell whether smaller, cogenerating, and non-utility units were fully included. We doubt they were, for two reasons. First, when we finally reached Ventyx on June 13, 2008, their specialist told us that their database fully includes only units of at least 25 MW, and omits smaller units to a degree he couldn’t specify. (We requested further details of what is or isn’t included in their database — in size, type, and ownership — but haven’t yet heard back.) Second, EIA data on U.S. net-summer-capacity additions (May 2008) show renewables added respectively in 2004, 2005, 2006, and 2007 about 14.5, 15.1, 8.2, and 12.6 GWe, but Mr. Bradish’s graph shows respective values, read off the graph, of only about <1, 2, 3, and 4 GWe. (We can't similarly check cogeneration additions because the Ventyx database doesn't break them out.) All these kinds of units are included, however, in the public-source data we analyzed in detail at pp. 23-34 of Small Is Profitable, to which we invite his attention.

Mr. Bradish’s assumption that micropower can’t be growing quickly because, he states, average U.S. plant size was ~150 MWe in 2007 appears to reflect (a) major omissions of small units from the Ventyx database, and (b) the United States’ lag in adopting micropower (~6 percent of ~2007 electricity, vs. ~16 percent globally and >50 percent for the highest adopter, Denmark). The high micropower adoption demonstrated by our global market data might require his claimed ~20-40 MWe average unit size if micropower had to be <10 MWe per unit, but under our definition, it doesn't.

Finally, Mr. Bradish claims our data don’t show thriving micropower in the rest of the world. Huh? Our empirical data document that micropower added ~43-58 GWe worldwide in 2006, vs. 1.44 GW net for nuclear (more than all of it from uprating old units). Micropower now provides a sixth of the world’s total electricity, about a third of the world’s new electricity, and from one-sixth to more than half of all electricity in a dozen industrial countries. Renewables other than big hydro attracted $71 billion of private risk capital in 2007; nuclear, zero. Mr. Bradish does not deny or even address any of these data; he simply tries to distract readers’ attention from them.

Big plants yield greater efficiencies and economies of scale than small plants.

Mr. Bradish’s claim (borrowed from Peter Huber and Mark Mills’s book The Bottomless Well) is simply shorthand for the high-temperature cycles with higher Carnot efficiencies that can be achieved in larger boilers with higher volume-to-surface ratios. This is true as far as it goes, but is grossly incomplete, reflecting a primitive understanding of scale economics:

  • An efficient plant discarding 2 GWt of waste heat — too much to use in most sites — has a lower fuel-to-useful-work efficiency and a lower economic efficiency than a small cogenerator matched to its thermal and electrical loads and achieving roughly twice the big plant’s system efficiency (our record is ~92 percent and the state of the art is probably ~93 percent, vs. Mr. Bradish’s cited 35-39 percent for advanced nuclear plants or ~50-60 percent for modern combined-cycle plants).
  • Unsuitability for cogeneration in most sites is only one of a huge array of diseconomies that offset the well-known Huber/Mills economies of scale. The history of how utilities that at first sought only economies of scale realized that diseconomies of scale were often more important is on pp. 11-34 of Small Is Profitable. Numerous phenomena come into play, such as the lower uptime, longer lead time, higher reserve margin and spinning reserve requirement, and higher financial risk of big units. Small Is Profitable exhaustively details both economies and diseconomies of scale. Mr. Bradish seems to think that only heat-to-electric efficiency matters, but owners know better.
  • The claim that big stations are more economical than small ones is flat wrong. Small Is Profitable, in over 400 pages of detailed analysis,documents 207 reasons how units the right size for the task can capture “distributed benefits” that often increase value by roughly an order of magnitude — enough to flip any investment decision.
  • Interestingly, and contrary to Mr. Bradish’s claim, it’s not even true that big gas turbines, the mainstay of today’s combined-cycle plants, are more efficient than small ones. Tom Casten’s Electricity Journal article in Dec. 1995 showed that at least at that time, the highest simple-cycle efficiency came from an aeroderivative 40 MWe unit (GE’s LM6000), not from the largest units at 250 MW, and that “many offerings below 50 MW compare well with 250 MW machines,” especially counting cogeneration potential and avoided high-voltage step-up. Of course, in the right applications, such as wind turbines, economies of unit scale can and do exist, but that does not contradict our thesis nor support Mr. Bradish’s.

Small plants are too slow to build to achieve a desired total capacity.

Mr. Bradish starts by wrongly assuming a maximum 10 MWe unit size, then ignores our empirical data showing that micropower is already achieving very large total capacities, far faster than GW-scale thermal plants can or do. For example, in 2006, global micropower added 30-40x more capacity than nuclear did.

His handwaving claim that building many small units quickly is somehow “not practical” flies in the face of the extensive data we present on what investors are actually buying and operators are actually installing and operating. Whatever exists is possible: Micropower is empirically outpacing nuclear, in capacity added per year, by factors of ten. Of course this ratio is smaller for electricity output because of differences of capacity factor. But nuclear remains a bit player in today’s global power market, where micropower (not to mention efficient end-use) is starting to put a dent in sales of all central thermal plants — nuclear, coal, gas-combined-cycle, and big hydro. Thus the U.S. in 2007 added more wind capacity than it has added coal capacity in the past five years combined. The U.S. or China or Spain each added more wind capacity in 2007 than the world added nuclear capacity. Utilities and investors do not appear to share Mr. Bradish’s theological commitment to the impracticality of micropower and the manifest virtue of central plants.

Contrary to what RMI believes, there is no one-size fits all solution.

Neither in the articles at issue nor elsewhere have we ever proposed such a solution. Rather, we suggest the right size for the job. Most jobs are small. For example, in 1993 (Small Is Profitable, p. 36), 75 percent of U.S. households had average loads <1.5 kW, and 75 percent of commercial buildings had average loads <12 kW. (A typical U.S. house uses, on average, only about one-twentieth as much electric power as the solar flux falling on that house.) Thus a single 1.4 GW generating unit could serve nearly a million typical households or more than 100,000 typical commercial buildings in the lower three quartiles of average usage. That customary practice no longer makes economic sense.

To be sure, a few very large industrial facilities do use ~10^9 to 10^10 W each, and as one of us (ABL) wrote in the 1970s, it would be just as silly to run an aluminum smelter on small wind turbines as it is to heat houses with a fast breeder reactor. But there is no technical or economic rationale for the many-orders-of-magnitude gap between gigawatt unit scale and the scale of the end-use devices important in our daily lives (typically 10^-1 to 10^3 W) or of our living and working units (usually 10^3 to 10^5 W). That extreme mismatch, as Small Is Profitable shows in detail, wastes money and energy. We hope Mr. Bradish will study that analysis before further promoting his “one-size-fits-all” solution.

Mr. Bradish has posted part three of his critique, claiming that RMI has overlooked Jevons Paradox, which undoes and reverses the intended energy savings from more efficient end-use. We have rebutted this invalid claim in a response to Mr. Bradish’s cited primary source — an article by Robert Bryce in his newsletter. Completion of our response was delayed by travel, but we expect to finish it shortly, and will then post it on RMI’s website, in this blog, and (Mr. Bryce has assured us) on his site.

Meanwhile, readers should know that the claimed “rebound” effect — phenomena that make net energy savings smaller than gross savings — is real but generally very small, and has no material effect on our conclusions. This is firmly established in the empirical literature, and is well-known to knowledgeable energy economists but evidently not to Mr. Bryce, Mr. Bradish, or the theory’s current standard-bearers, Dr. Peter Huber and Mr. Mark Mills. A brief introduction to some basic concepts is on Wikipedia.

We will address Mr. Bradish’s forthcoming posts on “nuclear and grid reliability” and “costs” as they appear.