In preparing for the future, it is very important to have a
wide range of options and to think in advance about how we are going to
react to the worst cases as well as the best.
The road to hell is paved with good intentions
Last week's column Steven Chu's Energy Miscalculations1 attempted to answer the question what is the plan? for dealing with potential shortfalls in the world's liquid fuels supply. The answer is let efficiency take care of it, at least in the foreseeable future 5, 10 or 15 years from now. Chu's reasoning is based on his miscalculation that we have between 10 and 40 years before oil & natural gas production, taken together, will peak and decline. Efficiency is supposed to double the time we have to find replacement fuels, so Chu has recast the problem to give himself the 20 to 80 years he requires to find a way to replace oil.
Chu's long range solution involves applications of synthetic biology to create 4th generation biofuels (Biopact, October 8, 2007). Last week I posed the questions is this a good solution? is it realistic? My critique should provide the answers.
You are probably asking what 4th generation biofuels are. Nobody actually knows yet. I can provide you with a conceptual overview and brief descriptions of the types of science problems that need to be solved. Beyond that there is only techno-optimism, a quasi-religious faith in the ability of cutting-edge science to solve our energy problems on the enormous scales required. Watch Charlie Rose's interview with Chu to get a feel for the energy secretary's faith. Here's a tidbit.
CHARLIE ROSE: Listening to you, what I’m struck with is the point -- is the connection between our energy future and our scientific know-how. The merger there, the convergence there is the essential link to get us out of this mess.
STEVEN CHU: I deeply believe that in my heart and soul, that science, when put to the task, can and will I think give us much better energy solutions.
[A transcript is also available.]
We are trying to accommodate exponential growth (and here) over long time-scales in domains like population and energy consumption. Dr. Chu promises that nifty science will provide solutions that allow us to keep the growth party going indefinitely. The math alone states that it will be impossible for us to achieve another doubling—13 billion people, 145 million barrels-per-day of oil—in the 21st century without experiencing major system breakdowns. No one in power has thought this through. I could simply say here at the outset that Chu's biofuels dream is also a pipe dream.
But where's the fun in that?3 Sooner or later—sooner would be better—we will have to acknowledge the inexorable mathematics and live more gently on our finite Earth.
A reader2 sent me a statement by Kenneth Boulding which I will use to put all that follows in context. Boulding was making an "individual statement" on pages 616-617 of the National Academy of Sciences study Energy in Transition, 1985-2010: Final Report of the Committee on Nuclear and Alternative Energy Systems. Remember as you read it that this was written in 1980 and it is now 19 years later.
... The great uncertainties here are in the area of the future of human knowledge, know-how, and skill. There is a nonexistence theorem about prediction in this area, in the sense that if we could predict what we are going to know at some time in the future, we would not have to wait, for we would know it now. It is not surprising, therefore, that the great technical changes have never been anticipated, neither the development of oil and gas, nor the automobile, nor the computer.
In preparing for the future, therefore, it is very important to have a wide range of options and to think in advance about how we are going to react to the worst cases as well as the best. The report does not quite do this. There is an underlying assumption throughout, for instance, that we will solve the problem of the development of large quantities of usable energy from constantly renewable sources, say, by 2010. Suppose, however, that in the next 50, 100, or 200 years we do not solve this problem; what then? It can hardly be doubted that there will be a deeply traumatic experience for the human race, which could well result in a catastrophe for which there is no historical parallel.
It is a fundamental principle that we cannot discover what is not there. For nearly 100 years, for instance, there have been very high payoffs for the discovery of a cheap, light, and capacious battery for storing electricity on a large scale; we have completely failed to solve this problem. It is very hard to prove that something is impossible, but this failure at least suggests that the problem is difficult. The trouble with all permanent or long-lasting sources of energy, like the sun or the earth’s internal heat, is that they are extremely diffuse and the cost of concentrating their energy may therefore be very high. Or with a bit of luck, it may not; we cannot be sure. To face a winding down of the extraordinary explosion of economic development that followed the rise of science and the discovery of fossil fuels would require extraordinary courage and sense of community on the part of the human race, which we could develop perhaps only under conditions of high perception of extreme challenge. I hope this may never have to take place, but it seems to me we cannot rule it out of our scenarios altogether.
Our failure to solve the problem of producing reasonable-cost liquid fuels at large scales from cellulosic feedstocks suggests that this problem is also very hard. Boulding also refers to the difficulty or impossibility of proving a thing's non-existence. I can not prove that alien civilizations don't exist, nor can I prove that none of Chu's synthetic biology experiments will work. I am allowed to speak in term of probabilities, however, just as Chu does.
4th Generation Biofuels
Figure 1 describes an envisioned 4th generation biofuels delivery system.
Figure 1 — A 4th generation biofuels conceptual architecture from Biopact (link above).
Figure 2 gives a conceptual overview of the Helios Project at the Lawrence Berkeley National Laboratory. Steven Chu is the former Berkeley Labs director.
Figure 2 — An overview of the Helios Project (pdf) from Berkeley Labs.
The multiple pathways shown in Figure 2 illustrate what Chu calls a "portfolio" approach to creating 4th generation biofuels. I prefer to describe it as a "scattershot" approach in which you randomly fire a lot of bullets in some general direction and hope you hit something.
In what general direction does Chu want to shoot his bullets? Chu wants nothing less than to alter the Earth's primary productivity, its energy flows, to achieve greater efficiency in the conversion of sunlight to chemical energy than Nature has after 3.5 billion years of evolution. What is primary production?
The general term "production" is the creation of new organic matter. When a crop of wheat grows, new organic matter is created by the process of photosynthesis, which converts light energy into energy stored in chemical bonds within plant tissue. This energy fuels the metabolic machinery of the plant...
Whether one measures the rate at which photosynthesis occurs, or the rate at which the individual plant increases in mass, one is concerned with primary production (definition: the synthesis and storage of organic molecules during the growth and reproduction of photosynthetic organisms). The core idea is that new chemical compounds and new plant tissue are produced. Over time, primary production results in the addition of new plant biomass to the system...
Is there an Upper Limit to Primary Production?
On average, plant gross primary production on earth is about 5.83 x 106 cal m-2 yr-1. This is about 0.06% of the amount of solar energy falling per square meter on the outer edge of the earth's atmosphere per year (defined as the solar constant...). After the costs of respiration, plant net primary production is reduced to ... about 0.05% of the solar constant. Note that this is the "average" efficiency, and in land plants this value can reach ~2-3% and in aquatic systems this value can reach ~1%. This relatively low efficiency of conversion of solar energy into energy in carbon compounds sets the overall amount of energy available to heterotrophs at all other trophic levels.
[Read the linked article for details.]
The mere fact that evolution has placed upper bounds on the efficiency of primary productivity in plants suggests that there are very deep reasons why this is so. Naively, plants with a higher conversion efficiency would effectively act like weedy species which out-compete their neighbors in ecosystems subject to human tampering. More efficient photosynthesis would thus be beneficial for plants. Nature might even have spurred an evolutionary arms race among plant species selected for more efficient photosynthesis, yet this did not occur. Natural (not human-influenced) ecosystems do not permit such species outside the limits described above. The situation is very complicated, not least because the Earth's Carbon Cycle is partly tied to photosynthesis rates (CO2 uptake) on land and in the oceans.
This suggests that tampering with plant productivity may be a grave mistake or impossible. I do grant the possibility that constraints on primary production were the result of an "accident" (mere chance) in the way life evolved on Earth. In this case plant engineering might well be possible. How are we to assess the likelihood of either position? We can't. We do know that 4th generation biofuels will require the creation of artificial ecosystems, i.e. systems which have been human-designed and -engineered for specific purposes.
Now we can interpret Figure 2. 4th generation biofuels might require microbes and plants which have been genetically altered to speed up or enhance photosynthesis. Such plants would take up more carbon and use solar energy more efficiently than Nature does. Alternatively, microbes or plants can be synthesized that use a more efficient human-designed photosynthesis. Or cellulose-degrading microbes found in nature would be altered—or again, synthesized—to allow us to break down lignin in corn stover or switchgrass more efficiently. Finally, more efficient photovoltaics (PV) that capture more of the sun's energy for useful work will drive the whole process. I have used to phrase "more efficient(ly)" four times in this paragraph—you get the idea.
Large-scale production of 4th generation biofuels is a form of geoengineering. We will plant energy crops on a land area of unknown size—this depends on the efficiency of the solar energy collection. (If no or only minor efficiency gains are achieved, there won't be enough land.) Then we will harvest those crops and transport them to biorefineries, where biomass will be converted to fuels as shown in Figure 1. It is unknown how much energy the entire pathway itself would require, so we don't know what the net energy will be. If we add carbon capture & storage (CCS) into the mix, the process could be carbon-negative at 100% efficiency. If we don't, the process might be carbon-neutral. The cost of CCS is 30% of the energy output.
The lack of humility before Nature displayed here is nothing short of astonishing.
In the video4 The Energy Problem: What the Helios Project Can Do About It, Chu speaks of the tantalizing benefits of solar energy efficiency.
Just to tantalize you on how important it is that we get these factors of 3 or 10 [in cost reductions for solar photovoltaics], or you let the price of electricity go up by a factor of 10, which I think most Americans aren't willing to do, you only need 0.2 to 0.3% of the world's desert to meet all [the world's] electricity needs at 20% efficiency ... [desert is only desert] ... so this is something that's out there in front of us as a challenge to scientists and technologists.
A little over half of the total solar energy passes through the atmosphere to reach the Earth's surface. This energy flux has an average power of 89,000 Terawatts (TW). In comparison: the total average energy flux of all the wind available on the planet is 370 TW and the average global power consumption 15 TW. In addition to being abundant, solar energy is reliable and has an outstanding energy payback...
Covering 1% of Continental U.S. with 20% efficient PV systems would provide all energy needs of the U.S. Covering 0.2% of the surface would generate all the electricity that the US consumes. To put this number in perspective: roads cover 1.5% of the U.S.
[1 terawatt = 1012 watts]
Experimental photovoltaic cell efficiencies have exceeded 40%, but no PV technology exists that achieves factor of 3 cost reductions (cents/kilowatt-hour) for solar electricity using common, cheap materials regardless of the efficiency. And because solar is "transient in nature" as Chu notes—you're good to go when the sun shines—we must actually reduce costs by a factor of 10 to achieve very-large scale deployment. Difficulty in getting cost reductions is thus tied to our failure to solve the very-large-scale energy storage problem, despite Chu's inappropriate statement that exponential learning curves for PVs will follow Moore's Law.
Thus loose talk of low-cost scalable 20% solar energy conversion efficiencies is an example of hand waving, a form of argument in which some hard problem is simply glossed over to make another point. If Chu's point is that the solar constant is quite large relative to our needs, then I completely agree. That doesn't get us anywhere unfortunately.
Kenneth Boulding alluded to "difficult" problems. Such problems appear to include very-large-scale energy storage (batteries) and cellulose-to-liquid fuel conversions. It should be clear by now, after decades of working to achieve highly efficient low-cost photovoltaics, that our failure to do so suggests that this problem too is very hard. Thus the costs of concentrating diffuse solar energy may be prohibitive or the problem may never get solved.
Portfolios and Probabilities
I talked about Chu's scattershot approach to outdoing Nature. Here is how Chu described his strategy at his confirmation hearing.
Such a multi-pronged approach looks to optimize all phases of biofuels production with no preconceived idea of which area is likely to offer the biggest payoff. And that, Chu said, “is why I’m so optimistic some real progress can be made.”
The uncertainty—"no preconceived idea" means "I have no idea"—surrounding which R&D project will offer a Big Payoff is more like playing the lottery than it is a reason for optimism that "real progress" can be made. This can be easily demonstrated.
Chu gets slightly more specific in the Energy Problem video. Discussing one idea published in the journal Advanced Materials for employing nanotechnology to improve photovoltaic (PV) solar cell efficiency, Chu talked in terms of probabilities at the 47:45 mark.
Now, if ideas like this work, it could transform the [solar] industry. What's the chance that a particular idea like this might work? Well, it could be 2%, 3% or something like that, but there's a portfolio of these ideas and I think the hope is that in the next 5 or 10 years one or several of these might come home to roost."
Suppose you have 10 potential breakthroughs5 (trials) where each is assigned a subjective probability of occurring (success) = 2%. Assume further that each trial is independent of the others. For example, what are the chances that using 2 dice you will roll a '5' on both? In our case, what are the chances that we will both attain 20% photovoltaic efficiency (in the sense above) and 5% photosynthetic efficiency? These problems are independent of each other. We use a binomial probability distribution to get some answers.
- The chance of a specific breakthrough is still 0.02 = 2%
- The chance of at least one breakthrough is 0.167 = 16.7%
- The chance of at least two breakthroughs is 0.015 = 1.5%
- The chance of no breakthroughs is 0.817 = 81.7%
Our probabilistic model for assessing success in Chu's science projects depends on 1) the subjective probability and 2) the number of trials. The more potential breakthroughs there are, the greater the probability of achieving success for one or two them is. This captures our intuition that the more bullets you fire in some general direction, assuming there is something out there to hit, the greater the chance that you will hit it. So if there are 100 potential breakthroughs, we will probably succeed on a few. This is why Chu calls his strategy for 4th generation biofuels a portfolio approach.
Would it help if the variables were not independent? Maybe yes, but we do not know in advance what depends on what or whether we will make the "right" breakthrough.
What if there are only a relatively few potential breakthroughs waiting for us to discover them? What if the actual (not subjective) probability of success is much lower than 2% or even zero? None of this can be known, for as Boulding noted in the opening quote, if we knew what we are going to know in the future, we would already know it!
How would Chu respond to this argument? I don't know, but I'm sure he would agree with the always optimistic Tom Friedman, who made the following argument on page 243 of his book Hot, Flat and Crowded.
Unfortunately, as noted earlier, we have not found that magic bullet—that form of energy production that will give us abundant, clean, reliable cheap electrons... Incremental breakthroughs are all we've had but exponential is what we desperately need...
[why have we failed up to now?]
The answer is two-fold. First energy innovation is real hard... But second, more important, and the subject of this chapter and the next: We haven't really tried. That's right, we haven't really tried.
One could certainly argue that we haven't tried hard enough. We must try harder and under Dr. Chu's direction we will. My unanswerable objection is that we can't Bet The Farm on the unknown outcome of all these science experiments. But that's precisely what the Obama administration seems to be doing.
Political Realities Are Human Realities
You might ask at this point why Chu is working on 4th generation biofuels instead of encouraging us to ease up on our driving by promoting urban living, light rail systems, more long-haul railroads, etc. Less driving means fewer pounds of CO2 emitted to the atmosphere and fewer barrels of oil consumed.
Higher fuel economy standards—this is Chu's notion of efficiency as far as I can see—have the political virtue of not requiring any behavioral change on the part of the American public. You simply drive a brand new Toyota Highlander hybrid off the dealer's lot instead of a Land Cruiser. You may even be eligible for a rebate that lowers the premium paid for your HEV SUV.
Put another way, why is geoengineering preferable to implementing sensible policies that promote liquid fuels frugality? Are these people crazy? Well, yes and no. I have thought about these questions for some time now. Here's part of my answer, which I have put into the form of a quote.
Once a beneficial mode of living is established, whatever its real cost, human beings are quite resistant to changing it. Solutions to problems requiring deep behavioral changes are thus politically impossible unless there is a mandate accompanied by a credible threat of coercion to enforce the required behavioral shifts.
In a non-totalitarian society like ours, such mandates are not credible. It is always possible to find new leaders who will get with the program, which essentially amounts to doing nothing. (Doing nothing is a choice: it allows events take their natural course.) Ease and expediency in a political system like ours always encourages politicians to maintain rather than change the status quo. Societal behavioral changes are therefore always gradual unless shocks occur that put large behavioral changes in motion. (World War II would be an example, but such examples will obviously be rare.)
Because human beings are drawn to and good at technology, the promise offered by the magical technological solution to any problem is almost irresistible. Although technological change drives behavioral change, this usually goes in the direction (young) people want to move in anyway (e.g. cell phones make it easier keep in touch.) If a problem is serious, as with energy, the more time a technological solution requires, the more popular it will be with politicians for whom gradual solutions are always good and shocks are always bad.
We started to experience a large energy shock in July, 2008. The oil price briefly exceeded $145/barrel. Among politicians there was widespread panic. As luck would have it, they were let off the hook when the global finance system almost collapsed. Now they are in an uproar about bonuses at AIG. So it goes.
Steven Chu is the scientist/technologist with a grandiose but non-disruptive, gradual solution to our energy problems. He has given himself 20 to 80 years to get the job done. Barack Obama is the politician who hired him. If he serves two terms, he will leave office in January, 2017. Now you see that it all makes sense politically. Of course, waging war on scientific reality must always fail. We do not want to pin all our hopes on what we don't know about the future.
Contact the author at firstname.lastname@example.org
1. Today's column title was taken from The Loom, a blog at Discover Magazine.
2. Thanks to Phyllis.
3. The world recently had some 'near-death" experiences accompanying the frenetic demand growth from 2003 to Q2 2008. For example, China could not supply itself with enough coal and turned to Australia and others to fill in shortfalls. In the United States, both the airlines and the long-haul trucking industries, which depend on a cheap, reliable supply of middle distillates, were on the verge of bankruptcy. if the global finance system, which carried the seeds of its own destruction, had not broken down, I wonder where we would be today. But that is only a thought experiment now.
4. This is the same presentation in which Chu described coal as his "worst nightmare."
5. Thanks to my friend Dan Friedman for pointing me in the right direction.