Abstract: The last study of the energy balance of producing ethanol from cellulose was back in 1979. It showed that the energy inputs required were 3.3 times the energy contained in the ethanol output. Nevertheless the belief has persisted that producing ethanol from cellulose may become viable. This paper attempts to interpret the most recent plans for improving the process. It concludes that the best plans on the drawing board, at the present time, indicate a positive energy balance, with inputs comprising about 63% of the ethanol output.
However, the net energy capture remains too low to be significantly useful in terms of supporting present levels of population. It is about 3 parts in 10,000 of the insolation (solar energy), or 0.5 kW/ha. Ethanol from corn, according to the more optimistic analyses, also achieves 3 parts in 10,000, and that of course is also non-viable.
Because of the reduced ecological damage caused by perennial crops, ethanol from cellulose may seem worth pursuing, if only a way could be found to make the inputs negligible in comparison to the output. The paper goes on to view the solar fraction of 3 parts per 10,000 within a wider context.
All sources of renewable liquid energy are inadequate when set against the net energy density that is achieved from extracting oil from wells, which we estimate as being the equivalent of capturing all 10,000 parts in 10,000 of insolation, or even from producing synthetic gasoline from coal — equivalent to capturing 2200 parts in 10,000 of insolation. 3 parts per 10,000 is a pale shadow of the fossil fuel net energy densities which have been the sine qua non of the 4400 million population growth in the last century. There is a manifest need for nations to reduce their populations before fossil fuels become scarce.
The production of ethanol from sugar cane and particularly from corn (maize) has been studied in great detail. One insuperable barrier to using corn is the amount of damage that is done to the soil, both by way of soil erosion and because of the need for heavy use of fertilizers and pesticides, which give rise to pollution of air and water (Pimentel, 2003). Even if we set aside those problems, the net energy capture is too low to be useful. Only about 3 parts in 10,000 of the sun’s energy is captured (Hayden, 2001, p. 98; OPTJ 3/1, April 2003, p. 11).
By way of comparison, we may note that the net energy density of producing synthetic gasoline (petrol) from coal is such that, to equal it, we would need to capture 2200 parts per 10,000 of the sun’s energy (see below).
Some people have tried to minimize the significance of the low net energy capture of ethanol from corn; for example Shapouri et al (2002, p 1) said, “In other words, abundant domestic feedstocks such as coal and natural gas can effectively be used to convert corn into a premium liquid fuel that replaces imported petroleum.”
However, with the United States already importing 61% of the oil it consumes, and about 15% of the natural gas it uses, and having reached the peak of its own natural gas production, and with the output to input ratio of coal extraction dropping rapidly, there is now a general awareness of the need to consider the net energy capture that can be achieved from the sun’s radiation (i.e. insolation), without relying on subsidy from fossil fuels.
Even were corn to be suitable on net energy grounds, it would be difficult to defend on other counts, as the feedstock for producing energy in liquid form. One obstacle is the need for cropland to feed the world. So we ought to consider the possibility of producing liquid energy from less fertile ground than cropland. For that, perennial woody plants fit the bill. Moreover they are less troublesome with respect to ecological damage.
Unfortunately there has been a lack of available data concerning the process of obtaining ethanol from cellulose. The only published study I know of — covering the most energy efficient system acid hydrolysis — goes back 25 years to Slesser and Lewis ( 1979, p. 108). The study estimates that it takes 98 MJ to produce a kg of pure ethanol, for which the High Heating Value of 29.6 MJ is given. In other words, the authors reckoned it took over 3.3 times as much energy to produce the ethanol, as the energy contained in the ethanol output. A substantial part of the reason for that is the energy needed for processing and distillation, for which the energy input (as hard coal) is shown as 2.1 times the ethanol output.
As I hope to show, the energy balance can be greatly improved, but before doing that perhaps we should note that the problem of inputs appears to have been ignored by some people; for example, when Lynd (1996, p. 412) estimated that 186 million tonnes of waste biomass (dry) could be collected in the US, and that this would yield 20 billion gallons of ethanol, he attempted no estimate of the inputs needed. What we may also note — without pausing to judge whether his assessment of the amount of waste biomass likely to be available is in the right ball park — is that he is implying a requirement for 2.46 kg of biomass to produce a liter of ethanol. Compared to the 2.69 kg of corn grain used to make a liter of ethanol that looks promising but, as already observed, it is essential to look at the energy balance.
It must be stressed that the data I am going to put forward regarding improvements are not nearly as secure as those used by Slesser and Lewis in their 1979 study. I merely suggest that they serve as a ball park assessment of the sort of improvements that may be possible. The source of my information is Jack Just, an engineer working in New Zealand (with a branch in Florida), who has been working on cellulose conversion processes for 25 years. First he ran a pilot plant to test developments of the old Madison method, the outcome of his tests was confirmation that more radical changes to the process were needed. The figures which are given below are estimates of what could be achieved with a revised procedure, which Just now has on the drawing board. He himself is wary of giving many definite figures, because secure data would only be available from a pilot plant of substantial size. Nevertheless his figures should serve to put us in the picture whether it appears to be worthwhile to pour substantial further funds into developing processes for producing ethanol from cellulose.
In order to generalise the study, we will also use data for the feedstock from a 1997 paper by Mario Giampietro, Sergio Ulgiati, and David Pimentel, Feasibility of Large-Scale Biofuel Production. The data we need are taken from one table there, “Table 4. Methanol production from wood”. Although the subject of that study was methanol, the paper will serve us perfectly well, since the agricultural data, which is what we will take from it, apply equally to ethanol production. Note, too, that the “Feasibility” paper is probably intended to apply to the United States. For the following analysis, for the purposes of assessing energy capture, we will use the mean annual insolation for the US, which we will take as 1900 kW/ha; but observe that we are thereby dealing in round figures, which would not precisely apply everywhere; somewhat different analyses could be made for other places.
In the United States itself, annual insolation varies from 2400 kW/ha in the southwest to about 1500 kW/ha in the north east (incidentally, Hayden, 2001xx, uses a mean figure of 2000 kW/ha for the US as a whole). 1500 kW/ha is about the right figure for southern France and New Zealand. The UK figure is about 1100 kW/ha. Assessing the efficiency of the process in terms of energy capture (the amount of insolation captured per hectare) is of course only one method of assessing the efficiency, but we have benchmarks for it, not only the already mentioned figure for ethanol from corn, 3 parts per 10,000, but also for ethanol from sugarcane, which achieves 9 parts per 10,000 (if we allow that the bagasse by-product can provide all the processing and distillation heat).
Returning now to the data from the “Feasibility” paper, we can note that a wood yield of 10 t/ha/yr is assumed there, based on “short-rotation woody crops.” . The calorific value used makes it evident that this is equivalent to 8.5 dry t/ha/yr. At the fairly standard calorific value of dry wood (it varies according to type), 20 GJ/t, this amounts to 170 GJ/ha/yr = 5.4 kW/ha. With a typical US insolation of 1900 kW/ha, that is a gross energy capture of 5.4 / 1900 = 28 parts per 10,000.
Jack Just says that for the continuous production “acid hydrolysis process” which he has on the drawing board, he will probably be able to improve on the yield from the old Madison process, perhaps by as much as 25%. Nevertheless, it would require a pilot plant to establish that fact for sure, so Just suggests using only the figures already established for the Madison process. They equate to about 253 liters of anhydrous ethanol per tonne of feedstock. Note I say “equate to” because while the ethanol has to be distilled to an acceptable degree of purity (say 95%), anhydrous ethanol is never the practical objective. Giving the anhydrous ethanol output is merely a way of making the heat value of the output easy to assess. Some scientific papers are imprecise (e.g. Shapouri et al, 2001) about the degree of distillation that has been accounted for in the input figures. With that point made, let’s proceed with Just’s figures.
They imply that 3.95 kg of dry wood produces 1 liter of anhydrous ethanol. Compared to ethanol from corn (for which 1 liter is produced from only 2.69 kg of corn as mentioned), and compared to Lynd’s 2.46 kg figure mentioned above, it may seem that we are needing a surprising amount of wood to produce a liter of ethanol, but note that this ethanol output is from the Hexose sugars alone. As we will observe later, the Pentose sugars are used to produce methane, and the methane output amounts to 83% (in energy terms) of the ethanol output. Let us now observe that the 3.95 kg of dry wood needed per liter of output indicates that about 27% of the energy in the wood is retained in the ethanol. Thus the amount of the insolation energy captured as ethanol is 28 x 0.27 = 7.6 parts per 10,000.
One of the main benefits of using woody crops, instead of corn or sugarcane, is that cropland is not required. However 8.5 dry tonnes of wood cannot be harvested every year from only moderately fertile land without replacing nutrients. Giampietro et al. (1997) indicate that the main fertilizers required, nitrogen, phosphorus and potassium, need to be applied at rates of 100, 20, and 60 kg/ha respectively. Additionally some pesticide application is required, and the authors estimate that requirement at 0.39 kg/ha/yr. Combining the two, they estimate the inputs as 9.1 GJ per ha. This amounts to 20% of the ethanol output.
Jack Just suggests that by returning the residue of the ethanol production process to the fields, it may be possible to either reduce or eliminate the additional artificial fertilizer, but that possibility would require a separate study.
Using the “Feasibility” data again, the energy inputs for wood harvesting and handling amount to 17% of the ethanol output.
As mentioned earlier, the Pentose sugars are used to produce methane. The energy content of this methane amounts to 38,000 MJ/ha, which is about 83% of the energy in the ethanol. Although the energy required for the basic distillation to take the ethanol up to 95% concentration is substantial, this is more than sufficient to supply the total heat for both hydrolysis and the basic distillation process. Thus no additional inputs are required to carry out those processes (this is of course the biggest difference with the process analysed by Slesser and Lewis). Although we mentioned that the amount of methane is “more than sufficient,” there is not so much excess energy that we need to take the excess energy from methane into account for the purpose of diminishing the next two inputs that we need to consider.
The next two inputs are both electrical: first the energy needed to chop up the feedstock to a suitable size, and second to run the pumps and other machinery. Jack Just estimates that, in the continuous process which he is proposing, in the course of 1 hour, an input of 495 kWh(e), 1500 kWh(th), is required, including the energy needed to chop up the feedstock. During that hour, the output of ethanol is 3049 liters, which is equal to about 18,000 kWh(th). So the thermal energy inputs, that are required to make the electricity, amount to 8% of the output. It should be noted that while this is a considerable improvement on the 21% of the ethanol output shown in Slesser and Lewis’s analysis, Just’s figure of 8% really requires further verification, as not only does he not yet have a pilot plant, but the figure he gives is his overall estimate, and it is without any breakdown to show the electricity needed to chop up the feedstock to a suitable size.
The final inputs comprise the chemical compounds, embodied energy, and water. One energy saving in Just’s planned procedure arises because he does not use CaO.
That leaves the following inputs which we take from Slesser and Lewis (Just does not have his own data):
kg/liter of ethanol MJ/per liter of ethanol
Superphosphate 0.004 0.05
Na3PO4 0.004 0.05
(NH4)2SO4 0.006 0.09
CaCO3 0.30 2.70
Antifoam 0.04 0.13
Stainless steel 0.003 0.20
Structural steel0.005 0.24
Cement 0.01 0.06
Water 125 0.24
The above listed inputs amount to 3.76 / 21.25 = 18% of the ethanol output. So the total inputs amount to 20 + 17 + 8 + 18 = 63% of the ethanol output, which leaves 37% as the net output of energy.
This means that energy capture net of all inputs (that is adjusting the wood solar fraction for the ethanol conversion and for the inputs) is 28 x 0.27 x 0.37 = 3 parts per 10,000. This is the same as the 3 parts per 10,000 achieved with corn. While the cellulose process would have the advantage of doing far less damage to the soil and ecology generally, 3 parts per 10,000 is still too weak a net energy capture to make a significant impact on our need to extract ‘liquid’ energy from the sun in ‘real time’, with anything like our present levels of population. The net energy capture amounts to 0.5 kilowatts per hectare.
Other forms of renewable energy
Liquid energy from renewable energy sources remains an unresolved problem because of very low net energy capture. It may be helpful to see how these processes compare with other renewable energy sources, even those which produce only electricity. Note that when the output is electricity, it is conventional to uprate the output to its thermal equivalent, which we will do (by dividing by 0.33), but it has to be said that whether that makes sense in a renewable energy world is open to challenge, since electricity is often as easy to generate, and transmit, as high temperatures (i.e. to provide energy to heat the house on a cold day).
Photovoltaics. Assuming that the shading ratio is two to one (i.e. as much space between the panels as the area of the panels), the gross thermal energy capture using photovoltaics is about 1500 parts per ten thousand of the insolation falling on the whole photovoltaic field. The net energy capture would be in the region of 1200 parts per ten thousand. That is a four hundred fold improvement on 3 parts per ten thousand, but photovoltaics have two problems of their own: (a) intermittency — based on average insolation of the United States, the modules only produce 14% of the energy they would if they were producing at their maximum output continuously; (b) cost — presently, for the same output, the capital cost of photovoltaics is about nine times that of wind power.
Coal and synthetic gasoline. The term ‘net energy capture’ is not entirely appropriate to coal and fossil fuels. More suitable, and more usual, is the term ‘net power density’. The concept and the units are essentially the same, except that with ‘net power density’, the early stages of energy capturing may have occurred during several million years, instead of all being captured by us while we are alive — that is in ‘real time’.
The land area needed to produce and transport coal is such that its extraction achieves a ‘net power density’ of about 830 kW/ha. In order to achieve, from ‘real time’ insolation, a net energy capture equal to the net power density that we get from coal mining, it would be necessary to capture 4400 parts per 10,000 of the insolation.
To make 1 t of synthetic gasoline from coal requires 3.65 t of coal, including the coal needed to carry out the conversion process (Durrant, 1953, p. 309). Thus to achieve a net energy capture to match the net power density from producing synthetic gasoline (petrol) from coal, we would need to be able to capture about 2200 parts per 10,000 of the insolation.
Oil and gas. Obviously oil and gas do not entail such extensive mining areas or transport areas as coal. Although I know of no attempt to assess their net power densities, we can confidently estimate them to be two or three times that of coal. The figure can only be approximate, and we may conveniently put it at 1900 kW/ha, thus equating it with the average insolation of the USA. What that means is that to achieve from ‘real time’ insolation, the same net energy capture as the net power density that we get from oil wells, it would be necessary to capture all 10,000 parts per 10,000 of insolation. That shows very clearly why the 3 parts per 10,000, which we achieve with ethanol from corn, and may achieve with ethanol from cellulose, is not going to be sufficient to maintain our present population.
3 parts per 10,000 of insolation — which there are somewhat loose reasons to think might be achieved by producing ethanol from woody plants — is almost insignificant compared to the 10,000 parts per 10,000, which is the ‘real time’ insolation capture that we would need to achieve in order to match the net power density that we enjoy when we tap into the ‘ancient sunlight’ of oil reservoirs, and even the 2200 parts per 10,000 that we achieve by producing synthetic gasoline from coal.
Nevertheless, on present evidence, 3 parts per 10,000 appears to be the best that is likely to be achievable in producing liquid fuels from ‘real time’ sunlight, with the exception of ethanol from sugar cane, which can achieve 9 parts per 10,000, but that is by providing the processing heat from the by-product bagasse, and the long term production of sugarcane over large areas is unlikely to be sustainable as it causes massive soil erosion, besides which growing sugarcane is confined to only suitable areas. The 4400 million expansion of human population in the last century was made possible by fossil fuels. The human race needs to take this into account, as it plans for the next century, during which the non-renewable fossil fuels, oil, gas, coal and uranium will gradually become exhausted.
(some references are contained in the endnotes)
Durrant, P. J. 1953. General and Inorganic Chemistry. London, UK and New York: Longmans, Green & Co. Ltd.
Giampietro, M., S. Ulgiati, D. Pimentel 1997. Feasibility of Large-Scale Biofuel Production: Does an enlargement of scale change the picture. Bioscience Vol. 47, No. 9, October 1997, pp. 587-600.
Lynd, L. .R. 1996. Overview and evaluation of fuel from cellulosic biomass, Annual Review of Energy and Environment, 21, 403-465.
OPTJ 3/1. 2003. Optimum Population Trust Journal, Vol. 3, No 1, April 2003. Manchester (U.K.): Optimum Population Trust. 32 pp. Archived at www.members.aol/optjournal2/optj31.doc
Pimentel D, Pimentel M. 1996. Food, Energy, and Society. Revised edition. Niwot Co., University Press of Colorado. 363 pp.
Pimentel, D., Giampietro, M. and S. G. F. Bukkens. 1998. An Optimum Population for North and Latin America. Population and Environment: a Journal of Interdisciplinary Studies, Vol. 20, No. 2.
Pimentel, D. 2003. Ethanol Fuels: Energy Balance, Economics, and Environmental Impacts are Negative. Natural Resources Research, Vol. 12, No. 2, pp. 127-134.
Reifsnyder W. E., and Lull H. W. 1965. Radiant energy in relation to forests. Technical Bulletin No. 1344, Washington, D. C.: U.S. Department of Agriculture, Forest Service.
Shapouri, H., Duffield, J.A., Wang, M. 2002. The Energy Balance of Corn Ethanol: An Update. United States Department of Agriculture (USDA), Agricultural Economic Report Number 813
Slesser, M., and C. Lewis. 1979. Biological Energy Resources. London: E. & F.N. Spon.