Kurt Cobb’s original article “Should we use net energy to measure global energy reserves?” is at Resource Insights and was republished at Energy Bulletin.
I am not sure what audience Kurt Cobb was writing his article for, but I think most Energy Bulletin readers would have found little difficulty in accepting the critical importance of what Cobb calls “net energy” to the world’s energy future. However it should be noted that net energy is a notoriously difficult figure to pin down in many cases, and when it is pinned down, the interpretation of the information is not always so straightforward.
Strictly speaking “net energy” is the difference between the energy input into a system (for example, an energy producing project like a power station and all the associated resources and infrastructure) and the energy output by the system over its lifetime. However the only way to properly compare projects of different sizes is to use the ratio of Energy Returned (ER) to Energy Input (EI), or ERoEI, instead of net energy. An ERoEI for an energy project is 1.00 when the process is at the break-even point, that corresponds to the net energy being zero.
Energy Returned
Energy Returned has to take account of the productive lifetime of the project, including down time for maintenance and repairs. If the energy driving the system is not constant, such as with wind, solar and tidal generation, this must be factored in, although peak generation capacity values are also important in other aspects of the design.
Solar PV panels are often sold with the sales pitch that they have a lifetime of 25 years, but the Solarhart 21W panel I bought in 1983 only lasted until 1996, when the junction box on the back of the unit corroded, due to dissimilar metals, and fell off – leaving the panel useless. The Heavy Duty Deep Cycle lead-acid battery I got with it lasted a lot less than that.
Given that our modern energy-intensive global society is very likely to suffer a major dislocation due to Peak Oil, with recessions, bankruptcies, buy-outs, mergers, asset-stripping and general economic upheaval, what are the chances that the 25 year warranty on your solar panel will be honoured 13 years down the track when the wires fall off ? Nil, I should think. So is it valid to calculate ER for the panel (Power x Time = Energy) with t = 25 years ? This is an essential question that doesn’t really have a good solid answer. It ought to be theoretically possible to ascertain a Mean Time to Failure and a probability distribution around the mean for early and late failures, but this is only going to be valuable when looking at the entire population of panels, not to a single panel or a small installation.
Solar PV at least has the simplicity that the fuel to run it, sun-light, is free. Most competing systems require a fuel, or a range of fuels, and availability of fuel could be the thing that determines ER. Is it fair to calculate the ERoEI of a gasoline generator on the understanding that gasoline is going to be available for the next 25 years ?
Energy Input
Turning to Energy Input, we have two kinds of inputs. The infrastructure energy costs for, say, a coal-fired power station include the cost of building the plant, on-going maintenance, and eventual decommissioning. These energy costs have to be shared out amongst all the units of output produced. And the ‘per unit output’ energy costs which include fuel, measured in tonnes(coal) per GW.h(electricity) , labour, and managerial overheads – accounting, advertising, tendering, insurance, office lighting, air-conditioning, computers, furniture, etc.
A coal-fired power plant cannot operate in isolation either. It needs a railway or conveyor belt to bring in the coal, and something has to happen to the ash. It is possible that some of the ash can be used by the concrete industry, in which case it might be worth something (energy-wise) as a by-product of electricity production. In a few places in the US, the local coal contains traces of the metallic element Germanium, which ends up in the ash and is extracted to be used in the fibre optic cable and semi-conductor industries.
In a similar way, the sugarcane industry produces crystal sugar, molasses, bagasse (the fibre), mill mud and co-generated electricity. In Australia, the bagasse is dried by stacking it near the furnace and then it is burnt in the furnace, providing all the process heat once the system is fired up at the start of the crushing season. The some of the electricity generated is also used on-site. The mill mud, which is mostly soil and fine organic particles, is sold for the cost of trucking the mud back to the farms.
So the question is, how does one allocate the energy input costs of growing the cane (preparing the field with tractors, planting, fertilising, pesticide spreading, harvesting, and transporting to the mill) to the various products of the milling process ? This has important implications for those who are trying to calculate the ERoEI for ethanol fuel made from sugarcane molasses.
One solution is to allocate the energy costs in proportion to the financial costs of the products, however this has the unsatisfactory result of the energy accounting changing when market factors change the price. Since the US has mandated the use of some ethanol in gasoline as an oxygenator, the demand for ethanol has risen, and this has driven up the price of corn and sugar all around the world, thus changing the ERoEI of ethanol when the energy accounting is done is this way.
Another way of allocating the EI costs is to consider how much energy is in the various output streams, based on how much heat they would produce if they were burnt. This has the advantage that the numbers don’t fluctuate with the rises and falls of the financial markets.
Outputs that are Inputs
Again using the sugar mill as an example, the bagasse is an output of the milling process, but although it is a good fibre, it has no commercial value as a fibre, so it is burnt to provide process heat for the mill. Thus is it an output and also an input to the process. The question then is, how should the bagasse be counted ? Should it be counted as both an output and an input, or should it just be left out of both ledgers ? To my way of thinking it should definitely be the former, but for those who are trying to make out a case for ethanol having a good ERoEI, consider this.
If EI is 100 energy units, of which 50% is the energy in the bagasse, and ER is 110 energy units, then ERoEI is 110/100 = 1.10 if the bagasse is counted both as an input and as an output, but the ERoEI becomes 60/50 = 1.20 if the bagasse is excluded. So what do you think unscrupulous statisticians do ?
A similar thing applies to the co-generated electricity, where some of the electricity is used by the mill itself, and some is returned to the grid, while some grid electricity is imported outside of the crushing season. Now the meters at the interface between the grid and the mill only measure the net energy transferred, so no one knows exactly what the quantity of re-used electricity is.
Forms of Energy
While it is true that all forms of energy are theoretically inter-convertible, in practice the conversions are always accompanied by losses (this is the Second Law of Thermodynamics in operation). Thus the chemical energy embedded in coal is only able to be converted into electrical energy with roughly 36% efficiency, and gas-to-liquid conversion is reputed to be 48% efficient ( Gas to Liquid Life Cycle Analysis Synthesis Report by Five Winds International for ConocoPhillips, Sasol-Chevron and Shell International Gas, 2004), at least in desktop calculations.
Similar figures are given for crude oil to gasoline refining, although the multiple co-products again makes hard numbers difficult to separate out.
Nevertheless, such processes with ERoEI of much less than 1 can be worthwhile conversions, if the output energy form is more useful than the input energy form. Claims of ‘it will never happen because the ERoEI is less than 1’ need to be considered carefully in this light before being accepted.
And finally let me extract from the Farrel Report (Dec. 2005), which hit the headlines a few months ago when the media decided to use the results below to claim that all argument has now been swept away and ethanol is proven to be a realistic proposition.
The report looks into six earlier papers where each tried to come up with an ERoEI for corn to ethanol in a US scenario. It tries to standardise the methodologies to the same system boundaries, correcting the results for perceived ‘errors and inconsistencies’. ( Always a dangerous thing to do to other peoples’ work ! Please note that some of these corrections are disputed. )
| Principal author | ERoEI | ERoEI corrected |
| Patzek | 0.81 | 0.94 |
| Pimentel | 0.78 | 0.86 |
| Shapouri | 1.5 | 1.4 |
| Graboski | 1.2 | 1.1 |
| de Oliviera | 1.1 | 1.2 |
| Wang | 1.4 | 1.3 |
As you can see, there is little agreement here even after the analyses have been ‘normalised’.
The report also finds that ‘average figures’ for the various input energy forms are as follows :
per 100 MJ of ethanol
coal : 41 MJ
NG : 28 MJ
petro : 5 MJ
nuclear : 4 MJ
total : 78 MJ
This is equivalent to an ERoEI of 1.28 . If these figures are accepted, it leaves the important question of whether this is a good outcome from the energy point of view. Is it a good return to spend coal and gas to get ethanol, rather than spending crude oil to get gasoline ? It is a question demanding a subtle answer.
There is a lot more in this report, and the whole subject revolves around critical definitions of system boundaries, so it should be read carefully before deciding what it means. Rather than settling the matter once and for all, the Farrel Report shows just how difficult it is to get agreement on such a ‘straightforward’ thing as ERoEI.
