Soy Biodiesel Review

The topic of discussion is Sheehan, J., V. Camobreco et al. (1998) . Life Cycle Inventory of Biodiesel and Petroleum Diesel for Use in an Urban Bus. Golden, National Renewable Energy Laboratory. It's a big, inclusive report (314 pages) on the the energy balance of soy derived biodiesel and fossil diesel.

Energy Return

Let's start on p. V of the executive summary:

Biodiesel yields 3.2 units of fuel product energy for every unit of fossil energy consumed in its life cycle. The production of B20 yields 0.98 units of fuel product energy for every unit of fossil energy consumed.

That first number, 3.2 units of fuel product energy for every unit of fossil energy. What does this mean? It's in a big bold block quote in the executive summary. It looks like the EROEI, right? Unfortunately it's not. On p. 207 we find that Fossil Energy Ratio = Fuel Energy/Fossil Energy Inputs. In other worlds, any power input that is not a fossil source is not accounted for. For the purposes of this study, this means the hydroelectric and nuclear power share. What we actually want to know is the total process energy required. The energy inputs for biodiesel are predominately electricity (to run machinery) and low grade steam (50 - 70 °C), along with natural gas (to produce methanol) in addition to the standard farm inputs.

Fortunately, the results are not that badly off due to this factor. The study divides the production of biodiesel into five stages:

1. Agriculture
2. Transport from farm to processing plant
3. Soybean crushing and oil separation operations
4. Conversion of soy oil to methyl ester fuel.
5. Transport and distribution of biodiesel to consumers.

The study also accounts for transport from the separation plant to the conversion plant but this energy input is left out of the end results. I agree with this because it is overly high due to the small number of biodiesel plants in the USA at the time of the study. For large scale biodiesel production the two operations would be naturally collocated.

I went through the study to attempt to figure out what the difference was between fossil and absolute energy inputs. Annoyingly, for Stage 4 (Conversion), the energy of the soy oil is incorporated as an input. What makes this frustrating is that the authors at no point in the study actually define what they consider the energy content of soy oil to be, which makes deconstructing this part of the study difficult.

Table 1: Energy Allocation to Biodiesel Production Stages

Activity	Energy (MJ/kg biodiesel)	Source
Agriculture	3.158	Table 62, p. 116
Transport	0.162	Table 63, p. 118
Separation	3.471	Table 83, p. 137
Conversion	5.572	Table 105, p. 166
Distribution	0.162	Table 106, p. 169
Total	12.526
Higher Heating Value	40.6	Table 108, p. 173
Lower Heating Value	37.0	Table 108, p. 173
ERR (HHV)	3.24

As it happens, I get a better result (3.24 > 3.2) but that's due to the fact that I employ the Higher Heating Value -- the original authors' calculation uses the LHV. As it happens, this is only a minor bone I have to pick with the study. The big problem is with what's called "Allocation of Lifecycle Flows." Anyone who has read into ethanol studies will know this as 'coproducts'.

Funny Coproduct Accounting

The first giant problem associated with 'coproducts' appears in the Separation (or crushing) stage. The oil content of soybeans is rather low − around 18.4 %. For this entire study, allocation of energy consumption between biodiesel and coproducts is done purely on a mass basis. Unfortunately, this leads to a silly assumption. Table 82 (p. 136) allocates 18 % of energy consumption for the Separation stage to soy oil and 82 % to meal. This in turn propagates back through the allocations for transport and agricultural energy consumption. Is this fair? Take a look at Table 64 (p. 121):

Table 2: Mass composition of soybeans

Oil	18.4 %
Dirt	0.8 %
Hulls	7.4 %
Water	16.0 %
Meal	57.4 %

That's right boys and girls, the authors are allocating the same value to oil as dirt and water. Realistically if the mass of dirt, water, and hulls were discarded then the oil would have to assume 24.2 % of energy use for the first three stages. Furthermore, it probably makes more sense to compare the ratio between the wholesale price of soy oil versus soy meal to determine the proper value of the coproducts. Free hint: the oil is worth more per kilogram than the meal.

This process is repeated for the Conversion stage is similar for the allocation between methyl ester (biodiesel) and glycerin. For this stage, 82 % of the energy is allocated to biodiesel and 18 % to glycerin. Once again this allocation is propagated back through the previous steps. However, this calculation is actually unfavourable to the biodiesel. Separating the glycerin and excess methanol consumes approximately 65 % of the energy for the Conversion stage (Table 96, p. 159). The reason is distillation. As anyone who has looked at ethanol systems will know, distillation is a killer because it requires so much energy to vapourize water. Also, the NREL numbers come out quite high compared to some European plants also presented in the report.

From my point of view, I want to know if biodiesel is energy positive, regardless of coproducts. For soy, the answer appears to be no. Going back and removing the coproduct credits appears to give the following results:

Table 3: Energy Consumption for Biodiesel Production
with Zero Coproduct Credits

Activity	Energy (MJ/kg biodiesel)
Agriculture	21.40
Transport	1.10
Separation	23.52
Conversion	6.80
Distribution	0.20
Total	53.01
ERR (HHV)	0.766

Before we all fall into a state of depression, it is fairly clear from the report that there is a lot of promise in reducing the energy inputs for the conversion stage. Methanol inputs constitute approximately half the energy inputs. I have previously hypothesized that anaerobic digestion of the meal could produce methane which in turn could be made into methanol in addition to providing heat energy. The NREL numbers require approximately three times as much energy as some quoted European operations (Table 98, p. 161).

There are potential improvements to be made to the efficiency of the Conversion stage as well. Research and development on catalysts offers the potential to reduce the reaction temperature. In particular I think a zeolite could be ideal for separating the glycerol from the methyl ester chains. Most of the energy (65 %) is used not for the actual conversion but for distilling out glycerin and excess methanol post-transesterfication − normally an excess of methanol is added to carry through the reaction to completion. Reducing the amount of water and methanol used will have a direct result on the distillation requirements.

Like ethanol, biodiesel would benefit significantly from combined heat and power generation. The temperature requirements for most processes is low enough (50 - 70 °C) that the use of solar thermal systems to augment the heat production is feasible.

To a certain extent glycerin might be the biodiesel analogue to sulfur for petroleum oil. Sulfur is a chemical with its uses, but oil refining produces mountains of the stuff. Will glycerin be a product worth distilling in a biodiesel nation, or should it just go into the anaerobic digester to make more methane?

Soy versus Rapeseed (Canola)

Any way you cut it soy is not an ideal crop for biofuel production. Soy does have one significant advantage in that it's a legume and hence fixes atmospheric nitrogen. As such, the energy requirements for fertilizer for soy is very low compared to everyone's favourite biomass villain, corn. However, the foremost quantity on my mind is the low oil content of soybeans. It's about 18 % (Table 64, p. 121) versus 40 % for rape and jatropha or 70 % for coconut. Rape appears to be the best temperate crop for biodiesel production. Its oil quality is high as its content, and its moisture content low.

The NREL study uses an average yield of 36 bushels/acre for soy which works out to 445 kg (oil)/hectare. (Here is a useful webpage for converting agricultural units from US Customary nonsense to more sensible metric units. Oh, and soybeans are 60 lbs./bushel, not 56 or 48 or 25 lbs./bushel but you all knew that, right? Next thing you know they'll be measuring the volume of biodiesel in barrels.) In comparison Canadian Canola yields about 640 kg (oil)/ha. The same source gives European Rapeseed a much higher yield of approximately 1280 kg (oil)/ha, largely due to the greater use of irrigation.

Aside from the oil content issue there are a number of other drawbacks for soy. For the most part, soy appears to take a great deal of work to get the oil separated from the meal. Soy has a high moisture content of 16.0 % water by mass (Table 64, p. 121) which necessitates drying.
In comparison, Canola is about half that if properly sun dried, and hence can be processed without drying. Soy also needs to be flaked into regular sized small pieces, which constitutes about a quarter of the electricity requirements for the Separation stage.