Life-cycle assessment

The plant's ethanol is sold not just by the litre but by its carbon intensity: the greenhouse gas emissions over its whole life cycle, from the corn field to the car's tailpipe, per megajoule of fuel. Under the federal Clean Fuel Regulations, a lower carbon intensity is worth real money. The number is calculated with the federal Fuel Life Cycle Assessment (LCA) Model that Environment and Climate Change Canada maintains for the Clean Fuel Regulations, from the plant's own data on corn, energy and co-products. Malik wants to know what the plant's current number is, what drives it, and which parts the expansion can change. Wren has done the annual carbon intensity filing for three years, but the expansion needs more than a filing: it needs a life-cycle assessment that the board and the plant's customers can rely on. Theo is asked to learn the method properly.

What a life-cycle assessment is

A life-cycle assessment (LCA) estimates the environmental impacts of a product over its whole life: raw material extraction, processing, manufacturing, distribution, use and end of life. Its value is that it shows where impacts actually occur, which is often not where people expect, and it prevents problem shifting: solving one impact by creating another somewhere else in the chain. The method is standardized in ISO 14040, which sets out the principles and framework, and ISO 14044, which gives the requirements and guidelines. A study that claims to follow them has four phases: goal and scope definition, inventory analysis, impact assessment and interpretation. The phases are iterative; findings in a later phase often send the analyst back to refine the scope.

Goal and scope

The goal states why the study is being done and who will use it. The scope defines the product system, its boundaries, the functional unit, the allocation rules, the impact categories and the data requirements. The functional unit is the quantified function the product delivers, against which everything is measured. For a transportation fuel the usual unit is 1 MJ of fuel energy, because fuels are compared by the energy they deliver to an engine, not by volume or mass. A litre of ethanol contains about two-thirds of the energy of a litre of gasoline, so comparing per litre would flatter ethanol.

The system boundary decides which processes are included. A cradle-to-gate study stops at the plant gate; a cradle-to-grave or well-to-wheels study, which fuel regulations use, includes distribution and combustion in the vehicle. For ethanol, the boundary includes corn farming (fertilizer production, nitrous oxide from soils, diesel for machinery, drying the grain), transport to the plant, the plant's energy and chemicals, distribution of the ethanol, and its combustion. The carbon dioxide from burning the ethanol, and from the fermentation, is biogenic: the corn took it from the air a few months earlier, so it is counted as zero. Methane and nitrous oxide from combustion are not biogenic in that sense and are counted.

Inventory analysis

The inventory is a mass and energy balance of the whole product system: every input of material and energy and every output of product, co-product, emission and waste, for each process in the boundary, scaled to the functional unit. Foreground data, from the processes the study owner controls, should be measured: the plant's own natural gas, electricity, chemicals, corn and products from its meters and records. Background data, for processes outside its control such as fertilizer production or the electricity grid, come from databases. The quality of an LCA depends heavily on the foreground data, and at Sarnia those are the same meters that feed the plant's mass balance.

Co-products and allocation

Most processes make more than one product, and the burden of the process must be shared among them. An ethanol plant makes about as much DDGS by mass as ethanol, plus corn oil. How much of the farming and plant emissions belongs to the ethanol? ISO 14044 gives a preference order. First, avoid allocation by dividing the process into sub-processes or by expanding the system to include the alternative production of the co-product (system expansion, or the displacement method): the DDGS replaces corn and soybean meal in animal feed, so the emissions avoided by not producing that feed are credited to the ethanol. If that is not possible, allocate by a physical relationship such as mass or energy content. Only as a last resort, allocate by another relationship, such as economic value.

# Carbon intensity of the plant's ethanol (g CO2e per MJ), with three ways of handling DDGS
LHV, L_per_t = 21.2, 393.0                  # MJ/L ethanol; L ethanol per tonne corn
MJ_t = LHV * L_per_t                        # MJ of ethanol per tonne of corn
GRID = 40.0                                 # g CO2e/kWh, Ontario grid (illustrative round value)
stages = {                                  # burdens before co-product treatment, g/MJ
 'corn farming (N2O, fertilizer, fuel)': 291e3 / MJ_t,
 'corn transport':                        6e3 / MJ_t,
 'plant natural gas (7.5 MJ/L)':          7.5 * 66.0 / LHV,
 'plant electricity (0.20 kWh/L)':        0.20 * GRID / LHV,
 'chemicals, enzymes and yeast':          2.0,
 'land-use change (model value)':         7.6,
}
after = {'ethanol distribution': 1.5, 'tailpipe CH4 and N2O': 0.8}
gate = sum(stages.values())
for k, v in {**stages, **after}.items(): print(f"{k:38s}{v:6.1f}")
ddgs_t, eth_t = 18.0, 18.8                 # t/h
credit = -(ddgs_t / 60.6) * 400e3 / MJ_t - 1.0   # displacement: DDGS 400 kg/t, corn oil 1 g/MJ
ci_disp = gate + credit + sum(after.values())
shares = {'mass': eth_t / (eth_t + ddgs_t),
          'energy': eth_t * 26.8 / (eth_t * 26.8 + ddgs_t * 19.0),
          'market value': eth_t * 700 / (eth_t * 700 + ddgs_t * 230)}
print(f"{'displacement (system expansion)':38s}{ci_disp:6.1f}")
for k, f in shares.items():
    print(f"{k + ' allocation':38s}{gate * f + sum(after.values()):6.1f}   (ethanol share {f:.0%})")
print(f"reduction from gasoline at 93 g/MJ, displacement case: {100*(1-ci_disp/93):.0f} %")

corn farming (N2O, fertilizer, fuel)    34.9
corn transport                           0.7
plant natural gas (7.5 MJ/L)            23.3
plant electricity (0.20 kWh/L)           0.4
chemicals, enzymes and yeast             2.0
land-use change (model value)            7.6
ethanol distribution                     1.5
tailpipe CH4 and N2O                     0.8
displacement (system expansion)         56.0
mass allocation                         37.5   (ethanol share 51%)
energy allocation                       43.4   (ethanol share 60%)
market value allocation                 54.8   (ethanol share 76%)
reduction from gasoline at 93 g/MJ, displacement case: 40 %

The choice of method moves the answer a long way. With displacement the plant's ethanol comes out at about 56 g CO₂e/MJ. Mass allocation, which gives the DDGS almost half the burden, gives nearly 20 g/MJ less. Market value allocation, which gives ethanol about three-quarters of the burden because it sells for much more per tonne, comes out close to the displacement result. None of these is wrong; they answer slightly different questions. What matters is that the method is stated, justified and applied consistently, and that regulatory numbers use the method the regulation prescribes. Fuel carbon intensity models generally use displacement for distillers grains.

Reading the result

Two stages dominate. Corn farming contributes about 35 g/MJ, more than half of the total before the co-product credit. Most of that is nitrous oxide released from soils as nitrogen fertilizer is transformed by microbes, plus the energy-intensive production of ammonia fertilizer. The plant's natural gas, for steam and for the dryers, contributes about 23 g/MJ. Electricity adds less than 1 g/MJ, with the Ontario grid at about 40 g CO₂e/kWh. Land-use change, explained below, adds a modelled value of a few grams. The DDGS and corn oil credit removes about 15 g/MJ. The total, about 56 g/MJ, is about 40 % below gasoline's roughly 93 g/MJ on this basis.

Theo notices what the plant can and cannot change. Farming emissions are outside the fence, though the plant can influence them through the corn it buys, and both fuel programs have been developing ways to credit low-carbon farming practices, measured field by field. Natural gas is inside the fence and is the largest lever the plant controls directly: every megajoule of gas saved per litre of ethanol lowers the carbon intensity by about 3 g/MJ. And there is one large stream the LCA treats as zero because it is biogenic: the fermentation carbon dioxide. If it were captured and stored permanently instead of released, the ethanol would be credited with removing it from the atmosphere. Lesson 11 shows that this is the single largest change available to the plant.

Land-use change

When demand for corn rises, farmers may plant more of it, and somewhere in the world land may be converted from forest or grassland to cropland, releasing carbon stored in soil and vegetation. Direct land-use change is conversion on the land that grew the crop; indirect land-use change (ILUC) is the global market response, estimated with economic models. ILUC estimates for corn ethanol have varied widely between models and over time, from a few to more than 30 g/MJ, and they remain the most debated number in biofuel LCA. Programs treat it differently.

Theo's calculation includes a modelled land-use change value of 7.6 g/MJ, a round value used for illustration. The Fuel LCA Model is the reference tool for the Clean Fuel Regulations, and the plant's filed carbon intensity follows its methods and default values; Wren always checks the current version of the model and its guidance, because both are updated.

Impact assessment

Life-cycle impact assessment converts the inventory into impact categories. Climate change is the best known: each greenhouse gas is multiplied by its global warming potential over 100 years (GWP100), which is 1 for carbon dioxide and roughly 28 to 30 for fossil methane and 265 to 273 for nitrous oxide in recent IPCC assessments, and the results are summed as CO₂ equivalents. Other categories include acidification (from sulfur dioxide, nitrogen oxides and ammonia), eutrophication (from nitrogen and phosphorus reaching water), water use, land use, particulate formation and toxicity. A fuel regulation looks only at climate change, but a full LCA looks wider, because corn ethanol scores well on fossil carbon and less well on eutrophication and land use than gasoline. Wren insists that the board see the whole picture, not just the number that earns credits.

Interpretation, uncertainty and critical review

The interpretation phase checks the results for completeness, sensitivity and consistency, and draws conclusions within the limits of the data. Sensitivity analysis is essential: the plant's carbon intensity moves by several grams with the nitrous oxide emission factor, the DDGS credit, the land-use change value and the grid factor, each of which is uncertain. ISO 14044 requires a critical review by an independent expert or panel when an LCA is used to support a comparative assertion disclosed to the public, such as a claim that one fuel is better than another. Wren plans an external review of the expansion LCA for exactly that reason; the plant intends to tell its customers that the expanded plant's ethanol has a lower carbon intensity, and that claim must stand up.

Using LCA in design

The most useful LCA is the one done before the plant is built, because it can still change the design. Theo builds a simple version of the model into a spreadsheet so the design team can test each option of the expansion as it is proposed: heat integration and vapour recompression on the evaporators and stills, recovering more heat in the RTO, burning biogas, buying renewable electricity, and capturing the fermentation carbon dioxide. Each option shows up as a change in g/MJ. Lesson 12 will put a cost beside each one, and the combination becomes the carbon reduction half of the board's plan.