35 Table 5: Ecoinvent data v2.0 contents and data generators.

Sector Database content Data generator

Energy Hard coal Paul Scherrer Institute

Oil ESU-services Ltd.

Natural gas ESU-services Ltd., Paul Scherrer Institute Nuclear power Paul Scherrer Institute

Hydroelectric power Paul Scherrer Institute

Wood energy Paul Scherrer Institute

Wind power Paul Scherrer Institute

Photovoltaics ESU-services Ltd.

Solar heat ESU-services Ltd.

Electricity supply and mixes ESU-services Ltd., Paul Scherrer Institute Small scale CHP systems Basler & Hofmann

Biofuels ESU-services Ltd., Carbotech, ENERS,

ETHZUNS1,Infras, LASEN/EPFL, Paul Scherrer Institute,Umwelt- und Kompostberatung

Materials Building materials Empa2, Bau- und Umweltchemie, ESU-services Ltd.

Metals Empa2, ESU-services Ltd.

Plastics Empa2

Paper and Board Empa2

Renewable materials Wood Empa2

Tropical wood Dr. Frank Werner Environment and Development

Renewable fibres Carbotech

Chemicals Basic Chemicals ETHZ-ICB3, Empa2, Chudacoff Ökoscience, ESUservicesLtd.

Petrochemical solvents ETHZ-ICB3

Detergents Empa2

Transport Transport services Paul Scherrer Institute, ESU-services Ltd. Waste management Waste treatment services Doka Life Cycle Assessments

Agriculture Agricultural products and processes ART4, Carbotech, ETHZ-ICB3

Electronics Electronics Empa2

Mechanical engineering Metals processing and compressed air ESU-services Ltd

1 Institute for Environmental Decisions, Natural and Social Science Interface, Swiss Federal Institute of Technology,

Zurich (ETHZ)

2 Swiss Federal Laboratories for Materials Testing and Research

3 Institute for Chemical and Bioengineering, Safety and Environmental Technology Group, Swiss Federal Institute

of Technology Zurich (ETHZ)

4 Agroscope Reckenholz-Tänikon Research Station, Life Cycle Assessment group

Ecoinvent datasets often serve as background data in specific LCA studies. The LCI and LCIA results of Ecoinvent datasets should not directly be compared with the aim to identify

environmentally preferable products or services. For comparative assessments, problem- and case-specific particularities need to be taken into account. Inventory data are in most cases collected on the level of national averages. Hence, no regional differentiation can be made. Data sources are assessed according to the six characteristics "reliability", "completeness", "temporal correlation", "geographic correlation", "further technological correlation" and "sample size." Each characteristic is divided into five quality levels with a score between 1 and 5. Accordingly, a set of six indicator scores is attributed to each individual input and output flow (except

reference product) reported in a data source (this set of six indicator scores is reported in the general comment field of each input and output). An uncertainty factor (expressed as a

V-36

contribution to the square of the geometric standard deviation) is attributed to each of the score of the six characteristics.

For well to fuel LCI, Ecoinvent’s cumulative LCI with maximum, mean, and minimum values are used. Ecoinvent provides high and low estimates for its inventory resource and emission data. In an effort to provided greater transparency with regard to likely variation of inputs and thus impact estimates, we present findings for all available LCI levels for each fuel pathway. Data considerations for each pathway are as follows:

Conventional Gasoline

For LCI of conventional gasoline, Ecoinvent’s unleaded gasoline produced by an average refining technology in Europe is employed. The inventory data is created based on in-site

specific data of 100 refineries in Europe (Jungbluth N. et al., 2007). Given the limited time frame of the study, and an effort to create US or MN specific LCI data, this relationship is assumed to represent domestic refining technology, similar to refining available in the US and distributed to local pumping stations in MN. The assumption might not be reasonable to represent the US averaged or MN specific refining. Approximately 83% of the total percentage of crude oil consumed within Minnesota is refined from Canadian oil sands. (GHG section of this report). However, the goal of our LCIA study to focus on not compiling whole LCI for fuel pathways but to develop a modeling analytical framework to serve as a means for further evaluation of LCFS options. Data was not available to characterize all emissions with MN specified characterization factors and identifying the significance of the impacts of all emissions within the state of MN. The framework included are all processes on the refinery site, all resources/emissions in all upstream processes and waste treatments, process emissions and direct discharges from refinery site to rivers. Excluded are the emissions from combustion facilities as GREET models finished gasoline products that are delivered to regional bulk terminal and thereafter distributed to regional fuel pumping station. The framework used GREET’s direct emissions from

transportation and the distance from refinery to bulk terminal and from bulk terminal to regional fuel pumping station. Process flow and system boundary of gasoline production is shown in process flow diagram in Figure 1.

Corn Ethanol

For corn ethanol LCI analysis, ‘ethanol, 99.7% in H2O, from biomass, at distillation’ is selected for cumulated LCI until distillery. Processes within system boundary for corn cultivation and corn ethanol production are in the US context. Ecoinvent includes the transport of corn grains to the distillery, and the processing of corn grains to hydrated ethanol (95%) and DDGS (92% dry matter). System boundary is at the distillery. The process described corresponds to the dry- milling technology. Hydrated ethanol input is corn-based ethanol, produced in the US context. The ratio of hydrated to anhydrous (wet basis) is equal to 0.997/0.95, i.e. 1.05 kg hydrated ethanol per kg of anhydrous ethanol. On a dry matter basis, the input of hydrated ethanol 95% is 1 kg per kg of anhydrous ethanol 99.7%. The energy use for the dehydration process are

electricity (8.8kWh) and steam (1002 MJ) per ton of anhydrous ethanol (Ecoinvent, 2007). The treatment of waste streams is also included.

V-37

Ecoinvent’s corn ethanol life cycle inventory database includes all resources and emissions produced from upstream processes and foreground processes are incorporated with GREET direct emissions data for fuel combustion for vehicle operation. The life cycle stages within corn ethanol production system are ‘corn cultivation and harvest at farm,’ ‘corn feedstock

transportation,’ ‘corn fermentation, hydrated ethanol production, and dehydration of hydrated ethanol at bio-refinery,’ and ‘corn ethanol combustion in vehicle operation,’ as shown in Figure 3. Even though the life cycle emissions inventory is not clearly separated along each process, the cumulated emissions inventory through entire life cycle stages are available for corn ethanol system under this study. It is assumed that regional fuel pumping stations are located within 30- mile distance from ethanol refinery.

The benefit of using corn ethanol as renewable fuels is seen primarily at the end use (tailpipe or fuel combustion in vehicle operation). During the corn cultivation at farm, corn absorbs CO2

from atmosphere (1.35 kg CO2 per kg corn fresh matter) to produce carbon required for corn

growth. Tailpipe CO2 emissions are generally calculated based on the assumption that the CO2

uptake from farm field equals to the tailpipe CO2 emissions so that the biogenic carbon emitted is

offset by the CO2 uptake resulting from corn growth. Thus, the amount of CO2 uptake is

subtracted from ethanol combustion stage. However, the combustion CO2 emission in our

GREET model result is 1.91 kg/ kg of corn ethanol while CO2 uptake of corn is 1.35 kg/kg of

corn ethanol based on carbon balance. Conventional Diesel

Life cycle inventory of conventional diesel production complied from Eco-invent. Due to time restriction to investigate MN-specific diesel production, life cycle inventory of diesel production system had to rely on Ecoinvent’s European average diesel production. Even though it is based on European context, for the purposes of developing an analytical modeling framework it was assumed that there is no significant variation in diesel production systems between U.S., Minnesota and Europe. This allowed the significant impacts of all emissions emitted due to production of conventional diesel to be identified. Ecoinvents system boundary of the production system includes: oil field extraction, crude oil production, long distance

transportation, and diesel refining;l processes at the refinery site including wastewater treatment, process emissions and direct discharges to rivers, and an environmental inventory that includes all emissions associated with upstream processes of production of resources, input products and energy used in the processes. It is important to understand, however, that although considered appropriate for developing a modeling analytical framework that can provide the means for comparing environmental impacts, the data inputs used were not specific to Minnesota.

Approximately 83% of the total percentage of crude oil consumed within Minnesota is refined from Canadian oil sands. (GHG section of this report).

As was outlined in the literature review, the choices available for developing a modeling

analytical framework for use as LCFS LCIA are evolving rapidly. Over the course of the project modeling options available in 2011 are different than those available when choices were

evaluated in 2009. A similar evolution is occurring regarding data.

Improved data availability of petroleum processes is needed to reduce geopgraphic assumption beyond typical energy source adjustments when applying Ecoinvent data in a US setting.

V-38