BIOFUELS, INDIRECT LAND USE CHANGE, & LIFE CYCLE ANALYSIS: DO WE NOW KNOW ENOUGH TO KNOW THAT WE DON’T KNOW? Bruce E. Dale University Distinguished Professor of Chemical Engineering Michigan State University Presented at: Low Carbon Fuels Webinar July 25, 2008

My Assumptions/Points of Departure Inexpensive plant raw materials will catalyze the growth of new and existing biofuel industries– this is absolutely going to happen We have a unique opportunity to design these industries for better environmental performance One important tool: life cycle analysis (LCA) LCA has significant value if used properly, but it is a limited tool LCA exists to make comparisons…LCA should not be done in the ideal or the abstract

Life Cycle Assessment Framework Goal & and Scope Definition Inventory Analysis Impact Assessment Interpretation & Stakeholder Participation Direct applications : - Product development - Marketing and improvement - Strategic planning - Public policy formation - Other No stakeholder involvement in either Science paper

It ain’t a life cycle analysis just because someone says it is. An LCA study must meet certain standards EISA 2007: renewable fuels must meet certain “lifecycle greenhouse gas” emission reductions

Some Life Cycle Analysis Standards: In Plain English Use the most recent/most accurate data possible If models are used to generate “data”, have the models been sufficiently tested & verified? Select the reference system/functional unit: what exactly are we comparing? Make it easy for others to check your data and methods= transparency (difficult for complex models) Set clear system boundaries (physical & temporal)— must be equal or comparable for reference system and/or reference product of interest Multi-product systems must allocate environmental costs among all products Perform sensitivity analysis: how much do results vary if assumptions or data change?

The Policy Related Tasks Are: 1.Use life cycle analysis to… 2.Determine the greenhouse gas impacts of... 3.Direct land use changes…and 4.Indirect land use changes (ILUC) Do the two ILUC studies (February 2008 Science papers) meet commonly accepted LCA standards and thereby satisfy the policy requirements or do they not meet these standards?

Let’s Examine the Recent Papers in Science using these Criteria Use the most recent/most accurate data possible If models are used to generate “data”, have the models been sufficiently tested & verified? Select the reference system/functional unit: what exactly are we comparing? Make it easy for others to check your data and methods= transparency (difficult for complex models) Set clear system boundaries (physical & temporal)— must be equal or comparable for reference system and/or reference product of interest Multi-product systems must allocate environmental costs among all products Perform sensitivity analysis: how much do results vary if assumptions or data change?

Use the most recent & most accurate data possible Land clearing from the 1990s—not checked by either modeling or more recent data Ignores literature on causes of land use change Four linked submodels…no empirical data at all 1.Ethanol demand to corn price 2.Corn price to corn or soybean supply 3.Corn or soybean supply to land use change 4.Land use change to greenhouse gas consequences 5.Land management post land use change not considered-apparently only plow tillage used Sensitivity analyses were generally incomplete or lacking (Monte Carlo simulation is the standard) No confirmation of model predictions by: 1) empirical data, 2) other models, or 3) back testing. An unverified, untested model is simply a guess.

Select the reference system or functional unit: what exactly are we comparing? Ethanol vs. Gasoline? Corn ethanol vs. cellulosic ethanol vs. tar sands “oil” to gasoline? Gasoline produced how, when and from what? (oil shale, tar sands, heavy crude???) Backwards looking or forward looking (temporal boundaries)? Corn for ethanol vs. corn for animal feed? Allocation would help resolve feed vs. fuel uses of land…this was apparently not done in either analysis

Set clear system boundaries (physical & temporal)—must be comparable for reference product of interest 1.Ethanol temporal: future (forward looking) 2.Ethanol physical: entire world land (indirect effects on GHG considered) 3.Petroleum fuels (or other alternatives) temporal: past (GREET model) 4.Petroleum fuels physical: restricted (indirect effects on GHG not considered)

Multi-product systems must allocate environmental costs among all products 1.System is land use in the entire world 2.Land produces: Animal feed (roughly 10x direct human food use) Human food Biofuels Pulp, paper, lumber…and lots more 3.Searchinger, et al, paper apparently allocated the entire incremental land use change “cost” of biofuel production to the biofuel 4.Ignores the fact that the “replaced” agricultural production went to provide animal feed 5.Without allocation, these analyses advantage animal feed from land vs. biofuels production. 6.Could have/should have dealt with this allocation issue in the sensitivity analysis

Perform sensitivity analysis: how much do results vary if assumptions or data change? Productive use of existing forest: make furniture or flooring from the tropical hardwoods or were the trees just burned? Decreased land clearing rates and/or different ecosystems converted. What if most land converted is pasture? Historical rates of corn yield increase in the U.S. & abroad “Carbon debt” compared with Athabasca oil sands/Colorado oil shale/Venezuelan heavy crude GHG in 2015 vs. GREET in ~1999 Increasing efficiency of future ethanol plants Uncertainties in global equilibrium models…test through Monte Carlo simulation? Tested with data? Other models? Allocation of environmental burdens among feed and fuel uses of corn—(livestock are responsible for 18% of worldwide GHG emissions) How is land managed after conversion? These & other factors were not adequately addressed during sensitivity analysis

Do the 2008 Science Papers Meet LCA Criteria? Data quality. Use the most recent/most accurate data possible? No. Models may be valid but that was not proven. Literature on causes of land use change ignored? Select the reference system/functional unit: what exactly are we comparing? Marginal. Make it easy for others to check your data and methods= transparency Acceptable

Do the 2008 Science Papers Meet LCA Criteria? Set clear system boundaries—must be equal or comparable for reference system and/or reference product of interest No. Temporal boundaries & physical boundaries are not comparable for ethanol & gasoline Multi-product systems must allocate environmental costs among all products No. No apparent or stated allocation of these costs among animal feed and biofuels Perform sensitivity analysis: how much do results vary if assumptions or data change? No. Sensitivity analysis lacked appropriate range of variables, especially for allocation No apparent stakeholder involvement

So What is My Bottom Line? GHG effects of direct land use change for biofuels (supply chain oriented) can and have been studied by LCA. Robust conclusions by LCA standards (+/- 30%) GHG effects of indirect land use change (market oriented) have not yet been successfully studied by LCA. Science papers are not (and probably were not intended to be) LCA studies. Existing ILUC papers do not meet the standards for “life cycle” studies. It is simply incorrect to use them as such. The system is so complex that it may never be possible to apply recognized LCA standards to ILUC (but that shouldn’t stop us from trying)

Land Management Post Land Use Change: Some Insights 1.Ethanol demand to corn price 2.Corn price to corn or soybean supply 3.Corn or soybean supply to land use change 4.Land use change to greenhouse gas consequences 5.Land management post land use change Land doesn’t cease to be managed once the land use change is executed. What are the GHG consequences of post land change management options?

Scenario*Description AConvert grassland to cornfield dedicated to ethanol production B Divert cornfield to ethanol production, Convert grassland to cornfield dedicated to animal feed production C Convert corn-soybean rotation to cornfield dedicated to ethanol production Convert grassland to corn-soybean rotation DConvert forest to cornfield dedicated to ethanol production E Divert cornfield to ethanol production, Convert forest to cornfield dedicated to animal feed production F Convert corn-soybean rotation to cornfield dedicated to ethanol production Convert forest to corn-soybean rotation Land Management Post Land Use Change: Tillage Practices & Cover Crops * Data for DAYCENT from 8 U. S. corn producing counties, different climates, etc.

Current tillagePlowing tillage No tillageCover crop

A Path Forward for LCA and ILUC? GTAP deals only with land for which rents are established: cropland, pasture & commercial forest (not Amazon rainforest). Abandoned and CRP lands are not in the model Expand GTAP (or related models) to include abandoned land (1 billion acres world wide) & CRP lands here Pasture (grassland) conversions do not seem to incur much “carbon debt” they may in fact get quickly to carbon credit Focus “carbon debt” analysis on forests (& savannah?) Use common tool (DAYCENT?) to model forest conversion and post conversion management (standing biomass?) Three forest issues are: 1) commercial forest, 2) non- commercial forests & reduced reforestation rates –If U.S. commercial forests, we can track & discourage conversion using specific policy instruments-not ILUC blunt force trauma –Carbon sequestration “lost” by reduced reforestation occurs over time & can be estimated if reliable reforestation rates are known –For non commercial forests, an academic land use literature exists but apparently has not been used in analysis to date: agricultural expansion is only one of several driving factors. Allocation?

One Tree (Study) Doth Not a Forest (Conclusion) Make In science, one or two studies are never enough to establish the facts. S&F papers began an important conversation Some key results of further (other) studies thus far: –Forests matter in “carbon debt”, grasslands may give “carbon credit” –Forest conversions driven by combined forces: agricultural expansion + timber utilization + road access explain 96% of observed cases but any single factor explains less than 20% –Land management post land change really affects GHG results –One billion acres of unused/abandoned land worldwide, not considered in S&F (nor were CRP lands) –Models relatively untested & do not validate each other well. Searchinger predicts land conversion in Latin America, China, U.S., etc Purdue (GTAP) predicts most conversion in U.S. (2/3 commercial forest) GTAP predicts ethanol expansion to date should have caused conversion of 2 million acres of forest. Did that happen? Duke (FASOMGHG) predicts mostly CRP & pasture conversion in U.S. –If FASOM is more correct, then ILUC may produce a carbon credit I believe we now know enough to know that we don’t know the sign (positive or negative) of ILUC, let alone its magnitude.

Questions ??