Upload
others
View
0
Download
0
Embed Size (px)
Citation preview
Manuscript Accepted January 22, 2013
Funding Acknowledgement: The David and Lucile Packard Foundation and the U.S. Department of Energy (Award# DE-EE0004396)
Motivation
• The U.S. has chosen to pursue a renewable fuel standard mandating 36 billion gallons of ethanol equivalent biofuels (EISA, 2007) – 16 billion gallons from cellulosic feedstocks
• Avoid food-vs-fuel concerns associated with grain ethanol • Potential environmental benefits
• Candidate feedstocks: Corn stover, perennial grasses, fast-growing tree species – Stover does not require additional land or new crops
• 60% greenhouse gas reduction requirement, but no other environmental performance requirements
Previous Work (1st and 2nd Gen. Biofuels)
• Life cycle assessment and greenhouse gases: Kim and Dale (2008), Kim et al. (2009), Fargione et al. (2008)
• Water resource impacts: Tilman et al. (2009), Secchi et al. (2010), Cibin et al. (2011)
• Limited amount of work considers economic and environmental dimensions together – Hill et al. (2009) and Searchinger et al. (2008) – ILUC work using GTAP, MIT models
• Gap: Cross-media (water and air) environmental impacts and economic cost
Research Questions
• Do GHG benefits and biofuels come at the expense of water quality?
• How are water quality and GHG emissions from soils related to stover removal?
• How are costs of stover collection and pollution related?
• Does purely economic optimization to supply biomass result in significantly different environmental outcomes than joint minimization of cost and pollution?
Cibin, et al. in Hydrological Processes Volume 26, Issue 11, pages 1629-1641, 28 SEP 2011 DOI: 10.1002/hyp.8280 http://onlinelibrary.wiley.com/doi/10.1002/hyp.8280/full#hyp8280-fig-0001
Study location: Wildcat Creek Watershed
Integrated modeling and optimization framework (Reeling and Gramig, 2012)
Baseline + 8 stover removal practices
SWAT
DAYCENT
Practice cost budgets
Soils, weather
INPUTS MODELS OUTPUTS
Yield, nitrate, total P, sediment [biomass or pollutant/ha]
GHG flux, CO2 and N20 [Mg CO2-e/ha]
[$/ha]
Multi-objective Optimization
Trade-off frontiers: • Cost-yield • Cost-pollutant loading
Water quality modeling
Soil and Water Assessment Tool (SWAT)
• Watershed divided into sub-basins
• Sub-basins divided into Hydrological Response Units which are ≅ farms
• Outputs: – Biomass yields
– Nitrate
– Total P
– Sediment
Greenhouse gas modeling
DayCent biogeochemical model
• Daily time-step version of the CENTURY model
• Soil carbon gain/loss and N20 flux
• Field-scale simulations based on soils, weather, crops, mgt (same as SWAT) – Simulations run for 41 soil types representing
majority watershed
– Area-weighted (by soil type) emissions of each practice on each HRU calculated from results
Economic modeling of practice costs
• The cost of each practice is calculated as the cost difference relative to the baseline – Baseline = Conventional tillage corn followed by
no-till soybeans
• Costs include: tillage, herbicide, nutrient (N) replacement, stover harvest+storage, average transportation, and cost of any yield change
• Cost of each practice considered is based on the area of the HRU Practice costs = ($/ha * total hectares) Cost of yield change = (Corn price)*(Δ yield/ha)*(total ha)
Scenario
Abbreviationa
Crop Rotation
(year1-year2) Tillageb
What is harvested in each
scenario
Increased annual cost of corn
production over baselinec
$/ha $/Mg Stoverd
CB (Baseline) Corn-Soybean Yes Grain only (no stover removal) $0 No stover removal
CB38 Corn-Soybean Yes Grain + 38% stover removal $61.55 $13.71
CB52 Corn-Soybean Yes Grain + 52% stover removal $92.76 $15.06
CC38 Continuous corn Yes Grain + 38% stover removal $194.70 $44.63
CC52 Continuous corn Yes Grain + 52% stover removal $259.47 $43.24
CBNT38 Corn-Soybean None Grain + 38% stover removal $64.79 $14.51
CBNT52 Corn-Soybean None Grain + 52% stover removal $95.42 $15.55
CCNT38 Continuous corn None Grain + 38% stover removal $163.35 $37.24
CCNT52 Continuous corn None Grain + 52% stover removal $228.13 $37.78
Stover removal modeling scenarios and costs per hectare
Stover removal rates, nutrient replacement rates, and cost assumptions adapted from Brechbill, Tyner and Ileleji (2011)
Scenario Abbreviationa
Average stover
yield in years when collected
(10yr avg for CB),
Mg/ha/yr b
Total harvested
biomass with N replacement,
Mg/ha c
Average annual pollutant contribution d
Nitrate,
kg/ha
Total
Phosphorus,
kg/ha
Sediment,
Mg/ha
Greenhouse gas
emissions from
cropland, CO2-ee
Mg/ha
CB (Baseline) No stover removal 10.07 8.49 1.83 2.23 5.98
CB38 4.490 (2.25) 14.56 8.38 1.74 2.29 6.25
CB52 6.16 (3.08) 16.24 8.24 1.73 2.35 6.40
CCf No stover removal 10.15 10.41 1.71 2.27 6.53
CC38 4.363 14.52 9.47 1.65 2.41 7.49
CC52 6.000 16.15 9.16 1.65 2.51 7.90
CBNTf No stover removal 10.04 8.77 2.39 2.13 5.87
CBNT38 4.465 (2.23) 14.50 8.57 2.23 2.22 6.06
CBNT52 6.136 (3.07) 16.17 8.45 2.19 2.28 6.23
CCNTf No stover removal 10.12 11.32 2.41 2.02 6.22
CCNT38 4.387 14.51 10.16 2.02 2.08 6.94
CCNT52 6.038 16.16 9.76 1.90 2.14 7.28
Simulation results (1995-2009)
Optimization: A Heuristic Approach
• Normative analysis to inform what is possible given cost and pollution (or stover) criteria
• A single solution is an allocation of 1 stover removal practice to each of 922 ‘fields’ in the watershed
• Genetic algorithm used to search for solutions
– Discrete practices aggregated over watershed
– Multiple maxima/minima, so gradient-based approach will not work
Pollution loading (or stover yield)
= load of a pollutant or supply of feedstock j per hectare produced from practice γk implemented on land unit i
a vector containing the size data for all of the individual land units in the watershed
each stover collection cropping practice (8 scenarios)
= watershed-average per hectare pollutant load or feedstock yield for each of the four pollutants and the feedstock
Watershed average cost per hectare of crop production practice
= net additional cost per hectare of implementing γk on land unit i relative to the baseline corn-soybean rotation with no stover collection
a vector containing the size data for all of the individual land units in the watershed
each stover collection cropping practice (8 scenarios)
Multi-objective optimization: Joint minimization
•Unconstrained optimization yields a trade-off frontier, rather than a single solution •Frontier illustrates a
range of outcomes •Useful because can’t
actually dictate what is done…
•Perform five separate optimizations to jointly minimize…
1. Cost and (-)stover yield 2. Cost and nitrate 3. Cost and TP 4. Cost and TSS 5. Cost and GWP
“Minimize the sum of watershed-average per hectare pollutant load (or minus feedstock yield) and the cost per hectare of crop production practices”
Conventional tillage corn-soy w/ 38% removal (CB38)
Conventional tillage corn-soy w/ 52% removal (CB52)
Optimization results: min cost-max stover
Stover Production vs. Nitrate Loss
Stover Production vs. Total Phosphorus(TP) Loss
Stover Production vs. Total Suspended Sediment (TSS) Loss
Stover Production vs. Global Warming Potential (GWP)
Average Stover Production (Mg/ha) Average Stover Production (Mg/ha)
Average Stover Production (Mg/ha) Average Stover Production (Mg/ha)
Nit
rate
(kg
/ha)
TP (
kg/h
a)
GW
P (
kg/h
a)
TSS
(Mg
/ha)
Relationship between stover production and practice cost per hectare for all
multi-objective optimizations
Nitrate minimization overlaps heavily with stover maximization
Small range of stover yields spanned by points selected by other optimizations, but very wide range of costs from $63 to $178 per ha along TP and sediment curves
Nitrate-Cost outcomes from different optimizations
Nitrate and stover move together, as cost increases
Nitrate and GWP always go in “opposite” directions, as cost increases
Combined trade-off frontiers for all cost-pollutant multi-objective optimizations
Take-away points from combined trade-off frontier plot
Stover maximization or NO3 minimization reduce nitrate loading
GWP and TSS minimization
objectives increase TP
loading
All stover removal scenarios result in
higher soil GHG flux than baseline
All objectives that don’t minimize TSS, increase it
Conclusions
• Maximizing stover results in lower nitrate and TP, but higher sediment and GWP
• All stover collection practices considered increase contribution to GWP from crop land
• A move toward more continuous corn will lead to larger environmental impacts compared to corn-soybean rotation w/o any stover removal
Next steps and what have you
• Stover removal – “Correct” N replacement rate…?
– Cover crops or other GHG and/or erosion mitigating practice(s)
• Farmers’ behavioral response will not be that selected by optimizations…
• Conditions under which perennial crops would be grown to supply cellulosic biomass
• Constrained optimization, like economists are supposed to do!