CHEMICAL ENGINEERING TRANSACTIONS  
 

VOL. 61, 2017 

A publication of 

 

The Italian Association 
of Chemical Engineering 
Online at www.aidic.it/cet 

Guest Editors: Petar S Varbanov, Rongxin Su, Hon Loong Lam, Xia Liu, Jiří J Klemeš 
Copyright © 2017, AIDIC Servizi S.r.l. 

ISBN 978-88-95608-51-8; ISSN 2283-9216 

Managing Land Use Change Greenhouse Gas Emissions of 

Bioethanol Production 

Daniel J. Garciaa, Fengqi You*b 

aNorthwestern University, 2145 Sheridan Road, Evanston, Illinois 60208, USA 
bCornell Univesrity, 318 Olin Hall, Ithaca, New York 14853, USA 

 fengqi.you@cornell.edu 

Indirect land use change (iLUC) must be considered for bioconversion technologies. These technologies, such 

as corn-to-ethanol technologies, require land not only for the processing facility but also indirectly require land 

for the cultivation of the input biomass. There are greenhouse gas (GHG) emissions associated with the 

transformation of the required land, such as forests or pastures, into cropland for the input biomass. iLUC 

emissions and effects have been examined on a cursory level in the literature, but deeper analysis is needed. 

We present an iLUC life cycle optimization model that integrates cutting-edge iLUC modeling techniques with 

optimization methodology. The model minimizes total GHG emissions, composed of direct and indirect 

processing emissions as well as iLUC emissions of the chosen biofuel production strategy. A case study 

considering production of 37.9 BL/y of bioethanol in the U.S. illustrates that iLUC emissions contribute to the life 

cycle GHG emissions of biofuels, even in optimal scenarios. The minimum realizable iLUC emissions rate is 1.5 

g CO2-eq/MJ at this production level using corn stover as a feedstock. 

1. Introduction and Background 

Biofuels are an important component of a renewable energy strategy (Yue et al., 2014). Life cycle greenhouse 

gas (GHG) emissions of biofuels are potentially lower than those of petroleum-based fuels (You et al., 2012). 

However, important drawbacks exist with biofuels (Alwi et al., 2016). Some biofuels cause indirect land use 

change (iLUC), and these concerns guide feedstock selection as well as biofuel process selection and design 

(Gong and You, 2015). Emissions due to LUC might severely diminish the environmental benefit gained from 

biofuels over conventional fuels (Plevin et al., 2015). Calculating the LUC emissions of biofuels is an active area 

of research. Many researchers present wide ranges for LUC emissions of certain biofuels. Recent estimates for 

LUC emissions of corn ethanol range from 7.6 (Canter et al., 2015) to 53 g CO2-eq/MJ (Plevin et al., 2015). 

These estimates guide global energy legislation. Minimizing the uncertainty involved in LUC emissions of 

biofuels by establishing a lowest realizable emission rate aids policymakers, industrialists, and scientists 

(Santibañez-Aguilar et al., 2015). Minimizing life cycle GHG emissions of biofuels must also include these 

emissions (Yue et al., 2013). iLUC emissions characterization and estimation methods are now quite advanced. 

The most sophisticated and advanced methods for estimating iLUC emissions from biofuels are based on 

computable general equilibrium (CGE) models. CGE models are highly detailed economic models of a large-

scale economy. Transactions and taxes between regions, firms, households, and governments in a number of 

economic sectors are accounted for. Price elasticities of commodities and factors of production are also in the 

model. Users of a CGE model can “shock” the initial, stable economic equilibrium by changing some variables, 

for example, increasing biofuel production in the US Solving the model with the shock means searching for a 

new equilibrium state. Economic equilibrium occurs when there are zero profits, zero surpluses, zero shortfalls, 

and zero losses among all agents in the economy and when market prices of all goods equal their production 

costs. A CGE model is useful when characterizing and estimating iLUC effects. One can identify the initial land 

type and use distributions, introduce the shock, solve for the new equilibrium, determine the new land type and 

use distributions, and infer what changes must have happened. Next, land carbon stocks and soil organic carbon 

(SOC) levels combined with the land use changes can be used to calculate LUC emissions.  

                               
 
 

 

 
   

                                                  
DOI: 10.3303/CET1761025

 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 

Please cite this article as: Garcia D.J., You F., 2017, Managing land use change greenhouse gas emissions of bioethanol production, Chemical 
Engineering Transactions, 61, 163-168  DOI:10.3303/CET1761025  

163



The most advanced iLUC characterization models use the Global Trade Analysis Project (GTAP) CGE model 

and data. The GTAP model is one of the most advanced CGE models of the global economy (Hertel, 1997). 

Taheripour et al. (2007) pioneered its use for calculating iLUC emissions of biofuels using computable general 

equilibrium (CGE) models (Taheripour et al., 2007). The Carbon Calculator for Land Use Change from Biofuels 

Production (CCLUB), developed by the U.S. DOE (Dunn et al., 2013), and an approach centred on a Net 

Displacement Factor (NDF) and IPCC carbon stocks data developed by Plevin et al. (Plevin et al., 2015) later 

built on this foundation and added other biofuels and feedstocks. Thanks to these advances, the uncertainty 

gap for the iLUC emissions rate of biofuels has narrowed. However, this type of method cannot determine an 

optimal biofuel production strategy that limits iLUC emissions. It can only determine the iLUC emissions for a 

given biofuel production scenario. Additional economic sectors must be added for each new biomass feedstock 

and ethanol production method. This requirement necessitates a variety of economic assumptions about the 

new sectors, increasing model uncertainties and computational burdens. We integrate cutting-edge iLUC 

calculation methodology into a life cycle optimization (LCO) framework (You and Wang, 2011), in order to 

produce an LUC LCO framework with 5 possible biomass feedstocks and 14 ethanol production options based 

on an existing bioconversion network (Garcia and You, 2015). This new framework allows us to circumvent 

previous uncertainties in biofuel iLUC emissions studies by integrating all equations and data into a single model. 

A minimum iLUC emissions rate from biofuels strategy is determined within the context of the global economy. 

The model also sets a lower bound on the iLUC emissions rate of biofuels. These new capabilities allow for 

more informed, targeted biofuel production policies and analysis of the life cycle emissions of biofuels. In the 

next section, we discuss the model’s formulation. We then present a case study where the U.S. produces 37.9 

BL of ethanol per year. We discuss the results of this case study and state the conclusions.  

2. Problem Statement 

The authors modeled a bioethanol production quota in the U.S. of 37.9 BL/y to be met with any biomass 

feedstock and any production pathway. Life cycle emissions, including LUC emissions, direct processing 

emissions, and indirect processing emissions are to be minimized. We assume that all biomass feedstock must 

be produced on new cropland (except stover), no region may more than double its biomass production, LUCs 

occur immediately, regions cannot create/lose land, biomass yields on new croplands start at the same level as 

previous yields in that region, only pasture and forest may be converted to cropland, and new biomass cannot 

be grown in regions where it was not initially grown. Major decision variables include which of the fourteen 

biofuel production pathway(s) to utilize, which of the 5 kinds of biomass to use, where to source the biomass for 

biofuel from the 19 regions, what kind of land to convert, and how much of each type of land to convert in each 

region. GTAP variables include commodity and factor prices, factor transformations, regional account balances, 

regional output as well as commodity export and import rates. Figure 1Error! Reference source not found. 

shows a visualization of this problem statement. 

3. Methodology 

Integrating an LCO framework with the GTAP model requires a variety of tasks and datasets. We briefly cover 

here GTAP, land use, and land value data, and the integration of a LUC LCO model based on GTAP into an 

optimization framework. Data on the different production pathways were obtained from various peer-reviewed 

literature (Gong et al., 2016). The production pathways chosen are as follows: for corn, traditional dry milling 

and wet milling to ethanol (Accardi et al., 2015); for sugarcane, the traditional process. For stover, the 

pretreatment methods differed: dilute acid (DA) pretreatment, two-stage DA pretreatment, AFEX pretreatment, 

and hot water pretreatment to distillation were considered (Humbird et al., 2011). Two more stover pathways of 

DA pretreatment to pervaporation and DA pretreatment with separate C5/C6 fermentation to distillation are also 

included (Kazi et al., 2010).  

 

Figure 1. Visualisation of the problem statement 

164



DA pretreatment of switchgrass to distillation is considered. Gasification of wood chips to mixed alcohol 

synthesis or acetic acid synthesis and hydrogenation round out the pathway options. The authors used the 

seventh version of the GTAP database which contains data for 112 regions and 57 sectors from 2004. 73 model 

equations comprise GTAP. The authors do not list them all here. They can be found in (Hertel, 1997). Many of 

the equations involve bilinear and/or nonconvex functions. For example, consider the unit cost function for 

endowment commodities f for sector g in region r: 

 
1
11

, ,f

gr fgr fgr
f

CF PPF g r



 

   
 
  (1) 

CFg rand PPFfgr are variables, and η is a price elasticity factor that can range anywhere between zero and a 

large positive number. The above nonlinear and nonconvex equation is a typical CGE model equation. Other 

unit cost functions calculate the unit cost of a bundle of domestic and imported commodities within a region 

where multiple commodity price variables are multiplied together, creating bilinear or trilinear terms. Integration 

of CGE equations will result in a large-scale nonlinear programming (NLP) problem. 

GTAP data is constantly updated, and users augment the model with data relevant to what they wish to study, 

but the core model equations do not change. The authors aggregated all data into 19 regions and 26 sectors, 

following the aggregation routine of previous iLUC research (Taheripour et al., 2007). Some regions include the 

U.S., EU-27, Brazil, etc., and some sectors include cereal grains, chemicals, sugarcane and sugar beet, etc. 

The authors followed Rutherford’s (2006) method to integrate the GTAP model into an optimization model. 

Following previous iLUC research, we assume ethanol produced from biomass will be part of the “chemicals, 

rubber, plastic products” sector. Integration of the GTAP database and model equations into an LCO framework 

involved data collection, modifying the core GTAP equations and data, and formulating a LUC calculation 

module that can be seamlessly  

 

Figure 2: Visualization of the integration of GTAP and LCO modelling frameworks 

integrated into the GTAP model. The core GTAP database has a single “Land” factor of production and does 

not include data related to land use, such as land cover, land type distribution, and economic values of different 

kinds of land. These data were taken from a number of sources, and integrated into the model. Land is divided 

into 18 Agro-Ecological Zones (AEZ) and 7 land types. The land types considered in this work include cropland, 

forest, pastureland, shrub-land, savannah/grassland, built-up land, and other unusable land. We expanded the 

core parameters to account for the 18 new AEZs instead of just one land factor of production. We also obtained 

data on the total cultivable land in each region. 

An iLUC LCO model was constructed to be integrated with the GTAP model, formulated below. A visualization 
of how the models interact is in Figure 2. Capitalized terms are variables, lowercase terms are parameters: 

min IP+DP+LP  

s.t.   , , ,lfr lfrj
j

TNL NL l f r  
(2) 

, , ,lfr lfrj
j

TLL LL l f r   
(3) 

, ,lfr lfr

l l

TLL TNL f r    (4) 

, ,lfr lfr

l l

TLL ld f r    (5) 

  ,lfr lfr r
cropland f

TNL TLL tcl r     
(6) 

165



, , , ,cj frcx cr crf frcx cr crf lfrj cj

c x corn x

cs C N iby C N ddgs iby NL cs f r j l cropland
 
           
 
   (7) 

r frcx cx

r f r c x

G C py    (8) 

r

r

G fdem  (9) 

0.33 , ,rfcx rf rfcx crf

x stover x corn

C sm C iby r f
 

     
 

    (10) 

r gr
r crp r

G fpr E    
(11) 

 1 1 1 , ,cr gr r

g

N P ype c r
 

     
 

  (12) 

    1 1 (1 )lfr lfr lfr lfr lfr lfr lfr lfr lfr fr

l f r

CE TLL soc slf soc slf sslf sslf unc TNL npp cef
 

             
 
  (13) 

 1lfr lfr lfr

l f r

NE TLL soc slf cnf nef
 

      
 
  (14) 

lfr lfr

Forest f r

FE TLL fg cef
 

   
 
   (15) 

  1 , , , ,
fr

lfrj lfr lj fr fjr frNL ld slm FT PS PF l f r j


       (16) 

  1 , , , ,
fr

lfrj lfr lj fr fjr frLL ld slm FT PS PF l f r j


        (17) 

  , ,gr gr gr gr r Cr Gr Ir igr irs ir
i g s

Y vom P E RA P vom vom DDFM DXMD dst g r


            
(18) 

 LP CE NE FE tf    (19) 

rfcx x

r f c x

IP C ipe   (20) 

rfcx x

r f c x

DP C dpe   (21) 

where set I,g,j represent GTAP commodities/sectors, l is the set of land types, f is the set of AEZs, r is the set 

of regions, c is the set of biomass feedstocks, and x represents the set of ethanol production pathways.IP 

represents the indirect processing emissions, DP the direct processing emissions, and LP the LUC emissions. 

TNLlfr represents the total new land, and TLLlfr represents the total lost land. NLlfrj is the amount of new land 

used by sector j, and LLlfrj is the amount of lost land that is no longer available to sector j. ldlfris the original land 

distribution, tclr is the total cultivable land of each region, cscj is an indicator that ties crop c to sector j, Cfrcxis the 

mass of crop c grown in AEZ f in region r used in ethanol production pathway x. Ncris the change in crop yields 

due to yield price elasticity. ibycrfis the yield of biomass feedstock, ddgs is the yield of ddgs in corn to ethanol 

processes, Gr is the volume of ethanol produced in region r, pycx is the process yield for converting crop c into 

ethanol in pathway x, and fdem is the volumetric quota for bioethanol. smr is the maximum amount of stover 

that can be harvested sustainably, fpr is the price of bioethanol (in 2004 USD), Egris the value of ethanol 

produced, Pgr is commodity price, and yper is yield price elasticity. CE represents the CO2-eq emissions from 

SOC loss, SOC loss from the subsoil, and from uncombusted biomass, nppfr is the net primary productivity of 

land, soclfr is the SOC level of the original land, slflfr is the SOC loss factor, sslflfr is the subsoil SOC loss factor, 

uncclfr is the mass of uncombusted biomass, and cef is the CO2-eq emission factor. NE represents the CO2-eq 

emissions from changes in soil composition, cnf is the C : N soil ratio, and nef is the N2O CO2-eq emissions 

factor. FE is the amount of foregone CO2 sequestration from forest growth, represented by fglfr. slmlj denotes 

which sectors use which types of land, FTfr represents factor transformation price changes; PSfjr and PFfr denote 

sector-specific factor prices and regional factor prices, and ηfr
ae is a factor price elasticity. vomgr, RAr, DDFMigr, 

DXMDirs, and dstir stand for total value of output at market prices, the regional account balance and market 

prices, the change in demand for commodity i, the change in exports/imports, and demand for international 

transportation services. Finally, ipex is the rate of indirect processing emissions of production pathway x and 

dpex is the rate of direct processing emissions of production pathway x. 

Constraint (4) states that the total land lost in AEZ f in region r must equal the total land repurposed in that AEZ 

and region. Eq(7) – Eq(12) determine how much of each type of biomass to grow in each region for use as a 

biofuel feedstock. The authors assumed 33 % of the corn stover may be harvested sustainably in constraint 

166



Eq(10). Constraint Eq(12) determines changes in crop yields in response to crop price changes. Eqs(12), (13), 

and (14) calculate LUC emissions based on changes in SOC, uncombusted biomass, and foregone forest 

sequestration. Constraints Eq(16) and Eq(17) state that the amount of each type of land gained or lost must be 

consistent with changes in demand for each type of land. Constraint (18) “shocks” the economy with production 

of bioethanol. Eq(19), Eq(20), and Eq(21) calculate LUC emissions, and indirect/direct processing emissions.  

4. Results and Discussion 

All experiments were performed on a DELL OPTIPLEX 790 desktop PC with an Intel(R) Core(TM) i5-2,400 CPU 

@ 3.10 GHz and 8 GB RAM. All models and solution procedures were coded in GAMS 24.7.1. The NLP 

consisted of more than 36,000 equations and 192,000 variables. There were over 440,000 nonlinear terms. The 

problem was solved with the KNITRO 10.0.1 solver to local optimality with feasibility tolerances of 10-5 and 

optimality tolerances of 10-4. The case study was solved within 420 CPU s.  

The demand of biofuel exactly equalled the required supply/quota of 37.9 BL/y. Two-stage dilute acid 

pretreatment and AFEX pretreatment of corn stover are the main chosen processing pathways to make the 

bioethanol. The emissions distribution for the case study is shown in Figure 3. Indirect emissions are negative 

due to sales of electricity from the process to the grid, displacing some energy input mix at the power plant. We 

assume the electricity that the plant displaces is generated from the average mix of the US electricity sources. 

Net emissions are also negative at -4.1 g CO2-eq/MJ. Small capacities of other technologies are also chosen to 

meet the increased production of each feedstock from the modelled economic shock. Corn stover is collected 

from the US (60 Мt), the EU (12 Мt), Brazil (7 Мt), and the China + Hong Kong region (23 Мt). iLUC emissions 

account for 1.5 g CO2-eq/MJ of life cycle emissions, and direct processing at the facility contributes 5.2 g CO2-

eq/MJ. 196,000 ha across the globe of pastureland and 8,100 ha of forests are converted to cropland. The 

additional cropland is largely cultivated to accommodate the increase in demand of various crops due to the 

increase in ethanol production. Prices of all commodities around the world rise slightly, increasing food prices 

slightly, and galvanizing additional crop production.     

 

Figure 3. Results of the case study 

Corn stover is the key feedstock as iLUC emissions of corn stover are marginal. No new cropland was needed 

– the current supply of corn stover in the large corn-producing regions is enough to produce 37.9 BL/y ethanol. 

Tthe iLUC emissions of corn stover ethanol represent the emissions from LUC brought about by changes in the 

economy. The LUC emissions rate of 1.5 g CO2-eq/MJ is in line with previous estimates. The estimate is on the 

lower end of the estimates presented by previous studies, which is not surprising considering our goal is 

minimization of emissions. Importantly, the life cycle GHG emissions are dominated by direct processing 

emissions and indirect processing emissions. The results provide two key findings. One, the iLUC emissions 

rate of corn stover is low. Second, if minimization of life cycle emissions is the goal, then more focus must be 

directed towards more efficient and cleaner processing methods, as direct processing emissions are more than 

three times larger than iLUC emissions. The process also produces extra electricity. If this electricity is used to 

sequester or capture CO2-eq emissions, then life cycle GHG emissions could decrease even further. Large 

decreases in forest cover for the U. (~ 7,000 ha) and Russian Federation (~16,000 ha) were observed. 

Afforestation occurred in Canada (~ 8,000 ha) as well as South and Other Americas (~3,000 ha). Conversion of 

pastureland in the U.S. (~125,000 ha), Brazil (~5,000 ha), China and Hong Kong (~3,000 ha), and South and 

Other Americas (~3,000 ha) to cropland, was by far the largest contributor to LUC. Price increases in cropping 

sectors outpaced those of livestock and animal product sectors, so the conversion of pastureland to cropland is 

expected. These anticipated changes in land use guide policymakers and relevant industries on what to expect. 

The more knowledge different actors in the biofuels life cycle have, the better they will be able to understand 

LUC effects of biofuels, and they will be better able to mitigate these effects. This model provides an optimal 

scenario for the LUC emissions of biofuels, so results can be seen as a “best case” scenario.   

167



5. Conclusion 

A novel land use change (LUC) life cycle optimization (LCO) modelling framework was developed. This model 

integrated LCO methodology with state-of-the-art LUC emissions characterization and calculation techniques. 

These techniques included the use of computable general equilibrium (CGE) models to determine LUCs before 

and after some economic shock. In this case, the economic shock was an increase in ethanol production. Five 

different biomass feedstocks and fourteen different ethanol production pathways were considered. Production 

of ethanol via two-stage dilute acid pretreatment and AFEX pretreatment and upgrading of corn stover was 

shown to have the lowest life cycle greenhouse gas (GHG) emissions at 37.9 BL/y US bioethanol. iLUC 

emissions from this scenario accounted for 1.5 g CO2-eq/MJ, and direct processing emissions contributed 5.2 

g CO2-eq/MJ. This result signals that improvements in processing efficiencies and/or emissions reduction 

technologies could have the largest impact on minimizing life cycle emissions of biofuels. The net emissions 

rate was -4.2 g CO2-eq/MJ, stemming from a GHG emissions “credit” by selling excess electricity produced 

back to the electrical grid. The developed model laid the groundwork for future, more detailed studies on the 

iLUC emissions of biofuels and other products and processes.  

References 

Accardi D.S., Russo P., Lauri R., Pietrangeli B., Palma L.D., 2015, From Soil Remediation to Biofuel: Process 

Simulation of Bioethanol Production from Arundo donax, Chemical Engineering Trans, 43, 2167-2172. 

Dunn J.B., Mueller S., Kwon H.-Y., Wang M.Q., 2013, Land-use change and greenhouse gas emissions from 

corn and cellulosic ethanol, Biotechnology for Biofuels, 6, 51. 

Garcia D.J., You F., 2015, Multiobjective optimization of product and process networks: General modeling 

framework, efficient global optimization algorithm, and case studies on bioconversion. AIChE J, 61,530-554. 

Garcia D.J., You F., 2015, Network-based Life Cycle Optimization of the Net Atmospheric CO2-eq Ratio (NACR) 

of Fuels and Chemicals Production from Biomass. ACS Sust Chemistry & Engineering, 3, 1732-1744. 

Gong J., Garcia D.J., You F., 2016, Unraveling Optimal Biomass Processing Routes from Bioconversion Product 

and Process Networks under Uncertainty: An Adaptive Robust Optimization Approach. ACS Sustainable 

Chemistry & Engineering, 4, 3160–3173. 

Gong J., You F., 2015, Sustainable Design and Synthesis of Energy Systems. Current Opinion in Chemical 

Engineering, 10, 77-86. 

Hertel T.W., 1997, Global trade analysis: modeling and applications, Cambridge Univ Press, Cambridge, UK. 

Humbird D., Davis R., Tao L., Kinchin C., Hsu D., Aden A., 2011, Process Design and Economics for 

Biochemical Conversion of Lignocellulosic Biomass to Ethanol: Dilute-Acid Pretreatment and Enzymatic 

Hydrolysis of Corn Stover, NREL, Golden, CO, United States. 

Kazi F.K., Fortman J., Anex R., Kothandaraman G., Hsu D., Aden A., Dutta A., 2010. Techno-economic analysis 

of biochemical scenarios for production of cellulosic ethanol, National Renewable Energy Laboratory, 

Golden, CO, United States, Technical Report No. NREL/TP-6A2-46588.  

Plevin R.J., Beckman J., Golub A.A., Witcover J., O’Hare M., 2015, Carbon accounting and economic model 

uncertainty of emissions from biofuels-induced land use change, Envir Scien & Technology, 49, 2656-2664. 

Santibañez-Aguilar J.E., Morales-Rodriguez R., González-Campos J.B., Ponce-Ortega J. M., 2016, Stochastic 

design of biorefinery supply chains considering economic and environmental objectives. Journal of Cleaner 

Production, 136, 224-245. 

Taheripour F., Birur D.K., Hertel T.W., Tyner W.E., 2007, Introducing liquid biofuels into the GTAP database, 

GTAP Research Memorandum, Center for Global Trade Analysis, Purdue University, West Lafayette, USA 

Wan Alwi S.R, Klemeš J.J., Varbanov P.S., 2016, Cleaner energy planning, management and technologies: 

Perspectives of supply-demand side and end-of-pipe management. J of Cleaner Production, 136, 1-13. 

You F., Tao L., Graziano D.J., Snyder S.W., 2012, Optimal design of sustainable cellulosic biofuel supply chains: 

Multiobjective optimization coupled with life cycle assessment and input–output analysis, AIChE Journal, 58, 

1157-1180. 

You F., Wang B., 2011, Life Cycle Optimization of Biomass-to-Liquids Supply Chains with Distributed-

Centralized Processing Networks. Industrial & Engineering Chemistry Research, 50, 10102–10127. 

Yue D., Kim M.A., You F., 2013, Design of Sustainable Product Systems and Supply Chains with Life Cycle 

Optimization Based on Functional Unit: General Modeling Framework, Mixed-Integer Nonlinear 

Programming Algorithms and Case Study on Hydrocabon Biofuels. ACS Sustainable Chemistry & 

Engineering, 1, 1003-1014. 

Yue D., Pandya S., You F., 2016, Integrating Hybrid Life Cycle Assessment with Multiobjective Optimization: A 

Modeling Framework. Environmental Science & Technology, 50, 1501–1509. 

Yue D., You F., Snyder S. W., 2014, Biomass-to-bioenergy and biofuel supply chain optimization: Overview, 

key issues and challenges, Computers & Chemical Engineering, 66, 36-56. 

168