(249j) A Generic MILP Modelling Framework for the Systematic Design of Lignocellulosic Biorefining Supply Chains

The development of sustainable biobased economies could help overcome the high societal dependency on fossil resources. Therefore, research has focused on the study of advanced biorefining networks. The complexity of such production systems requires the use of efficient decision-making tools, enabling a full exploitation of biomass (and its macrocomponents, i.e. cellulose, hemicellulose and lignin) for the production of biobased products and platform chemicals (Kokossis and Yang, 2010). Therefore, it is also essential to identify the most promising pretreatment process that fractionates biomass into cellulose, hemicellulose and lignin and usually represents the highest cost part of the entire biorefining system. In addition, the deployment of second-generation technologies is still hindered by high capital costs as well as the existence of uncertainties (e.g. demand and price of biobased products) in the so far immature biobased market. Consequently, one of the most important and challenging aspects in the quest of producing a set of sustainable biobased products, is the design of an integrated and economically viable biorefinery supply network (Akgul et al., 2011; Martín and Grossmann, 2010; Ä?uček et al., 2014). Optimisation tools could play a powerful role supporting decision in such novel production systems, through the identification of the major cost drivers, the performance of sensitivity analysis as well as the assessment of economic and technical uncertainties (Kim et al., 2013).

The aim of this work is the modelling and optimization of biorefining chain systems using an integrated approach to the modelling of all the entities involved across the technology chain, with the purpose of achieving a long-term, decision-making regarding the systematic design and planning of advanced biorefining networks.

Methodology

The design problem is formulated assuming as given, sets of lignocellulosic biomass types including their temporal and geographical availability, biobased product types and production facilities of different scale, transport logistics, economic and technical parameters as a function of biomass and product type as well as conversion technology and plant scale, biobased products market characteristics, demand over a fixed time horizon and geographical distribution.

The objective is the identification of the optimal network configurations that satisfy the target regional demand of the selected biobased products over the entire planning horizon, while maximising the overall financial profitability of the system. Therefore, the decision variables of the problem refer to the planning and operation of the biorefining network.

The biobased supply chain optimization problem is formulated as a mixed-integer linear programming (MILP) model, based on previous research works (Giarola et al., 2012) (Giarola et al., 2011) (Zamboni et al., 2009). In particular, the model represents a spatially-explicit, multi-feedstock, multi-period and multi-echelon lignocellulose-based supply chain. A maximum profit-based objective function is considered, including the capital investments, the operating costs and the revenues evaluated across the entire biorefining network.

A binary variable is introduced, accounting for whether a conversion facility is installed within a territorial element or not. Logical constraints depending on the value of the binary variable, impose that only one plant can be built in each geographical cell and production can only take place if a plant is already established.

In addition to the system, several other equality (e.g. mass balance) and inequality constraints were defined in the mathematical formulation of the optimization problem to characterize each supply chain node.

Case study

The proposed MILP model was implemented in the GAMS^® software tool using the CPLEX solver (Rosenthal, 2014). In particular, the model was used to solve a European case study involving the development of biorefining systems in the South-West of Hungary. The case study considered the arable land availability as well as agronomical factors in four Hungarian counties, i.e. Tolna, Baranya, Somogy and Zala. The examined region is discretized, for computational reasons, into 102 square grids, of 225 km²each. According to the Corine Land Cover database, truck was considered the most suitable transportation mode for this study. The model is optimized over a 1-year planning horizon, divided into 12 months, assuming seasonal availability of biomass for July, August, September and October. Additionally, the technical (i.e. technological yields) and economic (i.e. capital and operating costs) inputs, used in this work, are derived from the BIOCORE project (Oâ??Donohue, 2014).

Concluding remarks

Results show that the development of optimization models considering all the supply chain entities can shed the light onto the systematic design and planning of novel production infrastructures.

The model embedding the definition of the process technology superstructure for a portfolio of selected biobased products and platform chemicals, enables the evaluation of environmental impacts along with costs as well as the analysis of the key sources of uncertainty and their evolution over time.

Acknowledgement

The EC (Reneseng-607415 FP7-PEOPLE-2013-ITN) is gratefully acknowledged for supporting this work.

Prof. Kohlheb Norbert and the Szent Istvan University (Hungary) are thanked for providing support in the agronomical and economic characterization of the examined region.

References

Akgul, O., Shah, N., & Papageorgiou, L., G. (2011). ESCAPE21 Journal, 1799-1803.

Ä?uček, L., Martín, M., Grossmann, I., E., Kravanja, Z. (2014). Computers and Chemical Engineering, 66, 57-70.

Giarola, S., Shah, N., & Bezzo, F. (2012). Bioresource Technology, 107, 175-185.

Giarola, S., Zamboni, A., & Bezzo, F. (2011). Computers and Chemical Engineering, 35, 1782-1797.

Kim, J., Sen, S., M., & Maravelias, C., T. (2013). Energy Environ. Sci., 6, 1093-1104.

Kokossis, A., C., & Yang, A. (2010). Computers and Chemical Engineering, 34, 1397-1405.

Martín, M., & Grossmann, I., E. (2010). ESCAPE20 Journal, 943-948.

Oâ??Donohue, M. (2014). BIOCORE â?? FINAL REPORT, Institut National de la Recherche Agronomique (INRA), Toulouse, France.

Rosenthal (2014). GAMS â?? A Userâ??s Guide. GAMS Development Corporation, Washington, DC, USA.

Zamboni, A., Shah, N., & Bezzo, F. (2009). Energy & Fuels, 23, 5121-5133.