Methods for the Design of Spatially Explicit Biofuel Supply Chains

Methods for the Design of Spatially Explicit Biofuel Supply Chains

Rex T. L. Ng^1,2 and Christos T. Maravelias^1,2

¹Department of Chemical and Biological Engineering and ²DOE Great Lakes Bioenergy Research Center, University of Wisconsin-Madison, Madison, WI

Cellulosic biomass is seen as one of the promising sources for sustainable biofuels production. In recent years, much effort has been made to design efficient biomass-to-biofuel supply chains (SCs) in terms of economic, environmental and social aspects. In many studies, biomass feedstock availability is estimated based on county-level data, assuming a harvesting site is located at the centroid of a county [1,2], or a county is divided into rectangular or square sub-areas, and a harvesting site is located at the centroid of a sub-area [3,4]. Additionally, biomass is assumed to be uniformly distributed over the area. However, unlike fossil fuels, biomass availability is highly distributed, so the aforementioned assumptions often lead to biofuel SCs that are suboptimal or even infeasible from an energy consumption point of view, that is, the primary energy required to harvest and transport biomass is higher than the energy content of the produced biofuel.

To overcome this, we develop a framework that allows us to utilize high spatial resolution biomass data in designing biomass-to-biofuel SCs. We first develop a tool based on the Agricultural Model Intercomparison and Improvement Project (AgMIP) packages to generate feedstock data using climate data from the user-specified general circulation model and representative pathway [5]. This data is then combined with a historical cropland data layer (CDL) which is provided by the U.S. Department of Agriculture [6], to generate spatial data for an area based on a user-specified resolution (down to 30 m × 30 m). Note that this generates a complex SC network with a very large number of arcs, and therefore leads to a large-scale optimization problem.

Accordingly, we then develop a preprocessing algorithm to reduce the number of arcs. We first determine the feedstock, harvesting and handling (FHH) cost up to the biorefinery gate (includes transportation and depot installation and operation costs) for different SC configurations (with or without depot and different transportation modes). If depot is selected, we determine the maximum distance for collecting distributed biomass to a potential depot given the depot capacity. Note that the maximum distance greatly depends on biomass availability at every potential depot location. Next, we compare the FHH costs for all configurations and determine the threshold distances at which one configuration becomes the most economical. For example, in a corn stover-to-bioethanol SC, depot should only be installed if the distance between a harvesting site and the biorefinery is more than 120 km. Similarly, rail transportation of densified corn stover is cheaper than truck transportation when the distance between depot and biorefinery is greater than 225 km. Based on these pre-calculated distances, we remove arcs that will never be used in an optimal solution, and thus dramatically reduce the number of defines variables.

Next, we develop a multi-period mixed-integer linear programming (MILP) model that accounts for facility location, technology selection, and capacity planning decisions in terms of depots and biorefineries, as well as auxiliary decisions such as harvesting site and biomass feedstock selection, biomass allocation, transportation mode, and inventory planning decisions.

Finally, we illustrate the applicability of the proposed model by considering a corn stover-to-bioethanol SC in Wisconsin (420 km × 500 km). We consider a one-year horizon divided into 12 periods, and two transportation modes: truck and rail. The objective is to minimize the total annual cost, which includes biomass acquisition, inventory, operating, and annualized capital costs for depots and biorefineries.

References:

[1] F. You, B. Wang, Life Cycle Optimization of Biomass-to-Liquid Supply Chains with Distributed–Centralized Processing Networks, Ind. Eng. Chem. Res. 50 (2011) 10102–10127.

[2] W. Alex Marvin, L.D. Schmidt, S. Benjaafar, D.G. Tiffany, P. Daoutidis, Economic Optimization of a Lignocellulosic Biomass-to-Ethanol Supply Chain, Chem. Eng. Sci. 67 (2012) 68–79.

[3] R.T.L. Ng, C.T. Maravelias, Design of Cellulosic Ethanol Supply Chains with Regional Depots, Ind. Eng. Chem. Res. 55 (2016) 3420–3432.

[4] R.T.L. Ng, C.T. Maravelias, Design of biofuel supply chains with variable regional depot and biorefinery locations, Renew. Energy. 100 (2017) 90–102.

[5] N.B. Villoria, J. Elliott, C. Müller, J. Shin, L. Zhao, C. Song, Rapid aggregation of global gridded crop model outputs to facilitate cross-disciplinary analysis of climate change impacts in agriculture, Environ. Model. Softw. 75 (2016) 193–201.

[6] National Agricultural Statistics Service U.S. Department of Agriculture, CropScape - Cropland Data Layer, (2016). https://nassgeodata.gmu.edu/CropScape/ (accessed April 5, 2017).