(626h) Optimization Methods for the Integration of Spatially Explicit Landscape and Biofuel Supply Chain Network Design

The large-scale production of biofuels will require an efficiently designed and operated supply system. While many researchers have applied mathematical programming techniques to biofuel supply chain optimization, with a greater understanding of crop and landscape systems there is an opportunity to expand the system boundary to include upstream landscape design decisions simultaneously with biofuel supply chain network design to find integrated solutions that leverage the spatially explicit interactions. Landscape design is the process of deciding where in the landscape to establish bioenergy crops and how to manage that land to achieve a specific goal.

In the past, researchers have considered biomass availability as an exogenous parameter that the modeler has no influence over; however, by introducing biomass establishment locations and crop management levels (i.e. fertilization) as decision variables, the landscape and supply chain can be optimized simultaneously. We introduce methods for modelling landscape design and supply chain network design simultaneously. The constraints and data methodologies we introduce account for challenges resulting from such an integrated model.

First, biomass yield and soil organic carbon sequestration potential are highly spatially explicit and depend on factors such as soil quality and land management. We introduce spatially explicit constraints and a flexible data methodology that uses a gridded approximation to resolve the distribution of biomass in a computationally tractable way. Second, biomass yields can fluctuate considerably year-to-year. Yield fluctuations also depend on spatial factors like local weather and soil quality. We employ a two-stage stochastic model and use individual years of simulation data from biogeochemical crop models as stochastic scenarios to find solutions that perform well in expectation over the long term to ensure the supply chain configurations are resilient to especially ‘good’ or ‘bad’ years. Finally, we include constraints that account for greenhouse gas emissions including spatially explicit soil organic carbon sequestration using a ‘cost of carbon’ that lets decision makers adjust how heavily to weight environmental outcomes on a cost basis.

We present the optimization model, highlighting important constraints, and briefly demonstrate the usefulness of the model with a case study located in Michigan, USA using realistic high-resolution yield and soil carbon data obtained through biogeochemical crop model simulations performed over multiple years of weather data. Model complexity is sensitive to the number of uncertainty scenarios, the size of the study area, spatial resolution, and the number of supply chain nodes considered.