| dc.description.abstract | This study addresses optimal design and configuration of a supply chain for a lignocellulosic biorefinery. To do this, a comprehensive two-stage stochastic Mixed Integer Programming (MIP) model was developed and implemented to represent a multi-feedstock ethanol supply chain under feedstock yield uncertainty. The model minimizes the expected cost of construction and operation of the chain, choosing the facilities, feedstock production locations, monthly harvest, feedstock movement and handling, storage and refining activity. Two regional Texas case studies are conducted to examine the consequences of alternative supply chain elements and yield uncertainty. Additionally, the impact of using data resolution is studied. The study finds that incorporation of yield uncertainty is important and that its inclusion doubles feedstock contracting, resulting in substantial feedstock dumping costs when above average yields arise. In addition, using multiple (rather than single) feedstocks substantially lowers costs when there is inherent seasonality of feedstock harvest. The findings also indicate that remotely located storage depots with associated pellet plants allow exploitation of geographically stranded feedstocks. Our results in the Texas High Plains show the corn stover collection area moves from an 80 km radius to a 200 km radius when pellets can be exported at $150 per mg. Finally, the results show that use of higher resolution data improves the estimates of transportation costs and alters the supply chain design. | en |
