(483e) Technology Prioritization for Biofuel Supply Chain Design Via Stochastic Optimization

Technology is a critical component in biofuel supply chain. As a strategic decision, technology selection incurs large capital costs (Govindan, Fattahi, and Keyvanshokooh 2017), thus decision makers must commit to certain technologies under the budget constraint. At the same time, technology selection decisions are made under uncertain future budgets, along with other uncertainties surrounding the design and long-term planning setting. Consider a set of technologies, each converts one compound to another compound and requires continued funding over the coming years. Different technologies improve different compound conversion processes while competing for the uncertain annual budget. In addition, uncertainties around demand, price, and biomass yield should also be taken into account.

Most studies from literature address the aforementioned issue by solving for a portfolio of technologies via stochastic optimization, while decision makers in practice often form a rank-ordered list of technologies and select those with the highest priority until the budget is exhausted instead. Both approaches have their shortcomings: a portfolio of technologies based on assumptions on uncertain parameters can be fragile with respect to changes to those parameters not captured by uncertainty modeling, while a rank-ordered list is often based on the merits of individual technologies and does not capture the combined effects of technologies when they are selected together.

Koç and Morton (Koç and Morton 2014) proposed a prioritization approach based on stochastic optimization. This approach generates a rank-ordered list which recognizes that items in the list work together as a portfolio when selected. We generalize the work by Koç and Morton to a multi-stage multi-period setting for biofuel supply chain design problems. Specifically, we define binary variables that model the relative priority between technologies as first-stage variables, and the remaining variables enforce that a higher priority technology must be selected before selecting a lower priority technology and capture the relationship between other decision variables and technology selections.

We demonstrate the capability of our method to identify high-priority technologies through a case study located in the Midwest region of USA with realistic data. Our results provide insights into technology selection for biofuel supply chain design.

References

Govindan, Kannan, Mohammad Fattahi, and Esmaeil Keyvanshokooh. 2017. “Supply Chain Network Design under Uncertainty: A Comprehensive Review and Future Research Directions.” European Journal of Operational Research 263 (1): 108–41. https://doi.org/10.1016/J.EJOR.2017.04.009.

Koç, Ali, and David P. Morton. 2014. “Prioritization via Stochastic Optimization.” Management Science 61 (3): 586–603. https://doi.org/10.1287/MNSC.2013.1865.