Yet-to-develop infrastructures like biorefineries are exposed to many uncertainties compared to established systems such as fossil-based ones. The exposure to fluctuations of biomass supply is a growing concern due to the increasingly magnified consequences of climate change. This paper presents a two-stage stochastic mixed integer linear programming framework to design circular production systems using biomass wastes subjected to yield uncertainty. The modelling framework embeds an expected profit objective function in a spatially explicit, multi-echelon, multi-period, multi-feedstock, and multi-product lignocellulose-based biorefining supply chain network. The modelling framework integrates a risk-constrained formulation based on downside risk to represent decision-makers’ propensity towards risk. A case study based on real data from south-west Hungary is presented. Results show that biobased biorefining systems remain a risky capital-intensive investment, but profitable configurations of the network can be achieved, despite the inclusion of large variabilities in the biomass yields. Although they exhibit expected profits either comparable or slightly lower than risk-neutral configurations, the solutions subjected to risk-based regularisation (risk-constrained), are more stable than their stochastic counterpart. Furthermore, biomass supply chains, that can develop either a centralised or a decentralised configuration, would correspond to different risk profiles. While the localisation of centralised plants generates higher expected profits compared to sparsely distributed facilities, the latter, with a more diffuse presence of plants in the territory, can lead to a more stable system and to a more homogenous integration with local communities.
Panteli et al. (Thu,) studied this question.