We propose risk-averse models for a two-stage uncertain blood supply chain design problem, in the context of disaster management. At the first stage, one needs to choose the number and locations of permanent blood collection facilities, the amount of safety stocks at hospitals and the size of the fleet of mobile facilities. Once the uncertain blood demands are revealed, we decide, at the second stage, the route of each mobile facility as well as the flow of blood units that must be collected and dispatched to satisfy the demand.

We experimentally compare the solutions obtained using two discrete scenario-based approaches (worst-case and conditional value-at-risk) with a robust finite adaptable model, and with the solution robustness approach of Jabbarzadeh et al., 2014.