Two-stage robust bilevel optimisation model for facility location considering post-disruption demand fulfilment

The bilevel facility location problem is a core optimisation problem in hierarchical industrial systems. A typical example is a supply chain system in which an upper-level supplier constructs critical facilities to improve cost efficiency, while lower-level operators independently make service and allocation decisions to reduce unmet demand. Facility disruption risk, although often low in probability, can have a severe impact on the service capability. Traditional two-stage robust facility location models usually adopt a single-decision-maker perspective and may overlook the distinct operational objectives of lower-level decision makers. Therefore, this paper proposes a novel two-stage robust bilevel facility location model that captures strategic facility deployment before disruptions and operational demand fulfilment after disruptions occur, while accounting for the follower’s service-oriented response. Several structural properties are derived to clarify the relationship between the proposed model and the classical two-stage robust facility location model, and we develop an enhanced column-and-constraint generation algorithm for robust bilevel optimisation with decision-dependent uncertainty. Numerical results show that the proposed model usually achieves better post-disruption demand fulfilment, higher average supply capacity utilisation, and lower unit service cost. These findings demonstrate that explicitly modelling the hierarchical leader–follower structure can improve service efficiency for resilient supply network design.