Pre-positioning of relief items under road/facility vulnerability with concurrent restoration and relief transportation

Aslan E., Celik M.

IISE TRANSACTIONS, vol.51, no.8, pp.847-868, 2019 (SCI-Expanded) identifier identifier

  • Publication Type: Article / Article
  • Volume: 51 Issue: 8
  • Publication Date: 2019
  • Doi Number: 10.1080/24725854.2018.1540900
  • Journal Indexes: Science Citation Index Expanded (SCI-EXPANDED), Scopus
  • Page Numbers: pp.847-868
  • Keywords: Disaster preparedness, inventory pre-positioning, relief transportation, network restoration, stochastic programming, sample average approximation, DISASTER RESPONSE FACILITIES, LOGISTICS NETWORK DESIGN, LOCATION-ROUTING PROBLEM, HUMANITARIAN LOGISTICS, STOCHASTIC OPTIMIZATION, EMERGENCY SUPPLIES, OR/MS RESEARCH, MODEL, OPERATIONS, DECISIONS
  • Middle East Technical University Affiliated: Yes


Planning for response to sudden-onset disasters such as earthquakes, hurricanes, or floods needs to take into account the inherent uncertainties regarding the disaster and its impacts on the affected people as well as the logistics network. This article focuses on the design of a multi-echelon humanitarian response network, where the pre-disaster decisions of warehouse location and item pre-positioning are subject to uncertainties in relief item demand and vulnerability of roads and facilities following the disaster. Once the disaster strikes, relief transportation is accompanied by simultaneous repair of blocked roads, which delays the transportation process, but gradually increases the connectivity of the network at the same time. A two-stage stochastic program is formulated to model this system and a Sample Average Approximation (SAA) scheme is proposed for its heuristic solution. To enhance the efficiency of the SAA algorithm, we introduce a number of valid inequalities and bounds on the objective value. Computational experiments on a potential earthquake scenario in Istanbul, Turkey show that the SAA scheme is able to provide an accurate approximation of the objective function in reasonable time, and can help drive policy-based implications that may be applicable in preparation for similar potential disasters.