Prepositioning of relief goods under network uncertainty for tropical cyclones with information updates: A humanitarian supply chain

• Main Authors: Lahoz, Miguel, Ng, Mykhael, Sy, Derrick
• Format: text
• Language: English
• Published: Animo Repository 2015
• Subjects: Disaster relief
• Online Access: https://animorepository.dlsu.edu.ph/etd_bachelors/6674
• Institution: De La Salle University

Prepositioning decisions of where to preposition and how many to preposition at each location is very important as prepositioning is very costly and allows a little room for errors. Tropical cyclones is a type of natural disaster that is being monitored and forecasted ever since its development up until its landfall or dissipation. This paves the way for the possibility of using the forecast information in the prepositioning or relief goods in the areas forecasted to be hit by the tropical cyclones in order to help the post disaster distribution of relief goods in a humanitarian supply chain. However, since these forecasts are not perfect and are constantly updated in relation to the actual development of the tropical cyclone, there is value in waiting for an updated forecast information at the expense of an expedited logistics cost due to a shorter lead time. Thus, another concern aside from those of where and how many to preposition is when to preposition. The novelty of the study is the consideration of the multi-objective of cost and response times. The response times are affected by the possibility that the flow of goods in the post disaster links may be limited or reduced due to the effects of the tropical cyclone. These network uncertainties are modelled through scenarios. The previous concern of where and how many to preposition now becomes affected by the response times and the network uncertainty. Furthermore, the consideration of response times in the decision to wait or to preposition allows for the expected response times to play a role in the determining the optimal prepositioning strategy of when to preposition. Thus, the previous decisions of where, how many, and when to preposition become response driven and not cost driven only. Following a decision theory approach, four models were developed namely: models A (perfect information), B (under uncertainty), B0 (cost of waiting for more information), and C (decision model). These models are run every time there is a new forecast and stops before the storm landfalls. The models presented in this paper utilize the updating nature of tropical cyclone forecast information in the decision of minimizing the cost and response times of the prepositioning process. As such, linear physical programming was used for the optimization of the multiple objectives as it allows for the incorporation of the decision maker's preferences regarding the objectives. The results of the study show that network uncertainties are an important consideration when prepositioning as it affects the optimal prepositioning strategy.