Insights from two-stage stochastic programming in emergency logistics
Douglas Alem, Alistair Clark
This paper discusses the practical aspects and resulting insights of the results of a two-stage mathematical network flow model to help make the decisions required to get humanitarian aid quickly to needy recipients as part of a disaster relief operation. The aim of model is to plan where to best place aid inventory in preparation for possible disasters, and to make fast decisions about how best to channel aid to recipients as fast as possible. Humanitarian supply chains differ from commercial supply chains in their greater urgency of response and in the poor quality of data and increased uncertainty about important inputs such as transportation resources, aid availability, and the suddenness and degree of “demand”. The context is usually more chaotic with poor information feedback and a multiplicity of decision-makers in different aid organizations. The model attempts to handle this complexity by incorporating practical decisions, such as pre-allocation of emergency goods, transportation policy, fleet management and procurement, in an uncertainty environment featured by a scenario-based approach. Preliminary results based on the floods and landslides disaster of the Mountain Region of Rio de Janeiro state, Brazil, point to how to cope with these challenges by using the mathematical model.