Abstract
Spontaneous volunteers are ordinary citizens who assist in disaster relief efforts, while they are a great resource they also pose a difficult logistical challenge. Unlike classical labor assignment problems, the management of these volunteers is characterized by uncertainty regarding the size, availability, and commitment of the labor pool. We model this problem as a multi-server queueing system with both stochastic server arrival and abandonment. This model is intended to be applied to the relatively stable work associated with recovery efforts, e.g., debris clearing. We model this system as a continuous time Markov decision process and compare the optimal policy to several common-sense heuristics; one of which performs close to optimal and makes a practical alternative. We conduct extensive sensitivity analysis around model parameters and assumptions.
Notes
1 In all cases examined, a discount factor of was used. This value was chosen because it is widely accepted in economic evaluation for developed countries and is recommended by the US Panel on Cost-Effectiveness in Health and Medicine (Siegel et al, Citation1997).
2 Details of the percentage of states for which the MDP policy matched the FV or LD policy for several examples can be found in Wolczynski (Citation2015).