Using agent-based distillations to explore logistics support to urban, humanitarian assistance/disaster relief operations
Wolf, Eric S.
Sanchez, Susan M.
MetadataShow full item record
There are two motivations for studying Humanitarian Assistance/Disaster Relief (HA/DR) operations. First, the Marine Corps will be a first-responder in the future. Second, logistics support takes on a primary role. This thesis identifies the potential for using agent-based models to support logistical decision-making in an urban, HA/DR environment. We develop a simulation using Map Aware Non-uniform Automata (MANA). Our scenario depicts a relief convoy with security attachment, operating on urban terrain. The convoy moves to an HA/DR site where they distribute food to neutrals (locals) who have made their way to that site. We couple data farming with a Latin Hypercube design of experiment to explore very large data space. Forty variables are identified. We establish 640 different design settings and each setting is replicated 50 times producing a 32,000-point dataset. We use regression to fit several models. The conclusions from this thesis suggest: coupling intelligent designs with data farming is effective at exploring large data space; mission success in HA/DR operations may depend on only a handful of factors; understanding local communications is the key to mission success; success cannot be determined based solely on the factors the convoy controls.