Search and rescue management : modelling and development of heuristic strategies within a simulation environment : a thesis presented in partial fulfilment of the requirements for the degree of Doctor of Philosophy in Operations Research at Massey University, Palmerston North, New Zealand

Abstract

The search for a lost person on land has been the subject of relatively little research to date in comparison to other search problems. This thesis addresses this imbalance by examining the search for a stationary object that does not attempt to avoid detection. The problem is defined as a synthesis of the coverage, routing, and allocation problems that exist in the literature, and its complexity and unique aspects are discussed. A physical model of the search terrain is developed using a Triangulated Irregular Network (TIN). This model incorporates the vegetation and natural features of the terrain, and is extended to model access paths and traversal speeds between any two points. A visibility model is developed over the TIN in order to define a detection model for both a human subject and any clues placed by him. Correction factors are used to model visibility and traversal speeds under different search environments. Methods to define search regions as components of the elements of the TIN are described. Heuristic resource allocation methods are then developed for both the reconnaissance and general phases of a search operation. These methods allocate search tasks to resources individually or in parallel, and in real-time. Dynamic heuristic search strategies to respond to changing search conditions and the discovery of new information are then developed. A Discrete Event Simulation (DES) model of a Search and Rescue (SAR) operation is developed. This model incorporates: siting a search base; search resource deployment and searching; clue and subject detection; communication between resources and search management; flooding and resource deployment under adverse weather conditions; and responsiveness of the subject over time. The simulation model is used to perform some preliminary computational experiments on a restricted set of resource allocation methods and search strategies. Initial trends indicated from these experiments are: the general superiority of methods which do not restrict the set of regions to be allocated for searching to an initial primary search area; the dominance of a night searching strategy; the dominance of using a sound detection method when a subject is responsive; and the benefits of applying diversifying search strategies.