Route determination in disaster areas: Using predictions and introducing the option to wait to improve routing results

Disasters caused by human action or nature are part of life. Preventing disasters from occurring is often not possible. Lives can be saved or lost depending on our response to the disaster. Determining the fastest route to the people in need is therefore important. The determination of this fastest route is however not straight forward, since the environment the route is determined in has become chaotic as a result of the disaster. In this research attention focuses on ways to take the circumstances in the environment into consideration in an automated way. A distinction is made between changes that can be predicted and changes that cannot be predicted. Based on literature the latter turns out to be done best by choosing a process of re-evaluating the route determined. Based on the degree of dynamism a static, semi-static, iterative or dynamic approach can be taken to this re-evaluation. In a static approach the route is determined only once based on the most current information at that time. When the circumstances change re-evaluation of that determined route can take place every time a change occurs (semi-static), every couple of minutes (iterative) or constantly (dynamic). In this research a preference was given to use a combination between the semi-static and the iterative approach. This results in a re-evaluation after a number of changes have occurred and if there are only little changes the re-evaluation should take place after a set time. The incorporation of predictions in the algorithm is a second focus area. By incorporating predictions one is able to anticipate changes to the network and take them into consideration in the route determination process. Incorporating predictions on plume movement and bridge openings and closings introduces also a need to balance between travelling extra kilometres and waiting. To investigate the implications of incorporating predictions into the routing process a routing algorithm was designed. The Dijkstra algorithm was adapted to read the closing times from a file and decide whether it is better to wait or take an alternative route. Tests show that the estimation of travel times are more accurate when these are created with the algorithm that incorporates predictions into the routing process. Based on this research it is concluded that using the adapted algorithm routes can be determined that will prove to be faster and safer than the result of a shortest path calculation based only on the travel costs in the network.GIMAGIS TechnologyOTB Research Institute for the Built Environmen