Department of Information Systems, Statistics & Management Science
Permanent URI for this community
Browse
Browsing Department of Information Systems, Statistics & Management Science by Author "Capar, Ibrahim"
Now showing 1 - 1 of 1
Results Per Page
Sort Options
Item Online vehicle routing problems and analysis(University of Alabama Libraries, 2016) Capar, Ibrahim; Keskin, Burcu Baris; University of Alabama TuscaloosaTraditional optimization models, whether classified as deterministic or stochastic, require problem related parameters to be known with certainty or according to a probability distribution. However, in many real life problems, it is unrealistic to assume that problem parameters are known a priori. A problem is called online when all of the problem data is not immediately available and revealed incrementally. Hence, online optimization models help make decisions in the best possible way before all of the problem data is observed. This dissertation considers online vehicle routing problems in manufacturing and public safety domains. In the manufacturing domain, we consider outbound logistics problems of a company that serves a geographically dispersed customer base with a variety of products. At the end of each day, the company decides which orders to dispatch the next day via one of the three transportation modes, and which ones to delay into the future for better order consolidation and cost reduction opportunities. We solve this problem with two different methods. We propose two different solutions: i) a cluster-first, route-second type of heuristic; ii) a reduction heuristic based on a problem-specific lower bound. In the public safety domain, we investigate the online patrol routing problem given incrementally revealed accidents/incidents. In the traditional patrol routing problem, state troopers would follow the initial assigned routes without consideration for disruptive incidents, e.g., accidents. However, in real life, state troopers may have to be diverted to other tasks from their initial routing assignments such as responding to an accident. We use a simulation based analysis to compare the key performance indicators of the problem with other examples.