Solving the problem of cost-efficiently dispatching vehicles from a depot to deliver products to customers means determining: (1) which customers to assign to the same route, and (2) the sequence of deliveries to customers on the same route. When customer demands remain fixed for each period during the planning horizon, the solution developed at the start of the planning horizon will be optimal throughout the planning horizon. However, time-varying stochastic demand undermines this fixed-routes solution. First, it makes route failures possible; i.e., in every period, the demand on at least one of the routes can exceed vehicle capacity. Second, because the fixed-routes solution is not demand-responsive, it may not be optimal in every period. For management at the depot, these effects raise a key tradeoff question: Should we use the inherently stable fixed-routes delivery strategy and incur the negative customer service effects of route failures, or should we adopt more demand-responsive strategies which are unstable and incur greater information and/or dispatching resources?; This question is central to the research, and to provide some insights into possible answers, a set of regression-analytic formulae were calibrated with data from extensive computational experiments to answer four specific underlying questions: (I) how does the impact of stochastic demand on the transportation cost of controlling route failure vary in response to the following factors: (a) demand variability, (b) the tendency of demand locations to have zero demand, and (c) the capacity buffer built into the fixed-routes solution to control route failure? (II) how does customer service performance for the fixed-routes strategy vary in response to these factors? (III) how does the transportation cost of redressing route failures vary in response to these factors? (IV) how does the saving in transportation costs vary in response to the depot's use of updated demand information to modify its routes before dispatching vehicles each period?; The formulae proved accurate. To provide a computer-based structure for using these formulae in analyzing training-sized problems and in performing approximate analysis for more complex problems which depot managers face, four spreadsheet templates which incorporate the relevant research results were developed.
展开▼
