| The planning, scheduling, dispatching, and delivery of perishable items in a time-constrained environment are recognized as one of the most challenging problems in manufacturing. In the concrete industry, the challenge is dramatically increased due to overbooking and the requirement to always complete multi-truck, time-synchronized orders once they are started. Additionally, weather and traffic conditions can adversely affect expected travel-time in an environment where more than 90% of orders are modified during the day of delivery. The dynamic nature of the problem requires the schedule to be revised on a constant basis.; To solve this problem, a decision-support tool was created that consists of both planning and execution modules. These modules assist Customer Service Representatives and Dispatchers in evaluating thousands of possible delivery alternatives. This dissertation describes the series of optimization models required to implement a decision-support tool, the implications of imperfect data, and implementation issues associated with real-time requirements. The foundation for the solution is a time-space network representation of the problem incorporating a multitude of alternatives for delivering an order to a customer. Choosing a single delivery alternative for each customer adds restrictive integrality constraints to the network model. In addition to the time-space network formulation, a minimum-cost network flow model and a Tabu Search heuristic are utilized.; The solution of the model formulation assists customer service representatives and dispatchers in determining (1) the feasibility of accepting additional orders, (2) the arrival times for drivers reporting to work, (3) the scheduling of all orders, (4) the real-time assignment of drivers to delivery loads, (5) the dispatching of these drivers to customers and back to plants, and (6) the scheduling of plants. Most of these decisions are determined through the use of models incorporating exact optimization techniques. |
