| Summary: | Physical distribution is one of the key functions in logistics systems, involving the
flow of products from manufacturing plants or distribution centers through the
network of transportation to consumers. The function is very costly, especially for
the distribution industries. During the past decades, the significance of physical
distribution has increased enormously in both economic and managerial aspects.
Several firms have realized that an efficiency improvement and a cost control of
physical distribution would provide a potential for cost reduction and hence profit
improvement. However, the customer satisfaction cannot be disregarded, since the
focus on reducing cost often decreases the customer service level in many cases. A
compromised strategy which lowers the cost of physical distribution and accounts
for customer satisfaction must be sought.
In this research, an effort to improve distribution strategies and reduce the
distribution cost for the multi-product, multi-depot periodic distribution problem is
made, while maintaining the desired customer service levels. In industry practice,
distribution centers typically operate independently within their own territories
only. However, it may be beneficial to allow those distribution centers to operate
interdependently, particularly when the product supplies are limited at some
distribution centers. In such cases, the distributors may satisfy customers' requests
by delivering products from other distribution centers that hold more supplies. The
interaction of interdependent operations among distribution centers, which have not
been used in the industrial applications so far, is investigated. A mixed-integer
linear programming model is formulated to represent this multi-product, multi-depot
periodic distribution problem. Three tabu-search heuristics with different
applications of long-term memory are developed for solving the problem. The
performance of the heuristics is evaluated by comparing the solutions obtained with
the optimal solutions or lower bounds from the regular branch-and-bound method
and fast lower-bound finding techniques developed in this research. The heuristics
provide good-quality solutions in a much shorter time than those methods.
The performance of the three tabu-search heuristics is compared using statistical
experiments based on the randomized complete block design and Tukey's Honestly
Significant Differences. The experimental results show that the tabu-search
heuristic with the use of long-term memory for diversification process outperforms
the others. The heuristic is further applied to investigate the impact of
interdependent operations among distribution centers. The results reveal that the
interdependent operations among distribution centers provide a significant cost
saving over the independent operations among distribution centers, especially for
large-size problems. === Graduation date: 2004
|
|---|
