| dc.description.abstract | In today’s highly competitive market in Vietnam, each company has high
competition to attract and increase customer satisfaction. Their values and costs
play a crucial role in determining their positions in the market, so companies are
finding ways to optimize the delivery cost in the process of delivering products
to customers. The construction materials industry is a highly competitive industry
with many companies and stores having many different prices, suppliers struggle
with scheduling to deliver products promptly to clients. The research will be
expanded to the multi-depot VRP with the purpose is planning the optimal or
near-optimal routes to minimize the total cost of delivery. This study employs the
Mixed-integer Linear Programming mathematical model and heuristic Genetic
Algorithm to plan routes as well as ensure products are arranged according to
requirements. In this study, material stores deliver goods through a third-party
transport company, and the delivery planning considers some regulations such as
routing, capacity limitations, demand of customers, weight of products, or
distance. In this study, I compare the costs between two methods, called the MILP
method and the GA method, based on the results from the first run. The total cost
of delivery of the MILP method is 6,451,602 VND, while the GA method is
5,044,450 VND. The average difference between the two methods, calculated as
a percentage is 24.48%, representing the difference in total cost of these two
methods. The results show a clear difference between two research methods with
the same data set, but these two methods produce results based on the same
processing process. However, the results may be altered by changes in the values
of other factors that present these in the sensitivity analysis. In general, GA brings
a lower total cost of delivery than CPLEX, but in terms of aspects such as
accuracy and running time, CPLEX offers more advantages. How to minimize
the total cost of delivery in the MDVRP problem, by developing an effective
mathematical model and algorithm? | en_US |
