A decision support system for resilient vehicle route planning using mathematical modeling and artificial neural networks: a case study

Purpose The vehicle routing problem (VRP) is critical for the successful execution of logistics activities. However, there is strong evidence that efficiently solving the VRP is often complicated and requires more powerful – and possibly intelligent – support tools. In accordance with this necessity, the present study proposes a decision support system (DSS) applicable to the VRP, which includes both initial planning and replanning phases to support the real-time operations. Design/methodology/approach The proposed DSS lies at the basis of resilience thinking to provide a capacity to absorb and withstand the impact of disruptions, where resilience is connected with the factors of preparedness, flexibility and redundancy. These factors are approached in this study through a number of operational strategies in the reactive and proactive modes. The DSS includes a multi-layer perceptron neural network to predict changes that may arise in dynamic contexts, a modified k-means clustering algorithm to group customers with both static and dynamic attributes and two mixed-integer programming models to produce primary and alternate routing plans. Findings The research is motivated by the operational challenges faced by a collaborative networked clinical laboratory, which seeks to enhance efficiency and productivity in the daily management of medical sample collection and delivery through the implementation of increased automation. The findings reveal that centralized planning leads to heightened vulnerability in route planning and increased costs for replanning. Furthermore, the effectiveness of resilience-enhancement strategies varies based on the source and level of uncertainty. Originality/value The contributions of this paper are as follows: incorporating resilience thinking into the operational planning of logistics services, managing the decision-making of transport and collection companies through a DSS framework to ensure proper support to real-time operations, addressing the clustered VRP in a dynamic setting and adopting forecasting approaches to cover possible sources of dynamism.