Strategic design for dynamic multi-zone truckload shipments: A study of OTOP agricultural products in Thailand

- We develop the dynamic multi-zone dispatching model in order to achieve the ability to change in dynamic environments.
- We propose the novel algorithm based on the intelligent water drop and variable neighbourhood search algorithms embedded on the dynamic programming approach.
- The proposed algorithm has been tested on various benchmarking dynamic multi-zone dispatching problems on preliminary experimental designs to examine parameters of interest before on-line assigning service areas to zones for the best seller OTOP food products and beverages.
- We simultaneously examine various zone participation statistics of the best so far scenario from the dynamic multi-zone truckload shipments via a computer simulation.
- The mechanism of the proposed procedures has been programmed in forms of an interactive computer technique.

The one-tambon-one-product (OTOP) project in Thailand has been successful according to statistical data from the Interior Ministry's Department of Community Development. However, there is a concern that dispatchers in the supporting trucking industry carry out both tasks of truckload assignment and driver scheduling. It is necessary to investigate more efficient and reliable alternatives to handle the growing and fluctuating quantities of inbound and outbound goods in each service area. This paper focuses on determining alternative ways for enhancing the efficiency of the transportation management system based on multi-zone dispatching (MZD) with the objective of minimising total load imbalances. This problem is considered to be a special case of the MZD but it is complex when compared to the conventional MZD, especially since planning horizons are considered and modified as a dynamic MZD (dMZD). In this paper an intelligent water drop optimisation (IWD) was used to solve the problem. In order to improve the solution quality, the novel two-phase approximated algorithm (2-PAA) with two additional processes from the variable neighbourhood and dynamic programming algorithms, were proposed. Experimental results with preferable settings showed the 2-PAA was computationally effective and feasible to generate dispatching alternatives for both static and dynamic problems. Using a set of real data, the simulation suggested a 3-period dMZD configuration for driver scheduling with preferable levels of tour length and imbalance statistics.