Adaptive genetic algorithm for solving sugarcane loading stations with multi-facility services problem

• This study addresses the problem of the sugarcane supply through loading station to the mill factory.
• There exist different types of multi-facility services in each sugarcane loading station.
• Designing a mathematical model for location–allocation problem with multi-facility services.
• Proposing an efficient adaptive genetic algorithm to solve location–allocation problem with multi-facility services.
• Outperforming well by the proposed adaptive genetic algorithm in large-sized problems.

This paper presents a computational tool for operational planning of sugarcane loading stations. The loading stations are used to facilitate the supply of sugarcane to a sugar mill, especially for small-sized sugarcane growers whose fields are located over 30 km away from the sugar mill. The objective of this research is to solve the problems involved in transportation planning and allocation of sugarcane from grower’s fields to the loading stations, and from the loading stations to the sugar mill. The decisions consist basically of the determination of proper loading station’s locations, the optimal type and number of transloaders in each loading station, and sugarcane field allocation to guarantee a continuous and uniform feeding of sugarcane to the sugar mill. Furthermore, this study is different from that of the general location problem in that it determines suitable types and the number of different multi-facility services (i.e. transloaders) at each of the sugarcane loading stations. In order to solve the problem of sugarcane loading stations with multi-facility services, we apply a mixed-integer programming model that can handle small-scale problems (less than 300 sugarcane fields or nodes). Additionally for large-scale problems, we present a comprehensive decision support system (DSS) with geographical information system (GIS) based on the proposed method, adaptive genetic algorithm (AGA), to solve this problem. Numerical experimental results of the AGA were compared with those obtained from the MPL/CPLEX, traditional genetic algorithm (GA), and the current practices in the sugar mill of our case study. The results demonstrated that the AGA is not only useful for reducing cost when compared to the traditional GA and the current practices, but also for efficient management of a sugarcane supply system. Furthermore, the method of this research should prove beneficial to other similar agro-food sectors in Thailand and around the world.