Optimal scheduling and sequencing for large-scale seeding operations

We consider a scheduling and sequencing optimization problem with application to an agricultural seeding operation. The problem is motivated by potential deployment of a recently proposed mobile seed refilling system in large-scale field operations. In this setting, mobile refiller vehicles are used to minimize any interruptions to seeding that result from the need to refill the seed tanks on the planters. A feature of this refilling system is that the two types of vehicles can move together during the seed transfer operation. The optimization task seeks to determine locations where seed transfer should be initiated, as well as an ordering of visits of refiller vehicles to such locations, such that the total downtime is minimized. In this paper, we consider such an optimization task for the case of multiple planters and a single refiller that, in turn, can refill itself at a single, fixed, off-field location. We rely on a stochastic solution strategy for solving this problem by applying the genetic algorithm to a binary integer programming model that represents the decision of whether to initiate refilling or not at a discrete set of possible locations along each planter path. The results are compared with a conventional static refilling scheme (with no mobile refiller), as well as a greedy (but generally suboptimal) algorithm that produces a unique solution to the selection and ordering of rendezvous locations. The computational analysis provides deep and useful insights to this optimization problem.