Recent advances in vehicle routing with stochastic demands: Bayesian learning for correlated demands and elementary branch-price-and-cut

We consider the vehicle routing problem with stochastic demands (VRPSD), a variant of well-known VRP in which are only revealed upon arrival at each customer. Motivated by significant recent progress on VRPSD research, we begin this paper summarizing key new results and methods for solving problem. In doing so, discuss main challenges associated under chance-constraint restocking-based perspectives. Once cover current state-of-the-art, introduce two major methodological contributions. First, present branch-price-and-cut (BP&C) algorithm optimal restocking. The method, is based pricing elementary routes, compares favorably previous algorithms allows solution several open benchmark instances. Second, develop demand model dealing correlated customer demands. central concept “external factor”, represents unknown covariates that affect all Bayesian-based, iterated learning procedure to refine our knowledge about external factor as revealed. This updated then used prescribe replenishment decisions correlation. Computational demonstrate efficiency BP&C method show cost savings above 10% may be achieved when restocking take account Lastly, motivate few research perspectives that, believe, should shape future VRPSD.