Premium
An adjustable sample average approximation algorithm for the stochastic production‐inventory‐routing problem
Author(s) -
Agra Agostinho,
Requejo Cristina,
Rodrigues Filipe
Publication year - 2018
Publication title -
networks
Language(s) - English
Resource type - Journals
SCImago Journal Rank - 0.977
H-Index - 64
eISSN - 1097-0037
pISSN - 0028-3045
DOI - 10.1002/net.21796
Subject(s) - heuristic , mathematical optimization , sample (material) , set (abstract data type) , routing (electronic design automation) , computer science , production (economics) , sample size determination , mathematics , vehicle routing problem , order (exchange) , algorithm , statistics , computer network , chemistry , chromatography , economics , macroeconomics , finance , programming language
We consider a stochastic single item production‐inventory‐routing problem with a single producer, multiple clients, and multiple vehicles. At the clients, demand is allowed to be backlogged incurring a penalty cost. Demands are considered uncertain. A recourse model is presented, and valid inequalities are introduced to enhance the model. A new general approach that explores the sample average approximation (SAA) method is introduced. In the sample average approximation method, several sample sets are generated and solved independently in order to obtain a set of candidate solutions. Then, the candidate solutions are tested on a larger sample, and the best solution is selected among the candidates. In contrast to this approach, called static, we propose an adjustable approach that explores the candidate solutions in order to identify common structures. Using that information, part of the first‐stage decision variables is fixed, and the resulting restricted problem is solved for a larger size sample. Several heuristic algorithms based on the mathematical model are considered within each approach. Computational tests based on randomly generated instances are conducted to test several variants of the two approaches. The results show that the new adjustable SAA heuristic performs better than the static one for most of the instances.