Optimizing version release dates of research and development long-term processes

• Published in: European journal of operational research Vol. 259; no. 2; pp. 642 - 653
• Main Authors: Etgar, Ran, Gelbard, Roy, Cohen, Yuval
• Format: Journal Article
• Language: English
• Published: Elsevier B.V 01.06.2017
• Subjects: Genetic algorithm; Particle swarm; Release; Scheduling; Simulated annealing; Scheduling; Particle swarm; Genetic algorithm; Release; Simulated annealing
• ISSN: 0377-2217; 1872-6860
• DOI: 10.1016/j.ejor.2016.10.029

•We introduce a new practical problem optimizing version release dates and their feature contents.•We define a novel optimization model for adequately setting: release contents and release dates.•We offer a benchmark set of these new problems and their solutions for further research.•We validated our results against greedy heuristics and provided the results.•This is a seminal paper presenting a new practical problem to be tackled by future papers. This paper develops and compares several optimization approaches for the version planning and release problem. This problem is new, challenging for scholars and practitioners, and was not fully addressed in the OR literature. Version releases are part of a wide-spread phenomenon. Mobile phones, operating systems (e.g. MS-Windows) and digital printers are well known examples. However, version release can be found in many other product development fields, such as software products and games, and hardware versions (e.g. TV, screens, communication equipment etc.). In some fields (such as the automotive field) the version release is so well-established that it became an annual routine. An optimization formulation is developed for the total-value of a version-release policy throughout the relevant time-horizon. The novel formulation elements are release-features and release-dates. The value of each release is derived from the combination of features included in the specific released version, and the version release-dates. We developed several search techniques for solving this strongly NP-hard problem. We compared the results of (1) multiple particle swarm optimization (MPSO) (2) Genetic Algorithm (GA), (3) simulated annealing (SA), (4 & 5) two forms of greedy heuristics. A comprehensive computational experiment was performed. The study shows that GA and MPSO outperform the other methods. Moreover, for medium scale problems, GA better suits highly resource-constrained cases, while MPSO performs best for large scale problems disregarding the resource scarcity. This research may be a major reference point for future research on the version release problem.