Stochastic Joint Replenishment Planning Optimization with an Iterated Local Search - Archive ouverte HAL Accéder directement au contenu
Communication Dans Un Congrès Année : 2019

Stochastic Joint Replenishment Planning Optimization with an Iterated Local Search

Résumé

In this extended abstract, we discuss optimized operations of supply chains dealing with multiple items. Examples of such supply chains are omni-channel retailers, supermarkets and service supply chains, etc. Whenever more than one item is ordered from one supplier or they share the same transportation vehicle, first order interactions take place between the items. First order interaction can be due to a joint (common) fixed cost in addition to individual fixed costs. When the demands are known with certainty the problem is a deterministic Joint Replenishment Problem (JRP). On the other hand, if the demands are considered stochastic, it is called the stochastic joint replenishment problem. Other costs involved are the inventory holding costs and the shortage costs. The periodic deterministic JRP is strongly NP-hard for infinite horizon and NP-hard for finite horizon (Cohen-Hillel and Yedidsion, 2018). No such proof is available yet for the stochastic JRP. However, it is commonly regarded as very hard due to its huge solution space. Structural results for the stochastic JRP show that the optimal solution cannot be expressed as a simple replenishment policy even for the twoitem case (Ignall, 1969). Therefore, many authors have assumed some simple structures such as (s; c; S) (Johansen and Melchiors, 2003), (s; c; d; S) (Feng et al., 2015) and Periodic Review(s; S) (Viswanathan, 1997), to facilitate practical implementation (S = Order up to level; s = Reorder level; c, d = Can order levels). Even if the adopted method follows one of the above policies, any change in problem parameters requires re-evaluation of policy parameters. Such a process is also time consuming. The real-world applications of stochastic JRP are numerous. Usually, for the stochastic JRP, replenishment planning solutions use heuristics that are independent for each item under the JRP or follow one of the joint policies mentioned in the previous paragraph. These methods are not optimal for the global problem. Since the stochastic JRP are very hard to solve optimally, we propose to use a stochastic local search (SLS) approach to find good quality solutions in reasonable time. The ordering decisions are then implemented under a rolling horizon planning environment. The quality of a solution obtained using our proposed methodology is always better than what can be obtained using item level heuristics because, it uses the item level heuristics for generating the initial solution and then conducts further improvements. We also conduct numerical studies to attest its performance against the joint policies.
Fichier principal
Vignette du fichier
SLS2019_paper_6.pdf (238.1 Ko) Télécharger le fichier
Origine : Fichiers produits par l'(les) auteur(s)

Dates et versions

hal-02430774 , version 1 (11-12-2023)

Identifiants

  • HAL Id : hal-02430774 , version 1

Citer

Rabin Kumar Sahu, Nadarajen Veerapen. Stochastic Joint Replenishment Planning Optimization with an Iterated Local Search. SLS - International Workshop on Stochastic Local Search Algorithms, Sep 2019, Villeneuve d Ascq, France. ⟨hal-02430774⟩
35 Consultations
5 Téléchargements

Partager

Gmail Facebook X LinkedIn More