A column generation approach for discrete lotsizing and scheduling problem on identical parallel machines

Hiroaki Arai, Susumu Morito, Jun Imaizumi

    Research output: Contribution to journalArticle

    4 Citations (Scopus)

    Abstract

    In this paper the Discrete Lotsizing and Scheduling Problem (DLSP) on identical parallel machines with setup times is considered. The problem is determining the sequence and size of production batches for multiple items. The objective is to find a minimal cost production schedule such that deterministic, dynamic demand is fulfilled with backlogging. The problem is formulated as an integer programming problem and reformulated as a Generalized Set Partitioning Problem (GSPP). We show that a problem which checks optimality of LP relaxation problem of GSPP is equivalent to a single item, namely, a discrete lotsizing scheduling problem on parallel machines. It means that GSPP is decomposed to subproblems, and each one is solvable with dynamic programming. We obtain the lower bound of the reformulated problem by solving a LP relaxation problem of GSPP optimally and present a column generation heuristic to solve GSPP, which optimizes the restricted GSSP composed with columns generated at that time. The quality of the solutions can be measured, since the heuristic generates both lower and upper bounds. We compare our algorithm and a heuristic based on Lagrangian relaxation proposed by Campbell. Computational results on a personal computer show that our heuristic is rather effective in terms of quality of the solutions. One of the advantages of the column generation method is that the setting of parameters is not required in contrast to Lagrange relaxation.

    Original languageEnglish
    Pages (from-to)69-76
    Number of pages8
    JournalJournal of Japan Industrial Management Association
    Volume55
    Issue number2
    Publication statusPublished - 2004

    Fingerprint

    Identical Parallel Machines
    Column Generation
    Lot Sizing
    Scheduling Problem
    Scheduling
    Integer programming
    Set Partitioning
    Dynamic programming
    Personal computers
    Heuristics
    Costs
    LP Relaxation
    Column generation
    Lot sizing
    Parallel machines
    Set partitioning
    Backlogging
    Lagrangian Relaxation
    Setup Times
    Parallel Machines

    Keywords

    • Column Generation
    • Lotsizing
    • Parallel Machines
    • Scheduling
    • Set Partitioning Problem
    • Setup Times

    ASJC Scopus subject areas

    • Industrial and Manufacturing Engineering
    • Engineering (miscellaneous)

    Cite this

    A column generation approach for discrete lotsizing and scheduling problem on identical parallel machines. / Arai, Hiroaki; Morito, Susumu; Imaizumi, Jun.

    In: Journal of Japan Industrial Management Association, Vol. 55, No. 2, 2004, p. 69-76.

    Research output: Contribution to journalArticle

    @article{55d21817ab314d97a6ea094e60aefca6,
    title = "A column generation approach for discrete lotsizing and scheduling problem on identical parallel machines",
    abstract = "In this paper the Discrete Lotsizing and Scheduling Problem (DLSP) on identical parallel machines with setup times is considered. The problem is determining the sequence and size of production batches for multiple items. The objective is to find a minimal cost production schedule such that deterministic, dynamic demand is fulfilled with backlogging. The problem is formulated as an integer programming problem and reformulated as a Generalized Set Partitioning Problem (GSPP). We show that a problem which checks optimality of LP relaxation problem of GSPP is equivalent to a single item, namely, a discrete lotsizing scheduling problem on parallel machines. It means that GSPP is decomposed to subproblems, and each one is solvable with dynamic programming. We obtain the lower bound of the reformulated problem by solving a LP relaxation problem of GSPP optimally and present a column generation heuristic to solve GSPP, which optimizes the restricted GSSP composed with columns generated at that time. The quality of the solutions can be measured, since the heuristic generates both lower and upper bounds. We compare our algorithm and a heuristic based on Lagrangian relaxation proposed by Campbell. Computational results on a personal computer show that our heuristic is rather effective in terms of quality of the solutions. One of the advantages of the column generation method is that the setting of parameters is not required in contrast to Lagrange relaxation.",
    keywords = "Column Generation, Lotsizing, Parallel Machines, Scheduling, Set Partitioning Problem, Setup Times",
    author = "Hiroaki Arai and Susumu Morito and Jun Imaizumi",
    year = "2004",
    language = "English",
    volume = "55",
    pages = "69--76",
    journal = "Journal of Japan Industrial Management Association",
    issn = "0386-4812",
    publisher = "Nihon Keikei Kogakkai",
    number = "2",

    }

    TY - JOUR

    T1 - A column generation approach for discrete lotsizing and scheduling problem on identical parallel machines

    AU - Arai, Hiroaki

    AU - Morito, Susumu

    AU - Imaizumi, Jun

    PY - 2004

    Y1 - 2004

    N2 - In this paper the Discrete Lotsizing and Scheduling Problem (DLSP) on identical parallel machines with setup times is considered. The problem is determining the sequence and size of production batches for multiple items. The objective is to find a minimal cost production schedule such that deterministic, dynamic demand is fulfilled with backlogging. The problem is formulated as an integer programming problem and reformulated as a Generalized Set Partitioning Problem (GSPP). We show that a problem which checks optimality of LP relaxation problem of GSPP is equivalent to a single item, namely, a discrete lotsizing scheduling problem on parallel machines. It means that GSPP is decomposed to subproblems, and each one is solvable with dynamic programming. We obtain the lower bound of the reformulated problem by solving a LP relaxation problem of GSPP optimally and present a column generation heuristic to solve GSPP, which optimizes the restricted GSSP composed with columns generated at that time. The quality of the solutions can be measured, since the heuristic generates both lower and upper bounds. We compare our algorithm and a heuristic based on Lagrangian relaxation proposed by Campbell. Computational results on a personal computer show that our heuristic is rather effective in terms of quality of the solutions. One of the advantages of the column generation method is that the setting of parameters is not required in contrast to Lagrange relaxation.

    AB - In this paper the Discrete Lotsizing and Scheduling Problem (DLSP) on identical parallel machines with setup times is considered. The problem is determining the sequence and size of production batches for multiple items. The objective is to find a minimal cost production schedule such that deterministic, dynamic demand is fulfilled with backlogging. The problem is formulated as an integer programming problem and reformulated as a Generalized Set Partitioning Problem (GSPP). We show that a problem which checks optimality of LP relaxation problem of GSPP is equivalent to a single item, namely, a discrete lotsizing scheduling problem on parallel machines. It means that GSPP is decomposed to subproblems, and each one is solvable with dynamic programming. We obtain the lower bound of the reformulated problem by solving a LP relaxation problem of GSPP optimally and present a column generation heuristic to solve GSPP, which optimizes the restricted GSSP composed with columns generated at that time. The quality of the solutions can be measured, since the heuristic generates both lower and upper bounds. We compare our algorithm and a heuristic based on Lagrangian relaxation proposed by Campbell. Computational results on a personal computer show that our heuristic is rather effective in terms of quality of the solutions. One of the advantages of the column generation method is that the setting of parameters is not required in contrast to Lagrange relaxation.

    KW - Column Generation

    KW - Lotsizing

    KW - Parallel Machines

    KW - Scheduling

    KW - Set Partitioning Problem

    KW - Setup Times

    UR - http://www.scopus.com/inward/record.url?scp=22744448287&partnerID=8YFLogxK

    UR - http://www.scopus.com/inward/citedby.url?scp=22744448287&partnerID=8YFLogxK

    M3 - Article

    AN - SCOPUS:22744448287

    VL - 55

    SP - 69

    EP - 76

    JO - Journal of Japan Industrial Management Association

    JF - Journal of Japan Industrial Management Association

    SN - 0386-4812

    IS - 2

    ER -